* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-14 10:09 Greg KH
2016-09-14 17:36 ` Mark Rutland
` (3 more replies)
0 siblings, 4 replies; 84+ messages in thread
From: Greg KH @ 2016-09-14 10:09 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
"'tis but a scratch" - The Black Knight
Hi all,
As many of you know, I, and lots of other developers, have been working
on a new application level protocol for the Unipro hardware transport
called "Greybus". It has shown up in numerous demos from Google for the
Ara Phone platform, and an older version of the codebase is currently
shipping in a phone from Motorola any-day-now. There is a
semi-up-to-date version of the protocol specification available at:
https://github.com/projectara/greybus-spec
A newer version should be available soon, matching the code
implementation here, as soon as I work through a few final details with
some lawyers on the wording of the license.
Many other software packages to help out with greybus devices are on
that github project page (manfest generator, emulator, firmware
examples, etc.), if people are interested.
Given that it's never a good idea to keep subsystems out of the mainline
kernel, I've put together this pull request that adds the greybus driver
layer to drivers/greybus/. Because this was 2 1/2 years of work, with
many many developers contributing, I didn't want to flatten all of their
effort into a few small patches, as that wouldn't be very fair. So I've
built a git tree with all of the changes going back to the first commit,
and merged it into the kernel tree, just like btrfs was merged into the
kernel.
The protocol design, and implementation, was a group effort, and it would
not have turned out well at all without the contributions of all of
these talented developers.
I'll send out a follow-up set of "simple" patches that just add the
files to the kernel tree, to give people an idea of the code involved.
Overall, it's a tiny stand-alone driver subsystem, only 37k lines, that
implements a protocol which allows for "generic" cameras, audio devices,
and other class type devices, as well as a bridged "physical" layer
protocol to talk to serial, spi, uart, pwm, gpio, i2c, and even USB host
controllers. Included here is a USB bridge host controller driver that
interacts with a USB device that converts USB data to Unipro data,
allowing any system to talk to a Unipro platform (no special SoC
interface required.)
Unless people point out some major problems with this, I'd like to get
it merged into 4.9-rc1.
The shortlog and diffstat can be seen below.
If anyone has any questions about this, please let us know.
thanks,
greg k-h
---------------
The following changes since commit c6935931c1894ff857616ff8549b61236a19148f:
Linux 4.8-rc5 (2016-09-04 14:31:46 -0700)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc.git/ greybus
for you to fetch changes up to f86bfc90a401681838866b5f6826d86cc4c7010c:
greybus: remove CONFIG_PM_RUNTIME from kernel_ver.h (2016-09-09 13:27:19 +0200)
----------------------------------------------------------------
Akash Choudhari (1):
greybus: Revert "Added a sysfs entry to power down the SVC"
Alex Elder (345):
greybus: uart-gb: improve minor device number error checking
greybus: uart-gb: a few minor bug fixes
greybus: switch to the term "manifest"
greybus: define struct greybus_manifest
greybus: call put_device() on error
greybus: interpret descriptor type properly
greybus: validate descriptor sizes
greybus: fix manifest parsing size bug
greybus: simple fixes
greybus: rename struct greybus_device
greybus: quick fix for sysfs serial number
greybus: get rid of a weird idiom
greybus: embed workqueue structure in struct gbuf
greybus: encapsulate URB status checking
greybus: some more renames
greybus: update a few header file comments
greybus: fix document version number
greybus: fix the guard in "greybus_manifest.h"
greybus: descriptor type is 1 byte
greybus: fix version check
greybus: reorder greybus_svc_in() arguments
greybus: kill struct gmod_cport
greybus: clarify device vs module
greybus: descriptor type updates
greybus: isolate greybus module code
greybus: define greybus interface abstraction
greybus: define greybus function abstraction
greybus: introduce a connection abstraction
greybus: introduce an operation abstraction
greybus: start improving manifest parsing
greybus: manifest interface descriptor parsing
greybus: manifest cport descriptor parsing
greybus: kill off old manifest code
greybus: fix connection header declarations
greybus: October 1 updates
greybus: create host device cport id map
greybus: allocate connection host cport id
greybus: get rid of functions now...
greybus: record connection protocol
greybus: fix some hasty bugs
greybus: fix greybus_class_type symbol names
greybus: use ida for cport id allocation
greybus: fix module setup
greybus: gbuf: have caller set actual_length
greybus: specify type when creating an operation
greybus: add a connection->private field
greybus: define gb_connection_err()
greybus: fix two misnamed functions
greybus: bury some dead code
greybus: add bg_hd_connection_find()
greybus: look up connection for recevied messages
greybus: allocate gbufs using the connection
greybus: use alloc_gbuf_data() for both directions
greybus: kill __alloc_gbuf()
greybus: get rid of gbuf->hdpriv
greybus: don't assume subdevs are valid
greybus: unlock gbuf mutex on free
greybus: unlock hd mutex on free
greybus: don't restrict input buffer size
greybus: zero all data buffers
greybus: drop hd arg from submit_gbuf method
greybus: stop init_subdevs stuff
greybus: add response buffer to an operation
greybus: add gb_operation_find()
greybus: let operation layer examine incoming data
greybus: move receive handling to operation layer
greybus: add device initialization
greybus: kill off gbuf work queue
greybus: kill old cport handlers
greybus: add i2c driver
greybus: initial operations-based GPIO driver
greybus: add write retry support for i2c
greybus: update gbuf status for completion handlers
greybus: report gbuf errors
greybus: properly drop device reference
greybus: create a slab cache for operations
greybus: drop gb_* device fields from gb_module
greybus: remove cports and strings from gb_module
greybus: move ap_disconnect()
greybus: leverage ap_disconnect() in ap_probe()
greybus: fix op_cycle logic
greybus: stash power supply pointer in connection
greybus: improve module cleanup code
greybus: time out operation requests
greybus: define operation_cancel()
greybus: define connection state
greybus: make svc_set_route_send() public
greybus: only initialize interfaces when up
greybus: set route before sending packets
greybus: update AP id service message
greybus: fix repeated input errors
greybus: gpio: kill gpio_controller->gpio
greybus: embed the i2c adapter struct
greybus: begin abstracting connection operations
greybus: drop the cport id on error
greybus: use protocol_id for numeric values
greybus: introduce protocol abstraction
greybus: fix a bug in gb_operation_gbuf_complete()
greybus: drop a redundant assignment
greybus: make _gb_protocol_find() static
greybus: control printing message
greybus: improve some error messages
greybus: identify protocol by id *and* version
greybus: count rather than list protocol users
greybus: order the protocols list
greybus: register preallocated protocols
greybus: move methods into protocol
greybus: add an incoming request receive method
greybus: kill test_sink.c
greybus: kill old cport handler code
greybus: get rid of gbuf->actual_length
greybus: record gbuf->operation
greybus: remove gbuf->context
greybus: get rid of greybus_gbuf_finished()
greybus: move gb_operation_gbuf_complete()
greybus: kill gbuf->complete
greybus: don't assume PAGE_SIZE for URB size
greybus: fix request timeout bug
greybus: move timeout out of gb_operation_insert()
greybus: simplify pending operations tracking
greybus: op_cycle doesn't need to be atomic
greybus: pass operation type on request receive
greybus: move operation timeout teardown
greybus: delete some lines in "greybus.h"
greybus: kill gb_operation_gbuf_complete()
greybus: clean up gb_connection_operation_recv()
greybus: reference count operations
greybus: record a gbuf's destination CPort id
greybus: use gbuf's destination cport id
greybus: record the host device in a gbuf
greybus: use a simple list of hd connections
greybus: get rid of cport_id_map_lock
greybus: kill greybus_{get,put}_gbuf()
greybus: kill gbuf->kref
greybus: move the definition of struct gbuf
greybus: define struct gb_message
greybus: start using struct gb_message
greybus: move gbuf initialization to caller
greybus: use null gbuf->transfer_buffer
greybus: embed gbufs into operation message structure
greybus: kill the gbuf slab cache
greybus: kill off the last of gbuf.c
greybus: fix an allocation flag bug
greybus: prepend cport byte for all gbufs
greybus: improve data buffer alignment
greybus: fill in destination data at send time
greybus: allocate space without gbufs
greybus: free space without gbufs
greybus: cancel buffers via magic cookie
greybus: stash hd as context for all URBs
greybus: send buffers without gbufs
greybus: rework message initialization
greybus: send messages, not gbufs
greybus: cancel messages, not gbufs
greybus: rework receve handling
greybus: kill the last gbuf remnants
greybus: stop storing hd in message
greybus: stop storing dest_cport_id in message
greybus: pass gfp_flags for message allocation
greybus: explicitly mark cookies as opaque
greybus: tidy up svc_in_callback() and cport_in_callback()
greybus: get rid of message status
greybus: use "operation_id" for certain values
greybus: refactor gb_connection_recv()
greybus: fix battery_operation()
greybus: fix uart request_operation()
greybus: fix vibrator request_operation()
greybus: get rid of uart request_operation()
greybus: distinguish incoming from outgoing requests
greybus: send operation result in response message header
greybus: define gb_operation_status_map()
greybus: remove status from all responses
greybus: drop a now-empty structure
greybus: fix a timeout race
greybus: complete overflow responses
greybus: have greybus allocate its own buffers
greybus: rename message buffer fields
greybus: dynamically allocate requests and responses
greybus: embed message buffer into message structure
greybus: define greybus_data_sent()
greybus: rename greybus_cport_in()
greybus: use errno for operation result
greybus: abandon incoming requests for now
greybus: handle data send errors in workqueue
greybus: add a reference to pending operations
greybus: minor tweak in gb_connection_recv_response()
greybus: cancel operation on timeout
greybus: cancel whole operation on interrupt
greybus: kill gb_operation_wait()
greybus: rework synchronous operation completion
greybus: encapsulate operation result access
greybus: first operation error prevails
greybus: use special operation result valus
greybus: fix some error codes
greybus: enforce receive buffer size
greybus: update operation result atomically
greybus: ignore a null cookie when canceling buffer
greybus: protect cookie with a mutex
greybus: use outgoing flag when creating operation
greybus: enforce max representable message size
greybus: define -EILSEQ to mean implementation error
greybus: renumber operation result values
greybus: drop gfp_mask from gb_message_send()
greybus: always drop reference in gb_operation_work()
greybus: move copy of incoming request data
greybus: short message is OK for errors
greybus: pass result in gb_connection_recv_response()
greybus: enforce non-zero operation type requirement
greybus: use operation type 0 to signal incoming data
greybus: introduce gb_operation_message_init()
greybus: enforce a buffer headroom maximum size
greybus: create a slab cache for simple messages
greybus: set result in gb_operation_response_send()
greybus: activate incoming request handling
greybus: introduce gb_operation_errno_map()
greybus: send operation response messages
greybus: use little-endian in PWM requests
greybus: fix a bug in gb_operation_sync()
greybus: define the invalid operation type symbolically
greybus: don't use 0 as an operation id
greybus: get rid of pending operations list
greybus: make op_cycle atomic (again)
greybus: introduce gb_operation_request_send_sync()
greybus: fix an error message
greybus: set up connection->private properly
greybus: don't let i2c code assume non-null payload pointer
greybus: only record message payload size
greybus: use null pointer for empty payload
greybus: record type in operation structure
greybus: define GB_OP_NONEXISTENT
greybus: ENODEV can be an expected error too
greybus: switch cport id used for sends
greybus: add Linaro copyrights
greybus: clean up some small messes
greybus: get rid of {conceal,reveal}_urb()
greybus: manifest: use size_t for a size variable
greybus: es1: test apb1_log_task safely
greybus: es2: test apb1_log_task safely
greybus: reduce the ranting
greybus: bundle: use kstrdup() for state file
greybus: battery: free struct on error in caller
greybus: battery: use feature tag rather than kernel version
greybus: eliminate extra response flag definitions
greybus: loopback: fix the type attribute check
greybus: loopback: return the right error value
greybus: loopback: fix an incorrect comment
greybus: loopback: symbolically define max wait time
greybus: loopback: define loopback functions symbolically
greybus: loopback: support module-initiated requests
greybus: endo: rename gb_svc
greybus: endo: pass endo_id to gb_endo_create()
greybus: core: return error code when creating endo
greybus: core: return error code when creating host device
greybus: include "gpbridge.h" from "greybus.h"
greybus: rename "gpbridge.h"
greybus: endo: encapsulate computing the max interface id
greybus: endo: rework some attributes
greybus: endo: record AP interface id
greybus: core: don't set up endo until host device is initialized
greybus: add copyright statements
greybus: update copyrights
greybus: introduce SVC protocol
greybus: greybus_protocols: fix guard tag
greybus: add documentation for Endo sysfs files
greybus: update sysfs documentation files
greybus: endo: drop redundant prefixes from sysfs basenames
greybus: endo: delete "0xXXXX" portion in sysfs "endo" directory
greybus: svc: driver is basic to Greybus (not GP Bridge)
greybus: connection: make gb_connection_hd_find() private
greybus: uart: properly interpret receive data size
greybus: core: rename greybus_deregister()
greybus: manifest: clean up a few pr_err() calls
greybus: drop some unnecessary headers
greybus: tag core init and exit functions
greybus: endo: define endo_init() and endo_exit()
greybus: endo: clean up id assignment code
greybus: manifest: really minor cleanups
greybus: manifest: use bundle's embedded interface pointer
greybus: manifest: rework cport parsing
greybus: manifest: rework bundle parsing
greybus: bundle: check for duplicate bundle ids
greybus: connection: check for duplicate cport ids
greybus: connection: remove extra kfree() call
greybus: connection: drop unneeded gb_protocol_put() calls
greybus: connection: un-abstract host cport id allocation
greybus: manifest: clean up properly when parsing cports
greybus: bundle: refactor gb_bundle_find()
greybus: bundle: fix gb_bundle_destroy()
greybus: manifest: clean up properly when parsing bundles
greybus: introduce cport_id_valid()
greybus: esX: encapsulate packing cport id into header
greybus: rename HOST_DEV_CPORT_ID_MAX
greybus: reserve host cport id 0
greybus: esX: use one byte to encode cport ids in header
greybus: interface: declare gb_interface_destroy()
greybus: looback: fix two typos
greybus: loopback: drop unneeded casts for void pointers
greybus: loopback: use U64_MAX for initialization
greybus: loopback: use u32 for stats update
greybus: loopback: record 32-bit min and max
greybus: loopback: use a 32-bit count
greybus: loopback: error is an unsigned attribute
greybus: loopback: all read-only attributes are unsigned
greybus: loopback: use separate attribute macro for average
greybus: loopback: compute average stats on demand only
greybus: loopback: fix connection cleanup paths
greybus: kernel_ver.h: define U32_MAX and U64_MAX
greybus: db3-platform: fix code that fetches reset GPIO
greybus: db3-platform: fix some typos
greybus: db3-platform: get rid of redundant gpio tests
greybus: es2: rename es2 data structures
greybus: es2: move APB log task into the es2_ap_dev struct
greybus: es2: move logging fifo into es2 struct
greybus: es2: move logging dentries into es2 struct
greybus: es2: don't assume just one AP bridge
greybus: connection: verify disabled when destroyed
greybus: add operation traces
greybus: tracing: define events using macros
greybus: tracing: eliminate "location" comments
greybus: tracing: refine comments
greybus: tracing: fix a bad tracepoint
greybus: tracing: fix hd traces
greybus: tracing: add module traces
greybus: tracing: define interface traces
greybus: report right error value
greybus: fix unbalanced mutex
greybus: eliminate unneeded null check
greybus: fix pointless null check
greybus: use PTR_ERR_OR_ZERO()
greybus: drop a bogus semicolon
greybus: tracing: fix "make check" warnings
greybus: define BUNDLE_ID_NONE
greybus: tracing: fix host device num_cports
greybus: tracing: reorder trace definitions
greybus: tracing: fix module num_interfaces
greybus: tracing: assign "parent" id first
greybus: tracing: add interface mode_switch
greybus: tracing: define bundle traces
greybus: tracing: define connection traces
greybus: tracing: add timing traces
greybus: tracing: fix message traces
greybus: don't key on an enumerated constant
greybus: tracing: drop "greybus" prefix
greybus: get rid of a compile warning
greybus: use memdup_user()
Alexandre Bailon (29):
greybus: i2c-gb: fix bad message size in gb_i2c
greybus: protocol.c: fix a kernel panic caused by __gb_protocol_register
greybus: gpio.c: fix a bad irq number
greybus: Export greybus debugfs folder
greybus: Dump log from APB1
greybus: Add loopback protocol
greybus: es1.c: Don't use magic value for USB control request
greybus: es2.c: Don't use magic value for USB control request
greybus: es2.c: create dedicated struct for cport_in and cport_out
greybus: es2.c: Increase the number of bulk endpoints
greybus: es2.c: add a control request for endpoints mapping
greybus: es1/es2: set transfer flag to send a zero-length packet
greybus: Greybus driver: add a new callbacks to driver
greybus: es2: update the bulk_ep_set value accepted by map_to_cpor_ep()
greybus: es2: rename misnamed variables and methods
greybus: es2: change (un)map methods to static
greybus: es2: add some documentation about endpoints mapping
greybus: loopback: fix invalid response size
greybus: gb_loopback: Fix throughput calculations
greybus: loopback: Fix calculations error for ping transfers
greybus: loopback: Fix incoherency in calculations in the case of error
greybus: loopback: Fix throughput calculations
greybus: loopback: Fix averaging
greybus: loopback: Fix warning on 32-bit build
greybus: loopback: round closest the sixth decimal
greybus: loopback: Fix broken synchonous test
greybus: loopback: Fix broken loopback min values
greybus: es2: Add a new bulk in endpoint for APBridgeA RPC
greybus: es2: Implement APBridgeA RPC (ARPC)
Ann Chen (2):
greybus: vibrator: integrate runtime pm
greybus: Add workqueue to handle vibrator timeout
Axel Haslam (24):
greybus: greybus/loopback: remove mask attribute
greybus: greybus/loopback: register a struct device.
greybus: loopback_test: make output to csv file a parameter option
greybus: loopback_test: Decrease the max number of devices
greybus: greybus/loopback: Add reserved fields to transfer request
greybus: Notify user space only when the test finished.
greybus: Loopback_test: use poll instead of inotify
greybus: loopback_test: Use timeout argument
greybus: loopback_test: handle SIGINT signal
greybus: uart: Update line coding settings only when needed
greybus: uart fix missing negation on DTR setting
greybus: Fix loopback app after rename to gpphy
greybus: uart: Handle CRTSCTS flag in termios
greybus: uart: Implement dtr_rts callback.
greybus: uart: Use a fifo to send data to the modules
greybus: uart: Add credits based tracking for transmit path
greybus: uart: Implement flush_buffer
greybus: uart: wait for credits on shutdown
greybus: uart: Add runtime pm support
greybus: gpio: Add runtime_pm suppourt
greybus: pwm: Add runtime_pm support
greybus: spi: Add runtime_pm support
greybus: loopback: add runtime pm support
greybus: gbphy: fix compile error with CONFIG_PM_RUNTIME disabled
Bartosz Golaszewski (4):
greybus: loopback: allocate a response even for a 0-byte request
greybus: legacy: fix a null pointer dereference
greybus: audio_manager: add missing header
greybus: pm: add error handling to bundle activation
Bill Pemberton (3):
greybus: gb-vibrator: remove useless if in timeout_store()
greybus: Add FIXME warnings for possible NULL dereferences
greybus: gpio-gb: remove checks for negative offset variable
Bryan O'Donoghue (132):
greybus: Support building greybus on host PC
greybus: uart: Move UART protocol structs/defines to greybus_protocols.h
greybus: uart: Tidy naming convention to more closely match spec
greybus: uart: Reduce UART count from 255 to 16
greybus: uart: Update uart.c to register tty ports
greybus: uart: send_data should return size or error
greybus: uart: kmalloc for send_data once only
greybus: uart: Add gb_uart_request_recv for receiving async UART data
greybus: uart: Remove magic numbers make struct gb_tty variable names consistent
greybus: uart: Latch modem control signals for tciomget
greybus: greybus/uart: Fix sparse warning
greybus: greybus/uart: Relocate UART parity/overrun/framing/break signals
greybus: greybus/uart: Add support for UART error signals
greybus: greybus/uart: Update UART to reflect field size changes
greybus: greybus/loopback: Move loopback operation definitions
greybus: greybus/loopback: remove magic number in state-machine
greybus: greybus/loopback: remove spurious pr_err in sysfs store
greybus: greybus/loopback: make loopback type input equivalent to protocol type
greybus: greybus/loopback: add sink to loopback protocol
greybus: greybus/loopback: truncate maximum loop data to link size
greybus: greybus/loopback: update throughput metrics to improve granularity
greybus: greybus/loopback: fix broken tabs in greybus_protocols.h
greybus: greybus/loopback: fix 64bit printf format error
greybus: greybus/uart: fix typo in defintion
greybus: greybus/manifest: reserve control connection cport/bundle ids
greybus: greybus/manifest: convert pr_err to dev_err
greybus: greybus/loopback: timestamp seeding should not drop metrics
greybus: greybus/loopback: update loopback operation description comment
greybus: greybus/loopback: run operations a set number of times
greybus: greybus/loopback: add commentary to sysfs variables
greybus: greybus/loopback: rename frequency to requests
greybus: greybus/loopback: convert loopback wake/sleep to a waitqueue
greybus: greybus/loopback: remove redundant timestamping
greybus: greybus/loopback: make loopback code thread safe
greybus: greybus/loopback: provide interface to read all latency data-points
greybus: greybus/loopback: warn user if kfifo cannot log all data
greybus: greybus/connection: fix jump label on device_add failure
greybus: greybus/connection: add a timestamp kfifo to track connection handoff
greybus: greybus/es-drivers: add outbound timestamp to connection
greybus: greybus/loopback: functionally decompose calculation of turn-around times
greybus: greybus/loopback: handle timestamp roll-over
greybus: greybus/interface: change typo replicable => replaceable
greybus: greybus/loopback: add ability to graph greybus latency
greybus: greybus/loopback: convert sample report interface to debugfs
greybus: greybus/loopback: support synchronized tests over multiple cports
greybus: greybus/loopback: add gb_loopback_nsec_to_usec_latency
greybus: greybus/loopback: functionally decompose gb_loopback_calc_latency
greybus: greybus/loopback: graph round-trip time for all threads
greybus: greybus/loopback: add bitmask of connections to include in test
greybus: greybus/loopback: initialized ms_wait negate warning
greybus: greybus/loopback: remove checkpatch error causing macro
greybus: greybus/loopback: add module level sys/debug fs data points
greybus: greybus/loopback: fix typo in comment
greybus: greybus/loopback: make sure to list_del on connection_exit
greybus: greybus/loopback: ensure count decrement happens before sysfs_remove_groups
greybus: greybus/loopback: hold a coarse lock while init/exit run
greybus: greybus/loopback: exit kfree after mutex release
greybus: greybus/loopback: ensure debugfs entires are cleaned up on exit
greybus: greybus/loopback: ensure sysfs entries are cleaned up on exit
greybus: greybus/loopback: convert pr_info to dev_err
greybus: greybus/loopback: add response len to loopback protocol
greybus: greybus/loopback: sort list of connections for masking purposes
greybus: greybus/loopback: add gb_loopback_operation_sync
greybus: greybus/loopback: use dev_name to populate sysfsname
greybus: greybus/tracepoints: add tracepoint definitions
greybus: greybus/operation, core: hook tracepoints into message opertions
greybus: greybus/loopback: drop dependency on internal timestamps
greybus: greybus/tracepoints: add tracepoints for host_device tx/rx
greybus: greybus/es1, es2: hook tracepoints to hardware send/recv operations
greybus: greybus/loopback: move sysfs entries to /sys/bus/greybus/devices/endo0
greybus: greybus/loopback: masked out threads should sleep
greybus: greybus/loopback: drop redundant endo0 string from debugfs entry name
greybus: greybus/loopback: remove module specific identifier from debugfs name
greybus: greybus/loopback: add reserved fields to track firmware latencies
greybus: greybus/es1,es2: add USB vendor command to timestamp
greybus: greybus/connection: add latency tag enable/disable callbacks
greybus: greybus/loopback: send command to APBridge to tag throughput
greybus: greybus/loopback: add tracker variables to hold firmware timestamps
greybus: greybus/loopback: capture and present firmware supplied latencies
greybus: greybus/manifest: fix the placement of arguments to pr_err
greybus: greybus/audio-pcm: fix use of variable unitialized
greybus: greybus/loopback: Relax locking during loopback operations
greybus: greybus/loopback: Move latency_ts initialization to transfer routine
greybus: greybus/loopback: Drop unused timeval variables
greybus: greybus/loopback: drop bus aggregate calculation
greybus: greybus/loopback: Convert cross-thread mutex to spinlock while relaxing connect locks
greybus: greybus/loopback: Add asynchronous bi-directional support
greybus: greybus/loopback: Convert thread delay to microseconds
greybus: greybus/loopback: Ensure we reset stats once and once only
greybus: greybus/loopback: Drop NULL check on container_of pointer
greybus: greybus/loopback: Retrun -ENOMEM if operation allocation fails
greybus: greybus/loopback: Wait for all async operations to complete on exit
greybus: greybus/loopback: Add asynchronous backoff
greybus: svc: Change GB_SVC_TYPE_LINK_CONFIG to 0x10
greybus: greybus/timesync: Add Control and SVC TimeSync command/response data definitions
greybus: greybus/timesync: Add Control and SVC protocol TimeSync operation definitions
greybus: greybus/control: Add TimeSync control commands
greybus: Ensure gb->mutex is held when adding timer
greybus: Fixup __u64, __u32 to __le64, __le32 in timesync declarations
greybus: greybus_trace.h: Fix dodgy indentation
greybus: greybus_protocols.h Add SVC_TIMESYNC_WAKE_PINS_ACQUIRE/RELEASE
greybus: greybus_protocols.h Add SVC_TIMESYNC_PING
greybus: arche-platform: Rework platform/wd-line state transition logic
greybus: svc: Add TimeSync SVC commands
greybus: control: Add TimeSync get-last-event logic
greybus: control: Drop unused parameter from timesync_authoritative
greybus: hd: Add TimeSync APBridge commands
greybus: interface: Extract and store Interface feature byte
greybus: es2.c: Declare local __le64 not u64
greybus: Fix unbalanced irq_enable() backtrace
greybus: timesync: Add timesync core driver
greybus: timesync: Bind TimeSync into Greybus
greybus: tracepoints: Add standard Linux tracepoint for TimeSync event
greybus: timesync: Add gb_timesync_frame_time_to_timespec()
greybus: timesync: Add debugfs entry to display frame-ping in ktime
greybus: uart: Fix minor number leak
greybus: timesync: Do 64 bit divisions in a 32 friendly way
greybus: timesync: Do not hold mutex on cancel_delayed_work_sync
greybus: timesync: Fix transitions to the INACTIVE state
greybus: timesync: Rework timesync removal serialization logic
greybus: timesync: Enforce TimeSync locks as subordinate to Interface locks
greybus: timesync: Initialize the timesync ping fields to zero
greybus: timesync: Make printout consitent with other greybus messages
greybus: timesync/pm: Make synchronous call to restore FrameTime
greybus: greybus_protocols.h/es2: Ensure __le32 is used not u32
greybus: greybus_protocols.h: convert __u32 to __le32
greybus: timesync: Bugfix ping should not result in -EAGAIN
greybus: timesync: reduce initial startup time
greybus: timesync: probe shouldn't complete until FrameTime sync does
greybus: timesync: Implement a retry mechanism
greybus: timesync: Ensure parallel synchronous calls succeed
greybus: timesync: Printout strobe count on sync failure
David Lin (40):
greybus: arche-platform: fix incorrect gpio variable type
greybus: greybus_trace: Fix broken greybus ftrace
greybus: svc: add Interface power measurements support
greybus: arche-platform: Power-off unipro subsystem upon suspend
greybus: svc watchdog: Disable watchdog upon entering suspend
greybus: es2: Fix apb_log null pointer exception
greybus: svc: add AP power measurements debugfs support
greybus: svc: clean up gb_svc struct for pwrmon
greybus: svc: free pwrmon_rails memory upon exit
greybus: operation: fix an inconsistent indent
greybus: legacy: remove legacy driver support
greybus: legacy: remove protocol.o from the makefile
greybus: svc: pwrmon: validate svc protocol op status when getting rail names
greybus: audio: remove the unnecessary return statement
greybus: control: add bundle suspend and resume preparations
greybus: control: add bundle deactivate and activate operation
greybus: control: add interface suspend prepare operation
greybus: control: add interface deactivate prepare operation
greybus: control: add interface hibernate abort operation
greybus: svc: add interface resume operation
greybus: svc: add power mode call for link hibernation
greybus: interface: implement unipro link hibernate call
greybus: control: add connection suspend and resume calls
greybus: interface: send deactivate prepare when interface is disabled
greybus: bundle: add activate and deactivate
greybus: interface: add runtime pm support
greybus: bundle: add runtime pm support
greybus: audio: add runtime pm support
greybus: camera: add runtime pm support
greybus: gbphy: add gbphy runtime pm support
greybus: i2c: add runtime pm support
greybus: hd: arche-platform: implement greybus shutdown
greybus: timesync: do not print frametime by default
greybus: operation: print id when synchronous operation timeout
greybus: svc_watchdog: Add sysfs file to change the behavior of bite
greybus: audio: add runtime pm to enumerated control and DAPM widget
greybus: interface: delete control device upon enable failure
greybus: svc_watchdog: use schedule_delayed_work helper
greybus: interface: fix timesync registration sequencing
greybus: light: fix incorrect led attribute files allocation
Dinko Mironov (1):
greybus: audio: Fix incorrect counting of 'ida'
Eli Sennesh (5):
greybus: firmware and svc: detect the difference between ES2 and ES3 chips
greybus: firmware/bootrom: debug output from bootrom protocol
greybus: operation: rate-limit dev_err printing on the receive path
greybus: update UniPro Set Interface Power Mode operation to match spec
greybus: update UniPro Set Interface Power Mode operation to match spec
Evgeniy Borisov (6):
greybus: camera-gb: Extend the configure streams interface
greybus: camera-gb: Remove hardcode for CSI TX number of lanes
greybus: camera-gb: Add description of interface header
greybus: camera-gb: Extend gb camera module structure
greybus: camera-gb: Implement camera module reference counting as subject.
greybus: camera: Add RAW data format
Fabien Parent (15):
greybus: es2: rename misnamed CPORT_MAX into CPORT_COUNT
greybus: add num_cports field to greybus hd
greybus: es2: dynamically allocate array for cport <-> ep mapping
greybus: es2: get cport count from apb1 usb device
greybus: es{1,2}: remove control endpoint field
greybus: connection: fail to bind if connection init fails
greybus: es{1,2}: remove obselete define
greybus: es2: implement cport reset control request
greybus: connection: destroy connection on failing to bind it
greybus: connection: add api to {en,dis}able unipro fct flow
greybus: apba: add fct flow usb control requests
greybus: es2: implement the fct flow control requests
greybus: connection: {en,dis}able fct flow in connection management
greybus: camera: disable E2EFC on CSI connection
greybus: connection: remove CDSI1 hack
Georgi Dobrev (2):
greybus: Added a sysfs entry to power down the SVC
greybus: greybus-driver: Add intf_oops operation
Gjorgji Rosikopulos (11):
greybus: camera: HACK: Export GB camera interface
greybus: camera: HACK: Register gb camera to the HOST camera
greybus: camera: Fix backword compatibility in configure streams
greybus: camera: Add support for configure streams flag in gb interface
greybus: camera: Update configure stream based on new interface
greybus: camera: add semiplanar and planar formats
greybus: camera: Improve module registration mechanism
greybus: camera: Use pointer for gb camera module ops
greybus: camera: Add metadata format
greybus: camera: Add debug data format
greybus: camera: Rename debug and metadata mbus formats
Greg KH (8):
greybus: battery: update for 4.1 power supply api changes
greybus: gb-audio: fix build warning
greybus: gb-audio: fix build breakage on 4.1-rc1
greybus: es2: comment out unused functions
greybus: audio_codec: Don't be tricky with the driver model
greybus: add bundle class to the bundle uevent
greybus: svc: double the ping delay timeout
greybus: camera: only build module against msm kernel
Greg Kroah-Hartman (376):
greybus: Initial commit
greybus: Import most recent greybus code to new repo.
greybus: README and .gitignore updates
greybus: update README with info on how to build and contact me.
greybus: i2c-gb: actually add the i2c adapter properly...
greybus: greybus.h: tiny coding style cleanups
greybus: Greybus SD/MMC host driver
greybus: add framework for 'struct gbuf'
greybus: gpio driver
greybus: gpio-gb.c: it now builds properly
greybus: uart framework added, doesn't build
greybus: uart-gb: more work on tty functions
greybus: more uart work
greybus: uart-gb: now builds, more framework added
greybus: start moving the function types into the greybus core
greybus: static module_init/exit functions
greybus: hook up sdio, gpio, and tty into the greybus core.
greybus: i2c: tie to the proper place on the greybus_device
greybus: i2c: use same naming convention everywhere
greybus: gpio: tie into gb core properly
greybus: sdio: tie into gb core properly
greybus: uart: tie into gb core properly
greybus: initial framework for ES1 usb AP driver
greybus: can't use devm anymore, we aren't tieing into the driver model lifecycle :(
greybus: actually get the devm() change to build...
greybus: first framework for the es1 ap controller
greybus: es1-ap-usb: more init framework added.
greybus: register the bus with the driver core and add framework for debugfs files.
greybus: ap message loop added.
greybus: structures added
greybus: more structure definitions added
greybus: split svc msg out into separate header file
greybus: greybus_desc.h created
greybus: header file s/u8/__u8/g
greybus: greybus.h: tiny movement around
greybus: more changes due to name changes in the greybus document
greybus: s/greybus_device_id/greybus_module_id/g
greybus: get field names right for descriptors
greybus: turn off warnings for es1-ap-usb.c to make it easier to build for now...
greybus: export gb_new_ap_msg so that the es1 module can use it
greybus: start parsing descriptor structures
greybus: start parsing descriptor fields
greybus: sysfs attributes for functions and more driver core integration.
greybus: serial number attribute added
greybus: module id attributes
greybus: greybus_string()
greybus: split sysfs functions out to separate file.
greybus: minor checkpatch cleanups
greybus: add proper packing to all greybus message types
greybus: uart-gb: remove unneeded THIS_MODULE setting
greybus: add es1_ap_desc.c to describe the ES1 USB device descriptors
greybus: host controller additions
greybus: more hd work
greybus: first cut at parsing svc messages sent to the AP
greybus: add battery module
greybus: battery FIXME added
greybus: minor whitespace cleanups to make checkpatch.pl happy
greybus: es1_ap_desc.c: updated ES1 USB device descriptor
greybus: es1: finialized USB device structure
greybus: es1: forgot to free our urb on disconnect
greybus: pass appropriate type to create function
greybus: fix endian issue in sysfs.c
greybus: Fix build errors on older kernels.
greybus: Merge branch 'master' of github.com:gregkh/greybus
greybus: AP: move a bunch of svc message handling logic into ap.c
greybus: es1: add the start of cport urb handling.
greybus: es1: allocate cport out urbs properly
greybus: es1: handle cport data in and out
greybus: ap: convert to workqueue from thread
greybus: ap: cleanup of process ap message loop
greybus: Makefile: add 'check' option to run sparse with endian checks enabled
greybus: es1 endpoint descriptor: minor fixes to get the config right
greybus: tty driver fixes to get init working properly
greybus: uart-gb: let the core dynamically allocate the major number
greybus: core: verify major/minor number of greybus protocol
greybus: es1: functionally complete
greybus: devices: endpoint description of device
greybus: gbuf recieve path work, not done, dinner time...
greybus: gbuf: cport in buffer stream logic
greybus: add test_sink driver
greybus: fix hd init sequence of setting up parent and driver pointers properly
greybus: es1: set buffer sizes for messages based on email discussions
greybus: gbuf: implement submission logic
greybus: gbuf: clean up logic of who owns what "part" of the gbuf
greybus: svc_msg.h: add data for hotplug message
greybus: greybus.h: add function prototype for add/remove a module
greybus: core.c: create empty functions to keep linking working for hotplug/remove
greybus: ap: start validating the message better
greybus: ap: validate the rest of the svc message buffer sizes
greybus: core: hook up the hotplug message
greybus: fix up coding style issue I caused with the last patch...
greybus: es1: no BUG_ON() code, report an error and recover.
greybus: operation: fix endian issue in the operation message header size field.
greybus: manifest: some minor sparse warning fixups.
greybus: manifest.c: minor sparse cleanup
greybus: connection: properly lock idr
greybus: greybus.h: remove transfer_flags
greybus: greybus_id.h: checkpatch cleanup
greybus: svc_msg.h: add bsd license to file so that firmware can use it.
greybus: greybus_manifest.h: add BSD license so that firmware can share it.
greybus: greybus_manifest.h: fix up class protocol numbers to match the spec.
greybus: gpio-gb: allow it to build properly for all current kernel versions.
greybus: battery-gb: provide accessors for a few more functions
greybus: battery: some hooking up to the greybus core
greybus: add LED protocol numbers
greybus: battery-gb: Add battery communication with the module
greybus: Merge branch 'gregkh_work' into master
greybus: battery-gb.c: fix memory leak found by Viresh
greybus: battery-gb: Allow kernel values to get out of sync with greybus spec
greybus: module: fix double free of module
greybus: hook up greybus to the driver model
greybus: module: enable all sysfs attributes
greybus: i2c: point to the proper parent device
greybus: module: don't create duplicate module ids
greybus: operation: fix some sparse warnings
greybus: gpio-gb: fix some endian sparse warnings that were real.
greybus: gbuf: implement gbuf_kill_gbuf()
greybus: core: check for valid hcd callbacks
greybus: core: make greybus_kill_gbuf not return a value
greybus: operation: make the timeout a per-operation thing, not per-connection
greybus: uart-gb: convert over to the connection interface
greybus: uart-gb: remove global init functions
greybus: FIXME/XXX removals: We have proper reference counting now
greybus: module: remove unneeded XXX comment about module id values
greybus: kernel_ver.h: add ATTRIBUTE_GROUPS() macro for older kernels
greybus: uart-gb: mark some functions static
greybus: battery-gb: remove some #if 0 code
greybus: sdio-gb: convert to the connection interface.
greybus: Merge branch 'master' into driver_model_rework
greybus: #if 0 cleanups
greybus: Remove id_table usages
greybus: es1-ap-usb: document the lack of callback for the outgoing bulk urbs
greybus: connection: fix up error patch logic in gb_connection_create()
greybus: Revert "manifest: improve print message"
greybus: Revert "manifest: remove extra loop for finding module descriptor"
greybus: skeleton for future uevents.
greybus: Greybus UART connection driver
greybus: uart: handle NULL size requests in request_operation()
greybus: greybus_manifest.h: update with full list of protocols
greybus: vibrator-gb: add vibrator driver
greybus: vibrator-gb: fixes based on Marti's review comments.
greybus: vibrator-gb: proper allocate minor numbers
greybus: Merge branch 'master' into vibrator-gb
greybus: vibrator-gb: fix up api changes
greybus: Merge branch 'master' into vibrator-gb
greybus: vibrator-gb.c: fix up incorrect comment
greybus: uart-gb: handle throttle/unthrottle properly
greybus: uart-gb: fix up typo from previous patch moving the status variable around.
greybus: pwm-gb.c: fix up missing { } for else
greybus: add -Wall to the build flags
greybus: operation: create gb_operation_sync for sending "simple" messages
greybus: battery-gb: convert to use gb_operation_sync
greybus: vibrator-gb: convert to use gb_operation_sync
greybus: uart-gb: convert to use gb_operation_sync
greybus: gpio-gb: convert to use gb_operation_sync
greybus: i2c-gb: convert to use gb_operation_sync
greybus: pwm-gb: convert to use gb_operation_sync
greybus: uart-gb: clean up send_line_coding
greybus: usb-gb: import a "buildable" version of the usb-gb.c driver
greybus: operation: fix up sparse warning
greybus: battery-gb.c: add new functions from Greybus spec document.
greybus: greybus_manifest.h: add FIXME for version
greybus: uart-gb.c: don't include module.h
greybus: module versions: remove them
greybus: module: remove obsolete gb_tty pointer
greybus: interface_block: rename module.[c|h] to interface_block.[c|h]
greybus: interface_block: s/gb_module/gb_interface_block/
greybus: interface_block: rename the sysfs files to not have 'module' in them
greybus: sysfs documentation: Document the greybus interface block sysfs files.
greybus: interface_block: move sysfs files into the interface_block.c file
greybus: connection: document the sysfs files
greybus: interface: document sysfs files
greybus: driver matching: Greybus drivers bind to interface blocks, not modules
greybus: bundle: rename interface.[c|h] to bundle.[c|h]
greybus: es1-ap-usb: don't protest when getting -EPROTO USB errors
greybus: bundle: s/gb_interface/gb_bundle/g
greybus: interface_block.c: rename the "interfaces" list "bundles"
greybus: rename interface_block.[c|h] -> interface.[c|h]
greybus: interface: s/gb_interface_block/gb_interface/g
greybus: interface: rename greybus_interface_block_id to greybus_interface_id
greybus: interface.h: remove unused #defines
greybus: greybus_id.h: minor typo fixes
greybus: interface: we really are creating/destroying interfaces not modules.
greybus: greybus_host_device: rename modules -> interfaces
greybus: interface: rename gb_modules_lock -> gb_interfaces_lock
greybus: add module support
greybus: interface: remove global manifest_descs list
greybus: protocol: switch gb_protocol_register() to return an int
greybus: export needed symbols for protocols
greybus: protocol: split binding of prototcols to connections out of init
greybus: i2c-gb: split out into a stand-alone kernel module.
greybus: bundle: create GB_DEVICE_ID_BAD
greybus: protocol: add a module owner to a protocol
greybus: vibrator-gb: move vibrator protocol to a stand-alone module.
greybus: battery-gb: move the battery protocol out to a stand-alone module
greybus: gpb: Create a "GP Bridge" kernel module
greybus: i2c-gb: move i2c protocol into the gpbridge driver
greybus: module: get rid of global list of modules
greybus: protocol: name protocols.
greybus: protocol: fix oops when no protocol is assigned
greybus: sysfs: put a \n at the end of all sysfs files
greybus: Makefile: provide install target
greybus: rename gpbridge.ko to gb-phy.ko
greybus: rename vibrator-gb.ko to gb-vibrator.ko
greybus: rename battery-gb.c to gb-battery.c
greybus: rename es1-ap-usb.ko to gb-es1.ko
greybus: Move the es1_ap_desc.c file to Documentation directory
greybus: es1: no need to assign a variable we return directly
greybus: es2: add ES2 Greybus Host driver
greybus: Revert "protocol.c: fix a kernel panic caused by __gb_protocol_register"
greybus: Revert "protocol: dedup protocol find code"
greybus: ap.c: our workqueue should be ordered
greybus: es1.c: wait until the last possible minute to start the svc messages
greybus: es2: fix USB id to not be the same as ES1
greybus: es1: fix build warning for apb1_log_enable_write
greybus: es1: fix tiny whitespace issues
greybus: es1: use <linux/uaccess.h> and not <asm/uaccess.h>
greybus: es1: no need to check for NULL on debugfs_remove()
greybus: es1: struct file_operations needs to be const
greybus: es1: decimal modes are not what are wanted for debugfs
greybus: es1: move debugfs function to use kstrotoint_from_user()
greybus: es1: fix checkpatch warning about blank lines needed
greybus: es1: separate usb_log enable/disable logic into different functions
greybus: es1: allow the debug log to be stopped
greybus: Merge branch 'es1-fixes' into master
greybus: debugfs: we shouldn't care if debugfs is working or not
greybus: es2: sync up with recent es1.c changes
greybus: loopback: fix build breakage about SZ_4K
greybus: loopback: use the attribute groups, not group
greybus: kernel_ver.h: add sysfs_create_groups() and sysfs_remove_groups()
greybus: Documentation/sysfs: add a proposed sysfs tree for greybus
greybus: module.c: add attributes
greybus: endo: add endo structures and logic
greybus: endo: hook up endos into the device tree
greybus: Documentation/sysfs-bus-greybus: update kernel version and date
greybus: Documentation/sysfs-bus-greybus: document the endo and SVC
greybus: Documentation/sysfs-bus-greybus: document module sysfs files
greybus: gpbridge.h: whitespace cleanups
greybus: gpbridge.h: remove BIT() define
greybus: bundle: add state sysfs file
greybus: MODULE_LICENSE cleanup
greybus: kernel_ver.h: fix leading space coding style issues
greybus: kernel_ver.h: add a ' ' character to KERNEL_VERSION()
greybus: kernel_ver.h: add a blank line after a variable definition
greybus: protocol.h: macros should not have a trailing ';'
greybus: protocol.h: fix up long lines in define_get_version()
greybus: audio: fix some sparse static warnings
greybus: audio: fix a bunch of endian issues
greybus: vibrator: convert idr to be an ida
greybus: connection: remove lock around ida_simple_* functions
greybus: raw: add raw greybus kernel driver
greybus: Merge branch 'master' of github.com:gregkh/greybus
greybus: Merge branch 'master' of github.com:gregkh/greybus
greybus: es?: decrease buffer size to 2k
greybus: remove __init from .h files
greybus: Merge branch alex into Alexandre
greybus: Revert "greybus: reserve host cport id 0"
greybus: Revert "examples: add manifest file for sdio"
greybus: properly cleanup ida and idr structures when shutting down
greybus: connection: prevent oops for protocol_id sysfs file
greybus: loopback: add poll support to the iteration_count sysfs file
greybus: es1: remove svc endpoint message handling
greybus: es2: remove svc endpoint message handling
greybus: greybus.h: remove greybus_svc_in()
greybus: remove submit_svc from the host driver
greybus: interface: make gb_interface_destroy() static
greybus: core: clean up ida memory for host controller
greybus: svc: revert svc changes to keep things working for a while.
greybus: es1: remove svc endpoint message handling
greybus: es2: remove svc endpoint message handling
greybus: greybus.h: remove greybus_svc_in()
greybus: remove submit_svc from the host driver
greybus: interface: make gb_interface_destroy() static
greybus: Merge branch 'master' into svc to keep it up to date.
greybus: Revert "connection: remove special check for svc cport id"
greybus: Merge branch 'master' into svc
greybus: Merge branch 'master' into svc
greybus: Merge branch 'master' into svc
greybus: Merge branch 'master' into branch 'svc'.
greybus: Merge branch 'master' into branch 'svc'.
greybus: gpio: handle api change in generic_handle_irq_desc()
greybus: vibrator: use the bundle struct device instead of the connector
greybus: uart: use the bundle struct device instead of the connector
greybus: spi: use the bundle struct device instead of the connector
greybus: hid: use the bundle struct device instead of the connector
greybus: audio: use the bundle struct device instead of the connector
greybus: firmware: use the bundle struct device instead of the connector
greybus: gpio: use the bundle struct device instead of the connector
greybus: i2c: use the bundle struct device instead of the connector
greybus: pwm: use the bundle struct device instead of the connector
greybus: light : use the bundle struct device instead of the connector
greybus: sdio: use the bundle struct device instead of the connector
greybus: usb: use the bundle struct device instead of the connector
greybus: operation: use the bundle struct device instead of the connector
greybus: control: use the bundle struct device instead of the connector
greybus: raw: use the bundle struct device instead of the connector
greybus: connection: remove 'struct device' from 'struct gb_connection'
greybus: connection: use the bundle struct device instead of the connector
greybus: loopback: use the bundle struct device instead of the connector
greybus: greybus_trace: use the bundle struct device instead of the connector
greybus: protocol: use the bundle struct device instead of the connector
greybus: bundle.h: add a private field to struct gb_bundle
greybus: core: remove uevent handling for gb_connection
greybus: svc: don't trust any struct devices
greybus: es2: use official USB vendor/device id
greybus: hid: hid should not be part of the bridged-phy driver.
greybus: Camera: remove f_dentry usage
greybus: tools: Add tools directory to greybus repo and add loopback
greybus: loopback_test: fix warning about signed/unsigned comparison
greybus: loopback_test: null terminate the dict structure
greybus: tools: don't always build them.
greybus: arche-platform: merge arche-apb-ctrl and arche-platform
greybus: host: provide "generic" apbridge output calls
greybus: APBridge: move APBridge request protocol to a common .h file
greybus: uevent: add GREYBUS_ID to uevent
greybus: Makefile: fix up build test for gb-audio-codec.ko
greybus: Revert "camera: Fix backword compatibility in configure streams"
greybus: firmware: log the name of the firmware being requested
greybus: svc: add ping command
greybus: lsgb: add lsgb shell script
greybus: lsgb: Minor tweaks
greybus: svc: add a "watchdog" to check the network health
greybus: svc: return proper error code on watchdog create error
greybus: svc watchdog: allow it to be enabled/disabled from userspace
greybus: audio_manager: use an 'ida' for the module id
greybus: lsgb: remove it, it's in the gb-utils repo
greybus: fix sparse warning in manifest.c
greybus: only build arche platform driver if usb3613 is enabled
greybus: Revert "gb_loopback: Fix throughput calculations"
greybus: uart: properly calculate max buffer size
greybus: gpio: use bundle device for error messages
greybus: gpio: handle api changes for 4.5 kernel release
greybus: expose full 32 bits of vid/pid to userspace
greybus: gpbridge.h: move protocol init/exit prototypes
greybus: manifest: reserve the Bridged PHY class
greybus: properly annotate struct gb_control_timesync_enable_request
greybus: connection: add functions to get/set private data
greybus: convert drivers to use connection->private set/get
greybus: trivial: add checkpatch.pl to the tree
greybus: scripts: create subdir for checkpatch and spelling.txt
greybus: gpbridge: implement gpbridge "bus" logic
greybus: UART: convert to a gpbridge driver
greybus: PWM: convert to a gpbridge driver
greybus: I2C: convert to a gpbridge driver
greybus: GPIO: convert to a gpbridge driver
greybus: SDIO: convert to a gpbridge driver
greybus: SPI: convert to a gpbridge driver
greybus: USB: convert to a gpbridge driver
greybus: make greybus_gpbdev_type static
greybus: kernel_ver.h: provide pwm_is_enabled()
greybus: Revert "greybus: don't use spin_lock_irq()"
greybus: move all firmware prefix strings to one place
greybus: fix up copyright dates on arpc.h
greybus: Revert "svc: drop all symbol exports"
greybus: es2: remove #if 0 code
greybus: es2: remove struct cport_to_ep
greybus: es2: remove bulk_out array
greybus: es2: remove bulk_in array
greybus: es2: remove unneeded BULK_* #defines
greybus: es2: No need to check before freeing an urb
greybus: es2: fix up usb probe error messages
Merge greybus driver tree into 4.8-rc5
greybus: Add drivers/greybus to the build
greybus: remove local checkpatch.pl copy
greybus: kernel_ver.h: remove lots of stuff
greybus: Documentation: remove sysfs tree
greybus: remove README and LICENSE
greybus: remove .gitignore file
greybus: move the sysfs documentation to Documentation/ABI/testing
greybus: remove old es1 endpoint description
greybus: hid: remove KERNEL_VERSION checks
greybus: vibrator: remove KERNEL_VERSION checks
greybus: gpio: remove KERNEL_VERSION checks
greybus: spi: remove KERNEL_VERSION checks
greybus: sdio: fix min() type check
greybus: remove CONFIG_PM_RUNTIME from kernel_ver.h
Jackson Chang (1):
greybus: SDIO: Add runtime pm support
Jacopo Mondi (17):
greybus: camera: Add support for flags to stream_configure
greybus: camera: Stream config change unipro speed
greybus: camera: Reset link speed on failed config
greybus: camera: Add missing return value
greybus: camera: Add CSI configuration parameters
greybus: camera: Rename clock_mode to flags
greybus: camera: Register capabilities operation
greybus: camera: Early update num_streams and flags
greybus: camera: Fix number of configured streams
greybus: camera: Fix size of configure_streams(0)
greybus: camera: Add bpp to Greybus Format description
greybus: camera: Add wrapper for sync operation with flags
greybus: camera: Update Configure Streams Response
greybus: camera: Configure APB-A with new params
greybus: camera: Update CSI config parameters
greybus: camera: Remove reference to ara subdevice
greybus: camera: Remove support for legacy modules
Jeffrey Carlyle (8):
greybus: Revert "bootrom: Implement timeouts to detect Module failures"
greybus: svc: implement svc_intf_{vsys,refclk,unipro}_{enable,disable} operations
greybus: svc: implement svc_intf_activate
greybus: svc: support status in svc_intf_activate response
greybus: interface: retry enumeration of UniPro-only modules
greybus: Revert "update UniPro Set Interface Power Mode operation to match spec"
greybus: svc: implement connection_quiescing call
greybus: connection: switch to using spin_lock_irqsave/spin_lock_irqrestore excluisvely
Joel Porquet (2):
greybus: add support for the log protocol
greybus: firmware.h: rename TFTF firmware filenames
Johan Hovold (547):
greybus: connection: fix non-atomic allocations under spin lock
greybus: bundle: fix sleep-while-atomic in gb_bundle_destroy
greybus: gpio: fix memory leaks at init and exit
greybus: operation: fix locking issues
greybus: connection: fix locking in gb_hd_connection_find
greybus: pwm: fix memory leak in error path
greybus: vibrator: fix memory leak in error path
greybus: connection: fix oops after failed init
greybus: gpio: fix set-debounce request alignment
greybus: pwm: fix config-request alignment
greybus: uart: remove packed-attribute from line-coding struct
greybus: usb: silence compiler warning
greybus: ap: fix svc handshake protocol check
greybus: ap: fix typo in comment
greybus: ap: clean up svc link management error path
greybus: connection: replace custom error function with dev_err
greybus: operation: use dev_err in gb_operation_sync
greybus: operation: remove unnecessary cast
greybus: gpio: fix truncated debounce times
greybus: gpio: remove incorrect todo comments
greybus: gpio: remove redundant argument verification
greybus: gpio: remove overly defensive argument verification
greybus: gpio: remove unnecessary explicit cast
greybus: gpio: make gb_gpio_controller pointer naming consistent
greybus: gpio: fix error handling
greybus: gpio: use dev_err and dev_warn
greybus: gpio: clean up line-state allocation
greybus: gpio: refuse to set value of input pins
greybus: gpio: add error messages to callbacks not propagating errors
greybus: operation: fix missing symbol exports
greybus: operation: fix typo in comment
greybus: operation: fix callback handling and documentation
greybus: operation: fix use-after-free when sending responses
greybus: operation: fix memory leak in request_send error path
greybus: operation: fix use-after-free and infinite loop on unhandled requests
greybus: operation: fix null-deref on operation cancel
greybus: operation: fix null-deref on operation destroy
greybus: operation: fix incoming request payload size
greybus: operation: replace pr_err with dev_err
greybus: operation: make incomplete-message errors more informative
greybus: gpio: fix null-deref on short irq requests
greybus: gpio: fix null-deref on unexpected irq requests
greybus: gpio: fix missing response on request errors
greybus: hid: replace pr_err with dev_err
greybus: hid: fix success response being sent on errors
greybus: hid: fix null-deref on short report requests
greybus: hid: fix missing input verification of report events
greybus: hid: fix missing response on request errors
greybus: operation: refactor response handling
greybus: es1: drop unnecessary casts
greybus: es1: fix DMA-buffer on stack
greybus: es1: fix buffer-size limit
greybus: es2: sync up with recent es1 changes
greybus: operation: fix unaligned memory accesses in receive path
greybus: remove obsolete buffer-alignment requirement
greybus: operation: move message-header definition to header file
greybus: pass messages to host drivers
greybus: operation: fix potential message corruption
greybus: operation: add explicit padding to message header
greybus: es1: fix transfer-buffer alignment
greybus: es2: sync up with recent es1 changes
greybus: drop host-driver buffer headroom
greybus: loopback: fix 64-bit divisions
greybus: remove unnecessary greybus.h includes
greybus: remove unused prototypes from header
greybus: fix host-device buffer constraints
greybus: operation: allow maximum-sized messages
greybus: core: add lower-limit for host-device buffers
greybus: operation: add helper to retrieve max payload size
greybus: gpbridge: fix section mismatches
greybus: operation: add support for incoming unidirectional operations
greybus: gpio: fix debugfs output
greybus: gpio: remove unused irq-ack operation
greybus: gpio: don't call irq-flow handler directly
greybus: gpio: use irq-domain lookups
greybus: gpio: rename irq mask and unmask callbacks
greybus: gpio: fix atomic sleep when using interrupts
greybus: gpio: fix interrupt protocol
greybus: gpio: update operation types
greybus: gpio: remove unused debounce define
greybus: gpio: add irq-type defines
greybus: esx: fix null-deref on hotplug events
greybus: operation/esx: fix message-cancellation lifetime bugs
greybus: operation: allow atomic operation allocations
greybus: operation: allow atomic request submissions
greybus: operation: make response helper static
greybus: operation: fix cancellation use-after-free
greybus: operation: add incoming-operation flag
greybus: operation: add unidirectional-operation flag
greybus: operation: remove obsolete operation-field comments
greybus: operation: drop redundant oom message
greybus: operation: allocate response before setting result
greybus: operation: fix cancellation of responses
greybus: operation: fix operation-destroy race
greybus: operation: fix use-after-free in response receive path
greybus: operation: fix outgoing-response corruption
greybus: operation: fix incoming-response corruption
greybus: operation: fix another cancellation use-after-free
greybus: operation: clean up create-incoming error path
greybus: operation: add active counter
greybus: operation: make cancellation synchronous
greybus: operation: complete operations on cancellation
greybus: operation: clean up operation work function
greybus: operation: suppress response submission on connection tear down
greybus: connection: add connection-state locking
greybus: operation: fix connection tear down
greybus: connection: fix protocol tear-down race
greybus: operation: fix operation look-up race
greybus: operation: fix response-cancellation race
greybus: operation: split incoming and outgoing cancellation
greybus: operation: allow drivers to define custom timeouts
greybus: operation: add gb_operation_sync_timeout helper
greybus: kernel_ver: add list_last_entry for old kernels
greybus: operation: fix atomic response allocation
greybus: operation: fix atomic message submission
greybus: operation: fix operation ordering
greybus: operation: clean up greybus_message_sent
greybus: connection: fix connection initialisation
greybus: connection: clean up connection creation
greybus: connection: clean up creation error paths
greybus: operation: use per-connection work queues
greybus: operation: add completion work queue
greybus: usb: clean up driver
greybus: usb: renumber operation types
greybus: usb: fix hcd allocation, deregistration and deallocation
greybus: usb: fix hc_driver fields
greybus: usb: implement hub_control callback
greybus: usb: disable protocol driver
greybus: greybus_protocols: fix typo in comment
greybus: svc: fix message packing
greybus: uart: fix message packing
greybus: sdio: fix message packing
greybus: greybus_protocols: use only type attributes for message packing
greybus: firmware: fix potential stack corruption
greybus: endo: fix endo-id allocation flag
greybus: svc: fix device-id allocation flag
greybus: svc: fix hot-plug-state allocation flag
greybus: interface: fix potential attribute-buffer overflow
greybus: pwm: replace pr_err with dev_err
greybus: core: fix hd-creation error path
greybus: es1: fix build-time cport constraint
greybus: fix cport-id defines
greybus: vibrator: fix tear-down race
greybus: sdio: fix tear-down use-after-free
greybus: sdio: fix work-queue leak and use-after-free
greybus: svc: fix ida memory leak
greybus: svc: fix version response
greybus: greybus_protocols: add missing version-request definition
greybus: connection: clean up svc-connection creation
greybus: connection: clean up init error paths
greybus: connection: clean up control-disconnected helper
greybus: connection: add control-connected helper
greybus: connection: add protocol-version helper
greybus: hd: add optional cport enable and disable callbacks
greybus: hd: remove connection-create/destroy callbacks
greybus: es1/2: fix use-after-free in completion callback
greybus: operation: remove gb_operation_destroy
greybus: es1/es2: clean up error messages
greybus: connection: replace pr_err with dev_err
greybus: operation: drop OOM-response FIXME
greybus: operation: clean up error messages
greybus: protocol: warn on bad deregistration
greybus: protocol: warn on protocol put errors
greybus: protocol: make version debug message more informative
greybus: vibrator: add missing protocol-register error handling
greybus: protocol: make protocol-lookup error message more informative
greybus: connection: remove duplicate protocol lookup error message
greybus: build: add pwm to config options to enable
greybus: gpio: handle set_irq_flags api change
greybus: endo: fix use-after-free in error path
greybus: endo: fix device-id allocation
greybus: connection: remove unused time-stamp defines
greybus: control: remove unused request handler
greybus: es2: update obsolete bridge-device comments
greybus: remove unused host-device id
greybus: remove unused connection device type
greybus: connection: use bundle device for bundle error messages
greybus: connection: fix potential null-deref in hd_cport_enable
greybus: connection: fix potential null-deref in connection create
greybus: connection: fix potential null-deref when binding protocol
greybus: connection: kill gb_hd_connections_exit
greybus: endo: fix ida memory leak
greybus: remove obsolete comment
greybus: create host-device compilation unit
greybus: hd: rename host-device structure
greybus: hd: rename host-driver structure
greybus: hd: use common prefix for exported functions
greybus: es2: fix use-after-free at disconnect
greybus: es2: fix endpoint requirement check
greybus: es2: separate urb allocation and submission
greybus: es1: separate urb allocation and submission
greybus: es2: separate stopping and deallocating urbs
greybus: es1: separate stopping and deallocating urbs
greybus: es2: fix debugfs dentry leak
greybus: es1: fix debugfs dentry leak
greybus: es2: add more structure to probe error handling
greybus: es1: add more structure to probe error handling
greybus: hd: fix host-device life time issues
greybus: es2: clean up es2 destructor
greybus: loopback: fix layer violation
greybus: battery: fix class-device parent
greybus: light: fix class-device parent
greybus: interface: fix potential use-after-free at remove
greybus: interface: fix obsolete comment
greybus: use decimal notation for interface and cport ids
greybus: svc: use bit-macro for cport flags
greybus: firmware: fix information leak
greybus: firmware: break long lines
greybus: svc: fix missing version-request sanity checks
greybus: fix bundle-id match macros
greybus: remove unique-id matching
greybus: interface: remove unique id
greybus: fix vendor and product matching
greybus: move id-matching back to core
greybus: add class matching macro
greybus: connection: fix cport-id range
greybus: hd: fix cport-count check
greybus: Documentation/sysfs: update the example sysfs tree
greybus: Documentation/sysfs-bus-greybus: update the bus ABI documentation
greybus: interface: rename vendor and product attributes
greybus: interface: remove device_id attribute
greybus: interface: add interface_id attribute
greybus: bundle: rename class attribute
greybus: bundle: use hexadecimal notation for class attribute
greybus: bundle: add bundle_id attribute
greybus: hd: make host device a device
greybus: connection: drop parent parameter from connection create
greybus: interface: make interfaces children of host devices
greybus: bundle: rename bundle devices
greybus: svc: make svc a device
greybus: svc: clean up svc initialisation
greybus: svc: register svc device at hello
greybus: svc: move endo id and ap interface id to svc
greybus: interface: kill gb_create_bundle_connection
greybus: connection: clean up connection-creation interface
greybus: connection: fix potential null-deref on create
greybus: connection: kill GB_PROTOCOL_SKIP_SVC_CONNECTION
greybus: connection: add name field
greybus: connection: handle static connections
greybus: operation: handle static connections
greybus: protocol: handle static connections
greybus: hd: fix svc-connection handling
greybus: svc: clean up log messages
greybus: loopback: remove endo reference
greybus: kill the endo
greybus: connection: handle bundle-less connections in svc helpers
greybus: control: do not assume a control bundle
greybus: connection: unconditionally enable connections
greybus: manifest: refactor cport-descriptor release
greybus: interface: drop the control bundle
greybus: Documentation/sysfs: rename interface and bundle attributes
greybus: svc: clean up hotplug/unplug request handlers
greybus: svc: clean up deferred hotplug handler
greybus: svc: clean up interface-remove helper
greybus: svc: generalise deferred request handling
greybus: svc: defer processing of hot-unplug requests
greybus: svc: fix racy hotplug handling
greybus: svc: flush workqueue at connection exit
greybus: core: add bus id to uevents
greybus: core: add interface id to interface and bundle uevents
greybus: core: add bundle id to bundle uevents
greybus: svc: add missing boot-status error message
greybus: bundle: remove obsolete function prototype
greybus: bundle: kill gb_bundle_bind_protocols
greybus: connection: fail on missing protocol
greybus: connection: bind protocol at init
greybus: connection: unbind protocol at exit
greybus: connection: separate connection creation and enabling
greybus: hd: initialise device last
greybus: hd: make svc-connection life time coincide with host-device
greybus: svc: create svc along with host device
greybus: use decimal notation for interfaces, bundles and cports
greybus: interface: remove defensive WARN_ON
greybus: bundle: kill the bundle lock
greybus: bundle: use dev_err for bundle-creation errors
greybus: bundle: separate connection disabling and destruction
greybus: bundle: separate bundle creation and registration
greybus: interface: separate manifest parsing from bundle registration
greybus: interface: fix interface-registration race
greybus: interface: clean up error messages
greybus: es2: demote dynamic-urb error message
greybus: interface: clean up control-connection handling
greybus: interface: fix hot-unplug control-connection timeouts
greybus: svc: drop copy-pasted function header
greybus: es2: clean up csi-config request
greybus: camera: fix memory leak in configure-streams error path
greybus: camera: fix memory leak in capture-request handler
greybus: camera: destroy data connection on link-config errors
greybus: core: add drvdata accessors
greybus: core: fix greybus driver registration
greybus: core: fix greybus device matching
greybus: connection: fix version-request error handling
greybus: power_supply: fix lock imbalance in init error path
greybus: connection: fix connection-state handling
greybus: connection: refactor connection enable and disable
greybus: legacy: add legacy-protocol bundle driver
greybus: connection: rename legacy init and exit functions
greybus: connection: remove disable from destructor
greybus: connection: add per-connection request handlers
greybus: connection: always cancel operations on connection disable
greybus: connection: clean up operation cancellation on disable
greybus: connection: disable operations on enable errors
greybus: connection: make connection enable/disable thread safe
greybus: connection: add unidirectional enabled state
greybus: connection: add helper to disable incoming operations
greybus: core: disable incoming operations pre disconnect
greybus: core: add defensive connection disable post disconnect
greybus: firmware: remove skip-disconnected flag
greybus: connection: remove skip-connected legacy protocol flags
greybus: greybus_protocols: remove control-protocol version
greybus: control: drop legacy-protocol dependency
greybus: svc: drop legacy-protocol dependency
greybus: connection: move legacy-protocol handling to legacy driver
greybus: legacy: use protocol pointer to determine state
greybus: core: disable bundle connections on hot-unplug
greybus: svc: store protocol version
greybus: connection: remove broken protocol-version handling
greybus: control: add bundle-version operation
greybus: connection: remove WARN_ON from destroy
greybus: connection: destroy workqueue at unregister
greybus: connection: drop the connection_mutex
greybus: connection: fix lookup race
greybus: connection: serialise connection creation
greybus: connection: make cport lookup thread-safe
greybus: legacy: add private driver data
greybus: core: defer connection creation to driver probe
greybus: manifest: check for duplicate CPort descriptors when parsing
greybus: connection: drop protocol parameter from static interface
greybus: connection: add helper to create control connections
greybus: connection: drop the legacy protocol-id parameter
greybus: connection: simplify connection-creation interface
greybus: connection: rename connection-create interface
greybus: connection: return error-valued pointer on creation errors
greybus: legacy: refactor legacy-connection handling
greybus: legacy: add initialized flag
greybus: connection: remove legacy protocol id from core
greybus: legacy: look up protocol at connection creation
greybus: connection: set request handlers at creation
greybus: hid: clean up init error paths
greybus: hid: convert to bundle driver
greybus: vibrator: convert to bundle driver
greybus: firmware: abort if AP_READY fails
greybus: hd: fix host-device-removal race
greybus: legacy: remove unimplemented classes from id table
greybus: firmware: convert to bundle driver
greybus: manifest: add interface-device prefix to messages
greybus: core: add device prefix to error messages
greybus: i2c: add bundle-device prefix to error messages
greybus: loopback: add missing pr_fmt
greybus: spi: add bundle-device prefix to error messages
greybus: bundle: remove private data field
greybus: raw: fix memory leak on disconnect
greybus: es2: fix cport-count error handling
greybus: Documentation: remove svc unique_id attribute
greybus: uart: add max-payload sanity check
greybus: uart: add missing serial-state sanity check
greybus: uart: fix incomplete receive-data sanity checks
greybus: Documentation/sysfs: add module devices
greybus: Documentation/sysfs: move module 1-4 to position 5
greybus: Documentation/sysfs: make 1-5 a 2x2 module
greybus: operation: add temporary pointer to response handler
greybus: operation: add support for short responses
greybus: loopback: fix double error count
greybus: raw: use hexadecimal notation for request types
greybus: connection: add connection-flag interface
greybus: svc: clean up CPortFlags handling
greybus: connection: generalise CPortFlags handling
greybus: connection: add CSD connection flag
greybus: hd: rename CPort-features callbacks
greybus: greybus_manifest: remove unused AP class and protocol
greybus: greybus_manifest: remove unused SVC class
greybus: interface: remove unused function prototype
greybus: interface: remove unused drvdata helpers
greybus: interface: add comment about early control-connection disable
greybus: svc: refactor interface-route creation
greybus: svc: refactor interface-route destruction
greybus: interface: separate initialisation and registration
greybus: interface: free bundles on initialisation errors
greybus: svc: always register interfaces at hotplug
greybus: interface: disable control connection on initialisation errors
greybus: interface: remove useless spinlock
greybus: interface: move interface-removal helper
greybus: interface: rename initialisation function
greybus: interface: separate disabling from removal
greybus: svc: unexport eject helper
greybus: svc: print an error message on failed eject attempts
greybus: greybus_protocols: remove svc-eject timeout define
greybus: core: fix two container-of macros
greybus: greybus_protocols: align DME-attribute values
greybus: svc: remove bogus interface-reset helper
greybus: greybus_protocols: remove bogus comment about svc bundle
greybus: svc: indent CPort flag defines properly
greybus: interface: move boot-status clearing to interface enable
greybus: interface: fix es2 boot-status mask
greybus: interface: clean up and rename init-status helper
greybus: interface: clean up ES3-bootroom-quirk handling
greybus: interface: move route creation to interface activation
greybus: interface: deactivate interface on enumeration failure
greybus: interface: clean up device-id handling
greybus: greybus_protocols: rename NULL DME selector index
greybus: interface: read DME attributes at activation
greybus: interface: add ES2 init-status quirk flag
greybus: interface: clean up ES2 VID/PID hack
greybus: interface: clean up DME attribute handling
greybus: interface: add Ara serial-number support
greybus: manifest: fix illegal free in error path
greybus: Documentation: move the interface power attributes
greybus: Documentation/sysfs: remove interface unique_id attribute
greybus: Documentation/sysfs: rename the svc eject attribute
greybus: Documentation/sysfs: add example control devices
greybus: interface: disable interface on registration failures
greybus: control: move timesync-operation functions
greybus: core: make the control object be a device
greybus: interface: register control device at hotplug
greybus: interface: move vendor and product strings to control device
greybus: control: return error pointer when failing to create control device
greybus: interface: reduce control-device lifetime
greybus: svc: make sure to deactivate all interfaces on disconnect
greybus: svc: keep interfaces registered during mode switch
greybus: svc: remove interface-remove helper
greybus: svc: keep error messages uniform
greybus: svc: fix pwrmon return value
greybus: svc: fix function-parameter indentation
greybus: svc: use a common prefix for debugfs functions
greybus: Documentation/sysfs: add entry for control device
greybus: Documentation/sysfs: sort entries alphabetically
greybus: svc: refactor interface re-enable
greybus: core: add module abstraction
greybus: module: implement controlled module removal
greybus: interface: add active state flag
greybus: svc: add stub functions for v_sys, refclk and unipro
greybus: svc: add stub interface-activate function
greybus: interface: implement interface activation and power down
greybus: svc: implement module inserted and removed operations
greybus: svc: implement interface mailbox event
greybus: interface: fix broken UniPro flow control
greybus: interface: remove unused prototype
greybus: fix outdated kernel-doc comment
greybus: operation: update gb_operation_request_send() documentation
greybus: operation: add support for initiating unidirectional operations
greybus: operation: add helper functions for unidirectional operations
greybus: control: implement mode-switch operation
greybus: loopback: remove unsupported version request
greybus: svc: define the version request
greybus: operation: fix definition of the invalid type
greybus: interface: fix type in enable error message
greybus: interface: remove unused interface-version quirk
greybus: connection: add no-flow-control connection flag
greybus: connection: add offloaded connection flag
greybus: connection: add helper for creating offloaded connection
greybus: audio: mark the data connection as offloaded
greybus: hd: move CPort allocation to host-device code
greybus: es2: fix probe error handling
greybus: es2: clean up CDSI CPort reservation
greybus: hd: generalise cport allocation
greybus: es2: clean up cport-reset handling
greybus: es2: add support for CDSI1 allocation
greybus: connection: add camera-data connected workaround
greybus: camera: fix data-connection handling
greybus: operation: fix broken activation logic
greybus: svc: fix another pwrmon return value
greybus: audio_apbridgea: fix two return values
greybus: svc: use EREMOTEIO for remote errors
greybus: interface: fix svc-resource error messages
greybus: remove redundant latency-tag sanity checks
greybus: core: avoid I/O to disconnected interfaces
greybus: interface: clean up bootrom quirk handling
greybus: interface: avoid I/O to bootrom during removal
greybus: connection: add control connection flag
greybus: control: implement disconnecting operation
greybus: svc: add stub connection-quiescing operation
greybus: connection: log cport-enable error messages
greybus: es2: reset cports at disable
greybus: connection: rename an error label
greybus: connection: move CPort Buffer configuration out of svc helpers
greybus: connection: remove unused invalid state
greybus: operation: clean up early connection-state check
greybus: operation: restructure activation state handling
greybus: operation: add helper for creating core operations
greybus: connection: implement proper connection closure
greybus: connection: add flush host-device callback
greybus: kernel_ver: backport reinit_completion to pre-3.13
greybus: control: add error message to mode-switch helper
greybus: interface: implement generic mode-switch functionality
greybus: svc: remove deprecated hotplug operations
greybus: fix tracepoint-related whitespace issues
greybus: hd: fix gb_hd_release tracepoint
greybus: fix forced disable of offloaded connections
greybus: hd: add flag argument to cport_enable callback
greybus: connection: prevent drivers from specifying core flags
greybus: connection: add support for high-priority connections
greybus: es2: add support for greybus cport flags
greybus: Revert "connection: switch to using spin_lock_irqsave/spin_lock_irqrestore excluisvely"
greybus: Revert "greybus: gb_connections_lock: don't use spin_lock_irq()"
greybus: Revert "greybus: es2.c: don't use spin_lock_irq()"
greybus: Revert "greybus: uart: don't use spin_lock_irq()"
greybus: es2: fix arpc return value
greybus: es2: fix arpc response lookups
greybus: connection: drop the svc quiescing operation
greybus: interface: add interface-type attribute
greybus: interface: partition attribute group
greybus: interface: make attributes type dependent
greybus: interface: amend interface registration message
greybus: module: suppress activation error message for dummy interfaces
greybus: interface: use an enum for interface type
greybus: interface: do not read DME during reactivation
greybus: interface: clean up ES3 activation-retry hack
greybus: interface: make sure type is invariant during reactivation
greybus: interface: prevent reactivation during removal
greybus: interface: fix power_state documentation
greybus: interface: hibernate UniPro link in activation error path
greybus: interface: fix interface_id attribute group
greybus: control: suppress bundle_activate() for bootrom
greybus: operation: fix broken response error messages
greybus: operation: fix broken response tracepoint
greybus: operation: clean up request handler
greybus: Documentation/sysfs: replace Ara references
greybus: replace Ara references
greybus: es2: fix arpc response-allocation error handling
greybus: es2: fix arpc error message
greybus: es2: fix arpc active-list corruption
greybus: remove unused protocol-version messages
greybus: control: remove some braces
greybus: connection: fix offloaded-connection ping cport id
greybus: es2: clean up ARPC symbol names
greybus: es2: always set result value
greybus: es2: allow ARPC result to be ignored
greybus: es2: add define for ARPC CPort requests
greybus: es2: rename USB vendor-request timeout define
greybus: greybus_protocols: fix apba vendor-request comment
greybus: es2: fix USB vendor-request prefixes
greybus: arpc: move arpc definitions to their own header
greybus: hd/es2: add cport_connected callback and ARPC
greybus: hd/es2: add cport_quiesce callback and ARPC
greybus: hd/es2: add cport_clear callback and ARPC
greybus: hd: add cport_shutdown callback
greybus: control: make disconnecting a core operation
greybus: es2: fix memory leak in probe error path
greybus: svc: drop all symbol exports
greybus: svc: drop unnecessary symbol exports
greybus: es2: implement flush callback
greybus: es2: implement shutdown callback
greybus: connection: implement new connection handling
greybus: hd/es2: remove obsolete callbacks
greybus: es2: remove obsolete cport-reset ARPC
greybus: svc: fix timeout indentation
greybus: greybus_protocols: remove svc key-event type
John Stultz (4):
greybus: Changes required for integrating into the Android build
greybus: build: Add -fno-pic for 32bit arm as well
greybus: Move briged phy structure definitions into gpbridge.h
greybus: gb-audio: Add integrated greybus audio driver
Kris Huang (3):
greybus: lights: Add runtime pm support
greybus: lights: enable multi color LED support
greybus: lights: Control runtime pm suspend/resume on AP side
Laurent Pinchart (25):
greybus: es2: Reserve CPorts 16 and 17
greybus: Add camera protocol definition
greybus: Add driver for the camera class protocol
greybus: camera: Fix endian conversion issues
greybus: svc: Add support for the link config operation
greybus: camera: Configure link speed for the data connection
greybus: es2: Add support for CSI transmitter configuration
greybus: camera: Configure the bridge CSI transmitter
greybus: camera: Set link power mode to HS-G2 with 2 lanes
greybus: camera: Fix remaining endian conversion issues
greybus: es2: Fix endian conversion issues
greybus: es2: Don't use stack memory as USB request data
greybus: camera: Raise the CSI-2 bandwidth
greybus: svc: Replace link config hack with standard operation
greybus: camera: Return the result flags from the configure_streams response
greybus: camera: Factorize link power mode configuration code into a function
greybus: camera: Clean up when AP link power mode configuration failed
greybus: camera: Set power mode after configuring streams
greybus: camera: Don't configure CSI TX in test only mode
greybus: camera: Implement the capabilities operation
greybus: camera: Convert to bundle driver
greybus: camera: Clean up on stream configuration failure
greybus: camera: Fix data connection setup
greybus: camera: Create and destroy data connection on demand
greybus: camera: Hardcode the APB-A CSI-2 TX parameters
Mark A. Greer (11):
greybus: gb-audio: Set clock edges to match rt5647 codec requirements
greybus: gb-audio: Allocate space for 20 configurations
greybus: gb-audio: Clean up codec name generation
greybus: gb-audio: Fix DAI formats and master/slave settings
greybus: gb-audio: cpu_dai.fmt does not exist in v4.1
greybus: gb-audio: Add I2C device for rt5647 codec
greybus: gb-audio: Remove useless comment
greybus: gb-audio: Activate TX CPort in PCM workqueue
greybus: gb-audio: Activate TX connection Bundle CPort ID
greybus: gb-audio: Set I2S Configuration according to ASOC requests
greybus: gb-audio: Set samples per message during init
Mark Greer (16):
greybus: Initial I2S definitions
greybus: gb-audio: Reset sample count when CPort deactivated
greybus: gb-audio: Ensure i2c adapter struct exists before using
greybus: audio: Remove I2S Bridged-PHY Protcol based audio driver
greybus: i2s: Remove deprecated I2S Bridged-PHY Protocol
greybus: audio: Allocate protocol and class values
greybus: audio: Use underscore in file name
greybus: audio: Add Audio Device Class Protocol definitions
greybus: audio: Add Greybus Audio Device Class Protocol helper routines
greybus: audio: Add direct audio streaming control for APBridgeA
greybus: audio: Register CPorts for specific directions
greybus: audio: Split start and stop APBridgeA requests
greybus: audio: Use CSD instead of E2EFC for audio data connections
greybus: audio: apbridgea: Remove GET_TX/RX_DELAY message types
greybus: audio: Remove GET_TX/RX_DELAY message types
greybus: audio: Fix incorrect direction value when enabling RX
Marti Bolivar (4):
greybus: uart-gb.c: replace alloc_tty_driver with tty_alloc_driver
greybus: uart-gb.c: dynamically allocate device numbers
greybus: gb_hd_connection_find(): fix "not found" case
greybus: remove obsolete SVC result codes
Matt Porter (33):
greybus: update svc_msg_header fields and users to match spec
greybus: move versioning from svc message header to handshake function
greybus: update GREYBUS_VERSION_[MAJOR|MINOR] to match spec
greybus: es1-ap-usb: fix svc control pipe flags
greybus: initialize all fields in an SVC handshake message
greybus: update descriptor module_id->module to match spec
greybus: remove serial number descriptor to match spec
greybus: remove unused function descriptor fields and change class->function_type
greybus: update string descriptor length field to __u8 type to match spec
greybus: update descriptor type enums to match renumbering in spec
greybus: es1-ap-usb: adjust SVC buffer size to handle worst case
greybus: remove additional unused fields from the cport descriptor
greybus: fix error message on parse of cport descriptor size
greybus: fix gb_add_module() by enabling the device_add()
greybus: implement core module removal path
greybus: fix manifest parsing problem with descriptor payload
greybus: fix gb_manifest_parse() successful return path
greybus: es1-ap-usb: handle -EPROTO in check_urb_status()
greybus: gpio-gb: remove unused status field from struct gb_gpio_get_value_request
greybus: gpio-gb: fix offset error checking and usage
greybus: ap: add support for the AP Device ID unipro management function message
greybus: module: add gb_module_find()
greybus: interface: add gb_interface_find()
greybus: ap: process the UniPro link up message
greybus: svc: remove unneeded fields from the unipro set route message payload
greybus: ap: add svc_set_route_send() command and use it on a link up event
greybus: svc: remove the DDB function message support
greybus: module: set device_id when initializing an interface
greybus: connection: call connection_init hook after setting the handler
greybus: add pwm protocol driver
greybus: remove unused gb_connection_handler externs
greybus: operation: add missing gb_operation_response_send() export
greybus: gpio: add interrupt handling support
Michael Mogenson (2):
greybus: remove gb_i2c_timeout_operation
greybus: remove gb_i2c_retries_operation
Michael Scott (13):
greybus: build: android: replace hard-coded build destination with variable
greybus: build: android: add all greybus modules to the build
greybus: build: android: Fix script which locates .ko files and moves them into ramdisk
greybus: build: android: replace Linaro build specific logic with AOSP equivalents
greybus: build: android: fix strip module build step
greybus: build: remove Android makefile
greybus: firmware: replace colons with underscore in firmware file request
greybus: arche-platform: ensure wake-detect pin is deasserted
greybus: arche-platform: bring SVC out of reset later in probe
greybus: arche-platform: assert wake-detect to complete WAKE_OUT event
greybus: arche-platform: reduce wait between WAKE_OUT checks
greybus: interface: clear upper 16-bits of version_id and product_id
greybus: adjust kernel version check for led_sysfs_is_disabled
Mitchell Tasman (1):
greybus: svc: reconfig APBridgeA-Switch link to handle required load
Perry Hung (13):
greybus: gb_operation: replace timeout workqueue
greybus: gb_operation: drop operation refcount on cancel
greybus: makefile: fix unused cflags
greybus: battery: fix panic on operation error
greybus: connection: silence warning on unbound protocols
greybus: svc: add route create operation
greybus: svc: create bidirectional routes on hotplug request
greybus: svc: connection: ask SVC to create connections
greybus: svc: add flags and traffic class parameter to connection create op
greybus: svc: add route create operation
greybus: svc: create bidirectional routes on hotplug request
greybus: svc: connection: ask SVC to create connections
greybus: svc: add flags and traffic class parameter to connection create op
Philip Yang (2):
greybus: HID: Add runtime pm support
greybus: power_supply: Add runtime pm support
Phong Tran (12):
greybus: loopback: add more clean up when init connection fails
greybus: svc: Fix endian of hotplug request data
greybus: svc: use macro for init and exit protocol
greybus: audio: Fix typo for macro I2S data version minor
greybus: audio: Remove the MODULE_LICENSE()
greybus: uart: Fix the memory leak in connection init
greybus: endo: move endo id map declaration to endo.h
greybus: uart: remove the redundant unregister chrdev
greybus: uart: fix the clean up while uart initiates connection unsucessfully
greybus: sdio: remove the redefine of sdio major and minor
greybus: sdio: change the order of remove and free mmc host
greybus: sdio: correct the usage of mmc request in work queues handler
Rob Herring (1):
greybus: es2: fix driver name collision with es1
Rui Miguel Silva (90):
greybus: es1: remove useless statement
greybus: es1: release urb on error path
greybus: protocol: dedup protocol find code
greybus: manifest: be coherent with protocol name
greybus: make: check kernel configuration options
greybus: makefile: add MMC to the required option list
greybus: sdio: extend sdio implementation
greybus: sdio: fix defines for older kernels
greybus: kernel_ver: add sg copy operations for kernel < 3.11
greybus: examples: add manifest file for sdio
greybus: sdio: split cmd_flags to there meaning
greybus: sdio: rework of event handler
greybus: sdio: fix return of get_cd and get_ro
greybus: sdio: fix call to stop command if no data exist
greybus: sdio: pass only data pointer to tranfer funtion
greybus: sdio: check number of blocks in transfer
greybus: sdio: add need poll to host caps
greybus: sdio: fix transfer buffer handling and blocks counting
greybus: makefile: add needed config options for lights
greybus: lights: add lights implementation
greybus: makefile: use POSIX functions for kernel cmp
greybus: greybus_protocol: remove unused lights macros
greybus: sdio: fix command type defines
greybus: svc: fix endianness for svc message
greybus: gpio: fix generic_handle_irq_desc api change
greybus: sdio: add field to get_caps response
greybus: sdio: fix card removable detection
greybus: sdio: convert greybus ocr values to mmc ones
greybus: sdio: clarify operator precedence
greybus: sdio: send data block details at command request
greybus: sdio: some cleanups in command function
greybus: battery: move battery file to power_supply
greybus: battery: move implementation to power_supply
greybus: makefile: add power_supply config check
greybus: power_supply: rework and operation changes
greybus: manifest: fix bundle descriptor parse
greybus: spi: add rdwr field to transfer descriptor
greybus: spi: add master and device config operations
greybus: spi: fix transfers bigger than greybus payload
greybus: lights: add v4l2 flash operations
greybus: lights: avoid channel torch double free
greybus: lights: default value for v4l2 flash controls
greybus: spi: move chipselect to one byte size
greybus: spi: rename label to undo operation
greybus: spi: validate spi master register
greybus: spi: unregister master on device add fail
greybus: spi: fix and cleanup spi devices handling
greybus: camera: fix data types of operations
greybus: power_supply: do not release failed supply alloc
greybus: power_supply: fix unregister on error path
greybus: power_supply: free supplies at release
greybus: power_supply: fix error path at supplies setup
greybus: power_supply: fix leak getting string properties
greybus: power_supply: fix use after free of power supply
greybus: power_supply: fix check for configured supply
greybus: svc: add interface eject operation
greybus: svc: add intf_eject attribute
greybus: makefile: add requirement for CONFIG_INPUT
greybus: kernel_ver: add handle for lights in 4.5
greybus: lights: remove sync operation and work queue
greybus: svc: add key event handling
greybus: power_supply: rework get descriptors
greybus: uart: fix double free of tty port
greybus: spi: add device_type field to device config
greybus: lights: remove unnecessary checks
greybus: lights: remove has_flash on failure
greybus: lights: fix check for configured lights
greybus: lights: remove unnecessary check
greybus: spi: fix message transfer over greybus
greybus: spi: use timeout request send operation
greybus: spi: rename rdwr field to xfer_flags
greybus: spi: add inprogress bit to xfer_flags
greybus: sdio: avoid extra memory operation at data transfer
greybus: power_supply: change property values to integer
greybus: power_supply: convert greybus properties to power supply properties
greybus: power_supply: remove references to kernel header
greybus: power_supply: add missing defines present in the specification
greybus: sdio: convert vdd kernel values to greybus
greybus: sdio: add switch voltage operation
greybus: greybus_protocol: fix order of sdio get caps response
greybus: sdio: increase maximum segment size
greybus: power_supply: add callback to handle power supply changes
greybus: power_supply: fix update interval check at request handler
greybus: power_supply: invalidate cache at update request
greybus: power_supply: fix name setting location
greybus: light: remove KERNEL_VERSION checks
greybus: sdio: remove KERNEL_VERSION checks
greybus: power_supply: remove KERNEL_VERSION checks
greybus: makefile: fix dependency of spi to spilib
greybus: Kconfig: enable possibility to select light driver
Ryan Lim (3):
greybus: loopback_test: Cancel only the tests running on selected devices
greybus: loopback_test: Fix -z argument bug
greybus: loopback_test: Add -f argument to cancel all tests
Sachin Pandhare (1):
greybus: manifest: simplify descriptor address calculation
Sandeep Patil (12):
greybus: power_supply: reverse version check for new psy API
greybus: hd: add bus_id attribute
greybus: Documentation/sysfs: add entry for host device bus_id
greybus: Documentation/sysfs: fix sysfs paths
greybus: greybus/core: add MODULE uevent var for all control devices
greybus: gpbridge: add gpbridge device type
greybus: gpbridge: make gpbridge device ids start with 1
greybus: gpbridge: add uevent vars for gpbridge devices
greybus: arche-platform: Fix the MODULE_LICENSE to match with header
greybus: gpbridge: rename 'gpbridge' to 'gbphy' everywhere
greybus: svc: remove input device handling in SVC protocol
greybus: es2: use a single bulk ep pair for all greybus data
Svetlin Ankov (1):
greybus: audio: Add Audio Manager
Vaibhav Agarwal (72):
greybus: Audio: Add skeleton code for GB virtual codec driver
greybus: audio: Build audio module conditionally
greybus: audio: Use greybus connection device for codec registration
greybus: audio: Add topology parser for GB codec
greybus: audio: Add GB Audio class protocol functionality in GB codec DAI ops
greybus: audio: Report uevent on GB codec module insertion/removal
greybus: audio: cleanup unnecessary dev_err messages
greybus: audio: Cleanup GB protocol connections in case of abrupt codec removal
greybus: audio: Enable codec module detection on different slots
greybus: audio_codec: update codec_name as per driver->name
greybus: audio: codec driver cleanup
greybus: audio: use variable 'is_connected' to maintain module state
greybus: audio: Enable support for multiple codec modules
greybus: audio: schedule workqueue to perform codec cleanup on module removal
greybus: audio: return success for stop trigger if device removed
greybus: audio: Use single codec driver registration
greybus: audio: Add module specific driver
greybus: audio: fix to resolve multiple audio module playback issue
greybus: audio: Fix widget pointer update during control parsing
greybus: audio: Added jack support to audio module
greybus: audio: Add I2S_RX path related settings
greybus: audio_manager: Split device type into i/p & o/p devices
greybus: audio: Update device type based on widget types
greybus: audio: acquire wakelock during active playback
greybus: audio: acquire wakelock during active playback
greybus: audio: Changes in response to ASoC cleanup
greybus: audio: Reorder gb_deactivate sequence to avoid protocol error
greybus: audio: Rename widget_type to jack_attrubute
greybus: audio: Identify jack type based on attributes reported
greybus: audio:gb_manager: Use proper locking around kobject_xxx
greybus: audio: Cleanup GB Audio bundle driver.
greybus: audio: Remove redundant lock protection & is_connected field
greybus: audio: Reorganize sequence in GB audio bundle & codec driver
greybus: audio: Reduce codec->lock granularity
greybus: audio: Use standard API to set connections' private data
greybus: audio: gb_manager: Use valid argument while removing ida allocated id.
greybus: audio: Fix memory leak
greybus: hd: Add API to release reserved CPorts
greybus: es2: Release reserved cports CDSI0 and CDSI1
greybus: audio: Report warning in case module is already removed
greybus: audio: Report DISCONNECT event after resource cleanup.
greybus: audio: Ratelimit err messages
greybus: Use mute_stream callback to initiate GB data xfer
greybus: Remove apbridgea_shutdown_xx sequence if already done
greybus: audio: Ratelimit err messages in bundle, topology driver
greybus: audio: topology: Add helper API to map controlid with widget name
greybus: added warning message in case of missing widget
greybus: audio: topology: Use csize while computing next_ptr in parser
greybus: audio: topology: Enable enumerated control support
greybus: audio: Fix incorrect codec state modification
greybus: audio: Maintain proper codec state during shutdown sequence.
greybus: audio: Avoid using ARA keyword
greybus: audio: Remove un-necessary goto statement
greybus: audio: Enable audio path based on control switch state only
greybus: audio: Split helper APIs based on stream direction
greybus: audio: Update dai_driver table with appropriate fields
greybus: audio: Update parameters for gbaudio_module_update API
greybus: audio: Use AIF widget to enable path between module & APB
greybus: audio: Add id to identify data connection
greybus: audio: Find data connection based on id
greybus: Remove unused field from data_connection
greybus: audio: Update pm runtime support in dai_ops callback
greybus: audio: Maintain runtime stream params for each DAI
greybus: audio: Maintain module stream state for each data connection
greybus: audio: Report jack removal along with module removal
greybus: audio: Add check for invalid index while mapping control
greybus: Use valid control pointer while freeing memory
greybus: audio: Remove unnecessary num_jack field from module_info
greybus: audio: Added jack_type support in topology struct
greybus: audio: Fetch jack_mask, button_mask from module's topology data
greybus: audio: Report jack events conditionally
greybus: audio: Avoid reporting spurious button events
Vaibhav Hiremath (67):
greybus: platform: Add platform driver for DB3 AP bridge
greybus: db3(arche)-platform: Remove all APB control code
greybus: arche-platform: Rename db3-platform to arche-platform
greybus: arche-apb-ctrl: Add APB control driver
greybus: arche-apb-ctrl: Assert reset to APB at the end of probe
greybus: arche_platform: Remove child's platform device as part of _remove() fn
greybus: arche-platform: Export gpio (reset & sys_boot) to user
greybus: arche-apb-ctrl: Set wake_detect gpio to low initially
greybus: arche-apb-ctrl: Do not bring APB out of reset in probe
greybus: arche-apb-ctrl: Remove wake_detect gpio and related code
greybus: arche-apb-ctrl: deassert reset at the end of probe
greybus: arche-platform: Add wake/detect support along with handshaking with AP
greybus: arche-apb-ctrl: Enable the clocks after regulator enable
greybus: arche-platform: Disable clock as part of driver remove
greybus: arche-apb-ctrl: Remove extra delay in APB reset
greybus: Revert "arche-apb-ctrl: Remove extra delay in APB reset"
greybus: arche-platform: make apb_state common to both platform drivers
greybus: arche-platform: Introduce coldboot_seq fn
greybus: arche-platform: Cancel delayed_work in driver remove fn
greybus: arche-platform: Rename cleanup fn to poweroff_seq
greybus: arche-platform: Add state variable to driver data
greybus: arche-platform: Pull wake/detect low in poweroff
greybus: arche_platform: Add sysfs to allow user to change state
greybus: arche-platform: Introduce FW_FLASHING state
greybus: arche-apb-ctrl: Re-arrange init sequence
greybus: arche-apb-ctrl: Rename init_seq and cleanup fn
greybus: arche-apb-ctrl: Set default state value to OFF
greybus: arche-apb-ctrl: Pass only platform_device to _seq fns
greybus: arche-apb-ctrl: Add sysfs to allow user to change state
greybus: arche-apb-ctrl: Introduce ara,init-disable property for APB
greybus: arche-apb-ctrl: Rename internal operational state fns
greybus: arche-apb-ctrl: Export operational fns from driver
greybus: arche-platform: Fix boot, poweroff and fw_flashing seq with APBs
greybus: arche-platform: Do not export any gpios
greybus: arche-apb-ctrl: Do not export any gpios
greybus: arche-platform: Control usb3613 from arche-platform
greybus: arche-apb-ctrl: Do not coldboot APBs in probe
greybus: arche-platform: Return immediately if in same state from state change fns
greybus: arche-platform: Avoid doing same thing again in poweroff fn
greybus: arche-apb-ctrl: Return immediately if in same state from state change fns
greybus: arche-platform: Fix exit path in probe fn
greybus: arche-apb-ctrl: Remove extra delay in APB reset
greybus: arche-platform: Convert delayed work to do only hub3613 configuration
greybus: arche-platform: Add wake detect state based on functionality
greybus: arche-platform: Enable interrupt support on wake/detect line
greybus: arche-platform: Assert wake/detect after SVC reset without delay
greybus: arche-platform: Set direction of wake/detect gpio in poweroff fn
greybus: arche-platform: Disable wake/detect interrupt in poweroff
greybus: arche-platform: Make sure APB power cycles on coldboot
greybus: arche-platform: Disable HUB3613 only in APB poweroff fn
greybus: arche-platfrom: Get rid of 2sec delay in USB3613 configuration
greybus: gpbridge: Add gpbridge driver init/exit helper macros
greybus: connection: export gb_connection_disable_rx() fn
greybus: Documentation: Add sysfs information about bridged-phy device
greybus: arche-platform: Rename spinlock variable lock => wake_detect_lock
greybus: arche-platform: Export fn to allow timesync driver to change the state
greybus: arche-platform: Do not assert wake/detect signal to SVC
greybus: arche-platform: Rename assert_wakedetect=>arche_platform_wd_irq_en
greybus: arche-platform: Make fw_flashing_seq() return error
greybus: arche-platform: Enter ACTIVE state only from OFF state
greybus: arche-platform: Enable SVC clock during FW_FLASHING state
greybus: svc: Pass the correct pointer to input_free_device()
greybus: ratelimit errors usually seen on unipro_reset
greybus: arche-platform: Add support for init-off feature
greybus: arche-apb-ctrl: Rename ara,init-disable => arche,init-disable
greybus: arche-platform: Reset SVC & APB only if turned off by SUSPEND_PREPARE
greybus: arche-platform: Add support for SPI bus sharing for Mihi
Viresh Kumar (410):
greybus: Fix parameters of core_param()
greybus: .gitignore: minor updates
greybus: gitignore: few more additions
greybus: manifest: no need to initialize 'result'
greybus: manifest: initialize variable during definition
greybus: manifest: don't free unallocated resources
greybus: Fix missing gb_protocol_exit() on gb_exit()
greybus: don't set ->dev.driver to NULL when it is already NULL
greybus: debug: mark debug messages with pr_debug() instead of printk
greybus: sysfs: generalize gb_module_attr() to capture more cases
greybus: module: free resources properly on failures
greybus: interface: free resources properly on failures
greybus: connection: free resources properly on failures
greybus: manifest: remove extra loop for finding module descriptor
greybus: manifest: improve print message
greybus: core: remove unnecessary braces
greybus: core: use 'drv' instead of dev->driver
greybus: operation: free resources in the reverse order of allocation
greybus: nullify dangling pointers
greybus: connection: try cancelling operations only if list isn't empty
greybus: manifest: make sure manifest_descs list is empty before parsing manifest
greybus: connection: fix duplicating naming in _gb_hd_connection_insert()
greybus: core: re-use gb_module_find() in gb_remove_module()
greybus: module: move gb_module_find() to a more logical location
greybus: module: move module specific code to module.c
greybus: interface: move gb_module_interface_init() to interface.c
greybus: ap: remove extra parameter to convert_ap_message()
greybus: ap: keep all svc function routines together
greybus: Pass '*ptr' to sizeof() wherever possible
greybus: ap: function_id is already set by svc_msg_alloc()
greybus: operation: don't complete operation twice
greybus: Random spell fixes
greybus: es1-ap-usb: don't use 'es1' after it is freed
greybus: Remove "-gb" suffix from .c files
greybus: Remove "gb-" prefix from .c files
greybus: uart: s/REQ/TYPE
greybus: create get_version() routines with the help of a macro
greybus: remove unused version-response structs
greybus: i2c: fix name conflict between function and struct: gb_i2c_transfer_response
greybus: i2c: fix name conflict between function and struct: gb_i2c_transfer_request
greybus: spi: add bridged-PHY spi protocol driver
greybus: spi:fix sparse warnings
greybus: gpb: Fix print mistakes
greybus: hid: add HID class driver
greybus: operation: s/status/result to match field name
greybus: operation: Fix comment mistake
greybus: interface: remove double underscore from fn name
greybus: core: place module_{init|exit}() right below the routines
greybus: hid: don't support OUTPUT report over interrupt channel
greybus: core: Don't initialize greybus if it is disabled
greybus: manifest: descriptor size should be >= header size
greybus: manifest: don't need to check for desc_size == 0 anymore
greybus: Greybus: Place module_init/exit() right after respective routines
greybus: interface: put module->dev on failures
greybus: kernel_ver.h: include <linux/kernel.h> to fix warning
greybus: Unregister devices to get them freed
greybus: interface: Fetch interface id instead of module id during setup
greybus: Add bundle descriptor type
greybus: hid: Use payload-size to get report size
greybus: bundle: Create bundles using bundle descriptors
greybus: bundle: Initialize all bundles on link-up
greybus: manifest: Use interface descriptor instead of module descriptor to get information
greybus: drop module descriptors
greybus: driver corresponds to a bundle, not interface
greybus: bundle: s/class_type/class
greybus: endo: Use a real endo id
greybus: endo: Rename endo's 'type' as 'id'
greybus: Explicitly mark structures as packed
greybus: manifest: Account for padding in string descriptor
greybus: manifest: Warn if descriptor size > expected size
greybus: manifest: Remove vendor, product and unique-id from interface descriptor
greybus: manifest: drop interface-version field
greybus: Remove class descriptor
greybus: Explicitly add pad-bytes to manifest descriptors
greybus: Move id-field before bundle-field in CPort Descriptor.
greybus: endo: Add code to validate Endo ID
greybus: endo: Create modules after validating Endo ID
greybus: raw: include uaccess.h to fix warning
greybus: gpio: fix tab/space mistake
greybus: spi: Move structure definitions into gpbridge.h
greybus: protocol: remove leading underscore from _gb_protocol_find()
greybus: connection: name routines consistently
greybus: endo: name routines consistently
greybus: interface: name routines consistently
greybus: module: name routines consistently
greybus: raw: move module_{init|exit} to the end of functions
greybus: gpbridge: rename gpb.c to gpbridge.c
greybus: protocol: Add gb_gpbridge_protocol_driver()
greybus: Use gb_gpbridge_protocol_init()
greybus: endo: Add missing '\n' sprintf() for sysfs files
greybus: operation: Remove unused variable 'buffer'
greybus: s/bundle_cport_id/intf_cport_id
greybus: Generate greybus wide unique ids for endo devices
greybus: Prefix module-id with endo id
greybus: Tear down devices in the reverse order
greybus: endo: Fix compilation warning
greybus: manifest: Allow bundles/connections list to be non-empty at manifest parsing
greybus: Add control protocol support
greybus: interface: Get manifest using Control protocol
greybus: hotplug function doesn't need to contain manifest now
greybus: connection: send [dis]connected events over control CPort
greybus: control: Warn if non-control cport/bundles have control protocol/classes
greybus: define more greybus classes
greybus: svc: Fix doc-style comment
greybus: svc: Use macro's for major/minor numbers
greybus: Rename gb_gpbridge_protocol_driver() as gb_builtin_protocol_driver()
greybus: control: Use gb_builtin_protocol_driver()
greybus: connection: bind protocol after the connection is operational
greybus: connection: intf_cport_id is local to an interface
greybus: svc: Drop svc_set_route_send()
greybus: interface: device_id belongs to the interface
greybus: bundles: remove gb_bundle(s)_init()
greybus: connection: Exit connection before destroying it
greybus: interface: Fix comment mistake (s/add/init)
greybus: connection: Use connection->dev for printing messages
greybus: svc: SVC is part of greybus core
greybus: svc: save pointer to struct gb_svc in struct gb_interface
greybus: Cleanup connection leftovers
greybus: connections: comment style fix
greybus: interface: drop stale comment
greybus: sync protocol and class definitions
greybus: protocol: move version_response structure to greybus_protocols.h
greybus: Define device ids for AP/SVC/modules
greybus: Define cport/bundle for SVC protocol
greybus: connection: don't send connected/disconnected events for SVC connection
greybus: connection: Create gb_connection_create_range() to specify hd-cport-id range
greybus: connection: Allow a bundle-less connection
greybus: interface: Update gb_create_control_connection() to support SVC protocol
greybus: svc: Add helpers to create AP<->SVC connection
greybus: svc: Implement SVC requests
greybus: svc: No need of a separate version request
greybus: svc: Fix build warning on 64bit systems
greybus: svc: fully initialize the partially initialized connection
greybus: svc: connection is created only once now
greybus: svc: remove interface->svc pointer
greybus: interface: localize gb_interface_destroy()
greybus: connection: compare hd_cport_id for matching SVC protocol
greybus: es1/2: Lets start using svc protocol
greybus: Switch to Cport 0 for svc and control protocol
greybus: get rid of old svc-protocol
greybus: es1: create svc connection early enough
greybus: remove svc_msg.h
greybus: initialize svc connection while creating hd
greybus: es1/2: Lets start using svc protocol
greybus: Switch to Cport 0 for svc and control protocol
greybus: get rid of old svc-protocol
greybus: es1: create svc connection early enough
greybus: remove svc_msg.h
greybus: initialize svc connection while creating hd
greybus: connection: remove special check for svc cport id
greybus: operation: Drop alignment attribute from operation message header
greybus: operation: Move operation header to greybus_protocols.h
greybus: svc: Move '{' to a new line in function definitions
greybus: raw: Print expected/actual payload size on mismatch
greybus: gpio: Print expected/actual payload size on mismatch
greybus: svc: Print expected/actual payload size on mismatch
greybus: svc: Handle hotplug request from a workqueue
greybus: sdio: error out only for smaller payloads received
greybus: hid: spell fix (s/infomation/information)
greybus: battery: Use (already defined) major/minor macros
greybus: gpio: Use (already defined) major/minor macros
greybus: hid: Use (already defined) major/minor macros
greybus: i2c: Use (already defined) major/minor macros
greybus: pwm: Use (already defined) major/minor macros
greybus: spi: Use (already defined) major/minor macros
greybus: uart: Use (already defined) major/minor macros
greybus: usb: Use (already defined) major/minor macros
greybus: vibrator: Use (already defined) major/minor macros
greybus: svc: Remove FIXME for firmware download
greybus: audio: Add '<' to the print message for short messages
greybus: svc: error out only for smaller payloads received
greybus: connection: disconnect connection on failures during initialization
greybus: define greybus wide protocol request numbers
greybus: connection: Save major/minor supported by module
greybus: connection: request protocol version before initializing connection
greybus: audio: Drop get_version support
greybus: battery: Drop get_version support
greybus: control: Drop get_version support
greybus: gpio: Drop get_version support
greybus: hid: Drop get_version support
greybus: i2c: Drop get_version support
greybus: loopback: Drop get_version support
greybus: pwm: Drop get_version support
greybus: raw: Drop get_version support
greybus: sdio: Drop get_version support
greybus: spi: Drop get_version support
greybus: uart: Drop get_version support
greybus: usb: Drop get_version support
greybus: vibrator: Drop get_version support
greybus: svc: preserve major/minor of protocol supported by SVC
greybus: protocol: Drop define_get_version support
greybus: protocol: Remove unnecessary params of gb_protocol_get_version()
greybus: interface: Preserve data received during hotplug event
greybus: protocol: Create request structure from within gb_protocol_get_version()
greybus: Add firmware protocol driver
greybus: connection: Send protocol version for firmware protocol
greybus: protocol: send own protocol version while requesting it
greybus: battery: Move request/response structure/definitions to greybus_protocols.h
greybus: hid: Move request/response structure/definitions to greybus_protocols.h
greybus: raw: Move request/response structure/definitions to greybus_protocols.h
greybus: connection: Destroy wq on failure
greybus: connection: Don't free resources freed by gb_connection_release()
greybus: Drop protocol specific _PROTOCOL_VERSION and _INVALID macros
greybus: operation: print message type on errors
greybus: loopback: Send some sensible data
greybus: loopback: Print error on data mismatch
greybus: operation: Fix wrong order of arguments
greybus: svc: get hd directly from connection
greybus: svc: Include system headers at the top
greybus: svc: reject invalid state requests
greybus: greybus_protocols: svc: Add missing comment
greybus: greybus_protocols: Pack structure as required by the module firmware
greybus: greybus_protocols: Pack all request/response structure
greybus: svc: No need to return errors from [gb_]svc_connection_destroy()
greybus: svc: Kill unnecessary wrapper functions
greybus: svc: unexport few internal functions
greybus: bundle: kill unnecessary forward declaration of routine
greybus: connection: staticize gb_connection_init()
greybus: connection: convert connected dev_warn() to dev_err()
greybus: connection: no need to verify connection->protocol
greybus: connection: Propagate error properly
greybus: connection: call gb_connection_exit() from gb_connection_destroy()
greybus: connection: protocol can be NULL in gb_connection_exit()
greybus: spi: Allow spi-core to allocate bus numbers dynamically
greybus: manifest: don't reject the interface on failing to initialize a cport
greybus: manifest: release cport descriptors to avoid 'excess descriptors' warning
greybus: connection: call gb_svc_connection_create() from gb_connection_init()
greybus: connection: destroy svc connection on connection exit or errors
greybus: connection: Call connection_destroy() from gb_connection_svc_connection_destroy()
greybus: Add flags to struct gb_protocol
greybus: svc: No need of per-direction route-create requests
greybus: svc: destroy the route on module hot-unplug
greybus: connection: Add sysfs 'ap_cport_id' file for connections
greybus: svc: Implement DME peer get/set attributes helpers
greybus: firmware: firmware's file name is 32 characters long
greybus: firmware: Fix incorrect firmware file's name
greybus: firmware: firmware image name has .tftf instead of .fw
greybus: firmware: Send AP-ready operation
greybus: Documentation: Update sysfs-bus-greybus
greybus: svc: Use 'dev' instead '&op->connection->dev'
greybus: svc: Allow consecutive hotplug events for the same module
greybus: svc: Read and clear module's boot status
greybus: interface: Pass interface pointer to gb_interface_remove()
greybus: interface: gb_interface_create() isn't called for existing interfaces
greybus: firmware: drop 'stage' from ready-to-boot request
greybus: firmware: Don't send control-disconnected event for firmware protocol
greybus: svc: skip setting flags for boot over unipro
greybus: bundle: fix double freeing of bundle structure
greybus: interface: fix double freeing of interface structure
greybus: module: fix double freeing of module structure
greybus: endo: move greybus_endo_setup() to endo.c
greybus: db3-platform: Use dev_info for success messages
greybus: es2: Send cport-id in wValue field to usb_control_msg()
greybus: es2: remove fake vendor/product ids
greybus: es1: Remove the (now) unused es1 driver
greybus: firmware: Fetch es2 VID/PID to distinguish module vendors
greybus: Documentation: Fix N-svc directory name
greybus: Documentation: Arrange entries in alphabetical order
greybus: svc: Use -EIO instead of -EINVAL for unipro errors
greybus: don't use %h and %hh for printing short and char variables
greybus: Prefix hexadecimal values with 0x while printing them
greybus: interface: Prefix hexadecimal values with '0x'
greybus: interface: Prefix DDBL1 attributes with ddbl1_ instead of unipro_
greybus: interface: Expose DDBL1 manufacturing and production id in sysfs
greybus: control: Use Macro's instead of direct values for major/minor
greybus: svc: Set interface's hotplug attributes before using them
greybus: interface: Receive serial-number on hotplug event
greybus: interface: Fetch and expose version of interface's firmware
greybus: arche-platform: arche_pdata is guaranteed to be valid
greybus: arche-platform: Export GPIOs after populating APBs
greybus: arche-platform: Put APB in reset if of_platform_populate() fails
greybus: arche-platform: svc_reset_gpio can't be invalid in arche_platform_cleanup()
greybus: arche-platform: propagate errors returned by gpiolib
greybus: arche-apb: Spelling and whitespace fixes
greybus: arche-apb: platform data 'apb' is guaranteed to be valid
greybus: arche-apb: Replace gpio_is_valid() with gpio < 0 checks
greybus: arche-apb: devm_regulator_get() doesn't return NULL
greybus: arche-apb: NULL is a valid regulator
greybus: arche-apb: Properly use dev_err/info/warn
greybus: arche-apb: Don't use gpio after failing to request it
greybus: arche-apb: Do cleanup within apb_ctrl_init_seq() for error cases
greybus: arche-apb: Drop unnecessary checks
greybus: arche: Remove unwanted headers and rearrange others
greybus: svc: Expose and retain VID/PID received from bootrom for ES2
greybus: audio_codec: Free gccodec on codec probe failure
greybus: audio_codec: Free gccodec on dia probe failure
greybus: audio_codec: convert to bundle driver
greybus: audio: Rename Audio class and remove the unused one
greybus: audio: Fix sparse warnings
greybus: control: Fix sparse warnings
greybus: connection: Fix sparse warnings around locking
greybus: raw: Use consistent label names in connection_init()
greybus: raw: Don't use (possibly) uninitialized raw->device in gb_raw_receive()
greybus: raw: convert to bundle driver
greybus: loopback: convert to bundle driver
greybus: lights: Break light setup into two parts
greybus: lights: convert to bundle driver
greybus: power_supply: Break supply setup into two parts
greybus: power_supply: convert to bundle driver
greybus: manifest: Parse cports (within a bundle) in the order from manifest blob
greybus: interface: Add print messages on interface initialization/removal
greybus: hid: Don't disable connection-tx before destroying hid-device
greybus: hid: Replace WARN_ON() with dev_err()
greybus: checkpatch.pl: Mark --no-tree as default option
greybus: firmware: Rename to bootrom protocol
greybus: bootrom: fix typo
greybus: audio-codec: Staticize few routines to fix build warnings
greybus: es2: Allow proper release of greybus host device
greybus: es2: disable cport-in after remove greybus host device
greybus: es2: don't protest when getting -ENOENT/ESHUTDOWN USB errors
greybus: es2: move ap_disconnect() below ap_probe()
greybus: arche-ctrl: Don't expose driver internals to arche-platform driver
greybus: Bundle: Initialize dma_mask for bundle device
greybus: firmware: Add firmware management bundle driver
greybus: firmware: Add firmware-download protocol driver
greybus: fw-download: Create a macro for length of firmware file's name
greybus: bootrom: Implement timeouts to detect Module failures
greybus: Revert "interface: Fetch and expose version of interface's firmware"
greybus: uart: Kill reference_count hack
greybus: fw-download: Manage firmware requests with kref
greybus: fw-download: Introduce timeouts for firmware downloads
greybus: bootrom: Implement timeouts to detect Module failures
greybus: fw-download: Replace timer with delayed-work
greybus: gpbridge: Expose protocol_id in sysfs
greybus: core: Rename greybus_module_match()
greybus: Fix probing of gpbridge devices
greybus: gpbridge: Export few routines
greybus: gpbridge: Create module_gpbridge_driver()
greybus: gpio: Create separate module
greybus: i2c : Create separate module
greybus: pwm: Create separate module
greybus: sdio: Create separate module
greybus: spi: Create separate module
greybus: uart: Create separate module
greybus: usb: Create separate module
greybus: gpbridge: Rename gb-phy.ko module
greybus: gpbridge: Remove (now) unused macro gb_gpbridge_builtin_driver()
greybus: Remove bridge PHY protocol specific classes
greybus: spi: Restructure spi.c to share it with other bundle drivers
greybus: spi: Separate out spilib from spi bridged PHY bundle driver
greybus: fw-management: Add firmware-management protocol driver
greybus: fw-management: Add ioctl to initiate mode-switch
greybus: fw-management: Free fw-mgmt only after all users are gone
greybus: firmware: Add SPI protocol support
greybus: Documentation: Document firmware-management interfaces
greybus: fw-management: Replace double-tabs with space for structures
greybus: fw-management: Fix 'make check' warnings
greybus: fw-management: Update header's license
greybus: fw-core: destroy connections on error
greybus: fw-management: Fix error message's text
greybus: Documentation/sysfs: Rename gpbridge to gbphy
greybus: Documentation/sysfs: gbphy0 is an invalid name
greybus: Remove unused VERSION specific macros
greybus: gbphy: Remove protocol specific version handling
greybus: fw-mgmt: Add hooks to do mode-switch
greybus: control: Mode-switch is a 'core' operation
greybus: interface: Mode switch takes over a second
greybus: Add comment about the mode_switch_started flag
greybus: pwm: Fix compilation with v4.3 kernel
greybus: bootrom: Compile as a separate module
greybus: core: Make greybus_match_one_id() return bool
greybus: Use BIT(2) for GREYBUS_ID_MATCH_CLASS
greybus: Remove legacy suspend/resume callbacks
greybus: hd: Export host device tracepoint from hd.c
greybus: connection: Avoid unnecessary line breaks
greybus: firmware: Fix spelling mistake
greybus: Remove extra blank lines
greybus: connection: Return bool from gb_connection_intf_find()
greybus: camera: Initialize mutex before using it
greybus: svc: disable connection after all its users are gone
greybus: manifest: Disallow reuse of control cport
greybus: firmware: Add license and copyright header to application
greybus: firmware: prefix char-device with 'gb-'
greybus: connection: add trace events before disabling connection
greybus: Revert "greybus: ratelimit errors usually seen on unipro_reset"
greybus: don't use spin_lock_irq()
greybus: uart: don't use spin_lock_irq()
greybus: es2.c: don't use spin_lock_irq()
greybus: gb_connections_lock: don't use spin_lock_irq()
greybus: bootrom: Enhance timeout error message
greybus: bootrom: send timeout in milliseconds to gb_bootrom_set_timeout()
greybus: bootrom: Wait for 10 seconds for mode-switch
greybus: kernel_ver: Add kstrtobool()
greybus: light: Initialize mutex before using it
greybus: Add Component Authentication Protocol support
greybus: Documentation: Document Authentication interfaces
greybus: es2: Fix 'make check' warnings with arpc
greybus: audio_topology: Fix compile warning
greybus: bootrom: Use "s2l" instead of boot stage in package name
greybus: interface: add power_state attribute
greybus: firmware: Improve test application
greybus: spilib: make spilib independent of gbphy
greybus: control: Print bundle-id in print messages
greybus: bootrom: Skip setting timeout in failure path of size request
greybus: bootrom: Don't print duplicate error messages
greybus: bootrom: Set timeout before sending AP_READY
greybus: bootrom: Create gb_bootrom_cancel_timeout()
greybus: bootrom: Rename download_firmware() as find_firmware()
greybus: firmware: Add runtime PM support
greybus: interface: Rename *_NO_PM as *_NO_BUNDLE_ACTIVATE
greybus: interface: Add quirk for no PM for S2 Loader
greybus: firmware: Add new status types for backend updated request
greybus: firmware: Add 'status' byte to backend fw version response
greybus: firmware: Update Documentation and sample application
greybus: firmware: use 'tag' instead of 'firmware_tag'
greybus: firmware: Fix typo in documentation
greybus: firmware: Remove extra parenthesis
greybus: firmware: s/_LEN/_SIZE
greybus: firmware: s/should/must
greybus: firmware: add comment to show size calculations
Vishal Bhoj (1):
greybus: build: android: avoid building for targets that don't have kernel binary
Documentation/ABI/testing/sysfs-bus-greybus | 275 +++
drivers/Kconfig | 2 +
drivers/Makefile | 1 +
.../greybus/Documentation/firmware/authenticate.c | 139 ++
.../Documentation/firmware/firmware-management | 333 +++
drivers/greybus/Documentation/firmware/firmware.c | 262 +++
drivers/greybus/Kconfig | 219 ++
drivers/greybus/Makefile | 96 +
drivers/greybus/arche-apb-ctrl.c | 522 +++++
drivers/greybus/arche-platform.c | 828 +++++++
drivers/greybus/arche_platform.h | 39 +
drivers/greybus/arpc.h | 109 +
drivers/greybus/audio_apbridgea.c | 207 ++
drivers/greybus/audio_apbridgea.h | 156 ++
drivers/greybus/audio_codec.c | 1132 ++++++++++
drivers/greybus/audio_codec.h | 283 +++
drivers/greybus/audio_gb.c | 228 ++
drivers/greybus/audio_manager.c | 184 ++
drivers/greybus/audio_manager.h | 83 +
drivers/greybus/audio_manager_module.c | 258 +++
drivers/greybus/audio_manager_private.h | 28 +
drivers/greybus/audio_manager_sysfs.c | 102 +
drivers/greybus/audio_module.c | 482 +++++
drivers/greybus/audio_topology.c | 1442 +++++++++++++
drivers/greybus/authentication.c | 429 ++++
drivers/greybus/bootrom.c | 524 +++++
drivers/greybus/bundle.c | 253 +++
drivers/greybus/bundle.h | 90 +
drivers/greybus/camera.c | 1400 ++++++++++++
drivers/greybus/connection.c | 938 ++++++++
drivers/greybus/connection.h | 129 ++
drivers/greybus/control.c | 635 ++++++
drivers/greybus/control.h | 65 +
drivers/greybus/core.c | 361 ++++
drivers/greybus/debugfs.c | 31 +
drivers/greybus/devices | 11 +
drivers/greybus/es2.c | 1597 ++++++++++++++
drivers/greybus/firmware.h | 42 +
drivers/greybus/fw-core.c | 312 +++
drivers/greybus/fw-download.c | 465 ++++
drivers/greybus/fw-management.c | 721 +++++++
drivers/greybus/gb-camera.h | 127 ++
drivers/greybus/gbphy.c | 360 ++++
drivers/greybus/gbphy.h | 110 +
drivers/greybus/gpio.c | 767 +++++++
drivers/greybus/greybus.h | 154 ++
drivers/greybus/greybus_authentication.h | 120 ++
drivers/greybus/greybus_firmware.h | 120 ++
drivers/greybus/greybus_id.h | 26 +
drivers/greybus/greybus_manifest.h | 177 ++
drivers/greybus/greybus_protocols.h | 2268 ++++++++++++++++++++
drivers/greybus/greybus_trace.h | 531 +++++
drivers/greybus/hd.c | 257 +++
drivers/greybus/hd.h | 90 +
drivers/greybus/hid.c | 536 +++++
drivers/greybus/i2c.c | 343 +++
drivers/greybus/interface.c | 1316 ++++++++++++
drivers/greybus/interface.h | 88 +
drivers/greybus/light.c | 1359 ++++++++++++
drivers/greybus/log.c | 132 ++
drivers/greybus/loopback.c | 1365 ++++++++++++
drivers/greybus/manifest.c | 535 +++++
drivers/greybus/manifest.h | 16 +
drivers/greybus/module.c | 238 ++
drivers/greybus/module.h | 34 +
drivers/greybus/operation.c | 1239 +++++++++++
drivers/greybus/operation.h | 210 ++
drivers/greybus/power_supply.c | 1141 ++++++++++
drivers/greybus/pwm.c | 338 +++
drivers/greybus/raw.c | 381 ++++
drivers/greybus/sdio.c | 884 ++++++++
drivers/greybus/spi.c | 79 +
drivers/greybus/spilib.c | 565 +++++
drivers/greybus/spilib.h | 24 +
drivers/greybus/svc.c | 1486 +++++++++++++
drivers/greybus/svc.h | 109 +
drivers/greybus/svc_watchdog.c | 198 ++
drivers/greybus/timesync.c | 1357 ++++++++++++
drivers/greybus/timesync.h | 45 +
drivers/greybus/timesync_platform.c | 77 +
drivers/greybus/tools/.gitignore | 1 +
drivers/greybus/tools/Android.mk | 10 +
drivers/greybus/tools/Makefile | 31 +
drivers/greybus/tools/README.loopback | 198 ++
drivers/greybus/tools/lbtest | 168 ++
drivers/greybus/tools/loopback_test.c | 1000 +++++++++
drivers/greybus/uart.c | 1075 ++++++++++
drivers/greybus/usb.c | 247 +++
drivers/greybus/vibrator.c | 249 +++
89 files changed, 37594 insertions(+)
create mode 100644 Documentation/ABI/testing/sysfs-bus-greybus
create mode 100644 drivers/greybus/Documentation/firmware/authenticate.c
create mode 100644 drivers/greybus/Documentation/firmware/firmware-management
create mode 100644 drivers/greybus/Documentation/firmware/firmware.c
create mode 100644 drivers/greybus/Kconfig
create mode 100644 drivers/greybus/Makefile
create mode 100644 drivers/greybus/arche-apb-ctrl.c
create mode 100644 drivers/greybus/arche-platform.c
create mode 100644 drivers/greybus/arche_platform.h
create mode 100644 drivers/greybus/arpc.h
create mode 100644 drivers/greybus/audio_apbridgea.c
create mode 100644 drivers/greybus/audio_apbridgea.h
create mode 100644 drivers/greybus/audio_codec.c
create mode 100644 drivers/greybus/audio_codec.h
create mode 100644 drivers/greybus/audio_gb.c
create mode 100644 drivers/greybus/audio_manager.c
create mode 100644 drivers/greybus/audio_manager.h
create mode 100644 drivers/greybus/audio_manager_module.c
create mode 100644 drivers/greybus/audio_manager_private.h
create mode 100644 drivers/greybus/audio_manager_sysfs.c
create mode 100644 drivers/greybus/audio_module.c
create mode 100644 drivers/greybus/audio_topology.c
create mode 100644 drivers/greybus/authentication.c
create mode 100644 drivers/greybus/bootrom.c
create mode 100644 drivers/greybus/bundle.c
create mode 100644 drivers/greybus/bundle.h
create mode 100644 drivers/greybus/camera.c
create mode 100644 drivers/greybus/connection.c
create mode 100644 drivers/greybus/connection.h
create mode 100644 drivers/greybus/control.c
create mode 100644 drivers/greybus/control.h
create mode 100644 drivers/greybus/core.c
create mode 100644 drivers/greybus/debugfs.c
create mode 100644 drivers/greybus/devices
create mode 100644 drivers/greybus/es2.c
create mode 100644 drivers/greybus/firmware.h
create mode 100644 drivers/greybus/fw-core.c
create mode 100644 drivers/greybus/fw-download.c
create mode 100644 drivers/greybus/fw-management.c
create mode 100644 drivers/greybus/gb-camera.h
create mode 100644 drivers/greybus/gbphy.c
create mode 100644 drivers/greybus/gbphy.h
create mode 100644 drivers/greybus/gpio.c
create mode 100644 drivers/greybus/greybus.h
create mode 100644 drivers/greybus/greybus_authentication.h
create mode 100644 drivers/greybus/greybus_firmware.h
create mode 100644 drivers/greybus/greybus_id.h
create mode 100644 drivers/greybus/greybus_manifest.h
create mode 100644 drivers/greybus/greybus_protocols.h
create mode 100644 drivers/greybus/greybus_trace.h
create mode 100644 drivers/greybus/hd.c
create mode 100644 drivers/greybus/hd.h
create mode 100644 drivers/greybus/hid.c
create mode 100644 drivers/greybus/i2c.c
create mode 100644 drivers/greybus/interface.c
create mode 100644 drivers/greybus/interface.h
create mode 100644 drivers/greybus/light.c
create mode 100644 drivers/greybus/log.c
create mode 100644 drivers/greybus/loopback.c
create mode 100644 drivers/greybus/manifest.c
create mode 100644 drivers/greybus/manifest.h
create mode 100644 drivers/greybus/module.c
create mode 100644 drivers/greybus/module.h
create mode 100644 drivers/greybus/operation.c
create mode 100644 drivers/greybus/operation.h
create mode 100644 drivers/greybus/power_supply.c
create mode 100644 drivers/greybus/pwm.c
create mode 100644 drivers/greybus/raw.c
create mode 100644 drivers/greybus/sdio.c
create mode 100644 drivers/greybus/spi.c
create mode 100644 drivers/greybus/spilib.c
create mode 100644 drivers/greybus/spilib.h
create mode 100644 drivers/greybus/svc.c
create mode 100644 drivers/greybus/svc.h
create mode 100644 drivers/greybus/svc_watchdog.c
create mode 100644 drivers/greybus/timesync.c
create mode 100644 drivers/greybus/timesync.h
create mode 100644 drivers/greybus/timesync_platform.c
create mode 100644 drivers/greybus/tools/.gitignore
create mode 100644 drivers/greybus/tools/Android.mk
create mode 100644 drivers/greybus/tools/Makefile
create mode 100644 drivers/greybus/tools/README.loopback
create mode 100755 drivers/greybus/tools/lbtest
create mode 100644 drivers/greybus/tools/loopback_test.c
create mode 100644 drivers/greybus/uart.c
create mode 100644 drivers/greybus/usb.c
create mode 100644 drivers/greybus/vibrator.c
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-14 10:09 [GIT PULL] Greybus driver subsystem for 4.9-rc1 Greg KH
@ 2016-09-14 17:36 ` Mark Rutland
[not found] ` <20160915122141.650632149@bubbles.kroah.org>
` (2 subsequent siblings)
3 siblings, 0 replies; 84+ messages in thread
From: Mark Rutland @ 2016-09-14 17:36 UTC (permalink / raw)
To: Greg KH
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer, marc.zyngier,
linux-arm-kernel
Hi Greg,
On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
> Given that it's never a good idea to keep subsystems out of the mainline
> kernel, I've put together this pull request that adds the greybus driver
> layer to drivers/greybus/. Because this was 2 1/2 years of work, with
> many many developers contributing, I didn't want to flatten all of their
> effort into a few small patches, as that wouldn't be very fair. So I've
> built a git tree with all of the changes going back to the first commit,
> and merged it into the kernel tree, just like btrfs was merged into the
> kernel.
> Unless people point out some major problems with this, I'd like to get
> it merged into 4.9-rc1.
I'm extremely concerned that these patches have *never* seen upstream
review, and this pull request gives no real opportunity for people to
make a judgement regarding the code, as many relevant parties have not
been Cc'd.
>From a quick scan of the git tree, I can see code (that isn't even
placed under staging/) for which I have fundamental objections to as a
maintainer, and has not been Cc'd to a relevant list.
For example, I see commit 5a450477311fbfe2 ("greybus: timesync: Add
timesync core driver"). This states that it directly accesses the ARMv7
architected timer, though it's unclear as to precisely what it's doing
since it introduces an (undocumented) compatible string, and what should
be an unnecessary devicetree property.
That's never gone to the linux-arm-kernel mainline list, myself or Marc
(as maintainers of the arch timer driver), nor has the binding seen any
review on the devicetree mailing list.
Given that, for at least that patch, NAK.
Thanks,
Mark.
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-14 17:36 ` Mark Rutland
0 siblings, 0 replies; 84+ messages in thread
From: Mark Rutland @ 2016-09-14 17:36 UTC (permalink / raw)
To: linux-arm-kernel
Hi Greg,
On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
> Given that it's never a good idea to keep subsystems out of the mainline
> kernel, I've put together this pull request that adds the greybus driver
> layer to drivers/greybus/. Because this was 2 1/2 years of work, with
> many many developers contributing, I didn't want to flatten all of their
> effort into a few small patches, as that wouldn't be very fair. So I've
> built a git tree with all of the changes going back to the first commit,
> and merged it into the kernel tree, just like btrfs was merged into the
> kernel.
> Unless people point out some major problems with this, I'd like to get
> it merged into 4.9-rc1.
I'm extremely concerned that these patches have *never* seen upstream
review, and this pull request gives no real opportunity for people to
make a judgement regarding the code, as many relevant parties have not
been Cc'd.
>From a quick scan of the git tree, I can see code (that isn't even
placed under staging/) for which I have fundamental objections to as a
maintainer, and has not been Cc'd to a relevant list.
For example, I see commit 5a450477311fbfe2 ("greybus: timesync: Add
timesync core driver"). This states that it directly accesses the ARMv7
architected timer, though it's unclear as to precisely what it's doing
since it introduces an (undocumented) compatible string, and what should
be an unnecessary devicetree property.
That's never gone to the linux-arm-kernel mainline list, myself or Marc
(as maintainers of the arch timer driver), nor has the binding seen any
review on the devicetree mailing list.
Given that, for at least that patch, NAK.
Thanks,
Mark.
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-14 17:36 ` Mark Rutland
@ 2016-09-14 18:07 ` Greg KH
-1 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-14 18:07 UTC (permalink / raw)
To: Mark Rutland
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer, marc.zyngier,
linux-arm-kernel
On Wed, Sep 14, 2016 at 06:36:26PM +0100, Mark Rutland wrote:
> Hi Greg,
>
> On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
> > Given that it's never a good idea to keep subsystems out of the mainline
> > kernel, I've put together this pull request that adds the greybus driver
> > layer to drivers/greybus/. Because this was 2 1/2 years of work, with
> > many many developers contributing, I didn't want to flatten all of their
> > effort into a few small patches, as that wouldn't be very fair. So I've
> > built a git tree with all of the changes going back to the first commit,
> > and merged it into the kernel tree, just like btrfs was merged into the
> > kernel.
>
> > Unless people point out some major problems with this, I'd like to get
> > it merged into 4.9-rc1.
>
> I'm extremely concerned that these patches have *never* seen upstream
> review, and this pull request gives no real opportunity for people to
> make a judgement regarding the code, as many relevant parties have not
> been Cc'd.
As I said, I will send a set of simple patches, I wanted to get this out
as soon as possible and other things came up today. Will do it in the
morning, sorry.
> From a quick scan of the git tree, I can see code (that isn't even
> placed under staging/) for which I have fundamental objections to as a
> maintainer, and has not been Cc'd to a relevant list.
>
> For example, I see commit 5a450477311fbfe2 ("greybus: timesync: Add
> timesync core driver"). This states that it directly accesses the ARMv7
> architected timer, though it's unclear as to precisely what it's doing
> since it introduces an (undocumented) compatible string, and what should
> be an unnecessary devicetree property.
>
> That's never gone to the linux-arm-kernel mainline list, myself or Marc
> (as maintainers of the arch timer driver), nor has the binding seen any
> review on the devicetree mailing list.
Hm, odd, I thought we had Rob review all of the device tree bindings,
but maybe the timesync stuff missed him. And timesync is "odd" to say
the least, wait until you see the firmware side of it :)
Let me post the patches tomorrow and then we can review them.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-14 18:07 ` Greg KH
0 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-14 18:07 UTC (permalink / raw)
To: linux-arm-kernel
On Wed, Sep 14, 2016 at 06:36:26PM +0100, Mark Rutland wrote:
> Hi Greg,
>
> On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
> > Given that it's never a good idea to keep subsystems out of the mainline
> > kernel, I've put together this pull request that adds the greybus driver
> > layer to drivers/greybus/. Because this was 2 1/2 years of work, with
> > many many developers contributing, I didn't want to flatten all of their
> > effort into a few small patches, as that wouldn't be very fair. So I've
> > built a git tree with all of the changes going back to the first commit,
> > and merged it into the kernel tree, just like btrfs was merged into the
> > kernel.
>
> > Unless people point out some major problems with this, I'd like to get
> > it merged into 4.9-rc1.
>
> I'm extremely concerned that these patches have *never* seen upstream
> review, and this pull request gives no real opportunity for people to
> make a judgement regarding the code, as many relevant parties have not
> been Cc'd.
As I said, I will send a set of simple patches, I wanted to get this out
as soon as possible and other things came up today. Will do it in the
morning, sorry.
> From a quick scan of the git tree, I can see code (that isn't even
> placed under staging/) for which I have fundamental objections to as a
> maintainer, and has not been Cc'd to a relevant list.
>
> For example, I see commit 5a450477311fbfe2 ("greybus: timesync: Add
> timesync core driver"). This states that it directly accesses the ARMv7
> architected timer, though it's unclear as to precisely what it's doing
> since it introduces an (undocumented) compatible string, and what should
> be an unnecessary devicetree property.
>
> That's never gone to the linux-arm-kernel mainline list, myself or Marc
> (as maintainers of the arch timer driver), nor has the binding seen any
> review on the devicetree mailing list.
Hm, odd, I thought we had Rob review all of the device tree bindings,
but maybe the timesync stuff missed him. And timesync is "odd" to say
the least, wait until you see the firmware side of it :)
Let me post the patches tomorrow and then we can review them.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-14 18:07 ` Greg KH
@ 2016-09-14 18:29 ` Greg KH
-1 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-14 18:29 UTC (permalink / raw)
To: Mark Rutland
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer, marc.zyngier,
linux-arm-kernel
On Wed, Sep 14, 2016 at 08:07:54PM +0200, Greg KH wrote:
> On Wed, Sep 14, 2016 at 06:36:26PM +0100, Mark Rutland wrote:
> > Hi Greg,
> >
> > On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
> > > Given that it's never a good idea to keep subsystems out of the mainline
> > > kernel, I've put together this pull request that adds the greybus driver
> > > layer to drivers/greybus/. Because this was 2 1/2 years of work, with
> > > many many developers contributing, I didn't want to flatten all of their
> > > effort into a few small patches, as that wouldn't be very fair. So I've
> > > built a git tree with all of the changes going back to the first commit,
> > > and merged it into the kernel tree, just like btrfs was merged into the
> > > kernel.
> >
> > > Unless people point out some major problems with this, I'd like to get
> > > it merged into 4.9-rc1.
> >
> > I'm extremely concerned that these patches have *never* seen upstream
> > review, and this pull request gives no real opportunity for people to
> > make a judgement regarding the code, as many relevant parties have not
> > been Cc'd.
>
> As I said, I will send a set of simple patches, I wanted to get this out
> as soon as possible and other things came up today. Will do it in the
> morning, sorry.
Here's the timesync code pulled out into a simple patch if you want to
see it.
Bryan, any explanations you want to provide that would help in
clarifying Mark's issues?
thanks,
greg k-h
---
drivers/greybus/timesync.c | 1357 ++++++++++++++++++++++++++++++++++++
drivers/greybus/timesync.h | 45 +
drivers/greybus/timesync_platform.c | 77 ++
3 files changed, 1479 insertions(+)
--- /dev/null
+++ b/drivers/greybus/timesync.c
@@ -0,0 +1,1357 @@
+/*
+ * TimeSync API driver.
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+#include <linux/debugfs.h>
+#include <linux/hrtimer.h>
+#include "greybus.h"
+#include "timesync.h"
+#include "greybus_trace.h"
+
+/*
+ * Minimum inter-strobe value of one millisecond is chosen because it
+ * just-about fits the common definition of a jiffy.
+ *
+ * Maximum value OTOH is constrained by the number of bits the SVC can fit
+ * into a 16 bit up-counter. The SVC configures the timer in microseconds
+ * so the maximum allowable value is 65535 microseconds. We clip that value
+ * to 10000 microseconds for the sake of using nice round base 10 numbers
+ * and since right-now there's no imaginable use-case requiring anything
+ * other than a one millisecond inter-strobe time, let alone something
+ * higher than ten milliseconds.
+ */
+#define GB_TIMESYNC_STROBE_DELAY_US 1000
+#define GB_TIMESYNC_DEFAULT_OFFSET_US 1000
+
+/* Work queue timers long, short and SVC strobe timeout */
+#define GB_TIMESYNC_DELAYED_WORK_LONG msecs_to_jiffies(10)
+#define GB_TIMESYNC_DELAYED_WORK_SHORT msecs_to_jiffies(1)
+#define GB_TIMESYNC_MAX_WAIT_SVC msecs_to_jiffies(5000)
+#define GB_TIMESYNC_KTIME_UPDATE msecs_to_jiffies(1000)
+#define GB_TIMESYNC_MAX_KTIME_CONVERSION 15
+
+/* Maximum number of times we'll retry a failed synchronous sync */
+#define GB_TIMESYNC_MAX_RETRIES 5
+
+/* Reported nanoseconds/femtoseconds per clock */
+static u64 gb_timesync_ns_per_clock;
+static u64 gb_timesync_fs_per_clock;
+
+/* Maximum difference we will accept converting FrameTime to ktime */
+static u32 gb_timesync_max_ktime_diff;
+
+/* Reported clock rate */
+static unsigned long gb_timesync_clock_rate;
+
+/* Workqueue */
+static void gb_timesync_worker(struct work_struct *work);
+
+/* List of SVCs with one FrameTime per SVC */
+static LIST_HEAD(gb_timesync_svc_list);
+
+/* Synchronize parallel contexts accessing a valid timesync_svc pointer */
+static DEFINE_MUTEX(gb_timesync_svc_list_mutex);
+
+/* Structure to convert from FrameTime to timespec/ktime */
+struct gb_timesync_frame_time_data {
+ u64 frame_time;
+ struct timespec ts;
+};
+
+struct gb_timesync_svc {
+ struct list_head list;
+ struct list_head interface_list;
+ struct gb_svc *svc;
+ struct gb_timesync_host_device *timesync_hd;
+
+ spinlock_t spinlock; /* Per SVC spinlock to sync with ISR */
+ struct mutex mutex; /* Per SVC mutex for regular synchronization */
+
+ struct dentry *frame_time_dentry;
+ struct dentry *frame_ktime_dentry;
+ struct workqueue_struct *work_queue;
+ wait_queue_head_t wait_queue;
+ struct delayed_work delayed_work;
+ struct timer_list ktime_timer;
+
+ /* The current local FrameTime */
+ u64 frame_time_offset;
+ struct gb_timesync_frame_time_data strobe_data[GB_TIMESYNC_MAX_STROBES];
+ struct gb_timesync_frame_time_data ktime_data;
+
+ /* The SVC FrameTime and relative AP FrameTime @ last TIMESYNC_PING */
+ u64 svc_ping_frame_time;
+ u64 ap_ping_frame_time;
+
+ /* Transitory settings */
+ u32 strobe_mask;
+ bool offset_down;
+ bool print_ping;
+ bool capture_ping;
+ int strobe;
+
+ /* Current state */
+ int state;
+};
+
+struct gb_timesync_host_device {
+ struct list_head list;
+ struct gb_host_device *hd;
+ u64 ping_frame_time;
+};
+
+struct gb_timesync_interface {
+ struct list_head list;
+ struct gb_interface *interface;
+ u64 ping_frame_time;
+};
+
+enum gb_timesync_state {
+ GB_TIMESYNC_STATE_INVALID = 0,
+ GB_TIMESYNC_STATE_INACTIVE = 1,
+ GB_TIMESYNC_STATE_INIT = 2,
+ GB_TIMESYNC_STATE_WAIT_SVC = 3,
+ GB_TIMESYNC_STATE_AUTHORITATIVE = 4,
+ GB_TIMESYNC_STATE_PING = 5,
+ GB_TIMESYNC_STATE_ACTIVE = 6,
+};
+
+static void gb_timesync_ktime_timer_fn(unsigned long data);
+
+static u64 gb_timesync_adjust_count(struct gb_timesync_svc *timesync_svc,
+ u64 counts)
+{
+ if (timesync_svc->offset_down)
+ return counts - timesync_svc->frame_time_offset;
+ else
+ return counts + timesync_svc->frame_time_offset;
+}
+
+/*
+ * This function provides the authoritative FrameTime to a calling function. It
+ * is designed to be lockless and should remain that way the caller is assumed
+ * to be state-aware.
+ */
+static u64 __gb_timesync_get_frame_time(struct gb_timesync_svc *timesync_svc)
+{
+ u64 clocks = gb_timesync_platform_get_counter();
+
+ return gb_timesync_adjust_count(timesync_svc, clocks);
+}
+
+static void gb_timesync_schedule_svc_timeout(struct gb_timesync_svc
+ *timesync_svc)
+{
+ queue_delayed_work(timesync_svc->work_queue,
+ ×ync_svc->delayed_work,
+ GB_TIMESYNC_MAX_WAIT_SVC);
+}
+
+static void gb_timesync_set_state(struct gb_timesync_svc *timesync_svc,
+ int state)
+{
+ switch (state) {
+ case GB_TIMESYNC_STATE_INVALID:
+ timesync_svc->state = state;
+ wake_up(×ync_svc->wait_queue);
+ break;
+ case GB_TIMESYNC_STATE_INACTIVE:
+ timesync_svc->state = state;
+ wake_up(×ync_svc->wait_queue);
+ break;
+ case GB_TIMESYNC_STATE_INIT:
+ if (timesync_svc->state != GB_TIMESYNC_STATE_INVALID) {
+ timesync_svc->strobe = 0;
+ timesync_svc->frame_time_offset = 0;
+ timesync_svc->state = state;
+ cancel_delayed_work(×ync_svc->delayed_work);
+ queue_delayed_work(timesync_svc->work_queue,
+ ×ync_svc->delayed_work,
+ GB_TIMESYNC_DELAYED_WORK_LONG);
+ }
+ break;
+ case GB_TIMESYNC_STATE_WAIT_SVC:
+ if (timesync_svc->state == GB_TIMESYNC_STATE_INIT)
+ timesync_svc->state = state;
+ break;
+ case GB_TIMESYNC_STATE_AUTHORITATIVE:
+ if (timesync_svc->state == GB_TIMESYNC_STATE_WAIT_SVC) {
+ timesync_svc->state = state;
+ cancel_delayed_work(×ync_svc->delayed_work);
+ queue_delayed_work(timesync_svc->work_queue,
+ ×ync_svc->delayed_work, 0);
+ }
+ break;
+ case GB_TIMESYNC_STATE_PING:
+ if (timesync_svc->state == GB_TIMESYNC_STATE_ACTIVE) {
+ timesync_svc->state = state;
+ queue_delayed_work(timesync_svc->work_queue,
+ ×ync_svc->delayed_work,
+ GB_TIMESYNC_DELAYED_WORK_SHORT);
+ }
+ break;
+ case GB_TIMESYNC_STATE_ACTIVE:
+ if (timesync_svc->state == GB_TIMESYNC_STATE_AUTHORITATIVE ||
+ timesync_svc->state == GB_TIMESYNC_STATE_PING) {
+ timesync_svc->state = state;
+ wake_up(×ync_svc->wait_queue);
+ }
+ break;
+ }
+
+ if (WARN_ON(timesync_svc->state != state)) {
+ pr_err("Invalid state transition %d=>%d\n",
+ timesync_svc->state, state);
+ }
+}
+
+static void gb_timesync_set_state_atomic(struct gb_timesync_svc *timesync_svc,
+ int state)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+ gb_timesync_set_state(timesync_svc, state);
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+}
+
+static u64 gb_timesync_diff(u64 x, u64 y)
+{
+ if (x > y)
+ return x - y;
+ else
+ return y - x;
+}
+
+static void gb_timesync_adjust_to_svc(struct gb_timesync_svc *svc,
+ u64 svc_frame_time, u64 ap_frame_time)
+{
+ if (svc_frame_time > ap_frame_time) {
+ svc->frame_time_offset = svc_frame_time - ap_frame_time;
+ svc->offset_down = false;
+ } else {
+ svc->frame_time_offset = ap_frame_time - svc_frame_time;
+ svc->offset_down = true;
+ }
+}
+
+/*
+ * Associate a FrameTime with a ktime timestamp represented as struct timespec
+ * Requires the calling context to hold timesync_svc->mutex
+ */
+static void gb_timesync_store_ktime(struct gb_timesync_svc *timesync_svc,
+ struct timespec ts, u64 frame_time)
+{
+ timesync_svc->ktime_data.ts = ts;
+ timesync_svc->ktime_data.frame_time = frame_time;
+}
+
+/*
+ * Find the two pulses that best-match our expected inter-strobe gap and
+ * then calculate the difference between the SVC time at the second pulse
+ * to the local time at the second pulse.
+ */
+static void gb_timesync_collate_frame_time(struct gb_timesync_svc *timesync_svc,
+ u64 *frame_time)
+{
+ int i = 0;
+ u64 delta, ap_frame_time;
+ u64 strobe_delay_ns = GB_TIMESYNC_STROBE_DELAY_US * NSEC_PER_USEC;
+ u64 least = 0;
+
+ for (i = 1; i < GB_TIMESYNC_MAX_STROBES; i++) {
+ delta = timesync_svc->strobe_data[i].frame_time -
+ timesync_svc->strobe_data[i - 1].frame_time;
+ delta *= gb_timesync_ns_per_clock;
+ delta = gb_timesync_diff(delta, strobe_delay_ns);
+
+ if (!least || delta < least) {
+ least = delta;
+ gb_timesync_adjust_to_svc(timesync_svc, frame_time[i],
+ timesync_svc->strobe_data[i].frame_time);
+
+ ap_frame_time = timesync_svc->strobe_data[i].frame_time;
+ ap_frame_time = gb_timesync_adjust_count(timesync_svc,
+ ap_frame_time);
+ gb_timesync_store_ktime(timesync_svc,
+ timesync_svc->strobe_data[i].ts,
+ ap_frame_time);
+
+ pr_debug("adjust %s local %llu svc %llu delta %llu\n",
+ timesync_svc->offset_down ? "down" : "up",
+ timesync_svc->strobe_data[i].frame_time,
+ frame_time[i], delta);
+ }
+ }
+}
+
+static void gb_timesync_teardown(struct gb_timesync_svc *timesync_svc)
+{
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_interface *interface;
+ struct gb_host_device *hd;
+ int ret;
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ interface = timesync_interface->interface;
+ ret = gb_interface_timesync_disable(interface);
+ if (ret) {
+ dev_err(&interface->dev,
+ "interface timesync_disable %d\n", ret);
+ }
+ }
+
+ hd = timesync_svc->timesync_hd->hd;
+ ret = hd->driver->timesync_disable(hd);
+ if (ret < 0) {
+ dev_err(&hd->dev, "host timesync_disable %d\n",
+ ret);
+ }
+
+ gb_svc_timesync_wake_pins_release(svc);
+ gb_svc_timesync_disable(svc);
+ gb_timesync_platform_unlock_bus();
+
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_INACTIVE);
+}
+
+static void gb_timesync_platform_lock_bus_fail(struct gb_timesync_svc
+ *timesync_svc, int ret)
+{
+ if (ret == -EAGAIN) {
+ gb_timesync_set_state(timesync_svc, timesync_svc->state);
+ } else {
+ pr_err("Failed to lock timesync bus %d\n", ret);
+ gb_timesync_set_state(timesync_svc, GB_TIMESYNC_STATE_INACTIVE);
+ }
+}
+
+static void gb_timesync_enable(struct gb_timesync_svc *timesync_svc)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_interface *interface;
+ u64 init_frame_time;
+ unsigned long clock_rate = gb_timesync_clock_rate;
+ int ret;
+
+ /*
+ * Get access to the wake pins in the AP and SVC
+ * Release these pins either in gb_timesync_teardown() or in
+ * gb_timesync_authoritative()
+ */
+ ret = gb_timesync_platform_lock_bus(timesync_svc);
+ if (ret < 0) {
+ gb_timesync_platform_lock_bus_fail(timesync_svc, ret);
+ return;
+ }
+ ret = gb_svc_timesync_wake_pins_acquire(svc, timesync_svc->strobe_mask);
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_wake_pins_acquire %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+
+ /* Choose an initial time in the future */
+ init_frame_time = __gb_timesync_get_frame_time(timesync_svc) + 100000UL;
+
+ /* Send enable command to all relevant participants */
+ list_for_each_entry(timesync_interface, ×ync_svc->interface_list,
+ list) {
+ interface = timesync_interface->interface;
+ ret = gb_interface_timesync_enable(interface,
+ GB_TIMESYNC_MAX_STROBES,
+ init_frame_time,
+ GB_TIMESYNC_STROBE_DELAY_US,
+ clock_rate);
+ if (ret) {
+ dev_err(&interface->dev,
+ "interface timesync_enable %d\n", ret);
+ }
+ }
+
+ hd = timesync_svc->timesync_hd->hd;
+ ret = hd->driver->timesync_enable(hd, GB_TIMESYNC_MAX_STROBES,
+ init_frame_time,
+ GB_TIMESYNC_STROBE_DELAY_US,
+ clock_rate);
+ if (ret < 0) {
+ dev_err(&hd->dev, "host timesync_enable %d\n",
+ ret);
+ }
+
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_WAIT_SVC);
+ ret = gb_svc_timesync_enable(svc, GB_TIMESYNC_MAX_STROBES,
+ init_frame_time,
+ GB_TIMESYNC_STROBE_DELAY_US,
+ clock_rate);
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_enable %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+
+ /* Schedule a timeout waiting for SVC to complete strobing */
+ gb_timesync_schedule_svc_timeout(timesync_svc);
+}
+
+static void gb_timesync_authoritative(struct gb_timesync_svc *timesync_svc)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_interface *interface;
+ u64 svc_frame_time[GB_TIMESYNC_MAX_STROBES];
+ int ret;
+
+ /* Get authoritative time from SVC and adjust local clock */
+ ret = gb_svc_timesync_authoritative(svc, svc_frame_time);
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_authoritative %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+ gb_timesync_collate_frame_time(timesync_svc, svc_frame_time);
+
+ /* Transmit authoritative time to downstream slaves */
+ hd = timesync_svc->timesync_hd->hd;
+ ret = hd->driver->timesync_authoritative(hd, svc_frame_time);
+ if (ret < 0)
+ dev_err(&hd->dev, "host timesync_authoritative %d\n", ret);
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ interface = timesync_interface->interface;
+ ret = gb_interface_timesync_authoritative(
+ interface,
+ svc_frame_time);
+ if (ret) {
+ dev_err(&interface->dev,
+ "interface timesync_authoritative %d\n", ret);
+ }
+ }
+
+ /* Release wake pins */
+ gb_svc_timesync_wake_pins_release(svc);
+ gb_timesync_platform_unlock_bus();
+
+ /* Transition to state ACTIVE */
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_ACTIVE);
+
+ /* Schedule a ping to verify the synchronized system time */
+ timesync_svc->print_ping = true;
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_PING);
+}
+
+static int __gb_timesync_get_status(struct gb_timesync_svc *timesync_svc)
+{
+ int ret = -EINVAL;
+
+ switch (timesync_svc->state) {
+ case GB_TIMESYNC_STATE_INVALID:
+ case GB_TIMESYNC_STATE_INACTIVE:
+ ret = -ENODEV;
+ break;
+ case GB_TIMESYNC_STATE_INIT:
+ case GB_TIMESYNC_STATE_WAIT_SVC:
+ case GB_TIMESYNC_STATE_AUTHORITATIVE:
+ ret = -EAGAIN;
+ break;
+ case GB_TIMESYNC_STATE_PING:
+ case GB_TIMESYNC_STATE_ACTIVE:
+ ret = 0;
+ break;
+ }
+ return ret;
+}
+
+/*
+ * This routine takes a FrameTime and derives the difference with-respect
+ * to a reference FrameTime/ktime pair. It then returns the calculated
+ * ktime based on the difference between the supplied FrameTime and
+ * the reference FrameTime.
+ *
+ * The time difference is calculated to six decimal places. Taking 19.2MHz
+ * as an example this means we have 52.083333~ nanoseconds per clock or
+ * 52083333~ femtoseconds per clock.
+ *
+ * Naively taking the count difference and converting to
+ * seconds/nanoseconds would quickly see the 0.0833 component produce
+ * noticeable errors. For example a time difference of one second would
+ * loose 19200000 * 0.08333x nanoseconds or 1.59 seconds.
+ *
+ * In contrast calculating in femtoseconds the same example of 19200000 *
+ * 0.000000083333x nanoseconds per count of error is just 1.59 nanoseconds!
+ *
+ * Continuing the example of 19.2 MHz we cap the maximum error difference
+ * at a worst-case 0.3 microseconds over a potential calculation window of
+ * abount 15 seconds, meaning you can convert a FrameTime that is <= 15
+ * seconds older/younger than the reference time with a maximum error of
+ * 0.2385 useconds. Note 19.2MHz is an example frequency not a requirement.
+ */
+static int gb_timesync_to_timespec(struct gb_timesync_svc *timesync_svc,
+ u64 frame_time, struct timespec *ts)
+{
+ unsigned long flags;
+ u64 delta_fs, counts, sec, nsec;
+ bool add;
+ int ret = 0;
+
+ memset(ts, 0x00, sizeof(*ts));
+ mutex_lock(×ync_svc->mutex);
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+
+ ret = __gb_timesync_get_status(timesync_svc);
+ if (ret)
+ goto done;
+
+ /* Support calculating ktime upwards or downwards from the reference */
+ if (frame_time < timesync_svc->ktime_data.frame_time) {
+ add = false;
+ counts = timesync_svc->ktime_data.frame_time - frame_time;
+ } else {
+ add = true;
+ counts = frame_time - timesync_svc->ktime_data.frame_time;
+ }
+
+ /* Enforce the .23 of a usecond boundary @ 19.2MHz */
+ if (counts > gb_timesync_max_ktime_diff) {
+ ret = -EINVAL;
+ goto done;
+ }
+
+ /* Determine the time difference in femtoseconds */
+ delta_fs = counts * gb_timesync_fs_per_clock;
+
+ /* Convert to seconds */
+ sec = delta_fs;
+ do_div(sec, NSEC_PER_SEC);
+ do_div(sec, 1000000UL);
+
+ /* Get the nanosecond remainder */
+ nsec = do_div(delta_fs, sec);
+ do_div(nsec, 1000000UL);
+
+ if (add) {
+ /* Add the calculated offset - overflow nanoseconds upwards */
+ ts->tv_sec = timesync_svc->ktime_data.ts.tv_sec + sec;
+ ts->tv_nsec = timesync_svc->ktime_data.ts.tv_nsec + nsec;
+ if (ts->tv_nsec >= NSEC_PER_SEC) {
+ ts->tv_sec++;
+ ts->tv_nsec -= NSEC_PER_SEC;
+ }
+ } else {
+ /* Subtract the difference over/underflow as necessary */
+ if (nsec > timesync_svc->ktime_data.ts.tv_nsec) {
+ sec++;
+ nsec = nsec + timesync_svc->ktime_data.ts.tv_nsec;
+ nsec = do_div(nsec, NSEC_PER_SEC);
+ } else {
+ nsec = timesync_svc->ktime_data.ts.tv_nsec - nsec;
+ }
+ /* Cannot return a negative second value */
+ if (sec > timesync_svc->ktime_data.ts.tv_sec) {
+ ret = -EINVAL;
+ goto done;
+ }
+ ts->tv_sec = timesync_svc->ktime_data.ts.tv_sec - sec;
+ ts->tv_nsec = nsec;
+ }
+done:
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+ mutex_unlock(×ync_svc->mutex);
+ return ret;
+}
+
+static size_t gb_timesync_log_frame_time(struct gb_timesync_svc *timesync_svc,
+ char *buf, size_t buflen)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_interface *interface;
+ unsigned int len;
+ size_t off;
+
+ /* AP/SVC */
+ off = snprintf(buf, buflen, "%s frametime: ap=%llu %s=%llu ",
+ greybus_bus_type.name,
+ timesync_svc->ap_ping_frame_time, dev_name(&svc->dev),
+ timesync_svc->svc_ping_frame_time);
+ len = buflen - off;
+
+ /* APB/GPB */
+ if (len < buflen) {
+ hd = timesync_svc->timesync_hd->hd;
+ off += snprintf(&buf[off], len, "%s=%llu ", dev_name(&hd->dev),
+ timesync_svc->timesync_hd->ping_frame_time);
+ len = buflen - off;
+ }
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ if (len < buflen) {
+ interface = timesync_interface->interface;
+ off += snprintf(&buf[off], len, "%s=%llu ",
+ dev_name(&interface->dev),
+ timesync_interface->ping_frame_time);
+ len = buflen - off;
+ }
+ }
+ if (len < buflen)
+ off += snprintf(&buf[off], len, "\n");
+ return off;
+}
+
+static size_t gb_timesync_log_frame_ktime(struct gb_timesync_svc *timesync_svc,
+ char *buf, size_t buflen)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_interface *interface;
+ struct timespec ts;
+ unsigned int len;
+ size_t off;
+
+ /* AP */
+ gb_timesync_to_timespec(timesync_svc, timesync_svc->ap_ping_frame_time,
+ &ts);
+ off = snprintf(buf, buflen, "%s frametime: ap=%lu.%lu ",
+ greybus_bus_type.name, ts.tv_sec, ts.tv_nsec);
+ len = buflen - off;
+ if (len >= buflen)
+ goto done;
+
+ /* SVC */
+ gb_timesync_to_timespec(timesync_svc, timesync_svc->svc_ping_frame_time,
+ &ts);
+ off += snprintf(&buf[off], len, "%s=%lu.%lu ", dev_name(&svc->dev),
+ ts.tv_sec, ts.tv_nsec);
+ len = buflen - off;
+ if (len >= buflen)
+ goto done;
+
+ /* APB/GPB */
+ hd = timesync_svc->timesync_hd->hd;
+ gb_timesync_to_timespec(timesync_svc,
+ timesync_svc->timesync_hd->ping_frame_time,
+ &ts);
+ off += snprintf(&buf[off], len, "%s=%lu.%lu ",
+ dev_name(&hd->dev),
+ ts.tv_sec, ts.tv_nsec);
+ len = buflen - off;
+ if (len >= buflen)
+ goto done;
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ interface = timesync_interface->interface;
+ gb_timesync_to_timespec(timesync_svc,
+ timesync_interface->ping_frame_time,
+ &ts);
+ off += snprintf(&buf[off], len, "%s=%lu.%lu ",
+ dev_name(&interface->dev),
+ ts.tv_sec, ts.tv_nsec);
+ len = buflen - off;
+ if (len >= buflen)
+ goto done;
+ }
+ off += snprintf(&buf[off], len, "\n");
+done:
+ return off;
+}
+
+/*
+ * Send an SVC initiated wake 'ping' to each TimeSync participant.
+ * Get the FrameTime from each participant associated with the wake
+ * ping.
+ */
+static void gb_timesync_ping(struct gb_timesync_svc *timesync_svc)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_control *control;
+ u64 *ping_frame_time;
+ int ret;
+
+ /* Get access to the wake pins in the AP and SVC */
+ ret = gb_timesync_platform_lock_bus(timesync_svc);
+ if (ret < 0) {
+ gb_timesync_platform_lock_bus_fail(timesync_svc, ret);
+ return;
+ }
+ ret = gb_svc_timesync_wake_pins_acquire(svc, timesync_svc->strobe_mask);
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_wake_pins_acquire %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+
+ /* Have SVC generate a timesync ping */
+ timesync_svc->capture_ping = true;
+ timesync_svc->svc_ping_frame_time = 0;
+ ret = gb_svc_timesync_ping(svc, ×ync_svc->svc_ping_frame_time);
+ timesync_svc->capture_ping = false;
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_ping %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+
+ /* Get the ping FrameTime from each APB/GPB */
+ hd = timesync_svc->timesync_hd->hd;
+ timesync_svc->timesync_hd->ping_frame_time = 0;
+ ret = hd->driver->timesync_get_last_event(hd,
+ ×ync_svc->timesync_hd->ping_frame_time);
+ if (ret)
+ dev_err(&hd->dev, "host timesync_get_last_event %d\n", ret);
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ control = timesync_interface->interface->control;
+ timesync_interface->ping_frame_time = 0;
+ ping_frame_time = ×ync_interface->ping_frame_time;
+ ret = gb_control_timesync_get_last_event(control,
+ ping_frame_time);
+ if (ret) {
+ dev_err(×ync_interface->interface->dev,
+ "gb_control_timesync_get_last_event %d\n", ret);
+ }
+ }
+
+ /* Ping success - move to timesync active */
+ gb_svc_timesync_wake_pins_release(svc);
+ gb_timesync_platform_unlock_bus();
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_ACTIVE);
+}
+
+static void gb_timesync_log_ping_time(struct gb_timesync_svc *timesync_svc)
+{
+ char *buf;
+
+ if (!timesync_svc->print_ping)
+ return;
+
+ buf = kzalloc(PAGE_SIZE, GFP_KERNEL);
+ if (buf) {
+ gb_timesync_log_frame_time(timesync_svc, buf, PAGE_SIZE);
+ dev_dbg(×ync_svc->svc->dev, "%s", buf);
+ kfree(buf);
+ }
+}
+
+/*
+ * Perform the actual work of scheduled TimeSync logic.
+ */
+static void gb_timesync_worker(struct work_struct *work)
+{
+ struct delayed_work *delayed_work = to_delayed_work(work);
+ struct gb_timesync_svc *timesync_svc =
+ container_of(delayed_work, struct gb_timesync_svc, delayed_work);
+
+ mutex_lock(×ync_svc->mutex);
+
+ switch (timesync_svc->state) {
+ case GB_TIMESYNC_STATE_INIT:
+ gb_timesync_enable(timesync_svc);
+ break;
+
+ case GB_TIMESYNC_STATE_WAIT_SVC:
+ dev_err(×ync_svc->svc->dev,
+ "timeout SVC strobe completion %d/%d\n",
+ timesync_svc->strobe, GB_TIMESYNC_MAX_STROBES);
+ gb_timesync_teardown(timesync_svc);
+ break;
+
+ case GB_TIMESYNC_STATE_AUTHORITATIVE:
+ gb_timesync_authoritative(timesync_svc);
+ break;
+
+ case GB_TIMESYNC_STATE_PING:
+ gb_timesync_ping(timesync_svc);
+ gb_timesync_log_ping_time(timesync_svc);
+ break;
+
+ default:
+ pr_err("Invalid state %d for delayed work\n",
+ timesync_svc->state);
+ break;
+ }
+
+ mutex_unlock(×ync_svc->mutex);
+}
+
+/*
+ * Schedule a new TimeSync INIT or PING operation serialized w/r to
+ * gb_timesync_worker().
+ */
+static int gb_timesync_schedule(struct gb_timesync_svc *timesync_svc, int state)
+{
+ int ret = 0;
+
+ if (state != GB_TIMESYNC_STATE_INIT && state != GB_TIMESYNC_STATE_PING)
+ return -EINVAL;
+
+ mutex_lock(×ync_svc->mutex);
+ if (timesync_svc->state != GB_TIMESYNC_STATE_INVALID) {
+ gb_timesync_set_state_atomic(timesync_svc, state);
+ } else {
+ ret = -ENODEV;
+ }
+ mutex_unlock(×ync_svc->mutex);
+ return ret;
+}
+
+static int __gb_timesync_schedule_synchronous(
+ struct gb_timesync_svc *timesync_svc, int state)
+{
+ unsigned long flags;
+ int ret;
+
+ ret = gb_timesync_schedule(timesync_svc, state);
+ if (ret)
+ return ret;
+
+ ret = wait_event_interruptible(timesync_svc->wait_queue,
+ (timesync_svc->state == GB_TIMESYNC_STATE_ACTIVE ||
+ timesync_svc->state == GB_TIMESYNC_STATE_INACTIVE ||
+ timesync_svc->state == GB_TIMESYNC_STATE_INVALID));
+ if (ret)
+ return ret;
+
+ mutex_lock(×ync_svc->mutex);
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+
+ ret = __gb_timesync_get_status(timesync_svc);
+
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+ mutex_unlock(×ync_svc->mutex);
+
+ return ret;
+}
+
+static struct gb_timesync_svc *gb_timesync_find_timesync_svc(
+ struct gb_host_device *hd)
+{
+ struct gb_timesync_svc *timesync_svc;
+
+ list_for_each_entry(timesync_svc, &gb_timesync_svc_list, list) {
+ if (timesync_svc->svc == hd->svc)
+ return timesync_svc;
+ }
+ return NULL;
+}
+
+static struct gb_timesync_interface *gb_timesync_find_timesync_interface(
+ struct gb_timesync_svc *timesync_svc,
+ struct gb_interface *interface)
+{
+ struct gb_timesync_interface *timesync_interface;
+
+ list_for_each_entry(timesync_interface, ×ync_svc->interface_list, list) {
+ if (timesync_interface->interface == interface)
+ return timesync_interface;
+ }
+ return NULL;
+}
+
+int gb_timesync_schedule_synchronous(struct gb_interface *interface)
+{
+ int ret;
+ struct gb_timesync_svc *timesync_svc;
+ int retries;
+
+ if (!(interface->features & GREYBUS_INTERFACE_FEATURE_TIMESYNC))
+ return 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ for (retries = 0; retries < GB_TIMESYNC_MAX_RETRIES; retries++) {
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc) {
+ ret = -ENODEV;
+ goto done;
+ }
+
+ ret = __gb_timesync_schedule_synchronous(timesync_svc,
+ GB_TIMESYNC_STATE_INIT);
+ if (!ret)
+ break;
+ }
+ if (ret && retries == GB_TIMESYNC_MAX_RETRIES)
+ ret = -ETIMEDOUT;
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_schedule_synchronous);
+
+void gb_timesync_schedule_asynchronous(struct gb_interface *interface)
+{
+ struct gb_timesync_svc *timesync_svc;
+
+ if (!(interface->features & GREYBUS_INTERFACE_FEATURE_TIMESYNC))
+ return;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc)
+ goto done;
+
+ gb_timesync_schedule(timesync_svc, GB_TIMESYNC_STATE_INIT);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_schedule_asynchronous);
+
+static ssize_t gb_timesync_ping_read(struct file *file, char __user *ubuf,
+ size_t len, loff_t *offset, bool ktime)
+{
+ struct gb_timesync_svc *timesync_svc = file->f_inode->i_private;
+ char *buf;
+ ssize_t ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ mutex_lock(×ync_svc->mutex);
+ if (list_empty(×ync_svc->interface_list))
+ ret = -ENODEV;
+ timesync_svc->print_ping = false;
+ mutex_unlock(×ync_svc->mutex);
+ if (ret)
+ goto done;
+
+ ret = __gb_timesync_schedule_synchronous(timesync_svc,
+ GB_TIMESYNC_STATE_PING);
+ if (ret)
+ goto done;
+
+ buf = kzalloc(PAGE_SIZE, GFP_KERNEL);
+ if (!buf) {
+ ret = -ENOMEM;
+ goto done;
+ }
+
+ if (ktime)
+ ret = gb_timesync_log_frame_ktime(timesync_svc, buf, PAGE_SIZE);
+ else
+ ret = gb_timesync_log_frame_time(timesync_svc, buf, PAGE_SIZE);
+ if (ret > 0)
+ ret = simple_read_from_buffer(ubuf, len, offset, buf, ret);
+ kfree(buf);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+
+static ssize_t gb_timesync_ping_read_frame_time(struct file *file,
+ char __user *buf,
+ size_t len, loff_t *offset)
+{
+ return gb_timesync_ping_read(file, buf, len, offset, false);
+}
+
+static ssize_t gb_timesync_ping_read_frame_ktime(struct file *file,
+ char __user *buf,
+ size_t len, loff_t *offset)
+{
+ return gb_timesync_ping_read(file, buf, len, offset, true);
+}
+
+static const struct file_operations gb_timesync_debugfs_frame_time_ops = {
+ .read = gb_timesync_ping_read_frame_time,
+};
+
+static const struct file_operations gb_timesync_debugfs_frame_ktime_ops = {
+ .read = gb_timesync_ping_read_frame_ktime,
+};
+
+static int gb_timesync_hd_add(struct gb_timesync_svc *timesync_svc,
+ struct gb_host_device *hd)
+{
+ struct gb_timesync_host_device *timesync_hd;
+
+ timesync_hd = kzalloc(sizeof(*timesync_hd), GFP_KERNEL);
+ if (!timesync_hd)
+ return -ENOMEM;
+
+ WARN_ON(timesync_svc->timesync_hd);
+ timesync_hd->hd = hd;
+ timesync_svc->timesync_hd = timesync_hd;
+
+ return 0;
+}
+
+static void gb_timesync_hd_remove(struct gb_timesync_svc *timesync_svc,
+ struct gb_host_device *hd)
+{
+ if (timesync_svc->timesync_hd->hd == hd) {
+ kfree(timesync_svc->timesync_hd);
+ timesync_svc->timesync_hd = NULL;
+ return;
+ }
+ WARN_ON(1);
+}
+
+int gb_timesync_svc_add(struct gb_svc *svc)
+{
+ struct gb_timesync_svc *timesync_svc;
+ int ret;
+
+ timesync_svc = kzalloc(sizeof(*timesync_svc), GFP_KERNEL);
+ if (!timesync_svc)
+ return -ENOMEM;
+
+ timesync_svc->work_queue =
+ create_singlethread_workqueue("gb-timesync-work_queue");
+
+ if (!timesync_svc->work_queue) {
+ kfree(timesync_svc);
+ return -ENOMEM;
+ }
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ INIT_LIST_HEAD(×ync_svc->interface_list);
+ INIT_DELAYED_WORK(×ync_svc->delayed_work, gb_timesync_worker);
+ mutex_init(×ync_svc->mutex);
+ spin_lock_init(×ync_svc->spinlock);
+ init_waitqueue_head(×ync_svc->wait_queue);
+
+ timesync_svc->svc = svc;
+ timesync_svc->frame_time_offset = 0;
+ timesync_svc->capture_ping = false;
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_INACTIVE);
+
+ timesync_svc->frame_time_dentry =
+ debugfs_create_file("frame-time", S_IRUGO, svc->debugfs_dentry,
+ timesync_svc,
+ &gb_timesync_debugfs_frame_time_ops);
+ timesync_svc->frame_ktime_dentry =
+ debugfs_create_file("frame-ktime", S_IRUGO, svc->debugfs_dentry,
+ timesync_svc,
+ &gb_timesync_debugfs_frame_ktime_ops);
+
+ list_add(×ync_svc->list, &gb_timesync_svc_list);
+ ret = gb_timesync_hd_add(timesync_svc, svc->hd);
+ if (ret) {
+ list_del(×ync_svc->list);
+ debugfs_remove(timesync_svc->frame_ktime_dentry);
+ debugfs_remove(timesync_svc->frame_time_dentry);
+ destroy_workqueue(timesync_svc->work_queue);
+ kfree(timesync_svc);
+ goto done;
+ }
+
+ init_timer(×ync_svc->ktime_timer);
+ timesync_svc->ktime_timer.function = gb_timesync_ktime_timer_fn;
+ timesync_svc->ktime_timer.expires = jiffies + GB_TIMESYNC_KTIME_UPDATE;
+ timesync_svc->ktime_timer.data = (unsigned long)timesync_svc;
+ add_timer(×ync_svc->ktime_timer);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_svc_add);
+
+void gb_timesync_svc_remove(struct gb_svc *svc)
+{
+ struct gb_timesync_svc *timesync_svc;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_timesync_interface *next;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(svc->hd);
+ if (!timesync_svc)
+ goto done;
+
+ cancel_delayed_work_sync(×ync_svc->delayed_work);
+
+ mutex_lock(×ync_svc->mutex);
+
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_INVALID);
+ del_timer_sync(×ync_svc->ktime_timer);
+ gb_timesync_teardown(timesync_svc);
+
+ gb_timesync_hd_remove(timesync_svc, svc->hd);
+ list_for_each_entry_safe(timesync_interface, next,
+ ×ync_svc->interface_list, list) {
+ list_del(×ync_interface->list);
+ kfree(timesync_interface);
+ }
+ debugfs_remove(timesync_svc->frame_ktime_dentry);
+ debugfs_remove(timesync_svc->frame_time_dentry);
+ destroy_workqueue(timesync_svc->work_queue);
+ list_del(×ync_svc->list);
+
+ mutex_unlock(×ync_svc->mutex);
+
+ kfree(timesync_svc);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+}
+EXPORT_SYMBOL_GPL(gb_timesync_svc_remove);
+
+/*
+ * Add a Greybus Interface to the set of TimeSync Interfaces.
+ */
+int gb_timesync_interface_add(struct gb_interface *interface)
+{
+ struct gb_timesync_svc *timesync_svc;
+ struct gb_timesync_interface *timesync_interface;
+ int ret = 0;
+
+ if (!(interface->features & GREYBUS_INTERFACE_FEATURE_TIMESYNC))
+ return 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc) {
+ ret = -ENODEV;
+ goto done;
+ }
+
+ timesync_interface = kzalloc(sizeof(*timesync_interface), GFP_KERNEL);
+ if (!timesync_interface) {
+ ret = -ENOMEM;
+ goto done;
+ }
+
+ mutex_lock(×ync_svc->mutex);
+ timesync_interface->interface = interface;
+ list_add(×ync_interface->list, ×ync_svc->interface_list);
+ timesync_svc->strobe_mask |= 1 << interface->interface_id;
+ mutex_unlock(×ync_svc->mutex);
+
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_interface_add);
+
+/*
+ * Remove a Greybus Interface from the set of TimeSync Interfaces.
+ */
+void gb_timesync_interface_remove(struct gb_interface *interface)
+{
+ struct gb_timesync_svc *timesync_svc;
+ struct gb_timesync_interface *timesync_interface;
+
+ if (!(interface->features & GREYBUS_INTERFACE_FEATURE_TIMESYNC))
+ return;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc)
+ goto done;
+
+ timesync_interface = gb_timesync_find_timesync_interface(timesync_svc,
+ interface);
+ if (!timesync_interface)
+ goto done;
+
+ mutex_lock(×ync_svc->mutex);
+ timesync_svc->strobe_mask &= ~(1 << interface->interface_id);
+ list_del(×ync_interface->list);
+ kfree(timesync_interface);
+ mutex_unlock(×ync_svc->mutex);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+}
+EXPORT_SYMBOL_GPL(gb_timesync_interface_remove);
+
+/*
+ * Give the authoritative FrameTime to the calling function. Returns zero if we
+ * are not in GB_TIMESYNC_STATE_ACTIVE.
+ */
+static u64 gb_timesync_get_frame_time(struct gb_timesync_svc *timesync_svc)
+{
+ unsigned long flags;
+ u64 ret;
+
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+ if (timesync_svc->state == GB_TIMESYNC_STATE_ACTIVE)
+ ret = __gb_timesync_get_frame_time(timesync_svc);
+ else
+ ret = 0;
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+ return ret;
+}
+
+u64 gb_timesync_get_frame_time_by_interface(struct gb_interface *interface)
+{
+ struct gb_timesync_svc *timesync_svc;
+ u64 ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc)
+ goto done;
+
+ ret = gb_timesync_get_frame_time(timesync_svc);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_get_frame_time_by_interface);
+
+u64 gb_timesync_get_frame_time_by_svc(struct gb_svc *svc)
+{
+ struct gb_timesync_svc *timesync_svc;
+ u64 ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(svc->hd);
+ if (!timesync_svc)
+ goto done;
+
+ ret = gb_timesync_get_frame_time(timesync_svc);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_get_frame_time_by_svc);
+
+/* Incrementally updates the conversion base from FrameTime to ktime */
+static void gb_timesync_ktime_timer_fn(unsigned long data)
+{
+ struct gb_timesync_svc *timesync_svc =
+ (struct gb_timesync_svc *)data;
+ unsigned long flags;
+ u64 frame_time;
+ struct timespec ts;
+
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+
+ if (timesync_svc->state != GB_TIMESYNC_STATE_ACTIVE)
+ goto done;
+
+ ktime_get_ts(&ts);
+ frame_time = __gb_timesync_get_frame_time(timesync_svc);
+ gb_timesync_store_ktime(timesync_svc, ts, frame_time);
+
+done:
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+ mod_timer(×ync_svc->ktime_timer,
+ jiffies + GB_TIMESYNC_KTIME_UPDATE);
+}
+
+int gb_timesync_to_timespec_by_svc(struct gb_svc *svc, u64 frame_time,
+ struct timespec *ts)
+{
+ struct gb_timesync_svc *timesync_svc;
+ int ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(svc->hd);
+ if (!timesync_svc) {
+ ret = -ENODEV;
+ goto done;
+ }
+ ret = gb_timesync_to_timespec(timesync_svc, frame_time, ts);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_to_timespec_by_svc);
+
+int gb_timesync_to_timespec_by_interface(struct gb_interface *interface,
+ u64 frame_time, struct timespec *ts)
+{
+ struct gb_timesync_svc *timesync_svc;
+ int ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc) {
+ ret = -ENODEV;
+ goto done;
+ }
+
+ ret = gb_timesync_to_timespec(timesync_svc, frame_time, ts);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_to_timespec_by_interface);
+
+void gb_timesync_irq(struct gb_timesync_svc *timesync_svc)
+{
+ unsigned long flags;
+ u64 strobe_time;
+ bool strobe_is_ping = true;
+ struct timespec ts;
+
+ ktime_get_ts(&ts);
+ strobe_time = __gb_timesync_get_frame_time(timesync_svc);
+
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+
+ if (timesync_svc->state == GB_TIMESYNC_STATE_PING) {
+ if (!timesync_svc->capture_ping)
+ goto done_nolog;
+ timesync_svc->ap_ping_frame_time = strobe_time;
+ goto done_log;
+ } else if (timesync_svc->state != GB_TIMESYNC_STATE_WAIT_SVC) {
+ goto done_nolog;
+ }
+
+ timesync_svc->strobe_data[timesync_svc->strobe].frame_time = strobe_time;
+ timesync_svc->strobe_data[timesync_svc->strobe].ts = ts;
+
+ if (++timesync_svc->strobe == GB_TIMESYNC_MAX_STROBES) {
+ gb_timesync_set_state(timesync_svc,
+ GB_TIMESYNC_STATE_AUTHORITATIVE);
+ }
+ strobe_is_ping = false;
+done_log:
+ trace_gb_timesync_irq(strobe_is_ping, timesync_svc->strobe,
+ GB_TIMESYNC_MAX_STROBES, strobe_time);
+done_nolog:
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+}
+EXPORT_SYMBOL(gb_timesync_irq);
+
+int __init gb_timesync_init(void)
+{
+ int ret = 0;
+
+ ret = gb_timesync_platform_init();
+ if (ret) {
+ pr_err("timesync platform init fail!\n");
+ return ret;
+ }
+
+ gb_timesync_clock_rate = gb_timesync_platform_get_clock_rate();
+
+ /* Calculate nanoseconds and femtoseconds per clock */
+ gb_timesync_fs_per_clock = FSEC_PER_SEC;
+ do_div(gb_timesync_fs_per_clock, gb_timesync_clock_rate);
+ gb_timesync_ns_per_clock = NSEC_PER_SEC;
+ do_div(gb_timesync_ns_per_clock, gb_timesync_clock_rate);
+
+ /* Calculate the maximum number of clocks we will convert to ktime */
+ gb_timesync_max_ktime_diff =
+ GB_TIMESYNC_MAX_KTIME_CONVERSION * gb_timesync_clock_rate;
+
+ pr_info("Time-Sync @ %lu Hz max ktime conversion +/- %d seconds\n",
+ gb_timesync_clock_rate, GB_TIMESYNC_MAX_KTIME_CONVERSION);
+ return 0;
+}
+
+void gb_timesync_exit(void)
+{
+ gb_timesync_platform_exit();
+}
--- /dev/null
+++ b/drivers/greybus/timesync.h
@@ -0,0 +1,45 @@
+/*
+ * TimeSync API driver.
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __TIMESYNC_H
+#define __TIMESYNC_H
+
+struct gb_svc;
+struct gb_interface;
+struct gb_timesync_svc;
+
+/* Platform */
+u64 gb_timesync_platform_get_counter(void);
+u32 gb_timesync_platform_get_clock_rate(void);
+int gb_timesync_platform_lock_bus(struct gb_timesync_svc *pdata);
+void gb_timesync_platform_unlock_bus(void);
+
+int gb_timesync_platform_init(void);
+void gb_timesync_platform_exit(void);
+
+/* Core API */
+int gb_timesync_interface_add(struct gb_interface *interface);
+void gb_timesync_interface_remove(struct gb_interface *interface);
+int gb_timesync_svc_add(struct gb_svc *svc);
+void gb_timesync_svc_remove(struct gb_svc *svc);
+
+u64 gb_timesync_get_frame_time_by_interface(struct gb_interface *interface);
+u64 gb_timesync_get_frame_time_by_svc(struct gb_svc *svc);
+int gb_timesync_to_timespec_by_svc(struct gb_svc *svc, u64 frame_time,
+ struct timespec *ts);
+int gb_timesync_to_timespec_by_interface(struct gb_interface *interface,
+ u64 frame_time, struct timespec *ts);
+
+int gb_timesync_schedule_synchronous(struct gb_interface *intf);
+void gb_timesync_schedule_asynchronous(struct gb_interface *intf);
+void gb_timesync_irq(struct gb_timesync_svc *timesync_svc);
+int gb_timesync_init(void);
+void gb_timesync_exit(void);
+
+#endif /* __TIMESYNC_H */
--- /dev/null
+++ b/drivers/greybus/timesync_platform.c
@@ -0,0 +1,77 @@
+/*
+ * TimeSync API driver.
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ *
+ * This code reads directly from an ARMv7 memory-mapped timer that lives in
+ * MMIO space. Since this counter lives inside of MMIO space its shared between
+ * cores and that means we don't have to worry about issues like TSC on x86
+ * where each time-stamp-counter (TSC) is local to a particular core.
+ *
+ * Register-level access code is based on
+ * drivers/clocksource/arm_arch_timer.c
+ */
+#include <linux/cpufreq.h>
+#include <linux/of_platform.h>
+
+#include "greybus.h"
+#include "arche_platform.h"
+
+static u32 gb_timesync_clock_frequency;
+int (*arche_platform_change_state_cb)(enum arche_platform_state state,
+ struct gb_timesync_svc *pdata);
+EXPORT_SYMBOL_GPL(arche_platform_change_state_cb);
+
+u64 gb_timesync_platform_get_counter(void)
+{
+ return (u64)get_cycles();
+}
+
+u32 gb_timesync_platform_get_clock_rate(void)
+{
+ if (unlikely(!gb_timesync_clock_frequency))
+ return cpufreq_get(0);
+
+ return gb_timesync_clock_frequency;
+}
+
+int gb_timesync_platform_lock_bus(struct gb_timesync_svc *pdata)
+{
+ return arche_platform_change_state_cb(ARCHE_PLATFORM_STATE_TIME_SYNC,
+ pdata);
+}
+
+void gb_timesync_platform_unlock_bus(void)
+{
+ arche_platform_change_state_cb(ARCHE_PLATFORM_STATE_ACTIVE, NULL);
+}
+
+static const struct of_device_id arch_timer_of_match[] = {
+ { .compatible = "google,greybus-frame-time-counter", },
+ {},
+};
+
+int __init gb_timesync_platform_init(void)
+{
+ struct device_node *np;
+
+ np = of_find_matching_node(NULL, arch_timer_of_match);
+ if (!np) {
+ /* Tolerate not finding to allow BBB etc to continue */
+ pr_warn("Unable to find a compatible ARMv7 timer\n");
+ return 0;
+ }
+
+ if (of_property_read_u32(np, "clock-frequency",
+ &gb_timesync_clock_frequency)) {
+ pr_err("Unable to find timer clock-frequency\n");
+ return -ENODEV;
+ }
+
+ return 0;
+}
+
+void gb_timesync_platform_exit(void) {}
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-14 18:29 ` Greg KH
0 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-14 18:29 UTC (permalink / raw)
To: linux-arm-kernel
On Wed, Sep 14, 2016 at 08:07:54PM +0200, Greg KH wrote:
> On Wed, Sep 14, 2016 at 06:36:26PM +0100, Mark Rutland wrote:
> > Hi Greg,
> >
> > On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
> > > Given that it's never a good idea to keep subsystems out of the mainline
> > > kernel, I've put together this pull request that adds the greybus driver
> > > layer to drivers/greybus/. Because this was 2 1/2 years of work, with
> > > many many developers contributing, I didn't want to flatten all of their
> > > effort into a few small patches, as that wouldn't be very fair. So I've
> > > built a git tree with all of the changes going back to the first commit,
> > > and merged it into the kernel tree, just like btrfs was merged into the
> > > kernel.
> >
> > > Unless people point out some major problems with this, I'd like to get
> > > it merged into 4.9-rc1.
> >
> > I'm extremely concerned that these patches have *never* seen upstream
> > review, and this pull request gives no real opportunity for people to
> > make a judgement regarding the code, as many relevant parties have not
> > been Cc'd.
>
> As I said, I will send a set of simple patches, I wanted to get this out
> as soon as possible and other things came up today. Will do it in the
> morning, sorry.
Here's the timesync code pulled out into a simple patch if you want to
see it.
Bryan, any explanations you want to provide that would help in
clarifying Mark's issues?
thanks,
greg k-h
---
drivers/greybus/timesync.c | 1357 ++++++++++++++++++++++++++++++++++++
drivers/greybus/timesync.h | 45 +
drivers/greybus/timesync_platform.c | 77 ++
3 files changed, 1479 insertions(+)
--- /dev/null
+++ b/drivers/greybus/timesync.c
@@ -0,0 +1,1357 @@
+/*
+ * TimeSync API driver.
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+#include <linux/debugfs.h>
+#include <linux/hrtimer.h>
+#include "greybus.h"
+#include "timesync.h"
+#include "greybus_trace.h"
+
+/*
+ * Minimum inter-strobe value of one millisecond is chosen because it
+ * just-about fits the common definition of a jiffy.
+ *
+ * Maximum value OTOH is constrained by the number of bits the SVC can fit
+ * into a 16 bit up-counter. The SVC configures the timer in microseconds
+ * so the maximum allowable value is 65535 microseconds. We clip that value
+ * to 10000 microseconds for the sake of using nice round base 10 numbers
+ * and since right-now there's no imaginable use-case requiring anything
+ * other than a one millisecond inter-strobe time, let alone something
+ * higher than ten milliseconds.
+ */
+#define GB_TIMESYNC_STROBE_DELAY_US 1000
+#define GB_TIMESYNC_DEFAULT_OFFSET_US 1000
+
+/* Work queue timers long, short and SVC strobe timeout */
+#define GB_TIMESYNC_DELAYED_WORK_LONG msecs_to_jiffies(10)
+#define GB_TIMESYNC_DELAYED_WORK_SHORT msecs_to_jiffies(1)
+#define GB_TIMESYNC_MAX_WAIT_SVC msecs_to_jiffies(5000)
+#define GB_TIMESYNC_KTIME_UPDATE msecs_to_jiffies(1000)
+#define GB_TIMESYNC_MAX_KTIME_CONVERSION 15
+
+/* Maximum number of times we'll retry a failed synchronous sync */
+#define GB_TIMESYNC_MAX_RETRIES 5
+
+/* Reported nanoseconds/femtoseconds per clock */
+static u64 gb_timesync_ns_per_clock;
+static u64 gb_timesync_fs_per_clock;
+
+/* Maximum difference we will accept converting FrameTime to ktime */
+static u32 gb_timesync_max_ktime_diff;
+
+/* Reported clock rate */
+static unsigned long gb_timesync_clock_rate;
+
+/* Workqueue */
+static void gb_timesync_worker(struct work_struct *work);
+
+/* List of SVCs with one FrameTime per SVC */
+static LIST_HEAD(gb_timesync_svc_list);
+
+/* Synchronize parallel contexts accessing a valid timesync_svc pointer */
+static DEFINE_MUTEX(gb_timesync_svc_list_mutex);
+
+/* Structure to convert from FrameTime to timespec/ktime */
+struct gb_timesync_frame_time_data {
+ u64 frame_time;
+ struct timespec ts;
+};
+
+struct gb_timesync_svc {
+ struct list_head list;
+ struct list_head interface_list;
+ struct gb_svc *svc;
+ struct gb_timesync_host_device *timesync_hd;
+
+ spinlock_t spinlock; /* Per SVC spinlock to sync with ISR */
+ struct mutex mutex; /* Per SVC mutex for regular synchronization */
+
+ struct dentry *frame_time_dentry;
+ struct dentry *frame_ktime_dentry;
+ struct workqueue_struct *work_queue;
+ wait_queue_head_t wait_queue;
+ struct delayed_work delayed_work;
+ struct timer_list ktime_timer;
+
+ /* The current local FrameTime */
+ u64 frame_time_offset;
+ struct gb_timesync_frame_time_data strobe_data[GB_TIMESYNC_MAX_STROBES];
+ struct gb_timesync_frame_time_data ktime_data;
+
+ /* The SVC FrameTime and relative AP FrameTime @ last TIMESYNC_PING */
+ u64 svc_ping_frame_time;
+ u64 ap_ping_frame_time;
+
+ /* Transitory settings */
+ u32 strobe_mask;
+ bool offset_down;
+ bool print_ping;
+ bool capture_ping;
+ int strobe;
+
+ /* Current state */
+ int state;
+};
+
+struct gb_timesync_host_device {
+ struct list_head list;
+ struct gb_host_device *hd;
+ u64 ping_frame_time;
+};
+
+struct gb_timesync_interface {
+ struct list_head list;
+ struct gb_interface *interface;
+ u64 ping_frame_time;
+};
+
+enum gb_timesync_state {
+ GB_TIMESYNC_STATE_INVALID = 0,
+ GB_TIMESYNC_STATE_INACTIVE = 1,
+ GB_TIMESYNC_STATE_INIT = 2,
+ GB_TIMESYNC_STATE_WAIT_SVC = 3,
+ GB_TIMESYNC_STATE_AUTHORITATIVE = 4,
+ GB_TIMESYNC_STATE_PING = 5,
+ GB_TIMESYNC_STATE_ACTIVE = 6,
+};
+
+static void gb_timesync_ktime_timer_fn(unsigned long data);
+
+static u64 gb_timesync_adjust_count(struct gb_timesync_svc *timesync_svc,
+ u64 counts)
+{
+ if (timesync_svc->offset_down)
+ return counts - timesync_svc->frame_time_offset;
+ else
+ return counts + timesync_svc->frame_time_offset;
+}
+
+/*
+ * This function provides the authoritative FrameTime to a calling function. It
+ * is designed to be lockless and should remain that way the caller is assumed
+ * to be state-aware.
+ */
+static u64 __gb_timesync_get_frame_time(struct gb_timesync_svc *timesync_svc)
+{
+ u64 clocks = gb_timesync_platform_get_counter();
+
+ return gb_timesync_adjust_count(timesync_svc, clocks);
+}
+
+static void gb_timesync_schedule_svc_timeout(struct gb_timesync_svc
+ *timesync_svc)
+{
+ queue_delayed_work(timesync_svc->work_queue,
+ ×ync_svc->delayed_work,
+ GB_TIMESYNC_MAX_WAIT_SVC);
+}
+
+static void gb_timesync_set_state(struct gb_timesync_svc *timesync_svc,
+ int state)
+{
+ switch (state) {
+ case GB_TIMESYNC_STATE_INVALID:
+ timesync_svc->state = state;
+ wake_up(×ync_svc->wait_queue);
+ break;
+ case GB_TIMESYNC_STATE_INACTIVE:
+ timesync_svc->state = state;
+ wake_up(×ync_svc->wait_queue);
+ break;
+ case GB_TIMESYNC_STATE_INIT:
+ if (timesync_svc->state != GB_TIMESYNC_STATE_INVALID) {
+ timesync_svc->strobe = 0;
+ timesync_svc->frame_time_offset = 0;
+ timesync_svc->state = state;
+ cancel_delayed_work(×ync_svc->delayed_work);
+ queue_delayed_work(timesync_svc->work_queue,
+ ×ync_svc->delayed_work,
+ GB_TIMESYNC_DELAYED_WORK_LONG);
+ }
+ break;
+ case GB_TIMESYNC_STATE_WAIT_SVC:
+ if (timesync_svc->state == GB_TIMESYNC_STATE_INIT)
+ timesync_svc->state = state;
+ break;
+ case GB_TIMESYNC_STATE_AUTHORITATIVE:
+ if (timesync_svc->state == GB_TIMESYNC_STATE_WAIT_SVC) {
+ timesync_svc->state = state;
+ cancel_delayed_work(×ync_svc->delayed_work);
+ queue_delayed_work(timesync_svc->work_queue,
+ ×ync_svc->delayed_work, 0);
+ }
+ break;
+ case GB_TIMESYNC_STATE_PING:
+ if (timesync_svc->state == GB_TIMESYNC_STATE_ACTIVE) {
+ timesync_svc->state = state;
+ queue_delayed_work(timesync_svc->work_queue,
+ ×ync_svc->delayed_work,
+ GB_TIMESYNC_DELAYED_WORK_SHORT);
+ }
+ break;
+ case GB_TIMESYNC_STATE_ACTIVE:
+ if (timesync_svc->state == GB_TIMESYNC_STATE_AUTHORITATIVE ||
+ timesync_svc->state == GB_TIMESYNC_STATE_PING) {
+ timesync_svc->state = state;
+ wake_up(×ync_svc->wait_queue);
+ }
+ break;
+ }
+
+ if (WARN_ON(timesync_svc->state != state)) {
+ pr_err("Invalid state transition %d=>%d\n",
+ timesync_svc->state, state);
+ }
+}
+
+static void gb_timesync_set_state_atomic(struct gb_timesync_svc *timesync_svc,
+ int state)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+ gb_timesync_set_state(timesync_svc, state);
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+}
+
+static u64 gb_timesync_diff(u64 x, u64 y)
+{
+ if (x > y)
+ return x - y;
+ else
+ return y - x;
+}
+
+static void gb_timesync_adjust_to_svc(struct gb_timesync_svc *svc,
+ u64 svc_frame_time, u64 ap_frame_time)
+{
+ if (svc_frame_time > ap_frame_time) {
+ svc->frame_time_offset = svc_frame_time - ap_frame_time;
+ svc->offset_down = false;
+ } else {
+ svc->frame_time_offset = ap_frame_time - svc_frame_time;
+ svc->offset_down = true;
+ }
+}
+
+/*
+ * Associate a FrameTime with a ktime timestamp represented as struct timespec
+ * Requires the calling context to hold timesync_svc->mutex
+ */
+static void gb_timesync_store_ktime(struct gb_timesync_svc *timesync_svc,
+ struct timespec ts, u64 frame_time)
+{
+ timesync_svc->ktime_data.ts = ts;
+ timesync_svc->ktime_data.frame_time = frame_time;
+}
+
+/*
+ * Find the two pulses that best-match our expected inter-strobe gap and
+ * then calculate the difference between the SVC time at the second pulse
+ * to the local time at the second pulse.
+ */
+static void gb_timesync_collate_frame_time(struct gb_timesync_svc *timesync_svc,
+ u64 *frame_time)
+{
+ int i = 0;
+ u64 delta, ap_frame_time;
+ u64 strobe_delay_ns = GB_TIMESYNC_STROBE_DELAY_US * NSEC_PER_USEC;
+ u64 least = 0;
+
+ for (i = 1; i < GB_TIMESYNC_MAX_STROBES; i++) {
+ delta = timesync_svc->strobe_data[i].frame_time -
+ timesync_svc->strobe_data[i - 1].frame_time;
+ delta *= gb_timesync_ns_per_clock;
+ delta = gb_timesync_diff(delta, strobe_delay_ns);
+
+ if (!least || delta < least) {
+ least = delta;
+ gb_timesync_adjust_to_svc(timesync_svc, frame_time[i],
+ timesync_svc->strobe_data[i].frame_time);
+
+ ap_frame_time = timesync_svc->strobe_data[i].frame_time;
+ ap_frame_time = gb_timesync_adjust_count(timesync_svc,
+ ap_frame_time);
+ gb_timesync_store_ktime(timesync_svc,
+ timesync_svc->strobe_data[i].ts,
+ ap_frame_time);
+
+ pr_debug("adjust %s local %llu svc %llu delta %llu\n",
+ timesync_svc->offset_down ? "down" : "up",
+ timesync_svc->strobe_data[i].frame_time,
+ frame_time[i], delta);
+ }
+ }
+}
+
+static void gb_timesync_teardown(struct gb_timesync_svc *timesync_svc)
+{
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_interface *interface;
+ struct gb_host_device *hd;
+ int ret;
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ interface = timesync_interface->interface;
+ ret = gb_interface_timesync_disable(interface);
+ if (ret) {
+ dev_err(&interface->dev,
+ "interface timesync_disable %d\n", ret);
+ }
+ }
+
+ hd = timesync_svc->timesync_hd->hd;
+ ret = hd->driver->timesync_disable(hd);
+ if (ret < 0) {
+ dev_err(&hd->dev, "host timesync_disable %d\n",
+ ret);
+ }
+
+ gb_svc_timesync_wake_pins_release(svc);
+ gb_svc_timesync_disable(svc);
+ gb_timesync_platform_unlock_bus();
+
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_INACTIVE);
+}
+
+static void gb_timesync_platform_lock_bus_fail(struct gb_timesync_svc
+ *timesync_svc, int ret)
+{
+ if (ret == -EAGAIN) {
+ gb_timesync_set_state(timesync_svc, timesync_svc->state);
+ } else {
+ pr_err("Failed to lock timesync bus %d\n", ret);
+ gb_timesync_set_state(timesync_svc, GB_TIMESYNC_STATE_INACTIVE);
+ }
+}
+
+static void gb_timesync_enable(struct gb_timesync_svc *timesync_svc)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_interface *interface;
+ u64 init_frame_time;
+ unsigned long clock_rate = gb_timesync_clock_rate;
+ int ret;
+
+ /*
+ * Get access to the wake pins in the AP and SVC
+ * Release these pins either in gb_timesync_teardown() or in
+ * gb_timesync_authoritative()
+ */
+ ret = gb_timesync_platform_lock_bus(timesync_svc);
+ if (ret < 0) {
+ gb_timesync_platform_lock_bus_fail(timesync_svc, ret);
+ return;
+ }
+ ret = gb_svc_timesync_wake_pins_acquire(svc, timesync_svc->strobe_mask);
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_wake_pins_acquire %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+
+ /* Choose an initial time in the future */
+ init_frame_time = __gb_timesync_get_frame_time(timesync_svc) + 100000UL;
+
+ /* Send enable command to all relevant participants */
+ list_for_each_entry(timesync_interface, ×ync_svc->interface_list,
+ list) {
+ interface = timesync_interface->interface;
+ ret = gb_interface_timesync_enable(interface,
+ GB_TIMESYNC_MAX_STROBES,
+ init_frame_time,
+ GB_TIMESYNC_STROBE_DELAY_US,
+ clock_rate);
+ if (ret) {
+ dev_err(&interface->dev,
+ "interface timesync_enable %d\n", ret);
+ }
+ }
+
+ hd = timesync_svc->timesync_hd->hd;
+ ret = hd->driver->timesync_enable(hd, GB_TIMESYNC_MAX_STROBES,
+ init_frame_time,
+ GB_TIMESYNC_STROBE_DELAY_US,
+ clock_rate);
+ if (ret < 0) {
+ dev_err(&hd->dev, "host timesync_enable %d\n",
+ ret);
+ }
+
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_WAIT_SVC);
+ ret = gb_svc_timesync_enable(svc, GB_TIMESYNC_MAX_STROBES,
+ init_frame_time,
+ GB_TIMESYNC_STROBE_DELAY_US,
+ clock_rate);
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_enable %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+
+ /* Schedule a timeout waiting for SVC to complete strobing */
+ gb_timesync_schedule_svc_timeout(timesync_svc);
+}
+
+static void gb_timesync_authoritative(struct gb_timesync_svc *timesync_svc)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_interface *interface;
+ u64 svc_frame_time[GB_TIMESYNC_MAX_STROBES];
+ int ret;
+
+ /* Get authoritative time from SVC and adjust local clock */
+ ret = gb_svc_timesync_authoritative(svc, svc_frame_time);
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_authoritative %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+ gb_timesync_collate_frame_time(timesync_svc, svc_frame_time);
+
+ /* Transmit authoritative time to downstream slaves */
+ hd = timesync_svc->timesync_hd->hd;
+ ret = hd->driver->timesync_authoritative(hd, svc_frame_time);
+ if (ret < 0)
+ dev_err(&hd->dev, "host timesync_authoritative %d\n", ret);
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ interface = timesync_interface->interface;
+ ret = gb_interface_timesync_authoritative(
+ interface,
+ svc_frame_time);
+ if (ret) {
+ dev_err(&interface->dev,
+ "interface timesync_authoritative %d\n", ret);
+ }
+ }
+
+ /* Release wake pins */
+ gb_svc_timesync_wake_pins_release(svc);
+ gb_timesync_platform_unlock_bus();
+
+ /* Transition to state ACTIVE */
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_ACTIVE);
+
+ /* Schedule a ping to verify the synchronized system time */
+ timesync_svc->print_ping = true;
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_PING);
+}
+
+static int __gb_timesync_get_status(struct gb_timesync_svc *timesync_svc)
+{
+ int ret = -EINVAL;
+
+ switch (timesync_svc->state) {
+ case GB_TIMESYNC_STATE_INVALID:
+ case GB_TIMESYNC_STATE_INACTIVE:
+ ret = -ENODEV;
+ break;
+ case GB_TIMESYNC_STATE_INIT:
+ case GB_TIMESYNC_STATE_WAIT_SVC:
+ case GB_TIMESYNC_STATE_AUTHORITATIVE:
+ ret = -EAGAIN;
+ break;
+ case GB_TIMESYNC_STATE_PING:
+ case GB_TIMESYNC_STATE_ACTIVE:
+ ret = 0;
+ break;
+ }
+ return ret;
+}
+
+/*
+ * This routine takes a FrameTime and derives the difference with-respect
+ * to a reference FrameTime/ktime pair. It then returns the calculated
+ * ktime based on the difference between the supplied FrameTime and
+ * the reference FrameTime.
+ *
+ * The time difference is calculated to six decimal places. Taking 19.2MHz
+ * as an example this means we have 52.083333~ nanoseconds per clock or
+ * 52083333~ femtoseconds per clock.
+ *
+ * Naively taking the count difference and converting to
+ * seconds/nanoseconds would quickly see the 0.0833 component produce
+ * noticeable errors. For example a time difference of one second would
+ * loose 19200000 * 0.08333x nanoseconds or 1.59 seconds.
+ *
+ * In contrast calculating in femtoseconds the same example of 19200000 *
+ * 0.000000083333x nanoseconds per count of error is just 1.59 nanoseconds!
+ *
+ * Continuing the example of 19.2 MHz we cap the maximum error difference
+ * at a worst-case 0.3 microseconds over a potential calculation window of
+ * abount 15 seconds, meaning you can convert a FrameTime that is <= 15
+ * seconds older/younger than the reference time with a maximum error of
+ * 0.2385 useconds. Note 19.2MHz is an example frequency not a requirement.
+ */
+static int gb_timesync_to_timespec(struct gb_timesync_svc *timesync_svc,
+ u64 frame_time, struct timespec *ts)
+{
+ unsigned long flags;
+ u64 delta_fs, counts, sec, nsec;
+ bool add;
+ int ret = 0;
+
+ memset(ts, 0x00, sizeof(*ts));
+ mutex_lock(×ync_svc->mutex);
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+
+ ret = __gb_timesync_get_status(timesync_svc);
+ if (ret)
+ goto done;
+
+ /* Support calculating ktime upwards or downwards from the reference */
+ if (frame_time < timesync_svc->ktime_data.frame_time) {
+ add = false;
+ counts = timesync_svc->ktime_data.frame_time - frame_time;
+ } else {
+ add = true;
+ counts = frame_time - timesync_svc->ktime_data.frame_time;
+ }
+
+ /* Enforce the .23 of a usecond boundary @ 19.2MHz */
+ if (counts > gb_timesync_max_ktime_diff) {
+ ret = -EINVAL;
+ goto done;
+ }
+
+ /* Determine the time difference in femtoseconds */
+ delta_fs = counts * gb_timesync_fs_per_clock;
+
+ /* Convert to seconds */
+ sec = delta_fs;
+ do_div(sec, NSEC_PER_SEC);
+ do_div(sec, 1000000UL);
+
+ /* Get the nanosecond remainder */
+ nsec = do_div(delta_fs, sec);
+ do_div(nsec, 1000000UL);
+
+ if (add) {
+ /* Add the calculated offset - overflow nanoseconds upwards */
+ ts->tv_sec = timesync_svc->ktime_data.ts.tv_sec + sec;
+ ts->tv_nsec = timesync_svc->ktime_data.ts.tv_nsec + nsec;
+ if (ts->tv_nsec >= NSEC_PER_SEC) {
+ ts->tv_sec++;
+ ts->tv_nsec -= NSEC_PER_SEC;
+ }
+ } else {
+ /* Subtract the difference over/underflow as necessary */
+ if (nsec > timesync_svc->ktime_data.ts.tv_nsec) {
+ sec++;
+ nsec = nsec + timesync_svc->ktime_data.ts.tv_nsec;
+ nsec = do_div(nsec, NSEC_PER_SEC);
+ } else {
+ nsec = timesync_svc->ktime_data.ts.tv_nsec - nsec;
+ }
+ /* Cannot return a negative second value */
+ if (sec > timesync_svc->ktime_data.ts.tv_sec) {
+ ret = -EINVAL;
+ goto done;
+ }
+ ts->tv_sec = timesync_svc->ktime_data.ts.tv_sec - sec;
+ ts->tv_nsec = nsec;
+ }
+done:
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+ mutex_unlock(×ync_svc->mutex);
+ return ret;
+}
+
+static size_t gb_timesync_log_frame_time(struct gb_timesync_svc *timesync_svc,
+ char *buf, size_t buflen)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_interface *interface;
+ unsigned int len;
+ size_t off;
+
+ /* AP/SVC */
+ off = snprintf(buf, buflen, "%s frametime: ap=%llu %s=%llu ",
+ greybus_bus_type.name,
+ timesync_svc->ap_ping_frame_time, dev_name(&svc->dev),
+ timesync_svc->svc_ping_frame_time);
+ len = buflen - off;
+
+ /* APB/GPB */
+ if (len < buflen) {
+ hd = timesync_svc->timesync_hd->hd;
+ off += snprintf(&buf[off], len, "%s=%llu ", dev_name(&hd->dev),
+ timesync_svc->timesync_hd->ping_frame_time);
+ len = buflen - off;
+ }
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ if (len < buflen) {
+ interface = timesync_interface->interface;
+ off += snprintf(&buf[off], len, "%s=%llu ",
+ dev_name(&interface->dev),
+ timesync_interface->ping_frame_time);
+ len = buflen - off;
+ }
+ }
+ if (len < buflen)
+ off += snprintf(&buf[off], len, "\n");
+ return off;
+}
+
+static size_t gb_timesync_log_frame_ktime(struct gb_timesync_svc *timesync_svc,
+ char *buf, size_t buflen)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_interface *interface;
+ struct timespec ts;
+ unsigned int len;
+ size_t off;
+
+ /* AP */
+ gb_timesync_to_timespec(timesync_svc, timesync_svc->ap_ping_frame_time,
+ &ts);
+ off = snprintf(buf, buflen, "%s frametime: ap=%lu.%lu ",
+ greybus_bus_type.name, ts.tv_sec, ts.tv_nsec);
+ len = buflen - off;
+ if (len >= buflen)
+ goto done;
+
+ /* SVC */
+ gb_timesync_to_timespec(timesync_svc, timesync_svc->svc_ping_frame_time,
+ &ts);
+ off += snprintf(&buf[off], len, "%s=%lu.%lu ", dev_name(&svc->dev),
+ ts.tv_sec, ts.tv_nsec);
+ len = buflen - off;
+ if (len >= buflen)
+ goto done;
+
+ /* APB/GPB */
+ hd = timesync_svc->timesync_hd->hd;
+ gb_timesync_to_timespec(timesync_svc,
+ timesync_svc->timesync_hd->ping_frame_time,
+ &ts);
+ off += snprintf(&buf[off], len, "%s=%lu.%lu ",
+ dev_name(&hd->dev),
+ ts.tv_sec, ts.tv_nsec);
+ len = buflen - off;
+ if (len >= buflen)
+ goto done;
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ interface = timesync_interface->interface;
+ gb_timesync_to_timespec(timesync_svc,
+ timesync_interface->ping_frame_time,
+ &ts);
+ off += snprintf(&buf[off], len, "%s=%lu.%lu ",
+ dev_name(&interface->dev),
+ ts.tv_sec, ts.tv_nsec);
+ len = buflen - off;
+ if (len >= buflen)
+ goto done;
+ }
+ off += snprintf(&buf[off], len, "\n");
+done:
+ return off;
+}
+
+/*
+ * Send an SVC initiated wake 'ping' to each TimeSync participant.
+ * Get the FrameTime from each participant associated with the wake
+ * ping.
+ */
+static void gb_timesync_ping(struct gb_timesync_svc *timesync_svc)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_control *control;
+ u64 *ping_frame_time;
+ int ret;
+
+ /* Get access to the wake pins in the AP and SVC */
+ ret = gb_timesync_platform_lock_bus(timesync_svc);
+ if (ret < 0) {
+ gb_timesync_platform_lock_bus_fail(timesync_svc, ret);
+ return;
+ }
+ ret = gb_svc_timesync_wake_pins_acquire(svc, timesync_svc->strobe_mask);
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_wake_pins_acquire %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+
+ /* Have SVC generate a timesync ping */
+ timesync_svc->capture_ping = true;
+ timesync_svc->svc_ping_frame_time = 0;
+ ret = gb_svc_timesync_ping(svc, ×ync_svc->svc_ping_frame_time);
+ timesync_svc->capture_ping = false;
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_ping %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+
+ /* Get the ping FrameTime from each APB/GPB */
+ hd = timesync_svc->timesync_hd->hd;
+ timesync_svc->timesync_hd->ping_frame_time = 0;
+ ret = hd->driver->timesync_get_last_event(hd,
+ ×ync_svc->timesync_hd->ping_frame_time);
+ if (ret)
+ dev_err(&hd->dev, "host timesync_get_last_event %d\n", ret);
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ control = timesync_interface->interface->control;
+ timesync_interface->ping_frame_time = 0;
+ ping_frame_time = ×ync_interface->ping_frame_time;
+ ret = gb_control_timesync_get_last_event(control,
+ ping_frame_time);
+ if (ret) {
+ dev_err(×ync_interface->interface->dev,
+ "gb_control_timesync_get_last_event %d\n", ret);
+ }
+ }
+
+ /* Ping success - move to timesync active */
+ gb_svc_timesync_wake_pins_release(svc);
+ gb_timesync_platform_unlock_bus();
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_ACTIVE);
+}
+
+static void gb_timesync_log_ping_time(struct gb_timesync_svc *timesync_svc)
+{
+ char *buf;
+
+ if (!timesync_svc->print_ping)
+ return;
+
+ buf = kzalloc(PAGE_SIZE, GFP_KERNEL);
+ if (buf) {
+ gb_timesync_log_frame_time(timesync_svc, buf, PAGE_SIZE);
+ dev_dbg(×ync_svc->svc->dev, "%s", buf);
+ kfree(buf);
+ }
+}
+
+/*
+ * Perform the actual work of scheduled TimeSync logic.
+ */
+static void gb_timesync_worker(struct work_struct *work)
+{
+ struct delayed_work *delayed_work = to_delayed_work(work);
+ struct gb_timesync_svc *timesync_svc =
+ container_of(delayed_work, struct gb_timesync_svc, delayed_work);
+
+ mutex_lock(×ync_svc->mutex);
+
+ switch (timesync_svc->state) {
+ case GB_TIMESYNC_STATE_INIT:
+ gb_timesync_enable(timesync_svc);
+ break;
+
+ case GB_TIMESYNC_STATE_WAIT_SVC:
+ dev_err(×ync_svc->svc->dev,
+ "timeout SVC strobe completion %d/%d\n",
+ timesync_svc->strobe, GB_TIMESYNC_MAX_STROBES);
+ gb_timesync_teardown(timesync_svc);
+ break;
+
+ case GB_TIMESYNC_STATE_AUTHORITATIVE:
+ gb_timesync_authoritative(timesync_svc);
+ break;
+
+ case GB_TIMESYNC_STATE_PING:
+ gb_timesync_ping(timesync_svc);
+ gb_timesync_log_ping_time(timesync_svc);
+ break;
+
+ default:
+ pr_err("Invalid state %d for delayed work\n",
+ timesync_svc->state);
+ break;
+ }
+
+ mutex_unlock(×ync_svc->mutex);
+}
+
+/*
+ * Schedule a new TimeSync INIT or PING operation serialized w/r to
+ * gb_timesync_worker().
+ */
+static int gb_timesync_schedule(struct gb_timesync_svc *timesync_svc, int state)
+{
+ int ret = 0;
+
+ if (state != GB_TIMESYNC_STATE_INIT && state != GB_TIMESYNC_STATE_PING)
+ return -EINVAL;
+
+ mutex_lock(×ync_svc->mutex);
+ if (timesync_svc->state != GB_TIMESYNC_STATE_INVALID) {
+ gb_timesync_set_state_atomic(timesync_svc, state);
+ } else {
+ ret = -ENODEV;
+ }
+ mutex_unlock(×ync_svc->mutex);
+ return ret;
+}
+
+static int __gb_timesync_schedule_synchronous(
+ struct gb_timesync_svc *timesync_svc, int state)
+{
+ unsigned long flags;
+ int ret;
+
+ ret = gb_timesync_schedule(timesync_svc, state);
+ if (ret)
+ return ret;
+
+ ret = wait_event_interruptible(timesync_svc->wait_queue,
+ (timesync_svc->state == GB_TIMESYNC_STATE_ACTIVE ||
+ timesync_svc->state == GB_TIMESYNC_STATE_INACTIVE ||
+ timesync_svc->state == GB_TIMESYNC_STATE_INVALID));
+ if (ret)
+ return ret;
+
+ mutex_lock(×ync_svc->mutex);
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+
+ ret = __gb_timesync_get_status(timesync_svc);
+
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+ mutex_unlock(×ync_svc->mutex);
+
+ return ret;
+}
+
+static struct gb_timesync_svc *gb_timesync_find_timesync_svc(
+ struct gb_host_device *hd)
+{
+ struct gb_timesync_svc *timesync_svc;
+
+ list_for_each_entry(timesync_svc, &gb_timesync_svc_list, list) {
+ if (timesync_svc->svc == hd->svc)
+ return timesync_svc;
+ }
+ return NULL;
+}
+
+static struct gb_timesync_interface *gb_timesync_find_timesync_interface(
+ struct gb_timesync_svc *timesync_svc,
+ struct gb_interface *interface)
+{
+ struct gb_timesync_interface *timesync_interface;
+
+ list_for_each_entry(timesync_interface, ×ync_svc->interface_list, list) {
+ if (timesync_interface->interface == interface)
+ return timesync_interface;
+ }
+ return NULL;
+}
+
+int gb_timesync_schedule_synchronous(struct gb_interface *interface)
+{
+ int ret;
+ struct gb_timesync_svc *timesync_svc;
+ int retries;
+
+ if (!(interface->features & GREYBUS_INTERFACE_FEATURE_TIMESYNC))
+ return 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ for (retries = 0; retries < GB_TIMESYNC_MAX_RETRIES; retries++) {
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc) {
+ ret = -ENODEV;
+ goto done;
+ }
+
+ ret = __gb_timesync_schedule_synchronous(timesync_svc,
+ GB_TIMESYNC_STATE_INIT);
+ if (!ret)
+ break;
+ }
+ if (ret && retries == GB_TIMESYNC_MAX_RETRIES)
+ ret = -ETIMEDOUT;
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_schedule_synchronous);
+
+void gb_timesync_schedule_asynchronous(struct gb_interface *interface)
+{
+ struct gb_timesync_svc *timesync_svc;
+
+ if (!(interface->features & GREYBUS_INTERFACE_FEATURE_TIMESYNC))
+ return;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc)
+ goto done;
+
+ gb_timesync_schedule(timesync_svc, GB_TIMESYNC_STATE_INIT);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_schedule_asynchronous);
+
+static ssize_t gb_timesync_ping_read(struct file *file, char __user *ubuf,
+ size_t len, loff_t *offset, bool ktime)
+{
+ struct gb_timesync_svc *timesync_svc = file->f_inode->i_private;
+ char *buf;
+ ssize_t ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ mutex_lock(×ync_svc->mutex);
+ if (list_empty(×ync_svc->interface_list))
+ ret = -ENODEV;
+ timesync_svc->print_ping = false;
+ mutex_unlock(×ync_svc->mutex);
+ if (ret)
+ goto done;
+
+ ret = __gb_timesync_schedule_synchronous(timesync_svc,
+ GB_TIMESYNC_STATE_PING);
+ if (ret)
+ goto done;
+
+ buf = kzalloc(PAGE_SIZE, GFP_KERNEL);
+ if (!buf) {
+ ret = -ENOMEM;
+ goto done;
+ }
+
+ if (ktime)
+ ret = gb_timesync_log_frame_ktime(timesync_svc, buf, PAGE_SIZE);
+ else
+ ret = gb_timesync_log_frame_time(timesync_svc, buf, PAGE_SIZE);
+ if (ret > 0)
+ ret = simple_read_from_buffer(ubuf, len, offset, buf, ret);
+ kfree(buf);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+
+static ssize_t gb_timesync_ping_read_frame_time(struct file *file,
+ char __user *buf,
+ size_t len, loff_t *offset)
+{
+ return gb_timesync_ping_read(file, buf, len, offset, false);
+}
+
+static ssize_t gb_timesync_ping_read_frame_ktime(struct file *file,
+ char __user *buf,
+ size_t len, loff_t *offset)
+{
+ return gb_timesync_ping_read(file, buf, len, offset, true);
+}
+
+static const struct file_operations gb_timesync_debugfs_frame_time_ops = {
+ .read = gb_timesync_ping_read_frame_time,
+};
+
+static const struct file_operations gb_timesync_debugfs_frame_ktime_ops = {
+ .read = gb_timesync_ping_read_frame_ktime,
+};
+
+static int gb_timesync_hd_add(struct gb_timesync_svc *timesync_svc,
+ struct gb_host_device *hd)
+{
+ struct gb_timesync_host_device *timesync_hd;
+
+ timesync_hd = kzalloc(sizeof(*timesync_hd), GFP_KERNEL);
+ if (!timesync_hd)
+ return -ENOMEM;
+
+ WARN_ON(timesync_svc->timesync_hd);
+ timesync_hd->hd = hd;
+ timesync_svc->timesync_hd = timesync_hd;
+
+ return 0;
+}
+
+static void gb_timesync_hd_remove(struct gb_timesync_svc *timesync_svc,
+ struct gb_host_device *hd)
+{
+ if (timesync_svc->timesync_hd->hd == hd) {
+ kfree(timesync_svc->timesync_hd);
+ timesync_svc->timesync_hd = NULL;
+ return;
+ }
+ WARN_ON(1);
+}
+
+int gb_timesync_svc_add(struct gb_svc *svc)
+{
+ struct gb_timesync_svc *timesync_svc;
+ int ret;
+
+ timesync_svc = kzalloc(sizeof(*timesync_svc), GFP_KERNEL);
+ if (!timesync_svc)
+ return -ENOMEM;
+
+ timesync_svc->work_queue =
+ create_singlethread_workqueue("gb-timesync-work_queue");
+
+ if (!timesync_svc->work_queue) {
+ kfree(timesync_svc);
+ return -ENOMEM;
+ }
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ INIT_LIST_HEAD(×ync_svc->interface_list);
+ INIT_DELAYED_WORK(×ync_svc->delayed_work, gb_timesync_worker);
+ mutex_init(×ync_svc->mutex);
+ spin_lock_init(×ync_svc->spinlock);
+ init_waitqueue_head(×ync_svc->wait_queue);
+
+ timesync_svc->svc = svc;
+ timesync_svc->frame_time_offset = 0;
+ timesync_svc->capture_ping = false;
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_INACTIVE);
+
+ timesync_svc->frame_time_dentry =
+ debugfs_create_file("frame-time", S_IRUGO, svc->debugfs_dentry,
+ timesync_svc,
+ &gb_timesync_debugfs_frame_time_ops);
+ timesync_svc->frame_ktime_dentry =
+ debugfs_create_file("frame-ktime", S_IRUGO, svc->debugfs_dentry,
+ timesync_svc,
+ &gb_timesync_debugfs_frame_ktime_ops);
+
+ list_add(×ync_svc->list, &gb_timesync_svc_list);
+ ret = gb_timesync_hd_add(timesync_svc, svc->hd);
+ if (ret) {
+ list_del(×ync_svc->list);
+ debugfs_remove(timesync_svc->frame_ktime_dentry);
+ debugfs_remove(timesync_svc->frame_time_dentry);
+ destroy_workqueue(timesync_svc->work_queue);
+ kfree(timesync_svc);
+ goto done;
+ }
+
+ init_timer(×ync_svc->ktime_timer);
+ timesync_svc->ktime_timer.function = gb_timesync_ktime_timer_fn;
+ timesync_svc->ktime_timer.expires = jiffies + GB_TIMESYNC_KTIME_UPDATE;
+ timesync_svc->ktime_timer.data = (unsigned long)timesync_svc;
+ add_timer(×ync_svc->ktime_timer);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_svc_add);
+
+void gb_timesync_svc_remove(struct gb_svc *svc)
+{
+ struct gb_timesync_svc *timesync_svc;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_timesync_interface *next;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(svc->hd);
+ if (!timesync_svc)
+ goto done;
+
+ cancel_delayed_work_sync(×ync_svc->delayed_work);
+
+ mutex_lock(×ync_svc->mutex);
+
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_INVALID);
+ del_timer_sync(×ync_svc->ktime_timer);
+ gb_timesync_teardown(timesync_svc);
+
+ gb_timesync_hd_remove(timesync_svc, svc->hd);
+ list_for_each_entry_safe(timesync_interface, next,
+ ×ync_svc->interface_list, list) {
+ list_del(×ync_interface->list);
+ kfree(timesync_interface);
+ }
+ debugfs_remove(timesync_svc->frame_ktime_dentry);
+ debugfs_remove(timesync_svc->frame_time_dentry);
+ destroy_workqueue(timesync_svc->work_queue);
+ list_del(×ync_svc->list);
+
+ mutex_unlock(×ync_svc->mutex);
+
+ kfree(timesync_svc);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+}
+EXPORT_SYMBOL_GPL(gb_timesync_svc_remove);
+
+/*
+ * Add a Greybus Interface to the set of TimeSync Interfaces.
+ */
+int gb_timesync_interface_add(struct gb_interface *interface)
+{
+ struct gb_timesync_svc *timesync_svc;
+ struct gb_timesync_interface *timesync_interface;
+ int ret = 0;
+
+ if (!(interface->features & GREYBUS_INTERFACE_FEATURE_TIMESYNC))
+ return 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc) {
+ ret = -ENODEV;
+ goto done;
+ }
+
+ timesync_interface = kzalloc(sizeof(*timesync_interface), GFP_KERNEL);
+ if (!timesync_interface) {
+ ret = -ENOMEM;
+ goto done;
+ }
+
+ mutex_lock(×ync_svc->mutex);
+ timesync_interface->interface = interface;
+ list_add(×ync_interface->list, ×ync_svc->interface_list);
+ timesync_svc->strobe_mask |= 1 << interface->interface_id;
+ mutex_unlock(×ync_svc->mutex);
+
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_interface_add);
+
+/*
+ * Remove a Greybus Interface from the set of TimeSync Interfaces.
+ */
+void gb_timesync_interface_remove(struct gb_interface *interface)
+{
+ struct gb_timesync_svc *timesync_svc;
+ struct gb_timesync_interface *timesync_interface;
+
+ if (!(interface->features & GREYBUS_INTERFACE_FEATURE_TIMESYNC))
+ return;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc)
+ goto done;
+
+ timesync_interface = gb_timesync_find_timesync_interface(timesync_svc,
+ interface);
+ if (!timesync_interface)
+ goto done;
+
+ mutex_lock(×ync_svc->mutex);
+ timesync_svc->strobe_mask &= ~(1 << interface->interface_id);
+ list_del(×ync_interface->list);
+ kfree(timesync_interface);
+ mutex_unlock(×ync_svc->mutex);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+}
+EXPORT_SYMBOL_GPL(gb_timesync_interface_remove);
+
+/*
+ * Give the authoritative FrameTime to the calling function. Returns zero if we
+ * are not in GB_TIMESYNC_STATE_ACTIVE.
+ */
+static u64 gb_timesync_get_frame_time(struct gb_timesync_svc *timesync_svc)
+{
+ unsigned long flags;
+ u64 ret;
+
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+ if (timesync_svc->state == GB_TIMESYNC_STATE_ACTIVE)
+ ret = __gb_timesync_get_frame_time(timesync_svc);
+ else
+ ret = 0;
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+ return ret;
+}
+
+u64 gb_timesync_get_frame_time_by_interface(struct gb_interface *interface)
+{
+ struct gb_timesync_svc *timesync_svc;
+ u64 ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc)
+ goto done;
+
+ ret = gb_timesync_get_frame_time(timesync_svc);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_get_frame_time_by_interface);
+
+u64 gb_timesync_get_frame_time_by_svc(struct gb_svc *svc)
+{
+ struct gb_timesync_svc *timesync_svc;
+ u64 ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(svc->hd);
+ if (!timesync_svc)
+ goto done;
+
+ ret = gb_timesync_get_frame_time(timesync_svc);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_get_frame_time_by_svc);
+
+/* Incrementally updates the conversion base from FrameTime to ktime */
+static void gb_timesync_ktime_timer_fn(unsigned long data)
+{
+ struct gb_timesync_svc *timesync_svc =
+ (struct gb_timesync_svc *)data;
+ unsigned long flags;
+ u64 frame_time;
+ struct timespec ts;
+
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+
+ if (timesync_svc->state != GB_TIMESYNC_STATE_ACTIVE)
+ goto done;
+
+ ktime_get_ts(&ts);
+ frame_time = __gb_timesync_get_frame_time(timesync_svc);
+ gb_timesync_store_ktime(timesync_svc, ts, frame_time);
+
+done:
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+ mod_timer(×ync_svc->ktime_timer,
+ jiffies + GB_TIMESYNC_KTIME_UPDATE);
+}
+
+int gb_timesync_to_timespec_by_svc(struct gb_svc *svc, u64 frame_time,
+ struct timespec *ts)
+{
+ struct gb_timesync_svc *timesync_svc;
+ int ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(svc->hd);
+ if (!timesync_svc) {
+ ret = -ENODEV;
+ goto done;
+ }
+ ret = gb_timesync_to_timespec(timesync_svc, frame_time, ts);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_to_timespec_by_svc);
+
+int gb_timesync_to_timespec_by_interface(struct gb_interface *interface,
+ u64 frame_time, struct timespec *ts)
+{
+ struct gb_timesync_svc *timesync_svc;
+ int ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc) {
+ ret = -ENODEV;
+ goto done;
+ }
+
+ ret = gb_timesync_to_timespec(timesync_svc, frame_time, ts);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_to_timespec_by_interface);
+
+void gb_timesync_irq(struct gb_timesync_svc *timesync_svc)
+{
+ unsigned long flags;
+ u64 strobe_time;
+ bool strobe_is_ping = true;
+ struct timespec ts;
+
+ ktime_get_ts(&ts);
+ strobe_time = __gb_timesync_get_frame_time(timesync_svc);
+
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+
+ if (timesync_svc->state == GB_TIMESYNC_STATE_PING) {
+ if (!timesync_svc->capture_ping)
+ goto done_nolog;
+ timesync_svc->ap_ping_frame_time = strobe_time;
+ goto done_log;
+ } else if (timesync_svc->state != GB_TIMESYNC_STATE_WAIT_SVC) {
+ goto done_nolog;
+ }
+
+ timesync_svc->strobe_data[timesync_svc->strobe].frame_time = strobe_time;
+ timesync_svc->strobe_data[timesync_svc->strobe].ts = ts;
+
+ if (++timesync_svc->strobe == GB_TIMESYNC_MAX_STROBES) {
+ gb_timesync_set_state(timesync_svc,
+ GB_TIMESYNC_STATE_AUTHORITATIVE);
+ }
+ strobe_is_ping = false;
+done_log:
+ trace_gb_timesync_irq(strobe_is_ping, timesync_svc->strobe,
+ GB_TIMESYNC_MAX_STROBES, strobe_time);
+done_nolog:
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+}
+EXPORT_SYMBOL(gb_timesync_irq);
+
+int __init gb_timesync_init(void)
+{
+ int ret = 0;
+
+ ret = gb_timesync_platform_init();
+ if (ret) {
+ pr_err("timesync platform init fail!\n");
+ return ret;
+ }
+
+ gb_timesync_clock_rate = gb_timesync_platform_get_clock_rate();
+
+ /* Calculate nanoseconds and femtoseconds per clock */
+ gb_timesync_fs_per_clock = FSEC_PER_SEC;
+ do_div(gb_timesync_fs_per_clock, gb_timesync_clock_rate);
+ gb_timesync_ns_per_clock = NSEC_PER_SEC;
+ do_div(gb_timesync_ns_per_clock, gb_timesync_clock_rate);
+
+ /* Calculate the maximum number of clocks we will convert to ktime */
+ gb_timesync_max_ktime_diff =
+ GB_TIMESYNC_MAX_KTIME_CONVERSION * gb_timesync_clock_rate;
+
+ pr_info("Time-Sync @ %lu Hz max ktime conversion +/- %d seconds\n",
+ gb_timesync_clock_rate, GB_TIMESYNC_MAX_KTIME_CONVERSION);
+ return 0;
+}
+
+void gb_timesync_exit(void)
+{
+ gb_timesync_platform_exit();
+}
--- /dev/null
+++ b/drivers/greybus/timesync.h
@@ -0,0 +1,45 @@
+/*
+ * TimeSync API driver.
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __TIMESYNC_H
+#define __TIMESYNC_H
+
+struct gb_svc;
+struct gb_interface;
+struct gb_timesync_svc;
+
+/* Platform */
+u64 gb_timesync_platform_get_counter(void);
+u32 gb_timesync_platform_get_clock_rate(void);
+int gb_timesync_platform_lock_bus(struct gb_timesync_svc *pdata);
+void gb_timesync_platform_unlock_bus(void);
+
+int gb_timesync_platform_init(void);
+void gb_timesync_platform_exit(void);
+
+/* Core API */
+int gb_timesync_interface_add(struct gb_interface *interface);
+void gb_timesync_interface_remove(struct gb_interface *interface);
+int gb_timesync_svc_add(struct gb_svc *svc);
+void gb_timesync_svc_remove(struct gb_svc *svc);
+
+u64 gb_timesync_get_frame_time_by_interface(struct gb_interface *interface);
+u64 gb_timesync_get_frame_time_by_svc(struct gb_svc *svc);
+int gb_timesync_to_timespec_by_svc(struct gb_svc *svc, u64 frame_time,
+ struct timespec *ts);
+int gb_timesync_to_timespec_by_interface(struct gb_interface *interface,
+ u64 frame_time, struct timespec *ts);
+
+int gb_timesync_schedule_synchronous(struct gb_interface *intf);
+void gb_timesync_schedule_asynchronous(struct gb_interface *intf);
+void gb_timesync_irq(struct gb_timesync_svc *timesync_svc);
+int gb_timesync_init(void);
+void gb_timesync_exit(void);
+
+#endif /* __TIMESYNC_H */
--- /dev/null
+++ b/drivers/greybus/timesync_platform.c
@@ -0,0 +1,77 @@
+/*
+ * TimeSync API driver.
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ *
+ * This code reads directly from an ARMv7 memory-mapped timer that lives in
+ * MMIO space. Since this counter lives inside of MMIO space its shared between
+ * cores and that means we don't have to worry about issues like TSC on x86
+ * where each time-stamp-counter (TSC) is local to a particular core.
+ *
+ * Register-level access code is based on
+ * drivers/clocksource/arm_arch_timer.c
+ */
+#include <linux/cpufreq.h>
+#include <linux/of_platform.h>
+
+#include "greybus.h"
+#include "arche_platform.h"
+
+static u32 gb_timesync_clock_frequency;
+int (*arche_platform_change_state_cb)(enum arche_platform_state state,
+ struct gb_timesync_svc *pdata);
+EXPORT_SYMBOL_GPL(arche_platform_change_state_cb);
+
+u64 gb_timesync_platform_get_counter(void)
+{
+ return (u64)get_cycles();
+}
+
+u32 gb_timesync_platform_get_clock_rate(void)
+{
+ if (unlikely(!gb_timesync_clock_frequency))
+ return cpufreq_get(0);
+
+ return gb_timesync_clock_frequency;
+}
+
+int gb_timesync_platform_lock_bus(struct gb_timesync_svc *pdata)
+{
+ return arche_platform_change_state_cb(ARCHE_PLATFORM_STATE_TIME_SYNC,
+ pdata);
+}
+
+void gb_timesync_platform_unlock_bus(void)
+{
+ arche_platform_change_state_cb(ARCHE_PLATFORM_STATE_ACTIVE, NULL);
+}
+
+static const struct of_device_id arch_timer_of_match[] = {
+ { .compatible = "google,greybus-frame-time-counter", },
+ {},
+};
+
+int __init gb_timesync_platform_init(void)
+{
+ struct device_node *np;
+
+ np = of_find_matching_node(NULL, arch_timer_of_match);
+ if (!np) {
+ /* Tolerate not finding to allow BBB etc to continue */
+ pr_warn("Unable to find a compatible ARMv7 timer\n");
+ return 0;
+ }
+
+ if (of_property_read_u32(np, "clock-frequency",
+ &gb_timesync_clock_frequency)) {
+ pr_err("Unable to find timer clock-frequency\n");
+ return -ENODEV;
+ }
+
+ return 0;
+}
+
+void gb_timesync_platform_exit(void) {}
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-14 18:29 ` Greg KH
@ 2016-09-14 19:05 ` Joe Perches
-1 siblings, 0 replies; 84+ messages in thread
From: Joe Perches @ 2016-09-14 19:05 UTC (permalink / raw)
To: Greg KH, Mark Rutland
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer, marc.zyngier,
linux-arm-kernel
On Wed, 2016-09-14 at 20:29 +0200, Greg KH wrote:
trivial note:
> +static size_t gb_timesync_log_frame_time(struct gb_timesync_svc *timesync_svc,
> + char *buf, size_t buflen)
> +{
> + struct gb_svc *svc = timesync_svc->svc;
> + struct gb_host_device *hd;
> + struct gb_timesync_interface *timesync_interface;
> + struct gb_interface *interface;
> + unsigned int len;
> + size_t off;
> +
> + /* AP/SVC */
> + off = snprintf(buf, buflen, "%s frametime: ap=%llu %s=%llu ",
> + greybus_bus_type.name,
> + timesync_svc->ap_ping_frame_time, dev_name(&svc->dev),
> + timesync_svc->svc_ping_frame_time);
> + len = buflen - off;
> +
> + /* APB/GPB */
> + if (len < buflen) {
> + hd = timesync_svc->timesync_hd->hd;
> + off += snprintf(&buf[off], len, "%s=%llu ", dev_name(&hd->dev),
> + timesync_svc->timesync_hd->ping_frame_time);
> + len = buflen - off;
> + }
> +
> + list_for_each_entry(timesync_interface,
> + ×ync_svc->interface_list, list) {
> + if (len < buflen) {
> + interface = timesync_interface->interface;
> + off += snprintf(&buf[off], len, "%s=%llu ",
> + dev_name(&interface->dev),
> + timesync_interface->ping_frame_time);
> + len = buflen - off;
> + }
> + }
> + if (len < buflen)
> + off += snprintf(&buf[off], len, "\n");
> + return off;
> +}
The unnecessary trailing blank can be avoided by converting
snprintf(, "format ', ...);
for (...)
snprintf(, "format ", ...);
snprintf(, "\n");
to
snprintf(, "format', ...);
for (...)
snprintf(, " format", ...);
snprintf(, "\n");
>
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-14 19:05 ` Joe Perches
0 siblings, 0 replies; 84+ messages in thread
From: Joe Perches @ 2016-09-14 19:05 UTC (permalink / raw)
To: linux-arm-kernel
On Wed, 2016-09-14 at 20:29 +0200, Greg KH wrote:
trivial note:
> +static size_t gb_timesync_log_frame_time(struct gb_timesync_svc *timesync_svc,
> + char *buf, size_t buflen)
> +{
> + struct gb_svc *svc = timesync_svc->svc;
> + struct gb_host_device *hd;
> + struct gb_timesync_interface *timesync_interface;
> + struct gb_interface *interface;
> + unsigned int len;
> + size_t off;
> +
> + /* AP/SVC */
> + off = snprintf(buf, buflen, "%s frametime: ap=%llu %s=%llu ",
> + greybus_bus_type.name,
> + timesync_svc->ap_ping_frame_time, dev_name(&svc->dev),
> + timesync_svc->svc_ping_frame_time);
> + len = buflen - off;
> +
> + /* APB/GPB */
> + if (len < buflen) {
> + hd = timesync_svc->timesync_hd->hd;
> + off += snprintf(&buf[off], len, "%s=%llu ", dev_name(&hd->dev),
> + timesync_svc->timesync_hd->ping_frame_time);
> + len = buflen - off;
> + }
> +
> + list_for_each_entry(timesync_interface,
> + ×ync_svc->interface_list, list) {
> + if (len < buflen) {
> + interface = timesync_interface->interface;
> + off += snprintf(&buf[off], len, "%s=%llu ",
> + dev_name(&interface->dev),
> + timesync_interface->ping_frame_time);
> + len = buflen - off;
> + }
> + }
> + if (len < buflen)
> + off += snprintf(&buf[off], len, "\n");
> + return off;
> +}
The unnecessary trailing blank can be avoided by converting
snprintf(, "format ', ...);
for (...)
snprintf(, "format ", ...);
snprintf(, "\n");
to
snprintf(, "format', ...);
for (...)
snprintf(, " format", ...);
snprintf(, "\n");
>
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-14 18:07 ` Greg KH
@ 2016-09-14 20:07 ` Rob Herring
-1 siblings, 0 replies; 84+ messages in thread
From: Rob Herring @ 2016-09-14 20:07 UTC (permalink / raw)
To: Greg KH
Cc: Mark Rutland, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer, Marc Zyngier,
linux-arm-kernel
On Wed, Sep 14, 2016 at 1:07 PM, Greg KH <gregkh@linuxfoundation.org> wrote:
> On Wed, Sep 14, 2016 at 06:36:26PM +0100, Mark Rutland wrote:
>> Hi Greg,
>>
>> On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
>> > Given that it's never a good idea to keep subsystems out of the mainline
>> > kernel, I've put together this pull request that adds the greybus driver
>> > layer to drivers/greybus/. Because this was 2 1/2 years of work, with
>> > many many developers contributing, I didn't want to flatten all of their
>> > effort into a few small patches, as that wouldn't be very fair. So I've
>> > built a git tree with all of the changes going back to the first commit,
>> > and merged it into the kernel tree, just like btrfs was merged into the
>> > kernel.
>>
>> > Unless people point out some major problems with this, I'd like to get
>> > it merged into 4.9-rc1.
>>
>> I'm extremely concerned that these patches have *never* seen upstream
>> review, and this pull request gives no real opportunity for people to
>> make a judgement regarding the code, as many relevant parties have not
>> been Cc'd.
>
> As I said, I will send a set of simple patches, I wanted to get this out
> as soon as possible and other things came up today. Will do it in the
> morning, sorry.
>
>> From a quick scan of the git tree, I can see code (that isn't even
>> placed under staging/) for which I have fundamental objections to as a
>> maintainer, and has not been Cc'd to a relevant list.
>>
>> For example, I see commit 5a450477311fbfe2 ("greybus: timesync: Add
>> timesync core driver"). This states that it directly accesses the ARMv7
>> architected timer, though it's unclear as to precisely what it's doing
>> since it introduces an (undocumented) compatible string, and what should
>> be an unnecessary devicetree property.
>>
>> That's never gone to the linux-arm-kernel mainline list, myself or Marc
>> (as maintainers of the arch timer driver), nor has the binding seen any
>> review on the devicetree mailing list.
>
> Hm, odd, I thought we had Rob review all of the device tree bindings,
> but maybe the timesync stuff missed him. And timesync is "odd" to say
> the least, wait until you see the firmware side of it :)
I have not. There weren't any when I was involved (other than SOC
related bindings). Some like USB devices were discussed at least, but
at the time there was no common definition of how to deal with
soldered, on-board USB devices. And that was also what's good enough
to move forward with development, not ready for mainline. There's a
binding now for USB, but what's there for arche predates it IIRC. This
is the first I've heard of timesync having a binding. I can't imagine
why it needs one.
I was going to stay out of this, but now that I'm here...
I'm not all that worried about the quality of the code, but do
question whether this really makes sense to merge at this time. While
I worked on Ara and would like to see all this code be used, I'm
pretty doubtful it will be. Yes, Motorola is using a version of it,
but they forked it some time back and changed who knows what. So what
is upstream won't likely even work with any publicly available device.
And how long that Motorola phone lasts is unknown. Sure it is self
contained, but maintenance to keep it in tree is not 0.
There's also things that never got solved. Like how do you describe
devices on I2C, SPI, UART, etc. behind a greybus device? The plan was
to use DT overlays, but that was never solved and brings a whole set
of problems to solve upstream.
Rob
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-14 20:07 ` Rob Herring
0 siblings, 0 replies; 84+ messages in thread
From: Rob Herring @ 2016-09-14 20:07 UTC (permalink / raw)
To: linux-arm-kernel
On Wed, Sep 14, 2016 at 1:07 PM, Greg KH <gregkh@linuxfoundation.org> wrote:
> On Wed, Sep 14, 2016 at 06:36:26PM +0100, Mark Rutland wrote:
>> Hi Greg,
>>
>> On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
>> > Given that it's never a good idea to keep subsystems out of the mainline
>> > kernel, I've put together this pull request that adds the greybus driver
>> > layer to drivers/greybus/. Because this was 2 1/2 years of work, with
>> > many many developers contributing, I didn't want to flatten all of their
>> > effort into a few small patches, as that wouldn't be very fair. So I've
>> > built a git tree with all of the changes going back to the first commit,
>> > and merged it into the kernel tree, just like btrfs was merged into the
>> > kernel.
>>
>> > Unless people point out some major problems with this, I'd like to get
>> > it merged into 4.9-rc1.
>>
>> I'm extremely concerned that these patches have *never* seen upstream
>> review, and this pull request gives no real opportunity for people to
>> make a judgement regarding the code, as many relevant parties have not
>> been Cc'd.
>
> As I said, I will send a set of simple patches, I wanted to get this out
> as soon as possible and other things came up today. Will do it in the
> morning, sorry.
>
>> From a quick scan of the git tree, I can see code (that isn't even
>> placed under staging/) for which I have fundamental objections to as a
>> maintainer, and has not been Cc'd to a relevant list.
>>
>> For example, I see commit 5a450477311fbfe2 ("greybus: timesync: Add
>> timesync core driver"). This states that it directly accesses the ARMv7
>> architected timer, though it's unclear as to precisely what it's doing
>> since it introduces an (undocumented) compatible string, and what should
>> be an unnecessary devicetree property.
>>
>> That's never gone to the linux-arm-kernel mainline list, myself or Marc
>> (as maintainers of the arch timer driver), nor has the binding seen any
>> review on the devicetree mailing list.
>
> Hm, odd, I thought we had Rob review all of the device tree bindings,
> but maybe the timesync stuff missed him. And timesync is "odd" to say
> the least, wait until you see the firmware side of it :)
I have not. There weren't any when I was involved (other than SOC
related bindings). Some like USB devices were discussed at least, but
at the time there was no common definition of how to deal with
soldered, on-board USB devices. And that was also what's good enough
to move forward with development, not ready for mainline. There's a
binding now for USB, but what's there for arche predates it IIRC. This
is the first I've heard of timesync having a binding. I can't imagine
why it needs one.
I was going to stay out of this, but now that I'm here...
I'm not all that worried about the quality of the code, but do
question whether this really makes sense to merge at this time. While
I worked on Ara and would like to see all this code be used, I'm
pretty doubtful it will be. Yes, Motorola is using a version of it,
but they forked it some time back and changed who knows what. So what
is upstream won't likely even work with any publicly available device.
And how long that Motorola phone lasts is unknown. Sure it is self
contained, but maintenance to keep it in tree is not 0.
There's also things that never got solved. Like how do you describe
devices on I2C, SPI, UART, etc. behind a greybus device? The plan was
to use DT overlays, but that was never solved and brings a whole set
of problems to solve upstream.
Rob
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-14 18:29 ` Greg KH
@ 2016-09-15 9:35 ` Bryan O'Donoghue
-1 siblings, 0 replies; 84+ messages in thread
From: Bryan O'Donoghue @ 2016-09-15 9:35 UTC (permalink / raw)
To: Greg KH, Mark Rutland
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Vaibhav Agarwal, Mark Greer, marc.zyngier, linux-arm-kernel
On Wed, 2016-09-14 at 20:29 +0200, Greg KH wrote:
> On Wed, Sep 14, 2016 at 08:07:54PM +0200, Greg KH wrote:
> >
> > On Wed, Sep 14, 2016 at 06:36:26PM +0100, Mark Rutland wrote:
> > >
> > > Hi Greg,
> > >
> > > On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
> > > >
> > > > Given that it's never a good idea to keep subsystems out of the
> > > > mainline
> > > > kernel, I've put together this pull request that adds the
> > > > greybus driver
> > > > layer to drivers/greybus/. Because this was 2 1/2 years of
> > > > work, with
> > > > many many developers contributing, I didn't want to flatten all
> > > > of their
> > > > effort into a few small patches, as that wouldn't be very
> > > > fair. So I've
> > > > built a git tree with all of the changes going back to the
> > > > first commit,
> > > > and merged it into the kernel tree, just like btrfs was merged
> > > > into the
> > > > kernel.
> > > >
> > > > Unless people point out some major problems with this, I'd like
> > > > to get
> > > > it merged into 4.9-rc1.
> > > I'm extremely concerned that these patches have *never* seen
> > > upstream
> > > review, and this pull request gives no real opportunity for
> > > people to
> > > make a judgement regarding the code, as many relevant parties
> > > have not
> > > been Cc'd.
> > As I said, I will send a set of simple patches, I wanted to get
> > this out
> > as soon as possible and other things came up today. Will do it in
> > the
> > morning, sorry.
> Here's the timesync code pulled out into a simple patch if you want
> to
> see it.
>
> Bryan, any explanations you want to provide that would help in
> clarifying Mark's issues?
As Douglas Adams would say - "don't panic".
If you look at the final state the code ends up in - we're doing
get_cycles(); as opposed to reading an architectural timer directly.
u64 gb_timesync_platform_get_counter(void)
{
return (u64)get_cycles();
}
You have the entire git history - from the early days where we were reading one of the unused ARMv8 timers the MSM8994 has to the later days where we just do get_cycles()...
At the time when we first started writing the code it wasn't 100% clear if get_cycles() would do, so it was safer to allocate an unused architectural timer and read it directly. Later on and with some experimentation it was possible to switch to get_cycles().
---
bod
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-15 9:35 ` Bryan O'Donoghue
0 siblings, 0 replies; 84+ messages in thread
From: Bryan O'Donoghue @ 2016-09-15 9:35 UTC (permalink / raw)
To: linux-arm-kernel
On Wed, 2016-09-14 at 20:29 +0200, Greg KH wrote:
> On Wed, Sep 14, 2016 at 08:07:54PM +0200, Greg KH wrote:
> >
> > On Wed, Sep 14, 2016 at 06:36:26PM +0100, Mark Rutland wrote:
> > >
> > > Hi Greg,
> > >
> > > On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
> > > >
> > > > Given that it's never a good idea to keep subsystems out of the
> > > > mainline
> > > > kernel, I've put together this pull request that adds the
> > > > greybus driver
> > > > layer to drivers/greybus/.??Because this was 2 1/2 years of
> > > > work, with
> > > > many many developers contributing, I didn't want to flatten all
> > > > of their
> > > > effort into a few small patches, as that wouldn't be very
> > > > fair.??So I've
> > > > built a git tree with all of the changes going back to the
> > > > first commit,
> > > > and merged it into the kernel tree, just like btrfs was merged
> > > > into the
> > > > kernel.
> > > >
> > > > Unless people point out some major problems with this, I'd like
> > > > to get
> > > > it merged into 4.9-rc1.
> > > I'm extremely concerned that these patches have *never* seen
> > > upstream
> > > review, and this pull request gives no real opportunity for
> > > people to
> > > make a judgement regarding the code, as many relevant parties
> > > have not
> > > been Cc'd.
> > As I said, I will send a set of simple patches, I wanted to get
> > this out
> > as soon as possible and other things came up today.??Will do it in
> > the
> > morning, sorry.
> Here's the timesync code pulled out into a simple patch if you want
> to
> see it.
>
> Bryan, any explanations you want to provide that would help in
> clarifying Mark's issues?
As Douglas Adams would say - "don't panic".
If you look at the final state the code ends up in - we're doing
get_cycles(); as opposed to reading an architectural timer directly.
u64 gb_timesync_platform_get_counter(void)
{
????????return (u64)get_cycles();
}
You have the entire git history - from the early days where we were reading one of the unused ARMv8 timers the MSM8994 has to the later days where we just do get_cycles()...
At the time when we first started writing the code it wasn't 100% clear if get_cycles() would do, so it was safer to allocate an unused architectural timer and read it directly. Later on and with some experimentation it was possible to switch to get_cycles().
---
bod
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-15 9:35 ` Bryan O'Donoghue
@ 2016-09-15 10:13 ` Mark Rutland
-1 siblings, 0 replies; 84+ messages in thread
From: Mark Rutland @ 2016-09-15 10:13 UTC (permalink / raw)
To: Bryan O'Donoghue
Cc: Greg KH, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Vaibhav Agarwal, Mark Greer, marc.zyngier, linux-arm-kernel
On Thu, Sep 15, 2016 at 10:35:33AM +0100, Bryan O'Donoghue wrote:
> On Wed, 2016-09-14 at 20:29 +0200, Greg KH wrote:
> > Bryan, any explanations you want to provide that would help in
> > clarifying Mark's issues?
>
> As Douglas Adams would say - "don't panic".
>
> If you look at the final state the code ends up in - we're doing
> get_cycles(); as opposed to reading an architectural timer directly.
>
> u64 gb_timesync_platform_get_counter(void)
> {
> return (u64)get_cycles();
> }
I did in fact notice this, though my wording was somewhat unclear on
that part.
> You have the entire git history - from the early days where we were
> reading one of the unused ARMv8 timers the MSM8994 has to the later
> days where we just do get_cycles()...
>
> At the time when we first started writing the code it wasn't 100%
> clear if get_cycles() would do, so it was safer to allocate an unused
> architectural timer and read it directly. Later on and with some
> experimentation it was possible to switch to get_cycles().
I don't think the history matters, and I don't think that one can rely
on get_cycles() in this manner outside of arch code. Looking at the
state of the tree [1] as of the final commit [2] in the greybus branch,
my points still stand:
* The "google,greybus-frame-time-counter" node is superfluous. It does
not describe a particular device, and duplicates information we have
elsewhere. It does not explicitly define the relationship with the
underlying clocksource.
* The clock-frequency property isn't necessary. The architected timer
drivers know the frequency of the architected timers (MMIO or sysreg),
and they should be queried as to the frequency.
* In general, get_cycles() isn't guaranteed to be any clocksource or
cycle counter in particular, so even if the clock-frequency property
matches the architected timer, that doesn't help.
Beyond that, the fallback code using cpufreq and presumably an actual
cycle counter will be broken in a number of cases -- cycle counts will
not match across CPUs, and on CPUs whic gate clocks in WFI/WFE, they'll
drift randomly.
Per the comment at the top of the file, it looks like you want a
system-wide stable clocksource. If you want that, you need to use a
generic API that allows drivers and arch code to actually provide that,
rather than building one yourself that doesn't work.
If you're trying to synchronise with other agents in the system that are
reading from the MMIO arch timers, then that relationship should be
described explicitly in the DT.
Thanks,
Mark.
[1] https://git.kernel.org/cgit/linux/kernel/git/gregkh/char-misc.git/tree/drivers/greybus/timesync_platform.c?h=greybus&id=f86bfc90a401681838866b5f6826d86cc4c7010c
[2] https://git.kernel.org/cgit/linux/kernel/git/gregkh/char-misc.git/commit/?h=greybus&id=f86bfc90a401681838866b5f6826d86cc4c7010c
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-15 10:13 ` Mark Rutland
0 siblings, 0 replies; 84+ messages in thread
From: Mark Rutland @ 2016-09-15 10:13 UTC (permalink / raw)
To: linux-arm-kernel
On Thu, Sep 15, 2016 at 10:35:33AM +0100, Bryan O'Donoghue wrote:
> On Wed, 2016-09-14 at 20:29 +0200, Greg KH wrote:
> > Bryan, any explanations you want to provide that would help in
> > clarifying Mark's issues?
>
> As Douglas Adams would say - "don't panic".
>
> If you look at the final state the code ends up in - we're doing
> get_cycles(); as opposed to reading an architectural timer directly.
>
> u64 gb_timesync_platform_get_counter(void)
> {
> ????????return (u64)get_cycles();
> }
I did in fact notice this, though my wording was somewhat unclear on
that part.
> You have the entire git history - from the early days where we were
> reading one of the unused ARMv8 timers the MSM8994 has to the later
> days where we just do get_cycles()...
>
> At the time when we first started writing the code it wasn't 100%
> clear if get_cycles() would do, so it was safer to allocate an unused
> architectural timer and read it directly. Later on and with some
> experimentation it was possible to switch to get_cycles().
I don't think the history matters, and I don't think that one can rely
on get_cycles() in this manner outside of arch code. Looking at the
state of the tree [1] as of the final commit [2] in the greybus branch,
my points still stand:
* The "google,greybus-frame-time-counter" node is superfluous. It does
not describe a particular device, and duplicates information we have
elsewhere. It does not explicitly define the relationship with the
underlying clocksource.
* The clock-frequency property isn't necessary. The architected timer
drivers know the frequency of the architected timers (MMIO or sysreg),
and they should be queried as to the frequency.
* In general, get_cycles() isn't guaranteed to be any clocksource or
cycle counter in particular, so even if the clock-frequency property
matches the architected timer, that doesn't help.
Beyond that, the fallback code using cpufreq and presumably an actual
cycle counter will be broken in a number of cases -- cycle counts will
not match across CPUs, and on CPUs whic gate clocks in WFI/WFE, they'll
drift randomly.
Per the comment at the top of the file, it looks like you want a
system-wide stable clocksource. If you want that, you need to use a
generic API that allows drivers and arch code to actually provide that,
rather than building one yourself that doesn't work.
If you're trying to synchronise with other agents in the system that are
reading from the MMIO arch timers, then that relationship should be
described explicitly in the DT.
Thanks,
Mark.
[1] https://git.kernel.org/cgit/linux/kernel/git/gregkh/char-misc.git/tree/drivers/greybus/timesync_platform.c?h=greybus&id=f86bfc90a401681838866b5f6826d86cc4c7010c
[2] https://git.kernel.org/cgit/linux/kernel/git/gregkh/char-misc.git/commit/?h=greybus&id=f86bfc90a401681838866b5f6826d86cc4c7010c
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-14 20:07 ` Rob Herring
@ 2016-09-15 10:17 ` Greg KH
-1 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-15 10:17 UTC (permalink / raw)
To: Rob Herring
Cc: Mark Rutland, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer, Marc Zyngier,
linux-arm-kernel
On Wed, Sep 14, 2016 at 03:07:22PM -0500, Rob Herring wrote:
> On Wed, Sep 14, 2016 at 1:07 PM, Greg KH <gregkh@linuxfoundation.org> wrote:
> > On Wed, Sep 14, 2016 at 06:36:26PM +0100, Mark Rutland wrote:
> >> Hi Greg,
> >>
> >> On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
> >> > Given that it's never a good idea to keep subsystems out of the mainline
> >> > kernel, I've put together this pull request that adds the greybus driver
> >> > layer to drivers/greybus/. Because this was 2 1/2 years of work, with
> >> > many many developers contributing, I didn't want to flatten all of their
> >> > effort into a few small patches, as that wouldn't be very fair. So I've
> >> > built a git tree with all of the changes going back to the first commit,
> >> > and merged it into the kernel tree, just like btrfs was merged into the
> >> > kernel.
> >>
> >> > Unless people point out some major problems with this, I'd like to get
> >> > it merged into 4.9-rc1.
> >>
> >> I'm extremely concerned that these patches have *never* seen upstream
> >> review, and this pull request gives no real opportunity for people to
> >> make a judgement regarding the code, as many relevant parties have not
> >> been Cc'd.
> >
> > As I said, I will send a set of simple patches, I wanted to get this out
> > as soon as possible and other things came up today. Will do it in the
> > morning, sorry.
> >
> >> From a quick scan of the git tree, I can see code (that isn't even
> >> placed under staging/) for which I have fundamental objections to as a
> >> maintainer, and has not been Cc'd to a relevant list.
> >>
> >> For example, I see commit 5a450477311fbfe2 ("greybus: timesync: Add
> >> timesync core driver"). This states that it directly accesses the ARMv7
> >> architected timer, though it's unclear as to precisely what it's doing
> >> since it introduces an (undocumented) compatible string, and what should
> >> be an unnecessary devicetree property.
> >>
> >> That's never gone to the linux-arm-kernel mainline list, myself or Marc
> >> (as maintainers of the arch timer driver), nor has the binding seen any
> >> review on the devicetree mailing list.
> >
> > Hm, odd, I thought we had Rob review all of the device tree bindings,
> > but maybe the timesync stuff missed him. And timesync is "odd" to say
> > the least, wait until you see the firmware side of it :)
>
> I have not. There weren't any when I was involved (other than SOC
> related bindings). Some like USB devices were discussed at least, but
> at the time there was no common definition of how to deal with
> soldered, on-board USB devices. And that was also what's good enough
> to move forward with development, not ready for mainline. There's a
> binding now for USB, but what's there for arche predates it IIRC. This
> is the first I've heard of timesync having a binding. I can't imagine
> why it needs one.
Ah, I'll let Bryan answer that one :)
> I was going to stay out of this, but now that I'm here...
>
> I'm not all that worried about the quality of the code, but do
> question whether this really makes sense to merge at this time. While
> I worked on Ara and would like to see all this code be used, I'm
> pretty doubtful it will be. Yes, Motorola is using a version of it,
> but they forked it some time back and changed who knows what. So what
> is upstream won't likely even work with any publicly available device.
> And how long that Motorola phone lasts is unknown. Sure it is self
> contained, but maintenance to keep it in tree is not 0.
We have people who are willing to maintain it (me and Johan) and others
on the cc: have expressed interest in doing things with it (Rui has
great ideas about extending gbsim in ways to make it easier to test and
use).
And there is never a restriction on accepting kernel code for "publicly
availble devices only", remember, we have ripped CPU code out of the
kernel for chips that never shipped. It just has to not affect others
in ways that would affect them.
I merge new driver subsystems to the kernel every release that are much
larger than this, and have almost no users out there, so this shouldn't
be a surprise to anyone (who here has a most-bus device? A unisys
visorbus device? MCB device? I can go on :)
And with gbsim, you can run this code today on your laptop, or directly
on a target system like a beagle bone black or minnowboard. There's a
talk about how to use greybus over IP at ELC in a few weeks, and someone
is working on porting the firmware side to an arduino to make it simper
to control the devices on such a target from a host computer using the
greybus protocol.
And getting Motorola to merge back in with this upstream code is a good
goal, having it upstream is a neutral place where everyone can work
together on it. Keeping it in a random github project makes forks
almost inevitable, as we all know.
> There's also things that never got solved. Like how do you describe
> devices on I2C, SPI, UART, etc. behind a greybus device? The plan was
> to use DT overlays, but that was never solved and brings a whole set
> of problems to solve upstream.
That is only an issue if you want to bind a kernel driver for an
existing i2c/spi chip to an i2c/spi greybus device. With the code we
have today, we do it for a specific SPI chip (for firmware download),
but rely on everything to be userspace-only accesses to make it simpler
at this point in time.
When DT overlays get more settled down, yes, I want to revisit this idea
of how to do it for greybus devices, but that's a long-term goal and is
not required at all right now to have a working system and devices.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-15 10:17 ` Greg KH
0 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-15 10:17 UTC (permalink / raw)
To: linux-arm-kernel
On Wed, Sep 14, 2016 at 03:07:22PM -0500, Rob Herring wrote:
> On Wed, Sep 14, 2016 at 1:07 PM, Greg KH <gregkh@linuxfoundation.org> wrote:
> > On Wed, Sep 14, 2016 at 06:36:26PM +0100, Mark Rutland wrote:
> >> Hi Greg,
> >>
> >> On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
> >> > Given that it's never a good idea to keep subsystems out of the mainline
> >> > kernel, I've put together this pull request that adds the greybus driver
> >> > layer to drivers/greybus/. Because this was 2 1/2 years of work, with
> >> > many many developers contributing, I didn't want to flatten all of their
> >> > effort into a few small patches, as that wouldn't be very fair. So I've
> >> > built a git tree with all of the changes going back to the first commit,
> >> > and merged it into the kernel tree, just like btrfs was merged into the
> >> > kernel.
> >>
> >> > Unless people point out some major problems with this, I'd like to get
> >> > it merged into 4.9-rc1.
> >>
> >> I'm extremely concerned that these patches have *never* seen upstream
> >> review, and this pull request gives no real opportunity for people to
> >> make a judgement regarding the code, as many relevant parties have not
> >> been Cc'd.
> >
> > As I said, I will send a set of simple patches, I wanted to get this out
> > as soon as possible and other things came up today. Will do it in the
> > morning, sorry.
> >
> >> From a quick scan of the git tree, I can see code (that isn't even
> >> placed under staging/) for which I have fundamental objections to as a
> >> maintainer, and has not been Cc'd to a relevant list.
> >>
> >> For example, I see commit 5a450477311fbfe2 ("greybus: timesync: Add
> >> timesync core driver"). This states that it directly accesses the ARMv7
> >> architected timer, though it's unclear as to precisely what it's doing
> >> since it introduces an (undocumented) compatible string, and what should
> >> be an unnecessary devicetree property.
> >>
> >> That's never gone to the linux-arm-kernel mainline list, myself or Marc
> >> (as maintainers of the arch timer driver), nor has the binding seen any
> >> review on the devicetree mailing list.
> >
> > Hm, odd, I thought we had Rob review all of the device tree bindings,
> > but maybe the timesync stuff missed him. And timesync is "odd" to say
> > the least, wait until you see the firmware side of it :)
>
> I have not. There weren't any when I was involved (other than SOC
> related bindings). Some like USB devices were discussed at least, but
> at the time there was no common definition of how to deal with
> soldered, on-board USB devices. And that was also what's good enough
> to move forward with development, not ready for mainline. There's a
> binding now for USB, but what's there for arche predates it IIRC. This
> is the first I've heard of timesync having a binding. I can't imagine
> why it needs one.
Ah, I'll let Bryan answer that one :)
> I was going to stay out of this, but now that I'm here...
>
> I'm not all that worried about the quality of the code, but do
> question whether this really makes sense to merge at this time. While
> I worked on Ara and would like to see all this code be used, I'm
> pretty doubtful it will be. Yes, Motorola is using a version of it,
> but they forked it some time back and changed who knows what. So what
> is upstream won't likely even work with any publicly available device.
> And how long that Motorola phone lasts is unknown. Sure it is self
> contained, but maintenance to keep it in tree is not 0.
We have people who are willing to maintain it (me and Johan) and others
on the cc: have expressed interest in doing things with it (Rui has
great ideas about extending gbsim in ways to make it easier to test and
use).
And there is never a restriction on accepting kernel code for "publicly
availble devices only", remember, we have ripped CPU code out of the
kernel for chips that never shipped. It just has to not affect others
in ways that would affect them.
I merge new driver subsystems to the kernel every release that are much
larger than this, and have almost no users out there, so this shouldn't
be a surprise to anyone (who here has a most-bus device? A unisys
visorbus device? MCB device? I can go on :)
And with gbsim, you can run this code today on your laptop, or directly
on a target system like a beagle bone black or minnowboard. There's a
talk about how to use greybus over IP at ELC in a few weeks, and someone
is working on porting the firmware side to an arduino to make it simper
to control the devices on such a target from a host computer using the
greybus protocol.
And getting Motorola to merge back in with this upstream code is a good
goal, having it upstream is a neutral place where everyone can work
together on it. Keeping it in a random github project makes forks
almost inevitable, as we all know.
> There's also things that never got solved. Like how do you describe
> devices on I2C, SPI, UART, etc. behind a greybus device? The plan was
> to use DT overlays, but that was never solved and brings a whole set
> of problems to solve upstream.
That is only an issue if you want to bind a kernel driver for an
existing i2c/spi chip to an i2c/spi greybus device. With the code we
have today, we do it for a specific SPI chip (for firmware download),
but rely on everything to be userspace-only accesses to make it simpler
at this point in time.
When DT overlays get more settled down, yes, I want to revisit this idea
of how to do it for greybus devices, but that's a long-term goal and is
not required at all right now to have a working system and devices.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-15 10:13 ` Mark Rutland
@ 2016-09-15 10:35 ` Bryan O'Donoghue
-1 siblings, 0 replies; 84+ messages in thread
From: Bryan O'Donoghue @ 2016-09-15 10:35 UTC (permalink / raw)
To: Mark Rutland
Cc: Greg KH, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Vaibhav Agarwal, Mark Greer, marc.zyngier, linux-arm-kernel
On Thu, 2016-09-15 at 11:13 +0100, Mark Rutland wrote:
> On Thu, Sep 15, 2016 at 10:35:33AM +0100, Bryan O'Donoghue wrote:
> >
> I don't think the history matters,
Your comment seemed to indicate you thought we were reading a
architectural timer directly - which we aren't.
> and I don't think that one can rely
> on get_cycles() in this manner outside of arch code.
I don't follow your meaning. What's wrong with get_cycles() ? You've
already said you don't think reading an architectural timer directly is
correct.
The objective is to read one of the free-running counters in MSM8994,
clocked by the PMIC. The refclk provided by PMIC is distributed to each
processor in the system.
> Looking at the
> state of the tree [1] as of the final commit [2] in the greybus
> branch,
> my points still stand:
>
> * The "google,greybus-frame-time-counter" node is superfluous. It
> does
> not describe a particular device,
It describes a timer running @ 19.2MHz, clocked by PMIC refclk.
> and duplicates information we have
> elsewhere.
Can you give an example ?
> It does not explicitly define the relationship with the
> underlying clocksource.
> * The clock-frequency property isn't necessary. The architected timer
> drivers know the frequency of the architected timers (MMIO or
> sysreg),
> and they should be queried as to the frequency.
OK so if I'm understanding you. You think get_cycles() is fine but that
instead of encoding a "greybus-frame-time-counter" the platform code
should interrogate the frequency provided - is that correct ?
> Beyond that, the fallback code using cpufreq and presumably an actual
> cycle counter will be broken in a number of cases
Of course the fallback will be broken... it's not supposed to work if
you don't have a timer that can be used - just compile, run and print a
complaint - i.e., this won't really do FrameTime on an x86... then
again since so much of the underlying greybus/unipro hardware -
requires a 19.2MHz refclk - if you were to try to do greybus on x86
you'd need to solve that problem.
>
> Per the comment at the top of the file, it looks like you want a
> system-wide stable clocksource. If you want that, you need to use a
> generic API that allows drivers and arch code to actually provide
> that,
> rather than building one yourself that doesn't work.
Hmm. The objective is to read one of the timers clocked by the PMIC
refclk input. refclk is provided to each processor in the system - and
on MSM8994 clocks the MMIO timers. It's used to drive the PLL on the
other processors - which in turn drive the timers that the Modules use
to read their own local counters. We want to read that counter on MSM
directly - get_cycles() has worked nicely so far.
>
> If you're trying to synchronise with other agents in the system that
> are
> reading from the MMIO arch timers,
No. The MMIO timers are useful only to the MSM. We don't have any type
of parallel (or serial) bus that can access that on-chip resource.
MSM8994 -- > USB
APBridge (timer) -> UniPro bus
-> Module with a UART
-> Module with a GPIO
-> Module with an etc, etc
-> SPI bus
-> SVC
Owns FrameTime
So the SVC owns FrameTime and diseminates that to other entities in the
system by way of a GPIO and greybus. It's up to the MSM8994 to select a
timer that works for it - the other processors in the system are
responsible for their own timers.
---
bod
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-15 10:35 ` Bryan O'Donoghue
0 siblings, 0 replies; 84+ messages in thread
From: Bryan O'Donoghue @ 2016-09-15 10:35 UTC (permalink / raw)
To: linux-arm-kernel
On Thu, 2016-09-15 at 11:13 +0100, Mark Rutland wrote:
> On Thu, Sep 15, 2016 at 10:35:33AM +0100, Bryan O'Donoghue wrote:
> >?
> I don't think the history matters,
Your comment seemed to indicate you thought we were reading a
architectural timer directly - which we aren't.
> and I don't think that one can rely
> on get_cycles() in this manner outside of arch code.
I don't follow your meaning. What's wrong with get_cycles() ? You've
already said you don't think reading an architectural timer directly is
correct.
The objective is to read one of the free-running counters in MSM8994,
clocked by the PMIC. The refclk provided by PMIC is distributed to each
processor in the system.
> Looking at the
> state of the tree [1] as of the final commit [2] in the greybus
> branch,
> my points still stand:
>
> * The "google,greybus-frame-time-counter" node is superfluous. It
> does
> ? not describe a particular device,
It describes a timer running @ 19.2MHz, clocked by PMIC refclk.
> and duplicates information we have
> ? elsewhere.
Can you give an example ?
> It does not explicitly define the relationship with the
> ? underlying clocksource.
> * The clock-frequency property isn't necessary. The architected timer
> ? drivers know the frequency of the architected timers (MMIO or
> sysreg),
> ? and they should be queried as to the frequency.
OK so if I'm understanding you. You think get_cycles() is fine but that
instead of encoding a "greybus-frame-time-counter" the platform code
should interrogate the frequency provided - is that correct ?
> Beyond that, the fallback code using cpufreq and presumably an actual
> cycle counter will be broken in a number of cases
Of course the fallback will be broken... it's not supposed to work if
you don't have a timer that can be used - just compile, run and print a
complaint - i.e., this won't really do FrameTime on an x86... then
again since so much of the underlying greybus/unipro hardware -
requires a 19.2MHz refclk - if you were to try to do greybus on x86
you'd need to solve that problem.
>
> Per the comment at the top of the file, it looks like you want a
> system-wide stable clocksource. If you want that, you need to use a
> generic API that allows drivers and arch code to actually provide
> that,
> rather than building one yourself that doesn't work.
Hmm. The objective is to read one of the timers clocked by the PMIC
refclk input. refclk is provided to each processor in the system - and
on MSM8994 clocks the MMIO timers. It's used to drive the PLL on the
other processors - which in turn drive the timers that the Modules use
to read their own local counters. We want to read that counter on MSM
directly - get_cycles() has worked nicely so far.
>
> If you're trying to synchronise with other agents in the system that
> are
> reading from the MMIO arch timers,
No. The MMIO timers are useful only to the MSM. We don't have any type
of parallel (or serial) bus that can access that on-chip resource.
MSM8994 -- > USB
? ? ? ? ? ? ?APBridge (timer) -> UniPro bus
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? -> Module with a UART
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? -> Module with a GPIO
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? -> Module with an etc, etc
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? -> SPI bus
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? -> SVC
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?Owns FrameTime
So the SVC owns FrameTime and diseminates that to other entities in the
system by way of a GPIO and greybus. It's up to the MSM8994 to select a
timer that works for it - the other processors in the system are
responsible for their own timers.
---
bod
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-15 10:35 ` Bryan O'Donoghue
@ 2016-09-15 10:47 ` Bryan O'Donoghue
-1 siblings, 0 replies; 84+ messages in thread
From: Bryan O'Donoghue @ 2016-09-15 10:47 UTC (permalink / raw)
To: Mark Rutland
Cc: Greg KH, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Vaibhav Agarwal, Mark Greer, marc.zyngier, linux-arm-kernel
On Thu, 2016-09-15 at 11:35 +0100, Bryan O'Donoghue wrote:
>
Here's a slightly better diagram.
PMIC -> refclk provided to each (timer) element below.
MSM8994(timer) -- > USB
WD8a
APBridgeA (timer) -> UniPro bus
WD8a
-> Module(timer) with UART
WD1
-> Module(timer) with GPIO
WD2
-> Module(timer) with blah
WD3
-> SPI bus
-> SVC(timer)
Owns FrameTime
GPIO {WD0...WDn}
So yes, each processor has it's own timer. We aren't trying to read the
MSM's FrameTime.
---
bod
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-15 10:47 ` Bryan O'Donoghue
0 siblings, 0 replies; 84+ messages in thread
From: Bryan O'Donoghue @ 2016-09-15 10:47 UTC (permalink / raw)
To: linux-arm-kernel
On Thu, 2016-09-15 at 11:35 +0100, Bryan O'Donoghue wrote:
>
Here's a slightly better diagram.
PMIC -> refclk provided to each (timer) element below.
MSM8994(timer) -- > USB
WD8a
? ? ? ? ? ? ? ?APBridgeA (timer) -> UniPro bus
? ? ? ? ? ? ? ?WD8a
????????????????????????????????????????????-> Module(timer) with UART
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?WD1
????????????????????????????????????????????-> Module(timer) with GPIO
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?WD2
????????????????????????????????????????????-> Module(timer) with blah
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?WD3
??????????????????????????????-> SPI bus
????????????????????????????????????????????-> SVC(timer)
???????????????????????????????????????????????Owns FrameTime
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?GPIO {WD0...WDn}
So yes, each processor has it's own timer. We aren't trying to read the
MSM's FrameTime.
---
bod
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-15 10:17 ` Greg KH
@ 2016-09-15 11:02 ` Bryan O'Donoghue
-1 siblings, 0 replies; 84+ messages in thread
From: Bryan O'Donoghue @ 2016-09-15 11:02 UTC (permalink / raw)
To: Greg KH, Rob Herring
Cc: Mark Rutland, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Viresh Kumar, Alex Elder, David Lin,
Vaibhav Agarwal, Mark Greer, Marc Zyngier, linux-arm-kernel
On Thu, 2016-09-15 at 12:17 +0200, Greg KH wrote:
> This
> > is the first I've heard of timesync having a binding. I can't
> imagine
> > why it needs one.
>
> Ah, I'll let Bryan answer that one :)
It's possible we could drop the binding. It was needed to describe the
register location of the MMIO architectural register on MSM8994. I have
this binding ATM the describe fact that get_cycles() on MSM8994 returns
a free-running counter, clocked by refclk and that refclk is provided
to each processor that want to do FrameTime - i.e. the clock driving
get_cycles() comes from PMIC and drives the relevant PLLs on the
downstream processors clocking their respective TMR blocks.
static const struct of_device_id arch_timer_of_match[] = {
{ .compatible = "google,greybus-frame-time-counter", },
{},
};
I'm not aware of a corresponding kernel API that describes the
frequency get_cycles() operates at but if there is one then there's no
need for this binding.
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-15 11:02 ` Bryan O'Donoghue
0 siblings, 0 replies; 84+ messages in thread
From: Bryan O'Donoghue @ 2016-09-15 11:02 UTC (permalink / raw)
To: linux-arm-kernel
On Thu, 2016-09-15 at 12:17 +0200, Greg KH wrote:
> This
> > is the first I've heard of timesync having a binding. I can't
> imagine
> > why it needs one.
>
> Ah, I'll let Bryan answer that one :)
It's possible we could drop the binding. It was needed to describe the
register location of the MMIO architectural register on MSM8994. I have
this binding ATM the describe fact that get_cycles() on MSM8994 returns
a free-running counter, clocked by refclk and that refclk is provided
to each processor that want to do FrameTime - i.e. the clock driving
get_cycles() comes from PMIC and drives the relevant PLLs on the
downstream processors clocking their respective TMR blocks.
static const struct of_device_id arch_timer_of_match[] = {
????????{ .compatible???= "google,greybus-frame-time-counter", },
????????{},
};
I'm not aware of a corresponding kernel API that describes the
frequency get_cycles() operates at but if there is one then there's no
need for this binding.
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-15 10:35 ` Bryan O'Donoghue
@ 2016-09-15 11:20 ` Mark Rutland
-1 siblings, 0 replies; 84+ messages in thread
From: Mark Rutland @ 2016-09-15 11:20 UTC (permalink / raw)
To: Bryan O'Donoghue
Cc: Greg KH, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Vaibhav Agarwal, Mark Greer, marc.zyngier, linux-arm-kernel
Hi,
More questions below. Perhaps some of these will be implicitly answered
when the linearised patches appear, and I'm happy to wait until then to
continue the discussion, as I suspect otherwise we're all likely to end
up exasperated.
Please do Cc me on those.
Regardless, until those appear and a reasonable time has been given for
replies, my comments regarding the lack of review stand, as does my NAK
for the series.
On Thu, Sep 15, 2016 at 11:35:56AM +0100, Bryan O'Donoghue wrote:
> On Thu, 2016-09-15 at 11:13 +0100, Mark Rutland wrote:
> > On Thu, Sep 15, 2016 at 10:35:33AM +0100, Bryan O'Donoghue wrote:
> > >
> > I don't think the history matters,
>
> Your comment seemed to indicate you thought we were reading a
> architectural timer directly - which we aren't.
Sure, and as I pointed out, the comment in the HEAD commit still claims
it does, even if the code doesn't. This is at best, confusing, and the
history of how it came to be there doesn't really matter...
> > and I don't think that one can rely
> > on get_cycles() in this manner outside of arch code.
>
> I don't follow your meaning. What's wrong with get_cycles() ? You've
> already said you don't think reading an architectural timer directly is
> correct.
I pointed out a number of issues in my previous reply.
For example, you have absolutely no guarantee as to what backs
get_cycles(). Despite this, the code assumes that get_cycles() is backed
by something running at the frequency described in a
"google,greybus-frame-time-counter" node.
Even if this *happens* to match what some piece of arch code provides
today on some platform, it is in no way *guaranteed*.
> The objective is to read one of the free-running counters in MSM8994,
> clocked by the PMIC. The refclk provided by PMIC is distributed to each
> processor in the system.
>
> > Looking at the
> > state of the tree [1] as of the final commit [2] in the greybus
> > branch,
> > my points still stand:
> >
> > * The "google,greybus-frame-time-counter" node is superfluous. It
> > does
> > not describe a particular device,
>
> It describes a timer running @ 19.2MHz, clocked by PMIC refclk.
... which you assume is whatever backs get_cycles(), which you in
practice assume is the architected timer. For which we *already* have a
binding and driver.
Given that, as far as I can tell, "google,greybus-frame-time-counter"
describes a software construct that uses this, not an actual piece of
hardware.
> > and duplicates information we have elsewhere.
>
> Can you give an example ?
Trivially, the CNTFRQ register in the architected timer (which is common
across MMIO/sysreg), which you can query with arch_timer_get_rate().
Note that isn't guaranteed to match get_cycles() either. You need a
better API to call.
> > * The clock-frequency property isn't necessary. The architected timer
> > drivers know the frequency of the architected timers (MMIO or
> > sysreg),
> > and they should be queried as to the frequency.
>
> OK so if I'm understanding you. You think get_cycles() is fine but that
> instead of encoding a "greybus-frame-time-counter" the platform code
> should interrogate the frequency provided - is that correct ?
You should definitely interrogate the relevant driver, somehow.
Without a higher-level view of what you're trying to achieve, it's not
clear to me whether get_cycles() is the right interface.
> > Beyond that, the fallback code using cpufreq and presumably an actual
> > cycle counter will be broken in a number of cases
>
> Of course the fallback will be broken... it's not supposed to work if
> you don't have a timer that can be used - just compile, run and print a
> complaint - i.e., this won't really do FrameTime on an x86... then
> again since so much of the underlying greybus/unipro hardware -
> requires a 19.2MHz refclk - if you were to try to do greybus on x86
> you'd need to solve that problem.
If it's never going to work, why give the illusion that it might?
If you don't have the necessary prerequisites, fail to probe entirely.
Prevent drivers that depend on the non-existent functionality from
probing, or have them avoid the facility which is not available.
Don't provide them with something that can appear to work for a while,
yet will fall over in a breeze.
> > Per the comment at the top of the file, it looks like you want a
> > system-wide stable clocksource. If you want that, you need to use a
> > generic API that allows drivers and arch code to actually provide
> > that, rather than building one yourself that doesn't work.
>
> Hmm. The objective is to read one of the timers clocked by the PMIC
> refclk input. refclk is provided to each processor in the system - and
> on MSM8994 clocks the MMIO timers. It's used to drive the PLL on the
> other processors - which in turn drive the timers that the Modules use
> to read their own local counters. We want to read that counter on MSM
> directly - get_cycles() has worked nicely so far.
This is too low-level for me to see what you're actually trying to
achieve. The fact that you want to read a specific timer is an
implementation detail.
Why can't you use any other timer? Correctness? Performance? (Why) is
the fact that the PLL drives module timers important?
> > If you're trying to synchronise with other agents in the system that
> > are reading from the MMIO arch timers,
>
> No. The MMIO timers are useful only to the MSM. We don't have any type
> of parallel (or serial) bus that can access that on-chip resource.
To be clear, by "other agents in the system", I'm also asking about
other devices within the SoC (e.g. anything other than the CPUs running
this instance of Linux).
> MSM8994 -- > USB
> APBridge (timer) -> UniPro bus
> -> Module with a UART
> -> Module with a GPIO
> -> Module with an etc, etc
> -> SPI bus
> -> SVC
> Owns FrameTime
>
> So the SVC owns FrameTime and diseminates that to other entities in the
> system by way of a GPIO and greybus. It's up to the MSM8994 to select a
> timer that works for it - the other processors in the system are
> responsible for their own timers.
Sorry, but this doesn't clarify much from my PoV.
* What are the requirements for that timer?
* Is there_any_ implicit relationship with the module timers derived
from the fact they share a parent PLL? Is that a requirement somehow?
Thanks,
Mark.
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-15 11:20 ` Mark Rutland
0 siblings, 0 replies; 84+ messages in thread
From: Mark Rutland @ 2016-09-15 11:20 UTC (permalink / raw)
To: linux-arm-kernel
Hi,
More questions below. Perhaps some of these will be implicitly answered
when the linearised patches appear, and I'm happy to wait until then to
continue the discussion, as I suspect otherwise we're all likely to end
up exasperated.
Please do Cc me on those.
Regardless, until those appear and a reasonable time has been given for
replies, my comments regarding the lack of review stand, as does my NAK
for the series.
On Thu, Sep 15, 2016 at 11:35:56AM +0100, Bryan O'Donoghue wrote:
> On Thu, 2016-09-15 at 11:13 +0100, Mark Rutland wrote:
> > On Thu, Sep 15, 2016 at 10:35:33AM +0100, Bryan O'Donoghue wrote:
> > >?
> > I don't think the history matters,
>
> Your comment seemed to indicate you thought we were reading a
> architectural timer directly - which we aren't.
Sure, and as I pointed out, the comment in the HEAD commit still claims
it does, even if the code doesn't. This is at best, confusing, and the
history of how it came to be there doesn't really matter...
> > and I don't think that one can rely
> > on get_cycles() in this manner outside of arch code.
>
> I don't follow your meaning. What's wrong with get_cycles() ? You've
> already said you don't think reading an architectural timer directly is
> correct.
I pointed out a number of issues in my previous reply.
For example, you have absolutely no guarantee as to what backs
get_cycles(). Despite this, the code assumes that get_cycles() is backed
by something running at the frequency described in a
"google,greybus-frame-time-counter" node.
Even if this *happens* to match what some piece of arch code provides
today on some platform, it is in no way *guaranteed*.
> The objective is to read one of the free-running counters in MSM8994,
> clocked by the PMIC. The refclk provided by PMIC is distributed to each
> processor in the system.
>
> > Looking at the
> > state of the tree [1] as of the final commit [2] in the greybus
> > branch,
> > my points still stand:
> >
> > * The "google,greybus-frame-time-counter" node is superfluous. It
> > does
> > ? not describe a particular device,
>
> It describes a timer running @ 19.2MHz, clocked by PMIC refclk.
... which you assume is whatever backs get_cycles(), which you in
practice assume is the architected timer. For which we *already* have a
binding and driver.
Given that, as far as I can tell, "google,greybus-frame-time-counter"
describes a software construct that uses this, not an actual piece of
hardware.
> > and duplicates information we have ? elsewhere.
>
> Can you give an example ?
Trivially, the CNTFRQ register in the architected timer (which is common
across MMIO/sysreg), which you can query with arch_timer_get_rate().
Note that isn't guaranteed to match get_cycles() either. You need a
better API to call.
> > * The clock-frequency property isn't necessary. The architected timer
> > ? drivers know the frequency of the architected timers (MMIO or
> > sysreg),
> > ? and they should be queried as to the frequency.
>
> OK so if I'm understanding you. You think get_cycles() is fine but that
> instead of encoding a "greybus-frame-time-counter" the platform code
> should interrogate the frequency provided - is that correct ?
You should definitely interrogate the relevant driver, somehow.
Without a higher-level view of what you're trying to achieve, it's not
clear to me whether get_cycles() is the right interface.
> > Beyond that, the fallback code using cpufreq and presumably an actual
> > cycle counter will be broken in a number of cases
>
> Of course the fallback will be broken... it's not supposed to work if
> you don't have a timer that can be used - just compile, run and print a
> complaint - i.e., this won't really do FrameTime on an x86... then
> again since so much of the underlying greybus/unipro hardware -
> requires a 19.2MHz refclk - if you were to try to do greybus on x86
> you'd need to solve that problem.
If it's never going to work, why give the illusion that it might?
If you don't have the necessary prerequisites, fail to probe entirely.
Prevent drivers that depend on the non-existent functionality from
probing, or have them avoid the facility which is not available.
Don't provide them with something that can appear to work for a while,
yet will fall over in a breeze.
> > Per the comment at the top of the file, it looks like you want a
> > system-wide stable clocksource. If you want that, you need to use a
> > generic API that allows drivers and arch code to actually provide
> > that, rather than building one yourself that doesn't work.
>
> Hmm. The objective is to read one of the timers clocked by the PMIC
> refclk input. refclk is provided to each processor in the system - and
> on MSM8994 clocks the MMIO timers. It's used to drive the PLL on the
> other processors - which in turn drive the timers that the Modules use
> to read their own local counters. We want to read that counter on MSM
> directly - get_cycles() has worked nicely so far.
This is too low-level for me to see what you're actually trying to
achieve. The fact that you want to read a specific timer is an
implementation detail.
Why can't you use any other timer? Correctness? Performance? (Why) is
the fact that the PLL drives module timers important?
> > If you're trying to synchronise with other agents in the system that
> > are reading from the MMIO arch timers,
>
> No. The MMIO timers are useful only to the MSM. We don't have any type
> of parallel (or serial) bus that can access that on-chip resource.
To be clear, by "other agents in the system", I'm also asking about
other devices within the SoC (e.g. anything other than the CPUs running
this instance of Linux).
> MSM8994 -- > USB
> ? ? ? ? ? ? ?APBridge (timer) -> UniPro bus
> ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? -> Module with a UART
> ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? -> Module with a GPIO
> ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? -> Module with an etc, etc
> ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? -> SPI bus
> ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? -> SVC
> ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?Owns FrameTime
>
> So the SVC owns FrameTime and diseminates that to other entities in the
> system by way of a GPIO and greybus. It's up to the MSM8994 to select a
> timer that works for it - the other processors in the system are
> responsible for their own timers.
Sorry, but this doesn't clarify much from my PoV.
* What are the requirements for that timer?
* Is there_any_ implicit relationship with the module timers derived
from the fact they share a parent PLL? Is that a requirement somehow?
Thanks,
Mark.
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-15 11:20 ` Mark Rutland
@ 2016-09-15 11:48 ` Bryan O'Donoghue
-1 siblings, 0 replies; 84+ messages in thread
From: Bryan O'Donoghue @ 2016-09-15 11:48 UTC (permalink / raw)
To: Mark Rutland
Cc: Greg KH, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Vaibhav Agarwal, Mark Greer, marc.zyngier, linux-arm-kernel
On Thu, 2016-09-15 at 12:20 +0100, Mark Rutland wrote:
> Hi,
>
> More questions below. Perhaps some of these will be implicitly
> answered
> when the linearised patches appear, and I'm happy to wait until then
> to
> continue the discussion, as I suspect otherwise we're all likely to
> end
> up exasperated.
>
> Please do Cc me on those.
>
> Regardless, until those appear and a reasonable time has been given
> for
> replies, my comments regarding the lack of review stand, as does my
> NAK
> for the series.
>
> On Thu, Sep 15, 2016 at 11:35:56AM +0100, Bryan O'Donoghue wrote:
> >
> > On Thu, 2016-09-15 at 11:13 +0100, Mark Rutland wrote:
> > >
> > > On Thu, Sep 15, 2016 at 10:35:33AM +0100, Bryan O'Donoghue wrote:
> > > >
> > > >
> > > I don't think the history matters,
> > Your comment seemed to indicate you thought we were reading a
> > architectural timer directly - which we aren't.
> Sure, and as I pointed out, the comment in the HEAD commit still
> claims
> it does, even if the code doesn't. This is at best, confusing, and
> the
> history of how it came to be there doesn't really matter...
TBH a whole git history will invariably contain things developers did,
thought better of and then backed out as is the case here.
>
> >
> > >
> > > and I don't think that one can rely
> > > on get_cycles() in this manner outside of arch code.
> > I don't follow your meaning. What's wrong with get_cycles() ?
> > You've
> > already said you don't think reading an architectural timer
> > directly is
> > correct.
> I pointed out a number of issues in my previous reply.
On MSM8994 the timer backing get_cycles() is one of the MMIO
architectural timers (which is why I switched over in the end). There's
not much else that can be done bar custom silicon - this particular
timer is as good as it gets, more of a "how do we synchronise time with
the hardware we have" than a "lets design in a feature to synchronise
time" - which was something we were focusing in on for later
silicon...
>
> For example, you have absolutely no guarantee as to what backs
> get_cycles(). Despite this, the code assumes that get_cycles() is
> backed
> by something running at the frequency described in a
> "google,greybus-frame-time-counter" node.
>
> Even if this *happens* to match what some piece of arch code provides
> today on some platform, it is in no way *guaranteed*.
That's the point though, if you declare "google,greybus-frame-time-
counter" in your platform code - then you can use 'get_cycles()' in
this manner - if not - then you need to take steps in your own new
platform to provide that same level of functionality. You could switch
to an MSM8996 or an MSM8998 declare this node and bob's your uncle.
OTOH declaring this node on x86 would be a bit pointless. You'd be
better off providing a timer on a PCI bar, and binding that into
greybus with some x86/x86-platform code...
>
> >
> > The objective is to read one of the free-running counters in
> > MSM8994,
> > clocked by the PMIC. The refclk provided by PMIC is distributed to
> > each
> > processor in the system.
> >
> > >
> > > Looking at the
> > > state of the tree [1] as of the final commit [2] in the greybus
> > > branch,
> > > my points still stand:
> > >
> > > * The "google,greybus-frame-time-counter" node is superfluous. It
> > > does
> > > not describe a particular device,
> > It describes a timer running @ 19.2MHz, clocked by PMIC refclk.
> ... which you assume is whatever backs get_cycles(), which you in
> practice assume is the architected timer. For which we *already* have
> a
> binding and driver.
It's a requirement rather than assumption. If you declare that node,
it's assumed the timer driving get_cycles() does what it says on the
greybus-frame-time-counter tin.
> > > and duplicates information we have elsewhere.
> > Can you give an example ?
> Trivially, the CNTFRQ register in the architected timer (which is
> common
> across MMIO/sysreg), which you can query with arch_timer_get_rate().
>
> Note that isn't guaranteed to match get_cycles() either. You need a
> better API to call.
In that case a DT entry makes sense I'd say.
> You should definitely interrogate the relevant driver, somehow.
Hrmm. TBH if we are ruling out arch_timer_get_rate() then I think a DT
entry (which BTW is greybus specific) is the more intelligent way
forward. The greybus platform implementer needs to understand the
dependencies and take action to meet those dependencies should he or
she wish to support this feature.
I'm not opposed necessarily to calling arch_timer_get_rate() instead of
a DT binding, assuming it works, and drawing a line under it for
MSM8994. As I've said it's up to a system architect for other platforms
to go and do the necessary design to support this feature and this will
almost certainly require new platform code both here and in other
places anyway.
>
> Without a higher-level view of what you're trying to achieve, it's
> not
> clear to me whether get_cycles() is the right interface.
I appreciate that.
---
bod
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-15 11:48 ` Bryan O'Donoghue
0 siblings, 0 replies; 84+ messages in thread
From: Bryan O'Donoghue @ 2016-09-15 11:48 UTC (permalink / raw)
To: linux-arm-kernel
On Thu, 2016-09-15 at 12:20 +0100, Mark Rutland wrote:
> Hi,
>
> More questions below. Perhaps some of these will be implicitly
> answered
> when the linearised patches appear, and I'm happy to wait until then
> to
> continue the discussion, as I suspect otherwise we're all likely to
> end
> up exasperated.
>
> Please do Cc me on those.
>
> Regardless, until those appear and a reasonable time has been given
> for
> replies, my comments regarding the lack of review stand, as does my
> NAK
> for the series.
>
> On Thu, Sep 15, 2016 at 11:35:56AM +0100, Bryan O'Donoghue wrote:
> >
> > On Thu, 2016-09-15 at 11:13 +0100, Mark Rutland wrote:
> > >
> > > On Thu, Sep 15, 2016 at 10:35:33AM +0100, Bryan O'Donoghue wrote:
> > > >
> > > > ?
> > > I don't think the history matters,?
> > Your comment seemed to indicate you thought we were reading a
> > architectural timer directly - which we aren't.
> Sure, and as I pointed out, the comment in the HEAD commit still
> claims
> it does, even if the code doesn't. This is at best, confusing, and
> the
> history of how it came to be there doesn't really matter...
TBH a whole git history will invariably contain things developers did,
thought better of and then backed out as is the case here.
>
> >
> > >
> > > and I don't think that one can rely
> > > on get_cycles() in this manner outside of arch code.
> > I don't follow your meaning. What's wrong with get_cycles() ?
> > You've
> > already said you don't think reading an architectural timer
> > directly is
> > correct.
> I pointed out a number of issues in my previous reply.
On MSM8994 the timer backing get_cycles() is one of the MMIO
architectural timers (which is why I switched over in the end). There's
not much else that can be done bar custom silicon - this particular
timer is as good as it gets, more of a "how do we synchronise time with
the hardware we have" than a "lets design in a feature to synchronise
time" - which was something we were focusing in on for later
silicon...?
>
> For example, you have absolutely no guarantee as to what backs
> get_cycles(). Despite this, the code assumes that get_cycles() is
> backed
> by something running at the frequency described in a
> "google,greybus-frame-time-counter" node.
>
> Even if this *happens* to match what some piece of arch code provides
> today on some platform, it is in no way *guaranteed*.
That's the point though, if you declare "google,greybus-frame-time-
counter" in your platform code - then you can use 'get_cycles()' in
this manner - if not - then you need to take steps in your own new
platform to provide that same level of functionality. You could switch
to an MSM8996 or an MSM8998 declare this node and bob's your uncle.
OTOH declaring this node on x86 would be a bit pointless. You'd be
better off providing a timer on a PCI bar, and binding that into
greybus with some x86/x86-platform code...
>
> >
> > The objective is to read one of the free-running counters in
> > MSM8994,
> > clocked by the PMIC. The refclk provided by PMIC is distributed to
> > each
> > processor in the system.
> >
> > >
> > > ?Looking at the
> > > state of the tree [1] as of the final commit [2] in the greybus
> > > branch,
> > > my points still stand:
> > >
> > > * The "google,greybus-frame-time-counter" node is superfluous. It
> > > does
> > > ? not describe a particular device,
> > It describes a timer running @ 19.2MHz, clocked by PMIC refclk.
> ... which you assume is whatever backs get_cycles(), which you in
> practice assume is the architected timer. For which we *already* have
> a
> binding and driver.
It's a requirement rather than assumption. If you declare that node,
it's assumed the timer driving get_cycles() does what it says on the
greybus-frame-time-counter tin.
> > > ?and duplicates information we have ? elsewhere.
> > Can you give an example ?
> Trivially, the CNTFRQ register in the architected timer (which is
> common
> across MMIO/sysreg), which you can query with arch_timer_get_rate().
>
> Note that isn't guaranteed to match get_cycles() either. You need a
> better API to call.
In that case a DT entry makes sense I'd say.
> You should definitely interrogate the relevant driver, somehow.
Hrmm. TBH if we are ruling out arch_timer_get_rate() then I think a DT
entry (which BTW is greybus specific) is the more intelligent way
forward. The greybus platform implementer needs to understand the
dependencies and take action to meet those dependencies should he or
she wish to support this feature.
I'm not opposed necessarily to calling arch_timer_get_rate() instead of
a DT binding, assuming it works, and drawing a line under it for
MSM8994. As I've said it's up to a system architect for other platforms
to go and do the necessary design to support this feature and this will
almost certainly require new platform code both here and in other
places anyway.
>
> Without a higher-level view of what you're trying to achieve, it's
> not
> clear to me whether get_cycles() is the right interface.
I appreciate that.
---
bod
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-15 11:48 ` Bryan O'Donoghue
@ 2016-09-15 12:46 ` Mark Rutland
-1 siblings, 0 replies; 84+ messages in thread
From: Mark Rutland @ 2016-09-15 12:46 UTC (permalink / raw)
To: Bryan O'Donoghue
Cc: Greg KH, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Vaibhav Agarwal, Mark Greer, marc.zyngier, linux-arm-kernel
On Thu, Sep 15, 2016 at 12:48:08PM +0100, Bryan O'Donoghue wrote:
> On Thu, 2016-09-15 at 12:20 +0100, Mark Rutland wrote:
> > For example, you have absolutely no guarantee as to what backs
> > get_cycles(). Despite this, the code assumes that get_cycles() is
> > backed by something running at the frequency described in a
> > "google,greybus-frame-time-counter" node.
> >
> > Even if this *happens* to match what some piece of arch code provides
> > today on some platform, it is in no way *guaranteed*.
>
> That's the point though, if you declare "google,greybus-frame-time-
> counter" in your platform code - then you can use 'get_cycles()' in
> this manner
To be clear, *some* properties (and perhaps additional nodes) may need
to be in the DT, in order to capture the hardware property or
relationship that you are reliant upon.
However, the DT cannot possibly know anything about get_cycles(), as
get_cycles() is a kernel implementation details that's subject to
arbitrary change at any point in time, independent of the DT.
The "google,greybus-frame-time-counter" node, as it stands, does not
capture any relevant hardware detail, and does not belong in the DT.
> > Without a higher-level view of what you're trying to achieve, it's
> > not clear to me whether get_cycles() is the right interface.
>
> I appreciate that.
Until that's clarified, we won't make any progress here.
Thanks,
Mark.
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-15 12:46 ` Mark Rutland
0 siblings, 0 replies; 84+ messages in thread
From: Mark Rutland @ 2016-09-15 12:46 UTC (permalink / raw)
To: linux-arm-kernel
On Thu, Sep 15, 2016 at 12:48:08PM +0100, Bryan O'Donoghue wrote:
> On Thu, 2016-09-15 at 12:20 +0100, Mark Rutland wrote:
> > For example, you have absolutely no guarantee as to what backs
> > get_cycles(). Despite this, the code assumes that get_cycles() is
> > backed by something running at the frequency described in a
> > "google,greybus-frame-time-counter" node.
> >
> > Even if this *happens* to match what some piece of arch code provides
> > today on some platform, it is in no way *guaranteed*.
>
> That's the point though, if you declare "google,greybus-frame-time-
> counter" in your platform code - then you can use 'get_cycles()' in
> this manner
To be clear, *some* properties (and perhaps additional nodes) may need
to be in the DT, in order to capture the hardware property or
relationship that you are reliant upon.
However, the DT cannot possibly know anything about get_cycles(), as
get_cycles() is a kernel implementation details that's subject to
arbitrary change at any point in time, independent of the DT.
The "google,greybus-frame-time-counter" node, as it stands, does not
capture any relevant hardware detail, and does not belong in the DT.
> > Without a higher-level view of what you're trying to achieve, it's
> > not clear to me whether get_cycles() is the right interface.
>
> I appreciate that.
Until that's clarified, we won't make any progress here.
Thanks,
Mark.
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [patch 11/32] greybus: camera driver
[not found] ` <20160915122234.640367870@bubbles.kroah.org>
@ 2016-09-15 13:16 ` Laurent Pinchart
0 siblings, 0 replies; 84+ messages in thread
From: Laurent Pinchart @ 2016-09-15 13:16 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Sandeep Patil, Matt Porter, John Stultz, Rob Herring,
Viresh Kumar, Alex Elder, David Lin, Bryan O'Donoghue,
Vaibhav Agarwal, Mark Greer
Hi Greg,
Thank you for the patch.
On Thursday 15 Sep 2016 14:23:51 Greg Kroah-Hartman wrote:
> This driver implements the Greybus camera protocol.
>
> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> ---
> drivers/greybus/camera.c | 1400 ++++++++++++++++++++++++++++++++++++++++
> drivers/greybus/gb-camera.h | 127 +++
> 2 files changed, 1527 insertions(+)
>
> --- /dev/null
> +++ b/drivers/greybus/camera.c
> @@ -0,0 +1,1400 @@
[snip]
> +/* GB format to media code map */
> +static const struct gb_camera_fmt_info gb_fmt_info[] = {
> + {
> + .mbus_code = V4L2_MBUS_FMT_UYVY8_1X16,
> + .gb_format = 0x01,
> + .bpp = 16,
> + },
> + {
> + .mbus_code = V4L2_MBUS_FMT_NV12_1x8,
> + .gb_format = 0x12,
> + .bpp = 12,
> + },
> + {
> + .mbus_code = V4L2_MBUS_FMT_NV21_1x8,
> + .gb_format = 0x13,
> + .bpp = 12,
> + },
> + {
> + .mbus_code = V4L2_MBUS_FMT_YU12_1x8,
> + .gb_format = 0x16,
> + .bpp = 12,
> + },
> + {
> + .mbus_code = V4L2_MBUS_FMT_YV12_1x8,
> + .gb_format = 0x17,
> + .bpp = 12,
> + },
> + {
> + .mbus_code = V4L2_MBUS_FMT_JPEG_1X8,
> + .gb_format = 0x40,
> + .bpp = 0,
> + },
> + {
> + .mbus_code = V4L2_MBUS_FMT_GB_CAM_METADATA_1X8,
> + .gb_format = 0x41,
> + .bpp = 0,
> + },
> + {
> + .mbus_code = V4L2_MBUS_FMT_GB_CAM_DEBUG_DATA_1X8,
> + .gb_format = 0x42,
> + .bpp = 0,
> + },
These two formats will likely not be accepted in their current state in the
media subsystem, I would propose just dropping them.
> + {
> + .mbus_code = V4L2_MBUS_FMT_SBGGR10_1X10,
> + .gb_format = 0x80,
> + .bpp = 10,
> + },
> + {
> + .mbus_code = V4L2_MBUS_FMT_SGBRG10_1X10,
> + .gb_format = 0x81,
> + .bpp = 10,
> + },
> + {
> + .mbus_code = V4L2_MBUS_FMT_SGRBG10_1X10,
> + .gb_format = 0x82,
> + .bpp = 10,
> + },
> + {
> + .mbus_code = V4L2_MBUS_FMT_SRGGB10_1X10,
> + .gb_format = 0x83,
> + .bpp = 10,
> + },
> + {
> + .mbus_code = V4L2_MBUS_FMT_SBGGR12_1X12,
> + .gb_format = 0x84,
> + .bpp = 12,
> + },
> + {
> + .mbus_code = V4L2_MBUS_FMT_SGBRG12_1X12,
> + .gb_format = 0x85,
> + .bpp = 12,
> + },
> + {
> + .mbus_code = V4L2_MBUS_FMT_SGRBG12_1X12,
> + .gb_format = 0x86,
> + .bpp = 12,
> + },
> + {
> + .mbus_code = V4L2_MBUS_FMT_SRGGB12_1X12,
> + .gb_format = 0x87,
> + .bpp = 12,
> + },
> +};
[snip]
--
Regards,
Laurent Pinchart
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-14 10:09 [GIT PULL] Greybus driver subsystem for 4.9-rc1 Greg KH
2016-09-14 17:36 ` Mark Rutland
[not found] ` <20160915122141.650632149@bubbles.kroah.org>
@ 2016-09-15 14:45 ` Mark Brown
2016-09-16 6:05 ` Greg KH
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
3 siblings, 1 reply; 84+ messages in thread
From: Mark Brown @ 2016-09-15 14:45 UTC (permalink / raw)
To: Greg KH
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
[-- Attachment #1: Type: text/plain, Size: 1094 bytes --]
On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
> I'll send out a follow-up set of "simple" patches that just add the
> files to the kernel tree, to give people an idea of the code involved.
> Overall, it's a tiny stand-alone driver subsystem, only 37k lines, that
> implements a protocol which allows for "generic" cameras, audio devices,
> and other class type devices, as well as a bridged "physical" layer
> protocol to talk to serial, spi, uart, pwm, gpio, i2c, and even USB host
Just emphasizing what Mark Rutland said complete NACK, in particular the
drivers for functions have not been posted upstream at all (and it's
concerning that they're all being added under drivers/greybus rather
than within the relevant subsystem like we do normally). I've not
looked at the code as it has not been submitted but given that and that
the reason I found this pull request was an ASoC contributor who had
seen it and looked at the code going "oh dear, greybus..." on IRC I'm
very concerned.
Sending a pull request for code that's never been seen upstream seems
completely premature.
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 473 bytes --]
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-15 12:46 ` Mark Rutland
@ 2016-09-15 15:40 ` Bryan O'Donoghue
-1 siblings, 0 replies; 84+ messages in thread
From: Bryan O'Donoghue @ 2016-09-15 15:40 UTC (permalink / raw)
To: Mark Rutland
Cc: Greg KH, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Vaibhav Agarwal, Mark Greer, marc.zyngier, linux-arm-kernel
On Thu, 2016-09-15 at 13:46 +0100, Mark Rutland wrote:
> On Thu, Sep 15, 2016 at 12:48:08PM +0100, Bryan O'Donoghue wrote:
> >
> > On Thu, 2016-09-15 at 12:20 +0100, Mark Rutland wrote:
> > >
> > > For example, you have absolutely no guarantee as to what backs
> > > get_cycles(). Despite this, the code assumes that get_cycles() is
> > > backed by something running at the frequency described in a
> > > "google,greybus-frame-time-counter" node.
> > >
> > > Even if this *happens* to match what some piece of arch code
> > > provides
> > > today on some platform, it is in no way *guaranteed*.
> > That's the point though, if you declare "google,greybus-frame-time-
> > counter" in your platform code - then you can use 'get_cycles()' in
> > this manner
> To be clear, *some* properties (and perhaps additional nodes) may
> need
> to be in the DT, in order to capture the hardware property or
> relationship that you are reliant upon.
Sure but on the relevant platform we know
1. get_cycles() is derived from an architectured MMIO timer
2. What is clocking that timer
So any platform that declares that property must be aware of what its
doing.
For clarity this is the alternative to reading another register
directly bypassing get_cycles().
> >
> > >
> > > Without a higher-level view of what you're trying to achieve,
> > > it's
> > > not clear to me whether get_cycles() is the right interface.
> > I appreciate that.
> Until that's clarified, we won't make any progress here.
Let's see. We're synchronizing a set of distributed timers via a GPIO
pulse. Each processor on the system has at least one timer block
directly driven by the refclk@19.2MHz or a PLL driven by refclk@19.2MHz
FrameTime is a 64 bit free-running counter running @19.2MHz that is
used as a common source of reference for events. On the MSM side this
implies reading one of the timers driven by that refclk directly.
The choices on MSM8994 to access one of those timer blocks are
1. Read the register directly as is done in earlier patches or
2. Read get_cycles() as is done in later patches.
It's no more complex than that.
---
bod
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-15 15:40 ` Bryan O'Donoghue
0 siblings, 0 replies; 84+ messages in thread
From: Bryan O'Donoghue @ 2016-09-15 15:40 UTC (permalink / raw)
To: linux-arm-kernel
On Thu, 2016-09-15 at 13:46 +0100, Mark Rutland wrote:
> On Thu, Sep 15, 2016 at 12:48:08PM +0100, Bryan O'Donoghue wrote:
> >
> > On Thu, 2016-09-15 at 12:20 +0100, Mark Rutland wrote:
> > >
> > > For example, you have absolutely no guarantee as to what backs
> > > get_cycles(). Despite this, the code assumes that get_cycles() is
> > > backed by something running at the frequency described in a
> > > "google,greybus-frame-time-counter" node.
> > >
> > > Even if this *happens* to match what some piece of arch code
> > > provides
> > > today on some platform, it is in no way *guaranteed*.
> > That's the point though, if you declare "google,greybus-frame-time-
> > counter" in your platform code - then you can use 'get_cycles()' in
> > this manner
> To be clear, *some* properties (and perhaps additional nodes) may
> need
> to be in the DT, in order to capture the hardware property or
> relationship that you are reliant upon.
Sure but on the relevant platform we know?
1. get_cycles() is derived from an architectured MMIO timer
2. What is clocking that timer
So any platform that declares that property must be aware of what its
doing.
For clarity this is the alternative to reading another register
directly bypassing get_cycles().
> >
> > >
> > > Without a higher-level view of what you're trying to achieve,
> > > it's
> > > not clear to me whether get_cycles() is the right interface.
> > I appreciate that.
> Until that's clarified, we won't make any progress here.
Let's see. We're synchronizing a set of distributed timers via a GPIO
pulse. Each processor on the system has at least one timer block
directly driven by the refclk at 19.2MHz?or a PLL driven by refclk at 19.2MHz
FrameTime is a 64 bit free-running counter running @19.2MHz that is
used as a common source of reference for events. On the MSM side this
implies reading one of the timers driven by that refclk directly.
The choices on MSM8994 to access one of those timer blocks are
1. Read the register directly as is done in earlier patches or
2. Read get_cycles() as is done in later patches.
It's no more complex than that.
---
bod
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-15 15:40 ` Bryan O'Donoghue
@ 2016-09-15 15:47 ` Mark Rutland
-1 siblings, 0 replies; 84+ messages in thread
From: Mark Rutland @ 2016-09-15 15:47 UTC (permalink / raw)
To: Bryan O'Donoghue
Cc: Greg KH, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Vaibhav Agarwal, Mark Greer, marc.zyngier, linux-arm-kernel
On Thu, Sep 15, 2016 at 04:40:00PM +0100, Bryan O'Donoghue wrote:
> On Thu, 2016-09-15 at 13:46 +0100, Mark Rutland wrote:
> > On Thu, Sep 15, 2016 at 12:48:08PM +0100, Bryan O'Donoghue wrote:
> > > On Thu, 2016-09-15 at 12:20 +0100, Mark Rutland wrote:
> > > > For example, you have absolutely no guarantee as to what backs
> > > > get_cycles(). Despite this, the code assumes that get_cycles() is
> > > > backed by something running at the frequency described in a
> > > > "google,greybus-frame-time-counter" node.
> > > >
> > > > Even if this *happens* to match what some piece of arch code
> > > > provides
> > > > today on some platform, it is in no way *guaranteed*.
> > > That's the point though, if you declare "google,greybus-frame-time-
> > > counter" in your platform code - then you can use 'get_cycles()' in
> > > this manner
> > To be clear, *some* properties (and perhaps additional nodes) may
> > need
> > to be in the DT, in order to capture the hardware property or
> > relationship that you are reliant upon.
>
> Sure but on the relevant platform we know
>
> 1. get_cycles() is derived from an architectured MMIO timer
> 2. What is clocking that timer
>
> So any platform that declares that property must be aware of what its
> doing.
I can't say this any more explicitly:
****************************************************
* The DT *cannot* know anything about get_cycles() *
****************************************************
It's no more complex than that.
Thanks,
Mark.
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-15 15:47 ` Mark Rutland
0 siblings, 0 replies; 84+ messages in thread
From: Mark Rutland @ 2016-09-15 15:47 UTC (permalink / raw)
To: linux-arm-kernel
On Thu, Sep 15, 2016 at 04:40:00PM +0100, Bryan O'Donoghue wrote:
> On Thu, 2016-09-15 at 13:46 +0100, Mark Rutland wrote:
> > On Thu, Sep 15, 2016 at 12:48:08PM +0100, Bryan O'Donoghue wrote:
> > > On Thu, 2016-09-15 at 12:20 +0100, Mark Rutland wrote:
> > > > For example, you have absolutely no guarantee as to what backs
> > > > get_cycles(). Despite this, the code assumes that get_cycles() is
> > > > backed by something running at the frequency described in a
> > > > "google,greybus-frame-time-counter" node.
> > > >
> > > > Even if this *happens* to match what some piece of arch code
> > > > provides
> > > > today on some platform, it is in no way *guaranteed*.
> > > That's the point though, if you declare "google,greybus-frame-time-
> > > counter" in your platform code - then you can use 'get_cycles()' in
> > > this manner
> > To be clear, *some* properties (and perhaps additional nodes) may
> > need
> > to be in the DT, in order to capture the hardware property or
> > relationship that you are reliant upon.
>
> Sure but on the relevant platform we know?
>
> 1. get_cycles() is derived from an architectured MMIO timer
> 2. What is clocking that timer
>
> So any platform that declares that property must be aware of what its
> doing.
I can't say this any more explicitly:
****************************************************
* The DT *cannot* know anything about get_cycles() *
****************************************************
It's no more complex than that.
Thanks,
Mark.
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-15 15:47 ` Mark Rutland
@ 2016-09-15 16:09 ` Bryan O'Donoghue
-1 siblings, 0 replies; 84+ messages in thread
From: Bryan O'Donoghue @ 2016-09-15 16:09 UTC (permalink / raw)
To: Mark Rutland
Cc: Greg KH, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Vaibhav Agarwal, Mark Greer, marc.zyngier, linux-arm-kernel
On Thu, 2016-09-15 at 16:47 +0100, Mark Rutland wrote:
> On
> I can't say this any more explicitly:
>
> ****************************************************
> * The DT *cannot* know anything about get_cycles() *
> ****************************************************
>
> It's no more complex than that.
Sure think Mark I understand your point.
I think to satisfy your objection the most sensible change to make is
using arch_timer_get_rate() compared to a desired rate of 19200000.
---
bod
^ permalink raw reply [flat|nested] 84+ messages in thread
* [GIT PULL] Greybus driver subsystem for 4.9-rc1
@ 2016-09-15 16:09 ` Bryan O'Donoghue
0 siblings, 0 replies; 84+ messages in thread
From: Bryan O'Donoghue @ 2016-09-15 16:09 UTC (permalink / raw)
To: linux-arm-kernel
On Thu, 2016-09-15 at 16:47 +0100, Mark Rutland wrote:
> On?
> I can't say this any more explicitly:
>
> ****************************************************
> * The DT *cannot* know anything about get_cycles() *
> ****************************************************
>
> It's no more complex than that.
Sure think Mark I understand your point.
I think to satisfy your objection the most sensible change to make is
using arch_timer_get_rate() compared to a desired rate of 19200000.
---
bod
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-15 14:45 ` [GIT PULL] Greybus driver subsystem for 4.9-rc1 Mark Brown
@ 2016-09-16 6:05 ` Greg KH
2016-09-16 10:18 ` Mark Brown
` (2 more replies)
0 siblings, 3 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 6:05 UTC (permalink / raw)
To: Mark Brown
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
On Thu, Sep 15, 2016 at 03:45:53PM +0100, Mark Brown wrote:
> On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
>
> > I'll send out a follow-up set of "simple" patches that just add the
> > files to the kernel tree, to give people an idea of the code involved.
> > Overall, it's a tiny stand-alone driver subsystem, only 37k lines, that
> > implements a protocol which allows for "generic" cameras, audio devices,
> > and other class type devices, as well as a bridged "physical" layer
> > protocol to talk to serial, spi, uart, pwm, gpio, i2c, and even USB host
>
> Just emphasizing what Mark Rutland said complete NACK, in particular the
> drivers for functions have not been posted upstream at all (and it's
> concerning that they're all being added under drivers/greybus rather
> than within the relevant subsystem like we do normally). I've not
> looked at the code as it has not been submitted but given that and that
> the reason I found this pull request was an ASoC contributor who had
> seen it and looked at the code going "oh dear, greybus..." on IRC I'm
> very concerned.
>
> Sending a pull request for code that's never been seen upstream seems
> completely premature.
Hey, how does code get upstream then? :)
Anyway, as vger seems to be marking all of the greybus patches I send
out as "spam" according to a filter, it's making it a bit hard to send
patches out, but I'll try it again "by hand" later today...
As for the drivers all living under drivers/greybus/ I understand, but
we need the greybus core present first before we can get the drivers in.
How about we do what happened with IIO, we take the greybus core code in
drivers/greybus/ and put the drivers in staging, and then move them out
of staging into the "real" portion of the kernel as they get reviewed
and accepted?
Any objections to that workflow?
thanks,
greg k-h
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 00/32] Greybus driver subsystem
2016-09-14 10:09 [GIT PULL] Greybus driver subsystem for 4.9-rc1 Greg KH
` (2 preceding siblings ...)
2016-09-15 14:45 ` [GIT PULL] Greybus driver subsystem for 4.9-rc1 Mark Brown
@ 2016-09-16 6:40 ` Greg KH
2016-09-16 6:41 ` [patch 02/32] greybus: interface control logic Greg KH
` (30 more replies)
3 siblings, 31 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 6:40 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This is a simple set of patches, adding the greybus core and drivers to
the kernel tree.
It breaks the larger git tree submission up into individual files to
make it easier for reviewing.
Any questions or comments, please let us know.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 02/32] greybus: interface control logic
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
@ 2016-09-16 6:41 ` Greg KH
2016-09-16 13:22 ` [patch 03/32] greybus: operations logic Greg KH
` (29 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 6:41 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This is the Greybus interface control code. It handles the interactions
between the Greybus core and the Greybus interface devices in the
system.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/interface.c | 1316 ++++++++++++++++++++++++++++++++++++++++++++
drivers/greybus/interface.h | 88 ++
2 files changed, 1404 insertions(+)
--- /dev/null
+++ b/drivers/greybus/interface.c
@@ -0,0 +1,1316 @@
+/*
+ * Greybus interface code
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/delay.h>
+
+#include "greybus.h"
+#include "greybus_trace.h"
+
+#define GB_INTERFACE_MODE_SWITCH_TIMEOUT 2000
+
+#define GB_INTERFACE_DEVICE_ID_BAD 0xff
+
+#define GB_INTERFACE_AUTOSUSPEND_MS 3000
+
+/* Time required for interface to enter standby before disabling REFCLK */
+#define GB_INTERFACE_SUSPEND_HIBERNATE_DELAY_MS 20
+
+/* Don't-care selector index */
+#define DME_SELECTOR_INDEX_NULL 0
+
+/* DME attributes */
+/* FIXME: remove ES2 support and DME_T_TST_SRC_INCREMENT */
+#define DME_T_TST_SRC_INCREMENT 0x4083
+
+#define DME_DDBL1_MANUFACTURERID 0x5003
+#define DME_DDBL1_PRODUCTID 0x5004
+
+#define DME_TOSHIBA_GMP_VID 0x6000
+#define DME_TOSHIBA_GMP_PID 0x6001
+#define DME_TOSHIBA_GMP_SN0 0x6002
+#define DME_TOSHIBA_GMP_SN1 0x6003
+#define DME_TOSHIBA_GMP_INIT_STATUS 0x6101
+
+/* DDBL1 Manufacturer and Product ids */
+#define TOSHIBA_DMID 0x0126
+#define TOSHIBA_ES2_BRIDGE_DPID 0x1000
+#define TOSHIBA_ES3_APBRIDGE_DPID 0x1001
+#define TOSHIBA_ES3_GBPHY_DPID 0x1002
+
+static int gb_interface_hibernate_link(struct gb_interface *intf);
+static int gb_interface_refclk_set(struct gb_interface *intf, bool enable);
+
+static int gb_interface_dme_attr_get(struct gb_interface *intf,
+ u16 attr, u32 *val)
+{
+ return gb_svc_dme_peer_get(intf->hd->svc, intf->interface_id,
+ attr, DME_SELECTOR_INDEX_NULL, val);
+}
+
+static int gb_interface_read_ara_dme(struct gb_interface *intf)
+{
+ u32 sn0, sn1;
+ int ret;
+
+ /*
+ * Unless this is a Toshiba bridge, bail out until we have defined
+ * standard GMP attributes.
+ */
+ if (intf->ddbl1_manufacturer_id != TOSHIBA_DMID) {
+ dev_err(&intf->dev, "unknown manufacturer %08x\n",
+ intf->ddbl1_manufacturer_id);
+ return -ENODEV;
+ }
+
+ ret = gb_interface_dme_attr_get(intf, DME_TOSHIBA_GMP_VID,
+ &intf->vendor_id);
+ if (ret)
+ return ret;
+
+ ret = gb_interface_dme_attr_get(intf, DME_TOSHIBA_GMP_PID,
+ &intf->product_id);
+ if (ret)
+ return ret;
+
+ ret = gb_interface_dme_attr_get(intf, DME_TOSHIBA_GMP_SN0, &sn0);
+ if (ret)
+ return ret;
+
+ ret = gb_interface_dme_attr_get(intf, DME_TOSHIBA_GMP_SN1, &sn1);
+ if (ret)
+ return ret;
+
+ intf->serial_number = (u64)sn1 << 32 | sn0;
+
+ return 0;
+}
+
+static int gb_interface_read_dme(struct gb_interface *intf)
+{
+ int ret;
+
+ /* DME attributes have already been read */
+ if (intf->dme_read)
+ return 0;
+
+ ret = gb_interface_dme_attr_get(intf, DME_DDBL1_MANUFACTURERID,
+ &intf->ddbl1_manufacturer_id);
+ if (ret)
+ return ret;
+
+ ret = gb_interface_dme_attr_get(intf, DME_DDBL1_PRODUCTID,
+ &intf->ddbl1_product_id);
+ if (ret)
+ return ret;
+
+ if (intf->ddbl1_manufacturer_id == TOSHIBA_DMID &&
+ intf->ddbl1_product_id == TOSHIBA_ES2_BRIDGE_DPID) {
+ intf->quirks |= GB_INTERFACE_QUIRK_NO_GMP_IDS;
+ intf->quirks |= GB_INTERFACE_QUIRK_NO_INIT_STATUS;
+ }
+
+ ret = gb_interface_read_ara_dme(intf);
+ if (ret)
+ return ret;
+
+ intf->dme_read = true;
+
+ return 0;
+}
+
+static int gb_interface_route_create(struct gb_interface *intf)
+{
+ struct gb_svc *svc = intf->hd->svc;
+ u8 intf_id = intf->interface_id;
+ u8 device_id;
+ int ret;
+
+ /* Allocate an interface device id. */
+ ret = ida_simple_get(&svc->device_id_map,
+ GB_SVC_DEVICE_ID_MIN, GB_SVC_DEVICE_ID_MAX + 1,
+ GFP_KERNEL);
+ if (ret < 0) {
+ dev_err(&intf->dev, "failed to allocate device id: %d\n", ret);
+ return ret;
+ }
+ device_id = ret;
+
+ ret = gb_svc_intf_device_id(svc, intf_id, device_id);
+ if (ret) {
+ dev_err(&intf->dev, "failed to set device id %u: %d\n",
+ device_id, ret);
+ goto err_ida_remove;
+ }
+
+ /* FIXME: Hard-coded AP device id. */
+ ret = gb_svc_route_create(svc, svc->ap_intf_id, GB_SVC_DEVICE_ID_AP,
+ intf_id, device_id);
+ if (ret) {
+ dev_err(&intf->dev, "failed to create route: %d\n", ret);
+ goto err_svc_id_free;
+ }
+
+ intf->device_id = device_id;
+
+ return 0;
+
+err_svc_id_free:
+ /*
+ * XXX Should we tell SVC that this id doesn't belong to interface
+ * XXX anymore.
+ */
+err_ida_remove:
+ ida_simple_remove(&svc->device_id_map, device_id);
+
+ return ret;
+}
+
+static void gb_interface_route_destroy(struct gb_interface *intf)
+{
+ struct gb_svc *svc = intf->hd->svc;
+
+ if (intf->device_id == GB_INTERFACE_DEVICE_ID_BAD)
+ return;
+
+ gb_svc_route_destroy(svc, svc->ap_intf_id, intf->interface_id);
+ ida_simple_remove(&svc->device_id_map, intf->device_id);
+ intf->device_id = GB_INTERFACE_DEVICE_ID_BAD;
+}
+
+/* Locking: Caller holds the interface mutex. */
+static int gb_interface_legacy_mode_switch(struct gb_interface *intf)
+{
+ int ret;
+
+ dev_info(&intf->dev, "legacy mode switch detected\n");
+
+ /* Mark as disconnected to prevent I/O during disable. */
+ intf->disconnected = true;
+ gb_interface_disable(intf);
+ intf->disconnected = false;
+
+ ret = gb_interface_enable(intf);
+ if (ret) {
+ dev_err(&intf->dev, "failed to re-enable interface: %d\n", ret);
+ gb_interface_deactivate(intf);
+ }
+
+ return ret;
+}
+
+void gb_interface_mailbox_event(struct gb_interface *intf, u16 result,
+ u32 mailbox)
+{
+ mutex_lock(&intf->mutex);
+
+ if (result) {
+ dev_warn(&intf->dev,
+ "mailbox event with UniPro error: 0x%04x\n",
+ result);
+ goto err_disable;
+ }
+
+ if (mailbox != GB_SVC_INTF_MAILBOX_GREYBUS) {
+ dev_warn(&intf->dev,
+ "mailbox event with unexpected value: 0x%08x\n",
+ mailbox);
+ goto err_disable;
+ }
+
+ if (intf->quirks & GB_INTERFACE_QUIRK_LEGACY_MODE_SWITCH) {
+ gb_interface_legacy_mode_switch(intf);
+ goto out_unlock;
+ }
+
+ if (!intf->mode_switch) {
+ dev_warn(&intf->dev, "unexpected mailbox event: 0x%08x\n",
+ mailbox);
+ goto err_disable;
+ }
+
+ dev_info(&intf->dev, "mode switch detected\n");
+
+ complete(&intf->mode_switch_completion);
+
+out_unlock:
+ mutex_unlock(&intf->mutex);
+
+ return;
+
+err_disable:
+ gb_interface_disable(intf);
+ gb_interface_deactivate(intf);
+ mutex_unlock(&intf->mutex);
+}
+
+static void gb_interface_mode_switch_work(struct work_struct *work)
+{
+ struct gb_interface *intf;
+ struct gb_control *control;
+ unsigned long timeout;
+ int ret;
+
+ intf = container_of(work, struct gb_interface, mode_switch_work);
+
+ mutex_lock(&intf->mutex);
+ /* Make sure interface is still enabled. */
+ if (!intf->enabled) {
+ dev_dbg(&intf->dev, "mode switch aborted\n");
+ intf->mode_switch = false;
+ mutex_unlock(&intf->mutex);
+ goto out_interface_put;
+ }
+
+ /*
+ * Prepare the control device for mode switch and make sure to get an
+ * extra reference before it goes away during interface disable.
+ */
+ control = gb_control_get(intf->control);
+ gb_control_mode_switch_prepare(control);
+ gb_interface_disable(intf);
+ mutex_unlock(&intf->mutex);
+
+ timeout = msecs_to_jiffies(GB_INTERFACE_MODE_SWITCH_TIMEOUT);
+ ret = wait_for_completion_interruptible_timeout(
+ &intf->mode_switch_completion, timeout);
+
+ /* Finalise control-connection mode switch. */
+ gb_control_mode_switch_complete(control);
+ gb_control_put(control);
+
+ if (ret < 0) {
+ dev_err(&intf->dev, "mode switch interrupted\n");
+ goto err_deactivate;
+ } else if (ret == 0) {
+ dev_err(&intf->dev, "mode switch timed out\n");
+ goto err_deactivate;
+ }
+
+ /* Re-enable (re-enumerate) interface if still active. */
+ mutex_lock(&intf->mutex);
+ intf->mode_switch = false;
+ if (intf->active) {
+ ret = gb_interface_enable(intf);
+ if (ret) {
+ dev_err(&intf->dev, "failed to re-enable interface: %d\n",
+ ret);
+ gb_interface_deactivate(intf);
+ }
+ }
+ mutex_unlock(&intf->mutex);
+
+out_interface_put:
+ gb_interface_put(intf);
+
+ return;
+
+err_deactivate:
+ mutex_lock(&intf->mutex);
+ intf->mode_switch = false;
+ gb_interface_deactivate(intf);
+ mutex_unlock(&intf->mutex);
+
+ gb_interface_put(intf);
+}
+
+int gb_interface_request_mode_switch(struct gb_interface *intf)
+{
+ int ret = 0;
+
+ mutex_lock(&intf->mutex);
+ if (intf->mode_switch) {
+ ret = -EBUSY;
+ goto out_unlock;
+ }
+
+ intf->mode_switch = true;
+ reinit_completion(&intf->mode_switch_completion);
+
+ /*
+ * Get a reference to the interface device, which will be put once the
+ * mode switch is complete.
+ */
+ get_device(&intf->dev);
+
+ if (!queue_work(system_long_wq, &intf->mode_switch_work)) {
+ put_device(&intf->dev);
+ ret = -EBUSY;
+ goto out_unlock;
+ }
+
+out_unlock:
+ mutex_unlock(&intf->mutex);
+
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_interface_request_mode_switch);
+
+/*
+ * T_TstSrcIncrement is written by the module on ES2 as a stand-in for the
+ * init-status attribute DME_TOSHIBA_INIT_STATUS. The AP needs to read and
+ * clear it after reading a non-zero value from it.
+ *
+ * FIXME: This is module-hardware dependent and needs to be extended for every
+ * type of module we want to support.
+ */
+static int gb_interface_read_and_clear_init_status(struct gb_interface *intf)
+{
+ struct gb_host_device *hd = intf->hd;
+ unsigned long bootrom_quirks;
+ unsigned long s2l_quirks;
+ int ret;
+ u32 value;
+ u16 attr;
+ u8 init_status;
+
+ /*
+ * ES2 bridges use T_TstSrcIncrement for the init status.
+ *
+ * FIXME: Remove ES2 support
+ */
+ if (intf->quirks & GB_INTERFACE_QUIRK_NO_INIT_STATUS)
+ attr = DME_T_TST_SRC_INCREMENT;
+ else
+ attr = DME_TOSHIBA_GMP_INIT_STATUS;
+
+ ret = gb_svc_dme_peer_get(hd->svc, intf->interface_id, attr,
+ DME_SELECTOR_INDEX_NULL, &value);
+ if (ret)
+ return ret;
+
+ /*
+ * A nonzero init status indicates the module has finished
+ * initializing.
+ */
+ if (!value) {
+ dev_err(&intf->dev, "invalid init status\n");
+ return -ENODEV;
+ }
+
+ /*
+ * Extract the init status.
+ *
+ * For ES2: We need to check lowest 8 bits of 'value'.
+ * For ES3: We need to check highest 8 bits out of 32 of 'value'.
+ *
+ * FIXME: Remove ES2 support
+ */
+ if (intf->quirks & GB_INTERFACE_QUIRK_NO_INIT_STATUS)
+ init_status = value & 0xff;
+ else
+ init_status = value >> 24;
+
+ /*
+ * Check if the interface is executing the quirky ES3 bootrom that,
+ * for example, requires E2EFC, CSD and CSV to be disabled.
+ */
+ bootrom_quirks = GB_INTERFACE_QUIRK_NO_CPORT_FEATURES |
+ GB_INTERFACE_QUIRK_FORCED_DISABLE |
+ GB_INTERFACE_QUIRK_LEGACY_MODE_SWITCH |
+ GB_INTERFACE_QUIRK_NO_BUNDLE_ACTIVATE;
+
+ s2l_quirks = GB_INTERFACE_QUIRK_NO_PM;
+
+ switch (init_status) {
+ case GB_INIT_BOOTROM_UNIPRO_BOOT_STARTED:
+ case GB_INIT_BOOTROM_FALLBACK_UNIPRO_BOOT_STARTED:
+ intf->quirks |= bootrom_quirks;
+ break;
+ case GB_INIT_S2_LOADER_BOOT_STARTED:
+ /* S2 Loader doesn't support runtime PM */
+ intf->quirks &= ~bootrom_quirks;
+ intf->quirks |= s2l_quirks;
+ break;
+ default:
+ intf->quirks &= ~bootrom_quirks;
+ intf->quirks &= ~s2l_quirks;
+ }
+
+ /* Clear the init status. */
+ return gb_svc_dme_peer_set(hd->svc, intf->interface_id, attr,
+ DME_SELECTOR_INDEX_NULL, 0);
+}
+
+/* interface sysfs attributes */
+#define gb_interface_attr(field, type) \
+static ssize_t field##_show(struct device *dev, \
+ struct device_attribute *attr, \
+ char *buf) \
+{ \
+ struct gb_interface *intf = to_gb_interface(dev); \
+ return scnprintf(buf, PAGE_SIZE, type"\n", intf->field); \
+} \
+static DEVICE_ATTR_RO(field)
+
+gb_interface_attr(ddbl1_manufacturer_id, "0x%08x");
+gb_interface_attr(ddbl1_product_id, "0x%08x");
+gb_interface_attr(interface_id, "%u");
+gb_interface_attr(vendor_id, "0x%08x");
+gb_interface_attr(product_id, "0x%08x");
+gb_interface_attr(serial_number, "0x%016llx");
+
+static ssize_t voltage_now_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_interface *intf = to_gb_interface(dev);
+ int ret;
+ u32 measurement;
+
+ ret = gb_svc_pwrmon_intf_sample_get(intf->hd->svc, intf->interface_id,
+ GB_SVC_PWRMON_TYPE_VOL,
+ &measurement);
+ if (ret) {
+ dev_err(&intf->dev, "failed to get voltage sample (%d)\n", ret);
+ return ret;
+ }
+
+ return sprintf(buf, "%u\n", measurement);
+}
+static DEVICE_ATTR_RO(voltage_now);
+
+static ssize_t current_now_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_interface *intf = to_gb_interface(dev);
+ int ret;
+ u32 measurement;
+
+ ret = gb_svc_pwrmon_intf_sample_get(intf->hd->svc, intf->interface_id,
+ GB_SVC_PWRMON_TYPE_CURR,
+ &measurement);
+ if (ret) {
+ dev_err(&intf->dev, "failed to get current sample (%d)\n", ret);
+ return ret;
+ }
+
+ return sprintf(buf, "%u\n", measurement);
+}
+static DEVICE_ATTR_RO(current_now);
+
+static ssize_t power_now_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_interface *intf = to_gb_interface(dev);
+ int ret;
+ u32 measurement;
+
+ ret = gb_svc_pwrmon_intf_sample_get(intf->hd->svc, intf->interface_id,
+ GB_SVC_PWRMON_TYPE_PWR,
+ &measurement);
+ if (ret) {
+ dev_err(&intf->dev, "failed to get power sample (%d)\n", ret);
+ return ret;
+ }
+
+ return sprintf(buf, "%u\n", measurement);
+}
+static DEVICE_ATTR_RO(power_now);
+
+static ssize_t power_state_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_interface *intf = to_gb_interface(dev);
+
+ if (intf->active)
+ return scnprintf(buf, PAGE_SIZE, "on\n");
+ else
+ return scnprintf(buf, PAGE_SIZE, "off\n");
+}
+
+static ssize_t power_state_store(struct device *dev,
+ struct device_attribute *attr, const char *buf,
+ size_t len)
+{
+ struct gb_interface *intf = to_gb_interface(dev);
+ bool activate;
+ int ret = 0;
+
+ if (kstrtobool(buf, &activate))
+ return -EINVAL;
+
+ mutex_lock(&intf->mutex);
+
+ if (activate == intf->active)
+ goto unlock;
+
+ if (activate) {
+ ret = gb_interface_activate(intf);
+ if (ret) {
+ dev_err(&intf->dev,
+ "failed to activate interface: %d\n", ret);
+ goto unlock;
+ }
+
+ ret = gb_interface_enable(intf);
+ if (ret) {
+ dev_err(&intf->dev,
+ "failed to enable interface: %d\n", ret);
+ gb_interface_deactivate(intf);
+ goto unlock;
+ }
+ } else {
+ gb_interface_disable(intf);
+ gb_interface_deactivate(intf);
+ }
+
+unlock:
+ mutex_unlock(&intf->mutex);
+
+ if (ret)
+ return ret;
+
+ return len;
+}
+static DEVICE_ATTR_RW(power_state);
+
+static const char *gb_interface_type_string(struct gb_interface *intf)
+{
+ static const char * const types[] = {
+ [GB_INTERFACE_TYPE_INVALID] = "invalid",
+ [GB_INTERFACE_TYPE_UNKNOWN] = "unknown",
+ [GB_INTERFACE_TYPE_DUMMY] = "dummy",
+ [GB_INTERFACE_TYPE_UNIPRO] = "unipro",
+ [GB_INTERFACE_TYPE_GREYBUS] = "greybus",
+ };
+
+ return types[intf->type];
+}
+
+static ssize_t interface_type_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_interface *intf = to_gb_interface(dev);
+
+ return sprintf(buf, "%s\n", gb_interface_type_string(intf));
+}
+static DEVICE_ATTR_RO(interface_type);
+
+static struct attribute *interface_unipro_attrs[] = {
+ &dev_attr_ddbl1_manufacturer_id.attr,
+ &dev_attr_ddbl1_product_id.attr,
+ NULL
+};
+
+static struct attribute *interface_greybus_attrs[] = {
+ &dev_attr_vendor_id.attr,
+ &dev_attr_product_id.attr,
+ &dev_attr_serial_number.attr,
+ NULL
+};
+
+static struct attribute *interface_power_attrs[] = {
+ &dev_attr_voltage_now.attr,
+ &dev_attr_current_now.attr,
+ &dev_attr_power_now.attr,
+ &dev_attr_power_state.attr,
+ NULL
+};
+
+static struct attribute *interface_common_attrs[] = {
+ &dev_attr_interface_id.attr,
+ &dev_attr_interface_type.attr,
+ NULL
+};
+
+static umode_t interface_unipro_is_visible(struct kobject *kobj,
+ struct attribute *attr, int n)
+{
+ struct device *dev = container_of(kobj, struct device, kobj);
+ struct gb_interface *intf = to_gb_interface(dev);
+
+ switch (intf->type) {
+ case GB_INTERFACE_TYPE_UNIPRO:
+ case GB_INTERFACE_TYPE_GREYBUS:
+ return attr->mode;
+ default:
+ return 0;
+ }
+}
+
+static umode_t interface_greybus_is_visible(struct kobject *kobj,
+ struct attribute *attr, int n)
+{
+ struct device *dev = container_of(kobj, struct device, kobj);
+ struct gb_interface *intf = to_gb_interface(dev);
+
+ switch (intf->type) {
+ case GB_INTERFACE_TYPE_GREYBUS:
+ return attr->mode;
+ default:
+ return 0;
+ }
+}
+
+static umode_t interface_power_is_visible(struct kobject *kobj,
+ struct attribute *attr, int n)
+{
+ struct device *dev = container_of(kobj, struct device, kobj);
+ struct gb_interface *intf = to_gb_interface(dev);
+
+ switch (intf->type) {
+ case GB_INTERFACE_TYPE_UNIPRO:
+ case GB_INTERFACE_TYPE_GREYBUS:
+ return attr->mode;
+ default:
+ return 0;
+ }
+}
+
+static const struct attribute_group interface_unipro_group = {
+ .is_visible = interface_unipro_is_visible,
+ .attrs = interface_unipro_attrs,
+};
+
+static const struct attribute_group interface_greybus_group = {
+ .is_visible = interface_greybus_is_visible,
+ .attrs = interface_greybus_attrs,
+};
+
+static const struct attribute_group interface_power_group = {
+ .is_visible = interface_power_is_visible,
+ .attrs = interface_power_attrs,
+};
+
+static const struct attribute_group interface_common_group = {
+ .attrs = interface_common_attrs,
+};
+
+static const struct attribute_group *interface_groups[] = {
+ &interface_unipro_group,
+ &interface_greybus_group,
+ &interface_power_group,
+ &interface_common_group,
+ NULL
+};
+
+static void gb_interface_release(struct device *dev)
+{
+ struct gb_interface *intf = to_gb_interface(dev);
+
+ trace_gb_interface_release(intf);
+
+ kfree(intf);
+}
+
+#ifdef CONFIG_PM
+static int gb_interface_suspend(struct device *dev)
+{
+ struct gb_interface *intf = to_gb_interface(dev);
+ int ret, timesync_ret;
+
+ ret = gb_control_interface_suspend_prepare(intf->control);
+ if (ret)
+ return ret;
+
+ gb_timesync_interface_remove(intf);
+
+ ret = gb_control_suspend(intf->control);
+ if (ret)
+ goto err_hibernate_abort;
+
+ ret = gb_interface_hibernate_link(intf);
+ if (ret)
+ return ret;
+
+ /* Delay to allow interface to enter standby before disabling refclk */
+ msleep(GB_INTERFACE_SUSPEND_HIBERNATE_DELAY_MS);
+
+ ret = gb_interface_refclk_set(intf, false);
+ if (ret)
+ return ret;
+
+ return 0;
+
+err_hibernate_abort:
+ gb_control_interface_hibernate_abort(intf->control);
+
+ timesync_ret = gb_timesync_interface_add(intf);
+ if (timesync_ret) {
+ dev_err(dev, "failed to add to timesync: %d\n", timesync_ret);
+ return timesync_ret;
+ }
+
+ return ret;
+}
+
+static int gb_interface_resume(struct device *dev)
+{
+ struct gb_interface *intf = to_gb_interface(dev);
+ struct gb_svc *svc = intf->hd->svc;
+ int ret;
+
+ ret = gb_interface_refclk_set(intf, true);
+ if (ret)
+ return ret;
+
+ ret = gb_svc_intf_resume(svc, intf->interface_id);
+ if (ret)
+ return ret;
+
+ ret = gb_control_resume(intf->control);
+ if (ret)
+ return ret;
+
+ ret = gb_timesync_interface_add(intf);
+ if (ret) {
+ dev_err(dev, "failed to add to timesync: %d\n", ret);
+ return ret;
+ }
+
+ ret = gb_timesync_schedule_synchronous(intf);
+ if (ret) {
+ dev_err(dev, "failed to synchronize FrameTime: %d\n", ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int gb_interface_runtime_idle(struct device *dev)
+{
+ pm_runtime_mark_last_busy(dev);
+ pm_request_autosuspend(dev);
+
+ return 0;
+}
+#endif
+
+static const struct dev_pm_ops gb_interface_pm_ops = {
+ SET_RUNTIME_PM_OPS(gb_interface_suspend, gb_interface_resume,
+ gb_interface_runtime_idle)
+};
+
+struct device_type greybus_interface_type = {
+ .name = "greybus_interface",
+ .release = gb_interface_release,
+ .pm = &gb_interface_pm_ops,
+};
+
+/*
+ * A Greybus module represents a user-replaceable component on a GMP
+ * phone. An interface is the physical connection on that module. A
+ * module may have more than one interface.
+ *
+ * Create a gb_interface structure to represent a discovered interface.
+ * The position of interface within the Endo is encoded in "interface_id"
+ * argument.
+ *
+ * Returns a pointer to the new interfce or a null pointer if a
+ * failure occurs due to memory exhaustion.
+ */
+struct gb_interface *gb_interface_create(struct gb_module *module,
+ u8 interface_id)
+{
+ struct gb_host_device *hd = module->hd;
+ struct gb_interface *intf;
+
+ intf = kzalloc(sizeof(*intf), GFP_KERNEL);
+ if (!intf)
+ return NULL;
+
+ intf->hd = hd; /* XXX refcount? */
+ intf->module = module;
+ intf->interface_id = interface_id;
+ INIT_LIST_HEAD(&intf->bundles);
+ INIT_LIST_HEAD(&intf->manifest_descs);
+ mutex_init(&intf->mutex);
+ INIT_WORK(&intf->mode_switch_work, gb_interface_mode_switch_work);
+ init_completion(&intf->mode_switch_completion);
+
+ /* Invalid device id to start with */
+ intf->device_id = GB_INTERFACE_DEVICE_ID_BAD;
+
+ intf->dev.parent = &module->dev;
+ intf->dev.bus = &greybus_bus_type;
+ intf->dev.type = &greybus_interface_type;
+ intf->dev.groups = interface_groups;
+ intf->dev.dma_mask = module->dev.dma_mask;
+ device_initialize(&intf->dev);
+ dev_set_name(&intf->dev, "%s.%u", dev_name(&module->dev),
+ interface_id);
+
+ pm_runtime_set_autosuspend_delay(&intf->dev,
+ GB_INTERFACE_AUTOSUSPEND_MS);
+
+ trace_gb_interface_create(intf);
+
+ return intf;
+}
+
+static int gb_interface_vsys_set(struct gb_interface *intf, bool enable)
+{
+ struct gb_svc *svc = intf->hd->svc;
+ int ret;
+
+ dev_dbg(&intf->dev, "%s - %d\n", __func__, enable);
+
+ ret = gb_svc_intf_vsys_set(svc, intf->interface_id, enable);
+ if (ret) {
+ dev_err(&intf->dev, "failed to set v_sys: %d\n", ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int gb_interface_refclk_set(struct gb_interface *intf, bool enable)
+{
+ struct gb_svc *svc = intf->hd->svc;
+ int ret;
+
+ dev_dbg(&intf->dev, "%s - %d\n", __func__, enable);
+
+ ret = gb_svc_intf_refclk_set(svc, intf->interface_id, enable);
+ if (ret) {
+ dev_err(&intf->dev, "failed to set refclk: %d\n", ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int gb_interface_unipro_set(struct gb_interface *intf, bool enable)
+{
+ struct gb_svc *svc = intf->hd->svc;
+ int ret;
+
+ dev_dbg(&intf->dev, "%s - %d\n", __func__, enable);
+
+ ret = gb_svc_intf_unipro_set(svc, intf->interface_id, enable);
+ if (ret) {
+ dev_err(&intf->dev, "failed to set UniPro: %d\n", ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int gb_interface_activate_operation(struct gb_interface *intf,
+ enum gb_interface_type *intf_type)
+{
+ struct gb_svc *svc = intf->hd->svc;
+ u8 type;
+ int ret;
+
+ dev_dbg(&intf->dev, "%s\n", __func__);
+
+ ret = gb_svc_intf_activate(svc, intf->interface_id, &type);
+ if (ret) {
+ dev_err(&intf->dev, "failed to activate: %d\n", ret);
+ return ret;
+ }
+
+ switch (type) {
+ case GB_SVC_INTF_TYPE_DUMMY:
+ *intf_type = GB_INTERFACE_TYPE_DUMMY;
+ /* FIXME: handle as an error for now */
+ return -ENODEV;
+ case GB_SVC_INTF_TYPE_UNIPRO:
+ *intf_type = GB_INTERFACE_TYPE_UNIPRO;
+ dev_err(&intf->dev, "interface type UniPro not supported\n");
+ /* FIXME: handle as an error for now */
+ return -ENODEV;
+ case GB_SVC_INTF_TYPE_GREYBUS:
+ *intf_type = GB_INTERFACE_TYPE_GREYBUS;
+ break;
+ default:
+ dev_err(&intf->dev, "unknown interface type: %u\n", type);
+ *intf_type = GB_INTERFACE_TYPE_UNKNOWN;
+ return -ENODEV;
+ }
+
+ return 0;
+}
+
+static int gb_interface_hibernate_link(struct gb_interface *intf)
+{
+ struct gb_svc *svc = intf->hd->svc;
+
+ return gb_svc_intf_set_power_mode_hibernate(svc, intf->interface_id);
+}
+
+static int _gb_interface_activate(struct gb_interface *intf,
+ enum gb_interface_type *type)
+{
+ int ret;
+
+ *type = GB_INTERFACE_TYPE_UNKNOWN;
+
+ if (intf->ejected || intf->removed)
+ return -ENODEV;
+
+ ret = gb_interface_vsys_set(intf, true);
+ if (ret)
+ return ret;
+
+ ret = gb_interface_refclk_set(intf, true);
+ if (ret)
+ goto err_vsys_disable;
+
+ ret = gb_interface_unipro_set(intf, true);
+ if (ret)
+ goto err_refclk_disable;
+
+ ret = gb_interface_activate_operation(intf, type);
+ if (ret) {
+ switch (*type) {
+ case GB_INTERFACE_TYPE_UNIPRO:
+ case GB_INTERFACE_TYPE_GREYBUS:
+ goto err_hibernate_link;
+ default:
+ goto err_unipro_disable;
+ }
+ }
+
+ ret = gb_interface_read_dme(intf);
+ if (ret)
+ goto err_hibernate_link;
+
+ ret = gb_interface_route_create(intf);
+ if (ret)
+ goto err_hibernate_link;
+
+ intf->active = true;
+
+ trace_gb_interface_activate(intf);
+
+ return 0;
+
+err_hibernate_link:
+ gb_interface_hibernate_link(intf);
+err_unipro_disable:
+ gb_interface_unipro_set(intf, false);
+err_refclk_disable:
+ gb_interface_refclk_set(intf, false);
+err_vsys_disable:
+ gb_interface_vsys_set(intf, false);
+
+ return ret;
+}
+
+/*
+ * At present, we assume a UniPro-only module to be a Greybus module that
+ * failed to send its mailbox poke. There is some reason to believe that this
+ * is because of a bug in the ES3 bootrom.
+ *
+ * FIXME: Check if this is a Toshiba bridge before retrying?
+ */
+static int _gb_interface_activate_es3_hack(struct gb_interface *intf,
+ enum gb_interface_type *type)
+{
+ int retries = 3;
+ int ret;
+
+ while (retries--) {
+ ret = _gb_interface_activate(intf, type);
+ if (ret == -ENODEV && *type == GB_INTERFACE_TYPE_UNIPRO)
+ continue;
+
+ break;
+ }
+
+ return ret;
+}
+
+/*
+ * Activate an interface.
+ *
+ * Locking: Caller holds the interface mutex.
+ */
+int gb_interface_activate(struct gb_interface *intf)
+{
+ enum gb_interface_type type;
+ int ret;
+
+ switch (intf->type) {
+ case GB_INTERFACE_TYPE_INVALID:
+ case GB_INTERFACE_TYPE_GREYBUS:
+ ret = _gb_interface_activate_es3_hack(intf, &type);
+ break;
+ default:
+ ret = _gb_interface_activate(intf, &type);
+ }
+
+ /* Make sure type is detected correctly during reactivation. */
+ if (intf->type != GB_INTERFACE_TYPE_INVALID) {
+ if (type != intf->type) {
+ dev_err(&intf->dev, "failed to detect interface type\n");
+
+ if (!ret)
+ gb_interface_deactivate(intf);
+
+ return -EIO;
+ }
+ } else {
+ intf->type = type;
+ }
+
+ return ret;
+}
+
+/*
+ * Deactivate an interface.
+ *
+ * Locking: Caller holds the interface mutex.
+ */
+void gb_interface_deactivate(struct gb_interface *intf)
+{
+ if (!intf->active)
+ return;
+
+ trace_gb_interface_deactivate(intf);
+
+ /* Abort any ongoing mode switch. */
+ if (intf->mode_switch)
+ complete(&intf->mode_switch_completion);
+
+ gb_interface_route_destroy(intf);
+ gb_interface_hibernate_link(intf);
+ gb_interface_unipro_set(intf, false);
+ gb_interface_refclk_set(intf, false);
+ gb_interface_vsys_set(intf, false);
+
+ intf->active = false;
+}
+
+/*
+ * Enable an interface by enabling its control connection, fetching the
+ * manifest and other information over it, and finally registering its child
+ * devices.
+ *
+ * Locking: Caller holds the interface mutex.
+ */
+int gb_interface_enable(struct gb_interface *intf)
+{
+ struct gb_control *control;
+ struct gb_bundle *bundle, *tmp;
+ int ret, size;
+ void *manifest;
+
+ ret = gb_interface_read_and_clear_init_status(intf);
+ if (ret) {
+ dev_err(&intf->dev, "failed to clear init status: %d\n", ret);
+ return ret;
+ }
+
+ /* Establish control connection */
+ control = gb_control_create(intf);
+ if (IS_ERR(control)) {
+ dev_err(&intf->dev, "failed to create control device: %ld\n",
+ PTR_ERR(control));
+ return PTR_ERR(control);
+ }
+ intf->control = control;
+
+ ret = gb_control_enable(intf->control);
+ if (ret)
+ goto err_put_control;
+
+ /* Get manifest size using control protocol on CPort */
+ size = gb_control_get_manifest_size_operation(intf);
+ if (size <= 0) {
+ dev_err(&intf->dev, "failed to get manifest size: %d\n", size);
+
+ if (size)
+ ret = size;
+ else
+ ret = -EINVAL;
+
+ goto err_disable_control;
+ }
+
+ manifest = kmalloc(size, GFP_KERNEL);
+ if (!manifest) {
+ ret = -ENOMEM;
+ goto err_disable_control;
+ }
+
+ /* Get manifest using control protocol on CPort */
+ ret = gb_control_get_manifest_operation(intf, manifest, size);
+ if (ret) {
+ dev_err(&intf->dev, "failed to get manifest: %d\n", ret);
+ goto err_free_manifest;
+ }
+
+ /*
+ * Parse the manifest and build up our data structures representing
+ * what's in it.
+ */
+ if (!gb_manifest_parse(intf, manifest, size)) {
+ dev_err(&intf->dev, "failed to parse manifest\n");
+ ret = -EINVAL;
+ goto err_destroy_bundles;
+ }
+
+ ret = gb_control_get_bundle_versions(intf->control);
+ if (ret)
+ goto err_destroy_bundles;
+
+ ret = gb_timesync_interface_add(intf);
+ if (ret) {
+ dev_err(&intf->dev, "failed to add to timesync: %d\n", ret);
+ goto err_destroy_bundles;
+ }
+
+ /* Register the control device and any bundles */
+ ret = gb_control_add(intf->control);
+ if (ret)
+ goto err_remove_timesync;
+
+ pm_runtime_use_autosuspend(&intf->dev);
+ pm_runtime_get_noresume(&intf->dev);
+ pm_runtime_set_active(&intf->dev);
+ pm_runtime_enable(&intf->dev);
+
+ list_for_each_entry_safe_reverse(bundle, tmp, &intf->bundles, links) {
+ ret = gb_bundle_add(bundle);
+ if (ret) {
+ gb_bundle_destroy(bundle);
+ continue;
+ }
+ }
+
+ kfree(manifest);
+
+ intf->enabled = true;
+
+ pm_runtime_put(&intf->dev);
+
+ trace_gb_interface_enable(intf);
+
+ return 0;
+
+err_remove_timesync:
+ gb_timesync_interface_remove(intf);
+err_destroy_bundles:
+ list_for_each_entry_safe(bundle, tmp, &intf->bundles, links)
+ gb_bundle_destroy(bundle);
+err_free_manifest:
+ kfree(manifest);
+err_disable_control:
+ gb_control_disable(intf->control);
+err_put_control:
+ gb_control_put(intf->control);
+ intf->control = NULL;
+
+ return ret;
+}
+
+/*
+ * Disable an interface and destroy its bundles.
+ *
+ * Locking: Caller holds the interface mutex.
+ */
+void gb_interface_disable(struct gb_interface *intf)
+{
+ struct gb_bundle *bundle;
+ struct gb_bundle *next;
+
+ if (!intf->enabled)
+ return;
+
+ trace_gb_interface_disable(intf);
+
+ pm_runtime_get_sync(&intf->dev);
+
+ /* Set disconnected flag to avoid I/O during connection tear down. */
+ if (intf->quirks & GB_INTERFACE_QUIRK_FORCED_DISABLE)
+ intf->disconnected = true;
+
+ list_for_each_entry_safe(bundle, next, &intf->bundles, links)
+ gb_bundle_destroy(bundle);
+
+ if (!intf->mode_switch && !intf->disconnected)
+ gb_control_interface_deactivate_prepare(intf->control);
+
+ gb_control_del(intf->control);
+ gb_timesync_interface_remove(intf);
+ gb_control_disable(intf->control);
+ gb_control_put(intf->control);
+ intf->control = NULL;
+
+ intf->enabled = false;
+
+ pm_runtime_disable(&intf->dev);
+ pm_runtime_set_suspended(&intf->dev);
+ pm_runtime_dont_use_autosuspend(&intf->dev);
+ pm_runtime_put_noidle(&intf->dev);
+}
+
+/* Enable TimeSync on an Interface control connection. */
+int gb_interface_timesync_enable(struct gb_interface *intf, u8 count,
+ u64 frame_time, u32 strobe_delay, u32 refclk)
+{
+ return gb_control_timesync_enable(intf->control, count,
+ frame_time, strobe_delay,
+ refclk);
+}
+
+/* Disable TimeSync on an Interface control connection. */
+int gb_interface_timesync_disable(struct gb_interface *intf)
+{
+ return gb_control_timesync_disable(intf->control);
+}
+
+/* Transmit the Authoritative FrameTime via an Interface control connection. */
+int gb_interface_timesync_authoritative(struct gb_interface *intf,
+ u64 *frame_time)
+{
+ return gb_control_timesync_authoritative(intf->control,
+ frame_time);
+}
+
+/* Register an interface. */
+int gb_interface_add(struct gb_interface *intf)
+{
+ int ret;
+
+ ret = device_add(&intf->dev);
+ if (ret) {
+ dev_err(&intf->dev, "failed to register interface: %d\n", ret);
+ return ret;
+ }
+
+ trace_gb_interface_add(intf);
+
+ dev_info(&intf->dev, "Interface added (%s)\n",
+ gb_interface_type_string(intf));
+
+ switch (intf->type) {
+ case GB_INTERFACE_TYPE_GREYBUS:
+ dev_info(&intf->dev, "GMP VID=0x%08x, PID=0x%08x\n",
+ intf->vendor_id, intf->product_id);
+ /* fall-through */
+ case GB_INTERFACE_TYPE_UNIPRO:
+ dev_info(&intf->dev, "DDBL1 Manufacturer=0x%08x, Product=0x%08x\n",
+ intf->ddbl1_manufacturer_id,
+ intf->ddbl1_product_id);
+ break;
+ default:
+ break;
+ }
+
+ return 0;
+}
+
+/* Deregister an interface. */
+void gb_interface_del(struct gb_interface *intf)
+{
+ if (device_is_registered(&intf->dev)) {
+ trace_gb_interface_del(intf);
+
+ device_del(&intf->dev);
+ dev_info(&intf->dev, "Interface removed\n");
+ }
+}
+
+void gb_interface_put(struct gb_interface *intf)
+{
+ put_device(&intf->dev);
+}
--- /dev/null
+++ b/drivers/greybus/interface.h
@@ -0,0 +1,88 @@
+/*
+ * Greybus Interface Block code
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __INTERFACE_H
+#define __INTERFACE_H
+
+enum gb_interface_type {
+ GB_INTERFACE_TYPE_INVALID = 0,
+ GB_INTERFACE_TYPE_UNKNOWN,
+ GB_INTERFACE_TYPE_DUMMY,
+ GB_INTERFACE_TYPE_UNIPRO,
+ GB_INTERFACE_TYPE_GREYBUS,
+};
+
+#define GB_INTERFACE_QUIRK_NO_CPORT_FEATURES BIT(0)
+#define GB_INTERFACE_QUIRK_NO_INIT_STATUS BIT(1)
+#define GB_INTERFACE_QUIRK_NO_GMP_IDS BIT(2)
+#define GB_INTERFACE_QUIRK_FORCED_DISABLE BIT(3)
+#define GB_INTERFACE_QUIRK_LEGACY_MODE_SWITCH BIT(4)
+#define GB_INTERFACE_QUIRK_NO_BUNDLE_ACTIVATE BIT(5)
+#define GB_INTERFACE_QUIRK_NO_PM BIT(6)
+
+struct gb_interface {
+ struct device dev;
+ struct gb_control *control;
+
+ struct list_head bundles;
+ struct list_head module_node;
+ struct list_head manifest_descs;
+ u8 interface_id; /* Physical location within the Endo */
+ u8 device_id;
+ u8 features; /* Feature flags set in the manifest */
+
+ enum gb_interface_type type;
+
+ u32 ddbl1_manufacturer_id;
+ u32 ddbl1_product_id;
+ u32 vendor_id;
+ u32 product_id;
+ u64 serial_number;
+
+ struct gb_host_device *hd;
+ struct gb_module *module;
+
+ unsigned long quirks;
+
+ struct mutex mutex;
+
+ bool disconnected;
+
+ bool ejected;
+ bool removed;
+ bool active;
+ bool enabled;
+ bool mode_switch;
+ bool dme_read;
+
+ struct work_struct mode_switch_work;
+ struct completion mode_switch_completion;
+};
+#define to_gb_interface(d) container_of(d, struct gb_interface, dev)
+
+struct gb_interface *gb_interface_create(struct gb_module *module,
+ u8 interface_id);
+int gb_interface_activate(struct gb_interface *intf);
+void gb_interface_deactivate(struct gb_interface *intf);
+int gb_interface_enable(struct gb_interface *intf);
+void gb_interface_disable(struct gb_interface *intf);
+int gb_interface_timesync_enable(struct gb_interface *intf, u8 count,
+ u64 frame_time, u32 strobe_delay, u32 refclk);
+int gb_interface_timesync_authoritative(struct gb_interface *intf,
+ u64 *frame_time);
+int gb_interface_timesync_disable(struct gb_interface *intf);
+int gb_interface_add(struct gb_interface *intf);
+void gb_interface_del(struct gb_interface *intf);
+void gb_interface_put(struct gb_interface *intf);
+void gb_interface_mailbox_event(struct gb_interface *intf, u16 result,
+ u32 mailbox);
+
+int gb_interface_request_mode_switch(struct gb_interface *intf);
+
+#endif /* __INTERFACE_H */
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-16 6:05 ` Greg KH
@ 2016-09-16 10:18 ` Mark Brown
2016-09-16 13:22 ` Greg KH
2016-09-16 12:18 ` Arnd Bergmann
2016-09-21 13:02 ` Mark Rutland
2 siblings, 1 reply; 84+ messages in thread
From: Mark Brown @ 2016-09-16 10:18 UTC (permalink / raw)
To: Greg KH
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
[-- Attachment #1: Type: text/plain, Size: 755 bytes --]
On Fri, Sep 16, 2016 at 08:05:19AM +0200, Greg KH wrote:
> On Thu, Sep 15, 2016 at 03:45:53PM +0100, Mark Brown wrote:
> > Sending a pull request for code that's never been seen upstream seems
> > completely premature.
> Hey, how does code get upstream then? :)
By having the actual code to the mailing list.
> As for the drivers all living under drivers/greybus/ I understand, but
> we need the greybus core present first before we can get the drivers in.
> How about we do what happened with IIO, we take the greybus core code in
> drivers/greybus/ and put the drivers in staging, and then move them out
> of staging into the "real" portion of the kernel as they get reviewed
> and accepted?
Yes, staging would be appropriate for unreviewed code.
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 473 bytes --]
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-16 6:05 ` Greg KH
2016-09-16 10:18 ` Mark Brown
@ 2016-09-16 12:18 ` Arnd Bergmann
2016-09-21 13:02 ` Mark Rutland
2 siblings, 0 replies; 84+ messages in thread
From: Arnd Bergmann @ 2016-09-16 12:18 UTC (permalink / raw)
To: Greg KH
Cc: Mark Brown, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
On Friday, September 16, 2016 8:05:19 AM CEST Greg KH wrote:
> On Thu, Sep 15, 2016 at 03:45:53PM +0100, Mark Brown wrote:
> > On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
> >
> > > I'll send out a follow-up set of "simple" patches that just add the
> > > files to the kernel tree, to give people an idea of the code involved.
> > > Overall, it's a tiny stand-alone driver subsystem, only 37k lines, that
> > > implements a protocol which allows for "generic" cameras, audio devices,
> > > and other class type devices, as well as a bridged "physical" layer
> > > protocol to talk to serial, spi, uart, pwm, gpio, i2c, and even USB host
> >
> > Just emphasizing what Mark Rutland said complete NACK, in particular the
> > drivers for functions have not been posted upstream at all (and it's
> > concerning that they're all being added under drivers/greybus rather
> > than within the relevant subsystem like we do normally). I've not
> > looked at the code as it has not been submitted but given that and that
> > the reason I found this pull request was an ASoC contributor who had
> > seen it and looked at the code going "oh dear, greybus..." on IRC I'm
> > very concerned.
> >
> > Sending a pull request for code that's never been seen upstream seems
> > completely premature.
>
> Hey, how does code get upstream then? :)
>
> Anyway, as vger seems to be marking all of the greybus patches I send
> out as "spam" according to a filter, it's making it a bit hard to send
> patches out, but I'll try it again "by hand" later today...
>
> As for the drivers all living under drivers/greybus/ I understand, but
> we need the greybus core present first before we can get the drivers in.
> How about we do what happened with IIO, we take the greybus core code in
> drivers/greybus/ and put the drivers in staging, and then move them out
> of staging into the "real" portion of the kernel as they get reviewed
> and accepted?
>
> Any objections to that workflow?
I think that will work fine. If we can review just the core code,
the patch series probably gets small enough that more people
are inclined to take a closer look. At least speaking for me
personally, reviewing 32 nontrivial patches at once is unlikely
to happen.
>From a brief look, I found that at least the smaller device drivers
should be uncontroversial and won't have a problem getting merged
into the subsystem directories once they have be reviewed by the
respective maintainers.
Some drivers are larger (audio sticks out by quite a bit) and are
more likely to be controversial. Some other drivers (e.g. vibrator
or bootrom) don't have an obvious subsystem they belong into,
which can also cause problems, but having them in staging in the
meantime seems fine.
Arnd
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-16 10:18 ` Mark Brown
@ 2016-09-16 13:22 ` Greg KH
2016-09-16 14:24 ` Greg KH
0 siblings, 1 reply; 84+ messages in thread
From: Greg KH @ 2016-09-16 13:22 UTC (permalink / raw)
To: Mark Brown
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
On Fri, Sep 16, 2016 at 11:18:07AM +0100, Mark Brown wrote:
> On Fri, Sep 16, 2016 at 08:05:19AM +0200, Greg KH wrote:
> > On Thu, Sep 15, 2016 at 03:45:53PM +0100, Mark Brown wrote:
>
> > > Sending a pull request for code that's never been seen upstream seems
> > > completely premature.
>
> > Hey, how does code get upstream then? :)
>
> By having the actual code to the mailing list.
Ugh, vger keeps blocking the patches, I'm going to try for a third time
now...
> > As for the drivers all living under drivers/greybus/ I understand, but
> > we need the greybus core present first before we can get the drivers in.
> > How about we do what happened with IIO, we take the greybus core code in
> > drivers/greybus/ and put the drivers in staging, and then move them out
> > of staging into the "real" portion of the kernel as they get reviewed
> > and accepted?
>
> Yes, staging would be appropriate for unreviewed code.
Sounds good.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 03/32] greybus: operations logic
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
2016-09-16 6:41 ` [patch 02/32] greybus: interface control logic Greg KH
@ 2016-09-16 13:22 ` Greg KH
2016-09-16 13:23 ` [patch 04/32] greybus: host driver framework Greg KH
` (28 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 13:22 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This is the Greybus operations code, handling the logic for them.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/operation.c | 1239 ++++++++++++++++++++++++++++++++++++++++++++
drivers/greybus/operation.h | 210 +++++++
2 files changed, 1449 insertions(+)
--- /dev/null
+++ b/drivers/greybus/operation.c
@@ -0,0 +1,1239 @@
+/*
+ * Greybus operations
+ *
+ * Copyright 2014-2015 Google Inc.
+ * Copyright 2014-2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/wait.h>
+#include <linux/workqueue.h>
+
+#include "greybus.h"
+#include "greybus_trace.h"
+
+static struct kmem_cache *gb_operation_cache;
+static struct kmem_cache *gb_message_cache;
+
+/* Workqueue to handle Greybus operation completions. */
+static struct workqueue_struct *gb_operation_completion_wq;
+
+/* Wait queue for synchronous cancellations. */
+static DECLARE_WAIT_QUEUE_HEAD(gb_operation_cancellation_queue);
+
+/*
+ * Protects updates to operation->errno.
+ */
+static DEFINE_SPINLOCK(gb_operations_lock);
+
+static int gb_operation_response_send(struct gb_operation *operation,
+ int errno);
+
+/*
+ * Increment operation active count and add to connection list unless the
+ * connection is going away.
+ *
+ * Caller holds operation reference.
+ */
+static int gb_operation_get_active(struct gb_operation *operation)
+{
+ struct gb_connection *connection = operation->connection;
+ unsigned long flags;
+
+ spin_lock_irqsave(&connection->lock, flags);
+ switch (connection->state) {
+ case GB_CONNECTION_STATE_ENABLED:
+ break;
+ case GB_CONNECTION_STATE_ENABLED_TX:
+ if (gb_operation_is_incoming(operation))
+ goto err_unlock;
+ break;
+ case GB_CONNECTION_STATE_DISCONNECTING:
+ if (!gb_operation_is_core(operation))
+ goto err_unlock;
+ break;
+ default:
+ goto err_unlock;
+ }
+
+ if (operation->active++ == 0)
+ list_add_tail(&operation->links, &connection->operations);
+
+ trace_gb_operation_get_active(operation);
+
+ spin_unlock_irqrestore(&connection->lock, flags);
+
+ return 0;
+
+err_unlock:
+ spin_unlock_irqrestore(&connection->lock, flags);
+
+ return -ENOTCONN;
+}
+
+/* Caller holds operation reference. */
+static void gb_operation_put_active(struct gb_operation *operation)
+{
+ struct gb_connection *connection = operation->connection;
+ unsigned long flags;
+
+ spin_lock_irqsave(&connection->lock, flags);
+
+ trace_gb_operation_put_active(operation);
+
+ if (--operation->active == 0) {
+ list_del(&operation->links);
+ if (atomic_read(&operation->waiters))
+ wake_up(&gb_operation_cancellation_queue);
+ }
+ spin_unlock_irqrestore(&connection->lock, flags);
+}
+
+static bool gb_operation_is_active(struct gb_operation *operation)
+{
+ struct gb_connection *connection = operation->connection;
+ unsigned long flags;
+ bool ret;
+
+ spin_lock_irqsave(&connection->lock, flags);
+ ret = operation->active;
+ spin_unlock_irqrestore(&connection->lock, flags);
+
+ return ret;
+}
+
+/*
+ * Set an operation's result.
+ *
+ * Initially an outgoing operation's errno value is -EBADR.
+ * If no error occurs before sending the request message the only
+ * valid value operation->errno can be set to is -EINPROGRESS,
+ * indicating the request has been (or rather is about to be) sent.
+ * At that point nobody should be looking at the result until the
+ * response arrives.
+ *
+ * The first time the result gets set after the request has been
+ * sent, that result "sticks." That is, if two concurrent threads
+ * race to set the result, the first one wins. The return value
+ * tells the caller whether its result was recorded; if not the
+ * caller has nothing more to do.
+ *
+ * The result value -EILSEQ is reserved to signal an implementation
+ * error; if it's ever observed, the code performing the request has
+ * done something fundamentally wrong. It is an error to try to set
+ * the result to -EBADR, and attempts to do so result in a warning,
+ * and -EILSEQ is used instead. Similarly, the only valid result
+ * value to set for an operation in initial state is -EINPROGRESS.
+ * Attempts to do otherwise will also record a (successful) -EILSEQ
+ * operation result.
+ */
+static bool gb_operation_result_set(struct gb_operation *operation, int result)
+{
+ unsigned long flags;
+ int prev;
+
+ if (result == -EINPROGRESS) {
+ /*
+ * -EINPROGRESS is used to indicate the request is
+ * in flight. It should be the first result value
+ * set after the initial -EBADR. Issue a warning
+ * and record an implementation error if it's
+ * set at any other time.
+ */
+ spin_lock_irqsave(&gb_operations_lock, flags);
+ prev = operation->errno;
+ if (prev == -EBADR)
+ operation->errno = result;
+ else
+ operation->errno = -EILSEQ;
+ spin_unlock_irqrestore(&gb_operations_lock, flags);
+ WARN_ON(prev != -EBADR);
+
+ return true;
+ }
+
+ /*
+ * The first result value set after a request has been sent
+ * will be the final result of the operation. Subsequent
+ * attempts to set the result are ignored.
+ *
+ * Note that -EBADR is a reserved "initial state" result
+ * value. Attempts to set this value result in a warning,
+ * and the result code is set to -EILSEQ instead.
+ */
+ if (WARN_ON(result == -EBADR))
+ result = -EILSEQ; /* Nobody should be setting -EBADR */
+
+ spin_lock_irqsave(&gb_operations_lock, flags);
+ prev = operation->errno;
+ if (prev == -EINPROGRESS)
+ operation->errno = result; /* First and final result */
+ spin_unlock_irqrestore(&gb_operations_lock, flags);
+
+ return prev == -EINPROGRESS;
+}
+
+int gb_operation_result(struct gb_operation *operation)
+{
+ int result = operation->errno;
+
+ WARN_ON(result == -EBADR);
+ WARN_ON(result == -EINPROGRESS);
+
+ return result;
+}
+EXPORT_SYMBOL_GPL(gb_operation_result);
+
+/*
+ * Looks up an outgoing operation on a connection and returns a refcounted
+ * pointer if found, or NULL otherwise.
+ */
+static struct gb_operation *
+gb_operation_find_outgoing(struct gb_connection *connection, u16 operation_id)
+{
+ struct gb_operation *operation;
+ unsigned long flags;
+ bool found = false;
+
+ spin_lock_irqsave(&connection->lock, flags);
+ list_for_each_entry(operation, &connection->operations, links)
+ if (operation->id == operation_id &&
+ !gb_operation_is_incoming(operation)) {
+ gb_operation_get(operation);
+ found = true;
+ break;
+ }
+ spin_unlock_irqrestore(&connection->lock, flags);
+
+ return found ? operation : NULL;
+}
+
+static int gb_message_send(struct gb_message *message, gfp_t gfp)
+{
+ struct gb_connection *connection = message->operation->connection;
+
+ trace_gb_message_send(message);
+ return connection->hd->driver->message_send(connection->hd,
+ connection->hd_cport_id,
+ message,
+ gfp);
+}
+
+/*
+ * Cancel a message we have passed to the host device layer to be sent.
+ */
+static void gb_message_cancel(struct gb_message *message)
+{
+ struct gb_host_device *hd = message->operation->connection->hd;
+
+ hd->driver->message_cancel(message);
+}
+
+static void gb_operation_request_handle(struct gb_operation *operation)
+{
+ struct gb_connection *connection = operation->connection;
+ int status;
+ int ret;
+
+ if (connection->handler) {
+ status = connection->handler(operation);
+ } else {
+ dev_err(&connection->hd->dev,
+ "%s: unexpected incoming request of type 0x%02x\n",
+ connection->name, operation->type);
+
+ status = -EPROTONOSUPPORT;
+ }
+
+ ret = gb_operation_response_send(operation, status);
+ if (ret) {
+ dev_err(&connection->hd->dev,
+ "%s: failed to send response %d for type 0x%02x: %d\n",
+ connection->name, status, operation->type, ret);
+ return;
+ }
+}
+
+/*
+ * Process operation work.
+ *
+ * For incoming requests, call the protocol request handler. The operation
+ * result should be -EINPROGRESS at this point.
+ *
+ * For outgoing requests, the operation result value should have
+ * been set before queueing this. The operation callback function
+ * allows the original requester to know the request has completed
+ * and its result is available.
+ */
+static void gb_operation_work(struct work_struct *work)
+{
+ struct gb_operation *operation;
+
+ operation = container_of(work, struct gb_operation, work);
+
+ if (gb_operation_is_incoming(operation))
+ gb_operation_request_handle(operation);
+ else
+ operation->callback(operation);
+
+ gb_operation_put_active(operation);
+ gb_operation_put(operation);
+}
+
+static void gb_operation_message_init(struct gb_host_device *hd,
+ struct gb_message *message, u16 operation_id,
+ size_t payload_size, u8 type)
+{
+ struct gb_operation_msg_hdr *header;
+
+ header = message->buffer;
+
+ message->header = header;
+ message->payload = payload_size ? header + 1 : NULL;
+ message->payload_size = payload_size;
+
+ /*
+ * The type supplied for incoming message buffers will be
+ * GB_REQUEST_TYPE_INVALID. Such buffers will be overwritten by
+ * arriving data so there's no need to initialize the message header.
+ */
+ if (type != GB_REQUEST_TYPE_INVALID) {
+ u16 message_size = (u16)(sizeof(*header) + payload_size);
+
+ /*
+ * For a request, the operation id gets filled in
+ * when the message is sent. For a response, it
+ * will be copied from the request by the caller.
+ *
+ * The result field in a request message must be
+ * zero. It will be set just prior to sending for
+ * a response.
+ */
+ header->size = cpu_to_le16(message_size);
+ header->operation_id = 0;
+ header->type = type;
+ header->result = 0;
+ }
+}
+
+/*
+ * Allocate a message to be used for an operation request or response.
+ * Both types of message contain a common header. The request message
+ * for an outgoing operation is outbound, as is the response message
+ * for an incoming operation. The message header for an outbound
+ * message is partially initialized here.
+ *
+ * The headers for inbound messages don't need to be initialized;
+ * they'll be filled in by arriving data.
+ *
+ * Our message buffers have the following layout:
+ * message header \_ these combined are
+ * message payload / the message size
+ */
+static struct gb_message *
+gb_operation_message_alloc(struct gb_host_device *hd, u8 type,
+ size_t payload_size, gfp_t gfp_flags)
+{
+ struct gb_message *message;
+ struct gb_operation_msg_hdr *header;
+ size_t message_size = payload_size + sizeof(*header);
+
+ if (message_size > hd->buffer_size_max) {
+ dev_warn(&hd->dev, "requested message size too big (%zu > %zu)\n",
+ message_size, hd->buffer_size_max);
+ return NULL;
+ }
+
+ /* Allocate the message structure and buffer. */
+ message = kmem_cache_zalloc(gb_message_cache, gfp_flags);
+ if (!message)
+ return NULL;
+
+ message->buffer = kzalloc(message_size, gfp_flags);
+ if (!message->buffer)
+ goto err_free_message;
+
+ /* Initialize the message. Operation id is filled in later. */
+ gb_operation_message_init(hd, message, 0, payload_size, type);
+
+ return message;
+
+err_free_message:
+ kmem_cache_free(gb_message_cache, message);
+
+ return NULL;
+}
+
+static void gb_operation_message_free(struct gb_message *message)
+{
+ kfree(message->buffer);
+ kmem_cache_free(gb_message_cache, message);
+}
+
+/*
+ * Map an enum gb_operation_status value (which is represented in a
+ * message as a single byte) to an appropriate Linux negative errno.
+ */
+static int gb_operation_status_map(u8 status)
+{
+ switch (status) {
+ case GB_OP_SUCCESS:
+ return 0;
+ case GB_OP_INTERRUPTED:
+ return -EINTR;
+ case GB_OP_TIMEOUT:
+ return -ETIMEDOUT;
+ case GB_OP_NO_MEMORY:
+ return -ENOMEM;
+ case GB_OP_PROTOCOL_BAD:
+ return -EPROTONOSUPPORT;
+ case GB_OP_OVERFLOW:
+ return -EMSGSIZE;
+ case GB_OP_INVALID:
+ return -EINVAL;
+ case GB_OP_RETRY:
+ return -EAGAIN;
+ case GB_OP_NONEXISTENT:
+ return -ENODEV;
+ case GB_OP_MALFUNCTION:
+ return -EILSEQ;
+ case GB_OP_UNKNOWN_ERROR:
+ default:
+ return -EIO;
+ }
+}
+
+/*
+ * Map a Linux errno value (from operation->errno) into the value
+ * that should represent it in a response message status sent
+ * over the wire. Returns an enum gb_operation_status value (which
+ * is represented in a message as a single byte).
+ */
+static u8 gb_operation_errno_map(int errno)
+{
+ switch (errno) {
+ case 0:
+ return GB_OP_SUCCESS;
+ case -EINTR:
+ return GB_OP_INTERRUPTED;
+ case -ETIMEDOUT:
+ return GB_OP_TIMEOUT;
+ case -ENOMEM:
+ return GB_OP_NO_MEMORY;
+ case -EPROTONOSUPPORT:
+ return GB_OP_PROTOCOL_BAD;
+ case -EMSGSIZE:
+ return GB_OP_OVERFLOW; /* Could be underflow too */
+ case -EINVAL:
+ return GB_OP_INVALID;
+ case -EAGAIN:
+ return GB_OP_RETRY;
+ case -EILSEQ:
+ return GB_OP_MALFUNCTION;
+ case -ENODEV:
+ return GB_OP_NONEXISTENT;
+ case -EIO:
+ default:
+ return GB_OP_UNKNOWN_ERROR;
+ }
+}
+
+bool gb_operation_response_alloc(struct gb_operation *operation,
+ size_t response_size, gfp_t gfp)
+{
+ struct gb_host_device *hd = operation->connection->hd;
+ struct gb_operation_msg_hdr *request_header;
+ struct gb_message *response;
+ u8 type;
+
+ type = operation->type | GB_MESSAGE_TYPE_RESPONSE;
+ response = gb_operation_message_alloc(hd, type, response_size, gfp);
+ if (!response)
+ return false;
+ response->operation = operation;
+
+ /*
+ * Size and type get initialized when the message is
+ * allocated. The errno will be set before sending. All
+ * that's left is the operation id, which we copy from the
+ * request message header (as-is, in little-endian order).
+ */
+ request_header = operation->request->header;
+ response->header->operation_id = request_header->operation_id;
+ operation->response = response;
+
+ return true;
+}
+EXPORT_SYMBOL_GPL(gb_operation_response_alloc);
+
+/*
+ * Create a Greybus operation to be sent over the given connection.
+ * The request buffer will be big enough for a payload of the given
+ * size.
+ *
+ * For outgoing requests, the request message's header will be
+ * initialized with the type of the request and the message size.
+ * Outgoing operations must also specify the response buffer size,
+ * which must be sufficient to hold all expected response data. The
+ * response message header will eventually be overwritten, so there's
+ * no need to initialize it here.
+ *
+ * Request messages for incoming operations can arrive in interrupt
+ * context, so they must be allocated with GFP_ATOMIC. In this case
+ * the request buffer will be immediately overwritten, so there is
+ * no need to initialize the message header. Responsibility for
+ * allocating a response buffer lies with the incoming request
+ * handler for a protocol. So we don't allocate that here.
+ *
+ * Returns a pointer to the new operation or a null pointer if an
+ * error occurs.
+ */
+static struct gb_operation *
+gb_operation_create_common(struct gb_connection *connection, u8 type,
+ size_t request_size, size_t response_size,
+ unsigned long op_flags, gfp_t gfp_flags)
+{
+ struct gb_host_device *hd = connection->hd;
+ struct gb_operation *operation;
+
+ operation = kmem_cache_zalloc(gb_operation_cache, gfp_flags);
+ if (!operation)
+ return NULL;
+ operation->connection = connection;
+
+ operation->request = gb_operation_message_alloc(hd, type, request_size,
+ gfp_flags);
+ if (!operation->request)
+ goto err_cache;
+ operation->request->operation = operation;
+
+ /* Allocate the response buffer for outgoing operations */
+ if (!(op_flags & GB_OPERATION_FLAG_INCOMING)) {
+ if (!gb_operation_response_alloc(operation, response_size,
+ gfp_flags)) {
+ goto err_request;
+ }
+ }
+
+ operation->flags = op_flags;
+ operation->type = type;
+ operation->errno = -EBADR; /* Initial value--means "never set" */
+
+ INIT_WORK(&operation->work, gb_operation_work);
+ init_completion(&operation->completion);
+ kref_init(&operation->kref);
+ atomic_set(&operation->waiters, 0);
+
+ return operation;
+
+err_request:
+ gb_operation_message_free(operation->request);
+err_cache:
+ kmem_cache_free(gb_operation_cache, operation);
+
+ return NULL;
+}
+
+/*
+ * Create a new operation associated with the given connection. The
+ * request and response sizes provided are the number of bytes
+ * required to hold the request/response payload only. Both of
+ * these are allowed to be 0. Note that 0x00 is reserved as an
+ * invalid operation type for all protocols, and this is enforced
+ * here.
+ */
+struct gb_operation *
+gb_operation_create_flags(struct gb_connection *connection,
+ u8 type, size_t request_size,
+ size_t response_size, unsigned long flags,
+ gfp_t gfp)
+{
+ struct gb_operation *operation;
+
+ if (WARN_ON_ONCE(type == GB_REQUEST_TYPE_INVALID))
+ return NULL;
+ if (WARN_ON_ONCE(type & GB_MESSAGE_TYPE_RESPONSE))
+ type &= ~GB_MESSAGE_TYPE_RESPONSE;
+
+ if (WARN_ON_ONCE(flags & ~GB_OPERATION_FLAG_USER_MASK))
+ flags &= GB_OPERATION_FLAG_USER_MASK;
+
+ operation = gb_operation_create_common(connection, type,
+ request_size, response_size,
+ flags, gfp);
+ if (operation)
+ trace_gb_operation_create(operation);
+
+ return operation;
+}
+EXPORT_SYMBOL_GPL(gb_operation_create_flags);
+
+struct gb_operation *
+gb_operation_create_core(struct gb_connection *connection,
+ u8 type, size_t request_size,
+ size_t response_size, unsigned long flags,
+ gfp_t gfp)
+{
+ struct gb_operation *operation;
+
+ flags |= GB_OPERATION_FLAG_CORE;
+
+ operation = gb_operation_create_common(connection, type,
+ request_size, response_size,
+ flags, gfp);
+ if (operation)
+ trace_gb_operation_create_core(operation);
+
+ return operation;
+}
+/* Do not export this function. */
+
+size_t gb_operation_get_payload_size_max(struct gb_connection *connection)
+{
+ struct gb_host_device *hd = connection->hd;
+
+ return hd->buffer_size_max - sizeof(struct gb_operation_msg_hdr);
+}
+EXPORT_SYMBOL_GPL(gb_operation_get_payload_size_max);
+
+static struct gb_operation *
+gb_operation_create_incoming(struct gb_connection *connection, u16 id,
+ u8 type, void *data, size_t size)
+{
+ struct gb_operation *operation;
+ size_t request_size;
+ unsigned long flags = GB_OPERATION_FLAG_INCOMING;
+
+ /* Caller has made sure we at least have a message header. */
+ request_size = size - sizeof(struct gb_operation_msg_hdr);
+
+ if (!id)
+ flags |= GB_OPERATION_FLAG_UNIDIRECTIONAL;
+
+ operation = gb_operation_create_common(connection, type,
+ request_size,
+ GB_REQUEST_TYPE_INVALID,
+ flags, GFP_ATOMIC);
+ if (!operation)
+ return NULL;
+
+ operation->id = id;
+ memcpy(operation->request->header, data, size);
+ trace_gb_operation_create_incoming(operation);
+
+ return operation;
+}
+
+/*
+ * Get an additional reference on an operation.
+ */
+void gb_operation_get(struct gb_operation *operation)
+{
+ kref_get(&operation->kref);
+}
+EXPORT_SYMBOL_GPL(gb_operation_get);
+
+/*
+ * Destroy a previously created operation.
+ */
+static void _gb_operation_destroy(struct kref *kref)
+{
+ struct gb_operation *operation;
+
+ operation = container_of(kref, struct gb_operation, kref);
+
+ trace_gb_operation_destroy(operation);
+
+ if (operation->response)
+ gb_operation_message_free(operation->response);
+ gb_operation_message_free(operation->request);
+
+ kmem_cache_free(gb_operation_cache, operation);
+}
+
+/*
+ * Drop a reference on an operation, and destroy it when the last
+ * one is gone.
+ */
+void gb_operation_put(struct gb_operation *operation)
+{
+ if (WARN_ON(!operation))
+ return;
+
+ kref_put(&operation->kref, _gb_operation_destroy);
+}
+EXPORT_SYMBOL_GPL(gb_operation_put);
+
+/* Tell the requester we're done */
+static void gb_operation_sync_callback(struct gb_operation *operation)
+{
+ complete(&operation->completion);
+}
+
+/**
+ * gb_operation_request_send() - send an operation request message
+ * @operation: the operation to initiate
+ * @callback: the operation completion callback
+ * @gfp: the memory flags to use for any allocations
+ *
+ * The caller has filled in any payload so the request message is ready to go.
+ * The callback function supplied will be called when the response message has
+ * arrived, a unidirectional request has been sent, or the operation is
+ * cancelled, indicating that the operation is complete. The callback function
+ * can fetch the result of the operation using gb_operation_result() if
+ * desired.
+ *
+ * Return: 0 if the request was successfully queued in the host-driver queues,
+ * or a negative errno.
+ */
+int gb_operation_request_send(struct gb_operation *operation,
+ gb_operation_callback callback,
+ gfp_t gfp)
+{
+ struct gb_connection *connection = operation->connection;
+ struct gb_operation_msg_hdr *header;
+ unsigned int cycle;
+ int ret;
+
+ if (gb_connection_is_offloaded(connection))
+ return -EBUSY;
+
+ if (!callback)
+ return -EINVAL;
+
+ /*
+ * Record the callback function, which is executed in
+ * non-atomic (workqueue) context when the final result
+ * of an operation has been set.
+ */
+ operation->callback = callback;
+
+ /*
+ * Assign the operation's id, and store it in the request header.
+ * Zero is a reserved operation id for unidirectional operations.
+ */
+ if (gb_operation_is_unidirectional(operation)) {
+ operation->id = 0;
+ } else {
+ cycle = (unsigned int)atomic_inc_return(&connection->op_cycle);
+ operation->id = (u16)(cycle % U16_MAX + 1);
+ }
+
+ header = operation->request->header;
+ header->operation_id = cpu_to_le16(operation->id);
+
+ gb_operation_result_set(operation, -EINPROGRESS);
+
+ /*
+ * Get an extra reference on the operation. It'll be dropped when the
+ * operation completes.
+ */
+ gb_operation_get(operation);
+ ret = gb_operation_get_active(operation);
+ if (ret)
+ goto err_put;
+
+ ret = gb_message_send(operation->request, gfp);
+ if (ret)
+ goto err_put_active;
+
+ return 0;
+
+err_put_active:
+ gb_operation_put_active(operation);
+err_put:
+ gb_operation_put(operation);
+
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_operation_request_send);
+
+/*
+ * Send a synchronous operation. This function is expected to
+ * block, returning only when the response has arrived, (or when an
+ * error is detected. The return value is the result of the
+ * operation.
+ */
+int gb_operation_request_send_sync_timeout(struct gb_operation *operation,
+ unsigned int timeout)
+{
+ int ret;
+ unsigned long timeout_jiffies;
+
+ ret = gb_operation_request_send(operation, gb_operation_sync_callback,
+ GFP_KERNEL);
+ if (ret)
+ return ret;
+
+ if (timeout)
+ timeout_jiffies = msecs_to_jiffies(timeout);
+ else
+ timeout_jiffies = MAX_SCHEDULE_TIMEOUT;
+
+ ret = wait_for_completion_interruptible_timeout(&operation->completion,
+ timeout_jiffies);
+ if (ret < 0) {
+ /* Cancel the operation if interrupted */
+ gb_operation_cancel(operation, -ECANCELED);
+ } else if (ret == 0) {
+ /* Cancel the operation if op timed out */
+ gb_operation_cancel(operation, -ETIMEDOUT);
+ }
+
+ return gb_operation_result(operation);
+}
+EXPORT_SYMBOL_GPL(gb_operation_request_send_sync_timeout);
+
+/*
+ * Send a response for an incoming operation request. A non-zero
+ * errno indicates a failed operation.
+ *
+ * If there is any response payload, the incoming request handler is
+ * responsible for allocating the response message. Otherwise the
+ * it can simply supply the result errno; this function will
+ * allocate the response message if necessary.
+ */
+static int gb_operation_response_send(struct gb_operation *operation,
+ int errno)
+{
+ struct gb_connection *connection = operation->connection;
+ int ret;
+
+ if (!operation->response &&
+ !gb_operation_is_unidirectional(operation)) {
+ if (!gb_operation_response_alloc(operation, 0, GFP_KERNEL))
+ return -ENOMEM;
+ }
+
+ /* Record the result */
+ if (!gb_operation_result_set(operation, errno)) {
+ dev_err(&connection->hd->dev, "request result already set\n");
+ return -EIO; /* Shouldn't happen */
+ }
+
+ /* Sender of request does not care about response. */
+ if (gb_operation_is_unidirectional(operation))
+ return 0;
+
+ /* Reference will be dropped when message has been sent. */
+ gb_operation_get(operation);
+ ret = gb_operation_get_active(operation);
+ if (ret)
+ goto err_put;
+
+ /* Fill in the response header and send it */
+ operation->response->header->result = gb_operation_errno_map(errno);
+
+ ret = gb_message_send(operation->response, GFP_KERNEL);
+ if (ret)
+ goto err_put_active;
+
+ return 0;
+
+err_put_active:
+ gb_operation_put_active(operation);
+err_put:
+ gb_operation_put(operation);
+
+ return ret;
+}
+
+/*
+ * This function is called when a message send request has completed.
+ */
+void greybus_message_sent(struct gb_host_device *hd,
+ struct gb_message *message, int status)
+{
+ struct gb_operation *operation = message->operation;
+ struct gb_connection *connection = operation->connection;
+
+ /*
+ * If the message was a response, we just need to drop our
+ * reference to the operation. If an error occurred, report
+ * it.
+ *
+ * For requests, if there's no error and the operation in not
+ * unidirectional, there's nothing more to do until the response
+ * arrives. If an error occurred attempting to send it, or if the
+ * operation is unidrectional, record the result of the operation and
+ * schedule its completion.
+ */
+ if (message == operation->response) {
+ if (status) {
+ dev_err(&connection->hd->dev,
+ "%s: error sending response 0x%02x: %d\n",
+ connection->name, operation->type, status);
+ }
+
+ gb_operation_put_active(operation);
+ gb_operation_put(operation);
+ } else if (status || gb_operation_is_unidirectional(operation)) {
+ if (gb_operation_result_set(operation, status)) {
+ queue_work(gb_operation_completion_wq,
+ &operation->work);
+ }
+ }
+}
+EXPORT_SYMBOL_GPL(greybus_message_sent);
+
+/*
+ * We've received data on a connection, and it doesn't look like a
+ * response, so we assume it's a request.
+ *
+ * This is called in interrupt context, so just copy the incoming
+ * data into the request buffer and handle the rest via workqueue.
+ */
+static void gb_connection_recv_request(struct gb_connection *connection,
+ const struct gb_operation_msg_hdr *header,
+ void *data, size_t size)
+{
+ struct gb_operation *operation;
+ u16 operation_id;
+ u8 type;
+ int ret;
+
+ operation_id = le16_to_cpu(header->operation_id);
+ type = header->type;
+
+ operation = gb_operation_create_incoming(connection, operation_id,
+ type, data, size);
+ if (!operation) {
+ dev_err(&connection->hd->dev,
+ "%s: can't create incoming operation\n",
+ connection->name);
+ return;
+ }
+
+ ret = gb_operation_get_active(operation);
+ if (ret) {
+ gb_operation_put(operation);
+ return;
+ }
+ trace_gb_message_recv_request(operation->request);
+
+ /*
+ * The initial reference to the operation will be dropped when the
+ * request handler returns.
+ */
+ if (gb_operation_result_set(operation, -EINPROGRESS))
+ queue_work(connection->wq, &operation->work);
+}
+
+/*
+ * We've received data that appears to be an operation response
+ * message. Look up the operation, and record that we've received
+ * its response.
+ *
+ * This is called in interrupt context, so just copy the incoming
+ * data into the response buffer and handle the rest via workqueue.
+ */
+static void gb_connection_recv_response(struct gb_connection *connection,
+ const struct gb_operation_msg_hdr *header,
+ void *data, size_t size)
+{
+ struct gb_operation *operation;
+ struct gb_message *message;
+ size_t message_size;
+ u16 operation_id;
+ int errno;
+
+ operation_id = le16_to_cpu(header->operation_id);
+
+ if (!operation_id) {
+ dev_err_ratelimited(&connection->hd->dev,
+ "%s: invalid response id 0 received\n",
+ connection->name);
+ return;
+ }
+
+ operation = gb_operation_find_outgoing(connection, operation_id);
+ if (!operation) {
+ dev_err_ratelimited(&connection->hd->dev,
+ "%s: unexpected response id 0x%04x received\n",
+ connection->name, operation_id);
+ return;
+ }
+
+ errno = gb_operation_status_map(header->result);
+ message = operation->response;
+ message_size = sizeof(*header) + message->payload_size;
+ if (!errno && size > message_size) {
+ dev_err_ratelimited(&connection->hd->dev,
+ "%s: malformed response 0x%02x received (%zu > %zu)\n",
+ connection->name, header->type,
+ size, message_size);
+ errno = -EMSGSIZE;
+ } else if (!errno && size < message_size) {
+ if (gb_operation_short_response_allowed(operation)) {
+ message->payload_size = size - sizeof(*header);
+ } else {
+ dev_err_ratelimited(&connection->hd->dev,
+ "%s: short response 0x%02x received (%zu < %zu)\n",
+ connection->name, header->type,
+ size, message_size);
+ errno = -EMSGSIZE;
+ }
+ }
+
+ /* We must ignore the payload if a bad status is returned */
+ if (errno)
+ size = sizeof(*header);
+
+ /* The rest will be handled in work queue context */
+ if (gb_operation_result_set(operation, errno)) {
+ memcpy(message->buffer, data, size);
+
+ trace_gb_message_recv_response(message);
+
+ queue_work(gb_operation_completion_wq, &operation->work);
+ }
+
+ gb_operation_put(operation);
+}
+
+/*
+ * Handle data arriving on a connection. As soon as we return the
+ * supplied data buffer will be reused (so unless we do something
+ * with, it's effectively dropped).
+ */
+void gb_connection_recv(struct gb_connection *connection,
+ void *data, size_t size)
+{
+ struct gb_operation_msg_hdr header;
+ struct device *dev = &connection->hd->dev;
+ size_t msg_size;
+
+ if (connection->state == GB_CONNECTION_STATE_DISABLED ||
+ gb_connection_is_offloaded(connection)) {
+ dev_warn_ratelimited(dev, "%s: dropping %zu received bytes\n",
+ connection->name, size);
+ return;
+ }
+
+ if (size < sizeof(header)) {
+ dev_err_ratelimited(dev, "%s: short message received\n",
+ connection->name);
+ return;
+ }
+
+ /* Use memcpy as data may be unaligned */
+ memcpy(&header, data, sizeof(header));
+ msg_size = le16_to_cpu(header.size);
+ if (size < msg_size) {
+ dev_err_ratelimited(dev,
+ "%s: incomplete message 0x%04x of type 0x%02x received (%zu < %zu)\n",
+ connection->name,
+ le16_to_cpu(header.operation_id),
+ header.type, size, msg_size);
+ return; /* XXX Should still complete operation */
+ }
+
+ if (header.type & GB_MESSAGE_TYPE_RESPONSE) {
+ gb_connection_recv_response(connection, &header, data,
+ msg_size);
+ } else {
+ gb_connection_recv_request(connection, &header, data,
+ msg_size);
+ }
+}
+
+/*
+ * Cancel an outgoing operation synchronously, and record the given error to
+ * indicate why.
+ */
+void gb_operation_cancel(struct gb_operation *operation, int errno)
+{
+ if (WARN_ON(gb_operation_is_incoming(operation)))
+ return;
+
+ if (gb_operation_result_set(operation, errno)) {
+ gb_message_cancel(operation->request);
+ queue_work(gb_operation_completion_wq, &operation->work);
+ }
+ trace_gb_message_cancel_outgoing(operation->request);
+
+ atomic_inc(&operation->waiters);
+ wait_event(gb_operation_cancellation_queue,
+ !gb_operation_is_active(operation));
+ atomic_dec(&operation->waiters);
+}
+EXPORT_SYMBOL_GPL(gb_operation_cancel);
+
+/*
+ * Cancel an incoming operation synchronously. Called during connection tear
+ * down.
+ */
+void gb_operation_cancel_incoming(struct gb_operation *operation, int errno)
+{
+ if (WARN_ON(!gb_operation_is_incoming(operation)))
+ return;
+
+ if (!gb_operation_is_unidirectional(operation)) {
+ /*
+ * Make sure the request handler has submitted the response
+ * before cancelling it.
+ */
+ flush_work(&operation->work);
+ if (!gb_operation_result_set(operation, errno))
+ gb_message_cancel(operation->response);
+ }
+ trace_gb_message_cancel_incoming(operation->response);
+
+ atomic_inc(&operation->waiters);
+ wait_event(gb_operation_cancellation_queue,
+ !gb_operation_is_active(operation));
+ atomic_dec(&operation->waiters);
+}
+
+/**
+ * gb_operation_sync_timeout() - implement a "simple" synchronous operation
+ * @connection: the Greybus connection to send this to
+ * @type: the type of operation to send
+ * @request: pointer to a memory buffer to copy the request from
+ * @request_size: size of @request
+ * @response: pointer to a memory buffer to copy the response to
+ * @response_size: the size of @response.
+ * @timeout: operation timeout in milliseconds
+ *
+ * This function implements a simple synchronous Greybus operation. It sends
+ * the provided operation request and waits (sleeps) until the corresponding
+ * operation response message has been successfully received, or an error
+ * occurs. @request and @response are buffers to hold the request and response
+ * data respectively, and if they are not NULL, their size must be specified in
+ * @request_size and @response_size.
+ *
+ * If a response payload is to come back, and @response is not NULL,
+ * @response_size number of bytes will be copied into @response if the operation
+ * is successful.
+ *
+ * If there is an error, the response buffer is left alone.
+ */
+int gb_operation_sync_timeout(struct gb_connection *connection, int type,
+ void *request, int request_size,
+ void *response, int response_size,
+ unsigned int timeout)
+{
+ struct gb_operation *operation;
+ int ret;
+
+ if ((response_size && !response) ||
+ (request_size && !request))
+ return -EINVAL;
+
+ operation = gb_operation_create(connection, type,
+ request_size, response_size,
+ GFP_KERNEL);
+ if (!operation)
+ return -ENOMEM;
+
+ if (request_size)
+ memcpy(operation->request->payload, request, request_size);
+
+ ret = gb_operation_request_send_sync_timeout(operation, timeout);
+ if (ret) {
+ dev_err(&connection->hd->dev,
+ "%s: synchronous operation id 0x%04x of type 0x%02x failed: %d\n",
+ connection->name, operation->id, type, ret);
+ } else {
+ if (response_size) {
+ memcpy(response, operation->response->payload,
+ response_size);
+ }
+ }
+
+ gb_operation_put(operation);
+
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_operation_sync_timeout);
+
+/**
+ * gb_operation_unidirectional_timeout() - initiate a unidirectional operation
+ * @connection: connection to use
+ * @type: type of operation to send
+ * @request: memory buffer to copy the request from
+ * @request_size: size of @request
+ * @timeout: send timeout in milliseconds
+ *
+ * Initiate a unidirectional operation by sending a request message and
+ * waiting for it to be acknowledged as sent by the host device.
+ *
+ * Note that successful send of a unidirectional operation does not imply that
+ * the request as actually reached the remote end of the connection.
+ */
+int gb_operation_unidirectional_timeout(struct gb_connection *connection,
+ int type, void *request, int request_size,
+ unsigned int timeout)
+{
+ struct gb_operation *operation;
+ int ret;
+
+ if (request_size && !request)
+ return -EINVAL;
+
+ operation = gb_operation_create_flags(connection, type,
+ request_size, 0,
+ GB_OPERATION_FLAG_UNIDIRECTIONAL,
+ GFP_KERNEL);
+ if (!operation)
+ return -ENOMEM;
+
+ if (request_size)
+ memcpy(operation->request->payload, request, request_size);
+
+ ret = gb_operation_request_send_sync_timeout(operation, timeout);
+ if (ret) {
+ dev_err(&connection->hd->dev,
+ "%s: unidirectional operation of type 0x%02x failed: %d\n",
+ connection->name, type, ret);
+ }
+
+ gb_operation_put(operation);
+
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_operation_unidirectional_timeout);
+
+int __init gb_operation_init(void)
+{
+ gb_message_cache = kmem_cache_create("gb_message_cache",
+ sizeof(struct gb_message), 0, 0, NULL);
+ if (!gb_message_cache)
+ return -ENOMEM;
+
+ gb_operation_cache = kmem_cache_create("gb_operation_cache",
+ sizeof(struct gb_operation), 0, 0, NULL);
+ if (!gb_operation_cache)
+ goto err_destroy_message_cache;
+
+ gb_operation_completion_wq = alloc_workqueue("greybus_completion",
+ 0, 0);
+ if (!gb_operation_completion_wq)
+ goto err_destroy_operation_cache;
+
+ return 0;
+
+err_destroy_operation_cache:
+ kmem_cache_destroy(gb_operation_cache);
+ gb_operation_cache = NULL;
+err_destroy_message_cache:
+ kmem_cache_destroy(gb_message_cache);
+ gb_message_cache = NULL;
+
+ return -ENOMEM;
+}
+
+void gb_operation_exit(void)
+{
+ destroy_workqueue(gb_operation_completion_wq);
+ gb_operation_completion_wq = NULL;
+ kmem_cache_destroy(gb_operation_cache);
+ gb_operation_cache = NULL;
+ kmem_cache_destroy(gb_message_cache);
+ gb_message_cache = NULL;
+}
--- /dev/null
+++ b/drivers/greybus/operation.h
@@ -0,0 +1,210 @@
+/*
+ * Greybus operations
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __OPERATION_H
+#define __OPERATION_H
+
+#include <linux/completion.h>
+
+struct gb_operation;
+
+/* The default amount of time a request is given to complete */
+#define GB_OPERATION_TIMEOUT_DEFAULT 1000 /* milliseconds */
+
+/*
+ * The top bit of the type in an operation message header indicates
+ * whether the message is a request (bit clear) or response (bit set)
+ */
+#define GB_MESSAGE_TYPE_RESPONSE ((u8)0x80)
+
+enum gb_operation_result {
+ GB_OP_SUCCESS = 0x00,
+ GB_OP_INTERRUPTED = 0x01,
+ GB_OP_TIMEOUT = 0x02,
+ GB_OP_NO_MEMORY = 0x03,
+ GB_OP_PROTOCOL_BAD = 0x04,
+ GB_OP_OVERFLOW = 0x05,
+ GB_OP_INVALID = 0x06,
+ GB_OP_RETRY = 0x07,
+ GB_OP_NONEXISTENT = 0x08,
+ GB_OP_UNKNOWN_ERROR = 0xfe,
+ GB_OP_MALFUNCTION = 0xff,
+};
+
+#define GB_OPERATION_MESSAGE_SIZE_MIN sizeof(struct gb_operation_msg_hdr)
+#define GB_OPERATION_MESSAGE_SIZE_MAX U16_MAX
+
+/*
+ * Protocol code should only examine the payload and payload_size fields, and
+ * host-controller drivers may use the hcpriv field. All other fields are
+ * intended to be private to the operations core code.
+ */
+struct gb_message {
+ struct gb_operation *operation;
+ struct gb_operation_msg_hdr *header;
+
+ void *payload;
+ size_t payload_size;
+
+ void *buffer;
+
+ void *hcpriv;
+};
+
+#define GB_OPERATION_FLAG_INCOMING BIT(0)
+#define GB_OPERATION_FLAG_UNIDIRECTIONAL BIT(1)
+#define GB_OPERATION_FLAG_SHORT_RESPONSE BIT(2)
+#define GB_OPERATION_FLAG_CORE BIT(3)
+
+#define GB_OPERATION_FLAG_USER_MASK (GB_OPERATION_FLAG_SHORT_RESPONSE | \
+ GB_OPERATION_FLAG_UNIDIRECTIONAL)
+
+/*
+ * A Greybus operation is a remote procedure call performed over a
+ * connection between two UniPro interfaces.
+ *
+ * Every operation consists of a request message sent to the other
+ * end of the connection coupled with a reply message returned to
+ * the sender. Every operation has a type, whose interpretation is
+ * dependent on the protocol associated with the connection.
+ *
+ * Only four things in an operation structure are intended to be
+ * directly usable by protocol handlers: the operation's connection
+ * pointer; the operation type; the request message payload (and
+ * size); and the response message payload (and size). Note that a
+ * message with a 0-byte payload has a null message payload pointer.
+ *
+ * In addition, every operation has a result, which is an errno
+ * value. Protocol handlers access the operation result using
+ * gb_operation_result().
+ */
+typedef void (*gb_operation_callback)(struct gb_operation *);
+struct gb_operation {
+ struct gb_connection *connection;
+ struct gb_message *request;
+ struct gb_message *response;
+
+ unsigned long flags;
+ u8 type;
+ u16 id;
+ int errno; /* Operation result */
+
+ struct work_struct work;
+ gb_operation_callback callback;
+ struct completion completion;
+
+ struct kref kref;
+ atomic_t waiters;
+
+ int active;
+ struct list_head links; /* connection->operations */
+};
+
+static inline bool
+gb_operation_is_incoming(struct gb_operation *operation)
+{
+ return operation->flags & GB_OPERATION_FLAG_INCOMING;
+}
+
+static inline bool
+gb_operation_is_unidirectional(struct gb_operation *operation)
+{
+ return operation->flags & GB_OPERATION_FLAG_UNIDIRECTIONAL;
+}
+
+static inline bool
+gb_operation_short_response_allowed(struct gb_operation *operation)
+{
+ return operation->flags & GB_OPERATION_FLAG_SHORT_RESPONSE;
+}
+
+static inline bool gb_operation_is_core(struct gb_operation *operation)
+{
+ return operation->flags & GB_OPERATION_FLAG_CORE;
+}
+
+void gb_connection_recv(struct gb_connection *connection,
+ void *data, size_t size);
+
+int gb_operation_result(struct gb_operation *operation);
+
+size_t gb_operation_get_payload_size_max(struct gb_connection *connection);
+struct gb_operation *
+gb_operation_create_flags(struct gb_connection *connection,
+ u8 type, size_t request_size,
+ size_t response_size, unsigned long flags,
+ gfp_t gfp);
+
+static inline struct gb_operation *
+gb_operation_create(struct gb_connection *connection,
+ u8 type, size_t request_size,
+ size_t response_size, gfp_t gfp)
+{
+ return gb_operation_create_flags(connection, type, request_size,
+ response_size, 0, gfp);
+}
+
+struct gb_operation *
+gb_operation_create_core(struct gb_connection *connection,
+ u8 type, size_t request_size,
+ size_t response_size, unsigned long flags,
+ gfp_t gfp);
+
+void gb_operation_get(struct gb_operation *operation);
+void gb_operation_put(struct gb_operation *operation);
+
+bool gb_operation_response_alloc(struct gb_operation *operation,
+ size_t response_size, gfp_t gfp);
+
+int gb_operation_request_send(struct gb_operation *operation,
+ gb_operation_callback callback,
+ gfp_t gfp);
+int gb_operation_request_send_sync_timeout(struct gb_operation *operation,
+ unsigned int timeout);
+static inline int
+gb_operation_request_send_sync(struct gb_operation *operation)
+{
+ return gb_operation_request_send_sync_timeout(operation,
+ GB_OPERATION_TIMEOUT_DEFAULT);
+}
+
+void gb_operation_cancel(struct gb_operation *operation, int errno);
+void gb_operation_cancel_incoming(struct gb_operation *operation, int errno);
+
+void greybus_message_sent(struct gb_host_device *hd,
+ struct gb_message *message, int status);
+
+int gb_operation_sync_timeout(struct gb_connection *connection, int type,
+ void *request, int request_size,
+ void *response, int response_size,
+ unsigned int timeout);
+int gb_operation_unidirectional_timeout(struct gb_connection *connection,
+ int type, void *request, int request_size,
+ unsigned int timeout);
+
+static inline int gb_operation_sync(struct gb_connection *connection, int type,
+ void *request, int request_size,
+ void *response, int response_size)
+{
+ return gb_operation_sync_timeout(connection, type,
+ request, request_size, response, response_size,
+ GB_OPERATION_TIMEOUT_DEFAULT);
+}
+
+static inline int gb_operation_unidirectional(struct gb_connection *connection,
+ int type, void *request, int request_size)
+{
+ return gb_operation_unidirectional_timeout(connection, type,
+ request, request_size, GB_OPERATION_TIMEOUT_DEFAULT);
+}
+
+int gb_operation_init(void);
+void gb_operation_exit(void);
+
+#endif /* !__OPERATION_H */
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 04/32] greybus: host driver framework
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
2016-09-16 6:41 ` [patch 02/32] greybus: interface control logic Greg KH
2016-09-16 13:22 ` [patch 03/32] greybus: operations logic Greg KH
@ 2016-09-16 13:23 ` Greg KH
2016-09-16 13:23 ` [patch 05/32] greybus: trace.h Greg KH
` (27 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 13:23 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This is the Greybus host driver framework files and logic.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/hd.c | 257 +++++++++++++++++++++++++++++++++++++++++++++++++++
drivers/greybus/hd.h | 90 +++++++++++++++++
2 files changed, 347 insertions(+)
--- /dev/null
+++ b/drivers/greybus/hd.c
@@ -0,0 +1,257 @@
+/*
+ * Greybus Host Device
+ *
+ * Copyright 2014-2015 Google Inc.
+ * Copyright 2014-2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/kernel.h>
+#include <linux/slab.h>
+
+#include "greybus.h"
+#include "greybus_trace.h"
+
+EXPORT_TRACEPOINT_SYMBOL_GPL(gb_hd_create);
+EXPORT_TRACEPOINT_SYMBOL_GPL(gb_hd_release);
+EXPORT_TRACEPOINT_SYMBOL_GPL(gb_hd_add);
+EXPORT_TRACEPOINT_SYMBOL_GPL(gb_hd_del);
+EXPORT_TRACEPOINT_SYMBOL_GPL(gb_hd_in);
+EXPORT_TRACEPOINT_SYMBOL_GPL(gb_message_submit);
+
+static struct ida gb_hd_bus_id_map;
+
+int gb_hd_output(struct gb_host_device *hd, void *req, u16 size, u8 cmd,
+ bool async)
+{
+ if (!hd || !hd->driver || !hd->driver->output)
+ return -EINVAL;
+ return hd->driver->output(hd, req, size, cmd, async);
+}
+EXPORT_SYMBOL_GPL(gb_hd_output);
+
+static ssize_t bus_id_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_host_device *hd = to_gb_host_device(dev);
+
+ return sprintf(buf, "%d\n", hd->bus_id);
+}
+static DEVICE_ATTR_RO(bus_id);
+
+static struct attribute *bus_attrs[] = {
+ &dev_attr_bus_id.attr,
+ NULL
+};
+ATTRIBUTE_GROUPS(bus);
+
+int gb_hd_cport_reserve(struct gb_host_device *hd, u16 cport_id)
+{
+ struct ida *id_map = &hd->cport_id_map;
+ int ret;
+
+ ret = ida_simple_get(id_map, cport_id, cport_id + 1, GFP_KERNEL);
+ if (ret < 0) {
+ dev_err(&hd->dev, "failed to reserve cport %u\n", cport_id);
+ return ret;
+ }
+
+ return 0;
+}
+EXPORT_SYMBOL_GPL(gb_hd_cport_reserve);
+
+void gb_hd_cport_release_reserved(struct gb_host_device *hd, u16 cport_id)
+{
+ struct ida *id_map = &hd->cport_id_map;
+
+ ida_simple_remove(id_map, cport_id);
+}
+EXPORT_SYMBOL_GPL(gb_hd_cport_release_reserved);
+
+/* Locking: Caller guarantees serialisation */
+int gb_hd_cport_allocate(struct gb_host_device *hd, int cport_id,
+ unsigned long flags)
+{
+ struct ida *id_map = &hd->cport_id_map;
+ int ida_start, ida_end;
+
+ if (hd->driver->cport_allocate)
+ return hd->driver->cport_allocate(hd, cport_id, flags);
+
+ if (cport_id < 0) {
+ ida_start = 0;
+ ida_end = hd->num_cports;
+ } else if (cport_id < hd->num_cports) {
+ ida_start = cport_id;
+ ida_end = cport_id + 1;
+ } else {
+ dev_err(&hd->dev, "cport %d not available\n", cport_id);
+ return -EINVAL;
+ }
+
+ return ida_simple_get(id_map, ida_start, ida_end, GFP_KERNEL);
+}
+
+/* Locking: Caller guarantees serialisation */
+void gb_hd_cport_release(struct gb_host_device *hd, u16 cport_id)
+{
+ if (hd->driver->cport_release) {
+ hd->driver->cport_release(hd, cport_id);
+ return;
+ }
+
+ ida_simple_remove(&hd->cport_id_map, cport_id);
+}
+
+static void gb_hd_release(struct device *dev)
+{
+ struct gb_host_device *hd = to_gb_host_device(dev);
+
+ trace_gb_hd_release(hd);
+
+ if (hd->svc)
+ gb_svc_put(hd->svc);
+ ida_simple_remove(&gb_hd_bus_id_map, hd->bus_id);
+ ida_destroy(&hd->cport_id_map);
+ kfree(hd);
+}
+
+struct device_type greybus_hd_type = {
+ .name = "greybus_host_device",
+ .release = gb_hd_release,
+};
+
+struct gb_host_device *gb_hd_create(struct gb_hd_driver *driver,
+ struct device *parent,
+ size_t buffer_size_max,
+ size_t num_cports)
+{
+ struct gb_host_device *hd;
+ int ret;
+
+ /*
+ * Validate that the driver implements all of the callbacks
+ * so that we don't have to every time we make them.
+ */
+ if ((!driver->message_send) || (!driver->message_cancel)) {
+ dev_err(parent, "mandatory hd-callbacks missing\n");
+ return ERR_PTR(-EINVAL);
+ }
+
+ if (buffer_size_max < GB_OPERATION_MESSAGE_SIZE_MIN) {
+ dev_err(parent, "greybus host-device buffers too small\n");
+ return ERR_PTR(-EINVAL);
+ }
+
+ if (num_cports == 0 || num_cports > CPORT_ID_MAX + 1) {
+ dev_err(parent, "Invalid number of CPorts: %zu\n", num_cports);
+ return ERR_PTR(-EINVAL);
+ }
+
+ /*
+ * Make sure to never allocate messages larger than what the Greybus
+ * protocol supports.
+ */
+ if (buffer_size_max > GB_OPERATION_MESSAGE_SIZE_MAX) {
+ dev_warn(parent, "limiting buffer size to %u\n",
+ GB_OPERATION_MESSAGE_SIZE_MAX);
+ buffer_size_max = GB_OPERATION_MESSAGE_SIZE_MAX;
+ }
+
+ hd = kzalloc(sizeof(*hd) + driver->hd_priv_size, GFP_KERNEL);
+ if (!hd)
+ return ERR_PTR(-ENOMEM);
+
+ ret = ida_simple_get(&gb_hd_bus_id_map, 1, 0, GFP_KERNEL);
+ if (ret < 0) {
+ kfree(hd);
+ return ERR_PTR(ret);
+ }
+ hd->bus_id = ret;
+
+ hd->driver = driver;
+ INIT_LIST_HEAD(&hd->modules);
+ INIT_LIST_HEAD(&hd->connections);
+ ida_init(&hd->cport_id_map);
+ hd->buffer_size_max = buffer_size_max;
+ hd->num_cports = num_cports;
+
+ hd->dev.parent = parent;
+ hd->dev.bus = &greybus_bus_type;
+ hd->dev.type = &greybus_hd_type;
+ hd->dev.groups = bus_groups;
+ hd->dev.dma_mask = hd->dev.parent->dma_mask;
+ device_initialize(&hd->dev);
+ dev_set_name(&hd->dev, "greybus%d", hd->bus_id);
+
+ trace_gb_hd_create(hd);
+
+ hd->svc = gb_svc_create(hd);
+ if (!hd->svc) {
+ dev_err(&hd->dev, "failed to create svc\n");
+ put_device(&hd->dev);
+ return ERR_PTR(-ENOMEM);
+ }
+
+ return hd;
+}
+EXPORT_SYMBOL_GPL(gb_hd_create);
+
+int gb_hd_add(struct gb_host_device *hd)
+{
+ int ret;
+
+ ret = device_add(&hd->dev);
+ if (ret)
+ return ret;
+
+ ret = gb_svc_add(hd->svc);
+ if (ret) {
+ device_del(&hd->dev);
+ return ret;
+ }
+
+ trace_gb_hd_add(hd);
+
+ return 0;
+}
+EXPORT_SYMBOL_GPL(gb_hd_add);
+
+void gb_hd_del(struct gb_host_device *hd)
+{
+ trace_gb_hd_del(hd);
+
+ /*
+ * Tear down the svc and flush any on-going hotplug processing before
+ * removing the remaining interfaces.
+ */
+ gb_svc_del(hd->svc);
+
+ device_del(&hd->dev);
+}
+EXPORT_SYMBOL_GPL(gb_hd_del);
+
+void gb_hd_shutdown(struct gb_host_device *hd)
+{
+ gb_svc_del(hd->svc);
+}
+EXPORT_SYMBOL_GPL(gb_hd_shutdown);
+
+void gb_hd_put(struct gb_host_device *hd)
+{
+ put_device(&hd->dev);
+}
+EXPORT_SYMBOL_GPL(gb_hd_put);
+
+int __init gb_hd_init(void)
+{
+ ida_init(&gb_hd_bus_id_map);
+
+ return 0;
+}
+
+void gb_hd_exit(void)
+{
+ ida_destroy(&gb_hd_bus_id_map);
+}
--- /dev/null
+++ b/drivers/greybus/hd.h
@@ -0,0 +1,90 @@
+/*
+ * Greybus Host Device
+ *
+ * Copyright 2014-2015 Google Inc.
+ * Copyright 2014-2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __HD_H
+#define __HD_H
+
+struct gb_host_device;
+struct gb_message;
+
+struct gb_hd_driver {
+ size_t hd_priv_size;
+
+ int (*cport_allocate)(struct gb_host_device *hd, int cport_id,
+ unsigned long flags);
+ void (*cport_release)(struct gb_host_device *hd, u16 cport_id);
+ int (*cport_enable)(struct gb_host_device *hd, u16 cport_id,
+ unsigned long flags);
+ int (*cport_disable)(struct gb_host_device *hd, u16 cport_id);
+ int (*cport_connected)(struct gb_host_device *hd, u16 cport_id);
+ int (*cport_flush)(struct gb_host_device *hd, u16 cport_id);
+ int (*cport_shutdown)(struct gb_host_device *hd, u16 cport_id,
+ u8 phase, unsigned int timeout);
+ int (*cport_quiesce)(struct gb_host_device *hd, u16 cport_id,
+ size_t peer_space, unsigned int timeout);
+ int (*cport_clear)(struct gb_host_device *hd, u16 cport_id);
+
+ int (*message_send)(struct gb_host_device *hd, u16 dest_cport_id,
+ struct gb_message *message, gfp_t gfp_mask);
+ void (*message_cancel)(struct gb_message *message);
+ int (*latency_tag_enable)(struct gb_host_device *hd, u16 cport_id);
+ int (*latency_tag_disable)(struct gb_host_device *hd, u16 cport_id);
+ int (*output)(struct gb_host_device *hd, void *req, u16 size, u8 cmd,
+ bool async);
+ int (*timesync_enable)(struct gb_host_device *hd, u8 count,
+ u64 frame_time, u32 strobe_delay, u32 refclk);
+ int (*timesync_disable)(struct gb_host_device *hd);
+ int (*timesync_authoritative)(struct gb_host_device *hd,
+ u64 *frame_time);
+ int (*timesync_get_last_event)(struct gb_host_device *hd,
+ u64 *frame_time);
+};
+
+struct gb_host_device {
+ struct device dev;
+ int bus_id;
+ const struct gb_hd_driver *driver;
+
+ struct list_head modules;
+ struct list_head connections;
+ struct ida cport_id_map;
+
+ /* Number of CPorts supported by the UniPro IP */
+ size_t num_cports;
+
+ /* Host device buffer constraints */
+ size_t buffer_size_max;
+
+ struct gb_svc *svc;
+ /* Private data for the host driver */
+ unsigned long hd_priv[0] __aligned(sizeof(s64));
+};
+#define to_gb_host_device(d) container_of(d, struct gb_host_device, dev)
+
+int gb_hd_cport_reserve(struct gb_host_device *hd, u16 cport_id);
+void gb_hd_cport_release_reserved(struct gb_host_device *hd, u16 cport_id);
+int gb_hd_cport_allocate(struct gb_host_device *hd, int cport_id,
+ unsigned long flags);
+void gb_hd_cport_release(struct gb_host_device *hd, u16 cport_id);
+
+struct gb_host_device *gb_hd_create(struct gb_hd_driver *driver,
+ struct device *parent,
+ size_t buffer_size_max,
+ size_t num_cports);
+int gb_hd_add(struct gb_host_device *hd);
+void gb_hd_del(struct gb_host_device *hd);
+void gb_hd_shutdown(struct gb_host_device *hd);
+void gb_hd_put(struct gb_host_device *hd);
+int gb_hd_output(struct gb_host_device *hd, void *req, u16 size, u8 cmd,
+ bool in_irq);
+
+int gb_hd_init(void);
+void gb_hd_exit(void);
+
+#endif /* __HD_H */
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 05/32] greybus: trace.h
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (2 preceding siblings ...)
2016-09-16 13:23 ` [patch 04/32] greybus: host driver framework Greg KH
@ 2016-09-16 13:23 ` Greg KH
2016-09-16 13:23 ` [patch 06/32] greybus: svc driver/watchdog Greg KH
` (26 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 13:23 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This implements the Greybus trace macros that the system uses to see
what is going on when debugging.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/greybus_trace.h | 531 ++++++++++++++++++++++++++++++++++++++++
1 file changed, 531 insertions(+)
--- /dev/null
+++ b/drivers/greybus/greybus_trace.h
@@ -0,0 +1,531 @@
+/*
+ * Greybus driver and device API
+ *
+ * Copyright 2015 Google Inc.
+ * Copyright 2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM greybus
+
+#if !defined(_TRACE_GREYBUS_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_GREYBUS_H
+
+#include <linux/tracepoint.h>
+
+struct gb_message;
+struct gb_operation;
+struct gb_connection;
+struct gb_bundle;
+struct gb_host_device;
+
+DECLARE_EVENT_CLASS(gb_message,
+
+ TP_PROTO(struct gb_message *message),
+
+ TP_ARGS(message),
+
+ TP_STRUCT__entry(
+ __field(u16, size)
+ __field(u16, operation_id)
+ __field(u8, type)
+ __field(u8, result)
+ ),
+
+ TP_fast_assign(
+ __entry->size = le16_to_cpu(message->header->size);
+ __entry->operation_id =
+ le16_to_cpu(message->header->operation_id);
+ __entry->type = message->header->type;
+ __entry->result = message->header->result;
+ ),
+
+ TP_printk("size=%hu operation_id=0x%04x type=0x%02x result=0x%02x",
+ __entry->size, __entry->operation_id,
+ __entry->type, __entry->result)
+);
+
+#define DEFINE_MESSAGE_EVENT(name) \
+ DEFINE_EVENT(gb_message, name, \
+ TP_PROTO(struct gb_message *message), \
+ TP_ARGS(message))
+
+/*
+ * Occurs immediately before calling a host device's message_send()
+ * method.
+ */
+DEFINE_MESSAGE_EVENT(gb_message_send);
+
+/*
+ * Occurs after an incoming request message has been received
+ */
+DEFINE_MESSAGE_EVENT(gb_message_recv_request);
+
+/*
+ * Occurs after an incoming response message has been received,
+ * after its matching request has been found.
+ */
+DEFINE_MESSAGE_EVENT(gb_message_recv_response);
+
+/*
+ * Occurs after an operation has been canceled, possibly before the
+ * cancellation is complete.
+ */
+DEFINE_MESSAGE_EVENT(gb_message_cancel_outgoing);
+
+/*
+ * Occurs when an incoming request is cancelled; if the response has
+ * been queued for sending, this occurs after it is sent.
+ */
+DEFINE_MESSAGE_EVENT(gb_message_cancel_incoming);
+
+/*
+ * Occurs in the host driver message_send() function just prior to
+ * handing off the data to be processed by hardware.
+ */
+DEFINE_MESSAGE_EVENT(gb_message_submit);
+
+#undef DEFINE_MESSAGE_EVENT
+
+DECLARE_EVENT_CLASS(gb_operation,
+
+ TP_PROTO(struct gb_operation *operation),
+
+ TP_ARGS(operation),
+
+ TP_STRUCT__entry(
+ __field(u16, cport_id) /* CPort of HD side of connection */
+ __field(u16, id) /* Operation ID */
+ __field(u8, type)
+ __field(unsigned long, flags)
+ __field(int, active)
+ __field(int, waiters)
+ __field(int, errno)
+ ),
+
+ TP_fast_assign(
+ __entry->cport_id = operation->connection->hd_cport_id;
+ __entry->id = operation->id;
+ __entry->type = operation->type;
+ __entry->flags = operation->flags;
+ __entry->active = operation->active;
+ __entry->waiters = atomic_read(&operation->waiters);
+ __entry->errno = operation->errno;
+ ),
+
+ TP_printk("id=%04x type=0x%02x cport_id=%04x flags=0x%lx active=%d waiters=%d errno=%d",
+ __entry->id, __entry->cport_id, __entry->type, __entry->flags,
+ __entry->active, __entry->waiters, __entry->errno)
+);
+
+#define DEFINE_OPERATION_EVENT(name) \
+ DEFINE_EVENT(gb_operation, name, \
+ TP_PROTO(struct gb_operation *operation), \
+ TP_ARGS(operation))
+
+/*
+ * Occurs after a new operation is created for an outgoing request
+ * has been successfully created.
+ */
+DEFINE_OPERATION_EVENT(gb_operation_create);
+
+/*
+ * Occurs after a new core operation has been created.
+ */
+DEFINE_OPERATION_EVENT(gb_operation_create_core);
+
+/*
+ * Occurs after a new operation has been created for an incoming
+ * request has been successfully created and initialized.
+ */
+DEFINE_OPERATION_EVENT(gb_operation_create_incoming);
+
+/*
+ * Occurs when the last reference to an operation has been dropped,
+ * prior to freeing resources.
+ */
+DEFINE_OPERATION_EVENT(gb_operation_destroy);
+
+/*
+ * Occurs when an operation has been marked active, after updating
+ * its active count.
+ */
+DEFINE_OPERATION_EVENT(gb_operation_get_active);
+
+/*
+ * Occurs when an operation has been marked active, before updating
+ * its active count.
+ */
+DEFINE_OPERATION_EVENT(gb_operation_put_active);
+
+#undef DEFINE_OPERATION_EVENT
+
+DECLARE_EVENT_CLASS(gb_connection,
+
+ TP_PROTO(struct gb_connection *connection),
+
+ TP_ARGS(connection),
+
+ TP_STRUCT__entry(
+ __field(int, hd_bus_id)
+ __field(u8, bundle_id)
+ /* name contains "hd_cport_id/intf_id:cport_id" */
+ __dynamic_array(char, name, sizeof(connection->name))
+ __field(enum gb_connection_state, state)
+ __field(unsigned long, flags)
+ ),
+
+ TP_fast_assign(
+ __entry->hd_bus_id = connection->hd->bus_id;
+ __entry->bundle_id = connection->bundle ?
+ connection->bundle->id : BUNDLE_ID_NONE;
+ memcpy(__get_str(name), connection->name,
+ sizeof(connection->name));
+ __entry->state = connection->state;
+ __entry->flags = connection->flags;
+ ),
+
+ TP_printk("hd_bus_id=%d bundle_id=0x%02x name=\"%s\" state=%u flags=0x%lx",
+ __entry->hd_bus_id, __entry->bundle_id, __get_str(name),
+ (unsigned int)__entry->state, __entry->flags)
+);
+
+#define DEFINE_CONNECTION_EVENT(name) \
+ DEFINE_EVENT(gb_connection, name, \
+ TP_PROTO(struct gb_connection *connection), \
+ TP_ARGS(connection))
+
+/*
+ * Occurs after a new connection is successfully created.
+ */
+DEFINE_CONNECTION_EVENT(gb_connection_create);
+
+/*
+ * Occurs when the last reference to a connection has been dropped,
+ * before its resources are freed.
+ */
+DEFINE_CONNECTION_EVENT(gb_connection_release);
+
+/*
+ * Occurs when a new reference to connection is added, currently
+ * only when a message over the connection is received.
+ */
+DEFINE_CONNECTION_EVENT(gb_connection_get);
+
+/*
+ * Occurs when a new reference to connection is dropped, after a
+ * a received message is handled, or when the connection is
+ * destroyed.
+ */
+DEFINE_CONNECTION_EVENT(gb_connection_put);
+
+/*
+ * Occurs when a request to enable a connection is made, either for
+ * transmit only, or for both transmit and receive.
+ */
+DEFINE_CONNECTION_EVENT(gb_connection_enable);
+
+/*
+ * Occurs when a request to disable a connection is made, either for
+ * receive only, or for both transmit and receive. Also occurs when
+ * a request to forcefully disable a connection is made.
+ */
+DEFINE_CONNECTION_EVENT(gb_connection_disable);
+
+#undef DEFINE_CONNECTION_EVENT
+
+DECLARE_EVENT_CLASS(gb_bundle,
+
+ TP_PROTO(struct gb_bundle *bundle),
+
+ TP_ARGS(bundle),
+
+ TP_STRUCT__entry(
+ __field(u8, intf_id)
+ __field(u8, id)
+ __field(u8, class)
+ __field(size_t, num_cports)
+ ),
+
+ TP_fast_assign(
+ __entry->intf_id = bundle->intf->interface_id;
+ __entry->id = bundle->id;
+ __entry->class = bundle->class;
+ __entry->num_cports = bundle->num_cports;
+ ),
+
+ TP_printk("intf_id=0x%02x id=%02x class=0x%02x num_cports=%zu",
+ __entry->intf_id, __entry->id, __entry->class,
+ __entry->num_cports)
+);
+
+#define DEFINE_BUNDLE_EVENT(name) \
+ DEFINE_EVENT(gb_bundle, name, \
+ TP_PROTO(struct gb_bundle *bundle), \
+ TP_ARGS(bundle))
+
+/*
+ * Occurs after a new bundle is successfully created.
+ */
+DEFINE_BUNDLE_EVENT(gb_bundle_create);
+
+/*
+ * Occurs when the last reference to a bundle has been dropped,
+ * before its resources are freed.
+ */
+DEFINE_BUNDLE_EVENT(gb_bundle_release);
+
+/*
+ * Occurs when a bundle is added to an interface when the interface
+ * is enabled.
+ */
+DEFINE_BUNDLE_EVENT(gb_bundle_add);
+
+/*
+ * Occurs when a registered bundle gets destroyed, normally at the
+ * time an interface is disabled.
+ */
+DEFINE_BUNDLE_EVENT(gb_bundle_destroy);
+
+#undef DEFINE_BUNDLE_EVENT
+
+DECLARE_EVENT_CLASS(gb_interface,
+
+ TP_PROTO(struct gb_interface *intf),
+
+ TP_ARGS(intf),
+
+ TP_STRUCT__entry(
+ __field(u8, module_id)
+ __field(u8, id) /* Interface id */
+ __field(u8, device_id)
+ __field(int, disconnected) /* bool */
+ __field(int, ejected) /* bool */
+ __field(int, active) /* bool */
+ __field(int, enabled) /* bool */
+ __field(int, mode_switch) /* bool */
+ ),
+
+ TP_fast_assign(
+ __entry->module_id = intf->module->module_id;
+ __entry->id = intf->interface_id;
+ __entry->device_id = intf->device_id;
+ __entry->disconnected = intf->disconnected;
+ __entry->ejected = intf->ejected;
+ __entry->active = intf->active;
+ __entry->enabled = intf->enabled;
+ __entry->mode_switch = intf->mode_switch;
+ ),
+
+ TP_printk("intf_id=%hhu device_id=%hhu module_id=%hhu D=%d J=%d A=%d E=%d M=%d",
+ __entry->id, __entry->device_id, __entry->module_id,
+ __entry->disconnected, __entry->ejected, __entry->active,
+ __entry->enabled, __entry->mode_switch)
+);
+
+#define DEFINE_INTERFACE_EVENT(name) \
+ DEFINE_EVENT(gb_interface, name, \
+ TP_PROTO(struct gb_interface *intf), \
+ TP_ARGS(intf))
+
+/*
+ * Occurs after a new interface is successfully created.
+ */
+DEFINE_INTERFACE_EVENT(gb_interface_create);
+
+/*
+ * Occurs after the last reference to an interface has been dropped.
+ */
+DEFINE_INTERFACE_EVENT(gb_interface_release);
+
+/*
+ * Occurs after an interface been registerd.
+ */
+DEFINE_INTERFACE_EVENT(gb_interface_add);
+
+/*
+ * Occurs when a registered interface gets deregisterd.
+ */
+DEFINE_INTERFACE_EVENT(gb_interface_del);
+
+/*
+ * Occurs when a registered interface has been successfully
+ * activated.
+ */
+DEFINE_INTERFACE_EVENT(gb_interface_activate);
+
+/*
+ * Occurs when an activated interface is being deactivated.
+ */
+DEFINE_INTERFACE_EVENT(gb_interface_deactivate);
+
+/*
+ * Occurs when an interface has been successfully enabled.
+ */
+DEFINE_INTERFACE_EVENT(gb_interface_enable);
+
+/*
+ * Occurs when an enabled interface is being disabled.
+ */
+DEFINE_INTERFACE_EVENT(gb_interface_disable);
+
+#undef DEFINE_INTERFACE_EVENT
+
+DECLARE_EVENT_CLASS(gb_module,
+
+ TP_PROTO(struct gb_module *module),
+
+ TP_ARGS(module),
+
+ TP_STRUCT__entry(
+ __field(int, hd_bus_id)
+ __field(u8, module_id)
+ __field(size_t, num_interfaces)
+ __field(int, disconnected) /* bool */
+ ),
+
+ TP_fast_assign(
+ __entry->hd_bus_id = module->hd->bus_id;
+ __entry->module_id = module->module_id;
+ __entry->num_interfaces = module->num_interfaces;
+ __entry->disconnected = module->disconnected;
+ ),
+
+ TP_printk("hd_bus_id=%d module_id=%hhu num_interfaces=%zu disconnected=%d",
+ __entry->hd_bus_id, __entry->module_id,
+ __entry->num_interfaces, __entry->disconnected)
+);
+
+#define DEFINE_MODULE_EVENT(name) \
+ DEFINE_EVENT(gb_module, name, \
+ TP_PROTO(struct gb_module *module), \
+ TP_ARGS(module))
+
+/*
+ * Occurs after a new module is successfully created, before
+ * creating any of its interfaces.
+ */
+DEFINE_MODULE_EVENT(gb_module_create);
+
+/*
+ * Occurs after the last reference to a module has been dropped.
+ */
+DEFINE_MODULE_EVENT(gb_module_release);
+
+/*
+ * Occurs after a module is successfully created, before registering
+ * any of its interfaces.
+ */
+DEFINE_MODULE_EVENT(gb_module_add);
+
+/*
+ * Occurs when a module is deleted, before deregistering its
+ * interfaces.
+ */
+DEFINE_MODULE_EVENT(gb_module_del);
+
+#undef DEFINE_MODULE_EVENT
+
+DECLARE_EVENT_CLASS(gb_host_device,
+
+ TP_PROTO(struct gb_host_device *hd),
+
+ TP_ARGS(hd),
+
+ TP_STRUCT__entry(
+ __field(int, bus_id)
+ __field(size_t, num_cports)
+ __field(size_t, buffer_size_max)
+ ),
+
+ TP_fast_assign(
+ __entry->bus_id = hd->bus_id;
+ __entry->num_cports = hd->num_cports;
+ __entry->buffer_size_max = hd->buffer_size_max;
+ ),
+
+ TP_printk("bus_id=%d num_cports=%zu mtu=%zu",
+ __entry->bus_id, __entry->num_cports,
+ __entry->buffer_size_max)
+);
+
+#define DEFINE_HD_EVENT(name) \
+ DEFINE_EVENT(gb_host_device, name, \
+ TP_PROTO(struct gb_host_device *hd), \
+ TP_ARGS(hd))
+
+/*
+ * Occurs after a new host device is successfully created, before
+ * its SVC has been set up.
+ */
+DEFINE_HD_EVENT(gb_hd_create);
+
+/*
+ * Occurs after the last reference to a host device has been
+ * dropped.
+ */
+DEFINE_HD_EVENT(gb_hd_release);
+
+/*
+ * Occurs after a new host device has been added, after the
+ * connection to its SVC has been enabled.
+ */
+DEFINE_HD_EVENT(gb_hd_add);
+
+/*
+ * Occurs when a host device is being disconnected from the AP USB
+ * host controller.
+ */
+DEFINE_HD_EVENT(gb_hd_del);
+
+/*
+ * Occurs when a host device has passed received data to the Greybus
+ * core, after it has been determined it is destined for a valid
+ * CPort.
+ */
+DEFINE_HD_EVENT(gb_hd_in);
+
+#undef DEFINE_HD_EVENT
+
+/*
+ * Occurs on a TimeSync synchronization event or a TimeSync ping event.
+ */
+TRACE_EVENT(gb_timesync_irq,
+
+ TP_PROTO(u8 ping, u8 strobe, u8 count, u64 frame_time),
+
+ TP_ARGS(ping, strobe, count, frame_time),
+
+ TP_STRUCT__entry(
+ __field(u8, ping)
+ __field(u8, strobe)
+ __field(u8, count)
+ __field(u64, frame_time)
+ ),
+
+ TP_fast_assign(
+ __entry->ping = ping;
+ __entry->strobe = strobe;
+ __entry->count = count;
+ __entry->frame_time = frame_time;
+ ),
+
+ TP_printk("%s %d/%d frame-time %llu\n",
+ __entry->ping ? "ping" : "strobe", __entry->strobe,
+ __entry->count, __entry->frame_time)
+);
+
+#endif /* _TRACE_GREYBUS_H */
+
+/* This part must be outside protection */
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH .
+
+/*
+ * TRACE_INCLUDE_FILE is not needed if the filename and TRACE_SYSTEM are equal
+ */
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE greybus_trace
+#include <trace/define_trace.h>
+
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 06/32] greybus: svc driver/watchdog
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (3 preceding siblings ...)
2016-09-16 13:23 ` [patch 05/32] greybus: trace.h Greg KH
@ 2016-09-16 13:23 ` Greg KH
2016-09-16 13:23 ` [patch 07/32] greybus: core code Greg KH
` (25 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 13:23 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This is the Greybus SVC (a switch) driver. It includes a simple
watchdog that if it detects that there is no more network activity on
the Unipro system, userspace will be told to restart the system.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/svc.c | 1486 +++++++++++++++++++++++++++++++++++++++++
drivers/greybus/svc.h | 109 +++
drivers/greybus/svc_watchdog.c | 198 +++++
3 files changed, 1793 insertions(+)
--- /dev/null
+++ b/drivers/greybus/svc.c
@@ -0,0 +1,1486 @@
+/*
+ * SVC Greybus driver.
+ *
+ * Copyright 2015 Google Inc.
+ * Copyright 2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/debugfs.h>
+#include <linux/workqueue.h>
+
+#include "greybus.h"
+
+#define SVC_INTF_EJECT_TIMEOUT 9000
+#define SVC_INTF_ACTIVATE_TIMEOUT 6000
+#define SVC_INTF_RESUME_TIMEOUT 3000
+
+struct gb_svc_deferred_request {
+ struct work_struct work;
+ struct gb_operation *operation;
+};
+
+
+static int gb_svc_queue_deferred_request(struct gb_operation *operation);
+
+static ssize_t endo_id_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_svc *svc = to_gb_svc(dev);
+
+ return sprintf(buf, "0x%04x\n", svc->endo_id);
+}
+static DEVICE_ATTR_RO(endo_id);
+
+static ssize_t ap_intf_id_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_svc *svc = to_gb_svc(dev);
+
+ return sprintf(buf, "%u\n", svc->ap_intf_id);
+}
+static DEVICE_ATTR_RO(ap_intf_id);
+
+// FIXME
+// This is a hack, we need to do this "right" and clean the interface up
+// properly, not just forcibly yank the thing out of the system and hope for the
+// best. But for now, people want their modules to come out without having to
+// throw the thing to the ground or get out a screwdriver.
+static ssize_t intf_eject_store(struct device *dev,
+ struct device_attribute *attr, const char *buf,
+ size_t len)
+{
+ struct gb_svc *svc = to_gb_svc(dev);
+ unsigned short intf_id;
+ int ret;
+
+ ret = kstrtou16(buf, 10, &intf_id);
+ if (ret < 0)
+ return ret;
+
+ dev_warn(dev, "Forcibly trying to eject interface %d\n", intf_id);
+
+ ret = gb_svc_intf_eject(svc, intf_id);
+ if (ret < 0)
+ return ret;
+
+ return len;
+}
+static DEVICE_ATTR_WO(intf_eject);
+
+static ssize_t watchdog_show(struct device *dev, struct device_attribute *attr,
+ char *buf)
+{
+ struct gb_svc *svc = to_gb_svc(dev);
+
+ return sprintf(buf, "%s\n",
+ gb_svc_watchdog_enabled(svc) ? "enabled" : "disabled");
+}
+
+static ssize_t watchdog_store(struct device *dev,
+ struct device_attribute *attr, const char *buf,
+ size_t len)
+{
+ struct gb_svc *svc = to_gb_svc(dev);
+ int retval;
+ bool user_request;
+
+ retval = strtobool(buf, &user_request);
+ if (retval)
+ return retval;
+
+ if (user_request)
+ retval = gb_svc_watchdog_enable(svc);
+ else
+ retval = gb_svc_watchdog_disable(svc);
+ if (retval)
+ return retval;
+ return len;
+}
+static DEVICE_ATTR_RW(watchdog);
+
+static ssize_t watchdog_action_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_svc *svc = to_gb_svc(dev);
+
+ if (svc->action == GB_SVC_WATCHDOG_BITE_PANIC_KERNEL)
+ return sprintf(buf, "panic\n");
+ else if (svc->action == GB_SVC_WATCHDOG_BITE_RESET_UNIPRO)
+ return sprintf(buf, "reset\n");
+
+ return -EINVAL;
+}
+
+static ssize_t watchdog_action_store(struct device *dev,
+ struct device_attribute *attr,
+ const char *buf, size_t len)
+{
+ struct gb_svc *svc = to_gb_svc(dev);
+
+ if (sysfs_streq(buf, "panic"))
+ svc->action = GB_SVC_WATCHDOG_BITE_PANIC_KERNEL;
+ else if (sysfs_streq(buf, "reset"))
+ svc->action = GB_SVC_WATCHDOG_BITE_RESET_UNIPRO;
+ else
+ return -EINVAL;
+
+ return len;
+}
+static DEVICE_ATTR_RW(watchdog_action);
+
+static int gb_svc_pwrmon_rail_count_get(struct gb_svc *svc, u8 *value)
+{
+ struct gb_svc_pwrmon_rail_count_get_response response;
+ int ret;
+
+ ret = gb_operation_sync(svc->connection,
+ GB_SVC_TYPE_PWRMON_RAIL_COUNT_GET, NULL, 0,
+ &response, sizeof(response));
+ if (ret) {
+ dev_err(&svc->dev, "failed to get rail count: %d\n", ret);
+ return ret;
+ }
+
+ *value = response.rail_count;
+
+ return 0;
+}
+
+static int gb_svc_pwrmon_rail_names_get(struct gb_svc *svc,
+ struct gb_svc_pwrmon_rail_names_get_response *response,
+ size_t bufsize)
+{
+ int ret;
+
+ ret = gb_operation_sync(svc->connection,
+ GB_SVC_TYPE_PWRMON_RAIL_NAMES_GET, NULL, 0,
+ response, bufsize);
+ if (ret) {
+ dev_err(&svc->dev, "failed to get rail names: %d\n", ret);
+ return ret;
+ }
+
+ if (response->status != GB_SVC_OP_SUCCESS) {
+ dev_err(&svc->dev,
+ "SVC error while getting rail names: %u\n",
+ response->status);
+ return -EREMOTEIO;
+ }
+
+ return 0;
+}
+
+static int gb_svc_pwrmon_sample_get(struct gb_svc *svc, u8 rail_id,
+ u8 measurement_type, u32 *value)
+{
+ struct gb_svc_pwrmon_sample_get_request request;
+ struct gb_svc_pwrmon_sample_get_response response;
+ int ret;
+
+ request.rail_id = rail_id;
+ request.measurement_type = measurement_type;
+
+ ret = gb_operation_sync(svc->connection, GB_SVC_TYPE_PWRMON_SAMPLE_GET,
+ &request, sizeof(request),
+ &response, sizeof(response));
+ if (ret) {
+ dev_err(&svc->dev, "failed to get rail sample: %d\n", ret);
+ return ret;
+ }
+
+ if (response.result) {
+ dev_err(&svc->dev,
+ "UniPro error while getting rail power sample (%d %d): %d\n",
+ rail_id, measurement_type, response.result);
+ switch (response.result) {
+ case GB_SVC_PWRMON_GET_SAMPLE_INVAL:
+ return -EINVAL;
+ case GB_SVC_PWRMON_GET_SAMPLE_NOSUPP:
+ return -ENOMSG;
+ default:
+ return -EREMOTEIO;
+ }
+ }
+
+ *value = le32_to_cpu(response.measurement);
+
+ return 0;
+}
+
+int gb_svc_pwrmon_intf_sample_get(struct gb_svc *svc, u8 intf_id,
+ u8 measurement_type, u32 *value)
+{
+ struct gb_svc_pwrmon_intf_sample_get_request request;
+ struct gb_svc_pwrmon_intf_sample_get_response response;
+ int ret;
+
+ request.intf_id = intf_id;
+ request.measurement_type = measurement_type;
+
+ ret = gb_operation_sync(svc->connection,
+ GB_SVC_TYPE_PWRMON_INTF_SAMPLE_GET,
+ &request, sizeof(request),
+ &response, sizeof(response));
+ if (ret) {
+ dev_err(&svc->dev, "failed to get intf sample: %d\n", ret);
+ return ret;
+ }
+
+ if (response.result) {
+ dev_err(&svc->dev,
+ "UniPro error while getting intf power sample (%d %d): %d\n",
+ intf_id, measurement_type, response.result);
+ switch (response.result) {
+ case GB_SVC_PWRMON_GET_SAMPLE_INVAL:
+ return -EINVAL;
+ case GB_SVC_PWRMON_GET_SAMPLE_NOSUPP:
+ return -ENOMSG;
+ default:
+ return -EREMOTEIO;
+ }
+ }
+
+ *value = le32_to_cpu(response.measurement);
+
+ return 0;
+}
+
+static struct attribute *svc_attrs[] = {
+ &dev_attr_endo_id.attr,
+ &dev_attr_ap_intf_id.attr,
+ &dev_attr_intf_eject.attr,
+ &dev_attr_watchdog.attr,
+ &dev_attr_watchdog_action.attr,
+ NULL,
+};
+ATTRIBUTE_GROUPS(svc);
+
+int gb_svc_intf_device_id(struct gb_svc *svc, u8 intf_id, u8 device_id)
+{
+ struct gb_svc_intf_device_id_request request;
+
+ request.intf_id = intf_id;
+ request.device_id = device_id;
+
+ return gb_operation_sync(svc->connection, GB_SVC_TYPE_INTF_DEVICE_ID,
+ &request, sizeof(request), NULL, 0);
+}
+
+int gb_svc_intf_eject(struct gb_svc *svc, u8 intf_id)
+{
+ struct gb_svc_intf_eject_request request;
+ int ret;
+
+ request.intf_id = intf_id;
+
+ /*
+ * The pulse width for module release in svc is long so we need to
+ * increase the timeout so the operation will not return to soon.
+ */
+ ret = gb_operation_sync_timeout(svc->connection,
+ GB_SVC_TYPE_INTF_EJECT, &request,
+ sizeof(request), NULL, 0,
+ SVC_INTF_EJECT_TIMEOUT);
+ if (ret) {
+ dev_err(&svc->dev, "failed to eject interface %u\n", intf_id);
+ return ret;
+ }
+
+ return 0;
+}
+
+int gb_svc_intf_vsys_set(struct gb_svc *svc, u8 intf_id, bool enable)
+{
+ struct gb_svc_intf_vsys_request request;
+ struct gb_svc_intf_vsys_response response;
+ int type, ret;
+
+ request.intf_id = intf_id;
+
+ if (enable)
+ type = GB_SVC_TYPE_INTF_VSYS_ENABLE;
+ else
+ type = GB_SVC_TYPE_INTF_VSYS_DISABLE;
+
+ ret = gb_operation_sync(svc->connection, type,
+ &request, sizeof(request),
+ &response, sizeof(response));
+ if (ret < 0)
+ return ret;
+ if (response.result_code != GB_SVC_INTF_VSYS_OK)
+ return -EREMOTEIO;
+ return 0;
+}
+
+int gb_svc_intf_refclk_set(struct gb_svc *svc, u8 intf_id, bool enable)
+{
+ struct gb_svc_intf_refclk_request request;
+ struct gb_svc_intf_refclk_response response;
+ int type, ret;
+
+ request.intf_id = intf_id;
+
+ if (enable)
+ type = GB_SVC_TYPE_INTF_REFCLK_ENABLE;
+ else
+ type = GB_SVC_TYPE_INTF_REFCLK_DISABLE;
+
+ ret = gb_operation_sync(svc->connection, type,
+ &request, sizeof(request),
+ &response, sizeof(response));
+ if (ret < 0)
+ return ret;
+ if (response.result_code != GB_SVC_INTF_REFCLK_OK)
+ return -EREMOTEIO;
+ return 0;
+}
+
+int gb_svc_intf_unipro_set(struct gb_svc *svc, u8 intf_id, bool enable)
+{
+ struct gb_svc_intf_unipro_request request;
+ struct gb_svc_intf_unipro_response response;
+ int type, ret;
+
+ request.intf_id = intf_id;
+
+ if (enable)
+ type = GB_SVC_TYPE_INTF_UNIPRO_ENABLE;
+ else
+ type = GB_SVC_TYPE_INTF_UNIPRO_DISABLE;
+
+ ret = gb_operation_sync(svc->connection, type,
+ &request, sizeof(request),
+ &response, sizeof(response));
+ if (ret < 0)
+ return ret;
+ if (response.result_code != GB_SVC_INTF_UNIPRO_OK)
+ return -EREMOTEIO;
+ return 0;
+}
+
+int gb_svc_intf_activate(struct gb_svc *svc, u8 intf_id, u8 *intf_type)
+{
+ struct gb_svc_intf_activate_request request;
+ struct gb_svc_intf_activate_response response;
+ int ret;
+
+ request.intf_id = intf_id;
+
+ ret = gb_operation_sync_timeout(svc->connection,
+ GB_SVC_TYPE_INTF_ACTIVATE,
+ &request, sizeof(request),
+ &response, sizeof(response),
+ SVC_INTF_ACTIVATE_TIMEOUT);
+ if (ret < 0)
+ return ret;
+ if (response.status != GB_SVC_OP_SUCCESS) {
+ dev_err(&svc->dev, "failed to activate interface %u: %u\n",
+ intf_id, response.status);
+ return -EREMOTEIO;
+ }
+
+ *intf_type = response.intf_type;
+
+ return 0;
+}
+
+int gb_svc_intf_resume(struct gb_svc *svc, u8 intf_id)
+{
+ struct gb_svc_intf_resume_request request;
+ struct gb_svc_intf_resume_response response;
+ int ret;
+
+ request.intf_id = intf_id;
+
+ ret = gb_operation_sync_timeout(svc->connection,
+ GB_SVC_TYPE_INTF_RESUME,
+ &request, sizeof(request),
+ &response, sizeof(response),
+ SVC_INTF_RESUME_TIMEOUT);
+ if (ret < 0) {
+ dev_err(&svc->dev, "failed to send interface resume %u: %d\n",
+ intf_id, ret);
+ return ret;
+ }
+
+ if (response.status != GB_SVC_OP_SUCCESS) {
+ dev_err(&svc->dev, "failed to resume interface %u: %u\n",
+ intf_id, response.status);
+ return -EREMOTEIO;
+ }
+
+ return 0;
+}
+
+int gb_svc_dme_peer_get(struct gb_svc *svc, u8 intf_id, u16 attr, u16 selector,
+ u32 *value)
+{
+ struct gb_svc_dme_peer_get_request request;
+ struct gb_svc_dme_peer_get_response response;
+ u16 result;
+ int ret;
+
+ request.intf_id = intf_id;
+ request.attr = cpu_to_le16(attr);
+ request.selector = cpu_to_le16(selector);
+
+ ret = gb_operation_sync(svc->connection, GB_SVC_TYPE_DME_PEER_GET,
+ &request, sizeof(request),
+ &response, sizeof(response));
+ if (ret) {
+ dev_err(&svc->dev, "failed to get DME attribute (%u 0x%04x %u): %d\n",
+ intf_id, attr, selector, ret);
+ return ret;
+ }
+
+ result = le16_to_cpu(response.result_code);
+ if (result) {
+ dev_err(&svc->dev, "UniPro error while getting DME attribute (%u 0x%04x %u): %u\n",
+ intf_id, attr, selector, result);
+ return -EREMOTEIO;
+ }
+
+ if (value)
+ *value = le32_to_cpu(response.attr_value);
+
+ return 0;
+}
+
+int gb_svc_dme_peer_set(struct gb_svc *svc, u8 intf_id, u16 attr, u16 selector,
+ u32 value)
+{
+ struct gb_svc_dme_peer_set_request request;
+ struct gb_svc_dme_peer_set_response response;
+ u16 result;
+ int ret;
+
+ request.intf_id = intf_id;
+ request.attr = cpu_to_le16(attr);
+ request.selector = cpu_to_le16(selector);
+ request.value = cpu_to_le32(value);
+
+ ret = gb_operation_sync(svc->connection, GB_SVC_TYPE_DME_PEER_SET,
+ &request, sizeof(request),
+ &response, sizeof(response));
+ if (ret) {
+ dev_err(&svc->dev, "failed to set DME attribute (%u 0x%04x %u %u): %d\n",
+ intf_id, attr, selector, value, ret);
+ return ret;
+ }
+
+ result = le16_to_cpu(response.result_code);
+ if (result) {
+ dev_err(&svc->dev, "UniPro error while setting DME attribute (%u 0x%04x %u %u): %u\n",
+ intf_id, attr, selector, value, result);
+ return -EREMOTEIO;
+ }
+
+ return 0;
+}
+
+int gb_svc_connection_create(struct gb_svc *svc,
+ u8 intf1_id, u16 cport1_id,
+ u8 intf2_id, u16 cport2_id,
+ u8 cport_flags)
+{
+ struct gb_svc_conn_create_request request;
+
+ request.intf1_id = intf1_id;
+ request.cport1_id = cpu_to_le16(cport1_id);
+ request.intf2_id = intf2_id;
+ request.cport2_id = cpu_to_le16(cport2_id);
+ request.tc = 0; /* TC0 */
+ request.flags = cport_flags;
+
+ return gb_operation_sync(svc->connection, GB_SVC_TYPE_CONN_CREATE,
+ &request, sizeof(request), NULL, 0);
+}
+
+void gb_svc_connection_destroy(struct gb_svc *svc, u8 intf1_id, u16 cport1_id,
+ u8 intf2_id, u16 cport2_id)
+{
+ struct gb_svc_conn_destroy_request request;
+ struct gb_connection *connection = svc->connection;
+ int ret;
+
+ request.intf1_id = intf1_id;
+ request.cport1_id = cpu_to_le16(cport1_id);
+ request.intf2_id = intf2_id;
+ request.cport2_id = cpu_to_le16(cport2_id);
+
+ ret = gb_operation_sync(connection, GB_SVC_TYPE_CONN_DESTROY,
+ &request, sizeof(request), NULL, 0);
+ if (ret) {
+ dev_err(&svc->dev, "failed to destroy connection (%u:%u %u:%u): %d\n",
+ intf1_id, cport1_id, intf2_id, cport2_id, ret);
+ }
+}
+
+int gb_svc_timesync_enable(struct gb_svc *svc, u8 count, u64 frame_time,
+ u32 strobe_delay, u32 refclk)
+{
+ struct gb_connection *connection = svc->connection;
+ struct gb_svc_timesync_enable_request request;
+
+ request.count = count;
+ request.frame_time = cpu_to_le64(frame_time);
+ request.strobe_delay = cpu_to_le32(strobe_delay);
+ request.refclk = cpu_to_le32(refclk);
+ return gb_operation_sync(connection,
+ GB_SVC_TYPE_TIMESYNC_ENABLE,
+ &request, sizeof(request), NULL, 0);
+}
+
+int gb_svc_timesync_disable(struct gb_svc *svc)
+{
+ struct gb_connection *connection = svc->connection;
+
+ return gb_operation_sync(connection,
+ GB_SVC_TYPE_TIMESYNC_DISABLE,
+ NULL, 0, NULL, 0);
+}
+
+int gb_svc_timesync_authoritative(struct gb_svc *svc, u64 *frame_time)
+{
+ struct gb_connection *connection = svc->connection;
+ struct gb_svc_timesync_authoritative_response response;
+ int ret, i;
+
+ ret = gb_operation_sync(connection,
+ GB_SVC_TYPE_TIMESYNC_AUTHORITATIVE, NULL, 0,
+ &response, sizeof(response));
+ if (ret < 0)
+ return ret;
+
+ for (i = 0; i < GB_TIMESYNC_MAX_STROBES; i++)
+ frame_time[i] = le64_to_cpu(response.frame_time[i]);
+ return 0;
+}
+
+int gb_svc_timesync_ping(struct gb_svc *svc, u64 *frame_time)
+{
+ struct gb_connection *connection = svc->connection;
+ struct gb_svc_timesync_ping_response response;
+ int ret;
+
+ ret = gb_operation_sync(connection,
+ GB_SVC_TYPE_TIMESYNC_PING,
+ NULL, 0,
+ &response, sizeof(response));
+ if (ret < 0)
+ return ret;
+
+ *frame_time = le64_to_cpu(response.frame_time);
+ return 0;
+}
+
+int gb_svc_timesync_wake_pins_acquire(struct gb_svc *svc, u32 strobe_mask)
+{
+ struct gb_connection *connection = svc->connection;
+ struct gb_svc_timesync_wake_pins_acquire_request request;
+
+ request.strobe_mask = cpu_to_le32(strobe_mask);
+ return gb_operation_sync(connection,
+ GB_SVC_TYPE_TIMESYNC_WAKE_PINS_ACQUIRE,
+ &request, sizeof(request),
+ NULL, 0);
+}
+
+int gb_svc_timesync_wake_pins_release(struct gb_svc *svc)
+{
+ struct gb_connection *connection = svc->connection;
+
+ return gb_operation_sync(connection,
+ GB_SVC_TYPE_TIMESYNC_WAKE_PINS_RELEASE,
+ NULL, 0, NULL, 0);
+}
+
+/* Creates bi-directional routes between the devices */
+int gb_svc_route_create(struct gb_svc *svc, u8 intf1_id, u8 dev1_id,
+ u8 intf2_id, u8 dev2_id)
+{
+ struct gb_svc_route_create_request request;
+
+ request.intf1_id = intf1_id;
+ request.dev1_id = dev1_id;
+ request.intf2_id = intf2_id;
+ request.dev2_id = dev2_id;
+
+ return gb_operation_sync(svc->connection, GB_SVC_TYPE_ROUTE_CREATE,
+ &request, sizeof(request), NULL, 0);
+}
+
+/* Destroys bi-directional routes between the devices */
+void gb_svc_route_destroy(struct gb_svc *svc, u8 intf1_id, u8 intf2_id)
+{
+ struct gb_svc_route_destroy_request request;
+ int ret;
+
+ request.intf1_id = intf1_id;
+ request.intf2_id = intf2_id;
+
+ ret = gb_operation_sync(svc->connection, GB_SVC_TYPE_ROUTE_DESTROY,
+ &request, sizeof(request), NULL, 0);
+ if (ret) {
+ dev_err(&svc->dev, "failed to destroy route (%u %u): %d\n",
+ intf1_id, intf2_id, ret);
+ }
+}
+
+int gb_svc_intf_set_power_mode(struct gb_svc *svc, u8 intf_id, u8 hs_series,
+ u8 tx_mode, u8 tx_gear, u8 tx_nlanes,
+ u8 tx_amplitude, u8 tx_hs_equalizer,
+ u8 rx_mode, u8 rx_gear, u8 rx_nlanes,
+ u8 flags, u32 quirks,
+ struct gb_svc_l2_timer_cfg *local,
+ struct gb_svc_l2_timer_cfg *remote)
+{
+ struct gb_svc_intf_set_pwrm_request request;
+ struct gb_svc_intf_set_pwrm_response response;
+ int ret;
+ u16 result_code;
+
+ memset(&request, 0, sizeof(request));
+
+ request.intf_id = intf_id;
+ request.hs_series = hs_series;
+ request.tx_mode = tx_mode;
+ request.tx_gear = tx_gear;
+ request.tx_nlanes = tx_nlanes;
+ request.tx_amplitude = tx_amplitude;
+ request.tx_hs_equalizer = tx_hs_equalizer;
+ request.rx_mode = rx_mode;
+ request.rx_gear = rx_gear;
+ request.rx_nlanes = rx_nlanes;
+ request.flags = flags;
+ request.quirks = cpu_to_le32(quirks);
+ if (local)
+ request.local_l2timerdata = *local;
+ if (remote)
+ request.remote_l2timerdata = *remote;
+
+ ret = gb_operation_sync(svc->connection, GB_SVC_TYPE_INTF_SET_PWRM,
+ &request, sizeof(request),
+ &response, sizeof(response));
+ if (ret < 0)
+ return ret;
+
+ result_code = response.result_code;
+ if (result_code != GB_SVC_SETPWRM_PWR_LOCAL) {
+ dev_err(&svc->dev, "set power mode = %d\n", result_code);
+ return -EIO;
+ }
+
+ return 0;
+}
+EXPORT_SYMBOL_GPL(gb_svc_intf_set_power_mode);
+
+int gb_svc_intf_set_power_mode_hibernate(struct gb_svc *svc, u8 intf_id)
+{
+ struct gb_svc_intf_set_pwrm_request request;
+ struct gb_svc_intf_set_pwrm_response response;
+ int ret;
+ u16 result_code;
+
+ memset(&request, 0, sizeof(request));
+
+ request.intf_id = intf_id;
+ request.hs_series = GB_SVC_UNIPRO_HS_SERIES_A;
+ request.tx_mode = GB_SVC_UNIPRO_HIBERNATE_MODE;
+ request.rx_mode = GB_SVC_UNIPRO_HIBERNATE_MODE;
+
+ ret = gb_operation_sync(svc->connection, GB_SVC_TYPE_INTF_SET_PWRM,
+ &request, sizeof(request),
+ &response, sizeof(response));
+ if (ret < 0) {
+ dev_err(&svc->dev,
+ "failed to send set power mode operation to interface %u: %d\n",
+ intf_id, ret);
+ return ret;
+ }
+
+ result_code = response.result_code;
+ if (result_code != GB_SVC_SETPWRM_PWR_OK) {
+ dev_err(&svc->dev,
+ "failed to hibernate the link for interface %u: %u\n",
+ intf_id, result_code);
+ return -EIO;
+ }
+
+ return 0;
+}
+
+int gb_svc_ping(struct gb_svc *svc)
+{
+ return gb_operation_sync_timeout(svc->connection, GB_SVC_TYPE_PING,
+ NULL, 0, NULL, 0,
+ GB_OPERATION_TIMEOUT_DEFAULT * 2);
+}
+
+static int gb_svc_version_request(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct gb_svc *svc = gb_connection_get_data(connection);
+ struct gb_svc_version_request *request;
+ struct gb_svc_version_response *response;
+
+ if (op->request->payload_size < sizeof(*request)) {
+ dev_err(&svc->dev, "short version request (%zu < %zu)\n",
+ op->request->payload_size,
+ sizeof(*request));
+ return -EINVAL;
+ }
+
+ request = op->request->payload;
+
+ if (request->major > GB_SVC_VERSION_MAJOR) {
+ dev_warn(&svc->dev, "unsupported major version (%u > %u)\n",
+ request->major, GB_SVC_VERSION_MAJOR);
+ return -ENOTSUPP;
+ }
+
+ svc->protocol_major = request->major;
+ svc->protocol_minor = request->minor;
+
+ if (!gb_operation_response_alloc(op, sizeof(*response), GFP_KERNEL))
+ return -ENOMEM;
+
+ response = op->response->payload;
+ response->major = svc->protocol_major;
+ response->minor = svc->protocol_minor;
+
+ return 0;
+}
+
+static ssize_t pwr_debugfs_voltage_read(struct file *file, char __user *buf,
+ size_t len, loff_t *offset)
+{
+ struct svc_debugfs_pwrmon_rail *pwrmon_rails = file->f_inode->i_private;
+ struct gb_svc *svc = pwrmon_rails->svc;
+ int ret, desc;
+ u32 value;
+ char buff[16];
+
+ ret = gb_svc_pwrmon_sample_get(svc, pwrmon_rails->id,
+ GB_SVC_PWRMON_TYPE_VOL, &value);
+ if (ret) {
+ dev_err(&svc->dev,
+ "failed to get voltage sample %u: %d\n",
+ pwrmon_rails->id, ret);
+ return ret;
+ }
+
+ desc = scnprintf(buff, sizeof(buff), "%u\n", value);
+
+ return simple_read_from_buffer(buf, len, offset, buff, desc);
+}
+
+static ssize_t pwr_debugfs_current_read(struct file *file, char __user *buf,
+ size_t len, loff_t *offset)
+{
+ struct svc_debugfs_pwrmon_rail *pwrmon_rails = file->f_inode->i_private;
+ struct gb_svc *svc = pwrmon_rails->svc;
+ int ret, desc;
+ u32 value;
+ char buff[16];
+
+ ret = gb_svc_pwrmon_sample_get(svc, pwrmon_rails->id,
+ GB_SVC_PWRMON_TYPE_CURR, &value);
+ if (ret) {
+ dev_err(&svc->dev,
+ "failed to get current sample %u: %d\n",
+ pwrmon_rails->id, ret);
+ return ret;
+ }
+
+ desc = scnprintf(buff, sizeof(buff), "%u\n", value);
+
+ return simple_read_from_buffer(buf, len, offset, buff, desc);
+}
+
+static ssize_t pwr_debugfs_power_read(struct file *file, char __user *buf,
+ size_t len, loff_t *offset)
+{
+ struct svc_debugfs_pwrmon_rail *pwrmon_rails = file->f_inode->i_private;
+ struct gb_svc *svc = pwrmon_rails->svc;
+ int ret, desc;
+ u32 value;
+ char buff[16];
+
+ ret = gb_svc_pwrmon_sample_get(svc, pwrmon_rails->id,
+ GB_SVC_PWRMON_TYPE_PWR, &value);
+ if (ret) {
+ dev_err(&svc->dev, "failed to get power sample %u: %d\n",
+ pwrmon_rails->id, ret);
+ return ret;
+ }
+
+ desc = scnprintf(buff, sizeof(buff), "%u\n", value);
+
+ return simple_read_from_buffer(buf, len, offset, buff, desc);
+}
+
+static const struct file_operations pwrmon_debugfs_voltage_fops = {
+ .read = pwr_debugfs_voltage_read,
+};
+
+static const struct file_operations pwrmon_debugfs_current_fops = {
+ .read = pwr_debugfs_current_read,
+};
+
+static const struct file_operations pwrmon_debugfs_power_fops = {
+ .read = pwr_debugfs_power_read,
+};
+
+static void gb_svc_pwrmon_debugfs_init(struct gb_svc *svc)
+{
+ int i;
+ size_t bufsize;
+ struct dentry *dent;
+ struct gb_svc_pwrmon_rail_names_get_response *rail_names;
+ u8 rail_count;
+
+ dent = debugfs_create_dir("pwrmon", svc->debugfs_dentry);
+ if (IS_ERR_OR_NULL(dent))
+ return;
+
+ if (gb_svc_pwrmon_rail_count_get(svc, &rail_count))
+ goto err_pwrmon_debugfs;
+
+ if (!rail_count || rail_count > GB_SVC_PWRMON_MAX_RAIL_COUNT)
+ goto err_pwrmon_debugfs;
+
+ bufsize = sizeof(*rail_names) +
+ GB_SVC_PWRMON_RAIL_NAME_BUFSIZE * rail_count;
+
+ rail_names = kzalloc(bufsize, GFP_KERNEL);
+ if (!rail_names)
+ goto err_pwrmon_debugfs;
+
+ svc->pwrmon_rails = kcalloc(rail_count, sizeof(*svc->pwrmon_rails),
+ GFP_KERNEL);
+ if (!svc->pwrmon_rails)
+ goto err_pwrmon_debugfs_free;
+
+ if (gb_svc_pwrmon_rail_names_get(svc, rail_names, bufsize))
+ goto err_pwrmon_debugfs_free;
+
+ for (i = 0; i < rail_count; i++) {
+ struct dentry *dir;
+ struct svc_debugfs_pwrmon_rail *rail = &svc->pwrmon_rails[i];
+ char fname[GB_SVC_PWRMON_RAIL_NAME_BUFSIZE];
+
+ snprintf(fname, sizeof(fname), "%s",
+ (char *)&rail_names->name[i]);
+
+ rail->id = i;
+ rail->svc = svc;
+
+ dir = debugfs_create_dir(fname, dent);
+ debugfs_create_file("voltage_now", S_IRUGO, dir, rail,
+ &pwrmon_debugfs_voltage_fops);
+ debugfs_create_file("current_now", S_IRUGO, dir, rail,
+ &pwrmon_debugfs_current_fops);
+ debugfs_create_file("power_now", S_IRUGO, dir, rail,
+ &pwrmon_debugfs_power_fops);
+ }
+
+ kfree(rail_names);
+ return;
+
+err_pwrmon_debugfs_free:
+ kfree(rail_names);
+ kfree(svc->pwrmon_rails);
+ svc->pwrmon_rails = NULL;
+
+err_pwrmon_debugfs:
+ debugfs_remove(dent);
+}
+
+static void gb_svc_debugfs_init(struct gb_svc *svc)
+{
+ svc->debugfs_dentry = debugfs_create_dir(dev_name(&svc->dev),
+ gb_debugfs_get());
+ gb_svc_pwrmon_debugfs_init(svc);
+}
+
+static void gb_svc_debugfs_exit(struct gb_svc *svc)
+{
+ debugfs_remove_recursive(svc->debugfs_dentry);
+ kfree(svc->pwrmon_rails);
+ svc->pwrmon_rails = NULL;
+}
+
+static int gb_svc_hello(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct gb_svc *svc = gb_connection_get_data(connection);
+ struct gb_svc_hello_request *hello_request;
+ int ret;
+
+ if (op->request->payload_size < sizeof(*hello_request)) {
+ dev_warn(&svc->dev, "short hello request (%zu < %zu)\n",
+ op->request->payload_size,
+ sizeof(*hello_request));
+ return -EINVAL;
+ }
+
+ hello_request = op->request->payload;
+ svc->endo_id = le16_to_cpu(hello_request->endo_id);
+ svc->ap_intf_id = hello_request->interface_id;
+
+ ret = device_add(&svc->dev);
+ if (ret) {
+ dev_err(&svc->dev, "failed to register svc device: %d\n", ret);
+ return ret;
+ }
+
+ ret = gb_svc_watchdog_create(svc);
+ if (ret) {
+ dev_err(&svc->dev, "failed to create watchdog: %d\n", ret);
+ goto err_unregister_device;
+ }
+
+ gb_svc_debugfs_init(svc);
+
+ ret = gb_timesync_svc_add(svc);
+ if (ret) {
+ dev_err(&svc->dev, "failed to add SVC to timesync: %d\n", ret);
+ gb_svc_debugfs_exit(svc);
+ goto err_unregister_device;
+ }
+
+ return gb_svc_queue_deferred_request(op);
+
+err_unregister_device:
+ gb_svc_watchdog_destroy(svc);
+ device_del(&svc->dev);
+ return ret;
+}
+
+static struct gb_interface *gb_svc_interface_lookup(struct gb_svc *svc,
+ u8 intf_id)
+{
+ struct gb_host_device *hd = svc->hd;
+ struct gb_module *module;
+ size_t num_interfaces;
+ u8 module_id;
+
+ list_for_each_entry(module, &hd->modules, hd_node) {
+ module_id = module->module_id;
+ num_interfaces = module->num_interfaces;
+
+ if (intf_id >= module_id &&
+ intf_id < module_id + num_interfaces) {
+ return module->interfaces[intf_id - module_id];
+ }
+ }
+
+ return NULL;
+}
+
+static struct gb_module *gb_svc_module_lookup(struct gb_svc *svc, u8 module_id)
+{
+ struct gb_host_device *hd = svc->hd;
+ struct gb_module *module;
+
+ list_for_each_entry(module, &hd->modules, hd_node) {
+ if (module->module_id == module_id)
+ return module;
+ }
+
+ return NULL;
+}
+
+static void gb_svc_process_hello_deferred(struct gb_operation *operation)
+{
+ struct gb_connection *connection = operation->connection;
+ struct gb_svc *svc = gb_connection_get_data(connection);
+ int ret;
+
+ /*
+ * XXX This is a hack/work-around to reconfigure the APBridgeA-Switch
+ * link to PWM G2, 1 Lane, Slow Auto, so that it has sufficient
+ * bandwidth for 3 audio streams plus boot-over-UniPro of a hot-plugged
+ * module.
+ *
+ * The code should be removed once SW-2217, Heuristic for UniPro
+ * Power Mode Changes is resolved.
+ */
+ ret = gb_svc_intf_set_power_mode(svc, svc->ap_intf_id,
+ GB_SVC_UNIPRO_HS_SERIES_A,
+ GB_SVC_UNIPRO_SLOW_AUTO_MODE,
+ 2, 1,
+ GB_SVC_SMALL_AMPLITUDE, GB_SVC_NO_DE_EMPHASIS,
+ GB_SVC_UNIPRO_SLOW_AUTO_MODE,
+ 2, 1,
+ 0, 0,
+ NULL, NULL);
+
+ if (ret)
+ dev_warn(&svc->dev,
+ "power mode change failed on AP to switch link: %d\n",
+ ret);
+}
+
+static void gb_svc_process_module_inserted(struct gb_operation *operation)
+{
+ struct gb_svc_module_inserted_request *request;
+ struct gb_connection *connection = operation->connection;
+ struct gb_svc *svc = gb_connection_get_data(connection);
+ struct gb_host_device *hd = svc->hd;
+ struct gb_module *module;
+ size_t num_interfaces;
+ u8 module_id;
+ u16 flags;
+ int ret;
+
+ /* The request message size has already been verified. */
+ request = operation->request->payload;
+ module_id = request->primary_intf_id;
+ num_interfaces = request->intf_count;
+ flags = le16_to_cpu(request->flags);
+
+ dev_dbg(&svc->dev, "%s - id = %u, num_interfaces = %zu, flags = 0x%04x\n",
+ __func__, module_id, num_interfaces, flags);
+
+ if (flags & GB_SVC_MODULE_INSERTED_FLAG_NO_PRIMARY) {
+ dev_warn(&svc->dev, "no primary interface detected on module %u\n",
+ module_id);
+ }
+
+ module = gb_svc_module_lookup(svc, module_id);
+ if (module) {
+ dev_warn(&svc->dev, "unexpected module-inserted event %u\n",
+ module_id);
+ return;
+ }
+
+ module = gb_module_create(hd, module_id, num_interfaces);
+ if (!module) {
+ dev_err(&svc->dev, "failed to create module\n");
+ return;
+ }
+
+ ret = gb_module_add(module);
+ if (ret) {
+ gb_module_put(module);
+ return;
+ }
+
+ list_add(&module->hd_node, &hd->modules);
+}
+
+static void gb_svc_process_module_removed(struct gb_operation *operation)
+{
+ struct gb_svc_module_removed_request *request;
+ struct gb_connection *connection = operation->connection;
+ struct gb_svc *svc = gb_connection_get_data(connection);
+ struct gb_module *module;
+ u8 module_id;
+
+ /* The request message size has already been verified. */
+ request = operation->request->payload;
+ module_id = request->primary_intf_id;
+
+ dev_dbg(&svc->dev, "%s - id = %u\n", __func__, module_id);
+
+ module = gb_svc_module_lookup(svc, module_id);
+ if (!module) {
+ dev_warn(&svc->dev, "unexpected module-removed event %u\n",
+ module_id);
+ return;
+ }
+
+ module->disconnected = true;
+
+ gb_module_del(module);
+ list_del(&module->hd_node);
+ gb_module_put(module);
+}
+
+static void gb_svc_process_intf_oops(struct gb_operation *operation)
+{
+ struct gb_svc_intf_oops_request *request;
+ struct gb_connection *connection = operation->connection;
+ struct gb_svc *svc = gb_connection_get_data(connection);
+ struct gb_interface *intf;
+ u8 intf_id;
+ u8 reason;
+
+ /* The request message size has already been verified. */
+ request = operation->request->payload;
+ intf_id = request->intf_id;
+ reason = request->reason;
+
+ intf = gb_svc_interface_lookup(svc, intf_id);
+ if (!intf) {
+ dev_warn(&svc->dev, "unexpected interface-oops event %u\n",
+ intf_id);
+ return;
+ }
+
+ dev_info(&svc->dev, "Deactivating interface %u, interface oops reason = %u\n",
+ intf_id, reason);
+
+ mutex_lock(&intf->mutex);
+ intf->disconnected = true;
+ gb_interface_disable(intf);
+ gb_interface_deactivate(intf);
+ mutex_unlock(&intf->mutex);
+}
+
+static void gb_svc_process_intf_mailbox_event(struct gb_operation *operation)
+{
+ struct gb_svc_intf_mailbox_event_request *request;
+ struct gb_connection *connection = operation->connection;
+ struct gb_svc *svc = gb_connection_get_data(connection);
+ struct gb_interface *intf;
+ u8 intf_id;
+ u16 result_code;
+ u32 mailbox;
+
+ /* The request message size has already been verified. */
+ request = operation->request->payload;
+ intf_id = request->intf_id;
+ result_code = le16_to_cpu(request->result_code);
+ mailbox = le32_to_cpu(request->mailbox);
+
+ dev_dbg(&svc->dev, "%s - id = %u, result = 0x%04x, mailbox = 0x%08x\n",
+ __func__, intf_id, result_code, mailbox);
+
+ intf = gb_svc_interface_lookup(svc, intf_id);
+ if (!intf) {
+ dev_warn(&svc->dev, "unexpected mailbox event %u\n", intf_id);
+ return;
+ }
+
+ gb_interface_mailbox_event(intf, result_code, mailbox);
+}
+
+static void gb_svc_process_deferred_request(struct work_struct *work)
+{
+ struct gb_svc_deferred_request *dr;
+ struct gb_operation *operation;
+ struct gb_svc *svc;
+ u8 type;
+
+ dr = container_of(work, struct gb_svc_deferred_request, work);
+ operation = dr->operation;
+ svc = gb_connection_get_data(operation->connection);
+ type = operation->request->header->type;
+
+ switch (type) {
+ case GB_SVC_TYPE_SVC_HELLO:
+ gb_svc_process_hello_deferred(operation);
+ break;
+ case GB_SVC_TYPE_MODULE_INSERTED:
+ gb_svc_process_module_inserted(operation);
+ break;
+ case GB_SVC_TYPE_MODULE_REMOVED:
+ gb_svc_process_module_removed(operation);
+ break;
+ case GB_SVC_TYPE_INTF_MAILBOX_EVENT:
+ gb_svc_process_intf_mailbox_event(operation);
+ break;
+ case GB_SVC_TYPE_INTF_OOPS:
+ gb_svc_process_intf_oops(operation);
+ break;
+ default:
+ dev_err(&svc->dev, "bad deferred request type: 0x%02x\n", type);
+ }
+
+ gb_operation_put(operation);
+ kfree(dr);
+}
+
+static int gb_svc_queue_deferred_request(struct gb_operation *operation)
+{
+ struct gb_svc *svc = gb_connection_get_data(operation->connection);
+ struct gb_svc_deferred_request *dr;
+
+ dr = kmalloc(sizeof(*dr), GFP_KERNEL);
+ if (!dr)
+ return -ENOMEM;
+
+ gb_operation_get(operation);
+
+ dr->operation = operation;
+ INIT_WORK(&dr->work, gb_svc_process_deferred_request);
+
+ queue_work(svc->wq, &dr->work);
+
+ return 0;
+}
+
+static int gb_svc_intf_reset_recv(struct gb_operation *op)
+{
+ struct gb_svc *svc = gb_connection_get_data(op->connection);
+ struct gb_message *request = op->request;
+ struct gb_svc_intf_reset_request *reset;
+ u8 intf_id;
+
+ if (request->payload_size < sizeof(*reset)) {
+ dev_warn(&svc->dev, "short reset request received (%zu < %zu)\n",
+ request->payload_size, sizeof(*reset));
+ return -EINVAL;
+ }
+ reset = request->payload;
+
+ intf_id = reset->intf_id;
+
+ /* FIXME Reset the interface here */
+
+ return 0;
+}
+
+static int gb_svc_module_inserted_recv(struct gb_operation *op)
+{
+ struct gb_svc *svc = gb_connection_get_data(op->connection);
+ struct gb_svc_module_inserted_request *request;
+
+ if (op->request->payload_size < sizeof(*request)) {
+ dev_warn(&svc->dev, "short module-inserted request received (%zu < %zu)\n",
+ op->request->payload_size, sizeof(*request));
+ return -EINVAL;
+ }
+
+ request = op->request->payload;
+
+ dev_dbg(&svc->dev, "%s - id = %u\n", __func__,
+ request->primary_intf_id);
+
+ return gb_svc_queue_deferred_request(op);
+}
+
+static int gb_svc_module_removed_recv(struct gb_operation *op)
+{
+ struct gb_svc *svc = gb_connection_get_data(op->connection);
+ struct gb_svc_module_removed_request *request;
+
+ if (op->request->payload_size < sizeof(*request)) {
+ dev_warn(&svc->dev, "short module-removed request received (%zu < %zu)\n",
+ op->request->payload_size, sizeof(*request));
+ return -EINVAL;
+ }
+
+ request = op->request->payload;
+
+ dev_dbg(&svc->dev, "%s - id = %u\n", __func__,
+ request->primary_intf_id);
+
+ return gb_svc_queue_deferred_request(op);
+}
+
+static int gb_svc_intf_oops_recv(struct gb_operation *op)
+{
+ struct gb_svc *svc = gb_connection_get_data(op->connection);
+ struct gb_svc_intf_oops_request *request;
+
+ if (op->request->payload_size < sizeof(*request)) {
+ dev_warn(&svc->dev, "short intf-oops request received (%zu < %zu)\n",
+ op->request->payload_size, sizeof(*request));
+ return -EINVAL;
+ }
+
+ return gb_svc_queue_deferred_request(op);
+}
+
+static int gb_svc_intf_mailbox_event_recv(struct gb_operation *op)
+{
+ struct gb_svc *svc = gb_connection_get_data(op->connection);
+ struct gb_svc_intf_mailbox_event_request *request;
+
+ if (op->request->payload_size < sizeof(*request)) {
+ dev_warn(&svc->dev, "short mailbox request received (%zu < %zu)\n",
+ op->request->payload_size, sizeof(*request));
+ return -EINVAL;
+ }
+
+ request = op->request->payload;
+
+ dev_dbg(&svc->dev, "%s - id = %u\n", __func__, request->intf_id);
+
+ return gb_svc_queue_deferred_request(op);
+}
+
+static int gb_svc_request_handler(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct gb_svc *svc = gb_connection_get_data(connection);
+ u8 type = op->type;
+ int ret = 0;
+
+ /*
+ * SVC requests need to follow a specific order (at least initially) and
+ * below code takes care of enforcing that. The expected order is:
+ * - PROTOCOL_VERSION
+ * - SVC_HELLO
+ * - Any other request, but the earlier two.
+ *
+ * Incoming requests are guaranteed to be serialized and so we don't
+ * need to protect 'state' for any races.
+ */
+ switch (type) {
+ case GB_SVC_TYPE_PROTOCOL_VERSION:
+ if (svc->state != GB_SVC_STATE_RESET)
+ ret = -EINVAL;
+ break;
+ case GB_SVC_TYPE_SVC_HELLO:
+ if (svc->state != GB_SVC_STATE_PROTOCOL_VERSION)
+ ret = -EINVAL;
+ break;
+ default:
+ if (svc->state != GB_SVC_STATE_SVC_HELLO)
+ ret = -EINVAL;
+ break;
+ }
+
+ if (ret) {
+ dev_warn(&svc->dev, "unexpected request 0x%02x received (state %u)\n",
+ type, svc->state);
+ return ret;
+ }
+
+ switch (type) {
+ case GB_SVC_TYPE_PROTOCOL_VERSION:
+ ret = gb_svc_version_request(op);
+ if (!ret)
+ svc->state = GB_SVC_STATE_PROTOCOL_VERSION;
+ return ret;
+ case GB_SVC_TYPE_SVC_HELLO:
+ ret = gb_svc_hello(op);
+ if (!ret)
+ svc->state = GB_SVC_STATE_SVC_HELLO;
+ return ret;
+ case GB_SVC_TYPE_INTF_RESET:
+ return gb_svc_intf_reset_recv(op);
+ case GB_SVC_TYPE_MODULE_INSERTED:
+ return gb_svc_module_inserted_recv(op);
+ case GB_SVC_TYPE_MODULE_REMOVED:
+ return gb_svc_module_removed_recv(op);
+ case GB_SVC_TYPE_INTF_MAILBOX_EVENT:
+ return gb_svc_intf_mailbox_event_recv(op);
+ case GB_SVC_TYPE_INTF_OOPS:
+ return gb_svc_intf_oops_recv(op);
+ default:
+ dev_warn(&svc->dev, "unsupported request 0x%02x\n", type);
+ return -EINVAL;
+ }
+}
+
+static void gb_svc_release(struct device *dev)
+{
+ struct gb_svc *svc = to_gb_svc(dev);
+
+ if (svc->connection)
+ gb_connection_destroy(svc->connection);
+ ida_destroy(&svc->device_id_map);
+ destroy_workqueue(svc->wq);
+ kfree(svc);
+}
+
+struct device_type greybus_svc_type = {
+ .name = "greybus_svc",
+ .release = gb_svc_release,
+};
+
+struct gb_svc *gb_svc_create(struct gb_host_device *hd)
+{
+ struct gb_svc *svc;
+
+ svc = kzalloc(sizeof(*svc), GFP_KERNEL);
+ if (!svc)
+ return NULL;
+
+ svc->wq = alloc_workqueue("%s:svc", WQ_UNBOUND, 1, dev_name(&hd->dev));
+ if (!svc->wq) {
+ kfree(svc);
+ return NULL;
+ }
+
+ svc->dev.parent = &hd->dev;
+ svc->dev.bus = &greybus_bus_type;
+ svc->dev.type = &greybus_svc_type;
+ svc->dev.groups = svc_groups;
+ svc->dev.dma_mask = svc->dev.parent->dma_mask;
+ device_initialize(&svc->dev);
+
+ dev_set_name(&svc->dev, "%d-svc", hd->bus_id);
+
+ ida_init(&svc->device_id_map);
+ svc->state = GB_SVC_STATE_RESET;
+ svc->hd = hd;
+
+ svc->connection = gb_connection_create_static(hd, GB_SVC_CPORT_ID,
+ gb_svc_request_handler);
+ if (IS_ERR(svc->connection)) {
+ dev_err(&svc->dev, "failed to create connection: %ld\n",
+ PTR_ERR(svc->connection));
+ goto err_put_device;
+ }
+
+ gb_connection_set_data(svc->connection, svc);
+
+ return svc;
+
+err_put_device:
+ put_device(&svc->dev);
+ return NULL;
+}
+
+int gb_svc_add(struct gb_svc *svc)
+{
+ int ret;
+
+ /*
+ * The SVC protocol is currently driven by the SVC, so the SVC device
+ * is added from the connection request handler when enough
+ * information has been received.
+ */
+ ret = gb_connection_enable(svc->connection);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
+static void gb_svc_remove_modules(struct gb_svc *svc)
+{
+ struct gb_host_device *hd = svc->hd;
+ struct gb_module *module, *tmp;
+
+ list_for_each_entry_safe(module, tmp, &hd->modules, hd_node) {
+ gb_module_del(module);
+ list_del(&module->hd_node);
+ gb_module_put(module);
+ }
+}
+
+void gb_svc_del(struct gb_svc *svc)
+{
+ gb_connection_disable_rx(svc->connection);
+
+ /*
+ * The SVC device may have been registered from the request handler.
+ */
+ if (device_is_registered(&svc->dev)) {
+ gb_timesync_svc_remove(svc);
+ gb_svc_debugfs_exit(svc);
+ gb_svc_watchdog_destroy(svc);
+ device_del(&svc->dev);
+ }
+
+ flush_workqueue(svc->wq);
+
+ gb_svc_remove_modules(svc);
+
+ gb_connection_disable(svc->connection);
+}
+
+void gb_svc_put(struct gb_svc *svc)
+{
+ put_device(&svc->dev);
+}
--- /dev/null
+++ b/drivers/greybus/svc.h
@@ -0,0 +1,109 @@
+/*
+ * Greybus SVC code
+ *
+ * Copyright 2015 Google Inc.
+ * Copyright 2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __SVC_H
+#define __SVC_H
+
+#define GB_SVC_CPORT_FLAG_E2EFC BIT(0)
+#define GB_SVC_CPORT_FLAG_CSD_N BIT(1)
+#define GB_SVC_CPORT_FLAG_CSV_N BIT(2)
+
+enum gb_svc_state {
+ GB_SVC_STATE_RESET,
+ GB_SVC_STATE_PROTOCOL_VERSION,
+ GB_SVC_STATE_SVC_HELLO,
+};
+
+enum gb_svc_watchdog_bite {
+ GB_SVC_WATCHDOG_BITE_RESET_UNIPRO = 0,
+ GB_SVC_WATCHDOG_BITE_PANIC_KERNEL,
+};
+
+struct gb_svc_watchdog;
+
+struct svc_debugfs_pwrmon_rail {
+ u8 id;
+ struct gb_svc *svc;
+};
+
+struct gb_svc {
+ struct device dev;
+
+ struct gb_host_device *hd;
+ struct gb_connection *connection;
+ enum gb_svc_state state;
+ struct ida device_id_map;
+ struct workqueue_struct *wq;
+
+ u16 endo_id;
+ u8 ap_intf_id;
+
+ u8 protocol_major;
+ u8 protocol_minor;
+
+ struct gb_svc_watchdog *watchdog;
+ enum gb_svc_watchdog_bite action;
+
+ struct dentry *debugfs_dentry;
+ struct svc_debugfs_pwrmon_rail *pwrmon_rails;
+};
+#define to_gb_svc(d) container_of(d, struct gb_svc, dev)
+
+struct gb_svc *gb_svc_create(struct gb_host_device *hd);
+int gb_svc_add(struct gb_svc *svc);
+void gb_svc_del(struct gb_svc *svc);
+void gb_svc_put(struct gb_svc *svc);
+
+int gb_svc_pwrmon_intf_sample_get(struct gb_svc *svc, u8 intf_id,
+ u8 measurement_type, u32 *value);
+int gb_svc_intf_device_id(struct gb_svc *svc, u8 intf_id, u8 device_id);
+int gb_svc_route_create(struct gb_svc *svc, u8 intf1_id, u8 dev1_id,
+ u8 intf2_id, u8 dev2_id);
+void gb_svc_route_destroy(struct gb_svc *svc, u8 intf1_id, u8 intf2_id);
+int gb_svc_connection_create(struct gb_svc *svc, u8 intf1_id, u16 cport1_id,
+ u8 intf2_id, u16 cport2_id, u8 cport_flags);
+void gb_svc_connection_destroy(struct gb_svc *svc, u8 intf1_id, u16 cport1_id,
+ u8 intf2_id, u16 cport2_id);
+int gb_svc_intf_eject(struct gb_svc *svc, u8 intf_id);
+int gb_svc_intf_vsys_set(struct gb_svc *svc, u8 intf_id, bool enable);
+int gb_svc_intf_refclk_set(struct gb_svc *svc, u8 intf_id, bool enable);
+int gb_svc_intf_unipro_set(struct gb_svc *svc, u8 intf_id, bool enable);
+int gb_svc_intf_activate(struct gb_svc *svc, u8 intf_id, u8 *intf_type);
+int gb_svc_intf_resume(struct gb_svc *svc, u8 intf_id);
+
+int gb_svc_dme_peer_get(struct gb_svc *svc, u8 intf_id, u16 attr, u16 selector,
+ u32 *value);
+int gb_svc_dme_peer_set(struct gb_svc *svc, u8 intf_id, u16 attr, u16 selector,
+ u32 value);
+int gb_svc_intf_set_power_mode(struct gb_svc *svc, u8 intf_id, u8 hs_series,
+ u8 tx_mode, u8 tx_gear, u8 tx_nlanes,
+ u8 tx_amplitude, u8 tx_hs_equalizer,
+ u8 rx_mode, u8 rx_gear, u8 rx_nlanes,
+ u8 flags, u32 quirks,
+ struct gb_svc_l2_timer_cfg *local,
+ struct gb_svc_l2_timer_cfg *remote);
+int gb_svc_intf_set_power_mode_hibernate(struct gb_svc *svc, u8 intf_id);
+int gb_svc_ping(struct gb_svc *svc);
+int gb_svc_watchdog_create(struct gb_svc *svc);
+void gb_svc_watchdog_destroy(struct gb_svc *svc);
+bool gb_svc_watchdog_enabled(struct gb_svc *svc);
+int gb_svc_watchdog_enable(struct gb_svc *svc);
+int gb_svc_watchdog_disable(struct gb_svc *svc);
+int gb_svc_timesync_enable(struct gb_svc *svc, u8 count, u64 frame_time,
+ u32 strobe_delay, u32 refclk);
+int gb_svc_timesync_disable(struct gb_svc *svc);
+int gb_svc_timesync_authoritative(struct gb_svc *svc, u64 *frame_time);
+int gb_svc_timesync_ping(struct gb_svc *svc, u64 *frame_time);
+int gb_svc_timesync_wake_pins_acquire(struct gb_svc *svc, u32 strobe_mask);
+int gb_svc_timesync_wake_pins_release(struct gb_svc *svc);
+
+int gb_svc_protocol_init(void);
+void gb_svc_protocol_exit(void);
+
+#endif /* __SVC_H */
--- /dev/null
+++ b/drivers/greybus/svc_watchdog.c
@@ -0,0 +1,198 @@
+/*
+ * SVC Greybus "watchdog" driver.
+ *
+ * Copyright 2016 Google Inc.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/delay.h>
+#include <linux/suspend.h>
+#include <linux/workqueue.h>
+#include "greybus.h"
+
+#define SVC_WATCHDOG_PERIOD (2*HZ)
+
+struct gb_svc_watchdog {
+ struct delayed_work work;
+ struct gb_svc *svc;
+ bool enabled;
+ struct notifier_block pm_notifier;
+};
+
+static struct delayed_work reset_work;
+
+static int svc_watchdog_pm_notifier(struct notifier_block *notifier,
+ unsigned long pm_event, void *unused)
+{
+ struct gb_svc_watchdog *watchdog =
+ container_of(notifier, struct gb_svc_watchdog, pm_notifier);
+
+ switch (pm_event) {
+ case PM_SUSPEND_PREPARE:
+ gb_svc_watchdog_disable(watchdog->svc);
+ break;
+ case PM_POST_SUSPEND:
+ gb_svc_watchdog_enable(watchdog->svc);
+ break;
+ default:
+ break;
+ }
+
+ return NOTIFY_DONE;
+}
+
+static void greybus_reset(struct work_struct *work)
+{
+ static char start_path[256] = "/system/bin/start";
+ static char *envp[] = {
+ "HOME=/",
+ "PATH=/sbin:/vendor/bin:/system/sbin:/system/bin:/system/xbin",
+ NULL,
+ };
+ static char *argv[] = {
+ start_path,
+ "unipro_reset",
+ NULL,
+ };
+
+ printk(KERN_ERR "svc_watchdog: calling \"%s %s\" to reset greybus network!\n",
+ argv[0], argv[1]);
+ call_usermodehelper(start_path, argv, envp, UMH_WAIT_EXEC);
+}
+
+static void do_work(struct work_struct *work)
+{
+ struct gb_svc_watchdog *watchdog;
+ struct gb_svc *svc;
+ int retval;
+
+ watchdog = container_of(work, struct gb_svc_watchdog, work.work);
+ svc = watchdog->svc;
+
+ dev_dbg(&svc->dev, "%s: ping.\n", __func__);
+ retval = gb_svc_ping(svc);
+ if (retval) {
+ /*
+ * Something went really wrong, let's warn userspace and then
+ * pull the plug and reset the whole greybus network.
+ * We need to do this outside of this workqueue as we will be
+ * tearing down the svc device itself. So queue up
+ * yet-another-callback to do that.
+ */
+ dev_err(&svc->dev,
+ "SVC ping has returned %d, something is wrong!!!\n",
+ retval);
+
+ if (svc->action == GB_SVC_WATCHDOG_BITE_PANIC_KERNEL) {
+ panic("SVC is not responding\n");
+ } else if (svc->action == GB_SVC_WATCHDOG_BITE_RESET_UNIPRO) {
+ dev_err(&svc->dev, "Resetting the greybus network, watch out!!!\n");
+
+ INIT_DELAYED_WORK(&reset_work, greybus_reset);
+ schedule_delayed_work(&reset_work, HZ / 2);
+
+ /*
+ * Disable ourselves, we don't want to trip again unless
+ * userspace wants us to.
+ */
+ watchdog->enabled = false;
+ }
+ }
+
+ /* resubmit our work to happen again, if we are still "alive" */
+ if (watchdog->enabled)
+ schedule_delayed_work(&watchdog->work, SVC_WATCHDOG_PERIOD);
+}
+
+int gb_svc_watchdog_create(struct gb_svc *svc)
+{
+ struct gb_svc_watchdog *watchdog;
+ int retval;
+
+ if (svc->watchdog)
+ return 0;
+
+ watchdog = kmalloc(sizeof(*watchdog), GFP_KERNEL);
+ if (!watchdog)
+ return -ENOMEM;
+
+ watchdog->enabled = false;
+ watchdog->svc = svc;
+ INIT_DELAYED_WORK(&watchdog->work, do_work);
+ svc->watchdog = watchdog;
+
+ watchdog->pm_notifier.notifier_call = svc_watchdog_pm_notifier;
+ retval = register_pm_notifier(&watchdog->pm_notifier);
+ if (retval) {
+ dev_err(&svc->dev, "error registering pm notifier(%d)\n",
+ retval);
+ goto svc_watchdog_create_err;
+ }
+
+ retval = gb_svc_watchdog_enable(svc);
+ if (retval) {
+ dev_err(&svc->dev, "error enabling watchdog (%d)\n", retval);
+ unregister_pm_notifier(&watchdog->pm_notifier);
+ goto svc_watchdog_create_err;
+ }
+ return retval;
+
+svc_watchdog_create_err:
+ svc->watchdog = NULL;
+ kfree(watchdog);
+
+ return retval;
+}
+
+void gb_svc_watchdog_destroy(struct gb_svc *svc)
+{
+ struct gb_svc_watchdog *watchdog = svc->watchdog;
+
+ if (!watchdog)
+ return;
+
+ unregister_pm_notifier(&watchdog->pm_notifier);
+ gb_svc_watchdog_disable(svc);
+ svc->watchdog = NULL;
+ kfree(watchdog);
+}
+
+bool gb_svc_watchdog_enabled(struct gb_svc *svc)
+{
+ if (!svc || !svc->watchdog)
+ return false;
+ return svc->watchdog->enabled;
+}
+
+int gb_svc_watchdog_enable(struct gb_svc *svc)
+{
+ struct gb_svc_watchdog *watchdog;
+
+ if (!svc->watchdog)
+ return -ENODEV;
+
+ watchdog = svc->watchdog;
+ if (watchdog->enabled)
+ return 0;
+
+ watchdog->enabled = true;
+ schedule_delayed_work(&watchdog->work, SVC_WATCHDOG_PERIOD);
+ return 0;
+}
+
+int gb_svc_watchdog_disable(struct gb_svc *svc)
+{
+ struct gb_svc_watchdog *watchdog;
+
+ if (!svc->watchdog)
+ return -ENODEV;
+
+ watchdog = svc->watchdog;
+ if (!watchdog->enabled)
+ return 0;
+
+ watchdog->enabled = false;
+ cancel_delayed_work_sync(&watchdog->work);
+ return 0;
+}
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 07/32] greybus: core code
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (4 preceding siblings ...)
2016-09-16 13:23 ` [patch 06/32] greybus: svc driver/watchdog Greg KH
@ 2016-09-16 13:23 ` Greg KH
2016-09-16 13:24 ` [patch 08/32] greybus: bootrom driver Greg KH
` (24 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 13:23 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This is the core Greybus code. It implements the greybus protocol and
is the base of where all other greybus drivers tie into.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/arpc.h | 109 +++
drivers/greybus/authentication.c | 429 ++++++++++++++
drivers/greybus/bundle.c | 253 ++++++++
drivers/greybus/bundle.h | 90 ++
drivers/greybus/connection.c | 938 +++++++++++++++++++++++++++++++
drivers/greybus/connection.h | 129 ++++
drivers/greybus/control.c | 635 ++++++++++++++++++++
drivers/greybus/control.h | 65 ++
drivers/greybus/core.c | 361 +++++++++++
drivers/greybus/debugfs.c | 31 +
drivers/greybus/greybus.h | 154 +++++
drivers/greybus/greybus_authentication.h | 120 +++
drivers/greybus/greybus_id.h | 26
drivers/greybus/greybus_manifest.h | 177 +++++
drivers/greybus/manifest.c | 535 +++++++++++++++++
drivers/greybus/manifest.h | 16
drivers/greybus/module.c | 238 +++++++
drivers/greybus/module.h | 34 +
18 files changed, 4340 insertions(+)
--- /dev/null
+++ b/drivers/greybus/arpc.h
@@ -0,0 +1,109 @@
+/*
+ * This file is provided under a dual BSD/GPLv2 license. When using or
+ * redistributing this file, you may do so under either license.
+ *
+ * GPL LICENSE SUMMARY
+ *
+ * Copyright(c) 2016 Google Inc. All rights reserved.
+ * Copyright(c) 2016 Linaro Ltd. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License version 2 for more details.
+ *
+ * BSD LICENSE
+ *
+ * Copyright(c) 2016 Google Inc. All rights reserved.
+ * Copyright(c) 2016 Linaro Ltd. All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of Google Inc. or Linaro Ltd. nor the names of
+ * its contributors may be used to endorse or promote products
+ * derived from this software without specific prior written
+ * permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL GOOGLE INC. OR
+ * LINARO LTD. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+ * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef __ARPC_H
+#define __ARPC_H
+
+/* APBridgeA RPC (ARPC) */
+
+enum arpc_result {
+ ARPC_SUCCESS = 0x00,
+ ARPC_NO_MEMORY = 0x01,
+ ARPC_INVALID = 0x02,
+ ARPC_TIMEOUT = 0x03,
+ ARPC_UNKNOWN_ERROR = 0xff,
+};
+
+struct arpc_request_message {
+ __le16 id; /* RPC unique id */
+ __le16 size; /* Size in bytes of header + payload */
+ __u8 type; /* RPC type */
+ __u8 data[0]; /* ARPC data */
+} __packed;
+
+struct arpc_response_message {
+ __le16 id; /* RPC unique id */
+ __u8 result; /* Result of RPC */
+} __packed;
+
+
+/* ARPC requests */
+#define ARPC_TYPE_CPORT_CONNECTED 0x01
+#define ARPC_TYPE_CPORT_QUIESCE 0x02
+#define ARPC_TYPE_CPORT_CLEAR 0x03
+#define ARPC_TYPE_CPORT_FLUSH 0x04
+#define ARPC_TYPE_CPORT_SHUTDOWN 0x05
+
+struct arpc_cport_connected_req {
+ __le16 cport_id;
+} __packed;
+
+struct arpc_cport_quiesce_req {
+ __le16 cport_id;
+ __le16 peer_space;
+ __le16 timeout;
+} __packed;
+
+struct arpc_cport_clear_req {
+ __le16 cport_id;
+} __packed;
+
+struct arpc_cport_flush_req {
+ __le16 cport_id;
+} __packed;
+
+struct arpc_cport_shutdown_req {
+ __le16 cport_id;
+ __le16 timeout;
+ __u8 phase;
+} __packed;
+
+#endif /* __ARPC_H */
--- /dev/null
+++ b/drivers/greybus/authentication.c
@@ -0,0 +1,429 @@
+/*
+ * Greybus Component Authentication Protocol (CAP) Driver.
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include "greybus.h"
+
+#include <linux/cdev.h>
+#include <linux/fs.h>
+#include <linux/ioctl.h>
+#include <linux/uaccess.h>
+
+#include "greybus_authentication.h"
+#include "firmware.h"
+#include "greybus.h"
+
+#define CAP_TIMEOUT_MS 1000
+
+/*
+ * Number of minor devices this driver supports.
+ * There will be exactly one required per Interface.
+ */
+#define NUM_MINORS U8_MAX
+
+struct gb_cap {
+ struct device *parent;
+ struct gb_connection *connection;
+ struct kref kref;
+ struct list_head node;
+ bool disabled; /* connection getting disabled */
+
+ struct mutex mutex;
+ struct cdev cdev;
+ struct device *class_device;
+ dev_t dev_num;
+};
+
+static struct class *cap_class;
+static dev_t cap_dev_num;
+static DEFINE_IDA(cap_minors_map);
+static LIST_HEAD(cap_list);
+static DEFINE_MUTEX(list_mutex);
+
+static void cap_kref_release(struct kref *kref)
+{
+ struct gb_cap *cap = container_of(kref, struct gb_cap, kref);
+
+ kfree(cap);
+}
+
+/*
+ * All users of cap take a reference (from within list_mutex lock), before
+ * they get a pointer to play with. And the structure will be freed only after
+ * the last user has put the reference to it.
+ */
+static void put_cap(struct gb_cap *cap)
+{
+ kref_put(&cap->kref, cap_kref_release);
+}
+
+/* Caller must call put_cap() after using struct gb_cap */
+static struct gb_cap *get_cap(struct cdev *cdev)
+{
+ struct gb_cap *cap;
+
+ mutex_lock(&list_mutex);
+
+ list_for_each_entry(cap, &cap_list, node) {
+ if (&cap->cdev == cdev) {
+ kref_get(&cap->kref);
+ goto unlock;
+ }
+ }
+
+ cap = NULL;
+
+unlock:
+ mutex_unlock(&list_mutex);
+
+ return cap;
+}
+
+static int cap_get_endpoint_uid(struct gb_cap *cap, u8 *euid)
+{
+ struct gb_connection *connection = cap->connection;
+ struct gb_cap_get_endpoint_uid_response response;
+ int ret;
+
+ ret = gb_operation_sync(connection, GB_CAP_TYPE_GET_ENDPOINT_UID, NULL,
+ 0, &response, sizeof(response));
+ if (ret) {
+ dev_err(cap->parent, "failed to get endpoint uid (%d)\n", ret);
+ return ret;
+ }
+
+ memcpy(euid, response.uid, sizeof(response.uid));
+
+ return 0;
+}
+
+static int cap_get_ims_certificate(struct gb_cap *cap, u32 class, u32 id,
+ u8 *certificate, u32 *size, u8 *result)
+{
+ struct gb_connection *connection = cap->connection;
+ struct gb_cap_get_ims_certificate_request *request;
+ struct gb_cap_get_ims_certificate_response *response;
+ size_t max_size = gb_operation_get_payload_size_max(connection);
+ struct gb_operation *op;
+ int ret;
+
+ op = gb_operation_create_flags(connection,
+ GB_CAP_TYPE_GET_IMS_CERTIFICATE,
+ sizeof(*request), max_size,
+ GB_OPERATION_FLAG_SHORT_RESPONSE,
+ GFP_KERNEL);
+ if (!op)
+ return -ENOMEM;
+
+ request = op->request->payload;
+ request->certificate_class = cpu_to_le32(class);
+ request->certificate_id = cpu_to_le32(id);
+
+ ret = gb_operation_request_send_sync(op);
+ if (ret) {
+ dev_err(cap->parent, "failed to get certificate (%d)\n", ret);
+ goto done;
+ }
+
+ response = op->response->payload;
+ *result = response->result_code;
+ *size = op->response->payload_size - sizeof(*response);
+ memcpy(certificate, response->certificate, *size);
+
+done:
+ gb_operation_put(op);
+ return ret;
+}
+
+static int cap_authenticate(struct gb_cap *cap, u32 auth_type, u8 *uid,
+ u8 *challenge, u8 *result, u8 *auth_response,
+ u32 *signature_size, u8 *signature)
+{
+ struct gb_connection *connection = cap->connection;
+ struct gb_cap_authenticate_request *request;
+ struct gb_cap_authenticate_response *response;
+ size_t max_size = gb_operation_get_payload_size_max(connection);
+ struct gb_operation *op;
+ int ret;
+
+ op = gb_operation_create_flags(connection, GB_CAP_TYPE_AUTHENTICATE,
+ sizeof(*request), max_size,
+ GB_OPERATION_FLAG_SHORT_RESPONSE,
+ GFP_KERNEL);
+ if (!op)
+ return -ENOMEM;
+
+ request = op->request->payload;
+ request->auth_type = cpu_to_le32(auth_type);
+ memcpy(request->uid, uid, sizeof(request->uid));
+ memcpy(request->challenge, challenge, sizeof(request->challenge));
+
+ ret = gb_operation_request_send_sync(op);
+ if (ret) {
+ dev_err(cap->parent, "failed to authenticate (%d)\n", ret);
+ goto done;
+ }
+
+ response = op->response->payload;
+ *result = response->result_code;
+ *signature_size = op->response->payload_size - sizeof(*response);
+ memcpy(auth_response, response->response, sizeof(response->response));
+ memcpy(signature, response->signature, *signature_size);
+
+done:
+ gb_operation_put(op);
+ return ret;
+}
+
+/* Char device fops */
+
+static int cap_open(struct inode *inode, struct file *file)
+{
+ struct gb_cap *cap = get_cap(inode->i_cdev);
+
+ /* cap structure can't get freed until file descriptor is closed */
+ if (cap) {
+ file->private_data = cap;
+ return 0;
+ }
+
+ return -ENODEV;
+}
+
+static int cap_release(struct inode *inode, struct file *file)
+{
+ struct gb_cap *cap = file->private_data;
+
+ put_cap(cap);
+ return 0;
+}
+
+static int cap_ioctl(struct gb_cap *cap, unsigned int cmd,
+ void __user *buf)
+{
+ struct cap_ioc_get_endpoint_uid endpoint_uid;
+ struct cap_ioc_get_ims_certificate *ims_cert;
+ struct cap_ioc_authenticate *authenticate;
+ size_t size;
+ int ret;
+
+ switch (cmd) {
+ case CAP_IOC_GET_ENDPOINT_UID:
+ ret = cap_get_endpoint_uid(cap, endpoint_uid.uid);
+ if (ret)
+ return ret;
+
+ if (copy_to_user(buf, &endpoint_uid, sizeof(endpoint_uid)))
+ return -EFAULT;
+
+ return 0;
+ case CAP_IOC_GET_IMS_CERTIFICATE:
+ size = sizeof(*ims_cert);
+ ims_cert = memdup_user(buf, size);
+ if (IS_ERR(ims_cert))
+ return PTR_ERR(ims_cert);
+
+ ret = cap_get_ims_certificate(cap, ims_cert->certificate_class,
+ ims_cert->certificate_id,
+ ims_cert->certificate,
+ &ims_cert->cert_size,
+ &ims_cert->result_code);
+ if (!ret && copy_to_user(buf, ims_cert, size))
+ ret = -EFAULT;
+ kfree(ims_cert);
+
+ return ret;
+ case CAP_IOC_AUTHENTICATE:
+ size = sizeof(*authenticate);
+ authenticate = memdup_user(buf, size);
+ if (IS_ERR(authenticate))
+ return PTR_ERR(authenticate);
+
+ ret = cap_authenticate(cap, authenticate->auth_type,
+ authenticate->uid,
+ authenticate->challenge,
+ &authenticate->result_code,
+ authenticate->response,
+ &authenticate->signature_size,
+ authenticate->signature);
+ if (!ret && copy_to_user(buf, authenticate, size))
+ ret = -EFAULT;
+ kfree(authenticate);
+
+ return ret;
+ default:
+ return -ENOTTY;
+ }
+}
+
+static long cap_ioctl_unlocked(struct file *file, unsigned int cmd,
+ unsigned long arg)
+{
+ struct gb_cap *cap = file->private_data;
+ struct gb_bundle *bundle = cap->connection->bundle;
+ int ret = -ENODEV;
+
+ /*
+ * Serialize ioctls.
+ *
+ * We don't want the user to do multiple authentication operations in
+ * parallel.
+ *
+ * This is also used to protect ->disabled, which is used to check if
+ * the connection is getting disconnected, so that we don't start any
+ * new operations.
+ */
+ mutex_lock(&cap->mutex);
+ if (!cap->disabled) {
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (!ret) {
+ ret = cap_ioctl(cap, cmd, (void __user *)arg);
+ gb_pm_runtime_put_autosuspend(bundle);
+ }
+ }
+ mutex_unlock(&cap->mutex);
+
+ return ret;
+}
+
+static const struct file_operations cap_fops = {
+ .owner = THIS_MODULE,
+ .open = cap_open,
+ .release = cap_release,
+ .unlocked_ioctl = cap_ioctl_unlocked,
+};
+
+int gb_cap_connection_init(struct gb_connection *connection)
+{
+ struct gb_cap *cap;
+ int ret, minor;
+
+ if (!connection)
+ return 0;
+
+ cap = kzalloc(sizeof(*cap), GFP_KERNEL);
+ if (!cap)
+ return -ENOMEM;
+
+ cap->parent = &connection->bundle->dev;
+ cap->connection = connection;
+ mutex_init(&cap->mutex);
+ gb_connection_set_data(connection, cap);
+ kref_init(&cap->kref);
+
+ mutex_lock(&list_mutex);
+ list_add(&cap->node, &cap_list);
+ mutex_unlock(&list_mutex);
+
+ ret = gb_connection_enable(connection);
+ if (ret)
+ goto err_list_del;
+
+ minor = ida_simple_get(&cap_minors_map, 0, NUM_MINORS, GFP_KERNEL);
+ if (minor < 0) {
+ ret = minor;
+ goto err_connection_disable;
+ }
+
+ /* Add a char device to allow userspace to interact with cap */
+ cap->dev_num = MKDEV(MAJOR(cap_dev_num), minor);
+ cdev_init(&cap->cdev, &cap_fops);
+
+ ret = cdev_add(&cap->cdev, cap->dev_num, 1);
+ if (ret)
+ goto err_remove_ida;
+
+ /* Add a soft link to the previously added char-dev within the bundle */
+ cap->class_device = device_create(cap_class, cap->parent, cap->dev_num,
+ NULL, "gb-authenticate-%d", minor);
+ if (IS_ERR(cap->class_device)) {
+ ret = PTR_ERR(cap->class_device);
+ goto err_del_cdev;
+ }
+
+ return 0;
+
+err_del_cdev:
+ cdev_del(&cap->cdev);
+err_remove_ida:
+ ida_simple_remove(&cap_minors_map, minor);
+err_connection_disable:
+ gb_connection_disable(connection);
+err_list_del:
+ mutex_lock(&list_mutex);
+ list_del(&cap->node);
+ mutex_unlock(&list_mutex);
+
+ put_cap(cap);
+
+ return ret;
+}
+
+void gb_cap_connection_exit(struct gb_connection *connection)
+{
+ struct gb_cap *cap;
+
+ if (!connection)
+ return;
+
+ cap = gb_connection_get_data(connection);
+
+ device_destroy(cap_class, cap->dev_num);
+ cdev_del(&cap->cdev);
+ ida_simple_remove(&cap_minors_map, MINOR(cap->dev_num));
+
+ /*
+ * Disallow any new ioctl operations on the char device and wait for
+ * existing ones to finish.
+ */
+ mutex_lock(&cap->mutex);
+ cap->disabled = true;
+ mutex_unlock(&cap->mutex);
+
+ /* All pending greybus operations should have finished by now */
+ gb_connection_disable(cap->connection);
+
+ /* Disallow new users to get access to the cap structure */
+ mutex_lock(&list_mutex);
+ list_del(&cap->node);
+ mutex_unlock(&list_mutex);
+
+ /*
+ * All current users of cap would have taken a reference to it by
+ * now, we can drop our reference and wait the last user will get
+ * cap freed.
+ */
+ put_cap(cap);
+}
+
+int cap_init(void)
+{
+ int ret;
+
+ cap_class = class_create(THIS_MODULE, "gb_authenticate");
+ if (IS_ERR(cap_class))
+ return PTR_ERR(cap_class);
+
+ ret = alloc_chrdev_region(&cap_dev_num, 0, NUM_MINORS,
+ "gb_authenticate");
+ if (ret)
+ goto err_remove_class;
+
+ return 0;
+
+err_remove_class:
+ class_destroy(cap_class);
+ return ret;
+}
+
+void cap_exit(void)
+{
+ unregister_chrdev_region(cap_dev_num, NUM_MINORS);
+ class_destroy(cap_class);
+ ida_destroy(&cap_minors_map);
+}
--- /dev/null
+++ b/drivers/greybus/bundle.c
@@ -0,0 +1,253 @@
+/*
+ * Greybus bundles
+ *
+ * Copyright 2014-2015 Google Inc.
+ * Copyright 2014-2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include "greybus.h"
+#include "greybus_trace.h"
+
+static ssize_t bundle_class_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_bundle *bundle = to_gb_bundle(dev);
+
+ return sprintf(buf, "0x%02x\n", bundle->class);
+}
+static DEVICE_ATTR_RO(bundle_class);
+
+static ssize_t bundle_id_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_bundle *bundle = to_gb_bundle(dev);
+
+ return sprintf(buf, "%u\n", bundle->id);
+}
+static DEVICE_ATTR_RO(bundle_id);
+
+static ssize_t state_show(struct device *dev, struct device_attribute *attr,
+ char *buf)
+{
+ struct gb_bundle *bundle = to_gb_bundle(dev);
+
+ if (bundle->state == NULL)
+ return sprintf(buf, "\n");
+
+ return sprintf(buf, "%s\n", bundle->state);
+}
+
+static ssize_t state_store(struct device *dev, struct device_attribute *attr,
+ const char *buf, size_t size)
+{
+ struct gb_bundle *bundle = to_gb_bundle(dev);
+
+ kfree(bundle->state);
+ bundle->state = kstrdup(buf, GFP_KERNEL);
+ if (!bundle->state)
+ return -ENOMEM;
+
+ /* Tell userspace that the file contents changed */
+ sysfs_notify(&bundle->dev.kobj, NULL, "state");
+
+ return size;
+}
+static DEVICE_ATTR_RW(state);
+
+static struct attribute *bundle_attrs[] = {
+ &dev_attr_bundle_class.attr,
+ &dev_attr_bundle_id.attr,
+ &dev_attr_state.attr,
+ NULL,
+};
+
+ATTRIBUTE_GROUPS(bundle);
+
+static struct gb_bundle *gb_bundle_find(struct gb_interface *intf,
+ u8 bundle_id)
+{
+ struct gb_bundle *bundle;
+
+ list_for_each_entry(bundle, &intf->bundles, links) {
+ if (bundle->id == bundle_id)
+ return bundle;
+ }
+
+ return NULL;
+}
+
+static void gb_bundle_release(struct device *dev)
+{
+ struct gb_bundle *bundle = to_gb_bundle(dev);
+
+ trace_gb_bundle_release(bundle);
+
+ kfree(bundle->state);
+ kfree(bundle->cport_desc);
+ kfree(bundle);
+}
+
+#ifdef CONFIG_PM
+static void gb_bundle_disable_all_connections(struct gb_bundle *bundle)
+{
+ struct gb_connection *connection;
+
+ list_for_each_entry(connection, &bundle->connections, bundle_links)
+ gb_connection_disable(connection);
+}
+
+static void gb_bundle_enable_all_connections(struct gb_bundle *bundle)
+{
+ struct gb_connection *connection;
+
+ list_for_each_entry(connection, &bundle->connections, bundle_links)
+ gb_connection_enable(connection);
+}
+
+static int gb_bundle_suspend(struct device *dev)
+{
+ struct gb_bundle *bundle = to_gb_bundle(dev);
+ const struct dev_pm_ops *pm = dev->driver->pm;
+ int ret;
+
+ if (pm && pm->runtime_suspend) {
+ ret = pm->runtime_suspend(&bundle->dev);
+ if (ret)
+ return ret;
+ } else {
+ gb_bundle_disable_all_connections(bundle);
+ }
+
+ ret = gb_control_bundle_suspend(bundle->intf->control, bundle->id);
+ if (ret) {
+ if (pm && pm->runtime_resume)
+ ret = pm->runtime_resume(dev);
+ else
+ gb_bundle_enable_all_connections(bundle);
+
+ return ret;
+ }
+
+ return 0;
+}
+
+static int gb_bundle_resume(struct device *dev)
+{
+ struct gb_bundle *bundle = to_gb_bundle(dev);
+ const struct dev_pm_ops *pm = dev->driver->pm;
+ int ret;
+
+ ret = gb_control_bundle_resume(bundle->intf->control, bundle->id);
+ if (ret)
+ return ret;
+
+ if (pm && pm->runtime_resume) {
+ ret = pm->runtime_resume(dev);
+ if (ret)
+ return ret;
+ } else {
+ gb_bundle_enable_all_connections(bundle);
+ }
+
+ return 0;
+}
+
+static int gb_bundle_idle(struct device *dev)
+{
+ pm_runtime_mark_last_busy(dev);
+ pm_request_autosuspend(dev);
+
+ return 0;
+}
+#endif
+
+static const struct dev_pm_ops gb_bundle_pm_ops = {
+ SET_RUNTIME_PM_OPS(gb_bundle_suspend, gb_bundle_resume, gb_bundle_idle)
+};
+
+struct device_type greybus_bundle_type = {
+ .name = "greybus_bundle",
+ .release = gb_bundle_release,
+ .pm = &gb_bundle_pm_ops,
+};
+
+/*
+ * Create a gb_bundle structure to represent a discovered
+ * bundle. Returns a pointer to the new bundle or a null
+ * pointer if a failure occurs due to memory exhaustion.
+ */
+struct gb_bundle *gb_bundle_create(struct gb_interface *intf, u8 bundle_id,
+ u8 class)
+{
+ struct gb_bundle *bundle;
+
+ if (bundle_id == BUNDLE_ID_NONE) {
+ dev_err(&intf->dev, "can't use bundle id %u\n", bundle_id);
+ return NULL;
+ }
+
+ /*
+ * Reject any attempt to reuse a bundle id. We initialize
+ * these serially, so there's no need to worry about keeping
+ * the interface bundle list locked here.
+ */
+ if (gb_bundle_find(intf, bundle_id)) {
+ dev_err(&intf->dev, "duplicate bundle id %u\n", bundle_id);
+ return NULL;
+ }
+
+ bundle = kzalloc(sizeof(*bundle), GFP_KERNEL);
+ if (!bundle)
+ return NULL;
+
+ bundle->intf = intf;
+ bundle->id = bundle_id;
+ bundle->class = class;
+ INIT_LIST_HEAD(&bundle->connections);
+
+ bundle->dev.parent = &intf->dev;
+ bundle->dev.bus = &greybus_bus_type;
+ bundle->dev.type = &greybus_bundle_type;
+ bundle->dev.groups = bundle_groups;
+ bundle->dev.dma_mask = intf->dev.dma_mask;
+ device_initialize(&bundle->dev);
+ dev_set_name(&bundle->dev, "%s.%d", dev_name(&intf->dev), bundle_id);
+
+ list_add(&bundle->links, &intf->bundles);
+
+ trace_gb_bundle_create(bundle);
+
+ return bundle;
+}
+
+int gb_bundle_add(struct gb_bundle *bundle)
+{
+ int ret;
+
+ ret = device_add(&bundle->dev);
+ if (ret) {
+ dev_err(&bundle->dev, "failed to register bundle: %d\n", ret);
+ return ret;
+ }
+
+ trace_gb_bundle_add(bundle);
+
+ return 0;
+}
+
+/*
+ * Tear down a previously set up bundle.
+ */
+void gb_bundle_destroy(struct gb_bundle *bundle)
+{
+ trace_gb_bundle_destroy(bundle);
+
+ if (device_is_registered(&bundle->dev))
+ device_del(&bundle->dev);
+
+ list_del(&bundle->links);
+
+ put_device(&bundle->dev);
+}
--- /dev/null
+++ b/drivers/greybus/bundle.h
@@ -0,0 +1,90 @@
+/*
+ * Greybus bundles
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __BUNDLE_H
+#define __BUNDLE_H
+
+#include <linux/list.h>
+
+#define BUNDLE_ID_NONE U8_MAX
+
+/* Greybus "public" definitions" */
+struct gb_bundle {
+ struct device dev;
+ struct gb_interface *intf;
+
+ u8 id;
+ u8 class;
+ u8 class_major;
+ u8 class_minor;
+
+ size_t num_cports;
+ struct greybus_descriptor_cport *cport_desc;
+
+ struct list_head connections;
+ u8 *state;
+
+ struct list_head links; /* interface->bundles */
+};
+#define to_gb_bundle(d) container_of(d, struct gb_bundle, dev)
+
+/* Greybus "private" definitions" */
+struct gb_bundle *gb_bundle_create(struct gb_interface *intf, u8 bundle_id,
+ u8 class);
+int gb_bundle_add(struct gb_bundle *bundle);
+void gb_bundle_destroy(struct gb_bundle *bundle);
+
+/* Bundle Runtime PM wrappers */
+#ifdef CONFIG_PM
+static inline int gb_pm_runtime_get_sync(struct gb_bundle *bundle)
+{
+ int retval;
+
+ retval = pm_runtime_get_sync(&bundle->dev);
+ if (retval < 0) {
+ dev_err(&bundle->dev,
+ "pm_runtime_get_sync failed: %d\n", retval);
+ pm_runtime_put_noidle(&bundle->dev);
+ return retval;
+ }
+
+ return 0;
+}
+
+static inline int gb_pm_runtime_put_autosuspend(struct gb_bundle *bundle)
+{
+ int retval;
+
+ pm_runtime_mark_last_busy(&bundle->dev);
+ retval = pm_runtime_put_autosuspend(&bundle->dev);
+
+ return retval;
+}
+
+static inline void gb_pm_runtime_get_noresume(struct gb_bundle *bundle)
+{
+ pm_runtime_get_noresume(&bundle->dev);
+}
+
+static inline void gb_pm_runtime_put_noidle(struct gb_bundle *bundle)
+{
+ pm_runtime_put_noidle(&bundle->dev);
+}
+
+#else
+static inline int gb_pm_runtime_get_sync(struct gb_bundle *bundle)
+{ return 0; }
+static inline int gb_pm_runtime_put_autosuspend(struct gb_bundle *bundle)
+{ return 0; }
+
+static inline void gb_pm_runtime_get_noresume(struct gb_bundle *bundle) {}
+static inline void gb_pm_runtime_put_noidle(struct gb_bundle *bundle) {}
+#endif
+
+#endif /* __BUNDLE_H */
--- /dev/null
+++ b/drivers/greybus/connection.c
@@ -0,0 +1,938 @@
+/*
+ * Greybus connections
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/workqueue.h>
+
+#include "greybus.h"
+#include "greybus_trace.h"
+
+
+#define GB_CONNECTION_CPORT_QUIESCE_TIMEOUT 1000
+
+
+static void gb_connection_kref_release(struct kref *kref);
+
+
+static DEFINE_SPINLOCK(gb_connections_lock);
+static DEFINE_MUTEX(gb_connection_mutex);
+
+
+/* Caller holds gb_connection_mutex. */
+static bool gb_connection_cport_in_use(struct gb_interface *intf, u16 cport_id)
+{
+ struct gb_host_device *hd = intf->hd;
+ struct gb_connection *connection;
+
+ list_for_each_entry(connection, &hd->connections, hd_links) {
+ if (connection->intf == intf &&
+ connection->intf_cport_id == cport_id)
+ return true;
+ }
+
+ return false;
+}
+
+static void gb_connection_get(struct gb_connection *connection)
+{
+ kref_get(&connection->kref);
+
+ trace_gb_connection_get(connection);
+}
+
+static void gb_connection_put(struct gb_connection *connection)
+{
+ trace_gb_connection_put(connection);
+
+ kref_put(&connection->kref, gb_connection_kref_release);
+}
+
+/*
+ * Returns a reference-counted pointer to the connection if found.
+ */
+static struct gb_connection *
+gb_connection_hd_find(struct gb_host_device *hd, u16 cport_id)
+{
+ struct gb_connection *connection;
+ unsigned long flags;
+
+ spin_lock_irqsave(&gb_connections_lock, flags);
+ list_for_each_entry(connection, &hd->connections, hd_links)
+ if (connection->hd_cport_id == cport_id) {
+ gb_connection_get(connection);
+ goto found;
+ }
+ connection = NULL;
+found:
+ spin_unlock_irqrestore(&gb_connections_lock, flags);
+
+ return connection;
+}
+
+/*
+ * Callback from the host driver to let us know that data has been
+ * received on the bundle.
+ */
+void greybus_data_rcvd(struct gb_host_device *hd, u16 cport_id,
+ u8 *data, size_t length)
+{
+ struct gb_connection *connection;
+
+ trace_gb_hd_in(hd);
+
+ connection = gb_connection_hd_find(hd, cport_id);
+ if (!connection) {
+ dev_err(&hd->dev,
+ "nonexistent connection (%zu bytes dropped)\n", length);
+ return;
+ }
+ gb_connection_recv(connection, data, length);
+ gb_connection_put(connection);
+}
+EXPORT_SYMBOL_GPL(greybus_data_rcvd);
+
+static void gb_connection_kref_release(struct kref *kref)
+{
+ struct gb_connection *connection;
+
+ connection = container_of(kref, struct gb_connection, kref);
+
+ trace_gb_connection_release(connection);
+
+ kfree(connection);
+}
+
+static void gb_connection_init_name(struct gb_connection *connection)
+{
+ u16 hd_cport_id = connection->hd_cport_id;
+ u16 cport_id = 0;
+ u8 intf_id = 0;
+
+ if (connection->intf) {
+ intf_id = connection->intf->interface_id;
+ cport_id = connection->intf_cport_id;
+ }
+
+ snprintf(connection->name, sizeof(connection->name),
+ "%u/%u:%u", hd_cport_id, intf_id, cport_id);
+}
+
+/*
+ * _gb_connection_create() - create a Greybus connection
+ * @hd: host device of the connection
+ * @hd_cport_id: host-device cport id, or -1 for dynamic allocation
+ * @intf: remote interface, or NULL for static connections
+ * @bundle: remote-interface bundle (may be NULL)
+ * @cport_id: remote-interface cport id, or 0 for static connections
+ * @handler: request handler (may be NULL)
+ * @flags: connection flags
+ *
+ * Create a Greybus connection, representing the bidirectional link
+ * between a CPort on a (local) Greybus host device and a CPort on
+ * another Greybus interface.
+ *
+ * A connection also maintains the state of operations sent over the
+ * connection.
+ *
+ * Serialised against concurrent create and destroy using the
+ * gb_connection_mutex.
+ *
+ * Return: A pointer to the new connection if successful, or an ERR_PTR
+ * otherwise.
+ */
+static struct gb_connection *
+_gb_connection_create(struct gb_host_device *hd, int hd_cport_id,
+ struct gb_interface *intf,
+ struct gb_bundle *bundle, int cport_id,
+ gb_request_handler_t handler,
+ unsigned long flags)
+{
+ struct gb_connection *connection;
+ int ret;
+
+ mutex_lock(&gb_connection_mutex);
+
+ if (intf && gb_connection_cport_in_use(intf, cport_id)) {
+ dev_err(&intf->dev, "cport %u already in use\n", cport_id);
+ ret = -EBUSY;
+ goto err_unlock;
+ }
+
+ ret = gb_hd_cport_allocate(hd, hd_cport_id, flags);
+ if (ret < 0) {
+ dev_err(&hd->dev, "failed to allocate cport: %d\n", ret);
+ goto err_unlock;
+ }
+ hd_cport_id = ret;
+
+ connection = kzalloc(sizeof(*connection), GFP_KERNEL);
+ if (!connection) {
+ ret = -ENOMEM;
+ goto err_hd_cport_release;
+ }
+
+ connection->hd_cport_id = hd_cport_id;
+ connection->intf_cport_id = cport_id;
+ connection->hd = hd;
+ connection->intf = intf;
+ connection->bundle = bundle;
+ connection->handler = handler;
+ connection->flags = flags;
+ if (intf && (intf->quirks & GB_INTERFACE_QUIRK_NO_CPORT_FEATURES))
+ connection->flags |= GB_CONNECTION_FLAG_NO_FLOWCTRL;
+ connection->state = GB_CONNECTION_STATE_DISABLED;
+
+ atomic_set(&connection->op_cycle, 0);
+ mutex_init(&connection->mutex);
+ spin_lock_init(&connection->lock);
+ INIT_LIST_HEAD(&connection->operations);
+
+ connection->wq = alloc_workqueue("%s:%d", WQ_UNBOUND, 1,
+ dev_name(&hd->dev), hd_cport_id);
+ if (!connection->wq) {
+ ret = -ENOMEM;
+ goto err_free_connection;
+ }
+
+ kref_init(&connection->kref);
+
+ gb_connection_init_name(connection);
+
+ spin_lock_irq(&gb_connections_lock);
+ list_add(&connection->hd_links, &hd->connections);
+
+ if (bundle)
+ list_add(&connection->bundle_links, &bundle->connections);
+ else
+ INIT_LIST_HEAD(&connection->bundle_links);
+
+ spin_unlock_irq(&gb_connections_lock);
+
+ mutex_unlock(&gb_connection_mutex);
+
+ trace_gb_connection_create(connection);
+
+ return connection;
+
+err_free_connection:
+ kfree(connection);
+err_hd_cport_release:
+ gb_hd_cport_release(hd, hd_cport_id);
+err_unlock:
+ mutex_unlock(&gb_connection_mutex);
+
+ return ERR_PTR(ret);
+}
+
+struct gb_connection *
+gb_connection_create_static(struct gb_host_device *hd, u16 hd_cport_id,
+ gb_request_handler_t handler)
+{
+ return _gb_connection_create(hd, hd_cport_id, NULL, NULL, 0, handler,
+ GB_CONNECTION_FLAG_HIGH_PRIO);
+}
+
+struct gb_connection *
+gb_connection_create_control(struct gb_interface *intf)
+{
+ return _gb_connection_create(intf->hd, -1, intf, NULL, 0, NULL,
+ GB_CONNECTION_FLAG_CONTROL |
+ GB_CONNECTION_FLAG_HIGH_PRIO);
+}
+
+struct gb_connection *
+gb_connection_create(struct gb_bundle *bundle, u16 cport_id,
+ gb_request_handler_t handler)
+{
+ struct gb_interface *intf = bundle->intf;
+
+ return _gb_connection_create(intf->hd, -1, intf, bundle, cport_id,
+ handler, 0);
+}
+EXPORT_SYMBOL_GPL(gb_connection_create);
+
+struct gb_connection *
+gb_connection_create_flags(struct gb_bundle *bundle, u16 cport_id,
+ gb_request_handler_t handler,
+ unsigned long flags)
+{
+ struct gb_interface *intf = bundle->intf;
+
+ if (WARN_ON_ONCE(flags & GB_CONNECTION_FLAG_CORE_MASK))
+ flags &= ~GB_CONNECTION_FLAG_CORE_MASK;
+
+ return _gb_connection_create(intf->hd, -1, intf, bundle, cport_id,
+ handler, flags);
+}
+EXPORT_SYMBOL_GPL(gb_connection_create_flags);
+
+struct gb_connection *
+gb_connection_create_offloaded(struct gb_bundle *bundle, u16 cport_id,
+ unsigned long flags)
+{
+ flags |= GB_CONNECTION_FLAG_OFFLOADED;
+
+ return gb_connection_create_flags(bundle, cport_id, NULL, flags);
+}
+EXPORT_SYMBOL_GPL(gb_connection_create_offloaded);
+
+static int gb_connection_hd_cport_enable(struct gb_connection *connection)
+{
+ struct gb_host_device *hd = connection->hd;
+ int ret;
+
+ if (!hd->driver->cport_enable)
+ return 0;
+
+ ret = hd->driver->cport_enable(hd, connection->hd_cport_id,
+ connection->flags);
+ if (ret) {
+ dev_err(&hd->dev, "%s: failed to enable host cport: %d\n",
+ connection->name, ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static void gb_connection_hd_cport_disable(struct gb_connection *connection)
+{
+ struct gb_host_device *hd = connection->hd;
+ int ret;
+
+ if (!hd->driver->cport_disable)
+ return;
+
+ ret = hd->driver->cport_disable(hd, connection->hd_cport_id);
+ if (ret) {
+ dev_err(&hd->dev, "%s: failed to disable host cport: %d\n",
+ connection->name, ret);
+ }
+}
+
+static int gb_connection_hd_cport_connected(struct gb_connection *connection)
+{
+ struct gb_host_device *hd = connection->hd;
+ int ret;
+
+ if (!hd->driver->cport_connected)
+ return 0;
+
+ ret = hd->driver->cport_connected(hd, connection->hd_cport_id);
+ if (ret) {
+ dev_err(&hd->dev, "%s: failed to set connected state: %d\n",
+ connection->name, ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int gb_connection_hd_cport_flush(struct gb_connection *connection)
+{
+ struct gb_host_device *hd = connection->hd;
+ int ret;
+
+ if (!hd->driver->cport_flush)
+ return 0;
+
+ ret = hd->driver->cport_flush(hd, connection->hd_cport_id);
+ if (ret) {
+ dev_err(&hd->dev, "%s: failed to flush host cport: %d\n",
+ connection->name, ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int gb_connection_hd_cport_quiesce(struct gb_connection *connection)
+{
+ struct gb_host_device *hd = connection->hd;
+ size_t peer_space;
+ int ret;
+
+ peer_space = sizeof(struct gb_operation_msg_hdr) +
+ sizeof(struct gb_cport_shutdown_request);
+
+ if (connection->mode_switch)
+ peer_space += sizeof(struct gb_operation_msg_hdr);
+
+ ret = hd->driver->cport_quiesce(hd, connection->hd_cport_id,
+ peer_space,
+ GB_CONNECTION_CPORT_QUIESCE_TIMEOUT);
+ if (ret) {
+ dev_err(&hd->dev, "%s: failed to quiesce host cport: %d\n",
+ connection->name, ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int gb_connection_hd_cport_clear(struct gb_connection *connection)
+{
+ struct gb_host_device *hd = connection->hd;
+ int ret;
+
+ ret = hd->driver->cport_clear(hd, connection->hd_cport_id);
+ if (ret) {
+ dev_err(&hd->dev, "%s: failed to clear host cport: %d\n",
+ connection->name, ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+/*
+ * Request the SVC to create a connection from AP's cport to interface's
+ * cport.
+ */
+static int
+gb_connection_svc_connection_create(struct gb_connection *connection)
+{
+ struct gb_host_device *hd = connection->hd;
+ struct gb_interface *intf;
+ u8 cport_flags;
+ int ret;
+
+ if (gb_connection_is_static(connection))
+ return 0;
+
+ intf = connection->intf;
+
+ /*
+ * Enable either E2EFC or CSD, unless no flow control is requested.
+ */
+ cport_flags = GB_SVC_CPORT_FLAG_CSV_N;
+ if (gb_connection_flow_control_disabled(connection)) {
+ cport_flags |= GB_SVC_CPORT_FLAG_CSD_N;
+ } else if (gb_connection_e2efc_enabled(connection)) {
+ cport_flags |= GB_SVC_CPORT_FLAG_CSD_N |
+ GB_SVC_CPORT_FLAG_E2EFC;
+ }
+
+ ret = gb_svc_connection_create(hd->svc,
+ hd->svc->ap_intf_id,
+ connection->hd_cport_id,
+ intf->interface_id,
+ connection->intf_cport_id,
+ cport_flags);
+ if (ret) {
+ dev_err(&connection->hd->dev,
+ "%s: failed to create svc connection: %d\n",
+ connection->name, ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static void
+gb_connection_svc_connection_destroy(struct gb_connection *connection)
+{
+ if (gb_connection_is_static(connection))
+ return;
+
+ gb_svc_connection_destroy(connection->hd->svc,
+ connection->hd->svc->ap_intf_id,
+ connection->hd_cport_id,
+ connection->intf->interface_id,
+ connection->intf_cport_id);
+}
+
+/* Inform Interface about active CPorts */
+static int gb_connection_control_connected(struct gb_connection *connection)
+{
+ struct gb_control *control;
+ u16 cport_id = connection->intf_cport_id;
+ int ret;
+
+ if (gb_connection_is_static(connection))
+ return 0;
+
+ if (gb_connection_is_control(connection))
+ return 0;
+
+ control = connection->intf->control;
+
+ ret = gb_control_connected_operation(control, cport_id);
+ if (ret) {
+ dev_err(&connection->bundle->dev,
+ "failed to connect cport: %d\n", ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static void
+gb_connection_control_disconnecting(struct gb_connection *connection)
+{
+ struct gb_control *control;
+ u16 cport_id = connection->intf_cport_id;
+ int ret;
+
+ if (gb_connection_is_static(connection))
+ return;
+
+ control = connection->intf->control;
+
+ ret = gb_control_disconnecting_operation(control, cport_id);
+ if (ret) {
+ dev_err(&connection->hd->dev,
+ "%s: failed to send disconnecting: %d\n",
+ connection->name, ret);
+ }
+}
+
+static void
+gb_connection_control_disconnected(struct gb_connection *connection)
+{
+ struct gb_control *control;
+ u16 cport_id = connection->intf_cport_id;
+ int ret;
+
+ if (gb_connection_is_static(connection))
+ return;
+
+ control = connection->intf->control;
+
+ if (gb_connection_is_control(connection)) {
+ if (connection->mode_switch) {
+ ret = gb_control_mode_switch_operation(control);
+ if (ret) {
+ /*
+ * Allow mode switch to time out waiting for
+ * mailbox event.
+ */
+ return;
+ }
+ }
+
+ return;
+ }
+
+ ret = gb_control_disconnected_operation(control, cport_id);
+ if (ret) {
+ dev_warn(&connection->bundle->dev,
+ "failed to disconnect cport: %d\n", ret);
+ }
+}
+
+static int gb_connection_shutdown_operation(struct gb_connection *connection,
+ u8 phase)
+{
+ struct gb_cport_shutdown_request *req;
+ struct gb_operation *operation;
+ int ret;
+
+ operation = gb_operation_create_core(connection,
+ GB_REQUEST_TYPE_CPORT_SHUTDOWN,
+ sizeof(*req), 0, 0,
+ GFP_KERNEL);
+ if (!operation)
+ return -ENOMEM;
+
+ req = operation->request->payload;
+ req->phase = phase;
+
+ ret = gb_operation_request_send_sync(operation);
+
+ gb_operation_put(operation);
+
+ return ret;
+}
+
+static int gb_connection_cport_shutdown(struct gb_connection *connection,
+ u8 phase)
+{
+ struct gb_host_device *hd = connection->hd;
+ const struct gb_hd_driver *drv = hd->driver;
+ int ret;
+
+ if (gb_connection_is_static(connection))
+ return 0;
+
+ if (gb_connection_is_offloaded(connection)) {
+ if (!drv->cport_shutdown)
+ return 0;
+
+ ret = drv->cport_shutdown(hd, connection->hd_cport_id, phase,
+ GB_OPERATION_TIMEOUT_DEFAULT);
+ } else {
+ ret = gb_connection_shutdown_operation(connection, phase);
+ }
+
+ if (ret) {
+ dev_err(&hd->dev, "%s: failed to send cport shutdown (phase %d): %d\n",
+ connection->name, phase, ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int
+gb_connection_cport_shutdown_phase_1(struct gb_connection *connection)
+{
+ return gb_connection_cport_shutdown(connection, 1);
+}
+
+static int
+gb_connection_cport_shutdown_phase_2(struct gb_connection *connection)
+{
+ return gb_connection_cport_shutdown(connection, 2);
+}
+
+/*
+ * Cancel all active operations on a connection.
+ *
+ * Locking: Called with connection lock held and state set to DISABLED or
+ * DISCONNECTING.
+ */
+static void gb_connection_cancel_operations(struct gb_connection *connection,
+ int errno)
+ __must_hold(&connection->lock)
+{
+ struct gb_operation *operation;
+
+ while (!list_empty(&connection->operations)) {
+ operation = list_last_entry(&connection->operations,
+ struct gb_operation, links);
+ gb_operation_get(operation);
+ spin_unlock_irq(&connection->lock);
+
+ if (gb_operation_is_incoming(operation))
+ gb_operation_cancel_incoming(operation, errno);
+ else
+ gb_operation_cancel(operation, errno);
+
+ gb_operation_put(operation);
+
+ spin_lock_irq(&connection->lock);
+ }
+}
+
+/*
+ * Cancel all active incoming operations on a connection.
+ *
+ * Locking: Called with connection lock held and state set to ENABLED_TX.
+ */
+static void
+gb_connection_flush_incoming_operations(struct gb_connection *connection,
+ int errno)
+ __must_hold(&connection->lock)
+{
+ struct gb_operation *operation;
+ bool incoming;
+
+ while (!list_empty(&connection->operations)) {
+ incoming = false;
+ list_for_each_entry(operation, &connection->operations,
+ links) {
+ if (gb_operation_is_incoming(operation)) {
+ gb_operation_get(operation);
+ incoming = true;
+ break;
+ }
+ }
+
+ if (!incoming)
+ break;
+
+ spin_unlock_irq(&connection->lock);
+
+ /* FIXME: flush, not cancel? */
+ gb_operation_cancel_incoming(operation, errno);
+ gb_operation_put(operation);
+
+ spin_lock_irq(&connection->lock);
+ }
+}
+
+/*
+ * _gb_connection_enable() - enable a connection
+ * @connection: connection to enable
+ * @rx: whether to enable incoming requests
+ *
+ * Connection-enable helper for DISABLED->ENABLED, DISABLED->ENABLED_TX, and
+ * ENABLED_TX->ENABLED state transitions.
+ *
+ * Locking: Caller holds connection->mutex.
+ */
+static int _gb_connection_enable(struct gb_connection *connection, bool rx)
+{
+ int ret;
+
+ /* Handle ENABLED_TX -> ENABLED transitions. */
+ if (connection->state == GB_CONNECTION_STATE_ENABLED_TX) {
+ if (!(connection->handler && rx))
+ return 0;
+
+ spin_lock_irq(&connection->lock);
+ connection->state = GB_CONNECTION_STATE_ENABLED;
+ spin_unlock_irq(&connection->lock);
+
+ return 0;
+ }
+
+ ret = gb_connection_hd_cport_enable(connection);
+ if (ret)
+ return ret;
+
+ ret = gb_connection_svc_connection_create(connection);
+ if (ret)
+ goto err_hd_cport_clear;
+
+ ret = gb_connection_hd_cport_connected(connection);
+ if (ret)
+ goto err_svc_connection_destroy;
+
+ spin_lock_irq(&connection->lock);
+ if (connection->handler && rx)
+ connection->state = GB_CONNECTION_STATE_ENABLED;
+ else
+ connection->state = GB_CONNECTION_STATE_ENABLED_TX;
+ spin_unlock_irq(&connection->lock);
+
+ ret = gb_connection_control_connected(connection);
+ if (ret)
+ goto err_control_disconnecting;
+
+ return 0;
+
+err_control_disconnecting:
+ spin_lock_irq(&connection->lock);
+ connection->state = GB_CONNECTION_STATE_DISCONNECTING;
+ gb_connection_cancel_operations(connection, -ESHUTDOWN);
+ spin_unlock_irq(&connection->lock);
+
+ /* Transmit queue should already be empty. */
+ gb_connection_hd_cport_flush(connection);
+
+ gb_connection_control_disconnecting(connection);
+ gb_connection_cport_shutdown_phase_1(connection);
+ gb_connection_hd_cport_quiesce(connection);
+ gb_connection_cport_shutdown_phase_2(connection);
+ gb_connection_control_disconnected(connection);
+ connection->state = GB_CONNECTION_STATE_DISABLED;
+err_svc_connection_destroy:
+ gb_connection_svc_connection_destroy(connection);
+err_hd_cport_clear:
+ gb_connection_hd_cport_clear(connection);
+
+ gb_connection_hd_cport_disable(connection);
+
+ return ret;
+}
+
+int gb_connection_enable(struct gb_connection *connection)
+{
+ int ret = 0;
+
+ mutex_lock(&connection->mutex);
+
+ if (connection->state == GB_CONNECTION_STATE_ENABLED)
+ goto out_unlock;
+
+ ret = _gb_connection_enable(connection, true);
+ if (!ret)
+ trace_gb_connection_enable(connection);
+
+out_unlock:
+ mutex_unlock(&connection->mutex);
+
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_connection_enable);
+
+int gb_connection_enable_tx(struct gb_connection *connection)
+{
+ int ret = 0;
+
+ mutex_lock(&connection->mutex);
+
+ if (connection->state == GB_CONNECTION_STATE_ENABLED) {
+ ret = -EINVAL;
+ goto out_unlock;
+ }
+
+ if (connection->state == GB_CONNECTION_STATE_ENABLED_TX)
+ goto out_unlock;
+
+ ret = _gb_connection_enable(connection, false);
+ if (!ret)
+ trace_gb_connection_enable(connection);
+
+out_unlock:
+ mutex_unlock(&connection->mutex);
+
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_connection_enable_tx);
+
+void gb_connection_disable_rx(struct gb_connection *connection)
+{
+ mutex_lock(&connection->mutex);
+
+ spin_lock_irq(&connection->lock);
+ if (connection->state != GB_CONNECTION_STATE_ENABLED) {
+ spin_unlock_irq(&connection->lock);
+ goto out_unlock;
+ }
+ connection->state = GB_CONNECTION_STATE_ENABLED_TX;
+ gb_connection_flush_incoming_operations(connection, -ESHUTDOWN);
+ spin_unlock_irq(&connection->lock);
+
+ trace_gb_connection_disable(connection);
+
+out_unlock:
+ mutex_unlock(&connection->mutex);
+}
+EXPORT_SYMBOL_GPL(gb_connection_disable_rx);
+
+void gb_connection_mode_switch_prepare(struct gb_connection *connection)
+{
+ connection->mode_switch = true;
+}
+
+void gb_connection_mode_switch_complete(struct gb_connection *connection)
+{
+ gb_connection_svc_connection_destroy(connection);
+ gb_connection_hd_cport_clear(connection);
+
+ gb_connection_hd_cport_disable(connection);
+
+ connection->mode_switch = false;
+}
+
+void gb_connection_disable(struct gb_connection *connection)
+{
+ mutex_lock(&connection->mutex);
+
+ if (connection->state == GB_CONNECTION_STATE_DISABLED)
+ goto out_unlock;
+
+ trace_gb_connection_disable(connection);
+
+ spin_lock_irq(&connection->lock);
+ connection->state = GB_CONNECTION_STATE_DISCONNECTING;
+ gb_connection_cancel_operations(connection, -ESHUTDOWN);
+ spin_unlock_irq(&connection->lock);
+
+ gb_connection_hd_cport_flush(connection);
+
+ gb_connection_control_disconnecting(connection);
+ gb_connection_cport_shutdown_phase_1(connection);
+ gb_connection_hd_cport_quiesce(connection);
+ gb_connection_cport_shutdown_phase_2(connection);
+ gb_connection_control_disconnected(connection);
+
+ connection->state = GB_CONNECTION_STATE_DISABLED;
+
+ /* control-connection tear down is deferred when mode switching */
+ if (!connection->mode_switch) {
+ gb_connection_svc_connection_destroy(connection);
+ gb_connection_hd_cport_clear(connection);
+
+ gb_connection_hd_cport_disable(connection);
+ }
+
+out_unlock:
+ mutex_unlock(&connection->mutex);
+}
+EXPORT_SYMBOL_GPL(gb_connection_disable);
+
+/* Disable a connection without communicating with the remote end. */
+void gb_connection_disable_forced(struct gb_connection *connection)
+{
+ mutex_lock(&connection->mutex);
+
+ if (connection->state == GB_CONNECTION_STATE_DISABLED)
+ goto out_unlock;
+
+ trace_gb_connection_disable(connection);
+
+ spin_lock_irq(&connection->lock);
+ connection->state = GB_CONNECTION_STATE_DISABLED;
+ gb_connection_cancel_operations(connection, -ESHUTDOWN);
+ spin_unlock_irq(&connection->lock);
+
+ gb_connection_hd_cport_flush(connection);
+
+ gb_connection_svc_connection_destroy(connection);
+ gb_connection_hd_cport_clear(connection);
+
+ gb_connection_hd_cport_disable(connection);
+out_unlock:
+ mutex_unlock(&connection->mutex);
+}
+EXPORT_SYMBOL_GPL(gb_connection_disable_forced);
+
+/* Caller must have disabled the connection before destroying it. */
+void gb_connection_destroy(struct gb_connection *connection)
+{
+ if (!connection)
+ return;
+
+ if (WARN_ON(connection->state != GB_CONNECTION_STATE_DISABLED))
+ gb_connection_disable(connection);
+
+ mutex_lock(&gb_connection_mutex);
+
+ spin_lock_irq(&gb_connections_lock);
+ list_del(&connection->bundle_links);
+ list_del(&connection->hd_links);
+ spin_unlock_irq(&gb_connections_lock);
+
+ destroy_workqueue(connection->wq);
+
+ gb_hd_cport_release(connection->hd, connection->hd_cport_id);
+ connection->hd_cport_id = CPORT_ID_BAD;
+
+ mutex_unlock(&gb_connection_mutex);
+
+ gb_connection_put(connection);
+}
+EXPORT_SYMBOL_GPL(gb_connection_destroy);
+
+void gb_connection_latency_tag_enable(struct gb_connection *connection)
+{
+ struct gb_host_device *hd = connection->hd;
+ int ret;
+
+ if (!hd->driver->latency_tag_enable)
+ return;
+
+ ret = hd->driver->latency_tag_enable(hd, connection->hd_cport_id);
+ if (ret) {
+ dev_err(&connection->hd->dev,
+ "%s: failed to enable latency tag: %d\n",
+ connection->name, ret);
+ }
+}
+EXPORT_SYMBOL_GPL(gb_connection_latency_tag_enable);
+
+void gb_connection_latency_tag_disable(struct gb_connection *connection)
+{
+ struct gb_host_device *hd = connection->hd;
+ int ret;
+
+ if (!hd->driver->latency_tag_disable)
+ return;
+
+ ret = hd->driver->latency_tag_disable(hd, connection->hd_cport_id);
+ if (ret) {
+ dev_err(&connection->hd->dev,
+ "%s: failed to disable latency tag: %d\n",
+ connection->name, ret);
+ }
+}
+EXPORT_SYMBOL_GPL(gb_connection_latency_tag_disable);
--- /dev/null
+++ b/drivers/greybus/connection.h
@@ -0,0 +1,129 @@
+/*
+ * Greybus connections
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __CONNECTION_H
+#define __CONNECTION_H
+
+#include <linux/list.h>
+#include <linux/kfifo.h>
+
+#define GB_CONNECTION_FLAG_CSD BIT(0)
+#define GB_CONNECTION_FLAG_NO_FLOWCTRL BIT(1)
+#define GB_CONNECTION_FLAG_OFFLOADED BIT(2)
+#define GB_CONNECTION_FLAG_CDSI1 BIT(3)
+#define GB_CONNECTION_FLAG_CONTROL BIT(4)
+#define GB_CONNECTION_FLAG_HIGH_PRIO BIT(5)
+
+#define GB_CONNECTION_FLAG_CORE_MASK GB_CONNECTION_FLAG_CONTROL
+
+enum gb_connection_state {
+ GB_CONNECTION_STATE_DISABLED = 0,
+ GB_CONNECTION_STATE_ENABLED_TX = 1,
+ GB_CONNECTION_STATE_ENABLED = 2,
+ GB_CONNECTION_STATE_DISCONNECTING = 3,
+};
+
+struct gb_operation;
+
+typedef int (*gb_request_handler_t)(struct gb_operation *);
+
+struct gb_connection {
+ struct gb_host_device *hd;
+ struct gb_interface *intf;
+ struct gb_bundle *bundle;
+ struct kref kref;
+ u16 hd_cport_id;
+ u16 intf_cport_id;
+
+ struct list_head hd_links;
+ struct list_head bundle_links;
+
+ gb_request_handler_t handler;
+ unsigned long flags;
+
+ struct mutex mutex;
+ spinlock_t lock;
+ enum gb_connection_state state;
+ struct list_head operations;
+
+ char name[16];
+ struct workqueue_struct *wq;
+
+ atomic_t op_cycle;
+
+ void *private;
+
+ bool mode_switch;
+};
+
+struct gb_connection *gb_connection_create_static(struct gb_host_device *hd,
+ u16 hd_cport_id, gb_request_handler_t handler);
+struct gb_connection *gb_connection_create_control(struct gb_interface *intf);
+struct gb_connection *gb_connection_create(struct gb_bundle *bundle,
+ u16 cport_id, gb_request_handler_t handler);
+struct gb_connection *gb_connection_create_flags(struct gb_bundle *bundle,
+ u16 cport_id, gb_request_handler_t handler,
+ unsigned long flags);
+struct gb_connection *gb_connection_create_offloaded(struct gb_bundle *bundle,
+ u16 cport_id, unsigned long flags);
+void gb_connection_destroy(struct gb_connection *connection);
+
+static inline bool gb_connection_is_static(struct gb_connection *connection)
+{
+ return !connection->intf;
+}
+
+int gb_connection_enable(struct gb_connection *connection);
+int gb_connection_enable_tx(struct gb_connection *connection);
+void gb_connection_disable_rx(struct gb_connection *connection);
+void gb_connection_disable(struct gb_connection *connection);
+void gb_connection_disable_forced(struct gb_connection *connection);
+
+void gb_connection_mode_switch_prepare(struct gb_connection *connection);
+void gb_connection_mode_switch_complete(struct gb_connection *connection);
+
+void greybus_data_rcvd(struct gb_host_device *hd, u16 cport_id,
+ u8 *data, size_t length);
+
+void gb_connection_latency_tag_enable(struct gb_connection *connection);
+void gb_connection_latency_tag_disable(struct gb_connection *connection);
+
+static inline bool gb_connection_e2efc_enabled(struct gb_connection *connection)
+{
+ return !(connection->flags & GB_CONNECTION_FLAG_CSD);
+}
+
+static inline bool
+gb_connection_flow_control_disabled(struct gb_connection *connection)
+{
+ return connection->flags & GB_CONNECTION_FLAG_NO_FLOWCTRL;
+}
+
+static inline bool gb_connection_is_offloaded(struct gb_connection *connection)
+{
+ return connection->flags & GB_CONNECTION_FLAG_OFFLOADED;
+}
+
+static inline bool gb_connection_is_control(struct gb_connection *connection)
+{
+ return connection->flags & GB_CONNECTION_FLAG_CONTROL;
+}
+
+static inline void *gb_connection_get_data(struct gb_connection *connection)
+{
+ return connection->private;
+}
+
+static inline void gb_connection_set_data(struct gb_connection *connection,
+ void *data)
+{
+ connection->private = data;
+}
+
+#endif /* __CONNECTION_H */
--- /dev/null
+++ b/drivers/greybus/control.c
@@ -0,0 +1,635 @@
+/*
+ * Greybus CPort control protocol.
+ *
+ * Copyright 2015 Google Inc.
+ * Copyright 2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include "greybus.h"
+
+/* Highest control-protocol version supported */
+#define GB_CONTROL_VERSION_MAJOR 0
+#define GB_CONTROL_VERSION_MINOR 1
+
+
+static int gb_control_get_version(struct gb_control *control)
+{
+ struct gb_interface *intf = control->connection->intf;
+ struct gb_control_version_request request;
+ struct gb_control_version_response response;
+ int ret;
+
+ request.major = GB_CONTROL_VERSION_MAJOR;
+ request.minor = GB_CONTROL_VERSION_MINOR;
+
+ ret = gb_operation_sync(control->connection,
+ GB_CONTROL_TYPE_VERSION,
+ &request, sizeof(request), &response,
+ sizeof(response));
+ if (ret) {
+ dev_err(&intf->dev,
+ "failed to get control-protocol version: %d\n",
+ ret);
+ return ret;
+ }
+
+ if (response.major > request.major) {
+ dev_err(&intf->dev,
+ "unsupported major control-protocol version (%u > %u)\n",
+ response.major, request.major);
+ return -ENOTSUPP;
+ }
+
+ control->protocol_major = response.major;
+ control->protocol_minor = response.minor;
+
+ dev_dbg(&intf->dev, "%s - %u.%u\n", __func__, response.major,
+ response.minor);
+
+ return 0;
+}
+
+static int gb_control_get_bundle_version(struct gb_control *control,
+ struct gb_bundle *bundle)
+{
+ struct gb_interface *intf = control->connection->intf;
+ struct gb_control_bundle_version_request request;
+ struct gb_control_bundle_version_response response;
+ int ret;
+
+ request.bundle_id = bundle->id;
+
+ ret = gb_operation_sync(control->connection,
+ GB_CONTROL_TYPE_BUNDLE_VERSION,
+ &request, sizeof(request),
+ &response, sizeof(response));
+ if (ret) {
+ dev_err(&intf->dev,
+ "failed to get bundle %u class version: %d\n",
+ bundle->id, ret);
+ return ret;
+ }
+
+ bundle->class_major = response.major;
+ bundle->class_minor = response.minor;
+
+ dev_dbg(&intf->dev, "%s - %u: %u.%u\n", __func__, bundle->id,
+ response.major, response.minor);
+
+ return 0;
+}
+
+int gb_control_get_bundle_versions(struct gb_control *control)
+{
+ struct gb_interface *intf = control->connection->intf;
+ struct gb_bundle *bundle;
+ int ret;
+
+ if (!control->has_bundle_version)
+ return 0;
+
+ list_for_each_entry(bundle, &intf->bundles, links) {
+ ret = gb_control_get_bundle_version(control, bundle);
+ if (ret)
+ return ret;
+ }
+
+ return 0;
+}
+
+/* Get Manifest's size from the interface */
+int gb_control_get_manifest_size_operation(struct gb_interface *intf)
+{
+ struct gb_control_get_manifest_size_response response;
+ struct gb_connection *connection = intf->control->connection;
+ int ret;
+
+ ret = gb_operation_sync(connection, GB_CONTROL_TYPE_GET_MANIFEST_SIZE,
+ NULL, 0, &response, sizeof(response));
+ if (ret) {
+ dev_err(&connection->intf->dev,
+ "failed to get manifest size: %d\n", ret);
+ return ret;
+ }
+
+ return le16_to_cpu(response.size);
+}
+
+/* Reads Manifest from the interface */
+int gb_control_get_manifest_operation(struct gb_interface *intf, void *manifest,
+ size_t size)
+{
+ struct gb_connection *connection = intf->control->connection;
+
+ return gb_operation_sync(connection, GB_CONTROL_TYPE_GET_MANIFEST,
+ NULL, 0, manifest, size);
+}
+
+int gb_control_connected_operation(struct gb_control *control, u16 cport_id)
+{
+ struct gb_control_connected_request request;
+
+ request.cport_id = cpu_to_le16(cport_id);
+ return gb_operation_sync(control->connection, GB_CONTROL_TYPE_CONNECTED,
+ &request, sizeof(request), NULL, 0);
+}
+
+int gb_control_disconnected_operation(struct gb_control *control, u16 cport_id)
+{
+ struct gb_control_disconnected_request request;
+
+ request.cport_id = cpu_to_le16(cport_id);
+ return gb_operation_sync(control->connection,
+ GB_CONTROL_TYPE_DISCONNECTED, &request,
+ sizeof(request), NULL, 0);
+}
+
+int gb_control_disconnecting_operation(struct gb_control *control,
+ u16 cport_id)
+{
+ struct gb_control_disconnecting_request *request;
+ struct gb_operation *operation;
+ int ret;
+
+ operation = gb_operation_create_core(control->connection,
+ GB_CONTROL_TYPE_DISCONNECTING,
+ sizeof(*request), 0, 0,
+ GFP_KERNEL);
+ if (!operation)
+ return -ENOMEM;
+
+ request = operation->request->payload;
+ request->cport_id = cpu_to_le16(cport_id);
+
+ ret = gb_operation_request_send_sync(operation);
+ if (ret) {
+ dev_err(&control->dev, "failed to send disconnecting: %d\n",
+ ret);
+ }
+
+ gb_operation_put(operation);
+
+ return ret;
+}
+
+int gb_control_mode_switch_operation(struct gb_control *control)
+{
+ struct gb_operation *operation;
+ int ret;
+
+ operation = gb_operation_create_core(control->connection,
+ GB_CONTROL_TYPE_MODE_SWITCH,
+ 0, 0, GB_OPERATION_FLAG_UNIDIRECTIONAL,
+ GFP_KERNEL);
+ if (!operation)
+ return -ENOMEM;
+
+ ret = gb_operation_request_send_sync(operation);
+ if (ret)
+ dev_err(&control->dev, "failed to send mode switch: %d\n", ret);
+
+ gb_operation_put(operation);
+
+ return ret;
+}
+
+int gb_control_timesync_enable(struct gb_control *control, u8 count,
+ u64 frame_time, u32 strobe_delay, u32 refclk)
+{
+ struct gb_control_timesync_enable_request request;
+
+ request.count = count;
+ request.frame_time = cpu_to_le64(frame_time);
+ request.strobe_delay = cpu_to_le32(strobe_delay);
+ request.refclk = cpu_to_le32(refclk);
+ return gb_operation_sync(control->connection,
+ GB_CONTROL_TYPE_TIMESYNC_ENABLE, &request,
+ sizeof(request), NULL, 0);
+}
+
+int gb_control_timesync_disable(struct gb_control *control)
+{
+ return gb_operation_sync(control->connection,
+ GB_CONTROL_TYPE_TIMESYNC_DISABLE, NULL, 0,
+ NULL, 0);
+}
+
+int gb_control_timesync_get_last_event(struct gb_control *control,
+ u64 *frame_time)
+{
+ struct gb_control_timesync_get_last_event_response response;
+ int ret;
+
+ ret = gb_operation_sync(control->connection,
+ GB_CONTROL_TYPE_TIMESYNC_GET_LAST_EVENT,
+ NULL, 0, &response, sizeof(response));
+ if (!ret)
+ *frame_time = le64_to_cpu(response.frame_time);
+ return ret;
+}
+
+int gb_control_timesync_authoritative(struct gb_control *control,
+ u64 *frame_time)
+{
+ struct gb_control_timesync_authoritative_request request;
+ int i;
+
+ for (i = 0; i < GB_TIMESYNC_MAX_STROBES; i++)
+ request.frame_time[i] = cpu_to_le64(frame_time[i]);
+
+ return gb_operation_sync(control->connection,
+ GB_CONTROL_TYPE_TIMESYNC_AUTHORITATIVE,
+ &request, sizeof(request),
+ NULL, 0);
+}
+
+static int gb_control_bundle_pm_status_map(u8 status)
+{
+ switch (status) {
+ case GB_CONTROL_BUNDLE_PM_INVAL:
+ return -EINVAL;
+ case GB_CONTROL_BUNDLE_PM_BUSY:
+ return -EBUSY;
+ case GB_CONTROL_BUNDLE_PM_NA:
+ return -ENOMSG;
+ case GB_CONTROL_BUNDLE_PM_FAIL:
+ default:
+ return -EREMOTEIO;
+ }
+}
+
+int gb_control_bundle_suspend(struct gb_control *control, u8 bundle_id)
+{
+ struct gb_control_bundle_pm_request request;
+ struct gb_control_bundle_pm_response response;
+ int ret;
+
+ request.bundle_id = bundle_id;
+ ret = gb_operation_sync(control->connection,
+ GB_CONTROL_TYPE_BUNDLE_SUSPEND, &request,
+ sizeof(request), &response, sizeof(response));
+ if (ret) {
+ dev_err(&control->dev, "failed to send bundle %u suspend: %d\n",
+ bundle_id, ret);
+ return ret;
+ }
+
+ if (response.status != GB_CONTROL_BUNDLE_PM_OK) {
+ dev_err(&control->dev, "failed to suspend bundle %u: %d\n",
+ bundle_id, response.status);
+ return gb_control_bundle_pm_status_map(response.status);
+ }
+
+ return 0;
+}
+
+int gb_control_bundle_resume(struct gb_control *control, u8 bundle_id)
+{
+ struct gb_control_bundle_pm_request request;
+ struct gb_control_bundle_pm_response response;
+ int ret;
+
+ request.bundle_id = bundle_id;
+ ret = gb_operation_sync(control->connection,
+ GB_CONTROL_TYPE_BUNDLE_RESUME, &request,
+ sizeof(request), &response, sizeof(response));
+ if (ret) {
+ dev_err(&control->dev, "failed to send bundle %u resume: %d\n",
+ bundle_id, ret);
+ return ret;
+ }
+
+ if (response.status != GB_CONTROL_BUNDLE_PM_OK) {
+ dev_err(&control->dev, "failed to resume bundle %u: %d\n",
+ bundle_id, response.status);
+ return gb_control_bundle_pm_status_map(response.status);
+ }
+
+ return 0;
+}
+
+int gb_control_bundle_deactivate(struct gb_control *control, u8 bundle_id)
+{
+ struct gb_control_bundle_pm_request request;
+ struct gb_control_bundle_pm_response response;
+ int ret;
+
+ request.bundle_id = bundle_id;
+ ret = gb_operation_sync(control->connection,
+ GB_CONTROL_TYPE_BUNDLE_DEACTIVATE, &request,
+ sizeof(request), &response, sizeof(response));
+ if (ret) {
+ dev_err(&control->dev,
+ "failed to send bundle %u deactivate: %d\n", bundle_id,
+ ret);
+ return ret;
+ }
+
+ if (response.status != GB_CONTROL_BUNDLE_PM_OK) {
+ dev_err(&control->dev, "failed to deactivate bundle %u: %d\n",
+ bundle_id, response.status);
+ return gb_control_bundle_pm_status_map(response.status);
+ }
+
+ return 0;
+}
+
+int gb_control_bundle_activate(struct gb_control *control, u8 bundle_id)
+{
+ struct gb_control_bundle_pm_request request;
+ struct gb_control_bundle_pm_response response;
+ int ret;
+
+ if (!control->has_bundle_activate)
+ return 0;
+
+ request.bundle_id = bundle_id;
+ ret = gb_operation_sync(control->connection,
+ GB_CONTROL_TYPE_BUNDLE_ACTIVATE, &request,
+ sizeof(request), &response, sizeof(response));
+ if (ret) {
+ dev_err(&control->dev,
+ "failed to send bundle %u activate: %d\n", bundle_id,
+ ret);
+ return ret;
+ }
+
+ if (response.status != GB_CONTROL_BUNDLE_PM_OK) {
+ dev_err(&control->dev, "failed to activate bundle %u: %d\n",
+ bundle_id, response.status);
+ return gb_control_bundle_pm_status_map(response.status);
+ }
+
+ return 0;
+}
+
+static int gb_control_interface_pm_status_map(u8 status)
+{
+ switch (status) {
+ case GB_CONTROL_INTF_PM_BUSY:
+ return -EBUSY;
+ case GB_CONTROL_INTF_PM_NA:
+ return -ENOMSG;
+ default:
+ return -EREMOTEIO;
+ }
+}
+
+int gb_control_interface_suspend_prepare(struct gb_control *control)
+{
+ struct gb_control_intf_pm_response response;
+ int ret;
+
+ ret = gb_operation_sync(control->connection,
+ GB_CONTROL_TYPE_INTF_SUSPEND_PREPARE, NULL, 0,
+ &response, sizeof(response));
+ if (ret) {
+ dev_err(&control->dev,
+ "failed to send interface suspend prepare: %d\n", ret);
+ return ret;
+ }
+
+ if (response.status != GB_CONTROL_INTF_PM_OK) {
+ dev_err(&control->dev, "interface error while preparing suspend: %d\n",
+ response.status);
+ return gb_control_interface_pm_status_map(response.status);
+ }
+
+ return 0;
+}
+
+int gb_control_interface_deactivate_prepare(struct gb_control *control)
+{
+ struct gb_control_intf_pm_response response;
+ int ret;
+
+ ret = gb_operation_sync(control->connection,
+ GB_CONTROL_TYPE_INTF_DEACTIVATE_PREPARE, NULL,
+ 0, &response, sizeof(response));
+ if (ret) {
+ dev_err(&control->dev, "failed to send interface deactivate prepare: %d\n",
+ ret);
+ return ret;
+ }
+
+ if (response.status != GB_CONTROL_INTF_PM_OK) {
+ dev_err(&control->dev, "interface error while preparing deactivate: %d\n",
+ response.status);
+ return gb_control_interface_pm_status_map(response.status);
+ }
+
+ return 0;
+}
+
+int gb_control_interface_hibernate_abort(struct gb_control *control)
+{
+ struct gb_control_intf_pm_response response;
+ int ret;
+
+ ret = gb_operation_sync(control->connection,
+ GB_CONTROL_TYPE_INTF_HIBERNATE_ABORT, NULL, 0,
+ &response, sizeof(response));
+ if (ret) {
+ dev_err(&control->dev,
+ "failed to send interface aborting hibernate: %d\n",
+ ret);
+ return ret;
+ }
+
+ if (response.status != GB_CONTROL_INTF_PM_OK) {
+ dev_err(&control->dev, "interface error while aborting hibernate: %d\n",
+ response.status);
+ return gb_control_interface_pm_status_map(response.status);
+ }
+
+ return 0;
+}
+
+static ssize_t vendor_string_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_control *control = to_gb_control(dev);
+
+ return scnprintf(buf, PAGE_SIZE, "%s\n", control->vendor_string);
+}
+static DEVICE_ATTR_RO(vendor_string);
+
+static ssize_t product_string_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_control *control = to_gb_control(dev);
+
+ return scnprintf(buf, PAGE_SIZE, "%s\n", control->product_string);
+}
+static DEVICE_ATTR_RO(product_string);
+
+static struct attribute *control_attrs[] = {
+ &dev_attr_vendor_string.attr,
+ &dev_attr_product_string.attr,
+ NULL,
+};
+ATTRIBUTE_GROUPS(control);
+
+static void gb_control_release(struct device *dev)
+{
+ struct gb_control *control = to_gb_control(dev);
+
+ gb_connection_destroy(control->connection);
+
+ kfree(control->vendor_string);
+ kfree(control->product_string);
+
+ kfree(control);
+}
+
+struct device_type greybus_control_type = {
+ .name = "greybus_control",
+ .release = gb_control_release,
+};
+
+struct gb_control *gb_control_create(struct gb_interface *intf)
+{
+ struct gb_connection *connection;
+ struct gb_control *control;
+
+ control = kzalloc(sizeof(*control), GFP_KERNEL);
+ if (!control)
+ return ERR_PTR(-ENOMEM);
+
+ control->intf = intf;
+
+ connection = gb_connection_create_control(intf);
+ if (IS_ERR(connection)) {
+ dev_err(&intf->dev,
+ "failed to create control connection: %ld\n",
+ PTR_ERR(connection));
+ kfree(control);
+ return ERR_CAST(connection);
+ }
+
+ control->connection = connection;
+
+ control->dev.parent = &intf->dev;
+ control->dev.bus = &greybus_bus_type;
+ control->dev.type = &greybus_control_type;
+ control->dev.groups = control_groups;
+ control->dev.dma_mask = intf->dev.dma_mask;
+ device_initialize(&control->dev);
+ dev_set_name(&control->dev, "%s.ctrl", dev_name(&intf->dev));
+
+ gb_connection_set_data(control->connection, control);
+
+ return control;
+}
+
+int gb_control_enable(struct gb_control *control)
+{
+ int ret;
+
+ dev_dbg(&control->connection->intf->dev, "%s\n", __func__);
+
+ ret = gb_connection_enable_tx(control->connection);
+ if (ret) {
+ dev_err(&control->connection->intf->dev,
+ "failed to enable control connection: %d\n",
+ ret);
+ return ret;
+ }
+
+ ret = gb_control_get_version(control);
+ if (ret)
+ goto err_disable_connection;
+
+ if (control->protocol_major > 0 || control->protocol_minor > 1)
+ control->has_bundle_version = true;
+
+ /* FIXME: use protocol version instead */
+ if (!(control->intf->quirks & GB_INTERFACE_QUIRK_NO_BUNDLE_ACTIVATE))
+ control->has_bundle_activate = true;
+
+ return 0;
+
+err_disable_connection:
+ gb_connection_disable(control->connection);
+
+ return ret;
+}
+
+void gb_control_disable(struct gb_control *control)
+{
+ dev_dbg(&control->connection->intf->dev, "%s\n", __func__);
+
+ if (control->intf->disconnected)
+ gb_connection_disable_forced(control->connection);
+ else
+ gb_connection_disable(control->connection);
+}
+
+int gb_control_suspend(struct gb_control *control)
+{
+ gb_connection_disable(control->connection);
+
+ return 0;
+}
+
+int gb_control_resume(struct gb_control *control)
+{
+ int ret;
+
+ ret = gb_connection_enable_tx(control->connection);
+ if (ret) {
+ dev_err(&control->connection->intf->dev,
+ "failed to enable control connection: %d\n", ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+int gb_control_add(struct gb_control *control)
+{
+ int ret;
+
+ ret = device_add(&control->dev);
+ if (ret) {
+ dev_err(&control->dev,
+ "failed to register control device: %d\n",
+ ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+void gb_control_del(struct gb_control *control)
+{
+ if (device_is_registered(&control->dev))
+ device_del(&control->dev);
+}
+
+struct gb_control *gb_control_get(struct gb_control *control)
+{
+ get_device(&control->dev);
+
+ return control;
+}
+
+void gb_control_put(struct gb_control *control)
+{
+ put_device(&control->dev);
+}
+
+void gb_control_mode_switch_prepare(struct gb_control *control)
+{
+ gb_connection_mode_switch_prepare(control->connection);
+}
+
+void gb_control_mode_switch_complete(struct gb_control *control)
+{
+ gb_connection_mode_switch_complete(control->connection);
+}
--- /dev/null
+++ b/drivers/greybus/control.h
@@ -0,0 +1,65 @@
+/*
+ * Greybus CPort control protocol
+ *
+ * Copyright 2015 Google Inc.
+ * Copyright 2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __CONTROL_H
+#define __CONTROL_H
+
+struct gb_control {
+ struct device dev;
+ struct gb_interface *intf;
+
+ struct gb_connection *connection;
+
+ u8 protocol_major;
+ u8 protocol_minor;
+
+ bool has_bundle_activate;
+ bool has_bundle_version;
+
+ char *vendor_string;
+ char *product_string;
+};
+#define to_gb_control(d) container_of(d, struct gb_control, dev)
+
+struct gb_control *gb_control_create(struct gb_interface *intf);
+int gb_control_enable(struct gb_control *control);
+void gb_control_disable(struct gb_control *control);
+int gb_control_suspend(struct gb_control *control);
+int gb_control_resume(struct gb_control *control);
+int gb_control_add(struct gb_control *control);
+void gb_control_del(struct gb_control *control);
+struct gb_control *gb_control_get(struct gb_control *control);
+void gb_control_put(struct gb_control *control);
+
+int gb_control_get_bundle_versions(struct gb_control *control);
+int gb_control_connected_operation(struct gb_control *control, u16 cport_id);
+int gb_control_disconnected_operation(struct gb_control *control, u16 cport_id);
+int gb_control_disconnecting_operation(struct gb_control *control,
+ u16 cport_id);
+int gb_control_mode_switch_operation(struct gb_control *control);
+void gb_control_mode_switch_prepare(struct gb_control *control);
+void gb_control_mode_switch_complete(struct gb_control *control);
+int gb_control_get_manifest_size_operation(struct gb_interface *intf);
+int gb_control_get_manifest_operation(struct gb_interface *intf, void *manifest,
+ size_t size);
+int gb_control_timesync_enable(struct gb_control *control, u8 count,
+ u64 frame_time, u32 strobe_delay, u32 refclk);
+int gb_control_timesync_disable(struct gb_control *control);
+int gb_control_timesync_get_last_event(struct gb_control *control,
+ u64 *frame_time);
+int gb_control_timesync_authoritative(struct gb_control *control,
+ u64 *frame_time);
+int gb_control_bundle_suspend(struct gb_control *control, u8 bundle_id);
+int gb_control_bundle_resume(struct gb_control *control, u8 bundle_id);
+int gb_control_bundle_deactivate(struct gb_control *control, u8 bundle_id);
+int gb_control_bundle_activate(struct gb_control *control, u8 bundle_id);
+int gb_control_interface_suspend_prepare(struct gb_control *control);
+int gb_control_interface_deactivate_prepare(struct gb_control *control);
+int gb_control_interface_hibernate_abort(struct gb_control *control);
+#endif /* __CONTROL_H */
--- /dev/null
+++ b/drivers/greybus/core.c
@@ -0,0 +1,361 @@
+/*
+ * Greybus "Core"
+ *
+ * Copyright 2014-2015 Google Inc.
+ * Copyright 2014-2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#define CREATE_TRACE_POINTS
+#include "greybus.h"
+#include "greybus_trace.h"
+
+#define GB_BUNDLE_AUTOSUSPEND_MS 3000
+
+/* Allow greybus to be disabled at boot if needed */
+static bool nogreybus;
+#ifdef MODULE
+module_param(nogreybus, bool, 0444);
+#else
+core_param(nogreybus, nogreybus, bool, 0444);
+#endif
+int greybus_disabled(void)
+{
+ return nogreybus;
+}
+EXPORT_SYMBOL_GPL(greybus_disabled);
+
+static bool greybus_match_one_id(struct gb_bundle *bundle,
+ const struct greybus_bundle_id *id)
+{
+ if ((id->match_flags & GREYBUS_ID_MATCH_VENDOR) &&
+ (id->vendor != bundle->intf->vendor_id))
+ return false;
+
+ if ((id->match_flags & GREYBUS_ID_MATCH_PRODUCT) &&
+ (id->product != bundle->intf->product_id))
+ return false;
+
+ if ((id->match_flags & GREYBUS_ID_MATCH_CLASS) &&
+ (id->class != bundle->class))
+ return false;
+
+ return true;
+}
+
+static const struct greybus_bundle_id *
+greybus_match_id(struct gb_bundle *bundle, const struct greybus_bundle_id *id)
+{
+ if (id == NULL)
+ return NULL;
+
+ for (; id->vendor || id->product || id->class || id->driver_info;
+ id++) {
+ if (greybus_match_one_id(bundle, id))
+ return id;
+ }
+
+ return NULL;
+}
+
+static int greybus_match_device(struct device *dev, struct device_driver *drv)
+{
+ struct greybus_driver *driver = to_greybus_driver(drv);
+ struct gb_bundle *bundle;
+ const struct greybus_bundle_id *id;
+
+ if (!is_gb_bundle(dev))
+ return 0;
+
+ bundle = to_gb_bundle(dev);
+
+ id = greybus_match_id(bundle, driver->id_table);
+ if (id)
+ return 1;
+ /* FIXME - Dynamic ids? */
+ return 0;
+}
+
+static int greybus_uevent(struct device *dev, struct kobj_uevent_env *env)
+{
+ struct gb_host_device *hd;
+ struct gb_module *module = NULL;
+ struct gb_interface *intf = NULL;
+ struct gb_control *control = NULL;
+ struct gb_bundle *bundle = NULL;
+ struct gb_svc *svc = NULL;
+
+ if (is_gb_host_device(dev)) {
+ hd = to_gb_host_device(dev);
+ } else if (is_gb_module(dev)) {
+ module = to_gb_module(dev);
+ hd = module->hd;
+ } else if (is_gb_interface(dev)) {
+ intf = to_gb_interface(dev);
+ module = intf->module;
+ hd = intf->hd;
+ } else if (is_gb_control(dev)) {
+ control = to_gb_control(dev);
+ intf = control->intf;
+ module = intf->module;
+ hd = intf->hd;
+ } else if (is_gb_bundle(dev)) {
+ bundle = to_gb_bundle(dev);
+ intf = bundle->intf;
+ module = intf->module;
+ hd = intf->hd;
+ } else if (is_gb_svc(dev)) {
+ svc = to_gb_svc(dev);
+ hd = svc->hd;
+ } else {
+ dev_WARN(dev, "uevent for unknown greybus device \"type\"!\n");
+ return -EINVAL;
+ }
+
+ if (add_uevent_var(env, "BUS=%u", hd->bus_id))
+ return -ENOMEM;
+
+ if (module) {
+ if (add_uevent_var(env, "MODULE=%u", module->module_id))
+ return -ENOMEM;
+ }
+
+ if (intf) {
+ if (add_uevent_var(env, "INTERFACE=%u", intf->interface_id))
+ return -ENOMEM;
+ if (add_uevent_var(env, "GREYBUS_ID=%08x/%08x",
+ intf->vendor_id, intf->product_id))
+ return -ENOMEM;
+ }
+
+ if (bundle) {
+ // FIXME
+ // add a uevent that can "load" a bundle type
+ // This is what we need to bind a driver to so use the info
+ // in gmod here as well
+
+ if (add_uevent_var(env, "BUNDLE=%u", bundle->id))
+ return -ENOMEM;
+ if (add_uevent_var(env, "BUNDLE_CLASS=%02x", bundle->class))
+ return -ENOMEM;
+ }
+
+ return 0;
+}
+
+static void greybus_shutdown(struct device *dev)
+{
+ if (is_gb_host_device(dev)) {
+ struct gb_host_device *hd;
+
+ hd = to_gb_host_device(dev);
+ gb_hd_shutdown(hd);
+ }
+}
+
+struct bus_type greybus_bus_type = {
+ .name = "greybus",
+ .match = greybus_match_device,
+ .uevent = greybus_uevent,
+ .shutdown = greybus_shutdown,
+};
+
+static int greybus_probe(struct device *dev)
+{
+ struct greybus_driver *driver = to_greybus_driver(dev->driver);
+ struct gb_bundle *bundle = to_gb_bundle(dev);
+ const struct greybus_bundle_id *id;
+ int retval;
+
+ /* match id */
+ id = greybus_match_id(bundle, driver->id_table);
+ if (!id)
+ return -ENODEV;
+
+ retval = pm_runtime_get_sync(&bundle->intf->dev);
+ if (retval < 0) {
+ pm_runtime_put_noidle(&bundle->intf->dev);
+ return retval;
+ }
+
+ retval = gb_control_bundle_activate(bundle->intf->control, bundle->id);
+ if (retval) {
+ pm_runtime_put(&bundle->intf->dev);
+ return retval;
+ }
+
+ /*
+ * Unbound bundle devices are always deactivated. During probe, the
+ * Runtime PM is set to enabled and active and the usage count is
+ * incremented. If the driver supports runtime PM, it should call
+ * pm_runtime_put() in its probe routine and pm_runtime_get_sync()
+ * in remove routine.
+ */
+ pm_runtime_set_autosuspend_delay(dev, GB_BUNDLE_AUTOSUSPEND_MS);
+ pm_runtime_use_autosuspend(dev);
+ pm_runtime_get_noresume(dev);
+ pm_runtime_set_active(dev);
+ pm_runtime_enable(dev);
+
+ retval = driver->probe(bundle, id);
+ if (retval) {
+ /*
+ * Catch buggy drivers that fail to destroy their connections.
+ */
+ WARN_ON(!list_empty(&bundle->connections));
+
+ gb_control_bundle_deactivate(bundle->intf->control, bundle->id);
+
+ pm_runtime_disable(dev);
+ pm_runtime_set_suspended(dev);
+ pm_runtime_put_noidle(dev);
+ pm_runtime_dont_use_autosuspend(dev);
+ pm_runtime_put(&bundle->intf->dev);
+
+ return retval;
+ }
+
+ gb_timesync_schedule_synchronous(bundle->intf);
+
+ pm_runtime_put(&bundle->intf->dev);
+
+ return 0;
+}
+
+static int greybus_remove(struct device *dev)
+{
+ struct greybus_driver *driver = to_greybus_driver(dev->driver);
+ struct gb_bundle *bundle = to_gb_bundle(dev);
+ struct gb_connection *connection;
+ int retval;
+
+ retval = pm_runtime_get_sync(dev);
+ if (retval < 0)
+ dev_err(dev, "failed to resume bundle: %d\n", retval);
+
+ /*
+ * Disable (non-offloaded) connections early in case the interface is
+ * already gone to avoid unceccessary operation timeouts during
+ * driver disconnect. Otherwise, only disable incoming requests.
+ */
+ list_for_each_entry(connection, &bundle->connections, bundle_links) {
+ if (gb_connection_is_offloaded(connection))
+ continue;
+
+ if (bundle->intf->disconnected)
+ gb_connection_disable_forced(connection);
+ else
+ gb_connection_disable_rx(connection);
+ }
+
+ driver->disconnect(bundle);
+
+ /* Catch buggy drivers that fail to destroy their connections. */
+ WARN_ON(!list_empty(&bundle->connections));
+
+ if (!bundle->intf->disconnected)
+ gb_control_bundle_deactivate(bundle->intf->control, bundle->id);
+
+ pm_runtime_put_noidle(dev);
+ pm_runtime_disable(dev);
+ pm_runtime_set_suspended(dev);
+ pm_runtime_dont_use_autosuspend(dev);
+ pm_runtime_put_noidle(dev);
+
+ return 0;
+}
+
+int greybus_register_driver(struct greybus_driver *driver, struct module *owner,
+ const char *mod_name)
+{
+ int retval;
+
+ if (greybus_disabled())
+ return -ENODEV;
+
+ driver->driver.bus = &greybus_bus_type;
+ driver->driver.name = driver->name;
+ driver->driver.probe = greybus_probe;
+ driver->driver.remove = greybus_remove;
+ driver->driver.owner = owner;
+ driver->driver.mod_name = mod_name;
+
+ retval = driver_register(&driver->driver);
+ if (retval)
+ return retval;
+
+ pr_info("registered new driver %s\n", driver->name);
+ return 0;
+}
+EXPORT_SYMBOL_GPL(greybus_register_driver);
+
+void greybus_deregister_driver(struct greybus_driver *driver)
+{
+ driver_unregister(&driver->driver);
+}
+EXPORT_SYMBOL_GPL(greybus_deregister_driver);
+
+static int __init gb_init(void)
+{
+ int retval;
+
+ if (greybus_disabled())
+ return -ENODEV;
+
+ BUILD_BUG_ON(CPORT_ID_MAX >= (long)CPORT_ID_BAD);
+
+ gb_debugfs_init();
+
+ retval = bus_register(&greybus_bus_type);
+ if (retval) {
+ pr_err("bus_register failed (%d)\n", retval);
+ goto error_bus;
+ }
+
+ retval = gb_hd_init();
+ if (retval) {
+ pr_err("gb_hd_init failed (%d)\n", retval);
+ goto error_hd;
+ }
+
+ retval = gb_operation_init();
+ if (retval) {
+ pr_err("gb_operation_init failed (%d)\n", retval);
+ goto error_operation;
+ }
+
+ retval = gb_timesync_init();
+ if (retval) {
+ pr_err("gb_timesync_init failed\n");
+ goto error_timesync;
+ }
+ return 0; /* Success */
+
+error_timesync:
+ gb_operation_exit();
+error_operation:
+ gb_hd_exit();
+error_hd:
+ bus_unregister(&greybus_bus_type);
+error_bus:
+ gb_debugfs_cleanup();
+
+ return retval;
+}
+module_init(gb_init);
+
+static void __exit gb_exit(void)
+{
+ gb_timesync_exit();
+ gb_operation_exit();
+ gb_hd_exit();
+ bus_unregister(&greybus_bus_type);
+ gb_debugfs_cleanup();
+ tracepoint_synchronize_unregister();
+}
+module_exit(gb_exit);
+MODULE_LICENSE("GPL v2");
+MODULE_AUTHOR("Greg Kroah-Hartman <gregkh@linuxfoundation.org>");
--- /dev/null
+++ b/drivers/greybus/debugfs.c
@@ -0,0 +1,31 @@
+/*
+ * Greybus debugfs code
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/debugfs.h>
+
+#include "greybus.h"
+
+static struct dentry *gb_debug_root;
+
+void __init gb_debugfs_init(void)
+{
+ gb_debug_root = debugfs_create_dir("greybus", NULL);
+}
+
+void gb_debugfs_cleanup(void)
+{
+ debugfs_remove_recursive(gb_debug_root);
+ gb_debug_root = NULL;
+}
+
+struct dentry *gb_debugfs_get(void)
+{
+ return gb_debug_root;
+}
+EXPORT_SYMBOL_GPL(gb_debugfs_get);
--- /dev/null
+++ b/drivers/greybus/greybus.h
@@ -0,0 +1,154 @@
+/*
+ * Greybus driver and device API
+ *
+ * Copyright 2014-2015 Google Inc.
+ * Copyright 2014-2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __LINUX_GREYBUS_H
+#define __LINUX_GREYBUS_H
+
+#ifdef __KERNEL__
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/list.h>
+#include <linux/slab.h>
+#include <linux/device.h>
+#include <linux/module.h>
+#include <linux/pm_runtime.h>
+#include <linux/idr.h>
+
+#include "greybus_id.h"
+#include "greybus_manifest.h"
+#include "greybus_protocols.h"
+#include "manifest.h"
+#include "hd.h"
+#include "svc.h"
+#include "control.h"
+#include "module.h"
+#include "interface.h"
+#include "bundle.h"
+#include "connection.h"
+#include "operation.h"
+#include "timesync.h"
+
+/* Matches up with the Greybus Protocol specification document */
+#define GREYBUS_VERSION_MAJOR 0x00
+#define GREYBUS_VERSION_MINOR 0x01
+
+#define GREYBUS_ID_MATCH_DEVICE \
+ (GREYBUS_ID_MATCH_VENDOR | GREYBUS_ID_MATCH_PRODUCT)
+
+#define GREYBUS_DEVICE(v, p) \
+ .match_flags = GREYBUS_ID_MATCH_DEVICE, \
+ .vendor = (v), \
+ .product = (p),
+
+#define GREYBUS_DEVICE_CLASS(c) \
+ .match_flags = GREYBUS_ID_MATCH_CLASS, \
+ .class = (c),
+
+/* Maximum number of CPorts */
+#define CPORT_ID_MAX 4095 /* UniPro max id is 4095 */
+#define CPORT_ID_BAD U16_MAX
+
+struct greybus_driver {
+ const char *name;
+
+ int (*probe)(struct gb_bundle *bundle,
+ const struct greybus_bundle_id *id);
+ void (*disconnect)(struct gb_bundle *bundle);
+
+ const struct greybus_bundle_id *id_table;
+
+ struct device_driver driver;
+};
+#define to_greybus_driver(d) container_of(d, struct greybus_driver, driver)
+
+static inline void greybus_set_drvdata(struct gb_bundle *bundle, void *data)
+{
+ dev_set_drvdata(&bundle->dev, data);
+}
+
+static inline void *greybus_get_drvdata(struct gb_bundle *bundle)
+{
+ return dev_get_drvdata(&bundle->dev);
+}
+
+/* Don't call these directly, use the module_greybus_driver() macro instead */
+int greybus_register_driver(struct greybus_driver *driver,
+ struct module *module, const char *mod_name);
+void greybus_deregister_driver(struct greybus_driver *driver);
+
+/* define to get proper THIS_MODULE and KBUILD_MODNAME values */
+#define greybus_register(driver) \
+ greybus_register_driver(driver, THIS_MODULE, KBUILD_MODNAME)
+#define greybus_deregister(driver) \
+ greybus_deregister_driver(driver)
+
+/**
+ * module_greybus_driver() - Helper macro for registering a Greybus driver
+ * @__greybus_driver: greybus_driver structure
+ *
+ * Helper macro for Greybus drivers to set up proper module init / exit
+ * functions. Replaces module_init() and module_exit() and keeps people from
+ * printing pointless things to the kernel log when their driver is loaded.
+ */
+#define module_greybus_driver(__greybus_driver) \
+ module_driver(__greybus_driver, greybus_register, greybus_deregister)
+
+int greybus_disabled(void);
+
+void gb_debugfs_init(void);
+void gb_debugfs_cleanup(void);
+struct dentry *gb_debugfs_get(void);
+
+extern struct bus_type greybus_bus_type;
+
+extern struct device_type greybus_hd_type;
+extern struct device_type greybus_module_type;
+extern struct device_type greybus_interface_type;
+extern struct device_type greybus_control_type;
+extern struct device_type greybus_bundle_type;
+extern struct device_type greybus_svc_type;
+
+static inline int is_gb_host_device(const struct device *dev)
+{
+ return dev->type == &greybus_hd_type;
+}
+
+static inline int is_gb_module(const struct device *dev)
+{
+ return dev->type == &greybus_module_type;
+}
+
+static inline int is_gb_interface(const struct device *dev)
+{
+ return dev->type == &greybus_interface_type;
+}
+
+static inline int is_gb_control(const struct device *dev)
+{
+ return dev->type == &greybus_control_type;
+}
+
+static inline int is_gb_bundle(const struct device *dev)
+{
+ return dev->type == &greybus_bundle_type;
+}
+
+static inline int is_gb_svc(const struct device *dev)
+{
+ return dev->type == &greybus_svc_type;
+}
+
+static inline bool cport_id_valid(struct gb_host_device *hd, u16 cport_id)
+{
+ return cport_id != CPORT_ID_BAD && cport_id < hd->num_cports;
+}
+
+#endif /* __KERNEL__ */
+#endif /* __LINUX_GREYBUS_H */
--- /dev/null
+++ b/drivers/greybus/greybus_authentication.h
@@ -0,0 +1,120 @@
+/*
+ * Greybus Component Authentication User Header
+ *
+ * This file is provided under a dual BSD/GPLv2 license. When using or
+ * redistributing this file, you may do so under either license.
+ *
+ * GPL LICENSE SUMMARY
+ *
+ * Copyright(c) 2016 Google Inc. All rights reserved.
+ * Copyright(c) 2016 Linaro Ltd. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License version 2 for more details.
+ *
+ * BSD LICENSE
+ *
+ * Copyright(c) 2016 Google Inc. All rights reserved.
+ * Copyright(c) 2016 Linaro Ltd. All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of Google Inc. or Linaro Ltd. nor the names of
+ * its contributors may be used to endorse or promote products
+ * derived from this software without specific prior written
+ * permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL GOOGLE INC. OR
+ * LINARO LTD. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+ * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef __GREYBUS_AUTHENTICATION_USER_H
+#define __GREYBUS_AUTHENTICATION_USER_H
+
+#include <linux/ioctl.h>
+#include <linux/types.h>
+
+#define CAP_CERTIFICATE_MAX_SIZE 1600
+#define CAP_SIGNATURE_MAX_SIZE 320
+
+/* Certificate class types */
+#define CAP_CERT_IMS_EAPC 0x00000001
+#define CAP_CERT_IMS_EASC 0x00000002
+#define CAP_CERT_IMS_EARC 0x00000003
+#define CAP_CERT_IMS_IAPC 0x00000004
+#define CAP_CERT_IMS_IASC 0x00000005
+#define CAP_CERT_IMS_IARC 0x00000006
+
+/* IMS Certificate response result codes */
+#define CAP_IMS_RESULT_CERT_FOUND 0x00
+#define CAP_IMS_RESULT_CERT_CLASS_INVAL 0x01
+#define CAP_IMS_RESULT_CERT_CORRUPT 0x02
+#define CAP_IMS_RESULT_CERT_NOT_FOUND 0x03
+
+/* Authentication types */
+#define CAP_AUTH_IMS_PRI 0x00000001
+#define CAP_AUTH_IMS_SEC 0x00000002
+#define CAP_AUTH_IMS_RSA 0x00000003
+
+/* Authenticate response result codes */
+#define CAP_AUTH_RESULT_CR_SUCCESS 0x00
+#define CAP_AUTH_RESULT_CR_BAD_TYPE 0x01
+#define CAP_AUTH_RESULT_CR_WRONG_EP 0x02
+#define CAP_AUTH_RESULT_CR_NO_KEY 0x03
+#define CAP_AUTH_RESULT_CR_SIG_FAIL 0x04
+
+
+/* IOCTL support */
+struct cap_ioc_get_endpoint_uid {
+ __u8 uid[8];
+} __attribute__ ((__packed__));
+
+struct cap_ioc_get_ims_certificate {
+ __u32 certificate_class;
+ __u32 certificate_id;
+
+ __u8 result_code;
+ __u32 cert_size;
+ __u8 certificate[CAP_CERTIFICATE_MAX_SIZE];
+} __attribute__ ((__packed__));
+
+struct cap_ioc_authenticate {
+ __u32 auth_type;
+ __u8 uid[8];
+ __u8 challenge[32];
+
+ __u8 result_code;
+ __u8 response[64];
+ __u32 signature_size;
+ __u8 signature[CAP_SIGNATURE_MAX_SIZE];
+} __attribute__ ((__packed__));
+
+#define CAP_IOCTL_BASE 'C'
+#define CAP_IOC_GET_ENDPOINT_UID _IOR(CAP_IOCTL_BASE, 0, struct cap_ioc_get_endpoint_uid)
+#define CAP_IOC_GET_IMS_CERTIFICATE _IOWR(CAP_IOCTL_BASE, 1, struct cap_ioc_get_ims_certificate)
+#define CAP_IOC_AUTHENTICATE _IOWR(CAP_IOCTL_BASE, 2, struct cap_ioc_authenticate)
+
+#endif /* __GREYBUS_AUTHENTICATION_USER_H */
--- /dev/null
+++ b/drivers/greybus/greybus_id.h
@@ -0,0 +1,26 @@
+/* FIXME
+ * move this to include/linux/mod_devicetable.h when merging
+ */
+
+#ifndef __LINUX_GREYBUS_ID_H
+#define __LINUX_GREYBUS_ID_H
+
+#include <linux/types.h>
+#include <linux/mod_devicetable.h>
+
+
+struct greybus_bundle_id {
+ __u16 match_flags;
+ __u32 vendor;
+ __u32 product;
+ __u8 class;
+
+ kernel_ulong_t driver_info __aligned(sizeof(kernel_ulong_t));
+};
+
+/* Used to match the greybus_bundle_id */
+#define GREYBUS_ID_MATCH_VENDOR BIT(0)
+#define GREYBUS_ID_MATCH_PRODUCT BIT(1)
+#define GREYBUS_ID_MATCH_CLASS BIT(2)
+
+#endif /* __LINUX_GREYBUS_ID_H */
--- /dev/null
+++ b/drivers/greybus/greybus_manifest.h
@@ -0,0 +1,177 @@
+/*
+ * Greybus manifest definition
+ *
+ * See "Greybus Application Protocol" document (version 0.1) for
+ * details on these values and structures.
+ *
+ * Copyright 2014-2015 Google Inc.
+ * Copyright 2014-2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 and BSD licenses.
+ */
+
+#ifndef __GREYBUS_MANIFEST_H
+#define __GREYBUS_MANIFEST_H
+
+enum greybus_descriptor_type {
+ GREYBUS_TYPE_INVALID = 0x00,
+ GREYBUS_TYPE_INTERFACE = 0x01,
+ GREYBUS_TYPE_STRING = 0x02,
+ GREYBUS_TYPE_BUNDLE = 0x03,
+ GREYBUS_TYPE_CPORT = 0x04,
+};
+
+enum greybus_protocol {
+ GREYBUS_PROTOCOL_CONTROL = 0x00,
+ /* 0x01 is unused */
+ GREYBUS_PROTOCOL_GPIO = 0x02,
+ GREYBUS_PROTOCOL_I2C = 0x03,
+ GREYBUS_PROTOCOL_UART = 0x04,
+ GREYBUS_PROTOCOL_HID = 0x05,
+ GREYBUS_PROTOCOL_USB = 0x06,
+ GREYBUS_PROTOCOL_SDIO = 0x07,
+ GREYBUS_PROTOCOL_POWER_SUPPLY = 0x08,
+ GREYBUS_PROTOCOL_PWM = 0x09,
+ /* 0x0a is unused */
+ GREYBUS_PROTOCOL_SPI = 0x0b,
+ GREYBUS_PROTOCOL_DISPLAY = 0x0c,
+ GREYBUS_PROTOCOL_CAMERA_MGMT = 0x0d,
+ GREYBUS_PROTOCOL_SENSOR = 0x0e,
+ GREYBUS_PROTOCOL_LIGHTS = 0x0f,
+ GREYBUS_PROTOCOL_VIBRATOR = 0x10,
+ GREYBUS_PROTOCOL_LOOPBACK = 0x11,
+ GREYBUS_PROTOCOL_AUDIO_MGMT = 0x12,
+ GREYBUS_PROTOCOL_AUDIO_DATA = 0x13,
+ GREYBUS_PROTOCOL_SVC = 0x14,
+ GREYBUS_PROTOCOL_BOOTROM = 0x15,
+ GREYBUS_PROTOCOL_CAMERA_DATA = 0x16,
+ GREYBUS_PROTOCOL_FW_DOWNLOAD = 0x17,
+ GREYBUS_PROTOCOL_FW_MANAGEMENT = 0x18,
+ GREYBUS_PROTOCOL_AUTHENTICATION = 0x19,
+ GREYBUS_PROTOCOL_LOG = 0x1a,
+ /* ... */
+ GREYBUS_PROTOCOL_RAW = 0xfe,
+ GREYBUS_PROTOCOL_VENDOR = 0xff,
+};
+
+enum greybus_class_type {
+ GREYBUS_CLASS_CONTROL = 0x00,
+ /* 0x01 is unused */
+ /* 0x02 is unused */
+ /* 0x03 is unused */
+ /* 0x04 is unused */
+ GREYBUS_CLASS_HID = 0x05,
+ /* 0x06 is unused */
+ /* 0x07 is unused */
+ GREYBUS_CLASS_POWER_SUPPLY = 0x08,
+ /* 0x09 is unused */
+ GREYBUS_CLASS_BRIDGED_PHY = 0x0a,
+ /* 0x0b is unused */
+ GREYBUS_CLASS_DISPLAY = 0x0c,
+ GREYBUS_CLASS_CAMERA = 0x0d,
+ GREYBUS_CLASS_SENSOR = 0x0e,
+ GREYBUS_CLASS_LIGHTS = 0x0f,
+ GREYBUS_CLASS_VIBRATOR = 0x10,
+ GREYBUS_CLASS_LOOPBACK = 0x11,
+ GREYBUS_CLASS_AUDIO = 0x12,
+ /* 0x13 is unused */
+ /* 0x14 is unused */
+ GREYBUS_CLASS_BOOTROM = 0x15,
+ GREYBUS_CLASS_FW_MANAGEMENT = 0x16,
+ GREYBUS_CLASS_LOG = 0x17,
+ /* ... */
+ GREYBUS_CLASS_RAW = 0xfe,
+ GREYBUS_CLASS_VENDOR = 0xff,
+};
+
+enum {
+ GREYBUS_INTERFACE_FEATURE_TIMESYNC = BIT(0),
+};
+
+/*
+ * The string in a string descriptor is not NUL-terminated. The
+ * size of the descriptor will be rounded up to a multiple of 4
+ * bytes, by padding the string with 0x00 bytes if necessary.
+ */
+struct greybus_descriptor_string {
+ __u8 length;
+ __u8 id;
+ __u8 string[0];
+} __packed;
+
+/*
+ * An interface descriptor describes information about an interface as a whole,
+ * *not* the functions within it.
+ */
+struct greybus_descriptor_interface {
+ __u8 vendor_stringid;
+ __u8 product_stringid;
+ __u8 features;
+ __u8 pad;
+} __packed;
+
+/*
+ * An bundle descriptor defines an identification number and a class for
+ * each bundle.
+ *
+ * @id: Uniquely identifies a bundle within a interface, its sole purpose is to
+ * allow CPort descriptors to specify which bundle they are associated with.
+ * The first bundle will have id 0, second will have 1 and so on.
+ *
+ * The largest CPort id associated with an bundle (defined by a
+ * CPort descriptor in the manifest) is used to determine how to
+ * encode the device id and module number in UniPro packets
+ * that use the bundle.
+ *
+ * @class: It is used by kernel to know the functionality provided by the
+ * bundle and will be matched against drivers functinality while probing greybus
+ * driver. It should contain one of the values defined in
+ * 'enum greybus_class_type'.
+ *
+ */
+struct greybus_descriptor_bundle {
+ __u8 id; /* interface-relative id (0..) */
+ __u8 class;
+ __u8 pad[2];
+} __packed;
+
+/*
+ * A CPort descriptor indicates the id of the bundle within the
+ * module it's associated with, along with the CPort id used to
+ * address the CPort. The protocol id defines the format of messages
+ * exchanged using the CPort.
+ */
+struct greybus_descriptor_cport {
+ __le16 id;
+ __u8 bundle;
+ __u8 protocol_id; /* enum greybus_protocol */
+} __packed;
+
+struct greybus_descriptor_header {
+ __le16 size;
+ __u8 type; /* enum greybus_descriptor_type */
+ __u8 pad;
+} __packed;
+
+struct greybus_descriptor {
+ struct greybus_descriptor_header header;
+ union {
+ struct greybus_descriptor_string string;
+ struct greybus_descriptor_interface interface;
+ struct greybus_descriptor_bundle bundle;
+ struct greybus_descriptor_cport cport;
+ };
+} __packed;
+
+struct greybus_manifest_header {
+ __le16 size;
+ __u8 version_major;
+ __u8 version_minor;
+} __packed;
+
+struct greybus_manifest {
+ struct greybus_manifest_header header;
+ struct greybus_descriptor descriptors[0];
+} __packed;
+
+#endif /* __GREYBUS_MANIFEST_H */
--- /dev/null
+++ b/drivers/greybus/manifest.c
@@ -0,0 +1,535 @@
+/*
+ * Greybus manifest parsing
+ *
+ * Copyright 2014-2015 Google Inc.
+ * Copyright 2014-2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include "greybus.h"
+
+static const char *get_descriptor_type_string(u8 type)
+{
+ switch(type) {
+ case GREYBUS_TYPE_INVALID:
+ return "invalid";
+ case GREYBUS_TYPE_STRING:
+ return "string";
+ case GREYBUS_TYPE_INTERFACE:
+ return "interface";
+ case GREYBUS_TYPE_CPORT:
+ return "cport";
+ case GREYBUS_TYPE_BUNDLE:
+ return "bundle";
+ default:
+ WARN_ON(1);
+ return "unknown";
+ }
+}
+
+/*
+ * We scan the manifest once to identify where all the descriptors
+ * are. The result is a list of these manifest_desc structures. We
+ * then pick through them for what we're looking for (starting with
+ * the interface descriptor). As each is processed we remove it from
+ * the list. When we're done the list should (probably) be empty.
+ */
+struct manifest_desc {
+ struct list_head links;
+
+ size_t size;
+ void *data;
+ enum greybus_descriptor_type type;
+};
+
+static void release_manifest_descriptor(struct manifest_desc *descriptor)
+{
+ list_del(&descriptor->links);
+ kfree(descriptor);
+}
+
+static void release_manifest_descriptors(struct gb_interface *intf)
+{
+ struct manifest_desc *descriptor;
+ struct manifest_desc *next;
+
+ list_for_each_entry_safe(descriptor, next, &intf->manifest_descs, links)
+ release_manifest_descriptor(descriptor);
+}
+
+static void release_cport_descriptors(struct list_head *head, u8 bundle_id)
+{
+ struct manifest_desc *desc, *tmp;
+ struct greybus_descriptor_cport *desc_cport;
+
+ list_for_each_entry_safe(desc, tmp, head, links) {
+ desc_cport = desc->data;
+
+ if (desc->type != GREYBUS_TYPE_CPORT)
+ continue;
+
+ if (desc_cport->bundle == bundle_id)
+ release_manifest_descriptor(desc);
+ }
+}
+
+static struct manifest_desc *get_next_bundle_desc(struct gb_interface *intf)
+{
+ struct manifest_desc *descriptor;
+ struct manifest_desc *next;
+
+ list_for_each_entry_safe(descriptor, next, &intf->manifest_descs, links)
+ if (descriptor->type == GREYBUS_TYPE_BUNDLE)
+ return descriptor;
+
+ return NULL;
+}
+
+/*
+ * Validate the given descriptor. Its reported size must fit within
+ * the number of bytes remaining, and it must have a recognized
+ * type. Check that the reported size is at least as big as what
+ * we expect to see. (It could be bigger, perhaps for a new version
+ * of the format.)
+ *
+ * Returns the (non-zero) number of bytes consumed by the descriptor,
+ * or a negative errno.
+ */
+static int identify_descriptor(struct gb_interface *intf,
+ struct greybus_descriptor *desc, size_t size)
+{
+ struct greybus_descriptor_header *desc_header = &desc->header;
+ struct manifest_desc *descriptor;
+ size_t desc_size;
+ size_t expected_size;
+
+ if (size < sizeof(*desc_header)) {
+ dev_err(&intf->dev, "manifest too small (%zu < %zu)\n",
+ size, sizeof(*desc_header));
+ return -EINVAL; /* Must at least have header */
+ }
+
+ desc_size = le16_to_cpu(desc_header->size);
+ if (desc_size > size) {
+ dev_err(&intf->dev, "descriptor too big (%zu > %zu)\n",
+ desc_size, size);
+ return -EINVAL;
+ }
+
+ /* Descriptor needs to at least have a header */
+ expected_size = sizeof(*desc_header);
+
+ switch (desc_header->type) {
+ case GREYBUS_TYPE_STRING:
+ expected_size += sizeof(struct greybus_descriptor_string);
+ expected_size += desc->string.length;
+
+ /* String descriptors are padded to 4 byte boundaries */
+ expected_size = ALIGN(expected_size, 4);
+ break;
+ case GREYBUS_TYPE_INTERFACE:
+ expected_size += sizeof(struct greybus_descriptor_interface);
+ break;
+ case GREYBUS_TYPE_BUNDLE:
+ expected_size += sizeof(struct greybus_descriptor_bundle);
+ break;
+ case GREYBUS_TYPE_CPORT:
+ expected_size += sizeof(struct greybus_descriptor_cport);
+ break;
+ case GREYBUS_TYPE_INVALID:
+ default:
+ dev_err(&intf->dev, "invalid descriptor type (%u)\n",
+ desc_header->type);
+ return -EINVAL;
+ }
+
+ if (desc_size < expected_size) {
+ dev_err(&intf->dev, "%s descriptor too small (%zu < %zu)\n",
+ get_descriptor_type_string(desc_header->type),
+ desc_size, expected_size);
+ return -EINVAL;
+ }
+
+ /* Descriptor bigger than what we expect */
+ if (desc_size > expected_size) {
+ dev_warn(&intf->dev, "%s descriptor size mismatch (want %zu got %zu)\n",
+ get_descriptor_type_string(desc_header->type),
+ expected_size, desc_size);
+ }
+
+ descriptor = kzalloc(sizeof(*descriptor), GFP_KERNEL);
+ if (!descriptor)
+ return -ENOMEM;
+
+ descriptor->size = desc_size;
+ descriptor->data = (char *)desc + sizeof(*desc_header);
+ descriptor->type = desc_header->type;
+ list_add_tail(&descriptor->links, &intf->manifest_descs);
+
+ /* desc_size is positive and is known to fit in a signed int */
+
+ return desc_size;
+}
+
+/*
+ * Find the string descriptor having the given id, validate it, and
+ * allocate a duplicate copy of it. The duplicate has an extra byte
+ * which guarantees the returned string is NUL-terminated.
+ *
+ * String index 0 is valid (it represents "no string"), and for
+ * that a null pointer is returned.
+ *
+ * Otherwise returns a pointer to a newly-allocated copy of the
+ * descriptor string, or an error-coded pointer on failure.
+ */
+static char *gb_string_get(struct gb_interface *intf, u8 string_id)
+{
+ struct greybus_descriptor_string *desc_string;
+ struct manifest_desc *descriptor;
+ bool found = false;
+ char *string;
+
+ /* A zero string id means no string (but no error) */
+ if (!string_id)
+ return NULL;
+
+ list_for_each_entry(descriptor, &intf->manifest_descs, links) {
+ if (descriptor->type != GREYBUS_TYPE_STRING)
+ continue;
+
+ desc_string = descriptor->data;
+ if (desc_string->id == string_id) {
+ found = true;
+ break;
+ }
+ }
+ if (!found)
+ return ERR_PTR(-ENOENT);
+
+ /* Allocate an extra byte so we can guarantee it's NUL-terminated */
+ string = kmemdup(&desc_string->string, desc_string->length + 1,
+ GFP_KERNEL);
+ if (!string)
+ return ERR_PTR(-ENOMEM);
+ string[desc_string->length] = '\0';
+
+ /* Ok we've used this string, so we're done with it */
+ release_manifest_descriptor(descriptor);
+
+ return string;
+}
+
+/*
+ * Find cport descriptors in the manifest associated with the given
+ * bundle, and set up data structures for the functions that use
+ * them. Returns the number of cports set up for the bundle, or 0
+ * if there is an error.
+ */
+static u32 gb_manifest_parse_cports(struct gb_bundle *bundle)
+{
+ struct gb_interface *intf = bundle->intf;
+ struct greybus_descriptor_cport *desc_cport;
+ struct manifest_desc *desc, *next, *tmp;
+ LIST_HEAD(list);
+ u8 bundle_id = bundle->id;
+ u16 cport_id;
+ u32 count = 0;
+ int i;
+
+ /* Set up all cport descriptors associated with this bundle */
+ list_for_each_entry_safe(desc, next, &intf->manifest_descs, links) {
+ if (desc->type != GREYBUS_TYPE_CPORT)
+ continue;
+
+ desc_cport = desc->data;
+ if (desc_cport->bundle != bundle_id)
+ continue;
+
+ cport_id = le16_to_cpu(desc_cport->id);
+ if (cport_id > CPORT_ID_MAX)
+ goto exit;
+
+ /* Nothing else should have its cport_id as control cport id */
+ if (cport_id == GB_CONTROL_CPORT_ID) {
+ dev_err(&bundle->dev, "invalid cport id found (%02u)\n",
+ cport_id);
+ goto exit;
+ }
+
+ /*
+ * Found one, move it to our temporary list after checking for
+ * duplicates.
+ */
+ list_for_each_entry(tmp, &list, links) {
+ desc_cport = tmp->data;
+ if (cport_id == le16_to_cpu(desc_cport->id)) {
+ dev_err(&bundle->dev,
+ "duplicate CPort %u found\n",
+ cport_id);
+ goto exit;
+ }
+ }
+ list_move_tail(&desc->links, &list);
+ count++;
+ }
+
+ if (!count)
+ return 0;
+
+ bundle->cport_desc = kcalloc(count, sizeof(*bundle->cport_desc),
+ GFP_KERNEL);
+ if (!bundle->cport_desc)
+ goto exit;
+
+ bundle->num_cports = count;
+
+ i = 0;
+ list_for_each_entry_safe(desc, next, &list, links) {
+ desc_cport = desc->data;
+ memcpy(&bundle->cport_desc[i++], desc_cport,
+ sizeof(*desc_cport));
+
+ /* Release the cport descriptor */
+ release_manifest_descriptor(desc);
+ }
+
+ return count;
+exit:
+ release_cport_descriptors(&list, bundle_id);
+ /*
+ * Free all cports for this bundle to avoid 'excess descriptors'
+ * warnings.
+ */
+ release_cport_descriptors(&intf->manifest_descs, bundle_id);
+
+ return 0; /* Error; count should also be 0 */
+}
+
+/*
+ * Find bundle descriptors in the manifest and set up their data
+ * structures. Returns the number of bundles set up for the
+ * given interface.
+ */
+static u32 gb_manifest_parse_bundles(struct gb_interface *intf)
+{
+ struct manifest_desc *desc;
+ struct gb_bundle *bundle;
+ struct gb_bundle *bundle_next;
+ u32 count = 0;
+ u8 bundle_id;
+ u8 class;
+
+ while ((desc = get_next_bundle_desc(intf))) {
+ struct greybus_descriptor_bundle *desc_bundle;
+
+ /* Found one. Set up its bundle structure*/
+ desc_bundle = desc->data;
+ bundle_id = desc_bundle->id;
+ class = desc_bundle->class;
+
+ /* Done with this bundle descriptor */
+ release_manifest_descriptor(desc);
+
+ /* Ignore any legacy control bundles */
+ if (bundle_id == GB_CONTROL_BUNDLE_ID) {
+ dev_dbg(&intf->dev, "%s - ignoring control bundle\n",
+ __func__);
+ release_cport_descriptors(&intf->manifest_descs,
+ bundle_id);
+ continue;
+ }
+
+ /* Nothing else should have its class set to control class */
+ if (class == GREYBUS_CLASS_CONTROL) {
+ dev_err(&intf->dev,
+ "bundle %u cannot use control class\n",
+ bundle_id);
+ goto cleanup;
+ }
+
+ bundle = gb_bundle_create(intf, bundle_id, class);
+ if (!bundle)
+ goto cleanup;
+
+ /*
+ * Now go set up this bundle's functions and cports.
+ *
+ * A 'bundle' represents a device in greybus. It may require
+ * multiple cports for its functioning. If we fail to setup any
+ * cport of a bundle, we better reject the complete bundle as
+ * the device may not be able to function properly then.
+ *
+ * But, failing to setup a cport of bundle X doesn't mean that
+ * the device corresponding to bundle Y will not work properly.
+ * Bundles should be treated as separate independent devices.
+ *
+ * While parsing manifest for an interface, treat bundles as
+ * separate entities and don't reject entire interface and its
+ * bundles on failing to initialize a cport. But make sure the
+ * bundle which needs the cport, gets destroyed properly.
+ */
+ if (!gb_manifest_parse_cports(bundle)) {
+ gb_bundle_destroy(bundle);
+ continue;
+ }
+
+ count++;
+ }
+
+ return count;
+cleanup:
+ /* An error occurred; undo any changes we've made */
+ list_for_each_entry_safe(bundle, bundle_next, &intf->bundles, links) {
+ gb_bundle_destroy(bundle);
+ count--;
+ }
+ return 0; /* Error; count should also be 0 */
+}
+
+static bool gb_manifest_parse_interface(struct gb_interface *intf,
+ struct manifest_desc *interface_desc)
+{
+ struct greybus_descriptor_interface *desc_intf = interface_desc->data;
+ struct gb_control *control = intf->control;
+ char *str;
+
+ /* Handle the strings first--they can fail */
+ str = gb_string_get(intf, desc_intf->vendor_stringid);
+ if (IS_ERR(str))
+ return false;
+ control->vendor_string = str;
+
+ str = gb_string_get(intf, desc_intf->product_stringid);
+ if (IS_ERR(str))
+ goto out_free_vendor_string;
+ control->product_string = str;
+
+ /* Assign feature flags communicated via manifest */
+ intf->features = desc_intf->features;
+
+ /* Release the interface descriptor, now that we're done with it */
+ release_manifest_descriptor(interface_desc);
+
+ /* An interface must have at least one bundle descriptor */
+ if (!gb_manifest_parse_bundles(intf)) {
+ dev_err(&intf->dev, "manifest bundle descriptors not valid\n");
+ goto out_err;
+ }
+
+ return true;
+out_err:
+ kfree(control->product_string);
+ control->product_string = NULL;
+out_free_vendor_string:
+ kfree(control->vendor_string);
+ control->vendor_string = NULL;
+
+ return false;
+}
+
+/*
+ * Parse a buffer containing an interface manifest.
+ *
+ * If we find anything wrong with the content/format of the buffer
+ * we reject it.
+ *
+ * The first requirement is that the manifest's version is
+ * one we can parse.
+ *
+ * We make an initial pass through the buffer and identify all of
+ * the descriptors it contains, keeping track for each its type
+ * and the location size of its data in the buffer.
+ *
+ * Next we scan the descriptors, looking for an interface descriptor;
+ * there must be exactly one of those. When found, we record the
+ * information it contains, and then remove that descriptor (and any
+ * string descriptors it refers to) from further consideration.
+ *
+ * After that we look for the interface's bundles--there must be at
+ * least one of those.
+ *
+ * Returns true if parsing was successful, false otherwise.
+ */
+bool gb_manifest_parse(struct gb_interface *intf, void *data, size_t size)
+{
+ struct greybus_manifest *manifest;
+ struct greybus_manifest_header *header;
+ struct greybus_descriptor *desc;
+ struct manifest_desc *descriptor;
+ struct manifest_desc *interface_desc = NULL;
+ u16 manifest_size;
+ u32 found = 0;
+ bool result;
+
+ /* Manifest descriptor list should be empty here */
+ if (WARN_ON(!list_empty(&intf->manifest_descs)))
+ return false;
+
+ /* we have to have at _least_ the manifest header */
+ if (size < sizeof(*header)) {
+ dev_err(&intf->dev, "short manifest (%zu < %zu)\n",
+ size, sizeof(*header));
+ return false;
+ }
+
+ /* Make sure the size is right */
+ manifest = data;
+ header = &manifest->header;
+ manifest_size = le16_to_cpu(header->size);
+ if (manifest_size != size) {
+ dev_err(&intf->dev, "manifest size mismatch (%zu != %u)\n",
+ size, manifest_size);
+ return false;
+ }
+
+ /* Validate major/minor number */
+ if (header->version_major > GREYBUS_VERSION_MAJOR) {
+ dev_err(&intf->dev, "manifest version too new (%u.%u > %u.%u)\n",
+ header->version_major, header->version_minor,
+ GREYBUS_VERSION_MAJOR, GREYBUS_VERSION_MINOR);
+ return false;
+ }
+
+ /* OK, find all the descriptors */
+ desc = manifest->descriptors;
+ size -= sizeof(*header);
+ while (size) {
+ int desc_size;
+
+ desc_size = identify_descriptor(intf, desc, size);
+ if (desc_size < 0) {
+ result = false;
+ goto out;
+ }
+ desc = (struct greybus_descriptor *)((char *)desc + desc_size);
+ size -= desc_size;
+ }
+
+ /* There must be a single interface descriptor */
+ list_for_each_entry(descriptor, &intf->manifest_descs, links) {
+ if (descriptor->type == GREYBUS_TYPE_INTERFACE)
+ if (!found++)
+ interface_desc = descriptor;
+ }
+ if (found != 1) {
+ dev_err(&intf->dev, "manifest must have 1 interface descriptor (%u found)\n",
+ found);
+ result = false;
+ goto out;
+ }
+
+ /* Parse the manifest, starting with the interface descriptor */
+ result = gb_manifest_parse_interface(intf, interface_desc);
+
+ /*
+ * We really should have no remaining descriptors, but we
+ * don't know what newer format manifests might leave.
+ */
+ if (result && !list_empty(&intf->manifest_descs))
+ dev_info(&intf->dev, "excess descriptors in interface manifest\n");
+out:
+ release_manifest_descriptors(intf);
+
+ return result;
+}
--- /dev/null
+++ b/drivers/greybus/manifest.h
@@ -0,0 +1,16 @@
+/*
+ * Greybus manifest parsing
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __MANIFEST_H
+#define __MANIFEST_H
+
+struct gb_interface;
+bool gb_manifest_parse(struct gb_interface *intf, void *data, size_t size);
+
+#endif /* __MANIFEST_H */
--- /dev/null
+++ b/drivers/greybus/module.c
@@ -0,0 +1,238 @@
+/*
+ * Greybus Module code
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include "greybus.h"
+#include "greybus_trace.h"
+
+
+static ssize_t eject_store(struct device *dev,
+ struct device_attribute *attr,
+ const char *buf, size_t len)
+{
+ struct gb_module *module = to_gb_module(dev);
+ struct gb_interface *intf;
+ size_t i;
+ long val;
+ int ret;
+
+ ret = kstrtol(buf, 0, &val);
+ if (ret)
+ return ret;
+
+ if (!val)
+ return len;
+
+ for (i = 0; i < module->num_interfaces; ++i) {
+ intf = module->interfaces[i];
+
+ mutex_lock(&intf->mutex);
+ /* Set flag to prevent concurrent activation. */
+ intf->ejected = true;
+ gb_interface_disable(intf);
+ gb_interface_deactivate(intf);
+ mutex_unlock(&intf->mutex);
+ }
+
+ /* Tell the SVC to eject the primary interface. */
+ ret = gb_svc_intf_eject(module->hd->svc, module->module_id);
+ if (ret)
+ return ret;
+
+ return len;
+}
+static DEVICE_ATTR_WO(eject);
+
+static ssize_t module_id_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_module *module = to_gb_module(dev);
+
+ return sprintf(buf, "%u\n", module->module_id);
+}
+static DEVICE_ATTR_RO(module_id);
+
+static ssize_t num_interfaces_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gb_module *module = to_gb_module(dev);
+
+ return sprintf(buf, "%zu\n", module->num_interfaces);
+}
+static DEVICE_ATTR_RO(num_interfaces);
+
+static struct attribute *module_attrs[] = {
+ &dev_attr_eject.attr,
+ &dev_attr_module_id.attr,
+ &dev_attr_num_interfaces.attr,
+ NULL,
+};
+ATTRIBUTE_GROUPS(module);
+
+static void gb_module_release(struct device *dev)
+{
+ struct gb_module *module = to_gb_module(dev);
+
+ trace_gb_module_release(module);
+
+ kfree(module);
+}
+
+struct device_type greybus_module_type = {
+ .name = "greybus_module",
+ .release = gb_module_release,
+};
+
+struct gb_module *gb_module_create(struct gb_host_device *hd, u8 module_id,
+ size_t num_interfaces)
+{
+ struct gb_interface *intf;
+ struct gb_module *module;
+ int i;
+
+ module = kzalloc(sizeof(*module) + num_interfaces * sizeof(intf),
+ GFP_KERNEL);
+ if (!module)
+ return NULL;
+
+ module->hd = hd;
+ module->module_id = module_id;
+ module->num_interfaces = num_interfaces;
+
+ module->dev.parent = &hd->dev;
+ module->dev.bus = &greybus_bus_type;
+ module->dev.type = &greybus_module_type;
+ module->dev.groups = module_groups;
+ module->dev.dma_mask = hd->dev.dma_mask;
+ device_initialize(&module->dev);
+ dev_set_name(&module->dev, "%d-%u", hd->bus_id, module_id);
+
+ trace_gb_module_create(module);
+
+ for (i = 0; i < num_interfaces; ++i) {
+ intf = gb_interface_create(module, module_id + i);
+ if (!intf) {
+ dev_err(&module->dev, "failed to create interface %u\n",
+ module_id + i);
+ goto err_put_interfaces;
+ }
+ module->interfaces[i] = intf;
+ }
+
+ return module;
+
+err_put_interfaces:
+ for (--i; i > 0; --i)
+ gb_interface_put(module->interfaces[i]);
+
+ put_device(&module->dev);
+
+ return NULL;
+}
+
+/*
+ * Register and enable an interface after first attempting to activate it.
+ */
+static void gb_module_register_interface(struct gb_interface *intf)
+{
+ struct gb_module *module = intf->module;
+ u8 intf_id = intf->interface_id;
+ int ret;
+
+ mutex_lock(&intf->mutex);
+
+ ret = gb_interface_activate(intf);
+ if (ret) {
+ if (intf->type != GB_INTERFACE_TYPE_DUMMY) {
+ dev_err(&module->dev,
+ "failed to activate interface %u: %d\n",
+ intf_id, ret);
+ }
+
+ gb_interface_add(intf);
+ goto err_unlock;
+ }
+
+ ret = gb_interface_add(intf);
+ if (ret)
+ goto err_interface_deactivate;
+
+ ret = gb_interface_enable(intf);
+ if (ret) {
+ dev_err(&module->dev, "failed to enable interface %u: %d\n",
+ intf_id, ret);
+ goto err_interface_deactivate;
+ }
+
+ mutex_unlock(&intf->mutex);
+
+ return;
+
+err_interface_deactivate:
+ gb_interface_deactivate(intf);
+err_unlock:
+ mutex_unlock(&intf->mutex);
+}
+
+static void gb_module_deregister_interface(struct gb_interface *intf)
+{
+ /* Mark as disconnected to prevent I/O during disable. */
+ if (intf->module->disconnected)
+ intf->disconnected = true;
+
+ mutex_lock(&intf->mutex);
+ intf->removed = true;
+ gb_interface_disable(intf);
+ gb_interface_deactivate(intf);
+ mutex_unlock(&intf->mutex);
+
+ gb_interface_del(intf);
+}
+
+/* Register a module and its interfaces. */
+int gb_module_add(struct gb_module *module)
+{
+ size_t i;
+ int ret;
+
+ ret = device_add(&module->dev);
+ if (ret) {
+ dev_err(&module->dev, "failed to register module: %d\n", ret);
+ return ret;
+ }
+
+ trace_gb_module_add(module);
+
+ for (i = 0; i < module->num_interfaces; ++i)
+ gb_module_register_interface(module->interfaces[i]);
+
+ return 0;
+}
+
+/* Deregister a module and its interfaces. */
+void gb_module_del(struct gb_module *module)
+{
+ size_t i;
+
+ for (i = 0; i < module->num_interfaces; ++i)
+ gb_module_deregister_interface(module->interfaces[i]);
+
+ trace_gb_module_del(module);
+
+ device_del(&module->dev);
+}
+
+void gb_module_put(struct gb_module *module)
+{
+ size_t i;
+
+ for (i = 0; i < module->num_interfaces; ++i)
+ gb_interface_put(module->interfaces[i]);
+
+ put_device(&module->dev);
+}
--- /dev/null
+++ b/drivers/greybus/module.h
@@ -0,0 +1,34 @@
+/*
+ * Greybus Module code
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __MODULE_H
+#define __MODULE_H
+
+struct gb_module {
+ struct device dev;
+ struct gb_host_device *hd;
+
+ struct list_head hd_node;
+
+ u8 module_id;
+ size_t num_interfaces;
+
+ bool disconnected;
+
+ struct gb_interface *interfaces[0];
+};
+#define to_gb_module(d) container_of(d, struct gb_module, dev)
+
+struct gb_module *gb_module_create(struct gb_host_device *hd, u8 module_id,
+ size_t num_interfaces);
+int gb_module_add(struct gb_module *module);
+void gb_module_del(struct gb_module *module);
+void gb_module_put(struct gb_module *module);
+
+#endif /* __MODULE_H */
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 08/32] greybus: bootrom driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (5 preceding siblings ...)
2016-09-16 13:23 ` [patch 07/32] greybus: core code Greg KH
@ 2016-09-16 13:24 ` Greg KH
2016-09-16 13:24 ` [patch 09/32] greybus: firmware download class driver Greg KH
` (23 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 13:24 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus Bootrom protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/bootrom.c | 524 ++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 524 insertions(+)
--- /dev/null
+++ b/drivers/greybus/bootrom.c
@@ -0,0 +1,524 @@
+/*
+ * BOOTROM Greybus driver.
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/firmware.h>
+#include <linux/jiffies.h>
+#include <linux/mutex.h>
+#include <linux/workqueue.h>
+
+#include "greybus.h"
+#include "firmware.h"
+
+/* Timeout, in jiffies, within which the next request must be received */
+#define NEXT_REQ_TIMEOUT_MS 1000
+
+/*
+ * FIXME: Reduce this timeout once svc core handles parallel processing of
+ * events from the SVC, which are handled sequentially today.
+ */
+#define MODE_SWITCH_TIMEOUT_MS 10000
+
+enum next_request_type {
+ NEXT_REQ_FIRMWARE_SIZE,
+ NEXT_REQ_GET_FIRMWARE,
+ NEXT_REQ_READY_TO_BOOT,
+ NEXT_REQ_MODE_SWITCH,
+};
+
+struct gb_bootrom {
+ struct gb_connection *connection;
+ const struct firmware *fw;
+ u8 protocol_major;
+ u8 protocol_minor;
+ enum next_request_type next_request;
+ struct delayed_work dwork;
+ struct mutex mutex; /* Protects bootrom->fw */
+};
+
+static void free_firmware(struct gb_bootrom *bootrom)
+{
+ if (!bootrom->fw)
+ return;
+
+ release_firmware(bootrom->fw);
+ bootrom->fw = NULL;
+}
+
+static void gb_bootrom_timedout(struct work_struct *work)
+{
+ struct delayed_work *dwork = to_delayed_work(work);
+ struct gb_bootrom *bootrom = container_of(dwork, struct gb_bootrom, dwork);
+ struct device *dev = &bootrom->connection->bundle->dev;
+ const char *reason;
+
+ switch (bootrom->next_request) {
+ case NEXT_REQ_FIRMWARE_SIZE:
+ reason = "Firmware Size Request";
+ break;
+ case NEXT_REQ_GET_FIRMWARE:
+ reason = "Get Firmware Request";
+ break;
+ case NEXT_REQ_READY_TO_BOOT:
+ reason = "Ready to Boot Request";
+ break;
+ case NEXT_REQ_MODE_SWITCH:
+ reason = "Interface Mode Switch";
+ break;
+ default:
+ reason = NULL;
+ dev_err(dev, "Invalid next-request: %u", bootrom->next_request);
+ break;
+ }
+
+ dev_err(dev, "Timed out waiting for %s from the Module\n", reason);
+
+ mutex_lock(&bootrom->mutex);
+ free_firmware(bootrom);
+ mutex_unlock(&bootrom->mutex);
+
+ /* TODO: Power-off Module ? */
+}
+
+static void gb_bootrom_set_timeout(struct gb_bootrom *bootrom,
+ enum next_request_type next, unsigned long timeout)
+{
+ bootrom->next_request = next;
+ schedule_delayed_work(&bootrom->dwork, msecs_to_jiffies(timeout));
+}
+
+static void gb_bootrom_cancel_timeout(struct gb_bootrom *bootrom)
+{
+ cancel_delayed_work_sync(&bootrom->dwork);
+}
+
+/*
+ * The es2 chip doesn't have VID/PID programmed into the hardware and we need to
+ * hack that up to distinguish different modules and their firmware blobs.
+ *
+ * This fetches VID/PID (over bootrom protocol) for es2 chip only, when VID/PID
+ * already sent during hotplug are 0.
+ *
+ * Otherwise, we keep intf->vendor_id/product_id same as what's passed
+ * during hotplug.
+ */
+static void bootrom_es2_fixup_vid_pid(struct gb_bootrom *bootrom)
+{
+ struct gb_bootrom_get_vid_pid_response response;
+ struct gb_connection *connection = bootrom->connection;
+ struct gb_interface *intf = connection->bundle->intf;
+ int ret;
+
+ if (!(intf->quirks & GB_INTERFACE_QUIRK_NO_GMP_IDS))
+ return;
+
+ ret = gb_operation_sync(connection, GB_BOOTROM_TYPE_GET_VID_PID,
+ NULL, 0, &response, sizeof(response));
+ if (ret) {
+ dev_err(&connection->bundle->dev,
+ "Bootrom get vid/pid operation failed (%d)\n", ret);
+ return;
+ }
+
+ /*
+ * NOTE: This is hacked, so that the same values of VID/PID can be used
+ * by next firmware level as well. The uevent for bootrom will still
+ * have VID/PID as 0, though after this point the sysfs files will start
+ * showing the updated values. But yeah, that's a bit racy as the same
+ * sysfs files would be showing 0 before this point.
+ */
+ intf->vendor_id = le32_to_cpu(response.vendor_id);
+ intf->product_id = le32_to_cpu(response.product_id);
+
+ dev_dbg(&connection->bundle->dev, "Bootrom got vid (0x%x)/pid (0x%x)\n",
+ intf->vendor_id, intf->product_id);
+}
+
+/* This returns path of the firmware blob on the disk */
+static int find_firmware(struct gb_bootrom *bootrom, u8 stage)
+{
+ struct gb_connection *connection = bootrom->connection;
+ struct gb_interface *intf = connection->bundle->intf;
+ char firmware_name[49];
+ int rc;
+
+ /* Already have a firmware, free it */
+ free_firmware(bootrom);
+
+ /* Bootrom protocol is only supported for loading Stage 2 firmware */
+ if (stage != 2) {
+ dev_err(&connection->bundle->dev, "Invalid boot stage: %u\n",
+ stage);
+ return -EINVAL;
+ }
+
+ /*
+ * Create firmware name
+ *
+ * XXX Name it properly..
+ */
+ snprintf(firmware_name, sizeof(firmware_name),
+ FW_NAME_PREFIX "%08x_%08x_%08x_%08x_s2l.tftf",
+ intf->ddbl1_manufacturer_id, intf->ddbl1_product_id,
+ intf->vendor_id, intf->product_id);
+
+ // FIXME:
+ // Turn to dev_dbg later after everyone has valid bootloaders with good
+ // ids, but leave this as dev_info for now to make it easier to track
+ // down "empty" vid/pid modules.
+ dev_info(&connection->bundle->dev, "Firmware file '%s' requested\n",
+ firmware_name);
+
+ rc = request_firmware(&bootrom->fw, firmware_name,
+ &connection->bundle->dev);
+ if (rc) {
+ dev_err(&connection->bundle->dev,
+ "failed to find %s firmware (%d)\n", firmware_name, rc);
+ }
+
+ return rc;
+}
+
+static int gb_bootrom_firmware_size_request(struct gb_operation *op)
+{
+ struct gb_bootrom *bootrom = gb_connection_get_data(op->connection);
+ struct gb_bootrom_firmware_size_request *size_request = op->request->payload;
+ struct gb_bootrom_firmware_size_response *size_response;
+ struct device *dev = &op->connection->bundle->dev;
+ int ret;
+
+ /* Disable timeouts */
+ gb_bootrom_cancel_timeout(bootrom);
+
+ if (op->request->payload_size != sizeof(*size_request)) {
+ dev_err(dev, "%s: illegal size of firmware size request (%zu != %zu)\n",
+ __func__, op->request->payload_size,
+ sizeof(*size_request));
+ ret = -EINVAL;
+ goto queue_work;
+ }
+
+ mutex_lock(&bootrom->mutex);
+
+ ret = find_firmware(bootrom, size_request->stage);
+ if (ret)
+ goto unlock;
+
+ if (!gb_operation_response_alloc(op, sizeof(*size_response),
+ GFP_KERNEL)) {
+ dev_err(dev, "%s: error allocating response\n", __func__);
+ free_firmware(bootrom);
+ ret = -ENOMEM;
+ goto unlock;
+ }
+
+ size_response = op->response->payload;
+ size_response->size = cpu_to_le32(bootrom->fw->size);
+
+ dev_dbg(dev, "%s: firmware size %d bytes\n", __func__, size_response->size);
+
+unlock:
+ mutex_unlock(&bootrom->mutex);
+
+queue_work:
+ if (!ret) {
+ /* Refresh timeout */
+ gb_bootrom_set_timeout(bootrom, NEXT_REQ_GET_FIRMWARE,
+ NEXT_REQ_TIMEOUT_MS);
+ }
+
+ return ret;
+}
+
+static int gb_bootrom_get_firmware(struct gb_operation *op)
+{
+ struct gb_bootrom *bootrom = gb_connection_get_data(op->connection);
+ const struct firmware *fw;
+ struct gb_bootrom_get_firmware_request *firmware_request;
+ struct gb_bootrom_get_firmware_response *firmware_response;
+ struct device *dev = &op->connection->bundle->dev;
+ unsigned int offset, size;
+ enum next_request_type next_request;
+ int ret = 0;
+
+ /* Disable timeouts */
+ gb_bootrom_cancel_timeout(bootrom);
+
+ if (op->request->payload_size != sizeof(*firmware_request)) {
+ dev_err(dev, "%s: Illegal size of get firmware request (%zu %zu)\n",
+ __func__, op->request->payload_size,
+ sizeof(*firmware_request));
+ ret = -EINVAL;
+ goto queue_work;
+ }
+
+ mutex_lock(&bootrom->mutex);
+
+ fw = bootrom->fw;
+ if (!fw) {
+ dev_err(dev, "%s: firmware not available\n", __func__);
+ ret = -EINVAL;
+ goto unlock;
+ }
+
+ firmware_request = op->request->payload;
+ offset = le32_to_cpu(firmware_request->offset);
+ size = le32_to_cpu(firmware_request->size);
+
+ if (offset >= fw->size || size > fw->size - offset) {
+ dev_warn(dev, "bad firmware request (offs = %u, size = %u)\n",
+ offset, size);
+ ret = -EINVAL;
+ goto unlock;
+ }
+
+ if (!gb_operation_response_alloc(op, sizeof(*firmware_response) + size,
+ GFP_KERNEL)) {
+ dev_err(dev, "%s: error allocating response\n", __func__);
+ ret = -ENOMEM;
+ goto unlock;
+ }
+
+ firmware_response = op->response->payload;
+ memcpy(firmware_response->data, fw->data + offset, size);
+
+ dev_dbg(dev, "responding with firmware (offs = %u, size = %u)\n", offset,
+ size);
+
+unlock:
+ mutex_unlock(&bootrom->mutex);
+
+queue_work:
+ /* Refresh timeout */
+ if (!ret && (offset + size == fw->size))
+ next_request = NEXT_REQ_READY_TO_BOOT;
+ else
+ next_request = NEXT_REQ_GET_FIRMWARE;
+
+ gb_bootrom_set_timeout(bootrom, next_request, NEXT_REQ_TIMEOUT_MS);
+
+ return ret;
+}
+
+static int gb_bootrom_ready_to_boot(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct gb_bootrom *bootrom = gb_connection_get_data(connection);
+ struct gb_bootrom_ready_to_boot_request *rtb_request;
+ struct device *dev = &connection->bundle->dev;
+ u8 status;
+ int ret = 0;
+
+ /* Disable timeouts */
+ gb_bootrom_cancel_timeout(bootrom);
+
+ if (op->request->payload_size != sizeof(*rtb_request)) {
+ dev_err(dev, "%s: Illegal size of ready to boot request (%zu %zu)\n",
+ __func__, op->request->payload_size,
+ sizeof(*rtb_request));
+ ret = -EINVAL;
+ goto queue_work;
+ }
+
+ rtb_request = op->request->payload;
+ status = rtb_request->status;
+
+ /* Return error if the blob was invalid */
+ if (status == GB_BOOTROM_BOOT_STATUS_INVALID) {
+ ret = -EINVAL;
+ goto queue_work;
+ }
+
+ /*
+ * XXX Should we return error for insecure firmware?
+ */
+ dev_dbg(dev, "ready to boot: 0x%x, 0\n", status);
+
+queue_work:
+ /*
+ * Refresh timeout, the Interface shall load the new personality and
+ * send a new hotplug request, which shall get rid of the bootrom
+ * connection. As that can take some time, increase the timeout a bit.
+ */
+ gb_bootrom_set_timeout(bootrom, NEXT_REQ_MODE_SWITCH,
+ MODE_SWITCH_TIMEOUT_MS);
+
+ return ret;
+}
+
+static int gb_bootrom_request_handler(struct gb_operation *op)
+{
+ u8 type = op->type;
+
+ switch (type) {
+ case GB_BOOTROM_TYPE_FIRMWARE_SIZE:
+ return gb_bootrom_firmware_size_request(op);
+ case GB_BOOTROM_TYPE_GET_FIRMWARE:
+ return gb_bootrom_get_firmware(op);
+ case GB_BOOTROM_TYPE_READY_TO_BOOT:
+ return gb_bootrom_ready_to_boot(op);
+ default:
+ dev_err(&op->connection->bundle->dev,
+ "unsupported request: %u\n", type);
+ return -EINVAL;
+ }
+}
+
+static int gb_bootrom_get_version(struct gb_bootrom *bootrom)
+{
+ struct gb_bundle *bundle = bootrom->connection->bundle;
+ struct gb_bootrom_version_request request;
+ struct gb_bootrom_version_response response;
+ int ret;
+
+ request.major = GB_BOOTROM_VERSION_MAJOR;
+ request.minor = GB_BOOTROM_VERSION_MINOR;
+
+ ret = gb_operation_sync(bootrom->connection,
+ GB_BOOTROM_TYPE_VERSION,
+ &request, sizeof(request), &response,
+ sizeof(response));
+ if (ret) {
+ dev_err(&bundle->dev,
+ "failed to get protocol version: %d\n",
+ ret);
+ return ret;
+ }
+
+ if (response.major > request.major) {
+ dev_err(&bundle->dev,
+ "unsupported major protocol version (%u > %u)\n",
+ response.major, request.major);
+ return -ENOTSUPP;
+ }
+
+ bootrom->protocol_major = response.major;
+ bootrom->protocol_minor = response.minor;
+
+ dev_dbg(&bundle->dev, "%s - %u.%u\n", __func__, response.major,
+ response.minor);
+
+ return 0;
+}
+
+static int gb_bootrom_probe(struct gb_bundle *bundle,
+ const struct greybus_bundle_id *id)
+{
+ struct greybus_descriptor_cport *cport_desc;
+ struct gb_connection *connection;
+ struct gb_bootrom *bootrom;
+ int ret;
+
+ if (bundle->num_cports != 1)
+ return -ENODEV;
+
+ cport_desc = &bundle->cport_desc[0];
+ if (cport_desc->protocol_id != GREYBUS_PROTOCOL_BOOTROM)
+ return -ENODEV;
+
+ bootrom = kzalloc(sizeof(*bootrom), GFP_KERNEL);
+ if (!bootrom)
+ return -ENOMEM;
+
+ connection = gb_connection_create(bundle,
+ le16_to_cpu(cport_desc->id),
+ gb_bootrom_request_handler);
+ if (IS_ERR(connection)) {
+ ret = PTR_ERR(connection);
+ goto err_free_bootrom;
+ }
+
+ gb_connection_set_data(connection, bootrom);
+
+ bootrom->connection = connection;
+
+ mutex_init(&bootrom->mutex);
+ INIT_DELAYED_WORK(&bootrom->dwork, gb_bootrom_timedout);
+ greybus_set_drvdata(bundle, bootrom);
+
+ ret = gb_connection_enable_tx(connection);
+ if (ret)
+ goto err_connection_destroy;
+
+ ret = gb_bootrom_get_version(bootrom);
+ if (ret)
+ goto err_connection_disable;
+
+ bootrom_es2_fixup_vid_pid(bootrom);
+
+ ret = gb_connection_enable(connection);
+ if (ret)
+ goto err_connection_disable;
+
+ /* Refresh timeout */
+ gb_bootrom_set_timeout(bootrom, NEXT_REQ_FIRMWARE_SIZE,
+ NEXT_REQ_TIMEOUT_MS);
+
+ /* Tell bootrom we're ready. */
+ ret = gb_operation_sync(connection, GB_BOOTROM_TYPE_AP_READY, NULL, 0,
+ NULL, 0);
+ if (ret) {
+ dev_err(&connection->bundle->dev,
+ "failed to send AP READY: %d\n", ret);
+ goto err_cancel_timeout;
+ }
+
+ dev_dbg(&bundle->dev, "AP_READY sent\n");
+
+ return 0;
+
+err_cancel_timeout:
+ gb_bootrom_cancel_timeout(bootrom);
+err_connection_disable:
+ gb_connection_disable(connection);
+err_connection_destroy:
+ gb_connection_destroy(connection);
+err_free_bootrom:
+ kfree(bootrom);
+
+ return ret;
+}
+
+static void gb_bootrom_disconnect(struct gb_bundle *bundle)
+{
+ struct gb_bootrom *bootrom = greybus_get_drvdata(bundle);
+
+ dev_dbg(&bundle->dev, "%s\n", __func__);
+
+ gb_connection_disable(bootrom->connection);
+
+ /* Disable timeouts */
+ gb_bootrom_cancel_timeout(bootrom);
+
+ /*
+ * Release firmware:
+ *
+ * As the connection and the delayed work are already disabled, we don't
+ * need to lock access to bootrom->fw here.
+ */
+ free_firmware(bootrom);
+
+ gb_connection_destroy(bootrom->connection);
+ kfree(bootrom);
+}
+
+static const struct greybus_bundle_id gb_bootrom_id_table[] = {
+ { GREYBUS_DEVICE_CLASS(GREYBUS_CLASS_BOOTROM) },
+ { }
+};
+
+static struct greybus_driver gb_bootrom_driver = {
+ .name = "bootrom",
+ .probe = gb_bootrom_probe,
+ .disconnect = gb_bootrom_disconnect,
+ .id_table = gb_bootrom_id_table,
+};
+
+module_greybus_driver(gb_bootrom_driver);
+
+MODULE_LICENSE("GPL v2");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 09/32] greybus: firmware download class driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (6 preceding siblings ...)
2016-09-16 13:24 ` [patch 08/32] greybus: bootrom driver Greg KH
@ 2016-09-16 13:24 ` Greg KH
2016-09-16 13:24 ` [patch 10/32] greybus: audio driver Greg KH
` (22 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 13:24 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus firmware download protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/Documentation/firmware/authenticate.c | 139 ++
drivers/greybus/Documentation/firmware/firmware-management | 333 ++++++
drivers/greybus/Documentation/firmware/firmware.c | 262 ++++
drivers/greybus/firmware.h | 42
drivers/greybus/fw-core.c | 312 +++++
drivers/greybus/fw-download.c | 465 ++++++++
drivers/greybus/fw-management.c | 721 +++++++++++++
drivers/greybus/greybus_firmware.h | 120 ++
8 files changed, 2394 insertions(+)
--- /dev/null
+++ b/drivers/greybus/Documentation/firmware/authenticate.c
@@ -0,0 +1,139 @@
+/*
+ * Sample code to test CAP protocol
+ *
+ * This file is provided under a dual BSD/GPLv2 license. When using or
+ * redistributing this file, you may do so under either license.
+ *
+ * GPL LICENSE SUMMARY
+ *
+ * Copyright(c) 2016 Google Inc. All rights reserved.
+ * Copyright(c) 2016 Linaro Ltd. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License version 2 for more details.
+ *
+ * BSD LICENSE
+ *
+ * Copyright(c) 2016 Google Inc. All rights reserved.
+ * Copyright(c) 2016 Linaro Ltd. All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of Google Inc. or Linaro Ltd. nor the names of
+ * its contributors may be used to endorse or promote products
+ * derived from this software without specific prior written
+ * permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL GOOGLE INC. OR
+ * LINARO LTD. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+ * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <stdio.h>
+#include <string.h>
+#include <unistd.h>
+#include <sys/ioctl.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+
+#include "../../greybus_authentication.h"
+
+struct cap_ioc_get_endpoint_uid uid;
+struct cap_ioc_get_ims_certificate cert = {
+ .certificate_class = 0,
+ .certificate_id = 0,
+};
+
+struct cap_ioc_authenticate authenticate = {
+ .auth_type = 0,
+ .challenge = {0},
+};
+
+int main(int argc, char *argv[])
+{
+ unsigned int timeout = 10000;
+ char *capdev;
+ int fd, ret;
+
+ /* Make sure arguments are correct */
+ if (argc != 2) {
+ printf("\nUsage: ./firmware <Path of the gb-cap-X dev>\n");
+ return 0;
+ }
+
+ capdev = argv[1];
+
+ printf("Opening %s authentication device\n", capdev);
+
+ fd = open(capdev, O_RDWR);
+ if (fd < 0) {
+ printf("Failed to open: %s\n", capdev);
+ return -1;
+ }
+
+ /* Get UID */
+ printf("Get UID\n");
+
+ ret = ioctl(fd, CAP_IOC_GET_ENDPOINT_UID, &uid);
+ if (ret < 0) {
+ printf("Failed to get UID: %s (%d)\n", capdev, ret);
+ ret = -1;
+ goto close_fd;
+ }
+
+ printf("UID received: 0x%llx\n", *(long long unsigned int *)(uid.uid));
+
+ /* Get certificate */
+ printf("Get IMS certificate\n");
+
+ ret = ioctl(fd, CAP_IOC_GET_IMS_CERTIFICATE, &cert);
+ if (ret < 0) {
+ printf("Failed to get IMS certificate: %s (%d)\n", capdev, ret);
+ ret = -1;
+ goto close_fd;
+ }
+
+ printf("IMS Certificate size: %d\n", cert.cert_size);
+
+ /* Authenticate */
+ printf("Authenticate module\n");
+
+ memcpy(authenticate.uid, uid.uid, 8);
+
+ ret = ioctl(fd, CAP_IOC_AUTHENTICATE, &authenticate);
+ if (ret < 0) {
+ printf("Failed to authenticate module: %s (%d)\n", capdev, ret);
+ ret = -1;
+ goto close_fd;
+ }
+
+ printf("Authenticated, result (%02x), sig-size (%02x)\n",
+ authenticate.result_code, authenticate.signature_size);
+
+close_fd:
+ close(fd);
+
+ return ret;
+}
--- /dev/null
+++ b/drivers/greybus/Documentation/firmware/firmware-management
@@ -0,0 +1,333 @@
+
+Firmware Management
+-------------------
+ Copyright 2016 Google Inc.
+ Copyright 2016 Linaro Ltd.
+
+Interface-Manifest
+------------------
+
+All firmware packages on the Modules or Interfaces are managed by a special
+Firmware Management Protocol. To support Firmware Management by the AP, the
+Interface Manifest shall at least contain the Firmware Management Bundle and a
+Firmware Management Protocol CPort within it.
+
+The bundle may contain additional CPorts based on the extra functionality
+required to manage firmware packages.
+
+For example, this is how the Firmware Management part of the Interface Manifest
+may look like:
+
+ ; Firmware Management Bundle (Bundle 1):
+ [bundle-descriptor 1]
+ class = 0x16
+
+ ; (Mandatory) Firmware Management Protocol on CPort 1
+ [cport-descriptor 2]
+ bundle = 1
+ protocol = 0x18
+
+ ; (Optional) Firmware Download Protocol on CPort 2
+ [cport-descriptor 1]
+ bundle = 1
+ protocol = 0x17
+
+ ; (Optional) SPI protocol on CPort 3
+ [cport-descriptor 3]
+ bundle = 1
+ protocol = 0x0b
+
+ ; (Optional) Component Authentication Protocol (CAP) on CPort 4
+ [cport-descriptor 4]
+ bundle = 1
+ protocol = 0x19
+
+
+Sysfs Interfaces - Firmware Management
+--------------------------------------
+
+The Firmware Management Protocol interacts with Userspace using the character
+device interface. The character device will be present in /dev/ directory
+and will be named gb-fw-mgmt-<N>. The number <N> is assigned at runtime.
+
+Identifying the Character Device
+================================
+
+There can be multiple devices present in /dev/ directory with name gb-fw-mgmt-N
+and user first needs to identify the character device used for
+firmware-management for a particular interface.
+
+The Firmware Management core creates a device of class 'gb_fw_mgmt', which shall
+be used by the user to identify the right character device for it. The class
+device is created within the Bundle directory for a particular Interface.
+
+For example this is how the class-device can be present:
+
+/sys/bus/greybus/devices/1-1/1-1.1/1-1.1.1/gb_fw_mgmt/gb-fw-mgmt-0
+
+The last name in this path: gb-fw-mgmt-0 is precisely the name of the char
+device and so the device in this case will be:
+
+/dev/gb-fw-mgmt-0.
+
+Operations on the Char device
+=============================
+
+The Character device (gb-fw-mgmt-0 in example) can be opened by the userspace
+application and it can perform various 'ioctl' operations on the device. The
+device doesn't support any read/write operations.
+
+Following are the IOCTLs and their data structures available to the user:
+
+/* IOCTL support */
+#define GB_FW_LOAD_METHOD_UNIPRO 0x01
+#define GB_FW_LOAD_METHOD_INTERNAL 0x02
+
+#define GB_FW_LOAD_STATUS_FAILED 0x00
+#define GB_FW_LOAD_STATUS_UNVALIDATED 0x01
+#define GB_FW_LOAD_STATUS_VALIDATED 0x02
+#define GB_FW_LOAD_STATUS_VALIDATION_FAILED 0x03
+
+#define GB_FW_BACKEND_FW_STATUS_SUCCESS 0x01
+#define GB_FW_BACKEND_FW_STATUS_FAIL_FIND 0x02
+#define GB_FW_BACKEND_FW_STATUS_FAIL_FETCH 0x03
+#define GB_FW_BACKEND_FW_STATUS_FAIL_WRITE 0x04
+#define GB_FW_BACKEND_FW_STATUS_INT 0x05
+#define GB_FW_BACKEND_FW_STATUS_RETRY 0x06
+#define GB_FW_BACKEND_FW_STATUS_NOT_SUPPORTED 0x07
+
+#define GB_FW_BACKEND_VERSION_STATUS_SUCCESS 0x01
+#define GB_FW_BACKEND_VERSION_STATUS_NOT_AVAILABLE 0x02
+#define GB_FW_BACKEND_VERSION_STATUS_NOT_SUPPORTED 0x03
+#define GB_FW_BACKEND_VERSION_STATUS_RETRY 0x04
+#define GB_FW_BACKEND_VERSION_STATUS_FAIL_INT 0x05
+
+
+struct fw_mgmt_ioc_get_intf_version {
+ __u8 firmware_tag[GB_FIRMWARE_U_TAG_MAX_SIZE];
+ __u16 major;
+ __u16 minor;
+} __attribute__ ((__packed__));
+
+struct fw_mgmt_ioc_get_backend_version {
+ __u8 firmware_tag[GB_FIRMWARE_U_TAG_MAX_SIZE];
+ __u16 major;
+ __u16 minor;
+ __u8 status;
+} __attribute__ ((__packed__));
+
+struct fw_mgmt_ioc_intf_load_and_validate {
+ __u8 firmware_tag[GB_FIRMWARE_TAG_MAX_SIZE];
+ __u8 load_method;
+ __u8 status;
+ __u16 major;
+ __u16 minor;
+} __packed;
+
+struct fw_mgmt_ioc_backend_fw_update {
+ __u8 firmware_tag[GB_FIRMWARE_TAG_MAX_SIZE];
+ __u8 status;
+} __packed;
+
+#define FW_MGMT_IOCTL_BASE 'S'
+#define FW_MGMT_IOC_GET_INTF_FW _IOR(FW_MGMT_IOCTL_BASE, 0, struct fw_mgmt_ioc_get_intf_version)
+#define FW_MGMT_IOC_GET_BACKEND_FW _IOWR(FW_MGMT_IOCTL_BASE, 1, struct fw_mgmt_ioc_get_backend_version)
+#define FW_MGMT_IOC_INTF_LOAD_AND_VALIDATE _IOWR(FW_MGMT_IOCTL_BASE, 2, struct fw_mgmt_ioc_intf_load_and_validate)
+#define FW_MGMT_IOC_INTF_BACKEND_FW_UPDATE _IOWR(FW_MGMT_IOCTL_BASE, 3, struct fw_mgmt_ioc_backend_fw_update)
+#define FW_MGMT_IOC_SET_TIMEOUT_MS _IOW(FW_MGMT_IOCTL_BASE, 4, unsigned int)
+#define FW_MGMT_IOC_MODE_SWITCH _IO(FW_MGMT_IOCTL_BASE, 5)
+
+1. FW_MGMT_IOC_GET_INTF_FW:
+
+ This ioctl shall be used by the user to get the version and firmware-tag of
+ the currently running Interface Firmware. All the fields of the 'struct
+ fw_mgmt_ioc_get_fw' are filled by the kernel.
+
+2. FW_MGMT_IOC_GET_BACKEND_FW:
+
+ This ioctl shall be used by the user to get the version of a currently
+ running Backend Interface Firmware identified by a firmware-tag. The user is
+ required to fill the 'firmware_tag' field of the 'struct fw_mgmt_ioc_get_fw'
+ in this case. The 'major' and 'minor' fields are set by the kernel in
+ response.
+
+3. FW_MGMT_IOC_INTF_LOAD_AND_VALIDATE:
+
+ This ioctl shall be used by the user to load an Interface Firmware package on
+ an Interface. The user needs to fill the 'firmware_tag' and 'load_method'
+ fields of the 'struct fw_mgmt_ioc_intf_load_and_validate'. The 'status',
+ 'major' and 'minor' fields are set by the kernel in response.
+
+4. FW_MGMT_IOC_INTF_BACKEND_FW_UPDATE:
+
+ This ioctl shall be used by the user to request an Interface to update a
+ Backend Interface Firmware. The user is required to fill the 'firmware_tag'
+ field of the 'struct fw_mgmt_ioc_get_fw' in this case. The 'status' field is
+ set by the kernel in response.
+
+5. FW_MGMT_IOC_SET_TIMEOUT_MS:
+
+ This ioctl shall be used by the user to increase the timeout interval within
+ which the firmware must get loaded by the Module. The default timeout is 1
+ second. The user needs to pass the timeout in milliseconds.
+
+6. FW_MGMT_IOC_MODE_SWITCH:
+
+ This ioctl shall be used by the user to mode-switch the module to the
+ previously loaded interface firmware. If the interface firmware isn't loaded
+ previously, or if another unsuccessful FW_MGMT_IOC_INTF_LOAD_AND_VALIDATE
+ operation is started after loading interface firmware, then the firmware core
+ wouldn't allow mode-switch.
+
+
+Sysfs Interfaces - Authentication
+---------------------------------
+
+The Component Authentication Protocol interacts with Userspace using the
+character device interface. The character device will be present in /dev/
+directory and will be named gb-authenticate-<N>. The number <N> is assigned at
+runtime.
+
+Identifying the Character Device
+================================
+
+There can be multiple devices present in /dev/ directory with name
+gb-authenticate-N and user first needs to identify the character device used for
+authentication a of particular interface.
+
+The Authentication core creates a device of class 'gb_authenticate', which shall
+be used by the user to identify the right character device for it. The class
+device is created within the Bundle directory for a particular Interface.
+
+For example this is how the class-device can be present:
+
+/sys/bus/greybus/devices/1-1/1-1.1/1-1.1.1/gb_authenticate/gb-authenticate-0
+
+The last name in this path: gb-authenticate-0 is precisely the name of the char
+device and so the device in this case will be:
+
+/dev/gb-authenticate-0.
+
+Operations on the Char device
+=============================
+
+The Character device (/dev/gb-authenticate-0 in above example) can be opened by
+the userspace application and it can perform various 'ioctl' operations on the
+device. The device doesn't support any read/write operations.
+
+Following are the IOCTLs and their data structures available to the user:
+
+#define CAP_CERTIFICATE_MAX_SIZE 1600
+#define CAP_SIGNATURE_MAX_SIZE 320
+
+/* Certificate class types */
+#define CAP_CERT_IMS_EAPC 0x00000001
+#define CAP_CERT_IMS_EASC 0x00000002
+#define CAP_CERT_IMS_EARC 0x00000003
+#define CAP_CERT_IMS_IAPC 0x00000004
+#define CAP_CERT_IMS_IASC 0x00000005
+#define CAP_CERT_IMS_IARC 0x00000006
+
+/* IMS Certificate response result codes */
+#define CAP_IMS_RESULT_CERT_FOUND 0x00
+#define CAP_IMS_RESULT_CERT_CLASS_INVAL 0x01
+#define CAP_IMS_RESULT_CERT_CORRUPT 0x02
+#define CAP_IMS_RESULT_CERT_NOT_FOUND 0x03
+
+/* Authentication types */
+#define CAP_AUTH_IMS_PRI 0x00000001
+#define CAP_AUTH_IMS_SEC 0x00000002
+#define CAP_AUTH_IMS_RSA 0x00000003
+
+/* Authenticate response result codes */
+#define CAP_AUTH_RESULT_CR_SUCCESS 0x00
+#define CAP_AUTH_RESULT_CR_BAD_TYPE 0x01
+#define CAP_AUTH_RESULT_CR_WRONG_EP 0x02
+#define CAP_AUTH_RESULT_CR_NO_KEY 0x03
+#define CAP_AUTH_RESULT_CR_SIG_FAIL 0x04
+
+
+/* IOCTL support */
+struct cap_ioc_get_endpoint_uid {
+ __u8 uid[8];
+} __attribute__ ((__packed__));
+
+struct cap_ioc_get_ims_certificate {
+ __u32 certificate_class;
+ __u32 certificate_id;
+
+ __u8 result_code;
+ __u32 cert_size;
+ __u8 certificate[CAP_CERTIFICATE_MAX_SIZE];
+} __attribute__ ((__packed__));
+
+struct cap_ioc_authenticate {
+ __u32 auth_type;
+ __u8 uid[8];
+ __u8 challenge[32];
+
+ __u8 result_code;
+ __u8 response[64];
+ __u32 signature_size;
+ __u8 signature[CAP_SIGNATURE_MAX_SIZE];
+} __attribute__ ((__packed__));
+
+#define CAP_IOCTL_BASE 'C'
+#define CAP_IOC_GET_ENDPOINT_UID _IOR(CAP_IOCTL_BASE, 0, struct cap_ioc_get_endpoint_uid)
+#define CAP_IOC_GET_IMS_CERTIFICATE _IOWR(CAP_IOCTL_BASE, 1, struct cap_ioc_get_ims_certificate)
+#define CAP_IOC_AUTHENTICATE _IOWR(CAP_IOCTL_BASE, 2, struct cap_ioc_authenticate)
+
+
+1. CAP_IOC_GET_ENDPOINT_UID:
+
+ This ioctl shall be used by the user to get the endpoint UID associated with
+ the Interface. All the fields of the 'struct cap_ioc_get_endpoint_uid' are
+ filled by the kernel.
+
+2. CAP_IOC_GET_IMS_CERTIFICATE:
+
+ This ioctl shall be used by the user to retrieve one of the available
+ cryptographic certificates held by the Interface for use in Component
+ Authentication. The user is required to fill the 'certificate_class' and
+ 'certificate_id' field of the 'struct cap_ioc_get_ims_certificate' in this
+ case. The other fields will be set by the kernel in response. The first
+ 'cert_size' bytes of the 'certificate' shall be read by the user and others
+ must be discarded.
+
+3. CAP_IOC_AUTHENTICATE:
+
+ This ioctl shall be used by the user to authenticate the Module attached to
+ an Interface. The user needs to fill the 'auth_type', 'uid', and 'challenge'
+ fields of the 'struct cap_ioc_authenticate'. The other fields will be set by
+ the kernel in response. The first 'signature_size' bytes of the 'signature'
+ shall be read by the user and others must be discarded.
+
+
+Sysfs Interfaces - Firmware Download
+------------------------------------
+
+The Firmware Download Protocol uses the existing Linux Kernel's Firmware class
+and the interface provided to userspace are described in:
+Documentation/firmware_class/.
+
+
+Sysfs Interfaces - SPI Flash
+----------------------------
+
+The SPI flash is exposed in userspace as a MTD device and is created
+within the Bundle directory. For example, this is how the path may look like:
+
+$ ls /sys/bus/greybus/devices/1-1/1-1.1/1-1.1.1/spi_master/spi32766/spi32766.0/mtd
+mtd0 mtd0ro
+
+
+Sample Applications
+-------------------
+
+The current directory also provides a firmware.c test application, which can be
+referenced while developing userspace application to talk to firmware-management
+protocol.
+
+The current directory also provides a authenticate.c test application, which can
+be referenced while developing userspace application to talk to
+component authentication protocol.
--- /dev/null
+++ b/drivers/greybus/Documentation/firmware/firmware.c
@@ -0,0 +1,262 @@
+/*
+ * Sample code to test firmware-management protocol
+ *
+ * This file is provided under a dual BSD/GPLv2 license. When using or
+ * redistributing this file, you may do so under either license.
+ *
+ * GPL LICENSE SUMMARY
+ *
+ * Copyright(c) 2016 Google Inc. All rights reserved.
+ * Copyright(c) 2016 Linaro Ltd. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License version 2 for more details.
+ *
+ * BSD LICENSE
+ *
+ * Copyright(c) 2016 Google Inc. All rights reserved.
+ * Copyright(c) 2016 Linaro Ltd. All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of Google Inc. or Linaro Ltd. nor the names of
+ * its contributors may be used to endorse or promote products
+ * derived from this software without specific prior written
+ * permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL GOOGLE INC. OR
+ * LINARO LTD. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+ * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <stdio.h>
+#include <string.h>
+#include <unistd.h>
+#include <sys/ioctl.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+
+#include "../../greybus_firmware.h"
+
+#define FW_DEV_DEFAULT "/dev/gb-fw-mgmt-0"
+#define FW_TAG_INT_DEFAULT "s3f"
+#define FW_TAG_BCND_DEFAULT "bf_01"
+#define FW_UPDATE_TYPE_DEFAULT 0
+#define FW_TIMEOUT_DEFAULT 10000;
+
+static const char *firmware_tag;
+static const char *fwdev = FW_DEV_DEFAULT;
+static int fw_update_type = FW_UPDATE_TYPE_DEFAULT;
+static int fw_timeout = FW_TIMEOUT_DEFAULT;
+
+static struct fw_mgmt_ioc_get_intf_version intf_fw_info;
+static struct fw_mgmt_ioc_get_backend_version backend_fw_info;
+static struct fw_mgmt_ioc_intf_load_and_validate intf_load;
+static struct fw_mgmt_ioc_backend_fw_update backend_update;
+
+static void usage(void)
+{
+ printf("\nUsage: ./firmware <gb-fw-mgmt-X (default: gb-fw-mgmt-0)> <interface: 0, backend: 1 (default: 0)> <firmware-tag> (default: \"s3f\"/\"bf_01\") <timeout (default: 10000 ms)>\n");
+}
+
+static int update_intf_firmware(int fd)
+{
+ int ret;
+
+ /* Get Interface Firmware Version */
+ printf("Get Interface Firmware Version\n");
+
+ ret = ioctl(fd, FW_MGMT_IOC_GET_INTF_FW, &intf_fw_info);
+ if (ret < 0) {
+ printf("Failed to get interface firmware version: %s (%d)\n",
+ fwdev, ret);
+ return -1;
+ }
+
+ printf("Interface Firmware tag (%s), major (%d), minor (%d)\n",
+ intf_fw_info.firmware_tag, intf_fw_info.major,
+ intf_fw_info.minor);
+
+ /* Try Interface Firmware load over Unipro */
+ printf("Loading Interface Firmware\n");
+
+ intf_load.load_method = GB_FW_U_LOAD_METHOD_UNIPRO;
+ intf_load.status = 0;
+ intf_load.major = 0;
+ intf_load.minor = 0;
+
+ strncpy((char *)&intf_load.firmware_tag, firmware_tag,
+ GB_FIRMWARE_U_TAG_MAX_SIZE);
+
+ ret = ioctl(fd, FW_MGMT_IOC_INTF_LOAD_AND_VALIDATE, &intf_load);
+ if (ret < 0) {
+ printf("Failed to load interface firmware: %s (%d)\n", fwdev,
+ ret);
+ return -1;
+ }
+
+ if (intf_load.status != GB_FW_U_LOAD_STATUS_VALIDATED &&
+ intf_load.status != GB_FW_U_LOAD_STATUS_UNVALIDATED) {
+ printf("Load status says loading failed: %d\n",
+ intf_load.status);
+ return -1;
+ }
+
+ printf("Interface Firmware (%s) Load done: major: %d, minor: %d, status: %d\n",
+ firmware_tag, intf_load.major, intf_load.minor,
+ intf_load.status);
+
+ /* Initiate Mode-switch to the newly loaded firmware */
+ printf("Initiate Mode switch\n");
+
+ ret = ioctl(fd, FW_MGMT_IOC_MODE_SWITCH);
+ if (ret < 0)
+ printf("Failed to initiate mode-switch (%d)\n", ret);
+
+ return ret;
+}
+
+static int update_backend_firmware(int fd)
+{
+ int ret;
+
+ /* Get Backend Firmware Version */
+ printf("Getting Backend Firmware Version\n");
+
+ strncpy((char *)&backend_fw_info.firmware_tag, firmware_tag,
+ GB_FIRMWARE_U_TAG_MAX_SIZE);
+
+retry_fw_version:
+ ret = ioctl(fd, FW_MGMT_IOC_GET_BACKEND_FW, &backend_fw_info);
+ if (ret < 0) {
+ printf("Failed to get backend firmware version: %s (%d)\n",
+ fwdev, ret);
+ return -1;
+ }
+
+ printf("Backend Firmware tag (%s), major (%d), minor (%d), status (%d)\n",
+ backend_fw_info.firmware_tag, backend_fw_info.major,
+ backend_fw_info.minor, backend_fw_info.status);
+
+ if (backend_fw_info.status == GB_FW_U_BACKEND_VERSION_STATUS_RETRY)
+ goto retry_fw_version;
+
+ if ((backend_fw_info.status != GB_FW_U_BACKEND_VERSION_STATUS_SUCCESS)
+ && (backend_fw_info.status != GB_FW_U_BACKEND_VERSION_STATUS_NOT_AVAILABLE)) {
+ printf("Failed to get backend firmware version: %s (%d)\n",
+ fwdev, backend_fw_info.status);
+ return -1;
+ }
+
+ /* Try Backend Firmware Update over Unipro */
+ printf("Updating Backend Firmware\n");
+
+ strncpy((char *)&backend_update.firmware_tag, firmware_tag,
+ GB_FIRMWARE_U_TAG_MAX_SIZE);
+
+retry_fw_update:
+ backend_update.status = 0;
+
+ ret = ioctl(fd, FW_MGMT_IOC_INTF_BACKEND_FW_UPDATE, &backend_update);
+ if (ret < 0) {
+ printf("Failed to load backend firmware: %s (%d)\n", fwdev, ret);
+ return -1;
+ }
+
+ if (backend_update.status == GB_FW_U_BACKEND_FW_STATUS_RETRY) {
+ printf("Retrying firmware update: %d\n", backend_update.status);
+ goto retry_fw_update;
+ }
+
+ if (backend_update.status != GB_FW_U_BACKEND_FW_STATUS_SUCCESS) {
+ printf("Load status says loading failed: %d\n",
+ backend_update.status);
+ } else {
+ printf("Backend Firmware (%s) Load done: status: %d\n",
+ firmware_tag, backend_update.status);
+ }
+
+ return 0;
+}
+
+int main(int argc, char *argv[])
+{
+ int fd, ret;
+
+ if (argc > 1 &&
+ (!strcmp(argv[1], "-h") || !strcmp(argv[1], "--help"))) {
+ usage();
+ return -1;
+ }
+
+ if (argc > 1)
+ fwdev = argv[1];
+
+ if (argc > 2)
+ sscanf(argv[2], "%u", &fw_update_type);
+
+ if (argc > 3) {
+ firmware_tag = argv[3];
+ } else if (!fw_update_type) {
+ firmware_tag = FW_TAG_INT_DEFAULT;
+ } else {
+ firmware_tag = FW_TAG_BCND_DEFAULT;
+ }
+
+ if (argc > 4)
+ sscanf(argv[4], "%u", &fw_timeout);
+
+ printf("Trying Firmware update: fwdev: %s, type: %s, tag: %s, timeout: %d\n",
+ fwdev, fw_update_type == 0 ? "interface" : "backend",
+ firmware_tag, fw_timeout);
+
+ printf("Opening %s firmware management device\n", fwdev);
+
+ fd = open(fwdev, O_RDWR);
+ if (fd < 0) {
+ printf("Failed to open: %s\n", fwdev);
+ return -1;
+ }
+
+ /* Set Timeout */
+ printf("Setting timeout to %u ms\n", fw_timeout);
+
+ ret = ioctl(fd, FW_MGMT_IOC_SET_TIMEOUT_MS, &fw_timeout);
+ if (ret < 0) {
+ printf("Failed to set timeout: %s (%d)\n", fwdev, ret);
+ ret = -1;
+ goto close_fd;
+ }
+
+ if (!fw_update_type)
+ ret = update_intf_firmware(fd);
+ else
+ ret = update_backend_firmware(fd);
+
+close_fd:
+ close(fd);
+
+ return ret;
+}
--- /dev/null
+++ b/drivers/greybus/firmware.h
@@ -0,0 +1,42 @@
+/*
+ * Greybus Firmware Management Header
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __FIRMWARE_H
+#define __FIRMWARE_H
+
+#include "greybus.h"
+
+#define FW_NAME_PREFIX "gmp_"
+
+/*
+ * Length of the string in format: "FW_NAME_PREFIX""%08x_%08x_%08x_%08x_%s.tftf"
+ * (3 + 1 + 4 * (8 + 1) + 10 + 1 + 4 + 1)
+ */
+#define FW_NAME_SIZE 56
+
+/* Firmware Management Protocol specific functions */
+int fw_mgmt_init(void);
+void fw_mgmt_exit(void);
+struct gb_connection *to_fw_mgmt_connection(struct device *dev);
+int gb_fw_mgmt_request_handler(struct gb_operation *op);
+int gb_fw_mgmt_connection_init(struct gb_connection *connection);
+void gb_fw_mgmt_connection_exit(struct gb_connection *connection);
+
+/* Firmware Download Protocol specific functions */
+int gb_fw_download_request_handler(struct gb_operation *op);
+int gb_fw_download_connection_init(struct gb_connection *connection);
+void gb_fw_download_connection_exit(struct gb_connection *connection);
+
+/* CAP Protocol specific functions */
+int cap_init(void);
+void cap_exit(void);
+int gb_cap_connection_init(struct gb_connection *connection);
+void gb_cap_connection_exit(struct gb_connection *connection);
+
+#endif /* __FIRMWARE_H */
--- /dev/null
+++ b/drivers/greybus/fw-core.c
@@ -0,0 +1,312 @@
+/*
+ * Greybus Firmware Core Bundle Driver.
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/firmware.h>
+#include "firmware.h"
+#include "greybus.h"
+#include "spilib.h"
+
+struct gb_fw_core {
+ struct gb_connection *download_connection;
+ struct gb_connection *mgmt_connection;
+ struct gb_connection *spi_connection;
+ struct gb_connection *cap_connection;
+};
+
+static struct spilib_ops *spilib_ops;
+
+struct gb_connection *to_fw_mgmt_connection(struct device *dev)
+{
+ struct gb_fw_core *fw_core = dev_get_drvdata(dev);
+
+ return fw_core->mgmt_connection;
+}
+
+static int gb_fw_spi_connection_init(struct gb_connection *connection)
+{
+ int ret;
+
+ if (!connection)
+ return 0;
+
+ ret = gb_connection_enable(connection);
+ if (ret)
+ return ret;
+
+ ret = gb_spilib_master_init(connection, &connection->bundle->dev,
+ spilib_ops);
+ if (ret) {
+ gb_connection_disable(connection);
+ return ret;
+ }
+
+ return 0;
+}
+
+static void gb_fw_spi_connection_exit(struct gb_connection *connection)
+{
+ if (!connection)
+ return;
+
+ gb_spilib_master_exit(connection);
+ gb_connection_disable(connection);
+}
+
+static int gb_fw_core_probe(struct gb_bundle *bundle,
+ const struct greybus_bundle_id *id)
+{
+ struct greybus_descriptor_cport *cport_desc;
+ struct gb_connection *connection;
+ struct gb_fw_core *fw_core;
+ int ret, i;
+ u16 cport_id;
+ u8 protocol_id;
+
+ fw_core = kzalloc(sizeof(*fw_core), GFP_KERNEL);
+ if (!fw_core)
+ return -ENOMEM;
+
+ /* Parse CPorts and create connections */
+ for (i = 0; i < bundle->num_cports; i++) {
+ cport_desc = &bundle->cport_desc[i];
+ cport_id = le16_to_cpu(cport_desc->id);
+ protocol_id = cport_desc->protocol_id;
+
+ switch (protocol_id) {
+ case GREYBUS_PROTOCOL_FW_MANAGEMENT:
+ /* Disallow multiple Firmware Management CPorts */
+ if (fw_core->mgmt_connection) {
+ dev_err(&bundle->dev,
+ "multiple management CPorts found\n");
+ ret = -EINVAL;
+ goto err_destroy_connections;
+ }
+
+ connection = gb_connection_create(bundle, cport_id,
+ gb_fw_mgmt_request_handler);
+ if (IS_ERR(connection)) {
+ ret = PTR_ERR(connection);
+ dev_err(&bundle->dev,
+ "failed to create management connection (%d)\n",
+ ret);
+ goto err_destroy_connections;
+ }
+
+ fw_core->mgmt_connection = connection;
+ break;
+ case GREYBUS_PROTOCOL_FW_DOWNLOAD:
+ /* Disallow multiple Firmware Download CPorts */
+ if (fw_core->download_connection) {
+ dev_err(&bundle->dev,
+ "multiple download CPorts found\n");
+ ret = -EINVAL;
+ goto err_destroy_connections;
+ }
+
+ connection = gb_connection_create(bundle, cport_id,
+ gb_fw_download_request_handler);
+ if (IS_ERR(connection)) {
+ dev_err(&bundle->dev, "failed to create download connection (%ld)\n",
+ PTR_ERR(connection));
+ } else {
+ fw_core->download_connection = connection;
+ }
+
+ break;
+ case GREYBUS_PROTOCOL_SPI:
+ /* Disallow multiple SPI CPorts */
+ if (fw_core->spi_connection) {
+ dev_err(&bundle->dev,
+ "multiple SPI CPorts found\n");
+ ret = -EINVAL;
+ goto err_destroy_connections;
+ }
+
+ connection = gb_connection_create(bundle, cport_id,
+ NULL);
+ if (IS_ERR(connection)) {
+ dev_err(&bundle->dev, "failed to create SPI connection (%ld)\n",
+ PTR_ERR(connection));
+ } else {
+ fw_core->spi_connection = connection;
+ }
+
+ break;
+ case GREYBUS_PROTOCOL_AUTHENTICATION:
+ /* Disallow multiple CAP CPorts */
+ if (fw_core->cap_connection) {
+ dev_err(&bundle->dev, "multiple Authentication CPorts found\n");
+ ret = -EINVAL;
+ goto err_destroy_connections;
+ }
+
+ connection = gb_connection_create(bundle, cport_id,
+ NULL);
+ if (IS_ERR(connection)) {
+ dev_err(&bundle->dev, "failed to create Authentication connection (%ld)\n",
+ PTR_ERR(connection));
+ } else {
+ fw_core->cap_connection = connection;
+ }
+
+ break;
+ default:
+ dev_err(&bundle->dev, "invalid protocol id (0x%02x)\n",
+ protocol_id);
+ ret = -EINVAL;
+ goto err_destroy_connections;
+ }
+ }
+
+ /* Firmware Management connection is mandatory */
+ if (!fw_core->mgmt_connection) {
+ dev_err(&bundle->dev, "missing management connection\n");
+ ret = -ENODEV;
+ goto err_destroy_connections;
+ }
+
+ ret = gb_fw_download_connection_init(fw_core->download_connection);
+ if (ret) {
+ /* We may still be able to work with the Interface */
+ dev_err(&bundle->dev, "failed to initialize firmware download connection, disable it (%d)\n",
+ ret);
+ gb_connection_destroy(fw_core->download_connection);
+ fw_core->download_connection = NULL;
+ }
+
+ ret = gb_fw_spi_connection_init(fw_core->spi_connection);
+ if (ret) {
+ /* We may still be able to work with the Interface */
+ dev_err(&bundle->dev, "failed to initialize SPI connection, disable it (%d)\n",
+ ret);
+ gb_connection_destroy(fw_core->spi_connection);
+ fw_core->spi_connection = NULL;
+ }
+
+ ret = gb_cap_connection_init(fw_core->cap_connection);
+ if (ret) {
+ /* We may still be able to work with the Interface */
+ dev_err(&bundle->dev, "failed to initialize CAP connection, disable it (%d)\n",
+ ret);
+ gb_connection_destroy(fw_core->cap_connection);
+ fw_core->cap_connection = NULL;
+ }
+
+ ret = gb_fw_mgmt_connection_init(fw_core->mgmt_connection);
+ if (ret) {
+ /* We may still be able to work with the Interface */
+ dev_err(&bundle->dev, "failed to initialize firmware management connection, disable it (%d)\n",
+ ret);
+ goto err_exit_connections;
+ }
+
+ greybus_set_drvdata(bundle, fw_core);
+
+ /* FIXME: Remove this after S2 Loader gets runtime PM support */
+ if (!(bundle->intf->quirks & GB_INTERFACE_QUIRK_NO_PM))
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return 0;
+
+err_exit_connections:
+ gb_cap_connection_exit(fw_core->cap_connection);
+ gb_fw_spi_connection_exit(fw_core->spi_connection);
+ gb_fw_download_connection_exit(fw_core->download_connection);
+err_destroy_connections:
+ gb_connection_destroy(fw_core->mgmt_connection);
+ gb_connection_destroy(fw_core->cap_connection);
+ gb_connection_destroy(fw_core->spi_connection);
+ gb_connection_destroy(fw_core->download_connection);
+ kfree(fw_core);
+
+ return ret;
+}
+
+static void gb_fw_core_disconnect(struct gb_bundle *bundle)
+{
+ struct gb_fw_core *fw_core = greybus_get_drvdata(bundle);
+ int ret;
+
+ /* FIXME: Remove this after S2 Loader gets runtime PM support */
+ if (!(bundle->intf->quirks & GB_INTERFACE_QUIRK_NO_PM)) {
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ gb_pm_runtime_get_noresume(bundle);
+ }
+
+ gb_fw_mgmt_connection_exit(fw_core->mgmt_connection);
+ gb_cap_connection_exit(fw_core->cap_connection);
+ gb_fw_spi_connection_exit(fw_core->spi_connection);
+ gb_fw_download_connection_exit(fw_core->download_connection);
+
+ gb_connection_destroy(fw_core->mgmt_connection);
+ gb_connection_destroy(fw_core->cap_connection);
+ gb_connection_destroy(fw_core->spi_connection);
+ gb_connection_destroy(fw_core->download_connection);
+
+ kfree(fw_core);
+}
+
+static const struct greybus_bundle_id gb_fw_core_id_table[] = {
+ { GREYBUS_DEVICE_CLASS(GREYBUS_CLASS_FW_MANAGEMENT) },
+ { }
+};
+
+static struct greybus_driver gb_fw_core_driver = {
+ .name = "gb-firmware",
+ .probe = gb_fw_core_probe,
+ .disconnect = gb_fw_core_disconnect,
+ .id_table = gb_fw_core_id_table,
+};
+
+static int fw_core_init(void)
+{
+ int ret;
+
+ ret = fw_mgmt_init();
+ if (ret) {
+ pr_err("Failed to initialize fw-mgmt core (%d)\n", ret);
+ return ret;
+ }
+
+ ret = cap_init();
+ if (ret) {
+ pr_err("Failed to initialize component authentication core (%d)\n",
+ ret);
+ goto fw_mgmt_exit;
+ }
+
+ ret = greybus_register(&gb_fw_core_driver);
+ if (ret)
+ goto cap_exit;
+
+ return 0;
+
+cap_exit:
+ cap_exit();
+fw_mgmt_exit:
+ fw_mgmt_exit();
+
+ return ret;
+}
+module_init(fw_core_init);
+
+static void __exit fw_core_exit(void)
+{
+ greybus_deregister(&gb_fw_core_driver);
+ cap_exit();
+ fw_mgmt_exit();
+}
+module_exit(fw_core_exit);
+
+MODULE_ALIAS("greybus:firmware");
+MODULE_AUTHOR("Viresh Kumar <viresh.kumar@linaro.org>");
+MODULE_DESCRIPTION("Greybus Firmware Bundle Driver");
+MODULE_LICENSE("GPL v2");
--- /dev/null
+++ b/drivers/greybus/fw-download.c
@@ -0,0 +1,465 @@
+/*
+ * Greybus Firmware Download Protocol Driver.
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/firmware.h>
+#include <linux/jiffies.h>
+#include <linux/mutex.h>
+#include <linux/workqueue.h>
+#include "firmware.h"
+#include "greybus.h"
+
+/* Estimated minimum buffer size, actual size can be smaller than this */
+#define MIN_FETCH_SIZE 512
+/* Timeout, in jiffies, within which fetch or release firmware must be called */
+#define NEXT_REQ_TIMEOUT_J msecs_to_jiffies(1000)
+
+struct fw_request {
+ u8 firmware_id;
+ bool disabled;
+ bool timedout;
+ char name[FW_NAME_SIZE];
+ const struct firmware *fw;
+ struct list_head node;
+
+ struct delayed_work dwork;
+ /* Timeout, in jiffies, within which the firmware shall download */
+ unsigned long release_timeout_j;
+ struct kref kref;
+ struct fw_download *fw_download;
+};
+
+struct fw_download {
+ struct device *parent;
+ struct gb_connection *connection;
+ struct list_head fw_requests;
+ struct ida id_map;
+ struct mutex mutex;
+};
+
+static void fw_req_release(struct kref *kref)
+{
+ struct fw_request *fw_req = container_of(kref, struct fw_request, kref);
+
+ dev_dbg(fw_req->fw_download->parent, "firmware %s released\n",
+ fw_req->name);
+
+ release_firmware(fw_req->fw);
+
+ /*
+ * The request timed out and the module may send a fetch-fw or
+ * release-fw request later. Lets block the id we allocated for this
+ * request, so that the AP doesn't refer to a later fw-request (with
+ * same firmware_id) for the old timedout fw-request.
+ *
+ * NOTE:
+ *
+ * This also means that after 255 timeouts we will fail to service new
+ * firmware downloads. But what else can we do in that case anyway? Lets
+ * just hope that it never happens.
+ */
+ if (!fw_req->timedout)
+ ida_simple_remove(&fw_req->fw_download->id_map,
+ fw_req->firmware_id);
+
+ kfree(fw_req);
+}
+
+/*
+ * Incoming requests are serialized for a connection, and the only race possible
+ * is between the timeout handler freeing this and an incoming request.
+ *
+ * The operations on the fw-request list are protected by the mutex and
+ * get_fw_req() increments the reference count before returning a fw_req pointer
+ * to the users.
+ *
+ * free_firmware() also takes the mutex while removing an entry from the list,
+ * it guarantees that every user of fw_req has taken a kref-reference by now and
+ * we wouldn't have any new users.
+ *
+ * Once the last user drops the reference, the fw_req structure is freed.
+ */
+static void put_fw_req(struct fw_request *fw_req)
+{
+ kref_put(&fw_req->kref, fw_req_release);
+}
+
+/* Caller must call put_fw_req() after using struct fw_request */
+static struct fw_request *get_fw_req(struct fw_download *fw_download,
+ u8 firmware_id)
+{
+ struct fw_request *fw_req;
+
+ mutex_lock(&fw_download->mutex);
+
+ list_for_each_entry(fw_req, &fw_download->fw_requests, node) {
+ if (fw_req->firmware_id == firmware_id) {
+ kref_get(&fw_req->kref);
+ goto unlock;
+ }
+ }
+
+ fw_req = NULL;
+
+unlock:
+ mutex_unlock(&fw_download->mutex);
+
+ return fw_req;
+}
+
+static void free_firmware(struct fw_download *fw_download,
+ struct fw_request *fw_req)
+{
+ /* Already disabled from timeout handlers */
+ if (fw_req->disabled)
+ return;
+
+ mutex_lock(&fw_download->mutex);
+ list_del(&fw_req->node);
+ mutex_unlock(&fw_download->mutex);
+
+ fw_req->disabled = true;
+ put_fw_req(fw_req);
+}
+
+static void fw_request_timedout(struct work_struct *work)
+{
+ struct delayed_work *dwork = to_delayed_work(work);
+ struct fw_request *fw_req = container_of(dwork, struct fw_request, dwork);
+ struct fw_download *fw_download = fw_req->fw_download;
+
+ dev_err(fw_download->parent,
+ "Timed out waiting for fetch / release firmware requests: %u\n",
+ fw_req->firmware_id);
+
+ fw_req->timedout = true;
+ free_firmware(fw_download, fw_req);
+}
+
+static int exceeds_release_timeout(struct fw_request *fw_req)
+{
+ struct fw_download *fw_download = fw_req->fw_download;
+
+ if (time_before(jiffies, fw_req->release_timeout_j))
+ return 0;
+
+ dev_err(fw_download->parent,
+ "Firmware download didn't finish in time, abort: %d\n",
+ fw_req->firmware_id);
+
+ fw_req->timedout = true;
+ free_firmware(fw_download, fw_req);
+
+ return -ETIMEDOUT;
+}
+
+/* This returns path of the firmware blob on the disk */
+static struct fw_request *find_firmware(struct fw_download *fw_download,
+ const char *tag)
+{
+ struct gb_interface *intf = fw_download->connection->bundle->intf;
+ struct fw_request *fw_req;
+ int ret, req_count;
+
+ fw_req = kzalloc(sizeof(*fw_req), GFP_KERNEL);
+ if (!fw_req)
+ return ERR_PTR(-ENOMEM);
+
+ /* Allocate ids from 1 to 255 (u8-max), 0 is an invalid id */
+ ret = ida_simple_get(&fw_download->id_map, 1, 256, GFP_KERNEL);
+ if (ret < 0) {
+ dev_err(fw_download->parent,
+ "failed to allocate firmware id (%d)\n", ret);
+ goto err_free_req;
+ }
+ fw_req->firmware_id = ret;
+
+ snprintf(fw_req->name, sizeof(fw_req->name),
+ FW_NAME_PREFIX "%08x_%08x_%08x_%08x_%s.tftf",
+ intf->ddbl1_manufacturer_id, intf->ddbl1_product_id,
+ intf->vendor_id, intf->product_id, tag);
+
+ dev_info(fw_download->parent, "Requested firmware package '%s'\n",
+ fw_req->name);
+
+ ret = request_firmware(&fw_req->fw, fw_req->name, fw_download->parent);
+ if (ret) {
+ dev_err(fw_download->parent,
+ "firmware request failed for %s (%d)\n", fw_req->name,
+ ret);
+ goto err_free_id;
+ }
+
+ fw_req->fw_download = fw_download;
+ kref_init(&fw_req->kref);
+
+ mutex_lock(&fw_download->mutex);
+ list_add(&fw_req->node, &fw_download->fw_requests);
+ mutex_unlock(&fw_download->mutex);
+
+ /* Timeout, in jiffies, within which firmware should get loaded */
+ req_count = DIV_ROUND_UP(fw_req->fw->size, MIN_FETCH_SIZE);
+ fw_req->release_timeout_j = jiffies + req_count * NEXT_REQ_TIMEOUT_J;
+
+ INIT_DELAYED_WORK(&fw_req->dwork, fw_request_timedout);
+ schedule_delayed_work(&fw_req->dwork, NEXT_REQ_TIMEOUT_J);
+
+ return fw_req;
+
+err_free_id:
+ ida_simple_remove(&fw_download->id_map, fw_req->firmware_id);
+err_free_req:
+ kfree(fw_req);
+
+ return ERR_PTR(ret);
+}
+
+static int fw_download_find_firmware(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct fw_download *fw_download = gb_connection_get_data(connection);
+ struct gb_fw_download_find_firmware_request *request;
+ struct gb_fw_download_find_firmware_response *response;
+ struct fw_request *fw_req;
+ const char *tag;
+
+ if (op->request->payload_size != sizeof(*request)) {
+ dev_err(fw_download->parent,
+ "illegal size of find firmware request (%zu != %zu)\n",
+ op->request->payload_size, sizeof(*request));
+ return -EINVAL;
+ }
+
+ request = op->request->payload;
+ tag = (const char *)request->firmware_tag;
+
+ /* firmware_tag must be null-terminated */
+ if (strnlen(tag, GB_FIRMWARE_TAG_MAX_SIZE) == GB_FIRMWARE_TAG_MAX_SIZE) {
+ dev_err(fw_download->parent,
+ "firmware-tag is not null-terminated\n");
+ return -EINVAL;
+ }
+
+ fw_req = find_firmware(fw_download, tag);
+ if (IS_ERR(fw_req))
+ return PTR_ERR(fw_req);
+
+ if (!gb_operation_response_alloc(op, sizeof(*response), GFP_KERNEL)) {
+ dev_err(fw_download->parent, "error allocating response\n");
+ free_firmware(fw_download, fw_req);
+ return -ENOMEM;
+ }
+
+ response = op->response->payload;
+ response->firmware_id = fw_req->firmware_id;
+ response->size = cpu_to_le32(fw_req->fw->size);
+
+ dev_dbg(fw_download->parent,
+ "firmware size is %zu bytes\n", fw_req->fw->size);
+
+ return 0;
+}
+
+static int fw_download_fetch_firmware(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct fw_download *fw_download = gb_connection_get_data(connection);
+ struct gb_fw_download_fetch_firmware_request *request;
+ struct gb_fw_download_fetch_firmware_response *response;
+ struct fw_request *fw_req;
+ const struct firmware *fw;
+ unsigned int offset, size;
+ u8 firmware_id;
+ int ret = 0;
+
+ if (op->request->payload_size != sizeof(*request)) {
+ dev_err(fw_download->parent,
+ "Illegal size of fetch firmware request (%zu %zu)\n",
+ op->request->payload_size, sizeof(*request));
+ return -EINVAL;
+ }
+
+ request = op->request->payload;
+ offset = le32_to_cpu(request->offset);
+ size = le32_to_cpu(request->size);
+ firmware_id = request->firmware_id;
+
+ fw_req = get_fw_req(fw_download, firmware_id);
+ if (!fw_req) {
+ dev_err(fw_download->parent,
+ "firmware not available for id: %02u\n", firmware_id);
+ return -EINVAL;
+ }
+
+ /* Make sure work handler isn't running in parallel */
+ cancel_delayed_work_sync(&fw_req->dwork);
+
+ /* We timed-out before reaching here ? */
+ if (fw_req->disabled) {
+ ret = -ETIMEDOUT;
+ goto put_fw;
+ }
+
+ /*
+ * Firmware download must finish within a limited time interval. If it
+ * doesn't, then we might have a buggy Module on the other side. Abort
+ * download.
+ */
+ ret = exceeds_release_timeout(fw_req);
+ if (ret)
+ goto put_fw;
+
+ fw = fw_req->fw;
+
+ if (offset >= fw->size || size > fw->size - offset) {
+ dev_err(fw_download->parent,
+ "bad fetch firmware request (offs = %u, size = %u)\n",
+ offset, size);
+ ret = -EINVAL;
+ goto put_fw;
+ }
+
+ if (!gb_operation_response_alloc(op, sizeof(*response) + size,
+ GFP_KERNEL)) {
+ dev_err(fw_download->parent,
+ "error allocating fetch firmware response\n");
+ ret = -ENOMEM;
+ goto put_fw;
+ }
+
+ response = op->response->payload;
+ memcpy(response->data, fw->data + offset, size);
+
+ dev_dbg(fw_download->parent,
+ "responding with firmware (offs = %u, size = %u)\n", offset,
+ size);
+
+ /* Refresh timeout */
+ schedule_delayed_work(&fw_req->dwork, NEXT_REQ_TIMEOUT_J);
+
+put_fw:
+ put_fw_req(fw_req);
+
+ return ret;
+}
+
+static int fw_download_release_firmware(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct fw_download *fw_download = gb_connection_get_data(connection);
+ struct gb_fw_download_release_firmware_request *request;
+ struct fw_request *fw_req;
+ u8 firmware_id;
+
+ if (op->request->payload_size != sizeof(*request)) {
+ dev_err(fw_download->parent,
+ "Illegal size of release firmware request (%zu %zu)\n",
+ op->request->payload_size, sizeof(*request));
+ return -EINVAL;
+ }
+
+ request = op->request->payload;
+ firmware_id = request->firmware_id;
+
+ fw_req = get_fw_req(fw_download, firmware_id);
+ if (!fw_req) {
+ dev_err(fw_download->parent,
+ "firmware not available for id: %02u\n", firmware_id);
+ return -EINVAL;
+ }
+
+ cancel_delayed_work_sync(&fw_req->dwork);
+
+ free_firmware(fw_download, fw_req);
+ put_fw_req(fw_req);
+
+ dev_dbg(fw_download->parent, "release firmware\n");
+
+ return 0;
+}
+
+int gb_fw_download_request_handler(struct gb_operation *op)
+{
+ u8 type = op->type;
+
+ switch (type) {
+ case GB_FW_DOWNLOAD_TYPE_FIND_FIRMWARE:
+ return fw_download_find_firmware(op);
+ case GB_FW_DOWNLOAD_TYPE_FETCH_FIRMWARE:
+ return fw_download_fetch_firmware(op);
+ case GB_FW_DOWNLOAD_TYPE_RELEASE_FIRMWARE:
+ return fw_download_release_firmware(op);
+ default:
+ dev_err(&op->connection->bundle->dev,
+ "unsupported request: %u\n", type);
+ return -EINVAL;
+ }
+}
+
+int gb_fw_download_connection_init(struct gb_connection *connection)
+{
+ struct fw_download *fw_download;
+ int ret;
+
+ if (!connection)
+ return 0;
+
+ fw_download = kzalloc(sizeof(*fw_download), GFP_KERNEL);
+ if (!fw_download)
+ return -ENOMEM;
+
+ fw_download->parent = &connection->bundle->dev;
+ INIT_LIST_HEAD(&fw_download->fw_requests);
+ ida_init(&fw_download->id_map);
+ gb_connection_set_data(connection, fw_download);
+ fw_download->connection = connection;
+ mutex_init(&fw_download->mutex);
+
+ ret = gb_connection_enable(connection);
+ if (ret)
+ goto err_destroy_id_map;
+
+ return 0;
+
+err_destroy_id_map:
+ ida_destroy(&fw_download->id_map);
+ kfree(fw_download);
+
+ return ret;
+}
+
+void gb_fw_download_connection_exit(struct gb_connection *connection)
+{
+ struct fw_download *fw_download;
+ struct fw_request *fw_req, *tmp;
+
+ if (!connection)
+ return;
+
+ fw_download = gb_connection_get_data(connection);
+ gb_connection_disable(fw_download->connection);
+
+ /*
+ * Make sure we have a reference to the pending requests, before they
+ * are freed from the timeout handler.
+ */
+ mutex_lock(&fw_download->mutex);
+ list_for_each_entry(fw_req, &fw_download->fw_requests, node)
+ kref_get(&fw_req->kref);
+ mutex_unlock(&fw_download->mutex);
+
+ /* Release pending firmware packages */
+ list_for_each_entry_safe(fw_req, tmp, &fw_download->fw_requests, node) {
+ cancel_delayed_work_sync(&fw_req->dwork);
+ free_firmware(fw_download, fw_req);
+ put_fw_req(fw_req);
+ }
+
+ ida_destroy(&fw_download->id_map);
+ kfree(fw_download);
+}
--- /dev/null
+++ b/drivers/greybus/fw-management.c
@@ -0,0 +1,721 @@
+/*
+ * Greybus Firmware Management Protocol Driver.
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/cdev.h>
+#include <linux/completion.h>
+#include <linux/firmware.h>
+#include <linux/fs.h>
+#include <linux/idr.h>
+#include <linux/ioctl.h>
+#include <linux/uaccess.h>
+
+#include "firmware.h"
+#include "greybus_firmware.h"
+#include "greybus.h"
+
+#define FW_MGMT_TIMEOUT_MS 1000
+
+struct fw_mgmt {
+ struct device *parent;
+ struct gb_connection *connection;
+ struct kref kref;
+ struct list_head node;
+
+ /* Common id-map for interface and backend firmware requests */
+ struct ida id_map;
+ struct mutex mutex;
+ struct completion completion;
+ struct cdev cdev;
+ struct device *class_device;
+ dev_t dev_num;
+ unsigned int timeout_jiffies;
+ bool disabled; /* connection getting disabled */
+
+ /* Interface Firmware specific fields */
+ bool mode_switch_started;
+ bool intf_fw_loaded;
+ u8 intf_fw_request_id;
+ u8 intf_fw_status;
+ u16 intf_fw_major;
+ u16 intf_fw_minor;
+
+ /* Backend Firmware specific fields */
+ u8 backend_fw_request_id;
+ u8 backend_fw_status;
+};
+
+/*
+ * Number of minor devices this driver supports.
+ * There will be exactly one required per Interface.
+ */
+#define NUM_MINORS U8_MAX
+
+static struct class *fw_mgmt_class;
+static dev_t fw_mgmt_dev_num;
+static DEFINE_IDA(fw_mgmt_minors_map);
+static LIST_HEAD(fw_mgmt_list);
+static DEFINE_MUTEX(list_mutex);
+
+static void fw_mgmt_kref_release(struct kref *kref)
+{
+ struct fw_mgmt *fw_mgmt = container_of(kref, struct fw_mgmt, kref);
+
+ ida_destroy(&fw_mgmt->id_map);
+ kfree(fw_mgmt);
+}
+
+/*
+ * All users of fw_mgmt take a reference (from within list_mutex lock), before
+ * they get a pointer to play with. And the structure will be freed only after
+ * the last user has put the reference to it.
+ */
+static void put_fw_mgmt(struct fw_mgmt *fw_mgmt)
+{
+ kref_put(&fw_mgmt->kref, fw_mgmt_kref_release);
+}
+
+/* Caller must call put_fw_mgmt() after using struct fw_mgmt */
+static struct fw_mgmt *get_fw_mgmt(struct cdev *cdev)
+{
+ struct fw_mgmt *fw_mgmt;
+
+ mutex_lock(&list_mutex);
+
+ list_for_each_entry(fw_mgmt, &fw_mgmt_list, node) {
+ if (&fw_mgmt->cdev == cdev) {
+ kref_get(&fw_mgmt->kref);
+ goto unlock;
+ }
+ }
+
+ fw_mgmt = NULL;
+
+unlock:
+ mutex_unlock(&list_mutex);
+
+ return fw_mgmt;
+}
+
+static int fw_mgmt_interface_fw_version_operation(struct fw_mgmt *fw_mgmt,
+ struct fw_mgmt_ioc_get_intf_version *fw_info)
+{
+ struct gb_connection *connection = fw_mgmt->connection;
+ struct gb_fw_mgmt_interface_fw_version_response response;
+ int ret;
+
+ ret = gb_operation_sync(connection,
+ GB_FW_MGMT_TYPE_INTERFACE_FW_VERSION, NULL, 0,
+ &response, sizeof(response));
+ if (ret) {
+ dev_err(fw_mgmt->parent,
+ "failed to get interface firmware version (%d)\n", ret);
+ return ret;
+ }
+
+ fw_info->major = le16_to_cpu(response.major);
+ fw_info->minor = le16_to_cpu(response.minor);
+
+ strncpy(fw_info->firmware_tag, response.firmware_tag,
+ GB_FIRMWARE_TAG_MAX_SIZE);
+
+ /*
+ * The firmware-tag should be NULL terminated, otherwise throw error but
+ * don't fail.
+ */
+ if (fw_info->firmware_tag[GB_FIRMWARE_TAG_MAX_SIZE - 1] != '\0') {
+ dev_err(fw_mgmt->parent,
+ "fw-version: firmware-tag is not NULL terminated\n");
+ fw_info->firmware_tag[GB_FIRMWARE_TAG_MAX_SIZE - 1] = '\0';
+ }
+
+ return 0;
+}
+
+static int fw_mgmt_load_and_validate_operation(struct fw_mgmt *fw_mgmt,
+ u8 load_method, const char *tag)
+{
+ struct gb_fw_mgmt_load_and_validate_fw_request request;
+ int ret;
+
+ if (load_method != GB_FW_LOAD_METHOD_UNIPRO &&
+ load_method != GB_FW_LOAD_METHOD_INTERNAL) {
+ dev_err(fw_mgmt->parent,
+ "invalid load-method (%d)\n", load_method);
+ return -EINVAL;
+ }
+
+ request.load_method = load_method;
+ strncpy(request.firmware_tag, tag, GB_FIRMWARE_TAG_MAX_SIZE);
+
+ /*
+ * The firmware-tag should be NULL terminated, otherwise throw error and
+ * fail.
+ */
+ if (request.firmware_tag[GB_FIRMWARE_TAG_MAX_SIZE - 1] != '\0') {
+ dev_err(fw_mgmt->parent, "load-and-validate: firmware-tag is not NULL terminated\n");
+ return -EINVAL;
+ }
+
+ /* Allocate ids from 1 to 255 (u8-max), 0 is an invalid id */
+ ret = ida_simple_get(&fw_mgmt->id_map, 1, 256, GFP_KERNEL);
+ if (ret < 0) {
+ dev_err(fw_mgmt->parent, "failed to allocate request id (%d)\n",
+ ret);
+ return ret;
+ }
+
+ fw_mgmt->intf_fw_request_id = ret;
+ fw_mgmt->intf_fw_loaded = false;
+ request.request_id = ret;
+
+ ret = gb_operation_sync(fw_mgmt->connection,
+ GB_FW_MGMT_TYPE_LOAD_AND_VALIDATE_FW, &request,
+ sizeof(request), NULL, 0);
+ if (ret) {
+ ida_simple_remove(&fw_mgmt->id_map,
+ fw_mgmt->intf_fw_request_id);
+ fw_mgmt->intf_fw_request_id = 0;
+ dev_err(fw_mgmt->parent,
+ "load and validate firmware request failed (%d)\n",
+ ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int fw_mgmt_interface_fw_loaded_operation(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct fw_mgmt *fw_mgmt = gb_connection_get_data(connection);
+ struct gb_fw_mgmt_loaded_fw_request *request;
+
+ /* No pending load and validate request ? */
+ if (!fw_mgmt->intf_fw_request_id) {
+ dev_err(fw_mgmt->parent,
+ "unexpected firmware loaded request received\n");
+ return -ENODEV;
+ }
+
+ if (op->request->payload_size != sizeof(*request)) {
+ dev_err(fw_mgmt->parent, "illegal size of firmware loaded request (%zu != %zu)\n",
+ op->request->payload_size, sizeof(*request));
+ return -EINVAL;
+ }
+
+ request = op->request->payload;
+
+ /* Invalid request-id ? */
+ if (request->request_id != fw_mgmt->intf_fw_request_id) {
+ dev_err(fw_mgmt->parent, "invalid request id for firmware loaded request (%02u != %02u)\n",
+ fw_mgmt->intf_fw_request_id, request->request_id);
+ return -ENODEV;
+ }
+
+ ida_simple_remove(&fw_mgmt->id_map, fw_mgmt->intf_fw_request_id);
+ fw_mgmt->intf_fw_request_id = 0;
+ fw_mgmt->intf_fw_status = request->status;
+ fw_mgmt->intf_fw_major = le16_to_cpu(request->major);
+ fw_mgmt->intf_fw_minor = le16_to_cpu(request->minor);
+
+ if (fw_mgmt->intf_fw_status == GB_FW_LOAD_STATUS_FAILED)
+ dev_err(fw_mgmt->parent,
+ "failed to load interface firmware, status:%02x\n",
+ fw_mgmt->intf_fw_status);
+ else if (fw_mgmt->intf_fw_status == GB_FW_LOAD_STATUS_VALIDATION_FAILED)
+ dev_err(fw_mgmt->parent,
+ "failed to validate interface firmware, status:%02x\n",
+ fw_mgmt->intf_fw_status);
+ else
+ fw_mgmt->intf_fw_loaded = true;
+
+ complete(&fw_mgmt->completion);
+
+ return 0;
+}
+
+static int fw_mgmt_backend_fw_version_operation(struct fw_mgmt *fw_mgmt,
+ struct fw_mgmt_ioc_get_backend_version *fw_info)
+{
+ struct gb_connection *connection = fw_mgmt->connection;
+ struct gb_fw_mgmt_backend_fw_version_request request;
+ struct gb_fw_mgmt_backend_fw_version_response response;
+ int ret;
+
+ strncpy(request.firmware_tag, fw_info->firmware_tag,
+ GB_FIRMWARE_TAG_MAX_SIZE);
+
+ /*
+ * The firmware-tag should be NULL terminated, otherwise throw error and
+ * fail.
+ */
+ if (request.firmware_tag[GB_FIRMWARE_TAG_MAX_SIZE - 1] != '\0') {
+ dev_err(fw_mgmt->parent, "backend-version: firmware-tag is not NULL terminated\n");
+ return -EINVAL;
+ }
+
+ ret = gb_operation_sync(connection,
+ GB_FW_MGMT_TYPE_BACKEND_FW_VERSION, &request,
+ sizeof(request), &response, sizeof(response));
+ if (ret) {
+ dev_err(fw_mgmt->parent, "failed to get version of %s backend firmware (%d)\n",
+ fw_info->firmware_tag, ret);
+ return ret;
+ }
+
+ fw_info->status = response.status;
+
+ /* Reset version as that should be non-zero only for success case */
+ fw_info->major = 0;
+ fw_info->minor = 0;
+
+ switch (fw_info->status) {
+ case GB_FW_BACKEND_VERSION_STATUS_SUCCESS:
+ fw_info->major = le16_to_cpu(response.major);
+ fw_info->minor = le16_to_cpu(response.minor);
+ break;
+ case GB_FW_BACKEND_VERSION_STATUS_NOT_AVAILABLE:
+ case GB_FW_BACKEND_VERSION_STATUS_RETRY:
+ break;
+ case GB_FW_BACKEND_VERSION_STATUS_NOT_SUPPORTED:
+ dev_err(fw_mgmt->parent,
+ "Firmware with tag %s is not supported by Interface\n",
+ fw_info->firmware_tag);
+ break;
+ default:
+ dev_err(fw_mgmt->parent, "Invalid status received: %u\n",
+ fw_info->status);
+ }
+
+ return 0;
+}
+
+static int fw_mgmt_backend_fw_update_operation(struct fw_mgmt *fw_mgmt,
+ char *tag)
+{
+ struct gb_fw_mgmt_backend_fw_update_request request;
+ int ret;
+
+ strncpy(request.firmware_tag, tag, GB_FIRMWARE_TAG_MAX_SIZE);
+
+ /*
+ * The firmware-tag should be NULL terminated, otherwise throw error and
+ * fail.
+ */
+ if (request.firmware_tag[GB_FIRMWARE_TAG_MAX_SIZE - 1] != '\0') {
+ dev_err(fw_mgmt->parent, "backend-update: firmware-tag is not NULL terminated\n");
+ return -EINVAL;
+ }
+
+ /* Allocate ids from 1 to 255 (u8-max), 0 is an invalid id */
+ ret = ida_simple_get(&fw_mgmt->id_map, 1, 256, GFP_KERNEL);
+ if (ret < 0) {
+ dev_err(fw_mgmt->parent, "failed to allocate request id (%d)\n",
+ ret);
+ return ret;
+ }
+
+ fw_mgmt->backend_fw_request_id = ret;
+ request.request_id = ret;
+
+ ret = gb_operation_sync(fw_mgmt->connection,
+ GB_FW_MGMT_TYPE_BACKEND_FW_UPDATE, &request,
+ sizeof(request), NULL, 0);
+ if (ret) {
+ ida_simple_remove(&fw_mgmt->id_map,
+ fw_mgmt->backend_fw_request_id);
+ fw_mgmt->backend_fw_request_id = 0;
+ dev_err(fw_mgmt->parent,
+ "backend %s firmware update request failed (%d)\n", tag,
+ ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int fw_mgmt_backend_fw_updated_operation(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct fw_mgmt *fw_mgmt = gb_connection_get_data(connection);
+ struct gb_fw_mgmt_backend_fw_updated_request *request;
+
+ /* No pending load and validate request ? */
+ if (!fw_mgmt->backend_fw_request_id) {
+ dev_err(fw_mgmt->parent, "unexpected backend firmware updated request received\n");
+ return -ENODEV;
+ }
+
+ if (op->request->payload_size != sizeof(*request)) {
+ dev_err(fw_mgmt->parent, "illegal size of backend firmware updated request (%zu != %zu)\n",
+ op->request->payload_size, sizeof(*request));
+ return -EINVAL;
+ }
+
+ request = op->request->payload;
+
+ /* Invalid request-id ? */
+ if (request->request_id != fw_mgmt->backend_fw_request_id) {
+ dev_err(fw_mgmt->parent, "invalid request id for backend firmware updated request (%02u != %02u)\n",
+ fw_mgmt->backend_fw_request_id, request->request_id);
+ return -ENODEV;
+ }
+
+ ida_simple_remove(&fw_mgmt->id_map, fw_mgmt->backend_fw_request_id);
+ fw_mgmt->backend_fw_request_id = 0;
+ fw_mgmt->backend_fw_status = request->status;
+
+ if ((fw_mgmt->backend_fw_status != GB_FW_BACKEND_FW_STATUS_SUCCESS) &&
+ (fw_mgmt->backend_fw_status != GB_FW_BACKEND_FW_STATUS_RETRY))
+ dev_err(fw_mgmt->parent,
+ "failed to load backend firmware: %02x\n",
+ fw_mgmt->backend_fw_status);
+
+ complete(&fw_mgmt->completion);
+
+ return 0;
+}
+
+/* Char device fops */
+
+static int fw_mgmt_open(struct inode *inode, struct file *file)
+{
+ struct fw_mgmt *fw_mgmt = get_fw_mgmt(inode->i_cdev);
+
+ /* fw_mgmt structure can't get freed until file descriptor is closed */
+ if (fw_mgmt) {
+ file->private_data = fw_mgmt;
+ return 0;
+ }
+
+ return -ENODEV;
+}
+
+static int fw_mgmt_release(struct inode *inode, struct file *file)
+{
+ struct fw_mgmt *fw_mgmt = file->private_data;
+
+ put_fw_mgmt(fw_mgmt);
+ return 0;
+}
+
+static int fw_mgmt_ioctl(struct fw_mgmt *fw_mgmt, unsigned int cmd,
+ void __user *buf)
+{
+ struct fw_mgmt_ioc_get_intf_version intf_fw_info;
+ struct fw_mgmt_ioc_get_backend_version backend_fw_info;
+ struct fw_mgmt_ioc_intf_load_and_validate intf_load;
+ struct fw_mgmt_ioc_backend_fw_update backend_update;
+ unsigned int timeout;
+ int ret;
+
+ /* Reject any operations after mode-switch has started */
+ if (fw_mgmt->mode_switch_started)
+ return -EBUSY;
+
+ switch (cmd) {
+ case FW_MGMT_IOC_GET_INTF_FW:
+ ret = fw_mgmt_interface_fw_version_operation(fw_mgmt,
+ &intf_fw_info);
+ if (ret)
+ return ret;
+
+ if (copy_to_user(buf, &intf_fw_info, sizeof(intf_fw_info)))
+ return -EFAULT;
+
+ return 0;
+ case FW_MGMT_IOC_GET_BACKEND_FW:
+ if (copy_from_user(&backend_fw_info, buf,
+ sizeof(backend_fw_info)))
+ return -EFAULT;
+
+ ret = fw_mgmt_backend_fw_version_operation(fw_mgmt,
+ &backend_fw_info);
+ if (ret)
+ return ret;
+
+ if (copy_to_user(buf, &backend_fw_info,
+ sizeof(backend_fw_info)))
+ return -EFAULT;
+
+ return 0;
+ case FW_MGMT_IOC_INTF_LOAD_AND_VALIDATE:
+ if (copy_from_user(&intf_load, buf, sizeof(intf_load)))
+ return -EFAULT;
+
+ ret = fw_mgmt_load_and_validate_operation(fw_mgmt,
+ intf_load.load_method, intf_load.firmware_tag);
+ if (ret)
+ return ret;
+
+ if (!wait_for_completion_timeout(&fw_mgmt->completion,
+ fw_mgmt->timeout_jiffies)) {
+ dev_err(fw_mgmt->parent, "timed out waiting for firmware load and validation to finish\n");
+ return -ETIMEDOUT;
+ }
+
+ intf_load.status = fw_mgmt->intf_fw_status;
+ intf_load.major = fw_mgmt->intf_fw_major;
+ intf_load.minor = fw_mgmt->intf_fw_minor;
+
+ if (copy_to_user(buf, &intf_load, sizeof(intf_load)))
+ return -EFAULT;
+
+ return 0;
+ case FW_MGMT_IOC_INTF_BACKEND_FW_UPDATE:
+ if (copy_from_user(&backend_update, buf,
+ sizeof(backend_update)))
+ return -EFAULT;
+
+ ret = fw_mgmt_backend_fw_update_operation(fw_mgmt,
+ backend_update.firmware_tag);
+ if (ret)
+ return ret;
+
+ if (!wait_for_completion_timeout(&fw_mgmt->completion,
+ fw_mgmt->timeout_jiffies)) {
+ dev_err(fw_mgmt->parent, "timed out waiting for backend firmware update to finish\n");
+ return -ETIMEDOUT;
+ }
+
+ backend_update.status = fw_mgmt->backend_fw_status;
+
+ if (copy_to_user(buf, &backend_update, sizeof(backend_update)))
+ return -EFAULT;
+
+ return 0;
+ case FW_MGMT_IOC_SET_TIMEOUT_MS:
+ if (get_user(timeout, (unsigned int __user *)buf))
+ return -EFAULT;
+
+ if (!timeout) {
+ dev_err(fw_mgmt->parent, "timeout can't be zero\n");
+ return -EINVAL;
+ }
+
+ fw_mgmt->timeout_jiffies = msecs_to_jiffies(timeout);
+
+ return 0;
+ case FW_MGMT_IOC_MODE_SWITCH:
+ if (!fw_mgmt->intf_fw_loaded) {
+ dev_err(fw_mgmt->parent,
+ "Firmware not loaded for mode-switch\n");
+ return -EPERM;
+ }
+
+ /*
+ * Disallow new ioctls as the fw-core bundle driver is going to
+ * get disconnected soon and the character device will get
+ * removed.
+ */
+ fw_mgmt->mode_switch_started = true;
+
+ ret = gb_interface_request_mode_switch(fw_mgmt->connection->intf);
+ if (ret) {
+ dev_err(fw_mgmt->parent, "Mode-switch failed: %d\n",
+ ret);
+ fw_mgmt->mode_switch_started = false;
+ return ret;
+ }
+
+ return 0;
+ default:
+ return -ENOTTY;
+ }
+}
+
+static long fw_mgmt_ioctl_unlocked(struct file *file, unsigned int cmd,
+ unsigned long arg)
+{
+ struct fw_mgmt *fw_mgmt = file->private_data;
+ struct gb_bundle *bundle = fw_mgmt->connection->bundle;
+ int ret = -ENODEV;
+
+ /*
+ * Serialize ioctls.
+ *
+ * We don't want the user to do few operations in parallel. For example,
+ * updating Interface firmware in parallel for the same Interface. There
+ * is no need to do things in parallel for speed and we can avoid having
+ * complicated code for now.
+ *
+ * This is also used to protect ->disabled, which is used to check if
+ * the connection is getting disconnected, so that we don't start any
+ * new operations.
+ */
+ mutex_lock(&fw_mgmt->mutex);
+ if (!fw_mgmt->disabled) {
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (!ret) {
+ ret = fw_mgmt_ioctl(fw_mgmt, cmd, (void __user *)arg);
+ gb_pm_runtime_put_autosuspend(bundle);
+ }
+ }
+ mutex_unlock(&fw_mgmt->mutex);
+
+ return ret;
+}
+
+static const struct file_operations fw_mgmt_fops = {
+ .owner = THIS_MODULE,
+ .open = fw_mgmt_open,
+ .release = fw_mgmt_release,
+ .unlocked_ioctl = fw_mgmt_ioctl_unlocked,
+};
+
+int gb_fw_mgmt_request_handler(struct gb_operation *op)
+{
+ u8 type = op->type;
+
+ switch (type) {
+ case GB_FW_MGMT_TYPE_LOADED_FW:
+ return fw_mgmt_interface_fw_loaded_operation(op);
+ case GB_FW_MGMT_TYPE_BACKEND_FW_UPDATED:
+ return fw_mgmt_backend_fw_updated_operation(op);
+ default:
+ dev_err(&op->connection->bundle->dev,
+ "unsupported request: %u\n", type);
+ return -EINVAL;
+ }
+}
+
+int gb_fw_mgmt_connection_init(struct gb_connection *connection)
+{
+ struct fw_mgmt *fw_mgmt;
+ int ret, minor;
+
+ if (!connection)
+ return 0;
+
+ fw_mgmt = kzalloc(sizeof(*fw_mgmt), GFP_KERNEL);
+ if (!fw_mgmt)
+ return -ENOMEM;
+
+ fw_mgmt->parent = &connection->bundle->dev;
+ fw_mgmt->timeout_jiffies = msecs_to_jiffies(FW_MGMT_TIMEOUT_MS);
+ fw_mgmt->connection = connection;
+
+ gb_connection_set_data(connection, fw_mgmt);
+ init_completion(&fw_mgmt->completion);
+ ida_init(&fw_mgmt->id_map);
+ mutex_init(&fw_mgmt->mutex);
+ kref_init(&fw_mgmt->kref);
+
+ mutex_lock(&list_mutex);
+ list_add(&fw_mgmt->node, &fw_mgmt_list);
+ mutex_unlock(&list_mutex);
+
+ ret = gb_connection_enable(connection);
+ if (ret)
+ goto err_list_del;
+
+ minor = ida_simple_get(&fw_mgmt_minors_map, 0, NUM_MINORS, GFP_KERNEL);
+ if (minor < 0) {
+ ret = minor;
+ goto err_connection_disable;
+ }
+
+ /* Add a char device to allow userspace to interact with fw-mgmt */
+ fw_mgmt->dev_num = MKDEV(MAJOR(fw_mgmt_dev_num), minor);
+ cdev_init(&fw_mgmt->cdev, &fw_mgmt_fops);
+
+ ret = cdev_add(&fw_mgmt->cdev, fw_mgmt->dev_num, 1);
+ if (ret)
+ goto err_remove_ida;
+
+ /* Add a soft link to the previously added char-dev within the bundle */
+ fw_mgmt->class_device = device_create(fw_mgmt_class, fw_mgmt->parent,
+ fw_mgmt->dev_num, NULL,
+ "gb-fw-mgmt-%d", minor);
+ if (IS_ERR(fw_mgmt->class_device)) {
+ ret = PTR_ERR(fw_mgmt->class_device);
+ goto err_del_cdev;
+ }
+
+ return 0;
+
+err_del_cdev:
+ cdev_del(&fw_mgmt->cdev);
+err_remove_ida:
+ ida_simple_remove(&fw_mgmt_minors_map, minor);
+err_connection_disable:
+ gb_connection_disable(connection);
+err_list_del:
+ mutex_lock(&list_mutex);
+ list_del(&fw_mgmt->node);
+ mutex_unlock(&list_mutex);
+
+ put_fw_mgmt(fw_mgmt);
+
+ return ret;
+}
+
+void gb_fw_mgmt_connection_exit(struct gb_connection *connection)
+{
+ struct fw_mgmt *fw_mgmt;
+
+ if (!connection)
+ return;
+
+ fw_mgmt = gb_connection_get_data(connection);
+
+ device_destroy(fw_mgmt_class, fw_mgmt->dev_num);
+ cdev_del(&fw_mgmt->cdev);
+ ida_simple_remove(&fw_mgmt_minors_map, MINOR(fw_mgmt->dev_num));
+
+ /*
+ * Disallow any new ioctl operations on the char device and wait for
+ * existing ones to finish.
+ */
+ mutex_lock(&fw_mgmt->mutex);
+ fw_mgmt->disabled = true;
+ mutex_unlock(&fw_mgmt->mutex);
+
+ /* All pending greybus operations should have finished by now */
+ gb_connection_disable(fw_mgmt->connection);
+
+ /* Disallow new users to get access to the fw_mgmt structure */
+ mutex_lock(&list_mutex);
+ list_del(&fw_mgmt->node);
+ mutex_unlock(&list_mutex);
+
+ /*
+ * All current users of fw_mgmt would have taken a reference to it by
+ * now, we can drop our reference and wait the last user will get
+ * fw_mgmt freed.
+ */
+ put_fw_mgmt(fw_mgmt);
+}
+
+int fw_mgmt_init(void)
+{
+ int ret;
+
+ fw_mgmt_class = class_create(THIS_MODULE, "gb_fw_mgmt");
+ if (IS_ERR(fw_mgmt_class))
+ return PTR_ERR(fw_mgmt_class);
+
+ ret = alloc_chrdev_region(&fw_mgmt_dev_num, 0, NUM_MINORS,
+ "gb_fw_mgmt");
+ if (ret)
+ goto err_remove_class;
+
+ return 0;
+
+err_remove_class:
+ class_destroy(fw_mgmt_class);
+ return ret;
+}
+
+void fw_mgmt_exit(void)
+{
+ unregister_chrdev_region(fw_mgmt_dev_num, NUM_MINORS);
+ class_destroy(fw_mgmt_class);
+ ida_destroy(&fw_mgmt_minors_map);
+}
--- /dev/null
+++ b/drivers/greybus/greybus_firmware.h
@@ -0,0 +1,120 @@
+/*
+ * Greybus Firmware Management User Header
+ *
+ * This file is provided under a dual BSD/GPLv2 license. When using or
+ * redistributing this file, you may do so under either license.
+ *
+ * GPL LICENSE SUMMARY
+ *
+ * Copyright(c) 2016 Google Inc. All rights reserved.
+ * Copyright(c) 2016 Linaro Ltd. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License version 2 for more details.
+ *
+ * BSD LICENSE
+ *
+ * Copyright(c) 2016 Google Inc. All rights reserved.
+ * Copyright(c) 2016 Linaro Ltd. All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of Google Inc. or Linaro Ltd. nor the names of
+ * its contributors may be used to endorse or promote products
+ * derived from this software without specific prior written
+ * permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL GOOGLE INC. OR
+ * LINARO LTD. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+ * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef __GREYBUS_FIRMWARE_USER_H
+#define __GREYBUS_FIRMWARE_USER_H
+
+#include <linux/ioctl.h>
+#include <linux/types.h>
+
+#define GB_FIRMWARE_U_TAG_MAX_SIZE 10
+
+#define GB_FW_U_LOAD_METHOD_UNIPRO 0x01
+#define GB_FW_U_LOAD_METHOD_INTERNAL 0x02
+
+#define GB_FW_U_LOAD_STATUS_FAILED 0x00
+#define GB_FW_U_LOAD_STATUS_UNVALIDATED 0x01
+#define GB_FW_U_LOAD_STATUS_VALIDATED 0x02
+#define GB_FW_U_LOAD_STATUS_VALIDATION_FAILED 0x03
+
+#define GB_FW_U_BACKEND_FW_STATUS_SUCCESS 0x01
+#define GB_FW_U_BACKEND_FW_STATUS_FAIL_FIND 0x02
+#define GB_FW_U_BACKEND_FW_STATUS_FAIL_FETCH 0x03
+#define GB_FW_U_BACKEND_FW_STATUS_FAIL_WRITE 0x04
+#define GB_FW_U_BACKEND_FW_STATUS_INT 0x05
+#define GB_FW_U_BACKEND_FW_STATUS_RETRY 0x06
+#define GB_FW_U_BACKEND_FW_STATUS_NOT_SUPPORTED 0x07
+
+#define GB_FW_U_BACKEND_VERSION_STATUS_SUCCESS 0x01
+#define GB_FW_U_BACKEND_VERSION_STATUS_NOT_AVAILABLE 0x02
+#define GB_FW_U_BACKEND_VERSION_STATUS_NOT_SUPPORTED 0x03
+#define GB_FW_U_BACKEND_VERSION_STATUS_RETRY 0x04
+#define GB_FW_U_BACKEND_VERSION_STATUS_FAIL_INT 0x05
+
+/* IOCTL support */
+struct fw_mgmt_ioc_get_intf_version {
+ __u8 firmware_tag[GB_FIRMWARE_U_TAG_MAX_SIZE];
+ __u16 major;
+ __u16 minor;
+} __attribute__ ((__packed__));
+
+struct fw_mgmt_ioc_get_backend_version {
+ __u8 firmware_tag[GB_FIRMWARE_U_TAG_MAX_SIZE];
+ __u16 major;
+ __u16 minor;
+ __u8 status;
+} __attribute__ ((__packed__));
+
+struct fw_mgmt_ioc_intf_load_and_validate {
+ __u8 firmware_tag[GB_FIRMWARE_U_TAG_MAX_SIZE];
+ __u8 load_method;
+ __u8 status;
+ __u16 major;
+ __u16 minor;
+} __attribute__ ((__packed__));
+
+struct fw_mgmt_ioc_backend_fw_update {
+ __u8 firmware_tag[GB_FIRMWARE_U_TAG_MAX_SIZE];
+ __u8 status;
+} __attribute__ ((__packed__));
+
+#define FW_MGMT_IOCTL_BASE 'F'
+#define FW_MGMT_IOC_GET_INTF_FW _IOR(FW_MGMT_IOCTL_BASE, 0, struct fw_mgmt_ioc_get_intf_version)
+#define FW_MGMT_IOC_GET_BACKEND_FW _IOWR(FW_MGMT_IOCTL_BASE, 1, struct fw_mgmt_ioc_get_backend_version)
+#define FW_MGMT_IOC_INTF_LOAD_AND_VALIDATE _IOWR(FW_MGMT_IOCTL_BASE, 2, struct fw_mgmt_ioc_intf_load_and_validate)
+#define FW_MGMT_IOC_INTF_BACKEND_FW_UPDATE _IOWR(FW_MGMT_IOCTL_BASE, 3, struct fw_mgmt_ioc_backend_fw_update)
+#define FW_MGMT_IOC_SET_TIMEOUT_MS _IOW(FW_MGMT_IOCTL_BASE, 4, unsigned int)
+#define FW_MGMT_IOC_MODE_SWITCH _IO(FW_MGMT_IOCTL_BASE, 5)
+
+#endif /* __GREYBUS_FIRMWARE_USER_H */
+
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 10/32] greybus: audio driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (7 preceding siblings ...)
2016-09-16 13:24 ` [patch 09/32] greybus: firmware download class driver Greg KH
@ 2016-09-16 13:24 ` Greg KH
2016-09-16 13:25 ` [patch 11/32] greybus: camera driver Greg KH
` (21 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 13:24 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus audio protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/audio_apbridgea.c | 207 ++++
drivers/greybus/audio_apbridgea.h | 156 +++
drivers/greybus/audio_codec.c | 1132 +++++++++++++++++++++++++
drivers/greybus/audio_codec.h | 283 ++++++
drivers/greybus/audio_gb.c | 228 +++++
drivers/greybus/audio_manager.c | 184 ++++
drivers/greybus/audio_manager.h | 83 +
drivers/greybus/audio_manager_module.c | 258 +++++
drivers/greybus/audio_manager_private.h | 28
drivers/greybus/audio_manager_sysfs.c | 102 ++
drivers/greybus/audio_module.c | 482 ++++++++++
drivers/greybus/audio_topology.c | 1442 ++++++++++++++++++++++++++++++++
12 files changed, 4585 insertions(+)
--- /dev/null
+++ b/drivers/greybus/audio_apbridgea.c
@@ -0,0 +1,207 @@
+/*
+ * Greybus Audio Device Class Protocol helpers
+ *
+ * Copyright 2015-2016 Google Inc.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include "greybus.h"
+#include "greybus_protocols.h"
+#include "audio_apbridgea.h"
+#include "audio_codec.h"
+
+int gb_audio_apbridgea_set_config(struct gb_connection *connection,
+ __u16 i2s_port, __u32 format, __u32 rate,
+ __u32 mclk_freq)
+{
+ struct audio_apbridgea_set_config_request req;
+
+ req.hdr.type = AUDIO_APBRIDGEA_TYPE_SET_CONFIG;
+ req.hdr.i2s_port = cpu_to_le16(i2s_port);
+ req.format = cpu_to_le32(format);
+ req.rate = cpu_to_le32(rate);
+ req.mclk_freq = cpu_to_le32(mclk_freq);
+
+ return gb_hd_output(connection->hd, &req, sizeof(req),
+ GB_APB_REQUEST_AUDIO_CONTROL, true);
+}
+EXPORT_SYMBOL_GPL(gb_audio_apbridgea_set_config);
+
+int gb_audio_apbridgea_register_cport(struct gb_connection *connection,
+ __u16 i2s_port, __u16 cportid,
+ __u8 direction)
+{
+ struct audio_apbridgea_register_cport_request req;
+ int ret;
+
+ req.hdr.type = AUDIO_APBRIDGEA_TYPE_REGISTER_CPORT;
+ req.hdr.i2s_port = cpu_to_le16(i2s_port);
+ req.cport = cpu_to_le16(cportid);
+ req.direction = direction;
+
+ ret = gb_pm_runtime_get_sync(connection->bundle);
+ if (ret)
+ return ret;
+
+ return gb_hd_output(connection->hd, &req, sizeof(req),
+ GB_APB_REQUEST_AUDIO_CONTROL, true);
+}
+EXPORT_SYMBOL_GPL(gb_audio_apbridgea_register_cport);
+
+int gb_audio_apbridgea_unregister_cport(struct gb_connection *connection,
+ __u16 i2s_port, __u16 cportid,
+ __u8 direction)
+{
+ struct audio_apbridgea_unregister_cport_request req;
+ int ret;
+
+ req.hdr.type = AUDIO_APBRIDGEA_TYPE_UNREGISTER_CPORT;
+ req.hdr.i2s_port = cpu_to_le16(i2s_port);
+ req.cport = cpu_to_le16(cportid);
+ req.direction = direction;
+
+ ret = gb_hd_output(connection->hd, &req, sizeof(req),
+ GB_APB_REQUEST_AUDIO_CONTROL, true);
+
+ gb_pm_runtime_put_autosuspend(connection->bundle);
+
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_audio_apbridgea_unregister_cport);
+
+int gb_audio_apbridgea_set_tx_data_size(struct gb_connection *connection,
+ __u16 i2s_port, __u16 size)
+{
+ struct audio_apbridgea_set_tx_data_size_request req;
+
+ req.hdr.type = AUDIO_APBRIDGEA_TYPE_SET_TX_DATA_SIZE;
+ req.hdr.i2s_port = cpu_to_le16(i2s_port);
+ req.size = cpu_to_le16(size);
+
+ return gb_hd_output(connection->hd, &req, sizeof(req),
+ GB_APB_REQUEST_AUDIO_CONTROL, true);
+}
+EXPORT_SYMBOL_GPL(gb_audio_apbridgea_set_tx_data_size);
+
+int gb_audio_apbridgea_prepare_tx(struct gb_connection *connection,
+ __u16 i2s_port)
+{
+ struct audio_apbridgea_prepare_tx_request req;
+
+ req.hdr.type = AUDIO_APBRIDGEA_TYPE_PREPARE_TX;
+ req.hdr.i2s_port = cpu_to_le16(i2s_port);
+
+ return gb_hd_output(connection->hd, &req, sizeof(req),
+ GB_APB_REQUEST_AUDIO_CONTROL, true);
+}
+EXPORT_SYMBOL_GPL(gb_audio_apbridgea_prepare_tx);
+
+int gb_audio_apbridgea_start_tx(struct gb_connection *connection,
+ __u16 i2s_port, __u64 timestamp)
+{
+ struct audio_apbridgea_start_tx_request req;
+
+ req.hdr.type = AUDIO_APBRIDGEA_TYPE_START_TX;
+ req.hdr.i2s_port = cpu_to_le16(i2s_port);
+ req.timestamp = cpu_to_le64(timestamp);
+
+ return gb_hd_output(connection->hd, &req, sizeof(req),
+ GB_APB_REQUEST_AUDIO_CONTROL, true);
+}
+EXPORT_SYMBOL_GPL(gb_audio_apbridgea_start_tx);
+
+int gb_audio_apbridgea_stop_tx(struct gb_connection *connection, __u16 i2s_port)
+{
+ struct audio_apbridgea_stop_tx_request req;
+
+ req.hdr.type = AUDIO_APBRIDGEA_TYPE_STOP_TX;
+ req.hdr.i2s_port = cpu_to_le16(i2s_port);
+
+ return gb_hd_output(connection->hd, &req, sizeof(req),
+ GB_APB_REQUEST_AUDIO_CONTROL, true);
+}
+EXPORT_SYMBOL_GPL(gb_audio_apbridgea_stop_tx);
+
+int gb_audio_apbridgea_shutdown_tx(struct gb_connection *connection,
+ __u16 i2s_port)
+{
+ struct audio_apbridgea_shutdown_tx_request req;
+
+ req.hdr.type = AUDIO_APBRIDGEA_TYPE_SHUTDOWN_TX;
+ req.hdr.i2s_port = cpu_to_le16(i2s_port);
+
+ return gb_hd_output(connection->hd, &req, sizeof(req),
+ GB_APB_REQUEST_AUDIO_CONTROL, true);
+}
+EXPORT_SYMBOL_GPL(gb_audio_apbridgea_shutdown_tx);
+
+int gb_audio_apbridgea_set_rx_data_size(struct gb_connection *connection,
+ __u16 i2s_port, __u16 size)
+{
+ struct audio_apbridgea_set_rx_data_size_request req;
+
+ req.hdr.type = AUDIO_APBRIDGEA_TYPE_SET_RX_DATA_SIZE;
+ req.hdr.i2s_port = cpu_to_le16(i2s_port);
+ req.size = cpu_to_le16(size);
+
+ return gb_hd_output(connection->hd, &req, sizeof(req),
+ GB_APB_REQUEST_AUDIO_CONTROL, true);
+}
+EXPORT_SYMBOL_GPL(gb_audio_apbridgea_set_rx_data_size);
+
+int gb_audio_apbridgea_prepare_rx(struct gb_connection *connection,
+ __u16 i2s_port)
+{
+ struct audio_apbridgea_prepare_rx_request req;
+
+ req.hdr.type = AUDIO_APBRIDGEA_TYPE_PREPARE_RX;
+ req.hdr.i2s_port = cpu_to_le16(i2s_port);
+
+ return gb_hd_output(connection->hd, &req, sizeof(req),
+ GB_APB_REQUEST_AUDIO_CONTROL, true);
+}
+EXPORT_SYMBOL_GPL(gb_audio_apbridgea_prepare_rx);
+
+int gb_audio_apbridgea_start_rx(struct gb_connection *connection,
+ __u16 i2s_port)
+{
+ struct audio_apbridgea_start_rx_request req;
+
+ req.hdr.type = AUDIO_APBRIDGEA_TYPE_START_RX;
+ req.hdr.i2s_port = cpu_to_le16(i2s_port);
+
+ return gb_hd_output(connection->hd, &req, sizeof(req),
+ GB_APB_REQUEST_AUDIO_CONTROL, true);
+}
+EXPORT_SYMBOL_GPL(gb_audio_apbridgea_start_rx);
+
+int gb_audio_apbridgea_stop_rx(struct gb_connection *connection, __u16 i2s_port)
+{
+ struct audio_apbridgea_stop_rx_request req;
+
+ req.hdr.type = AUDIO_APBRIDGEA_TYPE_STOP_RX;
+ req.hdr.i2s_port = cpu_to_le16(i2s_port);
+
+ return gb_hd_output(connection->hd, &req, sizeof(req),
+ GB_APB_REQUEST_AUDIO_CONTROL, true);
+}
+EXPORT_SYMBOL_GPL(gb_audio_apbridgea_stop_rx);
+
+int gb_audio_apbridgea_shutdown_rx(struct gb_connection *connection,
+ __u16 i2s_port)
+{
+ struct audio_apbridgea_shutdown_rx_request req;
+
+ req.hdr.type = AUDIO_APBRIDGEA_TYPE_SHUTDOWN_RX;
+ req.hdr.i2s_port = cpu_to_le16(i2s_port);
+
+ return gb_hd_output(connection->hd, &req, sizeof(req),
+ GB_APB_REQUEST_AUDIO_CONTROL, true);
+}
+EXPORT_SYMBOL_GPL(gb_audio_apbridgea_shutdown_rx);
+
+MODULE_LICENSE("GPL v2");
+MODULE_ALIAS("greybus:audio-apbridgea");
+MODULE_DESCRIPTION("Greybus Special APBridgeA Audio Protocol library");
+MODULE_AUTHOR("Mark Greer <mgreer@animalcreek.com>");
--- /dev/null
+++ b/drivers/greybus/audio_apbridgea.h
@@ -0,0 +1,156 @@
+/**
+ * Copyright (c) 2015-2016 Google Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials provided with the distribution.
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from this
+ * software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+/*
+ * This is a special protocol for configuring communication over the
+ * I2S bus between the DSP on the MSM8994 and APBridgeA. Therefore,
+ * we can predefine several low-level attributes of the communication
+ * because we know that they are supported. In particular, the following
+ * assumptions are made:
+ * - there are two channels (i.e., stereo)
+ * - the low-level protocol is I2S as defined by Philips/NXP
+ * - the DSP on the MSM8994 is the clock master for MCLK, BCLK, and WCLK
+ * - WCLK changes on the falling edge of BCLK
+ * - WCLK low for left channel; high for right channel
+ * - TX data is sent on the falling edge of BCLK
+ * - RX data is received/latched on the rising edge of BCLK
+ */
+
+#ifndef __AUDIO_APBRIDGEA_H
+#define __AUDIO_APBRIDGEA_H
+
+#define AUDIO_APBRIDGEA_TYPE_SET_CONFIG 0x01
+#define AUDIO_APBRIDGEA_TYPE_REGISTER_CPORT 0x02
+#define AUDIO_APBRIDGEA_TYPE_UNREGISTER_CPORT 0x03
+#define AUDIO_APBRIDGEA_TYPE_SET_TX_DATA_SIZE 0x04
+ /* 0x05 unused */
+#define AUDIO_APBRIDGEA_TYPE_PREPARE_TX 0x06
+#define AUDIO_APBRIDGEA_TYPE_START_TX 0x07
+#define AUDIO_APBRIDGEA_TYPE_STOP_TX 0x08
+#define AUDIO_APBRIDGEA_TYPE_SHUTDOWN_TX 0x09
+#define AUDIO_APBRIDGEA_TYPE_SET_RX_DATA_SIZE 0x0a
+ /* 0x0b unused */
+#define AUDIO_APBRIDGEA_TYPE_PREPARE_RX 0x0c
+#define AUDIO_APBRIDGEA_TYPE_START_RX 0x0d
+#define AUDIO_APBRIDGEA_TYPE_STOP_RX 0x0e
+#define AUDIO_APBRIDGEA_TYPE_SHUTDOWN_RX 0x0f
+
+#define AUDIO_APBRIDGEA_PCM_FMT_8 BIT(0)
+#define AUDIO_APBRIDGEA_PCM_FMT_16 BIT(1)
+#define AUDIO_APBRIDGEA_PCM_FMT_24 BIT(2)
+#define AUDIO_APBRIDGEA_PCM_FMT_32 BIT(3)
+#define AUDIO_APBRIDGEA_PCM_FMT_64 BIT(4)
+
+#define AUDIO_APBRIDGEA_PCM_RATE_5512 BIT(0)
+#define AUDIO_APBRIDGEA_PCM_RATE_8000 BIT(1)
+#define AUDIO_APBRIDGEA_PCM_RATE_11025 BIT(2)
+#define AUDIO_APBRIDGEA_PCM_RATE_16000 BIT(3)
+#define AUDIO_APBRIDGEA_PCM_RATE_22050 BIT(4)
+#define AUDIO_APBRIDGEA_PCM_RATE_32000 BIT(5)
+#define AUDIO_APBRIDGEA_PCM_RATE_44100 BIT(6)
+#define AUDIO_APBRIDGEA_PCM_RATE_48000 BIT(7)
+#define AUDIO_APBRIDGEA_PCM_RATE_64000 BIT(8)
+#define AUDIO_APBRIDGEA_PCM_RATE_88200 BIT(9)
+#define AUDIO_APBRIDGEA_PCM_RATE_96000 BIT(10)
+#define AUDIO_APBRIDGEA_PCM_RATE_176400 BIT(11)
+#define AUDIO_APBRIDGEA_PCM_RATE_192000 BIT(12)
+
+#define AUDIO_APBRIDGEA_DIRECTION_TX BIT(0)
+#define AUDIO_APBRIDGEA_DIRECTION_RX BIT(1)
+
+/* The I2S port is passed in the 'index' parameter of the USB request */
+/* The CPort is passed in the 'value' parameter of the USB request */
+
+struct audio_apbridgea_hdr {
+ __u8 type;
+ __le16 i2s_port;
+ __u8 data[0];
+} __packed;
+
+struct audio_apbridgea_set_config_request {
+ struct audio_apbridgea_hdr hdr;
+ __le32 format; /* AUDIO_APBRIDGEA_PCM_FMT_* */
+ __le32 rate; /* AUDIO_APBRIDGEA_PCM_RATE_* */
+ __le32 mclk_freq; /* XXX Remove? */
+} __packed;
+
+struct audio_apbridgea_register_cport_request {
+ struct audio_apbridgea_hdr hdr;
+ __le16 cport;
+ __u8 direction;
+} __packed;
+
+struct audio_apbridgea_unregister_cport_request {
+ struct audio_apbridgea_hdr hdr;
+ __le16 cport;
+ __u8 direction;
+} __packed;
+
+struct audio_apbridgea_set_tx_data_size_request {
+ struct audio_apbridgea_hdr hdr;
+ __le16 size;
+} __packed;
+
+struct audio_apbridgea_prepare_tx_request {
+ struct audio_apbridgea_hdr hdr;
+} __packed;
+
+struct audio_apbridgea_start_tx_request {
+ struct audio_apbridgea_hdr hdr;
+ __le64 timestamp;
+} __packed;
+
+struct audio_apbridgea_stop_tx_request {
+ struct audio_apbridgea_hdr hdr;
+} __packed;
+
+struct audio_apbridgea_shutdown_tx_request {
+ struct audio_apbridgea_hdr hdr;
+} __packed;
+
+struct audio_apbridgea_set_rx_data_size_request {
+ struct audio_apbridgea_hdr hdr;
+ __le16 size;
+} __packed;
+
+struct audio_apbridgea_prepare_rx_request {
+ struct audio_apbridgea_hdr hdr;
+} __packed;
+
+struct audio_apbridgea_start_rx_request {
+ struct audio_apbridgea_hdr hdr;
+} __packed;
+
+struct audio_apbridgea_stop_rx_request {
+ struct audio_apbridgea_hdr hdr;
+} __packed;
+
+struct audio_apbridgea_shutdown_rx_request {
+ struct audio_apbridgea_hdr hdr;
+} __packed;
+
+#endif /*__AUDIO_APBRIDGEA_H */
--- /dev/null
+++ b/drivers/greybus/audio_codec.c
@@ -0,0 +1,1132 @@
+/*
+ * APBridge ALSA SoC dummy codec driver
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/pm_runtime.h>
+#include <sound/soc.h>
+#include <sound/pcm_params.h>
+#include <uapi/linux/input.h>
+
+#include "audio_codec.h"
+#include "audio_apbridgea.h"
+#include "audio_manager.h"
+
+static struct gbaudio_codec_info *gbcodec;
+
+static struct gbaudio_data_connection *
+find_data(struct gbaudio_module_info *module, int id)
+{
+ struct gbaudio_data_connection *data;
+
+ list_for_each_entry(data, &module->data_list, list) {
+ if (id == data->id)
+ return data;
+ }
+ return NULL;
+}
+
+static struct gbaudio_stream_params *
+find_dai_stream_params(struct gbaudio_codec_info *codec, int id, int stream)
+{
+ struct gbaudio_codec_dai *dai;
+
+ list_for_each_entry(dai, &codec->dai_list, list) {
+ if (dai->id == id)
+ return &dai->params[stream];
+ }
+ return NULL;
+}
+
+static int gbaudio_module_enable_tx(struct gbaudio_codec_info *codec,
+ struct gbaudio_module_info *module, int id)
+{
+ int module_state, ret = 0;
+ uint16_t data_cport, i2s_port, cportid;
+ uint8_t sig_bits, channels;
+ uint32_t format, rate;
+ struct gbaudio_data_connection *data;
+ struct gbaudio_stream_params *params;
+
+ /* find the dai */
+ data = find_data(module, id);
+ if (!data) {
+ dev_err(module->dev, "%d:DATA connection missing\n", id);
+ return -ENODEV;
+ }
+ module_state = data->state[SNDRV_PCM_STREAM_PLAYBACK];
+
+ params = find_dai_stream_params(codec, id, SNDRV_PCM_STREAM_PLAYBACK);
+ if (!params) {
+ dev_err(codec->dev, "Failed to fetch dai_stream pointer\n");
+ return -EINVAL;
+ }
+
+ /* register cport */
+ if (module_state < GBAUDIO_CODEC_STARTUP) {
+ i2s_port = 0; /* fixed for now */
+ cportid = data->connection->hd_cport_id;
+ ret = gb_audio_apbridgea_register_cport(data->connection,
+ i2s_port, cportid,
+ AUDIO_APBRIDGEA_DIRECTION_TX);
+ if (ret) {
+ dev_err_ratelimited(module->dev,
+ "reg_cport failed:%d\n", ret);
+ return ret;
+ }
+ data->state[SNDRV_PCM_STREAM_PLAYBACK] =
+ GBAUDIO_CODEC_STARTUP;
+ dev_dbg(module->dev, "Dynamic Register %d DAI\n", cportid);
+ }
+
+ /* hw_params */
+ if (module_state < GBAUDIO_CODEC_HWPARAMS) {
+ format = params->format;
+ channels = params->channels;
+ rate = params->rate;
+ sig_bits = params->sig_bits;
+ data_cport = data->connection->intf_cport_id;
+ ret = gb_audio_gb_set_pcm(module->mgmt_connection, data_cport,
+ format, rate, channels, sig_bits);
+ if (ret) {
+ dev_err_ratelimited(module->dev, "set_pcm failed:%d\n",
+ ret);
+ return ret;
+ }
+ data->state[SNDRV_PCM_STREAM_PLAYBACK] =
+ GBAUDIO_CODEC_HWPARAMS;
+ dev_dbg(module->dev, "Dynamic hw_params %d DAI\n", data_cport);
+ }
+
+ /* prepare */
+ if (module_state < GBAUDIO_CODEC_PREPARE) {
+ data_cport = data->connection->intf_cport_id;
+ ret = gb_audio_gb_set_tx_data_size(module->mgmt_connection,
+ data_cport, 192);
+ if (ret) {
+ dev_err_ratelimited(module->dev,
+ "set_tx_data_size failed:%d\n",
+ ret);
+ return ret;
+ }
+ ret = gb_audio_gb_activate_tx(module->mgmt_connection,
+ data_cport);
+ if (ret) {
+ dev_err_ratelimited(module->dev,
+ "activate_tx failed:%d\n", ret);
+ return ret;
+ }
+ data->state[SNDRV_PCM_STREAM_PLAYBACK] =
+ GBAUDIO_CODEC_PREPARE;
+ dev_dbg(module->dev, "Dynamic prepare %d DAI\n", data_cport);
+ }
+
+ return 0;
+}
+
+static int gbaudio_module_disable_tx(struct gbaudio_module_info *module, int id)
+{
+ int ret;
+ uint16_t data_cport, cportid, i2s_port;
+ int module_state;
+ struct gbaudio_data_connection *data;
+
+ /* find the dai */
+ data = find_data(module, id);
+ if (!data) {
+ dev_err(module->dev, "%d:DATA connection missing\n", id);
+ return -ENODEV;
+ }
+ module_state = data->state[SNDRV_PCM_STREAM_PLAYBACK];
+
+ if (module_state > GBAUDIO_CODEC_HWPARAMS) {
+ data_cport = data->connection->intf_cport_id;
+ ret = gb_audio_gb_deactivate_tx(module->mgmt_connection,
+ data_cport);
+ if (ret) {
+ dev_err_ratelimited(module->dev,
+ "deactivate_tx failed:%d\n", ret);
+ return ret;
+ }
+ dev_dbg(module->dev, "Dynamic deactivate %d DAI\n", data_cport);
+ data->state[SNDRV_PCM_STREAM_PLAYBACK] =
+ GBAUDIO_CODEC_HWPARAMS;
+ }
+
+ if (module_state > GBAUDIO_CODEC_SHUTDOWN) {
+ i2s_port = 0; /* fixed for now */
+ cportid = data->connection->hd_cport_id;
+ ret = gb_audio_apbridgea_unregister_cport(data->connection,
+ i2s_port, cportid,
+ AUDIO_APBRIDGEA_DIRECTION_TX);
+ if (ret) {
+ dev_err_ratelimited(module->dev,
+ "unregister_cport failed:%d\n",
+ ret);
+ return ret;
+ }
+ dev_dbg(module->dev, "Dynamic Unregister %d DAI\n", cportid);
+ data->state[SNDRV_PCM_STREAM_PLAYBACK] =
+ GBAUDIO_CODEC_SHUTDOWN;
+ }
+
+ return 0;
+}
+
+static int gbaudio_module_enable_rx(struct gbaudio_codec_info *codec,
+ struct gbaudio_module_info *module, int id)
+{
+ int module_state, ret = 0;
+ uint16_t data_cport, i2s_port, cportid;
+ uint8_t sig_bits, channels;
+ uint32_t format, rate;
+ struct gbaudio_data_connection *data;
+ struct gbaudio_stream_params *params;
+
+ /* find the dai */
+ data = find_data(module, id);
+ if (!data) {
+ dev_err(module->dev, "%d:DATA connection missing\n", id);
+ return -ENODEV;
+ }
+ module_state = data->state[SNDRV_PCM_STREAM_CAPTURE];
+
+ params = find_dai_stream_params(codec, id, SNDRV_PCM_STREAM_CAPTURE);
+ if (!params) {
+ dev_err(codec->dev, "Failed to fetch dai_stream pointer\n");
+ return -EINVAL;
+ }
+
+ /* register cport */
+ if (module_state < GBAUDIO_CODEC_STARTUP) {
+ i2s_port = 0; /* fixed for now */
+ cportid = data->connection->hd_cport_id;
+ ret = gb_audio_apbridgea_register_cport(data->connection,
+ i2s_port, cportid,
+ AUDIO_APBRIDGEA_DIRECTION_RX);
+ if (ret) {
+ dev_err_ratelimited(module->dev,
+ "reg_cport failed:%d\n", ret);
+ return ret;
+ }
+ data->state[SNDRV_PCM_STREAM_CAPTURE] =
+ GBAUDIO_CODEC_STARTUP;
+ dev_dbg(module->dev, "Dynamic Register %d DAI\n", cportid);
+ }
+
+ /* hw_params */
+ if (module_state < GBAUDIO_CODEC_HWPARAMS) {
+ format = params->format;
+ channels = params->channels;
+ rate = params->rate;
+ sig_bits = params->sig_bits;
+ data_cport = data->connection->intf_cport_id;
+ ret = gb_audio_gb_set_pcm(module->mgmt_connection, data_cport,
+ format, rate, channels, sig_bits);
+ if (ret) {
+ dev_err_ratelimited(module->dev, "set_pcm failed:%d\n",
+ ret);
+ return ret;
+ }
+ data->state[SNDRV_PCM_STREAM_CAPTURE] =
+ GBAUDIO_CODEC_HWPARAMS;
+ dev_dbg(module->dev, "Dynamic hw_params %d DAI\n", data_cport);
+ }
+
+ /* prepare */
+ if (module_state < GBAUDIO_CODEC_PREPARE) {
+ data_cport = data->connection->intf_cport_id;
+ ret = gb_audio_gb_set_rx_data_size(module->mgmt_connection,
+ data_cport, 192);
+ if (ret) {
+ dev_err_ratelimited(module->dev,
+ "set_rx_data_size failed:%d\n",
+ ret);
+ return ret;
+ }
+ ret = gb_audio_gb_activate_rx(module->mgmt_connection,
+ data_cport);
+ if (ret) {
+ dev_err_ratelimited(module->dev,
+ "activate_rx failed:%d\n", ret);
+ return ret;
+ }
+ data->state[SNDRV_PCM_STREAM_CAPTURE] =
+ GBAUDIO_CODEC_PREPARE;
+ dev_dbg(module->dev, "Dynamic prepare %d DAI\n", data_cport);
+ }
+
+ return 0;
+}
+
+static int gbaudio_module_disable_rx(struct gbaudio_module_info *module, int id)
+{
+ int ret;
+ uint16_t data_cport, cportid, i2s_port;
+ int module_state;
+ struct gbaudio_data_connection *data;
+
+ /* find the dai */
+ data = find_data(module, id);
+ if (!data) {
+ dev_err(module->dev, "%d:DATA connection missing\n", id);
+ return -ENODEV;
+ }
+ module_state = data->state[SNDRV_PCM_STREAM_CAPTURE];
+
+ if (module_state > GBAUDIO_CODEC_HWPARAMS) {
+ data_cport = data->connection->intf_cport_id;
+ ret = gb_audio_gb_deactivate_rx(module->mgmt_connection,
+ data_cport);
+ if (ret) {
+ dev_err_ratelimited(module->dev,
+ "deactivate_rx failed:%d\n", ret);
+ return ret;
+ }
+ dev_dbg(module->dev, "Dynamic deactivate %d DAI\n", data_cport);
+ data->state[SNDRV_PCM_STREAM_CAPTURE] =
+ GBAUDIO_CODEC_HWPARAMS;
+ }
+
+ if (module_state > GBAUDIO_CODEC_SHUTDOWN) {
+ i2s_port = 0; /* fixed for now */
+ cportid = data->connection->hd_cport_id;
+ ret = gb_audio_apbridgea_unregister_cport(data->connection,
+ i2s_port, cportid,
+ AUDIO_APBRIDGEA_DIRECTION_RX);
+ if (ret) {
+ dev_err_ratelimited(module->dev,
+ "unregister_cport failed:%d\n",
+ ret);
+ return ret;
+ }
+ dev_dbg(module->dev, "Dynamic Unregister %d DAI\n", cportid);
+ data->state[SNDRV_PCM_STREAM_CAPTURE] =
+ GBAUDIO_CODEC_SHUTDOWN;
+ }
+
+ return 0;
+}
+
+int gbaudio_module_update(struct gbaudio_codec_info *codec,
+ struct snd_soc_dapm_widget *w,
+ struct gbaudio_module_info *module, int enable)
+{
+ int dai_id, ret;
+ char intf_name[NAME_SIZE], dir[NAME_SIZE];
+
+ dev_dbg(module->dev, "%s:Module update %s sequence\n", w->name,
+ enable ? "Enable":"Disable");
+
+ if ((w->id != snd_soc_dapm_aif_in) && (w->id != snd_soc_dapm_aif_out)){
+ dev_dbg(codec->dev, "No action required for %s\n", w->name);
+ return 0;
+ }
+
+ /* parse dai_id from AIF widget's stream_name */
+ ret = sscanf(w->sname, "%s %d %s", intf_name, &dai_id, dir);
+ if (ret < 3) {
+ dev_err(codec->dev, "Error while parsing dai_id for %s\n",
+ w->name);
+ return -EINVAL;
+ }
+
+ mutex_lock(&codec->lock);
+ if (w->id == snd_soc_dapm_aif_in) {
+ if (enable)
+ ret = gbaudio_module_enable_tx(codec, module, dai_id);
+ else
+ ret = gbaudio_module_disable_tx(module, dai_id);
+ } else if (w->id == snd_soc_dapm_aif_out) {
+ if (enable)
+ ret = gbaudio_module_enable_rx(codec, module, dai_id);
+ else
+ ret = gbaudio_module_disable_rx(module, dai_id);
+ }
+
+ mutex_unlock(&codec->lock);
+
+ return ret;
+}
+EXPORT_SYMBOL(gbaudio_module_update);
+
+/*
+ * codec DAI ops
+ */
+static int gbcodec_startup(struct snd_pcm_substream *substream,
+ struct snd_soc_dai *dai)
+{
+ struct gbaudio_codec_info *codec = dev_get_drvdata(dai->dev);
+ struct gbaudio_stream_params *params;
+
+ mutex_lock(&codec->lock);
+
+ if (list_empty(&codec->module_list)) {
+ dev_err(codec->dev, "No codec module available\n");
+ mutex_unlock(&codec->lock);
+ return -ENODEV;
+ }
+
+ params = find_dai_stream_params(codec, dai->id, substream->stream);
+ if (!params) {
+ dev_err(codec->dev, "Failed to fetch dai_stream pointer\n");
+ mutex_unlock(&codec->lock);
+ return -EINVAL;
+ }
+ params->state = GBAUDIO_CODEC_STARTUP;
+ mutex_unlock(&codec->lock);
+ /* to prevent suspend in case of active audio */
+ pm_stay_awake(dai->dev);
+
+ return 0;
+}
+
+static void gbcodec_shutdown(struct snd_pcm_substream *substream,
+ struct snd_soc_dai *dai)
+{
+ struct gbaudio_codec_info *codec = dev_get_drvdata(dai->dev);
+ struct gbaudio_stream_params *params;
+
+ mutex_lock(&codec->lock);
+
+ if (list_empty(&codec->module_list))
+ dev_info(codec->dev, "No codec module available during shutdown\n");
+
+ params = find_dai_stream_params(codec, dai->id, substream->stream);
+ if (!params) {
+ dev_err(codec->dev, "Failed to fetch dai_stream pointer\n");
+ mutex_unlock(&codec->lock);
+ return;
+ }
+ params->state = GBAUDIO_CODEC_SHUTDOWN;
+ mutex_unlock(&codec->lock);
+ pm_relax(dai->dev);
+ return;
+}
+
+static int gbcodec_hw_params(struct snd_pcm_substream *substream,
+ struct snd_pcm_hw_params *hwparams,
+ struct snd_soc_dai *dai)
+{
+ int ret;
+ uint8_t sig_bits, channels;
+ uint32_t format, rate;
+ struct gbaudio_module_info *module;
+ struct gbaudio_data_connection *data;
+ struct gb_bundle *bundle;
+ struct gbaudio_codec_info *codec = dev_get_drvdata(dai->dev);
+ struct gbaudio_stream_params *params;
+
+ mutex_lock(&codec->lock);
+
+ if (list_empty(&codec->module_list)) {
+ dev_err(codec->dev, "No codec module available\n");
+ mutex_unlock(&codec->lock);
+ return -ENODEV;
+ }
+
+ /*
+ * assuming, currently only 48000 Hz, 16BIT_LE, stereo
+ * is supported, validate params before configuring codec
+ */
+ if (params_channels(hwparams) != 2) {
+ dev_err(dai->dev, "Invalid channel count:%d\n",
+ params_channels(hwparams));
+ mutex_unlock(&codec->lock);
+ return -EINVAL;
+ }
+ channels = params_channels(hwparams);
+
+ if (params_rate(hwparams) != 48000) {
+ dev_err(dai->dev, "Invalid sampling rate:%d\n",
+ params_rate(hwparams));
+ mutex_unlock(&codec->lock);
+ return -EINVAL;
+ }
+ rate = GB_AUDIO_PCM_RATE_48000;
+
+ if (params_format(hwparams) != SNDRV_PCM_FORMAT_S16_LE) {
+ dev_err(dai->dev, "Invalid format:%d\n",
+ params_format(hwparams));
+ mutex_unlock(&codec->lock);
+ return -EINVAL;
+ }
+ format = GB_AUDIO_PCM_FMT_S16_LE;
+
+ /* find the data connection */
+ list_for_each_entry(module, &codec->module_list, list) {
+ data = find_data(module, dai->id);
+ if (data)
+ break;
+ }
+
+ if (!data) {
+ dev_err(dai->dev, "DATA connection missing\n");
+ mutex_unlock(&codec->lock);
+ return -EINVAL;
+ }
+
+ params = find_dai_stream_params(codec, dai->id, substream->stream);
+ if (!params) {
+ dev_err(codec->dev, "Failed to fetch dai_stream pointer\n");
+ mutex_unlock(&codec->lock);
+ return -EINVAL;
+ }
+
+ bundle = to_gb_bundle(module->dev);
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret) {
+ mutex_unlock(&codec->lock);
+ return ret;
+ }
+
+ ret = gb_audio_apbridgea_set_config(data->connection, 0,
+ AUDIO_APBRIDGEA_PCM_FMT_16,
+ AUDIO_APBRIDGEA_PCM_RATE_48000,
+ 6144000);
+ if (ret) {
+ dev_err_ratelimited(dai->dev, "%d: Error during set_config\n",
+ ret);
+ mutex_unlock(&codec->lock);
+ return ret;
+ }
+
+ gb_pm_runtime_put_noidle(bundle);
+
+ params->state = GBAUDIO_CODEC_HWPARAMS;
+ params->format = format;
+ params->rate = rate;
+ params->channels = channels;
+ params->sig_bits = sig_bits;
+
+ mutex_unlock(&codec->lock);
+ return 0;
+}
+
+static int gbcodec_prepare(struct snd_pcm_substream *substream,
+ struct snd_soc_dai *dai)
+{
+ int ret;
+ struct gbaudio_module_info *module;
+ struct gbaudio_data_connection *data;
+ struct gb_bundle *bundle;
+ struct gbaudio_codec_info *codec = dev_get_drvdata(dai->dev);
+ struct gbaudio_stream_params *params;
+
+ mutex_lock(&codec->lock);
+
+ if (list_empty(&codec->module_list)) {
+ dev_err(codec->dev, "No codec module available\n");
+ mutex_unlock(&codec->lock);
+ return -ENODEV;
+ }
+
+ list_for_each_entry(module, &codec->module_list, list) {
+ /* find the dai */
+ data = find_data(module, dai->id);
+ if (data)
+ break;
+ }
+ if (!data) {
+ dev_err(dai->dev, "DATA connection missing\n");
+ mutex_unlock(&codec->lock);
+ return -ENODEV;
+ }
+
+ params = find_dai_stream_params(codec, dai->id, substream->stream);
+ if (!params) {
+ dev_err(codec->dev, "Failed to fetch dai_stream pointer\n");
+ mutex_unlock(&codec->lock);
+ return -EINVAL;
+ }
+
+ bundle = to_gb_bundle(module->dev);
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret) {
+ mutex_unlock(&codec->lock);
+ return ret;
+ }
+
+ switch (substream->stream) {
+ case SNDRV_PCM_STREAM_PLAYBACK:
+ ret = gb_audio_apbridgea_set_tx_data_size(data->connection, 0,
+ 192);
+ break;
+ case SNDRV_PCM_STREAM_CAPTURE:
+ ret = gb_audio_apbridgea_set_rx_data_size(data->connection, 0,
+ 192);
+ break;
+ }
+ if (ret) {
+ mutex_unlock(&codec->lock);
+ dev_err_ratelimited(dai->dev, "set_data_size failed:%d\n",
+ ret);
+ return ret;
+ }
+
+ gb_pm_runtime_put_noidle(bundle);
+
+ params->state = GBAUDIO_CODEC_PREPARE;
+ mutex_unlock(&codec->lock);
+ return 0;
+}
+
+static int gbcodec_mute_stream(struct snd_soc_dai *dai, int mute, int stream)
+{
+ int ret;
+ struct gbaudio_data_connection *data;
+ struct gbaudio_module_info *module;
+ struct gb_bundle *bundle;
+ struct gbaudio_codec_info *codec = dev_get_drvdata(dai->dev);
+ struct gbaudio_stream_params *params;
+
+
+ dev_dbg(dai->dev, "Mute:%d, Direction:%s\n", mute,
+ stream ? "CAPTURE":"PLAYBACK");
+
+ mutex_lock(&codec->lock);
+
+ params = find_dai_stream_params(codec, dai->id, stream);
+ if (!params) {
+ dev_err(codec->dev, "Failed to fetch dai_stream pointer\n");
+ mutex_unlock(&codec->lock);
+ return -EINVAL;
+ }
+
+ if (list_empty(&codec->module_list)) {
+ dev_err(codec->dev, "No codec module available\n");
+ if (mute) {
+ params->state = GBAUDIO_CODEC_STOP;
+ ret = 0;
+ } else {
+ ret = -ENODEV;
+ }
+ mutex_unlock(&codec->lock);
+ return ret;
+ }
+
+ list_for_each_entry(module, &codec->module_list, list) {
+ /* find the dai */
+ data = find_data(module, dai->id);
+ if (data)
+ break;
+ }
+ if (!data) {
+ dev_err(dai->dev, "%s:%s DATA connection missing\n",
+ dai->name, module->name);
+ mutex_unlock(&codec->lock);
+ return -ENODEV;
+ }
+
+ bundle = to_gb_bundle(module->dev);
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret) {
+ mutex_unlock(&codec->lock);
+ return ret;
+ }
+
+ if (!mute && !stream) {/* start playback */
+ ret = gb_audio_apbridgea_prepare_tx(data->connection,
+ 0);
+ if (!ret)
+ ret = gb_audio_apbridgea_start_tx(data->connection,
+ 0, 0);
+ params->state = GBAUDIO_CODEC_START;
+ } else if (!mute && stream) {/* start capture */
+ ret = gb_audio_apbridgea_prepare_rx(data->connection,
+ 0);
+ if (!ret)
+ ret = gb_audio_apbridgea_start_rx(data->connection,
+ 0);
+ params->state = GBAUDIO_CODEC_START;
+ } else if (mute && !stream) {/* stop playback */
+ ret = gb_audio_apbridgea_stop_tx(data->connection, 0);
+ if (!ret)
+ ret = gb_audio_apbridgea_shutdown_tx(data->connection,
+ 0);
+ params->state = GBAUDIO_CODEC_STOP;
+ } else if (mute && stream) {/* stop capture */
+ ret = gb_audio_apbridgea_stop_rx(data->connection, 0);
+ if (!ret)
+ ret = gb_audio_apbridgea_shutdown_rx(data->connection,
+ 0);
+ params->state = GBAUDIO_CODEC_STOP;
+ } else
+ ret = -EINVAL;
+ if (ret)
+ dev_err_ratelimited(dai->dev,
+ "%s:Error during %s %s stream:%d\n",
+ module->name, mute ? "Mute" : "Unmute",
+ stream ? "Capture" : "Playback", ret);
+
+ gb_pm_runtime_put_noidle(bundle);
+ mutex_unlock(&codec->lock);
+ return ret;
+}
+
+static struct snd_soc_dai_ops gbcodec_dai_ops = {
+ .startup = gbcodec_startup,
+ .shutdown = gbcodec_shutdown,
+ .hw_params = gbcodec_hw_params,
+ .prepare = gbcodec_prepare,
+ .mute_stream = gbcodec_mute_stream,
+};
+
+static struct snd_soc_dai_driver gbaudio_dai[] = {
+ {
+ .name = "apb-i2s0",
+ .id = 0,
+ .playback = {
+ .stream_name = "I2S 0 Playback",
+ .rates = SNDRV_PCM_RATE_48000,
+ .formats = SNDRV_PCM_FORMAT_S16_LE,
+ .rate_max = 48000,
+ .rate_min = 48000,
+ .channels_min = 1,
+ .channels_max = 2,
+ },
+ .capture = {
+ .stream_name = "I2S 0 Capture",
+ .rates = SNDRV_PCM_RATE_48000,
+ .formats = SNDRV_PCM_FORMAT_S16_LE,
+ .rate_max = 48000,
+ .rate_min = 48000,
+ .channels_min = 1,
+ .channels_max = 2,
+ },
+ .ops = &gbcodec_dai_ops,
+ },
+};
+
+static int gbaudio_init_jack(struct gbaudio_module_info *module,
+ struct snd_soc_codec *codec)
+{
+ int ret;
+
+ if (!module->jack_mask)
+ return 0;
+
+ snprintf(module->jack_name, NAME_SIZE, "GB %d Headset Jack",
+ module->dev_id);
+ ret = snd_soc_jack_new(codec, module->jack_name, module->jack_mask,
+ &module->headset_jack);
+ if (ret) {
+ dev_err(module->dev, "Failed to create new jack\n");
+ return ret;
+ }
+
+ if (!module->button_mask)
+ return 0;
+
+ snprintf(module->button_name, NAME_SIZE, "GB %d Button Jack",
+ module->dev_id);
+ ret = snd_soc_jack_new(codec, module->button_name, module->button_mask,
+ &module->button_jack);
+ if (ret) {
+ dev_err(module->dev, "Failed to create button jack\n");
+ return ret;
+ }
+
+ /*
+ * Currently, max 4 buttons are supported with following key mapping
+ * BTN_0 = KEY_MEDIA
+ * BTN_1 = KEY_VOICECOMMAND
+ * BTN_2 = KEY_VOLUMEUP
+ * BTN_3 = KEY_VOLUMEDOWN
+ */
+
+ if (module->button_mask & SND_JACK_BTN_0) {
+ ret = snd_jack_set_key(module->button_jack.jack, SND_JACK_BTN_0,
+ KEY_MEDIA);
+ if (ret) {
+ dev_err(module->dev, "Failed to set BTN_0\n");
+ return ret;
+ }
+ }
+
+ if (module->button_mask & SND_JACK_BTN_1) {
+ ret = snd_jack_set_key(module->button_jack.jack, SND_JACK_BTN_1,
+ KEY_VOICECOMMAND);
+ if (ret) {
+ dev_err(module->dev, "Failed to set BTN_1\n");
+ return ret;
+ }
+ }
+
+ if (module->button_mask & SND_JACK_BTN_2) {
+ ret = snd_jack_set_key(module->button_jack.jack, SND_JACK_BTN_2,
+ KEY_VOLUMEUP);
+ if (ret) {
+ dev_err(module->dev, "Failed to set BTN_2\n");
+ return ret;
+ }
+ }
+
+ if (module->button_mask & SND_JACK_BTN_3) {
+ ret = snd_jack_set_key(module->button_jack.jack, SND_JACK_BTN_3,
+ KEY_VOLUMEDOWN);
+ if (ret) {
+ dev_err(module->dev, "Failed to set BTN_0\n");
+ return ret;
+ }
+ }
+
+ /* FIXME
+ * verify if this is really required
+ set_bit(INPUT_PROP_NO_DUMMY_RELEASE,
+ module->button_jack.jack->input_dev->propbit);
+ */
+
+ return 0;
+}
+
+int gbaudio_register_module(struct gbaudio_module_info *module)
+{
+ int ret;
+ struct snd_soc_codec *codec;
+ struct snd_card *card;
+ struct snd_soc_jack *jack = NULL;
+
+ if (!gbcodec) {
+ dev_err(module->dev, "GB Codec not yet probed\n");
+ return -EAGAIN;
+ }
+
+ codec = gbcodec->codec;
+ card = codec->card->snd_card;
+
+ down_write(&card->controls_rwsem);
+
+ if (module->num_dais) {
+ dev_err(gbcodec->dev,
+ "%d:DAIs not supported via gbcodec driver\n",
+ module->num_dais);
+ up_write(&card->controls_rwsem);
+ return -EINVAL;
+ }
+
+ ret = gbaudio_init_jack(module, codec);
+ if (ret) {
+ up_write(&card->controls_rwsem);
+ return ret;
+ }
+
+ if (module->dapm_widgets)
+ snd_soc_dapm_new_controls(&codec->dapm, module->dapm_widgets,
+ module->num_dapm_widgets);
+ if (module->controls)
+ snd_soc_add_codec_controls(codec, module->controls,
+ module->num_controls);
+ if (module->dapm_routes)
+ snd_soc_dapm_add_routes(&codec->dapm, module->dapm_routes,
+ module->num_dapm_routes);
+
+ /* card already instantiated, create widgets here only */
+ if (codec->card->instantiated) {
+ snd_soc_dapm_link_component_dai_widgets(codec->card,
+ &codec->dapm);
+#ifdef CONFIG_SND_JACK
+ /* register jack devices for this module from codec->jack_list */
+ list_for_each_entry(jack, &codec->jack_list, list) {
+ if ((jack == &module->headset_jack)
+ || (jack == &module->button_jack))
+ snd_device_register(codec->card->snd_card,
+ jack->jack);
+ }
+#endif
+ }
+
+ mutex_lock(&gbcodec->lock);
+ list_add(&module->list, &gbcodec->module_list);
+ mutex_unlock(&gbcodec->lock);
+
+ if (codec->card->instantiated)
+ ret = snd_soc_dapm_new_widgets(&codec->dapm);
+ dev_dbg(codec->dev, "Registered %s module\n", module->name);
+
+ up_write(&card->controls_rwsem);
+ return ret;
+}
+EXPORT_SYMBOL(gbaudio_register_module);
+
+static void gbaudio_codec_clean_data_tx(struct gbaudio_data_connection *data)
+{
+ uint16_t i2s_port, cportid;
+ int ret;
+
+ if (list_is_singular(&gbcodec->module_list)) {
+ ret = gb_audio_apbridgea_stop_tx(data->connection, 0);
+ if (ret)
+ return;
+ ret = gb_audio_apbridgea_shutdown_tx(data->connection,
+ 0);
+ if (ret)
+ return;
+ }
+ i2s_port = 0; /* fixed for now */
+ cportid = data->connection->hd_cport_id;
+ ret = gb_audio_apbridgea_unregister_cport(data->connection,
+ i2s_port, cportid,
+ AUDIO_APBRIDGEA_DIRECTION_TX);
+ data->state[0] = GBAUDIO_CODEC_SHUTDOWN;
+}
+
+static void gbaudio_codec_clean_data_rx(struct gbaudio_data_connection *data)
+{
+ uint16_t i2s_port, cportid;
+ int ret;
+
+ if (list_is_singular(&gbcodec->module_list)) {
+ ret = gb_audio_apbridgea_stop_rx(data->connection, 0);
+ if (ret)
+ return;
+ ret = gb_audio_apbridgea_shutdown_rx(data->connection,
+ 0);
+ if (ret)
+ return;
+ }
+ i2s_port = 0; /* fixed for now */
+ cportid = data->connection->hd_cport_id;
+ ret = gb_audio_apbridgea_unregister_cport(data->connection,
+ i2s_port, cportid,
+ AUDIO_APBRIDGEA_DIRECTION_RX);
+ data->state[1] = GBAUDIO_CODEC_SHUTDOWN;
+}
+
+
+static void gbaudio_codec_cleanup(struct gbaudio_module_info *module)
+{
+ struct gbaudio_data_connection *data;
+ int pb_state, cap_state;
+
+ dev_dbg(gbcodec->dev, "%s: removed, cleanup APBridge\n", module->name);
+ list_for_each_entry(data, &module->data_list, list) {
+ pb_state = data->state[0];
+ cap_state = data->state[1];
+
+ if (pb_state > GBAUDIO_CODEC_SHUTDOWN)
+ gbaudio_codec_clean_data_tx(data);
+
+ if (cap_state > GBAUDIO_CODEC_SHUTDOWN)
+ gbaudio_codec_clean_data_rx(data);
+
+ }
+}
+
+void gbaudio_unregister_module(struct gbaudio_module_info *module)
+{
+ struct snd_soc_codec *codec = gbcodec->codec;
+ struct snd_card *card = codec->card->snd_card;
+ struct snd_soc_jack *jack, *next_j;
+ int mask;
+
+ dev_dbg(codec->dev, "Unregister %s module\n", module->name);
+
+ down_write(&card->controls_rwsem);
+ mutex_lock(&gbcodec->lock);
+ gbaudio_codec_cleanup(module);
+ list_del(&module->list);
+ dev_dbg(codec->dev, "Process Unregister %s module\n", module->name);
+ mutex_unlock(&gbcodec->lock);
+
+#ifdef CONFIG_SND_JACK
+ /* free jack devices for this module from codec->jack_list */
+ list_for_each_entry_safe(jack, next_j, &codec->jack_list, list) {
+ if (jack == &module->headset_jack)
+ mask = GBCODEC_JACK_MASK;
+ else if (jack == &module->button_jack)
+ mask = GBCODEC_JACK_BUTTON_MASK;
+ else
+ mask = 0;
+ if (mask) {
+ dev_dbg(module->dev, "Report %s removal\n",
+ jack->jack->id);
+ snd_soc_jack_report(jack, 0, mask);
+ snd_device_free(codec->card->snd_card, jack->jack);
+ list_del(&jack->list);
+ }
+ }
+#endif
+
+ if (module->dapm_routes) {
+ dev_dbg(codec->dev, "Removing %d routes\n",
+ module->num_dapm_routes);
+ snd_soc_dapm_del_routes(&codec->dapm, module->dapm_routes,
+ module->num_dapm_routes);
+ }
+ if (module->controls) {
+ dev_dbg(codec->dev, "Removing %d controls\n",
+ module->num_controls);
+ snd_soc_remove_codec_controls(codec, module->controls,
+ module->num_controls);
+ }
+ if (module->dapm_widgets) {
+ dev_dbg(codec->dev, "Removing %d widgets\n",
+ module->num_dapm_widgets);
+ snd_soc_dapm_free_controls(&codec->dapm, module->dapm_widgets,
+ module->num_dapm_widgets);
+ }
+
+ dev_dbg(codec->dev, "Unregistered %s module\n", module->name);
+
+ up_write(&card->controls_rwsem);
+}
+EXPORT_SYMBOL(gbaudio_unregister_module);
+
+/*
+ * codec driver ops
+ */
+static int gbcodec_probe(struct snd_soc_codec *codec)
+{
+ int i;
+ struct gbaudio_codec_info *info;
+ struct gbaudio_codec_dai *dai;
+
+ info = devm_kzalloc(codec->dev, sizeof(*info), GFP_KERNEL);
+ if (!info)
+ return -ENOMEM;
+
+ info->dev = codec->dev;
+ INIT_LIST_HEAD(&info->module_list);
+ mutex_init(&info->lock);
+ INIT_LIST_HEAD(&info->dai_list);
+
+ /* init dai_list used to maintain runtime stream info */
+ for (i = 0; i < ARRAY_SIZE(gbaudio_dai); i++) {
+ dai = devm_kzalloc(codec->dev, sizeof(*dai), GFP_KERNEL);
+ if (!dai)
+ return -ENOMEM;
+ dai->id = gbaudio_dai[i].id;
+ list_add(&dai->list, &info->dai_list);
+ }
+
+ info->codec = codec;
+ snd_soc_codec_set_drvdata(codec, info);
+ gbcodec = info;
+
+ device_init_wakeup(codec->dev, 1);
+ return 0;
+}
+
+static int gbcodec_remove(struct snd_soc_codec *codec)
+{
+ /* Empty function for now */
+ return 0;
+}
+
+static u8 gbcodec_reg[GBCODEC_REG_COUNT] = {
+ [GBCODEC_CTL_REG] = GBCODEC_CTL_REG_DEFAULT,
+ [GBCODEC_MUTE_REG] = GBCODEC_MUTE_REG_DEFAULT,
+ [GBCODEC_PB_LVOL_REG] = GBCODEC_PB_VOL_REG_DEFAULT,
+ [GBCODEC_PB_RVOL_REG] = GBCODEC_PB_VOL_REG_DEFAULT,
+ [GBCODEC_CAP_LVOL_REG] = GBCODEC_CAP_VOL_REG_DEFAULT,
+ [GBCODEC_CAP_RVOL_REG] = GBCODEC_CAP_VOL_REG_DEFAULT,
+ [GBCODEC_APB1_MUX_REG] = GBCODEC_APB1_MUX_REG_DEFAULT,
+ [GBCODEC_APB2_MUX_REG] = GBCODEC_APB2_MUX_REG_DEFAULT,
+};
+
+static int gbcodec_write(struct snd_soc_codec *codec, unsigned int reg,
+ unsigned int value)
+{
+ int ret = 0;
+
+ if (reg == SND_SOC_NOPM)
+ return 0;
+
+ BUG_ON(reg >= GBCODEC_REG_COUNT);
+
+ gbcodec_reg[reg] = value;
+ dev_dbg(codec->dev, "reg[%d] = 0x%x\n", reg, value);
+
+ return ret;
+}
+
+static unsigned int gbcodec_read(struct snd_soc_codec *codec,
+ unsigned int reg)
+{
+ unsigned int val = 0;
+
+ if (reg == SND_SOC_NOPM)
+ return 0;
+
+ BUG_ON(reg >= GBCODEC_REG_COUNT);
+
+ val = gbcodec_reg[reg];
+ dev_dbg(codec->dev, "reg[%d] = 0x%x\n", reg, val);
+
+ return val;
+}
+
+static struct snd_soc_codec_driver soc_codec_dev_gbaudio = {
+ .probe = gbcodec_probe,
+ .remove = gbcodec_remove,
+
+ .read = gbcodec_read,
+ .write = gbcodec_write,
+
+ .reg_cache_size = GBCODEC_REG_COUNT,
+ .reg_cache_default = gbcodec_reg_defaults,
+ .reg_word_size = 1,
+
+ .idle_bias_off = true,
+ .ignore_pmdown_time = 1,
+};
+
+#ifdef CONFIG_PM
+static int gbaudio_codec_suspend(struct device *dev)
+{
+ dev_dbg(dev, "%s: suspend\n", __func__);
+ return 0;
+}
+
+static int gbaudio_codec_resume(struct device *dev)
+{
+ dev_dbg(dev, "%s: resume\n", __func__);
+ return 0;
+}
+
+static const struct dev_pm_ops gbaudio_codec_pm_ops = {
+ .suspend = gbaudio_codec_suspend,
+ .resume = gbaudio_codec_resume,
+};
+#endif
+
+static int gbaudio_codec_probe(struct platform_device *pdev)
+{
+ return snd_soc_register_codec(&pdev->dev, &soc_codec_dev_gbaudio,
+ gbaudio_dai, ARRAY_SIZE(gbaudio_dai));
+}
+
+static int gbaudio_codec_remove(struct platform_device *pdev)
+{
+ snd_soc_unregister_codec(&pdev->dev);
+ return 0;
+}
+
+static const struct of_device_id greybus_asoc_machine_of_match[] = {
+ { .compatible = "toshiba,apb-dummy-codec", },
+ {},
+};
+
+static struct platform_driver gbaudio_codec_driver = {
+ .driver = {
+ .name = "apb-dummy-codec",
+ .owner = THIS_MODULE,
+#ifdef CONFIG_PM
+ .pm = &gbaudio_codec_pm_ops,
+#endif
+ .of_match_table = greybus_asoc_machine_of_match,
+ },
+ .probe = gbaudio_codec_probe,
+ .remove = gbaudio_codec_remove,
+};
+module_platform_driver(gbaudio_codec_driver);
+
+MODULE_DESCRIPTION("APBridge ALSA SoC dummy codec driver");
+MODULE_AUTHOR("Vaibhav Agarwal <vaibhav.agarwal@linaro.org>");
+MODULE_LICENSE("GPL v2");
+MODULE_ALIAS("platform:apb-dummy-codec");
--- /dev/null
+++ b/drivers/greybus/audio_codec.h
@@ -0,0 +1,283 @@
+/*
+ * Greybus audio driver
+ * Copyright 2015 Google Inc.
+ * Copyright 2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __LINUX_GBAUDIO_CODEC_H
+#define __LINUX_GBAUDIO_CODEC_H
+
+#include <sound/soc.h>
+#include <sound/jack.h>
+
+#include "greybus.h"
+#include "greybus_protocols.h"
+
+#define NAME_SIZE 32
+#define MAX_DAIS 2 /* APB1, APB2 */
+
+enum {
+ APB1_PCM = 0,
+ APB2_PCM,
+ NUM_CODEC_DAIS,
+};
+
+enum gbcodec_reg_index {
+ GBCODEC_CTL_REG,
+ GBCODEC_MUTE_REG,
+ GBCODEC_PB_LVOL_REG,
+ GBCODEC_PB_RVOL_REG,
+ GBCODEC_CAP_LVOL_REG,
+ GBCODEC_CAP_RVOL_REG,
+ GBCODEC_APB1_MUX_REG,
+ GBCODEC_APB2_MUX_REG,
+ GBCODEC_REG_COUNT
+};
+
+/* device_type should be same as defined in audio.h (Android media layer) */
+enum {
+ GBAUDIO_DEVICE_NONE = 0x0,
+ /* reserved bits */
+ GBAUDIO_DEVICE_BIT_IN = 0x80000000,
+ GBAUDIO_DEVICE_BIT_DEFAULT = 0x40000000,
+ /* output devices */
+ GBAUDIO_DEVICE_OUT_SPEAKER = 0x2,
+ GBAUDIO_DEVICE_OUT_WIRED_HEADSET = 0x4,
+ GBAUDIO_DEVICE_OUT_WIRED_HEADPHONE = 0x8,
+ /* input devices */
+ GBAUDIO_DEVICE_IN_BUILTIN_MIC = GBAUDIO_DEVICE_BIT_IN | 0x4,
+ GBAUDIO_DEVICE_IN_WIRED_HEADSET = GBAUDIO_DEVICE_BIT_IN | 0x10,
+};
+
+/* bit 0-SPK, 1-HP, 2-DAC,
+ * 4-MIC, 5-HSMIC, 6-MIC2
+ */
+#define GBCODEC_CTL_REG_DEFAULT 0x00
+
+/* bit 0,1 - APB1-PB-L/R
+ * bit 2,3 - APB2-PB-L/R
+ * bit 4,5 - APB1-Cap-L/R
+ * bit 6,7 - APB2-Cap-L/R
+ */
+#define GBCODEC_MUTE_REG_DEFAULT 0x00
+
+/* 0-127 steps */
+#define GBCODEC_PB_VOL_REG_DEFAULT 0x00
+#define GBCODEC_CAP_VOL_REG_DEFAULT 0x00
+
+/* bit 0,1,2 - PB stereo, left, right
+ * bit 8,9,10 - Cap stereo, left, right
+ */
+#define GBCODEC_APB1_MUX_REG_DEFAULT 0x00
+#define GBCODEC_APB2_MUX_REG_DEFAULT 0x00
+
+#define GBCODEC_JACK_MASK 0x0000FFFF
+#define GBCODEC_JACK_BUTTON_MASK 0xFFFF0000
+
+static const u8 gbcodec_reg_defaults[GBCODEC_REG_COUNT] = {
+ GBCODEC_CTL_REG_DEFAULT,
+ GBCODEC_MUTE_REG_DEFAULT,
+ GBCODEC_PB_VOL_REG_DEFAULT,
+ GBCODEC_PB_VOL_REG_DEFAULT,
+ GBCODEC_CAP_VOL_REG_DEFAULT,
+ GBCODEC_CAP_VOL_REG_DEFAULT,
+ GBCODEC_APB1_MUX_REG_DEFAULT,
+ GBCODEC_APB2_MUX_REG_DEFAULT,
+};
+
+enum gbaudio_codec_state {
+ GBAUDIO_CODEC_SHUTDOWN = 0,
+ GBAUDIO_CODEC_STARTUP,
+ GBAUDIO_CODEC_HWPARAMS,
+ GBAUDIO_CODEC_PREPARE,
+ GBAUDIO_CODEC_START,
+ GBAUDIO_CODEC_STOP,
+};
+
+struct gbaudio_stream_params {
+ int state;
+ uint8_t sig_bits, channels;
+ uint32_t format, rate;
+};
+
+struct gbaudio_codec_dai {
+ int id;
+ /* runtime params for playback/capture streams */
+ struct gbaudio_stream_params params[2];
+ struct list_head list;
+};
+
+struct gbaudio_codec_info {
+ struct device *dev;
+ struct snd_soc_codec *codec;
+ struct list_head module_list;
+ /* to maintain runtime stream params for each DAI */
+ struct list_head dai_list;
+ struct mutex lock;
+ u8 reg[GBCODEC_REG_COUNT];
+};
+
+struct gbaudio_widget {
+ __u8 id;
+ const char *name;
+ struct list_head list;
+};
+
+struct gbaudio_control {
+ __u8 id;
+ char *name;
+ char *wname;
+ const char * const *texts;
+ int items;
+ struct list_head list;
+};
+
+struct gbaudio_data_connection {
+ int id;
+ __le16 data_cport;
+ struct gb_connection *connection;
+ struct list_head list;
+ /* maintain runtime state for playback/capture stream */
+ int state[2];
+};
+
+/* stream direction */
+#define GB_PLAYBACK BIT(0)
+#define GB_CAPTURE BIT(1)
+
+enum gbaudio_module_state {
+ GBAUDIO_MODULE_OFF = 0,
+ GBAUDIO_MODULE_ON,
+};
+
+struct gbaudio_module_info {
+ /* module info */
+ struct device *dev;
+ int dev_id; /* check if it should be bundle_id/hd_cport_id */
+ int vid;
+ int pid;
+ int slot;
+ int type;
+ int set_uevent;
+ char vstr[NAME_SIZE];
+ char pstr[NAME_SIZE];
+ struct list_head list;
+ /* need to share this info to above user space */
+ int manager_id;
+ char name[NAME_SIZE];
+ unsigned int ip_devices;
+ unsigned int op_devices;
+
+ /* jack related */
+ char jack_name[NAME_SIZE];
+ char button_name[NAME_SIZE];
+ int jack_type;
+ int jack_mask;
+ int button_mask;
+ int button_status;
+ struct snd_soc_jack headset_jack;
+ struct snd_soc_jack button_jack;
+
+ /* connection info */
+ struct gb_connection *mgmt_connection;
+ size_t num_data_connections;
+ struct list_head data_list;
+
+ /* topology related */
+ int num_dais;
+ int num_controls;
+ int num_dapm_widgets;
+ int num_dapm_routes;
+ unsigned long dai_offset;
+ unsigned long widget_offset;
+ unsigned long control_offset;
+ unsigned long route_offset;
+ struct snd_kcontrol_new *controls;
+ struct snd_soc_dapm_widget *dapm_widgets;
+ struct snd_soc_dapm_route *dapm_routes;
+ struct snd_soc_dai_driver *dais;
+
+ struct list_head widget_list;
+ struct list_head ctl_list;
+ struct list_head widget_ctl_list;
+
+ struct gb_audio_topology *topology;
+};
+
+int gbaudio_tplg_parse_data(struct gbaudio_module_info *module,
+ struct gb_audio_topology *tplg_data);
+void gbaudio_tplg_release(struct gbaudio_module_info *module);
+
+int gbaudio_module_update(struct gbaudio_codec_info *codec,
+ struct snd_soc_dapm_widget *w,
+ struct gbaudio_module_info *module,
+ int enable);
+int gbaudio_register_module(struct gbaudio_module_info *module);
+void gbaudio_unregister_module(struct gbaudio_module_info *module);
+
+/* protocol related */
+extern int gb_audio_gb_get_topology(struct gb_connection *connection,
+ struct gb_audio_topology **topology);
+extern int gb_audio_gb_get_control(struct gb_connection *connection,
+ uint8_t control_id, uint8_t index,
+ struct gb_audio_ctl_elem_value *value);
+extern int gb_audio_gb_set_control(struct gb_connection *connection,
+ uint8_t control_id, uint8_t index,
+ struct gb_audio_ctl_elem_value *value);
+extern int gb_audio_gb_enable_widget(struct gb_connection *connection,
+ uint8_t widget_id);
+extern int gb_audio_gb_disable_widget(struct gb_connection *connection,
+ uint8_t widget_id);
+extern int gb_audio_gb_get_pcm(struct gb_connection *connection,
+ uint16_t data_cport, uint32_t *format,
+ uint32_t *rate, uint8_t *channels,
+ uint8_t *sig_bits);
+extern int gb_audio_gb_set_pcm(struct gb_connection *connection,
+ uint16_t data_cport, uint32_t format,
+ uint32_t rate, uint8_t channels,
+ uint8_t sig_bits);
+extern int gb_audio_gb_set_tx_data_size(struct gb_connection *connection,
+ uint16_t data_cport, uint16_t size);
+extern int gb_audio_gb_activate_tx(struct gb_connection *connection,
+ uint16_t data_cport);
+extern int gb_audio_gb_deactivate_tx(struct gb_connection *connection,
+ uint16_t data_cport);
+extern int gb_audio_gb_set_rx_data_size(struct gb_connection *connection,
+ uint16_t data_cport, uint16_t size);
+extern int gb_audio_gb_activate_rx(struct gb_connection *connection,
+ uint16_t data_cport);
+extern int gb_audio_gb_deactivate_rx(struct gb_connection *connection,
+ uint16_t data_cport);
+extern int gb_audio_apbridgea_set_config(struct gb_connection *connection,
+ __u16 i2s_port, __u32 format,
+ __u32 rate, __u32 mclk_freq);
+extern int gb_audio_apbridgea_register_cport(struct gb_connection *connection,
+ __u16 i2s_port, __u16 cportid,
+ __u8 direction);
+extern int gb_audio_apbridgea_unregister_cport(struct gb_connection *connection,
+ __u16 i2s_port, __u16 cportid,
+ __u8 direction);
+extern int gb_audio_apbridgea_set_tx_data_size(struct gb_connection *connection,
+ __u16 i2s_port, __u16 size);
+extern int gb_audio_apbridgea_prepare_tx(struct gb_connection *connection,
+ __u16 i2s_port);
+extern int gb_audio_apbridgea_start_tx(struct gb_connection *connection,
+ __u16 i2s_port, __u64 timestamp);
+extern int gb_audio_apbridgea_stop_tx(struct gb_connection *connection,
+ __u16 i2s_port);
+extern int gb_audio_apbridgea_shutdown_tx(struct gb_connection *connection,
+ __u16 i2s_port);
+extern int gb_audio_apbridgea_set_rx_data_size(struct gb_connection *connection,
+ __u16 i2s_port, __u16 size);
+extern int gb_audio_apbridgea_prepare_rx(struct gb_connection *connection,
+ __u16 i2s_port);
+extern int gb_audio_apbridgea_start_rx(struct gb_connection *connection,
+ __u16 i2s_port);
+extern int gb_audio_apbridgea_stop_rx(struct gb_connection *connection,
+ __u16 i2s_port);
+extern int gb_audio_apbridgea_shutdown_rx(struct gb_connection *connection,
+ __u16 i2s_port);
+
+#endif /* __LINUX_GBAUDIO_CODEC_H */
--- /dev/null
+++ b/drivers/greybus/audio_gb.c
@@ -0,0 +1,228 @@
+/*
+ * Greybus Audio Device Class Protocol helpers
+ *
+ * Copyright 2015-2016 Google Inc.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include "greybus.h"
+#include "greybus_protocols.h"
+#include "operation.h"
+#include "audio_codec.h"
+
+/* TODO: Split into separate calls */
+int gb_audio_gb_get_topology(struct gb_connection *connection,
+ struct gb_audio_topology **topology)
+{
+ struct gb_audio_get_topology_size_response size_resp;
+ struct gb_audio_topology *topo;
+ uint16_t size;
+ int ret;
+
+ ret = gb_operation_sync(connection, GB_AUDIO_TYPE_GET_TOPOLOGY_SIZE,
+ NULL, 0, &size_resp, sizeof(size_resp));
+ if (ret)
+ return ret;
+
+ size = le16_to_cpu(size_resp.size);
+ if (size < sizeof(*topo))
+ return -ENODATA;
+
+ topo = kzalloc(size, GFP_KERNEL);
+ if (!topo)
+ return -ENOMEM;
+
+ ret = gb_operation_sync(connection, GB_AUDIO_TYPE_GET_TOPOLOGY, NULL, 0,
+ topo, size);
+ if (ret) {
+ kfree(topo);
+ return ret;
+ }
+
+ *topology = topo;
+
+ return 0;
+}
+EXPORT_SYMBOL_GPL(gb_audio_gb_get_topology);
+
+int gb_audio_gb_get_control(struct gb_connection *connection,
+ uint8_t control_id, uint8_t index,
+ struct gb_audio_ctl_elem_value *value)
+{
+ struct gb_audio_get_control_request req;
+ struct gb_audio_get_control_response resp;
+ int ret;
+
+ req.control_id = control_id;
+ req.index = index;
+
+ ret = gb_operation_sync(connection, GB_AUDIO_TYPE_GET_CONTROL,
+ &req, sizeof(req), &resp, sizeof(resp));
+ if (ret)
+ return ret;
+
+ memcpy(value, &resp.value, sizeof(*value));
+
+ return 0;
+}
+EXPORT_SYMBOL_GPL(gb_audio_gb_get_control);
+
+int gb_audio_gb_set_control(struct gb_connection *connection,
+ uint8_t control_id, uint8_t index,
+ struct gb_audio_ctl_elem_value *value)
+{
+ struct gb_audio_set_control_request req;
+
+ req.control_id = control_id;
+ req.index = index;
+ memcpy(&req.value, value, sizeof(req.value));
+
+ return gb_operation_sync(connection, GB_AUDIO_TYPE_SET_CONTROL,
+ &req, sizeof(req), NULL, 0);
+}
+EXPORT_SYMBOL_GPL(gb_audio_gb_set_control);
+
+int gb_audio_gb_enable_widget(struct gb_connection *connection,
+ uint8_t widget_id)
+{
+ struct gb_audio_enable_widget_request req;
+
+ req.widget_id = widget_id;
+
+ return gb_operation_sync(connection, GB_AUDIO_TYPE_ENABLE_WIDGET,
+ &req, sizeof(req), NULL, 0);
+}
+EXPORT_SYMBOL_GPL(gb_audio_gb_enable_widget);
+
+int gb_audio_gb_disable_widget(struct gb_connection *connection,
+ uint8_t widget_id)
+{
+ struct gb_audio_disable_widget_request req;
+
+ req.widget_id = widget_id;
+
+ return gb_operation_sync(connection, GB_AUDIO_TYPE_DISABLE_WIDGET,
+ &req, sizeof(req), NULL, 0);
+}
+EXPORT_SYMBOL_GPL(gb_audio_gb_disable_widget);
+
+int gb_audio_gb_get_pcm(struct gb_connection *connection, uint16_t data_cport,
+ uint32_t *format, uint32_t *rate, uint8_t *channels,
+ uint8_t *sig_bits)
+{
+ struct gb_audio_get_pcm_request req;
+ struct gb_audio_get_pcm_response resp;
+ int ret;
+
+ req.data_cport = cpu_to_le16(data_cport);
+
+ ret = gb_operation_sync(connection, GB_AUDIO_TYPE_GET_PCM,
+ &req, sizeof(req), &resp, sizeof(resp));
+ if (ret)
+ return ret;
+
+ *format = le32_to_cpu(resp.format);
+ *rate = le32_to_cpu(resp.rate);
+ *channels = resp.channels;
+ *sig_bits = resp.sig_bits;
+
+ return 0;
+}
+EXPORT_SYMBOL_GPL(gb_audio_gb_get_pcm);
+
+int gb_audio_gb_set_pcm(struct gb_connection *connection, uint16_t data_cport,
+ uint32_t format, uint32_t rate, uint8_t channels,
+ uint8_t sig_bits)
+{
+ struct gb_audio_set_pcm_request req;
+
+ req.data_cport = cpu_to_le16(data_cport);
+ req.format = cpu_to_le32(format);
+ req.rate = cpu_to_le32(rate);
+ req.channels = channels;
+ req.sig_bits = sig_bits;
+
+ return gb_operation_sync(connection, GB_AUDIO_TYPE_SET_PCM,
+ &req, sizeof(req), NULL, 0);
+}
+EXPORT_SYMBOL_GPL(gb_audio_gb_set_pcm);
+
+int gb_audio_gb_set_tx_data_size(struct gb_connection *connection,
+ uint16_t data_cport, uint16_t size)
+{
+ struct gb_audio_set_tx_data_size_request req;
+
+ req.data_cport = cpu_to_le16(data_cport);
+ req.size = cpu_to_le16(size);
+
+ return gb_operation_sync(connection, GB_AUDIO_TYPE_SET_TX_DATA_SIZE,
+ &req, sizeof(req), NULL, 0);
+}
+EXPORT_SYMBOL_GPL(gb_audio_gb_set_tx_data_size);
+
+int gb_audio_gb_activate_tx(struct gb_connection *connection,
+ uint16_t data_cport)
+{
+ struct gb_audio_activate_tx_request req;
+
+ req.data_cport = cpu_to_le16(data_cport);
+
+ return gb_operation_sync(connection, GB_AUDIO_TYPE_ACTIVATE_TX,
+ &req, sizeof(req), NULL, 0);
+}
+EXPORT_SYMBOL_GPL(gb_audio_gb_activate_tx);
+
+int gb_audio_gb_deactivate_tx(struct gb_connection *connection,
+ uint16_t data_cport)
+{
+ struct gb_audio_deactivate_tx_request req;
+
+ req.data_cport = cpu_to_le16(data_cport);
+
+ return gb_operation_sync(connection, GB_AUDIO_TYPE_DEACTIVATE_TX,
+ &req, sizeof(req), NULL, 0);
+}
+EXPORT_SYMBOL_GPL(gb_audio_gb_deactivate_tx);
+
+int gb_audio_gb_set_rx_data_size(struct gb_connection *connection,
+ uint16_t data_cport, uint16_t size)
+{
+ struct gb_audio_set_rx_data_size_request req;
+
+ req.data_cport = cpu_to_le16(data_cport);
+ req.size = cpu_to_le16(size);
+
+ return gb_operation_sync(connection, GB_AUDIO_TYPE_SET_RX_DATA_SIZE,
+ &req, sizeof(req), NULL, 0);
+}
+EXPORT_SYMBOL_GPL(gb_audio_gb_set_rx_data_size);
+
+int gb_audio_gb_activate_rx(struct gb_connection *connection,
+ uint16_t data_cport)
+{
+ struct gb_audio_activate_rx_request req;
+
+ req.data_cport = cpu_to_le16(data_cport);
+
+ return gb_operation_sync(connection, GB_AUDIO_TYPE_ACTIVATE_RX,
+ &req, sizeof(req), NULL, 0);
+}
+EXPORT_SYMBOL_GPL(gb_audio_gb_activate_rx);
+
+int gb_audio_gb_deactivate_rx(struct gb_connection *connection,
+ uint16_t data_cport)
+{
+ struct gb_audio_deactivate_rx_request req;
+
+ req.data_cport = cpu_to_le16(data_cport);
+
+ return gb_operation_sync(connection, GB_AUDIO_TYPE_DEACTIVATE_RX,
+ &req, sizeof(req), NULL, 0);
+}
+EXPORT_SYMBOL_GPL(gb_audio_gb_deactivate_rx);
+
+MODULE_LICENSE("GPL v2");
+MODULE_ALIAS("greybus:audio-gb");
+MODULE_DESCRIPTION("Greybus Audio Device Class Protocol library");
+MODULE_AUTHOR("Mark Greer <mgreer@animalcreek.com>");
--- /dev/null
+++ b/drivers/greybus/audio_manager.c
@@ -0,0 +1,184 @@
+/*
+ * Greybus operations
+ *
+ * Copyright 2015-2016 Google Inc.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/string.h>
+#include <linux/sysfs.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/rwlock.h>
+#include <linux/idr.h>
+
+#include "audio_manager.h"
+#include "audio_manager_private.h"
+
+static struct kset *manager_kset;
+
+static LIST_HEAD(modules_list);
+static DECLARE_RWSEM(modules_rwsem);
+static DEFINE_IDA(module_id);
+
+/* helpers */
+static struct gb_audio_manager_module *gb_audio_manager_get_locked(int id)
+{
+ struct gb_audio_manager_module *module;
+
+ if (id < 0)
+ return NULL;
+
+ list_for_each_entry(module, &modules_list, list) {
+ if (module->id == id)
+ return module;
+ }
+
+ return NULL;
+}
+
+/* public API */
+int gb_audio_manager_add(struct gb_audio_manager_module_descriptor *desc)
+{
+ struct gb_audio_manager_module *module;
+ int id;
+ int err;
+
+ id = ida_simple_get(&module_id, 0, 0, GFP_KERNEL);
+ err = gb_audio_manager_module_create(&module, manager_kset,
+ id, desc);
+ if (err) {
+ ida_simple_remove(&module_id, id);
+ return err;
+ }
+
+ /* Add it to the list */
+ down_write(&modules_rwsem);
+ list_add_tail(&module->list, &modules_list);
+ up_write(&modules_rwsem);
+
+ return module->id;
+}
+EXPORT_SYMBOL_GPL(gb_audio_manager_add);
+
+int gb_audio_manager_remove(int id)
+{
+ struct gb_audio_manager_module *module;
+
+ down_write(&modules_rwsem);
+
+ module = gb_audio_manager_get_locked(id);
+ if (!module) {
+ up_write(&modules_rwsem);
+ return -EINVAL;
+ }
+ list_del(&module->list);
+ kobject_put(&module->kobj);
+ up_write(&modules_rwsem);
+ ida_simple_remove(&module_id, id);
+ return 0;
+}
+EXPORT_SYMBOL_GPL(gb_audio_manager_remove);
+
+void gb_audio_manager_remove_all(void)
+{
+ struct gb_audio_manager_module *module, *next;
+ int is_empty = 1;
+
+ down_write(&modules_rwsem);
+
+ list_for_each_entry_safe(module, next, &modules_list, list) {
+ list_del(&module->list);
+ kobject_put(&module->kobj);
+ ida_simple_remove(&module_id, module->id);
+ }
+
+ is_empty = list_empty(&modules_list);
+
+ up_write(&modules_rwsem);
+
+ if (!is_empty)
+ pr_warn("Not all nodes were deleted\n");
+}
+EXPORT_SYMBOL_GPL(gb_audio_manager_remove_all);
+
+struct gb_audio_manager_module *gb_audio_manager_get_module(int id)
+{
+ struct gb_audio_manager_module *module;
+
+ down_read(&modules_rwsem);
+ module = gb_audio_manager_get_locked(id);
+ kobject_get(&module->kobj);
+ up_read(&modules_rwsem);
+ return module;
+}
+EXPORT_SYMBOL_GPL(gb_audio_manager_get_module);
+
+void gb_audio_manager_put_module(struct gb_audio_manager_module *module)
+{
+ kobject_put(&module->kobj);
+}
+EXPORT_SYMBOL_GPL(gb_audio_manager_put_module);
+
+int gb_audio_manager_dump_module(int id)
+{
+ struct gb_audio_manager_module *module;
+
+ down_read(&modules_rwsem);
+ module = gb_audio_manager_get_locked(id);
+ up_read(&modules_rwsem);
+
+ if (!module)
+ return -EINVAL;
+
+ gb_audio_manager_module_dump(module);
+ return 0;
+}
+EXPORT_SYMBOL_GPL(gb_audio_manager_dump_module);
+
+void gb_audio_manager_dump_all(void)
+{
+ struct gb_audio_manager_module *module;
+ int count = 0;
+
+ down_read(&modules_rwsem);
+ list_for_each_entry(module, &modules_list, list) {
+ gb_audio_manager_module_dump(module);
+ count++;
+ }
+ up_read(&modules_rwsem);
+
+ pr_info("Number of connected modules: %d\n", count);
+}
+EXPORT_SYMBOL_GPL(gb_audio_manager_dump_all);
+
+/*
+ * module init/deinit
+ */
+static int __init manager_init(void)
+{
+ manager_kset = kset_create_and_add(GB_AUDIO_MANAGER_NAME, NULL,
+ kernel_kobj);
+ if (!manager_kset)
+ return -ENOMEM;
+
+#ifdef GB_AUDIO_MANAGER_SYSFS
+ gb_audio_manager_sysfs_init(&manager_kset->kobj);
+#endif
+
+ return 0;
+}
+
+static void __exit manager_exit(void)
+{
+ gb_audio_manager_remove_all();
+ kset_unregister(manager_kset);
+ ida_destroy(&module_id);
+}
+
+module_init(manager_init);
+module_exit(manager_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Svetlin Ankov <ankov_svetlin@projectara.com>");
--- /dev/null
+++ b/drivers/greybus/audio_manager.h
@@ -0,0 +1,83 @@
+/*
+ * Greybus operations
+ *
+ * Copyright 2015-2016 Google Inc.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef _GB_AUDIO_MANAGER_H_
+#define _GB_AUDIO_MANAGER_H_
+
+#include <linux/kobject.h>
+#include <linux/list.h>
+
+#define GB_AUDIO_MANAGER_NAME "gb_audio_manager"
+#define GB_AUDIO_MANAGER_MODULE_NAME_LEN 64
+#define GB_AUDIO_MANAGER_MODULE_NAME_LEN_SSCANF "63"
+
+struct gb_audio_manager_module_descriptor {
+ char name[GB_AUDIO_MANAGER_MODULE_NAME_LEN];
+ int slot;
+ int vid;
+ int pid;
+ int cport;
+ unsigned int ip_devices;
+ unsigned int op_devices;
+};
+
+struct gb_audio_manager_module {
+ struct kobject kobj;
+ struct list_head list;
+ int id;
+ struct gb_audio_manager_module_descriptor desc;
+};
+
+/*
+ * Creates a new gb_audio_manager_module_descriptor, using the specified
+ * descriptor.
+ *
+ * Returns a negative result on error, or the id of the newly created module.
+ *
+ */
+int gb_audio_manager_add(struct gb_audio_manager_module_descriptor *desc);
+
+/*
+ * Removes a connected gb_audio_manager_module_descriptor for the specified ID.
+ *
+ * Returns zero on success, or a negative value on error.
+ */
+int gb_audio_manager_remove(int id);
+
+/*
+ * Removes all connected gb_audio_modules
+ *
+ * Returns zero on success, or a negative value on error.
+ */
+void gb_audio_manager_remove_all(void);
+
+/*
+ * Retrieves a gb_audio_manager_module_descriptor for the specified id.
+ * Returns the gb_audio_manager_module_descriptor structure,
+ * or NULL if there is no module with the specified ID.
+ */
+struct gb_audio_manager_module *gb_audio_manager_get_module(int id);
+
+/*
+ * Decreases the refcount of the module, obtained by the get function.
+ * Modules are removed via gb_audio_manager_remove
+ */
+void gb_audio_manager_put_module(struct gb_audio_manager_module *module);
+
+/*
+ * Dumps the module for the specified id
+ * Return 0 on success
+ */
+int gb_audio_manager_dump_module(int id);
+
+/*
+ * Dumps all connected modules
+ */
+void gb_audio_manager_dump_all(void);
+
+#endif /* _GB_AUDIO_MANAGER_H_ */
--- /dev/null
+++ b/drivers/greybus/audio_manager_module.c
@@ -0,0 +1,258 @@
+/*
+ * Greybus operations
+ *
+ * Copyright 2015-2016 Google Inc.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/slab.h>
+
+#include "audio_manager.h"
+#include "audio_manager_private.h"
+
+#define to_gb_audio_module_attr(x) \
+ container_of(x, struct gb_audio_manager_module_attribute, attr)
+#define to_gb_audio_module(x) \
+ container_of(x, struct gb_audio_manager_module, kobj)
+
+struct gb_audio_manager_module_attribute {
+ struct attribute attr;
+ ssize_t (*show)(struct gb_audio_manager_module *module,
+ struct gb_audio_manager_module_attribute *attr,
+ char *buf);
+ ssize_t (*store)(struct gb_audio_manager_module *module,
+ struct gb_audio_manager_module_attribute *attr,
+ const char *buf, size_t count);
+};
+
+static ssize_t gb_audio_module_attr_show(
+ struct kobject *kobj, struct attribute *attr, char *buf)
+{
+ struct gb_audio_manager_module_attribute *attribute;
+ struct gb_audio_manager_module *module;
+
+ attribute = to_gb_audio_module_attr(attr);
+ module = to_gb_audio_module(kobj);
+
+ if (!attribute->show)
+ return -EIO;
+
+ return attribute->show(module, attribute, buf);
+}
+
+static ssize_t gb_audio_module_attr_store(struct kobject *kobj,
+ struct attribute *attr,
+ const char *buf, size_t len)
+{
+ struct gb_audio_manager_module_attribute *attribute;
+ struct gb_audio_manager_module *module;
+
+ attribute = to_gb_audio_module_attr(attr);
+ module = to_gb_audio_module(kobj);
+
+ if (!attribute->store)
+ return -EIO;
+
+ return attribute->store(module, attribute, buf, len);
+}
+
+static const struct sysfs_ops gb_audio_module_sysfs_ops = {
+ .show = gb_audio_module_attr_show,
+ .store = gb_audio_module_attr_store,
+};
+
+static void gb_audio_module_release(struct kobject *kobj)
+{
+ struct gb_audio_manager_module *module = to_gb_audio_module(kobj);
+
+ pr_info("Destroying audio module #%d\n", module->id);
+ /* TODO -> delete from list */
+ kfree(module);
+}
+
+static ssize_t gb_audio_module_name_show(
+ struct gb_audio_manager_module *module,
+ struct gb_audio_manager_module_attribute *attr, char *buf)
+{
+ return sprintf(buf, "%s", module->desc.name);
+}
+
+static struct gb_audio_manager_module_attribute gb_audio_module_name_attribute =
+ __ATTR(name, 0664, gb_audio_module_name_show, NULL);
+
+static ssize_t gb_audio_module_slot_show(
+ struct gb_audio_manager_module *module,
+ struct gb_audio_manager_module_attribute *attr, char *buf)
+{
+ return sprintf(buf, "%d", module->desc.slot);
+}
+
+static struct gb_audio_manager_module_attribute gb_audio_module_slot_attribute =
+ __ATTR(slot, 0664, gb_audio_module_slot_show, NULL);
+
+static ssize_t gb_audio_module_vid_show(
+ struct gb_audio_manager_module *module,
+ struct gb_audio_manager_module_attribute *attr, char *buf)
+{
+ return sprintf(buf, "%d", module->desc.vid);
+}
+
+static struct gb_audio_manager_module_attribute gb_audio_module_vid_attribute =
+ __ATTR(vid, 0664, gb_audio_module_vid_show, NULL);
+
+static ssize_t gb_audio_module_pid_show(
+ struct gb_audio_manager_module *module,
+ struct gb_audio_manager_module_attribute *attr, char *buf)
+{
+ return sprintf(buf, "%d", module->desc.pid);
+}
+
+static struct gb_audio_manager_module_attribute gb_audio_module_pid_attribute =
+ __ATTR(pid, 0664, gb_audio_module_pid_show, NULL);
+
+static ssize_t gb_audio_module_cport_show(
+ struct gb_audio_manager_module *module,
+ struct gb_audio_manager_module_attribute *attr, char *buf)
+{
+ return sprintf(buf, "%d", module->desc.cport);
+}
+
+static struct gb_audio_manager_module_attribute
+ gb_audio_module_cport_attribute =
+ __ATTR(cport, 0664, gb_audio_module_cport_show, NULL);
+
+static ssize_t gb_audio_module_ip_devices_show(
+ struct gb_audio_manager_module *module,
+ struct gb_audio_manager_module_attribute *attr, char *buf)
+{
+ return sprintf(buf, "0x%X", module->desc.ip_devices);
+}
+
+static struct gb_audio_manager_module_attribute
+ gb_audio_module_ip_devices_attribute =
+ __ATTR(ip_devices, 0664, gb_audio_module_ip_devices_show, NULL);
+
+static ssize_t gb_audio_module_op_devices_show(
+ struct gb_audio_manager_module *module,
+ struct gb_audio_manager_module_attribute *attr, char *buf)
+{
+ return sprintf(buf, "0x%X", module->desc.op_devices);
+}
+
+static struct gb_audio_manager_module_attribute
+ gb_audio_module_op_devices_attribute =
+ __ATTR(op_devices, 0664, gb_audio_module_op_devices_show, NULL);
+
+static struct attribute *gb_audio_module_default_attrs[] = {
+ &gb_audio_module_name_attribute.attr,
+ &gb_audio_module_slot_attribute.attr,
+ &gb_audio_module_vid_attribute.attr,
+ &gb_audio_module_pid_attribute.attr,
+ &gb_audio_module_cport_attribute.attr,
+ &gb_audio_module_ip_devices_attribute.attr,
+ &gb_audio_module_op_devices_attribute.attr,
+ NULL, /* need to NULL terminate the list of attributes */
+};
+
+static struct kobj_type gb_audio_module_type = {
+ .sysfs_ops = &gb_audio_module_sysfs_ops,
+ .release = gb_audio_module_release,
+ .default_attrs = gb_audio_module_default_attrs,
+};
+
+static void send_add_uevent(struct gb_audio_manager_module *module)
+{
+ char name_string[128];
+ char slot_string[64];
+ char vid_string[64];
+ char pid_string[64];
+ char cport_string[64];
+ char ip_devices_string[64];
+ char op_devices_string[64];
+
+ char *envp[] = {
+ name_string,
+ slot_string,
+ vid_string,
+ pid_string,
+ cport_string,
+ ip_devices_string,
+ op_devices_string,
+ NULL
+ };
+
+ snprintf(name_string, 128, "NAME=%s", module->desc.name);
+ snprintf(slot_string, 64, "SLOT=%d", module->desc.slot);
+ snprintf(vid_string, 64, "VID=%d", module->desc.vid);
+ snprintf(pid_string, 64, "PID=%d", module->desc.pid);
+ snprintf(cport_string, 64, "CPORT=%d", module->desc.cport);
+ snprintf(ip_devices_string, 64, "I/P DEVICES=0x%X",
+ module->desc.ip_devices);
+ snprintf(op_devices_string, 64, "O/P DEVICES=0x%X",
+ module->desc.op_devices);
+
+ kobject_uevent_env(&module->kobj, KOBJ_ADD, envp);
+}
+
+int gb_audio_manager_module_create(
+ struct gb_audio_manager_module **module,
+ struct kset *manager_kset,
+ int id, struct gb_audio_manager_module_descriptor *desc)
+{
+ int err;
+ struct gb_audio_manager_module *m;
+
+ m = kzalloc(sizeof(*m), GFP_ATOMIC);
+ if (!m)
+ return -ENOMEM;
+
+ /* Initialize the node */
+ INIT_LIST_HEAD(&m->list);
+
+ /* Set the module id */
+ m->id = id;
+
+ /* Copy the provided descriptor */
+ memcpy(&m->desc, desc, sizeof(*desc));
+
+ /* set the kset */
+ m->kobj.kset = manager_kset;
+
+ /*
+ * Initialize and add the kobject to the kernel. All the default files
+ * will be created here. As we have already specified a kset for this
+ * kobject, we don't have to set a parent for the kobject, the kobject
+ * will be placed beneath that kset automatically.
+ */
+ err = kobject_init_and_add(&m->kobj, &gb_audio_module_type, NULL, "%d",
+ id);
+ if (err) {
+ pr_err("failed initializing kobject for audio module #%d\n",
+ id);
+ kobject_put(&m->kobj);
+ return err;
+ }
+
+ /*
+ * Notify the object was created
+ */
+ send_add_uevent(m);
+
+ *module = m;
+ pr_info("Created audio module #%d\n", id);
+ return 0;
+}
+
+void gb_audio_manager_module_dump(struct gb_audio_manager_module *module)
+{
+ pr_info("audio module #%d name=%s slot=%d vid=%d pid=%d cport=%d i/p devices=0x%X o/p devices=0x%X\n",
+ module->id,
+ module->desc.name,
+ module->desc.slot,
+ module->desc.vid,
+ module->desc.pid,
+ module->desc.cport,
+ module->desc.ip_devices,
+ module->desc.op_devices);
+}
--- /dev/null
+++ b/drivers/greybus/audio_manager_private.h
@@ -0,0 +1,28 @@
+/*
+ * Greybus operations
+ *
+ * Copyright 2015-2016 Google Inc.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef _GB_AUDIO_MANAGER_PRIVATE_H_
+#define _GB_AUDIO_MANAGER_PRIVATE_H_
+
+#include <linux/kobject.h>
+
+#include "audio_manager.h"
+
+int gb_audio_manager_module_create(
+ struct gb_audio_manager_module **module,
+ struct kset *manager_kset,
+ int id, struct gb_audio_manager_module_descriptor *desc);
+
+/* module destroyed via kobject_put */
+
+void gb_audio_manager_module_dump(struct gb_audio_manager_module *module);
+
+/* sysfs control */
+void gb_audio_manager_sysfs_init(struct kobject *kobj);
+
+#endif /* _GB_AUDIO_MANAGER_PRIVATE_H_ */
--- /dev/null
+++ b/drivers/greybus/audio_manager_sysfs.c
@@ -0,0 +1,102 @@
+/*
+ * Greybus operations
+ *
+ * Copyright 2015-2016 Google Inc.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/string.h>
+#include <linux/sysfs.h>
+
+#include "audio_manager.h"
+#include "audio_manager_private.h"
+
+static ssize_t manager_sysfs_add_store(
+ struct kobject *kobj, struct kobj_attribute *attr,
+ const char *buf, size_t count)
+{
+ struct gb_audio_manager_module_descriptor desc = { {0} };
+
+ int num = sscanf(buf,
+ "name=%" GB_AUDIO_MANAGER_MODULE_NAME_LEN_SSCANF "s "
+ "slot=%d vid=%d pid=%d cport=%d i/p devices=0x%X"
+ "o/p devices=0x%X",
+ desc.name, &desc.slot, &desc.vid, &desc.pid,
+ &desc.cport, &desc.ip_devices, &desc.op_devices);
+
+ if (num != 7)
+ return -EINVAL;
+
+ num = gb_audio_manager_add(&desc);
+ if (num < 0)
+ return -EINVAL;
+
+ return count;
+}
+
+static struct kobj_attribute manager_add_attribute =
+ __ATTR(add, 0664, NULL, manager_sysfs_add_store);
+
+static ssize_t manager_sysfs_remove_store(
+ struct kobject *kobj, struct kobj_attribute *attr,
+ const char *buf, size_t count)
+{
+ int id;
+
+ int num = sscanf(buf, "%d", &id);
+
+ if (num != 1)
+ return -EINVAL;
+
+ num = gb_audio_manager_remove(id);
+ if (num)
+ return num;
+
+ return count;
+}
+
+static struct kobj_attribute manager_remove_attribute =
+ __ATTR(remove, 0664, NULL, manager_sysfs_remove_store);
+
+static ssize_t manager_sysfs_dump_store(
+ struct kobject *kobj, struct kobj_attribute *attr,
+ const char *buf, size_t count)
+{
+ int id;
+
+ int num = sscanf(buf, "%d", &id);
+
+ if (num == 1) {
+ num = gb_audio_manager_dump_module(id);
+ if (num)
+ return num;
+ } else if (!strncmp("all", buf, 3))
+ gb_audio_manager_dump_all();
+ else
+ return -EINVAL;
+
+ return count;
+}
+
+static struct kobj_attribute manager_dump_attribute =
+ __ATTR(dump, 0664, NULL, manager_sysfs_dump_store);
+
+static void manager_sysfs_init_attribute(
+ struct kobject *kobj, struct kobj_attribute *kattr)
+{
+ int err;
+
+ err = sysfs_create_file(kobj, &kattr->attr);
+ if (err) {
+ pr_warn("creating the sysfs entry for %s failed: %d\n",
+ kattr->attr.name, err);
+ }
+}
+
+void gb_audio_manager_sysfs_init(struct kobject *kobj)
+{
+ manager_sysfs_init_attribute(kobj, &manager_add_attribute);
+ manager_sysfs_init_attribute(kobj, &manager_remove_attribute);
+ manager_sysfs_init_attribute(kobj, &manager_dump_attribute);
+}
--- /dev/null
+++ b/drivers/greybus/audio_module.c
@@ -0,0 +1,482 @@
+/*
+ * Greybus audio driver
+ * Copyright 2015 Google Inc.
+ * Copyright 2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <sound/soc.h>
+#include <sound/pcm_params.h>
+
+#include "audio_codec.h"
+#include "audio_apbridgea.h"
+#include "audio_manager.h"
+
+/*
+ * gb_snd management functions
+ */
+
+static int gbaudio_request_jack(struct gbaudio_module_info *module,
+ struct gb_audio_jack_event_request *req)
+{
+ int report;
+ struct snd_jack *jack = module->headset_jack.jack;
+ struct snd_jack *btn_jack = module->button_jack.jack;
+
+ if (!jack) {
+ dev_err_ratelimited(module->dev,
+ "Invalid jack event received:type: %u, event: %u\n",
+ req->jack_attribute, req->event);
+ return -EINVAL;
+ }
+
+ dev_warn_ratelimited(module->dev,
+ "Jack Event received: type: %u, event: %u\n",
+ req->jack_attribute, req->event);
+
+ if (req->event == GB_AUDIO_JACK_EVENT_REMOVAL) {
+ module->jack_type = 0;
+ if (btn_jack && module->button_status) {
+ snd_soc_jack_report(&module->button_jack, 0,
+ module->button_mask);
+ module->button_status = 0;
+ }
+ snd_soc_jack_report(&module->headset_jack, 0,
+ module->jack_mask);
+ return 0;
+ }
+
+ report = req->jack_attribute & module->jack_mask;
+ if (!report) {
+ dev_err_ratelimited(module->dev,
+ "Invalid jack event received:type: %u, event: %u\n",
+ req->jack_attribute, req->event);
+ return -EINVAL;
+ }
+
+ if (module->jack_type)
+ dev_warn_ratelimited(module->dev,
+ "Modifying jack from %d to %d\n",
+ module->jack_type, report);
+
+ module->jack_type = report;
+ snd_soc_jack_report(&module->headset_jack, report, module->jack_mask);
+
+ return 0;
+}
+
+static int gbaudio_request_button(struct gbaudio_module_info *module,
+ struct gb_audio_button_event_request *req)
+{
+ int soc_button_id, report;
+ struct snd_jack *btn_jack = module->button_jack.jack;
+
+ if (!btn_jack) {
+ dev_err_ratelimited(module->dev,
+ "Invalid button event received:type: %u, event: %u\n",
+ req->button_id, req->event);
+ return -EINVAL;
+ }
+
+ dev_warn_ratelimited(module->dev,
+ "Button Event received: id: %u, event: %u\n",
+ req->button_id, req->event);
+
+ /* currently supports 4 buttons only */
+ if (!module->jack_type) {
+ dev_err_ratelimited(module->dev,
+ "Jack not present. Bogus event!!\n");
+ return -EINVAL;
+ }
+
+ report = module->button_status & module->button_mask;
+ soc_button_id = 0;
+
+ switch (req->button_id) {
+ case 1:
+ soc_button_id = SND_JACK_BTN_0 & module->button_mask;
+ break;
+
+ case 2:
+ soc_button_id = SND_JACK_BTN_1 & module->button_mask;
+ break;
+
+ case 3:
+ soc_button_id = SND_JACK_BTN_2 & module->button_mask;
+ break;
+
+ case 4:
+ soc_button_id = SND_JACK_BTN_3 & module->button_mask;
+ break;
+ }
+
+ if (!soc_button_id) {
+ dev_err_ratelimited(module->dev,
+ "Invalid button request received\n");
+ return -EINVAL;
+ }
+
+ if (req->event == GB_AUDIO_BUTTON_EVENT_PRESS)
+ report = report | soc_button_id;
+ else
+ report = report & ~soc_button_id;
+
+ module->button_status = report;
+
+ snd_soc_jack_report(&module->button_jack, report, module->button_mask);
+
+ return 0;
+}
+
+static int gbaudio_request_stream(struct gbaudio_module_info *module,
+ struct gb_audio_streaming_event_request *req)
+{
+ dev_warn(module->dev, "Audio Event received: cport: %u, event: %u\n",
+ req->data_cport, req->event);
+
+ return 0;
+}
+
+static int gbaudio_codec_request_handler(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct gbaudio_module_info *module =
+ greybus_get_drvdata(connection->bundle);
+ struct gb_operation_msg_hdr *header = op->request->header;
+ struct gb_audio_streaming_event_request *stream_req;
+ struct gb_audio_jack_event_request *jack_req;
+ struct gb_audio_button_event_request *button_req;
+ int ret;
+
+ switch (header->type) {
+ case GB_AUDIO_TYPE_STREAMING_EVENT:
+ stream_req = op->request->payload;
+ ret = gbaudio_request_stream(module, stream_req);
+ break;
+
+ case GB_AUDIO_TYPE_JACK_EVENT:
+ jack_req = op->request->payload;
+ ret = gbaudio_request_jack(module, jack_req);
+ break;
+
+ case GB_AUDIO_TYPE_BUTTON_EVENT:
+ button_req = op->request->payload;
+ ret = gbaudio_request_button(module, button_req);
+ break;
+
+ default:
+ dev_err_ratelimited(&connection->bundle->dev,
+ "Invalid Audio Event received\n");
+ return -EINVAL;
+ }
+
+ return ret;
+}
+
+static int gb_audio_add_mgmt_connection(struct gbaudio_module_info *gbmodule,
+ struct greybus_descriptor_cport *cport_desc,
+ struct gb_bundle *bundle)
+{
+ struct gb_connection *connection;
+
+ /* Management Cport */
+ if (gbmodule->mgmt_connection) {
+ dev_err(&bundle->dev,
+ "Can't have multiple Management connections\n");
+ return -ENODEV;
+ }
+
+ connection = gb_connection_create(bundle, le16_to_cpu(cport_desc->id),
+ gbaudio_codec_request_handler);
+ if (IS_ERR(connection))
+ return PTR_ERR(connection);
+
+ greybus_set_drvdata(bundle, gbmodule);
+ gbmodule->mgmt_connection = connection;
+
+ return 0;
+}
+
+static int gb_audio_add_data_connection(struct gbaudio_module_info *gbmodule,
+ struct greybus_descriptor_cport *cport_desc,
+ struct gb_bundle *bundle)
+{
+ struct gb_connection *connection;
+ struct gbaudio_data_connection *dai;
+
+ dai = devm_kzalloc(gbmodule->dev, sizeof(*dai), GFP_KERNEL);
+ if (!dai) {
+ dev_err(gbmodule->dev, "DAI Malloc failure\n");
+ return -ENOMEM;
+ }
+
+ connection = gb_connection_create_offloaded(bundle,
+ le16_to_cpu(cport_desc->id),
+ GB_CONNECTION_FLAG_CSD);
+ if (IS_ERR(connection)) {
+ devm_kfree(gbmodule->dev, dai);
+ return PTR_ERR(connection);
+ }
+
+ greybus_set_drvdata(bundle, gbmodule);
+ dai->id = 0;
+ dai->data_cport = connection->intf_cport_id;
+ dai->connection = connection;
+ list_add(&dai->list, &gbmodule->data_list);
+
+ return 0;
+}
+
+/*
+ * This is the basic hook get things initialized and registered w/ gb
+ */
+
+static int gb_audio_probe(struct gb_bundle *bundle,
+ const struct greybus_bundle_id *id)
+{
+ struct device *dev = &bundle->dev;
+ struct gbaudio_module_info *gbmodule;
+ struct greybus_descriptor_cport *cport_desc;
+ struct gb_audio_manager_module_descriptor desc;
+ struct gbaudio_data_connection *dai, *_dai;
+ int ret, i;
+ struct gb_audio_topology *topology;
+
+ /* There should be at least one Management and one Data cport */
+ if (bundle->num_cports < 2)
+ return -ENODEV;
+
+ /*
+ * There can be only one Management connection and any number of data
+ * connections.
+ */
+ gbmodule = devm_kzalloc(dev, sizeof(*gbmodule), GFP_KERNEL);
+ if (!gbmodule)
+ return -ENOMEM;
+
+ gbmodule->num_data_connections = bundle->num_cports - 1;
+ INIT_LIST_HEAD(&gbmodule->data_list);
+ INIT_LIST_HEAD(&gbmodule->widget_list);
+ INIT_LIST_HEAD(&gbmodule->ctl_list);
+ INIT_LIST_HEAD(&gbmodule->widget_ctl_list);
+ gbmodule->dev = dev;
+ snprintf(gbmodule->name, NAME_SIZE, "%s.%s", dev->driver->name,
+ dev_name(dev));
+ greybus_set_drvdata(bundle, gbmodule);
+
+ /* Create all connections */
+ for (i = 0; i < bundle->num_cports; i++) {
+ cport_desc = &bundle->cport_desc[i];
+
+ switch (cport_desc->protocol_id) {
+ case GREYBUS_PROTOCOL_AUDIO_MGMT:
+ ret = gb_audio_add_mgmt_connection(gbmodule, cport_desc,
+ bundle);
+ if (ret)
+ goto destroy_connections;
+ break;
+ case GREYBUS_PROTOCOL_AUDIO_DATA:
+ ret = gb_audio_add_data_connection(gbmodule, cport_desc,
+ bundle);
+ if (ret)
+ goto destroy_connections;
+ break;
+ default:
+ dev_err(dev, "Unsupported protocol: 0x%02x\n",
+ cport_desc->protocol_id);
+ ret = -ENODEV;
+ goto destroy_connections;
+ }
+ }
+
+ /* There must be a management cport */
+ if (!gbmodule->mgmt_connection) {
+ ret = -EINVAL;
+ dev_err(dev, "Missing management connection\n");
+ goto destroy_connections;
+ }
+
+ /* Initialize management connection */
+ ret = gb_connection_enable(gbmodule->mgmt_connection);
+ if (ret) {
+ dev_err(dev, "%d: Error while enabling mgmt connection\n", ret);
+ goto destroy_connections;
+ }
+ gbmodule->dev_id = gbmodule->mgmt_connection->intf->interface_id;
+
+ /*
+ * FIXME: malloc for topology happens via audio_gb driver
+ * should be done within codec driver itself
+ */
+ ret = gb_audio_gb_get_topology(gbmodule->mgmt_connection, &topology);
+ if (ret) {
+ dev_err(dev, "%d:Error while fetching topology\n", ret);
+ goto disable_connection;
+ }
+
+ /* process topology data */
+ ret = gbaudio_tplg_parse_data(gbmodule, topology);
+ if (ret) {
+ dev_err(dev, "%d:Error while parsing topology data\n",
+ ret);
+ goto free_topology;
+ }
+ gbmodule->topology = topology;
+
+ /* Initialize data connections */
+ list_for_each_entry(dai, &gbmodule->data_list, list) {
+ ret = gb_connection_enable(dai->connection);
+ if (ret) {
+ dev_err(dev,
+ "%d:Error while enabling %d:data connection\n",
+ ret, dai->data_cport);
+ goto disable_data_connection;
+ }
+ }
+
+ /* register module with gbcodec */
+ ret = gbaudio_register_module(gbmodule);
+ if (ret)
+ goto disable_data_connection;
+
+ /* inform above layer for uevent */
+ dev_dbg(dev, "Inform set_event:%d to above layer\n", 1);
+ /* prepare for the audio manager */
+ strlcpy(desc.name, gbmodule->name, GB_AUDIO_MANAGER_MODULE_NAME_LEN);
+ desc.slot = 1; /* todo */
+ desc.vid = 2; /* todo */
+ desc.pid = 3; /* todo */
+ desc.cport = gbmodule->dev_id;
+ desc.op_devices = gbmodule->op_devices;
+ desc.ip_devices = gbmodule->ip_devices;
+ gbmodule->manager_id = gb_audio_manager_add(&desc);
+
+ dev_dbg(dev, "Add GB Audio device:%s\n", gbmodule->name);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return 0;
+
+disable_data_connection:
+ list_for_each_entry_safe(dai, _dai, &gbmodule->data_list, list)
+ gb_connection_disable(dai->connection);
+ gbaudio_tplg_release(gbmodule);
+ gbmodule->topology = NULL;
+
+free_topology:
+ kfree(topology);
+
+disable_connection:
+ gb_connection_disable(gbmodule->mgmt_connection);
+
+destroy_connections:
+ list_for_each_entry_safe(dai, _dai, &gbmodule->data_list, list) {
+ gb_connection_destroy(dai->connection);
+ list_del(&dai->list);
+ devm_kfree(dev, dai);
+ }
+
+ if (gbmodule->mgmt_connection)
+ gb_connection_destroy(gbmodule->mgmt_connection);
+
+ devm_kfree(dev, gbmodule);
+
+ return ret;
+}
+
+static void gb_audio_disconnect(struct gb_bundle *bundle)
+{
+ struct gbaudio_module_info *gbmodule = greybus_get_drvdata(bundle);
+ struct gbaudio_data_connection *dai, *_dai;
+
+ gb_pm_runtime_get_sync(bundle);
+
+ /* cleanup module related resources first */
+ gbaudio_unregister_module(gbmodule);
+
+ /* inform uevent to above layers */
+ gb_audio_manager_remove(gbmodule->manager_id);
+
+ gbaudio_tplg_release(gbmodule);
+ kfree(gbmodule->topology);
+ gbmodule->topology = NULL;
+ gb_connection_disable(gbmodule->mgmt_connection);
+ list_for_each_entry_safe(dai, _dai, &gbmodule->data_list, list) {
+ gb_connection_disable(dai->connection);
+ gb_connection_destroy(dai->connection);
+ list_del(&dai->list);
+ devm_kfree(gbmodule->dev, dai);
+ }
+ gb_connection_destroy(gbmodule->mgmt_connection);
+ gbmodule->mgmt_connection = NULL;
+
+ devm_kfree(&bundle->dev, gbmodule);
+}
+
+static const struct greybus_bundle_id gb_audio_id_table[] = {
+ { GREYBUS_DEVICE_CLASS(GREYBUS_CLASS_AUDIO) },
+ { }
+};
+MODULE_DEVICE_TABLE(greybus, gb_audio_id_table);
+
+#ifdef CONFIG_PM
+static int gb_audio_suspend(struct device *dev)
+{
+ struct gb_bundle *bundle = to_gb_bundle(dev);
+ struct gbaudio_module_info *gbmodule = greybus_get_drvdata(bundle);
+ struct gbaudio_data_connection *dai;
+
+ list_for_each_entry(dai, &gbmodule->data_list, list)
+ gb_connection_disable(dai->connection);
+
+ gb_connection_disable(gbmodule->mgmt_connection);
+
+ return 0;
+}
+
+static int gb_audio_resume(struct device *dev)
+{
+ struct gb_bundle *bundle = to_gb_bundle(dev);
+ struct gbaudio_module_info *gbmodule = greybus_get_drvdata(bundle);
+ struct gbaudio_data_connection *dai;
+ int ret;
+
+ ret = gb_connection_enable(gbmodule->mgmt_connection);
+ if (ret) {
+ dev_err(dev, "%d:Error while enabling mgmt connection\n", ret);
+ return ret;
+ }
+
+ list_for_each_entry(dai, &gbmodule->data_list, list) {
+ ret = gb_connection_enable(dai->connection);
+ if (ret) {
+ dev_err(dev,
+ "%d:Error while enabling %d:data connection\n",
+ ret, dai->data_cport);
+ return ret;
+ }
+ }
+
+ return 0;
+}
+#endif
+
+static const struct dev_pm_ops gb_audio_pm_ops = {
+ SET_RUNTIME_PM_OPS(gb_audio_suspend, gb_audio_resume, NULL)
+};
+
+static struct greybus_driver gb_audio_driver = {
+ .name = "gb-audio",
+ .probe = gb_audio_probe,
+ .disconnect = gb_audio_disconnect,
+ .id_table = gb_audio_id_table,
+ .driver.pm = &gb_audio_pm_ops,
+};
+module_greybus_driver(gb_audio_driver);
+
+MODULE_DESCRIPTION("Greybus Audio module driver");
+MODULE_AUTHOR("Vaibhav Agarwal <vaibhav.agarwal@linaro.org>");
+MODULE_LICENSE("GPL v2");
+MODULE_ALIAS("platform:gbaudio-module");
--- /dev/null
+++ b/drivers/greybus/audio_topology.c
@@ -0,0 +1,1442 @@
+/*
+ * Greybus audio driver
+ * Copyright 2015-2016 Google Inc.
+ * Copyright 2015-2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include "audio_codec.h"
+#include "greybus_protocols.h"
+
+#define GBAUDIO_INVALID_ID 0xFF
+
+/* mixer control */
+struct gb_mixer_control {
+ int min, max;
+ unsigned int reg, rreg, shift, rshift, invert;
+};
+
+struct gbaudio_ctl_pvt {
+ unsigned int ctl_id;
+ unsigned int data_cport;
+ unsigned int access;
+ unsigned int vcount;
+ struct gb_audio_ctl_elem_info *info;
+};
+
+static struct gbaudio_module_info *find_gb_module(
+ struct gbaudio_codec_info *codec,
+ char const *name)
+{
+ int dev_id, ret;
+ char begin[NAME_SIZE];
+ struct gbaudio_module_info *module;
+
+ if (!name)
+ return NULL;
+
+ ret = sscanf(name, "%s %d", begin, &dev_id);
+ dev_dbg(codec->dev, "%s:Find module#%d\n", __func__, dev_id);
+
+ mutex_lock(&codec->lock);
+ list_for_each_entry(module, &codec->module_list, list) {
+ if (module->dev_id == dev_id) {
+ mutex_unlock(&codec->lock);
+ return module;
+ }
+ }
+ mutex_unlock(&codec->lock);
+ dev_warn(codec->dev, "%s: module#%d missing in codec list\n", name,
+ dev_id);
+ return NULL;
+}
+
+static const char *gbaudio_map_controlid(struct gbaudio_module_info *module,
+ __u8 control_id, __u8 index)
+{
+ struct gbaudio_control *control;
+
+ if (control_id == GBAUDIO_INVALID_ID)
+ return NULL;
+
+ list_for_each_entry(control, &module->ctl_list, list) {
+ if (control->id == control_id) {
+ if (index == GBAUDIO_INVALID_ID)
+ return control->name;
+ if (index >= control->items)
+ return NULL;
+ return control->texts[index];
+ }
+ }
+ list_for_each_entry(control, &module->widget_ctl_list, list) {
+ if (control->id == control_id) {
+ if (index == GBAUDIO_INVALID_ID)
+ return control->name;
+ if (index >= control->items)
+ return NULL;
+ return control->texts[index];
+ }
+ }
+ return NULL;
+}
+
+static int gbaudio_map_controlname(struct gbaudio_module_info *module,
+ const char *name)
+{
+ struct gbaudio_control *control;
+
+ list_for_each_entry(control, &module->ctl_list, list) {
+ if (!strncmp(control->name, name, NAME_SIZE))
+ return control->id;
+ }
+
+ dev_warn(module->dev, "%s: missing in modules controls list\n", name);
+
+ return -EINVAL;
+}
+
+static int gbaudio_map_wcontrolname(struct gbaudio_module_info *module,
+ const char *name)
+{
+ struct gbaudio_control *control;
+
+ list_for_each_entry(control, &module->widget_ctl_list, list) {
+ if (!strncmp(control->wname, name, NAME_SIZE))
+ return control->id;
+ }
+ dev_warn(module->dev, "%s: missing in modules controls list\n", name);
+
+ return -EINVAL;
+}
+
+static int gbaudio_map_widgetname(struct gbaudio_module_info *module,
+ const char *name)
+{
+ struct gbaudio_widget *widget;
+ list_for_each_entry(widget, &module->widget_list, list) {
+ if (!strncmp(widget->name, name, NAME_SIZE))
+ return widget->id;
+ }
+ dev_warn(module->dev, "%s: missing in modules widgets list\n", name);
+
+ return -EINVAL;
+}
+
+static const char *gbaudio_map_widgetid(struct gbaudio_module_info *module,
+ __u8 widget_id)
+{
+ struct gbaudio_widget *widget;
+
+ list_for_each_entry(widget, &module->widget_list, list) {
+ if (widget->id == widget_id)
+ return widget->name;
+ }
+ return NULL;
+}
+
+static const char **gb_generate_enum_strings(struct gbaudio_module_info *gb,
+ struct gb_audio_enumerated *gbenum)
+{
+ const char **strings;
+ int i;
+ __u8 *data;
+
+ strings = devm_kzalloc(gb->dev, sizeof(char *) * gbenum->items,
+ GFP_KERNEL);
+ data = gbenum->names;
+
+ for (i = 0; i < gbenum->items; i++) {
+ strings[i] = (const char *)data;
+ while (*data != '\0')
+ data++;
+ data++;
+ }
+
+ return strings;
+}
+
+static int gbcodec_mixer_ctl_info(struct snd_kcontrol *kcontrol,
+ struct snd_ctl_elem_info *uinfo)
+{
+ unsigned int max;
+ const char *name;
+ struct gbaudio_ctl_pvt *data;
+ struct gb_audio_ctl_elem_info *info;
+ struct gbaudio_module_info *module;
+ struct snd_soc_codec *codec = snd_kcontrol_chip(kcontrol);
+ struct gbaudio_codec_info *gbcodec = snd_soc_codec_get_drvdata(codec);
+
+ dev_dbg(codec->dev, "Entered %s:%s\n", __func__, kcontrol->id.name);
+ data = (struct gbaudio_ctl_pvt *)kcontrol->private_value;
+ info = (struct gb_audio_ctl_elem_info *)data->info;
+
+ if (!info) {
+ dev_err(module->dev, "NULL info for %s\n", uinfo->id.name);
+ return -EINVAL;
+ }
+
+ /* update uinfo */
+ uinfo->access = data->access;
+ uinfo->count = data->vcount;
+ uinfo->type = (snd_ctl_elem_type_t)info->type;
+
+ switch (info->type) {
+ case GB_AUDIO_CTL_ELEM_TYPE_BOOLEAN:
+ case GB_AUDIO_CTL_ELEM_TYPE_INTEGER:
+ uinfo->value.integer.min = info->value.integer.min;
+ uinfo->value.integer.max = info->value.integer.max;
+ break;
+ case GB_AUDIO_CTL_ELEM_TYPE_ENUMERATED:
+ max = info->value.enumerated.items;
+ uinfo->value.enumerated.items = max;
+ if (uinfo->value.enumerated.item > max - 1)
+ uinfo->value.enumerated.item = max - 1;
+ module = find_gb_module(gbcodec, kcontrol->id.name);
+ if (!module)
+ return -EINVAL;
+ name = gbaudio_map_controlid(module, data->ctl_id,
+ uinfo->value.enumerated.item);
+ strlcpy(uinfo->value.enumerated.name, name, NAME_SIZE);
+ break;
+ default:
+ dev_err(codec->dev, "Invalid type: %d for %s:kcontrol\n",
+ info->type, kcontrol->id.name);
+ break;
+ }
+ return 0;
+}
+
+static int gbcodec_mixer_ctl_get(struct snd_kcontrol *kcontrol,
+ struct snd_ctl_elem_value *ucontrol)
+{
+ int ret;
+ struct gb_audio_ctl_elem_info *info;
+ struct gbaudio_ctl_pvt *data;
+ struct gb_audio_ctl_elem_value gbvalue;
+ struct gbaudio_module_info *module;
+ struct snd_soc_codec *codec = snd_kcontrol_chip(kcontrol);
+ struct gbaudio_codec_info *gb = snd_soc_codec_get_drvdata(codec);
+ struct gb_bundle *bundle;
+
+ dev_dbg(codec->dev, "Entered %s:%s\n", __func__, kcontrol->id.name);
+ module = find_gb_module(gb, kcontrol->id.name);
+ if (!module)
+ return -EINVAL;
+
+ data = (struct gbaudio_ctl_pvt *)kcontrol->private_value;
+ info = (struct gb_audio_ctl_elem_info *)data->info;
+ bundle = to_gb_bundle(module->dev);
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ return ret;
+
+ ret = gb_audio_gb_get_control(module->mgmt_connection, data->ctl_id,
+ GB_AUDIO_INVALID_INDEX, &gbvalue);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ if (ret) {
+ dev_err_ratelimited(codec->dev, "%d:Error in %s for %s\n", ret,
+ __func__, kcontrol->id.name);
+ return ret;
+ }
+
+ /* update ucontrol */
+ switch (info->type) {
+ case GB_AUDIO_CTL_ELEM_TYPE_BOOLEAN:
+ case GB_AUDIO_CTL_ELEM_TYPE_INTEGER:
+ ucontrol->value.integer.value[0] =
+ gbvalue.value.integer_value[0];
+ if (data->vcount == 2)
+ ucontrol->value.integer.value[1] =
+ gbvalue.value.integer_value[1];
+ break;
+ case GB_AUDIO_CTL_ELEM_TYPE_ENUMERATED:
+ ucontrol->value.enumerated.item[0] =
+ gbvalue.value.enumerated_item[0];
+ if (data->vcount == 2)
+ ucontrol->value.enumerated.item[1] =
+ gbvalue.value.enumerated_item[1];
+ break;
+ default:
+ dev_err(codec->dev, "Invalid type: %d for %s:kcontrol\n",
+ info->type, kcontrol->id.name);
+ ret = -EINVAL;
+ break;
+ }
+ return ret;
+}
+
+static int gbcodec_mixer_ctl_put(struct snd_kcontrol *kcontrol,
+ struct snd_ctl_elem_value *ucontrol)
+{
+ int ret = 0;
+ struct gb_audio_ctl_elem_info *info;
+ struct gbaudio_ctl_pvt *data;
+ struct gb_audio_ctl_elem_value gbvalue;
+ struct gbaudio_module_info *module;
+ struct snd_soc_codec *codec = snd_kcontrol_chip(kcontrol);
+ struct gbaudio_codec_info *gb = snd_soc_codec_get_drvdata(codec);
+ struct gb_bundle *bundle;
+
+ dev_dbg(codec->dev, "Entered %s:%s\n", __func__, kcontrol->id.name);
+ module = find_gb_module(gb, kcontrol->id.name);
+ if (!module)
+ return -EINVAL;
+
+ data = (struct gbaudio_ctl_pvt *)kcontrol->private_value;
+ info = (struct gb_audio_ctl_elem_info *)data->info;
+ bundle = to_gb_bundle(module->dev);
+
+ /* update ucontrol */
+ switch (info->type) {
+ case GB_AUDIO_CTL_ELEM_TYPE_BOOLEAN:
+ case GB_AUDIO_CTL_ELEM_TYPE_INTEGER:
+ gbvalue.value.integer_value[0] =
+ ucontrol->value.integer.value[0];
+ if (data->vcount == 2)
+ gbvalue.value.integer_value[1] =
+ ucontrol->value.integer.value[1];
+ break;
+ case GB_AUDIO_CTL_ELEM_TYPE_ENUMERATED:
+ gbvalue.value.enumerated_item[0] =
+ ucontrol->value.enumerated.item[0];
+ if (data->vcount == 2)
+ gbvalue.value.enumerated_item[1] =
+ ucontrol->value.enumerated.item[1];
+ break;
+ default:
+ dev_err(codec->dev, "Invalid type: %d for %s:kcontrol\n",
+ info->type, kcontrol->id.name);
+ ret = -EINVAL;
+ break;
+ }
+
+ if (ret)
+ return ret;
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ return ret;
+
+ ret = gb_audio_gb_set_control(module->mgmt_connection, data->ctl_id,
+ GB_AUDIO_INVALID_INDEX, &gbvalue);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ if (ret) {
+ dev_err_ratelimited(codec->dev, "%d:Error in %s for %s\n", ret,
+ __func__, kcontrol->id.name);
+ }
+
+ return ret;
+}
+
+#define SOC_MIXER_GB(xname, kcount, data) \
+{ .iface = SNDRV_CTL_ELEM_IFACE_MIXER, .name = xname, \
+ .count = kcount, .info = gbcodec_mixer_ctl_info, \
+ .get = gbcodec_mixer_ctl_get, .put = gbcodec_mixer_ctl_put, \
+ .private_value = (unsigned long)data }
+
+/*
+ * although below callback functions seems redundant to above functions.
+ * same are kept to allow provision for different handling in case
+ * of DAPM related sequencing, etc.
+ */
+static int gbcodec_mixer_dapm_ctl_info(struct snd_kcontrol *kcontrol,
+ struct snd_ctl_elem_info *uinfo)
+{
+ int platform_max, platform_min;
+ struct gbaudio_ctl_pvt *data;
+ struct gb_audio_ctl_elem_info *info;
+ struct snd_soc_dapm_widget_list *wlist = snd_kcontrol_chip(kcontrol);
+ struct snd_soc_dapm_widget *widget = wlist->widgets[0];
+ struct snd_soc_codec *codec = widget->codec;
+
+ dev_dbg(codec->dev, "Entered %s:%s\n", __func__, kcontrol->id.name);
+ data = (struct gbaudio_ctl_pvt *)kcontrol->private_value;
+ info = (struct gb_audio_ctl_elem_info *)data->info;
+
+ /* update uinfo */
+ platform_max = info->value.integer.max;
+ platform_min = info->value.integer.min;
+
+ if (platform_max == 1 &&
+ !strnstr(kcontrol->id.name, " Volume", NAME_SIZE))
+ uinfo->type = SNDRV_CTL_ELEM_TYPE_BOOLEAN;
+ else
+ uinfo->type = SNDRV_CTL_ELEM_TYPE_INTEGER;
+
+ uinfo->count = data->vcount;
+ uinfo->value.integer.min = 0;
+ if (info->value.integer.min < 0 &&
+ (uinfo->type == SNDRV_CTL_ELEM_TYPE_INTEGER))
+ uinfo->value.integer.max = platform_max - platform_min;
+ else
+ uinfo->value.integer.max = platform_max;
+
+ return 0;
+}
+
+static int gbcodec_mixer_dapm_ctl_get(struct snd_kcontrol *kcontrol,
+ struct snd_ctl_elem_value *ucontrol)
+{
+ int ret;
+ struct gb_audio_ctl_elem_info *info;
+ struct gbaudio_ctl_pvt *data;
+ struct gb_audio_ctl_elem_value gbvalue;
+ struct gbaudio_module_info *module;
+ struct snd_soc_dapm_widget_list *wlist = snd_kcontrol_chip(kcontrol);
+ struct snd_soc_dapm_widget *widget = wlist->widgets[0];
+ struct snd_soc_codec *codec = widget->codec;
+ struct gbaudio_codec_info *gb = snd_soc_codec_get_drvdata(codec);
+ struct gb_bundle *bundle;
+
+ dev_dbg(codec->dev, "Entered %s:%s\n", __func__, kcontrol->id.name);
+ module = find_gb_module(gb, kcontrol->id.name);
+ if (!module)
+ return -EINVAL;
+
+ data = (struct gbaudio_ctl_pvt *)kcontrol->private_value;
+ info = (struct gb_audio_ctl_elem_info *)data->info;
+ bundle = to_gb_bundle(module->dev);
+
+ if (data->vcount == 2)
+ dev_warn(widget->dapm->dev,
+ "GB: Control '%s' is stereo, which is not supported\n",
+ kcontrol->id.name);
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ return ret;
+
+ ret = gb_audio_gb_get_control(module->mgmt_connection, data->ctl_id,
+ GB_AUDIO_INVALID_INDEX, &gbvalue);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ if (ret) {
+ dev_err_ratelimited(codec->dev, "%d:Error in %s for %s\n", ret,
+ __func__, kcontrol->id.name);
+ return ret;
+ }
+ /* update ucontrol */
+ ucontrol->value.integer.value[0] = gbvalue.value.integer_value[0];
+
+ return ret;
+}
+
+static int gbcodec_mixer_dapm_ctl_put(struct snd_kcontrol *kcontrol,
+ struct snd_ctl_elem_value *ucontrol)
+{
+ int ret, wi, max, connect;
+ unsigned int mask, val;
+ struct gb_audio_ctl_elem_info *info;
+ struct gbaudio_ctl_pvt *data;
+ struct gb_audio_ctl_elem_value gbvalue;
+ struct gbaudio_module_info *module;
+ struct snd_soc_dapm_widget_list *wlist = snd_kcontrol_chip(kcontrol);
+ struct snd_soc_dapm_widget *widget = wlist->widgets[0];
+ struct snd_soc_codec *codec = widget->codec;
+ struct gbaudio_codec_info *gb = snd_soc_codec_get_drvdata(codec);
+ struct gb_bundle *bundle;
+
+ dev_dbg(codec->dev, "Entered %s:%s\n", __func__, kcontrol->id.name);
+ module = find_gb_module(gb, kcontrol->id.name);
+ if (!module)
+ return -EINVAL;
+
+ data = (struct gbaudio_ctl_pvt *)kcontrol->private_value;
+ info = (struct gb_audio_ctl_elem_info *)data->info;
+ bundle = to_gb_bundle(module->dev);
+
+ if (data->vcount == 2)
+ dev_warn(widget->dapm->dev,
+ "GB: Control '%s' is stereo, which is not supported\n",
+ kcontrol->id.name);
+
+ max = info->value.integer.max;
+ mask = (1 << fls(max)) - 1;
+ val = (ucontrol->value.integer.value[0] & mask);
+ connect = !!val;
+
+ /* update ucontrol */
+ if (gbvalue.value.integer_value[0] != val) {
+ for (wi = 0; wi < wlist->num_widgets; wi++) {
+ widget = wlist->widgets[wi];
+
+ widget->value = val;
+ widget->dapm->update = NULL;
+ snd_soc_dapm_mixer_update_power(widget, kcontrol,
+ connect);
+ }
+ gbvalue.value.integer_value[0] =
+ ucontrol->value.integer.value[0];
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ return ret;
+
+ ret = gb_audio_gb_set_control(module->mgmt_connection,
+ data->ctl_id,
+ GB_AUDIO_INVALID_INDEX, &gbvalue);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ if (ret) {
+ dev_err_ratelimited(codec->dev,
+ "%d:Error in %s for %s\n", ret,
+ __func__, kcontrol->id.name);
+ }
+ }
+
+ return ret;
+}
+
+#define SOC_DAPM_MIXER_GB(xname, kcount, data) \
+{ .iface = SNDRV_CTL_ELEM_IFACE_MIXER, .name = xname, \
+ .count = kcount, .info = gbcodec_mixer_dapm_ctl_info, \
+ .get = gbcodec_mixer_dapm_ctl_get, .put = gbcodec_mixer_dapm_ctl_put, \
+ .private_value = (unsigned long)data}
+
+static int gbcodec_event_spk(struct snd_soc_dapm_widget *w,
+ struct snd_kcontrol *k, int event)
+{
+ /* Ensure GB speaker is connected */
+
+ return 0;
+}
+
+static int gbcodec_event_hp(struct snd_soc_dapm_widget *w,
+ struct snd_kcontrol *k, int event)
+{
+ /* Ensure GB module supports jack slot */
+
+ return 0;
+}
+
+static int gbcodec_event_int_mic(struct snd_soc_dapm_widget *w,
+ struct snd_kcontrol *k, int event)
+{
+ /* Ensure GB module supports jack slot */
+
+ return 0;
+}
+
+static int gbaudio_validate_kcontrol_count(struct gb_audio_widget *w)
+{
+ int ret = 0;
+
+ switch (w->type) {
+ case snd_soc_dapm_spk:
+ case snd_soc_dapm_hp:
+ case snd_soc_dapm_mic:
+ case snd_soc_dapm_output:
+ case snd_soc_dapm_input:
+ if (w->ncontrols)
+ ret = -EINVAL;
+ break;
+ case snd_soc_dapm_switch:
+ case snd_soc_dapm_mux:
+ if (w->ncontrols != 1)
+ ret = -EINVAL;
+ break;
+ default:
+ break;
+ }
+
+ return ret;
+}
+
+static int gbcodec_enum_ctl_get(struct snd_kcontrol *kcontrol,
+ struct snd_ctl_elem_value *ucontrol)
+{
+ int ret, ctl_id;
+ struct snd_soc_codec *codec = snd_kcontrol_chip(kcontrol);
+ struct soc_enum *e = (struct soc_enum *)kcontrol->private_value;
+ struct gb_audio_ctl_elem_value gbvalue;
+ struct gbaudio_module_info *module;
+ struct gbaudio_codec_info *gb = snd_soc_codec_get_drvdata(codec);
+ struct gb_bundle *bundle;
+
+ module = find_gb_module(gb, kcontrol->id.name);
+ if (!module)
+ return -EINVAL;
+
+ ctl_id = gbaudio_map_controlname(module, kcontrol->id.name);
+ if (ctl_id < 0)
+ return -EINVAL;
+
+ bundle = to_gb_bundle(module->dev);
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ return ret;
+
+ ret = gb_audio_gb_get_control(module->mgmt_connection, ctl_id,
+ GB_AUDIO_INVALID_INDEX, &gbvalue);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ if (ret) {
+ dev_err_ratelimited(codec->dev, "%d:Error in %s for %s\n", ret,
+ __func__, kcontrol->id.name);
+ return ret;
+ }
+
+ ucontrol->value.enumerated.item[0] = gbvalue.value.enumerated_item[0];
+ if (e->shift_l != e->shift_r)
+ ucontrol->value.enumerated.item[1] =
+ gbvalue.value.enumerated_item[1];
+
+ return 0;
+}
+
+static int gbcodec_enum_ctl_put(struct snd_kcontrol *kcontrol,
+ struct snd_ctl_elem_value *ucontrol)
+{
+ int ret, ctl_id;
+ struct snd_soc_codec *codec = snd_kcontrol_chip(kcontrol);
+ struct soc_enum *e = (struct soc_enum *)kcontrol->private_value;
+ struct gb_audio_ctl_elem_value gbvalue;
+ struct gbaudio_module_info *module;
+ struct gbaudio_codec_info *gb = snd_soc_codec_get_drvdata(codec);
+ struct gb_bundle *bundle;
+
+ module = find_gb_module(gb, kcontrol->id.name);
+ if (!module)
+ return -EINVAL;
+
+ ctl_id = gbaudio_map_controlname(module, kcontrol->id.name);
+ if (ctl_id < 0)
+ return -EINVAL;
+
+ if (ucontrol->value.enumerated.item[0] > e->max - 1)
+ return -EINVAL;
+ gbvalue.value.enumerated_item[0] = ucontrol->value.enumerated.item[0];
+
+ if (e->shift_l != e->shift_r) {
+ if (ucontrol->value.enumerated.item[1] > e->max - 1)
+ return -EINVAL;
+ gbvalue.value.enumerated_item[1] =
+ ucontrol->value.enumerated.item[1];
+ }
+
+ bundle = to_gb_bundle(module->dev);
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ return ret;
+
+ ret = gb_audio_gb_set_control(module->mgmt_connection, ctl_id,
+ GB_AUDIO_INVALID_INDEX, &gbvalue);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ if (ret) {
+ dev_err_ratelimited(codec->dev, "%d:Error in %s for %s\n", ret,
+ __func__, kcontrol->id.name);
+ }
+
+ return ret;
+}
+
+static int gbaudio_tplg_create_enum_kctl(struct gbaudio_module_info *gb,
+ struct snd_kcontrol_new *kctl,
+ struct gb_audio_control *ctl)
+{
+ struct soc_enum *gbe;
+ struct gb_audio_enumerated *gb_enum;
+ int i;
+
+ gbe = devm_kzalloc(gb->dev, sizeof(*gbe), GFP_KERNEL);
+ if (!gbe)
+ return -ENOMEM;
+
+ gb_enum = &ctl->info.value.enumerated;
+
+ /* since count=1, and reg is dummy */
+ gbe->max = gb_enum->items;
+ gbe->texts = gb_generate_enum_strings(gb, gb_enum);
+
+ /* debug enum info */
+ dev_dbg(gb->dev, "Max:%d, name_length:%d\n", gb_enum->items,
+ gb_enum->names_length);
+ for (i = 0; i < gb_enum->items; i++)
+ dev_dbg(gb->dev, "src[%d]: %s\n", i, gbe->texts[i]);
+
+ *kctl = (struct snd_kcontrol_new)
+ SOC_ENUM_EXT(ctl->name, *gbe, gbcodec_enum_ctl_get,
+ gbcodec_enum_ctl_put);
+ return 0;
+}
+
+static int gbaudio_tplg_create_kcontrol(struct gbaudio_module_info *gb,
+ struct snd_kcontrol_new *kctl,
+ struct gb_audio_control *ctl)
+{
+ int ret = 0;
+ struct gbaudio_ctl_pvt *ctldata;
+
+ switch (ctl->iface) {
+ case SNDRV_CTL_ELEM_IFACE_MIXER:
+ switch (ctl->info.type) {
+ case GB_AUDIO_CTL_ELEM_TYPE_ENUMERATED:
+ ret = gbaudio_tplg_create_enum_kctl(gb, kctl, ctl);
+ break;
+ default:
+ ctldata = devm_kzalloc(gb->dev,
+ sizeof(struct gbaudio_ctl_pvt),
+ GFP_KERNEL);
+ if (!ctldata)
+ return -ENOMEM;
+ ctldata->ctl_id = ctl->id;
+ ctldata->data_cport = ctl->data_cport;
+ ctldata->access = ctl->access;
+ ctldata->vcount = ctl->count_values;
+ ctldata->info = &ctl->info;
+ *kctl = (struct snd_kcontrol_new)
+ SOC_MIXER_GB(ctl->name, ctl->count, ctldata);
+ ctldata = NULL;
+ break;
+ }
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ dev_dbg(gb->dev, "%s:%d control created\n", ctl->name, ctl->id);
+ return ret;
+}
+
+static int gbcodec_enum_dapm_ctl_get(struct snd_kcontrol *kcontrol,
+ struct snd_ctl_elem_value *ucontrol)
+{
+ int ret, ctl_id;
+ struct snd_soc_dapm_widget_list *wlist = snd_kcontrol_chip(kcontrol);
+ struct snd_soc_dapm_widget *widget = wlist->widgets[0];
+ struct gbaudio_module_info *module;
+ struct gb_audio_ctl_elem_value gbvalue;
+ struct snd_soc_codec *codec = widget->codec;
+ struct gbaudio_codec_info *gb = snd_soc_codec_get_drvdata(codec);
+ struct soc_enum *e = (struct soc_enum *)kcontrol->private_value;
+ struct gb_bundle *bundle;
+
+ module = find_gb_module(gb, kcontrol->id.name);
+ if (!module)
+ return -EINVAL;
+
+ ctl_id = gbaudio_map_wcontrolname(module, kcontrol->id.name);
+ if (ctl_id < 0)
+ return -EINVAL;
+
+ bundle = to_gb_bundle(module->dev);
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ return ret;
+
+ ret = gb_audio_gb_get_control(module->mgmt_connection, ctl_id,
+ GB_AUDIO_INVALID_INDEX, &gbvalue);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ if (ret) {
+ dev_err_ratelimited(codec->dev, "%d:Error in %s for %s\n", ret,
+ __func__, kcontrol->id.name);
+ return ret;
+ }
+
+ ucontrol->value.enumerated.item[0] = gbvalue.value.enumerated_item[0];
+ if (e->shift_l != e->shift_r)
+ ucontrol->value.enumerated.item[1] =
+ gbvalue.value.enumerated_item[1];
+
+ return 0;
+}
+
+static int gbcodec_enum_dapm_ctl_put(struct snd_kcontrol *kcontrol,
+ struct snd_ctl_elem_value *ucontrol)
+{
+ int ret, wi, ctl_id;
+ unsigned int val, mux, change;
+ unsigned int mask;
+ struct snd_soc_dapm_widget_list *wlist = snd_kcontrol_chip(kcontrol);
+ struct snd_soc_dapm_widget *widget = wlist->widgets[0];
+ struct gb_audio_ctl_elem_value gbvalue;
+ struct gbaudio_module_info *module;
+ struct snd_soc_codec *codec = widget->codec;
+ struct gbaudio_codec_info *gb = snd_soc_codec_get_drvdata(codec);
+ struct soc_enum *e = (struct soc_enum *)kcontrol->private_value;
+ struct gb_bundle *bundle;
+
+ if (ucontrol->value.enumerated.item[0] > e->max - 1)
+ return -EINVAL;
+
+ module = find_gb_module(gb, kcontrol->id.name);
+ if (!module)
+ return -EINVAL;
+
+ ctl_id = gbaudio_map_wcontrolname(module, kcontrol->id.name);
+ if (ctl_id < 0)
+ return -EINVAL;
+
+ change = 0;
+ bundle = to_gb_bundle(module->dev);
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ return ret;
+
+ ret = gb_audio_gb_get_control(module->mgmt_connection, ctl_id,
+ GB_AUDIO_INVALID_INDEX, &gbvalue);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ if (ret) {
+ dev_err_ratelimited(codec->dev, "%d:Error in %s for %s\n", ret,
+ __func__, kcontrol->id.name);
+ return ret;
+ }
+
+ mux = ucontrol->value.enumerated.item[0];
+ val = mux << e->shift_l;
+ mask = e->mask << e->shift_l;
+
+ if (gbvalue.value.enumerated_item[0] !=
+ ucontrol->value.enumerated.item[0]) {
+ change = 1;
+ gbvalue.value.enumerated_item[0] =
+ ucontrol->value.enumerated.item[0];
+ }
+
+ if (e->shift_l != e->shift_r) {
+ if (ucontrol->value.enumerated.item[1] > e->max - 1)
+ return -EINVAL;
+ val |= ucontrol->value.enumerated.item[1] << e->shift_r;
+ mask |= e->mask << e->shift_r;
+ if (gbvalue.value.enumerated_item[1] !=
+ ucontrol->value.enumerated.item[1]) {
+ change = 1;
+ gbvalue.value.enumerated_item[1] =
+ ucontrol->value.enumerated.item[1];
+ }
+ }
+
+ if (change) {
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ return ret;
+
+ ret = gb_audio_gb_set_control(module->mgmt_connection, ctl_id,
+ GB_AUDIO_INVALID_INDEX, &gbvalue);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ if (ret) {
+ dev_err_ratelimited(codec->dev,
+ "%d:Error in %s for %s\n", ret,
+ __func__, kcontrol->id.name);
+ }
+ for (wi = 0; wi < wlist->num_widgets; wi++) {
+ widget = wlist->widgets[wi];
+
+ widget->value = val;
+ widget->dapm->update = NULL;
+ snd_soc_dapm_mux_update_power(widget, kcontrol, mux, e);
+ }
+ }
+
+ return change;
+}
+
+static int gbaudio_tplg_create_enum_ctl(struct gbaudio_module_info *gb,
+ struct snd_kcontrol_new *kctl,
+ struct gb_audio_control *ctl)
+{
+ struct soc_enum *gbe;
+ struct gb_audio_enumerated *gb_enum;
+ int i;
+
+ gbe = devm_kzalloc(gb->dev, sizeof(*gbe), GFP_KERNEL);
+ if (!gbe)
+ return -ENOMEM;
+
+ gb_enum = &ctl->info.value.enumerated;
+
+ /* since count=1, and reg is dummy */
+ gbe->max = gb_enum->items;
+ gbe->texts = gb_generate_enum_strings(gb, gb_enum);
+
+ /* debug enum info */
+ dev_dbg(gb->dev, "Max:%d, name_length:%d\n", gb_enum->items,
+ gb_enum->names_length);
+ for (i = 0; i < gb_enum->items; i++)
+ dev_dbg(gb->dev, "src[%d]: %s\n", i, gbe->texts[i]);
+
+ *kctl = (struct snd_kcontrol_new)
+ SOC_DAPM_ENUM_EXT(ctl->name, *gbe, gbcodec_enum_dapm_ctl_get,
+ gbcodec_enum_dapm_ctl_put);
+ return 0;
+}
+
+static int gbaudio_tplg_create_mixer_ctl(struct gbaudio_module_info *gb,
+ struct snd_kcontrol_new *kctl,
+ struct gb_audio_control *ctl)
+{
+ struct gbaudio_ctl_pvt *ctldata;
+
+ ctldata = devm_kzalloc(gb->dev, sizeof(struct gbaudio_ctl_pvt),
+ GFP_KERNEL);
+ if (!ctldata)
+ return -ENOMEM;
+ ctldata->ctl_id = ctl->id;
+ ctldata->data_cport = ctl->data_cport;
+ ctldata->access = ctl->access;
+ ctldata->vcount = ctl->count_values;
+ ctldata->info = &ctl->info;
+ *kctl = (struct snd_kcontrol_new)
+ SOC_DAPM_MIXER_GB(ctl->name, ctl->count, ctldata);
+
+ return 0;
+}
+
+static int gbaudio_tplg_create_wcontrol(struct gbaudio_module_info *gb,
+ struct snd_kcontrol_new *kctl,
+ struct gb_audio_control *ctl)
+{
+ int ret;
+
+ switch (ctl->iface) {
+ case SNDRV_CTL_ELEM_IFACE_MIXER:
+ switch (ctl->info.type) {
+ case GB_AUDIO_CTL_ELEM_TYPE_ENUMERATED:
+ ret = gbaudio_tplg_create_enum_ctl(gb, kctl, ctl);
+ break;
+ default:
+ ret = gbaudio_tplg_create_mixer_ctl(gb, kctl, ctl);
+ break;
+ }
+ break;
+ default:
+ return -EINVAL;
+
+ }
+
+ dev_dbg(gb->dev, "%s:%d DAPM control created, ret:%d\n", ctl->name,
+ ctl->id, ret);
+ return ret;
+}
+
+static int gbaudio_widget_event(struct snd_soc_dapm_widget *w,
+ struct snd_kcontrol *kcontrol, int event)
+{
+ int wid;
+ int ret;
+ struct snd_soc_codec *codec = w->codec;
+ struct gbaudio_codec_info *gbcodec = snd_soc_codec_get_drvdata(codec);
+ struct gbaudio_module_info *module;
+ struct gb_bundle *bundle;
+
+ dev_dbg(codec->dev, "%s %s %d\n", __func__, w->name, event);
+
+ /* Find relevant module */
+ module = find_gb_module(gbcodec, w->name);
+ if (!module)
+ return -EINVAL;
+
+ /* map name to widget id */
+ wid = gbaudio_map_widgetname(module, w->name);
+ if (wid < 0) {
+ dev_err(codec->dev, "Invalid widget name:%s\n", w->name);
+ return -EINVAL;
+ }
+
+ bundle = to_gb_bundle(module->dev);
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ return ret;
+
+ switch (event) {
+ case SND_SOC_DAPM_PRE_PMU:
+ ret = gb_audio_gb_enable_widget(module->mgmt_connection, wid);
+ if (!ret)
+ ret = gbaudio_module_update(gbcodec, w, module, 1);
+ break;
+ case SND_SOC_DAPM_POST_PMD:
+ ret = gb_audio_gb_disable_widget(module->mgmt_connection, wid);
+ if (!ret)
+ ret = gbaudio_module_update(gbcodec, w, module, 0);
+ break;
+ }
+ if (ret)
+ dev_err_ratelimited(codec->dev,
+ "%d: widget, event:%d failed:%d\n", wid,
+ event, ret);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return ret;
+}
+
+static int gbaudio_tplg_create_widget(struct gbaudio_module_info *module,
+ struct snd_soc_dapm_widget *dw,
+ struct gb_audio_widget *w, int *w_size)
+{
+ int i, ret, csize;
+ struct snd_kcontrol_new *widget_kctls;
+ struct gb_audio_control *curr;
+ struct gbaudio_control *control, *_control;
+ size_t size;
+ char temp_name[NAME_SIZE];
+
+ ret = gbaudio_validate_kcontrol_count(w);
+ if (ret) {
+ dev_err(module->dev, "Inavlid kcontrol count=%d for %s\n",
+ w->ncontrols, w->name);
+ return ret;
+ }
+
+ /* allocate memory for kcontrol */
+ if (w->ncontrols) {
+ size = sizeof(struct snd_kcontrol_new) * w->ncontrols;
+ widget_kctls = devm_kzalloc(module->dev, size, GFP_KERNEL);
+ if (!widget_kctls)
+ return -ENOMEM;
+ }
+
+ *w_size = sizeof(struct gb_audio_widget);
+
+ /* create relevant kcontrols */
+ curr = w->ctl;
+ for (i = 0; i < w->ncontrols; i++) {
+ ret = gbaudio_tplg_create_wcontrol(module, &widget_kctls[i],
+ curr);
+ if (ret) {
+ dev_err(module->dev,
+ "%s:%d type widget_ctl not supported\n",
+ curr->name, curr->iface);
+ goto error;
+ }
+ control = devm_kzalloc(module->dev,
+ sizeof(struct gbaudio_control),
+ GFP_KERNEL);
+ if (!control) {
+ ret = -ENOMEM;
+ goto error;
+ }
+ control->id = curr->id;
+ control->name = curr->name;
+ control->wname = w->name;
+
+ if (curr->info.type == GB_AUDIO_CTL_ELEM_TYPE_ENUMERATED) {
+ struct gb_audio_enumerated *gbenum =
+ &curr->info.value.enumerated;
+
+ csize = offsetof(struct gb_audio_control, info);
+ csize += offsetof(struct gb_audio_ctl_elem_info, value);
+ csize += offsetof(struct gb_audio_enumerated, names);
+ csize += gbenum->names_length;
+ control->texts = (const char * const *)
+ gb_generate_enum_strings(module, gbenum);
+ control->items = gbenum->items;
+ } else
+ csize = sizeof(struct gb_audio_control);
+ *w_size += csize;
+ curr = (void *)curr + csize;
+ list_add(&control->list, &module->widget_ctl_list);
+ dev_dbg(module->dev, "%s: control of type %d created\n",
+ widget_kctls[i].name, widget_kctls[i].iface);
+ }
+
+ /* Prefix dev_id to widget control_name */
+ strlcpy(temp_name, w->name, NAME_SIZE);
+ snprintf(w->name, NAME_SIZE, "GB %d %s", module->dev_id, temp_name);
+
+ switch (w->type) {
+ case snd_soc_dapm_spk:
+ *dw = (struct snd_soc_dapm_widget)
+ SND_SOC_DAPM_SPK(w->name, gbcodec_event_spk);
+ module->op_devices |= GBAUDIO_DEVICE_OUT_SPEAKER;
+ break;
+ case snd_soc_dapm_hp:
+ *dw = (struct snd_soc_dapm_widget)
+ SND_SOC_DAPM_HP(w->name, gbcodec_event_hp);
+ module->op_devices |= (GBAUDIO_DEVICE_OUT_WIRED_HEADSET
+ | GBAUDIO_DEVICE_OUT_WIRED_HEADPHONE);
+ module->ip_devices |= GBAUDIO_DEVICE_IN_WIRED_HEADSET;
+ break;
+ case snd_soc_dapm_mic:
+ *dw = (struct snd_soc_dapm_widget)
+ SND_SOC_DAPM_MIC(w->name, gbcodec_event_int_mic);
+ module->ip_devices |= GBAUDIO_DEVICE_IN_BUILTIN_MIC;
+ break;
+ case snd_soc_dapm_output:
+ *dw = (struct snd_soc_dapm_widget)SND_SOC_DAPM_OUTPUT(w->name);
+ break;
+ case snd_soc_dapm_input:
+ *dw = (struct snd_soc_dapm_widget)SND_SOC_DAPM_INPUT(w->name);
+ break;
+ case snd_soc_dapm_switch:
+ *dw = (struct snd_soc_dapm_widget)
+ SND_SOC_DAPM_SWITCH_E(w->name, SND_SOC_NOPM, 0, 0,
+ widget_kctls, gbaudio_widget_event,
+ SND_SOC_DAPM_PRE_PMU |
+ SND_SOC_DAPM_POST_PMD);
+ break;
+ case snd_soc_dapm_pga:
+ *dw = (struct snd_soc_dapm_widget)
+ SND_SOC_DAPM_PGA_E(w->name, SND_SOC_NOPM, 0, 0, NULL, 0,
+ gbaudio_widget_event,
+ SND_SOC_DAPM_PRE_PMU |
+ SND_SOC_DAPM_POST_PMD);
+ break;
+ case snd_soc_dapm_mixer:
+ *dw = (struct snd_soc_dapm_widget)
+ SND_SOC_DAPM_MIXER_E(w->name, SND_SOC_NOPM, 0, 0, NULL,
+ 0, gbaudio_widget_event,
+ SND_SOC_DAPM_PRE_PMU |
+ SND_SOC_DAPM_POST_PMD);
+ break;
+ case snd_soc_dapm_mux:
+ *dw = (struct snd_soc_dapm_widget)
+ SND_SOC_DAPM_MUX_E(w->name, SND_SOC_NOPM, 0, 0,
+ widget_kctls, gbaudio_widget_event,
+ SND_SOC_DAPM_PRE_PMU |
+ SND_SOC_DAPM_POST_PMD);
+ break;
+ case snd_soc_dapm_aif_in:
+ *dw = (struct snd_soc_dapm_widget)
+ SND_SOC_DAPM_AIF_IN_E(w->name, w->sname, 0,
+ SND_SOC_NOPM,
+ 0, 0, gbaudio_widget_event,
+ SND_SOC_DAPM_PRE_PMU |
+ SND_SOC_DAPM_POST_PMD);
+ break;
+ case snd_soc_dapm_aif_out:
+ *dw = (struct snd_soc_dapm_widget)
+ SND_SOC_DAPM_AIF_OUT_E(w->name, w->sname, 0,
+ SND_SOC_NOPM,
+ 0, 0, gbaudio_widget_event,
+ SND_SOC_DAPM_PRE_PMU |
+ SND_SOC_DAPM_POST_PMD);
+ break;
+ default:
+ ret = -EINVAL;
+ goto error;
+ }
+
+ dev_dbg(module->dev, "%s: widget of type %d created\n", dw->name,
+ dw->id);
+ return 0;
+error:
+ list_for_each_entry_safe(control, _control, &module->widget_ctl_list,
+ list) {
+ list_del(&control->list);
+ devm_kfree(module->dev, control);
+ }
+ return ret;
+}
+
+static int gbaudio_tplg_process_kcontrols(struct gbaudio_module_info *module,
+ struct gb_audio_control *controls)
+{
+ int i, csize, ret;
+ struct snd_kcontrol_new *dapm_kctls;
+ struct gb_audio_control *curr;
+ struct gbaudio_control *control, *_control;
+ size_t size;
+ char temp_name[NAME_SIZE];
+
+ size = sizeof(struct snd_kcontrol_new) * module->num_controls;
+ dapm_kctls = devm_kzalloc(module->dev, size, GFP_KERNEL);
+ if (!dapm_kctls)
+ return -ENOMEM;
+
+ curr = controls;
+ for (i = 0; i < module->num_controls; i++) {
+ ret = gbaudio_tplg_create_kcontrol(module, &dapm_kctls[i],
+ curr);
+ if (ret) {
+ dev_err(module->dev, "%s:%d type not supported\n",
+ curr->name, curr->iface);
+ goto error;
+ }
+ control = devm_kzalloc(module->dev, sizeof(struct
+ gbaudio_control),
+ GFP_KERNEL);
+ if (!control) {
+ ret = -ENOMEM;
+ goto error;
+ }
+ control->id = curr->id;
+ /* Prefix dev_id to widget_name */
+ strlcpy(temp_name, curr->name, NAME_SIZE);
+ snprintf(curr->name, NAME_SIZE, "GB %d %s", module->dev_id,
+ temp_name);
+ control->name = curr->name;
+ if (curr->info.type == GB_AUDIO_CTL_ELEM_TYPE_ENUMERATED) {
+ struct gb_audio_enumerated *gbenum =
+ &curr->info.value.enumerated;
+
+ csize = offsetof(struct gb_audio_control, info);
+ csize += offsetof(struct gb_audio_ctl_elem_info, value);
+ csize += offsetof(struct gb_audio_enumerated, names);
+ csize += gbenum->names_length;
+ control->texts = (const char * const *)
+ gb_generate_enum_strings(module, gbenum);
+ control->items = gbenum->items;
+ } else
+ csize = sizeof(struct gb_audio_control);
+
+ list_add(&control->list, &module->ctl_list);
+ dev_dbg(module->dev, "%d:%s created of type %d\n", curr->id,
+ curr->name, curr->info.type);
+ curr = (void *)curr + csize;
+ }
+ module->controls = dapm_kctls;
+
+ return 0;
+error:
+ list_for_each_entry_safe(control, _control, &module->ctl_list,
+ list) {
+ list_del(&control->list);
+ devm_kfree(module->dev, control);
+ }
+ devm_kfree(module->dev, dapm_kctls);
+ return ret;
+}
+
+static int gbaudio_tplg_process_widgets(struct gbaudio_module_info *module,
+ struct gb_audio_widget *widgets)
+{
+ int i, ret, w_size;
+ struct snd_soc_dapm_widget *dapm_widgets;
+ struct gb_audio_widget *curr;
+ struct gbaudio_widget *widget, *_widget;
+ size_t size;
+
+ size = sizeof(struct snd_soc_dapm_widget) * module->num_dapm_widgets;
+ dapm_widgets = devm_kzalloc(module->dev, size, GFP_KERNEL);
+ if (!dapm_widgets)
+ return -ENOMEM;
+
+ curr = widgets;
+ for (i = 0; i < module->num_dapm_widgets; i++) {
+ ret = gbaudio_tplg_create_widget(module, &dapm_widgets[i],
+ curr, &w_size);
+ if (ret) {
+ dev_err(module->dev, "%s:%d type not supported\n",
+ curr->name, curr->type);
+ goto error;
+ }
+ widget = devm_kzalloc(module->dev, sizeof(struct
+ gbaudio_widget),
+ GFP_KERNEL);
+ if (!widget) {
+ ret = -ENOMEM;
+ goto error;
+ }
+ widget->id = curr->id;
+ widget->name = curr->name;
+ list_add(&widget->list, &module->widget_list);
+ curr = (void *)curr + w_size;
+ }
+ module->dapm_widgets = dapm_widgets;
+
+ return 0;
+
+error:
+ list_for_each_entry_safe(widget, _widget, &module->widget_list,
+ list) {
+ list_del(&widget->list);
+ devm_kfree(module->dev, widget);
+ }
+ devm_kfree(module->dev, dapm_widgets);
+ return ret;
+}
+
+static int gbaudio_tplg_process_routes(struct gbaudio_module_info *module,
+ struct gb_audio_route *routes)
+{
+ int i, ret;
+ struct snd_soc_dapm_route *dapm_routes;
+ struct gb_audio_route *curr;
+ size_t size;
+
+ size = sizeof(struct snd_soc_dapm_route) * module->num_dapm_routes;
+ dapm_routes = devm_kzalloc(module->dev, size, GFP_KERNEL);
+ if (!dapm_routes)
+ return -ENOMEM;
+
+ module->dapm_routes = dapm_routes;
+ curr = routes;
+
+ for (i = 0; i < module->num_dapm_routes; i++) {
+ dapm_routes->sink =
+ gbaudio_map_widgetid(module, curr->destination_id);
+ if (!dapm_routes->sink) {
+ dev_err(module->dev, "%d:%d:%d:%d - Invalid sink\n",
+ curr->source_id, curr->destination_id,
+ curr->control_id, curr->index);
+ ret = -EINVAL;
+ goto error;
+ }
+ dapm_routes->source =
+ gbaudio_map_widgetid(module, curr->source_id);
+ if (!dapm_routes->source) {
+ dev_err(module->dev, "%d:%d:%d:%d - Invalid source\n",
+ curr->source_id, curr->destination_id,
+ curr->control_id, curr->index);
+ ret = -EINVAL;
+ goto error;
+ }
+ dapm_routes->control =
+ gbaudio_map_controlid(module,
+ curr->control_id,
+ curr->index);
+ if ((curr->control_id != GBAUDIO_INVALID_ID) &&
+ !dapm_routes->control) {
+ dev_err(module->dev, "%d:%d:%d:%d - Invalid control\n",
+ curr->source_id, curr->destination_id,
+ curr->control_id, curr->index);
+ ret = -EINVAL;
+ goto error;
+ }
+ dev_dbg(module->dev, "Route {%s, %s, %s}\n", dapm_routes->sink,
+ (dapm_routes->control) ? dapm_routes->control:"NULL",
+ dapm_routes->source);
+ dapm_routes++;
+ curr++;
+ }
+
+ return 0;
+
+error:
+ devm_kfree(module->dev, module->dapm_routes);
+ return ret;
+}
+
+static int gbaudio_tplg_process_header(struct gbaudio_module_info *module,
+ struct gb_audio_topology *tplg_data)
+{
+ /* fetch no. of kcontrols, widgets & routes */
+ module->num_controls = tplg_data->num_controls;
+ module->num_dapm_widgets = tplg_data->num_widgets;
+ module->num_dapm_routes = tplg_data->num_routes;
+
+ /* update block offset */
+ module->dai_offset = (unsigned long)&tplg_data->data;
+ module->control_offset = module->dai_offset + tplg_data->size_dais;
+ module->widget_offset = module->control_offset +
+ tplg_data->size_controls;
+ module->route_offset = module->widget_offset +
+ tplg_data->size_widgets;
+
+ dev_dbg(module->dev, "DAI offset is 0x%lx\n", module->dai_offset);
+ dev_dbg(module->dev, "control offset is %lx\n",
+ module->control_offset);
+ dev_dbg(module->dev, "widget offset is %lx\n", module->widget_offset);
+ dev_dbg(module->dev, "route offset is %lx\n", module->route_offset);
+
+ return 0;
+}
+
+int gbaudio_tplg_parse_data(struct gbaudio_module_info *module,
+ struct gb_audio_topology *tplg_data)
+{
+ int ret;
+ struct gb_audio_control *controls;
+ struct gb_audio_widget *widgets;
+ struct gb_audio_route *routes;
+
+ if (!tplg_data)
+ return -EINVAL;
+
+ ret = gbaudio_tplg_process_header(module, tplg_data);
+ if (ret) {
+ dev_err(module->dev, "%d: Error in parsing topology header\n",
+ ret);
+ return ret;
+ }
+
+ /* process control */
+ controls = (struct gb_audio_control *)module->control_offset;
+ ret = gbaudio_tplg_process_kcontrols(module, controls);
+ if (ret) {
+ dev_err(module->dev,
+ "%d: Error in parsing controls data\n", ret);
+ return ret;
+ }
+ dev_dbg(module->dev, "Control parsing finished\n");
+
+ /* process widgets */
+ widgets = (struct gb_audio_widget *)module->widget_offset;
+ ret = gbaudio_tplg_process_widgets(module, widgets);
+ if (ret) {
+ dev_err(module->dev,
+ "%d: Error in parsing widgets data\n", ret);
+ return ret;
+ }
+ dev_dbg(module->dev, "Widget parsing finished\n");
+
+ /* process route */
+ routes = (struct gb_audio_route *)module->route_offset;
+ ret = gbaudio_tplg_process_routes(module, routes);
+ if (ret) {
+ dev_err(module->dev,
+ "%d: Error in parsing routes data\n", ret);
+ return ret;
+ }
+ dev_dbg(module->dev, "Route parsing finished\n");
+
+ /* parse jack capabilities */
+ if (tplg_data->jack_type) {
+ module->jack_mask = tplg_data->jack_type & GBCODEC_JACK_MASK;
+ module->button_mask = tplg_data->jack_type &
+ GBCODEC_JACK_BUTTON_MASK;
+ }
+
+ return ret;
+}
+
+void gbaudio_tplg_release(struct gbaudio_module_info *module)
+{
+ struct gbaudio_control *control, *_control;
+ struct gbaudio_widget *widget, *_widget;
+
+ if (!module->topology)
+ return;
+
+ /* release kcontrols */
+ list_for_each_entry_safe(control, _control, &module->ctl_list,
+ list) {
+ list_del(&control->list);
+ devm_kfree(module->dev, control);
+ }
+ if (module->controls)
+ devm_kfree(module->dev, module->controls);
+
+ /* release widget controls */
+ list_for_each_entry_safe(control, _control, &module->widget_ctl_list,
+ list) {
+ list_del(&control->list);
+ devm_kfree(module->dev, control);
+ }
+
+ /* release widgets */
+ list_for_each_entry_safe(widget, _widget, &module->widget_list,
+ list) {
+ list_del(&widget->list);
+ devm_kfree(module->dev, widget);
+ }
+ if (module->dapm_widgets)
+ devm_kfree(module->dev, module->dapm_widgets);
+
+ /* release routes */
+ if (module->dapm_routes)
+ devm_kfree(module->dev, module->dapm_routes);
+}
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 11/32] greybus: camera driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (8 preceding siblings ...)
2016-09-16 13:24 ` [patch 10/32] greybus: audio driver Greg KH
@ 2016-09-16 13:25 ` Greg KH
2016-09-16 13:25 ` [patch 12/32] greybus: es2 host driver Greg KH
` (20 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 13:25 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus camera protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/camera.c | 1400 ++++++++++++++++++++++++++++++++++++++++++++
drivers/greybus/gb-camera.h | 127 +++
2 files changed, 1527 insertions(+)
--- /dev/null
+++ b/drivers/greybus/camera.c
@@ -0,0 +1,1400 @@
+/*
+ * Greybus Camera protocol driver.
+ *
+ * Copyright 2015 Google Inc.
+ * Copyright 2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/debugfs.h>
+#include <linux/fs.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/string.h>
+#include <linux/uaccess.h>
+#include <linux/vmalloc.h>
+
+#include "gb-camera.h"
+#include "greybus.h"
+#include "greybus_protocols.h"
+
+enum gb_camera_debugs_buffer_id {
+ GB_CAMERA_DEBUGFS_BUFFER_CAPABILITIES,
+ GB_CAMERA_DEBUGFS_BUFFER_STREAMS,
+ GB_CAMERA_DEBUGFS_BUFFER_CAPTURE,
+ GB_CAMERA_DEBUGFS_BUFFER_FLUSH,
+ GB_CAMERA_DEBUGFS_BUFFER_MAX,
+};
+
+struct gb_camera_debugfs_buffer {
+ char data[PAGE_SIZE];
+ size_t length;
+};
+
+enum gb_camera_state {
+ GB_CAMERA_STATE_UNCONFIGURED,
+ GB_CAMERA_STATE_CONFIGURED,
+};
+
+/**
+ * struct gb_camera - A Greybus Camera Device
+ * @connection: the greybus connection for camera management
+ * @data_connection: the greybus connection for camera data
+ * @data_cport_id: the data CPort ID on the module side
+ * @mutex: protects the connection and state fields
+ * @state: the current module state
+ * @debugfs: debugfs entries for camera protocol operations testing
+ * @module: Greybus camera module registered to HOST processor.
+ */
+struct gb_camera {
+ struct gb_bundle *bundle;
+ struct gb_connection *connection;
+ struct gb_connection *data_connection;
+ u16 data_cport_id;
+
+ struct mutex mutex;
+ enum gb_camera_state state;
+
+ struct {
+ struct dentry *root;
+ struct gb_camera_debugfs_buffer *buffers;
+ } debugfs;
+
+ struct gb_camera_module module;
+};
+
+struct gb_camera_stream_config {
+ unsigned int width;
+ unsigned int height;
+ unsigned int format;
+ unsigned int vc;
+ unsigned int dt[2];
+ unsigned int max_size;
+};
+
+struct gb_camera_fmt_info {
+ enum v4l2_mbus_pixelcode mbus_code;
+ unsigned int gb_format;
+ unsigned int bpp;
+};
+
+/* GB format to media code map */
+static const struct gb_camera_fmt_info gb_fmt_info[] = {
+ {
+ .mbus_code = V4L2_MBUS_FMT_UYVY8_1X16,
+ .gb_format = 0x01,
+ .bpp = 16,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_NV12_1x8,
+ .gb_format = 0x12,
+ .bpp = 12,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_NV21_1x8,
+ .gb_format = 0x13,
+ .bpp = 12,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_YU12_1x8,
+ .gb_format = 0x16,
+ .bpp = 12,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_YV12_1x8,
+ .gb_format = 0x17,
+ .bpp = 12,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_JPEG_1X8,
+ .gb_format = 0x40,
+ .bpp = 0,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_GB_CAM_METADATA_1X8,
+ .gb_format = 0x41,
+ .bpp = 0,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_GB_CAM_DEBUG_DATA_1X8,
+ .gb_format = 0x42,
+ .bpp = 0,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_SBGGR10_1X10,
+ .gb_format = 0x80,
+ .bpp = 10,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_SGBRG10_1X10,
+ .gb_format = 0x81,
+ .bpp = 10,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_SGRBG10_1X10,
+ .gb_format = 0x82,
+ .bpp = 10,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_SRGGB10_1X10,
+ .gb_format = 0x83,
+ .bpp = 10,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_SBGGR12_1X12,
+ .gb_format = 0x84,
+ .bpp = 12,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_SGBRG12_1X12,
+ .gb_format = 0x85,
+ .bpp = 12,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_SGRBG12_1X12,
+ .gb_format = 0x86,
+ .bpp = 12,
+ },
+ {
+ .mbus_code = V4L2_MBUS_FMT_SRGGB12_1X12,
+ .gb_format = 0x87,
+ .bpp = 12,
+ },
+};
+
+static const struct gb_camera_fmt_info *gb_camera_get_format_info(u16 gb_fmt)
+{
+ unsigned int i;
+
+ for (i = 0; i < ARRAY_SIZE(gb_fmt_info); i++) {
+ if (gb_fmt_info[i].gb_format == gb_fmt)
+ return &gb_fmt_info[i];
+ }
+
+ return NULL;
+}
+
+#define ES2_APB_CDSI0_CPORT 16
+#define ES2_APB_CDSI1_CPORT 17
+
+#define GB_CAMERA_MAX_SETTINGS_SIZE 8192
+
+#define gcam_dbg(gcam, format...) dev_dbg(&gcam->bundle->dev, format)
+#define gcam_info(gcam, format...) dev_info(&gcam->bundle->dev, format)
+#define gcam_err(gcam, format...) dev_err(&gcam->bundle->dev, format)
+
+static int gb_camera_operation_sync_flags(struct gb_connection *connection,
+ int type, unsigned int flags,
+ void *request, size_t request_size,
+ void *response, size_t *response_size)
+{
+ struct gb_operation *operation;
+ int ret;
+
+ operation = gb_operation_create_flags(connection, type, request_size,
+ *response_size, flags,
+ GFP_KERNEL);
+ if (!operation)
+ return -ENOMEM;
+
+ if (request_size)
+ memcpy(operation->request->payload, request, request_size);
+
+ ret = gb_operation_request_send_sync(operation);
+ if (ret) {
+ dev_err(&connection->hd->dev,
+ "%s: synchronous operation of type 0x%02x failed: %d\n",
+ connection->name, type, ret);
+ } else {
+ *response_size = operation->response->payload_size;
+
+ if (operation->response->payload_size)
+ memcpy(response, operation->response->payload,
+ operation->response->payload_size);
+ }
+
+ gb_operation_put(operation);
+
+ return ret;
+}
+
+static int gb_camera_get_max_pkt_size(struct gb_camera *gcam,
+ struct gb_camera_configure_streams_response *resp)
+{
+ unsigned int max_pkt_size = 0;
+ unsigned int i;
+
+ for (i = 0; i < resp->num_streams; i++) {
+ struct gb_camera_stream_config_response *cfg = &resp->config[i];
+ const struct gb_camera_fmt_info *fmt_info;
+ unsigned int pkt_size;
+
+ fmt_info = gb_camera_get_format_info(cfg->format);
+ if (!fmt_info) {
+ gcam_err(gcam, "unsupported greybus image format: %d\n",
+ cfg->format);
+ return -EIO;
+ }
+
+ if (fmt_info->bpp == 0) {
+ pkt_size = le32_to_cpu(cfg->max_pkt_size);
+
+ if (pkt_size == 0) {
+ gcam_err(gcam,
+ "Stream %u: invalid zero maximum packet size\n",
+ i);
+ return -EIO;
+ }
+ } else {
+ pkt_size = le16_to_cpu(cfg->width) * fmt_info->bpp / 8;
+
+ if (pkt_size != le32_to_cpu(cfg->max_pkt_size)) {
+ gcam_err(gcam,
+ "Stream %u: maximum packet size mismatch (%u/%u)\n",
+ i, pkt_size, cfg->max_pkt_size);
+ return -EIO;
+ }
+ }
+
+ max_pkt_size = max(pkt_size, max_pkt_size);
+ }
+
+ return max_pkt_size;
+}
+
+/*
+ * Validate the stream configuration response verifying padding is correctly
+ * set and the returned number of streams is supported
+ */
+static const int gb_camera_configure_streams_validate_response(
+ struct gb_camera *gcam,
+ struct gb_camera_configure_streams_response *resp,
+ unsigned int nstreams)
+{
+ unsigned int i;
+
+ /* Validate the returned response structure */
+ if (resp->padding[0] || resp->padding[1]) {
+ gcam_err(gcam, "response padding != 0\n");
+ return -EIO;
+ }
+
+ if (resp->num_streams > nstreams) {
+ gcam_err(gcam, "got #streams %u > request %u\n",
+ resp->num_streams, nstreams);
+ return -EIO;
+ }
+
+ for (i = 0; i < resp->num_streams; i++) {
+ struct gb_camera_stream_config_response *cfg = &resp->config[i];
+ if (cfg->padding) {
+ gcam_err(gcam, "stream #%u padding != 0\n", i);
+ return -EIO;
+ }
+ }
+
+ return 0;
+}
+
+/* -----------------------------------------------------------------------------
+ * Hardware Configuration
+ */
+
+static int gb_camera_set_intf_power_mode(struct gb_camera *gcam, u8 intf_id,
+ bool hs)
+{
+ struct gb_svc *svc = gcam->connection->hd->svc;
+ int ret;
+
+ if (hs)
+ ret = gb_svc_intf_set_power_mode(svc, intf_id,
+ GB_SVC_UNIPRO_HS_SERIES_A,
+ GB_SVC_UNIPRO_FAST_MODE, 2, 2,
+ GB_SVC_SMALL_AMPLITUDE,
+ GB_SVC_NO_DE_EMPHASIS,
+ GB_SVC_UNIPRO_FAST_MODE, 2, 2,
+ GB_SVC_PWRM_RXTERMINATION |
+ GB_SVC_PWRM_TXTERMINATION, 0,
+ NULL, NULL);
+ else
+ ret = gb_svc_intf_set_power_mode(svc, intf_id,
+ GB_SVC_UNIPRO_HS_SERIES_A,
+ GB_SVC_UNIPRO_SLOW_AUTO_MODE,
+ 2, 1,
+ GB_SVC_SMALL_AMPLITUDE,
+ GB_SVC_NO_DE_EMPHASIS,
+ GB_SVC_UNIPRO_SLOW_AUTO_MODE,
+ 2, 1,
+ 0, 0,
+ NULL, NULL);
+
+ return ret;
+}
+
+static int gb_camera_set_power_mode(struct gb_camera *gcam, bool hs)
+{
+ struct gb_interface *intf = gcam->connection->intf;
+ struct gb_svc *svc = gcam->connection->hd->svc;
+ int ret;
+
+ ret = gb_camera_set_intf_power_mode(gcam, intf->interface_id, hs);
+ if (ret < 0) {
+ gcam_err(gcam, "failed to set module interface to %s (%d)\n",
+ hs ? "HS" : "PWM", ret);
+ return ret;
+ }
+
+ ret = gb_camera_set_intf_power_mode(gcam, svc->ap_intf_id, hs);
+ if (ret < 0) {
+ gb_camera_set_intf_power_mode(gcam, intf->interface_id, !hs);
+ gcam_err(gcam, "failed to set AP interface to %s (%d)\n",
+ hs ? "HS" : "PWM", ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+struct ap_csi_config_request {
+ __u8 csi_id;
+ __u8 flags;
+#define GB_CAMERA_CSI_FLAG_CLOCK_CONTINUOUS 0x01
+ __u8 num_lanes;
+ __u8 padding;
+ __le32 csi_clk_freq;
+ __le32 max_pkt_size;
+} __packed;
+
+/*
+ * TODO: Compute the number of lanes dynamically based on bandwidth
+ * requirements.
+ */
+#define GB_CAMERA_CSI_NUM_DATA_LANES 4
+
+#define GB_CAMERA_CSI_CLK_FREQ_MAX 999000000U
+#define GB_CAMERA_CSI_CLK_FREQ_MIN 100000000U
+#define GB_CAMERA_CSI_CLK_FREQ_MARGIN 150000000U
+
+static int gb_camera_setup_data_connection(struct gb_camera *gcam,
+ struct gb_camera_configure_streams_response *resp,
+ struct gb_camera_csi_params *csi_params)
+{
+ struct ap_csi_config_request csi_cfg;
+ struct gb_connection *conn;
+ unsigned int clk_freq;
+ int ret;
+
+ /*
+ * Create the data connection between the camera module data CPort and
+ * APB CDSI1. The CDSI1 CPort ID is hardcoded by the ES2 bridge.
+ */
+ conn = gb_connection_create_offloaded(gcam->bundle, gcam->data_cport_id,
+ GB_CONNECTION_FLAG_NO_FLOWCTRL |
+ GB_CONNECTION_FLAG_CDSI1);
+ if (IS_ERR(conn))
+ return PTR_ERR(conn);
+
+ gcam->data_connection = conn;
+ gb_connection_set_data(conn, gcam);
+
+ ret = gb_connection_enable(conn);
+ if (ret)
+ goto error_conn_destroy;
+
+ /* Set the UniPro link to high speed mode. */
+ ret = gb_camera_set_power_mode(gcam, true);
+ if (ret < 0)
+ goto error_conn_disable;
+
+ /*
+ * Configure the APB-A CSI-2 transmitter.
+ *
+ * Hardcode the number of lanes to 4 and compute the bus clock frequency
+ * based on the module bandwidth requirements with a safety margin.
+ */
+ memset(&csi_cfg, 0, sizeof(csi_cfg));
+ csi_cfg.csi_id = 1;
+ csi_cfg.flags = 0;
+ csi_cfg.num_lanes = GB_CAMERA_CSI_NUM_DATA_LANES;
+
+ clk_freq = resp->data_rate / 2 / GB_CAMERA_CSI_NUM_DATA_LANES;
+ clk_freq = clamp(clk_freq + GB_CAMERA_CSI_CLK_FREQ_MARGIN,
+ GB_CAMERA_CSI_CLK_FREQ_MIN,
+ GB_CAMERA_CSI_CLK_FREQ_MAX);
+ csi_cfg.csi_clk_freq = clk_freq;
+
+ ret = gb_camera_get_max_pkt_size(gcam, resp);
+ if (ret < 0) {
+ ret = -EIO;
+ goto error_power;
+ }
+ csi_cfg.max_pkt_size = ret;
+
+ ret = gb_hd_output(gcam->connection->hd, &csi_cfg,
+ sizeof(csi_cfg),
+ GB_APB_REQUEST_CSI_TX_CONTROL, false);
+ if (ret < 0) {
+ gcam_err(gcam, "failed to start the CSI transmitter\n");
+ goto error_power;
+ }
+
+ if (csi_params) {
+ csi_params->clk_freq = csi_cfg.csi_clk_freq;
+ csi_params->num_lanes = csi_cfg.num_lanes;
+ }
+
+ return 0;
+
+error_power:
+ gb_camera_set_power_mode(gcam, false);
+error_conn_disable:
+ gb_connection_disable(gcam->data_connection);
+error_conn_destroy:
+ gb_connection_destroy(gcam->data_connection);
+ gcam->data_connection = NULL;
+ return ret;
+}
+
+static void gb_camera_teardown_data_connection(struct gb_camera *gcam)
+{
+ struct ap_csi_config_request csi_cfg;
+ int ret;
+
+ /* Stop the APB1 CSI transmitter. */
+ memset(&csi_cfg, 0, sizeof(csi_cfg));
+ csi_cfg.csi_id = 1;
+
+ ret = gb_hd_output(gcam->connection->hd, &csi_cfg,
+ sizeof(csi_cfg),
+ GB_APB_REQUEST_CSI_TX_CONTROL, false);
+
+ if (ret < 0)
+ gcam_err(gcam, "failed to stop the CSI transmitter\n");
+
+ /* Set the UniPro link to low speed mode. */
+ gb_camera_set_power_mode(gcam, false);
+
+ /* Destroy the data connection. */
+ gb_connection_disable(gcam->data_connection);
+ gb_connection_destroy(gcam->data_connection);
+ gcam->data_connection = NULL;
+}
+
+/* -----------------------------------------------------------------------------
+ * Camera Protocol Operations
+ */
+
+static int gb_camera_capabilities(struct gb_camera *gcam,
+ u8 *capabilities, size_t *size)
+{
+ int ret;
+
+ ret = gb_pm_runtime_get_sync(gcam->bundle);
+ if (ret)
+ return ret;
+
+ mutex_lock(&gcam->mutex);
+
+ if (!gcam->connection) {
+ ret = -EINVAL;
+ goto done;
+ }
+
+ ret = gb_camera_operation_sync_flags(gcam->connection,
+ GB_CAMERA_TYPE_CAPABILITIES,
+ GB_OPERATION_FLAG_SHORT_RESPONSE,
+ NULL, 0,
+ (void *)capabilities, size);
+ if (ret)
+ gcam_err(gcam, "failed to retrieve capabilities: %d\n", ret);
+
+done:
+ mutex_unlock(&gcam->mutex);
+
+ gb_pm_runtime_put_autosuspend(gcam->bundle);
+
+ return ret;
+}
+
+static int gb_camera_configure_streams(struct gb_camera *gcam,
+ unsigned int *num_streams,
+ unsigned int *flags,
+ struct gb_camera_stream_config *streams,
+ struct gb_camera_csi_params *csi_params)
+{
+ struct gb_camera_configure_streams_request *req;
+ struct gb_camera_configure_streams_response *resp;
+ unsigned int nstreams = *num_streams;
+ unsigned int i;
+ size_t req_size;
+ size_t resp_size;
+ int ret;
+
+ if (nstreams > GB_CAMERA_MAX_STREAMS)
+ return -EINVAL;
+
+ req_size = sizeof(*req) + nstreams * sizeof(req->config[0]);
+ resp_size = sizeof(*resp) + nstreams * sizeof(resp->config[0]);
+
+ req = kmalloc(req_size, GFP_KERNEL);
+ resp = kmalloc(resp_size, GFP_KERNEL);
+ if (!req || !resp) {
+ kfree(req);
+ kfree(resp);
+ return -ENOMEM;
+ }
+
+ req->num_streams = nstreams;
+ req->flags = *flags;
+ req->padding = 0;
+
+ for (i = 0; i < nstreams; ++i) {
+ struct gb_camera_stream_config_request *cfg = &req->config[i];
+
+ cfg->width = cpu_to_le16(streams[i].width);
+ cfg->height = cpu_to_le16(streams[i].height);
+ cfg->format = cpu_to_le16(streams[i].format);
+ cfg->padding = 0;
+ }
+
+ mutex_lock(&gcam->mutex);
+
+ ret = gb_pm_runtime_get_sync(gcam->bundle);
+ if (ret)
+ goto done_skip_pm_put;
+
+ if (!gcam->connection) {
+ ret = -EINVAL;
+ goto done;
+ }
+
+ ret = gb_camera_operation_sync_flags(gcam->connection,
+ GB_CAMERA_TYPE_CONFIGURE_STREAMS,
+ GB_OPERATION_FLAG_SHORT_RESPONSE,
+ req, req_size,
+ resp, &resp_size);
+ if (ret < 0)
+ goto done;
+
+ ret = gb_camera_configure_streams_validate_response(gcam, resp,
+ nstreams);
+ if (ret < 0)
+ goto done;
+
+ *flags = resp->flags;
+ *num_streams = resp->num_streams;
+
+ for (i = 0; i < resp->num_streams; ++i) {
+ struct gb_camera_stream_config_response *cfg = &resp->config[i];
+
+ streams[i].width = le16_to_cpu(cfg->width);
+ streams[i].height = le16_to_cpu(cfg->height);
+ streams[i].format = le16_to_cpu(cfg->format);
+ streams[i].vc = cfg->virtual_channel;
+ streams[i].dt[0] = cfg->data_type[0];
+ streams[i].dt[1] = cfg->data_type[1];
+ streams[i].max_size = le32_to_cpu(cfg->max_size);
+ }
+
+ if ((resp->flags & GB_CAMERA_CONFIGURE_STREAMS_ADJUSTED) ||
+ (req->flags & GB_CAMERA_CONFIGURE_STREAMS_TEST_ONLY))
+ goto done;
+
+ if (gcam->state == GB_CAMERA_STATE_CONFIGURED) {
+ gb_camera_teardown_data_connection(gcam);
+ gcam->state = GB_CAMERA_STATE_UNCONFIGURED;
+
+ /*
+ * When unconfiguring streams release the PM runtime reference
+ * that was acquired when streams were configured. The bundle
+ * won't be suspended until the PM runtime reference acquired at
+ * the beginning of this function gets released right before
+ * returning.
+ */
+ gb_pm_runtime_put_noidle(gcam->bundle);
+ }
+
+ if (resp->num_streams == 0)
+ goto done;
+
+ /*
+ * Make sure the bundle won't be suspended until streams get
+ * unconfigured after the stream is configured successfully
+ */
+ gb_pm_runtime_get_noresume(gcam->bundle);
+
+ /* Setup CSI-2 connection from APB-A to AP */
+ ret = gb_camera_setup_data_connection(gcam, resp, csi_params);
+ if (ret < 0) {
+ memset(req, 0, sizeof(*req));
+ gb_operation_sync(gcam->connection,
+ GB_CAMERA_TYPE_CONFIGURE_STREAMS,
+ req, sizeof(*req),
+ resp, sizeof(*resp));
+ *flags = 0;
+ *num_streams = 0;
+ gb_pm_runtime_put_noidle(gcam->bundle);
+ goto done;
+ }
+
+ gcam->state = GB_CAMERA_STATE_CONFIGURED;
+
+done:
+ gb_pm_runtime_put_autosuspend(gcam->bundle);
+
+done_skip_pm_put:
+ mutex_unlock(&gcam->mutex);
+ kfree(req);
+ kfree(resp);
+ return ret;
+}
+
+static int gb_camera_capture(struct gb_camera *gcam, u32 request_id,
+ unsigned int streams, unsigned int num_frames,
+ size_t settings_size, const void *settings)
+{
+ struct gb_camera_capture_request *req;
+ size_t req_size;
+ int ret;
+
+ if (settings_size > GB_CAMERA_MAX_SETTINGS_SIZE)
+ return -EINVAL;
+
+ req_size = sizeof(*req) + settings_size;
+ req = kmalloc(req_size, GFP_KERNEL);
+ if (!req)
+ return -ENOMEM;
+
+ req->request_id = cpu_to_le32(request_id);
+ req->streams = streams;
+ req->padding = 0;
+ req->num_frames = cpu_to_le16(num_frames);
+ memcpy(req->settings, settings, settings_size);
+
+ mutex_lock(&gcam->mutex);
+
+ if (!gcam->connection) {
+ ret = -EINVAL;
+ goto done;
+ }
+
+ ret = gb_operation_sync(gcam->connection, GB_CAMERA_TYPE_CAPTURE,
+ req, req_size, NULL, 0);
+done:
+ mutex_unlock(&gcam->mutex);
+
+ kfree(req);
+
+ return ret;
+}
+
+static int gb_camera_flush(struct gb_camera *gcam, u32 *request_id)
+{
+ struct gb_camera_flush_response resp;
+ int ret;
+
+ mutex_lock(&gcam->mutex);
+
+ if (!gcam->connection) {
+ ret = -EINVAL;
+ goto done;
+ }
+
+ ret = gb_operation_sync(gcam->connection, GB_CAMERA_TYPE_FLUSH, NULL, 0,
+ &resp, sizeof(resp));
+
+ if (ret < 0)
+ goto done;
+
+ if (request_id)
+ *request_id = le32_to_cpu(resp.request_id);
+
+done:
+ mutex_unlock(&gcam->mutex);
+
+ return ret;
+}
+
+static int gb_camera_request_handler(struct gb_operation *op)
+{
+ struct gb_camera *gcam = gb_connection_get_data(op->connection);
+ struct gb_camera_metadata_request *payload;
+ struct gb_message *request;
+
+ if (op->type != GB_CAMERA_TYPE_METADATA) {
+ gcam_err(gcam, "Unsupported unsolicited event: %u\n", op->type);
+ return -EINVAL;
+ }
+
+ request = op->request;
+
+ if (request->payload_size < sizeof(*payload)) {
+ gcam_err(gcam, "Wrong event size received (%zu < %zu)\n",
+ request->payload_size, sizeof(*payload));
+ return -EINVAL;
+ }
+
+ payload = request->payload;
+
+ gcam_dbg(gcam, "received metadata for request %u, frame %u, stream %u\n",
+ payload->request_id, payload->frame_number, payload->stream);
+
+ return 0;
+}
+
+/* -----------------------------------------------------------------------------
+ * Interface with HOST gmp camera.
+ */
+static unsigned int gb_camera_mbus_to_gb(enum v4l2_mbus_pixelcode mbus_code)
+{
+ unsigned int i;
+
+ for (i = 0; i < ARRAY_SIZE(gb_fmt_info); i++) {
+ if (gb_fmt_info[i].mbus_code == mbus_code)
+ return gb_fmt_info[i].gb_format;
+ }
+ return gb_fmt_info[0].gb_format;
+}
+
+static enum v4l2_mbus_pixelcode gb_camera_gb_to_mbus(u16 gb_fmt)
+{
+ unsigned int i;
+
+ for (i = 0; i < ARRAY_SIZE(gb_fmt_info); i++) {
+ if (gb_fmt_info[i].gb_format == gb_fmt)
+ return gb_fmt_info[i].mbus_code;
+ }
+ return gb_fmt_info[0].mbus_code;
+}
+
+static ssize_t gb_camera_op_capabilities(void *priv, char *data, size_t len)
+{
+ struct gb_camera *gcam = priv;
+ size_t capabilities_len = len;
+ int ret;
+
+ ret = gb_camera_capabilities(gcam, data, &capabilities_len);
+ if (ret)
+ return ret;
+
+ return capabilities_len;
+}
+
+static int gb_camera_op_configure_streams(void *priv, unsigned int *nstreams,
+ unsigned int *flags, struct gb_camera_stream *streams,
+ struct gb_camera_csi_params *csi_params)
+{
+ struct gb_camera *gcam = priv;
+ struct gb_camera_stream_config *gb_streams;
+ unsigned int gb_flags = 0;
+ unsigned int gb_nstreams = *nstreams;
+ unsigned int i;
+ int ret;
+
+ if (gb_nstreams > GB_CAMERA_MAX_STREAMS)
+ return -EINVAL;
+
+ gb_streams = kzalloc(gb_nstreams * sizeof(*gb_streams), GFP_KERNEL);
+ if (!gb_streams)
+ return -ENOMEM;
+
+ for (i = 0; i < gb_nstreams; i++) {
+ gb_streams[i].width = streams[i].width;
+ gb_streams[i].height = streams[i].height;
+ gb_streams[i].format =
+ gb_camera_mbus_to_gb(streams[i].pixel_code);
+ }
+
+ if (*flags & GB_CAMERA_IN_FLAG_TEST)
+ gb_flags |= GB_CAMERA_CONFIGURE_STREAMS_TEST_ONLY;
+
+ ret = gb_camera_configure_streams(gcam, &gb_nstreams,
+ &gb_flags, gb_streams, csi_params);
+ if (ret < 0)
+ goto done;
+ if (gb_nstreams > *nstreams) {
+ ret = -EINVAL;
+ goto done;
+ }
+
+ *flags = 0;
+ if (gb_flags & GB_CAMERA_CONFIGURE_STREAMS_ADJUSTED)
+ *flags |= GB_CAMERA_OUT_FLAG_ADJUSTED;
+
+ for (i = 0; i < gb_nstreams; i++) {
+ streams[i].width = gb_streams[i].width;
+ streams[i].height = gb_streams[i].height;
+ streams[i].vc = gb_streams[i].vc;
+ streams[i].dt[0] = gb_streams[i].dt[0];
+ streams[i].dt[1] = gb_streams[i].dt[1];
+ streams[i].max_size = gb_streams[i].max_size;
+ streams[i].pixel_code =
+ gb_camera_gb_to_mbus(gb_streams[i].format);
+ }
+ *nstreams = gb_nstreams;
+
+done:
+ kfree(gb_streams);
+ return ret;
+}
+
+static int gb_camera_op_capture(void *priv, u32 request_id,
+ unsigned int streams, unsigned int num_frames,
+ size_t settings_size, const void *settings)
+{
+ struct gb_camera *gcam = priv;
+
+ return gb_camera_capture(gcam, request_id, streams, num_frames,
+ settings_size, settings);
+}
+
+static int gb_camera_op_flush(void *priv, u32 *request_id)
+{
+ struct gb_camera *gcam = priv;
+
+ return gb_camera_flush(gcam, request_id);
+}
+
+static const struct gb_camera_ops gb_cam_ops = {
+ .capabilities = gb_camera_op_capabilities,
+ .configure_streams = gb_camera_op_configure_streams,
+ .capture = gb_camera_op_capture,
+ .flush = gb_camera_op_flush,
+};
+
+/* -----------------------------------------------------------------------------
+ * DebugFS
+ */
+
+static ssize_t gb_camera_debugfs_capabilities(struct gb_camera *gcam,
+ char *buf, size_t len)
+{
+ struct gb_camera_debugfs_buffer *buffer =
+ &gcam->debugfs.buffers[GB_CAMERA_DEBUGFS_BUFFER_CAPABILITIES];
+ size_t size = 1024;
+ unsigned int i;
+ u8 *caps;
+ int ret;
+
+ caps = kmalloc(size, GFP_KERNEL);
+ if (!caps)
+ return -ENOMEM;
+
+ ret = gb_camera_capabilities(gcam, caps, &size);
+ if (ret < 0)
+ goto done;
+
+ /*
+ * hex_dump_to_buffer() doesn't return the number of bytes dumped prior
+ * to v4.0, we need our own implementation :-(
+ */
+ buffer->length = 0;
+
+ for (i = 0; i < size; i += 16) {
+ unsigned int nbytes = min_t(unsigned int, size - i, 16);
+
+ buffer->length += sprintf(buffer->data + buffer->length,
+ "%*ph\n", nbytes, caps + i);
+ }
+
+done:
+ kfree(caps);
+ return ret;
+}
+
+static ssize_t gb_camera_debugfs_configure_streams(struct gb_camera *gcam,
+ char *buf, size_t len)
+{
+ struct gb_camera_debugfs_buffer *buffer =
+ &gcam->debugfs.buffers[GB_CAMERA_DEBUGFS_BUFFER_STREAMS];
+ struct gb_camera_stream_config *streams;
+ unsigned int nstreams;
+ unsigned int flags;
+ unsigned int i;
+ char *token;
+ int ret;
+
+ /* Retrieve number of streams to configure */
+ token = strsep(&buf, ";");
+ if (token == NULL)
+ return -EINVAL;
+
+ ret = kstrtouint(token, 10, &nstreams);
+ if (ret < 0)
+ return ret;
+
+ if (nstreams > GB_CAMERA_MAX_STREAMS)
+ return -EINVAL;
+
+ token = strsep(&buf, ";");
+ if (token == NULL)
+ return -EINVAL;
+
+ ret = kstrtouint(token, 10, &flags);
+ if (ret < 0)
+ return ret;
+
+ /* For each stream to configure parse width, height and format */
+ streams = kzalloc(nstreams * sizeof(*streams), GFP_KERNEL);
+ if (!streams)
+ return -ENOMEM;
+
+ for (i = 0; i < nstreams; ++i) {
+ struct gb_camera_stream_config *stream = &streams[i];
+
+ /* width */
+ token = strsep(&buf, ";");
+ if (token == NULL) {
+ ret = -EINVAL;
+ goto done;
+ }
+ ret = kstrtouint(token, 10, &stream->width);
+ if (ret < 0)
+ goto done;
+
+ /* height */
+ token = strsep(&buf, ";");
+ if (token == NULL)
+ goto done;
+
+ ret = kstrtouint(token, 10, &stream->height);
+ if (ret < 0)
+ goto done;
+
+ /* Image format code */
+ token = strsep(&buf, ";");
+ if (token == NULL)
+ goto done;
+
+ ret = kstrtouint(token, 16, &stream->format);
+ if (ret < 0)
+ goto done;
+ }
+
+ ret = gb_camera_configure_streams(gcam, &nstreams, &flags, streams,
+ NULL);
+ if (ret < 0)
+ goto done;
+
+ buffer->length = sprintf(buffer->data, "%u;%u;", nstreams, flags);
+
+ for (i = 0; i < nstreams; ++i) {
+ struct gb_camera_stream_config *stream = &streams[i];
+
+ buffer->length += sprintf(buffer->data + buffer->length,
+ "%u;%u;%u;%u;%u;%u;%u;",
+ stream->width, stream->height,
+ stream->format, stream->vc,
+ stream->dt[0], stream->dt[1],
+ stream->max_size);
+ }
+
+ ret = len;
+
+done:
+ kfree(streams);
+ return ret;
+};
+
+static ssize_t gb_camera_debugfs_capture(struct gb_camera *gcam,
+ char *buf, size_t len)
+{
+ unsigned int request_id;
+ unsigned int streams_mask;
+ unsigned int num_frames;
+ char *token;
+ int ret;
+
+ /* Request id */
+ token = strsep(&buf, ";");
+ if (token == NULL)
+ return -EINVAL;
+ ret = kstrtouint(token, 10, &request_id);
+ if (ret < 0)
+ return ret;
+
+ /* Stream mask */
+ token = strsep(&buf, ";");
+ if (token == NULL)
+ return -EINVAL;
+ ret = kstrtouint(token, 16, &streams_mask);
+ if (ret < 0)
+ return ret;
+
+ /* number of frames */
+ token = strsep(&buf, ";");
+ if (token == NULL)
+ return -EINVAL;
+ ret = kstrtouint(token, 10, &num_frames);
+ if (ret < 0)
+ return ret;
+
+ ret = gb_camera_capture(gcam, request_id, streams_mask, num_frames, 0,
+ NULL);
+ if (ret < 0)
+ return ret;
+
+ return len;
+}
+
+static ssize_t gb_camera_debugfs_flush(struct gb_camera *gcam,
+ char *buf, size_t len)
+{
+ struct gb_camera_debugfs_buffer *buffer =
+ &gcam->debugfs.buffers[GB_CAMERA_DEBUGFS_BUFFER_FLUSH];
+ unsigned int req_id;
+ int ret;
+
+ ret = gb_camera_flush(gcam, &req_id);
+ if (ret < 0)
+ return ret;
+
+ buffer->length = sprintf(buffer->data, "%u", req_id);
+
+ return len;
+}
+
+struct gb_camera_debugfs_entry {
+ const char *name;
+ unsigned int mask;
+ unsigned int buffer;
+ ssize_t (*execute)(struct gb_camera *gcam, char *buf, size_t len);
+};
+
+static const struct gb_camera_debugfs_entry gb_camera_debugfs_entries[] = {
+ {
+ .name = "capabilities",
+ .mask = S_IFREG | S_IRUGO,
+ .buffer = GB_CAMERA_DEBUGFS_BUFFER_CAPABILITIES,
+ .execute = gb_camera_debugfs_capabilities,
+ }, {
+ .name = "configure_streams",
+ .mask = S_IFREG | S_IRUGO | S_IWUGO,
+ .buffer = GB_CAMERA_DEBUGFS_BUFFER_STREAMS,
+ .execute = gb_camera_debugfs_configure_streams,
+ }, {
+ .name = "capture",
+ .mask = S_IFREG | S_IRUGO | S_IWUGO,
+ .buffer = GB_CAMERA_DEBUGFS_BUFFER_CAPTURE,
+ .execute = gb_camera_debugfs_capture,
+ }, {
+ .name = "flush",
+ .mask = S_IFREG | S_IRUGO | S_IWUGO,
+ .buffer = GB_CAMERA_DEBUGFS_BUFFER_FLUSH,
+ .execute = gb_camera_debugfs_flush,
+ },
+};
+
+static ssize_t gb_camera_debugfs_read(struct file *file, char __user *buf,
+ size_t len, loff_t *offset)
+{
+ const struct gb_camera_debugfs_entry *op = file->private_data;
+ struct gb_camera *gcam = file->f_inode->i_private;
+ struct gb_camera_debugfs_buffer *buffer;
+ ssize_t ret;
+
+ /* For read-only entries the operation is triggered by a read. */
+ if (!(op->mask & S_IWUGO)) {
+ ret = op->execute(gcam, NULL, 0);
+ if (ret < 0)
+ return ret;
+ }
+
+ buffer = &gcam->debugfs.buffers[op->buffer];
+
+ return simple_read_from_buffer(buf, len, offset, buffer->data,
+ buffer->length);
+}
+
+static ssize_t gb_camera_debugfs_write(struct file *file,
+ const char __user *buf, size_t len,
+ loff_t *offset)
+{
+ const struct gb_camera_debugfs_entry *op = file->private_data;
+ struct gb_camera *gcam = file->f_inode->i_private;
+ ssize_t ret;
+ char *kbuf;
+
+ if (len > 1024)
+ return -EINVAL;
+
+ kbuf = kmalloc(len + 1, GFP_KERNEL);
+ if (kbuf == NULL)
+ return -ENOMEM;
+
+ if (copy_from_user(kbuf, buf, len)) {
+ ret = -EFAULT;
+ goto done;
+ }
+
+ kbuf[len] = '\0';
+
+ ret = op->execute(gcam, kbuf, len);
+
+done:
+ kfree(kbuf);
+ return ret;
+}
+
+static int gb_camera_debugfs_open(struct inode *inode, struct file *file)
+{
+ unsigned int i;
+
+ for (i = 0; i < ARRAY_SIZE(gb_camera_debugfs_entries); ++i) {
+ const struct gb_camera_debugfs_entry *entry =
+ &gb_camera_debugfs_entries[i];
+
+ if (!strcmp(file->f_path.dentry->d_iname, entry->name)) {
+ file->private_data = (void *)entry;
+ break;
+ }
+ }
+
+ return 0;
+}
+
+static const struct file_operations gb_camera_debugfs_ops = {
+ .open = gb_camera_debugfs_open,
+ .read = gb_camera_debugfs_read,
+ .write = gb_camera_debugfs_write,
+};
+
+static int gb_camera_debugfs_init(struct gb_camera *gcam)
+{
+ struct gb_connection *connection = gcam->connection;
+ char dirname[27];
+ unsigned int i;
+
+ /*
+ * Create root debugfs entry and a file entry for each camera operation.
+ */
+ snprintf(dirname, 27, "camera-%u.%u", connection->intf->interface_id,
+ gcam->bundle->id);
+
+ gcam->debugfs.root = debugfs_create_dir(dirname, gb_debugfs_get());
+ if (IS_ERR(gcam->debugfs.root)) {
+ gcam_err(gcam, "debugfs root create failed (%ld)\n",
+ PTR_ERR(gcam->debugfs.root));
+ return PTR_ERR(gcam->debugfs.root);
+ }
+
+ gcam->debugfs.buffers = vmalloc(sizeof(*gcam->debugfs.buffers) *
+ GB_CAMERA_DEBUGFS_BUFFER_MAX);
+ if (!gcam->debugfs.buffers)
+ return -ENOMEM;
+
+ for (i = 0; i < ARRAY_SIZE(gb_camera_debugfs_entries); ++i) {
+ const struct gb_camera_debugfs_entry *entry =
+ &gb_camera_debugfs_entries[i];
+ struct dentry *dentry;
+
+ gcam->debugfs.buffers[i].length = 0;
+
+ dentry = debugfs_create_file(entry->name, entry->mask,
+ gcam->debugfs.root, gcam,
+ &gb_camera_debugfs_ops);
+ if (IS_ERR(dentry)) {
+ gcam_err(gcam,
+ "debugfs operation %s create failed (%ld)\n",
+ entry->name, PTR_ERR(dentry));
+ return PTR_ERR(dentry);
+ }
+ }
+
+ return 0;
+}
+
+static void gb_camera_debugfs_cleanup(struct gb_camera *gcam)
+{
+ debugfs_remove_recursive(gcam->debugfs.root);
+
+ vfree(gcam->debugfs.buffers);
+}
+
+/* -----------------------------------------------------------------------------
+ * Init & Cleanup
+ */
+
+static void gb_camera_cleanup(struct gb_camera *gcam)
+{
+ gb_camera_debugfs_cleanup(gcam);
+
+ mutex_lock(&gcam->mutex);
+ if (gcam->data_connection) {
+ gb_connection_disable(gcam->data_connection);
+ gb_connection_destroy(gcam->data_connection);
+ gcam->data_connection = NULL;
+ }
+
+ if (gcam->connection) {
+ gb_connection_disable(gcam->connection);
+ gb_connection_destroy(gcam->connection);
+ gcam->connection = NULL;
+ }
+ mutex_unlock(&gcam->mutex);
+}
+
+static void gb_camera_release_module(struct kref *ref)
+{
+ struct gb_camera_module *cam_mod =
+ container_of(ref, struct gb_camera_module, refcount);
+ kfree(cam_mod->priv);
+}
+
+static int gb_camera_probe(struct gb_bundle *bundle,
+ const struct greybus_bundle_id *id)
+{
+ struct gb_connection *conn;
+ struct gb_camera *gcam;
+ u16 mgmt_cport_id = 0;
+ u16 data_cport_id = 0;
+ unsigned int i;
+ int ret;
+
+ /*
+ * The camera bundle must contain exactly two CPorts, one for the
+ * camera management protocol and one for the camera data protocol.
+ */
+ if (bundle->num_cports != 2)
+ return -ENODEV;
+
+ for (i = 0; i < bundle->num_cports; ++i) {
+ struct greybus_descriptor_cport *desc = &bundle->cport_desc[i];
+
+ switch (desc->protocol_id) {
+ case GREYBUS_PROTOCOL_CAMERA_MGMT:
+ mgmt_cport_id = le16_to_cpu(desc->id);
+ break;
+ case GREYBUS_PROTOCOL_CAMERA_DATA:
+ data_cport_id = le16_to_cpu(desc->id);
+ break;
+ default:
+ return -ENODEV;
+ }
+ }
+
+ if (!mgmt_cport_id || !data_cport_id)
+ return -ENODEV;
+
+ gcam = kzalloc(sizeof(*gcam), GFP_KERNEL);
+ if (!gcam)
+ return -ENOMEM;
+
+ mutex_init(&gcam->mutex);
+
+ gcam->bundle = bundle;
+ gcam->state = GB_CAMERA_STATE_UNCONFIGURED;
+ gcam->data_cport_id = data_cport_id;
+
+ conn = gb_connection_create(bundle, mgmt_cport_id,
+ gb_camera_request_handler);
+ if (IS_ERR(conn)) {
+ ret = PTR_ERR(conn);
+ goto error;
+ }
+
+ gcam->connection = conn;
+ gb_connection_set_data(conn, gcam);
+
+ ret = gb_connection_enable(conn);
+ if (ret)
+ goto error;
+
+ ret = gb_camera_debugfs_init(gcam);
+ if (ret < 0)
+ goto error;
+
+ gcam->module.priv = gcam;
+ gcam->module.ops = &gb_cam_ops;
+ gcam->module.interface_id = gcam->connection->intf->interface_id;
+ gcam->module.release = gb_camera_release_module;
+ ret = gb_camera_register(&gcam->module);
+ if (ret < 0)
+ goto error;
+
+ greybus_set_drvdata(bundle, gcam);
+
+ gb_pm_runtime_put_autosuspend(gcam->bundle);
+
+ return 0;
+
+error:
+ gb_camera_cleanup(gcam);
+ kfree(gcam);
+ return ret;
+}
+
+static void gb_camera_disconnect(struct gb_bundle *bundle)
+{
+ struct gb_camera *gcam = greybus_get_drvdata(bundle);
+ int ret;
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ gb_pm_runtime_get_noresume(bundle);
+
+ gb_camera_cleanup(gcam);
+ gb_camera_unregister(&gcam->module);
+}
+
+static const struct greybus_bundle_id gb_camera_id_table[] = {
+ { GREYBUS_DEVICE_CLASS(GREYBUS_CLASS_CAMERA) },
+ { },
+};
+
+#ifdef CONFIG_PM
+static int gb_camera_suspend(struct device *dev)
+{
+ struct gb_bundle *bundle = to_gb_bundle(dev);
+ struct gb_camera *gcam = greybus_get_drvdata(bundle);
+
+ if (gcam->data_connection)
+ gb_connection_disable(gcam->data_connection);
+
+ gb_connection_disable(gcam->connection);
+
+ return 0;
+}
+
+static int gb_camera_resume(struct device *dev)
+{
+ struct gb_bundle *bundle = to_gb_bundle(dev);
+ struct gb_camera *gcam = greybus_get_drvdata(bundle);
+ int ret;
+
+ ret = gb_connection_enable(gcam->connection);
+ if (ret) {
+ gcam_err(gcam, "failed to enable connection: %d\n", ret);
+ return ret;
+ }
+
+ if (gcam->data_connection) {
+ ret = gb_connection_enable(gcam->data_connection);
+ if (ret) {
+ gcam_err(gcam,
+ "failed to enable data connection: %d\n", ret);
+ return ret;
+ }
+ }
+
+ return 0;
+}
+#endif
+
+static const struct dev_pm_ops gb_camera_pm_ops = {
+ SET_RUNTIME_PM_OPS(gb_camera_suspend, gb_camera_resume, NULL)
+};
+
+static struct greybus_driver gb_camera_driver = {
+ .name = "camera",
+ .probe = gb_camera_probe,
+ .disconnect = gb_camera_disconnect,
+ .id_table = gb_camera_id_table,
+ .driver.pm = &gb_camera_pm_ops,
+};
+
+module_greybus_driver(gb_camera_driver);
+
+MODULE_LICENSE("GPL v2");
--- /dev/null
+++ b/drivers/greybus/gb-camera.h
@@ -0,0 +1,127 @@
+/*
+ * Greybus Camera protocol driver.
+ *
+ * Copyright 2015 Google Inc.
+ *
+ * Released under the GPLv2 only.
+ */
+#ifndef __GB_CAMERA_H
+#define __GB_CAMERA_H
+
+#include <linux/v4l2-mediabus.h>
+
+/* Input flags need to be set from the caller */
+#define GB_CAMERA_IN_FLAG_TEST (1 << 0)
+/* Output flags returned */
+#define GB_CAMERA_OUT_FLAG_ADJUSTED (1 << 0)
+
+/**
+ * struct gb_camera_stream - Represents greybus camera stream.
+ * @width: Stream width in pixels.
+ * @height: Stream height in pixels.
+ * @pixel_code: Media bus pixel code.
+ * @vc: MIPI CSI virtual channel.
+ * @dt: MIPI CSI data types. Most formats use a single data type, in which case
+ * the second element will be ignored.
+ * @max_size: Maximum size of a frame in bytes. The camera module guarantees
+ * that all data between the Frame Start and Frame End packet for
+ * the associated virtual channel and data type(s) will not exceed
+ * this size.
+ */
+struct gb_camera_stream {
+ unsigned int width;
+ unsigned int height;
+ enum v4l2_mbus_pixelcode pixel_code;
+ unsigned int vc;
+ unsigned int dt[2];
+ unsigned int max_size;
+};
+
+/**
+ * struct gb_camera_csi_params - CSI configuration parameters
+ * @num_lanes: number of CSI data lanes
+ * @clk_freq: CSI clock frequency in Hz
+ */
+struct gb_camera_csi_params {
+ unsigned int num_lanes;
+ unsigned int clk_freq;
+};
+
+/**
+ * struct gb_camera_ops - Greybus camera operations, used by the Greybus camera
+ * driver to expose operations to the host camera driver.
+ * @capabilities: Retrieve camera capabilities and store them in the buffer
+ * 'buf' capabilities. The buffer maximum size is specified by
+ * the caller in the 'size' parameter, and the effective
+ * capabilities size is returned from the function. If the buffer
+ * size is too small to hold the capabilities an error is
+ * returned and the buffer is left untouched.
+ *
+ * @configure_streams: Negotiate configuration and prepare the module for video
+ * capture. The caller specifies the number of streams it
+ * requests in the 'nstreams' argument and the associated
+ * streams configurations in the 'streams' argument. The
+ * GB_CAMERA_IN_FLAG_TEST 'flag' can be set to test a
+ * configuration without applying it, otherwise the
+ * configuration is applied by the module. The module can
+ * decide to modify the requested configuration, including
+ * using a different number of streams. In that case the
+ * modified configuration won't be applied, the
+ * GB_CAMERA_OUT_FLAG_ADJUSTED 'flag' will be set upon
+ * return, and the modified configuration and number of
+ * streams stored in 'streams' and 'array'. The module
+ * returns its CSI-2 bus parameters in the 'csi_params'
+ * structure in all cases.
+ *
+ * @capture: Submit a capture request. The supplied 'request_id' must be unique
+ * and higher than the IDs of all the previously submitted requests.
+ * The 'streams' argument specifies which streams are affected by the
+ * request in the form of a bitmask, with bits corresponding to the
+ * configured streams indexes. If the request contains settings, the
+ * 'settings' argument points to the settings buffer and its size is
+ * specified by the 'settings_size' argument. Otherwise the 'settings'
+ * argument should be set to NULL and 'settings_size' to 0.
+ *
+ * @flush: Flush the capture requests queue. Return the ID of the last request
+ * that will processed by the device before it stops transmitting video
+ * frames. All queued capture requests with IDs higher than the returned
+ * ID will be dropped without being processed.
+ */
+struct gb_camera_ops {
+ ssize_t (*capabilities)(void *priv, char *buf, size_t len);
+ int (*configure_streams)(void *priv, unsigned int *nstreams,
+ unsigned int *flags, struct gb_camera_stream *streams,
+ struct gb_camera_csi_params *csi_params);
+ int (*capture)(void *priv, u32 request_id,
+ unsigned int streams, unsigned int num_frames,
+ size_t settings_size, const void *settings);
+ int (*flush)(void *priv, u32 *request_id);
+};
+
+/**
+ * struct gb_camera_module - Represents greybus camera module.
+ * @priv: Module private data, passed to all camera operations.
+ * @ops: Greybus camera operation callbacks.
+ * @interface_id: Interface id of the module.
+ * @refcount: Reference counting object.
+ * @release: Module release function.
+ * @list: List entry in the camera modules list.
+ */
+struct gb_camera_module {
+ void *priv;
+ const struct gb_camera_ops *ops;
+
+ unsigned int interface_id;
+ struct kref refcount;
+ void (*release)(struct kref *kref);
+ struct list_head list; /* Global list */
+};
+
+#define gb_camera_call(f, op, args...) \
+ (!(f) ? -ENODEV : (((f)->ops->op) ? \
+ (f)->ops->op((f)->priv, ##args) : -ENOIOCTLCMD))
+
+int gb_camera_register(struct gb_camera_module *module);
+int gb_camera_unregister(struct gb_camera_module *module);
+
+#endif /* __GB_CAMERA_H */
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 12/32] greybus: es2 host driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (9 preceding siblings ...)
2016-09-16 13:25 ` [patch 11/32] greybus: camera driver Greg KH
@ 2016-09-16 13:25 ` Greg KH
2016-10-07 13:43 ` Pavel Machek
2016-09-16 14:09 ` [patch 13/32] greybus: HID driver Greg KH
` (19 subsequent siblings)
30 siblings, 1 reply; 84+ messages in thread
From: Greg KH @ 2016-09-16 13:25 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This is a Greybus host driver for the ES2/3 bridge chip from Toshiba.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/es2.c | 1597 ++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 1597 insertions(+)
--- /dev/null
+++ b/drivers/greybus/es2.c
@@ -0,0 +1,1597 @@
+/*
+ * Greybus "AP" USB driver for "ES2" controller chips
+ *
+ * Copyright 2014-2015 Google Inc.
+ * Copyright 2014-2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+#include <linux/kthread.h>
+#include <linux/sizes.h>
+#include <linux/usb.h>
+#include <linux/kfifo.h>
+#include <linux/debugfs.h>
+#include <linux/list.h>
+#include <asm/unaligned.h>
+
+#include "arpc.h"
+#include "greybus.h"
+#include "greybus_trace.h"
+#include "connection.h"
+
+
+/* Default timeout for USB vendor requests. */
+#define ES2_USB_CTRL_TIMEOUT 500
+
+/* Default timeout for ARPC CPort requests */
+#define ES2_ARPC_CPORT_TIMEOUT 500
+
+/* Fixed CPort numbers */
+#define ES2_CPORT_CDSI0 16
+#define ES2_CPORT_CDSI1 17
+
+/* Memory sizes for the buffers sent to/from the ES2 controller */
+#define ES2_GBUF_MSG_SIZE_MAX 2048
+
+/* Memory sizes for the ARPC buffers */
+#define ARPC_OUT_SIZE_MAX U16_MAX
+#define ARPC_IN_SIZE_MAX 128
+
+static const struct usb_device_id id_table[] = {
+ { USB_DEVICE(0x18d1, 0x1eaf) },
+ { },
+};
+MODULE_DEVICE_TABLE(usb, id_table);
+
+#define APB1_LOG_SIZE SZ_16K
+
+/*
+ * Number of CPort IN urbs in flight at any point in time.
+ * Adjust if we are having stalls in the USB buffer due to not enough urbs in
+ * flight.
+ */
+#define NUM_CPORT_IN_URB 4
+
+/* Number of CPort OUT urbs in flight at any point in time.
+ * Adjust if we get messages saying we are out of urbs in the system log.
+ */
+#define NUM_CPORT_OUT_URB 8
+
+/*
+ * Number of ARPC in urbs in flight at any point in time.
+ */
+#define NUM_ARPC_IN_URB 2
+
+/*
+ * @endpoint: bulk in endpoint for CPort data
+ * @urb: array of urbs for the CPort in messages
+ * @buffer: array of buffers for the @cport_in_urb urbs
+ */
+struct es2_cport_in {
+ __u8 endpoint;
+ struct urb *urb[NUM_CPORT_IN_URB];
+ u8 *buffer[NUM_CPORT_IN_URB];
+};
+
+/**
+ * es2_ap_dev - ES2 USB Bridge to AP structure
+ * @usb_dev: pointer to the USB device we are.
+ * @usb_intf: pointer to the USB interface we are bound to.
+ * @hd: pointer to our gb_host_device structure
+
+ * @cport_in: endpoint, urbs and buffer for cport in messages
+ * @cport_out_endpoint: endpoint for for cport out messages
+ * @cport_out_urb: array of urbs for the CPort out messages
+ * @cport_out_urb_busy: array of flags to see if the @cport_out_urb is busy or
+ * not.
+ * @cport_out_urb_cancelled: array of flags indicating whether the
+ * corresponding @cport_out_urb is being cancelled
+ * @cport_out_urb_lock: locks the @cport_out_urb_busy "list"
+ *
+ * @apb_log_task: task pointer for logging thread
+ * @apb_log_dentry: file system entry for the log file interface
+ * @apb_log_enable_dentry: file system entry for enabling logging
+ * @apb_log_fifo: kernel FIFO to carry logged data
+ * @arpc_urb: array of urbs for the ARPC in messages
+ * @arpc_buffer: array of buffers for the @arpc_urb urbs
+ * @arpc_endpoint_in: bulk in endpoint for APBridgeA RPC
+ * @arpc_id_cycle: gives an unique id to ARPC
+ * @arpc_lock: locks ARPC list
+ * @arpcs: list of in progress ARPCs
+ */
+struct es2_ap_dev {
+ struct usb_device *usb_dev;
+ struct usb_interface *usb_intf;
+ struct gb_host_device *hd;
+
+ struct es2_cport_in cport_in;
+ __u8 cport_out_endpoint;
+ struct urb *cport_out_urb[NUM_CPORT_OUT_URB];
+ bool cport_out_urb_busy[NUM_CPORT_OUT_URB];
+ bool cport_out_urb_cancelled[NUM_CPORT_OUT_URB];
+ spinlock_t cport_out_urb_lock;
+
+ bool cdsi1_in_use;
+
+ struct task_struct *apb_log_task;
+ struct dentry *apb_log_dentry;
+ struct dentry *apb_log_enable_dentry;
+ DECLARE_KFIFO(apb_log_fifo, char, APB1_LOG_SIZE);
+
+ __u8 arpc_endpoint_in;
+ struct urb *arpc_urb[NUM_ARPC_IN_URB];
+ u8 *arpc_buffer[NUM_ARPC_IN_URB];
+
+ int arpc_id_cycle;
+ spinlock_t arpc_lock;
+ struct list_head arpcs;
+};
+
+/**
+ * timesync_enable_request - Enable timesync in an APBridge
+ * @count: number of TimeSync Pulses to expect
+ * @frame_time: the initial FrameTime at the first TimeSync Pulse
+ * @strobe_delay: the expected delay in microseconds between each TimeSync Pulse
+ * @refclk: The AP mandated reference clock to run FrameTime at
+ */
+struct timesync_enable_request {
+ __u8 count;
+ __le64 frame_time;
+ __le32 strobe_delay;
+ __le32 refclk;
+} __packed;
+
+/**
+ * timesync_authoritative_request - Transmit authoritative FrameTime to APBridge
+ * @frame_time: An array of authoritative FrameTimes provided by the SVC
+ * and relayed to the APBridge by the AP
+ */
+struct timesync_authoritative_request {
+ __le64 frame_time[GB_TIMESYNC_MAX_STROBES];
+} __packed;
+
+struct arpc {
+ struct list_head list;
+ struct arpc_request_message *req;
+ struct arpc_response_message *resp;
+ struct completion response_received;
+ bool active;
+};
+
+static inline struct es2_ap_dev *hd_to_es2(struct gb_host_device *hd)
+{
+ return (struct es2_ap_dev *)&hd->hd_priv;
+}
+
+static void cport_out_callback(struct urb *urb);
+static void usb_log_enable(struct es2_ap_dev *es2);
+static void usb_log_disable(struct es2_ap_dev *es2);
+static int arpc_sync(struct es2_ap_dev *es2, u8 type, void *payload,
+ size_t size, int *result, unsigned int timeout);
+
+static int output_sync(struct es2_ap_dev *es2, void *req, u16 size, u8 cmd)
+{
+ struct usb_device *udev = es2->usb_dev;
+ u8 *data;
+ int retval;
+
+ data = kmalloc(size, GFP_KERNEL);
+ if (!data)
+ return -ENOMEM;
+ memcpy(data, req, size);
+
+ retval = usb_control_msg(udev, usb_sndctrlpipe(udev, 0),
+ cmd,
+ USB_DIR_OUT | USB_TYPE_VENDOR |
+ USB_RECIP_INTERFACE,
+ 0, 0, data, size, ES2_USB_CTRL_TIMEOUT);
+ if (retval < 0)
+ dev_err(&udev->dev, "%s: return error %d\n", __func__, retval);
+ else
+ retval = 0;
+
+ kfree(data);
+ return retval;
+}
+
+static void ap_urb_complete(struct urb *urb)
+{
+ struct usb_ctrlrequest *dr = urb->context;
+
+ kfree(dr);
+ usb_free_urb(urb);
+}
+
+static int output_async(struct es2_ap_dev *es2, void *req, u16 size, u8 cmd)
+{
+ struct usb_device *udev = es2->usb_dev;
+ struct urb *urb;
+ struct usb_ctrlrequest *dr;
+ u8 *buf;
+ int retval;
+
+ urb = usb_alloc_urb(0, GFP_ATOMIC);
+ if (!urb)
+ return -ENOMEM;
+
+ dr = kmalloc(sizeof(*dr) + size, GFP_ATOMIC);
+ if (!dr) {
+ usb_free_urb(urb);
+ return -ENOMEM;
+ }
+
+ buf = (u8 *)dr + sizeof(*dr);
+ memcpy(buf, req, size);
+
+ dr->bRequest = cmd;
+ dr->bRequestType = USB_DIR_OUT | USB_TYPE_VENDOR | USB_RECIP_INTERFACE;
+ dr->wValue = 0;
+ dr->wIndex = 0;
+ dr->wLength = cpu_to_le16(size);
+
+ usb_fill_control_urb(urb, udev, usb_sndctrlpipe(udev, 0),
+ (unsigned char *)dr, buf, size,
+ ap_urb_complete, dr);
+ retval = usb_submit_urb(urb, GFP_ATOMIC);
+ if (retval) {
+ usb_free_urb(urb);
+ kfree(dr);
+ }
+ return retval;
+}
+
+static int output(struct gb_host_device *hd, void *req, u16 size, u8 cmd,
+ bool async)
+{
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+
+ if (async)
+ return output_async(es2, req, size, cmd);
+
+ return output_sync(es2, req, size, cmd);
+}
+
+static int es2_cport_in_enable(struct es2_ap_dev *es2,
+ struct es2_cport_in *cport_in)
+{
+ struct urb *urb;
+ int ret;
+ int i;
+
+ for (i = 0; i < NUM_CPORT_IN_URB; ++i) {
+ urb = cport_in->urb[i];
+
+ ret = usb_submit_urb(urb, GFP_KERNEL);
+ if (ret) {
+ dev_err(&es2->usb_dev->dev,
+ "failed to submit in-urb: %d\n", ret);
+ goto err_kill_urbs;
+ }
+ }
+
+ return 0;
+
+err_kill_urbs:
+ for (--i; i >= 0; --i) {
+ urb = cport_in->urb[i];
+ usb_kill_urb(urb);
+ }
+
+ return ret;
+}
+
+static void es2_cport_in_disable(struct es2_ap_dev *es2,
+ struct es2_cport_in *cport_in)
+{
+ struct urb *urb;
+ int i;
+
+ for (i = 0; i < NUM_CPORT_IN_URB; ++i) {
+ urb = cport_in->urb[i];
+ usb_kill_urb(urb);
+ }
+}
+
+static int es2_arpc_in_enable(struct es2_ap_dev *es2)
+{
+ struct urb *urb;
+ int ret;
+ int i;
+
+ for (i = 0; i < NUM_ARPC_IN_URB; ++i) {
+ urb = es2->arpc_urb[i];
+
+ ret = usb_submit_urb(urb, GFP_KERNEL);
+ if (ret) {
+ dev_err(&es2->usb_dev->dev,
+ "failed to submit arpc in-urb: %d\n", ret);
+ goto err_kill_urbs;
+ }
+ }
+
+ return 0;
+
+err_kill_urbs:
+ for (--i; i >= 0; --i) {
+ urb = es2->arpc_urb[i];
+ usb_kill_urb(urb);
+ }
+
+ return ret;
+}
+
+static void es2_arpc_in_disable(struct es2_ap_dev *es2)
+{
+ struct urb *urb;
+ int i;
+
+ for (i = 0; i < NUM_ARPC_IN_URB; ++i) {
+ urb = es2->arpc_urb[i];
+ usb_kill_urb(urb);
+ }
+}
+
+static struct urb *next_free_urb(struct es2_ap_dev *es2, gfp_t gfp_mask)
+{
+ struct urb *urb = NULL;
+ unsigned long flags;
+ int i;
+
+ spin_lock_irqsave(&es2->cport_out_urb_lock, flags);
+
+ /* Look in our pool of allocated urbs first, as that's the "fastest" */
+ for (i = 0; i < NUM_CPORT_OUT_URB; ++i) {
+ if (es2->cport_out_urb_busy[i] == false &&
+ es2->cport_out_urb_cancelled[i] == false) {
+ es2->cport_out_urb_busy[i] = true;
+ urb = es2->cport_out_urb[i];
+ break;
+ }
+ }
+ spin_unlock_irqrestore(&es2->cport_out_urb_lock, flags);
+ if (urb)
+ return urb;
+
+ /*
+ * Crap, pool is empty, complain to the syslog and go allocate one
+ * dynamically as we have to succeed.
+ */
+ dev_dbg(&es2->usb_dev->dev,
+ "No free CPort OUT urbs, having to dynamically allocate one!\n");
+ return usb_alloc_urb(0, gfp_mask);
+}
+
+static void free_urb(struct es2_ap_dev *es2, struct urb *urb)
+{
+ unsigned long flags;
+ int i;
+ /*
+ * See if this was an urb in our pool, if so mark it "free", otherwise
+ * we need to free it ourselves.
+ */
+ spin_lock_irqsave(&es2->cport_out_urb_lock, flags);
+ for (i = 0; i < NUM_CPORT_OUT_URB; ++i) {
+ if (urb == es2->cport_out_urb[i]) {
+ es2->cport_out_urb_busy[i] = false;
+ urb = NULL;
+ break;
+ }
+ }
+ spin_unlock_irqrestore(&es2->cport_out_urb_lock, flags);
+
+ /* If urb is not NULL, then we need to free this urb */
+ usb_free_urb(urb);
+}
+
+/*
+ * We (ab)use the operation-message header pad bytes to transfer the
+ * cport id in order to minimise overhead.
+ */
+static void
+gb_message_cport_pack(struct gb_operation_msg_hdr *header, u16 cport_id)
+{
+ header->pad[0] = cport_id;
+}
+
+/* Clear the pad bytes used for the CPort id */
+static void gb_message_cport_clear(struct gb_operation_msg_hdr *header)
+{
+ header->pad[0] = 0;
+}
+
+/* Extract the CPort id packed into the header, and clear it */
+static u16 gb_message_cport_unpack(struct gb_operation_msg_hdr *header)
+{
+ u16 cport_id = header->pad[0];
+
+ gb_message_cport_clear(header);
+
+ return cport_id;
+}
+
+/*
+ * Returns zero if the message was successfully queued, or a negative errno
+ * otherwise.
+ */
+static int message_send(struct gb_host_device *hd, u16 cport_id,
+ struct gb_message *message, gfp_t gfp_mask)
+{
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct usb_device *udev = es2->usb_dev;
+ size_t buffer_size;
+ int retval;
+ struct urb *urb;
+ unsigned long flags;
+
+ /*
+ * The data actually transferred will include an indication
+ * of where the data should be sent. Do one last check of
+ * the target CPort id before filling it in.
+ */
+ if (!cport_id_valid(hd, cport_id)) {
+ dev_err(&udev->dev, "invalid cport %u\n", cport_id);
+ return -EINVAL;
+ }
+
+ /* Find a free urb */
+ urb = next_free_urb(es2, gfp_mask);
+ if (!urb)
+ return -ENOMEM;
+
+ spin_lock_irqsave(&es2->cport_out_urb_lock, flags);
+ message->hcpriv = urb;
+ spin_unlock_irqrestore(&es2->cport_out_urb_lock, flags);
+
+ /* Pack the cport id into the message header */
+ gb_message_cport_pack(message->header, cport_id);
+
+ buffer_size = sizeof(*message->header) + message->payload_size;
+
+ usb_fill_bulk_urb(urb, udev,
+ usb_sndbulkpipe(udev,
+ es2->cport_out_endpoint),
+ message->buffer, buffer_size,
+ cport_out_callback, message);
+ urb->transfer_flags |= URB_ZERO_PACKET;
+
+ trace_gb_message_submit(message);
+
+ retval = usb_submit_urb(urb, gfp_mask);
+ if (retval) {
+ dev_err(&udev->dev, "failed to submit out-urb: %d\n", retval);
+
+ spin_lock_irqsave(&es2->cport_out_urb_lock, flags);
+ message->hcpriv = NULL;
+ spin_unlock_irqrestore(&es2->cport_out_urb_lock, flags);
+
+ free_urb(es2, urb);
+ gb_message_cport_clear(message->header);
+
+ return retval;
+ }
+
+ return 0;
+}
+
+/*
+ * Can not be called in atomic context.
+ */
+static void message_cancel(struct gb_message *message)
+{
+ struct gb_host_device *hd = message->operation->connection->hd;
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct urb *urb;
+ int i;
+
+ might_sleep();
+
+ spin_lock_irq(&es2->cport_out_urb_lock);
+ urb = message->hcpriv;
+
+ /* Prevent dynamically allocated urb from being deallocated. */
+ usb_get_urb(urb);
+
+ /* Prevent pre-allocated urb from being reused. */
+ for (i = 0; i < NUM_CPORT_OUT_URB; ++i) {
+ if (urb == es2->cport_out_urb[i]) {
+ es2->cport_out_urb_cancelled[i] = true;
+ break;
+ }
+ }
+ spin_unlock_irq(&es2->cport_out_urb_lock);
+
+ usb_kill_urb(urb);
+
+ if (i < NUM_CPORT_OUT_URB) {
+ spin_lock_irq(&es2->cport_out_urb_lock);
+ es2->cport_out_urb_cancelled[i] = false;
+ spin_unlock_irq(&es2->cport_out_urb_lock);
+ }
+
+ usb_free_urb(urb);
+}
+
+static int es2_cport_allocate(struct gb_host_device *hd, int cport_id,
+ unsigned long flags)
+{
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct ida *id_map = &hd->cport_id_map;
+ int ida_start, ida_end;
+
+ switch (cport_id) {
+ case ES2_CPORT_CDSI0:
+ case ES2_CPORT_CDSI1:
+ dev_err(&hd->dev, "cport %d not available\n", cport_id);
+ return -EBUSY;
+ }
+
+ if (flags & GB_CONNECTION_FLAG_OFFLOADED &&
+ flags & GB_CONNECTION_FLAG_CDSI1) {
+ if (es2->cdsi1_in_use) {
+ dev_err(&hd->dev, "CDSI1 already in use\n");
+ return -EBUSY;
+ }
+
+ es2->cdsi1_in_use = true;
+
+ return ES2_CPORT_CDSI1;
+ }
+
+ if (cport_id < 0) {
+ ida_start = 0;
+ ida_end = hd->num_cports;
+ } else if (cport_id < hd->num_cports) {
+ ida_start = cport_id;
+ ida_end = cport_id + 1;
+ } else {
+ dev_err(&hd->dev, "cport %d not available\n", cport_id);
+ return -EINVAL;
+ }
+
+ return ida_simple_get(id_map, ida_start, ida_end, GFP_KERNEL);
+}
+
+static void es2_cport_release(struct gb_host_device *hd, u16 cport_id)
+{
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+
+ switch (cport_id) {
+ case ES2_CPORT_CDSI1:
+ es2->cdsi1_in_use = false;
+ return;
+ }
+
+ ida_simple_remove(&hd->cport_id_map, cport_id);
+}
+
+static int cport_enable(struct gb_host_device *hd, u16 cport_id,
+ unsigned long flags)
+{
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct usb_device *udev = es2->usb_dev;
+ struct gb_apb_request_cport_flags *req;
+ u32 connection_flags;
+ int ret;
+
+ req = kzalloc(sizeof(*req), GFP_KERNEL);
+ if (!req)
+ return -ENOMEM;
+
+ connection_flags = 0;
+ if (flags & GB_CONNECTION_FLAG_CONTROL)
+ connection_flags |= GB_APB_CPORT_FLAG_CONTROL;
+ if (flags & GB_CONNECTION_FLAG_HIGH_PRIO)
+ connection_flags |= GB_APB_CPORT_FLAG_HIGH_PRIO;
+
+ req->flags = cpu_to_le32(connection_flags);
+
+ dev_dbg(&hd->dev, "%s - cport = %u, flags = %02x\n", __func__,
+ cport_id, connection_flags);
+
+ ret = usb_control_msg(udev, usb_sndctrlpipe(udev, 0),
+ GB_APB_REQUEST_CPORT_FLAGS,
+ USB_DIR_OUT | USB_TYPE_VENDOR |
+ USB_RECIP_INTERFACE, cport_id, 0,
+ req, sizeof(*req), ES2_USB_CTRL_TIMEOUT);
+ if (ret != sizeof(*req)) {
+ dev_err(&udev->dev, "failed to set cport flags for port %d\n",
+ cport_id);
+ if (ret >= 0)
+ ret = -EIO;
+
+ goto out;
+ }
+
+ ret = 0;
+out:
+ kfree(req);
+
+ return ret;
+}
+
+static int es2_cport_connected(struct gb_host_device *hd, u16 cport_id)
+{
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct device *dev = &es2->usb_dev->dev;
+ struct arpc_cport_connected_req req;
+ int ret;
+
+ req.cport_id = cpu_to_le16(cport_id);
+ ret = arpc_sync(es2, ARPC_TYPE_CPORT_CONNECTED, &req, sizeof(req),
+ NULL, ES2_ARPC_CPORT_TIMEOUT);
+ if (ret) {
+ dev_err(dev, "failed to set connected state for cport %u: %d\n",
+ cport_id, ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int es2_cport_flush(struct gb_host_device *hd, u16 cport_id)
+{
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct device *dev = &es2->usb_dev->dev;
+ struct arpc_cport_flush_req req;
+ int ret;
+
+ req.cport_id = cpu_to_le16(cport_id);
+ ret = arpc_sync(es2, ARPC_TYPE_CPORT_FLUSH, &req, sizeof(req),
+ NULL, ES2_ARPC_CPORT_TIMEOUT);
+ if (ret) {
+ dev_err(dev, "failed to flush cport %u: %d\n", cport_id, ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int es2_cport_shutdown(struct gb_host_device *hd, u16 cport_id,
+ u8 phase, unsigned int timeout)
+{
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct device *dev = &es2->usb_dev->dev;
+ struct arpc_cport_shutdown_req req;
+ int result;
+ int ret;
+
+ if (timeout > U16_MAX)
+ return -EINVAL;
+
+ req.cport_id = cpu_to_le16(cport_id);
+ req.timeout = cpu_to_le16(timeout);
+ req.phase = phase;
+ ret = arpc_sync(es2, ARPC_TYPE_CPORT_SHUTDOWN, &req, sizeof(req),
+ &result, ES2_ARPC_CPORT_TIMEOUT + timeout);
+ if (ret) {
+ dev_err(dev, "failed to send shutdown over cport %u: %d (%d)\n",
+ cport_id, ret, result);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int es2_cport_quiesce(struct gb_host_device *hd, u16 cport_id,
+ size_t peer_space, unsigned int timeout)
+{
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct device *dev = &es2->usb_dev->dev;
+ struct arpc_cport_quiesce_req req;
+ int result;
+ int ret;
+
+ if (peer_space > U16_MAX)
+ return -EINVAL;
+
+ if (timeout > U16_MAX)
+ return -EINVAL;
+
+ req.cport_id = cpu_to_le16(cport_id);
+ req.peer_space = cpu_to_le16(peer_space);
+ req.timeout = cpu_to_le16(timeout);
+ ret = arpc_sync(es2, ARPC_TYPE_CPORT_QUIESCE, &req, sizeof(req),
+ &result, ES2_ARPC_CPORT_TIMEOUT + timeout);
+ if (ret) {
+ dev_err(dev, "failed to quiesce cport %u: %d (%d)\n",
+ cport_id, ret, result);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int es2_cport_clear(struct gb_host_device *hd, u16 cport_id)
+{
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct device *dev = &es2->usb_dev->dev;
+ struct arpc_cport_clear_req req;
+ int ret;
+
+ req.cport_id = cpu_to_le16(cport_id);
+ ret = arpc_sync(es2, ARPC_TYPE_CPORT_CLEAR, &req, sizeof(req),
+ NULL, ES2_ARPC_CPORT_TIMEOUT);
+ if (ret) {
+ dev_err(dev, "failed to clear cport %u: %d\n", cport_id, ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static int latency_tag_enable(struct gb_host_device *hd, u16 cport_id)
+{
+ int retval;
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct usb_device *udev = es2->usb_dev;
+
+ retval = usb_control_msg(udev, usb_sndctrlpipe(udev, 0),
+ GB_APB_REQUEST_LATENCY_TAG_EN,
+ USB_DIR_OUT | USB_TYPE_VENDOR |
+ USB_RECIP_INTERFACE, cport_id, 0, NULL,
+ 0, ES2_USB_CTRL_TIMEOUT);
+
+ if (retval < 0)
+ dev_err(&udev->dev, "Cannot enable latency tag for cport %d\n",
+ cport_id);
+ return retval;
+}
+
+static int latency_tag_disable(struct gb_host_device *hd, u16 cport_id)
+{
+ int retval;
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct usb_device *udev = es2->usb_dev;
+
+ retval = usb_control_msg(udev, usb_sndctrlpipe(udev, 0),
+ GB_APB_REQUEST_LATENCY_TAG_DIS,
+ USB_DIR_OUT | USB_TYPE_VENDOR |
+ USB_RECIP_INTERFACE, cport_id, 0, NULL,
+ 0, ES2_USB_CTRL_TIMEOUT);
+
+ if (retval < 0)
+ dev_err(&udev->dev, "Cannot disable latency tag for cport %d\n",
+ cport_id);
+ return retval;
+}
+
+static int timesync_enable(struct gb_host_device *hd, u8 count,
+ u64 frame_time, u32 strobe_delay, u32 refclk)
+{
+ int retval;
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct usb_device *udev = es2->usb_dev;
+ struct gb_control_timesync_enable_request *request;
+
+ request = kzalloc(sizeof(*request), GFP_KERNEL);
+ if (!request)
+ return -ENOMEM;
+
+ request->count = count;
+ request->frame_time = cpu_to_le64(frame_time);
+ request->strobe_delay = cpu_to_le32(strobe_delay);
+ request->refclk = cpu_to_le32(refclk);
+ retval = usb_control_msg(udev, usb_sndctrlpipe(udev, 0),
+ GB_APB_REQUEST_TIMESYNC_ENABLE,
+ USB_DIR_OUT | USB_TYPE_VENDOR |
+ USB_RECIP_INTERFACE, 0, 0, request,
+ sizeof(*request), ES2_USB_CTRL_TIMEOUT);
+ if (retval < 0)
+ dev_err(&udev->dev, "Cannot enable timesync %d\n", retval);
+
+ kfree(request);
+ return retval;
+}
+
+static int timesync_disable(struct gb_host_device *hd)
+{
+ int retval;
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct usb_device *udev = es2->usb_dev;
+
+ retval = usb_control_msg(udev, usb_sndctrlpipe(udev, 0),
+ GB_APB_REQUEST_TIMESYNC_DISABLE,
+ USB_DIR_OUT | USB_TYPE_VENDOR |
+ USB_RECIP_INTERFACE, 0, 0, NULL,
+ 0, ES2_USB_CTRL_TIMEOUT);
+ if (retval < 0)
+ dev_err(&udev->dev, "Cannot disable timesync %d\n", retval);
+
+ return retval;
+}
+
+static int timesync_authoritative(struct gb_host_device *hd, u64 *frame_time)
+{
+ int retval, i;
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct usb_device *udev = es2->usb_dev;
+ struct timesync_authoritative_request *request;
+
+ request = kzalloc(sizeof(*request), GFP_KERNEL);
+ if (!request)
+ return -ENOMEM;
+
+ for (i = 0; i < GB_TIMESYNC_MAX_STROBES; i++)
+ request->frame_time[i] = cpu_to_le64(frame_time[i]);
+
+ retval = usb_control_msg(udev, usb_sndctrlpipe(udev, 0),
+ GB_APB_REQUEST_TIMESYNC_AUTHORITATIVE,
+ USB_DIR_OUT | USB_TYPE_VENDOR |
+ USB_RECIP_INTERFACE, 0, 0, request,
+ sizeof(*request), ES2_USB_CTRL_TIMEOUT);
+ if (retval < 0)
+ dev_err(&udev->dev, "Cannot timesync authoritative out %d\n", retval);
+
+ kfree(request);
+ return retval;
+}
+
+static int timesync_get_last_event(struct gb_host_device *hd, u64 *frame_time)
+{
+ int retval;
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ struct usb_device *udev = es2->usb_dev;
+ __le64 *response_frame_time;
+
+ response_frame_time = kzalloc(sizeof(*response_frame_time), GFP_KERNEL);
+ if (!response_frame_time)
+ return -ENOMEM;
+
+ retval = usb_control_msg(udev, usb_rcvctrlpipe(udev, 0),
+ GB_APB_REQUEST_TIMESYNC_GET_LAST_EVENT,
+ USB_DIR_IN | USB_TYPE_VENDOR |
+ USB_RECIP_INTERFACE, 0, 0, response_frame_time,
+ sizeof(*response_frame_time),
+ ES2_USB_CTRL_TIMEOUT);
+
+ if (retval != sizeof(*response_frame_time)) {
+ dev_err(&udev->dev, "Cannot get last TimeSync event: %d\n",
+ retval);
+
+ if (retval >= 0)
+ retval = -EIO;
+
+ goto out;
+ }
+ *frame_time = le64_to_cpu(*response_frame_time);
+ retval = 0;
+out:
+ kfree(response_frame_time);
+ return retval;
+}
+
+static struct gb_hd_driver es2_driver = {
+ .hd_priv_size = sizeof(struct es2_ap_dev),
+ .message_send = message_send,
+ .message_cancel = message_cancel,
+ .cport_allocate = es2_cport_allocate,
+ .cport_release = es2_cport_release,
+ .cport_enable = cport_enable,
+ .cport_connected = es2_cport_connected,
+ .cport_flush = es2_cport_flush,
+ .cport_shutdown = es2_cport_shutdown,
+ .cport_quiesce = es2_cport_quiesce,
+ .cport_clear = es2_cport_clear,
+ .latency_tag_enable = latency_tag_enable,
+ .latency_tag_disable = latency_tag_disable,
+ .output = output,
+ .timesync_enable = timesync_enable,
+ .timesync_disable = timesync_disable,
+ .timesync_authoritative = timesync_authoritative,
+ .timesync_get_last_event = timesync_get_last_event,
+};
+
+/* Common function to report consistent warnings based on URB status */
+static int check_urb_status(struct urb *urb)
+{
+ struct device *dev = &urb->dev->dev;
+ int status = urb->status;
+
+ switch (status) {
+ case 0:
+ return 0;
+
+ case -EOVERFLOW:
+ dev_err(dev, "%s: overflow actual length is %d\n",
+ __func__, urb->actual_length);
+ case -ECONNRESET:
+ case -ENOENT:
+ case -ESHUTDOWN:
+ case -EILSEQ:
+ case -EPROTO:
+ /* device is gone, stop sending */
+ return status;
+ }
+ dev_err(dev, "%s: unknown status %d\n", __func__, status);
+
+ return -EAGAIN;
+}
+
+static void es2_destroy(struct es2_ap_dev *es2)
+{
+ struct usb_device *udev;
+ struct urb *urb;
+ int i;
+
+ debugfs_remove(es2->apb_log_enable_dentry);
+ usb_log_disable(es2);
+
+ /* Tear down everything! */
+ for (i = 0; i < NUM_CPORT_OUT_URB; ++i) {
+ urb = es2->cport_out_urb[i];
+ usb_kill_urb(urb);
+ usb_free_urb(urb);
+ es2->cport_out_urb[i] = NULL;
+ es2->cport_out_urb_busy[i] = false; /* just to be anal */
+ }
+
+ for (i = 0; i < NUM_ARPC_IN_URB; ++i) {
+ usb_free_urb(es2->arpc_urb[i]);
+ kfree(es2->arpc_buffer[i]);
+ es2->arpc_buffer[i] = NULL;
+ }
+
+ for (i = 0; i < NUM_CPORT_IN_URB; ++i) {
+ usb_free_urb(es2->cport_in.urb[i]);
+ kfree(es2->cport_in.buffer[i]);
+ es2->cport_in.buffer[i] = NULL;
+ }
+
+ /* release reserved CDSI0 and CDSI1 cports */
+ gb_hd_cport_release_reserved(es2->hd, ES2_CPORT_CDSI1);
+ gb_hd_cport_release_reserved(es2->hd, ES2_CPORT_CDSI0);
+
+ udev = es2->usb_dev;
+ gb_hd_put(es2->hd);
+
+ usb_put_dev(udev);
+}
+
+static void cport_in_callback(struct urb *urb)
+{
+ struct gb_host_device *hd = urb->context;
+ struct device *dev = &urb->dev->dev;
+ struct gb_operation_msg_hdr *header;
+ int status = check_urb_status(urb);
+ int retval;
+ u16 cport_id;
+
+ if (status) {
+ if ((status == -EAGAIN) || (status == -EPROTO))
+ goto exit;
+
+ /* The urb is being unlinked */
+ if (status == -ENOENT || status == -ESHUTDOWN)
+ return;
+
+ dev_err(dev, "urb cport in error %d (dropped)\n", status);
+ return;
+ }
+
+ if (urb->actual_length < sizeof(*header)) {
+ dev_err(dev, "short message received\n");
+ goto exit;
+ }
+
+ /* Extract the CPort id, which is packed in the message header */
+ header = urb->transfer_buffer;
+ cport_id = gb_message_cport_unpack(header);
+
+ if (cport_id_valid(hd, cport_id)) {
+ greybus_data_rcvd(hd, cport_id, urb->transfer_buffer,
+ urb->actual_length);
+ } else {
+ dev_err(dev, "invalid cport id %u received\n", cport_id);
+ }
+exit:
+ /* put our urb back in the request pool */
+ retval = usb_submit_urb(urb, GFP_ATOMIC);
+ if (retval)
+ dev_err(dev, "failed to resubmit in-urb: %d\n", retval);
+}
+
+static void cport_out_callback(struct urb *urb)
+{
+ struct gb_message *message = urb->context;
+ struct gb_host_device *hd = message->operation->connection->hd;
+ struct es2_ap_dev *es2 = hd_to_es2(hd);
+ int status = check_urb_status(urb);
+ unsigned long flags;
+
+ gb_message_cport_clear(message->header);
+
+ spin_lock_irqsave(&es2->cport_out_urb_lock, flags);
+ message->hcpriv = NULL;
+ spin_unlock_irqrestore(&es2->cport_out_urb_lock, flags);
+
+ /*
+ * Tell the submitter that the message send (attempt) is
+ * complete, and report the status.
+ */
+ greybus_message_sent(hd, message, status);
+
+ free_urb(es2, urb);
+}
+
+static struct arpc *arpc_alloc(void *payload, u16 size, u8 type)
+{
+ struct arpc *rpc;
+
+ if (size + sizeof(*rpc->req) > ARPC_OUT_SIZE_MAX)
+ return NULL;
+
+ rpc = kzalloc(sizeof(*rpc), GFP_KERNEL);
+ if (!rpc)
+ return NULL;
+
+ INIT_LIST_HEAD(&rpc->list);
+ rpc->req = kzalloc(sizeof(*rpc->req) + size, GFP_KERNEL);
+ if (!rpc->req)
+ goto err_free_rpc;
+
+ rpc->resp = kzalloc(sizeof(*rpc->resp), GFP_KERNEL);
+ if (!rpc->resp)
+ goto err_free_req;
+
+ rpc->req->type = type;
+ rpc->req->size = cpu_to_le16(sizeof(rpc->req) + size);
+ memcpy(rpc->req->data, payload, size);
+
+ init_completion(&rpc->response_received);
+
+ return rpc;
+
+err_free_req:
+ kfree(rpc->req);
+err_free_rpc:
+ kfree(rpc);
+
+ return NULL;
+}
+
+static void arpc_free(struct arpc *rpc)
+{
+ kfree(rpc->req);
+ kfree(rpc->resp);
+ kfree(rpc);
+}
+
+static struct arpc *arpc_find(struct es2_ap_dev *es2, __le16 id)
+{
+ struct arpc *rpc;
+
+ list_for_each_entry(rpc, &es2->arpcs, list) {
+ if (rpc->req->id == id)
+ return rpc;
+ }
+
+ return NULL;
+}
+
+static void arpc_add(struct es2_ap_dev *es2, struct arpc *rpc)
+{
+ rpc->active = true;
+ rpc->req->id = cpu_to_le16(es2->arpc_id_cycle++);
+ list_add_tail(&rpc->list, &es2->arpcs);
+}
+
+static void arpc_del(struct es2_ap_dev *es2, struct arpc *rpc)
+{
+ if (rpc->active) {
+ rpc->active = false;
+ list_del(&rpc->list);
+ }
+}
+
+static int arpc_send(struct es2_ap_dev *es2, struct arpc *rpc, int timeout)
+{
+ struct usb_device *udev = es2->usb_dev;
+ int retval;
+
+ retval = usb_control_msg(udev, usb_sndctrlpipe(udev, 0),
+ GB_APB_REQUEST_ARPC_RUN,
+ USB_DIR_OUT | USB_TYPE_VENDOR |
+ USB_RECIP_INTERFACE,
+ 0, 0,
+ rpc->req, le16_to_cpu(rpc->req->size),
+ ES2_USB_CTRL_TIMEOUT);
+ if (retval != le16_to_cpu(rpc->req->size)) {
+ dev_err(&udev->dev,
+ "failed to send ARPC request %d: %d\n",
+ rpc->req->type, retval);
+ if (retval > 0)
+ retval = -EIO;
+ return retval;
+ }
+
+ return 0;
+}
+
+static int arpc_sync(struct es2_ap_dev *es2, u8 type, void *payload,
+ size_t size, int *result, unsigned int timeout)
+{
+ struct arpc *rpc;
+ unsigned long flags;
+ int retval;
+
+ if (result)
+ *result = 0;
+
+ rpc = arpc_alloc(payload, size, type);
+ if (!rpc)
+ return -ENOMEM;
+
+ spin_lock_irqsave(&es2->arpc_lock, flags);
+ arpc_add(es2, rpc);
+ spin_unlock_irqrestore(&es2->arpc_lock, flags);
+
+ retval = arpc_send(es2, rpc, timeout);
+ if (retval)
+ goto out_arpc_del;
+
+ retval = wait_for_completion_interruptible_timeout(
+ &rpc->response_received,
+ msecs_to_jiffies(timeout));
+ if (retval <= 0) {
+ if (!retval)
+ retval = -ETIMEDOUT;
+ goto out_arpc_del;
+ }
+
+ if (rpc->resp->result) {
+ retval = -EREMOTEIO;
+ if (result)
+ *result = rpc->resp->result;
+ } else {
+ retval = 0;
+ }
+
+out_arpc_del:
+ spin_lock_irqsave(&es2->arpc_lock, flags);
+ arpc_del(es2, rpc);
+ spin_unlock_irqrestore(&es2->arpc_lock, flags);
+ arpc_free(rpc);
+
+ if (retval < 0 && retval != -EREMOTEIO) {
+ dev_err(&es2->usb_dev->dev,
+ "failed to execute ARPC: %d\n", retval);
+ }
+
+ return retval;
+}
+
+static void arpc_in_callback(struct urb *urb)
+{
+ struct es2_ap_dev *es2 = urb->context;
+ struct device *dev = &urb->dev->dev;
+ int status = check_urb_status(urb);
+ struct arpc *rpc;
+ struct arpc_response_message *resp;
+ unsigned long flags;
+ int retval;
+
+ if (status) {
+ if ((status == -EAGAIN) || (status == -EPROTO))
+ goto exit;
+
+ /* The urb is being unlinked */
+ if (status == -ENOENT || status == -ESHUTDOWN)
+ return;
+
+ dev_err(dev, "arpc in-urb error %d (dropped)\n", status);
+ return;
+ }
+
+ if (urb->actual_length < sizeof(*resp)) {
+ dev_err(dev, "short aprc response received\n");
+ goto exit;
+ }
+
+ resp = urb->transfer_buffer;
+ spin_lock_irqsave(&es2->arpc_lock, flags);
+ rpc = arpc_find(es2, resp->id);
+ if (!rpc) {
+ dev_err(dev, "invalid arpc response id received: %u\n",
+ le16_to_cpu(resp->id));
+ spin_unlock_irqrestore(&es2->arpc_lock, flags);
+ goto exit;
+ }
+
+ arpc_del(es2, rpc);
+ memcpy(rpc->resp, resp, sizeof(*resp));
+ complete(&rpc->response_received);
+ spin_unlock_irqrestore(&es2->arpc_lock, flags);
+
+exit:
+ /* put our urb back in the request pool */
+ retval = usb_submit_urb(urb, GFP_ATOMIC);
+ if (retval)
+ dev_err(dev, "failed to resubmit arpc in-urb: %d\n", retval);
+}
+
+#define APB1_LOG_MSG_SIZE 64
+static void apb_log_get(struct es2_ap_dev *es2, char *buf)
+{
+ int retval;
+
+ do {
+ retval = usb_control_msg(es2->usb_dev,
+ usb_rcvctrlpipe(es2->usb_dev, 0),
+ GB_APB_REQUEST_LOG,
+ USB_DIR_IN | USB_TYPE_VENDOR | USB_RECIP_INTERFACE,
+ 0x00, 0x00,
+ buf,
+ APB1_LOG_MSG_SIZE,
+ ES2_USB_CTRL_TIMEOUT);
+ if (retval > 0)
+ kfifo_in(&es2->apb_log_fifo, buf, retval);
+ } while (retval > 0);
+}
+
+static int apb_log_poll(void *data)
+{
+ struct es2_ap_dev *es2 = data;
+ char *buf;
+
+ buf = kmalloc(APB1_LOG_MSG_SIZE, GFP_KERNEL);
+ if (!buf)
+ return -ENOMEM;
+
+ while (!kthread_should_stop()) {
+ msleep(1000);
+ apb_log_get(es2, buf);
+ }
+
+ kfree(buf);
+
+ return 0;
+}
+
+static ssize_t apb_log_read(struct file *f, char __user *buf,
+ size_t count, loff_t *ppos)
+{
+ struct es2_ap_dev *es2 = f->f_inode->i_private;
+ ssize_t ret;
+ size_t copied;
+ char *tmp_buf;
+
+ if (count > APB1_LOG_SIZE)
+ count = APB1_LOG_SIZE;
+
+ tmp_buf = kmalloc(count, GFP_KERNEL);
+ if (!tmp_buf)
+ return -ENOMEM;
+
+ copied = kfifo_out(&es2->apb_log_fifo, tmp_buf, count);
+ ret = simple_read_from_buffer(buf, count, ppos, tmp_buf, copied);
+
+ kfree(tmp_buf);
+
+ return ret;
+}
+
+static const struct file_operations apb_log_fops = {
+ .read = apb_log_read,
+};
+
+static void usb_log_enable(struct es2_ap_dev *es2)
+{
+ if (!IS_ERR_OR_NULL(es2->apb_log_task))
+ return;
+
+ /* get log from APB1 */
+ es2->apb_log_task = kthread_run(apb_log_poll, es2, "apb_log");
+ if (IS_ERR(es2->apb_log_task))
+ return;
+ /* XXX We will need to rename this per APB */
+ es2->apb_log_dentry = debugfs_create_file("apb_log", S_IRUGO,
+ gb_debugfs_get(), es2,
+ &apb_log_fops);
+}
+
+static void usb_log_disable(struct es2_ap_dev *es2)
+{
+ if (IS_ERR_OR_NULL(es2->apb_log_task))
+ return;
+
+ debugfs_remove(es2->apb_log_dentry);
+ es2->apb_log_dentry = NULL;
+
+ kthread_stop(es2->apb_log_task);
+ es2->apb_log_task = NULL;
+}
+
+static ssize_t apb_log_enable_read(struct file *f, char __user *buf,
+ size_t count, loff_t *ppos)
+{
+ struct es2_ap_dev *es2 = f->f_inode->i_private;
+ int enable = !IS_ERR_OR_NULL(es2->apb_log_task);
+ char tmp_buf[3];
+
+ sprintf(tmp_buf, "%d\n", enable);
+ return simple_read_from_buffer(buf, count, ppos, tmp_buf, 3);
+}
+
+static ssize_t apb_log_enable_write(struct file *f, const char __user *buf,
+ size_t count, loff_t *ppos)
+{
+ int enable;
+ ssize_t retval;
+ struct es2_ap_dev *es2 = f->f_inode->i_private;
+
+ retval = kstrtoint_from_user(buf, count, 10, &enable);
+ if (retval)
+ return retval;
+
+ if (enable)
+ usb_log_enable(es2);
+ else
+ usb_log_disable(es2);
+
+ return count;
+}
+
+static const struct file_operations apb_log_enable_fops = {
+ .read = apb_log_enable_read,
+ .write = apb_log_enable_write,
+};
+
+static int apb_get_cport_count(struct usb_device *udev)
+{
+ int retval;
+ __le16 *cport_count;
+
+ cport_count = kzalloc(sizeof(*cport_count), GFP_KERNEL);
+ if (!cport_count)
+ return -ENOMEM;
+
+ retval = usb_control_msg(udev, usb_rcvctrlpipe(udev, 0),
+ GB_APB_REQUEST_CPORT_COUNT,
+ USB_DIR_IN | USB_TYPE_VENDOR |
+ USB_RECIP_INTERFACE, 0, 0, cport_count,
+ sizeof(*cport_count), ES2_USB_CTRL_TIMEOUT);
+ if (retval != sizeof(*cport_count)) {
+ dev_err(&udev->dev, "Cannot retrieve CPort count: %d\n",
+ retval);
+
+ if (retval >= 0)
+ retval = -EIO;
+
+ goto out;
+ }
+
+ retval = le16_to_cpu(*cport_count);
+
+ /* We need to fit a CPort ID in one byte of a message header */
+ if (retval > U8_MAX) {
+ retval = U8_MAX;
+ dev_warn(&udev->dev, "Limiting number of CPorts to U8_MAX\n");
+ }
+
+out:
+ kfree(cport_count);
+ return retval;
+}
+
+/*
+ * The ES2 USB Bridge device has 15 endpoints
+ * 1 Control - usual USB stuff + AP -> APBridgeA messages
+ * 7 Bulk IN - CPort data in
+ * 7 Bulk OUT - CPort data out
+ */
+static int ap_probe(struct usb_interface *interface,
+ const struct usb_device_id *id)
+{
+ struct es2_ap_dev *es2;
+ struct gb_host_device *hd;
+ struct usb_device *udev;
+ struct usb_host_interface *iface_desc;
+ struct usb_endpoint_descriptor *endpoint;
+ __u8 ep_addr;
+ int retval;
+ int i;
+ int num_cports;
+ bool bulk_out_found = false;
+ bool bulk_in_found = false;
+ bool arpc_in_found = false;
+
+ udev = usb_get_dev(interface_to_usbdev(interface));
+
+ num_cports = apb_get_cport_count(udev);
+ if (num_cports < 0) {
+ usb_put_dev(udev);
+ dev_err(&udev->dev, "Cannot retrieve CPort count: %d\n",
+ num_cports);
+ return num_cports;
+ }
+
+ hd = gb_hd_create(&es2_driver, &udev->dev, ES2_GBUF_MSG_SIZE_MAX,
+ num_cports);
+ if (IS_ERR(hd)) {
+ usb_put_dev(udev);
+ return PTR_ERR(hd);
+ }
+
+ es2 = hd_to_es2(hd);
+ es2->hd = hd;
+ es2->usb_intf = interface;
+ es2->usb_dev = udev;
+ spin_lock_init(&es2->cport_out_urb_lock);
+ INIT_KFIFO(es2->apb_log_fifo);
+ usb_set_intfdata(interface, es2);
+
+ /*
+ * Reserve the CDSI0 and CDSI1 CPorts so they won't be allocated
+ * dynamically.
+ */
+ retval = gb_hd_cport_reserve(hd, ES2_CPORT_CDSI0);
+ if (retval)
+ goto error;
+ retval = gb_hd_cport_reserve(hd, ES2_CPORT_CDSI1);
+ if (retval)
+ goto error;
+
+ /* find all bulk endpoints */
+ iface_desc = interface->cur_altsetting;
+ for (i = 0; i < iface_desc->desc.bNumEndpoints; ++i) {
+ endpoint = &iface_desc->endpoint[i].desc;
+ ep_addr = endpoint->bEndpointAddress;
+
+ if (usb_endpoint_is_bulk_in(endpoint)) {
+ if (!bulk_in_found) {
+ es2->cport_in.endpoint = ep_addr;
+ bulk_in_found = true;
+ } else if (!arpc_in_found) {
+ es2->arpc_endpoint_in = ep_addr;
+ arpc_in_found = true;
+ } else {
+ dev_warn(&udev->dev,
+ "Unused bulk IN endpoint found: 0x%02x\n",
+ ep_addr);
+ }
+ continue;
+ }
+ if (usb_endpoint_is_bulk_out(endpoint)) {
+ if (!bulk_out_found) {
+ es2->cport_out_endpoint = ep_addr;
+ bulk_out_found = true;
+ } else {
+ dev_warn(&udev->dev,
+ "Unused bulk OUT endpoint found: 0x%02x\n",
+ ep_addr);
+ }
+ continue;
+ }
+ dev_warn(&udev->dev,
+ "Unknown endpoint type found, address 0x%02x\n",
+ ep_addr);
+ }
+ if (!bulk_in_found || !arpc_in_found || !bulk_out_found) {
+ dev_err(&udev->dev, "Not enough endpoints found in device, aborting!\n");
+ retval = -ENODEV;
+ goto error;
+ }
+
+ /* Allocate buffers for our cport in messages */
+ for (i = 0; i < NUM_CPORT_IN_URB; ++i) {
+ struct urb *urb;
+ u8 *buffer;
+
+ urb = usb_alloc_urb(0, GFP_KERNEL);
+ if (!urb) {
+ retval = -ENOMEM;
+ goto error;
+ }
+ es2->cport_in.urb[i] = urb;
+
+ buffer = kmalloc(ES2_GBUF_MSG_SIZE_MAX, GFP_KERNEL);
+ if (!buffer) {
+ retval = -ENOMEM;
+ goto error;
+ }
+
+ usb_fill_bulk_urb(urb, udev,
+ usb_rcvbulkpipe(udev, es2->cport_in.endpoint),
+ buffer, ES2_GBUF_MSG_SIZE_MAX,
+ cport_in_callback, hd);
+
+ es2->cport_in.buffer[i] = buffer;
+ }
+
+ /* Allocate buffers for ARPC in messages */
+ for (i = 0; i < NUM_ARPC_IN_URB; ++i) {
+ struct urb *urb;
+ u8 *buffer;
+
+ urb = usb_alloc_urb(0, GFP_KERNEL);
+ if (!urb) {
+ retval = -ENOMEM;
+ goto error;
+ }
+ es2->arpc_urb[i] = urb;
+
+ buffer = kmalloc(ARPC_IN_SIZE_MAX, GFP_KERNEL);
+ if (!buffer) {
+ retval = -ENOMEM;
+ goto error;
+ }
+
+ usb_fill_bulk_urb(urb, udev,
+ usb_rcvbulkpipe(udev,
+ es2->arpc_endpoint_in),
+ buffer, ARPC_IN_SIZE_MAX,
+ arpc_in_callback, es2);
+
+ es2->arpc_buffer[i] = buffer;
+ }
+
+ /* Allocate urbs for our CPort OUT messages */
+ for (i = 0; i < NUM_CPORT_OUT_URB; ++i) {
+ struct urb *urb;
+
+ urb = usb_alloc_urb(0, GFP_KERNEL);
+ if (!urb) {
+ retval = -ENOMEM;
+ goto error;
+ }
+
+ es2->cport_out_urb[i] = urb;
+ es2->cport_out_urb_busy[i] = false; /* just to be anal */
+ }
+
+ /* XXX We will need to rename this per APB */
+ es2->apb_log_enable_dentry = debugfs_create_file("apb_log_enable",
+ (S_IWUSR | S_IRUGO),
+ gb_debugfs_get(), es2,
+ &apb_log_enable_fops);
+
+ INIT_LIST_HEAD(&es2->arpcs);
+ spin_lock_init(&es2->arpc_lock);
+
+ if (es2_arpc_in_enable(es2))
+ goto error;
+
+ retval = gb_hd_add(hd);
+ if (retval)
+ goto err_disable_arpc_in;
+
+ retval = es2_cport_in_enable(es2, &es2->cport_in);
+ if (retval)
+ goto err_hd_del;
+
+ return 0;
+
+err_hd_del:
+ gb_hd_del(hd);
+err_disable_arpc_in:
+ es2_arpc_in_disable(es2);
+error:
+ es2_destroy(es2);
+
+ return retval;
+}
+
+static void ap_disconnect(struct usb_interface *interface)
+{
+ struct es2_ap_dev *es2 = usb_get_intfdata(interface);
+
+ gb_hd_del(es2->hd);
+
+ es2_cport_in_disable(es2, &es2->cport_in);
+ es2_arpc_in_disable(es2);
+
+ es2_destroy(es2);
+}
+
+static struct usb_driver es2_ap_driver = {
+ .name = "es2_ap_driver",
+ .probe = ap_probe,
+ .disconnect = ap_disconnect,
+ .id_table = id_table,
+ .soft_unbind = 1,
+};
+
+module_usb_driver(es2_ap_driver);
+
+MODULE_LICENSE("GPL v2");
+MODULE_AUTHOR("Greg Kroah-Hartman <gregkh@linuxfoundation.org>");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 13/32] greybus: HID driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (10 preceding siblings ...)
2016-09-16 13:25 ` [patch 12/32] greybus: es2 host driver Greg KH
@ 2016-09-16 14:09 ` Greg KH
2016-09-16 14:10 ` [patch 14/32] greybus: LED driver Greg KH
` (18 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:09 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus HID protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/hid.c | 536 ++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 536 insertions(+)
--- /dev/null
+++ b/drivers/greybus/hid.c
@@ -0,0 +1,536 @@
+/*
+ * HID class driver for the Greybus.
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/bitops.h>
+#include <linux/hid.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/slab.h>
+
+#include "greybus.h"
+
+/* Greybus HID device's structure */
+struct gb_hid {
+ struct gb_bundle *bundle;
+ struct gb_connection *connection;
+
+ struct hid_device *hid;
+ struct gb_hid_desc_response hdesc;
+
+ unsigned long flags;
+#define GB_HID_STARTED 0x01
+#define GB_HID_READ_PENDING 0x04
+
+ unsigned int bufsize;
+ char *inbuf;
+};
+
+static DEFINE_MUTEX(gb_hid_open_mutex);
+
+/* Routines to get controller's information over greybus */
+
+/* Operations performed on greybus */
+static int gb_hid_get_desc(struct gb_hid *ghid)
+{
+ return gb_operation_sync(ghid->connection, GB_HID_TYPE_GET_DESC, NULL,
+ 0, &ghid->hdesc, sizeof(ghid->hdesc));
+}
+
+static int gb_hid_get_report_desc(struct gb_hid *ghid, char *rdesc)
+{
+ int ret;
+
+ ret = gb_pm_runtime_get_sync(ghid->bundle);
+ if (ret)
+ return ret;
+
+ ret = gb_operation_sync(ghid->connection, GB_HID_TYPE_GET_REPORT_DESC,
+ NULL, 0, rdesc,
+ le16_to_cpu(ghid->hdesc.wReportDescLength));
+
+ gb_pm_runtime_put_autosuspend(ghid->bundle);
+
+ return ret;
+}
+
+static int gb_hid_set_power(struct gb_hid *ghid, int type)
+{
+ int ret;
+
+ ret = gb_pm_runtime_get_sync(ghid->bundle);
+ if (ret)
+ return ret;
+
+ ret = gb_operation_sync(ghid->connection, type, NULL, 0, NULL, 0);
+
+ gb_pm_runtime_put_autosuspend(ghid->bundle);
+
+ return ret;
+}
+
+static int gb_hid_get_report(struct gb_hid *ghid, u8 report_type, u8 report_id,
+ unsigned char *buf, int len)
+{
+ struct gb_hid_get_report_request request;
+ int ret;
+
+ ret = gb_pm_runtime_get_sync(ghid->bundle);
+ if (ret)
+ return ret;
+
+ request.report_type = report_type;
+ request.report_id = report_id;
+
+ ret = gb_operation_sync(ghid->connection, GB_HID_TYPE_GET_REPORT,
+ &request, sizeof(request), buf, len);
+
+ gb_pm_runtime_put_autosuspend(ghid->bundle);
+
+ return ret;
+}
+
+static int gb_hid_set_report(struct gb_hid *ghid, u8 report_type, u8 report_id,
+ unsigned char *buf, int len)
+{
+ struct gb_hid_set_report_request *request;
+ struct gb_operation *operation;
+ int ret, size = sizeof(*request) + len - 1;
+
+ ret = gb_pm_runtime_get_sync(ghid->bundle);
+ if (ret)
+ return ret;
+
+ operation = gb_operation_create(ghid->connection,
+ GB_HID_TYPE_SET_REPORT, size, 0,
+ GFP_KERNEL);
+ if (!operation) {
+ gb_pm_runtime_put_autosuspend(ghid->bundle);
+ return -ENOMEM;
+ }
+
+ request = operation->request->payload;
+ request->report_type = report_type;
+ request->report_id = report_id;
+ memcpy(request->report, buf, len);
+
+ ret = gb_operation_request_send_sync(operation);
+ if (ret) {
+ dev_err(&operation->connection->bundle->dev,
+ "failed to set report: %d\n", ret);
+ } else {
+ ret = len;
+ }
+
+ gb_operation_put(operation);
+ gb_pm_runtime_put_autosuspend(ghid->bundle);
+
+ return ret;
+}
+
+static int gb_hid_request_handler(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct gb_hid *ghid = gb_connection_get_data(connection);
+ struct gb_hid_input_report_request *request = op->request->payload;
+
+ if (op->type != GB_HID_TYPE_IRQ_EVENT) {
+ dev_err(&connection->bundle->dev,
+ "unsupported unsolicited request\n");
+ return -EINVAL;
+ }
+
+ if (test_bit(GB_HID_STARTED, &ghid->flags))
+ hid_input_report(ghid->hid, HID_INPUT_REPORT,
+ request->report, op->request->payload_size, 1);
+
+ return 0;
+}
+
+static int gb_hid_report_len(struct hid_report *report)
+{
+ return ((report->size - 1) >> 3) + 1 +
+ report->device->report_enum[report->type].numbered;
+}
+
+static void gb_hid_find_max_report(struct hid_device *hid, unsigned int type,
+ unsigned int *max)
+{
+ struct hid_report *report;
+ unsigned int size;
+
+ list_for_each_entry(report, &hid->report_enum[type].report_list, list) {
+ size = gb_hid_report_len(report);
+ if (*max < size)
+ *max = size;
+ }
+}
+
+static void gb_hid_free_buffers(struct gb_hid *ghid)
+{
+ kfree(ghid->inbuf);
+ ghid->inbuf = NULL;
+ ghid->bufsize = 0;
+}
+
+static int gb_hid_alloc_buffers(struct gb_hid *ghid, size_t bufsize)
+{
+ ghid->inbuf = kzalloc(bufsize, GFP_KERNEL);
+ if (!ghid->inbuf)
+ return -ENOMEM;
+
+ ghid->bufsize = bufsize;
+
+ return 0;
+}
+
+/* Routines dealing with reports */
+static void gb_hid_init_report(struct gb_hid *ghid, struct hid_report *report)
+{
+ unsigned int size;
+
+ size = gb_hid_report_len(report);
+ if (gb_hid_get_report(ghid, report->type, report->id, ghid->inbuf,
+ size))
+ return;
+
+ /*
+ * hid->driver_lock is held as we are in probe function,
+ * we just need to setup the input fields, so using
+ * hid_report_raw_event is safe.
+ */
+ hid_report_raw_event(ghid->hid, report->type, ghid->inbuf, size, 1);
+}
+
+static void gb_hid_init_reports(struct gb_hid *ghid)
+{
+ struct hid_device *hid = ghid->hid;
+ struct hid_report *report;
+
+ list_for_each_entry(report,
+ &hid->report_enum[HID_INPUT_REPORT].report_list, list)
+ gb_hid_init_report(ghid, report);
+
+ list_for_each_entry(report,
+ &hid->report_enum[HID_FEATURE_REPORT].report_list, list)
+ gb_hid_init_report(ghid, report);
+}
+
+static int __gb_hid_get_raw_report(struct hid_device *hid,
+ unsigned char report_number, __u8 *buf, size_t count,
+ unsigned char report_type)
+{
+ struct gb_hid *ghid = hid->driver_data;
+ int ret;
+
+ if (report_type == HID_OUTPUT_REPORT)
+ return -EINVAL;
+
+ ret = gb_hid_get_report(ghid, report_type, report_number, buf, count);
+ if (!ret)
+ ret = count;
+
+ return ret;
+}
+
+static int __gb_hid_output_raw_report(struct hid_device *hid, __u8 *buf,
+ size_t len, unsigned char report_type)
+{
+ struct gb_hid *ghid = hid->driver_data;
+ int report_id = buf[0];
+ int ret;
+
+ if (report_type == HID_INPUT_REPORT)
+ return -EINVAL;
+
+ if (report_id) {
+ buf++;
+ len--;
+ }
+
+ ret = gb_hid_set_report(ghid, report_type, report_id, buf, len);
+ if (report_id && ret >= 0)
+ ret++; /* add report_id to the number of transfered bytes */
+
+ return 0;
+}
+
+static int gb_hid_raw_request(struct hid_device *hid, unsigned char reportnum,
+ __u8 *buf, size_t len, unsigned char rtype,
+ int reqtype)
+{
+ switch (reqtype) {
+ case HID_REQ_GET_REPORT:
+ return __gb_hid_get_raw_report(hid, reportnum, buf, len, rtype);
+ case HID_REQ_SET_REPORT:
+ if (buf[0] != reportnum)
+ return -EINVAL;
+ return __gb_hid_output_raw_report(hid, buf, len, rtype);
+ default:
+ return -EIO;
+ }
+}
+
+/* HID Callbacks */
+static int gb_hid_parse(struct hid_device *hid)
+{
+ struct gb_hid *ghid = hid->driver_data;
+ unsigned int rsize;
+ char *rdesc;
+ int ret;
+
+ rsize = le16_to_cpu(ghid->hdesc.wReportDescLength);
+ if (!rsize || rsize > HID_MAX_DESCRIPTOR_SIZE) {
+ dbg_hid("weird size of report descriptor (%u)\n", rsize);
+ return -EINVAL;
+ }
+
+ rdesc = kzalloc(rsize, GFP_KERNEL);
+ if (!rdesc) {
+ dbg_hid("couldn't allocate rdesc memory\n");
+ return -ENOMEM;
+ }
+
+ ret = gb_hid_get_report_desc(ghid, rdesc);
+ if (ret) {
+ hid_err(hid, "reading report descriptor failed\n");
+ goto free_rdesc;
+ }
+
+ ret = hid_parse_report(hid, rdesc, rsize);
+ if (ret)
+ dbg_hid("parsing report descriptor failed\n");
+
+free_rdesc:
+ kfree(rdesc);
+
+ return ret;
+}
+
+static int gb_hid_start(struct hid_device *hid)
+{
+ struct gb_hid *ghid = hid->driver_data;
+ unsigned int bufsize = HID_MIN_BUFFER_SIZE;
+ int ret;
+
+ gb_hid_find_max_report(hid, HID_INPUT_REPORT, &bufsize);
+ gb_hid_find_max_report(hid, HID_OUTPUT_REPORT, &bufsize);
+ gb_hid_find_max_report(hid, HID_FEATURE_REPORT, &bufsize);
+
+ if (bufsize > HID_MAX_BUFFER_SIZE)
+ bufsize = HID_MAX_BUFFER_SIZE;
+
+ ret = gb_hid_alloc_buffers(ghid, bufsize);
+ if (ret)
+ return ret;
+
+ if (!(hid->quirks & HID_QUIRK_NO_INIT_REPORTS))
+ gb_hid_init_reports(ghid);
+
+ return 0;
+}
+
+static void gb_hid_stop(struct hid_device *hid)
+{
+ struct gb_hid *ghid = hid->driver_data;
+
+ gb_hid_free_buffers(ghid);
+}
+
+static int gb_hid_open(struct hid_device *hid)
+{
+ struct gb_hid *ghid = hid->driver_data;
+ int ret = 0;
+
+ mutex_lock(&gb_hid_open_mutex);
+ if (!hid->open++) {
+ ret = gb_hid_set_power(ghid, GB_HID_TYPE_PWR_ON);
+ if (ret < 0)
+ hid->open--;
+ else
+ set_bit(GB_HID_STARTED, &ghid->flags);
+ }
+ mutex_unlock(&gb_hid_open_mutex);
+
+ return ret;
+}
+
+static void gb_hid_close(struct hid_device *hid)
+{
+ struct gb_hid *ghid = hid->driver_data;
+ int ret;
+
+ /*
+ * Protecting hid->open to make sure we don't restart data acquistion
+ * due to a resumption we no longer care about..
+ */
+ mutex_lock(&gb_hid_open_mutex);
+ if (!--hid->open) {
+ clear_bit(GB_HID_STARTED, &ghid->flags);
+
+ /* Save some power */
+ ret = gb_hid_set_power(ghid, GB_HID_TYPE_PWR_OFF);
+ if (ret)
+ dev_err(&ghid->connection->bundle->dev,
+ "failed to power off (%d)\n", ret);
+ }
+ mutex_unlock(&gb_hid_open_mutex);
+}
+
+static int gb_hid_power(struct hid_device *hid, int lvl)
+{
+ struct gb_hid *ghid = hid->driver_data;
+
+ switch (lvl) {
+ case PM_HINT_FULLON:
+ return gb_hid_set_power(ghid, GB_HID_TYPE_PWR_ON);
+ case PM_HINT_NORMAL:
+ return gb_hid_set_power(ghid, GB_HID_TYPE_PWR_OFF);
+ }
+
+ return 0;
+}
+
+/* HID structure to pass callbacks */
+static struct hid_ll_driver gb_hid_ll_driver = {
+ .parse = gb_hid_parse,
+ .start = gb_hid_start,
+ .stop = gb_hid_stop,
+ .open = gb_hid_open,
+ .close = gb_hid_close,
+ .power = gb_hid_power,
+ .raw_request = gb_hid_raw_request,
+};
+
+static int gb_hid_init(struct gb_hid *ghid)
+{
+ struct hid_device *hid = ghid->hid;
+ int ret;
+
+ ret = gb_hid_get_desc(ghid);
+ if (ret)
+ return ret;
+
+ hid->version = le16_to_cpu(ghid->hdesc.bcdHID);
+ hid->vendor = le16_to_cpu(ghid->hdesc.wVendorID);
+ hid->product = le16_to_cpu(ghid->hdesc.wProductID);
+ hid->country = ghid->hdesc.bCountryCode;
+
+ hid->driver_data = ghid;
+ hid->ll_driver = &gb_hid_ll_driver;
+ hid->dev.parent = &ghid->connection->bundle->dev;
+// hid->bus = BUS_GREYBUS; /* Need a bustype for GREYBUS in <linux/input.h> */
+
+ /* Set HID device's name */
+ snprintf(hid->name, sizeof(hid->name), "%s %04X:%04X",
+ dev_name(&ghid->connection->bundle->dev),
+ hid->vendor, hid->product);
+
+ return 0;
+}
+
+static int gb_hid_probe(struct gb_bundle *bundle,
+ const struct greybus_bundle_id *id)
+{
+ struct greybus_descriptor_cport *cport_desc;
+ struct gb_connection *connection;
+ struct hid_device *hid;
+ struct gb_hid *ghid;
+ int ret;
+
+ if (bundle->num_cports != 1)
+ return -ENODEV;
+
+ cport_desc = &bundle->cport_desc[0];
+ if (cport_desc->protocol_id != GREYBUS_PROTOCOL_HID)
+ return -ENODEV;
+
+ ghid = kzalloc(sizeof(*ghid), GFP_KERNEL);
+ if (!ghid)
+ return -ENOMEM;
+
+ connection = gb_connection_create(bundle, le16_to_cpu(cport_desc->id),
+ gb_hid_request_handler);
+ if (IS_ERR(connection)) {
+ ret = PTR_ERR(connection);
+ goto err_free_ghid;
+ }
+
+ gb_connection_set_data(connection, ghid);
+ ghid->connection = connection;
+
+ hid = hid_allocate_device();
+ if (IS_ERR(hid)) {
+ ret = PTR_ERR(hid);
+ goto err_connection_destroy;
+ }
+
+ ghid->hid = hid;
+ ghid->bundle = bundle;
+
+ greybus_set_drvdata(bundle, ghid);
+
+ ret = gb_connection_enable(connection);
+ if (ret)
+ goto err_destroy_hid;
+
+ ret = gb_hid_init(ghid);
+ if (ret)
+ goto err_connection_disable;
+
+ ret = hid_add_device(hid);
+ if (ret) {
+ hid_err(hid, "can't add hid device: %d\n", ret);
+ goto err_connection_disable;
+ }
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return 0;
+
+err_connection_disable:
+ gb_connection_disable(connection);
+err_destroy_hid:
+ hid_destroy_device(hid);
+err_connection_destroy:
+ gb_connection_destroy(connection);
+err_free_ghid:
+ kfree(ghid);
+
+ return ret;
+}
+
+static void gb_hid_disconnect(struct gb_bundle *bundle)
+{
+ struct gb_hid *ghid = greybus_get_drvdata(bundle);
+
+ if (gb_pm_runtime_get_sync(bundle))
+ gb_pm_runtime_get_noresume(bundle);
+
+ hid_destroy_device(ghid->hid);
+ gb_connection_disable(ghid->connection);
+ gb_connection_destroy(ghid->connection);
+ kfree(ghid);
+}
+
+static const struct greybus_bundle_id gb_hid_id_table[] = {
+ { GREYBUS_DEVICE_CLASS(GREYBUS_CLASS_HID) },
+ { }
+};
+MODULE_DEVICE_TABLE(greybus, gb_hid_id_table);
+
+static struct greybus_driver gb_hid_driver = {
+ .name = "hid",
+ .probe = gb_hid_probe,
+ .disconnect = gb_hid_disconnect,
+ .id_table = gb_hid_id_table,
+};
+module_greybus_driver(gb_hid_driver);
+
+MODULE_LICENSE("GPL v2");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 14/32] greybus: LED driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (11 preceding siblings ...)
2016-09-16 14:09 ` [patch 13/32] greybus: HID driver Greg KH
@ 2016-09-16 14:10 ` Greg KH
2016-10-07 13:36 ` Pavel Machek
2016-09-16 14:10 ` [patch 15/32] greybus: logging driver Greg KH
` (17 subsequent siblings)
30 siblings, 1 reply; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:10 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus LED protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/light.c | 1359 ++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 1359 insertions(+)
--- /dev/null
+++ b/drivers/greybus/light.c
@@ -0,0 +1,1359 @@
+/*
+ * Greybus Lights protocol driver.
+ *
+ * Copyright 2015 Google Inc.
+ * Copyright 2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/kernel.h>
+#include <linux/leds.h>
+#include <linux/led-class-flash.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/version.h>
+#include <media/v4l2-flash-led-class.h>
+
+#include "greybus.h"
+#include "greybus_protocols.h"
+
+#define NAMES_MAX 32
+
+struct gb_channel {
+ u8 id;
+ u32 flags;
+ u32 color;
+ char *color_name;
+ u8 fade_in;
+ u8 fade_out;
+ u32 mode;
+ char *mode_name;
+ struct attribute **attrs;
+ struct attribute_group *attr_group;
+ const struct attribute_group **attr_groups;
+ struct led_classdev *led;
+#if IS_REACHABLE(CONFIG_LEDS_CLASS_FLASH)
+ struct led_classdev_flash fled;
+ struct led_flash_setting intensity_uA;
+ struct led_flash_setting timeout_us;
+#else
+ struct led_classdev cled;
+#endif
+ struct gb_light *light;
+ bool is_registered;
+ bool releasing;
+ bool strobe_state;
+ bool active;
+ struct mutex lock;
+};
+
+struct gb_light {
+ u8 id;
+ char *name;
+ struct gb_lights *glights;
+ u32 flags;
+ u8 channels_count;
+ struct gb_channel *channels;
+ bool has_flash;
+ bool ready;
+#if IS_REACHABLE(CONFIG_V4L2_FLASH_LED_CLASS)
+ struct v4l2_flash *v4l2_flash;
+#endif
+};
+
+struct gb_lights {
+ struct gb_connection *connection;
+ u8 lights_count;
+ struct gb_light *lights;
+ struct mutex lights_lock;
+};
+
+static void gb_lights_channel_free(struct gb_channel *channel);
+
+static struct gb_connection *get_conn_from_channel(struct gb_channel *channel)
+{
+ return channel->light->glights->connection;
+}
+
+static struct gb_connection *get_conn_from_light(struct gb_light *light)
+{
+ return light->glights->connection;
+}
+
+static bool is_channel_flash(struct gb_channel *channel)
+{
+ return !!(channel->mode & (GB_CHANNEL_MODE_FLASH | GB_CHANNEL_MODE_TORCH
+ | GB_CHANNEL_MODE_INDICATOR));
+}
+
+#if IS_REACHABLE(CONFIG_LEDS_CLASS_FLASH)
+static struct gb_channel *get_channel_from_cdev(struct led_classdev *cdev)
+{
+ struct led_classdev_flash *fled_cdev = lcdev_to_flcdev(cdev);
+
+ return container_of(fled_cdev, struct gb_channel, fled);
+}
+
+static struct led_classdev *get_channel_cdev(struct gb_channel *channel)
+{
+ return &channel->fled.led_cdev;
+}
+
+static struct gb_channel *get_channel_from_mode(struct gb_light *light,
+ u32 mode)
+{
+ struct gb_channel *channel = NULL;
+ int i;
+
+ for (i = 0; i < light->channels_count; i++) {
+ channel = &light->channels[i];
+ if (channel && channel->mode == mode)
+ break;
+ }
+ return channel;
+}
+
+static int __gb_lights_flash_intensity_set(struct gb_channel *channel,
+ u32 intensity)
+{
+ struct gb_connection *connection = get_conn_from_channel(channel);
+ struct gb_bundle *bundle = connection->bundle;
+ struct gb_lights_set_flash_intensity_request req;
+ int ret;
+
+ if (channel->releasing)
+ return -ESHUTDOWN;
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret < 0)
+ return ret;
+
+ req.light_id = channel->light->id;
+ req.channel_id = channel->id;
+ req.intensity_uA = cpu_to_le32(intensity);
+
+ ret = gb_operation_sync(connection, GB_LIGHTS_TYPE_SET_FLASH_INTENSITY,
+ &req, sizeof(req), NULL, 0);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return ret;
+}
+
+static int __gb_lights_flash_brightness_set(struct gb_channel *channel)
+{
+ u32 intensity;
+
+ /* If the channel is flash we need to get the attached torch channel */
+ if (channel->mode & GB_CHANNEL_MODE_FLASH)
+ channel = get_channel_from_mode(channel->light,
+ GB_CHANNEL_MODE_TORCH);
+
+ /* For not flash we need to convert brightness to intensity */
+ intensity = channel->intensity_uA.min +
+ (channel->intensity_uA.step * channel->led->brightness);
+
+ return __gb_lights_flash_intensity_set(channel, intensity);
+}
+#else
+static struct gb_channel *get_channel_from_cdev(struct led_classdev *cdev)
+{
+ return container_of(cdev, struct gb_channel, cled);
+}
+
+static struct led_classdev *get_channel_cdev(struct gb_channel *channel)
+{
+ return &channel->cled;
+}
+
+static int __gb_lights_flash_brightness_set(struct gb_channel *channel)
+{
+ return 0;
+}
+#endif
+
+static int gb_lights_color_set(struct gb_channel *channel, u32 color);
+static int gb_lights_fade_set(struct gb_channel *channel);
+
+static void led_lock(struct led_classdev *cdev)
+{
+ mutex_lock(&cdev->led_access);
+}
+
+static void led_unlock(struct led_classdev *cdev)
+{
+ mutex_unlock(&cdev->led_access);
+}
+
+#define gb_lights_fade_attr(__dir) \
+static ssize_t fade_##__dir##_show(struct device *dev, \
+ struct device_attribute *attr, \
+ char *buf) \
+{ \
+ struct led_classdev *cdev = dev_get_drvdata(dev); \
+ struct gb_channel *channel = get_channel_from_cdev(cdev); \
+ \
+ return sprintf(buf, "%u\n", channel->fade_##__dir); \
+} \
+ \
+static ssize_t fade_##__dir##_store(struct device *dev, \
+ struct device_attribute *attr, \
+ const char *buf, size_t size) \
+{ \
+ struct led_classdev *cdev = dev_get_drvdata(dev); \
+ struct gb_channel *channel = get_channel_from_cdev(cdev); \
+ u8 fade; \
+ int ret; \
+ \
+ led_lock(cdev); \
+ if (led_sysfs_is_disabled(cdev)) { \
+ ret = -EBUSY; \
+ goto unlock; \
+ } \
+ \
+ ret = kstrtou8(buf, 0, &fade); \
+ if (ret < 0) { \
+ dev_err(dev, "could not parse fade value %d\n", ret); \
+ goto unlock; \
+ } \
+ if (channel->fade_##__dir == fade) \
+ goto unlock; \
+ channel->fade_##__dir = fade; \
+ \
+ ret = gb_lights_fade_set(channel); \
+ if (ret < 0) \
+ goto unlock; \
+ \
+ ret = size; \
+unlock: \
+ led_unlock(cdev); \
+ return ret; \
+} \
+static DEVICE_ATTR_RW(fade_##__dir)
+
+gb_lights_fade_attr(in);
+gb_lights_fade_attr(out);
+
+static ssize_t color_show(struct device *dev, struct device_attribute *attr,
+ char *buf)
+{
+ struct led_classdev *cdev = dev_get_drvdata(dev);
+ struct gb_channel *channel = get_channel_from_cdev(cdev);
+
+ return sprintf(buf, "0x%08x\n", channel->color);
+}
+
+static ssize_t color_store(struct device *dev, struct device_attribute *attr,
+ const char *buf, size_t size)
+{
+ struct led_classdev *cdev = dev_get_drvdata(dev);
+ struct gb_channel *channel = get_channel_from_cdev(cdev);
+ u32 color;
+ int ret;
+
+ led_lock(cdev);
+ if (led_sysfs_is_disabled(cdev)) {
+ ret = -EBUSY;
+ goto unlock;
+ }
+ ret = kstrtou32(buf, 0, &color);
+ if (ret < 0) {
+ dev_err(dev, "could not parse color value %d\n", ret);
+ goto unlock;
+ }
+
+ ret = gb_lights_color_set(channel, color);
+ if (ret < 0)
+ goto unlock;
+
+ channel->color = color;
+ ret = size;
+unlock:
+ led_unlock(cdev);
+ return ret;
+}
+static DEVICE_ATTR_RW(color);
+
+static int channel_attr_groups_set(struct gb_channel *channel,
+ struct led_classdev *cdev)
+{
+ int attr = 0;
+ int size = 0;
+
+ if (channel->flags & GB_LIGHT_CHANNEL_MULTICOLOR)
+ size++;
+ if (channel->flags & GB_LIGHT_CHANNEL_FADER)
+ size += 2;
+
+ if (!size)
+ return 0;
+
+ /* Set attributes based in the channel flags */
+ channel->attrs = kcalloc(size + 1, sizeof(**channel->attrs),
+ GFP_KERNEL);
+ if (!channel->attrs)
+ return -ENOMEM;
+ channel->attr_group = kcalloc(1, sizeof(*channel->attr_group),
+ GFP_KERNEL);
+ if (!channel->attr_group)
+ return -ENOMEM;
+ channel->attr_groups = kcalloc(2, sizeof(*channel->attr_groups),
+ GFP_KERNEL);
+ if (!channel->attr_groups)
+ return -ENOMEM;
+
+ if (channel->flags & GB_LIGHT_CHANNEL_MULTICOLOR)
+ channel->attrs[attr++] = &dev_attr_color.attr;
+ if (channel->flags & GB_LIGHT_CHANNEL_FADER) {
+ channel->attrs[attr++] = &dev_attr_fade_in.attr;
+ channel->attrs[attr++] = &dev_attr_fade_out.attr;
+ }
+
+ channel->attr_group->attrs = channel->attrs;
+
+ channel->attr_groups[0] = channel->attr_group;
+
+ cdev->groups = channel->attr_groups;
+
+ return 0;
+}
+
+static int gb_lights_fade_set(struct gb_channel *channel)
+{
+ struct gb_connection *connection = get_conn_from_channel(channel);
+ struct gb_bundle *bundle = connection->bundle;
+ struct gb_lights_set_fade_request req;
+ int ret;
+
+ if (channel->releasing)
+ return -ESHUTDOWN;
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret < 0)
+ return ret;
+
+ req.light_id = channel->light->id;
+ req.channel_id = channel->id;
+ req.fade_in = channel->fade_in;
+ req.fade_out = channel->fade_out;
+ ret = gb_operation_sync(connection, GB_LIGHTS_TYPE_SET_FADE,
+ &req, sizeof(req), NULL, 0);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return ret;
+}
+
+static int gb_lights_color_set(struct gb_channel *channel, u32 color)
+{
+ struct gb_connection *connection = get_conn_from_channel(channel);
+ struct gb_bundle *bundle = connection->bundle;
+ struct gb_lights_set_color_request req;
+ int ret;
+
+ if (channel->releasing)
+ return -ESHUTDOWN;
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret < 0)
+ return ret;
+
+ req.light_id = channel->light->id;
+ req.channel_id = channel->id;
+ req.color = cpu_to_le32(color);
+ ret = gb_operation_sync(connection, GB_LIGHTS_TYPE_SET_COLOR,
+ &req, sizeof(req), NULL, 0);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return ret;
+}
+
+static int __gb_lights_led_brightness_set(struct gb_channel *channel)
+{
+ struct gb_lights_set_brightness_request req;
+ struct gb_connection *connection = get_conn_from_channel(channel);
+ struct gb_bundle *bundle = connection->bundle;
+ bool old_active;
+ int ret;
+
+ mutex_lock(&channel->lock);
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret < 0)
+ goto out_unlock;
+
+ old_active = channel->active;
+
+ req.light_id = channel->light->id;
+ req.channel_id = channel->id;
+ req.brightness = (u8)channel->led->brightness;
+
+ ret = gb_operation_sync(connection, GB_LIGHTS_TYPE_SET_BRIGHTNESS,
+ &req, sizeof(req), NULL, 0);
+ if (ret < 0)
+ goto out_pm_put;
+
+ if (channel->led->brightness)
+ channel->active = true;
+ else
+ channel->active = false;
+
+ /* we need to keep module alive when turning to active state */
+ if (!old_active && channel->active)
+ goto out_unlock;
+
+ /*
+ * on the other hand if going to inactive we still hold a reference and
+ * need to put it, so we could go to suspend.
+ */
+ if (old_active && !channel->active)
+ gb_pm_runtime_put_autosuspend(bundle);
+
+out_pm_put:
+ gb_pm_runtime_put_autosuspend(bundle);
+out_unlock:
+ mutex_unlock(&channel->lock);
+
+ return ret;
+}
+
+static int __gb_lights_brightness_set(struct gb_channel *channel)
+{
+ int ret;
+
+ if (channel->releasing)
+ return 0;
+
+ if (is_channel_flash(channel))
+ ret = __gb_lights_flash_brightness_set(channel);
+ else
+ ret = __gb_lights_led_brightness_set(channel);
+
+ return ret;
+}
+
+static int gb_brightness_set(struct led_classdev *cdev,
+ enum led_brightness value)
+{
+ struct gb_channel *channel = get_channel_from_cdev(cdev);
+
+ channel->led->brightness = value;
+
+ return __gb_lights_brightness_set(channel);
+}
+
+static enum led_brightness gb_brightness_get(struct led_classdev *cdev)
+
+{
+ struct gb_channel *channel = get_channel_from_cdev(cdev);
+
+ return channel->led->brightness;
+}
+
+static int gb_blink_set(struct led_classdev *cdev, unsigned long *delay_on,
+ unsigned long *delay_off)
+{
+ struct gb_channel *channel = get_channel_from_cdev(cdev);
+ struct gb_connection *connection = get_conn_from_channel(channel);
+ struct gb_bundle *bundle = connection->bundle;
+ struct gb_lights_blink_request req;
+ bool old_active;
+ int ret;
+
+ if (channel->releasing)
+ return -ESHUTDOWN;
+
+ mutex_lock(&channel->lock);
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret < 0)
+ goto out_unlock;
+
+ old_active = channel->active;
+
+ req.light_id = channel->light->id;
+ req.channel_id = channel->id;
+ req.time_on_ms = cpu_to_le16(*delay_on);
+ req.time_off_ms = cpu_to_le16(*delay_off);
+
+ ret = gb_operation_sync(connection, GB_LIGHTS_TYPE_SET_BLINK, &req,
+ sizeof(req), NULL, 0);
+ if (ret < 0)
+ goto out_pm_put;
+
+ if (delay_on)
+ channel->active = true;
+ else
+ channel->active = false;
+
+ /* we need to keep module alive when turning to active state */
+ if (!old_active && channel->active)
+ goto out_unlock;
+
+ /*
+ * on the other hand if going to inactive we still hold a reference and
+ * need to put it, so we could go to suspend.
+ */
+ if (old_active && !channel->active)
+ gb_pm_runtime_put_autosuspend(bundle);
+
+out_pm_put:
+ gb_pm_runtime_put_autosuspend(bundle);
+out_unlock:
+ mutex_unlock(&channel->lock);
+
+ return ret;
+}
+
+static void gb_lights_led_operations_set(struct gb_channel *channel,
+ struct led_classdev *cdev)
+{
+ cdev->brightness_get = gb_brightness_get;
+ cdev->brightness_set_blocking = gb_brightness_set;
+
+ if (channel->flags & GB_LIGHT_CHANNEL_BLINK)
+ cdev->blink_set = gb_blink_set;
+}
+
+#if IS_REACHABLE(CONFIG_V4L2_FLASH_LED_CLASS)
+/* V4L2 specific helpers */
+static const struct v4l2_flash_ops v4l2_flash_ops;
+
+static void __gb_lights_channel_v4l2_config(struct led_flash_setting *channel_s,
+ struct led_flash_setting *v4l2_s)
+{
+ v4l2_s->min = channel_s->min;
+ v4l2_s->max = channel_s->max;
+ v4l2_s->step = channel_s->step;
+ /* For v4l2 val is the default value */
+ v4l2_s->val = channel_s->max;
+}
+
+static int gb_lights_light_v4l2_register(struct gb_light *light)
+{
+ struct gb_connection *connection = get_conn_from_light(light);
+ struct device *dev = &connection->bundle->dev;
+ struct v4l2_flash_config *sd_cfg;
+ struct led_classdev_flash *fled;
+ struct led_classdev_flash *iled = NULL;
+ struct gb_channel *channel_torch, *channel_ind, *channel_flash;
+ int ret = 0;
+
+ sd_cfg = kcalloc(1, sizeof(*sd_cfg), GFP_KERNEL);
+ if (!sd_cfg)
+ return -ENOMEM;
+
+ channel_torch = get_channel_from_mode(light, GB_CHANNEL_MODE_TORCH);
+ if (channel_torch)
+ __gb_lights_channel_v4l2_config(&channel_torch->intensity_uA,
+ &sd_cfg->torch_intensity);
+
+ channel_ind = get_channel_from_mode(light, GB_CHANNEL_MODE_INDICATOR);
+ if (channel_ind) {
+ __gb_lights_channel_v4l2_config(&channel_ind->intensity_uA,
+ &sd_cfg->indicator_intensity);
+ iled = &channel_ind->fled;
+ }
+
+ channel_flash = get_channel_from_mode(light, GB_CHANNEL_MODE_FLASH);
+ WARN_ON(!channel_flash);
+
+ fled = &channel_flash->fled;
+
+ snprintf(sd_cfg->dev_name, sizeof(sd_cfg->dev_name), "%s", light->name);
+
+ /* Set the possible values to faults, in our case all faults */
+ sd_cfg->flash_faults = LED_FAULT_OVER_VOLTAGE | LED_FAULT_TIMEOUT |
+ LED_FAULT_OVER_TEMPERATURE | LED_FAULT_SHORT_CIRCUIT |
+ LED_FAULT_OVER_CURRENT | LED_FAULT_INDICATOR |
+ LED_FAULT_UNDER_VOLTAGE | LED_FAULT_INPUT_VOLTAGE |
+ LED_FAULT_LED_OVER_TEMPERATURE;
+
+ light->v4l2_flash = v4l2_flash_init(dev, NULL, fled, iled,
+ &v4l2_flash_ops, sd_cfg);
+ if (IS_ERR_OR_NULL(light->v4l2_flash)) {
+ ret = PTR_ERR(light->v4l2_flash);
+ goto out_free;
+ }
+
+ return ret;
+
+out_free:
+ kfree(sd_cfg);
+ return ret;
+}
+
+static void gb_lights_light_v4l2_unregister(struct gb_light *light)
+{
+ v4l2_flash_release(light->v4l2_flash);
+}
+#else
+static int gb_lights_light_v4l2_register(struct gb_light *light)
+{
+ struct gb_connection *connection = get_conn_from_light(light);
+
+ dev_err(&connection->bundle->dev, "no support for v4l2 subdevices\n");
+ return 0;
+}
+
+static void gb_lights_light_v4l2_unregister(struct gb_light *light)
+{
+}
+#endif
+
+#if IS_REACHABLE(CONFIG_LEDS_CLASS_FLASH)
+/* Flash specific operations */
+static int gb_lights_flash_intensity_set(struct led_classdev_flash *fcdev,
+ u32 brightness)
+{
+ struct gb_channel *channel = container_of(fcdev, struct gb_channel,
+ fled);
+ int ret;
+
+ ret = __gb_lights_flash_intensity_set(channel, brightness);
+ if (ret < 0)
+ return ret;
+
+ fcdev->brightness.val = brightness;
+
+ return 0;
+}
+
+static int gb_lights_flash_intensity_get(struct led_classdev_flash *fcdev,
+ u32 *brightness)
+{
+ *brightness = fcdev->brightness.val;
+
+ return 0;
+}
+
+static int gb_lights_flash_strobe_set(struct led_classdev_flash *fcdev,
+ bool state)
+{
+ struct gb_channel *channel = container_of(fcdev, struct gb_channel,
+ fled);
+ struct gb_connection *connection = get_conn_from_channel(channel);
+ struct gb_bundle *bundle = connection->bundle;
+ struct gb_lights_set_flash_strobe_request req;
+ int ret;
+
+ if (channel->releasing)
+ return -ESHUTDOWN;
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret < 0)
+ return ret;
+
+ req.light_id = channel->light->id;
+ req.channel_id = channel->id;
+ req.state = state ? 1 : 0;
+
+ ret = gb_operation_sync(connection, GB_LIGHTS_TYPE_SET_FLASH_STROBE,
+ &req, sizeof(req), NULL, 0);
+ if (!ret)
+ channel->strobe_state = state;
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return ret;
+}
+
+static int gb_lights_flash_strobe_get(struct led_classdev_flash *fcdev,
+ bool *state)
+{
+ struct gb_channel *channel = container_of(fcdev, struct gb_channel,
+ fled);
+
+ *state = channel->strobe_state;
+ return 0;
+}
+
+static int gb_lights_flash_timeout_set(struct led_classdev_flash *fcdev,
+ u32 timeout)
+{
+ struct gb_channel *channel = container_of(fcdev, struct gb_channel,
+ fled);
+ struct gb_connection *connection = get_conn_from_channel(channel);
+ struct gb_bundle *bundle = connection->bundle;
+ struct gb_lights_set_flash_timeout_request req;
+ int ret;
+
+ if (channel->releasing)
+ return -ESHUTDOWN;
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret < 0)
+ return ret;
+
+ req.light_id = channel->light->id;
+ req.channel_id = channel->id;
+ req.timeout_us = cpu_to_le32(timeout);
+
+ ret = gb_operation_sync(connection, GB_LIGHTS_TYPE_SET_FLASH_TIMEOUT,
+ &req, sizeof(req), NULL, 0);
+ if (!ret)
+ fcdev->timeout.val = timeout;
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return ret;
+}
+
+static int gb_lights_flash_fault_get(struct led_classdev_flash *fcdev,
+ u32 *fault)
+{
+ struct gb_channel *channel = container_of(fcdev, struct gb_channel,
+ fled);
+ struct gb_connection *connection = get_conn_from_channel(channel);
+ struct gb_bundle *bundle = connection->bundle;
+ struct gb_lights_get_flash_fault_request req;
+ struct gb_lights_get_flash_fault_response resp;
+ int ret;
+
+ if (channel->releasing)
+ return -ESHUTDOWN;
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret < 0)
+ return ret;
+
+ req.light_id = channel->light->id;
+ req.channel_id = channel->id;
+
+ ret = gb_operation_sync(connection, GB_LIGHTS_TYPE_GET_FLASH_FAULT,
+ &req, sizeof(req), &resp, sizeof(resp));
+ if (!ret)
+ *fault = le32_to_cpu(resp.fault);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return ret;
+}
+
+static const struct led_flash_ops gb_lights_flash_ops = {
+ .flash_brightness_set = gb_lights_flash_intensity_set,
+ .flash_brightness_get = gb_lights_flash_intensity_get,
+ .strobe_set = gb_lights_flash_strobe_set,
+ .strobe_get = gb_lights_flash_strobe_get,
+ .timeout_set = gb_lights_flash_timeout_set,
+ .fault_get = gb_lights_flash_fault_get,
+};
+
+static int __gb_lights_channel_torch_attach(struct gb_channel *channel,
+ struct gb_channel *channel_torch)
+{
+ char *name;
+
+ /* we can only attach torch to a flash channel */
+ if (!(channel->mode & GB_CHANNEL_MODE_FLASH))
+ return 0;
+
+ /* Move torch brightness to the destination */
+ channel->led->max_brightness = channel_torch->led->max_brightness;
+
+ /* append mode name to flash name */
+ name = kasprintf(GFP_KERNEL, "%s_%s", channel->led->name,
+ channel_torch->mode_name);
+ if (!name)
+ return -ENOMEM;
+ kfree(channel->led->name);
+ channel->led->name = name;
+
+ channel_torch->led = channel->led;
+
+ return 0;
+}
+
+static int __gb_lights_flash_led_register(struct gb_channel *channel)
+{
+ struct gb_connection *connection = get_conn_from_channel(channel);
+ struct led_classdev_flash *fled = &channel->fled;
+ struct led_flash_setting *fset;
+ struct gb_channel *channel_torch;
+ int ret;
+
+ fled->ops = &gb_lights_flash_ops;
+
+ fled->led_cdev.flags |= LED_DEV_CAP_FLASH;
+
+ fset = &fled->brightness;
+ fset->min = channel->intensity_uA.min;
+ fset->max = channel->intensity_uA.max;
+ fset->step = channel->intensity_uA.step;
+ fset->val = channel->intensity_uA.max;
+
+ /* Only the flash mode have the timeout constraints settings */
+ if (channel->mode & GB_CHANNEL_MODE_FLASH) {
+ fset = &fled->timeout;
+ fset->min = channel->timeout_us.min;
+ fset->max = channel->timeout_us.max;
+ fset->step = channel->timeout_us.step;
+ fset->val = channel->timeout_us.max;
+ }
+
+ /*
+ * If light have torch mode channel, this channel will be the led
+ * classdev of the registered above flash classdev
+ */
+ channel_torch = get_channel_from_mode(channel->light,
+ GB_CHANNEL_MODE_TORCH);
+ if (channel_torch) {
+ ret = __gb_lights_channel_torch_attach(channel, channel_torch);
+ if (ret < 0)
+ goto fail;
+ }
+
+ ret = led_classdev_flash_register(&connection->bundle->dev, fled);
+ if (ret < 0)
+ goto fail;
+
+ channel->is_registered = true;
+ return 0;
+fail:
+ channel->led = NULL;
+ return ret;
+}
+
+static void __gb_lights_flash_led_unregister(struct gb_channel *channel)
+{
+ if (!channel->is_registered)
+ return;
+
+ led_classdev_flash_unregister(&channel->fled);
+}
+
+static int gb_lights_channel_flash_config(struct gb_channel *channel)
+{
+ struct gb_connection *connection = get_conn_from_channel(channel);
+ struct gb_lights_get_channel_flash_config_request req;
+ struct gb_lights_get_channel_flash_config_response conf;
+ struct led_flash_setting *fset;
+ int ret;
+
+ req.light_id = channel->light->id;
+ req.channel_id = channel->id;
+
+ ret = gb_operation_sync(connection,
+ GB_LIGHTS_TYPE_GET_CHANNEL_FLASH_CONFIG,
+ &req, sizeof(req), &conf, sizeof(conf));
+ if (ret < 0)
+ return ret;
+
+ /*
+ * Intensity constraints for flash related modes: flash, torch,
+ * indicator. They will be needed for v4l2 registration.
+ */
+ fset = &channel->intensity_uA;
+ fset->min = le32_to_cpu(conf.intensity_min_uA);
+ fset->max = le32_to_cpu(conf.intensity_max_uA);
+ fset->step = le32_to_cpu(conf.intensity_step_uA);
+
+ /*
+ * On flash type, max brightness is set as the number of intensity steps
+ * available.
+ */
+ channel->led->max_brightness = (fset->max - fset->min) / fset->step;
+
+ /* Only the flash mode have the timeout constraints settings */
+ if (channel->mode & GB_CHANNEL_MODE_FLASH) {
+ fset = &channel->timeout_us;
+ fset->min = le32_to_cpu(conf.timeout_min_us);
+ fset->max = le32_to_cpu(conf.timeout_max_us);
+ fset->step = le32_to_cpu(conf.timeout_step_us);
+ }
+
+ return 0;
+}
+#else
+static int gb_lights_channel_flash_config(struct gb_channel *channel)
+{
+ struct gb_connection *connection = get_conn_from_channel(channel);
+
+ dev_err(&connection->bundle->dev, "no support for flash devices\n");
+ return 0;
+}
+
+static int __gb_lights_flash_led_register(struct gb_channel *channel)
+{
+ return 0;
+}
+
+static void __gb_lights_flash_led_unregister(struct gb_channel *channel)
+{
+}
+
+#endif
+
+static int __gb_lights_led_register(struct gb_channel *channel)
+{
+ struct gb_connection *connection = get_conn_from_channel(channel);
+ struct led_classdev *cdev = get_channel_cdev(channel);
+ int ret;
+
+ ret = led_classdev_register(&connection->bundle->dev, cdev);
+ if (ret < 0)
+ channel->led = NULL;
+ else
+ channel->is_registered = true;
+ return ret;
+}
+
+static int gb_lights_channel_register(struct gb_channel *channel)
+{
+ /* Normal LED channel, just register in led classdev and we are done */
+ if (!is_channel_flash(channel))
+ return __gb_lights_led_register(channel);
+
+ /*
+ * Flash Type need more work, register flash classdev, indicator as
+ * flash classdev, torch will be led classdev of the flash classdev.
+ */
+ if (!(channel->mode & GB_CHANNEL_MODE_TORCH))
+ return __gb_lights_flash_led_register(channel);
+
+ return 0;
+}
+
+static void __gb_lights_led_unregister(struct gb_channel *channel)
+{
+ struct led_classdev *cdev = get_channel_cdev(channel);
+
+ if (!channel->is_registered)
+ return;
+
+ led_classdev_unregister(cdev);
+ channel->led = NULL;
+}
+
+static void gb_lights_channel_unregister(struct gb_channel *channel)
+{
+ /* The same as register, handle channels differently */
+ if (!is_channel_flash(channel)) {
+ __gb_lights_led_unregister(channel);
+ return;
+ }
+
+ if (channel->mode & GB_CHANNEL_MODE_TORCH)
+ __gb_lights_led_unregister(channel);
+ else
+ __gb_lights_flash_led_unregister(channel);
+}
+
+static int gb_lights_channel_config(struct gb_light *light,
+ struct gb_channel *channel)
+{
+ struct gb_lights_get_channel_config_response conf;
+ struct gb_lights_get_channel_config_request req;
+ struct gb_connection *connection = get_conn_from_light(light);
+ struct led_classdev *cdev = get_channel_cdev(channel);
+ char *name;
+ int ret;
+
+ req.light_id = light->id;
+ req.channel_id = channel->id;
+
+ ret = gb_operation_sync(connection, GB_LIGHTS_TYPE_GET_CHANNEL_CONFIG,
+ &req, sizeof(req), &conf, sizeof(conf));
+ if (ret < 0)
+ return ret;
+
+ channel->light = light;
+ channel->mode = le32_to_cpu(conf.mode);
+ channel->flags = le32_to_cpu(conf.flags);
+ channel->color = le32_to_cpu(conf.color);
+ channel->color_name = kstrndup(conf.color_name, NAMES_MAX, GFP_KERNEL);
+ if (!channel->color_name)
+ return -ENOMEM;
+ channel->mode_name = kstrndup(conf.mode_name, NAMES_MAX, GFP_KERNEL);
+ if (!channel->mode_name)
+ return -ENOMEM;
+
+ channel->led = cdev;
+
+ name = kasprintf(GFP_KERNEL, "%s:%s:%s", light->name,
+ channel->color_name, channel->mode_name);
+ if (!name)
+ return -ENOMEM;
+
+ cdev->name = name;
+
+ cdev->max_brightness = conf.max_brightness;
+
+ ret = channel_attr_groups_set(channel, cdev);
+ if (ret < 0)
+ return ret;
+
+ gb_lights_led_operations_set(channel, cdev);
+
+ /*
+ * If it is not a flash related channel (flash, torch or indicator) we
+ * are done here. If not, continue and fetch flash related
+ * configurations.
+ */
+ if (!is_channel_flash(channel))
+ return ret;
+
+ light->has_flash = true;
+
+ ret = gb_lights_channel_flash_config(channel);
+ if (ret < 0)
+ return ret;
+
+ return ret;
+}
+
+static int gb_lights_light_config(struct gb_lights *glights, u8 id)
+{
+ struct gb_light *light = &glights->lights[id];
+ struct gb_lights_get_light_config_request req;
+ struct gb_lights_get_light_config_response conf;
+ int ret;
+ int i;
+
+ light->glights = glights;
+ light->id = id;
+
+ req.id = id;
+
+ ret = gb_operation_sync(glights->connection,
+ GB_LIGHTS_TYPE_GET_LIGHT_CONFIG,
+ &req, sizeof(req), &conf, sizeof(conf));
+ if (ret < 0)
+ return ret;
+
+ if (!conf.channel_count)
+ return -EINVAL;
+ if (!strlen(conf.name))
+ return -EINVAL;
+
+ light->channels_count = conf.channel_count;
+ light->name = kstrndup(conf.name, NAMES_MAX, GFP_KERNEL);
+
+ light->channels = kzalloc(light->channels_count *
+ sizeof(struct gb_channel), GFP_KERNEL);
+ if (!light->channels)
+ return -ENOMEM;
+
+ /* First we collect all the configurations for all channels */
+ for (i = 0; i < light->channels_count; i++) {
+ light->channels[i].id = i;
+ ret = gb_lights_channel_config(light, &light->channels[i]);
+ if (ret < 0)
+ return ret;
+ }
+
+ return 0;
+}
+
+static int gb_lights_light_register(struct gb_light *light)
+{
+ int ret;
+ int i;
+
+ /*
+ * Then, if everything went ok in getting configurations, we register
+ * the classdev, flash classdev and v4l2 subsystem, if a flash device is
+ * found.
+ */
+ for (i = 0; i < light->channels_count; i++) {
+ ret = gb_lights_channel_register(&light->channels[i]);
+ if (ret < 0)
+ return ret;
+
+ mutex_init(&light->channels[i].lock);
+ }
+
+ light->ready = true;
+
+ if (light->has_flash) {
+ ret = gb_lights_light_v4l2_register(light);
+ if (ret < 0) {
+ light->has_flash = false;
+ return ret;
+ }
+ }
+
+ return 0;
+}
+
+static void gb_lights_channel_free(struct gb_channel *channel)
+{
+ kfree(channel->attrs);
+ kfree(channel->attr_group);
+ kfree(channel->attr_groups);
+ kfree(channel->color_name);
+ kfree(channel->mode_name);
+ mutex_destroy(&channel->lock);
+}
+
+static void gb_lights_channel_release(struct gb_channel *channel)
+{
+ channel->releasing = true;
+
+ gb_lights_channel_unregister(channel);
+
+ gb_lights_channel_free(channel);
+}
+
+static void gb_lights_light_release(struct gb_light *light)
+{
+ int i;
+ int count;
+
+ light->ready = false;
+
+ count = light->channels_count;
+
+ if (light->has_flash)
+ gb_lights_light_v4l2_unregister(light);
+
+ for (i = 0; i < count; i++) {
+ gb_lights_channel_release(&light->channels[i]);
+ light->channels_count--;
+ }
+ kfree(light->channels);
+ kfree(light->name);
+}
+
+static void gb_lights_release(struct gb_lights *glights)
+{
+ int i;
+
+ if (!glights)
+ return;
+
+ mutex_lock(&glights->lights_lock);
+ if (!glights->lights)
+ goto free_glights;
+
+ for (i = 0; i < glights->lights_count; i++)
+ gb_lights_light_release(&glights->lights[i]);
+
+ kfree(glights->lights);
+
+free_glights:
+ mutex_unlock(&glights->lights_lock);
+ mutex_destroy(&glights->lights_lock);
+ kfree(glights);
+}
+
+static int gb_lights_get_count(struct gb_lights *glights)
+{
+ struct gb_lights_get_lights_response resp;
+ int ret;
+
+ ret = gb_operation_sync(glights->connection, GB_LIGHTS_TYPE_GET_LIGHTS,
+ NULL, 0, &resp, sizeof(resp));
+ if (ret < 0)
+ return ret;
+
+ if (!resp.lights_count)
+ return -EINVAL;
+
+ glights->lights_count = resp.lights_count;
+
+ return 0;
+}
+
+static int gb_lights_create_all(struct gb_lights *glights)
+{
+ struct gb_connection *connection = glights->connection;
+ int ret;
+ int i;
+
+ mutex_lock(&glights->lights_lock);
+ ret = gb_lights_get_count(glights);
+ if (ret < 0)
+ goto out;
+
+ glights->lights = kzalloc(glights->lights_count *
+ sizeof(struct gb_light), GFP_KERNEL);
+ if (!glights->lights) {
+ ret = -ENOMEM;
+ goto out;
+ }
+
+ for (i = 0; i < glights->lights_count; i++) {
+ ret = gb_lights_light_config(glights, i);
+ if (ret < 0) {
+ dev_err(&connection->bundle->dev,
+ "Fail to configure lights device\n");
+ goto out;
+ }
+ }
+
+out:
+ mutex_unlock(&glights->lights_lock);
+ return ret;
+}
+
+static int gb_lights_register_all(struct gb_lights *glights)
+{
+ struct gb_connection *connection = glights->connection;
+ int ret = 0;
+ int i;
+
+ mutex_lock(&glights->lights_lock);
+ for (i = 0; i < glights->lights_count; i++) {
+ ret = gb_lights_light_register(&glights->lights[i]);
+ if (ret < 0) {
+ dev_err(&connection->bundle->dev,
+ "Fail to enable lights device\n");
+ break;
+ }
+ }
+
+ mutex_unlock(&glights->lights_lock);
+ return ret;
+}
+
+static int gb_lights_request_handler(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct device *dev = &connection->bundle->dev;
+ struct gb_lights *glights = gb_connection_get_data(connection);
+ struct gb_light *light;
+ struct gb_message *request;
+ struct gb_lights_event_request *payload;
+ int ret = 0;
+ u8 light_id;
+ u8 event;
+
+ if (op->type != GB_LIGHTS_TYPE_EVENT) {
+ dev_err(dev, "Unsupported unsolicited event: %u\n", op->type);
+ return -EINVAL;
+ }
+
+ request = op->request;
+
+ if (request->payload_size < sizeof(*payload)) {
+ dev_err(dev, "Wrong event size received (%zu < %zu)\n",
+ request->payload_size, sizeof(*payload));
+ return -EINVAL;
+ }
+
+ payload = request->payload;
+ light_id = payload->light_id;
+
+ if (light_id >= glights->lights_count ||
+ !glights->lights[light_id].ready) {
+ dev_err(dev, "Event received for unconfigured light id: %d\n",
+ light_id);
+ return -EINVAL;
+ }
+
+ event = payload->event;
+
+ if (event & GB_LIGHTS_LIGHT_CONFIG) {
+ light = &glights->lights[light_id];
+
+ mutex_lock(&glights->lights_lock);
+ gb_lights_light_release(light);
+ ret = gb_lights_light_config(glights, light_id);
+ if (!ret)
+ ret = gb_lights_light_register(light);
+ if (ret < 0)
+ gb_lights_light_release(light);
+ mutex_unlock(&glights->lights_lock);
+ }
+
+ return ret;
+}
+
+static int gb_lights_probe(struct gb_bundle *bundle,
+ const struct greybus_bundle_id *id)
+{
+ struct greybus_descriptor_cport *cport_desc;
+ struct gb_connection *connection;
+ struct gb_lights *glights;
+ int ret;
+
+ if (bundle->num_cports != 1)
+ return -ENODEV;
+
+ cport_desc = &bundle->cport_desc[0];
+ if (cport_desc->protocol_id != GREYBUS_PROTOCOL_LIGHTS)
+ return -ENODEV;
+
+ glights = kzalloc(sizeof(*glights), GFP_KERNEL);
+ if (!glights)
+ return -ENOMEM;
+
+ mutex_init(&glights->lights_lock);
+
+ connection = gb_connection_create(bundle, le16_to_cpu(cport_desc->id),
+ gb_lights_request_handler);
+ if (IS_ERR(connection)) {
+ ret = PTR_ERR(connection);
+ goto out;
+ }
+
+ glights->connection = connection;
+ gb_connection_set_data(connection, glights);
+
+ greybus_set_drvdata(bundle, glights);
+
+ /* We aren't ready to receive an incoming request yet */
+ ret = gb_connection_enable_tx(connection);
+ if (ret)
+ goto error_connection_destroy;
+
+ /*
+ * Setup all the lights devices over this connection, if anything goes
+ * wrong tear down all lights
+ */
+ ret = gb_lights_create_all(glights);
+ if (ret < 0)
+ goto error_connection_disable;
+
+ /* We are ready to receive an incoming request now, enable RX as well */
+ ret = gb_connection_enable(connection);
+ if (ret)
+ goto error_connection_disable;
+
+ /* Enable & register lights */
+ ret = gb_lights_register_all(glights);
+ if (ret < 0)
+ goto error_connection_disable;
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return 0;
+
+error_connection_disable:
+ gb_connection_disable(connection);
+error_connection_destroy:
+ gb_connection_destroy(connection);
+out:
+ gb_lights_release(glights);
+ return ret;
+}
+
+static void gb_lights_disconnect(struct gb_bundle *bundle)
+{
+ struct gb_lights *glights = greybus_get_drvdata(bundle);
+
+ if (gb_pm_runtime_get_sync(bundle))
+ gb_pm_runtime_get_noresume(bundle);
+
+ gb_connection_disable(glights->connection);
+ gb_connection_destroy(glights->connection);
+
+ gb_lights_release(glights);
+}
+
+static const struct greybus_bundle_id gb_lights_id_table[] = {
+ { GREYBUS_DEVICE_CLASS(GREYBUS_CLASS_LIGHTS) },
+ { }
+};
+MODULE_DEVICE_TABLE(greybus, gb_lights_id_table);
+
+static struct greybus_driver gb_lights_driver = {
+ .name = "lights",
+ .probe = gb_lights_probe,
+ .disconnect = gb_lights_disconnect,
+ .id_table = gb_lights_id_table,
+};
+module_greybus_driver(gb_lights_driver);
+
+MODULE_LICENSE("GPL v2");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 15/32] greybus: logging driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (12 preceding siblings ...)
2016-09-16 14:10 ` [patch 14/32] greybus: LED driver Greg KH
@ 2016-09-16 14:10 ` Greg KH
2016-09-16 14:10 ` [patch 16/32] greybus: loopback driver Greg KH
` (16 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:10 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus logging protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/log.c | 132 ++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 132 insertions(+)
--- /dev/null
+++ b/drivers/greybus/log.c
@@ -0,0 +1,132 @@
+/*
+ * Greybus driver for the log protocol
+ *
+ * Copyright 2016 Google Inc.
+ *
+ * Released under the GPLv2 only.
+ */
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/sizes.h>
+#include <linux/uaccess.h>
+
+#include "greybus.h"
+
+struct gb_log {
+ struct gb_connection *connection;
+};
+
+static int gb_log_request_handler(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct device *dev = &connection->bundle->dev;
+ struct gb_log_send_log_request *receive;
+ u16 len;
+
+ if (op->type != GB_LOG_TYPE_SEND_LOG) {
+ dev_err(dev, "unknown request type 0x%02x\n", op->type);
+ return -EINVAL;
+ }
+
+ /* Verify size of payload */
+ if (op->request->payload_size < sizeof(*receive)) {
+ dev_err(dev, "log request too small (%zu < %zu)\n",
+ op->request->payload_size, sizeof(*receive));
+ return -EINVAL;
+ }
+ receive = op->request->payload;
+ len = le16_to_cpu(receive->len);
+ if (len != (int)(op->request->payload_size - sizeof(*receive))) {
+ dev_err(dev, "log request wrong size %d vs %d\n", len,
+ (int)(op->request->payload_size - sizeof(*receive)));
+ return -EINVAL;
+ }
+ if (len == 0) {
+ dev_err(dev, "log request of 0 bytes?\n");
+ return -EINVAL;
+ }
+
+ if (len > GB_LOG_MAX_LEN) {
+ dev_err(dev, "log request too big: %d\n", len);
+ return -EINVAL;
+ }
+
+ /* Ensure the buffer is 0 terminated */
+ receive->msg[len - 1] = '\0';
+
+ /* Print with dev_dbg() so that it can be easily turned off using
+ * dynamic debugging (and prevent any DoS) */
+ dev_dbg(dev, "%s", receive->msg);
+
+ return 0;
+}
+
+static int gb_log_probe(struct gb_bundle *bundle,
+ const struct greybus_bundle_id *id)
+{
+ struct greybus_descriptor_cport *cport_desc;
+ struct gb_connection *connection;
+ struct gb_log *log;
+ int retval;
+
+ if (bundle->num_cports != 1)
+ return -ENODEV;
+
+ cport_desc = &bundle->cport_desc[0];
+ if (cport_desc->protocol_id != GREYBUS_PROTOCOL_LOG)
+ return -ENODEV;
+
+ log = kzalloc(sizeof(*log), GFP_KERNEL);
+ if (!log)
+ return -ENOMEM;
+
+ connection = gb_connection_create(bundle, le16_to_cpu(cport_desc->id),
+ gb_log_request_handler);
+ if (IS_ERR(connection)) {
+ retval = PTR_ERR(connection);
+ goto error_free;
+ }
+
+ log->connection = connection;
+ greybus_set_drvdata(bundle, log);
+
+ retval = gb_connection_enable(connection);
+ if (retval)
+ goto error_connection_destroy;
+
+ return 0;
+
+error_connection_destroy:
+ gb_connection_destroy(connection);
+error_free:
+ kfree(log);
+ return retval;
+}
+
+static void gb_log_disconnect(struct gb_bundle *bundle)
+{
+ struct gb_log *log = greybus_get_drvdata(bundle);
+ struct gb_connection *connection = log->connection;
+
+ gb_connection_disable(connection);
+ gb_connection_destroy(connection);
+
+ kfree(log);
+}
+
+static const struct greybus_bundle_id gb_log_id_table[] = {
+ { GREYBUS_DEVICE_CLASS(GREYBUS_CLASS_LOG) },
+ { }
+};
+MODULE_DEVICE_TABLE(greybus, gb_log_id_table);
+
+static struct greybus_driver gb_log_driver = {
+ .name = "log",
+ .probe = gb_log_probe,
+ .disconnect = gb_log_disconnect,
+ .id_table = gb_log_id_table,
+};
+module_greybus_driver(gb_log_driver);
+
+MODULE_LICENSE("GPL v2");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 16/32] greybus: loopback driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (13 preceding siblings ...)
2016-09-16 14:10 ` [patch 15/32] greybus: logging driver Greg KH
@ 2016-09-16 14:10 ` Greg KH
2016-09-16 14:10 ` [patch 17/32] greybus: power supply driver Greg KH
` (15 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:10 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus loopback protocol.
It's useful for profiling hardware to determine problems and latencies in the system.
Yes, our most complex driver is a debugging tool, go figure...
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/loopback.c | 1365 +++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 1365 insertions(+)
--- /dev/null
+++ b/drivers/greybus/loopback.c
@@ -0,0 +1,1365 @@
+/*
+ * Loopback bridge driver for the Greybus loopback module.
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/slab.h>
+#include <linux/kthread.h>
+#include <linux/delay.h>
+#include <linux/random.h>
+#include <linux/sizes.h>
+#include <linux/cdev.h>
+#include <linux/fs.h>
+#include <linux/kfifo.h>
+#include <linux/debugfs.h>
+#include <linux/list_sort.h>
+#include <linux/spinlock.h>
+#include <linux/workqueue.h>
+#include <linux/atomic.h>
+#include <linux/pm_runtime.h>
+
+#include <asm/div64.h>
+
+#include "greybus.h"
+#include "connection.h"
+
+#define NSEC_PER_DAY 86400000000000ULL
+
+struct gb_loopback_stats {
+ u32 min;
+ u32 max;
+ u64 sum;
+ u32 count;
+};
+
+struct gb_loopback_device {
+ struct dentry *root;
+ u32 count;
+ size_t size_max;
+
+ /* We need to take a lock in atomic context */
+ spinlock_t lock;
+ struct list_head list;
+ struct list_head list_op_async;
+ wait_queue_head_t wq;
+};
+
+static struct gb_loopback_device gb_dev;
+
+struct gb_loopback_async_operation {
+ struct gb_loopback *gb;
+ struct gb_operation *operation;
+ struct timeval ts;
+ struct timer_list timer;
+ struct list_head entry;
+ struct work_struct work;
+ struct kref kref;
+ bool pending;
+ int (*completion)(struct gb_loopback_async_operation *op_async);
+};
+
+struct gb_loopback {
+ struct gb_connection *connection;
+
+ struct dentry *file;
+ struct kfifo kfifo_lat;
+ struct kfifo kfifo_ts;
+ struct mutex mutex;
+ struct task_struct *task;
+ struct list_head entry;
+ struct device *dev;
+ wait_queue_head_t wq;
+ wait_queue_head_t wq_completion;
+ atomic_t outstanding_operations;
+
+ /* Per connection stats */
+ struct timeval ts;
+ struct gb_loopback_stats latency;
+ struct gb_loopback_stats throughput;
+ struct gb_loopback_stats requests_per_second;
+ struct gb_loopback_stats apbridge_unipro_latency;
+ struct gb_loopback_stats gbphy_firmware_latency;
+
+ int type;
+ int async;
+ int id;
+ u32 size;
+ u32 iteration_max;
+ u32 iteration_count;
+ int us_wait;
+ u32 error;
+ u32 requests_completed;
+ u32 requests_timedout;
+ u32 timeout;
+ u32 jiffy_timeout;
+ u32 timeout_min;
+ u32 timeout_max;
+ u32 outstanding_operations_max;
+ u32 lbid;
+ u64 elapsed_nsecs;
+ u32 apbridge_latency_ts;
+ u32 gbphy_latency_ts;
+
+ u32 send_count;
+};
+
+static struct class loopback_class = {
+ .name = "gb_loopback",
+ .owner = THIS_MODULE,
+};
+static DEFINE_IDA(loopback_ida);
+
+/* Min/max values in jiffies */
+#define GB_LOOPBACK_TIMEOUT_MIN 1
+#define GB_LOOPBACK_TIMEOUT_MAX 10000
+
+#define GB_LOOPBACK_FIFO_DEFAULT 8192
+
+static unsigned kfifo_depth = GB_LOOPBACK_FIFO_DEFAULT;
+module_param(kfifo_depth, uint, 0444);
+
+/* Maximum size of any one send data buffer we support */
+#define MAX_PACKET_SIZE (PAGE_SIZE * 2)
+
+#define GB_LOOPBACK_US_WAIT_MAX 1000000
+
+/* interface sysfs attributes */
+#define gb_loopback_ro_attr(field) \
+static ssize_t field##_show(struct device *dev, \
+ struct device_attribute *attr, \
+ char *buf) \
+{ \
+ struct gb_loopback *gb = dev_get_drvdata(dev); \
+ return sprintf(buf, "%u\n", gb->field); \
+} \
+static DEVICE_ATTR_RO(field)
+
+#define gb_loopback_ro_stats_attr(name, field, type) \
+static ssize_t name##_##field##_show(struct device *dev, \
+ struct device_attribute *attr, \
+ char *buf) \
+{ \
+ struct gb_loopback *gb = dev_get_drvdata(dev); \
+ /* Report 0 for min and max if no transfer successed */ \
+ if (!gb->requests_completed) \
+ return sprintf(buf, "0\n"); \
+ return sprintf(buf, "%"#type"\n", gb->name.field); \
+} \
+static DEVICE_ATTR_RO(name##_##field)
+
+#define gb_loopback_ro_avg_attr(name) \
+static ssize_t name##_avg_show(struct device *dev, \
+ struct device_attribute *attr, \
+ char *buf) \
+{ \
+ struct gb_loopback_stats *stats; \
+ struct gb_loopback *gb; \
+ u64 avg, rem; \
+ u32 count; \
+ gb = dev_get_drvdata(dev); \
+ stats = &gb->name; \
+ count = stats->count ? stats->count : 1; \
+ avg = stats->sum + count / 2000000; /* round closest */ \
+ rem = do_div(avg, count); \
+ rem *= 1000000; \
+ do_div(rem, count); \
+ return sprintf(buf, "%llu.%06u\n", avg, (u32)rem); \
+} \
+static DEVICE_ATTR_RO(name##_avg)
+
+#define gb_loopback_stats_attrs(field) \
+ gb_loopback_ro_stats_attr(field, min, u); \
+ gb_loopback_ro_stats_attr(field, max, u); \
+ gb_loopback_ro_avg_attr(field)
+
+#define gb_loopback_attr(field, type) \
+static ssize_t field##_show(struct device *dev, \
+ struct device_attribute *attr, \
+ char *buf) \
+{ \
+ struct gb_loopback *gb = dev_get_drvdata(dev); \
+ return sprintf(buf, "%"#type"\n", gb->field); \
+} \
+static ssize_t field##_store(struct device *dev, \
+ struct device_attribute *attr, \
+ const char *buf, \
+ size_t len) \
+{ \
+ int ret; \
+ struct gb_loopback *gb = dev_get_drvdata(dev); \
+ mutex_lock(&gb->mutex); \
+ ret = sscanf(buf, "%"#type, &gb->field); \
+ if (ret != 1) \
+ len = -EINVAL; \
+ else \
+ gb_loopback_check_attr(gb, bundle); \
+ mutex_unlock(&gb->mutex); \
+ return len; \
+} \
+static DEVICE_ATTR_RW(field)
+
+#define gb_dev_loopback_ro_attr(field, conn) \
+static ssize_t field##_show(struct device *dev, \
+ struct device_attribute *attr, \
+ char *buf) \
+{ \
+ struct gb_loopback *gb = dev_get_drvdata(dev); \
+ return sprintf(buf, "%u\n", gb->field); \
+} \
+static DEVICE_ATTR_RO(field)
+
+#define gb_dev_loopback_rw_attr(field, type) \
+static ssize_t field##_show(struct device *dev, \
+ struct device_attribute *attr, \
+ char *buf) \
+{ \
+ struct gb_loopback *gb = dev_get_drvdata(dev); \
+ return sprintf(buf, "%"#type"\n", gb->field); \
+} \
+static ssize_t field##_store(struct device *dev, \
+ struct device_attribute *attr, \
+ const char *buf, \
+ size_t len) \
+{ \
+ int ret; \
+ struct gb_loopback *gb = dev_get_drvdata(dev); \
+ mutex_lock(&gb->mutex); \
+ ret = sscanf(buf, "%"#type, &gb->field); \
+ if (ret != 1) \
+ len = -EINVAL; \
+ else \
+ gb_loopback_check_attr(gb); \
+ mutex_unlock(&gb->mutex); \
+ return len; \
+} \
+static DEVICE_ATTR_RW(field)
+
+static void gb_loopback_reset_stats(struct gb_loopback *gb);
+static void gb_loopback_check_attr(struct gb_loopback *gb)
+{
+ if (gb->us_wait > GB_LOOPBACK_US_WAIT_MAX)
+ gb->us_wait = GB_LOOPBACK_US_WAIT_MAX;
+ if (gb->size > gb_dev.size_max)
+ gb->size = gb_dev.size_max;
+ gb->requests_timedout = 0;
+ gb->requests_completed = 0;
+ gb->iteration_count = 0;
+ gb->send_count = 0;
+ gb->error = 0;
+
+ if (kfifo_depth < gb->iteration_max) {
+ dev_warn(gb->dev,
+ "cannot log bytes %u kfifo_depth %u\n",
+ gb->iteration_max, kfifo_depth);
+ }
+ kfifo_reset_out(&gb->kfifo_lat);
+ kfifo_reset_out(&gb->kfifo_ts);
+
+ switch (gb->type) {
+ case GB_LOOPBACK_TYPE_PING:
+ case GB_LOOPBACK_TYPE_TRANSFER:
+ case GB_LOOPBACK_TYPE_SINK:
+ gb->jiffy_timeout = usecs_to_jiffies(gb->timeout);
+ if (!gb->jiffy_timeout)
+ gb->jiffy_timeout = GB_LOOPBACK_TIMEOUT_MIN;
+ else if (gb->jiffy_timeout > GB_LOOPBACK_TIMEOUT_MAX)
+ gb->jiffy_timeout = GB_LOOPBACK_TIMEOUT_MAX;
+ gb_loopback_reset_stats(gb);
+ wake_up(&gb->wq);
+ break;
+ default:
+ gb->type = 0;
+ break;
+ }
+}
+
+/* Time to send and receive one message */
+gb_loopback_stats_attrs(latency);
+/* Number of requests sent per second on this cport */
+gb_loopback_stats_attrs(requests_per_second);
+/* Quantity of data sent and received on this cport */
+gb_loopback_stats_attrs(throughput);
+/* Latency across the UniPro link from APBridge's perspective */
+gb_loopback_stats_attrs(apbridge_unipro_latency);
+/* Firmware induced overhead in the GPBridge */
+gb_loopback_stats_attrs(gbphy_firmware_latency);
+
+/* Number of errors encountered during loop */
+gb_loopback_ro_attr(error);
+/* Number of requests successfully completed async */
+gb_loopback_ro_attr(requests_completed);
+/* Number of requests timed out async */
+gb_loopback_ro_attr(requests_timedout);
+/* Timeout minimum in useconds */
+gb_loopback_ro_attr(timeout_min);
+/* Timeout minimum in useconds */
+gb_loopback_ro_attr(timeout_max);
+
+/*
+ * Type of loopback message to send based on protocol type definitions
+ * 0 => Don't send message
+ * 2 => Send ping message continuously (message without payload)
+ * 3 => Send transfer message continuously (message with payload,
+ * payload returned in response)
+ * 4 => Send a sink message (message with payload, no payload in response)
+ */
+gb_dev_loopback_rw_attr(type, d);
+/* Size of transfer message payload: 0-4096 bytes */
+gb_dev_loopback_rw_attr(size, u);
+/* Time to wait between two messages: 0-1000 ms */
+gb_dev_loopback_rw_attr(us_wait, d);
+/* Maximum iterations for a given operation: 1-(2^32-1), 0 implies infinite */
+gb_dev_loopback_rw_attr(iteration_max, u);
+/* The current index of the for (i = 0; i < iteration_max; i++) loop */
+gb_dev_loopback_ro_attr(iteration_count, false);
+/* A flag to indicate synchronous or asynchronous operations */
+gb_dev_loopback_rw_attr(async, u);
+/* Timeout of an individual asynchronous request */
+gb_dev_loopback_rw_attr(timeout, u);
+/* Maximum number of in-flight operations before back-off */
+gb_dev_loopback_rw_attr(outstanding_operations_max, u);
+
+static struct attribute *loopback_attrs[] = {
+ &dev_attr_latency_min.attr,
+ &dev_attr_latency_max.attr,
+ &dev_attr_latency_avg.attr,
+ &dev_attr_requests_per_second_min.attr,
+ &dev_attr_requests_per_second_max.attr,
+ &dev_attr_requests_per_second_avg.attr,
+ &dev_attr_throughput_min.attr,
+ &dev_attr_throughput_max.attr,
+ &dev_attr_throughput_avg.attr,
+ &dev_attr_apbridge_unipro_latency_min.attr,
+ &dev_attr_apbridge_unipro_latency_max.attr,
+ &dev_attr_apbridge_unipro_latency_avg.attr,
+ &dev_attr_gbphy_firmware_latency_min.attr,
+ &dev_attr_gbphy_firmware_latency_max.attr,
+ &dev_attr_gbphy_firmware_latency_avg.attr,
+ &dev_attr_type.attr,
+ &dev_attr_size.attr,
+ &dev_attr_us_wait.attr,
+ &dev_attr_iteration_count.attr,
+ &dev_attr_iteration_max.attr,
+ &dev_attr_async.attr,
+ &dev_attr_error.attr,
+ &dev_attr_requests_completed.attr,
+ &dev_attr_requests_timedout.attr,
+ &dev_attr_timeout.attr,
+ &dev_attr_outstanding_operations_max.attr,
+ &dev_attr_timeout_min.attr,
+ &dev_attr_timeout_max.attr,
+ NULL,
+};
+ATTRIBUTE_GROUPS(loopback);
+
+static void gb_loopback_calculate_stats(struct gb_loopback *gb, bool error);
+
+static u32 gb_loopback_nsec_to_usec_latency(u64 elapsed_nsecs)
+{
+ u32 lat;
+
+ do_div(elapsed_nsecs, NSEC_PER_USEC);
+ lat = elapsed_nsecs;
+ return lat;
+}
+
+static u64 __gb_loopback_calc_latency(u64 t1, u64 t2)
+{
+ if (t2 > t1)
+ return t2 - t1;
+ else
+ return NSEC_PER_DAY - t2 + t1;
+}
+
+static u64 gb_loopback_calc_latency(struct timeval *ts, struct timeval *te)
+{
+ u64 t1, t2;
+
+ t1 = timeval_to_ns(ts);
+ t2 = timeval_to_ns(te);
+
+ return __gb_loopback_calc_latency(t1, t2);
+}
+
+static void gb_loopback_push_latency_ts(struct gb_loopback *gb,
+ struct timeval *ts, struct timeval *te)
+{
+ kfifo_in(&gb->kfifo_ts, (unsigned char *)ts, sizeof(*ts));
+ kfifo_in(&gb->kfifo_ts, (unsigned char *)te, sizeof(*te));
+}
+
+static int gb_loopback_operation_sync(struct gb_loopback *gb, int type,
+ void *request, int request_size,
+ void *response, int response_size)
+{
+ struct gb_operation *operation;
+ struct timeval ts, te;
+ int ret;
+
+ do_gettimeofday(&ts);
+ operation = gb_operation_create(gb->connection, type, request_size,
+ response_size, GFP_KERNEL);
+ if (!operation)
+ return -ENOMEM;
+
+ if (request_size)
+ memcpy(operation->request->payload, request, request_size);
+
+ ret = gb_operation_request_send_sync(operation);
+ if (ret) {
+ dev_err(&gb->connection->bundle->dev,
+ "synchronous operation failed: %d\n", ret);
+ goto out_put_operation;
+ } else {
+ if (response_size == operation->response->payload_size) {
+ memcpy(response, operation->response->payload,
+ response_size);
+ } else {
+ dev_err(&gb->connection->bundle->dev,
+ "response size %zu expected %d\n",
+ operation->response->payload_size,
+ response_size);
+ ret = -EINVAL;
+ goto out_put_operation;
+ }
+ }
+
+ do_gettimeofday(&te);
+
+ /* Calculate the total time the message took */
+ gb_loopback_push_latency_ts(gb, &ts, &te);
+ gb->elapsed_nsecs = gb_loopback_calc_latency(&ts, &te);
+
+out_put_operation:
+ gb_operation_put(operation);
+
+ return ret;
+}
+
+static void __gb_loopback_async_operation_destroy(struct kref *kref)
+{
+ struct gb_loopback_async_operation *op_async;
+
+ op_async = container_of(kref, struct gb_loopback_async_operation, kref);
+
+ list_del(&op_async->entry);
+ if (op_async->operation)
+ gb_operation_put(op_async->operation);
+ atomic_dec(&op_async->gb->outstanding_operations);
+ wake_up(&op_async->gb->wq_completion);
+ kfree(op_async);
+}
+
+static void gb_loopback_async_operation_get(struct gb_loopback_async_operation
+ *op_async)
+{
+ kref_get(&op_async->kref);
+}
+
+static void gb_loopback_async_operation_put(struct gb_loopback_async_operation
+ *op_async)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(&gb_dev.lock, flags);
+ kref_put(&op_async->kref, __gb_loopback_async_operation_destroy);
+ spin_unlock_irqrestore(&gb_dev.lock, flags);
+}
+
+static struct gb_loopback_async_operation *
+ gb_loopback_operation_find(u16 id)
+{
+ struct gb_loopback_async_operation *op_async;
+ bool found = false;
+ unsigned long flags;
+
+ spin_lock_irqsave(&gb_dev.lock, flags);
+ list_for_each_entry(op_async, &gb_dev.list_op_async, entry) {
+ if (op_async->operation->id == id) {
+ gb_loopback_async_operation_get(op_async);
+ found = true;
+ break;
+ }
+ }
+ spin_unlock_irqrestore(&gb_dev.lock, flags);
+
+ return found ? op_async : NULL;
+}
+
+static void gb_loopback_async_wait_all(struct gb_loopback *gb)
+{
+ wait_event(gb->wq_completion,
+ !atomic_read(&gb->outstanding_operations));
+}
+
+static void gb_loopback_async_operation_callback(struct gb_operation *operation)
+{
+ struct gb_loopback_async_operation *op_async;
+ struct gb_loopback *gb;
+ struct timeval te;
+ bool err = false;
+
+ do_gettimeofday(&te);
+ op_async = gb_loopback_operation_find(operation->id);
+ if (!op_async)
+ return;
+
+ gb = op_async->gb;
+ mutex_lock(&gb->mutex);
+
+ if (!op_async->pending || gb_operation_result(operation)) {
+ err = true;
+ } else {
+ if (op_async->completion)
+ if (op_async->completion(op_async))
+ err = true;
+ }
+
+ if (!err) {
+ gb_loopback_push_latency_ts(gb, &op_async->ts, &te);
+ gb->elapsed_nsecs = gb_loopback_calc_latency(&op_async->ts,
+ &te);
+ }
+
+ if (op_async->pending) {
+ if (err)
+ gb->error++;
+ gb->iteration_count++;
+ op_async->pending = false;
+ del_timer_sync(&op_async->timer);
+ gb_loopback_async_operation_put(op_async);
+ gb_loopback_calculate_stats(gb, err);
+ }
+ mutex_unlock(&gb->mutex);
+
+ dev_dbg(&gb->connection->bundle->dev, "complete operation %d\n",
+ operation->id);
+
+ gb_loopback_async_operation_put(op_async);
+}
+
+static void gb_loopback_async_operation_work(struct work_struct *work)
+{
+ struct gb_loopback *gb;
+ struct gb_operation *operation;
+ struct gb_loopback_async_operation *op_async;
+
+ op_async = container_of(work, struct gb_loopback_async_operation, work);
+ gb = op_async->gb;
+ operation = op_async->operation;
+
+ mutex_lock(&gb->mutex);
+ if (op_async->pending) {
+ gb->requests_timedout++;
+ gb->error++;
+ gb->iteration_count++;
+ op_async->pending = false;
+ gb_loopback_async_operation_put(op_async);
+ gb_loopback_calculate_stats(gb, true);
+ }
+ mutex_unlock(&gb->mutex);
+
+ dev_dbg(&gb->connection->bundle->dev, "timeout operation %d\n",
+ operation->id);
+
+ gb_operation_cancel(operation, -ETIMEDOUT);
+ gb_loopback_async_operation_put(op_async);
+}
+
+static void gb_loopback_async_operation_timeout(unsigned long data)
+{
+ struct gb_loopback_async_operation *op_async;
+ u16 id = data;
+
+ op_async = gb_loopback_operation_find(id);
+ if (!op_async) {
+ pr_err("operation %d not found - time out ?\n", id);
+ return;
+ }
+ schedule_work(&op_async->work);
+}
+
+static int gb_loopback_async_operation(struct gb_loopback *gb, int type,
+ void *request, int request_size,
+ int response_size,
+ void *completion)
+{
+ struct gb_loopback_async_operation *op_async;
+ struct gb_operation *operation;
+ int ret;
+ unsigned long flags;
+
+ op_async = kzalloc(sizeof(*op_async), GFP_KERNEL);
+ if (!op_async)
+ return -ENOMEM;
+
+ INIT_WORK(&op_async->work, gb_loopback_async_operation_work);
+ init_timer(&op_async->timer);
+ kref_init(&op_async->kref);
+
+ operation = gb_operation_create(gb->connection, type, request_size,
+ response_size, GFP_KERNEL);
+ if (!operation) {
+ kfree(op_async);
+ return -ENOMEM;
+ }
+
+ if (request_size)
+ memcpy(operation->request->payload, request, request_size);
+
+ op_async->gb = gb;
+ op_async->operation = operation;
+ op_async->completion = completion;
+
+ spin_lock_irqsave(&gb_dev.lock, flags);
+ list_add_tail(&op_async->entry, &gb_dev.list_op_async);
+ spin_unlock_irqrestore(&gb_dev.lock, flags);
+
+ do_gettimeofday(&op_async->ts);
+ op_async->pending = true;
+ atomic_inc(&gb->outstanding_operations);
+ mutex_lock(&gb->mutex);
+ ret = gb_operation_request_send(operation,
+ gb_loopback_async_operation_callback,
+ GFP_KERNEL);
+ if (ret)
+ goto error;
+
+ op_async->timer.function = gb_loopback_async_operation_timeout;
+ op_async->timer.expires = jiffies + gb->jiffy_timeout;
+ op_async->timer.data = (unsigned long)operation->id;
+ add_timer(&op_async->timer);
+
+ goto done;
+error:
+ gb_loopback_async_operation_put(op_async);
+done:
+ mutex_unlock(&gb->mutex);
+ return ret;
+}
+
+static int gb_loopback_sync_sink(struct gb_loopback *gb, u32 len)
+{
+ struct gb_loopback_transfer_request *request;
+ int retval;
+
+ request = kmalloc(len + sizeof(*request), GFP_KERNEL);
+ if (!request)
+ return -ENOMEM;
+
+ request->len = cpu_to_le32(len);
+ retval = gb_loopback_operation_sync(gb, GB_LOOPBACK_TYPE_SINK,
+ request, len + sizeof(*request),
+ NULL, 0);
+ kfree(request);
+ return retval;
+}
+
+static int gb_loopback_sync_transfer(struct gb_loopback *gb, u32 len)
+{
+ struct gb_loopback_transfer_request *request;
+ struct gb_loopback_transfer_response *response;
+ int retval;
+
+ gb->apbridge_latency_ts = 0;
+ gb->gbphy_latency_ts = 0;
+
+ request = kmalloc(len + sizeof(*request), GFP_KERNEL);
+ if (!request)
+ return -ENOMEM;
+ response = kmalloc(len + sizeof(*response), GFP_KERNEL);
+ if (!response) {
+ kfree(request);
+ return -ENOMEM;
+ }
+
+ memset(request->data, 0x5A, len);
+
+ request->len = cpu_to_le32(len);
+ retval = gb_loopback_operation_sync(gb, GB_LOOPBACK_TYPE_TRANSFER,
+ request, len + sizeof(*request),
+ response, len + sizeof(*response));
+ if (retval)
+ goto gb_error;
+
+ if (memcmp(request->data, response->data, len)) {
+ dev_err(&gb->connection->bundle->dev,
+ "Loopback Data doesn't match\n");
+ retval = -EREMOTEIO;
+ }
+ gb->apbridge_latency_ts = (u32)__le32_to_cpu(response->reserved0);
+ gb->gbphy_latency_ts = (u32)__le32_to_cpu(response->reserved1);
+
+gb_error:
+ kfree(request);
+ kfree(response);
+
+ return retval;
+}
+
+static int gb_loopback_sync_ping(struct gb_loopback *gb)
+{
+ return gb_loopback_operation_sync(gb, GB_LOOPBACK_TYPE_PING,
+ NULL, 0, NULL, 0);
+}
+
+static int gb_loopback_async_sink(struct gb_loopback *gb, u32 len)
+{
+ struct gb_loopback_transfer_request *request;
+ int retval;
+
+ request = kmalloc(len + sizeof(*request), GFP_KERNEL);
+ if (!request)
+ return -ENOMEM;
+
+ request->len = cpu_to_le32(len);
+ retval = gb_loopback_async_operation(gb, GB_LOOPBACK_TYPE_SINK,
+ request, len + sizeof(*request),
+ 0, NULL);
+ kfree(request);
+ return retval;
+}
+
+static int gb_loopback_async_transfer_complete(
+ struct gb_loopback_async_operation *op_async)
+{
+ struct gb_loopback *gb;
+ struct gb_operation *operation;
+ struct gb_loopback_transfer_request *request;
+ struct gb_loopback_transfer_response *response;
+ size_t len;
+ int retval = 0;
+
+ gb = op_async->gb;
+ operation = op_async->operation;
+ request = operation->request->payload;
+ response = operation->response->payload;
+ len = le32_to_cpu(request->len);
+
+ if (memcmp(request->data, response->data, len)) {
+ dev_err(&gb->connection->bundle->dev,
+ "Loopback Data doesn't match operation id %d\n",
+ operation->id);
+ retval = -EREMOTEIO;
+ } else {
+ gb->apbridge_latency_ts =
+ (u32)__le32_to_cpu(response->reserved0);
+ gb->gbphy_latency_ts =
+ (u32)__le32_to_cpu(response->reserved1);
+ }
+
+ return retval;
+}
+
+static int gb_loopback_async_transfer(struct gb_loopback *gb, u32 len)
+{
+ struct gb_loopback_transfer_request *request;
+ int retval, response_len;
+
+ request = kmalloc(len + sizeof(*request), GFP_KERNEL);
+ if (!request)
+ return -ENOMEM;
+
+ memset(request->data, 0x5A, len);
+
+ request->len = cpu_to_le32(len);
+ response_len = sizeof(struct gb_loopback_transfer_response);
+ retval = gb_loopback_async_operation(gb, GB_LOOPBACK_TYPE_TRANSFER,
+ request, len + sizeof(*request),
+ len + response_len,
+ gb_loopback_async_transfer_complete);
+ if (retval)
+ goto gb_error;
+
+gb_error:
+ kfree(request);
+ return retval;
+}
+
+static int gb_loopback_async_ping(struct gb_loopback *gb)
+{
+ return gb_loopback_async_operation(gb, GB_LOOPBACK_TYPE_PING,
+ NULL, 0, 0, NULL);
+}
+
+static int gb_loopback_request_handler(struct gb_operation *operation)
+{
+ struct gb_connection *connection = operation->connection;
+ struct gb_loopback_transfer_request *request;
+ struct gb_loopback_transfer_response *response;
+ struct device *dev = &connection->bundle->dev;
+ size_t len;
+
+ /* By convention, the AP initiates the version operation */
+ switch (operation->type) {
+ case GB_LOOPBACK_TYPE_PING:
+ case GB_LOOPBACK_TYPE_SINK:
+ return 0;
+ case GB_LOOPBACK_TYPE_TRANSFER:
+ if (operation->request->payload_size < sizeof(*request)) {
+ dev_err(dev, "transfer request too small (%zu < %zu)\n",
+ operation->request->payload_size,
+ sizeof(*request));
+ return -EINVAL; /* -EMSGSIZE */
+ }
+ request = operation->request->payload;
+ len = le32_to_cpu(request->len);
+ if (len > gb_dev.size_max) {
+ dev_err(dev, "transfer request too large (%zu > %zu)\n",
+ len, gb_dev.size_max);
+ return -EINVAL;
+ }
+
+ if (!gb_operation_response_alloc(operation,
+ len + sizeof(*response), GFP_KERNEL)) {
+ dev_err(dev, "error allocating response\n");
+ return -ENOMEM;
+ }
+ response = operation->response->payload;
+ response->len = cpu_to_le32(len);
+ if (len)
+ memcpy(response->data, request->data, len);
+
+ return 0;
+ default:
+ dev_err(dev, "unsupported request: %u\n", operation->type);
+ return -EINVAL;
+ }
+}
+
+static void gb_loopback_reset_stats(struct gb_loopback *gb)
+{
+ struct gb_loopback_stats reset = {
+ .min = U32_MAX,
+ };
+
+ /* Reset per-connection stats */
+ memcpy(&gb->latency, &reset,
+ sizeof(struct gb_loopback_stats));
+ memcpy(&gb->throughput, &reset,
+ sizeof(struct gb_loopback_stats));
+ memcpy(&gb->requests_per_second, &reset,
+ sizeof(struct gb_loopback_stats));
+ memcpy(&gb->apbridge_unipro_latency, &reset,
+ sizeof(struct gb_loopback_stats));
+ memcpy(&gb->gbphy_firmware_latency, &reset,
+ sizeof(struct gb_loopback_stats));
+
+ /* Should be initialized at least once per transaction set */
+ gb->apbridge_latency_ts = 0;
+ gb->gbphy_latency_ts = 0;
+ memset(&gb->ts, 0, sizeof(struct timeval));
+}
+
+static void gb_loopback_update_stats(struct gb_loopback_stats *stats, u32 val)
+{
+ if (stats->min > val)
+ stats->min = val;
+ if (stats->max < val)
+ stats->max = val;
+ stats->sum += val;
+ stats->count++;
+}
+
+static void gb_loopback_update_stats_window(struct gb_loopback_stats *stats,
+ u64 val, u32 count)
+{
+ stats->sum += val;
+ stats->count += count;
+
+ do_div(val, count);
+ if (stats->min > val)
+ stats->min = val;
+ if (stats->max < val)
+ stats->max = val;
+}
+
+static void gb_loopback_requests_update(struct gb_loopback *gb, u32 latency)
+{
+ u64 req = gb->requests_completed * USEC_PER_SEC;
+
+ gb_loopback_update_stats_window(&gb->requests_per_second, req, latency);
+}
+
+static void gb_loopback_throughput_update(struct gb_loopback *gb, u32 latency)
+{
+ u64 aggregate_size = sizeof(struct gb_operation_msg_hdr) * 2;
+
+ switch (gb->type) {
+ case GB_LOOPBACK_TYPE_PING:
+ break;
+ case GB_LOOPBACK_TYPE_SINK:
+ aggregate_size += sizeof(struct gb_loopback_transfer_request) +
+ gb->size;
+ break;
+ case GB_LOOPBACK_TYPE_TRANSFER:
+ aggregate_size += sizeof(struct gb_loopback_transfer_request) +
+ sizeof(struct gb_loopback_transfer_response) +
+ gb->size * 2;
+ break;
+ default:
+ return;
+ }
+
+ aggregate_size *= gb->requests_completed;
+ aggregate_size *= USEC_PER_SEC;
+ gb_loopback_update_stats_window(&gb->throughput, aggregate_size,
+ latency);
+}
+
+static void gb_loopback_calculate_latency_stats(struct gb_loopback *gb)
+{
+ u32 lat;
+
+ /* Express latency in terms of microseconds */
+ lat = gb_loopback_nsec_to_usec_latency(gb->elapsed_nsecs);
+
+ /* Log latency stastic */
+ gb_loopback_update_stats(&gb->latency, lat);
+
+ /* Raw latency log on a per thread basis */
+ kfifo_in(&gb->kfifo_lat, (unsigned char *)&lat, sizeof(lat));
+
+ /* Log the firmware supplied latency values */
+ gb_loopback_update_stats(&gb->apbridge_unipro_latency,
+ gb->apbridge_latency_ts);
+ gb_loopback_update_stats(&gb->gbphy_firmware_latency,
+ gb->gbphy_latency_ts);
+}
+
+static void gb_loopback_calculate_stats(struct gb_loopback *gb, bool error)
+{
+ u64 nlat;
+ u32 lat;
+ struct timeval te;
+
+ if (!error) {
+ gb->requests_completed++;
+ gb_loopback_calculate_latency_stats(gb);
+ }
+
+ do_gettimeofday(&te);
+ nlat = gb_loopback_calc_latency(&gb->ts, &te);
+ if (nlat >= NSEC_PER_SEC || gb->iteration_count == gb->iteration_max) {
+ lat = gb_loopback_nsec_to_usec_latency(nlat);
+
+ gb_loopback_throughput_update(gb, lat);
+ gb_loopback_requests_update(gb, lat);
+
+ if (gb->iteration_count != gb->iteration_max) {
+ gb->ts = te;
+ gb->requests_completed = 0;
+ }
+ }
+}
+
+static void gb_loopback_async_wait_to_send(struct gb_loopback *gb)
+{
+ if (!(gb->async && gb->outstanding_operations_max))
+ return;
+ wait_event_interruptible(gb->wq_completion,
+ (atomic_read(&gb->outstanding_operations) <
+ gb->outstanding_operations_max) ||
+ kthread_should_stop());
+}
+
+static int gb_loopback_fn(void *data)
+{
+ int error = 0;
+ int us_wait = 0;
+ int type;
+ int ret;
+ u32 size;
+
+ struct gb_loopback *gb = data;
+ struct gb_bundle *bundle = gb->connection->bundle;
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ return ret;
+
+ while (1) {
+ if (!gb->type) {
+ gb_pm_runtime_put_autosuspend(bundle);
+ wait_event_interruptible(gb->wq, gb->type ||
+ kthread_should_stop());
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ return ret;
+ }
+
+ if (kthread_should_stop())
+ break;
+
+ /* Limit the maximum number of in-flight async operations */
+ gb_loopback_async_wait_to_send(gb);
+ if (kthread_should_stop())
+ break;
+
+ mutex_lock(&gb->mutex);
+
+ /* Optionally terminate */
+ if (gb->send_count == gb->iteration_max) {
+ if (gb->iteration_count == gb->iteration_max) {
+ gb->type = 0;
+ gb->send_count = 0;
+ sysfs_notify(&gb->dev->kobj, NULL,
+ "iteration_count");
+ }
+ mutex_unlock(&gb->mutex);
+ continue;
+ }
+ size = gb->size;
+ us_wait = gb->us_wait;
+ type = gb->type;
+ if (gb->ts.tv_usec == 0 && gb->ts.tv_sec == 0)
+ do_gettimeofday(&gb->ts);
+ mutex_unlock(&gb->mutex);
+
+ /* Else operations to perform */
+ if (gb->async) {
+ if (type == GB_LOOPBACK_TYPE_PING) {
+ error = gb_loopback_async_ping(gb);
+ } else if (type == GB_LOOPBACK_TYPE_TRANSFER) {
+ error = gb_loopback_async_transfer(gb, size);
+ } else if (type == GB_LOOPBACK_TYPE_SINK) {
+ error = gb_loopback_async_sink(gb, size);
+ }
+
+ if (error)
+ gb->error++;
+ } else {
+ /* We are effectively single threaded here */
+ if (type == GB_LOOPBACK_TYPE_PING)
+ error = gb_loopback_sync_ping(gb);
+ else if (type == GB_LOOPBACK_TYPE_TRANSFER)
+ error = gb_loopback_sync_transfer(gb, size);
+ else if (type == GB_LOOPBACK_TYPE_SINK)
+ error = gb_loopback_sync_sink(gb, size);
+
+ if (error)
+ gb->error++;
+ gb->iteration_count++;
+ gb_loopback_calculate_stats(gb, !!error);
+ }
+ gb->send_count++;
+ if (us_wait)
+ udelay(us_wait);
+ }
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return 0;
+}
+
+static int gb_loopback_dbgfs_latency_show_common(struct seq_file *s,
+ struct kfifo *kfifo,
+ struct mutex *mutex)
+{
+ u32 latency;
+ int retval;
+
+ if (kfifo_len(kfifo) == 0) {
+ retval = -EAGAIN;
+ goto done;
+ }
+
+ mutex_lock(mutex);
+ retval = kfifo_out(kfifo, &latency, sizeof(latency));
+ if (retval > 0) {
+ seq_printf(s, "%u", latency);
+ retval = 0;
+ }
+ mutex_unlock(mutex);
+done:
+ return retval;
+}
+
+static int gb_loopback_dbgfs_latency_show(struct seq_file *s, void *unused)
+{
+ struct gb_loopback *gb = s->private;
+
+ return gb_loopback_dbgfs_latency_show_common(s, &gb->kfifo_lat,
+ &gb->mutex);
+}
+
+static int gb_loopback_latency_open(struct inode *inode, struct file *file)
+{
+ return single_open(file, gb_loopback_dbgfs_latency_show,
+ inode->i_private);
+}
+
+static const struct file_operations gb_loopback_debugfs_latency_ops = {
+ .open = gb_loopback_latency_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = single_release,
+};
+
+static int gb_loopback_bus_id_compare(void *priv, struct list_head *lha,
+ struct list_head *lhb)
+{
+ struct gb_loopback *a = list_entry(lha, struct gb_loopback, entry);
+ struct gb_loopback *b = list_entry(lhb, struct gb_loopback, entry);
+ struct gb_connection *ca = a->connection;
+ struct gb_connection *cb = b->connection;
+
+ if (ca->bundle->intf->interface_id < cb->bundle->intf->interface_id)
+ return -1;
+ if (cb->bundle->intf->interface_id < ca->bundle->intf->interface_id)
+ return 1;
+ if (ca->bundle->id < cb->bundle->id)
+ return -1;
+ if (cb->bundle->id < ca->bundle->id)
+ return 1;
+ if (ca->intf_cport_id < cb->intf_cport_id)
+ return -1;
+ else if (cb->intf_cport_id < ca->intf_cport_id)
+ return 1;
+
+ return 0;
+}
+
+static void gb_loopback_insert_id(struct gb_loopback *gb)
+{
+ struct gb_loopback *gb_list;
+ u32 new_lbid = 0;
+
+ /* perform an insertion sort */
+ list_add_tail(&gb->entry, &gb_dev.list);
+ list_sort(NULL, &gb_dev.list, gb_loopback_bus_id_compare);
+ list_for_each_entry(gb_list, &gb_dev.list, entry) {
+ gb_list->lbid = 1 << new_lbid;
+ new_lbid++;
+ }
+}
+
+#define DEBUGFS_NAMELEN 32
+
+static int gb_loopback_probe(struct gb_bundle *bundle,
+ const struct greybus_bundle_id *id)
+{
+ struct greybus_descriptor_cport *cport_desc;
+ struct gb_connection *connection;
+ struct gb_loopback *gb;
+ struct device *dev;
+ int retval;
+ char name[DEBUGFS_NAMELEN];
+ unsigned long flags;
+
+ if (bundle->num_cports != 1)
+ return -ENODEV;
+
+ cport_desc = &bundle->cport_desc[0];
+ if (cport_desc->protocol_id != GREYBUS_PROTOCOL_LOOPBACK)
+ return -ENODEV;
+
+ gb = kzalloc(sizeof(*gb), GFP_KERNEL);
+ if (!gb)
+ return -ENOMEM;
+
+ connection = gb_connection_create(bundle, le16_to_cpu(cport_desc->id),
+ gb_loopback_request_handler);
+ if (IS_ERR(connection)) {
+ retval = PTR_ERR(connection);
+ goto out_kzalloc;
+ }
+
+ gb->connection = connection;
+ greybus_set_drvdata(bundle, gb);
+
+ init_waitqueue_head(&gb->wq);
+ init_waitqueue_head(&gb->wq_completion);
+ atomic_set(&gb->outstanding_operations, 0);
+ gb_loopback_reset_stats(gb);
+
+ /* Reported values to user-space for min/max timeouts */
+ gb->timeout_min = jiffies_to_usecs(GB_LOOPBACK_TIMEOUT_MIN);
+ gb->timeout_max = jiffies_to_usecs(GB_LOOPBACK_TIMEOUT_MAX);
+
+ if (!gb_dev.count) {
+ /* Calculate maximum payload */
+ gb_dev.size_max = gb_operation_get_payload_size_max(connection);
+ if (gb_dev.size_max <=
+ sizeof(struct gb_loopback_transfer_request)) {
+ retval = -EINVAL;
+ goto out_connection_destroy;
+ }
+ gb_dev.size_max -= sizeof(struct gb_loopback_transfer_request);
+ }
+
+ /* Create per-connection sysfs and debugfs data-points */
+ snprintf(name, sizeof(name), "raw_latency_%s",
+ dev_name(&connection->bundle->dev));
+ gb->file = debugfs_create_file(name, S_IFREG | S_IRUGO, gb_dev.root, gb,
+ &gb_loopback_debugfs_latency_ops);
+
+ gb->id = ida_simple_get(&loopback_ida, 0, 0, GFP_KERNEL);
+ if (gb->id < 0) {
+ retval = gb->id;
+ goto out_debugfs_remove;
+ }
+
+ retval = gb_connection_enable(connection);
+ if (retval)
+ goto out_ida_remove;
+
+ dev = device_create_with_groups(&loopback_class,
+ &connection->bundle->dev,
+ MKDEV(0, 0), gb, loopback_groups,
+ "gb_loopback%d", gb->id);
+ if (IS_ERR(dev)) {
+ retval = PTR_ERR(dev);
+ goto out_connection_disable;
+ }
+ gb->dev = dev;
+
+ /* Allocate kfifo */
+ if (kfifo_alloc(&gb->kfifo_lat, kfifo_depth * sizeof(u32),
+ GFP_KERNEL)) {
+ retval = -ENOMEM;
+ goto out_conn;
+ }
+ if (kfifo_alloc(&gb->kfifo_ts, kfifo_depth * sizeof(struct timeval) * 2,
+ GFP_KERNEL)) {
+ retval = -ENOMEM;
+ goto out_kfifo0;
+ }
+
+ /* Fork worker thread */
+ mutex_init(&gb->mutex);
+ gb->task = kthread_run(gb_loopback_fn, gb, "gb_loopback");
+ if (IS_ERR(gb->task)) {
+ retval = PTR_ERR(gb->task);
+ goto out_kfifo1;
+ }
+
+ spin_lock_irqsave(&gb_dev.lock, flags);
+ gb_loopback_insert_id(gb);
+ gb_dev.count++;
+ spin_unlock_irqrestore(&gb_dev.lock, flags);
+
+ gb_connection_latency_tag_enable(connection);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return 0;
+
+out_kfifo1:
+ kfifo_free(&gb->kfifo_ts);
+out_kfifo0:
+ kfifo_free(&gb->kfifo_lat);
+out_conn:
+ device_unregister(dev);
+out_connection_disable:
+ gb_connection_disable(connection);
+out_ida_remove:
+ ida_simple_remove(&loopback_ida, gb->id);
+out_debugfs_remove:
+ debugfs_remove(gb->file);
+out_connection_destroy:
+ gb_connection_destroy(connection);
+out_kzalloc:
+ kfree(gb);
+
+ return retval;
+}
+
+static void gb_loopback_disconnect(struct gb_bundle *bundle)
+{
+ struct gb_loopback *gb = greybus_get_drvdata(bundle);
+ unsigned long flags;
+ int ret;
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ gb_pm_runtime_get_noresume(bundle);
+
+ gb_connection_disable(gb->connection);
+
+ if (!IS_ERR_OR_NULL(gb->task))
+ kthread_stop(gb->task);
+
+ kfifo_free(&gb->kfifo_lat);
+ kfifo_free(&gb->kfifo_ts);
+ gb_connection_latency_tag_disable(gb->connection);
+ debugfs_remove(gb->file);
+
+ /*
+ * FIXME: gb_loopback_async_wait_all() is redundant now, as connection
+ * is disabled at the beginning and so we can't have any more
+ * incoming/outgoing requests.
+ */
+ gb_loopback_async_wait_all(gb);
+
+ spin_lock_irqsave(&gb_dev.lock, flags);
+ gb_dev.count--;
+ list_del(&gb->entry);
+ spin_unlock_irqrestore(&gb_dev.lock, flags);
+
+ device_unregister(gb->dev);
+ ida_simple_remove(&loopback_ida, gb->id);
+
+ gb_connection_destroy(gb->connection);
+ kfree(gb);
+}
+
+static const struct greybus_bundle_id gb_loopback_id_table[] = {
+ { GREYBUS_DEVICE_CLASS(GREYBUS_CLASS_LOOPBACK) },
+ { }
+};
+MODULE_DEVICE_TABLE(greybus, gb_loopback_id_table);
+
+static struct greybus_driver gb_loopback_driver = {
+ .name = "loopback",
+ .probe = gb_loopback_probe,
+ .disconnect = gb_loopback_disconnect,
+ .id_table = gb_loopback_id_table,
+};
+
+static int loopback_init(void)
+{
+ int retval;
+
+ INIT_LIST_HEAD(&gb_dev.list);
+ INIT_LIST_HEAD(&gb_dev.list_op_async);
+ spin_lock_init(&gb_dev.lock);
+ gb_dev.root = debugfs_create_dir("gb_loopback", NULL);
+
+ retval = class_register(&loopback_class);
+ if (retval)
+ goto err;
+
+ retval = greybus_register(&gb_loopback_driver);
+ if (retval)
+ goto err_unregister;
+
+ return 0;
+
+err_unregister:
+ class_unregister(&loopback_class);
+err:
+ debugfs_remove_recursive(gb_dev.root);
+ return retval;
+}
+module_init(loopback_init);
+
+static void __exit loopback_exit(void)
+{
+ debugfs_remove_recursive(gb_dev.root);
+ greybus_deregister(&gb_loopback_driver);
+ class_unregister(&loopback_class);
+ ida_destroy(&loopback_ida);
+}
+module_exit(loopback_exit);
+
+MODULE_LICENSE("GPL v2");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 17/32] greybus: power supply driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (14 preceding siblings ...)
2016-09-16 14:10 ` [patch 16/32] greybus: loopback driver Greg KH
@ 2016-09-16 14:10 ` Greg KH
2016-10-07 13:49 ` Pavel Machek
2016-09-16 14:11 ` [patch 18/32] greybus: raw driver Greg KH
` (14 subsequent siblings)
30 siblings, 1 reply; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:10 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus power supply class protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/power_supply.c | 1141 +++++++++++++++++++++++++++++++++++++++++
1 file changed, 1141 insertions(+)
--- /dev/null
+++ b/drivers/greybus/power_supply.c
@@ -0,0 +1,1141 @@
+/*
+ * Power Supply driver for a Greybus module.
+ *
+ * Copyright 2014-2015 Google Inc.
+ * Copyright 2014-2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/power_supply.h>
+#include <linux/slab.h>
+
+#include "greybus.h"
+
+#define PROP_MAX 32
+
+struct gb_power_supply_prop {
+ enum power_supply_property prop;
+ u8 gb_prop;
+ int val;
+ int previous_val;
+ bool is_writeable;
+};
+
+struct gb_power_supply {
+ u8 id;
+ bool registered;
+ struct power_supply *psy;
+ struct power_supply_desc desc;
+ char name[64];
+ struct gb_power_supplies *supplies;
+ struct delayed_work work;
+ char *manufacturer;
+ char *model_name;
+ char *serial_number;
+ u8 type;
+ u8 properties_count;
+ u8 properties_count_str;
+ unsigned long last_update;
+ u8 cache_invalid;
+ unsigned int update_interval;
+ bool changed;
+ struct gb_power_supply_prop *props;
+ enum power_supply_property *props_raw;
+ bool pm_acquired;
+ struct mutex supply_lock;
+};
+
+struct gb_power_supplies {
+ struct gb_connection *connection;
+ u8 supplies_count;
+ struct gb_power_supply *supply;
+ struct mutex supplies_lock;
+};
+
+#define to_gb_power_supply(x) power_supply_get_drvdata(x)
+
+/*
+ * General power supply properties that could be absent from various reasons,
+ * like kernel versions or vendor specific versions
+ */
+#ifndef POWER_SUPPLY_PROP_VOLTAGE_BOOT
+ #define POWER_SUPPLY_PROP_VOLTAGE_BOOT -1
+#endif
+#ifndef POWER_SUPPLY_PROP_CURRENT_BOOT
+ #define POWER_SUPPLY_PROP_CURRENT_BOOT -1
+#endif
+#ifndef POWER_SUPPLY_PROP_CALIBRATE
+ #define POWER_SUPPLY_PROP_CALIBRATE -1
+#endif
+
+/* cache time in milliseconds, if cache_time is set to 0 cache is disable */
+static unsigned int cache_time = 1000;
+/*
+ * update interval initial and maximum value, between the two will
+ * back-off exponential
+ */
+static unsigned int update_interval_init = 1 * HZ;
+static unsigned int update_interval_max = 30 * HZ;
+
+struct gb_power_supply_changes {
+ enum power_supply_property prop;
+ u32 tolerance_change;
+ void (*prop_changed)(struct gb_power_supply *gbpsy,
+ struct gb_power_supply_prop *prop);
+};
+
+static void gb_power_supply_state_change(struct gb_power_supply *gbpsy,
+ struct gb_power_supply_prop *prop);
+
+static const struct gb_power_supply_changes psy_props_changes[] = {
+ { .prop = GB_POWER_SUPPLY_PROP_STATUS,
+ .tolerance_change = 0,
+ .prop_changed = gb_power_supply_state_change,
+ },
+ { .prop = GB_POWER_SUPPLY_PROP_TEMP,
+ .tolerance_change = 500,
+ .prop_changed = NULL,
+ },
+ { .prop = GB_POWER_SUPPLY_PROP_ONLINE,
+ .tolerance_change = 0,
+ .prop_changed = NULL,
+ },
+};
+
+static int get_psp_from_gb_prop(int gb_prop, enum power_supply_property *psp)
+{
+ int prop;
+
+ switch (gb_prop) {
+ case GB_POWER_SUPPLY_PROP_STATUS:
+ prop = POWER_SUPPLY_PROP_STATUS;
+ break;
+ case GB_POWER_SUPPLY_PROP_CHARGE_TYPE:
+ prop = POWER_SUPPLY_PROP_CHARGE_TYPE;
+ break;
+ case GB_POWER_SUPPLY_PROP_HEALTH:
+ prop = POWER_SUPPLY_PROP_HEALTH;
+ break;
+ case GB_POWER_SUPPLY_PROP_PRESENT:
+ prop = POWER_SUPPLY_PROP_PRESENT;
+ break;
+ case GB_POWER_SUPPLY_PROP_ONLINE:
+ prop = POWER_SUPPLY_PROP_ONLINE;
+ break;
+ case GB_POWER_SUPPLY_PROP_AUTHENTIC:
+ prop = POWER_SUPPLY_PROP_AUTHENTIC;
+ break;
+ case GB_POWER_SUPPLY_PROP_TECHNOLOGY:
+ prop = POWER_SUPPLY_PROP_TECHNOLOGY;
+ break;
+ case GB_POWER_SUPPLY_PROP_CYCLE_COUNT:
+ prop = POWER_SUPPLY_PROP_CYCLE_COUNT;
+ break;
+ case GB_POWER_SUPPLY_PROP_VOLTAGE_MAX:
+ prop = POWER_SUPPLY_PROP_VOLTAGE_MAX;
+ break;
+ case GB_POWER_SUPPLY_PROP_VOLTAGE_MIN:
+ prop = POWER_SUPPLY_PROP_VOLTAGE_MIN;
+ break;
+ case GB_POWER_SUPPLY_PROP_VOLTAGE_MAX_DESIGN:
+ prop = POWER_SUPPLY_PROP_VOLTAGE_MAX_DESIGN;
+ break;
+ case GB_POWER_SUPPLY_PROP_VOLTAGE_MIN_DESIGN:
+ prop = POWER_SUPPLY_PROP_VOLTAGE_MIN_DESIGN;
+ break;
+ case GB_POWER_SUPPLY_PROP_VOLTAGE_NOW:
+ prop = POWER_SUPPLY_PROP_VOLTAGE_NOW;
+ break;
+ case GB_POWER_SUPPLY_PROP_VOLTAGE_AVG:
+ prop = POWER_SUPPLY_PROP_VOLTAGE_AVG;
+ break;
+ case GB_POWER_SUPPLY_PROP_VOLTAGE_OCV:
+ prop = POWER_SUPPLY_PROP_VOLTAGE_OCV;
+ break;
+ case GB_POWER_SUPPLY_PROP_VOLTAGE_BOOT:
+ prop = POWER_SUPPLY_PROP_VOLTAGE_BOOT;
+ break;
+ case GB_POWER_SUPPLY_PROP_CURRENT_MAX:
+ prop = POWER_SUPPLY_PROP_CURRENT_MAX;
+ break;
+ case GB_POWER_SUPPLY_PROP_CURRENT_NOW:
+ prop = POWER_SUPPLY_PROP_CURRENT_NOW;
+ break;
+ case GB_POWER_SUPPLY_PROP_CURRENT_AVG:
+ prop = POWER_SUPPLY_PROP_CURRENT_AVG;
+ break;
+ case GB_POWER_SUPPLY_PROP_CURRENT_BOOT:
+ prop = POWER_SUPPLY_PROP_CURRENT_BOOT;
+ break;
+ case GB_POWER_SUPPLY_PROP_POWER_NOW:
+ prop = POWER_SUPPLY_PROP_POWER_NOW;
+ break;
+ case GB_POWER_SUPPLY_PROP_POWER_AVG:
+ prop = POWER_SUPPLY_PROP_POWER_AVG;
+ break;
+ case GB_POWER_SUPPLY_PROP_CHARGE_FULL_DESIGN:
+ prop = POWER_SUPPLY_PROP_CHARGE_FULL_DESIGN;
+ break;
+ case GB_POWER_SUPPLY_PROP_CHARGE_EMPTY_DESIGN:
+ prop = POWER_SUPPLY_PROP_CHARGE_EMPTY_DESIGN;
+ break;
+ case GB_POWER_SUPPLY_PROP_CHARGE_FULL:
+ prop = POWER_SUPPLY_PROP_CHARGE_FULL;
+ break;
+ case GB_POWER_SUPPLY_PROP_CHARGE_EMPTY:
+ prop = POWER_SUPPLY_PROP_CHARGE_EMPTY;
+ break;
+ case GB_POWER_SUPPLY_PROP_CHARGE_NOW:
+ prop = POWER_SUPPLY_PROP_CHARGE_NOW;
+ break;
+ case GB_POWER_SUPPLY_PROP_CHARGE_AVG:
+ prop = POWER_SUPPLY_PROP_CHARGE_AVG;
+ break;
+ case GB_POWER_SUPPLY_PROP_CHARGE_COUNTER:
+ prop = POWER_SUPPLY_PROP_CHARGE_COUNTER;
+ break;
+ case GB_POWER_SUPPLY_PROP_CONSTANT_CHARGE_CURRENT:
+ prop = POWER_SUPPLY_PROP_CONSTANT_CHARGE_CURRENT;
+ break;
+ case GB_POWER_SUPPLY_PROP_CONSTANT_CHARGE_CURRENT_MAX:
+ prop = POWER_SUPPLY_PROP_CONSTANT_CHARGE_CURRENT_MAX;
+ break;
+ case GB_POWER_SUPPLY_PROP_CONSTANT_CHARGE_VOLTAGE:
+ prop = POWER_SUPPLY_PROP_CONSTANT_CHARGE_VOLTAGE;
+ break;
+ case GB_POWER_SUPPLY_PROP_CONSTANT_CHARGE_VOLTAGE_MAX:
+ prop = POWER_SUPPLY_PROP_CONSTANT_CHARGE_VOLTAGE_MAX;
+ break;
+ case GB_POWER_SUPPLY_PROP_CHARGE_CONTROL_LIMIT:
+ prop = POWER_SUPPLY_PROP_CHARGE_CONTROL_LIMIT;
+ break;
+ case GB_POWER_SUPPLY_PROP_CHARGE_CONTROL_LIMIT_MAX:
+ prop = POWER_SUPPLY_PROP_CHARGE_CONTROL_LIMIT_MAX;
+ break;
+ case GB_POWER_SUPPLY_PROP_INPUT_CURRENT_LIMIT:
+ prop = POWER_SUPPLY_PROP_INPUT_CURRENT_LIMIT;
+ break;
+ case GB_POWER_SUPPLY_PROP_ENERGY_FULL_DESIGN:
+ prop = POWER_SUPPLY_PROP_ENERGY_FULL_DESIGN;
+ break;
+ case GB_POWER_SUPPLY_PROP_ENERGY_EMPTY_DESIGN:
+ prop = POWER_SUPPLY_PROP_ENERGY_EMPTY_DESIGN;
+ break;
+ case GB_POWER_SUPPLY_PROP_ENERGY_FULL:
+ prop = POWER_SUPPLY_PROP_ENERGY_FULL;
+ break;
+ case GB_POWER_SUPPLY_PROP_ENERGY_EMPTY:
+ prop = POWER_SUPPLY_PROP_ENERGY_EMPTY;
+ break;
+ case GB_POWER_SUPPLY_PROP_ENERGY_NOW:
+ prop = POWER_SUPPLY_PROP_ENERGY_NOW;
+ break;
+ case GB_POWER_SUPPLY_PROP_ENERGY_AVG:
+ prop = POWER_SUPPLY_PROP_ENERGY_AVG;
+ break;
+ case GB_POWER_SUPPLY_PROP_CAPACITY:
+ prop = POWER_SUPPLY_PROP_CAPACITY;
+ break;
+ case GB_POWER_SUPPLY_PROP_CAPACITY_ALERT_MIN:
+ prop = POWER_SUPPLY_PROP_CAPACITY_ALERT_MIN;
+ break;
+ case GB_POWER_SUPPLY_PROP_CAPACITY_ALERT_MAX:
+ prop = POWER_SUPPLY_PROP_CAPACITY_ALERT_MAX;
+ break;
+ case GB_POWER_SUPPLY_PROP_CAPACITY_LEVEL:
+ prop = POWER_SUPPLY_PROP_CAPACITY_LEVEL;
+ break;
+ case GB_POWER_SUPPLY_PROP_TEMP:
+ prop = POWER_SUPPLY_PROP_TEMP;
+ break;
+ case GB_POWER_SUPPLY_PROP_TEMP_MAX:
+ prop = POWER_SUPPLY_PROP_TEMP_MAX;
+ break;
+ case GB_POWER_SUPPLY_PROP_TEMP_MIN:
+ prop = POWER_SUPPLY_PROP_TEMP_MIN;
+ break;
+ case GB_POWER_SUPPLY_PROP_TEMP_ALERT_MIN:
+ prop = POWER_SUPPLY_PROP_TEMP_ALERT_MIN;
+ break;
+ case GB_POWER_SUPPLY_PROP_TEMP_ALERT_MAX:
+ prop = POWER_SUPPLY_PROP_TEMP_ALERT_MAX;
+ break;
+ case GB_POWER_SUPPLY_PROP_TEMP_AMBIENT:
+ prop = POWER_SUPPLY_PROP_TEMP_AMBIENT;
+ break;
+ case GB_POWER_SUPPLY_PROP_TEMP_AMBIENT_ALERT_MIN:
+ prop = POWER_SUPPLY_PROP_TEMP_AMBIENT_ALERT_MIN;
+ break;
+ case GB_POWER_SUPPLY_PROP_TEMP_AMBIENT_ALERT_MAX:
+ prop = POWER_SUPPLY_PROP_TEMP_AMBIENT_ALERT_MAX;
+ break;
+ case GB_POWER_SUPPLY_PROP_TIME_TO_EMPTY_NOW:
+ prop = POWER_SUPPLY_PROP_TIME_TO_EMPTY_NOW;
+ break;
+ case GB_POWER_SUPPLY_PROP_TIME_TO_EMPTY_AVG:
+ prop = POWER_SUPPLY_PROP_TIME_TO_EMPTY_AVG;
+ break;
+ case GB_POWER_SUPPLY_PROP_TIME_TO_FULL_NOW:
+ prop = POWER_SUPPLY_PROP_TIME_TO_FULL_NOW;
+ break;
+ case GB_POWER_SUPPLY_PROP_TIME_TO_FULL_AVG:
+ prop = POWER_SUPPLY_PROP_TIME_TO_FULL_AVG;
+ break;
+ case GB_POWER_SUPPLY_PROP_TYPE:
+ prop = POWER_SUPPLY_PROP_TYPE;
+ break;
+ case GB_POWER_SUPPLY_PROP_SCOPE:
+ prop = POWER_SUPPLY_PROP_SCOPE;
+ break;
+ case GB_POWER_SUPPLY_PROP_CHARGE_TERM_CURRENT:
+ prop = POWER_SUPPLY_PROP_CHARGE_TERM_CURRENT;
+ break;
+ case GB_POWER_SUPPLY_PROP_CALIBRATE:
+ prop = POWER_SUPPLY_PROP_CALIBRATE;
+ break;
+ default:
+ prop = -1;
+ break;
+ }
+
+ if (prop < 0)
+ return prop;
+
+ *psp = (enum power_supply_property)prop;
+
+ return 0;
+}
+
+static struct gb_connection *get_conn_from_psy(struct gb_power_supply *gbpsy)
+{
+ return gbpsy->supplies->connection;
+}
+
+static struct gb_power_supply_prop *get_psy_prop(struct gb_power_supply *gbpsy,
+ enum power_supply_property psp)
+{
+ int i;
+
+ for (i = 0; i < gbpsy->properties_count; i++)
+ if (gbpsy->props[i].prop == psp)
+ return &gbpsy->props[i];
+ return NULL;
+}
+
+static int is_psy_prop_writeable(struct gb_power_supply *gbpsy,
+ enum power_supply_property psp)
+{
+ struct gb_power_supply_prop *prop;
+
+ prop = get_psy_prop(gbpsy, psp);
+ if (!prop)
+ return -ENOENT;
+ return prop->is_writeable ? 1 : 0;
+}
+
+static int is_prop_valint(enum power_supply_property psp)
+{
+ return ((psp < POWER_SUPPLY_PROP_MODEL_NAME) ? 1 : 0);
+}
+
+static void next_interval(struct gb_power_supply *gbpsy)
+{
+ if (gbpsy->update_interval == update_interval_max)
+ return;
+
+ /* do some exponential back-off in the update interval */
+ gbpsy->update_interval *= 2;
+ if (gbpsy->update_interval > update_interval_max)
+ gbpsy->update_interval = update_interval_max;
+}
+
+static void __gb_power_supply_changed(struct gb_power_supply *gbpsy)
+{
+ power_supply_changed(gbpsy->psy);
+}
+
+static void gb_power_supply_state_change(struct gb_power_supply *gbpsy,
+ struct gb_power_supply_prop *prop)
+{
+ struct gb_connection *connection = get_conn_from_psy(gbpsy);
+ int ret;
+
+ /*
+ * Check gbpsy->pm_acquired to make sure only one pair of 'get_sync'
+ * and 'put_autosuspend' runtime pm call for state property change.
+ */
+ mutex_lock(&gbpsy->supply_lock);
+
+ if ((prop->val == GB_POWER_SUPPLY_STATUS_CHARGING) &&
+ !gbpsy->pm_acquired) {
+ ret = gb_pm_runtime_get_sync(connection->bundle);
+ if (ret)
+ dev_err(&connection->bundle->dev,
+ "Fail to set wake lock for charging state\n");
+ else
+ gbpsy->pm_acquired = true;
+ } else {
+ if (gbpsy->pm_acquired) {
+ ret = gb_pm_runtime_put_autosuspend(connection->bundle);
+ if (ret)
+ dev_err(&connection->bundle->dev,
+ "Fail to set wake unlock for none charging\n");
+ else
+ gbpsy->pm_acquired = false;
+ }
+ }
+
+ mutex_unlock(&gbpsy->supply_lock);
+}
+
+static void check_changed(struct gb_power_supply *gbpsy,
+ struct gb_power_supply_prop *prop)
+{
+ const struct gb_power_supply_changes *psyc;
+ int val = prop->val;
+ int prev_val = prop->previous_val;
+ bool changed = false;
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(psy_props_changes); i++) {
+ psyc = &psy_props_changes[i];
+ if (prop->prop == psyc->prop) {
+ if (!psyc->tolerance_change)
+ changed = true;
+ else if (val < prev_val &&
+ prev_val - val > psyc->tolerance_change)
+ changed = true;
+ else if (val > prev_val &&
+ val - prev_val > psyc->tolerance_change)
+ changed = true;
+
+ if (changed && psyc->prop_changed)
+ psyc->prop_changed(gbpsy, prop);
+
+ if (changed)
+ gbpsy->changed = true;
+ break;
+ }
+ }
+}
+
+static int total_props(struct gb_power_supply *gbpsy)
+{
+ /* this return the intval plus the strval properties */
+ return (gbpsy->properties_count + gbpsy->properties_count_str);
+}
+
+static void prop_append(struct gb_power_supply *gbpsy,
+ enum power_supply_property prop)
+{
+ enum power_supply_property *new_props_raw;
+
+ gbpsy->properties_count_str++;
+ new_props_raw = krealloc(gbpsy->props_raw, total_props(gbpsy) *
+ sizeof(enum power_supply_property),
+ GFP_KERNEL);
+ if (!new_props_raw)
+ return;
+ gbpsy->props_raw = new_props_raw;
+ gbpsy->props_raw[total_props(gbpsy) - 1] = prop;
+}
+
+static int __gb_power_supply_set_name(char *init_name, char *name, size_t len)
+{
+ unsigned int i = 0;
+ int ret = 0;
+ struct power_supply *psy;
+
+ if (!strlen(init_name))
+ init_name = "gb_power_supply";
+ strlcpy(name, init_name, len);
+
+ while ((ret < len) && (psy = power_supply_get_by_name(name))) {
+ power_supply_put(psy);
+
+ ret = snprintf(name, len, "%s_%u", init_name, ++i);
+ }
+ if (ret >= len)
+ return -ENOMEM;
+ return i;
+}
+
+static void _gb_power_supply_append_props(struct gb_power_supply *gbpsy)
+{
+ if (strlen(gbpsy->manufacturer))
+ prop_append(gbpsy, POWER_SUPPLY_PROP_MANUFACTURER);
+ if (strlen(gbpsy->model_name))
+ prop_append(gbpsy, POWER_SUPPLY_PROP_MODEL_NAME);
+ if (strlen(gbpsy->serial_number))
+ prop_append(gbpsy, POWER_SUPPLY_PROP_SERIAL_NUMBER);
+}
+
+static int gb_power_supply_description_get(struct gb_power_supply *gbpsy)
+{
+ struct gb_connection *connection = get_conn_from_psy(gbpsy);
+ struct gb_power_supply_get_description_request req;
+ struct gb_power_supply_get_description_response resp;
+ int ret;
+
+ req.psy_id = gbpsy->id;
+
+ ret = gb_operation_sync(connection,
+ GB_POWER_SUPPLY_TYPE_GET_DESCRIPTION,
+ &req, sizeof(req), &resp, sizeof(resp));
+ if (ret < 0)
+ return ret;
+
+ gbpsy->manufacturer = kstrndup(resp.manufacturer, PROP_MAX, GFP_KERNEL);
+ if (!gbpsy->manufacturer)
+ return -ENOMEM;
+ gbpsy->model_name = kstrndup(resp.model, PROP_MAX, GFP_KERNEL);
+ if (!gbpsy->model_name)
+ return -ENOMEM;
+ gbpsy->serial_number = kstrndup(resp.serial_number, PROP_MAX,
+ GFP_KERNEL);
+ if (!gbpsy->serial_number)
+ return -ENOMEM;
+
+ gbpsy->type = le16_to_cpu(resp.type);
+ gbpsy->properties_count = resp.properties_count;
+
+ return 0;
+}
+
+static int gb_power_supply_prop_descriptors_get(struct gb_power_supply *gbpsy)
+{
+ struct gb_connection *connection = get_conn_from_psy(gbpsy);
+ struct gb_power_supply_get_property_descriptors_request *req;
+ struct gb_power_supply_get_property_descriptors_response *resp;
+ struct gb_operation *op;
+ u8 props_count = gbpsy->properties_count;
+ enum power_supply_property psp;
+ int ret;
+ int i, r = 0;
+
+ if (props_count == 0)
+ return 0;
+
+ op = gb_operation_create(connection,
+ GB_POWER_SUPPLY_TYPE_GET_PROP_DESCRIPTORS,
+ sizeof(req), sizeof(*resp) + props_count *
+ sizeof(struct gb_power_supply_props_desc),
+ GFP_KERNEL);
+ if (!op)
+ return -ENOMEM;
+
+ req = op->request->payload;
+ req->psy_id = gbpsy->id;
+
+ ret = gb_operation_request_send_sync(op);
+ if (ret < 0)
+ goto out_put_operation;
+
+ resp = op->response->payload;
+
+ /* validate received properties */
+ for (i = 0; i < props_count; i++) {
+ ret = get_psp_from_gb_prop(resp->props[i].property, &psp);
+ if (ret < 0) {
+ dev_warn(&connection->bundle->dev,
+ "greybus property %u it is not supported by this kernel, dropped\n",
+ resp->props[i].property);
+ gbpsy->properties_count--;
+ }
+ }
+
+ gbpsy->props = kcalloc(gbpsy->properties_count, sizeof(*gbpsy->props),
+ GFP_KERNEL);
+ if (!gbpsy->props) {
+ ret = -ENOMEM;
+ goto out_put_operation;
+ }
+
+ gbpsy->props_raw = kcalloc(gbpsy->properties_count,
+ sizeof(*gbpsy->props_raw), GFP_KERNEL);
+ if (!gbpsy->props_raw) {
+ ret = -ENOMEM;
+ goto out_put_operation;
+ }
+
+ /* Store available properties, skip the ones we do not support */
+ for (i = 0; i < props_count; i++) {
+ ret = get_psp_from_gb_prop(resp->props[i].property, &psp);
+ if (ret < 0) {
+ r++;
+ continue;
+ }
+ gbpsy->props[i - r].prop = psp;
+ gbpsy->props[i - r].gb_prop = resp->props[i].property;
+ gbpsy->props_raw[i - r] = psp;
+ if (resp->props[i].is_writeable)
+ gbpsy->props[i - r].is_writeable = true;
+ }
+
+ /*
+ * now append the properties that we already got information in the
+ * get_description operation. (char * ones)
+ */
+ _gb_power_supply_append_props(gbpsy);
+
+ ret = 0;
+out_put_operation:
+ gb_operation_put(op);
+
+ return ret;
+}
+
+static int __gb_power_supply_property_update(struct gb_power_supply *gbpsy,
+ enum power_supply_property psp)
+{
+ struct gb_connection *connection = get_conn_from_psy(gbpsy);
+ struct gb_power_supply_prop *prop;
+ struct gb_power_supply_get_property_request req;
+ struct gb_power_supply_get_property_response resp;
+ int val;
+ int ret;
+
+ prop = get_psy_prop(gbpsy, psp);
+ if (!prop)
+ return -EINVAL;
+ req.psy_id = gbpsy->id;
+ req.property = prop->gb_prop;
+
+ ret = gb_operation_sync(connection, GB_POWER_SUPPLY_TYPE_GET_PROPERTY,
+ &req, sizeof(req), &resp, sizeof(resp));
+ if (ret < 0)
+ return ret;
+
+ val = le32_to_cpu(resp.prop_val);
+ if (val == prop->val)
+ return 0;
+
+ prop->previous_val = prop->val;
+ prop->val = val;
+
+ check_changed(gbpsy, prop);
+
+ return 0;
+}
+
+static int __gb_power_supply_property_get(struct gb_power_supply *gbpsy,
+ enum power_supply_property psp,
+ union power_supply_propval *val)
+{
+ struct gb_power_supply_prop *prop;
+
+ prop = get_psy_prop(gbpsy, psp);
+ if (!prop)
+ return -EINVAL;
+
+ val->intval = prop->val;
+ return 0;
+}
+
+static int __gb_power_supply_property_strval_get(struct gb_power_supply *gbpsy,
+ enum power_supply_property psp,
+ union power_supply_propval *val)
+{
+ switch (psp) {
+ case POWER_SUPPLY_PROP_MODEL_NAME:
+ val->strval = gbpsy->model_name;
+ break;
+ case POWER_SUPPLY_PROP_MANUFACTURER:
+ val->strval = gbpsy->manufacturer;
+ break;
+ case POWER_SUPPLY_PROP_SERIAL_NUMBER:
+ val->strval = gbpsy->serial_number;
+ break;
+ default:
+ break;
+ }
+
+ return 0;
+}
+
+static int _gb_power_supply_property_get(struct gb_power_supply *gbpsy,
+ enum power_supply_property psp,
+ union power_supply_propval *val)
+{
+ struct gb_connection *connection = get_conn_from_psy(gbpsy);
+ int ret;
+
+ /*
+ * Properties of type const char *, were already fetched on
+ * get_description operation and should be cached in gb
+ */
+ if (is_prop_valint(psp))
+ ret = __gb_power_supply_property_get(gbpsy, psp, val);
+ else
+ ret = __gb_power_supply_property_strval_get(gbpsy, psp, val);
+
+ if (ret < 0)
+ dev_err(&connection->bundle->dev, "get property %u\n", psp);
+
+ return 0;
+}
+
+static int is_cache_valid(struct gb_power_supply *gbpsy)
+{
+ /* check if cache is good enough or it has expired */
+ if (gbpsy->cache_invalid) {
+ gbpsy->cache_invalid = 0;
+ return 0;
+ }
+
+ if (gbpsy->last_update &&
+ time_is_after_jiffies(gbpsy->last_update +
+ msecs_to_jiffies(cache_time)))
+ return 1;
+
+ return 0;
+}
+
+static int gb_power_supply_status_get(struct gb_power_supply *gbpsy)
+{
+ struct gb_connection *connection = get_conn_from_psy(gbpsy);
+ int ret = 0;
+ int i;
+
+ if (is_cache_valid(gbpsy))
+ return 0;
+
+ ret = gb_pm_runtime_get_sync(connection->bundle);
+ if (ret)
+ return ret;
+
+ for (i = 0; i < gbpsy->properties_count; i++) {
+ ret = __gb_power_supply_property_update(gbpsy,
+ gbpsy->props[i].prop);
+ if (ret < 0)
+ break;
+ }
+
+ if (ret == 0)
+ gbpsy->last_update = jiffies;
+
+ gb_pm_runtime_put_autosuspend(connection->bundle);
+ return ret;
+}
+
+static void gb_power_supply_status_update(struct gb_power_supply *gbpsy)
+{
+ /* check if there a change that need to be reported */
+ gb_power_supply_status_get(gbpsy);
+
+ if (!gbpsy->changed)
+ return;
+
+ gbpsy->update_interval = update_interval_init;
+ __gb_power_supply_changed(gbpsy);
+ gbpsy->changed = false;
+}
+
+static void gb_power_supply_work(struct work_struct *work)
+{
+ struct gb_power_supply *gbpsy = container_of(work,
+ struct gb_power_supply,
+ work.work);
+
+ /*
+ * if the poll interval is not set, disable polling, this is helpful
+ * specially at unregister time.
+ */
+ if (!gbpsy->update_interval)
+ return;
+
+ gb_power_supply_status_update(gbpsy);
+ next_interval(gbpsy);
+ schedule_delayed_work(&gbpsy->work, gbpsy->update_interval);
+}
+
+static int get_property(struct power_supply *b,
+ enum power_supply_property psp,
+ union power_supply_propval *val)
+{
+ struct gb_power_supply *gbpsy = to_gb_power_supply(b);
+
+ gb_power_supply_status_get(gbpsy);
+
+ return _gb_power_supply_property_get(gbpsy, psp, val);
+}
+
+static int gb_power_supply_property_set(struct gb_power_supply *gbpsy,
+ enum power_supply_property psp,
+ int val)
+{
+ struct gb_connection *connection = get_conn_from_psy(gbpsy);
+ struct gb_power_supply_prop *prop;
+ struct gb_power_supply_set_property_request req;
+ int ret;
+
+ ret = gb_pm_runtime_get_sync(connection->bundle);
+ if (ret)
+ return ret;
+
+ prop = get_psy_prop(gbpsy, psp);
+ if (!prop) {
+ ret = -EINVAL;
+ goto out;
+ }
+
+ req.psy_id = gbpsy->id;
+ req.property = prop->gb_prop;
+ req.prop_val = cpu_to_le32((s32)val);
+
+ ret = gb_operation_sync(connection, GB_POWER_SUPPLY_TYPE_SET_PROPERTY,
+ &req, sizeof(req), NULL, 0);
+ if (ret < 0)
+ goto out;
+
+ /* cache immediately the new value */
+ prop->val = val;
+
+out:
+ gb_pm_runtime_put_autosuspend(connection->bundle);
+ return ret;
+}
+
+static int set_property(struct power_supply *b,
+ enum power_supply_property psp,
+ const union power_supply_propval *val)
+{
+ struct gb_power_supply *gbpsy = to_gb_power_supply(b);
+
+ return gb_power_supply_property_set(gbpsy, psp, val->intval);
+}
+
+static int property_is_writeable(struct power_supply *b,
+ enum power_supply_property psp)
+{
+ struct gb_power_supply *gbpsy = to_gb_power_supply(b);
+
+ return is_psy_prop_writeable(gbpsy, psp);
+}
+
+static int gb_power_supply_register(struct gb_power_supply *gbpsy)
+{
+ struct gb_connection *connection = get_conn_from_psy(gbpsy);
+ struct power_supply_config cfg = {};
+
+ cfg.drv_data = gbpsy;
+
+ gbpsy->desc.name = gbpsy->name;
+ gbpsy->desc.type = gbpsy->type;
+ gbpsy->desc.properties = gbpsy->props_raw;
+ gbpsy->desc.num_properties = total_props(gbpsy);
+ gbpsy->desc.get_property = get_property;
+ gbpsy->desc.set_property = set_property;
+ gbpsy->desc.property_is_writeable = property_is_writeable;
+
+ gbpsy->psy = power_supply_register(&connection->bundle->dev,
+ &gbpsy->desc, &cfg);
+ return PTR_ERR_OR_ZERO(gbpsy->psy);
+}
+
+static void _gb_power_supply_free(struct gb_power_supply *gbpsy)
+{
+ kfree(gbpsy->serial_number);
+ kfree(gbpsy->model_name);
+ kfree(gbpsy->manufacturer);
+ kfree(gbpsy->props_raw);
+ kfree(gbpsy->props);
+}
+
+static void _gb_power_supply_release(struct gb_power_supply *gbpsy)
+{
+ gbpsy->update_interval = 0;
+
+ cancel_delayed_work_sync(&gbpsy->work);
+
+ if (gbpsy->registered)
+ power_supply_unregister(gbpsy->psy);
+
+ _gb_power_supply_free(gbpsy);
+}
+
+static void _gb_power_supplies_release(struct gb_power_supplies *supplies)
+{
+ int i;
+
+ if (!supplies->supply)
+ return;
+
+ mutex_lock(&supplies->supplies_lock);
+ for (i = 0; i < supplies->supplies_count; i++)
+ _gb_power_supply_release(&supplies->supply[i]);
+ kfree(supplies->supply);
+ mutex_unlock(&supplies->supplies_lock);
+ kfree(supplies);
+}
+
+static int gb_power_supplies_get_count(struct gb_power_supplies *supplies)
+{
+ struct gb_power_supply_get_supplies_response resp;
+ int ret;
+
+ ret = gb_operation_sync(supplies->connection,
+ GB_POWER_SUPPLY_TYPE_GET_SUPPLIES,
+ NULL, 0, &resp, sizeof(resp));
+ if (ret < 0)
+ return ret;
+
+ if (!resp.supplies_count)
+ return -EINVAL;
+
+ supplies->supplies_count = resp.supplies_count;
+
+ return ret;
+}
+
+static int gb_power_supply_config(struct gb_power_supplies *supplies, int id)
+{
+ struct gb_power_supply *gbpsy = &supplies->supply[id];
+ int ret;
+
+ gbpsy->supplies = supplies;
+ gbpsy->id = id;
+
+ ret = gb_power_supply_description_get(gbpsy);
+ if (ret < 0)
+ return ret;
+
+ return gb_power_supply_prop_descriptors_get(gbpsy);
+}
+
+static int gb_power_supply_enable(struct gb_power_supply *gbpsy)
+{
+ int ret;
+
+ /* guarantee that we have an unique name, before register */
+ ret = __gb_power_supply_set_name(gbpsy->model_name, gbpsy->name,
+ sizeof(gbpsy->name));
+ if (ret < 0)
+ return ret;
+
+ mutex_init(&gbpsy->supply_lock);
+
+ ret = gb_power_supply_register(gbpsy);
+ if (ret < 0)
+ return ret;
+
+ gbpsy->update_interval = update_interval_init;
+ INIT_DELAYED_WORK(&gbpsy->work, gb_power_supply_work);
+ schedule_delayed_work(&gbpsy->work, 0);
+
+ /* everything went fine, mark it for release code to know */
+ gbpsy->registered = true;
+
+ return 0;
+}
+
+static int gb_power_supplies_setup(struct gb_power_supplies *supplies)
+{
+ struct gb_connection *connection = supplies->connection;
+ int ret;
+ int i;
+
+ mutex_lock(&supplies->supplies_lock);
+
+ ret = gb_power_supplies_get_count(supplies);
+ if (ret < 0)
+ goto out;
+
+ supplies->supply = kzalloc(supplies->supplies_count *
+ sizeof(struct gb_power_supply),
+ GFP_KERNEL);
+
+ if (!supplies->supply) {
+ ret = -ENOMEM;
+ goto out;
+ }
+
+ for (i = 0; i < supplies->supplies_count; i++) {
+ ret = gb_power_supply_config(supplies, i);
+ if (ret < 0) {
+ dev_err(&connection->bundle->dev,
+ "Fail to configure supplies devices\n");
+ goto out;
+ }
+ }
+out:
+ mutex_unlock(&supplies->supplies_lock);
+ return ret;
+}
+
+static int gb_power_supplies_register(struct gb_power_supplies *supplies)
+{
+ struct gb_connection *connection = supplies->connection;
+ int ret = 0;
+ int i;
+
+ mutex_lock(&supplies->supplies_lock);
+
+ for (i = 0; i < supplies->supplies_count; i++) {
+ ret = gb_power_supply_enable(&supplies->supply[i]);
+ if (ret < 0) {
+ dev_err(&connection->bundle->dev,
+ "Fail to enable supplies devices\n");
+ break;
+ }
+ }
+
+ mutex_unlock(&supplies->supplies_lock);
+ return ret;
+}
+
+static int gb_supplies_request_handler(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct gb_power_supplies *supplies = gb_connection_get_data(connection);
+ struct gb_power_supply *gbpsy;
+ struct gb_message *request;
+ struct gb_power_supply_event_request *payload;
+ u8 psy_id;
+ u8 event;
+ int ret = 0;
+
+ if (op->type != GB_POWER_SUPPLY_TYPE_EVENT) {
+ dev_err(&connection->bundle->dev,
+ "Unsupported unsolicited event: %u\n", op->type);
+ return -EINVAL;
+ }
+
+ request = op->request;
+
+ if (request->payload_size < sizeof(*payload)) {
+ dev_err(&connection->bundle->dev,
+ "Wrong event size received (%zu < %zu)\n",
+ request->payload_size, sizeof(*payload));
+ return -EINVAL;
+ }
+
+ payload = request->payload;
+ psy_id = payload->psy_id;
+ mutex_lock(&supplies->supplies_lock);
+ if (psy_id >= supplies->supplies_count ||
+ !supplies->supply[psy_id].registered) {
+ dev_err(&connection->bundle->dev,
+ "Event received for unconfigured power_supply id: %d\n",
+ psy_id);
+ ret = -EINVAL;
+ goto out_unlock;
+ }
+
+ event = payload->event;
+ /*
+ * we will only handle events after setup is done and before release is
+ * running. For that just check update_interval.
+ */
+ gbpsy = &supplies->supply[psy_id];
+ if (!gbpsy->update_interval) {
+ ret = -ESHUTDOWN;
+ goto out_unlock;
+ }
+
+ if (event & GB_POWER_SUPPLY_UPDATE) {
+ /*
+ * we need to make sure we invalidate cache, if not no new
+ * values for the properties will be fetch and the all propose
+ * of this event is missed
+ */
+ gbpsy->cache_invalid = 1;
+ gb_power_supply_status_update(gbpsy);
+ }
+
+out_unlock:
+ mutex_unlock(&supplies->supplies_lock);
+ return ret;
+}
+
+static int gb_power_supply_probe(struct gb_bundle *bundle,
+ const struct greybus_bundle_id *id)
+{
+ struct greybus_descriptor_cport *cport_desc;
+ struct gb_connection *connection;
+ struct gb_power_supplies *supplies;
+ int ret;
+
+ if (bundle->num_cports != 1)
+ return -ENODEV;
+
+ cport_desc = &bundle->cport_desc[0];
+ if (cport_desc->protocol_id != GREYBUS_PROTOCOL_POWER_SUPPLY)
+ return -ENODEV;
+
+ supplies = kzalloc(sizeof(*supplies), GFP_KERNEL);
+ if (!supplies)
+ return -ENOMEM;
+
+ connection = gb_connection_create(bundle, le16_to_cpu(cport_desc->id),
+ gb_supplies_request_handler);
+ if (IS_ERR(connection)) {
+ ret = PTR_ERR(connection);
+ goto out;
+ }
+
+ supplies->connection = connection;
+ gb_connection_set_data(connection, supplies);
+
+ mutex_init(&supplies->supplies_lock);
+
+ greybus_set_drvdata(bundle, supplies);
+
+ /* We aren't ready to receive an incoming request yet */
+ ret = gb_connection_enable_tx(connection);
+ if (ret)
+ goto error_connection_destroy;
+
+ ret = gb_power_supplies_setup(supplies);
+ if (ret < 0)
+ goto error_connection_disable;
+
+ /* We are ready to receive an incoming request now, enable RX as well */
+ ret = gb_connection_enable(connection);
+ if (ret)
+ goto error_connection_disable;
+
+ ret = gb_power_supplies_register(supplies);
+ if (ret < 0)
+ goto error_connection_disable;
+
+ gb_pm_runtime_put_autosuspend(bundle);
+ return 0;
+
+error_connection_disable:
+ gb_connection_disable(connection);
+error_connection_destroy:
+ gb_connection_destroy(connection);
+out:
+ _gb_power_supplies_release(supplies);
+ return ret;
+}
+
+static void gb_power_supply_disconnect(struct gb_bundle *bundle)
+{
+ struct gb_power_supplies *supplies = greybus_get_drvdata(bundle);
+
+ gb_connection_disable(supplies->connection);
+ gb_connection_destroy(supplies->connection);
+
+ _gb_power_supplies_release(supplies);
+}
+
+static const struct greybus_bundle_id gb_power_supply_id_table[] = {
+ { GREYBUS_DEVICE_CLASS(GREYBUS_CLASS_POWER_SUPPLY) },
+ { }
+};
+MODULE_DEVICE_TABLE(greybus, gb_power_supply_id_table);
+
+static struct greybus_driver gb_power_supply_driver = {
+ .name = "power_supply",
+ .probe = gb_power_supply_probe,
+ .disconnect = gb_power_supply_disconnect,
+ .id_table = gb_power_supply_id_table,
+};
+module_greybus_driver(gb_power_supply_driver);
+
+MODULE_LICENSE("GPL v2");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 18/32] greybus: raw driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (15 preceding siblings ...)
2016-09-16 14:10 ` [patch 17/32] greybus: power supply driver Greg KH
@ 2016-09-16 14:11 ` Greg KH
2016-09-16 14:11 ` [patch 19/32] greybus: timesync driver Greg KH
` (13 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:11 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus raw protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/raw.c | 381 ++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 381 insertions(+)
--- /dev/null
+++ b/drivers/greybus/raw.c
@@ -0,0 +1,381 @@
+/*
+ * Greybus driver for the Raw protocol
+ *
+ * Copyright 2015 Google Inc.
+ * Copyright 2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/sizes.h>
+#include <linux/cdev.h>
+#include <linux/fs.h>
+#include <linux/idr.h>
+#include <linux/uaccess.h>
+
+#include "greybus.h"
+
+struct gb_raw {
+ struct gb_connection *connection;
+
+ struct list_head list;
+ int list_data;
+ struct mutex list_lock;
+ dev_t dev;
+ struct cdev cdev;
+ struct device *device;
+};
+
+struct raw_data {
+ struct list_head entry;
+ u32 len;
+ u8 data[0];
+};
+
+static struct class *raw_class;
+static int raw_major;
+static const struct file_operations raw_fops;
+static DEFINE_IDA(minors);
+
+/* Number of minor devices this driver supports */
+#define NUM_MINORS 256
+
+/* Maximum size of any one send data buffer we support */
+#define MAX_PACKET_SIZE (PAGE_SIZE * 2)
+
+/*
+ * Maximum size of the data in the receive buffer we allow before we start to
+ * drop messages on the floor
+ */
+#define MAX_DATA_SIZE (MAX_PACKET_SIZE * 8)
+
+/*
+ * Add the raw data message to the list of received messages.
+ */
+static int receive_data(struct gb_raw *raw, u32 len, u8 *data)
+{
+ struct raw_data *raw_data;
+ struct device *dev = &raw->connection->bundle->dev;
+ int retval = 0;
+
+ if (len > MAX_PACKET_SIZE) {
+ dev_err(dev, "Too big of a data packet, rejected\n");
+ return -EINVAL;
+ }
+
+ mutex_lock(&raw->list_lock);
+ if ((raw->list_data + len) > MAX_DATA_SIZE) {
+ dev_err(dev, "Too much data in receive buffer, now dropping packets\n");
+ retval = -EINVAL;
+ goto exit;
+ }
+
+ raw_data = kmalloc(sizeof(*raw_data) + len, GFP_KERNEL);
+ if (!raw_data) {
+ retval = -ENOMEM;
+ goto exit;
+ }
+
+ raw->list_data += len;
+ raw_data->len = len;
+ memcpy(&raw_data->data[0], data, len);
+
+ list_add_tail(&raw_data->entry, &raw->list);
+exit:
+ mutex_unlock(&raw->list_lock);
+ return retval;
+}
+
+static int gb_raw_request_handler(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct device *dev = &connection->bundle->dev;
+ struct gb_raw *raw = greybus_get_drvdata(connection->bundle);
+ struct gb_raw_send_request *receive;
+ u32 len;
+
+ if (op->type != GB_RAW_TYPE_SEND) {
+ dev_err(dev, "unknown request type 0x%02x\n", op->type);
+ return -EINVAL;
+ }
+
+ /* Verify size of payload */
+ if (op->request->payload_size < sizeof(*receive)) {
+ dev_err(dev, "raw receive request too small (%zu < %zu)\n",
+ op->request->payload_size, sizeof(*receive));
+ return -EINVAL;
+ }
+ receive = op->request->payload;
+ len = le32_to_cpu(receive->len);
+ if (len != (int)(op->request->payload_size - sizeof(__le32))) {
+ dev_err(dev, "raw receive request wrong size %d vs %d\n", len,
+ (int)(op->request->payload_size - sizeof(__le32)));
+ return -EINVAL;
+ }
+ if (len == 0) {
+ dev_err(dev, "raw receive request of 0 bytes?\n");
+ return -EINVAL;
+ }
+
+ return receive_data(raw, len, receive->data);
+}
+
+static int gb_raw_send(struct gb_raw *raw, u32 len, const char __user *data)
+{
+ struct gb_connection *connection = raw->connection;
+ struct gb_raw_send_request *request;
+ int retval;
+
+ request = kmalloc(len + sizeof(*request), GFP_KERNEL);
+ if (!request)
+ return -ENOMEM;
+
+ if (copy_from_user(&request->data[0], data, len)) {
+ kfree(request);
+ return -EFAULT;
+ }
+
+ request->len = cpu_to_le32(len);
+
+ retval = gb_operation_sync(connection, GB_RAW_TYPE_SEND,
+ request, len + sizeof(*request),
+ NULL, 0);
+
+ kfree(request);
+ return retval;
+}
+
+static int gb_raw_probe(struct gb_bundle *bundle,
+ const struct greybus_bundle_id *id)
+{
+ struct greybus_descriptor_cport *cport_desc;
+ struct gb_connection *connection;
+ struct gb_raw *raw;
+ int retval;
+ int minor;
+
+ if (bundle->num_cports != 1)
+ return -ENODEV;
+
+ cport_desc = &bundle->cport_desc[0];
+ if (cport_desc->protocol_id != GREYBUS_PROTOCOL_RAW)
+ return -ENODEV;
+
+ raw = kzalloc(sizeof(*raw), GFP_KERNEL);
+ if (!raw)
+ return -ENOMEM;
+
+ connection = gb_connection_create(bundle, le16_to_cpu(cport_desc->id),
+ gb_raw_request_handler);
+ if (IS_ERR(connection)) {
+ retval = PTR_ERR(connection);
+ goto error_free;
+ }
+
+ INIT_LIST_HEAD(&raw->list);
+ mutex_init(&raw->list_lock);
+
+ raw->connection = connection;
+ greybus_set_drvdata(bundle, raw);
+
+ minor = ida_simple_get(&minors, 0, 0, GFP_KERNEL);
+ if (minor < 0) {
+ retval = minor;
+ goto error_connection_destroy;
+ }
+
+ raw->dev = MKDEV(raw_major, minor);
+ cdev_init(&raw->cdev, &raw_fops);
+
+ retval = gb_connection_enable(connection);
+ if (retval)
+ goto error_remove_ida;
+
+ retval = cdev_add(&raw->cdev, raw->dev, 1);
+ if (retval)
+ goto error_connection_disable;
+
+ raw->device = device_create(raw_class, &connection->bundle->dev,
+ raw->dev, raw, "gb!raw%d", minor);
+ if (IS_ERR(raw->device)) {
+ retval = PTR_ERR(raw->device);
+ goto error_del_cdev;
+ }
+
+ return 0;
+
+error_del_cdev:
+ cdev_del(&raw->cdev);
+
+error_connection_disable:
+ gb_connection_disable(connection);
+
+error_remove_ida:
+ ida_simple_remove(&minors, minor);
+
+error_connection_destroy:
+ gb_connection_destroy(connection);
+
+error_free:
+ kfree(raw);
+ return retval;
+}
+
+static void gb_raw_disconnect(struct gb_bundle *bundle)
+{
+ struct gb_raw *raw = greybus_get_drvdata(bundle);
+ struct gb_connection *connection = raw->connection;
+ struct raw_data *raw_data;
+ struct raw_data *temp;
+
+ // FIXME - handle removing a connection when the char device node is open.
+ device_destroy(raw_class, raw->dev);
+ cdev_del(&raw->cdev);
+ gb_connection_disable(connection);
+ ida_simple_remove(&minors, MINOR(raw->dev));
+ gb_connection_destroy(connection);
+
+ mutex_lock(&raw->list_lock);
+ list_for_each_entry_safe(raw_data, temp, &raw->list, entry) {
+ list_del(&raw_data->entry);
+ kfree(raw_data);
+ }
+ mutex_unlock(&raw->list_lock);
+
+ kfree(raw);
+}
+
+/*
+ * Character device node interfaces.
+ *
+ * Note, we are using read/write to only allow a single read/write per message.
+ * This means for read(), you have to provide a big enough buffer for the full
+ * message to be copied into. If the buffer isn't big enough, the read() will
+ * fail with -ENOSPC.
+ */
+
+static int raw_open(struct inode *inode, struct file *file)
+{
+ struct cdev *cdev = inode->i_cdev;
+ struct gb_raw *raw = container_of(cdev, struct gb_raw, cdev);
+
+ file->private_data = raw;
+ return 0;
+}
+
+static ssize_t raw_write(struct file *file, const char __user *buf,
+ size_t count, loff_t *ppos)
+{
+ struct gb_raw *raw = file->private_data;
+ int retval;
+
+ if (!count)
+ return 0;
+
+ if (count > MAX_PACKET_SIZE)
+ return -E2BIG;
+
+ retval = gb_raw_send(raw, count, buf);
+ if (retval)
+ return retval;
+
+ return count;
+}
+
+static ssize_t raw_read(struct file *file, char __user *buf, size_t count,
+ loff_t *ppos)
+{
+ struct gb_raw *raw = file->private_data;
+ int retval = 0;
+ struct raw_data *raw_data;
+
+ mutex_lock(&raw->list_lock);
+ if (list_empty(&raw->list))
+ goto exit;
+
+ raw_data = list_first_entry(&raw->list, struct raw_data, entry);
+ if (raw_data->len > count) {
+ retval = -ENOSPC;
+ goto exit;
+ }
+
+ if (copy_to_user(buf, &raw_data->data[0], raw_data->len)) {
+ retval = -EFAULT;
+ goto exit;
+ }
+
+ list_del(&raw_data->entry);
+ raw->list_data -= raw_data->len;
+ retval = raw_data->len;
+ kfree(raw_data);
+
+exit:
+ mutex_unlock(&raw->list_lock);
+ return retval;
+}
+
+static const struct file_operations raw_fops = {
+ .owner = THIS_MODULE,
+ .write = raw_write,
+ .read = raw_read,
+ .open = raw_open,
+ .llseek = noop_llseek,
+};
+
+static const struct greybus_bundle_id gb_raw_id_table[] = {
+ { GREYBUS_DEVICE_CLASS(GREYBUS_CLASS_RAW) },
+ { }
+};
+MODULE_DEVICE_TABLE(greybus, gb_raw_id_table);
+
+static struct greybus_driver gb_raw_driver = {
+ .name = "raw",
+ .probe = gb_raw_probe,
+ .disconnect = gb_raw_disconnect,
+ .id_table = gb_raw_id_table,
+};
+
+static int raw_init(void)
+{
+ dev_t dev;
+ int retval;
+
+ raw_class = class_create(THIS_MODULE, "gb_raw");
+ if (IS_ERR(raw_class)) {
+ retval = PTR_ERR(raw_class);
+ goto error_class;
+ }
+
+ retval = alloc_chrdev_region(&dev, 0, NUM_MINORS, "gb_raw");
+ if (retval < 0)
+ goto error_chrdev;
+
+ raw_major = MAJOR(dev);
+
+ retval = greybus_register(&gb_raw_driver);
+ if (retval)
+ goto error_gb;
+
+ return 0;
+
+error_gb:
+ unregister_chrdev_region(dev, NUM_MINORS);
+error_chrdev:
+ class_destroy(raw_class);
+error_class:
+ return retval;
+}
+module_init(raw_init);
+
+static void __exit raw_exit(void)
+{
+ greybus_deregister(&gb_raw_driver);
+ unregister_chrdev_region(MKDEV(raw_major, 0), NUM_MINORS);
+ class_destroy(raw_class);
+ ida_destroy(&minors);
+}
+module_exit(raw_exit);
+
+MODULE_LICENSE("GPL v2");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 19/32] greybus: timesync driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (16 preceding siblings ...)
2016-09-16 14:11 ` [patch 18/32] greybus: raw driver Greg KH
@ 2016-09-16 14:11 ` Greg KH
2016-09-16 14:11 ` [patch 20/32] greybus: vibrator driver Greg KH
` (12 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:11 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This code implements the Greybus timesync logic. It has the ability to
sync a time signal across all Unipro devices allowing them to stay in
sync properly.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/timesync.c | 1357 ++++++++++++++++++++++++++++++++++++
drivers/greybus/timesync.h | 45 +
drivers/greybus/timesync_platform.c | 77 ++
3 files changed, 1479 insertions(+)
--- /dev/null
+++ b/drivers/greybus/timesync.c
@@ -0,0 +1,1357 @@
+/*
+ * TimeSync API driver.
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+#include <linux/debugfs.h>
+#include <linux/hrtimer.h>
+#include "greybus.h"
+#include "timesync.h"
+#include "greybus_trace.h"
+
+/*
+ * Minimum inter-strobe value of one millisecond is chosen because it
+ * just-about fits the common definition of a jiffy.
+ *
+ * Maximum value OTOH is constrained by the number of bits the SVC can fit
+ * into a 16 bit up-counter. The SVC configures the timer in microseconds
+ * so the maximum allowable value is 65535 microseconds. We clip that value
+ * to 10000 microseconds for the sake of using nice round base 10 numbers
+ * and since right-now there's no imaginable use-case requiring anything
+ * other than a one millisecond inter-strobe time, let alone something
+ * higher than ten milliseconds.
+ */
+#define GB_TIMESYNC_STROBE_DELAY_US 1000
+#define GB_TIMESYNC_DEFAULT_OFFSET_US 1000
+
+/* Work queue timers long, short and SVC strobe timeout */
+#define GB_TIMESYNC_DELAYED_WORK_LONG msecs_to_jiffies(10)
+#define GB_TIMESYNC_DELAYED_WORK_SHORT msecs_to_jiffies(1)
+#define GB_TIMESYNC_MAX_WAIT_SVC msecs_to_jiffies(5000)
+#define GB_TIMESYNC_KTIME_UPDATE msecs_to_jiffies(1000)
+#define GB_TIMESYNC_MAX_KTIME_CONVERSION 15
+
+/* Maximum number of times we'll retry a failed synchronous sync */
+#define GB_TIMESYNC_MAX_RETRIES 5
+
+/* Reported nanoseconds/femtoseconds per clock */
+static u64 gb_timesync_ns_per_clock;
+static u64 gb_timesync_fs_per_clock;
+
+/* Maximum difference we will accept converting FrameTime to ktime */
+static u32 gb_timesync_max_ktime_diff;
+
+/* Reported clock rate */
+static unsigned long gb_timesync_clock_rate;
+
+/* Workqueue */
+static void gb_timesync_worker(struct work_struct *work);
+
+/* List of SVCs with one FrameTime per SVC */
+static LIST_HEAD(gb_timesync_svc_list);
+
+/* Synchronize parallel contexts accessing a valid timesync_svc pointer */
+static DEFINE_MUTEX(gb_timesync_svc_list_mutex);
+
+/* Structure to convert from FrameTime to timespec/ktime */
+struct gb_timesync_frame_time_data {
+ u64 frame_time;
+ struct timespec ts;
+};
+
+struct gb_timesync_svc {
+ struct list_head list;
+ struct list_head interface_list;
+ struct gb_svc *svc;
+ struct gb_timesync_host_device *timesync_hd;
+
+ spinlock_t spinlock; /* Per SVC spinlock to sync with ISR */
+ struct mutex mutex; /* Per SVC mutex for regular synchronization */
+
+ struct dentry *frame_time_dentry;
+ struct dentry *frame_ktime_dentry;
+ struct workqueue_struct *work_queue;
+ wait_queue_head_t wait_queue;
+ struct delayed_work delayed_work;
+ struct timer_list ktime_timer;
+
+ /* The current local FrameTime */
+ u64 frame_time_offset;
+ struct gb_timesync_frame_time_data strobe_data[GB_TIMESYNC_MAX_STROBES];
+ struct gb_timesync_frame_time_data ktime_data;
+
+ /* The SVC FrameTime and relative AP FrameTime @ last TIMESYNC_PING */
+ u64 svc_ping_frame_time;
+ u64 ap_ping_frame_time;
+
+ /* Transitory settings */
+ u32 strobe_mask;
+ bool offset_down;
+ bool print_ping;
+ bool capture_ping;
+ int strobe;
+
+ /* Current state */
+ int state;
+};
+
+struct gb_timesync_host_device {
+ struct list_head list;
+ struct gb_host_device *hd;
+ u64 ping_frame_time;
+};
+
+struct gb_timesync_interface {
+ struct list_head list;
+ struct gb_interface *interface;
+ u64 ping_frame_time;
+};
+
+enum gb_timesync_state {
+ GB_TIMESYNC_STATE_INVALID = 0,
+ GB_TIMESYNC_STATE_INACTIVE = 1,
+ GB_TIMESYNC_STATE_INIT = 2,
+ GB_TIMESYNC_STATE_WAIT_SVC = 3,
+ GB_TIMESYNC_STATE_AUTHORITATIVE = 4,
+ GB_TIMESYNC_STATE_PING = 5,
+ GB_TIMESYNC_STATE_ACTIVE = 6,
+};
+
+static void gb_timesync_ktime_timer_fn(unsigned long data);
+
+static u64 gb_timesync_adjust_count(struct gb_timesync_svc *timesync_svc,
+ u64 counts)
+{
+ if (timesync_svc->offset_down)
+ return counts - timesync_svc->frame_time_offset;
+ else
+ return counts + timesync_svc->frame_time_offset;
+}
+
+/*
+ * This function provides the authoritative FrameTime to a calling function. It
+ * is designed to be lockless and should remain that way the caller is assumed
+ * to be state-aware.
+ */
+static u64 __gb_timesync_get_frame_time(struct gb_timesync_svc *timesync_svc)
+{
+ u64 clocks = gb_timesync_platform_get_counter();
+
+ return gb_timesync_adjust_count(timesync_svc, clocks);
+}
+
+static void gb_timesync_schedule_svc_timeout(struct gb_timesync_svc
+ *timesync_svc)
+{
+ queue_delayed_work(timesync_svc->work_queue,
+ ×ync_svc->delayed_work,
+ GB_TIMESYNC_MAX_WAIT_SVC);
+}
+
+static void gb_timesync_set_state(struct gb_timesync_svc *timesync_svc,
+ int state)
+{
+ switch (state) {
+ case GB_TIMESYNC_STATE_INVALID:
+ timesync_svc->state = state;
+ wake_up(×ync_svc->wait_queue);
+ break;
+ case GB_TIMESYNC_STATE_INACTIVE:
+ timesync_svc->state = state;
+ wake_up(×ync_svc->wait_queue);
+ break;
+ case GB_TIMESYNC_STATE_INIT:
+ if (timesync_svc->state != GB_TIMESYNC_STATE_INVALID) {
+ timesync_svc->strobe = 0;
+ timesync_svc->frame_time_offset = 0;
+ timesync_svc->state = state;
+ cancel_delayed_work(×ync_svc->delayed_work);
+ queue_delayed_work(timesync_svc->work_queue,
+ ×ync_svc->delayed_work,
+ GB_TIMESYNC_DELAYED_WORK_LONG);
+ }
+ break;
+ case GB_TIMESYNC_STATE_WAIT_SVC:
+ if (timesync_svc->state == GB_TIMESYNC_STATE_INIT)
+ timesync_svc->state = state;
+ break;
+ case GB_TIMESYNC_STATE_AUTHORITATIVE:
+ if (timesync_svc->state == GB_TIMESYNC_STATE_WAIT_SVC) {
+ timesync_svc->state = state;
+ cancel_delayed_work(×ync_svc->delayed_work);
+ queue_delayed_work(timesync_svc->work_queue,
+ ×ync_svc->delayed_work, 0);
+ }
+ break;
+ case GB_TIMESYNC_STATE_PING:
+ if (timesync_svc->state == GB_TIMESYNC_STATE_ACTIVE) {
+ timesync_svc->state = state;
+ queue_delayed_work(timesync_svc->work_queue,
+ ×ync_svc->delayed_work,
+ GB_TIMESYNC_DELAYED_WORK_SHORT);
+ }
+ break;
+ case GB_TIMESYNC_STATE_ACTIVE:
+ if (timesync_svc->state == GB_TIMESYNC_STATE_AUTHORITATIVE ||
+ timesync_svc->state == GB_TIMESYNC_STATE_PING) {
+ timesync_svc->state = state;
+ wake_up(×ync_svc->wait_queue);
+ }
+ break;
+ }
+
+ if (WARN_ON(timesync_svc->state != state)) {
+ pr_err("Invalid state transition %d=>%d\n",
+ timesync_svc->state, state);
+ }
+}
+
+static void gb_timesync_set_state_atomic(struct gb_timesync_svc *timesync_svc,
+ int state)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+ gb_timesync_set_state(timesync_svc, state);
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+}
+
+static u64 gb_timesync_diff(u64 x, u64 y)
+{
+ if (x > y)
+ return x - y;
+ else
+ return y - x;
+}
+
+static void gb_timesync_adjust_to_svc(struct gb_timesync_svc *svc,
+ u64 svc_frame_time, u64 ap_frame_time)
+{
+ if (svc_frame_time > ap_frame_time) {
+ svc->frame_time_offset = svc_frame_time - ap_frame_time;
+ svc->offset_down = false;
+ } else {
+ svc->frame_time_offset = ap_frame_time - svc_frame_time;
+ svc->offset_down = true;
+ }
+}
+
+/*
+ * Associate a FrameTime with a ktime timestamp represented as struct timespec
+ * Requires the calling context to hold timesync_svc->mutex
+ */
+static void gb_timesync_store_ktime(struct gb_timesync_svc *timesync_svc,
+ struct timespec ts, u64 frame_time)
+{
+ timesync_svc->ktime_data.ts = ts;
+ timesync_svc->ktime_data.frame_time = frame_time;
+}
+
+/*
+ * Find the two pulses that best-match our expected inter-strobe gap and
+ * then calculate the difference between the SVC time at the second pulse
+ * to the local time at the second pulse.
+ */
+static void gb_timesync_collate_frame_time(struct gb_timesync_svc *timesync_svc,
+ u64 *frame_time)
+{
+ int i = 0;
+ u64 delta, ap_frame_time;
+ u64 strobe_delay_ns = GB_TIMESYNC_STROBE_DELAY_US * NSEC_PER_USEC;
+ u64 least = 0;
+
+ for (i = 1; i < GB_TIMESYNC_MAX_STROBES; i++) {
+ delta = timesync_svc->strobe_data[i].frame_time -
+ timesync_svc->strobe_data[i - 1].frame_time;
+ delta *= gb_timesync_ns_per_clock;
+ delta = gb_timesync_diff(delta, strobe_delay_ns);
+
+ if (!least || delta < least) {
+ least = delta;
+ gb_timesync_adjust_to_svc(timesync_svc, frame_time[i],
+ timesync_svc->strobe_data[i].frame_time);
+
+ ap_frame_time = timesync_svc->strobe_data[i].frame_time;
+ ap_frame_time = gb_timesync_adjust_count(timesync_svc,
+ ap_frame_time);
+ gb_timesync_store_ktime(timesync_svc,
+ timesync_svc->strobe_data[i].ts,
+ ap_frame_time);
+
+ pr_debug("adjust %s local %llu svc %llu delta %llu\n",
+ timesync_svc->offset_down ? "down" : "up",
+ timesync_svc->strobe_data[i].frame_time,
+ frame_time[i], delta);
+ }
+ }
+}
+
+static void gb_timesync_teardown(struct gb_timesync_svc *timesync_svc)
+{
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_interface *interface;
+ struct gb_host_device *hd;
+ int ret;
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ interface = timesync_interface->interface;
+ ret = gb_interface_timesync_disable(interface);
+ if (ret) {
+ dev_err(&interface->dev,
+ "interface timesync_disable %d\n", ret);
+ }
+ }
+
+ hd = timesync_svc->timesync_hd->hd;
+ ret = hd->driver->timesync_disable(hd);
+ if (ret < 0) {
+ dev_err(&hd->dev, "host timesync_disable %d\n",
+ ret);
+ }
+
+ gb_svc_timesync_wake_pins_release(svc);
+ gb_svc_timesync_disable(svc);
+ gb_timesync_platform_unlock_bus();
+
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_INACTIVE);
+}
+
+static void gb_timesync_platform_lock_bus_fail(struct gb_timesync_svc
+ *timesync_svc, int ret)
+{
+ if (ret == -EAGAIN) {
+ gb_timesync_set_state(timesync_svc, timesync_svc->state);
+ } else {
+ pr_err("Failed to lock timesync bus %d\n", ret);
+ gb_timesync_set_state(timesync_svc, GB_TIMESYNC_STATE_INACTIVE);
+ }
+}
+
+static void gb_timesync_enable(struct gb_timesync_svc *timesync_svc)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_interface *interface;
+ u64 init_frame_time;
+ unsigned long clock_rate = gb_timesync_clock_rate;
+ int ret;
+
+ /*
+ * Get access to the wake pins in the AP and SVC
+ * Release these pins either in gb_timesync_teardown() or in
+ * gb_timesync_authoritative()
+ */
+ ret = gb_timesync_platform_lock_bus(timesync_svc);
+ if (ret < 0) {
+ gb_timesync_platform_lock_bus_fail(timesync_svc, ret);
+ return;
+ }
+ ret = gb_svc_timesync_wake_pins_acquire(svc, timesync_svc->strobe_mask);
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_wake_pins_acquire %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+
+ /* Choose an initial time in the future */
+ init_frame_time = __gb_timesync_get_frame_time(timesync_svc) + 100000UL;
+
+ /* Send enable command to all relevant participants */
+ list_for_each_entry(timesync_interface, ×ync_svc->interface_list,
+ list) {
+ interface = timesync_interface->interface;
+ ret = gb_interface_timesync_enable(interface,
+ GB_TIMESYNC_MAX_STROBES,
+ init_frame_time,
+ GB_TIMESYNC_STROBE_DELAY_US,
+ clock_rate);
+ if (ret) {
+ dev_err(&interface->dev,
+ "interface timesync_enable %d\n", ret);
+ }
+ }
+
+ hd = timesync_svc->timesync_hd->hd;
+ ret = hd->driver->timesync_enable(hd, GB_TIMESYNC_MAX_STROBES,
+ init_frame_time,
+ GB_TIMESYNC_STROBE_DELAY_US,
+ clock_rate);
+ if (ret < 0) {
+ dev_err(&hd->dev, "host timesync_enable %d\n",
+ ret);
+ }
+
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_WAIT_SVC);
+ ret = gb_svc_timesync_enable(svc, GB_TIMESYNC_MAX_STROBES,
+ init_frame_time,
+ GB_TIMESYNC_STROBE_DELAY_US,
+ clock_rate);
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_enable %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+
+ /* Schedule a timeout waiting for SVC to complete strobing */
+ gb_timesync_schedule_svc_timeout(timesync_svc);
+}
+
+static void gb_timesync_authoritative(struct gb_timesync_svc *timesync_svc)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_interface *interface;
+ u64 svc_frame_time[GB_TIMESYNC_MAX_STROBES];
+ int ret;
+
+ /* Get authoritative time from SVC and adjust local clock */
+ ret = gb_svc_timesync_authoritative(svc, svc_frame_time);
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_authoritative %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+ gb_timesync_collate_frame_time(timesync_svc, svc_frame_time);
+
+ /* Transmit authoritative time to downstream slaves */
+ hd = timesync_svc->timesync_hd->hd;
+ ret = hd->driver->timesync_authoritative(hd, svc_frame_time);
+ if (ret < 0)
+ dev_err(&hd->dev, "host timesync_authoritative %d\n", ret);
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ interface = timesync_interface->interface;
+ ret = gb_interface_timesync_authoritative(
+ interface,
+ svc_frame_time);
+ if (ret) {
+ dev_err(&interface->dev,
+ "interface timesync_authoritative %d\n", ret);
+ }
+ }
+
+ /* Release wake pins */
+ gb_svc_timesync_wake_pins_release(svc);
+ gb_timesync_platform_unlock_bus();
+
+ /* Transition to state ACTIVE */
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_ACTIVE);
+
+ /* Schedule a ping to verify the synchronized system time */
+ timesync_svc->print_ping = true;
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_PING);
+}
+
+static int __gb_timesync_get_status(struct gb_timesync_svc *timesync_svc)
+{
+ int ret = -EINVAL;
+
+ switch (timesync_svc->state) {
+ case GB_TIMESYNC_STATE_INVALID:
+ case GB_TIMESYNC_STATE_INACTIVE:
+ ret = -ENODEV;
+ break;
+ case GB_TIMESYNC_STATE_INIT:
+ case GB_TIMESYNC_STATE_WAIT_SVC:
+ case GB_TIMESYNC_STATE_AUTHORITATIVE:
+ ret = -EAGAIN;
+ break;
+ case GB_TIMESYNC_STATE_PING:
+ case GB_TIMESYNC_STATE_ACTIVE:
+ ret = 0;
+ break;
+ }
+ return ret;
+}
+
+/*
+ * This routine takes a FrameTime and derives the difference with-respect
+ * to a reference FrameTime/ktime pair. It then returns the calculated
+ * ktime based on the difference between the supplied FrameTime and
+ * the reference FrameTime.
+ *
+ * The time difference is calculated to six decimal places. Taking 19.2MHz
+ * as an example this means we have 52.083333~ nanoseconds per clock or
+ * 52083333~ femtoseconds per clock.
+ *
+ * Naively taking the count difference and converting to
+ * seconds/nanoseconds would quickly see the 0.0833 component produce
+ * noticeable errors. For example a time difference of one second would
+ * loose 19200000 * 0.08333x nanoseconds or 1.59 seconds.
+ *
+ * In contrast calculating in femtoseconds the same example of 19200000 *
+ * 0.000000083333x nanoseconds per count of error is just 1.59 nanoseconds!
+ *
+ * Continuing the example of 19.2 MHz we cap the maximum error difference
+ * at a worst-case 0.3 microseconds over a potential calculation window of
+ * abount 15 seconds, meaning you can convert a FrameTime that is <= 15
+ * seconds older/younger than the reference time with a maximum error of
+ * 0.2385 useconds. Note 19.2MHz is an example frequency not a requirement.
+ */
+static int gb_timesync_to_timespec(struct gb_timesync_svc *timesync_svc,
+ u64 frame_time, struct timespec *ts)
+{
+ unsigned long flags;
+ u64 delta_fs, counts, sec, nsec;
+ bool add;
+ int ret = 0;
+
+ memset(ts, 0x00, sizeof(*ts));
+ mutex_lock(×ync_svc->mutex);
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+
+ ret = __gb_timesync_get_status(timesync_svc);
+ if (ret)
+ goto done;
+
+ /* Support calculating ktime upwards or downwards from the reference */
+ if (frame_time < timesync_svc->ktime_data.frame_time) {
+ add = false;
+ counts = timesync_svc->ktime_data.frame_time - frame_time;
+ } else {
+ add = true;
+ counts = frame_time - timesync_svc->ktime_data.frame_time;
+ }
+
+ /* Enforce the .23 of a usecond boundary @ 19.2MHz */
+ if (counts > gb_timesync_max_ktime_diff) {
+ ret = -EINVAL;
+ goto done;
+ }
+
+ /* Determine the time difference in femtoseconds */
+ delta_fs = counts * gb_timesync_fs_per_clock;
+
+ /* Convert to seconds */
+ sec = delta_fs;
+ do_div(sec, NSEC_PER_SEC);
+ do_div(sec, 1000000UL);
+
+ /* Get the nanosecond remainder */
+ nsec = do_div(delta_fs, sec);
+ do_div(nsec, 1000000UL);
+
+ if (add) {
+ /* Add the calculated offset - overflow nanoseconds upwards */
+ ts->tv_sec = timesync_svc->ktime_data.ts.tv_sec + sec;
+ ts->tv_nsec = timesync_svc->ktime_data.ts.tv_nsec + nsec;
+ if (ts->tv_nsec >= NSEC_PER_SEC) {
+ ts->tv_sec++;
+ ts->tv_nsec -= NSEC_PER_SEC;
+ }
+ } else {
+ /* Subtract the difference over/underflow as necessary */
+ if (nsec > timesync_svc->ktime_data.ts.tv_nsec) {
+ sec++;
+ nsec = nsec + timesync_svc->ktime_data.ts.tv_nsec;
+ nsec = do_div(nsec, NSEC_PER_SEC);
+ } else {
+ nsec = timesync_svc->ktime_data.ts.tv_nsec - nsec;
+ }
+ /* Cannot return a negative second value */
+ if (sec > timesync_svc->ktime_data.ts.tv_sec) {
+ ret = -EINVAL;
+ goto done;
+ }
+ ts->tv_sec = timesync_svc->ktime_data.ts.tv_sec - sec;
+ ts->tv_nsec = nsec;
+ }
+done:
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+ mutex_unlock(×ync_svc->mutex);
+ return ret;
+}
+
+static size_t gb_timesync_log_frame_time(struct gb_timesync_svc *timesync_svc,
+ char *buf, size_t buflen)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_interface *interface;
+ unsigned int len;
+ size_t off;
+
+ /* AP/SVC */
+ off = snprintf(buf, buflen, "%s frametime: ap=%llu %s=%llu ",
+ greybus_bus_type.name,
+ timesync_svc->ap_ping_frame_time, dev_name(&svc->dev),
+ timesync_svc->svc_ping_frame_time);
+ len = buflen - off;
+
+ /* APB/GPB */
+ if (len < buflen) {
+ hd = timesync_svc->timesync_hd->hd;
+ off += snprintf(&buf[off], len, "%s=%llu ", dev_name(&hd->dev),
+ timesync_svc->timesync_hd->ping_frame_time);
+ len = buflen - off;
+ }
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ if (len < buflen) {
+ interface = timesync_interface->interface;
+ off += snprintf(&buf[off], len, "%s=%llu ",
+ dev_name(&interface->dev),
+ timesync_interface->ping_frame_time);
+ len = buflen - off;
+ }
+ }
+ if (len < buflen)
+ off += snprintf(&buf[off], len, "\n");
+ return off;
+}
+
+static size_t gb_timesync_log_frame_ktime(struct gb_timesync_svc *timesync_svc,
+ char *buf, size_t buflen)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_interface *interface;
+ struct timespec ts;
+ unsigned int len;
+ size_t off;
+
+ /* AP */
+ gb_timesync_to_timespec(timesync_svc, timesync_svc->ap_ping_frame_time,
+ &ts);
+ off = snprintf(buf, buflen, "%s frametime: ap=%lu.%lu ",
+ greybus_bus_type.name, ts.tv_sec, ts.tv_nsec);
+ len = buflen - off;
+ if (len >= buflen)
+ goto done;
+
+ /* SVC */
+ gb_timesync_to_timespec(timesync_svc, timesync_svc->svc_ping_frame_time,
+ &ts);
+ off += snprintf(&buf[off], len, "%s=%lu.%lu ", dev_name(&svc->dev),
+ ts.tv_sec, ts.tv_nsec);
+ len = buflen - off;
+ if (len >= buflen)
+ goto done;
+
+ /* APB/GPB */
+ hd = timesync_svc->timesync_hd->hd;
+ gb_timesync_to_timespec(timesync_svc,
+ timesync_svc->timesync_hd->ping_frame_time,
+ &ts);
+ off += snprintf(&buf[off], len, "%s=%lu.%lu ",
+ dev_name(&hd->dev),
+ ts.tv_sec, ts.tv_nsec);
+ len = buflen - off;
+ if (len >= buflen)
+ goto done;
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ interface = timesync_interface->interface;
+ gb_timesync_to_timespec(timesync_svc,
+ timesync_interface->ping_frame_time,
+ &ts);
+ off += snprintf(&buf[off], len, "%s=%lu.%lu ",
+ dev_name(&interface->dev),
+ ts.tv_sec, ts.tv_nsec);
+ len = buflen - off;
+ if (len >= buflen)
+ goto done;
+ }
+ off += snprintf(&buf[off], len, "\n");
+done:
+ return off;
+}
+
+/*
+ * Send an SVC initiated wake 'ping' to each TimeSync participant.
+ * Get the FrameTime from each participant associated with the wake
+ * ping.
+ */
+static void gb_timesync_ping(struct gb_timesync_svc *timesync_svc)
+{
+ struct gb_svc *svc = timesync_svc->svc;
+ struct gb_host_device *hd;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_control *control;
+ u64 *ping_frame_time;
+ int ret;
+
+ /* Get access to the wake pins in the AP and SVC */
+ ret = gb_timesync_platform_lock_bus(timesync_svc);
+ if (ret < 0) {
+ gb_timesync_platform_lock_bus_fail(timesync_svc, ret);
+ return;
+ }
+ ret = gb_svc_timesync_wake_pins_acquire(svc, timesync_svc->strobe_mask);
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_wake_pins_acquire %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+
+ /* Have SVC generate a timesync ping */
+ timesync_svc->capture_ping = true;
+ timesync_svc->svc_ping_frame_time = 0;
+ ret = gb_svc_timesync_ping(svc, ×ync_svc->svc_ping_frame_time);
+ timesync_svc->capture_ping = false;
+ if (ret) {
+ dev_err(&svc->dev,
+ "gb_svc_timesync_ping %d\n", ret);
+ gb_timesync_teardown(timesync_svc);
+ return;
+ }
+
+ /* Get the ping FrameTime from each APB/GPB */
+ hd = timesync_svc->timesync_hd->hd;
+ timesync_svc->timesync_hd->ping_frame_time = 0;
+ ret = hd->driver->timesync_get_last_event(hd,
+ ×ync_svc->timesync_hd->ping_frame_time);
+ if (ret)
+ dev_err(&hd->dev, "host timesync_get_last_event %d\n", ret);
+
+ list_for_each_entry(timesync_interface,
+ ×ync_svc->interface_list, list) {
+ control = timesync_interface->interface->control;
+ timesync_interface->ping_frame_time = 0;
+ ping_frame_time = ×ync_interface->ping_frame_time;
+ ret = gb_control_timesync_get_last_event(control,
+ ping_frame_time);
+ if (ret) {
+ dev_err(×ync_interface->interface->dev,
+ "gb_control_timesync_get_last_event %d\n", ret);
+ }
+ }
+
+ /* Ping success - move to timesync active */
+ gb_svc_timesync_wake_pins_release(svc);
+ gb_timesync_platform_unlock_bus();
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_ACTIVE);
+}
+
+static void gb_timesync_log_ping_time(struct gb_timesync_svc *timesync_svc)
+{
+ char *buf;
+
+ if (!timesync_svc->print_ping)
+ return;
+
+ buf = kzalloc(PAGE_SIZE, GFP_KERNEL);
+ if (buf) {
+ gb_timesync_log_frame_time(timesync_svc, buf, PAGE_SIZE);
+ dev_dbg(×ync_svc->svc->dev, "%s", buf);
+ kfree(buf);
+ }
+}
+
+/*
+ * Perform the actual work of scheduled TimeSync logic.
+ */
+static void gb_timesync_worker(struct work_struct *work)
+{
+ struct delayed_work *delayed_work = to_delayed_work(work);
+ struct gb_timesync_svc *timesync_svc =
+ container_of(delayed_work, struct gb_timesync_svc, delayed_work);
+
+ mutex_lock(×ync_svc->mutex);
+
+ switch (timesync_svc->state) {
+ case GB_TIMESYNC_STATE_INIT:
+ gb_timesync_enable(timesync_svc);
+ break;
+
+ case GB_TIMESYNC_STATE_WAIT_SVC:
+ dev_err(×ync_svc->svc->dev,
+ "timeout SVC strobe completion %d/%d\n",
+ timesync_svc->strobe, GB_TIMESYNC_MAX_STROBES);
+ gb_timesync_teardown(timesync_svc);
+ break;
+
+ case GB_TIMESYNC_STATE_AUTHORITATIVE:
+ gb_timesync_authoritative(timesync_svc);
+ break;
+
+ case GB_TIMESYNC_STATE_PING:
+ gb_timesync_ping(timesync_svc);
+ gb_timesync_log_ping_time(timesync_svc);
+ break;
+
+ default:
+ pr_err("Invalid state %d for delayed work\n",
+ timesync_svc->state);
+ break;
+ }
+
+ mutex_unlock(×ync_svc->mutex);
+}
+
+/*
+ * Schedule a new TimeSync INIT or PING operation serialized w/r to
+ * gb_timesync_worker().
+ */
+static int gb_timesync_schedule(struct gb_timesync_svc *timesync_svc, int state)
+{
+ int ret = 0;
+
+ if (state != GB_TIMESYNC_STATE_INIT && state != GB_TIMESYNC_STATE_PING)
+ return -EINVAL;
+
+ mutex_lock(×ync_svc->mutex);
+ if (timesync_svc->state != GB_TIMESYNC_STATE_INVALID) {
+ gb_timesync_set_state_atomic(timesync_svc, state);
+ } else {
+ ret = -ENODEV;
+ }
+ mutex_unlock(×ync_svc->mutex);
+ return ret;
+}
+
+static int __gb_timesync_schedule_synchronous(
+ struct gb_timesync_svc *timesync_svc, int state)
+{
+ unsigned long flags;
+ int ret;
+
+ ret = gb_timesync_schedule(timesync_svc, state);
+ if (ret)
+ return ret;
+
+ ret = wait_event_interruptible(timesync_svc->wait_queue,
+ (timesync_svc->state == GB_TIMESYNC_STATE_ACTIVE ||
+ timesync_svc->state == GB_TIMESYNC_STATE_INACTIVE ||
+ timesync_svc->state == GB_TIMESYNC_STATE_INVALID));
+ if (ret)
+ return ret;
+
+ mutex_lock(×ync_svc->mutex);
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+
+ ret = __gb_timesync_get_status(timesync_svc);
+
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+ mutex_unlock(×ync_svc->mutex);
+
+ return ret;
+}
+
+static struct gb_timesync_svc *gb_timesync_find_timesync_svc(
+ struct gb_host_device *hd)
+{
+ struct gb_timesync_svc *timesync_svc;
+
+ list_for_each_entry(timesync_svc, &gb_timesync_svc_list, list) {
+ if (timesync_svc->svc == hd->svc)
+ return timesync_svc;
+ }
+ return NULL;
+}
+
+static struct gb_timesync_interface *gb_timesync_find_timesync_interface(
+ struct gb_timesync_svc *timesync_svc,
+ struct gb_interface *interface)
+{
+ struct gb_timesync_interface *timesync_interface;
+
+ list_for_each_entry(timesync_interface, ×ync_svc->interface_list, list) {
+ if (timesync_interface->interface == interface)
+ return timesync_interface;
+ }
+ return NULL;
+}
+
+int gb_timesync_schedule_synchronous(struct gb_interface *interface)
+{
+ int ret;
+ struct gb_timesync_svc *timesync_svc;
+ int retries;
+
+ if (!(interface->features & GREYBUS_INTERFACE_FEATURE_TIMESYNC))
+ return 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ for (retries = 0; retries < GB_TIMESYNC_MAX_RETRIES; retries++) {
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc) {
+ ret = -ENODEV;
+ goto done;
+ }
+
+ ret = __gb_timesync_schedule_synchronous(timesync_svc,
+ GB_TIMESYNC_STATE_INIT);
+ if (!ret)
+ break;
+ }
+ if (ret && retries == GB_TIMESYNC_MAX_RETRIES)
+ ret = -ETIMEDOUT;
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_schedule_synchronous);
+
+void gb_timesync_schedule_asynchronous(struct gb_interface *interface)
+{
+ struct gb_timesync_svc *timesync_svc;
+
+ if (!(interface->features & GREYBUS_INTERFACE_FEATURE_TIMESYNC))
+ return;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc)
+ goto done;
+
+ gb_timesync_schedule(timesync_svc, GB_TIMESYNC_STATE_INIT);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_schedule_asynchronous);
+
+static ssize_t gb_timesync_ping_read(struct file *file, char __user *ubuf,
+ size_t len, loff_t *offset, bool ktime)
+{
+ struct gb_timesync_svc *timesync_svc = file->f_inode->i_private;
+ char *buf;
+ ssize_t ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ mutex_lock(×ync_svc->mutex);
+ if (list_empty(×ync_svc->interface_list))
+ ret = -ENODEV;
+ timesync_svc->print_ping = false;
+ mutex_unlock(×ync_svc->mutex);
+ if (ret)
+ goto done;
+
+ ret = __gb_timesync_schedule_synchronous(timesync_svc,
+ GB_TIMESYNC_STATE_PING);
+ if (ret)
+ goto done;
+
+ buf = kzalloc(PAGE_SIZE, GFP_KERNEL);
+ if (!buf) {
+ ret = -ENOMEM;
+ goto done;
+ }
+
+ if (ktime)
+ ret = gb_timesync_log_frame_ktime(timesync_svc, buf, PAGE_SIZE);
+ else
+ ret = gb_timesync_log_frame_time(timesync_svc, buf, PAGE_SIZE);
+ if (ret > 0)
+ ret = simple_read_from_buffer(ubuf, len, offset, buf, ret);
+ kfree(buf);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+
+static ssize_t gb_timesync_ping_read_frame_time(struct file *file,
+ char __user *buf,
+ size_t len, loff_t *offset)
+{
+ return gb_timesync_ping_read(file, buf, len, offset, false);
+}
+
+static ssize_t gb_timesync_ping_read_frame_ktime(struct file *file,
+ char __user *buf,
+ size_t len, loff_t *offset)
+{
+ return gb_timesync_ping_read(file, buf, len, offset, true);
+}
+
+static const struct file_operations gb_timesync_debugfs_frame_time_ops = {
+ .read = gb_timesync_ping_read_frame_time,
+};
+
+static const struct file_operations gb_timesync_debugfs_frame_ktime_ops = {
+ .read = gb_timesync_ping_read_frame_ktime,
+};
+
+static int gb_timesync_hd_add(struct gb_timesync_svc *timesync_svc,
+ struct gb_host_device *hd)
+{
+ struct gb_timesync_host_device *timesync_hd;
+
+ timesync_hd = kzalloc(sizeof(*timesync_hd), GFP_KERNEL);
+ if (!timesync_hd)
+ return -ENOMEM;
+
+ WARN_ON(timesync_svc->timesync_hd);
+ timesync_hd->hd = hd;
+ timesync_svc->timesync_hd = timesync_hd;
+
+ return 0;
+}
+
+static void gb_timesync_hd_remove(struct gb_timesync_svc *timesync_svc,
+ struct gb_host_device *hd)
+{
+ if (timesync_svc->timesync_hd->hd == hd) {
+ kfree(timesync_svc->timesync_hd);
+ timesync_svc->timesync_hd = NULL;
+ return;
+ }
+ WARN_ON(1);
+}
+
+int gb_timesync_svc_add(struct gb_svc *svc)
+{
+ struct gb_timesync_svc *timesync_svc;
+ int ret;
+
+ timesync_svc = kzalloc(sizeof(*timesync_svc), GFP_KERNEL);
+ if (!timesync_svc)
+ return -ENOMEM;
+
+ timesync_svc->work_queue =
+ create_singlethread_workqueue("gb-timesync-work_queue");
+
+ if (!timesync_svc->work_queue) {
+ kfree(timesync_svc);
+ return -ENOMEM;
+ }
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ INIT_LIST_HEAD(×ync_svc->interface_list);
+ INIT_DELAYED_WORK(×ync_svc->delayed_work, gb_timesync_worker);
+ mutex_init(×ync_svc->mutex);
+ spin_lock_init(×ync_svc->spinlock);
+ init_waitqueue_head(×ync_svc->wait_queue);
+
+ timesync_svc->svc = svc;
+ timesync_svc->frame_time_offset = 0;
+ timesync_svc->capture_ping = false;
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_INACTIVE);
+
+ timesync_svc->frame_time_dentry =
+ debugfs_create_file("frame-time", S_IRUGO, svc->debugfs_dentry,
+ timesync_svc,
+ &gb_timesync_debugfs_frame_time_ops);
+ timesync_svc->frame_ktime_dentry =
+ debugfs_create_file("frame-ktime", S_IRUGO, svc->debugfs_dentry,
+ timesync_svc,
+ &gb_timesync_debugfs_frame_ktime_ops);
+
+ list_add(×ync_svc->list, &gb_timesync_svc_list);
+ ret = gb_timesync_hd_add(timesync_svc, svc->hd);
+ if (ret) {
+ list_del(×ync_svc->list);
+ debugfs_remove(timesync_svc->frame_ktime_dentry);
+ debugfs_remove(timesync_svc->frame_time_dentry);
+ destroy_workqueue(timesync_svc->work_queue);
+ kfree(timesync_svc);
+ goto done;
+ }
+
+ init_timer(×ync_svc->ktime_timer);
+ timesync_svc->ktime_timer.function = gb_timesync_ktime_timer_fn;
+ timesync_svc->ktime_timer.expires = jiffies + GB_TIMESYNC_KTIME_UPDATE;
+ timesync_svc->ktime_timer.data = (unsigned long)timesync_svc;
+ add_timer(×ync_svc->ktime_timer);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_svc_add);
+
+void gb_timesync_svc_remove(struct gb_svc *svc)
+{
+ struct gb_timesync_svc *timesync_svc;
+ struct gb_timesync_interface *timesync_interface;
+ struct gb_timesync_interface *next;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(svc->hd);
+ if (!timesync_svc)
+ goto done;
+
+ cancel_delayed_work_sync(×ync_svc->delayed_work);
+
+ mutex_lock(×ync_svc->mutex);
+
+ gb_timesync_set_state_atomic(timesync_svc, GB_TIMESYNC_STATE_INVALID);
+ del_timer_sync(×ync_svc->ktime_timer);
+ gb_timesync_teardown(timesync_svc);
+
+ gb_timesync_hd_remove(timesync_svc, svc->hd);
+ list_for_each_entry_safe(timesync_interface, next,
+ ×ync_svc->interface_list, list) {
+ list_del(×ync_interface->list);
+ kfree(timesync_interface);
+ }
+ debugfs_remove(timesync_svc->frame_ktime_dentry);
+ debugfs_remove(timesync_svc->frame_time_dentry);
+ destroy_workqueue(timesync_svc->work_queue);
+ list_del(×ync_svc->list);
+
+ mutex_unlock(×ync_svc->mutex);
+
+ kfree(timesync_svc);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+}
+EXPORT_SYMBOL_GPL(gb_timesync_svc_remove);
+
+/*
+ * Add a Greybus Interface to the set of TimeSync Interfaces.
+ */
+int gb_timesync_interface_add(struct gb_interface *interface)
+{
+ struct gb_timesync_svc *timesync_svc;
+ struct gb_timesync_interface *timesync_interface;
+ int ret = 0;
+
+ if (!(interface->features & GREYBUS_INTERFACE_FEATURE_TIMESYNC))
+ return 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc) {
+ ret = -ENODEV;
+ goto done;
+ }
+
+ timesync_interface = kzalloc(sizeof(*timesync_interface), GFP_KERNEL);
+ if (!timesync_interface) {
+ ret = -ENOMEM;
+ goto done;
+ }
+
+ mutex_lock(×ync_svc->mutex);
+ timesync_interface->interface = interface;
+ list_add(×ync_interface->list, ×ync_svc->interface_list);
+ timesync_svc->strobe_mask |= 1 << interface->interface_id;
+ mutex_unlock(×ync_svc->mutex);
+
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_interface_add);
+
+/*
+ * Remove a Greybus Interface from the set of TimeSync Interfaces.
+ */
+void gb_timesync_interface_remove(struct gb_interface *interface)
+{
+ struct gb_timesync_svc *timesync_svc;
+ struct gb_timesync_interface *timesync_interface;
+
+ if (!(interface->features & GREYBUS_INTERFACE_FEATURE_TIMESYNC))
+ return;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc)
+ goto done;
+
+ timesync_interface = gb_timesync_find_timesync_interface(timesync_svc,
+ interface);
+ if (!timesync_interface)
+ goto done;
+
+ mutex_lock(×ync_svc->mutex);
+ timesync_svc->strobe_mask &= ~(1 << interface->interface_id);
+ list_del(×ync_interface->list);
+ kfree(timesync_interface);
+ mutex_unlock(×ync_svc->mutex);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+}
+EXPORT_SYMBOL_GPL(gb_timesync_interface_remove);
+
+/*
+ * Give the authoritative FrameTime to the calling function. Returns zero if we
+ * are not in GB_TIMESYNC_STATE_ACTIVE.
+ */
+static u64 gb_timesync_get_frame_time(struct gb_timesync_svc *timesync_svc)
+{
+ unsigned long flags;
+ u64 ret;
+
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+ if (timesync_svc->state == GB_TIMESYNC_STATE_ACTIVE)
+ ret = __gb_timesync_get_frame_time(timesync_svc);
+ else
+ ret = 0;
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+ return ret;
+}
+
+u64 gb_timesync_get_frame_time_by_interface(struct gb_interface *interface)
+{
+ struct gb_timesync_svc *timesync_svc;
+ u64 ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc)
+ goto done;
+
+ ret = gb_timesync_get_frame_time(timesync_svc);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_get_frame_time_by_interface);
+
+u64 gb_timesync_get_frame_time_by_svc(struct gb_svc *svc)
+{
+ struct gb_timesync_svc *timesync_svc;
+ u64 ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(svc->hd);
+ if (!timesync_svc)
+ goto done;
+
+ ret = gb_timesync_get_frame_time(timesync_svc);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_get_frame_time_by_svc);
+
+/* Incrementally updates the conversion base from FrameTime to ktime */
+static void gb_timesync_ktime_timer_fn(unsigned long data)
+{
+ struct gb_timesync_svc *timesync_svc =
+ (struct gb_timesync_svc *)data;
+ unsigned long flags;
+ u64 frame_time;
+ struct timespec ts;
+
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+
+ if (timesync_svc->state != GB_TIMESYNC_STATE_ACTIVE)
+ goto done;
+
+ ktime_get_ts(&ts);
+ frame_time = __gb_timesync_get_frame_time(timesync_svc);
+ gb_timesync_store_ktime(timesync_svc, ts, frame_time);
+
+done:
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+ mod_timer(×ync_svc->ktime_timer,
+ jiffies + GB_TIMESYNC_KTIME_UPDATE);
+}
+
+int gb_timesync_to_timespec_by_svc(struct gb_svc *svc, u64 frame_time,
+ struct timespec *ts)
+{
+ struct gb_timesync_svc *timesync_svc;
+ int ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(svc->hd);
+ if (!timesync_svc) {
+ ret = -ENODEV;
+ goto done;
+ }
+ ret = gb_timesync_to_timespec(timesync_svc, frame_time, ts);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_to_timespec_by_svc);
+
+int gb_timesync_to_timespec_by_interface(struct gb_interface *interface,
+ u64 frame_time, struct timespec *ts)
+{
+ struct gb_timesync_svc *timesync_svc;
+ int ret = 0;
+
+ mutex_lock(&gb_timesync_svc_list_mutex);
+ timesync_svc = gb_timesync_find_timesync_svc(interface->hd);
+ if (!timesync_svc) {
+ ret = -ENODEV;
+ goto done;
+ }
+
+ ret = gb_timesync_to_timespec(timesync_svc, frame_time, ts);
+done:
+ mutex_unlock(&gb_timesync_svc_list_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_timesync_to_timespec_by_interface);
+
+void gb_timesync_irq(struct gb_timesync_svc *timesync_svc)
+{
+ unsigned long flags;
+ u64 strobe_time;
+ bool strobe_is_ping = true;
+ struct timespec ts;
+
+ ktime_get_ts(&ts);
+ strobe_time = __gb_timesync_get_frame_time(timesync_svc);
+
+ spin_lock_irqsave(×ync_svc->spinlock, flags);
+
+ if (timesync_svc->state == GB_TIMESYNC_STATE_PING) {
+ if (!timesync_svc->capture_ping)
+ goto done_nolog;
+ timesync_svc->ap_ping_frame_time = strobe_time;
+ goto done_log;
+ } else if (timesync_svc->state != GB_TIMESYNC_STATE_WAIT_SVC) {
+ goto done_nolog;
+ }
+
+ timesync_svc->strobe_data[timesync_svc->strobe].frame_time = strobe_time;
+ timesync_svc->strobe_data[timesync_svc->strobe].ts = ts;
+
+ if (++timesync_svc->strobe == GB_TIMESYNC_MAX_STROBES) {
+ gb_timesync_set_state(timesync_svc,
+ GB_TIMESYNC_STATE_AUTHORITATIVE);
+ }
+ strobe_is_ping = false;
+done_log:
+ trace_gb_timesync_irq(strobe_is_ping, timesync_svc->strobe,
+ GB_TIMESYNC_MAX_STROBES, strobe_time);
+done_nolog:
+ spin_unlock_irqrestore(×ync_svc->spinlock, flags);
+}
+EXPORT_SYMBOL(gb_timesync_irq);
+
+int __init gb_timesync_init(void)
+{
+ int ret = 0;
+
+ ret = gb_timesync_platform_init();
+ if (ret) {
+ pr_err("timesync platform init fail!\n");
+ return ret;
+ }
+
+ gb_timesync_clock_rate = gb_timesync_platform_get_clock_rate();
+
+ /* Calculate nanoseconds and femtoseconds per clock */
+ gb_timesync_fs_per_clock = FSEC_PER_SEC;
+ do_div(gb_timesync_fs_per_clock, gb_timesync_clock_rate);
+ gb_timesync_ns_per_clock = NSEC_PER_SEC;
+ do_div(gb_timesync_ns_per_clock, gb_timesync_clock_rate);
+
+ /* Calculate the maximum number of clocks we will convert to ktime */
+ gb_timesync_max_ktime_diff =
+ GB_TIMESYNC_MAX_KTIME_CONVERSION * gb_timesync_clock_rate;
+
+ pr_info("Time-Sync @ %lu Hz max ktime conversion +/- %d seconds\n",
+ gb_timesync_clock_rate, GB_TIMESYNC_MAX_KTIME_CONVERSION);
+ return 0;
+}
+
+void gb_timesync_exit(void)
+{
+ gb_timesync_platform_exit();
+}
--- /dev/null
+++ b/drivers/greybus/timesync.h
@@ -0,0 +1,45 @@
+/*
+ * TimeSync API driver.
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __TIMESYNC_H
+#define __TIMESYNC_H
+
+struct gb_svc;
+struct gb_interface;
+struct gb_timesync_svc;
+
+/* Platform */
+u64 gb_timesync_platform_get_counter(void);
+u32 gb_timesync_platform_get_clock_rate(void);
+int gb_timesync_platform_lock_bus(struct gb_timesync_svc *pdata);
+void gb_timesync_platform_unlock_bus(void);
+
+int gb_timesync_platform_init(void);
+void gb_timesync_platform_exit(void);
+
+/* Core API */
+int gb_timesync_interface_add(struct gb_interface *interface);
+void gb_timesync_interface_remove(struct gb_interface *interface);
+int gb_timesync_svc_add(struct gb_svc *svc);
+void gb_timesync_svc_remove(struct gb_svc *svc);
+
+u64 gb_timesync_get_frame_time_by_interface(struct gb_interface *interface);
+u64 gb_timesync_get_frame_time_by_svc(struct gb_svc *svc);
+int gb_timesync_to_timespec_by_svc(struct gb_svc *svc, u64 frame_time,
+ struct timespec *ts);
+int gb_timesync_to_timespec_by_interface(struct gb_interface *interface,
+ u64 frame_time, struct timespec *ts);
+
+int gb_timesync_schedule_synchronous(struct gb_interface *intf);
+void gb_timesync_schedule_asynchronous(struct gb_interface *intf);
+void gb_timesync_irq(struct gb_timesync_svc *timesync_svc);
+int gb_timesync_init(void);
+void gb_timesync_exit(void);
+
+#endif /* __TIMESYNC_H */
--- /dev/null
+++ b/drivers/greybus/timesync_platform.c
@@ -0,0 +1,77 @@
+/*
+ * TimeSync API driver.
+ *
+ * Copyright 2016 Google Inc.
+ * Copyright 2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ *
+ * This code reads directly from an ARMv7 memory-mapped timer that lives in
+ * MMIO space. Since this counter lives inside of MMIO space its shared between
+ * cores and that means we don't have to worry about issues like TSC on x86
+ * where each time-stamp-counter (TSC) is local to a particular core.
+ *
+ * Register-level access code is based on
+ * drivers/clocksource/arm_arch_timer.c
+ */
+#include <linux/cpufreq.h>
+#include <linux/of_platform.h>
+
+#include "greybus.h"
+#include "arche_platform.h"
+
+static u32 gb_timesync_clock_frequency;
+int (*arche_platform_change_state_cb)(enum arche_platform_state state,
+ struct gb_timesync_svc *pdata);
+EXPORT_SYMBOL_GPL(arche_platform_change_state_cb);
+
+u64 gb_timesync_platform_get_counter(void)
+{
+ return (u64)get_cycles();
+}
+
+u32 gb_timesync_platform_get_clock_rate(void)
+{
+ if (unlikely(!gb_timesync_clock_frequency))
+ return cpufreq_get(0);
+
+ return gb_timesync_clock_frequency;
+}
+
+int gb_timesync_platform_lock_bus(struct gb_timesync_svc *pdata)
+{
+ return arche_platform_change_state_cb(ARCHE_PLATFORM_STATE_TIME_SYNC,
+ pdata);
+}
+
+void gb_timesync_platform_unlock_bus(void)
+{
+ arche_platform_change_state_cb(ARCHE_PLATFORM_STATE_ACTIVE, NULL);
+}
+
+static const struct of_device_id arch_timer_of_match[] = {
+ { .compatible = "google,greybus-frame-time-counter", },
+ {},
+};
+
+int __init gb_timesync_platform_init(void)
+{
+ struct device_node *np;
+
+ np = of_find_matching_node(NULL, arch_timer_of_match);
+ if (!np) {
+ /* Tolerate not finding to allow BBB etc to continue */
+ pr_warn("Unable to find a compatible ARMv7 timer\n");
+ return 0;
+ }
+
+ if (of_property_read_u32(np, "clock-frequency",
+ &gb_timesync_clock_frequency)) {
+ pr_err("Unable to find timer clock-frequency\n");
+ return -ENODEV;
+ }
+
+ return 0;
+}
+
+void gb_timesync_platform_exit(void) {}
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 20/32] greybus: vibrator driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (17 preceding siblings ...)
2016-09-16 14:11 ` [patch 19/32] greybus: timesync driver Greg KH
@ 2016-09-16 14:11 ` Greg KH
2016-09-16 14:19 ` [patch 21/32] greybus: arche platform driver Greg KH
` (11 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:11 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus vibrator class protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/vibrator.c | 249 +++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 249 insertions(+)
--- /dev/null
+++ b/drivers/greybus/vibrator.c
@@ -0,0 +1,249 @@
+/*
+ * Greybus Vibrator protocol driver.
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/device.h>
+#include <linux/kdev_t.h>
+#include <linux/idr.h>
+#include <linux/pm_runtime.h>
+
+#include "greybus.h"
+
+struct gb_vibrator_device {
+ struct gb_connection *connection;
+ struct device *dev;
+ int minor; /* vibrator minor number */
+ struct delayed_work delayed_work;
+};
+
+/* Greybus Vibrator operation types */
+#define GB_VIBRATOR_TYPE_ON 0x02
+#define GB_VIBRATOR_TYPE_OFF 0x03
+
+static int turn_off(struct gb_vibrator_device *vib)
+{
+ struct gb_bundle *bundle = vib->connection->bundle;
+ int ret;
+
+ ret = gb_operation_sync(vib->connection, GB_VIBRATOR_TYPE_OFF,
+ NULL, 0, NULL, 0);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return ret;
+}
+
+static int turn_on(struct gb_vibrator_device *vib, u16 timeout_ms)
+{
+ struct gb_bundle *bundle = vib->connection->bundle;
+ int ret;
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ return ret;
+
+ /* Vibrator was switched ON earlier */
+ if (cancel_delayed_work_sync(&vib->delayed_work))
+ turn_off(vib);
+
+ ret = gb_operation_sync(vib->connection, GB_VIBRATOR_TYPE_ON,
+ NULL, 0, NULL, 0);
+ if (ret) {
+ gb_pm_runtime_put_autosuspend(bundle);
+ return ret;
+ }
+
+ schedule_delayed_work(&vib->delayed_work, msecs_to_jiffies(timeout_ms));
+
+ return 0;
+}
+
+static void gb_vibrator_worker(struct work_struct *work)
+{
+ struct delayed_work *delayed_work = to_delayed_work(work);
+ struct gb_vibrator_device *vib =
+ container_of(delayed_work, struct gb_vibrator_device, delayed_work);
+
+ turn_off(vib);
+}
+
+static ssize_t timeout_store(struct device *dev, struct device_attribute *attr,
+ const char *buf, size_t count)
+{
+ struct gb_vibrator_device *vib = dev_get_drvdata(dev);
+ unsigned long val;
+ int retval;
+
+ retval = kstrtoul(buf, 10, &val);
+ if (retval < 0) {
+ dev_err(dev, "could not parse timeout value %d\n", retval);
+ return retval;
+ }
+
+ if (val)
+ retval = turn_on(vib, (u16)val);
+ else
+ retval = turn_off(vib);
+ if (retval)
+ return retval;
+
+ return count;
+}
+static DEVICE_ATTR_WO(timeout);
+
+static struct attribute *vibrator_attrs[] = {
+ &dev_attr_timeout.attr,
+ NULL,
+};
+ATTRIBUTE_GROUPS(vibrator);
+
+static struct class vibrator_class = {
+ .name = "vibrator",
+ .owner = THIS_MODULE,
+ .dev_groups = vibrator_groups,
+};
+
+static DEFINE_IDA(minors);
+
+static int gb_vibrator_probe(struct gb_bundle *bundle,
+ const struct greybus_bundle_id *id)
+{
+ struct greybus_descriptor_cport *cport_desc;
+ struct gb_connection *connection;
+ struct gb_vibrator_device *vib;
+ struct device *dev;
+ int retval;
+
+ if (bundle->num_cports != 1)
+ return -ENODEV;
+
+ cport_desc = &bundle->cport_desc[0];
+ if (cport_desc->protocol_id != GREYBUS_PROTOCOL_VIBRATOR)
+ return -ENODEV;
+
+ vib = kzalloc(sizeof(*vib), GFP_KERNEL);
+ if (!vib)
+ return -ENOMEM;
+
+ connection = gb_connection_create(bundle, le16_to_cpu(cport_desc->id),
+ NULL);
+ if (IS_ERR(connection)) {
+ retval = PTR_ERR(connection);
+ goto err_free_vib;
+ }
+ gb_connection_set_data(connection, vib);
+
+ vib->connection = connection;
+
+ greybus_set_drvdata(bundle, vib);
+
+ retval = gb_connection_enable(connection);
+ if (retval)
+ goto err_connection_destroy;
+
+ /*
+ * For now we create a device in sysfs for the vibrator, but odds are
+ * there is a "real" device somewhere in the kernel for this, but I
+ * can't find it at the moment...
+ */
+ vib->minor = ida_simple_get(&minors, 0, 0, GFP_KERNEL);
+ if (vib->minor < 0) {
+ retval = vib->minor;
+ goto err_connection_disable;
+ }
+ dev = device_create(&vibrator_class, &bundle->dev,
+ MKDEV(0, 0), vib, "vibrator%d", vib->minor);
+ if (IS_ERR(dev)) {
+ retval = -EINVAL;
+ goto err_ida_remove;
+ }
+ vib->dev = dev;
+
+ INIT_DELAYED_WORK(&vib->delayed_work, gb_vibrator_worker);
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return 0;
+
+err_ida_remove:
+ ida_simple_remove(&minors, vib->minor);
+err_connection_disable:
+ gb_connection_disable(connection);
+err_connection_destroy:
+ gb_connection_destroy(connection);
+err_free_vib:
+ kfree(vib);
+
+ return retval;
+}
+
+static void gb_vibrator_disconnect(struct gb_bundle *bundle)
+{
+ struct gb_vibrator_device *vib = greybus_get_drvdata(bundle);
+ int ret;
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret)
+ gb_pm_runtime_get_noresume(bundle);
+
+ if (cancel_delayed_work_sync(&vib->delayed_work))
+ turn_off(vib);
+
+ device_unregister(vib->dev);
+ ida_simple_remove(&minors, vib->minor);
+ gb_connection_disable(vib->connection);
+ gb_connection_destroy(vib->connection);
+ kfree(vib);
+}
+
+static const struct greybus_bundle_id gb_vibrator_id_table[] = {
+ { GREYBUS_DEVICE_CLASS(GREYBUS_CLASS_VIBRATOR) },
+ { }
+};
+MODULE_DEVICE_TABLE(greybus, gb_vibrator_id_table);
+
+static struct greybus_driver gb_vibrator_driver = {
+ .name = "vibrator",
+ .probe = gb_vibrator_probe,
+ .disconnect = gb_vibrator_disconnect,
+ .id_table = gb_vibrator_id_table,
+};
+
+static __init int gb_vibrator_init(void)
+{
+ int retval;
+
+ retval = class_register(&vibrator_class);
+ if (retval)
+ return retval;
+
+ retval = greybus_register(&gb_vibrator_driver);
+ if (retval)
+ goto err_class_unregister;
+
+ return 0;
+
+err_class_unregister:
+ class_unregister(&vibrator_class);
+
+ return retval;
+}
+module_init(gb_vibrator_init);
+
+static __exit void gb_vibrator_exit(void)
+{
+ greybus_deregister(&gb_vibrator_driver);
+ class_unregister(&vibrator_class);
+ ida_destroy(&minors);
+}
+module_exit(gb_vibrator_exit);
+
+MODULE_LICENSE("GPL v2");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 21/32] greybus: arche platform driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (18 preceding siblings ...)
2016-09-16 14:11 ` [patch 20/32] greybus: vibrator driver Greg KH
@ 2016-09-16 14:19 ` Greg KH
2016-09-16 14:20 ` [patch 22/32] greybus: bridged phy bus code Greg KH
` (10 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:19 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This is a platform driver for the arche platform controller that hooks
into the USB hub that talks to the USB<->Unipro bridge chip.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/arche-apb-ctrl.c | 522 ++++++++++++++++++++++++
drivers/greybus/arche-platform.c | 828 +++++++++++++++++++++++++++++++++++++++
drivers/greybus/arche_platform.h | 39 +
3 files changed, 1389 insertions(+)
--- /dev/null
+++ b/drivers/greybus/arche-apb-ctrl.c
@@ -0,0 +1,522 @@
+/*
+ * Arche Platform driver to control APB.
+ *
+ * Copyright 2014-2015 Google Inc.
+ * Copyright 2014-2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/clk.h>
+#include <linux/delay.h>
+#include <linux/gpio.h>
+#include <linux/interrupt.h>
+#include <linux/of_gpio.h>
+#include <linux/of_irq.h>
+#include <linux/module.h>
+#include <linux/pinctrl/consumer.h>
+#include <linux/platform_device.h>
+#include <linux/pm.h>
+#include <linux/regulator/consumer.h>
+#include <linux/spinlock.h>
+#include "arche_platform.h"
+
+
+struct arche_apb_ctrl_drvdata {
+ /* Control GPIO signals to and from AP <=> AP Bridges */
+ int resetn_gpio;
+ int boot_ret_gpio;
+ int pwroff_gpio;
+ int wake_in_gpio;
+ int wake_out_gpio;
+ int pwrdn_gpio;
+
+ enum arche_platform_state state;
+ bool init_disabled;
+
+ struct regulator *vcore;
+ struct regulator *vio;
+
+ int clk_en_gpio;
+ struct clk *clk;
+
+ struct pinctrl *pinctrl;
+ struct pinctrl_state *pin_default;
+
+ /* V2: SPI Bus control */
+ int spi_en_gpio;
+ bool spi_en_polarity_high;
+};
+
+/*
+ * Note that these low level api's are active high
+ */
+static inline void deassert_reset(unsigned int gpio)
+{
+ gpio_set_value(gpio, 1);
+}
+
+static inline void assert_reset(unsigned int gpio)
+{
+ gpio_set_value(gpio, 0);
+}
+
+/*
+ * Note: Please do not modify the below sequence, as it is as per the spec
+ */
+static int coldboot_seq(struct platform_device *pdev)
+{
+ struct device *dev = &pdev->dev;
+ struct arche_apb_ctrl_drvdata *apb = platform_get_drvdata(pdev);
+ int ret;
+
+ if (apb->init_disabled ||
+ apb->state == ARCHE_PLATFORM_STATE_ACTIVE)
+ return 0;
+
+ /* Hold APB in reset state */
+ assert_reset(apb->resetn_gpio);
+
+ if (apb->state == ARCHE_PLATFORM_STATE_FW_FLASHING &&
+ gpio_is_valid(apb->spi_en_gpio))
+ devm_gpio_free(dev, apb->spi_en_gpio);
+
+ /* Enable power to APB */
+ if (!IS_ERR(apb->vcore)) {
+ ret = regulator_enable(apb->vcore);
+ if (ret) {
+ dev_err(dev, "failed to enable core regulator\n");
+ return ret;
+ }
+ }
+
+ if (!IS_ERR(apb->vio)) {
+ ret = regulator_enable(apb->vio);
+ if (ret) {
+ dev_err(dev, "failed to enable IO regulator\n");
+ return ret;
+ }
+ }
+
+ apb_bootret_deassert(dev);
+
+ /* On DB3 clock was not mandatory */
+ if (gpio_is_valid(apb->clk_en_gpio))
+ gpio_set_value(apb->clk_en_gpio, 1);
+
+ usleep_range(100, 200);
+
+ /* deassert reset to APB : Active-low signal */
+ deassert_reset(apb->resetn_gpio);
+
+ apb->state = ARCHE_PLATFORM_STATE_ACTIVE;
+
+ return 0;
+}
+
+static int fw_flashing_seq(struct platform_device *pdev)
+{
+ struct device *dev = &pdev->dev;
+ struct arche_apb_ctrl_drvdata *apb = platform_get_drvdata(pdev);
+ int ret;
+
+ if (apb->init_disabled ||
+ apb->state == ARCHE_PLATFORM_STATE_FW_FLASHING)
+ return 0;
+
+ ret = regulator_enable(apb->vcore);
+ if (ret) {
+ dev_err(dev, "failed to enable core regulator\n");
+ return ret;
+ }
+
+ ret = regulator_enable(apb->vio);
+ if (ret) {
+ dev_err(dev, "failed to enable IO regulator\n");
+ return ret;
+ }
+
+ if (gpio_is_valid(apb->spi_en_gpio)) {
+ unsigned long flags;
+
+ if (apb->spi_en_polarity_high)
+ flags = GPIOF_OUT_INIT_HIGH;
+ else
+ flags = GPIOF_OUT_INIT_LOW;
+
+ ret = devm_gpio_request_one(dev, apb->spi_en_gpio,
+ flags, "apb_spi_en");
+ if (ret) {
+ dev_err(dev, "Failed requesting SPI bus en gpio %d\n",
+ apb->spi_en_gpio);
+ return ret;
+ }
+ }
+
+ /* for flashing device should be in reset state */
+ assert_reset(apb->resetn_gpio);
+ apb->state = ARCHE_PLATFORM_STATE_FW_FLASHING;
+
+ return 0;
+}
+
+static int standby_boot_seq(struct platform_device *pdev)
+{
+ struct device *dev = &pdev->dev;
+ struct arche_apb_ctrl_drvdata *apb = platform_get_drvdata(pdev);
+
+ if (apb->init_disabled)
+ return 0;
+
+ /* Even if it is in OFF state, then we do not want to change the state */
+ if (apb->state == ARCHE_PLATFORM_STATE_STANDBY ||
+ apb->state == ARCHE_PLATFORM_STATE_OFF)
+ return 0;
+
+ if (apb->state == ARCHE_PLATFORM_STATE_FW_FLASHING &&
+ gpio_is_valid(apb->spi_en_gpio))
+ devm_gpio_free(dev, apb->spi_en_gpio);
+
+ /*
+ * As per WDM spec, do nothing
+ *
+ * Pasted from WDM spec,
+ * - A falling edge on POWEROFF_L is detected (a)
+ * - WDM enters standby mode, but no output signals are changed
+ * */
+
+ /* TODO: POWEROFF_L is input to WDM module */
+ apb->state = ARCHE_PLATFORM_STATE_STANDBY;
+ return 0;
+}
+
+static void poweroff_seq(struct platform_device *pdev)
+{
+ struct device *dev = &pdev->dev;
+ struct arche_apb_ctrl_drvdata *apb = platform_get_drvdata(pdev);
+
+ if (apb->init_disabled || apb->state == ARCHE_PLATFORM_STATE_OFF)
+ return;
+
+ if (apb->state == ARCHE_PLATFORM_STATE_FW_FLASHING &&
+ gpio_is_valid(apb->spi_en_gpio))
+ devm_gpio_free(dev, apb->spi_en_gpio);
+
+ /* disable the clock */
+ if (gpio_is_valid(apb->clk_en_gpio))
+ gpio_set_value(apb->clk_en_gpio, 0);
+
+ if (!IS_ERR(apb->vcore) && regulator_is_enabled(apb->vcore) > 0)
+ regulator_disable(apb->vcore);
+
+ if (!IS_ERR(apb->vio) && regulator_is_enabled(apb->vio) > 0)
+ regulator_disable(apb->vio);
+
+ /* As part of exit, put APB back in reset state */
+ assert_reset(apb->resetn_gpio);
+ apb->state = ARCHE_PLATFORM_STATE_OFF;
+
+ /* TODO: May have to send an event to SVC about this exit */
+}
+
+void apb_bootret_assert(struct device *dev)
+{
+ struct arche_apb_ctrl_drvdata *apb = dev_get_drvdata(dev);
+
+ gpio_set_value(apb->boot_ret_gpio, 1);
+}
+
+void apb_bootret_deassert(struct device *dev)
+{
+ struct arche_apb_ctrl_drvdata *apb = dev_get_drvdata(dev);
+
+ gpio_set_value(apb->boot_ret_gpio, 0);
+}
+
+int apb_ctrl_coldboot(struct device *dev)
+{
+ return coldboot_seq(to_platform_device(dev));
+}
+
+int apb_ctrl_fw_flashing(struct device *dev)
+{
+ return fw_flashing_seq(to_platform_device(dev));
+}
+
+int apb_ctrl_standby_boot(struct device *dev)
+{
+ return standby_boot_seq(to_platform_device(dev));
+}
+
+void apb_ctrl_poweroff(struct device *dev)
+{
+ poweroff_seq(to_platform_device(dev));
+}
+
+static ssize_t state_store(struct device *dev,
+ struct device_attribute *attr, const char *buf, size_t count)
+{
+ struct platform_device *pdev = to_platform_device(dev);
+ struct arche_apb_ctrl_drvdata *apb = platform_get_drvdata(pdev);
+ int ret = 0;
+ bool is_disabled;
+
+ if (sysfs_streq(buf, "off")) {
+ if (apb->state == ARCHE_PLATFORM_STATE_OFF)
+ return count;
+
+ poweroff_seq(pdev);
+ } else if (sysfs_streq(buf, "active")) {
+ if (apb->state == ARCHE_PLATFORM_STATE_ACTIVE)
+ return count;
+
+ poweroff_seq(pdev);
+ is_disabled = apb->init_disabled;
+ apb->init_disabled = false;
+ ret = coldboot_seq(pdev);
+ if (ret)
+ apb->init_disabled = is_disabled;
+ } else if (sysfs_streq(buf, "standby")) {
+ if (apb->state == ARCHE_PLATFORM_STATE_STANDBY)
+ return count;
+
+ ret = standby_boot_seq(pdev);
+ } else if (sysfs_streq(buf, "fw_flashing")) {
+ if (apb->state == ARCHE_PLATFORM_STATE_FW_FLASHING)
+ return count;
+
+ /* First we want to make sure we power off everything
+ * and then enter FW flashing state */
+ poweroff_seq(pdev);
+ ret = fw_flashing_seq(pdev);
+ } else {
+ dev_err(dev, "unknown state\n");
+ ret = -EINVAL;
+ }
+
+ return ret ? ret : count;
+}
+
+static ssize_t state_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct arche_apb_ctrl_drvdata *apb = dev_get_drvdata(dev);
+
+ switch (apb->state) {
+ case ARCHE_PLATFORM_STATE_OFF:
+ return sprintf(buf, "off%s\n",
+ apb->init_disabled ? ",disabled" : "");
+ case ARCHE_PLATFORM_STATE_ACTIVE:
+ return sprintf(buf, "active\n");
+ case ARCHE_PLATFORM_STATE_STANDBY:
+ return sprintf(buf, "standby\n");
+ case ARCHE_PLATFORM_STATE_FW_FLASHING:
+ return sprintf(buf, "fw_flashing\n");
+ default:
+ return sprintf(buf, "unknown state\n");
+ }
+}
+
+static DEVICE_ATTR_RW(state);
+
+static int apb_ctrl_get_devtree_data(struct platform_device *pdev,
+ struct arche_apb_ctrl_drvdata *apb)
+{
+ struct device *dev = &pdev->dev;
+ struct device_node *np = dev->of_node;
+ int ret;
+
+ apb->resetn_gpio = of_get_named_gpio(np, "reset-gpios", 0);
+ if (apb->resetn_gpio < 0) {
+ dev_err(dev, "failed to get reset gpio\n");
+ return apb->resetn_gpio;
+ }
+ ret = devm_gpio_request_one(dev, apb->resetn_gpio,
+ GPIOF_OUT_INIT_LOW, "apb-reset");
+ if (ret) {
+ dev_err(dev, "Failed requesting reset gpio %d\n",
+ apb->resetn_gpio);
+ return ret;
+ }
+
+ apb->boot_ret_gpio = of_get_named_gpio(np, "boot-ret-gpios", 0);
+ if (apb->boot_ret_gpio < 0) {
+ dev_err(dev, "failed to get boot retention gpio\n");
+ return apb->boot_ret_gpio;
+ }
+ ret = devm_gpio_request_one(dev, apb->boot_ret_gpio,
+ GPIOF_OUT_INIT_LOW, "boot retention");
+ if (ret) {
+ dev_err(dev, "Failed requesting bootret gpio %d\n",
+ apb->boot_ret_gpio);
+ return ret;
+ }
+
+ /* It's not mandatory to support power management interface */
+ apb->pwroff_gpio = of_get_named_gpio(np, "pwr-off-gpios", 0);
+ if (apb->pwroff_gpio < 0) {
+ dev_err(dev, "failed to get power off gpio\n");
+ return apb->pwroff_gpio;
+ }
+ ret = devm_gpio_request_one(dev, apb->pwroff_gpio,
+ GPIOF_IN, "pwroff_n");
+ if (ret) {
+ dev_err(dev, "Failed requesting pwroff_n gpio %d\n",
+ apb->pwroff_gpio);
+ return ret;
+ }
+
+ /* Do not make clock mandatory as of now (for DB3) */
+ apb->clk_en_gpio = of_get_named_gpio(np, "clock-en-gpio", 0);
+ if (apb->clk_en_gpio < 0) {
+ dev_warn(dev, "failed to get clock en gpio\n");
+ } else if (gpio_is_valid(apb->clk_en_gpio)) {
+ ret = devm_gpio_request_one(dev, apb->clk_en_gpio,
+ GPIOF_OUT_INIT_LOW, "apb_clk_en");
+ if (ret) {
+ dev_warn(dev, "Failed requesting APB clock en gpio %d\n",
+ apb->clk_en_gpio);
+ return ret;
+ }
+ }
+
+ apb->pwrdn_gpio = of_get_named_gpio(np, "pwr-down-gpios", 0);
+ if (apb->pwrdn_gpio < 0)
+ dev_warn(dev, "failed to get power down gpio\n");
+
+ /* Regulators are optional, as we may have fixed supply coming in */
+ apb->vcore = devm_regulator_get(dev, "vcore");
+ if (IS_ERR(apb->vcore))
+ dev_warn(dev, "no core regulator found\n");
+
+ apb->vio = devm_regulator_get(dev, "vio");
+ if (IS_ERR(apb->vio))
+ dev_warn(dev, "no IO regulator found\n");
+
+ apb->pinctrl = devm_pinctrl_get(&pdev->dev);
+ if (IS_ERR(apb->pinctrl)) {
+ dev_err(&pdev->dev, "could not get pinctrl handle\n");
+ return PTR_ERR(apb->pinctrl);
+ }
+ apb->pin_default = pinctrl_lookup_state(apb->pinctrl, "default");
+ if (IS_ERR(apb->pin_default)) {
+ dev_err(&pdev->dev, "could not get default pin state\n");
+ return PTR_ERR(apb->pin_default);
+ }
+
+ /* Only applicable for platform >= V2 */
+ apb->spi_en_gpio = of_get_named_gpio(np, "spi-en-gpio", 0);
+ if (apb->spi_en_gpio >= 0) {
+ if (of_property_read_bool(pdev->dev.of_node,
+ "spi-en-active-high"))
+ apb->spi_en_polarity_high = true;
+ }
+
+ return 0;
+}
+
+static int arche_apb_ctrl_probe(struct platform_device *pdev)
+{
+ int ret;
+ struct arche_apb_ctrl_drvdata *apb;
+ struct device *dev = &pdev->dev;
+
+ apb = devm_kzalloc(&pdev->dev, sizeof(*apb), GFP_KERNEL);
+ if (!apb)
+ return -ENOMEM;
+
+ ret = apb_ctrl_get_devtree_data(pdev, apb);
+ if (ret) {
+ dev_err(dev, "failed to get apb devicetree data %d\n", ret);
+ return ret;
+ }
+
+ /* Initially set APB to OFF state */
+ apb->state = ARCHE_PLATFORM_STATE_OFF;
+ /* Check whether device needs to be enabled on boot */
+ if (of_property_read_bool(pdev->dev.of_node, "arche,init-disable"))
+ apb->init_disabled = true;
+
+ platform_set_drvdata(pdev, apb);
+
+ /* Create sysfs interface to allow user to change state dynamically */
+ ret = device_create_file(dev, &dev_attr_state);
+ if (ret) {
+ dev_err(dev, "failed to create state file in sysfs\n");
+ return ret;
+ }
+
+ dev_info(&pdev->dev, "Device registered successfully\n");
+ return 0;
+}
+
+static int arche_apb_ctrl_remove(struct platform_device *pdev)
+{
+ device_remove_file(&pdev->dev, &dev_attr_state);
+ poweroff_seq(pdev);
+ platform_set_drvdata(pdev, NULL);
+
+ return 0;
+}
+
+static int arche_apb_ctrl_suspend(struct device *dev)
+{
+ /*
+ * If timing profile permits, we may shutdown bridge
+ * completely
+ *
+ * TODO: sequence ??
+ *
+ * Also, need to make sure we meet precondition for unipro suspend
+ * Precondition: Definition ???
+ */
+ return 0;
+}
+
+static int arche_apb_ctrl_resume(struct device *dev)
+{
+ /*
+ * Atleast for ES2 we have to meet the delay requirement between
+ * unipro switch and AP bridge init, depending on whether bridge is in
+ * OFF state or standby state.
+ *
+ * Based on whether bridge is in standby or OFF state we may have to
+ * assert multiple signals. Please refer to WDM spec, for more info.
+ *
+ */
+ return 0;
+}
+
+static void arche_apb_ctrl_shutdown(struct platform_device *pdev)
+{
+ apb_ctrl_poweroff(&pdev->dev);
+}
+
+static SIMPLE_DEV_PM_OPS(arche_apb_ctrl_pm_ops, arche_apb_ctrl_suspend,
+ arche_apb_ctrl_resume);
+
+static struct of_device_id arche_apb_ctrl_of_match[] = {
+ { .compatible = "usbffff,2", },
+ { },
+};
+
+static struct platform_driver arche_apb_ctrl_device_driver = {
+ .probe = arche_apb_ctrl_probe,
+ .remove = arche_apb_ctrl_remove,
+ .shutdown = arche_apb_ctrl_shutdown,
+ .driver = {
+ .name = "arche-apb-ctrl",
+ .pm = &arche_apb_ctrl_pm_ops,
+ .of_match_table = arche_apb_ctrl_of_match,
+ }
+};
+
+int __init arche_apb_init(void)
+{
+ return platform_driver_register(&arche_apb_ctrl_device_driver);
+}
+
+void __exit arche_apb_exit(void)
+{
+ platform_driver_unregister(&arche_apb_ctrl_device_driver);
+}
--- /dev/null
+++ b/drivers/greybus/arche-platform.c
@@ -0,0 +1,828 @@
+/*
+ * Arche Platform driver to enable Unipro link.
+ *
+ * Copyright 2014-2015 Google Inc.
+ * Copyright 2014-2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/clk.h>
+#include <linux/delay.h>
+#include <linux/gpio.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/of_gpio.h>
+#include <linux/of_platform.h>
+#include <linux/pinctrl/consumer.h>
+#include <linux/platform_device.h>
+#include <linux/pm.h>
+#include <linux/interrupt.h>
+#include <linux/irq.h>
+#include <linux/suspend.h>
+#include <linux/time.h>
+#include "arche_platform.h"
+#include "greybus.h"
+
+#include <linux/usb/usb3613.h>
+
+#define WD_COLDBOOT_PULSE_WIDTH_MS 30
+
+enum svc_wakedetect_state {
+ WD_STATE_IDLE, /* Default state = pulled high/low */
+ WD_STATE_BOOT_INIT, /* WD = falling edge (low) */
+ WD_STATE_COLDBOOT_TRIG, /* WD = rising edge (high), > 30msec */
+ WD_STATE_STANDBYBOOT_TRIG, /* As of now not used ?? */
+ WD_STATE_COLDBOOT_START, /* Cold boot process started */
+ WD_STATE_STANDBYBOOT_START, /* Not used */
+ WD_STATE_TIMESYNC,
+};
+
+struct arche_platform_drvdata {
+ /* Control GPIO signals to and from AP <=> SVC */
+ int svc_reset_gpio;
+ bool is_reset_act_hi;
+ int svc_sysboot_gpio;
+ int wake_detect_gpio; /* bi-dir,maps to WAKE_MOD & WAKE_FRAME signals */
+
+ enum arche_platform_state state;
+
+ int svc_refclk_req;
+ struct clk *svc_ref_clk;
+
+ struct pinctrl *pinctrl;
+ struct pinctrl_state *pin_default;
+
+ int num_apbs;
+
+ enum svc_wakedetect_state wake_detect_state;
+ int wake_detect_irq;
+ spinlock_t wake_lock; /* Protect wake_detect_state */
+ struct mutex platform_state_mutex; /* Protect state */
+ wait_queue_head_t wq; /* WQ for arche_pdata->state */
+ unsigned long wake_detect_start;
+ struct notifier_block pm_notifier;
+
+ struct device *dev;
+ struct gb_timesync_svc *timesync_svc_pdata;
+};
+
+static int arche_apb_bootret_assert(struct device *dev, void *data)
+{
+ apb_bootret_assert(dev);
+ return 0;
+}
+
+static int arche_apb_bootret_deassert(struct device *dev, void *data)
+{
+ apb_bootret_deassert(dev);
+ return 0;
+}
+
+/* Requires calling context to hold arche_pdata->platform_state_mutex */
+static void arche_platform_set_state(struct arche_platform_drvdata *arche_pdata,
+ enum arche_platform_state state)
+{
+ arche_pdata->state = state;
+}
+
+/*
+ * arche_platform_change_state: Change the operational state
+ *
+ * This exported function allows external drivers to change the state
+ * of the arche-platform driver.
+ * Note that this function only supports transitions between two states
+ * with limited functionality.
+ *
+ * - ARCHE_PLATFORM_STATE_TIME_SYNC:
+ * Once set, allows timesync operations between SVC <=> AP and makes
+ * sure that arche-platform driver ignores any subsequent events/pulses
+ * from SVC over wake/detect.
+ *
+ * - ARCHE_PLATFORM_STATE_ACTIVE:
+ * Puts back driver to active state, where any pulse from SVC on wake/detect
+ * line would trigger either cold/standby boot.
+ * Note: Transition request from this function does not trigger cold/standby
+ * boot. It just puts back driver book keeping variable back to ACTIVE
+ * state and restores the interrupt.
+ *
+ * Returns -ENODEV if device not found, -EAGAIN if the driver cannot currently
+ * satisfy the requested state-transition or -EINVAL for all other
+ * state-transition requests.
+ */
+int arche_platform_change_state(enum arche_platform_state state,
+ struct gb_timesync_svc *timesync_svc_pdata)
+{
+ struct arche_platform_drvdata *arche_pdata;
+ struct platform_device *pdev;
+ struct device_node *np;
+ int ret = -EAGAIN;
+ unsigned long flags;
+
+ np = of_find_compatible_node(NULL, NULL, "google,arche-platform");
+ if (!np) {
+ pr_err("google,arche-platform device node not found\n");
+ return -ENODEV;
+ }
+
+ pdev = of_find_device_by_node(np);
+ if (!pdev) {
+ pr_err("arche-platform device not found\n");
+ return -ENODEV;
+ }
+
+ arche_pdata = platform_get_drvdata(pdev);
+
+ mutex_lock(&arche_pdata->platform_state_mutex);
+ spin_lock_irqsave(&arche_pdata->wake_lock, flags);
+
+ if (arche_pdata->state == state) {
+ ret = 0;
+ goto exit;
+ }
+
+ switch (state) {
+ case ARCHE_PLATFORM_STATE_TIME_SYNC:
+ if (arche_pdata->state != ARCHE_PLATFORM_STATE_ACTIVE) {
+ ret = -EINVAL;
+ goto exit;
+ }
+ if (arche_pdata->wake_detect_state != WD_STATE_IDLE) {
+ dev_err(arche_pdata->dev,
+ "driver busy with wake/detect line ops\n");
+ goto exit;
+ }
+ device_for_each_child(arche_pdata->dev, NULL,
+ arche_apb_bootret_assert);
+ arche_pdata->wake_detect_state = WD_STATE_TIMESYNC;
+ break;
+ case ARCHE_PLATFORM_STATE_ACTIVE:
+ if (arche_pdata->state != ARCHE_PLATFORM_STATE_TIME_SYNC) {
+ ret = -EINVAL;
+ goto exit;
+ }
+ device_for_each_child(arche_pdata->dev, NULL,
+ arche_apb_bootret_deassert);
+ arche_pdata->wake_detect_state = WD_STATE_IDLE;
+ break;
+ case ARCHE_PLATFORM_STATE_OFF:
+ case ARCHE_PLATFORM_STATE_STANDBY:
+ case ARCHE_PLATFORM_STATE_FW_FLASHING:
+ dev_err(arche_pdata->dev, "busy, request to retry later\n");
+ goto exit;
+ default:
+ ret = -EINVAL;
+ dev_err(arche_pdata->dev,
+ "invalid state transition request\n");
+ goto exit;
+ }
+ arche_pdata->timesync_svc_pdata = timesync_svc_pdata;
+ arche_platform_set_state(arche_pdata, state);
+ if (state == ARCHE_PLATFORM_STATE_ACTIVE)
+ wake_up(&arche_pdata->wq);
+
+ ret = 0;
+exit:
+ spin_unlock_irqrestore(&arche_pdata->wake_lock, flags);
+ mutex_unlock(&arche_pdata->platform_state_mutex);
+ of_node_put(np);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(arche_platform_change_state);
+
+/* Requires arche_pdata->wake_lock is held by calling context */
+static void arche_platform_set_wake_detect_state(
+ struct arche_platform_drvdata *arche_pdata,
+ enum svc_wakedetect_state state)
+{
+ arche_pdata->wake_detect_state = state;
+}
+
+static inline void svc_reset_onoff(unsigned int gpio, bool onoff)
+{
+ gpio_set_value(gpio, onoff);
+}
+
+static int apb_cold_boot(struct device *dev, void *data)
+{
+ int ret;
+
+ ret = apb_ctrl_coldboot(dev);
+ if (ret)
+ dev_warn(dev, "failed to coldboot\n");
+
+ /*Child nodes are independent, so do not exit coldboot operation */
+ return 0;
+}
+
+static int apb_poweroff(struct device *dev, void *data)
+{
+ apb_ctrl_poweroff(dev);
+
+ /* Enable HUB3613 into HUB mode. */
+ if (usb3613_hub_mode_ctrl(false))
+ dev_warn(dev, "failed to control hub device\n");
+
+ return 0;
+}
+
+static void arche_platform_wd_irq_en(struct arche_platform_drvdata *arche_pdata)
+{
+ /* Enable interrupt here, to read event back from SVC */
+ gpio_direction_input(arche_pdata->wake_detect_gpio);
+ enable_irq(arche_pdata->wake_detect_irq);
+}
+
+static irqreturn_t arche_platform_wd_irq_thread(int irq, void *devid)
+{
+ struct arche_platform_drvdata *arche_pdata = devid;
+ unsigned long flags;
+
+ spin_lock_irqsave(&arche_pdata->wake_lock, flags);
+ if (arche_pdata->wake_detect_state != WD_STATE_COLDBOOT_TRIG) {
+ /* Something is wrong */
+ spin_unlock_irqrestore(&arche_pdata->wake_lock, flags);
+ return IRQ_HANDLED;
+ }
+
+ arche_platform_set_wake_detect_state(arche_pdata,
+ WD_STATE_COLDBOOT_START);
+ spin_unlock_irqrestore(&arche_pdata->wake_lock, flags);
+
+ /* It should complete power cycle, so first make sure it is poweroff */
+ device_for_each_child(arche_pdata->dev, NULL, apb_poweroff);
+
+ /* Bring APB out of reset: cold boot sequence */
+ device_for_each_child(arche_pdata->dev, NULL, apb_cold_boot);
+
+ /* Enable HUB3613 into HUB mode. */
+ if (usb3613_hub_mode_ctrl(true))
+ dev_warn(arche_pdata->dev, "failed to control hub device\n");
+
+ spin_lock_irqsave(&arche_pdata->wake_lock, flags);
+ arche_platform_set_wake_detect_state(arche_pdata, WD_STATE_IDLE);
+ spin_unlock_irqrestore(&arche_pdata->wake_lock, flags);
+
+ return IRQ_HANDLED;
+}
+
+static irqreturn_t arche_platform_wd_irq(int irq, void *devid)
+{
+ struct arche_platform_drvdata *arche_pdata = devid;
+ unsigned long flags;
+
+ spin_lock_irqsave(&arche_pdata->wake_lock, flags);
+
+ if (arche_pdata->wake_detect_state == WD_STATE_TIMESYNC) {
+ gb_timesync_irq(arche_pdata->timesync_svc_pdata);
+ goto exit;
+ }
+
+ if (gpio_get_value(arche_pdata->wake_detect_gpio)) {
+ /* wake/detect rising */
+
+ /*
+ * If wake/detect line goes high after low, within less than
+ * 30msec, then standby boot sequence is initiated, which is not
+ * supported/implemented as of now. So ignore it.
+ */
+ if (arche_pdata->wake_detect_state == WD_STATE_BOOT_INIT) {
+ if (time_before(jiffies,
+ arche_pdata->wake_detect_start +
+ msecs_to_jiffies(WD_COLDBOOT_PULSE_WIDTH_MS))) {
+ arche_platform_set_wake_detect_state(arche_pdata,
+ WD_STATE_IDLE);
+ } else {
+ /* Check we are not in middle of irq thread already */
+ if (arche_pdata->wake_detect_state !=
+ WD_STATE_COLDBOOT_START) {
+ arche_platform_set_wake_detect_state(arche_pdata,
+ WD_STATE_COLDBOOT_TRIG);
+ spin_unlock_irqrestore(
+ &arche_pdata->wake_lock,
+ flags);
+ return IRQ_WAKE_THREAD;
+ }
+ }
+ }
+ } else {
+ /* wake/detect falling */
+ if (arche_pdata->wake_detect_state == WD_STATE_IDLE) {
+ arche_pdata->wake_detect_start = jiffies;
+ /*
+ * In the begining, when wake/detect goes low (first time), we assume
+ * it is meant for coldboot and set the flag. If wake/detect line stays low
+ * beyond 30msec, then it is coldboot else fallback to standby boot.
+ */
+ arche_platform_set_wake_detect_state(arche_pdata,
+ WD_STATE_BOOT_INIT);
+ }
+ }
+
+exit:
+ spin_unlock_irqrestore(&arche_pdata->wake_lock, flags);
+
+ return IRQ_HANDLED;
+}
+
+/*
+ * Requires arche_pdata->platform_state_mutex to be held
+ */
+static int arche_platform_coldboot_seq(struct arche_platform_drvdata *arche_pdata)
+{
+ int ret;
+
+ if (arche_pdata->state == ARCHE_PLATFORM_STATE_ACTIVE)
+ return 0;
+
+ dev_info(arche_pdata->dev, "Booting from cold boot state\n");
+
+ svc_reset_onoff(arche_pdata->svc_reset_gpio,
+ arche_pdata->is_reset_act_hi);
+
+ gpio_set_value(arche_pdata->svc_sysboot_gpio, 0);
+ usleep_range(100, 200);
+
+ ret = clk_prepare_enable(arche_pdata->svc_ref_clk);
+ if (ret) {
+ dev_err(arche_pdata->dev, "failed to enable svc_ref_clk: %d\n",
+ ret);
+ return ret;
+ }
+
+ /* bring SVC out of reset */
+ svc_reset_onoff(arche_pdata->svc_reset_gpio,
+ !arche_pdata->is_reset_act_hi);
+
+ arche_platform_set_state(arche_pdata, ARCHE_PLATFORM_STATE_ACTIVE);
+
+ return 0;
+}
+
+/*
+ * Requires arche_pdata->platform_state_mutex to be held
+ */
+static int arche_platform_fw_flashing_seq(struct arche_platform_drvdata *arche_pdata)
+{
+ int ret;
+
+ if (arche_pdata->state == ARCHE_PLATFORM_STATE_FW_FLASHING)
+ return 0;
+
+ dev_info(arche_pdata->dev, "Switching to FW flashing state\n");
+
+ svc_reset_onoff(arche_pdata->svc_reset_gpio,
+ arche_pdata->is_reset_act_hi);
+
+ gpio_set_value(arche_pdata->svc_sysboot_gpio, 1);
+
+ usleep_range(100, 200);
+
+ ret = clk_prepare_enable(arche_pdata->svc_ref_clk);
+ if (ret) {
+ dev_err(arche_pdata->dev, "failed to enable svc_ref_clk: %d\n",
+ ret);
+ return ret;
+ }
+
+ svc_reset_onoff(arche_pdata->svc_reset_gpio,
+ !arche_pdata->is_reset_act_hi);
+
+ arche_platform_set_state(arche_pdata, ARCHE_PLATFORM_STATE_FW_FLASHING);
+
+ return 0;
+}
+
+/*
+ * Requires arche_pdata->platform_state_mutex to be held
+ */
+static void arche_platform_poweroff_seq(struct arche_platform_drvdata *arche_pdata)
+{
+ unsigned long flags;
+
+ if (arche_pdata->state == ARCHE_PLATFORM_STATE_OFF)
+ return;
+
+ /* If in fw_flashing mode, then no need to repeate things again */
+ if (arche_pdata->state != ARCHE_PLATFORM_STATE_FW_FLASHING) {
+ disable_irq(arche_pdata->wake_detect_irq);
+
+ spin_lock_irqsave(&arche_pdata->wake_lock, flags);
+ arche_platform_set_wake_detect_state(arche_pdata,
+ WD_STATE_IDLE);
+ spin_unlock_irqrestore(&arche_pdata->wake_lock, flags);
+ }
+
+ clk_disable_unprepare(arche_pdata->svc_ref_clk);
+
+ /* As part of exit, put APB back in reset state */
+ svc_reset_onoff(arche_pdata->svc_reset_gpio,
+ arche_pdata->is_reset_act_hi);
+
+ arche_platform_set_state(arche_pdata, ARCHE_PLATFORM_STATE_OFF);
+}
+
+static ssize_t state_store(struct device *dev,
+ struct device_attribute *attr, const char *buf, size_t count)
+{
+ struct platform_device *pdev = to_platform_device(dev);
+ struct arche_platform_drvdata *arche_pdata = platform_get_drvdata(pdev);
+ int ret = 0;
+
+retry:
+ mutex_lock(&arche_pdata->platform_state_mutex);
+ if (arche_pdata->state == ARCHE_PLATFORM_STATE_TIME_SYNC) {
+ mutex_unlock(&arche_pdata->platform_state_mutex);
+ ret = wait_event_interruptible(
+ arche_pdata->wq,
+ arche_pdata->state != ARCHE_PLATFORM_STATE_TIME_SYNC);
+ if (ret)
+ return ret;
+ goto retry;
+ }
+
+ if (sysfs_streq(buf, "off")) {
+ if (arche_pdata->state == ARCHE_PLATFORM_STATE_OFF)
+ goto exit;
+
+ /* If SVC goes down, bring down APB's as well */
+ device_for_each_child(arche_pdata->dev, NULL, apb_poweroff);
+
+ arche_platform_poweroff_seq(arche_pdata);
+
+ } else if (sysfs_streq(buf, "active")) {
+ if (arche_pdata->state == ARCHE_PLATFORM_STATE_ACTIVE)
+ goto exit;
+
+ /* First we want to make sure we power off everything
+ * and then activate back again */
+ device_for_each_child(arche_pdata->dev, NULL, apb_poweroff);
+ arche_platform_poweroff_seq(arche_pdata);
+
+ arche_platform_wd_irq_en(arche_pdata);
+ ret = arche_platform_coldboot_seq(arche_pdata);
+ if (ret)
+ goto exit;
+
+ } else if (sysfs_streq(buf, "standby")) {
+ if (arche_pdata->state == ARCHE_PLATFORM_STATE_STANDBY)
+ goto exit;
+
+ dev_warn(arche_pdata->dev, "standby state not supported\n");
+ } else if (sysfs_streq(buf, "fw_flashing")) {
+ if (arche_pdata->state == ARCHE_PLATFORM_STATE_FW_FLASHING)
+ goto exit;
+
+ /*
+ * Here we only control SVC.
+ *
+ * In case of FW_FLASHING mode we do not want to control
+ * APBs, as in case of V2, SPI bus is shared between both
+ * the APBs. So let user chose which APB he wants to flash.
+ */
+ arche_platform_poweroff_seq(arche_pdata);
+
+ ret = arche_platform_fw_flashing_seq(arche_pdata);
+ if (ret)
+ goto exit;
+ } else {
+ dev_err(arche_pdata->dev, "unknown state\n");
+ ret = -EINVAL;
+ }
+
+exit:
+ mutex_unlock(&arche_pdata->platform_state_mutex);
+ return ret ? ret : count;
+}
+
+static ssize_t state_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct arche_platform_drvdata *arche_pdata = dev_get_drvdata(dev);
+
+ switch (arche_pdata->state) {
+ case ARCHE_PLATFORM_STATE_OFF:
+ return sprintf(buf, "off\n");
+ case ARCHE_PLATFORM_STATE_ACTIVE:
+ return sprintf(buf, "active\n");
+ case ARCHE_PLATFORM_STATE_STANDBY:
+ return sprintf(buf, "standby\n");
+ case ARCHE_PLATFORM_STATE_FW_FLASHING:
+ return sprintf(buf, "fw_flashing\n");
+ case ARCHE_PLATFORM_STATE_TIME_SYNC:
+ return sprintf(buf, "time_sync\n");
+ default:
+ return sprintf(buf, "unknown state\n");
+ }
+}
+
+static DEVICE_ATTR_RW(state);
+
+static int arche_platform_pm_notifier(struct notifier_block *notifier,
+ unsigned long pm_event, void *unused)
+{
+ struct arche_platform_drvdata *arche_pdata =
+ container_of(notifier, struct arche_platform_drvdata,
+ pm_notifier);
+ int ret = NOTIFY_DONE;
+
+ mutex_lock(&arche_pdata->platform_state_mutex);
+ switch (pm_event) {
+ case PM_SUSPEND_PREPARE:
+ if (arche_pdata->state != ARCHE_PLATFORM_STATE_ACTIVE) {
+ ret = NOTIFY_STOP;
+ break;
+ }
+ device_for_each_child(arche_pdata->dev, NULL, apb_poweroff);
+ arche_platform_poweroff_seq(arche_pdata);
+ break;
+ case PM_POST_SUSPEND:
+ if (arche_pdata->state != ARCHE_PLATFORM_STATE_OFF)
+ break;
+
+ arche_platform_wd_irq_en(arche_pdata);
+ arche_platform_coldboot_seq(arche_pdata);
+ break;
+ default:
+ break;
+ }
+ mutex_unlock(&arche_pdata->platform_state_mutex);
+
+ return ret;
+}
+
+static int arche_platform_probe(struct platform_device *pdev)
+{
+ struct arche_platform_drvdata *arche_pdata;
+ struct device *dev = &pdev->dev;
+ struct device_node *np = dev->of_node;
+ int ret;
+
+ arche_pdata = devm_kzalloc(&pdev->dev, sizeof(*arche_pdata), GFP_KERNEL);
+ if (!arche_pdata)
+ return -ENOMEM;
+
+ /* setup svc reset gpio */
+ arche_pdata->is_reset_act_hi = of_property_read_bool(np,
+ "svc,reset-active-high");
+ arche_pdata->svc_reset_gpio = of_get_named_gpio(np, "svc,reset-gpio", 0);
+ if (arche_pdata->svc_reset_gpio < 0) {
+ dev_err(dev, "failed to get reset-gpio\n");
+ return arche_pdata->svc_reset_gpio;
+ }
+ ret = devm_gpio_request(dev, arche_pdata->svc_reset_gpio, "svc-reset");
+ if (ret) {
+ dev_err(dev, "failed to request svc-reset gpio:%d\n", ret);
+ return ret;
+ }
+ ret = gpio_direction_output(arche_pdata->svc_reset_gpio,
+ arche_pdata->is_reset_act_hi);
+ if (ret) {
+ dev_err(dev, "failed to set svc-reset gpio dir:%d\n", ret);
+ return ret;
+ }
+ arche_platform_set_state(arche_pdata, ARCHE_PLATFORM_STATE_OFF);
+
+ arche_pdata->svc_sysboot_gpio = of_get_named_gpio(np,
+ "svc,sysboot-gpio", 0);
+ if (arche_pdata->svc_sysboot_gpio < 0) {
+ dev_err(dev, "failed to get sysboot gpio\n");
+ return arche_pdata->svc_sysboot_gpio;
+ }
+ ret = devm_gpio_request(dev, arche_pdata->svc_sysboot_gpio, "sysboot0");
+ if (ret) {
+ dev_err(dev, "failed to request sysboot0 gpio:%d\n", ret);
+ return ret;
+ }
+ ret = gpio_direction_output(arche_pdata->svc_sysboot_gpio, 0);
+ if (ret) {
+ dev_err(dev, "failed to set svc-reset gpio dir:%d\n", ret);
+ return ret;
+ }
+
+ /* setup the clock request gpio first */
+ arche_pdata->svc_refclk_req = of_get_named_gpio(np,
+ "svc,refclk-req-gpio", 0);
+ if (arche_pdata->svc_refclk_req < 0) {
+ dev_err(dev, "failed to get svc clock-req gpio\n");
+ return arche_pdata->svc_refclk_req;
+ }
+ ret = devm_gpio_request(dev, arche_pdata->svc_refclk_req, "svc-clk-req");
+ if (ret) {
+ dev_err(dev, "failed to request svc-clk-req gpio: %d\n", ret);
+ return ret;
+ }
+ ret = gpio_direction_input(arche_pdata->svc_refclk_req);
+ if (ret) {
+ dev_err(dev, "failed to set svc-clk-req gpio dir :%d\n", ret);
+ return ret;
+ }
+
+ /* setup refclk2 to follow the pin */
+ arche_pdata->svc_ref_clk = devm_clk_get(dev, "svc_ref_clk");
+ if (IS_ERR(arche_pdata->svc_ref_clk)) {
+ ret = PTR_ERR(arche_pdata->svc_ref_clk);
+ dev_err(dev, "failed to get svc_ref_clk: %d\n", ret);
+ return ret;
+ }
+
+ platform_set_drvdata(pdev, arche_pdata);
+
+ arche_pdata->num_apbs = of_get_child_count(np);
+ dev_dbg(dev, "Number of APB's available - %d\n", arche_pdata->num_apbs);
+
+ arche_pdata->wake_detect_gpio = of_get_named_gpio(np, "svc,wake-detect-gpio", 0);
+ if (arche_pdata->wake_detect_gpio < 0) {
+ dev_err(dev, "failed to get wake detect gpio\n");
+ ret = arche_pdata->wake_detect_gpio;
+ return ret;
+ }
+
+ ret = devm_gpio_request(dev, arche_pdata->wake_detect_gpio, "wake detect");
+ if (ret) {
+ dev_err(dev, "Failed requesting wake_detect gpio %d\n",
+ arche_pdata->wake_detect_gpio);
+ return ret;
+ }
+
+ arche_platform_set_wake_detect_state(arche_pdata, WD_STATE_IDLE);
+
+ arche_pdata->dev = &pdev->dev;
+
+ spin_lock_init(&arche_pdata->wake_lock);
+ mutex_init(&arche_pdata->platform_state_mutex);
+ init_waitqueue_head(&arche_pdata->wq);
+ arche_pdata->wake_detect_irq =
+ gpio_to_irq(arche_pdata->wake_detect_gpio);
+
+ ret = devm_request_threaded_irq(dev, arche_pdata->wake_detect_irq,
+ arche_platform_wd_irq,
+ arche_platform_wd_irq_thread,
+ IRQF_TRIGGER_FALLING | IRQF_TRIGGER_RISING | IRQF_ONESHOT,
+ dev_name(dev), arche_pdata);
+ if (ret) {
+ dev_err(dev, "failed to request wake detect IRQ %d\n", ret);
+ return ret;
+ }
+ disable_irq(arche_pdata->wake_detect_irq);
+
+ ret = device_create_file(dev, &dev_attr_state);
+ if (ret) {
+ dev_err(dev, "failed to create state file in sysfs\n");
+ return ret;
+ }
+
+ ret = of_platform_populate(np, NULL, NULL, dev);
+ if (ret) {
+ dev_err(dev, "failed to populate child nodes %d\n", ret);
+ goto err_device_remove;
+ }
+
+ arche_pdata->pm_notifier.notifier_call = arche_platform_pm_notifier;
+ ret = register_pm_notifier(&arche_pdata->pm_notifier);
+
+ if (ret) {
+ dev_err(dev, "failed to register pm notifier %d\n", ret);
+ goto err_device_remove;
+ }
+
+ /* Register callback pointer */
+ arche_platform_change_state_cb = arche_platform_change_state;
+
+ /* Explicitly power off if requested */
+ if (!of_property_read_bool(pdev->dev.of_node, "arche,init-off")) {
+ mutex_lock(&arche_pdata->platform_state_mutex);
+ ret = arche_platform_coldboot_seq(arche_pdata);
+ if (ret) {
+ dev_err(dev, "Failed to cold boot svc %d\n", ret);
+ goto err_coldboot;
+ }
+ arche_platform_wd_irq_en(arche_pdata);
+ mutex_unlock(&arche_pdata->platform_state_mutex);
+ }
+
+ dev_info(dev, "Device registered successfully\n");
+ return 0;
+
+err_coldboot:
+ mutex_unlock(&arche_pdata->platform_state_mutex);
+err_device_remove:
+ device_remove_file(&pdev->dev, &dev_attr_state);
+ return ret;
+}
+
+static int arche_remove_child(struct device *dev, void *unused)
+{
+ struct platform_device *pdev = to_platform_device(dev);
+
+ platform_device_unregister(pdev);
+
+ return 0;
+}
+
+static int arche_platform_remove(struct platform_device *pdev)
+{
+ struct arche_platform_drvdata *arche_pdata = platform_get_drvdata(pdev);
+
+ unregister_pm_notifier(&arche_pdata->pm_notifier);
+ device_remove_file(&pdev->dev, &dev_attr_state);
+ device_for_each_child(&pdev->dev, NULL, arche_remove_child);
+ arche_platform_poweroff_seq(arche_pdata);
+ platform_set_drvdata(pdev, NULL);
+
+ if (usb3613_hub_mode_ctrl(false))
+ dev_warn(arche_pdata->dev, "failed to control hub device\n");
+ /* TODO: Should we do anything more here ?? */
+ return 0;
+}
+
+static int arche_platform_suspend(struct device *dev)
+{
+ /*
+ * If timing profile premits, we may shutdown bridge
+ * completely
+ *
+ * TODO: sequence ??
+ *
+ * Also, need to make sure we meet precondition for unipro suspend
+ * Precondition: Definition ???
+ */
+ return 0;
+}
+
+static int arche_platform_resume(struct device *dev)
+{
+ /*
+ * Atleast for ES2 we have to meet the delay requirement between
+ * unipro switch and AP bridge init, depending on whether bridge is in
+ * OFF state or standby state.
+ *
+ * Based on whether bridge is in standby or OFF state we may have to
+ * assert multiple signals. Please refer to WDM spec, for more info.
+ *
+ */
+ return 0;
+}
+
+static void arche_platform_shutdown(struct platform_device *pdev)
+{
+ struct arche_platform_drvdata *arche_pdata = platform_get_drvdata(pdev);
+
+ arche_platform_poweroff_seq(arche_pdata);
+
+ usb3613_hub_mode_ctrl(false);
+}
+
+static SIMPLE_DEV_PM_OPS(arche_platform_pm_ops,
+ arche_platform_suspend,
+ arche_platform_resume);
+
+static struct of_device_id arche_platform_of_match[] = {
+ { .compatible = "google,arche-platform", }, /* Use PID/VID of SVC device */
+ { },
+};
+
+static struct of_device_id arche_combined_id[] = {
+ { .compatible = "google,arche-platform", }, /* Use PID/VID of SVC device */
+ { .compatible = "usbffff,2", },
+ { },
+};
+MODULE_DEVICE_TABLE(of, arche_combined_id);
+
+static struct platform_driver arche_platform_device_driver = {
+ .probe = arche_platform_probe,
+ .remove = arche_platform_remove,
+ .shutdown = arche_platform_shutdown,
+ .driver = {
+ .name = "arche-platform-ctrl",
+ .pm = &arche_platform_pm_ops,
+ .of_match_table = arche_platform_of_match,
+ }
+};
+
+static int __init arche_init(void)
+{
+ int retval;
+
+ retval = platform_driver_register(&arche_platform_device_driver);
+ if (retval)
+ return retval;
+
+ retval = arche_apb_init();
+ if (retval)
+ platform_driver_unregister(&arche_platform_device_driver);
+
+ return retval;
+}
+module_init(arche_init);
+
+static void __exit arche_exit(void)
+{
+ arche_apb_exit();
+ platform_driver_unregister(&arche_platform_device_driver);
+}
+module_exit(arche_exit);
+
+MODULE_LICENSE("GPL v2");
+MODULE_AUTHOR("Vaibhav Hiremath <vaibhav.hiremath@linaro.org>");
+MODULE_DESCRIPTION("Arche Platform Driver");
--- /dev/null
+++ b/drivers/greybus/arche_platform.h
@@ -0,0 +1,39 @@
+/*
+ * Arche Platform driver to enable Unipro link.
+ *
+ * Copyright 2015-2016 Google Inc.
+ * Copyright 2015-2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __ARCHE_PLATFORM_H
+#define __ARCHE_PLATFORM_H
+
+#include "timesync.h"
+
+enum arche_platform_state {
+ ARCHE_PLATFORM_STATE_OFF,
+ ARCHE_PLATFORM_STATE_ACTIVE,
+ ARCHE_PLATFORM_STATE_STANDBY,
+ ARCHE_PLATFORM_STATE_FW_FLASHING,
+ ARCHE_PLATFORM_STATE_TIME_SYNC,
+};
+
+int arche_platform_change_state(enum arche_platform_state state,
+ struct gb_timesync_svc *pdata);
+
+extern int (*arche_platform_change_state_cb)(enum arche_platform_state state,
+ struct gb_timesync_svc *pdata);
+int __init arche_apb_init(void);
+void __exit arche_apb_exit(void);
+
+/* Operational states for the APB device */
+int apb_ctrl_coldboot(struct device *dev);
+int apb_ctrl_fw_flashing(struct device *dev);
+int apb_ctrl_standby_boot(struct device *dev);
+void apb_ctrl_poweroff(struct device *dev);
+void apb_bootret_assert(struct device *dev);
+void apb_bootret_deassert(struct device *dev);
+
+#endif /* __ARCHE_PLATFORM_H */
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 22/32] greybus: bridged phy bus code
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (19 preceding siblings ...)
2016-09-16 14:19 ` [patch 21/32] greybus: arche platform driver Greg KH
@ 2016-09-16 14:20 ` Greg KH
2016-09-16 14:20 ` [patch 23/32] greybus: bridged phy gpio driver Greg KH
` (9 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:20 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver is for the Greybus Bridged Phy class. It implements a bus
where all bridged phy devices hang off of, and provides the framework
for the Greybus Bridged phy drivers to plug into.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/gbphy.c | 360 ++++++++++++++++++++++++++++++++++++++++++++++++
drivers/greybus/gbphy.h | 110 ++++++++++++++
2 files changed, 470 insertions(+)
--- /dev/null
+++ b/drivers/greybus/gbphy.c
@@ -0,0 +1,360 @@
+/*
+ * Greybus Bridged-Phy Bus driver
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/types.h>
+#include <linux/module.h>
+#include <linux/moduleparam.h>
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/device.h>
+
+#include "greybus.h"
+#include "gbphy.h"
+
+#define GB_GBPHY_AUTOSUSPEND_MS 3000
+
+struct gbphy_host {
+ struct gb_bundle *bundle;
+ struct list_head devices;
+};
+
+static DEFINE_IDA(gbphy_id);
+
+static ssize_t protocol_id_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct gbphy_device *gbphy_dev = to_gbphy_dev(dev);
+
+ return sprintf(buf, "0x%02x\n", gbphy_dev->cport_desc->protocol_id);
+}
+static DEVICE_ATTR_RO(protocol_id);
+
+static struct attribute *gbphy_dev_attrs[] = {
+ &dev_attr_protocol_id.attr,
+ NULL,
+};
+
+ATTRIBUTE_GROUPS(gbphy_dev);
+
+static void gbphy_dev_release(struct device *dev)
+{
+ struct gbphy_device *gbphy_dev = to_gbphy_dev(dev);
+
+ ida_simple_remove(&gbphy_id, gbphy_dev->id);
+ kfree(gbphy_dev);
+}
+
+#ifdef CONFIG_PM
+static int gb_gbphy_idle(struct device *dev)
+{
+ pm_runtime_mark_last_busy(dev);
+ pm_request_autosuspend(dev);
+ return 0;
+}
+#endif
+
+static const struct dev_pm_ops gb_gbphy_pm_ops = {
+ SET_RUNTIME_PM_OPS(pm_generic_runtime_suspend,
+ pm_generic_runtime_resume,
+ gb_gbphy_idle)
+};
+
+static struct device_type greybus_gbphy_dev_type = {
+ .name = "gbphy_device",
+ .release = gbphy_dev_release,
+ .pm = &gb_gbphy_pm_ops,
+};
+
+static int gbphy_dev_uevent(struct device *dev, struct kobj_uevent_env *env)
+{
+ struct gbphy_device *gbphy_dev = to_gbphy_dev(dev);
+ struct greybus_descriptor_cport *cport_desc = gbphy_dev->cport_desc;
+ struct gb_bundle *bundle = gbphy_dev->bundle;
+ struct gb_interface *intf = bundle->intf;
+ struct gb_module *module = intf->module;
+ struct gb_host_device *hd = intf->hd;
+
+ if (add_uevent_var(env, "BUS=%u", hd->bus_id))
+ return -ENOMEM;
+ if (add_uevent_var(env, "MODULE=%u", module->module_id))
+ return -ENOMEM;
+ if (add_uevent_var(env, "INTERFACE=%u", intf->interface_id))
+ return -ENOMEM;
+ if (add_uevent_var(env, "GREYBUS_ID=%08x/%08x",
+ intf->vendor_id, intf->product_id))
+ return -ENOMEM;
+ if (add_uevent_var(env, "BUNDLE=%u", gbphy_dev->bundle->id))
+ return -ENOMEM;
+ if (add_uevent_var(env, "BUNDLE_CLASS=%02x", bundle->class))
+ return -ENOMEM;
+ if (add_uevent_var(env, "GBPHY=%u", gbphy_dev->id))
+ return -ENOMEM;
+ if (add_uevent_var(env, "PROTOCOL_ID=%02x", cport_desc->protocol_id))
+ return -ENOMEM;
+
+ return 0;
+}
+
+static const struct gbphy_device_id *
+gbphy_dev_match_id(struct gbphy_device *gbphy_dev, struct gbphy_driver *gbphy_drv)
+{
+ const struct gbphy_device_id *id = gbphy_drv->id_table;
+
+ if (!id)
+ return NULL;
+
+ for (; id->protocol_id; id++)
+ if (id->protocol_id == gbphy_dev->cport_desc->protocol_id)
+ return id;
+
+ return NULL;
+}
+
+static int gbphy_dev_match(struct device *dev, struct device_driver *drv)
+{
+ struct gbphy_driver *gbphy_drv = to_gbphy_driver(drv);
+ struct gbphy_device *gbphy_dev = to_gbphy_dev(dev);
+ const struct gbphy_device_id *id;
+
+ id = gbphy_dev_match_id(gbphy_dev, gbphy_drv);
+ if (id)
+ return 1;
+
+ return 0;
+}
+
+static int gbphy_dev_probe(struct device *dev)
+{
+ struct gbphy_driver *gbphy_drv = to_gbphy_driver(dev->driver);
+ struct gbphy_device *gbphy_dev = to_gbphy_dev(dev);
+ const struct gbphy_device_id *id;
+ int ret;
+
+ id = gbphy_dev_match_id(gbphy_dev, gbphy_drv);
+ if (!id)
+ return -ENODEV;
+
+ /* for old kernels we need get_sync to resume parent devices */
+ ret = gb_pm_runtime_get_sync(gbphy_dev->bundle);
+ if (ret < 0)
+ return ret;
+
+ pm_runtime_set_autosuspend_delay(dev, GB_GBPHY_AUTOSUSPEND_MS);
+ pm_runtime_use_autosuspend(dev);
+ pm_runtime_get_noresume(dev);
+ pm_runtime_set_active(dev);
+ pm_runtime_enable(dev);
+
+ /*
+ * Drivers should call put on the gbphy dev before returning
+ * from probe if they support runtime pm.
+ */
+ ret = gbphy_drv->probe(gbphy_dev, id);
+ if (ret) {
+ pm_runtime_disable(dev);
+ pm_runtime_set_suspended(dev);
+ pm_runtime_put_noidle(dev);
+ pm_runtime_dont_use_autosuspend(dev);
+ }
+
+ gb_pm_runtime_put_autosuspend(gbphy_dev->bundle);
+
+ return ret;
+}
+
+static int gbphy_dev_remove(struct device *dev)
+{
+ struct gbphy_driver *gbphy_drv = to_gbphy_driver(dev->driver);
+ struct gbphy_device *gbphy_dev = to_gbphy_dev(dev);
+
+ gbphy_drv->remove(gbphy_dev);
+
+ pm_runtime_disable(dev);
+ pm_runtime_set_suspended(dev);
+ pm_runtime_put_noidle(dev);
+ pm_runtime_dont_use_autosuspend(dev);
+
+ return 0;
+}
+
+static struct bus_type gbphy_bus_type = {
+ .name = "gbphy",
+ .match = gbphy_dev_match,
+ .probe = gbphy_dev_probe,
+ .remove = gbphy_dev_remove,
+ .uevent = gbphy_dev_uevent,
+};
+
+int gb_gbphy_register_driver(struct gbphy_driver *driver,
+ struct module *owner, const char *mod_name)
+{
+ int retval;
+
+ if (greybus_disabled())
+ return -ENODEV;
+
+ driver->driver.bus = &gbphy_bus_type;
+ driver->driver.name = driver->name;
+ driver->driver.owner = owner;
+ driver->driver.mod_name = mod_name;
+
+ retval = driver_register(&driver->driver);
+ if (retval)
+ return retval;
+
+ pr_info("registered new driver %s\n", driver->name);
+ return 0;
+}
+EXPORT_SYMBOL_GPL(gb_gbphy_register_driver);
+
+void gb_gbphy_deregister_driver(struct gbphy_driver *driver)
+{
+ driver_unregister(&driver->driver);
+}
+EXPORT_SYMBOL_GPL(gb_gbphy_deregister_driver);
+
+static struct gbphy_device *gb_gbphy_create_dev(struct gb_bundle *bundle,
+ struct greybus_descriptor_cport *cport_desc)
+{
+ struct gbphy_device *gbphy_dev;
+ int retval;
+ int id;
+
+ id = ida_simple_get(&gbphy_id, 1, 0, GFP_KERNEL);
+ if (id < 0)
+ return ERR_PTR(id);
+
+ gbphy_dev = kzalloc(sizeof(*gbphy_dev), GFP_KERNEL);
+ if (!gbphy_dev) {
+ ida_simple_remove(&gbphy_id, id);
+ return ERR_PTR(-ENOMEM);
+ }
+
+ gbphy_dev->id = id;
+ gbphy_dev->bundle = bundle;
+ gbphy_dev->cport_desc = cport_desc;
+ gbphy_dev->dev.parent = &bundle->dev;
+ gbphy_dev->dev.bus = &gbphy_bus_type;
+ gbphy_dev->dev.type = &greybus_gbphy_dev_type;
+ gbphy_dev->dev.groups = gbphy_dev_groups;
+ gbphy_dev->dev.dma_mask = bundle->dev.dma_mask;
+ dev_set_name(&gbphy_dev->dev, "gbphy%d", id);
+
+ retval = device_register(&gbphy_dev->dev);
+ if (retval) {
+ put_device(&gbphy_dev->dev);
+ return ERR_PTR(retval);
+ }
+
+ return gbphy_dev;
+}
+
+static void gb_gbphy_disconnect(struct gb_bundle *bundle)
+{
+ struct gbphy_host *gbphy_host = greybus_get_drvdata(bundle);
+ struct gbphy_device *gbphy_dev, *temp;
+ int ret;
+
+ ret = gb_pm_runtime_get_sync(bundle);
+ if (ret < 0)
+ gb_pm_runtime_get_noresume(bundle);
+
+ list_for_each_entry_safe(gbphy_dev, temp, &gbphy_host->devices, list) {
+ list_del(&gbphy_dev->list);
+ device_unregister(&gbphy_dev->dev);
+ }
+
+ kfree(gbphy_host);
+}
+
+static int gb_gbphy_probe(struct gb_bundle *bundle,
+ const struct greybus_bundle_id *id)
+{
+ struct gbphy_host *gbphy_host;
+ struct gbphy_device *gbphy_dev;
+ int i;
+
+ if (bundle->num_cports == 0)
+ return -ENODEV;
+
+ gbphy_host = kzalloc(sizeof(*gbphy_host), GFP_KERNEL);
+ if (!gbphy_host)
+ return -ENOMEM;
+
+ gbphy_host->bundle = bundle;
+ INIT_LIST_HEAD(&gbphy_host->devices);
+ greybus_set_drvdata(bundle, gbphy_host);
+
+ /*
+ * Create a bunch of children devices, one per cport, and bind the
+ * bridged phy drivers to them.
+ */
+ for (i = 0; i < bundle->num_cports; ++i) {
+ gbphy_dev = gb_gbphy_create_dev(bundle, &bundle->cport_desc[i]);
+ if (IS_ERR(gbphy_dev)) {
+ gb_gbphy_disconnect(bundle);
+ return PTR_ERR(gbphy_dev);
+ }
+ list_add(&gbphy_dev->list, &gbphy_host->devices);
+ }
+
+ gb_pm_runtime_put_autosuspend(bundle);
+
+ return 0;
+}
+
+static const struct greybus_bundle_id gb_gbphy_id_table[] = {
+ { GREYBUS_DEVICE_CLASS(GREYBUS_CLASS_BRIDGED_PHY) },
+ { },
+};
+MODULE_DEVICE_TABLE(greybus, gb_gbphy_id_table);
+
+static struct greybus_driver gb_gbphy_driver = {
+ .name = "gbphy",
+ .probe = gb_gbphy_probe,
+ .disconnect = gb_gbphy_disconnect,
+ .id_table = gb_gbphy_id_table,
+};
+
+static int __init gbphy_init(void)
+{
+ int retval;
+
+ retval = bus_register(&gbphy_bus_type);
+ if (retval) {
+ pr_err("gbphy bus register failed (%d)\n", retval);
+ return retval;
+ }
+
+ retval = greybus_register(&gb_gbphy_driver);
+ if (retval) {
+ pr_err("error registering greybus driver\n");
+ goto error_gbphy;
+ }
+
+ return 0;
+
+error_gbphy:
+ bus_unregister(&gbphy_bus_type);
+ ida_destroy(&gbphy_id);
+ return retval;
+}
+module_init(gbphy_init);
+
+static void __exit gbphy_exit(void)
+{
+ greybus_deregister(&gb_gbphy_driver);
+ bus_unregister(&gbphy_bus_type);
+ ida_destroy(&gbphy_id);
+}
+module_exit(gbphy_exit);
+
+MODULE_LICENSE("GPL v2");
--- /dev/null
+++ b/drivers/greybus/gbphy.h
@@ -0,0 +1,110 @@
+/*
+ * Greybus Bridged-Phy Bus driver
+ *
+ * Copyright 2016 Google Inc.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#ifndef __GBPHY_H
+#define __GBPHY_H
+
+struct gbphy_device {
+ u32 id;
+ struct greybus_descriptor_cport *cport_desc;
+ struct gb_bundle *bundle;
+ struct list_head list;
+ struct device dev;
+};
+#define to_gbphy_dev(d) container_of(d, struct gbphy_device, dev)
+
+static inline void *gb_gbphy_get_data(struct gbphy_device *gdev)
+{
+ return dev_get_drvdata(&gdev->dev);
+}
+
+static inline void gb_gbphy_set_data(struct gbphy_device *gdev, void *data)
+{
+ dev_set_drvdata(&gdev->dev, data);
+}
+
+struct gbphy_device_id {
+ __u8 protocol_id;
+};
+
+#define GBPHY_PROTOCOL(p) \
+ .protocol_id = (p),
+
+struct gbphy_driver {
+ const char *name;
+ int (*probe)(struct gbphy_device *,
+ const struct gbphy_device_id *id);
+ void (*remove)(struct gbphy_device *);
+ const struct gbphy_device_id *id_table;
+
+ struct device_driver driver;
+};
+#define to_gbphy_driver(d) container_of(d, struct gbphy_driver, driver)
+
+int gb_gbphy_register_driver(struct gbphy_driver *driver,
+ struct module *owner, const char *mod_name);
+void gb_gbphy_deregister_driver(struct gbphy_driver *driver);
+
+#define gb_gbphy_register(driver) \
+ gb_gbphy_register_driver(driver, THIS_MODULE, KBUILD_MODNAME)
+#define gb_gbphy_deregister(driver) \
+ gb_gbphy_deregister_driver(driver)
+
+/**
+ * module_gbphy_driver() - Helper macro for registering a gbphy driver
+ * @__gbphy_driver: gbphy_driver structure
+ *
+ * Helper macro for gbphy drivers to set up proper module init / exit
+ * functions. Replaces module_init() and module_exit() and keeps people from
+ * printing pointless things to the kernel log when their driver is loaded.
+ */
+#define module_gbphy_driver(__gbphy_driver) \
+ module_driver(__gbphy_driver, gb_gbphy_register, gb_gbphy_deregister)
+
+#ifdef CONFIG_PM
+static inline int gbphy_runtime_get_sync(struct gbphy_device *gbphy_dev)
+{
+ struct device *dev = &gbphy_dev->dev;
+ int ret;
+
+ ret = pm_runtime_get_sync(dev);
+ if (ret < 0) {
+ dev_err(dev, "pm_runtime_get_sync failed: %d\n", ret);
+ pm_runtime_put_noidle(dev);
+ return ret;
+ }
+
+ return 0;
+}
+
+static inline void gbphy_runtime_put_autosuspend(struct gbphy_device *gbphy_dev)
+{
+ struct device *dev = &gbphy_dev->dev;
+
+ pm_runtime_mark_last_busy(dev);
+ pm_runtime_put_autosuspend(dev);
+}
+
+static inline void gbphy_runtime_get_noresume(struct gbphy_device *gbphy_dev)
+{
+ pm_runtime_get_noresume(&gbphy_dev->dev);
+}
+
+static inline void gbphy_runtime_put_noidle(struct gbphy_device *gbphy_dev)
+{
+ pm_runtime_put_noidle(&gbphy_dev->dev);
+}
+#else
+static inline int gbphy_runtime_get_sync(struct gbphy_device *gbphy_dev) { return 0; }
+static inline void gbphy_runtime_put_autosuspend(struct gbphy_device *gbphy_dev) {}
+static inline void gbphy_runtime_get_noresume(struct gbphy_device *gbphy_dev) {}
+static inline void gbphy_runtime_put_noidle(struct gbphy_device *gbphy_dev) {}
+#endif
+
+#endif /* __GBPHY_H */
+
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 23/32] greybus: bridged phy gpio driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (20 preceding siblings ...)
2016-09-16 14:20 ` [patch 22/32] greybus: bridged phy bus code Greg KH
@ 2016-09-16 14:20 ` Greg KH
2016-09-16 14:20 ` [patch 24/32] greybus: bridged phy i2c driver Greg KH
` (8 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:20 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus bridged phy gpio class protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/gpio.c | 767 +++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 767 insertions(+)
--- /dev/null
+++ b/drivers/greybus/gpio.c
@@ -0,0 +1,767 @@
+/*
+ * GPIO Greybus driver.
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/gpio.h>
+#include <linux/irq.h>
+#include <linux/irqdomain.h>
+#include <linux/mutex.h>
+
+#include "greybus.h"
+#include "gbphy.h"
+
+struct gb_gpio_line {
+ /* The following has to be an array of line_max entries */
+ /* --> make them just a flags field */
+ u8 active: 1,
+ direction: 1, /* 0 = output, 1 = input */
+ value: 1; /* 0 = low, 1 = high */
+ u16 debounce_usec;
+
+ u8 irq_type;
+ bool irq_type_pending;
+ bool masked;
+ bool masked_pending;
+};
+
+struct gb_gpio_controller {
+ struct gbphy_device *gbphy_dev;
+ struct gb_connection *connection;
+ u8 line_max; /* max line number */
+ struct gb_gpio_line *lines;
+
+ struct gpio_chip chip;
+ struct irq_chip irqc;
+ struct irq_chip *irqchip;
+ struct irq_domain *irqdomain;
+ unsigned int irq_base;
+ irq_flow_handler_t irq_handler;
+ unsigned int irq_default_type;
+ struct mutex irq_lock;
+};
+#define gpio_chip_to_gb_gpio_controller(chip) \
+ container_of(chip, struct gb_gpio_controller, chip)
+#define irq_data_to_gpio_chip(d) (d->domain->host_data)
+
+static int gb_gpio_line_count_operation(struct gb_gpio_controller *ggc)
+{
+ struct gb_gpio_line_count_response response;
+ int ret;
+
+ ret = gb_operation_sync(ggc->connection, GB_GPIO_TYPE_LINE_COUNT,
+ NULL, 0, &response, sizeof(response));
+ if (!ret)
+ ggc->line_max = response.count;
+ return ret;
+}
+
+static int gb_gpio_activate_operation(struct gb_gpio_controller *ggc, u8 which)
+{
+ struct gb_gpio_activate_request request;
+ struct gbphy_device *gbphy_dev = ggc->gbphy_dev;
+ int ret;
+
+ ret = gbphy_runtime_get_sync(gbphy_dev);
+ if (ret)
+ return ret;
+
+ request.which = which;
+ ret = gb_operation_sync(ggc->connection, GB_GPIO_TYPE_ACTIVATE,
+ &request, sizeof(request), NULL, 0);
+ if (ret) {
+ gbphy_runtime_put_autosuspend(gbphy_dev);
+ return ret;
+ }
+
+ ggc->lines[which].active = true;
+
+ return 0;
+}
+
+static void gb_gpio_deactivate_operation(struct gb_gpio_controller *ggc,
+ u8 which)
+{
+ struct gbphy_device *gbphy_dev = ggc->gbphy_dev;
+ struct device *dev = &gbphy_dev->dev;
+ struct gb_gpio_deactivate_request request;
+ int ret;
+
+ request.which = which;
+ ret = gb_operation_sync(ggc->connection, GB_GPIO_TYPE_DEACTIVATE,
+ &request, sizeof(request), NULL, 0);
+ if (ret) {
+ dev_err(dev, "failed to deactivate gpio %u\n", which);
+ goto out_pm_put;
+ }
+
+ ggc->lines[which].active = false;
+
+out_pm_put:
+ gbphy_runtime_put_autosuspend(gbphy_dev);
+}
+
+static int gb_gpio_get_direction_operation(struct gb_gpio_controller *ggc,
+ u8 which)
+{
+ struct device *dev = &ggc->gbphy_dev->dev;
+ struct gb_gpio_get_direction_request request;
+ struct gb_gpio_get_direction_response response;
+ int ret;
+ u8 direction;
+
+ request.which = which;
+ ret = gb_operation_sync(ggc->connection, GB_GPIO_TYPE_GET_DIRECTION,
+ &request, sizeof(request),
+ &response, sizeof(response));
+ if (ret)
+ return ret;
+
+ direction = response.direction;
+ if (direction && direction != 1) {
+ dev_warn(dev, "gpio %u direction was %u (should be 0 or 1)\n",
+ which, direction);
+ }
+ ggc->lines[which].direction = direction ? 1 : 0;
+ return 0;
+}
+
+static int gb_gpio_direction_in_operation(struct gb_gpio_controller *ggc,
+ u8 which)
+{
+ struct gb_gpio_direction_in_request request;
+ int ret;
+
+ request.which = which;
+ ret = gb_operation_sync(ggc->connection, GB_GPIO_TYPE_DIRECTION_IN,
+ &request, sizeof(request), NULL, 0);
+ if (!ret)
+ ggc->lines[which].direction = 1;
+ return ret;
+}
+
+static int gb_gpio_direction_out_operation(struct gb_gpio_controller *ggc,
+ u8 which, bool value_high)
+{
+ struct gb_gpio_direction_out_request request;
+ int ret;
+
+ request.which = which;
+ request.value = value_high ? 1 : 0;
+ ret = gb_operation_sync(ggc->connection, GB_GPIO_TYPE_DIRECTION_OUT,
+ &request, sizeof(request), NULL, 0);
+ if (!ret)
+ ggc->lines[which].direction = 0;
+ return ret;
+}
+
+static int gb_gpio_get_value_operation(struct gb_gpio_controller *ggc,
+ u8 which)
+{
+ struct device *dev = &ggc->gbphy_dev->dev;
+ struct gb_gpio_get_value_request request;
+ struct gb_gpio_get_value_response response;
+ int ret;
+ u8 value;
+
+ request.which = which;
+ ret = gb_operation_sync(ggc->connection, GB_GPIO_TYPE_GET_VALUE,
+ &request, sizeof(request),
+ &response, sizeof(response));
+ if (ret) {
+ dev_err(dev, "failed to get value of gpio %u\n", which);
+ return ret;
+ }
+
+ value = response.value;
+ if (value && value != 1) {
+ dev_warn(dev, "gpio %u value was %u (should be 0 or 1)\n",
+ which, value);
+ }
+ ggc->lines[which].value = value ? 1 : 0;
+ return 0;
+}
+
+static void gb_gpio_set_value_operation(struct gb_gpio_controller *ggc,
+ u8 which, bool value_high)
+{
+ struct device *dev = &ggc->gbphy_dev->dev;
+ struct gb_gpio_set_value_request request;
+ int ret;
+
+ if (ggc->lines[which].direction == 1) {
+ dev_warn(dev, "refusing to set value of input gpio %u\n",
+ which);
+ return;
+ }
+
+ request.which = which;
+ request.value = value_high ? 1 : 0;
+ ret = gb_operation_sync(ggc->connection, GB_GPIO_TYPE_SET_VALUE,
+ &request, sizeof(request), NULL, 0);
+ if (ret) {
+ dev_err(dev, "failed to set value of gpio %u\n", which);
+ return;
+ }
+
+ ggc->lines[which].value = request.value;
+}
+
+static int gb_gpio_set_debounce_operation(struct gb_gpio_controller *ggc,
+ u8 which, u16 debounce_usec)
+{
+ struct gb_gpio_set_debounce_request request;
+ int ret;
+
+ request.which = which;
+ request.usec = cpu_to_le16(debounce_usec);
+ ret = gb_operation_sync(ggc->connection, GB_GPIO_TYPE_SET_DEBOUNCE,
+ &request, sizeof(request), NULL, 0);
+ if (!ret)
+ ggc->lines[which].debounce_usec = debounce_usec;
+ return ret;
+}
+
+static void _gb_gpio_irq_mask(struct gb_gpio_controller *ggc, u8 hwirq)
+{
+ struct device *dev = &ggc->gbphy_dev->dev;
+ struct gb_gpio_irq_mask_request request;
+ int ret;
+
+ request.which = hwirq;
+ ret = gb_operation_sync(ggc->connection,
+ GB_GPIO_TYPE_IRQ_MASK,
+ &request, sizeof(request), NULL, 0);
+ if (ret)
+ dev_err(dev, "failed to mask irq: %d\n", ret);
+}
+
+static void _gb_gpio_irq_unmask(struct gb_gpio_controller *ggc, u8 hwirq)
+{
+ struct device *dev = &ggc->gbphy_dev->dev;
+ struct gb_gpio_irq_unmask_request request;
+ int ret;
+
+ request.which = hwirq;
+ ret = gb_operation_sync(ggc->connection,
+ GB_GPIO_TYPE_IRQ_UNMASK,
+ &request, sizeof(request), NULL, 0);
+ if (ret)
+ dev_err(dev, "failed to unmask irq: %d\n", ret);
+}
+
+static void _gb_gpio_irq_set_type(struct gb_gpio_controller *ggc,
+ u8 hwirq, u8 type)
+{
+ struct device *dev = &ggc->gbphy_dev->dev;
+ struct gb_gpio_irq_type_request request;
+ int ret;
+
+ request.which = hwirq;
+ request.type = type;
+
+ ret = gb_operation_sync(ggc->connection,
+ GB_GPIO_TYPE_IRQ_TYPE,
+ &request, sizeof(request), NULL, 0);
+ if (ret)
+ dev_err(dev, "failed to set irq type: %d\n", ret);
+}
+
+static void gb_gpio_irq_mask(struct irq_data *d)
+{
+ struct gpio_chip *chip = irq_data_to_gpio_chip(d);
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+ struct gb_gpio_line *line = &ggc->lines[d->hwirq];
+
+ line->masked = true;
+ line->masked_pending = true;
+}
+
+static void gb_gpio_irq_unmask(struct irq_data *d)
+{
+ struct gpio_chip *chip = irq_data_to_gpio_chip(d);
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+ struct gb_gpio_line *line = &ggc->lines[d->hwirq];
+
+ line->masked = false;
+ line->masked_pending = true;
+}
+
+static int gb_gpio_irq_set_type(struct irq_data *d, unsigned int type)
+{
+ struct gpio_chip *chip = irq_data_to_gpio_chip(d);
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+ struct gb_gpio_line *line = &ggc->lines[d->hwirq];
+ struct device *dev = &ggc->gbphy_dev->dev;
+ u8 irq_type;
+
+ switch (type) {
+ case IRQ_TYPE_NONE:
+ irq_type = GB_GPIO_IRQ_TYPE_NONE;
+ break;
+ case IRQ_TYPE_EDGE_RISING:
+ irq_type = GB_GPIO_IRQ_TYPE_EDGE_RISING;
+ break;
+ case IRQ_TYPE_EDGE_FALLING:
+ irq_type = GB_GPIO_IRQ_TYPE_EDGE_FALLING;
+ break;
+ case IRQ_TYPE_EDGE_BOTH:
+ irq_type = GB_GPIO_IRQ_TYPE_EDGE_BOTH;
+ break;
+ case IRQ_TYPE_LEVEL_LOW:
+ irq_type = GB_GPIO_IRQ_TYPE_LEVEL_LOW;
+ break;
+ case IRQ_TYPE_LEVEL_HIGH:
+ irq_type = GB_GPIO_IRQ_TYPE_LEVEL_HIGH;
+ break;
+ default:
+ dev_err(dev, "unsupported irq type: %u\n", type);
+ return -EINVAL;
+ }
+
+ line->irq_type = irq_type;
+ line->irq_type_pending = true;
+
+ return 0;
+}
+
+static void gb_gpio_irq_bus_lock(struct irq_data *d)
+{
+ struct gpio_chip *chip = irq_data_to_gpio_chip(d);
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+
+ mutex_lock(&ggc->irq_lock);
+}
+
+static void gb_gpio_irq_bus_sync_unlock(struct irq_data *d)
+{
+ struct gpio_chip *chip = irq_data_to_gpio_chip(d);
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+ struct gb_gpio_line *line = &ggc->lines[d->hwirq];
+
+ if (line->irq_type_pending) {
+ _gb_gpio_irq_set_type(ggc, d->hwirq, line->irq_type);
+ line->irq_type_pending = false;
+ }
+
+ if (line->masked_pending) {
+ if (line->masked)
+ _gb_gpio_irq_mask(ggc, d->hwirq);
+ else
+ _gb_gpio_irq_unmask(ggc, d->hwirq);
+ line->masked_pending = false;
+ }
+
+ mutex_unlock(&ggc->irq_lock);
+}
+
+static int gb_gpio_request_handler(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct gb_gpio_controller *ggc = gb_connection_get_data(connection);
+ struct device *dev = &ggc->gbphy_dev->dev;
+ struct gb_message *request;
+ struct gb_gpio_irq_event_request *event;
+ u8 type = op->type;
+ int irq;
+ struct irq_desc *desc;
+
+ if (type != GB_GPIO_TYPE_IRQ_EVENT) {
+ dev_err(dev, "unsupported unsolicited request: %u\n", type);
+ return -EINVAL;
+ }
+
+ request = op->request;
+
+ if (request->payload_size < sizeof(*event)) {
+ dev_err(dev, "short event received (%zu < %zu)\n",
+ request->payload_size, sizeof(*event));
+ return -EINVAL;
+ }
+
+ event = request->payload;
+ if (event->which > ggc->line_max) {
+ dev_err(dev, "invalid hw irq: %d\n", event->which);
+ return -EINVAL;
+ }
+
+ irq = irq_find_mapping(ggc->irqdomain, event->which);
+ if (!irq) {
+ dev_err(dev, "failed to find IRQ\n");
+ return -EINVAL;
+ }
+ desc = irq_to_desc(irq);
+ if (!desc) {
+ dev_err(dev, "failed to look up irq\n");
+ return -EINVAL;
+ }
+
+ local_irq_disable();
+ generic_handle_irq_desc(desc);
+ local_irq_enable();
+
+ return 0;
+}
+
+static int gb_gpio_request(struct gpio_chip *chip, unsigned offset)
+{
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+
+ return gb_gpio_activate_operation(ggc, (u8)offset);
+}
+
+static void gb_gpio_free(struct gpio_chip *chip, unsigned offset)
+{
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+
+ gb_gpio_deactivate_operation(ggc, (u8)offset);
+}
+
+static int gb_gpio_get_direction(struct gpio_chip *chip, unsigned offset)
+{
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+ u8 which;
+ int ret;
+
+ which = (u8)offset;
+ ret = gb_gpio_get_direction_operation(ggc, which);
+ if (ret)
+ return ret;
+
+ return ggc->lines[which].direction ? 1 : 0;
+}
+
+static int gb_gpio_direction_input(struct gpio_chip *chip, unsigned offset)
+{
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+
+ return gb_gpio_direction_in_operation(ggc, (u8)offset);
+}
+
+static int gb_gpio_direction_output(struct gpio_chip *chip, unsigned offset,
+ int value)
+{
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+
+ return gb_gpio_direction_out_operation(ggc, (u8)offset, !!value);
+}
+
+static int gb_gpio_get(struct gpio_chip *chip, unsigned offset)
+{
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+ u8 which;
+ int ret;
+
+ which = (u8)offset;
+ ret = gb_gpio_get_value_operation(ggc, which);
+ if (ret)
+ return ret;
+
+ return ggc->lines[which].value;
+}
+
+static void gb_gpio_set(struct gpio_chip *chip, unsigned offset, int value)
+{
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+
+ gb_gpio_set_value_operation(ggc, (u8)offset, !!value);
+}
+
+static int gb_gpio_set_debounce(struct gpio_chip *chip, unsigned offset,
+ unsigned debounce)
+{
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+ u16 usec;
+
+ if (debounce > U16_MAX)
+ return -EINVAL;
+ usec = (u16)debounce;
+
+ return gb_gpio_set_debounce_operation(ggc, (u8)offset, usec);
+}
+
+static int gb_gpio_controller_setup(struct gb_gpio_controller *ggc)
+{
+ int ret;
+
+ /* Now find out how many lines there are */
+ ret = gb_gpio_line_count_operation(ggc);
+ if (ret)
+ return ret;
+
+ ggc->lines = kcalloc(ggc->line_max + 1, sizeof(*ggc->lines),
+ GFP_KERNEL);
+ if (!ggc->lines)
+ return -ENOMEM;
+
+ return ret;
+}
+
+/**
+ * gb_gpio_irq_map() - maps an IRQ into a GB gpio irqchip
+ * @d: the irqdomain used by this irqchip
+ * @irq: the global irq number used by this GB gpio irqchip irq
+ * @hwirq: the local IRQ/GPIO line offset on this GB gpio
+ *
+ * This function will set up the mapping for a certain IRQ line on a
+ * GB gpio by assigning the GB gpio as chip data, and using the irqchip
+ * stored inside the GB gpio.
+ */
+static int gb_gpio_irq_map(struct irq_domain *domain, unsigned int irq,
+ irq_hw_number_t hwirq)
+{
+ struct gpio_chip *chip = domain->host_data;
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+
+ irq_set_chip_data(irq, ggc);
+ irq_set_chip_and_handler(irq, ggc->irqchip, ggc->irq_handler);
+ irq_set_noprobe(irq);
+ /*
+ * No set-up of the hardware will happen if IRQ_TYPE_NONE
+ * is passed as default type.
+ */
+ if (ggc->irq_default_type != IRQ_TYPE_NONE)
+ irq_set_irq_type(irq, ggc->irq_default_type);
+
+ return 0;
+}
+
+static void gb_gpio_irq_unmap(struct irq_domain *d, unsigned int irq)
+{
+ irq_set_chip_and_handler(irq, NULL, NULL);
+ irq_set_chip_data(irq, NULL);
+}
+
+static const struct irq_domain_ops gb_gpio_domain_ops = {
+ .map = gb_gpio_irq_map,
+ .unmap = gb_gpio_irq_unmap,
+};
+
+/**
+ * gb_gpio_irqchip_remove() - removes an irqchip added to a gb_gpio_controller
+ * @ggc: the gb_gpio_controller to remove the irqchip from
+ *
+ * This is called only from gb_gpio_remove()
+ */
+static void gb_gpio_irqchip_remove(struct gb_gpio_controller *ggc)
+{
+ unsigned int offset;
+
+ /* Remove all IRQ mappings and delete the domain */
+ if (ggc->irqdomain) {
+ for (offset = 0; offset < (ggc->line_max + 1); offset++)
+ irq_dispose_mapping(irq_find_mapping(ggc->irqdomain, offset));
+ irq_domain_remove(ggc->irqdomain);
+ }
+
+ if (ggc->irqchip) {
+ ggc->irqchip = NULL;
+ }
+}
+
+/**
+ * gb_gpio_irqchip_add() - adds an irqchip to a gpio chip
+ * @chip: the gpio chip to add the irqchip to
+ * @irqchip: the irqchip to add to the adapter
+ * @first_irq: if not dynamically assigned, the base (first) IRQ to
+ * allocate gpio irqs from
+ * @handler: the irq handler to use (often a predefined irq core function)
+ * @type: the default type for IRQs on this irqchip, pass IRQ_TYPE_NONE
+ * to have the core avoid setting up any default type in the hardware.
+ *
+ * This function closely associates a certain irqchip with a certain
+ * gpio chip, providing an irq domain to translate the local IRQs to
+ * global irqs, and making sure that the gpio chip
+ * is passed as chip data to all related functions. Driver callbacks
+ * need to use container_of() to get their local state containers back
+ * from the gpio chip passed as chip data. An irqdomain will be stored
+ * in the gpio chip that shall be used by the driver to handle IRQ number
+ * translation. The gpio chip will need to be initialized and registered
+ * before calling this function.
+ */
+static int gb_gpio_irqchip_add(struct gpio_chip *chip,
+ struct irq_chip *irqchip,
+ unsigned int first_irq,
+ irq_flow_handler_t handler,
+ unsigned int type)
+{
+ struct gb_gpio_controller *ggc;
+ unsigned int offset;
+ unsigned irq_base;
+
+ if (!chip || !irqchip)
+ return -EINVAL;
+
+ ggc = gpio_chip_to_gb_gpio_controller(chip);
+
+ ggc->irqchip = irqchip;
+ ggc->irq_handler = handler;
+ ggc->irq_default_type = type;
+ ggc->irqdomain = irq_domain_add_simple(NULL,
+ ggc->line_max + 1, first_irq,
+ &gb_gpio_domain_ops, chip);
+ if (!ggc->irqdomain) {
+ ggc->irqchip = NULL;
+ return -EINVAL;
+ }
+
+ /*
+ * Prepare the mapping since the irqchip shall be orthogonal to
+ * any gpio calls. If the first_irq was zero, this is
+ * necessary to allocate descriptors for all IRQs.
+ */
+ for (offset = 0; offset < (ggc->line_max + 1); offset++) {
+ irq_base = irq_create_mapping(ggc->irqdomain, offset);
+ if (offset == 0)
+ ggc->irq_base = irq_base;
+ }
+
+ return 0;
+}
+
+static int gb_gpio_to_irq(struct gpio_chip *chip, unsigned offset)
+{
+ struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
+
+ return irq_find_mapping(ggc->irqdomain, offset);
+}
+
+static int gb_gpio_probe(struct gbphy_device *gbphy_dev,
+ const struct gbphy_device_id *id)
+{
+ struct gb_connection *connection;
+ struct gb_gpio_controller *ggc;
+ struct gpio_chip *gpio;
+ struct irq_chip *irqc;
+ int ret;
+
+ ggc = kzalloc(sizeof(*ggc), GFP_KERNEL);
+ if (!ggc)
+ return -ENOMEM;
+
+ connection = gb_connection_create(gbphy_dev->bundle,
+ le16_to_cpu(gbphy_dev->cport_desc->id),
+ gb_gpio_request_handler);
+ if (IS_ERR(connection)) {
+ ret = PTR_ERR(connection);
+ goto exit_ggc_free;
+ }
+
+ ggc->connection = connection;
+ gb_connection_set_data(connection, ggc);
+ ggc->gbphy_dev = gbphy_dev;
+ gb_gbphy_set_data(gbphy_dev, ggc);
+
+ ret = gb_connection_enable_tx(connection);
+ if (ret)
+ goto exit_connection_destroy;
+
+ ret = gb_gpio_controller_setup(ggc);
+ if (ret)
+ goto exit_connection_disable;
+
+ irqc = &ggc->irqc;
+ irqc->irq_mask = gb_gpio_irq_mask;
+ irqc->irq_unmask = gb_gpio_irq_unmask;
+ irqc->irq_set_type = gb_gpio_irq_set_type;
+ irqc->irq_bus_lock = gb_gpio_irq_bus_lock;
+ irqc->irq_bus_sync_unlock = gb_gpio_irq_bus_sync_unlock;
+ irqc->name = "greybus_gpio";
+
+ mutex_init(&ggc->irq_lock);
+
+ gpio = &ggc->chip;
+
+ gpio->label = "greybus_gpio";
+ gpio->parent = &gbphy_dev->dev;
+ gpio->owner = THIS_MODULE;
+
+ gpio->request = gb_gpio_request;
+ gpio->free = gb_gpio_free;
+ gpio->get_direction = gb_gpio_get_direction;
+ gpio->direction_input = gb_gpio_direction_input;
+ gpio->direction_output = gb_gpio_direction_output;
+ gpio->get = gb_gpio_get;
+ gpio->set = gb_gpio_set;
+ gpio->set_debounce = gb_gpio_set_debounce;
+ gpio->to_irq = gb_gpio_to_irq;
+ gpio->base = -1; /* Allocate base dynamically */
+ gpio->ngpio = ggc->line_max + 1;
+ gpio->can_sleep = true;
+
+ ret = gb_connection_enable(connection);
+ if (ret)
+ goto exit_line_free;
+
+ ret = gpiochip_add(gpio);
+ if (ret) {
+ dev_err(&connection->bundle->dev,
+ "failed to add gpio chip: %d\n", ret);
+ goto exit_line_free;
+ }
+
+ ret = gb_gpio_irqchip_add(gpio, irqc, 0,
+ handle_level_irq, IRQ_TYPE_NONE);
+ if (ret) {
+ dev_err(&connection->bundle->dev,
+ "failed to add irq chip: %d\n", ret);
+ goto exit_gpiochip_remove;
+ }
+
+ gbphy_runtime_put_autosuspend(gbphy_dev);
+ return 0;
+
+exit_gpiochip_remove:
+ gpiochip_remove(gpio);
+exit_line_free:
+ kfree(ggc->lines);
+exit_connection_disable:
+ gb_connection_disable(connection);
+exit_connection_destroy:
+ gb_connection_destroy(connection);
+exit_ggc_free:
+ kfree(ggc);
+ return ret;
+}
+
+static void gb_gpio_remove(struct gbphy_device *gbphy_dev)
+{
+ struct gb_gpio_controller *ggc = gb_gbphy_get_data(gbphy_dev);
+ struct gb_connection *connection = ggc->connection;
+ int ret;
+
+ ret = gbphy_runtime_get_sync(gbphy_dev);
+ if (ret)
+ gbphy_runtime_get_noresume(gbphy_dev);
+
+ gb_connection_disable_rx(connection);
+ gb_gpio_irqchip_remove(ggc);
+ gpiochip_remove(&ggc->chip);
+ gb_connection_disable(connection);
+ gb_connection_destroy(connection);
+ kfree(ggc->lines);
+ kfree(ggc);
+}
+
+static const struct gbphy_device_id gb_gpio_id_table[] = {
+ { GBPHY_PROTOCOL(GREYBUS_PROTOCOL_GPIO) },
+ { },
+};
+MODULE_DEVICE_TABLE(gbphy, gb_gpio_id_table);
+
+static struct gbphy_driver gpio_driver = {
+ .name = "gpio",
+ .probe = gb_gpio_probe,
+ .remove = gb_gpio_remove,
+ .id_table = gb_gpio_id_table,
+};
+
+module_gbphy_driver(gpio_driver);
+MODULE_LICENSE("GPL v2");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 24/32] greybus: bridged phy i2c driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (21 preceding siblings ...)
2016-09-16 14:20 ` [patch 23/32] greybus: bridged phy gpio driver Greg KH
@ 2016-09-16 14:20 ` Greg KH
2016-09-16 14:20 ` [patch 25/32] greybus: bridged phy pwm driver Greg KH
` (7 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:20 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus bridged phy i2c class protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/i2c.c | 343 ++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 343 insertions(+)
--- /dev/null
+++ b/drivers/greybus/i2c.c
@@ -0,0 +1,343 @@
+/*
+ * I2C bridge driver for the Greybus "generic" I2C module.
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/i2c.h>
+
+#include "greybus.h"
+#include "gbphy.h"
+
+struct gb_i2c_device {
+ struct gb_connection *connection;
+ struct gbphy_device *gbphy_dev;
+
+ u32 functionality;
+
+ struct i2c_adapter adapter;
+};
+
+/*
+ * Map Greybus i2c functionality bits into Linux ones
+ */
+static u32 gb_i2c_functionality_map(u32 gb_i2c_functionality)
+{
+ return gb_i2c_functionality; /* All bits the same for now */
+}
+
+static int gb_i2c_functionality_operation(struct gb_i2c_device *gb_i2c_dev)
+{
+ struct gb_i2c_functionality_response response;
+ u32 functionality;
+ int ret;
+
+ ret = gb_operation_sync(gb_i2c_dev->connection,
+ GB_I2C_TYPE_FUNCTIONALITY,
+ NULL, 0, &response, sizeof(response));
+ if (ret)
+ return ret;
+
+ functionality = le32_to_cpu(response.functionality);
+ gb_i2c_dev->functionality = gb_i2c_functionality_map(functionality);
+
+ return 0;
+}
+
+/*
+ * Map Linux i2c_msg flags into Greybus i2c transfer op flags.
+ */
+static u16 gb_i2c_transfer_op_flags_map(u16 flags)
+{
+ return flags; /* All flags the same for now */
+}
+
+static void
+gb_i2c_fill_transfer_op(struct gb_i2c_transfer_op *op, struct i2c_msg *msg)
+{
+ u16 flags = gb_i2c_transfer_op_flags_map(msg->flags);
+
+ op->addr = cpu_to_le16(msg->addr);
+ op->flags = cpu_to_le16(flags);
+ op->size = cpu_to_le16(msg->len);
+}
+
+static struct gb_operation *
+gb_i2c_operation_create(struct gb_connection *connection,
+ struct i2c_msg *msgs, u32 msg_count)
+{
+ struct gb_i2c_device *gb_i2c_dev = gb_connection_get_data(connection);
+ struct gb_i2c_transfer_request *request;
+ struct gb_operation *operation;
+ struct gb_i2c_transfer_op *op;
+ struct i2c_msg *msg;
+ u32 data_out_size = 0;
+ u32 data_in_size = 0;
+ size_t request_size;
+ void *data;
+ u16 op_count;
+ u32 i;
+
+ if (msg_count > (u32)U16_MAX) {
+ dev_err(&gb_i2c_dev->gbphy_dev->dev, "msg_count (%u) too big\n",
+ msg_count);
+ return NULL;
+ }
+ op_count = (u16)msg_count;
+
+ /*
+ * In addition to space for all message descriptors we need
+ * to have enough to hold all outbound message data.
+ */
+ msg = msgs;
+ for (i = 0; i < msg_count; i++, msg++)
+ if (msg->flags & I2C_M_RD)
+ data_in_size += (u32)msg->len;
+ else
+ data_out_size += (u32)msg->len;
+
+ request_size = sizeof(*request);
+ request_size += msg_count * sizeof(*op);
+ request_size += data_out_size;
+
+ /* Response consists only of incoming data */
+ operation = gb_operation_create(connection, GB_I2C_TYPE_TRANSFER,
+ request_size, data_in_size, GFP_KERNEL);
+ if (!operation)
+ return NULL;
+
+ request = operation->request->payload;
+ request->op_count = cpu_to_le16(op_count);
+ /* Fill in the ops array */
+ op = &request->ops[0];
+ msg = msgs;
+ for (i = 0; i < msg_count; i++)
+ gb_i2c_fill_transfer_op(op++, msg++);
+
+ if (!data_out_size)
+ return operation;
+
+ /* Copy over the outgoing data; it starts after the last op */
+ data = op;
+ msg = msgs;
+ for (i = 0; i < msg_count; i++) {
+ if (!(msg->flags & I2C_M_RD)) {
+ memcpy(data, msg->buf, msg->len);
+ data += msg->len;
+ }
+ msg++;
+ }
+
+ return operation;
+}
+
+static void gb_i2c_decode_response(struct i2c_msg *msgs, u32 msg_count,
+ struct gb_i2c_transfer_response *response)
+{
+ struct i2c_msg *msg = msgs;
+ u8 *data;
+ u32 i;
+
+ if (!response)
+ return;
+ data = response->data;
+ for (i = 0; i < msg_count; i++) {
+ if (msg->flags & I2C_M_RD) {
+ memcpy(msg->buf, data, msg->len);
+ data += msg->len;
+ }
+ msg++;
+ }
+}
+
+/*
+ * Some i2c transfer operations return results that are expected.
+ */
+static bool gb_i2c_expected_transfer_error(int errno)
+{
+ return errno == -EAGAIN || errno == -ENODEV;
+}
+
+static int gb_i2c_transfer_operation(struct gb_i2c_device *gb_i2c_dev,
+ struct i2c_msg *msgs, u32 msg_count)
+{
+ struct gb_connection *connection = gb_i2c_dev->connection;
+ struct device *dev = &gb_i2c_dev->gbphy_dev->dev;
+ struct gb_operation *operation;
+ int ret;
+
+ operation = gb_i2c_operation_create(connection, msgs, msg_count);
+ if (!operation)
+ return -ENOMEM;
+
+ ret = gbphy_runtime_get_sync(gb_i2c_dev->gbphy_dev);
+ if (ret)
+ goto exit_operation_put;
+
+ ret = gb_operation_request_send_sync(operation);
+ if (!ret) {
+ struct gb_i2c_transfer_response *response;
+
+ response = operation->response->payload;
+ gb_i2c_decode_response(msgs, msg_count, response);
+ ret = msg_count;
+ } else if (!gb_i2c_expected_transfer_error(ret)) {
+ dev_err(dev, "transfer operation failed (%d)\n", ret);
+ }
+
+ gbphy_runtime_put_autosuspend(gb_i2c_dev->gbphy_dev);
+
+exit_operation_put:
+ gb_operation_put(operation);
+
+ return ret;
+}
+
+static int gb_i2c_master_xfer(struct i2c_adapter *adap, struct i2c_msg *msgs,
+ int msg_count)
+{
+ struct gb_i2c_device *gb_i2c_dev;
+
+ gb_i2c_dev = i2c_get_adapdata(adap);
+
+ return gb_i2c_transfer_operation(gb_i2c_dev, msgs, msg_count);
+}
+
+#if 0
+/* Later */
+static int gb_i2c_smbus_xfer(struct i2c_adapter *adap,
+ u16 addr, unsigned short flags, char read_write,
+ u8 command, int size, union i2c_smbus_data *data)
+{
+ struct gb_i2c_device *gb_i2c_dev;
+
+ gb_i2c_dev = i2c_get_adapdata(adap);
+
+ return 0;
+}
+#endif
+
+static u32 gb_i2c_functionality(struct i2c_adapter *adap)
+{
+ struct gb_i2c_device *gb_i2c_dev = i2c_get_adapdata(adap);
+
+ return gb_i2c_dev->functionality;
+}
+
+static const struct i2c_algorithm gb_i2c_algorithm = {
+ .master_xfer = gb_i2c_master_xfer,
+ /* .smbus_xfer = gb_i2c_smbus_xfer, */
+ .functionality = gb_i2c_functionality,
+};
+
+/*
+ * Do initial setup of the i2c device. This includes verifying we
+ * can support it (based on the protocol version it advertises).
+ * If that's OK, we get and cached its functionality bits.
+ *
+ * Note: gb_i2c_dev->connection is assumed to have been valid.
+ */
+static int gb_i2c_device_setup(struct gb_i2c_device *gb_i2c_dev)
+{
+ /* Assume the functionality never changes, just get it once */
+ return gb_i2c_functionality_operation(gb_i2c_dev);
+}
+
+static int gb_i2c_probe(struct gbphy_device *gbphy_dev,
+ const struct gbphy_device_id *id)
+{
+ struct gb_connection *connection;
+ struct gb_i2c_device *gb_i2c_dev;
+ struct i2c_adapter *adapter;
+ int ret;
+
+ gb_i2c_dev = kzalloc(sizeof(*gb_i2c_dev), GFP_KERNEL);
+ if (!gb_i2c_dev)
+ return -ENOMEM;
+
+ connection = gb_connection_create(gbphy_dev->bundle,
+ le16_to_cpu(gbphy_dev->cport_desc->id),
+ NULL);
+ if (IS_ERR(connection)) {
+ ret = PTR_ERR(connection);
+ goto exit_i2cdev_free;
+ }
+
+ gb_i2c_dev->connection = connection;
+ gb_connection_set_data(connection, gb_i2c_dev);
+ gb_i2c_dev->gbphy_dev = gbphy_dev;
+ gb_gbphy_set_data(gbphy_dev, gb_i2c_dev);
+
+ ret = gb_connection_enable(connection);
+ if (ret)
+ goto exit_connection_destroy;
+
+ ret = gb_i2c_device_setup(gb_i2c_dev);
+ if (ret)
+ goto exit_connection_disable;
+
+ /* Looks good; up our i2c adapter */
+ adapter = &gb_i2c_dev->adapter;
+ adapter->owner = THIS_MODULE;
+ adapter->class = I2C_CLASS_HWMON | I2C_CLASS_SPD;
+ adapter->algo = &gb_i2c_algorithm;
+ /* adapter->algo_data = what? */
+
+ adapter->dev.parent = &gbphy_dev->dev;
+ snprintf(adapter->name, sizeof(adapter->name), "Greybus i2c adapter");
+ i2c_set_adapdata(adapter, gb_i2c_dev);
+
+ ret = i2c_add_adapter(adapter);
+ if (ret)
+ goto exit_connection_disable;
+
+ gbphy_runtime_put_autosuspend(gbphy_dev);
+ return 0;
+
+exit_connection_disable:
+ gb_connection_disable(connection);
+exit_connection_destroy:
+ gb_connection_destroy(connection);
+exit_i2cdev_free:
+ kfree(gb_i2c_dev);
+
+ return ret;
+}
+
+static void gb_i2c_remove(struct gbphy_device *gbphy_dev)
+{
+ struct gb_i2c_device *gb_i2c_dev = gb_gbphy_get_data(gbphy_dev);
+ struct gb_connection *connection = gb_i2c_dev->connection;
+ int ret;
+
+ ret = gbphy_runtime_get_sync(gbphy_dev);
+ if (ret)
+ gbphy_runtime_get_noresume(gbphy_dev);
+
+ i2c_del_adapter(&gb_i2c_dev->adapter);
+ gb_connection_disable(connection);
+ gb_connection_destroy(connection);
+ kfree(gb_i2c_dev);
+}
+
+static const struct gbphy_device_id gb_i2c_id_table[] = {
+ { GBPHY_PROTOCOL(GREYBUS_PROTOCOL_I2C) },
+ { },
+};
+MODULE_DEVICE_TABLE(gbphy, gb_i2c_id_table);
+
+static struct gbphy_driver i2c_driver = {
+ .name = "i2c",
+ .probe = gb_i2c_probe,
+ .remove = gb_i2c_remove,
+ .id_table = gb_i2c_id_table,
+};
+
+module_gbphy_driver(i2c_driver);
+MODULE_LICENSE("GPL v2");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 25/32] greybus: bridged phy pwm driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (22 preceding siblings ...)
2016-09-16 14:20 ` [patch 24/32] greybus: bridged phy i2c driver Greg KH
@ 2016-09-16 14:20 ` Greg KH
2016-09-16 14:21 ` [patch 26/32] greybus: bridged phy sdio driver Greg KH
` (6 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:20 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus bridged phy pwm class protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/pwm.c | 338 ++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 338 insertions(+)
--- /dev/null
+++ b/drivers/greybus/pwm.c
@@ -0,0 +1,338 @@
+/*
+ * PWM Greybus driver.
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/pwm.h>
+
+#include "greybus.h"
+#include "gbphy.h"
+
+struct gb_pwm_chip {
+ struct gb_connection *connection;
+ u8 pwm_max; /* max pwm number */
+
+ struct pwm_chip chip;
+ struct pwm_chip *pwm;
+};
+#define pwm_chip_to_gb_pwm_chip(chip) \
+ container_of(chip, struct gb_pwm_chip, chip)
+
+
+static int gb_pwm_count_operation(struct gb_pwm_chip *pwmc)
+{
+ struct gb_pwm_count_response response;
+ int ret;
+
+ ret = gb_operation_sync(pwmc->connection, GB_PWM_TYPE_PWM_COUNT,
+ NULL, 0, &response, sizeof(response));
+ if (ret)
+ return ret;
+ pwmc->pwm_max = response.count;
+ return 0;
+}
+
+static int gb_pwm_activate_operation(struct gb_pwm_chip *pwmc,
+ u8 which)
+{
+ struct gb_pwm_activate_request request;
+ struct gbphy_device *gbphy_dev;
+ int ret;
+
+ if (which > pwmc->pwm_max)
+ return -EINVAL;
+
+ request.which = which;
+
+ gbphy_dev = to_gbphy_dev(pwmc->chip.dev);
+ ret = gbphy_runtime_get_sync(gbphy_dev);
+ if (ret)
+ return ret;
+
+ ret = gb_operation_sync(pwmc->connection, GB_PWM_TYPE_ACTIVATE,
+ &request, sizeof(request), NULL, 0);
+
+ gbphy_runtime_put_autosuspend(gbphy_dev);
+
+ return ret;
+}
+
+static int gb_pwm_deactivate_operation(struct gb_pwm_chip *pwmc,
+ u8 which)
+{
+ struct gb_pwm_deactivate_request request;
+ struct gbphy_device *gbphy_dev;
+ int ret;
+
+ if (which > pwmc->pwm_max)
+ return -EINVAL;
+
+ request.which = which;
+
+ gbphy_dev = to_gbphy_dev(pwmc->chip.dev);
+ ret = gbphy_runtime_get_sync(gbphy_dev);
+ if (ret)
+ return ret;
+
+ ret = gb_operation_sync(pwmc->connection, GB_PWM_TYPE_DEACTIVATE,
+ &request, sizeof(request), NULL, 0);
+
+ gbphy_runtime_put_autosuspend(gbphy_dev);
+
+ return ret;
+}
+
+static int gb_pwm_config_operation(struct gb_pwm_chip *pwmc,
+ u8 which, u32 duty, u32 period)
+{
+ struct gb_pwm_config_request request;
+ struct gbphy_device *gbphy_dev;
+ int ret;
+
+ if (which > pwmc->pwm_max)
+ return -EINVAL;
+
+ request.which = which;
+ request.duty = cpu_to_le32(duty);
+ request.period = cpu_to_le32(period);
+
+ gbphy_dev = to_gbphy_dev(pwmc->chip.dev);
+ ret = gbphy_runtime_get_sync(gbphy_dev);
+ if (ret)
+ return ret;
+
+ ret = gb_operation_sync(pwmc->connection, GB_PWM_TYPE_CONFIG,
+ &request, sizeof(request), NULL, 0);
+
+ gbphy_runtime_put_autosuspend(gbphy_dev);
+
+ return ret;
+}
+
+static int gb_pwm_set_polarity_operation(struct gb_pwm_chip *pwmc,
+ u8 which, u8 polarity)
+{
+ struct gb_pwm_polarity_request request;
+ struct gbphy_device *gbphy_dev;
+ int ret;
+
+ if (which > pwmc->pwm_max)
+ return -EINVAL;
+
+ request.which = which;
+ request.polarity = polarity;
+
+ gbphy_dev = to_gbphy_dev(pwmc->chip.dev);
+ ret = gbphy_runtime_get_sync(gbphy_dev);
+ if (ret)
+ return ret;
+
+ ret = gb_operation_sync(pwmc->connection, GB_PWM_TYPE_POLARITY,
+ &request, sizeof(request), NULL, 0);
+
+ gbphy_runtime_put_autosuspend(gbphy_dev);
+
+ return ret;
+}
+
+static int gb_pwm_enable_operation(struct gb_pwm_chip *pwmc,
+ u8 which)
+{
+ struct gb_pwm_enable_request request;
+ struct gbphy_device *gbphy_dev;
+ int ret;
+
+ if (which > pwmc->pwm_max)
+ return -EINVAL;
+
+ request.which = which;
+
+ gbphy_dev = to_gbphy_dev(pwmc->chip.dev);
+ ret = gbphy_runtime_get_sync(gbphy_dev);
+ if (ret)
+ return ret;
+
+ ret = gb_operation_sync(pwmc->connection, GB_PWM_TYPE_ENABLE,
+ &request, sizeof(request), NULL, 0);
+ if (ret)
+ gbphy_runtime_put_autosuspend(gbphy_dev);
+
+ return ret;
+}
+
+static int gb_pwm_disable_operation(struct gb_pwm_chip *pwmc,
+ u8 which)
+{
+ struct gb_pwm_disable_request request;
+ struct gbphy_device *gbphy_dev;
+ int ret;
+
+ if (which > pwmc->pwm_max)
+ return -EINVAL;
+
+ request.which = which;
+
+ ret = gb_operation_sync(pwmc->connection, GB_PWM_TYPE_DISABLE,
+ &request, sizeof(request), NULL, 0);
+
+ gbphy_dev = to_gbphy_dev(pwmc->chip.dev);
+ gbphy_runtime_put_autosuspend(gbphy_dev);
+
+ return ret;
+}
+
+static int gb_pwm_request(struct pwm_chip *chip, struct pwm_device *pwm)
+{
+ struct gb_pwm_chip *pwmc = pwm_chip_to_gb_pwm_chip(chip);
+
+ return gb_pwm_activate_operation(pwmc, pwm->hwpwm);
+};
+
+static void gb_pwm_free(struct pwm_chip *chip, struct pwm_device *pwm)
+{
+ struct gb_pwm_chip *pwmc = pwm_chip_to_gb_pwm_chip(chip);
+
+ if (pwm_is_enabled(pwm))
+ dev_warn(chip->dev, "freeing PWM device without disabling\n");
+
+ gb_pwm_deactivate_operation(pwmc, pwm->hwpwm);
+}
+
+static int gb_pwm_config(struct pwm_chip *chip, struct pwm_device *pwm,
+ int duty_ns, int period_ns)
+{
+ struct gb_pwm_chip *pwmc = pwm_chip_to_gb_pwm_chip(chip);
+
+ return gb_pwm_config_operation(pwmc, pwm->hwpwm, duty_ns, period_ns);
+};
+
+static int gb_pwm_set_polarity(struct pwm_chip *chip, struct pwm_device *pwm,
+ enum pwm_polarity polarity)
+{
+ struct gb_pwm_chip *pwmc = pwm_chip_to_gb_pwm_chip(chip);
+
+ return gb_pwm_set_polarity_operation(pwmc, pwm->hwpwm, polarity);
+};
+
+static int gb_pwm_enable(struct pwm_chip *chip, struct pwm_device *pwm)
+{
+ struct gb_pwm_chip *pwmc = pwm_chip_to_gb_pwm_chip(chip);
+
+ return gb_pwm_enable_operation(pwmc, pwm->hwpwm);
+};
+
+static void gb_pwm_disable(struct pwm_chip *chip, struct pwm_device *pwm)
+{
+ struct gb_pwm_chip *pwmc = pwm_chip_to_gb_pwm_chip(chip);
+
+ gb_pwm_disable_operation(pwmc, pwm->hwpwm);
+};
+
+static const struct pwm_ops gb_pwm_ops = {
+ .request = gb_pwm_request,
+ .free = gb_pwm_free,
+ .config = gb_pwm_config,
+ .set_polarity = gb_pwm_set_polarity,
+ .enable = gb_pwm_enable,
+ .disable = gb_pwm_disable,
+ .owner = THIS_MODULE,
+};
+
+static int gb_pwm_probe(struct gbphy_device *gbphy_dev,
+ const struct gbphy_device_id *id)
+{
+ struct gb_connection *connection;
+ struct gb_pwm_chip *pwmc;
+ struct pwm_chip *pwm;
+ int ret;
+
+ pwmc = kzalloc(sizeof(*pwmc), GFP_KERNEL);
+ if (!pwmc)
+ return -ENOMEM;
+
+ connection = gb_connection_create(gbphy_dev->bundle,
+ le16_to_cpu(gbphy_dev->cport_desc->id),
+ NULL);
+ if (IS_ERR(connection)) {
+ ret = PTR_ERR(connection);
+ goto exit_pwmc_free;
+ }
+
+ pwmc->connection = connection;
+ gb_connection_set_data(connection, pwmc);
+ gb_gbphy_set_data(gbphy_dev, pwmc);
+
+ ret = gb_connection_enable(connection);
+ if (ret)
+ goto exit_connection_destroy;
+
+ /* Query number of pwms present */
+ ret = gb_pwm_count_operation(pwmc);
+ if (ret)
+ goto exit_connection_disable;
+
+ pwm = &pwmc->chip;
+
+ pwm->dev = &gbphy_dev->dev;
+ pwm->ops = &gb_pwm_ops;
+ pwm->base = -1; /* Allocate base dynamically */
+ pwm->npwm = pwmc->pwm_max + 1;
+ pwm->can_sleep = true; /* FIXME */
+
+ ret = pwmchip_add(pwm);
+ if (ret) {
+ dev_err(&gbphy_dev->dev,
+ "failed to register PWM: %d\n", ret);
+ goto exit_connection_disable;
+ }
+
+ gbphy_runtime_put_autosuspend(gbphy_dev);
+ return 0;
+
+exit_connection_disable:
+ gb_connection_disable(connection);
+exit_connection_destroy:
+ gb_connection_destroy(connection);
+exit_pwmc_free:
+ kfree(pwmc);
+ return ret;
+}
+
+static void gb_pwm_remove(struct gbphy_device *gbphy_dev)
+{
+ struct gb_pwm_chip *pwmc = gb_gbphy_get_data(gbphy_dev);
+ struct gb_connection *connection = pwmc->connection;
+ int ret;
+
+ ret = gbphy_runtime_get_sync(gbphy_dev);
+ if (ret)
+ gbphy_runtime_get_noresume(gbphy_dev);
+
+ pwmchip_remove(&pwmc->chip);
+ gb_connection_disable(connection);
+ gb_connection_destroy(connection);
+ kfree(pwmc);
+}
+
+static const struct gbphy_device_id gb_pwm_id_table[] = {
+ { GBPHY_PROTOCOL(GREYBUS_PROTOCOL_PWM) },
+ { },
+};
+MODULE_DEVICE_TABLE(gbphy, gb_pwm_id_table);
+
+static struct gbphy_driver pwm_driver = {
+ .name = "pwm",
+ .probe = gb_pwm_probe,
+ .remove = gb_pwm_remove,
+ .id_table = gb_pwm_id_table,
+};
+
+module_gbphy_driver(pwm_driver);
+MODULE_LICENSE("GPL v2");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 26/32] greybus: bridged phy sdio driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (23 preceding siblings ...)
2016-09-16 14:20 ` [patch 25/32] greybus: bridged phy pwm driver Greg KH
@ 2016-09-16 14:21 ` Greg KH
2016-09-16 14:21 ` [patch 27/32] greybus: bridged phy spi driver Greg KH
` (5 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:21 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus bridged phy sdio class protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/sdio.c | 884 +++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 884 insertions(+)
--- /dev/null
+++ b/drivers/greybus/sdio.c
@@ -0,0 +1,884 @@
+/*
+ * SD/MMC Greybus driver.
+ *
+ * Copyright 2014-2015 Google Inc.
+ * Copyright 2014-2015 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/kernel.h>
+#include <linux/mmc/core.h>
+#include <linux/mmc/host.h>
+#include <linux/mmc/mmc.h>
+#include <linux/scatterlist.h>
+#include <linux/workqueue.h>
+
+#include "greybus.h"
+#include "gbphy.h"
+
+struct gb_sdio_host {
+ struct gb_connection *connection;
+ struct gbphy_device *gbphy_dev;
+ struct mmc_host *mmc;
+ struct mmc_request *mrq;
+ struct mutex lock; /* lock for this host */
+ size_t data_max;
+ spinlock_t xfer; /* lock to cancel ongoing transfer */
+ bool xfer_stop;
+ struct workqueue_struct *mrq_workqueue;
+ struct work_struct mrqwork;
+ u8 queued_events;
+ bool removed;
+ bool card_present;
+ bool read_only;
+};
+
+
+#define GB_SDIO_RSP_R1_R5_R6_R7 (GB_SDIO_RSP_PRESENT | GB_SDIO_RSP_CRC | \
+ GB_SDIO_RSP_OPCODE)
+#define GB_SDIO_RSP_R3_R4 (GB_SDIO_RSP_PRESENT)
+#define GB_SDIO_RSP_R2 (GB_SDIO_RSP_PRESENT | GB_SDIO_RSP_CRC | \
+ GB_SDIO_RSP_136)
+#define GB_SDIO_RSP_R1B (GB_SDIO_RSP_PRESENT | GB_SDIO_RSP_CRC | \
+ GB_SDIO_RSP_OPCODE | GB_SDIO_RSP_BUSY)
+
+/* kernel vdd starts at 0x80 and we need to translate to greybus ones 0x01 */
+#define GB_SDIO_VDD_SHIFT 8
+
+#ifndef MMC_CAP2_CORE_RUNTIME_PM
+#define MMC_CAP2_CORE_RUNTIME_PM 0
+#endif
+
+static inline bool single_op(struct mmc_command *cmd)
+{
+ uint32_t opcode = cmd->opcode;
+
+ return opcode == MMC_WRITE_BLOCK ||
+ opcode == MMC_READ_SINGLE_BLOCK;
+}
+
+static void _gb_sdio_set_host_caps(struct gb_sdio_host *host, u32 r)
+{
+ u32 caps = 0;
+ u32 caps2 = 0;
+
+ caps = ((r & GB_SDIO_CAP_NONREMOVABLE) ? MMC_CAP_NONREMOVABLE : 0) |
+ ((r & GB_SDIO_CAP_4_BIT_DATA) ? MMC_CAP_4_BIT_DATA : 0) |
+ ((r & GB_SDIO_CAP_8_BIT_DATA) ? MMC_CAP_8_BIT_DATA : 0) |
+ ((r & GB_SDIO_CAP_MMC_HS) ? MMC_CAP_MMC_HIGHSPEED : 0) |
+ ((r & GB_SDIO_CAP_SD_HS) ? MMC_CAP_SD_HIGHSPEED : 0) |
+ ((r & GB_SDIO_CAP_ERASE) ? MMC_CAP_ERASE : 0) |
+ ((r & GB_SDIO_CAP_1_2V_DDR) ? MMC_CAP_1_2V_DDR : 0) |
+ ((r & GB_SDIO_CAP_1_8V_DDR) ? MMC_CAP_1_8V_DDR : 0) |
+ ((r & GB_SDIO_CAP_POWER_OFF_CARD) ? MMC_CAP_POWER_OFF_CARD : 0) |
+ ((r & GB_SDIO_CAP_UHS_SDR12) ? MMC_CAP_UHS_SDR12 : 0) |
+ ((r & GB_SDIO_CAP_UHS_SDR25) ? MMC_CAP_UHS_SDR25 : 0) |
+ ((r & GB_SDIO_CAP_UHS_SDR50) ? MMC_CAP_UHS_SDR50 : 0) |
+ ((r & GB_SDIO_CAP_UHS_SDR104) ? MMC_CAP_UHS_SDR104 : 0) |
+ ((r & GB_SDIO_CAP_UHS_DDR50) ? MMC_CAP_UHS_DDR50 : 0) |
+ ((r & GB_SDIO_CAP_DRIVER_TYPE_A) ? MMC_CAP_DRIVER_TYPE_A : 0) |
+ ((r & GB_SDIO_CAP_DRIVER_TYPE_C) ? MMC_CAP_DRIVER_TYPE_C : 0) |
+ ((r & GB_SDIO_CAP_DRIVER_TYPE_D) ? MMC_CAP_DRIVER_TYPE_D : 0);
+
+ caps2 = ((r & GB_SDIO_CAP_HS200_1_2V) ? MMC_CAP2_HS200_1_2V_SDR : 0) |
+ ((r & GB_SDIO_CAP_HS400_1_2V) ? MMC_CAP2_HS400_1_2V : 0) |
+ ((r & GB_SDIO_CAP_HS400_1_8V) ? MMC_CAP2_HS400_1_8V : 0) |
+ ((r & GB_SDIO_CAP_HS200_1_8V) ? MMC_CAP2_HS200_1_8V_SDR : 0);
+
+ host->mmc->caps = caps;
+ host->mmc->caps2 = caps2 | MMC_CAP2_CORE_RUNTIME_PM;
+
+ if (caps & MMC_CAP_NONREMOVABLE)
+ host->card_present = true;
+}
+
+static u32 _gb_sdio_get_host_ocr(u32 ocr)
+{
+ return (((ocr & GB_SDIO_VDD_165_195) ? MMC_VDD_165_195 : 0) |
+ ((ocr & GB_SDIO_VDD_20_21) ? MMC_VDD_20_21 : 0) |
+ ((ocr & GB_SDIO_VDD_21_22) ? MMC_VDD_21_22 : 0) |
+ ((ocr & GB_SDIO_VDD_22_23) ? MMC_VDD_22_23 : 0) |
+ ((ocr & GB_SDIO_VDD_23_24) ? MMC_VDD_23_24 : 0) |
+ ((ocr & GB_SDIO_VDD_24_25) ? MMC_VDD_24_25 : 0) |
+ ((ocr & GB_SDIO_VDD_25_26) ? MMC_VDD_25_26 : 0) |
+ ((ocr & GB_SDIO_VDD_26_27) ? MMC_VDD_26_27 : 0) |
+ ((ocr & GB_SDIO_VDD_27_28) ? MMC_VDD_27_28 : 0) |
+ ((ocr & GB_SDIO_VDD_28_29) ? MMC_VDD_28_29 : 0) |
+ ((ocr & GB_SDIO_VDD_29_30) ? MMC_VDD_29_30 : 0) |
+ ((ocr & GB_SDIO_VDD_30_31) ? MMC_VDD_30_31 : 0) |
+ ((ocr & GB_SDIO_VDD_31_32) ? MMC_VDD_31_32 : 0) |
+ ((ocr & GB_SDIO_VDD_32_33) ? MMC_VDD_32_33 : 0) |
+ ((ocr & GB_SDIO_VDD_33_34) ? MMC_VDD_33_34 : 0) |
+ ((ocr & GB_SDIO_VDD_34_35) ? MMC_VDD_34_35 : 0) |
+ ((ocr & GB_SDIO_VDD_35_36) ? MMC_VDD_35_36 : 0)
+ );
+}
+
+static int gb_sdio_get_caps(struct gb_sdio_host *host)
+{
+ struct gb_sdio_get_caps_response response;
+ struct mmc_host *mmc = host->mmc;
+ u16 data_max;
+ u32 blksz;
+ u32 ocr;
+ u32 r;
+ int ret;
+
+ ret = gb_operation_sync(host->connection, GB_SDIO_TYPE_GET_CAPABILITIES,
+ NULL, 0, &response, sizeof(response));
+ if (ret < 0)
+ return ret;
+ r = le32_to_cpu(response.caps);
+
+ _gb_sdio_set_host_caps(host, r);
+
+ /* get the max block size that could fit our payload */
+ data_max = gb_operation_get_payload_size_max(host->connection);
+ data_max = min(data_max - sizeof(struct gb_sdio_transfer_request),
+ data_max - sizeof(struct gb_sdio_transfer_response));
+
+ blksz = min_t(u16, le16_to_cpu(response.max_blk_size), data_max);
+ blksz = max_t(u32, 512, blksz);
+
+ mmc->max_blk_size = rounddown_pow_of_two(blksz);
+ mmc->max_blk_count = le16_to_cpu(response.max_blk_count);
+ host->data_max = data_max;
+
+ /* get ocr supported values */
+ ocr = _gb_sdio_get_host_ocr(le32_to_cpu(response.ocr));
+ mmc->ocr_avail = ocr;
+ mmc->ocr_avail_sdio = mmc->ocr_avail;
+ mmc->ocr_avail_sd = mmc->ocr_avail;
+ mmc->ocr_avail_mmc = mmc->ocr_avail;
+
+ /* get frequency range values */
+ mmc->f_min = le32_to_cpu(response.f_min);
+ mmc->f_max = le32_to_cpu(response.f_max);
+
+ return 0;
+}
+
+static void _gb_queue_event(struct gb_sdio_host *host, u8 event)
+{
+ if (event & GB_SDIO_CARD_INSERTED)
+ host->queued_events &= ~GB_SDIO_CARD_REMOVED;
+ else if (event & GB_SDIO_CARD_REMOVED)
+ host->queued_events &= ~GB_SDIO_CARD_INSERTED;
+
+ host->queued_events |= event;
+}
+
+static int _gb_sdio_process_events(struct gb_sdio_host *host, u8 event)
+{
+ u8 state_changed = 0;
+
+ if (event & GB_SDIO_CARD_INSERTED) {
+ if (host->mmc->caps & MMC_CAP_NONREMOVABLE)
+ return 0;
+ if (host->card_present)
+ return 0;
+ host->card_present = true;
+ state_changed = 1;
+ }
+
+ if (event & GB_SDIO_CARD_REMOVED) {
+ if (host->mmc->caps & MMC_CAP_NONREMOVABLE)
+ return 0;
+ if (!(host->card_present))
+ return 0;
+ host->card_present = false;
+ state_changed = 1;
+ }
+
+ if (event & GB_SDIO_WP) {
+ host->read_only = true;
+ }
+
+ if (state_changed) {
+ dev_info(mmc_dev(host->mmc), "card %s now event\n",
+ (host->card_present ? "inserted" : "removed"));
+ mmc_detect_change(host->mmc, 0);
+ }
+
+ return 0;
+}
+
+static int gb_sdio_request_handler(struct gb_operation *op)
+{
+ struct gb_sdio_host *host = gb_connection_get_data(op->connection);
+ struct gb_message *request;
+ struct gb_sdio_event_request *payload;
+ u8 type = op->type;
+ int ret = 0;
+ u8 event;
+
+ if (type != GB_SDIO_TYPE_EVENT) {
+ dev_err(mmc_dev(host->mmc),
+ "unsupported unsolicited event: %u\n", type);
+ return -EINVAL;
+ }
+
+ request = op->request;
+
+ if (request->payload_size < sizeof(*payload)) {
+ dev_err(mmc_dev(host->mmc), "wrong event size received (%zu < %zu)\n",
+ request->payload_size, sizeof(*payload));
+ return -EINVAL;
+ }
+
+ payload = request->payload;
+ event = payload->event;
+
+ if (host->removed)
+ _gb_queue_event(host, event);
+ else
+ ret = _gb_sdio_process_events(host, event);
+
+ return ret;
+}
+
+static int gb_sdio_set_ios(struct gb_sdio_host *host,
+ struct gb_sdio_set_ios_request *request)
+{
+ int ret;
+
+ ret = gbphy_runtime_get_sync(host->gbphy_dev);
+ if (ret)
+ return ret;
+
+ ret = gb_operation_sync(host->connection, GB_SDIO_TYPE_SET_IOS, request,
+ sizeof(*request), NULL, 0);
+
+ gbphy_runtime_put_autosuspend(host->gbphy_dev);
+
+ return ret;
+}
+
+static int _gb_sdio_send(struct gb_sdio_host *host, struct mmc_data *data,
+ size_t len, u16 nblocks, off_t skip)
+{
+ struct gb_sdio_transfer_request *request;
+ struct gb_sdio_transfer_response *response;
+ struct gb_operation *operation;
+ struct scatterlist *sg = data->sg;
+ unsigned int sg_len = data->sg_len;
+ size_t copied;
+ u16 send_blksz;
+ u16 send_blocks;
+ int ret;
+
+ WARN_ON(len > host->data_max);
+
+ operation = gb_operation_create(host->connection, GB_SDIO_TYPE_TRANSFER,
+ len + sizeof(*request),
+ sizeof(*response), GFP_KERNEL);
+ if (!operation)
+ return -ENOMEM;
+
+ request = operation->request->payload;
+ request->data_flags = (data->flags >> 8);
+ request->data_blocks = cpu_to_le16(nblocks);
+ request->data_blksz = cpu_to_le16(data->blksz);
+
+ copied = sg_pcopy_to_buffer(sg, sg_len, &request->data[0], len, skip);
+
+ if (copied != len) {
+ ret = -EINVAL;
+ goto err_put_operation;
+ }
+
+ ret = gb_operation_request_send_sync(operation);
+ if (ret < 0)
+ goto err_put_operation;
+
+ response = operation->response->payload;
+
+ send_blocks = le16_to_cpu(response->data_blocks);
+ send_blksz = le16_to_cpu(response->data_blksz);
+
+ if (len != send_blksz * send_blocks) {
+ dev_err(mmc_dev(host->mmc), "send: size received: %zu != %d\n",
+ len, send_blksz * send_blocks);
+ ret = -EINVAL;
+ }
+
+err_put_operation:
+ gb_operation_put(operation);
+
+ return ret;
+}
+
+static int _gb_sdio_recv(struct gb_sdio_host *host, struct mmc_data *data,
+ size_t len, u16 nblocks, off_t skip)
+{
+ struct gb_sdio_transfer_request *request;
+ struct gb_sdio_transfer_response *response;
+ struct gb_operation *operation;
+ struct scatterlist *sg = data->sg;
+ unsigned int sg_len = data->sg_len;
+ size_t copied;
+ u16 recv_blksz;
+ u16 recv_blocks;
+ int ret;
+
+ WARN_ON(len > host->data_max);
+
+ operation = gb_operation_create(host->connection, GB_SDIO_TYPE_TRANSFER,
+ sizeof(*request),
+ len + sizeof(*response), GFP_KERNEL);
+ if (!operation)
+ return -ENOMEM;
+
+ request = operation->request->payload;
+ request->data_flags = (data->flags >> 8);
+ request->data_blocks = cpu_to_le16(nblocks);
+ request->data_blksz = cpu_to_le16(data->blksz);
+
+ ret = gb_operation_request_send_sync(operation);
+ if (ret < 0)
+ goto err_put_operation;
+
+ response = operation->response->payload;
+ recv_blocks = le16_to_cpu(response->data_blocks);
+ recv_blksz = le16_to_cpu(response->data_blksz);
+
+ if (len != recv_blksz * recv_blocks) {
+ dev_err(mmc_dev(host->mmc), "recv: size received: %d != %zu\n",
+ recv_blksz * recv_blocks, len);
+ ret = -EINVAL;
+ goto err_put_operation;
+ }
+
+ copied = sg_pcopy_from_buffer(sg, sg_len, &response->data[0], len,
+ skip);
+ if (copied != len)
+ ret = -EINVAL;
+
+err_put_operation:
+ gb_operation_put(operation);
+
+ return ret;
+}
+
+static int gb_sdio_transfer(struct gb_sdio_host *host, struct mmc_data *data)
+{
+ size_t left, len;
+ off_t skip = 0;
+ int ret = 0;
+ u16 nblocks;
+
+ if (single_op(data->mrq->cmd) && data->blocks > 1) {
+ ret = -ETIMEDOUT;
+ goto out;
+ }
+
+ left = data->blksz * data->blocks;
+
+ while (left) {
+ /* check is a stop transmission is pending */
+ spin_lock(&host->xfer);
+ if (host->xfer_stop) {
+ host->xfer_stop = false;
+ spin_unlock(&host->xfer);
+ ret = -EINTR;
+ goto out;
+ }
+ spin_unlock(&host->xfer);
+ len = min(left, host->data_max);
+ nblocks = len / data->blksz;
+ len = nblocks * data->blksz;
+
+ if (data->flags & MMC_DATA_READ) {
+ ret = _gb_sdio_recv(host, data, len, nblocks, skip);
+ if (ret < 0)
+ goto out;
+ } else {
+ ret = _gb_sdio_send(host, data, len, nblocks, skip);
+ if (ret < 0)
+ goto out;
+ }
+ data->bytes_xfered += len;
+ left -= len;
+ skip += len;
+ }
+
+out:
+ data->error = ret;
+ return ret;
+}
+
+static int gb_sdio_command(struct gb_sdio_host *host, struct mmc_command *cmd)
+{
+ struct gb_sdio_command_request request = {0};
+ struct gb_sdio_command_response response;
+ struct mmc_data *data = host->mrq->data;
+ u8 cmd_flags;
+ u8 cmd_type;
+ int i;
+ int ret;
+
+ switch (mmc_resp_type(cmd)) {
+ case MMC_RSP_NONE:
+ cmd_flags = GB_SDIO_RSP_NONE;
+ break;
+ case MMC_RSP_R1:
+ cmd_flags = GB_SDIO_RSP_R1_R5_R6_R7;
+ break;
+ case MMC_RSP_R1B:
+ cmd_flags = GB_SDIO_RSP_R1B;
+ break;
+ case MMC_RSP_R2:
+ cmd_flags = GB_SDIO_RSP_R2;
+ break;
+ case MMC_RSP_R3:
+ cmd_flags = GB_SDIO_RSP_R3_R4;
+ break;
+ default:
+ dev_err(mmc_dev(host->mmc), "cmd flag invalid 0x%04x\n",
+ mmc_resp_type(cmd));
+ ret = -EINVAL;
+ goto out;
+ }
+
+ switch (mmc_cmd_type(cmd)) {
+ case MMC_CMD_BC:
+ cmd_type = GB_SDIO_CMD_BC;
+ break;
+ case MMC_CMD_BCR:
+ cmd_type = GB_SDIO_CMD_BCR;
+ break;
+ case MMC_CMD_AC:
+ cmd_type = GB_SDIO_CMD_AC;
+ break;
+ case MMC_CMD_ADTC:
+ cmd_type = GB_SDIO_CMD_ADTC;
+ break;
+ default:
+ dev_err(mmc_dev(host->mmc), "cmd type invalid 0x%04x\n",
+ mmc_cmd_type(cmd));
+ ret = -EINVAL;
+ goto out;
+ }
+
+ request.cmd = cmd->opcode;
+ request.cmd_flags = cmd_flags;
+ request.cmd_type = cmd_type;
+ request.cmd_arg = cpu_to_le32(cmd->arg);
+ /* some controllers need to know at command time data details */
+ if (data) {
+ request.data_blocks = cpu_to_le16(data->blocks);
+ request.data_blksz = cpu_to_le16(data->blksz);
+ }
+
+ ret = gb_operation_sync(host->connection, GB_SDIO_TYPE_COMMAND,
+ &request, sizeof(request), &response,
+ sizeof(response));
+ if (ret < 0)
+ goto out;
+
+ /* no response expected */
+ if (cmd_flags & GB_SDIO_RSP_NONE)
+ goto out;
+
+ /* long response expected */
+ if (cmd_flags & GB_SDIO_RSP_R2)
+ for (i = 0; i < 4; i++)
+ cmd->resp[i] = le32_to_cpu(response.resp[i]);
+ else
+ cmd->resp[0] = le32_to_cpu(response.resp[0]);
+
+out:
+ cmd->error = ret;
+ return ret;
+}
+
+static void gb_sdio_mrq_work(struct work_struct *work)
+{
+ struct gb_sdio_host *host;
+ struct mmc_request *mrq;
+ int ret;
+
+ host = container_of(work, struct gb_sdio_host, mrqwork);
+
+ ret = gbphy_runtime_get_sync(host->gbphy_dev);
+ if (ret)
+ return;
+
+ mutex_lock(&host->lock);
+ mrq = host->mrq;
+ if (!mrq) {
+ mutex_unlock(&host->lock);
+ gbphy_runtime_put_autosuspend(host->gbphy_dev);
+ dev_err(mmc_dev(host->mmc), "mmc request is NULL");
+ return;
+ }
+
+ if (host->removed) {
+ mrq->cmd->error = -ESHUTDOWN;
+ goto done;
+ }
+
+ if (mrq->sbc) {
+ ret = gb_sdio_command(host, mrq->sbc);
+ if (ret < 0)
+ goto done;
+ }
+
+ ret = gb_sdio_command(host, mrq->cmd);
+ if (ret < 0)
+ goto done;
+
+ if (mrq->data) {
+ ret = gb_sdio_transfer(host, mrq->data);
+ if (ret < 0)
+ goto done;
+ }
+
+ if (mrq->stop) {
+ ret = gb_sdio_command(host, mrq->stop);
+ if (ret < 0)
+ goto done;
+ }
+
+done:
+ host->mrq = NULL;
+ mutex_unlock(&host->lock);
+ mmc_request_done(host->mmc, mrq);
+ gbphy_runtime_put_autosuspend(host->gbphy_dev);
+}
+
+static void gb_mmc_request(struct mmc_host *mmc, struct mmc_request *mrq)
+{
+ struct gb_sdio_host *host = mmc_priv(mmc);
+ struct mmc_command *cmd = mrq->cmd;
+
+ /* Check if it is a cancel to ongoing transfer */
+ if (cmd->opcode == MMC_STOP_TRANSMISSION) {
+ spin_lock(&host->xfer);
+ host->xfer_stop = true;
+ spin_unlock(&host->xfer);
+ }
+
+ mutex_lock(&host->lock);
+
+ WARN_ON(host->mrq);
+ host->mrq = mrq;
+
+ if (host->removed) {
+ mrq->cmd->error = -ESHUTDOWN;
+ goto out;
+ }
+ if (!host->card_present) {
+ mrq->cmd->error = -ENOMEDIUM;
+ goto out;
+ }
+
+ queue_work(host->mrq_workqueue, &host->mrqwork);
+
+ mutex_unlock(&host->lock);
+ return;
+
+out:
+ host->mrq = NULL;
+ mutex_unlock(&host->lock);
+ mmc_request_done(mmc, mrq);
+}
+
+static void gb_mmc_set_ios(struct mmc_host *mmc, struct mmc_ios *ios)
+{
+ struct gb_sdio_host *host = mmc_priv(mmc);
+ struct gb_sdio_set_ios_request request;
+ int ret;
+ u8 power_mode;
+ u8 bus_width;
+ u8 timing;
+ u8 signal_voltage;
+ u8 drv_type;
+ u32 vdd = 0;
+
+ mutex_lock(&host->lock);
+ request.clock = cpu_to_le32(ios->clock);
+
+ if (ios->vdd)
+ vdd = 1 << (ios->vdd - GB_SDIO_VDD_SHIFT);
+ request.vdd = cpu_to_le32(vdd);
+
+ request.bus_mode = (ios->bus_mode == MMC_BUSMODE_OPENDRAIN ?
+ GB_SDIO_BUSMODE_OPENDRAIN :
+ GB_SDIO_BUSMODE_PUSHPULL);
+
+ switch (ios->power_mode) {
+ case MMC_POWER_OFF:
+ default:
+ power_mode = GB_SDIO_POWER_OFF;
+ break;
+ case MMC_POWER_UP:
+ power_mode = GB_SDIO_POWER_UP;
+ break;
+ case MMC_POWER_ON:
+ power_mode = GB_SDIO_POWER_ON;
+ break;
+ case MMC_POWER_UNDEFINED:
+ power_mode = GB_SDIO_POWER_UNDEFINED;
+ break;
+ }
+ request.power_mode = power_mode;
+
+ switch (ios->bus_width) {
+ case MMC_BUS_WIDTH_1:
+ bus_width = GB_SDIO_BUS_WIDTH_1;
+ break;
+ case MMC_BUS_WIDTH_4:
+ default:
+ bus_width = GB_SDIO_BUS_WIDTH_4;
+ break;
+ case MMC_BUS_WIDTH_8:
+ bus_width = GB_SDIO_BUS_WIDTH_8;
+ break;
+ }
+ request.bus_width = bus_width;
+
+ switch (ios->timing) {
+ case MMC_TIMING_LEGACY:
+ default:
+ timing = GB_SDIO_TIMING_LEGACY;
+ break;
+ case MMC_TIMING_MMC_HS:
+ timing = GB_SDIO_TIMING_MMC_HS;
+ break;
+ case MMC_TIMING_SD_HS:
+ timing = GB_SDIO_TIMING_SD_HS;
+ break;
+ case MMC_TIMING_UHS_SDR12:
+ timing = GB_SDIO_TIMING_UHS_SDR12;
+ break;
+ case MMC_TIMING_UHS_SDR25:
+ timing = GB_SDIO_TIMING_UHS_SDR25;
+ break;
+ case MMC_TIMING_UHS_SDR50:
+ timing = GB_SDIO_TIMING_UHS_SDR50;
+ break;
+ case MMC_TIMING_UHS_SDR104:
+ timing = GB_SDIO_TIMING_UHS_SDR104;
+ break;
+ case MMC_TIMING_UHS_DDR50:
+ timing = GB_SDIO_TIMING_UHS_DDR50;
+ break;
+ case MMC_TIMING_MMC_DDR52:
+ timing = GB_SDIO_TIMING_MMC_DDR52;
+ break;
+ case MMC_TIMING_MMC_HS200:
+ timing = GB_SDIO_TIMING_MMC_HS200;
+ break;
+ case MMC_TIMING_MMC_HS400:
+ timing = GB_SDIO_TIMING_MMC_HS400;
+ break;
+ }
+ request.timing = timing;
+
+ switch (ios->signal_voltage) {
+ case MMC_SIGNAL_VOLTAGE_330:
+ signal_voltage = GB_SDIO_SIGNAL_VOLTAGE_330;
+ break;
+ case MMC_SIGNAL_VOLTAGE_180:
+ default:
+ signal_voltage = GB_SDIO_SIGNAL_VOLTAGE_180;
+ break;
+ case MMC_SIGNAL_VOLTAGE_120:
+ signal_voltage = GB_SDIO_SIGNAL_VOLTAGE_120;
+ break;
+ }
+ request.signal_voltage = signal_voltage;
+
+ switch (ios->drv_type) {
+ case MMC_SET_DRIVER_TYPE_A:
+ drv_type = GB_SDIO_SET_DRIVER_TYPE_A;
+ break;
+ case MMC_SET_DRIVER_TYPE_C:
+ drv_type = GB_SDIO_SET_DRIVER_TYPE_C;
+ break;
+ case MMC_SET_DRIVER_TYPE_D:
+ drv_type = GB_SDIO_SET_DRIVER_TYPE_D;
+ break;
+ case MMC_SET_DRIVER_TYPE_B:
+ default:
+ drv_type = GB_SDIO_SET_DRIVER_TYPE_B;
+ break;
+ }
+ request.drv_type = drv_type;
+
+ ret = gb_sdio_set_ios(host, &request);
+ if (ret < 0)
+ goto out;
+
+ memcpy(&mmc->ios, ios, sizeof(mmc->ios));
+
+out:
+ mutex_unlock(&host->lock);
+}
+
+static int gb_mmc_get_ro(struct mmc_host *mmc)
+{
+ struct gb_sdio_host *host = mmc_priv(mmc);
+
+ mutex_lock(&host->lock);
+ if (host->removed) {
+ mutex_unlock(&host->lock);
+ return -ESHUTDOWN;
+ }
+ mutex_unlock(&host->lock);
+
+ return host->read_only;
+}
+
+static int gb_mmc_get_cd(struct mmc_host *mmc)
+{
+ struct gb_sdio_host *host = mmc_priv(mmc);
+
+ mutex_lock(&host->lock);
+ if (host->removed) {
+ mutex_unlock(&host->lock);
+ return -ESHUTDOWN;
+ }
+ mutex_unlock(&host->lock);
+
+ return host->card_present;
+}
+
+static int gb_mmc_switch_voltage(struct mmc_host *mmc, struct mmc_ios *ios)
+{
+ return 0;
+}
+
+static const struct mmc_host_ops gb_sdio_ops = {
+ .request = gb_mmc_request,
+ .set_ios = gb_mmc_set_ios,
+ .get_ro = gb_mmc_get_ro,
+ .get_cd = gb_mmc_get_cd,
+ .start_signal_voltage_switch = gb_mmc_switch_voltage,
+};
+
+static int gb_sdio_probe(struct gbphy_device *gbphy_dev,
+ const struct gbphy_device_id *id)
+{
+ struct gb_connection *connection;
+ struct mmc_host *mmc;
+ struct gb_sdio_host *host;
+ int ret = 0;
+
+ mmc = mmc_alloc_host(sizeof(*host), &gbphy_dev->dev);
+ if (!mmc)
+ return -ENOMEM;
+
+ connection = gb_connection_create(gbphy_dev->bundle,
+ le16_to_cpu(gbphy_dev->cport_desc->id),
+ gb_sdio_request_handler);
+ if (IS_ERR(connection)) {
+ ret = PTR_ERR(connection);
+ goto exit_mmc_free;
+ }
+
+ host = mmc_priv(mmc);
+ host->mmc = mmc;
+ host->removed = true;
+
+ host->connection = connection;
+ gb_connection_set_data(connection, host);
+ host->gbphy_dev = gbphy_dev;
+ gb_gbphy_set_data(gbphy_dev, host);
+
+ ret = gb_connection_enable_tx(connection);
+ if (ret)
+ goto exit_connection_destroy;
+
+ ret = gb_sdio_get_caps(host);
+ if (ret < 0)
+ goto exit_connection_disable;
+
+ mmc->ops = &gb_sdio_ops;
+
+ mmc->max_segs = host->mmc->max_blk_count;
+
+ /* for now we make a map 1:1 between max request and segment size */
+ mmc->max_req_size = mmc->max_blk_size * mmc->max_blk_count;
+ mmc->max_seg_size = mmc->max_req_size;
+
+ mutex_init(&host->lock);
+ spin_lock_init(&host->xfer);
+ host->mrq_workqueue = alloc_workqueue("mmc-%s", 0, 1,
+ dev_name(&gbphy_dev->dev));
+ if (!host->mrq_workqueue) {
+ ret = -ENOMEM;
+ goto exit_connection_disable;
+ }
+ INIT_WORK(&host->mrqwork, gb_sdio_mrq_work);
+
+ ret = gb_connection_enable(connection);
+ if (ret)
+ goto exit_wq_destroy;
+
+ ret = mmc_add_host(mmc);
+ if (ret < 0)
+ goto exit_wq_destroy;
+ host->removed = false;
+ ret = _gb_sdio_process_events(host, host->queued_events);
+ host->queued_events = 0;
+
+ gbphy_runtime_put_autosuspend(gbphy_dev);
+
+ return ret;
+
+exit_wq_destroy:
+ destroy_workqueue(host->mrq_workqueue);
+exit_connection_disable:
+ gb_connection_disable(connection);
+exit_connection_destroy:
+ gb_connection_destroy(connection);
+exit_mmc_free:
+ mmc_free_host(mmc);
+
+ return ret;
+}
+
+static void gb_sdio_remove(struct gbphy_device *gbphy_dev)
+{
+ struct gb_sdio_host *host = gb_gbphy_get_data(gbphy_dev);
+ struct gb_connection *connection = host->connection;
+ struct mmc_host *mmc;
+ int ret;
+
+ ret = gbphy_runtime_get_sync(gbphy_dev);
+ if (ret)
+ gbphy_runtime_get_noresume(gbphy_dev);
+
+ mutex_lock(&host->lock);
+ host->removed = true;
+ mmc = host->mmc;
+ gb_connection_set_data(connection, NULL);
+ mutex_unlock(&host->lock);
+
+ flush_workqueue(host->mrq_workqueue);
+ destroy_workqueue(host->mrq_workqueue);
+ gb_connection_disable_rx(connection);
+ mmc_remove_host(mmc);
+ gb_connection_disable(connection);
+ gb_connection_destroy(connection);
+ mmc_free_host(mmc);
+}
+
+static const struct gbphy_device_id gb_sdio_id_table[] = {
+ { GBPHY_PROTOCOL(GREYBUS_PROTOCOL_SDIO) },
+ { },
+};
+MODULE_DEVICE_TABLE(gbphy, gb_sdio_id_table);
+
+static struct gbphy_driver sdio_driver = {
+ .name = "sdio",
+ .probe = gb_sdio_probe,
+ .remove = gb_sdio_remove,
+ .id_table = gb_sdio_id_table,
+};
+
+module_gbphy_driver(sdio_driver);
+MODULE_LICENSE("GPL v2");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 27/32] greybus: bridged phy spi driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (24 preceding siblings ...)
2016-09-16 14:21 ` [patch 26/32] greybus: bridged phy sdio driver Greg KH
@ 2016-09-16 14:21 ` Greg KH
2016-09-16 14:21 ` [patch 28/32] greybus: bridged phy uart driver Greg KH
` (4 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:21 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus bridged phy spi class protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/spi.c | 79 ++++++
drivers/greybus/spilib.c | 565 +++++++++++++++++++++++++++++++++++++++++++++++
drivers/greybus/spilib.h | 24 +
3 files changed, 668 insertions(+)
--- /dev/null
+++ b/drivers/greybus/spi.c
@@ -0,0 +1,79 @@
+/*
+ * SPI bridge PHY driver.
+ *
+ * Copyright 2014-2016 Google Inc.
+ * Copyright 2014-2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/module.h>
+
+#include "greybus.h"
+#include "gbphy.h"
+#include "spilib.h"
+
+static struct spilib_ops *spilib_ops;
+
+static int gb_spi_probe(struct gbphy_device *gbphy_dev,
+ const struct gbphy_device_id *id)
+{
+ struct gb_connection *connection;
+ int ret;
+
+ connection = gb_connection_create(gbphy_dev->bundle,
+ le16_to_cpu(gbphy_dev->cport_desc->id),
+ NULL);
+ if (IS_ERR(connection))
+ return PTR_ERR(connection);
+
+ ret = gb_connection_enable(connection);
+ if (ret)
+ goto exit_connection_destroy;
+
+ ret = gb_spilib_master_init(connection, &gbphy_dev->dev, spilib_ops);
+ if (ret)
+ goto exit_connection_disable;
+
+ gb_gbphy_set_data(gbphy_dev, connection);
+
+ gbphy_runtime_put_autosuspend(gbphy_dev);
+ return 0;
+
+exit_connection_disable:
+ gb_connection_disable(connection);
+exit_connection_destroy:
+ gb_connection_destroy(connection);
+
+ return ret;
+}
+
+static void gb_spi_remove(struct gbphy_device *gbphy_dev)
+{
+ struct gb_connection *connection = gb_gbphy_get_data(gbphy_dev);
+ int ret;
+
+ ret = gbphy_runtime_get_sync(gbphy_dev);
+ if (ret)
+ gbphy_runtime_get_noresume(gbphy_dev);
+
+ gb_spilib_master_exit(connection);
+ gb_connection_disable(connection);
+ gb_connection_destroy(connection);
+}
+
+static const struct gbphy_device_id gb_spi_id_table[] = {
+ { GBPHY_PROTOCOL(GREYBUS_PROTOCOL_SPI) },
+ { },
+};
+MODULE_DEVICE_TABLE(gbphy, gb_spi_id_table);
+
+static struct gbphy_driver spi_driver = {
+ .name = "spi",
+ .probe = gb_spi_probe,
+ .remove = gb_spi_remove,
+ .id_table = gb_spi_id_table,
+};
+
+module_gbphy_driver(spi_driver);
+MODULE_LICENSE("GPL v2");
--- /dev/null
+++ b/drivers/greybus/spilib.c
@@ -0,0 +1,565 @@
+/*
+ * Greybus SPI library
+ *
+ * Copyright 2014-2016 Google Inc.
+ * Copyright 2014-2016 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ */
+
+#include <linux/bitops.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/spi/spi.h>
+
+#include "greybus.h"
+#include "spilib.h"
+
+struct gb_spilib {
+ struct gb_connection *connection;
+ struct device *parent;
+ struct spi_transfer *first_xfer;
+ struct spi_transfer *last_xfer;
+ struct spilib_ops *ops;
+ u32 rx_xfer_offset;
+ u32 tx_xfer_offset;
+ u32 last_xfer_size;
+ unsigned int op_timeout;
+ u16 mode;
+ u16 flags;
+ u32 bits_per_word_mask;
+ u8 num_chipselect;
+ u32 min_speed_hz;
+ u32 max_speed_hz;
+};
+
+#define GB_SPI_STATE_MSG_DONE ((void *)0)
+#define GB_SPI_STATE_MSG_IDLE ((void *)1)
+#define GB_SPI_STATE_MSG_RUNNING ((void *)2)
+#define GB_SPI_STATE_OP_READY ((void *)3)
+#define GB_SPI_STATE_OP_DONE ((void *)4)
+#define GB_SPI_STATE_MSG_ERROR ((void *)-1)
+
+#define XFER_TIMEOUT_TOLERANCE 200
+
+static struct spi_master *get_master_from_spi(struct gb_spilib *spi)
+{
+ return gb_connection_get_data(spi->connection);
+}
+
+static int tx_header_fit_operation(u32 tx_size, u32 count, size_t data_max)
+{
+ size_t headers_size;
+
+ data_max -= sizeof(struct gb_spi_transfer_request);
+ headers_size = (count + 1) * sizeof(struct gb_spi_transfer);
+
+ return tx_size + headers_size > data_max ? 0 : 1;
+}
+
+static size_t calc_rx_xfer_size(u32 rx_size, u32 *tx_xfer_size, u32 len,
+ size_t data_max)
+{
+ size_t rx_xfer_size;
+
+ data_max -= sizeof(struct gb_spi_transfer_response);
+
+ if (rx_size + len > data_max)
+ rx_xfer_size = data_max - rx_size;
+ else
+ rx_xfer_size = len;
+
+ /* if this is a write_read, for symmetry read the same as write */
+ if (*tx_xfer_size && rx_xfer_size > *tx_xfer_size)
+ rx_xfer_size = *tx_xfer_size;
+ if (*tx_xfer_size && rx_xfer_size < *tx_xfer_size)
+ *tx_xfer_size = rx_xfer_size;
+
+ return rx_xfer_size;
+}
+
+static size_t calc_tx_xfer_size(u32 tx_size, u32 count, size_t len,
+ size_t data_max)
+{
+ size_t headers_size;
+
+ data_max -= sizeof(struct gb_spi_transfer_request);
+ headers_size = (count + 1) * sizeof(struct gb_spi_transfer);
+
+ if (tx_size + headers_size + len > data_max)
+ return data_max - (tx_size + sizeof(struct gb_spi_transfer));
+
+ return len;
+}
+
+static void clean_xfer_state(struct gb_spilib *spi)
+{
+ spi->first_xfer = NULL;
+ spi->last_xfer = NULL;
+ spi->rx_xfer_offset = 0;
+ spi->tx_xfer_offset = 0;
+ spi->last_xfer_size = 0;
+ spi->op_timeout = 0;
+}
+
+static bool is_last_xfer_done(struct gb_spilib *spi)
+{
+ struct spi_transfer *last_xfer = spi->last_xfer;
+
+ if ((spi->tx_xfer_offset + spi->last_xfer_size == last_xfer->len) ||
+ (spi->rx_xfer_offset + spi->last_xfer_size == last_xfer->len))
+ return true;
+
+ return false;
+}
+
+static int setup_next_xfer(struct gb_spilib *spi, struct spi_message *msg)
+{
+ struct spi_transfer *last_xfer = spi->last_xfer;
+
+ if (msg->state != GB_SPI_STATE_OP_DONE)
+ return 0;
+
+ /*
+ * if we transferred all content of the last transfer, reset values and
+ * check if this was the last transfer in the message
+ */
+ if (is_last_xfer_done(spi)) {
+ spi->tx_xfer_offset = 0;
+ spi->rx_xfer_offset = 0;
+ spi->op_timeout = 0;
+ if (last_xfer == list_last_entry(&msg->transfers,
+ struct spi_transfer,
+ transfer_list))
+ msg->state = GB_SPI_STATE_MSG_DONE;
+ else
+ spi->first_xfer = list_next_entry(last_xfer,
+ transfer_list);
+ return 0;
+ }
+
+ spi->first_xfer = last_xfer;
+ if (last_xfer->tx_buf)
+ spi->tx_xfer_offset += spi->last_xfer_size;
+
+ if (last_xfer->rx_buf)
+ spi->rx_xfer_offset += spi->last_xfer_size;
+
+ return 0;
+}
+
+static struct spi_transfer *get_next_xfer(struct spi_transfer *xfer,
+ struct spi_message *msg)
+{
+ if (xfer == list_last_entry(&msg->transfers, struct spi_transfer,
+ transfer_list))
+ return NULL;
+
+ return list_next_entry(xfer, transfer_list);
+}
+
+/* Routines to transfer data */
+static struct gb_operation *gb_spi_operation_create(struct gb_spilib *spi,
+ struct gb_connection *connection, struct spi_message *msg)
+{
+ struct gb_spi_transfer_request *request;
+ struct spi_device *dev = msg->spi;
+ struct spi_transfer *xfer;
+ struct gb_spi_transfer *gb_xfer;
+ struct gb_operation *operation;
+ u32 tx_size = 0, rx_size = 0, count = 0, xfer_len = 0, request_size;
+ u32 tx_xfer_size = 0, rx_xfer_size = 0, len;
+ u32 total_len = 0;
+ unsigned int xfer_timeout;
+ size_t data_max;
+ void *tx_data;
+
+ data_max = gb_operation_get_payload_size_max(connection);
+ xfer = spi->first_xfer;
+
+ /* Find number of transfers queued and tx/rx length in the message */
+
+ while (msg->state != GB_SPI_STATE_OP_READY) {
+ msg->state = GB_SPI_STATE_MSG_RUNNING;
+ spi->last_xfer = xfer;
+
+ if (!xfer->tx_buf && !xfer->rx_buf) {
+ dev_err(spi->parent,
+ "bufferless transfer, length %u\n", xfer->len);
+ msg->state = GB_SPI_STATE_MSG_ERROR;
+ return NULL;
+ }
+
+ tx_xfer_size = 0;
+ rx_xfer_size = 0;
+
+ if (xfer->tx_buf) {
+ len = xfer->len - spi->tx_xfer_offset;
+ if (!tx_header_fit_operation(tx_size, count, data_max))
+ break;
+ tx_xfer_size = calc_tx_xfer_size(tx_size, count,
+ len, data_max);
+ spi->last_xfer_size = tx_xfer_size;
+ }
+
+ if (xfer->rx_buf) {
+ len = xfer->len - spi->rx_xfer_offset;
+ rx_xfer_size = calc_rx_xfer_size(rx_size, &tx_xfer_size,
+ len, data_max);
+ spi->last_xfer_size = rx_xfer_size;
+ }
+
+ tx_size += tx_xfer_size;
+ rx_size += rx_xfer_size;
+
+ total_len += spi->last_xfer_size;
+ count++;
+
+ xfer = get_next_xfer(xfer, msg);
+ if (!xfer || total_len >= data_max)
+ msg->state = GB_SPI_STATE_OP_READY;
+ }
+
+ /*
+ * In addition to space for all message descriptors we need
+ * to have enough to hold all tx data.
+ */
+ request_size = sizeof(*request);
+ request_size += count * sizeof(*gb_xfer);
+ request_size += tx_size;
+
+ /* Response consists only of incoming data */
+ operation = gb_operation_create(connection, GB_SPI_TYPE_TRANSFER,
+ request_size, rx_size, GFP_KERNEL);
+ if (!operation)
+ return NULL;
+
+ request = operation->request->payload;
+ request->count = cpu_to_le16(count);
+ request->mode = dev->mode;
+ request->chip_select = dev->chip_select;
+
+ gb_xfer = &request->transfers[0];
+ tx_data = gb_xfer + count; /* place tx data after last gb_xfer */
+
+ /* Fill in the transfers array */
+ xfer = spi->first_xfer;
+ while (msg->state != GB_SPI_STATE_OP_DONE) {
+ if (xfer == spi->last_xfer)
+ xfer_len = spi->last_xfer_size;
+ else
+ xfer_len = xfer->len;
+
+ /* make sure we do not timeout in a slow transfer */
+ xfer_timeout = xfer_len * 8 * MSEC_PER_SEC / xfer->speed_hz;
+ xfer_timeout += GB_OPERATION_TIMEOUT_DEFAULT;
+
+ if (xfer_timeout > spi->op_timeout)
+ spi->op_timeout = xfer_timeout;
+
+ gb_xfer->speed_hz = cpu_to_le32(xfer->speed_hz);
+ gb_xfer->len = cpu_to_le32(xfer_len);
+ gb_xfer->delay_usecs = cpu_to_le16(xfer->delay_usecs);
+ gb_xfer->cs_change = xfer->cs_change;
+ gb_xfer->bits_per_word = xfer->bits_per_word;
+
+ /* Copy tx data */
+ if (xfer->tx_buf) {
+ gb_xfer->xfer_flags |= GB_SPI_XFER_WRITE;
+ memcpy(tx_data, xfer->tx_buf + spi->tx_xfer_offset,
+ xfer_len);
+ tx_data += xfer_len;
+ }
+
+ if (xfer->rx_buf)
+ gb_xfer->xfer_flags |= GB_SPI_XFER_READ;
+
+ if (xfer == spi->last_xfer) {
+ if (!is_last_xfer_done(spi))
+ gb_xfer->xfer_flags |= GB_SPI_XFER_INPROGRESS;
+ msg->state = GB_SPI_STATE_OP_DONE;
+ continue;
+ }
+
+ gb_xfer++;
+ xfer = get_next_xfer(xfer, msg);
+ }
+
+ msg->actual_length += total_len;
+
+ return operation;
+}
+
+static void gb_spi_decode_response(struct gb_spilib *spi,
+ struct spi_message *msg,
+ struct gb_spi_transfer_response *response)
+{
+ struct spi_transfer *xfer = spi->first_xfer;
+ void *rx_data = response->data;
+ u32 xfer_len;
+
+ while (xfer) {
+ /* Copy rx data */
+ if (xfer->rx_buf) {
+ if (xfer == spi->first_xfer)
+ xfer_len = xfer->len - spi->rx_xfer_offset;
+ else if (xfer == spi->last_xfer)
+ xfer_len = spi->last_xfer_size;
+ else
+ xfer_len = xfer->len;
+
+ memcpy(xfer->rx_buf + spi->rx_xfer_offset, rx_data,
+ xfer_len);
+ rx_data += xfer_len;
+ }
+
+ if (xfer == spi->last_xfer)
+ break;
+
+ xfer = list_next_entry(xfer, transfer_list);
+ }
+}
+
+static int gb_spi_transfer_one_message(struct spi_master *master,
+ struct spi_message *msg)
+{
+ struct gb_spilib *spi = spi_master_get_devdata(master);
+ struct gb_connection *connection = spi->connection;
+ struct gb_spi_transfer_response *response;
+ struct gb_operation *operation;
+ int ret = 0;
+
+ spi->first_xfer = list_first_entry_or_null(&msg->transfers,
+ struct spi_transfer,
+ transfer_list);
+ if (!spi->first_xfer) {
+ ret = -ENOMEM;
+ goto out;
+ }
+
+ msg->state = GB_SPI_STATE_MSG_IDLE;
+
+ while (msg->state != GB_SPI_STATE_MSG_DONE &&
+ msg->state != GB_SPI_STATE_MSG_ERROR) {
+ operation = gb_spi_operation_create(spi, connection, msg);
+ if (!operation) {
+ msg->state = GB_SPI_STATE_MSG_ERROR;
+ ret = -EINVAL;
+ continue;
+ }
+
+ ret = gb_operation_request_send_sync_timeout(operation,
+ spi->op_timeout);
+ if (!ret) {
+ response = operation->response->payload;
+ if (response)
+ gb_spi_decode_response(spi, msg, response);
+ } else {
+ dev_err(spi->parent,
+ "transfer operation failed: %d\n", ret);
+ msg->state = GB_SPI_STATE_MSG_ERROR;
+ }
+
+ gb_operation_put(operation);
+ setup_next_xfer(spi, msg);
+ }
+
+out:
+ msg->status = ret;
+ clean_xfer_state(spi);
+ spi_finalize_current_message(master);
+
+ return ret;
+}
+
+static int gb_spi_prepare_transfer_hardware(struct spi_master *master)
+{
+ struct gb_spilib *spi = spi_master_get_devdata(master);
+
+ return spi->ops->prepare_transfer_hardware(spi->parent);
+}
+
+static int gb_spi_unprepare_transfer_hardware(struct spi_master *master)
+{
+ struct gb_spilib *spi = spi_master_get_devdata(master);
+
+ spi->ops->unprepare_transfer_hardware(spi->parent);
+
+ return 0;
+}
+
+static int gb_spi_setup(struct spi_device *spi)
+{
+ /* Nothing to do for now */
+ return 0;
+}
+
+static void gb_spi_cleanup(struct spi_device *spi)
+{
+ /* Nothing to do for now */
+}
+
+/* Routines to get controller information */
+
+/*
+ * Map Greybus spi mode bits/flags/bpw into Linux ones.
+ * All bits are same for now and so these macro's return same values.
+ */
+#define gb_spi_mode_map(mode) mode
+#define gb_spi_flags_map(flags) flags
+
+static int gb_spi_get_master_config(struct gb_spilib *spi)
+{
+ struct gb_spi_master_config_response response;
+ u16 mode, flags;
+ int ret;
+
+ ret = gb_operation_sync(spi->connection, GB_SPI_TYPE_MASTER_CONFIG,
+ NULL, 0, &response, sizeof(response));
+ if (ret < 0)
+ return ret;
+
+ mode = le16_to_cpu(response.mode);
+ spi->mode = gb_spi_mode_map(mode);
+
+ flags = le16_to_cpu(response.flags);
+ spi->flags = gb_spi_flags_map(flags);
+
+ spi->bits_per_word_mask = le32_to_cpu(response.bits_per_word_mask);
+ spi->num_chipselect = response.num_chipselect;
+
+ spi->min_speed_hz = le32_to_cpu(response.min_speed_hz);
+ spi->max_speed_hz = le32_to_cpu(response.max_speed_hz);
+
+ return 0;
+}
+
+static int gb_spi_setup_device(struct gb_spilib *spi, u8 cs)
+{
+ struct spi_master *master = get_master_from_spi(spi);
+ struct gb_spi_device_config_request request;
+ struct gb_spi_device_config_response response;
+ struct spi_board_info spi_board = { {0} };
+ struct spi_device *spidev;
+ int ret;
+ u8 dev_type;
+
+ request.chip_select = cs;
+
+ ret = gb_operation_sync(spi->connection, GB_SPI_TYPE_DEVICE_CONFIG,
+ &request, sizeof(request),
+ &response, sizeof(response));
+ if (ret < 0)
+ return ret;
+
+ dev_type = response.device_type;
+
+ if (dev_type == GB_SPI_SPI_DEV)
+ strlcpy(spi_board.modalias, "spidev",
+ sizeof(spi_board.modalias));
+ else if (dev_type == GB_SPI_SPI_NOR)
+ strlcpy(spi_board.modalias, "spi-nor",
+ sizeof(spi_board.modalias));
+ else if (dev_type == GB_SPI_SPI_MODALIAS)
+ memcpy(spi_board.modalias, response.name,
+ sizeof(spi_board.modalias));
+ else
+ return -EINVAL;
+
+ spi_board.mode = le16_to_cpu(response.mode);
+ spi_board.bus_num = master->bus_num;
+ spi_board.chip_select = cs;
+ spi_board.max_speed_hz = le32_to_cpu(response.max_speed_hz);
+
+ spidev = spi_new_device(master, &spi_board);
+ if (!spidev)
+ return -EINVAL;
+
+ return 0;
+}
+
+int gb_spilib_master_init(struct gb_connection *connection, struct device *dev,
+ struct spilib_ops *ops)
+{
+ struct gb_spilib *spi;
+ struct spi_master *master;
+ int ret;
+ u8 i;
+
+ /* Allocate master with space for data */
+ master = spi_alloc_master(dev, sizeof(*spi));
+ if (!master) {
+ dev_err(dev, "cannot alloc SPI master\n");
+ return -ENOMEM;
+ }
+
+ spi = spi_master_get_devdata(master);
+ spi->connection = connection;
+ gb_connection_set_data(connection, master);
+ spi->parent = dev;
+ spi->ops = ops;
+
+ /* get master configuration */
+ ret = gb_spi_get_master_config(spi);
+ if (ret)
+ goto exit_spi_put;
+
+ master->bus_num = -1; /* Allow spi-core to allocate it dynamically */
+ master->num_chipselect = spi->num_chipselect;
+ master->mode_bits = spi->mode;
+ master->flags = spi->flags;
+ master->bits_per_word_mask = spi->bits_per_word_mask;
+
+ /* Attach methods */
+ master->cleanup = gb_spi_cleanup;
+ master->setup = gb_spi_setup;
+ master->transfer_one_message = gb_spi_transfer_one_message;
+
+ if (ops && ops->prepare_transfer_hardware) {
+ master->prepare_transfer_hardware =
+ gb_spi_prepare_transfer_hardware;
+ }
+
+ if (ops && ops->unprepare_transfer_hardware) {
+ master->unprepare_transfer_hardware =
+ gb_spi_unprepare_transfer_hardware;
+ }
+
+ master->auto_runtime_pm = true;
+
+ ret = spi_register_master(master);
+ if (ret < 0)
+ goto exit_spi_put;
+
+ /* now, fetch the devices configuration */
+ for (i = 0; i < spi->num_chipselect; i++) {
+ ret = gb_spi_setup_device(spi, i);
+ if (ret < 0) {
+ dev_err(dev, "failed to allocate spi device %d: %d\n",
+ i, ret);
+ goto exit_spi_unregister;
+ }
+ }
+
+ return 0;
+
+exit_spi_unregister:
+ spi_unregister_master(master);
+exit_spi_put:
+ spi_master_put(master);
+
+ return ret;
+}
+EXPORT_SYMBOL_GPL(gb_spilib_master_init);
+
+void gb_spilib_master_exit(struct gb_connection *connection)
+{
+ struct spi_master *master = gb_connection_get_data(connection);
+
+ spi_unregister_master(master);
+ spi_master_put(master);
+}
+EXPORT_SYMBOL_GPL(gb_spilib_master_exit);
+
+MODULE_LICENSE("GPL v2");
--- /dev/null
+++ b/drivers/greybus/spilib.h
@@ -0,0 +1,24 @@
+/*
+ * Greybus SPI library header
+ *
+ * copyright 2016 google inc.
+ * copyright 2016 linaro ltd.
+ *
+ * released under the gplv2 only.
+ */
+
+#ifndef __SPILIB_H
+#define __SPILIB_H
+
+struct device;
+struct gb_connection;
+
+struct spilib_ops {
+ int (*prepare_transfer_hardware)(struct device *dev);
+ void (*unprepare_transfer_hardware)(struct device *dev);
+};
+
+int gb_spilib_master_init(struct gb_connection *connection, struct device *dev, struct spilib_ops *ops);
+void gb_spilib_master_exit(struct gb_connection *connection);
+
+#endif /* __SPILIB_H */
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 28/32] greybus: bridged phy uart driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (25 preceding siblings ...)
2016-09-16 14:21 ` [patch 27/32] greybus: bridged phy spi driver Greg KH
@ 2016-09-16 14:21 ` Greg KH
2016-09-16 14:21 ` [patch 29/32] greybus: bridged phy usb driver Greg KH
` (3 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:21 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus bridged phy tty/uart class protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/uart.c | 1075 +++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 1075 insertions(+)
--- /dev/null
+++ b/drivers/greybus/uart.c
@@ -0,0 +1,1075 @@
+/*
+ * UART driver for the Greybus "generic" UART module.
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ *
+ * Heavily based on drivers/usb/class/cdc-acm.c and
+ * drivers/usb/serial/usb-serial.c.
+ */
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/wait.h>
+#include <linux/slab.h>
+#include <linux/uaccess.h>
+#include <linux/mutex.h>
+#include <linux/tty.h>
+#include <linux/serial.h>
+#include <linux/tty_driver.h>
+#include <linux/tty_flip.h>
+#include <linux/serial.h>
+#include <linux/idr.h>
+#include <linux/fs.h>
+#include <linux/kdev_t.h>
+#include <linux/kfifo.h>
+#include <linux/workqueue.h>
+#include <linux/completion.h>
+
+#include "greybus.h"
+#include "gbphy.h"
+
+#define GB_NUM_MINORS 16 /* 16 is is more than enough */
+#define GB_NAME "ttyGB"
+
+#define GB_UART_WRITE_FIFO_SIZE PAGE_SIZE
+#define GB_UART_WRITE_ROOM_MARGIN 1 /* leave some space in fifo */
+#define GB_UART_FIRMWARE_CREDITS 4096
+#define GB_UART_CREDIT_WAIT_TIMEOUT_MSEC 10000
+
+struct gb_tty_line_coding {
+ __le32 rate;
+ __u8 format;
+ __u8 parity;
+ __u8 data_bits;
+ __u8 flow_control;
+};
+
+struct gb_tty {
+ struct gbphy_device *gbphy_dev;
+ struct tty_port port;
+ void *buffer;
+ size_t buffer_payload_max;
+ struct gb_connection *connection;
+ u16 cport_id;
+ unsigned int minor;
+ unsigned char clocal;
+ bool disconnected;
+ spinlock_t read_lock;
+ spinlock_t write_lock;
+ struct async_icount iocount;
+ struct async_icount oldcount;
+ wait_queue_head_t wioctl;
+ struct mutex mutex;
+ u8 ctrlin; /* input control lines */
+ u8 ctrlout; /* output control lines */
+ struct gb_tty_line_coding line_coding;
+ struct work_struct tx_work;
+ struct kfifo write_fifo;
+ bool close_pending;
+ unsigned int credits;
+ struct completion credits_complete;
+};
+
+static struct tty_driver *gb_tty_driver;
+static DEFINE_IDR(tty_minors);
+static DEFINE_MUTEX(table_lock);
+
+static int gb_uart_receive_data_handler(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct gb_tty *gb_tty = gb_connection_get_data(connection);
+ struct tty_port *port = &gb_tty->port;
+ struct gb_message *request = op->request;
+ struct gb_uart_recv_data_request *receive_data;
+ u16 recv_data_size;
+ int count;
+ unsigned long tty_flags = TTY_NORMAL;
+
+ if (request->payload_size < sizeof(*receive_data)) {
+ dev_err(&gb_tty->gbphy_dev->dev,
+ "short receive-data request received (%zu < %zu)\n",
+ request->payload_size, sizeof(*receive_data));
+ return -EINVAL;
+ }
+
+ receive_data = op->request->payload;
+ recv_data_size = le16_to_cpu(receive_data->size);
+
+ if (recv_data_size != request->payload_size - sizeof(*receive_data)) {
+ dev_err(&gb_tty->gbphy_dev->dev,
+ "malformed receive-data request received (%u != %zu)\n",
+ recv_data_size,
+ request->payload_size - sizeof(*receive_data));
+ return -EINVAL;
+ }
+
+ if (!recv_data_size)
+ return -EINVAL;
+
+ if (receive_data->flags) {
+ if (receive_data->flags & GB_UART_RECV_FLAG_BREAK)
+ tty_flags = TTY_BREAK;
+ else if (receive_data->flags & GB_UART_RECV_FLAG_PARITY)
+ tty_flags = TTY_PARITY;
+ else if (receive_data->flags & GB_UART_RECV_FLAG_FRAMING)
+ tty_flags = TTY_FRAME;
+
+ /* overrun is special, not associated with a char */
+ if (receive_data->flags & GB_UART_RECV_FLAG_OVERRUN)
+ tty_insert_flip_char(port, 0, TTY_OVERRUN);
+ }
+ count = tty_insert_flip_string_fixed_flag(port, receive_data->data,
+ tty_flags, recv_data_size);
+ if (count != recv_data_size) {
+ dev_err(&gb_tty->gbphy_dev->dev,
+ "UART: RX 0x%08x bytes only wrote 0x%08x\n",
+ recv_data_size, count);
+ }
+ if (count)
+ tty_flip_buffer_push(port);
+ return 0;
+}
+
+static int gb_uart_serial_state_handler(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct gb_tty *gb_tty = gb_connection_get_data(connection);
+ struct gb_message *request = op->request;
+ struct gb_uart_serial_state_request *serial_state;
+
+ if (request->payload_size < sizeof(*serial_state)) {
+ dev_err(&gb_tty->gbphy_dev->dev,
+ "short serial-state event received (%zu < %zu)\n",
+ request->payload_size, sizeof(*serial_state));
+ return -EINVAL;
+ }
+
+ serial_state = request->payload;
+ gb_tty->ctrlin = serial_state->control;
+
+ return 0;
+}
+
+static int gb_uart_receive_credits_handler(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct gb_tty *gb_tty = gb_connection_get_data(connection);
+ struct gb_message *request = op->request;
+ struct gb_uart_receive_credits_request *credit_request;
+ unsigned long flags;
+ unsigned int incoming_credits;
+ int ret = 0;
+
+ if (request->payload_size < sizeof(*credit_request)) {
+ dev_err(&gb_tty->gbphy_dev->dev,
+ "short receive_credits event received (%zu < %zu)\n",
+ request->payload_size,
+ sizeof(*credit_request));
+ return -EINVAL;
+ }
+
+ credit_request = request->payload;
+ incoming_credits = le16_to_cpu(credit_request->count);
+
+ spin_lock_irqsave(&gb_tty->write_lock, flags);
+ gb_tty->credits += incoming_credits;
+ if (gb_tty->credits > GB_UART_FIRMWARE_CREDITS) {
+ gb_tty->credits -= incoming_credits;
+ ret = -EINVAL;
+ }
+ spin_unlock_irqrestore(&gb_tty->write_lock, flags);
+
+ if (ret) {
+ dev_err(&gb_tty->gbphy_dev->dev,
+ "invalid number of incoming credits: %d\n",
+ incoming_credits);
+ return ret;
+ }
+
+ if (!gb_tty->close_pending)
+ schedule_work(&gb_tty->tx_work);
+
+ /*
+ * the port the tty layer may be waiting for credits
+ */
+ tty_port_tty_wakeup(&gb_tty->port);
+
+ if (gb_tty->credits == GB_UART_FIRMWARE_CREDITS)
+ complete(&gb_tty->credits_complete);
+
+ return ret;
+}
+
+static int gb_uart_request_handler(struct gb_operation *op)
+{
+ struct gb_connection *connection = op->connection;
+ struct gb_tty *gb_tty = gb_connection_get_data(connection);
+ int type = op->type;
+ int ret;
+
+ switch (type) {
+ case GB_UART_TYPE_RECEIVE_DATA:
+ ret = gb_uart_receive_data_handler(op);
+ break;
+ case GB_UART_TYPE_SERIAL_STATE:
+ ret = gb_uart_serial_state_handler(op);
+ break;
+ case GB_UART_TYPE_RECEIVE_CREDITS:
+ ret = gb_uart_receive_credits_handler(op);
+ break;
+ default:
+ dev_err(&gb_tty->gbphy_dev->dev,
+ "unsupported unsolicited request: 0x%02x\n", type);
+ ret = -EINVAL;
+ }
+
+ return ret;
+}
+
+static void gb_uart_tx_write_work(struct work_struct *work)
+{
+ struct gb_uart_send_data_request *request;
+ struct gb_tty *gb_tty;
+ unsigned long flags;
+ unsigned int send_size;
+ int ret;
+
+ gb_tty = container_of(work, struct gb_tty, tx_work);
+ request = gb_tty->buffer;
+
+ while (1) {
+ if (gb_tty->close_pending)
+ break;
+
+ spin_lock_irqsave(&gb_tty->write_lock, flags);
+ send_size = gb_tty->buffer_payload_max;
+ if (send_size > gb_tty->credits)
+ send_size = gb_tty->credits;
+
+ send_size = kfifo_out_peek(&gb_tty->write_fifo,
+ &request->data[0],
+ send_size);
+ if (!send_size) {
+ spin_unlock_irqrestore(&gb_tty->write_lock, flags);
+ break;
+ }
+
+ gb_tty->credits -= send_size;
+ spin_unlock_irqrestore(&gb_tty->write_lock, flags);
+
+ request->size = cpu_to_le16(send_size);
+ ret = gb_operation_sync(gb_tty->connection,
+ GB_UART_TYPE_SEND_DATA,
+ request, sizeof(*request) + send_size,
+ NULL, 0);
+ if (ret) {
+ dev_err(&gb_tty->gbphy_dev->dev,
+ "send data error: %d\n", ret);
+ spin_lock_irqsave(&gb_tty->write_lock, flags);
+ gb_tty->credits += send_size;
+ spin_unlock_irqrestore(&gb_tty->write_lock, flags);
+ if (!gb_tty->close_pending)
+ schedule_work(work);
+ return;
+ }
+
+ spin_lock_irqsave(&gb_tty->write_lock, flags);
+ ret = kfifo_out(&gb_tty->write_fifo, &request->data[0],
+ send_size);
+ spin_unlock_irqrestore(&gb_tty->write_lock, flags);
+
+ tty_port_tty_wakeup(&gb_tty->port);
+ }
+}
+
+static int send_line_coding(struct gb_tty *tty)
+{
+ struct gb_uart_set_line_coding_request request;
+
+ memcpy(&request, &tty->line_coding,
+ sizeof(tty->line_coding));
+ return gb_operation_sync(tty->connection, GB_UART_TYPE_SET_LINE_CODING,
+ &request, sizeof(request), NULL, 0);
+}
+
+static int send_control(struct gb_tty *gb_tty, u8 control)
+{
+ struct gb_uart_set_control_line_state_request request;
+
+ request.control = control;
+ return gb_operation_sync(gb_tty->connection,
+ GB_UART_TYPE_SET_CONTROL_LINE_STATE,
+ &request, sizeof(request), NULL, 0);
+}
+
+static int send_break(struct gb_tty *gb_tty, u8 state)
+{
+ struct gb_uart_set_break_request request;
+
+ if ((state != 0) && (state != 1)) {
+ dev_err(&gb_tty->gbphy_dev->dev,
+ "invalid break state of %d\n", state);
+ return -EINVAL;
+ }
+
+ request.state = state;
+ return gb_operation_sync(gb_tty->connection, GB_UART_TYPE_SEND_BREAK,
+ &request, sizeof(request), NULL, 0);
+}
+
+static int gb_uart_wait_for_all_credits(struct gb_tty *gb_tty)
+{
+ int ret;
+
+ if (gb_tty->credits == GB_UART_FIRMWARE_CREDITS)
+ return 0;
+
+ ret = wait_for_completion_timeout(&gb_tty->credits_complete,
+ msecs_to_jiffies(GB_UART_CREDIT_WAIT_TIMEOUT_MSEC));
+ if (!ret) {
+ dev_err(&gb_tty->gbphy_dev->dev,
+ "time out waiting for credits\n");
+ return -ETIMEDOUT;
+ }
+
+ return 0;
+}
+
+static int gb_uart_flush(struct gb_tty *gb_tty, u8 flags)
+{
+ struct gb_uart_serial_flush_request request;
+
+ request.flags = flags;
+ return gb_operation_sync(gb_tty->connection, GB_UART_TYPE_FLUSH_FIFOS,
+ &request, sizeof(request), NULL, 0);
+}
+
+static struct gb_tty *get_gb_by_minor(unsigned minor)
+{
+ struct gb_tty *gb_tty;
+
+ mutex_lock(&table_lock);
+ gb_tty = idr_find(&tty_minors, minor);
+ if (gb_tty) {
+ mutex_lock(&gb_tty->mutex);
+ if (gb_tty->disconnected) {
+ mutex_unlock(&gb_tty->mutex);
+ gb_tty = NULL;
+ } else {
+ tty_port_get(&gb_tty->port);
+ mutex_unlock(&gb_tty->mutex);
+ }
+ }
+ mutex_unlock(&table_lock);
+ return gb_tty;
+}
+
+static int alloc_minor(struct gb_tty *gb_tty)
+{
+ int minor;
+
+ mutex_lock(&table_lock);
+ minor = idr_alloc(&tty_minors, gb_tty, 0, GB_NUM_MINORS, GFP_KERNEL);
+ mutex_unlock(&table_lock);
+ if (minor >= 0)
+ gb_tty->minor = minor;
+ return minor;
+}
+
+static void release_minor(struct gb_tty *gb_tty)
+{
+ int minor = gb_tty->minor;
+
+ gb_tty->minor = 0; /* Maybe should use an invalid value instead */
+ mutex_lock(&table_lock);
+ idr_remove(&tty_minors, minor);
+ mutex_unlock(&table_lock);
+}
+
+static int gb_tty_install(struct tty_driver *driver, struct tty_struct *tty)
+{
+ struct gb_tty *gb_tty;
+ int retval;
+
+ gb_tty = get_gb_by_minor(tty->index);
+ if (!gb_tty)
+ return -ENODEV;
+
+ retval = tty_standard_install(driver, tty);
+ if (retval)
+ goto error;
+
+ tty->driver_data = gb_tty;
+ return 0;
+error:
+ tty_port_put(&gb_tty->port);
+ return retval;
+}
+
+static int gb_tty_open(struct tty_struct *tty, struct file *file)
+{
+ struct gb_tty *gb_tty = tty->driver_data;
+
+ return tty_port_open(&gb_tty->port, tty, file);
+}
+
+static void gb_tty_close(struct tty_struct *tty, struct file *file)
+{
+ struct gb_tty *gb_tty = tty->driver_data;
+
+ tty_port_close(&gb_tty->port, tty, file);
+}
+
+static void gb_tty_cleanup(struct tty_struct *tty)
+{
+ struct gb_tty *gb_tty = tty->driver_data;
+
+ tty_port_put(&gb_tty->port);
+}
+
+static void gb_tty_hangup(struct tty_struct *tty)
+{
+ struct gb_tty *gb_tty = tty->driver_data;
+
+ tty_port_hangup(&gb_tty->port);
+}
+
+static int gb_tty_write(struct tty_struct *tty, const unsigned char *buf,
+ int count)
+{
+ struct gb_tty *gb_tty = tty->driver_data;
+
+ count = kfifo_in_spinlocked(&gb_tty->write_fifo, buf, count,
+ &gb_tty->write_lock);
+ if (count && !gb_tty->close_pending)
+ schedule_work(&gb_tty->tx_work);
+
+ return count;
+}
+
+static int gb_tty_write_room(struct tty_struct *tty)
+{
+ struct gb_tty *gb_tty = tty->driver_data;
+ unsigned long flags;
+ int room;
+
+ spin_lock_irqsave(&gb_tty->write_lock, flags);
+ room = kfifo_avail(&gb_tty->write_fifo);
+ spin_unlock_irqrestore(&gb_tty->write_lock, flags);
+
+ room -= GB_UART_WRITE_ROOM_MARGIN;
+ if (room < 0)
+ return 0;
+
+ return room;
+}
+
+static int gb_tty_chars_in_buffer(struct tty_struct *tty)
+{
+ struct gb_tty *gb_tty = tty->driver_data;
+ unsigned long flags;
+ int chars;
+
+ spin_lock_irqsave(&gb_tty->write_lock, flags);
+ chars = kfifo_len(&gb_tty->write_fifo);
+ if (gb_tty->credits < GB_UART_FIRMWARE_CREDITS)
+ chars += GB_UART_FIRMWARE_CREDITS - gb_tty->credits;
+ spin_unlock_irqrestore(&gb_tty->write_lock, flags);
+
+ return chars;
+}
+
+static int gb_tty_break_ctl(struct tty_struct *tty, int state)
+{
+ struct gb_tty *gb_tty = tty->driver_data;
+
+ return send_break(gb_tty, state ? 1 : 0);
+}
+
+static void gb_tty_set_termios(struct tty_struct *tty,
+ struct ktermios *termios_old)
+{
+ struct gb_tty *gb_tty = tty->driver_data;
+ struct ktermios *termios = &tty->termios;
+ struct gb_tty_line_coding newline;
+ u8 newctrl = gb_tty->ctrlout;
+
+ newline.rate = cpu_to_le32(tty_get_baud_rate(tty));
+ newline.format = termios->c_cflag & CSTOPB ?
+ GB_SERIAL_2_STOP_BITS : GB_SERIAL_1_STOP_BITS;
+ newline.parity = termios->c_cflag & PARENB ?
+ (termios->c_cflag & PARODD ? 1 : 2) +
+ (termios->c_cflag & CMSPAR ? 2 : 0) : 0;
+
+ switch (termios->c_cflag & CSIZE) {
+ case CS5:
+ newline.data_bits = 5;
+ break;
+ case CS6:
+ newline.data_bits = 6;
+ break;
+ case CS7:
+ newline.data_bits = 7;
+ break;
+ case CS8:
+ default:
+ newline.data_bits = 8;
+ break;
+ }
+
+ /* FIXME: needs to clear unsupported bits in the termios */
+ gb_tty->clocal = ((termios->c_cflag & CLOCAL) != 0);
+
+ if (C_BAUD(tty) == B0) {
+ newline.rate = gb_tty->line_coding.rate;
+ newctrl &= ~(GB_UART_CTRL_DTR | GB_UART_CTRL_RTS);
+ } else if (termios_old && (termios_old->c_cflag & CBAUD) == B0) {
+ newctrl |= (GB_UART_CTRL_DTR | GB_UART_CTRL_RTS);
+ }
+
+ if (newctrl != gb_tty->ctrlout) {
+ gb_tty->ctrlout = newctrl;
+ send_control(gb_tty, newctrl);
+ }
+
+ if (C_CRTSCTS(tty) && C_BAUD(tty) != B0)
+ newline.flow_control |= GB_SERIAL_AUTO_RTSCTS_EN;
+ else
+ newline.flow_control &= ~GB_SERIAL_AUTO_RTSCTS_EN;
+
+ if (memcmp(&gb_tty->line_coding, &newline, sizeof(newline))) {
+ memcpy(&gb_tty->line_coding, &newline, sizeof(newline));
+ send_line_coding(gb_tty);
+ }
+}
+
+static int gb_tty_tiocmget(struct tty_struct *tty)
+{
+ struct gb_tty *gb_tty = tty->driver_data;
+
+ return (gb_tty->ctrlout & GB_UART_CTRL_DTR ? TIOCM_DTR : 0) |
+ (gb_tty->ctrlout & GB_UART_CTRL_RTS ? TIOCM_RTS : 0) |
+ (gb_tty->ctrlin & GB_UART_CTRL_DSR ? TIOCM_DSR : 0) |
+ (gb_tty->ctrlin & GB_UART_CTRL_RI ? TIOCM_RI : 0) |
+ (gb_tty->ctrlin & GB_UART_CTRL_DCD ? TIOCM_CD : 0) |
+ TIOCM_CTS;
+}
+
+static int gb_tty_tiocmset(struct tty_struct *tty, unsigned int set,
+ unsigned int clear)
+{
+ struct gb_tty *gb_tty = tty->driver_data;
+ u8 newctrl = gb_tty->ctrlout;
+
+ set = (set & TIOCM_DTR ? GB_UART_CTRL_DTR : 0) |
+ (set & TIOCM_RTS ? GB_UART_CTRL_RTS : 0);
+ clear = (clear & TIOCM_DTR ? GB_UART_CTRL_DTR : 0) |
+ (clear & TIOCM_RTS ? GB_UART_CTRL_RTS : 0);
+
+ newctrl = (newctrl & ~clear) | set;
+ if (gb_tty->ctrlout == newctrl)
+ return 0;
+
+ gb_tty->ctrlout = newctrl;
+ return send_control(gb_tty, newctrl);
+}
+
+static void gb_tty_throttle(struct tty_struct *tty)
+{
+ struct gb_tty *gb_tty = tty->driver_data;
+ unsigned char stop_char;
+ int retval;
+
+ if (I_IXOFF(tty)) {
+ stop_char = STOP_CHAR(tty);
+ retval = gb_tty_write(tty, &stop_char, 1);
+ if (retval <= 0)
+ return;
+ }
+
+ if (tty->termios.c_cflag & CRTSCTS) {
+ gb_tty->ctrlout &= ~GB_UART_CTRL_RTS;
+ retval = send_control(gb_tty, gb_tty->ctrlout);
+ }
+}
+
+static void gb_tty_unthrottle(struct tty_struct *tty)
+{
+ struct gb_tty *gb_tty = tty->driver_data;
+ unsigned char start_char;
+ int retval;
+
+ if (I_IXOFF(tty)) {
+ start_char = START_CHAR(tty);
+ retval = gb_tty_write(tty, &start_char, 1);
+ if (retval <= 0)
+ return;
+ }
+
+ if (tty->termios.c_cflag & CRTSCTS) {
+ gb_tty->ctrlout |= GB_UART_CTRL_RTS;
+ retval = send_control(gb_tty, gb_tty->ctrlout);
+ }
+}
+
+static int get_serial_info(struct gb_tty *gb_tty,
+ struct serial_struct __user *info)
+{
+ struct serial_struct tmp;
+
+ if (!info)
+ return -EINVAL;
+
+ memset(&tmp, 0, sizeof(tmp));
+ tmp.flags = ASYNC_LOW_LATENCY | ASYNC_SKIP_TEST;
+ tmp.type = PORT_16550A;
+ tmp.line = gb_tty->minor;
+ tmp.xmit_fifo_size = 16;
+ tmp.baud_base = 9600;
+ tmp.close_delay = gb_tty->port.close_delay / 10;
+ tmp.closing_wait = gb_tty->port.closing_wait == ASYNC_CLOSING_WAIT_NONE ?
+ ASYNC_CLOSING_WAIT_NONE : gb_tty->port.closing_wait / 10;
+
+ if (copy_to_user(info, &tmp, sizeof(tmp)))
+ return -EFAULT;
+ return 0;
+}
+
+static int set_serial_info(struct gb_tty *gb_tty,
+ struct serial_struct __user *newinfo)
+{
+ struct serial_struct new_serial;
+ unsigned int closing_wait;
+ unsigned int close_delay;
+ int retval = 0;
+
+ if (copy_from_user(&new_serial, newinfo, sizeof(new_serial)))
+ return -EFAULT;
+
+ close_delay = new_serial.close_delay * 10;
+ closing_wait = new_serial.closing_wait == ASYNC_CLOSING_WAIT_NONE ?
+ ASYNC_CLOSING_WAIT_NONE : new_serial.closing_wait * 10;
+
+ mutex_lock(&gb_tty->port.mutex);
+ if (!capable(CAP_SYS_ADMIN)) {
+ if ((close_delay != gb_tty->port.close_delay) ||
+ (closing_wait != gb_tty->port.closing_wait))
+ retval = -EPERM;
+ else
+ retval = -EOPNOTSUPP;
+ } else {
+ gb_tty->port.close_delay = close_delay;
+ gb_tty->port.closing_wait = closing_wait;
+ }
+ mutex_unlock(&gb_tty->port.mutex);
+ return retval;
+}
+
+static int wait_serial_change(struct gb_tty *gb_tty, unsigned long arg)
+{
+ int retval = 0;
+ DECLARE_WAITQUEUE(wait, current);
+ struct async_icount old;
+ struct async_icount new;
+
+ if (!(arg & (TIOCM_DSR | TIOCM_RI | TIOCM_CD)))
+ return -EINVAL;
+
+ do {
+ spin_lock_irq(&gb_tty->read_lock);
+ old = gb_tty->oldcount;
+ new = gb_tty->iocount;
+ gb_tty->oldcount = new;
+ spin_unlock_irq(&gb_tty->read_lock);
+
+ if ((arg & TIOCM_DSR) && (old.dsr != new.dsr))
+ break;
+ if ((arg & TIOCM_CD) && (old.dcd != new.dcd))
+ break;
+ if ((arg & TIOCM_RI) && (old.rng != new.rng))
+ break;
+
+ add_wait_queue(&gb_tty->wioctl, &wait);
+ set_current_state(TASK_INTERRUPTIBLE);
+ schedule();
+ remove_wait_queue(&gb_tty->wioctl, &wait);
+ if (gb_tty->disconnected) {
+ if (arg & TIOCM_CD)
+ break;
+ retval = -ENODEV;
+ } else if (signal_pending(current)) {
+ retval = -ERESTARTSYS;
+ }
+ } while (!retval);
+
+ return retval;
+}
+
+static int get_serial_usage(struct gb_tty *gb_tty,
+ struct serial_icounter_struct __user *count)
+{
+ struct serial_icounter_struct icount;
+ int retval = 0;
+
+ memset(&icount, 0, sizeof(icount));
+ icount.dsr = gb_tty->iocount.dsr;
+ icount.rng = gb_tty->iocount.rng;
+ icount.dcd = gb_tty->iocount.dcd;
+ icount.frame = gb_tty->iocount.frame;
+ icount.overrun = gb_tty->iocount.overrun;
+ icount.parity = gb_tty->iocount.parity;
+ icount.brk = gb_tty->iocount.brk;
+
+ if (copy_to_user(count, &icount, sizeof(icount)) > 0)
+ retval = -EFAULT;
+
+ return retval;
+}
+
+static int gb_tty_ioctl(struct tty_struct *tty, unsigned int cmd,
+ unsigned long arg)
+{
+ struct gb_tty *gb_tty = tty->driver_data;
+
+ switch (cmd) {
+ case TIOCGSERIAL:
+ return get_serial_info(gb_tty,
+ (struct serial_struct __user *)arg);
+ case TIOCSSERIAL:
+ return set_serial_info(gb_tty,
+ (struct serial_struct __user *)arg);
+ case TIOCMIWAIT:
+ return wait_serial_change(gb_tty, arg);
+ case TIOCGICOUNT:
+ return get_serial_usage(gb_tty,
+ (struct serial_icounter_struct __user *)arg);
+ }
+
+ return -ENOIOCTLCMD;
+}
+
+static void gb_tty_dtr_rts(struct tty_port *port, int on)
+{
+ struct gb_tty *gb_tty;
+ u8 newctrl;
+
+ gb_tty = container_of(port, struct gb_tty, port);
+ newctrl = gb_tty->ctrlout;
+
+ if (on)
+ newctrl |= (GB_UART_CTRL_DTR | GB_UART_CTRL_RTS);
+ else
+ newctrl &= ~(GB_UART_CTRL_DTR | GB_UART_CTRL_RTS);
+
+ gb_tty->ctrlout = newctrl;
+ send_control(gb_tty, newctrl);
+}
+
+static int gb_tty_port_activate(struct tty_port *port,
+ struct tty_struct *tty)
+{
+ struct gb_tty *gb_tty;
+
+ gb_tty = container_of(port, struct gb_tty, port);
+
+ return gbphy_runtime_get_sync(gb_tty->gbphy_dev);
+}
+
+static void gb_tty_port_shutdown(struct tty_port *port)
+{
+ struct gb_tty *gb_tty;
+ unsigned long flags;
+ int ret;
+
+ gb_tty = container_of(port, struct gb_tty, port);
+
+ gb_tty->close_pending = true;
+
+ cancel_work_sync(&gb_tty->tx_work);
+
+ spin_lock_irqsave(&gb_tty->write_lock, flags);
+ kfifo_reset_out(&gb_tty->write_fifo);
+ spin_unlock_irqrestore(&gb_tty->write_lock, flags);
+
+ if (gb_tty->credits == GB_UART_FIRMWARE_CREDITS)
+ goto out;
+
+ ret = gb_uart_flush(gb_tty, GB_SERIAL_FLAG_FLUSH_TRANSMITTER);
+ if (ret) {
+ dev_err(&gb_tty->gbphy_dev->dev,
+ "error flushing transmitter: %d\n", ret);
+ }
+
+ gb_uart_wait_for_all_credits(gb_tty);
+
+out:
+ gb_tty->close_pending = false;
+
+ gbphy_runtime_put_autosuspend(gb_tty->gbphy_dev);
+}
+
+static const struct tty_operations gb_ops = {
+ .install = gb_tty_install,
+ .open = gb_tty_open,
+ .close = gb_tty_close,
+ .cleanup = gb_tty_cleanup,
+ .hangup = gb_tty_hangup,
+ .write = gb_tty_write,
+ .write_room = gb_tty_write_room,
+ .ioctl = gb_tty_ioctl,
+ .throttle = gb_tty_throttle,
+ .unthrottle = gb_tty_unthrottle,
+ .chars_in_buffer = gb_tty_chars_in_buffer,
+ .break_ctl = gb_tty_break_ctl,
+ .set_termios = gb_tty_set_termios,
+ .tiocmget = gb_tty_tiocmget,
+ .tiocmset = gb_tty_tiocmset,
+};
+
+static struct tty_port_operations gb_port_ops = {
+ .dtr_rts = gb_tty_dtr_rts,
+ .activate = gb_tty_port_activate,
+ .shutdown = gb_tty_port_shutdown,
+};
+
+static int gb_uart_probe(struct gbphy_device *gbphy_dev,
+ const struct gbphy_device_id *id)
+{
+ struct gb_connection *connection;
+ size_t max_payload;
+ struct gb_tty *gb_tty;
+ struct device *tty_dev;
+ int retval;
+ int minor;
+
+ gb_tty = kzalloc(sizeof(*gb_tty), GFP_KERNEL);
+ if (!gb_tty)
+ return -ENOMEM;
+
+ connection = gb_connection_create(gbphy_dev->bundle,
+ le16_to_cpu(gbphy_dev->cport_desc->id),
+ gb_uart_request_handler);
+ if (IS_ERR(connection)) {
+ retval = PTR_ERR(connection);
+ goto exit_tty_free;
+ }
+
+ max_payload = gb_operation_get_payload_size_max(connection);
+ if (max_payload < sizeof(struct gb_uart_send_data_request)) {
+ retval = -EINVAL;
+ goto exit_connection_destroy;
+ }
+
+ gb_tty->buffer_payload_max = max_payload -
+ sizeof(struct gb_uart_send_data_request);
+
+ gb_tty->buffer = kzalloc(gb_tty->buffer_payload_max, GFP_KERNEL);
+ if (!gb_tty->buffer) {
+ retval = -ENOMEM;
+ goto exit_connection_destroy;
+ }
+
+ INIT_WORK(&gb_tty->tx_work, gb_uart_tx_write_work);
+
+ retval = kfifo_alloc(&gb_tty->write_fifo, GB_UART_WRITE_FIFO_SIZE,
+ GFP_KERNEL);
+ if (retval)
+ goto exit_buf_free;
+
+ gb_tty->credits = GB_UART_FIRMWARE_CREDITS;
+ init_completion(&gb_tty->credits_complete);
+
+ minor = alloc_minor(gb_tty);
+ if (minor < 0) {
+ if (minor == -ENOSPC) {
+ dev_err(&connection->bundle->dev,
+ "no more free minor numbers\n");
+ retval = -ENODEV;
+ } else {
+ retval = minor;
+ }
+ goto exit_kfifo_free;
+ }
+
+ gb_tty->minor = minor;
+ spin_lock_init(&gb_tty->write_lock);
+ spin_lock_init(&gb_tty->read_lock);
+ init_waitqueue_head(&gb_tty->wioctl);
+ mutex_init(&gb_tty->mutex);
+
+ tty_port_init(&gb_tty->port);
+ gb_tty->port.ops = &gb_port_ops;
+
+ gb_tty->connection = connection;
+ gb_tty->gbphy_dev = gbphy_dev;
+ gb_connection_set_data(connection, gb_tty);
+ gb_gbphy_set_data(gbphy_dev, gb_tty);
+
+ retval = gb_connection_enable_tx(connection);
+ if (retval)
+ goto exit_release_minor;
+
+ send_control(gb_tty, gb_tty->ctrlout);
+
+ /* initialize the uart to be 9600n81 */
+ gb_tty->line_coding.rate = cpu_to_le32(9600);
+ gb_tty->line_coding.format = GB_SERIAL_1_STOP_BITS;
+ gb_tty->line_coding.parity = GB_SERIAL_NO_PARITY;
+ gb_tty->line_coding.data_bits = 8;
+ send_line_coding(gb_tty);
+
+ retval = gb_connection_enable(connection);
+ if (retval)
+ goto exit_connection_disable;
+
+ tty_dev = tty_port_register_device(&gb_tty->port, gb_tty_driver, minor,
+ &gbphy_dev->dev);
+ if (IS_ERR(tty_dev)) {
+ retval = PTR_ERR(tty_dev);
+ goto exit_connection_disable;
+ }
+
+ gbphy_runtime_put_autosuspend(gbphy_dev);
+ return 0;
+
+exit_connection_disable:
+ gb_connection_disable(connection);
+exit_release_minor:
+ release_minor(gb_tty);
+exit_kfifo_free:
+ kfifo_free(&gb_tty->write_fifo);
+exit_buf_free:
+ kfree(gb_tty->buffer);
+exit_connection_destroy:
+ gb_connection_destroy(connection);
+exit_tty_free:
+ kfree(gb_tty);
+
+ return retval;
+}
+
+static void gb_uart_remove(struct gbphy_device *gbphy_dev)
+{
+ struct gb_tty *gb_tty = gb_gbphy_get_data(gbphy_dev);
+ struct gb_connection *connection = gb_tty->connection;
+ struct tty_struct *tty;
+ int ret;
+
+ ret = gbphy_runtime_get_sync(gbphy_dev);
+ if (ret)
+ gbphy_runtime_get_noresume(gbphy_dev);
+
+ mutex_lock(&gb_tty->mutex);
+ gb_tty->disconnected = true;
+
+ wake_up_all(&gb_tty->wioctl);
+ mutex_unlock(&gb_tty->mutex);
+
+ tty = tty_port_tty_get(&gb_tty->port);
+ if (tty) {
+ tty_vhangup(tty);
+ tty_kref_put(tty);
+ }
+
+ gb_connection_disable_rx(connection);
+ tty_unregister_device(gb_tty_driver, gb_tty->minor);
+
+ /* FIXME - free transmit / receive buffers */
+
+ gb_connection_disable(connection);
+ tty_port_destroy(&gb_tty->port);
+ gb_connection_destroy(connection);
+ release_minor(gb_tty);
+ kfifo_free(&gb_tty->write_fifo);
+ kfree(gb_tty->buffer);
+ kfree(gb_tty);
+}
+
+static int gb_tty_init(void)
+{
+ int retval = 0;
+
+ gb_tty_driver = tty_alloc_driver(GB_NUM_MINORS, 0);
+ if (IS_ERR(gb_tty_driver)) {
+ pr_err("Can not allocate tty driver\n");
+ retval = -ENOMEM;
+ goto fail_unregister_dev;
+ }
+
+ gb_tty_driver->driver_name = "gb";
+ gb_tty_driver->name = GB_NAME;
+ gb_tty_driver->major = 0;
+ gb_tty_driver->minor_start = 0;
+ gb_tty_driver->type = TTY_DRIVER_TYPE_SERIAL;
+ gb_tty_driver->subtype = SERIAL_TYPE_NORMAL;
+ gb_tty_driver->flags = TTY_DRIVER_REAL_RAW | TTY_DRIVER_DYNAMIC_DEV;
+ gb_tty_driver->init_termios = tty_std_termios;
+ gb_tty_driver->init_termios.c_cflag = B9600 | CS8 | CREAD | HUPCL | CLOCAL;
+ tty_set_operations(gb_tty_driver, &gb_ops);
+
+ retval = tty_register_driver(gb_tty_driver);
+ if (retval) {
+ pr_err("Can not register tty driver: %d\n", retval);
+ goto fail_put_gb_tty;
+ }
+
+ return 0;
+
+fail_put_gb_tty:
+ put_tty_driver(gb_tty_driver);
+fail_unregister_dev:
+ return retval;
+}
+
+static void gb_tty_exit(void)
+{
+ tty_unregister_driver(gb_tty_driver);
+ put_tty_driver(gb_tty_driver);
+ idr_destroy(&tty_minors);
+}
+
+static const struct gbphy_device_id gb_uart_id_table[] = {
+ { GBPHY_PROTOCOL(GREYBUS_PROTOCOL_UART) },
+ { },
+};
+MODULE_DEVICE_TABLE(gbphy, gb_uart_id_table);
+
+static struct gbphy_driver uart_driver = {
+ .name = "uart",
+ .probe = gb_uart_probe,
+ .remove = gb_uart_remove,
+ .id_table = gb_uart_id_table,
+};
+
+static int gb_uart_driver_init(void)
+{
+ int ret;
+
+ ret = gb_tty_init();
+ if (ret)
+ return ret;
+
+ ret = gb_gbphy_register(&uart_driver);
+ if (ret) {
+ gb_tty_exit();
+ return ret;
+ }
+
+ return 0;
+}
+module_init(gb_uart_driver_init);
+
+static void gb_uart_driver_exit(void)
+{
+ gb_gbphy_deregister(&uart_driver);
+ gb_tty_exit();
+}
+
+module_exit(gb_uart_driver_exit);
+MODULE_LICENSE("GPL v2");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 29/32] greybus: bridged phy usb driver
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (26 preceding siblings ...)
2016-09-16 14:21 ` [patch 28/32] greybus: bridged phy uart driver Greg KH
@ 2016-09-16 14:21 ` Greg KH
2016-09-16 14:22 ` [patch 30/32] greybus: tools Greg KH
` (2 subsequent siblings)
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:21 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This driver implements the Greybus bridged phy USB class protocol.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/usb.c | 247 ++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 247 insertions(+)
--- /dev/null
+++ b/drivers/greybus/usb.c
@@ -0,0 +1,247 @@
+/*
+ * USB host driver for the Greybus "generic" USB module.
+ *
+ * Copyright 2014 Google Inc.
+ * Copyright 2014 Linaro Ltd.
+ *
+ * Released under the GPLv2 only.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/usb.h>
+#include <linux/usb/hcd.h>
+
+#include "greybus.h"
+#include "gbphy.h"
+
+/* Greybus USB request types */
+#define GB_USB_TYPE_HCD_START 0x02
+#define GB_USB_TYPE_HCD_STOP 0x03
+#define GB_USB_TYPE_HUB_CONTROL 0x04
+
+struct gb_usb_hub_control_request {
+ __le16 typeReq;
+ __le16 wValue;
+ __le16 wIndex;
+ __le16 wLength;
+};
+
+struct gb_usb_hub_control_response {
+ u8 buf[0];
+};
+
+struct gb_usb_device {
+ struct gb_connection *connection;
+ struct gbphy_device *gbphy_dev;
+};
+
+static inline struct gb_usb_device *to_gb_usb_device(struct usb_hcd *hcd)
+{
+ return (struct gb_usb_device *)hcd->hcd_priv;
+}
+
+static inline struct usb_hcd *gb_usb_device_to_hcd(struct gb_usb_device *dev)
+{
+ return container_of((void *)dev, struct usb_hcd, hcd_priv);
+}
+
+static void hcd_stop(struct usb_hcd *hcd)
+{
+ struct gb_usb_device *dev = to_gb_usb_device(hcd);
+ int ret;
+
+ ret = gb_operation_sync(dev->connection, GB_USB_TYPE_HCD_STOP,
+ NULL, 0, NULL, 0);
+ if (ret)
+ dev_err(&dev->gbphy_dev->dev, "HCD stop failed '%d'\n", ret);
+}
+
+static int hcd_start(struct usb_hcd *hcd)
+{
+ struct usb_bus *bus = hcd_to_bus(hcd);
+ struct gb_usb_device *dev = to_gb_usb_device(hcd);
+ int ret;
+
+ ret = gb_operation_sync(dev->connection, GB_USB_TYPE_HCD_START,
+ NULL, 0, NULL, 0);
+ if (ret) {
+ dev_err(&dev->gbphy_dev->dev, "HCD start failed '%d'\n", ret);
+ return ret;
+ }
+
+ hcd->state = HC_STATE_RUNNING;
+ if (bus->root_hub)
+ usb_hcd_resume_root_hub(hcd);
+ return 0;
+}
+
+static int urb_enqueue(struct usb_hcd *hcd, struct urb *urb, gfp_t mem_flags)
+{
+ return -ENXIO;
+}
+
+static int urb_dequeue(struct usb_hcd *hcd, struct urb *urb, int status)
+{
+ return -ENXIO;
+}
+
+static int get_frame_number(struct usb_hcd *hcd)
+{
+ return 0;
+}
+
+static int hub_status_data(struct usb_hcd *hcd, char *buf)
+{
+ return 0;
+}
+
+static int hub_control(struct usb_hcd *hcd, u16 typeReq, u16 wValue, u16 wIndex,
+ char *buf, u16 wLength)
+{
+ struct gb_usb_device *dev = to_gb_usb_device(hcd);
+ struct gb_operation *operation;
+ struct gb_usb_hub_control_request *request;
+ struct gb_usb_hub_control_response *response;
+ size_t response_size;
+ int ret;
+
+ /* FIXME: handle unspecified lengths */
+ response_size = sizeof(*response) + wLength;
+
+ operation = gb_operation_create(dev->connection,
+ GB_USB_TYPE_HUB_CONTROL,
+ sizeof(*request),
+ response_size,
+ GFP_KERNEL);
+ if (!operation)
+ return -ENOMEM;
+
+ request = operation->request->payload;
+ request->typeReq = cpu_to_le16(typeReq);
+ request->wValue = cpu_to_le16(wValue);
+ request->wIndex = cpu_to_le16(wIndex);
+ request->wLength = cpu_to_le16(wLength);
+
+ ret = gb_operation_request_send_sync(operation);
+ if (ret)
+ goto out;
+
+ if (wLength) {
+ /* Greybus core has verified response size */
+ response = operation->response->payload;
+ memcpy(buf, response->buf, wLength);
+ }
+out:
+ gb_operation_put(operation);
+
+ return ret;
+}
+
+static struct hc_driver usb_gb_hc_driver = {
+ .description = "greybus-hcd",
+ .product_desc = "Greybus USB Host Controller",
+ .hcd_priv_size = sizeof(struct gb_usb_device),
+
+ .flags = HCD_USB2,
+
+ .start = hcd_start,
+ .stop = hcd_stop,
+
+ .urb_enqueue = urb_enqueue,
+ .urb_dequeue = urb_dequeue,
+
+ .get_frame_number = get_frame_number,
+ .hub_status_data = hub_status_data,
+ .hub_control = hub_control,
+};
+
+static int gb_usb_probe(struct gbphy_device *gbphy_dev,
+ const struct gbphy_device_id *id)
+{
+ struct gb_connection *connection;
+ struct device *dev = &gbphy_dev->dev;
+ struct gb_usb_device *gb_usb_dev;
+ struct usb_hcd *hcd;
+ int retval;
+
+ hcd = usb_create_hcd(&usb_gb_hc_driver, dev, dev_name(dev));
+ if (!hcd)
+ return -ENOMEM;
+
+ connection = gb_connection_create(gbphy_dev->bundle,
+ le16_to_cpu(gbphy_dev->cport_desc->id),
+ NULL);
+ if (IS_ERR(connection)) {
+ retval = PTR_ERR(connection);
+ goto exit_usb_put;
+ }
+
+ gb_usb_dev = to_gb_usb_device(hcd);
+ gb_usb_dev->connection = connection;
+ gb_connection_set_data(connection, gb_usb_dev);
+ gb_usb_dev->gbphy_dev = gbphy_dev;
+ gb_gbphy_set_data(gbphy_dev, gb_usb_dev);
+
+ hcd->has_tt = 1;
+
+ retval = gb_connection_enable(connection);
+ if (retval)
+ goto exit_connection_destroy;
+
+ /*
+ * FIXME: The USB bridged-PHY protocol driver depends on changes to
+ * USB core which are not yet upstream.
+ *
+ * Disable for now.
+ */
+ if (1) {
+ dev_warn(dev, "USB protocol disabled\n");
+ retval = -EPROTONOSUPPORT;
+ goto exit_connection_disable;
+ }
+
+ retval = usb_add_hcd(hcd, 0, 0);
+ if (retval)
+ goto exit_connection_disable;
+
+ return 0;
+
+exit_connection_disable:
+ gb_connection_disable(connection);
+exit_connection_destroy:
+ gb_connection_destroy(connection);
+exit_usb_put:
+ usb_put_hcd(hcd);
+
+ return retval;
+}
+
+static void gb_usb_remove(struct gbphy_device *gbphy_dev)
+{
+ struct gb_usb_device *gb_usb_dev = gb_gbphy_get_data(gbphy_dev);
+ struct gb_connection *connection = gb_usb_dev->connection;
+ struct usb_hcd *hcd = gb_usb_device_to_hcd(gb_usb_dev);
+
+ usb_remove_hcd(hcd);
+ gb_connection_disable(connection);
+ gb_connection_destroy(connection);
+ usb_put_hcd(hcd);
+}
+
+static const struct gbphy_device_id gb_usb_id_table[] = {
+ { GBPHY_PROTOCOL(GREYBUS_PROTOCOL_USB) },
+ { },
+};
+MODULE_DEVICE_TABLE(gbphy, gb_usb_id_table);
+
+static struct gbphy_driver usb_driver = {
+ .name = "usb",
+ .probe = gb_usb_probe,
+ .remove = gb_usb_remove,
+ .id_table = gb_usb_id_table,
+};
+
+module_gbphy_driver(usb_driver);
+MODULE_LICENSE("GPL v2");
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 30/32] greybus: tools
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (27 preceding siblings ...)
2016-09-16 14:21 ` [patch 29/32] greybus: bridged phy usb driver Greg KH
@ 2016-09-16 14:22 ` Greg KH
2016-09-16 14:22 ` [patch 31/32] greybus: documentation Greg KH
2016-09-16 14:22 ` [patch 32/32] greybus: add to the build Greg KH
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:22 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
Here are a few Greybus userspace tools that are useful for testing.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/greybus/tools/.gitignore | 1
drivers/greybus/tools/Android.mk | 10
drivers/greybus/tools/Makefile | 31 +
drivers/greybus/tools/README.loopback | 198 ++++++
drivers/greybus/tools/lbtest | 168 +++++
drivers/greybus/tools/loopback_test.c | 1000 ++++++++++++++++++++++++++++++++++
6 files changed, 1408 insertions(+)
--- /dev/null
+++ b/drivers/greybus/tools/.gitignore
@@ -0,0 +1 @@
+loopback_test
--- /dev/null
+++ b/drivers/greybus/tools/Android.mk
@@ -0,0 +1,10 @@
+LOCAL_PATH:= $(call my-dir)
+
+include $(CLEAR_VARS)
+
+LOCAL_SRC_FILES:= loopback_test.c
+LOCAL_MODULE_TAGS := optional
+LOCAL_MODULE := gb_loopback_test
+
+include $(BUILD_EXECUTABLE)
+
--- /dev/null
+++ b/drivers/greybus/tools/Makefile
@@ -0,0 +1,31 @@
+ifeq ($(strip $(V)), 1)
+ Q =
+else
+ Q = @
+endif
+
+CFLAGS += -std=gnu99 -Wall -Wextra -g \
+ -D_GNU_SOURCE \
+ -Wno-unused-parameter \
+ -Wmaybe-uninitialized \
+ -Wredundant-decls \
+ -Wcast-align \
+ -Wsign-compare \
+ -Wno-missing-field-initializers
+
+CC := $(CROSS_COMPILE)gcc
+
+TOOLS = loopback_test
+
+all: $(TOOLS)
+
+%.o: %.c ../greybus_protocols.h
+ @echo ' TARGET_CC $@'
+ $(Q)$(CC) $(CFLAGS) -c $< -o $@
+
+loopback_%: loopback_%.o
+ @echo ' TARGET_LD $@'
+ $(Q)$(CC) $(CFLAGS) $(LDFLAGS) $^ -o $@
+
+clean::
+ rm -f *.o $(TOOLS)
--- /dev/null
+++ b/drivers/greybus/tools/README.loopback
@@ -0,0 +1,198 @@
+
+
+ 1 - LOOPBACK DRIVER
+
+The driver implements the main logic of the loopback test and provides
+sysfs files to configure the test and retrieve the results.
+A user could run a test without the need of the test application given
+that he understands the sysfs interface of the loopback driver.
+
+The loopback kernel driver needs to be loaded and at least one module
+with the loopback feature enabled must be present for the sysfs files to be
+created and for the loopback test application to be able to run.
+
+To load the module:
+# modprobe gb-loopback
+
+
+When the module is probed, New files are available on the sysfs
+directory of the detected loopback device.
+(typically under "/sys/bus/graybus/devices").
+
+Here is a short summary of the sysfs interface files that should be visible:
+
+* Loopback Configuration Files:
+ async - Use asynchronous operations.
+ iteration_max - Number of tests iterations to perform.
+ size - payload size of the transfer.
+ timeout - The number of microseconds to give an individual
+ asynchronous request before timing out.
+ us_wait - Time to wait between 2 messages
+ type - By writing the test type to this file, the test starts.
+ Valid tests are:
+ 0 stop the test
+ 2 - ping
+ 3 - transfer
+ 4 - sink
+
+* Loopback feedback files:
+ error - number of errors that have occurred.
+ iteration_count - Number of iterations performed.
+ requests_completed - Number of requests successfully completed.
+ requests_timedout - Number of requests that have timed out.
+ timeout_max - Max allowed timeout
+ timeout_min - Min allowed timeout.
+
+* Loopback result files:
+ apbridge_unipro_latency_avg
+ apbridge_unipro_latency_max
+ apbridge_unipro_latency_min
+ gpbridge_firmware_latency_avg
+ gpbridge_firmware_latency_max
+ gpbridge_firmware_latency_min
+ requests_per_second_avg
+ requests_per_second_max
+ requests_per_second_min
+ latency_avg
+ latency_max
+ latency_min
+ throughput_avg
+ throughput_max
+ throughput_min
+
+
+
+ 2 - LOOPBACK TEST APPLICATION
+
+The loopback test application manages and formats the results provided by
+the loopback kernel module. The purpose of this application
+is to:
+ - Start and manage multiple loopback device tests concurrently.
+ - Calculate the aggregate results for multiple devices.
+ - Gather and format test results (csv or human readable).
+
+The best way to get up to date usage information for the application is
+usually to pass the "-h" parameter.
+Here is the summary of the available options:
+
+ Mandatory arguments
+ -t must be one of the test names - sink, transfer or ping
+ -i iteration count - the number of iterations to run the test over
+ Optional arguments
+ -S sysfs location - location for greybus 'endo' entires default /sys/bus/greybus/devices/
+ -D debugfs location - location for loopback debugfs entries default /sys/kernel/debug/gb_loopback/
+ -s size of data packet to send during test - defaults to zero
+ -m mask - a bit mask of connections to include example: -m 8 = 4th connection -m 9 = 1st and 4th connection etc
+ default is zero which means broadcast to all connections
+ -v verbose output
+ -d debug output
+ -r raw data output - when specified the full list of latency values are included in the output CSV
+ -p porcelain - when specified printout is in a user-friendly non-CSV format. This option suppresses writing to CSV file
+ -a aggregate - show aggregation of all enabled devies
+ -l list found loopback devices and exit.
+ -x Async - Enable async transfers.
+ -o Timeout - Timeout in microseconds for async operations.
+
+
+
+ 3 - REAL WORLD EXAMPLE USAGES
+
+ 3.1 - Using the driver sysfs files to run a test on a single device:
+
+* Run a 1000 transfers of a 100 byte packet. Each transfer is started only
+after the previous one finished successfully:
+ echo 0 > /sys/bus/greybus/devices/1-2.17/type
+ echo 0 > /sys/bus/greybus/devices/1-2.17/async
+ echo 2000 > /sys/bus/greybus/devices/1-2.17/us_wait
+ echo 100 > /sys/bus/greybus/devices/1-2.17/size
+ echo 1000 > /sys/bus/greybus/devices/1-2.17/iteration_max
+ echo 0 > /sys/bus/greybus/devices/1-2.17/mask
+ echo 200000 > /sys/bus/greybus/devices/1-2.17/timeout
+ echo 3 > /sys/bus/greybus/devices/1-2.17/type
+
+* Run a 1000 transfers of a 100 byte packet. Transfers are started without
+waiting for the previous one to finish:
+ echo 0 > /sys/bus/greybus/devices/1-2.17/type
+ echo 3 > /sys/bus/greybus/devices/1-2.17/async
+ echo 0 > /sys/bus/greybus/devices/1-2.17/us_wait
+ echo 100 > /sys/bus/greybus/devices/1-2.17/size
+ echo 1000 > /sys/bus/greybus/devices/1-2.17/iteration_max
+ echo 0 > /sys/bus/greybus/devices/1-2.17/mask
+ echo 200000 > /sys/bus/greybus/devices/1-2.17/timeout
+ echo 3 > /sys/bus/greybus/devices/1-2.17/type
+
+* Read the results from sysfs:
+ cat /sys/bus/greybus/devices/1-2.17/requests_per_second_min
+ cat /sys/bus/greybus/devices/1-2.17/requests_per_second_max
+ cat /sys/bus/greybus/devices/1-2.17/requests_per_second_avg
+
+ cat /sys/bus/greybus/devices/1-2.17/latency_min
+ cat /sys/bus/greybus/devices/1-2.17/latency_max
+ cat /sys/bus/greybus/devices/1-2.17/latency_avg
+
+ cat /sys/bus/greybus/devices/1-2.17/apbridge_unipro_latency_min
+ cat /sys/bus/greybus/devices/1-2.17/apbridge_unipro_latency_max
+ cat /sys/bus/greybus/devices/1-2.17/apbridge_unipro_latency_avg
+
+ cat /sys/bus/greybus/devices/1-2.17/gpbridge_firmware_latency_min
+ cat /sys/bus/greybus/devices/1-2.17/gpbridge_firmware_latency_max
+ cat /sys/bus/greybus/devices/1-2.17/gpbridge_firmware_latency_avg
+
+ cat /sys/bus/greybus/devices/1-2.17/error
+ cat /sys/bus/greybus/devices/1-2.17/requests_completed
+ cat /sys/bus/greybus/devices/1-2.17/requests_timedout
+
+
+3.2 - using the test application:
+
+* Run a transfer test 10 iterations of size 100 bytes on all available devices
+ #/loopback_test -t transfer -i 10 -s 100
+ 1970-1-1 0:10:7,transfer,1-4.17,100,10,0,443,509,471.700012,66,1963,2256,2124.600098,293,102776,118088,109318.898438,15312,1620,1998,1894.099976,378,56,57,56.799999,1
+ 1970-1-1 0:10:7,transfer,1-5.17,100,10,0,399,542,463.399994,143,1845,2505,2175.800049,660,92568,125744,107393.296875,33176,1469,2305,1806.500000,836,56,57,56.799999,1
+
+
+* Show the aggregate results of both devices. ("-a")
+ #/loopback_test -t transfer -i 10 -s 100 -a
+ 1970-1-1 0:10:35,transfer,1-4.17,100,10,0,448,580,494.100006,132,1722,2230,2039.400024,508,103936,134560,114515.703125,30624,1513,1980,1806.900024,467,56,57,57.299999,1
+ 1970-1-1 0:10:35,transfer,1-5.17,100,10,0,383,558,478.600006,175,1791,2606,2115.199951,815,88856,129456,110919.703125,40600,1457,2246,1773.599976,789,56,57,57.099998,1
+ 1970-1-1 0:10:35,transfer,aggregate,100,10,0,383,580,486.000000,197,1722,2606,2077.000000,884,88856,134560,112717.000000,45704,1457,2246,1789.000000,789,56,57,57.000000,1
+
+* Example usage of the mask option to select which devices will
+ run the test (1st, 2nd, or both devices):
+ # /loopback_test -t transfer -i 10 -s 100 -m 1
+ 1970-1-1 0:11:56,transfer,1-4.17,100,10,0,514,558,544.900024,44,1791,1943,1836.599976,152,119248,129456,126301.296875,10208,1600,1001609,101613.601562,1000009,56,57,56.900002,1
+ # /loopback_test -t transfer -i 10 -s 100 -m 2
+ 1970-1-1 0:12:0,transfer,1-5.17,100,10,0,468,554,539.000000,86,1804,2134,1859.500000,330,108576,128528,124932.500000,19952,1606,1626,1619.300049,20,56,57,57.400002,1
+ # /loopback_test -t transfer -i 10 -s 100 -m 3
+ 1970-1-1 0:12:3,transfer,1-4.17,100,10,0,432,510,469.399994,78,1959,2313,2135.800049,354,100224,118320,108785.296875,18096,1610,2024,1893.500000,414,56,57,57.200001,1
+ 1970-1-1 0:12:3,transfer,1-5.17,100,10,0,404,542,468.799988,138,1843,2472,2152.500000,629,93728,125744,108646.101562,32016,1504,2247,1853.099976,743,56,57,57.099998,1
+
+* Show output in human readable format ("-p")
+ # /loopback_test -t transfer -i 10 -s 100 -m 3 -p
+
+ 1970-1-1 0:12:37
+ test: transfer
+ path: 1-4.17
+ size: 100
+ iterations: 10
+ errors: 0
+ async: Disabled
+ requests per-sec: min=390, max=547, average=469.299988, jitter=157
+ ap-throughput B/s: min=90480 max=126904 average=108762.101562 jitter=36424
+ ap-latency usec: min=1826 max=2560 average=2146.000000 jitter=734
+ apbridge-latency usec: min=1620 max=1982 average=1882.099976 jitter=362
+ gpbridge-latency usec: min=56 max=57 average=57.099998 jitter=1
+
+
+ 1970-1-1 0:12:37
+ test: transfer
+ path: 1-5.17
+ size: 100
+ iterations: 10
+ errors: 0
+ async: Disabled
+ requests per-sec: min=397, max=538, average=461.700012, jitter=141
+ ap-throughput B/s: min=92104 max=124816 average=106998.898438 jitter=32712
+ ap-latency usec: min=1856 max=2514 average=2185.699951 jitter=658
+ apbridge-latency usec: min=1460 max=2296 average=1828.599976 jitter=836
+ gpbridge-latency usec: min=56 max=57 average=57.099998 jitter=1
--- /dev/null
+++ b/drivers/greybus/tools/lbtest
@@ -0,0 +1,168 @@
+#!/usr/bin/env python
+
+# Copyright (c) 2015 Google, Inc.
+# Copyright (c) 2015 Linaro, Ltd.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+# 1. Redistributions of source code must retain the above copyright notice,
+# this list of conditions and the following disclaimer.
+# 2. Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+# 3. Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from this
+# software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+# THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+# OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+# WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+# OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+# ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+from __future__ import print_function
+import csv
+import datetime
+import sys
+import time
+
+dict = {'ping': '2', 'transfer': '3', 'sink': '4'}
+verbose = 1
+
+def abort():
+ sys.exit(1)
+
+def usage():
+ print('Usage: looptest TEST SIZE ITERATIONS PATH\n\n'
+ ' Run TEST for a number of ITERATIONS with operation data SIZE bytes\n'
+ ' TEST may be \'ping\' \'transfer\' or \'sink\'\n'
+ ' SIZE indicates the size of transfer <= greybus max payload bytes\n'
+ ' ITERATIONS indicates the number of times to execute TEST at SIZE bytes\n'
+ ' Note if ITERATIONS is set to zero then this utility will\n'
+ ' initiate an infinite (non terminating) test and exit\n'
+ ' without logging any metrics data\n'
+ ' PATH indicates the sysfs path for the loopback greybus entries e.g.\n'
+ ' /sys/bus/greybus/devices/endo0:1:1:1:1/\n'
+ 'Examples:\n'
+ ' looptest transfer 128 10000\n'
+ ' looptest ping 0 128\n'
+ ' looptest sink 2030 32768\n'
+ .format(sys.argv[0]), file=sys.stderr)
+
+ abort()
+
+def read_sysfs_int(path):
+ try:
+ f = open(path, "r");
+ val = f.read();
+ f.close()
+ return int(val)
+ except IOError as e:
+ print("I/O error({0}): {1}".format(e.errno, e.strerror))
+ print("Invalid path %s" % path)
+
+def write_sysfs_val(path, val):
+ try:
+ f = open(path, "r+")
+ f.write(val)
+ f.close()
+ except IOError as e:
+ print("I/O error({0}): {1}".format(e.errno, e.strerror))
+ print("Invalid path %s" % path)
+
+def log_csv(test_name, size, iteration_max, sys_pfx):
+ # file name will test_name_size_iteration_max.csv
+ # every time the same test with the same parameters is run we will then
+ # append to the same CSV with datestamp - representing each test dataset
+ fname = test_name + '_' + size + '_' + str(iteration_max) + '.csv'
+
+ try:
+ # gather data set
+ date = str(datetime.datetime.now())
+ error = read_sysfs_int(sys_pfx + 'error')
+ request_min = read_sysfs_int(sys_pfx + 'requests_per_second_min')
+ request_max = read_sysfs_int(sys_pfx + 'requests_per_second_max')
+ request_avg = read_sysfs_int(sys_pfx + 'requests_per_second_avg')
+ latency_min = read_sysfs_int(sys_pfx + 'latency_min')
+ latency_max = read_sysfs_int(sys_pfx + 'latency_max')
+ latency_avg = read_sysfs_int(sys_pfx + 'latency_avg')
+ throughput_min = read_sysfs_int(sys_pfx + 'throughput_min')
+ throughput_max = read_sysfs_int(sys_pfx + 'throughput_max')
+ throughput_avg = read_sysfs_int(sys_pfx + 'throughput_avg')
+
+ # derive jitter
+ request_jitter = request_max - request_min
+ latency_jitter = latency_max - latency_min
+ throughput_jitter = throughput_max - throughput_min
+
+ # append data set to file
+ with open(fname, 'a') as csvf:
+ row = csv.writer(csvf, delimiter=",", quotechar="'",
+ quoting=csv.QUOTE_MINIMAL)
+ row.writerow([date, test_name, size, iteration_max, error,
+ request_min, request_max, request_avg, request_jitter,
+ latency_min, latency_max, latency_avg, latency_jitter,
+ throughput_min, throughput_max, throughput_avg, throughput_jitter])
+ except IOError as e:
+ print("I/O error({0}): {1}".format(e.errno, e.strerror))
+
+def loopback_run(test_name, size, iteration_max, sys_pfx):
+ test_id = dict[test_name]
+ try:
+ # Terminate any currently running test
+ write_sysfs_val(sys_pfx + 'type', '0')
+ # Set parameter for no wait between messages
+ write_sysfs_val(sys_pfx + 'ms_wait', '0')
+ # Set operation size
+ write_sysfs_val(sys_pfx + 'size', size)
+ # Set iterations
+ write_sysfs_val(sys_pfx + 'iteration_max', str(iteration_max))
+ # Initiate by setting loopback operation type
+ write_sysfs_val(sys_pfx + 'type', test_id)
+ time.sleep(1)
+
+ if iteration_max == 0:
+ print ("Infinite test initiated CSV won't be logged\n")
+ return
+
+ previous = 0
+ err = 0
+ while True:
+ # get current count bail out if it hasn't changed
+ iteration_count = read_sysfs_int(sys_pfx + 'iteration_count')
+ if previous == iteration_count:
+ err = 1
+ break
+ elif iteration_count == iteration_max:
+ break
+ previous = iteration_count
+ if verbose:
+ print('%02d%% complete %d of %d ' %
+ (100 * iteration_count / iteration_max,
+ iteration_count, iteration_max))
+ time.sleep(1)
+ if err:
+ print ('\nError executing test\n')
+ else:
+ log_csv(test_name, size, iteration_max, sys_pfx)
+ except ValueError as ve:
+ print("Error: %s " % format(e.strerror), file=sys.stderr)
+ abort()
+
+def main():
+ if len(sys.argv) < 5:
+ usage()
+
+ if sys.argv[1] in dict.keys():
+ loopback_run(sys.argv[1], sys.argv[2], int(sys.argv[3]), sys.argv[4])
+ else:
+ usage()
+if __name__ == '__main__':
+ main()
--- /dev/null
+++ b/drivers/greybus/tools/loopback_test.c
@@ -0,0 +1,1000 @@
+/*
+ * Loopback test application
+ *
+ * Copyright 2015 Google Inc.
+ * Copyright 2015 Linaro Ltd.
+ *
+ * Provided under the three clause BSD license found in the LICENSE file.
+ */
+#include <errno.h>
+#include <fcntl.h>
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <poll.h>
+#include <sys/types.h>
+#include <time.h>
+#include <unistd.h>
+#include <dirent.h>
+#include <signal.h>
+
+#define MAX_NUM_DEVICES 10
+#define MAX_SYSFS_PATH 0x200
+#define CSV_MAX_LINE 0x1000
+#define SYSFS_MAX_INT 0x20
+#define MAX_STR_LEN 255
+#define DEFAULT_ASYNC_TIMEOUT 200000
+
+struct dict {
+ char *name;
+ int type;
+};
+
+static struct dict dict[] = {
+ {"ping", 2},
+ {"transfer", 3},
+ {"sink", 4},
+ {NULL,} /* list termination */
+};
+
+struct loopback_results {
+ float latency_avg;
+ uint32_t latency_max;
+ uint32_t latency_min;
+ uint32_t latency_jitter;
+
+ float request_avg;
+ uint32_t request_max;
+ uint32_t request_min;
+ uint32_t request_jitter;
+
+ float throughput_avg;
+ uint32_t throughput_max;
+ uint32_t throughput_min;
+ uint32_t throughput_jitter;
+
+ float apbridge_unipro_latency_avg;
+ uint32_t apbridge_unipro_latency_max;
+ uint32_t apbridge_unipro_latency_min;
+ uint32_t apbridge_unipro_latency_jitter;
+
+ float gbphy_firmware_latency_avg;
+ uint32_t gbphy_firmware_latency_max;
+ uint32_t gbphy_firmware_latency_min;
+ uint32_t gbphy_firmware_latency_jitter;
+
+ uint32_t error;
+};
+
+struct loopback_device {
+ char name[MAX_SYSFS_PATH];
+ char sysfs_entry[MAX_SYSFS_PATH];
+ char debugfs_entry[MAX_SYSFS_PATH];
+ struct loopback_results results;
+};
+
+struct loopback_test {
+ int verbose;
+ int debug;
+ int raw_data_dump;
+ int porcelain;
+ int mask;
+ int size;
+ int iteration_max;
+ int aggregate_output;
+ int test_id;
+ int device_count;
+ int list_devices;
+ int use_async;
+ int async_timeout;
+ int async_outstanding_operations;
+ int us_wait;
+ int file_output;
+ int stop_all;
+ int poll_count;
+ char test_name[MAX_STR_LEN];
+ char sysfs_prefix[MAX_SYSFS_PATH];
+ char debugfs_prefix[MAX_SYSFS_PATH];
+ struct timespec poll_timeout;
+ struct loopback_device devices[MAX_NUM_DEVICES];
+ struct loopback_results aggregate_results;
+ struct pollfd fds[MAX_NUM_DEVICES];
+};
+
+struct loopback_test t;
+
+/* Helper macros to calculate the aggregate results for all devices */
+static inline int device_enabled(struct loopback_test *t, int dev_idx);
+
+#define GET_MAX(field) \
+static int get_##field##_aggregate(struct loopback_test *t) \
+{ \
+ uint32_t max = 0; \
+ int i; \
+ for (i = 0; i < t->device_count; i++) { \
+ if (!device_enabled(t, i)) \
+ continue; \
+ if (t->devices[i].results.field > max) \
+ max = t->devices[i].results.field; \
+ } \
+ return max; \
+} \
+
+#define GET_MIN(field) \
+static int get_##field##_aggregate(struct loopback_test *t) \
+{ \
+ uint32_t min = ~0; \
+ int i; \
+ for (i = 0; i < t->device_count; i++) { \
+ if (!device_enabled(t, i)) \
+ continue; \
+ if (t->devices[i].results.field < min) \
+ min = t->devices[i].results.field; \
+ } \
+ return min; \
+} \
+
+#define GET_AVG(field) \
+static int get_##field##_aggregate(struct loopback_test *t) \
+{ \
+ uint32_t val = 0; \
+ uint32_t count = 0; \
+ int i; \
+ for (i = 0; i < t->device_count; i++) { \
+ if (!device_enabled(t, i)) \
+ continue; \
+ count++; \
+ val += t->devices[i].results.field; \
+ } \
+ if (count) \
+ val /= count; \
+ return val; \
+} \
+
+GET_MAX(throughput_max);
+GET_MAX(request_max);
+GET_MAX(latency_max);
+GET_MAX(apbridge_unipro_latency_max);
+GET_MAX(gbphy_firmware_latency_max);
+GET_MIN(throughput_min);
+GET_MIN(request_min);
+GET_MIN(latency_min);
+GET_MIN(apbridge_unipro_latency_min);
+GET_MIN(gbphy_firmware_latency_min);
+GET_AVG(throughput_avg);
+GET_AVG(request_avg);
+GET_AVG(latency_avg);
+GET_AVG(apbridge_unipro_latency_avg);
+GET_AVG(gbphy_firmware_latency_avg);
+
+void abort()
+{
+ _exit(1);
+}
+
+void usage(void)
+{
+ fprintf(stderr, "Usage: loopback_test TEST [SIZE] ITERATIONS [SYSPATH] [DBGPATH]\n\n"
+ " Run TEST for a number of ITERATIONS with operation data SIZE bytes\n"
+ " TEST may be \'ping\' \'transfer\' or \'sink\'\n"
+ " SIZE indicates the size of transfer <= greybus max payload bytes\n"
+ " ITERATIONS indicates the number of times to execute TEST at SIZE bytes\n"
+ " Note if ITERATIONS is set to zero then this utility will\n"
+ " initiate an infinite (non terminating) test and exit\n"
+ " without logging any metrics data\n"
+ " SYSPATH indicates the sysfs path for the loopback greybus entries e.g.\n"
+ " /sys/bus/greybus/devices\n"
+ " DBGPATH indicates the debugfs path for the loopback greybus entries e.g.\n"
+ " /sys/kernel/debug/gb_loopback/\n"
+ " Mandatory arguments\n"
+ " -t must be one of the test names - sink, transfer or ping\n"
+ " -i iteration count - the number of iterations to run the test over\n"
+ " Optional arguments\n"
+ " -S sysfs location - location for greybus 'endo' entires default /sys/bus/greybus/devices/\n"
+ " -D debugfs location - location for loopback debugfs entries default /sys/kernel/debug/gb_loopback/\n"
+ " -s size of data packet to send during test - defaults to zero\n"
+ " -m mask - a bit mask of connections to include example: -m 8 = 4th connection -m 9 = 1st and 4th connection etc\n"
+ " default is zero which means broadcast to all connections\n"
+ " -v verbose output\n"
+ " -d debug output\n"
+ " -r raw data output - when specified the full list of latency values are included in the output CSV\n"
+ " -p porcelain - when specified printout is in a user-friendly non-CSV format. This option suppresses writing to CSV file\n"
+ " -a aggregate - show aggregation of all enabled devices\n"
+ " -l list found loopback devices and exit\n"
+ " -x Async - Enable async transfers\n"
+ " -o Async Timeout - Timeout in uSec for async operations\n"
+ " -O Poll loop time out in seconds(max time a test is expected to last, default: 30sec)\n"
+ " -c Max number of outstanding operations for async operations\n"
+ " -w Wait in uSec between operations\n"
+ " -z Enable output to a CSV file (incompatible with -p)\n"
+ " -f When starting new loopback test, stop currently running tests on all devices\n"
+ "Examples:\n"
+ " Send 10000 transfers with a packet size of 128 bytes to all active connections\n"
+ " loopback_test -t transfer -s 128 -i 10000 -S /sys/bus/greybus/devices/ -D /sys/kernel/debug/gb_loopback/\n"
+ " loopback_test -t transfer -s 128 -i 10000 -m 0\n"
+ " Send 10000 transfers with a packet size of 128 bytes to connection 1 and 4\n"
+ " loopback_test -t transfer -s 128 -i 10000 -m 9\n"
+ " loopback_test -t ping -s 0 128 -i -S /sys/bus/greybus/devices/ -D /sys/kernel/debug/gb_loopback/\n"
+ " loopback_test -t sink -s 2030 -i 32768 -S /sys/bus/greybus/devices/ -D /sys/kernel/debug/gb_loopback/\n");
+ abort();
+}
+
+static inline int device_enabled(struct loopback_test *t, int dev_idx)
+{
+ if (!t->mask || (t->mask & (1 << dev_idx)))
+ return 1;
+
+ return 0;
+}
+
+static void show_loopback_devices(struct loopback_test *t)
+{
+ int i;
+
+ if (t->device_count == 0) {
+ printf("No loopback devices.\n");
+ return;
+ }
+
+ for (i = 0; i < t->device_count; i++)
+ printf("device[%d] = %s\n", i, t->devices[i].name);
+
+}
+
+int open_sysfs(const char *sys_pfx, const char *node, int flags)
+{
+ int fd;
+ char path[MAX_SYSFS_PATH];
+
+ snprintf(path, sizeof(path), "%s%s", sys_pfx, node);
+ fd = open(path, flags);
+ if (fd < 0) {
+ fprintf(stderr, "unable to open %s\n", path);
+ abort();
+ }
+ return fd;
+}
+
+int read_sysfs_int_fd(int fd, const char *sys_pfx, const char *node)
+{
+ char buf[SYSFS_MAX_INT];
+
+ if (read(fd, buf, sizeof(buf)) < 0) {
+ fprintf(stderr, "unable to read from %s%s %s\n", sys_pfx, node,
+ strerror(errno));
+ close(fd);
+ abort();
+ }
+ return atoi(buf);
+}
+
+float read_sysfs_float_fd(int fd, const char *sys_pfx, const char *node)
+{
+ char buf[SYSFS_MAX_INT];
+
+ if (read(fd, buf, sizeof(buf)) < 0) {
+
+ fprintf(stderr, "unable to read from %s%s %s\n", sys_pfx, node,
+ strerror(errno));
+ close(fd);
+ abort();
+ }
+ return atof(buf);
+}
+
+int read_sysfs_int(const char *sys_pfx, const char *node)
+{
+ int fd, val;
+
+ fd = open_sysfs(sys_pfx, node, O_RDONLY);
+ val = read_sysfs_int_fd(fd, sys_pfx, node);
+ close(fd);
+ return val;
+}
+
+float read_sysfs_float(const char *sys_pfx, const char *node)
+{
+ int fd;
+ float val;
+
+ fd = open_sysfs(sys_pfx, node, O_RDONLY);
+ val = read_sysfs_float_fd(fd, sys_pfx, node);
+ close(fd);
+ return val;
+}
+
+void write_sysfs_val(const char *sys_pfx, const char *node, int val)
+{
+ int fd, len;
+ char buf[SYSFS_MAX_INT];
+
+ fd = open_sysfs(sys_pfx, node, O_RDWR);
+ len = snprintf(buf, sizeof(buf), "%d", val);
+ if (write(fd, buf, len) < 0) {
+ fprintf(stderr, "unable to write to %s%s %s\n", sys_pfx, node,
+ strerror(errno));
+ close(fd);
+ abort();
+ }
+ close(fd);
+}
+
+static int get_results(struct loopback_test *t)
+{
+ struct loopback_device *d;
+ struct loopback_results *r;
+ int i;
+
+ for (i = 0; i < t->device_count; i++) {
+ if (!device_enabled(t, i))
+ continue;
+
+ d = &t->devices[i];
+ r = &d->results;
+
+ r->error = read_sysfs_int(d->sysfs_entry, "error");
+ r->request_min = read_sysfs_int(d->sysfs_entry, "requests_per_second_min");
+ r->request_max = read_sysfs_int(d->sysfs_entry, "requests_per_second_max");
+ r->request_avg = read_sysfs_float(d->sysfs_entry, "requests_per_second_avg");
+
+ r->latency_min = read_sysfs_int(d->sysfs_entry, "latency_min");
+ r->latency_max = read_sysfs_int(d->sysfs_entry, "latency_max");
+ r->latency_avg = read_sysfs_float(d->sysfs_entry, "latency_avg");
+
+ r->throughput_min = read_sysfs_int(d->sysfs_entry, "throughput_min");
+ r->throughput_max = read_sysfs_int(d->sysfs_entry, "throughput_max");
+ r->throughput_avg = read_sysfs_float(d->sysfs_entry, "throughput_avg");
+
+ r->apbridge_unipro_latency_min =
+ read_sysfs_int(d->sysfs_entry, "apbridge_unipro_latency_min");
+ r->apbridge_unipro_latency_max =
+ read_sysfs_int(d->sysfs_entry, "apbridge_unipro_latency_max");
+ r->apbridge_unipro_latency_avg =
+ read_sysfs_float(d->sysfs_entry, "apbridge_unipro_latency_avg");
+
+ r->gbphy_firmware_latency_min =
+ read_sysfs_int(d->sysfs_entry, "gbphy_firmware_latency_min");
+ r->gbphy_firmware_latency_max =
+ read_sysfs_int(d->sysfs_entry, "gbphy_firmware_latency_max");
+ r->gbphy_firmware_latency_avg =
+ read_sysfs_float(d->sysfs_entry, "gbphy_firmware_latency_avg");
+
+ r->request_jitter = r->request_max - r->request_min;
+ r->latency_jitter = r->latency_max - r->latency_min;
+ r->throughput_jitter = r->throughput_max - r->throughput_min;
+ r->apbridge_unipro_latency_jitter =
+ r->apbridge_unipro_latency_max - r->apbridge_unipro_latency_min;
+ r->gbphy_firmware_latency_jitter =
+ r->gbphy_firmware_latency_max - r->gbphy_firmware_latency_min;
+
+ }
+
+ /*calculate the aggregate results of all enabled devices */
+ if (t->aggregate_output) {
+ r = &t->aggregate_results;
+
+ r->request_min = get_request_min_aggregate(t);
+ r->request_max = get_request_max_aggregate(t);
+ r->request_avg = get_request_avg_aggregate(t);
+
+ r->latency_min = get_latency_min_aggregate(t);
+ r->latency_max = get_latency_max_aggregate(t);
+ r->latency_avg = get_latency_avg_aggregate(t);
+
+ r->throughput_min = get_throughput_min_aggregate(t);
+ r->throughput_max = get_throughput_max_aggregate(t);
+ r->throughput_avg = get_throughput_avg_aggregate(t);
+
+ r->apbridge_unipro_latency_min =
+ get_apbridge_unipro_latency_min_aggregate(t);
+ r->apbridge_unipro_latency_max =
+ get_apbridge_unipro_latency_max_aggregate(t);
+ r->apbridge_unipro_latency_avg =
+ get_apbridge_unipro_latency_avg_aggregate(t);
+
+ r->gbphy_firmware_latency_min =
+ get_gbphy_firmware_latency_min_aggregate(t);
+ r->gbphy_firmware_latency_max =
+ get_gbphy_firmware_latency_max_aggregate(t);
+ r->gbphy_firmware_latency_avg =
+ get_gbphy_firmware_latency_avg_aggregate(t);
+
+ r->request_jitter = r->request_max - r->request_min;
+ r->latency_jitter = r->latency_max - r->latency_min;
+ r->throughput_jitter = r->throughput_max - r->throughput_min;
+ r->apbridge_unipro_latency_jitter =
+ r->apbridge_unipro_latency_max - r->apbridge_unipro_latency_min;
+ r->gbphy_firmware_latency_jitter =
+ r->gbphy_firmware_latency_max - r->gbphy_firmware_latency_min;
+
+ }
+
+ return 0;
+}
+
+void log_csv_error(int len, int err)
+{
+ fprintf(stderr, "unable to write %d bytes to csv %s\n", len,
+ strerror(err));
+}
+
+int format_output(struct loopback_test *t,
+ struct loopback_results *r,
+ const char *dev_name,
+ char *buf, int buf_len,
+ struct tm *tm)
+{
+ int len = 0;
+
+ memset(buf, 0x00, buf_len);
+ len = snprintf(buf, buf_len, "%u-%u-%u %u:%u:%u",
+ tm->tm_year + 1900, tm->tm_mon + 1, tm->tm_mday,
+ tm->tm_hour, tm->tm_min, tm->tm_sec);
+
+ if (t->porcelain) {
+ len += snprintf(&buf[len], buf_len - len,
+ "\n test:\t\t\t%s\n path:\t\t\t%s\n size:\t\t\t%u\n iterations:\t\t%u\n errors:\t\t%u\n async:\t\t\t%s\n",
+ t->test_name,
+ dev_name,
+ t->size,
+ t->iteration_max,
+ r->error,
+ t->use_async ? "Enabled" : "Disabled");
+
+ len += snprintf(&buf[len], buf_len - len,
+ " requests per-sec:\tmin=%u, max=%u, average=%f, jitter=%u\n",
+ r->request_min,
+ r->request_max,
+ r->request_avg,
+ r->request_jitter);
+
+ len += snprintf(&buf[len], buf_len - len,
+ " ap-throughput B/s:\tmin=%u max=%u average=%f jitter=%u\n",
+ r->throughput_min,
+ r->throughput_max,
+ r->throughput_avg,
+ r->throughput_jitter);
+ len += snprintf(&buf[len], buf_len - len,
+ " ap-latency usec:\tmin=%u max=%u average=%f jitter=%u\n",
+ r->latency_min,
+ r->latency_max,
+ r->latency_avg,
+ r->latency_jitter);
+ len += snprintf(&buf[len], buf_len - len,
+ " apbridge-latency usec:\tmin=%u max=%u average=%f jitter=%u\n",
+ r->apbridge_unipro_latency_min,
+ r->apbridge_unipro_latency_max,
+ r->apbridge_unipro_latency_avg,
+ r->apbridge_unipro_latency_jitter);
+
+ len += snprintf(&buf[len], buf_len - len,
+ " gbphy-latency usec:\tmin=%u max=%u average=%f jitter=%u\n",
+ r->gbphy_firmware_latency_min,
+ r->gbphy_firmware_latency_max,
+ r->gbphy_firmware_latency_avg,
+ r->gbphy_firmware_latency_jitter);
+
+ } else {
+ len += snprintf(&buf[len], buf_len- len, ",%s,%s,%u,%u,%u",
+ t->test_name, dev_name, t->size, t->iteration_max,
+ r->error);
+
+ len += snprintf(&buf[len], buf_len - len, ",%u,%u,%f,%u",
+ r->request_min,
+ r->request_max,
+ r->request_avg,
+ r->request_jitter);
+
+ len += snprintf(&buf[len], buf_len - len, ",%u,%u,%f,%u",
+ r->latency_min,
+ r->latency_max,
+ r->latency_avg,
+ r->latency_jitter);
+
+ len += snprintf(&buf[len], buf_len - len, ",%u,%u,%f,%u",
+ r->throughput_min,
+ r->throughput_max,
+ r->throughput_avg,
+ r->throughput_jitter);
+
+ len += snprintf(&buf[len], buf_len - len, ",%u,%u,%f,%u",
+ r->apbridge_unipro_latency_min,
+ r->apbridge_unipro_latency_max,
+ r->apbridge_unipro_latency_avg,
+ r->apbridge_unipro_latency_jitter);
+
+ len += snprintf(&buf[len], buf_len - len, ",%u,%u,%f,%u",
+ r->gbphy_firmware_latency_min,
+ r->gbphy_firmware_latency_max,
+ r->gbphy_firmware_latency_avg,
+ r->gbphy_firmware_latency_jitter);
+ }
+
+ printf("\n%s\n", buf);
+
+ return len;
+}
+
+static int log_results(struct loopback_test *t)
+{
+ int fd, i, len, ret;
+ struct tm tm;
+ time_t local_time;
+ mode_t mode = S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH;
+ char file_name[MAX_SYSFS_PATH];
+ char data[CSV_MAX_LINE];
+
+ local_time = time(NULL);
+ tm = *localtime(&local_time);
+
+ /*
+ * file name will test_name_size_iteration_max.csv
+ * every time the same test with the same parameters is run we will then
+ * append to the same CSV with datestamp - representing each test
+ * dataset.
+ */
+ if (t->file_output && !t->porcelain) {
+ snprintf(file_name, sizeof(file_name), "%s_%d_%d.csv",
+ t->test_name, t->size, t->iteration_max);
+
+ fd = open(file_name, O_WRONLY | O_CREAT | O_APPEND, mode);
+ if (fd < 0) {
+ fprintf(stderr, "unable to open %s for appendation\n", file_name);
+ abort();
+ }
+
+ }
+ for (i = 0; i < t->device_count; i++) {
+ if (!device_enabled(t, i))
+ continue;
+
+ len = format_output(t, &t->devices[i].results,
+ t->devices[i].name,
+ data, sizeof(data), &tm);
+ if (t->file_output && !t->porcelain) {
+ ret = write(fd, data, len);
+ if (ret == -1)
+ fprintf(stderr, "unable to write %d bytes to csv.\n", len);
+ }
+
+ }
+
+
+ if (t->aggregate_output) {
+ len = format_output(t, &t->aggregate_results, "aggregate",
+ data, sizeof(data), &tm);
+ if (t->file_output && !t->porcelain) {
+ ret = write(fd, data, len);
+ if (ret == -1)
+ fprintf(stderr, "unable to write %d bytes to csv.\n", len);
+ }
+ }
+
+ if (t->file_output && !t->porcelain)
+ close(fd);
+
+ return 0;
+}
+
+int is_loopback_device(const char *path, const char *node)
+{
+ char file[MAX_SYSFS_PATH];
+
+ snprintf(file, MAX_SYSFS_PATH, "%s%s/iteration_count", path, node);
+ if (access(file, F_OK) == 0)
+ return 1;
+ return 0;
+}
+
+int find_loopback_devices(struct loopback_test *t)
+{
+ struct dirent **namelist;
+ int i, n, ret;
+ unsigned int dev_id;
+ struct loopback_device *d;
+
+ n = scandir(t->sysfs_prefix, &namelist, NULL, alphasort);
+ if (n < 0) {
+ perror("scandir");
+ ret = -ENODEV;
+ goto baddir;
+ }
+
+ /* Don't include '.' and '..' */
+ if (n <= 2) {
+ ret = -ENOMEM;
+ goto done;
+ }
+
+ for (i = 0; i < n; i++) {
+ ret = sscanf(namelist[i]->d_name, "gb_loopback%u", &dev_id);
+ if (ret != 1)
+ continue;
+
+ if (!is_loopback_device(t->sysfs_prefix, namelist[i]->d_name))
+ continue;
+
+ if (t->device_count == MAX_NUM_DEVICES) {
+ fprintf(stderr, "max number of devices reached!\n");
+ break;
+ }
+
+ d = &t->devices[t->device_count++];
+ snprintf(d->name, MAX_STR_LEN, "gb_loopback%u", dev_id);
+
+ snprintf(d->sysfs_entry, MAX_SYSFS_PATH, "%s%s/",
+ t->sysfs_prefix, d->name);
+
+ snprintf(d->debugfs_entry, MAX_SYSFS_PATH, "%sraw_latency_%s",
+ t->debugfs_prefix, d->name);
+
+ if (t->debug)
+ printf("add %s %s\n", d->sysfs_entry,
+ d->debugfs_entry);
+ }
+
+ ret = 0;
+done:
+ for (i = 0; i < n; i++)
+ free(namelist[n]);
+ free(namelist);
+baddir:
+ return ret;
+}
+
+static int open_poll_files(struct loopback_test *t)
+{
+ struct loopback_device *dev;
+ char buf[MAX_STR_LEN];
+ char dummy;
+ int fds_idx = 0;
+ int i;
+
+ for (i = 0; i < t->device_count; i++) {
+ dev = &t->devices[i];
+
+ if (!device_enabled(t, i))
+ continue;
+
+ snprintf(buf, sizeof(buf), "%s%s", dev->sysfs_entry, "iteration_count");
+ t->fds[fds_idx].fd = open(buf, O_RDONLY);
+ if (t->fds[fds_idx].fd < 0) {
+ fprintf(stderr, "Error opening poll file!\n");
+ goto err;
+ }
+ read(t->fds[fds_idx].fd, &dummy, 1);
+ t->fds[fds_idx].events = POLLERR|POLLPRI;
+ t->fds[fds_idx].revents = 0;
+ fds_idx++;
+ }
+
+ t->poll_count = fds_idx;
+
+ return 0;
+
+err:
+ for (i = 0; i < fds_idx; i++)
+ close(t->fds[fds_idx].fd);
+
+ return -1;
+}
+
+static int close_poll_files(struct loopback_test *t)
+{
+ int i;
+ for (i = 0; i < t->poll_count; i++)
+ close(t->fds[i].fd);
+
+ return 0;
+}
+static int is_complete(struct loopback_test *t)
+{
+ int iteration_count;
+ int i;
+
+ for (i = 0; i < t->device_count; i++) {
+ if (!device_enabled(t, i))
+ continue;
+
+ iteration_count = read_sysfs_int(t->devices[i].sysfs_entry,
+ "iteration_count");
+
+ /* at least one device did not finish yet */
+ if (iteration_count != t->iteration_max)
+ return 0;
+ }
+
+ return 1;
+}
+
+static void stop_tests(struct loopback_test *t)
+{
+ int i;
+
+ for (i = 0; i < t->device_count; i++) {
+ if (!device_enabled(t, i))
+ continue;
+ write_sysfs_val(t->devices[i].sysfs_entry, "type", 0);
+ }
+}
+
+static void handler(int sig) { /* do nothing */ }
+
+static int wait_for_complete(struct loopback_test *t)
+{
+ int number_of_events = 0;
+ char dummy;
+ int ret;
+ int i;
+ struct timespec *ts = NULL;
+ struct sigaction sa;
+ sigset_t mask_old, mask;
+
+ sigemptyset(&mask);
+ sigemptyset(&mask_old);
+ sigaddset(&mask, SIGINT);
+ sigprocmask(SIG_BLOCK, &mask, &mask_old);
+
+ sa.sa_handler = handler;
+ sa.sa_flags = 0;
+ sigemptyset(&sa.sa_mask);
+ if (sigaction(SIGINT, &sa, NULL) == -1) {
+ fprintf(stderr, "sigaction error\n");
+ return -1;
+ }
+
+ if (t->poll_timeout.tv_sec != 0)
+ ts = &t->poll_timeout;
+
+ while (1) {
+
+ ret = ppoll(t->fds, t->poll_count, ts, &mask_old);
+ if (ret <= 0) {
+ stop_tests(t);
+ fprintf(stderr, "Poll exit with errno %d\n", errno);
+ return -1;
+ }
+
+ for (i = 0; i < t->poll_count; i++) {
+ if (t->fds[i].revents & POLLPRI) {
+ /* Dummy read to clear the event */
+ read(t->fds[i].fd, &dummy, 1);
+ number_of_events++;
+ }
+ }
+
+ if (number_of_events == t->poll_count)
+ break;
+ }
+
+ if (!is_complete(t)) {
+ fprintf(stderr, "Iteration count did not finish!\n");
+ return -1;
+ }
+
+ return 0;
+}
+
+static void prepare_devices(struct loopback_test *t)
+{
+ int i;
+
+ /* Cancel any running tests on enabled devices. If
+ * stop_all option is given, stop test on all devices.
+ */
+ for (i = 0; i < t->device_count; i++)
+ if (t->stop_all || device_enabled(t, i))
+ write_sysfs_val(t->devices[i].sysfs_entry, "type", 0);
+
+
+ for (i = 0; i < t->device_count; i++) {
+ if (!device_enabled(t, i))
+ continue;
+
+ write_sysfs_val(t->devices[i].sysfs_entry, "us_wait",
+ t->us_wait);
+
+ /* Set operation size */
+ write_sysfs_val(t->devices[i].sysfs_entry, "size", t->size);
+
+ /* Set iterations */
+ write_sysfs_val(t->devices[i].sysfs_entry, "iteration_max",
+ t->iteration_max);
+
+ if (t->use_async) {
+ write_sysfs_val(t->devices[i].sysfs_entry,
+ "async", 1);
+ write_sysfs_val(t->devices[i].sysfs_entry,
+ "timeout", t->async_timeout);
+ write_sysfs_val(t->devices[i].sysfs_entry,
+ "outstanding_operations_max",
+ t->async_outstanding_operations);
+ } else
+ write_sysfs_val(t->devices[i].sysfs_entry,
+ "async", 0);
+ }
+}
+
+static int start(struct loopback_test *t)
+{
+ int i;
+
+ /* the test starts by writing test_id to the type file. */
+ for (i = 0; i < t->device_count; i++) {
+ if (!device_enabled(t, i))
+ continue;
+
+ write_sysfs_val(t->devices[i].sysfs_entry, "type", t->test_id);
+ }
+
+ return 0;
+}
+
+
+void loopback_run(struct loopback_test *t)
+{
+ int i;
+ int ret;
+
+ for (i = 0; dict[i].name != NULL; i++) {
+ if (strstr(dict[i].name, t->test_name))
+ t->test_id = dict[i].type;
+ }
+ if (!t->test_id) {
+ fprintf(stderr, "invalid test %s\n", t->test_name);
+ usage();
+ return;
+ }
+
+ prepare_devices(t);
+
+ ret = open_poll_files(t);
+ if (ret)
+ goto err;
+
+ start(t);
+
+ ret = wait_for_complete(t);
+ close_poll_files(t);
+ if (ret)
+ goto err;
+
+
+ get_results(t);
+
+ log_results(t);
+
+ return;
+
+err:
+ printf("Error running test\n");
+ return;
+}
+
+static int sanity_check(struct loopback_test *t)
+{
+ int i;
+
+ if (t->device_count == 0) {
+ fprintf(stderr, "No loopback devices found\n");
+ return -1;
+ }
+
+ for (i = 0; i < MAX_NUM_DEVICES; i++) {
+ if (!device_enabled(t, i))
+ continue;
+
+ if (t->mask && !strcmp(t->devices[i].name, "")) {
+ fprintf(stderr, "Bad device mask %x\n", (1 << i));
+ return -1;
+ }
+
+ }
+
+
+ return 0;
+}
+
+int main(int argc, char *argv[])
+{
+ int o, ret;
+ char *sysfs_prefix = "/sys/class/gb_loopback/";
+ char *debugfs_prefix = "/sys/kernel/debug/gb_loopback/";
+
+ memset(&t, 0, sizeof(t));
+
+ while ((o = getopt(argc, argv,
+ "t:s:i:S:D:m:v::d::r::p::a::l::x::o:O:c:w:z::f::")) != -1) {
+ switch (o) {
+ case 't':
+ snprintf(t.test_name, MAX_STR_LEN, "%s", optarg);
+ break;
+ case 's':
+ t.size = atoi(optarg);
+ break;
+ case 'i':
+ t.iteration_max = atoi(optarg);
+ break;
+ case 'S':
+ snprintf(t.sysfs_prefix, MAX_SYSFS_PATH, "%s", optarg);
+ break;
+ case 'D':
+ snprintf(t.debugfs_prefix, MAX_SYSFS_PATH, "%s", optarg);
+ break;
+ case 'm':
+ t.mask = atol(optarg);
+ break;
+ case 'v':
+ t.verbose = 1;
+ break;
+ case 'd':
+ t.debug = 1;
+ break;
+ case 'r':
+ t.raw_data_dump = 1;
+ break;
+ case 'p':
+ t.porcelain = 1;
+ break;
+ case 'a':
+ t.aggregate_output = 1;
+ break;
+ case 'l':
+ t.list_devices = 1;
+ break;
+ case 'x':
+ t.use_async = 1;
+ break;
+ case 'o':
+ t.async_timeout = atoi(optarg);
+ break;
+ case 'O':
+ t.poll_timeout.tv_sec = atoi(optarg);
+ break;
+ case 'c':
+ t.async_outstanding_operations = atoi(optarg);
+ break;
+ case 'w':
+ t.us_wait = atoi(optarg);
+ break;
+ case 'z':
+ t.file_output = 1;
+ break;
+ case 'f':
+ t.stop_all = 1;
+ break;
+ default:
+ usage();
+ return -EINVAL;
+ }
+ }
+
+ if (!strcmp(t.sysfs_prefix, ""))
+ snprintf(t.sysfs_prefix, MAX_SYSFS_PATH, "%s", sysfs_prefix);
+
+ if (!strcmp(t.debugfs_prefix, ""))
+ snprintf(t.debugfs_prefix, MAX_SYSFS_PATH, "%s", debugfs_prefix);
+
+ ret = find_loopback_devices(&t);
+ if (ret)
+ return ret;
+ ret = sanity_check(&t);
+ if (ret)
+ return ret;
+
+ if (t.list_devices) {
+ show_loopback_devices(&t);
+ return 0;
+ }
+
+ if (t.test_name[0] == '\0' || t.iteration_max == 0)
+ usage();
+
+ if (t.async_timeout == 0)
+ t.async_timeout = DEFAULT_ASYNC_TIMEOUT;
+
+ loopback_run(&t);
+
+ return 0;
+}
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 31/32] greybus: documentation
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (28 preceding siblings ...)
2016-09-16 14:22 ` [patch 30/32] greybus: tools Greg KH
@ 2016-09-16 14:22 ` Greg KH
2016-09-16 14:22 ` [patch 32/32] greybus: add to the build Greg KH
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:22 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This describes all of the sysfs files that greybus exposes to userspace
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
Documentation/ABI/testing/sysfs-bus-greybus | 275 ++++++++++++++++++++++++++++
1 file changed, 275 insertions(+)
--- /dev/null
+++ b/Documentation/ABI/testing/sysfs-bus-greybus
@@ -0,0 +1,275 @@
+What: /sys/bus/greybus/devices/greybusN
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ The "root" greybus device for the Greybus device tree, or bus,
+ where N is a dynamically assigned 1-based id.
+
+What: /sys/bus/greybus/devices/greybusN/bus_id
+Date: April 2016
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ The ID of the "root" greybus device, or bus.
+
+What: /sys/bus/greybus/devices/N-M
+Date: March 2016
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ A Module M on the bus N, where M is the 1-byte interface
+ ID of the module's primary interface.
+
+What: /sys/bus/greybus/devices/N-M/eject
+Date: March 2016
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ Writing a non-zero argument to this attibute disables the
+ module's interfaces before physically ejecting it.
+
+What: /sys/bus/greybus/devices/N-M/module_id
+Date: March 2016
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ The ID of a Greybus module, corresponding to the ID of its
+ primary interface.
+
+What: /sys/bus/greybus/devices/N-M/num_interfaces
+Date: March 2016
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ The number of interfaces of a module.
+
+What: /sys/bus/greybus/devices/N-M.I
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ An Interface I on the bus N and module N-M, where I is the
+ 1-byte interface ID.
+
+What: /sys/bus/greybus/devices/N-M.I/current_now
+Date: March 2016
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ Current measurement of the interface in microamps (uA)
+
+What: /sys/bus/greybus/devices/N-M.I/ddbl1_manufacturer_id
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ Unipro Device Descriptor Block Level 1 manufacturer ID for the
+ greybus Interface.
+
+What: /sys/bus/greybus/devices/N-M.I/ddbl1_product_id
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ Unipro Device Descriptor Block Level 1 product ID for the
+ greybus Interface.
+
+What: /sys/bus/greybus/devices/N-M.I/interface_id
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ The ID of a Greybus interface.
+
+What: /sys/bus/greybus/devices/N-M.I/interface_type
+Date: June 2016
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ The type of a Greybus interface; "dummy", "unipro", "greybus",
+ or "unknown".
+
+What: /sys/bus/greybus/devices/N-M.I/power_now
+Date: March 2016
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ Power measurement of the interface in microwatts (uW)
+
+What: /sys/bus/greybus/devices/N-M.I/power_state
+Date: March 2016
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ This file reflects the power state of a Greybus interface. If
+ the value read from it is "on", then power is currently
+ supplied to the interface. Otherwise it will read "off" and
+ power is currently not supplied to the interface.
+
+ If the value read is "off", then writing "on" (or '1', 'y',
+ 'Y') to this file will enable power to the interface and an
+ attempt to boot and possibly enumerate it will be made. Note
+ that on errors, the interface will again be powered down.
+
+ If the value read is "on", then writing "off" (or '0', 'n',
+ 'N') to this file will power down the interface.
+
+What: /sys/bus/greybus/devices/N-M.I/product_id
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ Product ID of a Greybus interface.
+
+What: /sys/bus/greybus/devices/N-M.I/serial_number
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ Serial Number of the Greybus interface, represented by a 64 bit
+ hexadecimal number.
+
+What: /sys/bus/greybus/devices/N-M.I/vendor_id
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ Vendor ID of a Greybus interface.
+
+What: /sys/bus/greybus/devices/N-M.I/voltage_now
+Date: March 2016
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ Voltage measurement of the interface in microvolts (uV)
+
+What: /sys/bus/greybus/devices/N-M.I.ctrl
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ Abstract control device for interface I that represents the
+ current mode of an enumerated Greybus interface.
+
+What: /sys/bus/greybus/devices/N-M.I.ctrl/product_string
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ Product ID string of a Greybus interface.
+
+What: /sys/bus/greybus/devices/N-M.I.ctrl/vendor_string
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ Vendor ID string of a Greybus interface.
+
+What: /sys/bus/greybus/devices/N-M.I.B
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ A bundle B on the Interface I, B is replaced by a 1-byte
+ number representing the bundle.
+
+What: /sys/bus/greybus/devices/N-M.I.B/bundle_class
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ The greybus class of the bundle B.
+
+What: /sys/bus/greybus/devices/N-M.I.B/bundle_id
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ The interface-unique id of the bundle B.
+
+What: /sys/bus/greybus/devices/N-M.I.B/gpbX
+Date: April 2016
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ The General Purpose Bridged PHY device of the bundle B,
+ where X is a dynamically assigned 0-based id.
+
+What: /sys/bus/greybus/devices/N-M.I.B/state
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ A bundle has a state that is managed by the userspace
+ Endo process. This file allows that Endo to signal
+ other Android HALs that the state of the bundle has
+ changed to a specific value. When written to, any
+ process watching the file will be woken up, and the new
+ value can be read. It's a "poor-man's IPC", yes, but
+ simplifies the Android userspace code immensely.
+
+What: /sys/bus/greybus/devices/N-svc
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ The singleton SVC device of bus N.
+
+What: /sys/bus/greybus/devices/N-svc/ap_intf_id
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ The AP interface ID, a 1-byte non-zero integer which
+ defines the position of the AP module on the frame.
+ The interface positions are defined in the GMP
+ Module Developer Kit.
+
+What: /sys/bus/greybus/devices/N-svc/endo_id
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ The Endo ID, which is a 2-byte hexadecimal value
+ defined by the Endo layout scheme, documented in
+ the GMP Module Developer Kit.
+
+What: /sys/bus/greybus/devices/N-svc/intf_eject
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ Write the number of the interface that you wish to
+ forcibly eject from the system.
+
+What: /sys/bus/greybus/devices/N-svc/version
+Date: October 2015
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ The version number of the firmware in the SVC device.
+
+What: /sys/bus/greybus/devices/N-svc/watchdog
+Date: October 2016
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ If the SVC watchdog is enabled or not. Writing 0 to this
+ file will disable the watchdog, writing 1 will enable it.
+
+What: /sys/bus/greybus/devices/N-svc/watchdog_action
+Date: July 2016
+KernelVersion: 4.9
+Contact: Greg Kroah-Hartman <greg@kroah.com>
+Description:
+ This attribute indicates the action to be performed upon SVC
+ watchdog bite.
+
+ The action can be one of the "reset" or "panic". Writing either
+ one of the "reset" or "panic" will change the behavior of SVC
+ watchdog bite. Default value is "reset".
+
+ "reset" means the UniPro subsystem is to be reset.
+
+ "panic" means SVC watchdog bite will cause kernel to panic.
^ permalink raw reply [flat|nested] 84+ messages in thread
* [patch 32/32] greybus: add to the build
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
` (29 preceding siblings ...)
2016-09-16 14:22 ` [patch 31/32] greybus: documentation Greg KH
@ 2016-09-16 14:22 ` Greg KH
30 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:22 UTC (permalink / raw)
To: Arnd Bergmann, linux-kernel
Cc: Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
This adds the drivers/greybus/ directory to the build and provides a
valid Kconfig file.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/Kconfig | 2
drivers/Makefile | 1
drivers/greybus/Kconfig | 219 +++++++++++++++++++++++++++++++++++++++++++++++
drivers/greybus/Makefile | 96 ++++++++++++++++++++
4 files changed, 318 insertions(+)
--- a/drivers/Kconfig
+++ b/drivers/Kconfig
@@ -202,4 +202,6 @@ source "drivers/hwtracing/intel_th/Kconf
source "drivers/fpga/Kconfig"
+source "drivers/greybus/Kconfig"
+
endmenu
--- a/drivers/Makefile
+++ b/drivers/Makefile
@@ -173,3 +173,4 @@ obj-$(CONFIG_STM) += hwtracing/stm/
obj-$(CONFIG_ANDROID) += android/
obj-$(CONFIG_NVMEM) += nvmem/
obj-$(CONFIG_FPGA) += fpga/
+obj-$(CONFIG_GREYBUS) += greybus/
--- /dev/null
+++ b/drivers/greybus/Kconfig
@@ -0,0 +1,219 @@
+menuconfig GREYBUS
+ tristate "Greybus support"
+ depends on SYSFS
+ ---help---
+ This option enables the Greybus driver core. Greybus is an
+ hardware protocol that was designed to provide Unipro with a
+ sane application layer. It was originally designed for the
+ ARA project, a module phone system, but has shown up in other
+ phones, and can be tunneled over other busses in order to
+ control hardware devices.
+
+ Say Y here to enable support for these types of drivers.
+
+ To compile this code as a module, chose M here: the module
+ will be called greybus.ko
+
+if GREYBUS
+
+config GREYBUS_ES2
+ tristate "Greybus ES3 USB host controller"
+ depends on USB
+ ---help---
+ Select this option if you have a Toshiba ES3 USB device that
+ acts as a Greybus "host controller". This device is a bridge
+ from a USB device to a Unipro network.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-es2.ko
+
+config GREYBUS_AUDIO
+ tristate "Greybus Audio Class driver"
+ depends on SOUND
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus Audio Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-audio.ko
+
+config GREYBUS_BOOTROM
+ tristate "Greybus Bootrom Class driver"
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus Bootrom Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-bootrom.ko
+
+config GREYBUS_CAMERA
+ tristate "Greybus Camera Class driver"
+ depends on MEDIA && LEDS_CLASS_FLASH && BROKEN
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus Camera Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-camera.ko
+
+config GREYBUS_FIRMWARE
+ tristate "Greybus Firmware Download Class driver"
+ depends on SPI
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus Firmware Download Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-firmware.ko
+
+config GREYBUS_HID
+ tristate "Greybus HID Class driver"
+ depends on HID && INPUT
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus HID Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-hid.ko
+
+config GREYBUS_LIGHT
+ tristate "Greybus LED Class driver"
+ depends on LEDS_CLASS
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus LED Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-light.ko
+
+config GREYBUS_LOG
+ tristate "Greybus Debug Log Class driver"
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus Debug Log Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-log.ko
+
+config GREYBUS_LOOPBACK
+ tristate "Greybus Loopback Class driver"
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus Debug Log Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-log.ko
+
+config GREYBUS_POWER
+ tristate "Greybus Powersupply Class driver"
+ depends on POWER_SUPPLY
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus Powersupply Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-power-supply.ko
+
+config GREYBUS_RAW
+ tristate "Greybus Raw Class driver"
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus Raw Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-raw.ko
+
+config GREYBUS_VIBRATOR
+ tristate "Greybus Vibrator Motor Class driver"
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus Vibrator Motor Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-vibrator.ko
+
+menuconfig GREYBUS_BRIDGED_PHY
+ tristate "Greybus Bridged PHY Class drivers"
+ ---help---
+ Select this option to pick from a variety of Greybus Bridged
+ PHY class drivers. These drivers emulate a number of
+ different "traditional" busses by tunneling them over Greybus.
+ Examples of this include serial, SPI, USB, and others.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-phy.ko
+
+if GREYBUS_BRIDGED_PHY
+
+config GREYBUS_GPIO
+ tristate "Greybus GPIO Bridged PHY driver"
+ depends on GPIO
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus GPIO Bridged PHY Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-gpio.ko
+
+config GREYBUS_I2C
+ tristate "Greybus I2C Bridged PHY driver"
+ depends on I2C
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus I2C Bridged PHY Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-i2c.ko
+
+config GREYBUS_PWM
+ tristate "Greybus PWM Bridged PHY driver"
+ depends on PWM
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus PWM Bridged PHY Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-pwm.ko
+
+config GREYBUS_SDIO
+ tristate "Greybus SDIO Bridged PHY driver"
+ depends on MMC
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus SDIO Bridged PHY Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-sdio.ko
+
+config GREYBUS_SPI
+ tristate "Greybus SPI Bridged PHY driver"
+ depends on SPI
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus SPI Bridged PHY Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-spi.ko
+
+config GREYBUS_UART
+ tristate "Greybus UART Bridged PHY driver"
+ depends on TTY
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus UART Bridged PHY Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-uart.ko
+
+config GREYBUS_USB
+ tristate "Greybus USB Host Bridged PHY driver"
+ depends on USB
+ ---help---
+ Select this option if you have a device that follows the
+ Greybus USB Host Bridged PHY Class specification.
+
+ To compile this code as a module, chose M here: the module
+ will be called gb-usb.ko
+
+endif # GREYBUS_BRIDGED_PHY
+endif # GREYBUS
--- /dev/null
+++ b/drivers/greybus/Makefile
@@ -0,0 +1,96 @@
+# Greybus core
+greybus-y := core.o \
+ debugfs.o \
+ hd.o \
+ manifest.o \
+ module.o \
+ interface.o \
+ bundle.o \
+ connection.o \
+ control.o \
+ svc.o \
+ svc_watchdog.o \
+ operation.o \
+ timesync.o \
+ timesync_platform.o
+
+obj-$(CONFIG_GREYBUS) += greybus.o
+
+# needed for trace events
+ccflags-y += -I$(src)
+
+
+# Greybus Host controller drivers
+gb-es2-y := es2.o
+
+obj-$(CONFIG_GREYBUS_ES2) += gb-es2.o
+
+# Greybus class drivers
+gb-bootrom-y := bootrom.o
+gb-camera-y := camera.o
+gb-firmware-y := fw-core.o fw-download.o fw-management.o authentication.o
+gb-spilib-y := spilib.o
+gb-hid-y := hid.o
+gb-light-y := light.o
+gb-log-y := log.o
+gb-loopback-y := loopback.o
+gb-power-supply-y := power_supply.o
+gb-raw-y := raw.o
+gb-vibrator-y := vibrator.o
+
+obj-$(CONFIG_GREYBUS_BOOTROM) += gb-bootrom.o
+obj-$(CONFIG_GREYBUS_CAMERA) += gb-camera.o
+obj-$(CONFIG_GREYBUS_FIRMWARE) += gb-firmware.o gb-spilib.o
+obj-$(CONFIG_GREYBUS_HID) += gb-hid.o
+obj-$(CONFIG_GREYBUS_LIGHT) += gb-light.o
+obj-$(CONFIG_GREYBUS_LOG) += gb-log.o
+obj-$(CONFIG_GREYBUS_LOOPBACK) += gb-loopback.o
+obj-$(CONFIG_GREYBUS_POWER) += gb-power-supply.o
+obj-$(CONFIG_GREYBUS_RAW) += gb-raw.o
+obj-$(CONFIG_GREYBUS_VIBRATOR) += gb-vibrator.o
+
+# Greybus Audio is a bunch of modules
+gb-audio-module-y := audio_module.o audio_topology.o
+gb-audio-codec-y := audio_codec.o
+gb-audio-gb-y := audio_gb.o
+gb-audio-apbridgea-y := audio_apbridgea.o
+gb-audio-manager-y := audio_manager.o audio_manager_module.o
+
+# Greybus Audio sysfs helpers can be useful when debugging
+#GB_AUDIO_MANAGER_SYSFS ?= true
+#ifeq ($(GB_AUDIO_MANAGER_SYSFS),true)
+#gb-audio-manager-y += audio_manager_sysfs.o
+#ccflags-y += -DGB_AUDIO_MANAGER_SYSFS
+#endif
+
+obj-$(CONFIG_GREYBUS_AUDIO_MSM8994) += gb-audio-codec.o
+obj-$(CONFIG_GREYBUS_AUDIO_MSM8994) += gb-audio-module.o
+obj-$(CONFIG_GREYBUS_AUDIO) += gb-audio-gb.o
+obj-$(CONFIG_GREYBUS_AUDIO) += gb-audio-apbridgea.o
+obj-$(CONFIG_GREYBUS_AUDIO) += gb-audio-manager.o
+
+
+# Greybus Bridged PHY drivers
+gb-gbphy-y := gbphy.o
+gb-gpio-y := gpio.o
+gb-i2c-y := i2c.o
+gb-pwm-y := pwm.o
+gb-sdio-y := sdio.o
+gb-spi-y := spi.o
+gb-uart-y := uart.o
+gb-usb-y := usb.o
+
+obj-$(CONFIG_GREYBUS_BRIDGED_PHY) += gb-gbphy.o
+obj-$(CONFIG_GREYBUS_GPIO) += gb-gpio.o
+obj-$(CONFIG_GREYBUS_I2C) += gb-i2c.o
+obj-$(CONFIG_GREYBUS_PWM) += gb-pwm.o
+obj-$(CONFIG_GREYBUS_SDIO) += gb-sdio.o
+obj-$(CONFIG_GREYBUS_SPI) += gb-spi.o gb-spilib.o
+obj-$(CONFIG_GREYBUS_UART) += gb-uart.o
+obj-$(CONFIG_GREYBUS_USB) += gb-usb.o
+
+
+# Greybus Platform driver
+gb-arche-y := arche-platform.o arche-apb-ctrl.o
+
+obj-$(CONFIG_USB_HSIC_USB3613) += gb-arche.o
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-16 13:22 ` Greg KH
@ 2016-09-16 14:24 ` Greg KH
2016-09-20 6:41 ` Greg KH
0 siblings, 1 reply; 84+ messages in thread
From: Greg KH @ 2016-09-16 14:24 UTC (permalink / raw)
To: Mark Brown
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
On Fri, Sep 16, 2016 at 03:22:08PM +0200, Greg KH wrote:
> On Fri, Sep 16, 2016 at 11:18:07AM +0100, Mark Brown wrote:
> > On Fri, Sep 16, 2016 at 08:05:19AM +0200, Greg KH wrote:
> > > On Thu, Sep 15, 2016 at 03:45:53PM +0100, Mark Brown wrote:
> >
> > > > Sending a pull request for code that's never been seen upstream seems
> > > > completely premature.
> >
> > > Hey, how does code get upstream then? :)
> >
> > By having the actual code to the mailing list.
>
> Ugh, vger keeps blocking the patches, I'm going to try for a third time
> now...
third time seems to have worked, except for patch 01/32, which was only
a single .h file. That file can be seen here if people are curious:
https://git.kernel.org/cgit/linux/kernel/git/gregkh/char-misc.git/tree/drivers/greybus/greybus_protocols.h?h=greybus
I'll now respin the series, moving everything into
drivers/staging/greybus/ and leave it there for 4.9. Then we can work
on moving the core out, and then the drivers to the needed locations.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-16 14:24 ` Greg KH
@ 2016-09-20 6:41 ` Greg KH
2016-09-20 7:12 ` Vaibhav Agarwal
0 siblings, 1 reply; 84+ messages in thread
From: Greg KH @ 2016-09-20 6:41 UTC (permalink / raw)
To: Mark Brown
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
On Fri, Sep 16, 2016 at 04:24:59PM +0200, Greg KH wrote:
> On Fri, Sep 16, 2016 at 03:22:08PM +0200, Greg KH wrote:
> > On Fri, Sep 16, 2016 at 11:18:07AM +0100, Mark Brown wrote:
> > > On Fri, Sep 16, 2016 at 08:05:19AM +0200, Greg KH wrote:
> > > > On Thu, Sep 15, 2016 at 03:45:53PM +0100, Mark Brown wrote:
> > >
> > > > > Sending a pull request for code that's never been seen upstream seems
> > > > > completely premature.
> > >
> > > > Hey, how does code get upstream then? :)
> > >
> > > By having the actual code to the mailing list.
> >
> > Ugh, vger keeps blocking the patches, I'm going to try for a third time
> > now...
>
> third time seems to have worked, except for patch 01/32, which was only
> a single .h file. That file can be seen here if people are curious:
> https://git.kernel.org/cgit/linux/kernel/git/gregkh/char-misc.git/tree/drivers/greybus/greybus_protocols.h?h=greybus
>
> I'll now respin the series, moving everything into
> drivers/staging/greybus/ and leave it there for 4.9. Then we can work
> on moving the core out, and then the drivers to the needed locations.
This is now all in drivers/staging/greybus/ and will show up in the next
linux-next.
And because of this, we already have a checkpatch.pl cleanup patch sent
to us for the code for things we missed :)
thanks,
greg k-h
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-20 6:41 ` Greg KH
@ 2016-09-20 7:12 ` Vaibhav Agarwal
0 siblings, 0 replies; 84+ messages in thread
From: Vaibhav Agarwal @ 2016-09-20 7:12 UTC (permalink / raw)
To: Greg KH
Cc: Mark Brown, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Mark Greer, Vaibhav Agarwal,
hvaibhav.linux
+new email ids for myself & Vaibhav H
--
thanks,
./va
On 20 September 2016 at 12:11, Greg KH <gregkh@linuxfoundation.org> wrote:
> On Fri, Sep 16, 2016 at 04:24:59PM +0200, Greg KH wrote:
>> On Fri, Sep 16, 2016 at 03:22:08PM +0200, Greg KH wrote:
>> > On Fri, Sep 16, 2016 at 11:18:07AM +0100, Mark Brown wrote:
>> > > On Fri, Sep 16, 2016 at 08:05:19AM +0200, Greg KH wrote:
>> > > > On Thu, Sep 15, 2016 at 03:45:53PM +0100, Mark Brown wrote:
>> > >
>> > > > > Sending a pull request for code that's never been seen upstream seems
>> > > > > completely premature.
>> > >
>> > > > Hey, how does code get upstream then? :)
>> > >
>> > > By having the actual code to the mailing list.
>> >
>> > Ugh, vger keeps blocking the patches, I'm going to try for a third time
>> > now...
>>
>> third time seems to have worked, except for patch 01/32, which was only
>> a single .h file. That file can be seen here if people are curious:
>> https://git.kernel.org/cgit/linux/kernel/git/gregkh/char-misc.git/tree/drivers/greybus/greybus_protocols.h?h=greybus
>>
>> I'll now respin the series, moving everything into
>> drivers/staging/greybus/ and leave it there for 4.9. Then we can work
>> on moving the core out, and then the drivers to the needed locations.
>
> This is now all in drivers/staging/greybus/ and will show up in the next
> linux-next.
>
> And because of this, we already have a checkpatch.pl cleanup patch sent
> to us for the code for things we missed :)
>
> thanks,
>
> greg k-h
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-16 6:05 ` Greg KH
2016-09-16 10:18 ` Mark Brown
2016-09-16 12:18 ` Arnd Bergmann
@ 2016-09-21 13:02 ` Mark Rutland
2016-09-21 14:13 ` Greg KH
2 siblings, 1 reply; 84+ messages in thread
From: Mark Rutland @ 2016-09-21 13:02 UTC (permalink / raw)
To: Greg KH
Cc: Mark Brown, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
Hi,
Apologies for the late reply to this; I don't subscribe to LKML with my
work address and didn't spot this sub-thread until now.
On Fri, Sep 16, 2016 at 08:05:19AM +0200, Greg KH wrote:
> On Thu, Sep 15, 2016 at 03:45:53PM +0100, Mark Brown wrote:
> > On Wed, Sep 14, 2016 at 12:09:49PM +0200, Greg KH wrote:
> >
> > > I'll send out a follow-up set of "simple" patches that just add the
> > > files to the kernel tree, to give people an idea of the code involved.
> > > Overall, it's a tiny stand-alone driver subsystem, only 37k lines, that
> > > implements a protocol which allows for "generic" cameras, audio devices,
> > > and other class type devices, as well as a bridged "physical" layer
> > > protocol to talk to serial, spi, uart, pwm, gpio, i2c, and even USB host
> >
> > Just emphasizing what Mark Rutland said complete NACK, in particular the
> > drivers for functions have not been posted upstream at all (and it's
> > concerning that they're all being added under drivers/greybus rather
> > than within the relevant subsystem like we do normally). I've not
> > looked at the code as it has not been submitted but given that and that
> > the reason I found this pull request was an ASoC contributor who had
> > seen it and looked at the code going "oh dear, greybus..." on IRC I'm
> > very concerned.
> >
> > Sending a pull request for code that's never been seen upstream seems
> > completely premature.
>
> Hey, how does code get upstream then? :)
As Mark Brown said, after review.
> Anyway, as vger seems to be marking all of the greybus patches I send
> out as "spam" according to a filter, it's making it a bit hard to send
> patches out, but I'll try it again "by hand" later today...
>
> As for the drivers all living under drivers/greybus/ I understand, but
> we need the greybus core present first before we can get the drivers in.
> How about we do what happened with IIO, we take the greybus core code in
> drivers/greybus/ and put the drivers in staging, and then move them out
> of staging into the "real" portion of the kernel as they get reviewed
> and accepted?
>
> Any objections to that workflow?
Personally, yes.
I would rather see the few core patches go through some level of review
first. It's vastly easier to handle that than to reverse-engineer an
understanding of the code when "move XXX out of staging/" patches
appear.
I'm more than happy to review a reasonably-size, linearised series for
that core code.
My concerns previously mentioned still apply to the patches queued in
linux-next, regardless of whether these patches are under staging.
Especially given the ABI implications of the devicetree bindings.
Thanks,
Mark.
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [GIT PULL] Greybus driver subsystem for 4.9-rc1
2016-09-21 13:02 ` Mark Rutland
@ 2016-09-21 14:13 ` Greg KH
0 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-09-21 14:13 UTC (permalink / raw)
To: Mark Rutland
Cc: Mark Brown, Arnd Bergmann, linux-kernel, Johan Hovold,
Rui Miguel Silva, Laurent Pinchart, Sandeep Patil, Matt Porter,
John Stultz, Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
On Wed, Sep 21, 2016 at 02:02:10PM +0100, Mark Rutland wrote:
> > As for the drivers all living under drivers/greybus/ I understand, but
> > we need the greybus core present first before we can get the drivers in.
> > How about we do what happened with IIO, we take the greybus core code in
> > drivers/greybus/ and put the drivers in staging, and then move them out
> > of staging into the "real" portion of the kernel as they get reviewed
> > and accepted?
> >
> > Any objections to that workflow?
>
> Personally, yes.
>
> I would rather see the few core patches go through some level of review
> first. It's vastly easier to handle that than to reverse-engineer an
> understanding of the code when "move XXX out of staging/" patches
> appear.
>
> I'm more than happy to review a reasonably-size, linearised series for
> that core code.
>
> My concerns previously mentioned still apply to the patches queued in
> linux-next, regardless of whether these patches are under staging.
> Especially given the ABI implications of the devicetree bindings.
There are no ABI implications just yet, we are free to change anything
we want. It's all in drivers/staging/ now, where we will clean it up
some more over the next few weeks and then post reviewable sets of
patches to move portions out of staging (greybus core first, then some
class drivers, then the subsystem-specific things). That will give you
your set of clean patches to review, and with the first rounds, no
device tree bindings at all :)
thanks,
greg k-h
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [patch 14/32] greybus: LED driver
2016-09-16 14:10 ` [patch 14/32] greybus: LED driver Greg KH
@ 2016-10-07 13:36 ` Pavel Machek
2016-10-07 13:41 ` Greg KH
0 siblings, 1 reply; 84+ messages in thread
From: Pavel Machek @ 2016-10-07 13:36 UTC (permalink / raw)
To: Greg KH
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer, rpurdie,
j.anaszewski, linux-leds
[-- Attachment #1: Type: text/plain, Size: 409 bytes --]
On Fri 2016-09-16 16:10:14, Greg KH wrote:
>
> This driver implements the Greybus LED protocol.
I guess this should be copied to LED mainainters?
Havingstandartized phone hardware to work would be great. Is it
possible to get device using this?
Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [patch 14/32] greybus: LED driver
2016-10-07 13:36 ` Pavel Machek
@ 2016-10-07 13:41 ` Greg KH
0 siblings, 0 replies; 84+ messages in thread
From: Greg KH @ 2016-10-07 13:41 UTC (permalink / raw)
To: Pavel Machek
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer, rpurdie,
j.anaszewski, linux-leds
On Fri, Oct 07, 2016 at 03:36:56PM +0200, Pavel Machek wrote:
> On Fri 2016-09-16 16:10:14, Greg KH wrote:
> >
> > This driver implements the Greybus LED protocol.
>
> I guess this should be copied to LED mainainters?
It will when it comes out of the staging directory.
> Havingstandartized phone hardware to work would be great. Is it
> possible to get device using this?
Yes, Motorolla ships a phone with this hardware in it.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [patch 12/32] greybus: es2 host driver
2016-09-16 13:25 ` [patch 12/32] greybus: es2 host driver Greg KH
@ 2016-10-07 13:43 ` Pavel Machek
0 siblings, 0 replies; 84+ messages in thread
From: Pavel Machek @ 2016-10-07 13:43 UTC (permalink / raw)
To: Greg KH
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
[-- Attachment #1: Type: text/plain, Size: 787 bytes --]
On Fri 2016-09-16 15:25:25, Greg KH wrote:
> This is a Greybus host driver for the ES2/3 bridge chip from Toshiba.
>
> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> ---
> +/*
> + * @endpoint: bulk in endpoint for CPort data
> + * @urb: array of urbs for the CPort in messages
> + * @buffer: array of buffers for the @cport_in_urb urbs
> + */
> +struct es2_cport_in {
> + __u8 endpoint;
> + struct urb *urb[NUM_CPORT_IN_URB];
> + u8 *buffer[NUM_CPORT_IN_URB];
Underscored versions of __u8 and friends are unneccessary here... as
shown later in the struct. Makes sense to fix in whole file.
pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [patch 17/32] greybus: power supply driver
2016-09-16 14:10 ` [patch 17/32] greybus: power supply driver Greg KH
@ 2016-10-07 13:49 ` Pavel Machek
2016-10-07 14:12 ` Greg KH
0 siblings, 1 reply; 84+ messages in thread
From: Pavel Machek @ 2016-10-07 13:49 UTC (permalink / raw)
To: Greg KH, sre, dbaryshkov, dwmw2, linux-pm
Cc: Arnd Bergmann, linux-kernel, Johan Hovold, Rui Miguel Silva,
Laurent Pinchart, Sandeep Patil, Matt Porter, John Stultz,
Rob Herring, Viresh Kumar, Alex Elder, David Lin,
Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
[-- Attachment #1: Type: text/plain, Size: 758 bytes --]
On Fri 2016-09-16 16:10:58, Greg KH wrote:
>
> This driver implements the Greybus power supply class protocol.
Probably
POWER SUPPLY CLASS/SUBSYSTEM and DRIVERS
M: Sebastian Reichel <sre@kernel.org>
M: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
M: David Woodhouse <dwmw2@infradead.org>
L: linux-pm@vger.kernel.org
should be cc-ed?
> +static int is_cache_valid(struct gb_power_supply *gbpsy)
> +{
...
> + if (gbpsy->last_update &&
> + time_is_after_jiffies(gbpsy->last_update +
> + msecs_to_jiffies(cache_time)))
> + return 1;
Are you sure?
Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [patch 17/32] greybus: power supply driver
2016-10-07 13:49 ` Pavel Machek
@ 2016-10-07 14:12 ` Greg KH
2016-10-07 18:15 ` Pavel Machek
0 siblings, 1 reply; 84+ messages in thread
From: Greg KH @ 2016-10-07 14:12 UTC (permalink / raw)
To: Pavel Machek
Cc: sre, dbaryshkov, dwmw2, linux-pm, Arnd Bergmann, linux-kernel,
Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
On Fri, Oct 07, 2016 at 03:49:02PM +0200, Pavel Machek wrote:
> On Fri 2016-09-16 16:10:58, Greg KH wrote:
> >
> > This driver implements the Greybus power supply class protocol.
>
> Probably
>
> POWER SUPPLY CLASS/SUBSYSTEM and DRIVERS
> M: Sebastian Reichel <sre@kernel.org>
> M: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
> M: David Woodhouse <dwmw2@infradead.org>
> L: linux-pm@vger.kernel.org
>
> should be cc-ed?
You didn't read the 00 email, or anything else in this thread, did you?
thanks,
greg k-h
^ permalink raw reply [flat|nested] 84+ messages in thread
* Re: [patch 17/32] greybus: power supply driver
2016-10-07 14:12 ` Greg KH
@ 2016-10-07 18:15 ` Pavel Machek
0 siblings, 0 replies; 84+ messages in thread
From: Pavel Machek @ 2016-10-07 18:15 UTC (permalink / raw)
To: Greg KH
Cc: sre, dbaryshkov, dwmw2, linux-pm, Arnd Bergmann, linux-kernel,
Johan Hovold, Rui Miguel Silva, Laurent Pinchart, Sandeep Patil,
Matt Porter, John Stultz, Rob Herring, Viresh Kumar, Alex Elder,
David Lin, Bryan O'Donoghue, Vaibhav Agarwal, Mark Greer
[-- Attachment #1: Type: text/plain, Size: 1136 bytes --]
On Fri 2016-10-07 16:12:58, Greg KH wrote:
> On Fri, Oct 07, 2016 at 03:49:02PM +0200, Pavel Machek wrote:
> > On Fri 2016-09-16 16:10:58, Greg KH wrote:
> > >
> > > This driver implements the Greybus power supply class protocol.
> >
> > Probably
> >
> > POWER SUPPLY CLASS/SUBSYSTEM and DRIVERS
> > M: Sebastian Reichel <sre@kernel.org>
> > M: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
> > M: David Woodhouse <dwmw2@infradead.org>
> > L: linux-pm@vger.kernel.org
> >
> > should be cc-ed?
>
> You didn't read the 00 email, or anything else in this thread, did you?
You mean this one?
Pavel
Subject: [patch 00/32] Greybus driver subsystem
User-Agent: Mutt/1.7.0 (2016-08-17)
This is a simple set of patches, adding the greybus core and drivers
to
the kernel tree.
It breaks the larger git tree submission up into individual files to
make it easier for reviewing.
Any questions or comments, please let us know.
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]
^ permalink raw reply [flat|nested] 84+ messages in thread
end of thread, other threads:[~2016-10-07 18:15 UTC | newest]
Thread overview: 84+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-09-14 10:09 [GIT PULL] Greybus driver subsystem for 4.9-rc1 Greg KH
2016-09-14 17:36 ` Mark Rutland
2016-09-14 17:36 ` Mark Rutland
2016-09-14 18:07 ` Greg KH
2016-09-14 18:07 ` Greg KH
2016-09-14 18:29 ` Greg KH
2016-09-14 18:29 ` Greg KH
2016-09-14 19:05 ` Joe Perches
2016-09-14 19:05 ` Joe Perches
2016-09-15 9:35 ` Bryan O'Donoghue
2016-09-15 9:35 ` Bryan O'Donoghue
2016-09-15 10:13 ` Mark Rutland
2016-09-15 10:13 ` Mark Rutland
2016-09-15 10:35 ` Bryan O'Donoghue
2016-09-15 10:35 ` Bryan O'Donoghue
2016-09-15 10:47 ` Bryan O'Donoghue
2016-09-15 10:47 ` Bryan O'Donoghue
2016-09-15 11:20 ` Mark Rutland
2016-09-15 11:20 ` Mark Rutland
2016-09-15 11:48 ` Bryan O'Donoghue
2016-09-15 11:48 ` Bryan O'Donoghue
2016-09-15 12:46 ` Mark Rutland
2016-09-15 12:46 ` Mark Rutland
2016-09-15 15:40 ` Bryan O'Donoghue
2016-09-15 15:40 ` Bryan O'Donoghue
2016-09-15 15:47 ` Mark Rutland
2016-09-15 15:47 ` Mark Rutland
2016-09-15 16:09 ` Bryan O'Donoghue
2016-09-15 16:09 ` Bryan O'Donoghue
2016-09-14 20:07 ` Rob Herring
2016-09-14 20:07 ` Rob Herring
2016-09-15 10:17 ` Greg KH
2016-09-15 10:17 ` Greg KH
2016-09-15 11:02 ` Bryan O'Donoghue
2016-09-15 11:02 ` Bryan O'Donoghue
[not found] ` <20160915122141.650632149@bubbles.kroah.org>
[not found] ` <20160915122234.640367870@bubbles.kroah.org>
2016-09-15 13:16 ` [patch 11/32] greybus: camera driver Laurent Pinchart
2016-09-15 14:45 ` [GIT PULL] Greybus driver subsystem for 4.9-rc1 Mark Brown
2016-09-16 6:05 ` Greg KH
2016-09-16 10:18 ` Mark Brown
2016-09-16 13:22 ` Greg KH
2016-09-16 14:24 ` Greg KH
2016-09-20 6:41 ` Greg KH
2016-09-20 7:12 ` Vaibhav Agarwal
2016-09-16 12:18 ` Arnd Bergmann
2016-09-21 13:02 ` Mark Rutland
2016-09-21 14:13 ` Greg KH
2016-09-16 6:40 ` [patch 00/32] Greybus driver subsystem Greg KH
2016-09-16 6:41 ` [patch 02/32] greybus: interface control logic Greg KH
2016-09-16 13:22 ` [patch 03/32] greybus: operations logic Greg KH
2016-09-16 13:23 ` [patch 04/32] greybus: host driver framework Greg KH
2016-09-16 13:23 ` [patch 05/32] greybus: trace.h Greg KH
2016-09-16 13:23 ` [patch 06/32] greybus: svc driver/watchdog Greg KH
2016-09-16 13:23 ` [patch 07/32] greybus: core code Greg KH
2016-09-16 13:24 ` [patch 08/32] greybus: bootrom driver Greg KH
2016-09-16 13:24 ` [patch 09/32] greybus: firmware download class driver Greg KH
2016-09-16 13:24 ` [patch 10/32] greybus: audio driver Greg KH
2016-09-16 13:25 ` [patch 11/32] greybus: camera driver Greg KH
2016-09-16 13:25 ` [patch 12/32] greybus: es2 host driver Greg KH
2016-10-07 13:43 ` Pavel Machek
2016-09-16 14:09 ` [patch 13/32] greybus: HID driver Greg KH
2016-09-16 14:10 ` [patch 14/32] greybus: LED driver Greg KH
2016-10-07 13:36 ` Pavel Machek
2016-10-07 13:41 ` Greg KH
2016-09-16 14:10 ` [patch 15/32] greybus: logging driver Greg KH
2016-09-16 14:10 ` [patch 16/32] greybus: loopback driver Greg KH
2016-09-16 14:10 ` [patch 17/32] greybus: power supply driver Greg KH
2016-10-07 13:49 ` Pavel Machek
2016-10-07 14:12 ` Greg KH
2016-10-07 18:15 ` Pavel Machek
2016-09-16 14:11 ` [patch 18/32] greybus: raw driver Greg KH
2016-09-16 14:11 ` [patch 19/32] greybus: timesync driver Greg KH
2016-09-16 14:11 ` [patch 20/32] greybus: vibrator driver Greg KH
2016-09-16 14:19 ` [patch 21/32] greybus: arche platform driver Greg KH
2016-09-16 14:20 ` [patch 22/32] greybus: bridged phy bus code Greg KH
2016-09-16 14:20 ` [patch 23/32] greybus: bridged phy gpio driver Greg KH
2016-09-16 14:20 ` [patch 24/32] greybus: bridged phy i2c driver Greg KH
2016-09-16 14:20 ` [patch 25/32] greybus: bridged phy pwm driver Greg KH
2016-09-16 14:21 ` [patch 26/32] greybus: bridged phy sdio driver Greg KH
2016-09-16 14:21 ` [patch 27/32] greybus: bridged phy spi driver Greg KH
2016-09-16 14:21 ` [patch 28/32] greybus: bridged phy uart driver Greg KH
2016-09-16 14:21 ` [patch 29/32] greybus: bridged phy usb driver Greg KH
2016-09-16 14:22 ` [patch 30/32] greybus: tools Greg KH
2016-09-16 14:22 ` [patch 31/32] greybus: documentation Greg KH
2016-09-16 14:22 ` [patch 32/32] greybus: add to the build Greg KH
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.