[GIT PATCH] SCSI updates for 2.6.25

[GIT PATCH] SCSI updates for 2.6.25

[James Bottomley]

This is the first piece of our merge for the merge window.  I have a bit
of other stuff with dependent patches, and I'm sure there will be
something else crawling out of the woodwork.

It's the usual grab bag of driver updates and mid-layer changes.
There's no new stand out features in this one.

The patch is available here:

master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6.git

The short change log is:

Adrian Bunk (5):
      qla1280: remove version check
      sun3_scsi_vme: add MODULE_LICENSE
      st: rename flush_write_buffer()
      aic94xx: cleanups
      make scsi_end_bidi_request() static

Al Viro (1):
      libsas: fix endianness bug in sas_ata

Andi Kleen (1):
      Remove random noop unchecked_isa_dma users

Andrew Vasquez (15):
      qla2xxx: Update version number to 8.02.01-k1.
      qla2xxx: Remove unused member (dma_handle) from srb_t structure.
      qla2xxx: Add midlayer target/device reset support.
      qla2xxx: Check alternate 'reason' code during GPSC status handling.
      qla2xxx: Add Flash Descriptor Table layout support.
      qla2xxx: Use PCI-SIG nomenclature for PCIe bandwidth units.
      qla2xxx: Cruft cleanup of functions and structures.
      qla2xxx: Add hardware trace-logging support.
      qla2xxx: Add FC-transport Asynchronous Event Notification support.
      qla2xxx: Simplify interrupt handler locking.
      qla2xxx: Use an rport's scsi_target_id member consistently throughout driv
      qla2xxx: Remove unused and obsolete #define's.
      qla2xxx: Add support for host supported speeds FC transport attribute.
      qla2xxx: Update copyright banner.
      qla2xxx: Update firmware filenames for new ISP parts.

Boaz Harrosh (4):
      iscsi: bidi support for iscsi_tcp
      iscsi: bidi support at the generic libiscsi level
      iscsi: extended cdb support
      gdth: remove command accessors

Christof Schmitt (5):
      zfcp: Fix error handling for blocked unit for send FCP command
      zfcp: Remove zfcp_erp_wait from slave destory handler to fix deadlock
      zfcp: Move DBF definitions to private header file
      zfcp: Fix handling for boxed port after physical close
      zfcp: convert zfcp to use target reset and device reset handler

Darrick J. Wong (2):
      aic94xx: Use sas_request_addr() to provide SAS WWN if the adapter lacks on
      libsas: Provide a transport-level facility to request SAS addrs

David C Somayajulu (1):
      qla4xxx: fix scsi command completion, lun reset and target reset code

Denis Cheng (2):
      hosts.c: removed one unneeded shost->active_mode assignment
      3w-9xxx, 3w-xxxx: memset not needed in probe

FUJITA Tomonori (40):
      bsg: no need to set BSG_F_BLOCK bit in bsg_complete_all_commands
      bsg: remove minor in struct bsg_device
      bsg: use better helper list functions
      bsg: replace kobject_get with blk_get_queue
      bsg: takes a ref to struct device in fops->open
      tgt: use KMEM_CACHE macro
      scsi_debug: remove unnecessary function declarations
      scsi_debug: support large non-fake virtual disk
      scsi_debug: remove the duplicated code in resp_read and resp_write
      scsi_debug: sweep up sdebug_capacity calculation
      scsi_debug: remove unnecessary sdebug_store_size
      scsi_debug: fix lba and data length calculation bugs
      ps3rom: use scsi_build_sense_buffer
      stex: use scsi_build_sense_buffer
      libata: use scsi_build_sense_buffer
      scsi_debug: use scsi_build_sense_buffer
      add scsi_build_sense_buffer helper function
      scsi_debug: remove unnecessary function declarations
      scsi_debug: use list_for_each_entry_safe
      scsi_debug: remove unnecessary condition test in devInfoReg
      scsi_debug: create new scsi_debug devices at a single place
      scsi_debug: remove temporary hack around sscanf for negative values
      3w-9xxx: use sg buffer copy helper functions
      3w-xxxx: use sg buffer copy helper functions
      stex: use sg buffer copy helper functions
      aacraid: use sg buffer copy helper functions
      ips: use sg buffer copy helper funcitons
      simscsi: use sg buffer copy helper funcitons
      ps3rom: use sg buffer copy helper funcitons
      scsi_debug: use sg buffer copy helper functions
      scsi: add wrapper functions for sg buffer copy helper functions
      block: add sg buffer copy helper functions
      aic79xx: fix IOMMU mapping failure handling
      aic7xxx: fix IOMMU mapping failure handling
      scsi_debug: use shost_priv macro
      scsi_debug: remove unnecessary checking
      scsi_debug: remove scsi_debug.h
      scsi_debug: stop including drivers/scsi/scsi.h
      aacraid: READ_CAPACITY_16 shouldn't trust allocation length in cdb
      ips: sg chaining support to the path to non I/O commands

Geert Uytterhoeven (1):
      ps3rom: Simplify fill_from_dev_buffer()

Grant Grundler (1):
      initio: fix big endian problems for auto request sense

Harihara Kadayam (1):
      qla2xxx: Add ISP84XX support.

Harvey Harrison (1):
      ch: fix sparse shadowed variable warnings

James Bottomley (9):
      ips: remove spurious cpu_to_leX on outX statements
      libsas: fix missing inlines in header file
      transport_class: BUG if we can't release the attribute container
      fix barrier failure issue
      hptiop: fix header.context usage
      wd33c93: fix up cut and paste error
      mpt fusion: fix up msi_enable in mpt_suspend
      export command allocation and freeing functions independently of the host
      consolidate command allocation in a single place

James Smart (4):
      lpfc 8.2.6 : Update lpfc driver version to 8.2.6
      lpfc 8.2.6 : Miscellaneous Fixes
      lpfc 8.2.6 : PCI Parity and EEH handling fixes
      lpfc 8.2.6 : Multiple discovery fixes

Jeff Garzik (2):
      gdth: convert to PCI hotplug API
      gdth: PCI probe cleanups, prep for PCI hotplug API conversion

Kai Makisara (2):
      st: show options currently set in sysfs
      st: add option to use SILI in variable block reads

Marcin Slusarz (1):
      aacraid, ips: leX_add_cpu conversion

Mark Salyzyn (1):
      aacraid: Fix down_interruptible() to check the return value

Martin Peschke (21):
      zfcp: fix 31 bit compile warnings
      zfcp: fix compiler warning caused by poking inside new semaphore (linux-ne
      zfcp: Add docbook comments to debug trace.
      zfcp: Cleanup line breaks in debug trace.
      zfcp: Shorten excessive names in debug trace.
      zfcp: Simplify zfcp_dbf_tag and related functions in debug trace.
      zfcp: Simplify usage of hex dump output function for debug trace.
      zfcp: Remove obsolete output function from debug trace.
      zfcp: Cleanup debug trace view functions.
      zfcp: simplify zfcp_dbf_timestamp()
      zfcp: Remove obsolete erp_dbf trace
      zfcp: Add trace records for recovery actions.
      zfcp: Trace all triggers of error recovery activity
      zfcp: Add traces for state changes.
      zfcp: Add trace records for recovery thread and its queues
      zfcp: Register new recovery trace.
      zfcp: Introduce printf helper functions for debug trace.
      zfcp: Add qtcb dump to hba debug trace
      zfcp: Remove qtcb dump to kernel log
      zfcp: Clean up _zfcp_san_dbf_event_common_els
      zfcp: Introduce a helper function that dumps hex data to a zfcp trace.

Matthew Wilcox (1):
      BusLogic: make FlashPoint support x86-32 only

Matthias Kaehlcke (1):
      mpt fusion: convert inactive_list_mutex to a mutex

Michael Reed (1):
      mptsas: do not use ioc->handle to locate hba portinfo structure

Mike Christie (2):
      qla4xxx: Add target reset functionality
      scsi_error: add target reset handler

Paul Bolle (1):
      aacraid: Do not describe check_reset parameter with its value

Prakash, Sathya (1):
      mpt fusion: Enable MSI by default for SAS controllers

Ravi Anand (1):
      qla2xxx: Assign mailbox command timeout values in a consistent manner.

Roel Kluin (1):
      aic7xxx: Test opcode, not definition in aicasm:type_check()

Seokmann Ju (4):
      qla2xxx: Correct vport configuration-change handling.
      qla2xxx: Use proper HA during asynchrounous event handling.
      qla2xxx: Check DFLG_NO_CABLE only on physical port.
      qla2xxx: Consistently access the physical HA port.

Thomas Bogendoerfer (1):
      WD33C93: let platform stub override no_sync/fast/dma_mode

bo yang (3):
      megaraid_sas: Add the new controller(1078DE) support to the driver
      megaraid_sas: Fix the frame count calculation
      megaraid_sas: rollback the sense info implementation

and the diffstat:

 Documentation/scsi/st.txt                 |   12 
 arch/ia64/hp/sim/simscsi.c                |   23 
 block/bsg.c                               |   52 -
 drivers/ata/libata-scsi.c                 |    6 
 drivers/base/transport_class.c            |    3 
 drivers/message/fusion/mptbase.c          |   29 
 drivers/message/fusion/mptbase.h          |    5 
 drivers/message/fusion/mptsas.c           |   22 
 drivers/message/fusion/mptscsih.c         |    8 
 drivers/s390/scsi/zfcp_aux.c              |   33 
 drivers/s390/scsi/zfcp_ccw.c              |   24 
 drivers/s390/scsi/zfcp_dbf.c              | 1281 ++++++++++++++++++------------
 drivers/s390/scsi/zfcp_dbf.h              |  228 +++++
 drivers/s390/scsi/zfcp_def.h              |  169 ---
 drivers/s390/scsi/zfcp_erp.c              |  688 +++++-----------
 drivers/s390/scsi/zfcp_ext.h              |   59 -
 drivers/s390/scsi/zfcp_fsf.c              |  397 ++-------
 drivers/s390/scsi/zfcp_qdio.c             |    7 
 drivers/s390/scsi/zfcp_scsi.c             |   69 -
 drivers/s390/scsi/zfcp_sysfs_adapter.c    |   11 
 drivers/s390/scsi/zfcp_sysfs_port.c       |    9 
 drivers/s390/scsi/zfcp_sysfs_unit.c       |    5 
 drivers/scsi/3w-9xxx.c                    |   23 
 drivers/scsi/3w-xxxx.c                    |   14 
 drivers/scsi/BusLogic.c                   |    5 
 drivers/scsi/BusLogic.h                   |   21 
 drivers/scsi/FlashPoint.c                 |    6 
 drivers/scsi/Kconfig                      |   14 
 drivers/scsi/a2091.c                      |    3 
 drivers/scsi/a3000.c                      |    3 
 drivers/scsi/aacraid/aachba.c             |   69 -
 drivers/scsi/aacraid/commsup.c            |   10 
 drivers/scsi/aic7xxx/aic79xx_osm.c        |    7 
 drivers/scsi/aic7xxx/aic7xxx_osm.c        |   10 
 drivers/scsi/aic7xxx/aicasm/aicasm_gram.y |    2 
 drivers/scsi/aic94xx/aic94xx.h            |   16 
 drivers/scsi/aic94xx/aic94xx_dev.c        |    8 
 drivers/scsi/aic94xx/aic94xx_dump.c       |   10 
 drivers/scsi/aic94xx/aic94xx_dump.h       |    9 
 drivers/scsi/aic94xx/aic94xx_hwi.c        |   44 -
 drivers/scsi/aic94xx/aic94xx_hwi.h        |    2 
 drivers/scsi/aic94xx/aic94xx_init.c       |   10 
 drivers/scsi/aic94xx/aic94xx_reg.c        |   53 -
 drivers/scsi/aic94xx/aic94xx_scb.c        |   33 
 drivers/scsi/aic94xx/aic94xx_sds.c        |    4 
 drivers/scsi/aic94xx/aic94xx_seq.c        |   31 
 drivers/scsi/aic94xx/aic94xx_seq.h        |    4 
 drivers/scsi/aic94xx/aic94xx_task.c       |   12 
 drivers/scsi/aic94xx/aic94xx_tmf.c        |    2 
 drivers/scsi/arm/acornscsi.c              |    1 
 drivers/scsi/arm/cumana_1.c               |    1 
 drivers/scsi/ch.c                         |   33 
 drivers/scsi/dc395x.c                     |    1 
 drivers/scsi/eata_pio.c                   |    2 
 drivers/scsi/gdth.c                       |  320 +++----
 drivers/scsi/gdth.h                       |    2 
 drivers/scsi/gvp11.c                      |    3 
 drivers/scsi/hosts.c                      |    1 
 drivers/scsi/hptiop.c                     |    7 
 drivers/scsi/initio.c                     |    9 
 drivers/scsi/ips.c                        |   89 --
 drivers/scsi/iscsi_tcp.c                  |   31 
 drivers/scsi/libiscsi.c                   |  140 ++-
 drivers/scsi/libsas/sas_ata.c             |    2 
 drivers/scsi/libsas/sas_scsi_host.c       |   41 
 drivers/scsi/lpfc/lpfc.h                  |    5 
 drivers/scsi/lpfc/lpfc_attr.c             |   10 
 drivers/scsi/lpfc/lpfc_ct.c               |   48 -
 drivers/scsi/lpfc/lpfc_debugfs.c          |    2 
 drivers/scsi/lpfc/lpfc_els.c              |  121 +-
 drivers/scsi/lpfc/lpfc_hbadisc.c          |   73 -
 drivers/scsi/lpfc/lpfc_init.c             |   98 +-
 drivers/scsi/lpfc/lpfc_nportdisc.c        |   40 
 drivers/scsi/lpfc/lpfc_scsi.c             |   34 
 drivers/scsi/lpfc/lpfc_sli.c              |  112 +-
 drivers/scsi/lpfc/lpfc_version.h          |    2 
 drivers/scsi/lpfc/lpfc_vport.c            |    3 
 drivers/scsi/mac_scsi.c                   |    1 
 drivers/scsi/megaraid/megaraid_sas.c      |   49 -
 drivers/scsi/megaraid/megaraid_sas.h      |    5 
 drivers/scsi/mvme147.c                    |    3 
 drivers/scsi/ps3rom.c                     |  101 --
 drivers/scsi/qla1280.c                    |    5 
 drivers/scsi/qla2xxx/Kconfig              |    3 
 drivers/scsi/qla2xxx/qla_attr.c           |   36 
 drivers/scsi/qla2xxx/qla_dbg.c            |  124 --
 drivers/scsi/qla2xxx/qla_dbg.h            |   23 
 drivers/scsi/qla2xxx/qla_def.h            |   78 +
 drivers/scsi/qla2xxx/qla_dfs.c            |    2 
 drivers/scsi/qla2xxx/qla_fw.h             |  173 +++-
 drivers/scsi/qla2xxx/qla_gbl.h            |   33 
 drivers/scsi/qla2xxx/qla_gs.c             |   16 
 drivers/scsi/qla2xxx/qla_init.c           |  192 +++-
 drivers/scsi/qla2xxx/qla_inline.h         |   87 --
 drivers/scsi/qla2xxx/qla_iocb.c           |    5 
 drivers/scsi/qla2xxx/qla_isr.c            |  223 ++---
 drivers/scsi/qla2xxx/qla_mbx.c            |  318 +++++--
 drivers/scsi/qla2xxx/qla_mid.c            |   30 
 drivers/scsi/qla2xxx/qla_os.c             |  443 +++++-----
 drivers/scsi/qla2xxx/qla_settings.h       |   16 
 drivers/scsi/qla2xxx/qla_sup.c            |  315 +++++--
 drivers/scsi/qla2xxx/qla_version.h        |    6 
 drivers/scsi/qla4xxx/ql4_fw.h             |    4 
 drivers/scsi/qla4xxx/ql4_glbl.h           |    4 
 drivers/scsi/qla4xxx/ql4_iocb.c           |   14 
 drivers/scsi/qla4xxx/ql4_isr.c            |   40 
 drivers/scsi/qla4xxx/ql4_mbx.c            |   39 
 drivers/scsi/qla4xxx/ql4_os.c             |   96 +-
 drivers/scsi/raid_class.c                 |    2 
 drivers/scsi/scsi.c                       |  229 +++--
 drivers/scsi/scsi_debug.c                 | 1219 ++++++++++++----------------
 drivers/scsi/scsi_debug.h                 |   24 
 drivers/scsi/scsi_error.c                 |  150 +++
 drivers/scsi/scsi_lib.c                   |   12 
 drivers/scsi/scsi_tgt_lib.c               |    4 
 drivers/scsi/sgiwd93.c                    |    7 
 drivers/scsi/st.c                         |   89 +-
 drivers/scsi/st.h                         |    3 
 drivers/scsi/st_options.h                 |    6 
 drivers/scsi/stex.c                       |   83 -
 drivers/scsi/sun3_scsi_vme.c              |    1 
 drivers/scsi/wd33c93.c                    |    3 
 include/linux/attribute_container.h       |    2 
 include/linux/mtio.h                      |    1 
 include/linux/scatterlist.h               |    5 
 include/linux/transport_class.h           |    5 
 include/scsi/iscsi_proto.h                |    6 
 include/scsi/libsas.h                     |    2 
 include/scsi/sas_ata.h                    |    4 
 include/scsi/scsi_cmnd.h                  |   17 
 include/scsi/scsi_eh.h                    |    5 
 include/scsi/scsi_host.h                  |    1 
 lib/scatterlist.c                         |  102 ++
 133 files changed, 4873 insertions(+), 4389 deletions(-)

James

Re: [GIT PATCH] SCSI updates for 2.6.25

[Ingo Molnar]

* James Bottomley <James.Bottomley@HansenPartnership.com> wrote:

>       libsas: Provide a transport-level facility to request SAS addrs

this (or a nearby) commit caused a build regression:

 drivers/built-in.o: In function `sas_request_addr':
 : undefined reference to `request_firmware'
 drivers/built-in.o: In function `sas_request_addr':
 : undefined reference to `release_firmware'

config can be found at:

   http://redhat.com/~mingo/misc/config-Sat_Apr_19_16_58_35_CEST_2008.bad

... brought to you by x86.git's randconfig build and boot service ;-)

	Ingo

Re: [GIT PATCH] SCSI updates for 2.6.25

[James Bottomley]

On Sat, 2008-04-19 at 17:19 +0200, Ingo Molnar wrote:
> * James Bottomley <James.Bottomley@HansenPartnership.com> wrote:
> 
> >       libsas: Provide a transport-level facility to request SAS addrs
> 
> this (or a nearby) commit caused a build regression:
> 
>  drivers/built-in.o: In function `sas_request_addr':
>  : undefined reference to `request_firmware'
>  drivers/built-in.o: In function `sas_request_addr':
>  : undefined reference to `release_firmware'
> 
> config can be found at:
> 
>    http://redhat.com/~mingo/misc/config-Sat_Apr_19_16_58_35_CEST_2008.bad
> 
> ... brought to you by x86.git's randconfig build and boot service ;-)

This one's fun.  The root cause is 

CONFIG_SCSI_SAS_LIBSAS=y
CONFIG_FW_LOADER=m

The problem is that libsas doesn't depend on the FW loader and doesn't
want to.  It just wants to use it if it's available.  The definitions in
include/linux/firmware.h have stubs to facilitate this.

However, CONFIG_FW_LOADER=m defeats the stubs.

This is a bit nasty to fix; however, I think this patch does.  I've also
put a large comment in to explain what's going on.

Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>

---

diff --git a/include/linux/firmware.h b/include/linux/firmware.h
index 4d10c73..01e39a1 100644
--- a/include/linux/firmware.h
+++ b/include/linux/firmware.h
@@ -13,7 +13,16 @@ struct firmware {
 
 struct device;
 
-#if defined(CONFIG_FW_LOADER) || defined(CONFIG_FW_LOADER_MODULE)
+
+/*
+ * This is very subtle.  If the Firmware loader is built in then the
+ * request/release calls can be accessed by anything.  If it's built
+ * as a module then only other modules can access it.  The check on
+ * MODULE specifically fixes the case where the firmware loader is
+ * built as a module, but a built in kernel component tries to use the
+ * request/release functions.
+ */
+#if defined(CONFIG_FW_LOADER) || (defined(CONFIG_FW_LOADER_MODULE) && defined(MODULE))
 int request_firmware(const struct firmware **fw, const char *name,
 		     struct device *device);
 int request_firmware_nowait(

Re: [GIT PATCH] SCSI updates for 2.6.25

[Ingo Molnar]

* James Bottomley <James.Bottomley@HansenPartnership.com> wrote:

> >  drivers/built-in.o: In function `sas_request_addr':
> >  : undefined reference to `release_firmware'
> > 
> > config can be found at:
> > 
> >    http://redhat.com/~mingo/misc/config-Sat_Apr_19_16_58_35_CEST_2008.bad
> > 
> > ... brought to you by x86.git's randconfig build and boot service ;-)
> 
> This one's fun.  The root cause is 
> 
> CONFIG_SCSI_SAS_LIBSAS=y
> CONFIG_FW_LOADER=m
> 
> The problem is that libsas doesn't depend on the FW loader and doesn't 
> want to.  It just wants to use it if it's available.  The definitions 
> in include/linux/firmware.h have stubs to facilitate this.
> 
> However, CONFIG_FW_LOADER=m defeats the stubs.
> 
> This is a bit nasty to fix; however, I think this patch does.  I've 
> also put a large comment in to explain what's going on.
> 
> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>

thanks James - your fix works fine here.

Tested-by: Ingo Molnar <mingo@elte.hu>

	Ingo

Re: [GIT PATCH] SCSI updates for 2.6.25

[Adrian Bunk]

On Sat, Apr 19, 2008 at 10:54:44AM -0500, James Bottomley wrote:
> 
> On Sat, 2008-04-19 at 17:19 +0200, Ingo Molnar wrote:
> > * James Bottomley <James.Bottomley@HansenPartnership.com> wrote:
> > 
> > >       libsas: Provide a transport-level facility to request SAS addrs
> > 
> > this (or a nearby) commit caused a build regression:
> > 
> >  drivers/built-in.o: In function `sas_request_addr':
> >  : undefined reference to `request_firmware'
> >  drivers/built-in.o: In function `sas_request_addr':
> >  : undefined reference to `release_firmware'
> > 
> > config can be found at:
> > 
> >    http://redhat.com/~mingo/misc/config-Sat_Apr_19_16_58_35_CEST_2008.bad
> > 
> > ... brought to you by x86.git's randconfig build and boot service ;-)
> 
> This one's fun.  The root cause is 
> 
> CONFIG_SCSI_SAS_LIBSAS=y
> CONFIG_FW_LOADER=m
> 
> The problem is that libsas doesn't depend on the FW loader and doesn't
> want to.  It just wants to use it if it's available.  The definitions in
> include/linux/firmware.h have stubs to facilitate this.
> 
> However, CONFIG_FW_LOADER=m defeats the stubs.
> 
> This is a bit nasty to fix; however, I think this patch does.  I've also
> put a large comment in to explain what's going on.
>...

Your patch fixes the build error, but I'm not sure whether it's the
correct fix:

CONFIG_SCSI_SAS_LIBSAS=y, CONFIG_FW_LOADER=m, CONFIG_SCSI_AIC94XX=m
will now pass all randconfig tests, but the effects of the
"select FW_LOADER" in the SCSI_AIC94XX driver would be lost in this 
configuration, and we might get bug reports like "driver works built-in 
but fails with 'No SAS Address provided for %s\n' when built modular."

IMHO SCSI_SAS_LIBSAS should simply select FW_LOADER.

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

Re: [GIT PATCH] SCSI updates for 2.6.25

[James Bottomley]

On Sat, 2008-04-19 at 20:05 +0300, Adrian Bunk wrote:
> On Sat, Apr 19, 2008 at 10:54:44AM -0500, James Bottomley wrote:
> > 
> > On Sat, 2008-04-19 at 17:19 +0200, Ingo Molnar wrote:
> > > * James Bottomley <James.Bottomley@HansenPartnership.com> wrote:
> > > 
> > > >       libsas: Provide a transport-level facility to request SAS addrs
> > > 
> > > this (or a nearby) commit caused a build regression:
> > > 
> > >  drivers/built-in.o: In function `sas_request_addr':
> > >  : undefined reference to `request_firmware'
> > >  drivers/built-in.o: In function `sas_request_addr':
> > >  : undefined reference to `release_firmware'
> > > 
> > > config can be found at:
> > > 
> > >    http://redhat.com/~mingo/misc/config-Sat_Apr_19_16_58_35_CEST_2008.bad
> > > 
> > > ... brought to you by x86.git's randconfig build and boot service ;-)
> > 
> > This one's fun.  The root cause is 
> > 
> > CONFIG_SCSI_SAS_LIBSAS=y
> > CONFIG_FW_LOADER=m
> > 
> > The problem is that libsas doesn't depend on the FW loader and doesn't
> > want to.  It just wants to use it if it's available.  The definitions in
> > include/linux/firmware.h have stubs to facilitate this.
> > 
> > However, CONFIG_FW_LOADER=m defeats the stubs.
> > 
> > This is a bit nasty to fix; however, I think this patch does.  I've also
> > put a large comment in to explain what's going on.
> >...
> 
> Your patch fixes the build error, but I'm not sure whether it's the
> correct fix:
> 
> CONFIG_SCSI_SAS_LIBSAS=y, CONFIG_FW_LOADER=m, CONFIG_SCSI_AIC94XX=m
> will now pass all randconfig tests, but the effects of the
> "select FW_LOADER" in the SCSI_AIC94XX driver would be lost in this 
> configuration, and we might get bug reports like "driver works built-in 
> but fails with 'No SAS Address provided for %s\n' when built modular."
> 
> IMHO SCSI_SAS_LIBSAS should simply select FW_LOADER.

Well, we already had this argument on linux-scsi when the interface was
first proposed:  Apart from select being very nasty and should only be
sparingly used, this interface is really only provided for specific OS
defined cases.  It's users are expected to know the pitfalls and the
dangers (it's only for boards with no fw in either pre-test or embedded
situations), so the casual user should never have to use it.

Even if I changed it to select, a user could still defeat it by making
CONFIG_SCSI_AIC94XX=y (because now the initialisation is before early
boot userland, so the request always fails).

James

Re: [GIT PATCH] SCSI updates for 2.6.25

[Linus Torvalds]

On Sat, 19 Apr 2008, James Bottomley wrote:
> 
> Well, we already had this argument on linux-scsi when the interface was
> first proposed:  Apart from select being very nasty and should only be
> sparingly used

I'm still not understanding people who say that.

Yes, select has problems, in that it doesn't guarantee dependencies.

But that said, select is a whole lot better than having simply *different* 
behaviour (or build errors) depending on totally unrelated config options.

So people: stop this total *idiocy* with "select is bad". It's not. The 
lack of select is *much* worse.

If you want LIBSAS to have two different modes, how about just making that 
explicit in the configuration, ie using something like

	config LIBSAS_SET_ADDR
		bool "Provide user-settable SAS address" 
		depends on SCSI_SAS_LIBSAS
		help
		  This allows a runtime "firmware" loader that loads the 
		  SAS address

	config LIBSAS_SET_ADDR_FWLOAD
		tristate
		depends on SCSI_SAS_LIBSAS && LIBSAS_SET_ADDR
		default y
		select FW_LOADER

and now you have the explicit option to turn it on or off - and FW_LOADER 
is loaded appropriately.

> Even if I changed it to select, a user could still defeat it by making
> CONFIG_SCSI_AIC94XX=y (because now the initialisation is before early
> boot userland, so the request always fails).

Not true. That's why we have initrd and ramdisks etc.

Not that any normal people will ever care about LIBSAS, but anyway.. The 
ones that do care about these kinds of things are probably happy to have 
special initrd images.

		Linus

Re: [GIT PATCH] SCSI updates for 2.6.25

[Matthew Wilcox]

On Sat, Apr 19, 2008 at 01:14:23PM -0700, Linus Torvalds wrote:
> So people: stop this total *idiocy* with "select is bad". It's not. The 
> lack of select is *much* worse.

Could we consider applying this patch then?

----

'select' not Considered Evil

While select should be used with care, it is not actually evil.

Signed-off-by: Matthew Wilcox <willy@linux.intel.com>

diff --git a/Documentation/kbuild/kconfig-language.txt b/Documentation/kbuild/kconfig-language.txt
index 649cb87..dacfcf2 100644
--- a/Documentation/kbuild/kconfig-language.txt
+++ b/Documentation/kbuild/kconfig-language.txt
@@ -104,9 +104,9 @@ applicable everywhere (see syntax).
   Reverse dependencies can only be used with boolean or tristate
   symbols.
   Note:
-	select is evil.... select will by brute force set a symbol
-	equal to 'y' without visiting the dependencies. So abusing
-	select you are able to select a symbol FOO even if FOO depends
+	select should be used with care.  It can set a symbol
+	without visiting the dependencies.  By abusing select you are
+	able to select a symbol FOO even if FOO depends
 	on BAR that is not set. In general use select only for
 	non-visible symbols (no prompts anywhere) and for symbols with
 	no dependencies. That will limit the usefulness but on the

-- 
Intel are signing my paycheques ... these opinions are still mine
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."

Re: [GIT PATCH] SCSI updates for 2.6.25

[James Bottomley]

On Sat, 2008-04-19 at 13:14 -0700, Linus Torvalds wrote:
> On Sat, 19 Apr 2008, James Bottomley wrote:
> > 
> > Well, we already had this argument on linux-scsi when the interface was
> > first proposed:  Apart from select being very nasty and should only be
> > sparingly used
> 
> I'm still not understanding people who say that.

Well, I can ... even though SCSI is one of the major users of select in
the tree.

The argument goes that abuse of select is very easy to do ... just over
select on a lot of options.  The result is pieces of the kernel get
built in and you have a devil of a time trying to work out how to get
them to be turned off (this is annoying when you get bugs that say
without feature X enabled, Y bug happens).

Conversely depends has the opposite problem: config options disappear
for no readily apparent reason (at least until you read the Kconfig
file, then it becomes apparent).

What I try to do in SCSI is sort our Kconfig features into primary and
subsidiary, so primary features (like HBA drivers) are allowed to select
subsidiary features (like transport classes).  However, for most other
cases, we use depends instead.


> Yes, select has problems, in that it doesn't guarantee dependencies.

Yes, over use of select trips across this fairly quickly.

> But that said, select is a whole lot better than having simply *different* 
> behaviour (or build errors) depending on totally unrelated config options.
> 
> So people: stop this total *idiocy* with "select is bad". It's not. The 
> lack of select is *much* worse.
> 
> If you want LIBSAS to have two different modes, how about just making that 
> explicit in the configuration, ie using something like
> 
> 	config LIBSAS_SET_ADDR
> 		bool "Provide user-settable SAS address" 
> 		depends on SCSI_SAS_LIBSAS
> 		help
> 		  This allows a runtime "firmware" loader that loads the 
> 		  SAS address
> 
> 	config LIBSAS_SET_ADDR_FWLOAD
> 		tristate
> 		depends on SCSI_SAS_LIBSAS && LIBSAS_SET_ADDR
> 		default y
> 		select FW_LOADER

Yes, that works for libsas ... but what do we do about the prototypes in
firmware.h ... just take away the stub and force everything relying on
it to break so we can put in a Kconfig sequence like the above?

> and now you have the explicit option to turn it on or off - and FW_LOADER 
> is loaded appropriately.
> 
> > Even if I changed it to select, a user could still defeat it by making
> > CONFIG_SCSI_AIC94XX=y (because now the initialisation is before early
> > boot userland, so the request always fails).
> 
> Not true. That's why we have initrd and ramdisks etc.

They don't currently seem to work for compiled in drivers ... we have
the same issue with drivers that actually need firmware (like aic94xx or
qla2xxx): no-one seems to have figured out how to provide it for them;
it seems the device is already initialised and has failed to find its
firmware before the initrd/initramfs is running.

> Not that any normal people will ever care about LIBSAS, but anyway.. The 
> ones that do care about these kinds of things are probably happy to have 
> special initrd images.

Right ... that's more or less my argument about how far do we really
need to go to make Kconfig foolproof.

James

Re: [GIT PATCH] SCSI updates for 2.6.25

[Matthew Wilcox]

On Sat, Apr 19, 2008 at 05:27:56PM -0500, James Bottomley wrote:
> > Not true. That's why we have initrd and ramdisks etc.
> 
> They don't currently seem to work for compiled in drivers ... we have
> the same issue with drivers that actually need firmware (like aic94xx or
> qla2xxx): no-one seems to have figured out how to provide it for them;
> it seems the device is already initialised and has failed to find its
> firmware before the initrd/initramfs is running.

WorksForMe ... or it did a few months ago.  I can re-test it with a
modern kernel ... wonder which machine I had that needed it ...

-- 
Intel are signing my paycheques ... these opinions are still mine
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."

Re: [GIT PATCH] SCSI updates for 2.6.25

[Stefan Richter]

Matthew Wilcox wrote:
> --- a/Documentation/kbuild/kconfig-language.txt
> +++ b/Documentation/kbuild/kconfig-language.txt
> @@ -104,9 +104,9 @@ applicable everywhere (see syntax).
>    Reverse dependencies can only be used with boolean or tristate
>    symbols.
>    Note:
> -	select is evil.... select will by brute force set a symbol
> -	equal to 'y' without visiting the dependencies. So abusing
> -	select you are able to select a symbol FOO even if FOO depends
> +	select should be used with care.  It can set a symbol
> +	without visiting the dependencies.  By abusing select you are
> +	able to select a symbol FOO even if FOO depends
>  	on BAR that is not set. In general use select only for
>  	non-visible symbols (no prompts anywhere) and for symbols with
>  	no dependencies. That will limit the usefulness but on the

"It can set a symbol without visiting the dependencies." = imprecise

"select will by brute force set a symbol [...] without visiting the 
dependencies." = clear.  So why not leave this portion of the text as it is.

Maybe change
	"equal to 'y'"
to
	"equal to 'y' or 'm'"
or to
	""
to make it entirely correct.
-- 
Stefan Richter
-=====-==--- -=-- =-=--
http://arcgraph.de/sr/

Re: [GIT PATCH] SCSI updates for 2.6.25

[Matthew Wilcox]

On Sun, Apr 20, 2008 at 04:09:27PM +0200, Stefan Richter wrote:
> Matthew Wilcox wrote:
> >--- a/Documentation/kbuild/kconfig-language.txt
> >+++ b/Documentation/kbuild/kconfig-language.txt
> >   Note:
> >-	select is evil.... select will by brute force set a symbol
> >-	equal to 'y' without visiting the dependencies. So abusing
> >-	select you are able to select a symbol FOO even if FOO depends
> >+	select should be used with care.  It can set a symbol
> >+	without visiting the dependencies.  By abusing select you are
> >+	able to select a symbol FOO even if FOO depends
> > 	on BAR that is not set. In general use select only for
> > 	non-visible symbols (no prompts anywhere) and for symbols with
> > 	no dependencies. That will limit the usefulness but on the
> 
> "It can set a symbol without visiting the dependencies." = imprecise
> 
> "select will by brute force set a symbol [...] without visiting the 
> dependencies." = clear.  So why not leave this portion of the text as it is.

"brute force" is pejorative, not clear.  I'm not really interested in
bikeshedding this issue.

> Maybe change
> 	"equal to 'y'"
> to
> 	"equal to 'y' or 'm'"
> or to
> 	""
> to make it entirely correct.
> -- 
> Stefan Richter
> -=====-==--- -=-- =-=--
> http://arcgraph.de/sr/

-- 
Intel are signing my paycheques ... these opinions are still mine
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."

Re: [GIT PATCH] SCSI updates for 2.6.25

[Stefan Richter]

Matthew Wilcox wrote:
>>> -	select is evil.... select will by brute force set a symbol
...
>>> +	select should be used with care.  It can set a symbol
...
> I'm not really interested in bikeshedding this issue.

At least don't replace "will" by "can".
-- 
Stefan Richter
-=====-==--- -=-- =-=--
http://arcgraph.de/sr/

Re: [GIT PATCH] SCSI updates for 2.6.25

[Ingo Molnar]

* James Bottomley <James.Bottomley@HansenPartnership.com> wrote:

> On Sat, 2008-04-19 at 13:14 -0700, Linus Torvalds wrote:
> > On Sat, 19 Apr 2008, James Bottomley wrote:
> > > 
> > > Well, we already had this argument on linux-scsi when the interface was
> > > first proposed:  Apart from select being very nasty and should only be
> > > sparingly used
> > 
> > I'm still not understanding people who say that.
> 
> Well, I can ... even though SCSI is one of the major users of select 
> in the tree.

you are handling it well then :) I tend to be amongst the first ones to 
notice select trouble (randconfig is quite sensitive to it) and by far 
the leading area of failure "in practice" is video drivers and sound 
modules. Networking is an occasional blip on the radar (but that's 
really just due to its sheer natural size and complexity) and i dont 
remember any deep SCSI trouble there ever.

pie-in-the-sky: a fundamentally simpler Kconfig space would be nice, 
with strict hierarchies and where enabling a lower component 
automatically selects all the upper components as well. The Kconfig 
language should detect loops in hierarchies. (it already does so in a 
number of situations)

i.e. we should only describe what depends on what - and the rest would 
be figured out automatically. In that sense "select" should be a _user 
action_ that would never pop up in the language at all - and the 
"selection" would all trickle through automatically via the 'depends on' 
links without having to mention it in the Kconfig language.

whenever due to a selection that a user made something becomes 
unavailable, the kconfig tool/GUI should warn about that with an extra 
confirmation step. I.e. it should all really be analogous to normal 
user-space package management which has its own dependency resolution 
process.

	Ingo

Re: [GIT PATCH] SCSI updates for 2.6.25

[Sam Ravnborg]

On Sat, Apr 19, 2008 at 02:45:11PM -0600, Matthew Wilcox wrote:
> On Sat, Apr 19, 2008 at 01:14:23PM -0700, Linus Torvalds wrote:
> > So people: stop this total *idiocy* with "select is bad". It's not. The 
> > lack of select is *much* worse.
> 
> Could we consider applying this patch then?

I applied the following.

	Sam

commit dfecbec8b54038ef02835d2f8181e1f44bd080d2
Author: Matthew Wilcox <matthew@wil.cx>
Date:   Sat Apr 19 14:45:11 2008 -0600

    kconifg: 'select' considered less evil
    
    While select should be used with care, it is not actually evil.
    
    Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
    Signed-off-by: Sam Ravnborg <sam@ravnborg.org>

diff --git a/Documentation/kbuild/kconfig-language.txt b/Documentation/kbuild/kconfig-language.txt
index 649cb87..00b950d 100644
--- a/Documentation/kbuild/kconfig-language.txt
+++ b/Documentation/kbuild/kconfig-language.txt
@@ -104,14 +104,15 @@ applicable everywhere (see syntax).
   Reverse dependencies can only be used with boolean or tristate
   symbols.
   Note:
-	select is evil.... select will by brute force set a symbol
-	equal to 'y' without visiting the dependencies. So abusing
-	select you are able to select a symbol FOO even if FOO depends
-	on BAR that is not set. In general use select only for
-	non-visible symbols (no prompts anywhere) and for symbols with
-	no dependencies. That will limit the usefulness but on the
-	other hand avoid the illegal configurations all over. kconfig
-	should one day warn about such things.
+	select should be used with care. select will force
+	a symbol to a value without visiting the dependencies.
+	By abusing select you are able to select a symbol FOO even
+	if FOO depends on BAR that is not set.
+	In general use select only for non-visible symbols
+	(no prompts anywhere) and for symbols with no dependencies.
+	That will limit the usefulness but on the other hand avoid
+	the illegal configurations all over.
+	kconfig should one day warn about such things.
 
 - numerical ranges: "range" <symbol> <symbol> ["if" <expr>]
   This allows to limit the range of possible input values for int

Re: [GIT PATCH] SCSI updates for 2.6.25

[Matthew Wilcox]

On Mon, Apr 28, 2008 at 11:00:34PM +0200, Sam Ravnborg wrote:
> On Sat, Apr 19, 2008 at 02:45:11PM -0600, Matthew Wilcox wrote:
> > On Sat, Apr 19, 2008 at 01:14:23PM -0700, Linus Torvalds wrote:
> > > So people: stop this total *idiocy* with "select is bad". It's not. The 
> > > lack of select is *much* worse.
> > 
> > Could we consider applying this patch then?
> 
> I applied the following.

Reads much better.  Thanks!

-- 
Intel are signing my paycheques ... these opinions are still mine
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."