[PATCH v5 0/8] Introduce on-chip interconnect API

[PATCH v5 0/8] Introduce on-chip interconnect API

[Georgi Djakov]

Modern SoCs have multiple processors and various dedicated cores (video, gpu,
graphics, modem). These cores are talking to each other and can generate a
lot of data flowing through the on-chip interconnects. These interconnect
buses could form different topologies such as crossbar, point to point buses,
hierarchical buses or use the network-on-chip concept.

These buses have been sized usually to handle use cases with high data
throughput but it is not necessary all the time and consume a lot of power.
Furthermore, the priority between masters can vary depending on the running
use case like video playback or CPU intensive tasks.

Having an API to control the requirement of the system in terms of bandwidth
and QoS, so we can adapt the interconnect configuration to match those by
scaling the frequencies, setting link priority and tuning QoS parameters.
This configuration can be a static, one-time operation done at boot for some
platforms or a dynamic set of operations that happen at run-time.

This patchset introduce a new API to get the requirement and configure the
interconnect buses across the entire chipset to fit with the current demand.
The API is NOT for changing the performance of the endpoint devices, but only
the interconnect path in between them.

The API is using a consumer/provider-based model, where the providers are
the interconnect buses and the consumers could be various drivers.
The consumers request interconnect resources (path) to an endpoint and set
the desired constraints on this data flow path. The provider(s) receive
requests from consumers and aggregate these requests for all master-slave
pairs on that path. Then the providers configure each participating in the
topology node according to the requested data flow path, physical links and
constraints. The topology could be complicated and multi-tiered and is SoC
specific.

Below is a simplified diagram of a real-world SoC topology. The interconnect
providers are the NoCs.

+----------------+    +----------------+
| HW Accelerator |--->|      M NoC     |<---------------+
+----------------+    +----------------+                |
                        |      |                    +------------+
 +-----+  +-------------+      V       +------+     |            |
 | DDR |  |                +--------+  | PCIe |     |            |
 +-----+  |                | Slaves |  +------+     |            |
   ^ ^    |                +--------+     |         |   C NoC    |
   | |    V                               V         |            |
+------------------+   +------------------------+   |            |   +-----+
|                  |-->|                        |-->|            |-->| CPU |
|                  |-->|                        |<--|            |   +-----+
|     Mem NoC      |   |         S NoC          |   +------------+
|                  |<--|                        |---------+    |
|                  |<--|                        |<------+ |    |   +--------+
+------------------+   +------------------------+       | |    +-->| Slaves |
  ^  ^    ^    ^          ^                             | |        +--------+
  |  |    |    |          |                             | V
+------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
| CPUs |  |  | GPU |   | DSP |  | Masters |-->|       P NoC    |-->| Slaves |
+------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
          |
      +-------+
      | Modem |
      +-------+

TODO:
 * Create icc_set_extended() to handle parameters such as latency and other
   QoS values.
 * Convert from using global node identifiers to local per provider ids.
 * Cache the path between the nodes instead of walking the graph on each get().
 * Sync interconnect requests with the idle state of the device.

Changes since patchset v4 (https://lkml.org/lkml/2018/3/9/856)
* Simplified locking by using a single global mutex. (Evan)
* Changed the aggregation function interface.
* Implemented functions for node, link, provider removal. (Evan)
* Naming changes on variables and functions, removed redundant code. (Evan)
* Fixes and clarifications in the docs. (Matthias, Evan, Amit, Alexandre)
* Removed mandatory reg DT property, made interconnect-names optional. (Bjorn)
* Made interconnect-cells property required to align with other bindings. (Neil)
* Moved msm8916 specific bindings into a separate file and patch. (Bjorn)
* Use the names, instead of the hardcoded ids for topology. (Matthias)
* Init the node before creating the links. (Evan)
* Added icc_units_to_bps macro. (Amit)

Changes since patchset v3 (https://lkml.org/lkml/2017/9/8/544)
* Refactored the constraints aggregation.
* Use the IDR API.
* Split the provider and consumer bindings into separate patches and propose
  new bindings for consumers, which allows to specify the local source port.
* Adopted the icc_ prefix for API functions.
* Introduced separate API functions for creating interconnect nodes and links.
* Added DT lookup support in addition to platform data.
* Dropped the event tracing patch for now.
* Added a patch to provide summary via debugfs.
* Use macro for the list of topology definitions in the platform driver.
* Various minor changes.

Changes since patchset v2 (https://lkml.org/lkml/2017/7/20/825)
* Split the aggregation into per node and per provider. Cache the
  aggregated values.
* Various small refactorings and cleanups in the framework.
* Added a patch introducing basic tracepoint support for monitoring
  the time required to update the interconnect nodes.

Changes since patchset v1 (https://lkml.org/lkml/2017/6/27/890)
* Updates in the documentation.
* Changes in request aggregation, locking.
* Dropped the aggregate() callback and use the default as it currently
  sufficient for the single vendor driver. Will add it later when needed.
* Dropped the dt-bindings draft patch for now.

Changes since RFC v2 (https://lkml.org/lkml/2017/6/12/316)
* Converted documentation to rst format.
* Fixed an incorrect call to mutex_lock. Renamed max_bw to peak_bw.

Changes since RFC v1 (https://lkml.org/lkml/2017/5/15/605)
* Refactored code into shorter functions.
* Added a new aggregate() API function.
* Rearranged some structs to reduce padding bytes.

Changes since RFC v0 (https://lkml.org/lkml/2017/3/1/599)
* Removed DT support and added optional Patch 3 with new bindings proposal.
* Converted the topology into internal driver data.
* Made the framework modular.
* interconnect_get() now takes (src and dst ports as arguments).
* Removed public declarations of some structs.
* Now passing prev/next nodes to the vendor driver.
* Properly remove requests on _put().
* Added refcounting.
* Updated documentation.
* Changed struct interconnect_path to use array instead of linked list.

Georgi Djakov (8):
  interconnect: Add generic on-chip interconnect API
  dt-bindings: Introduce interconnect provider bindings
  interconnect: Add debugfs support
  interconnect: qcom: Add RPM communication
  dt-bindings: interconnect: Document qcom,msm8916 NoC bindings
  interconnect: qcom: Add msm8916 interconnect provider driver
  dt-bindings: Introduce interconnect consumers bindings
  interconnect: Allow endpoints translation via DT

 .../bindings/interconnect/interconnect.txt    |  60 ++
 .../bindings/interconnect/qcom-msm8916.txt    |  39 +
 .../bindings/interconnect/qcom-smd.txt        |  32 +
 Documentation/interconnect/interconnect.rst   |  96 +++
 drivers/Kconfig                               |   2 +
 drivers/Makefile                              |   1 +
 drivers/interconnect/Kconfig                  |  15 +
 drivers/interconnect/Makefile                 |   3 +
 drivers/interconnect/core.c                   | 726 ++++++++++++++++++
 drivers/interconnect/qcom/Kconfig             |  11 +
 drivers/interconnect/qcom/Makefile            |   4 +
 drivers/interconnect/qcom/msm8916.c           | 495 ++++++++++++
 drivers/interconnect/qcom/smd-rpm.c           |  90 +++
 drivers/interconnect/qcom/smd-rpm.h           |  15 +
 include/dt-bindings/interconnect/qcom.h       | 350 +++++++++
 include/linux/interconnect-provider.h         | 127 +++
 include/linux/interconnect.h                  |  49 ++
 17 files changed, 2115 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/interconnect/interconnect.txt
 create mode 100644 Documentation/devicetree/bindings/interconnect/qcom-msm8916.txt
 create mode 100644 Documentation/devicetree/bindings/interconnect/qcom-smd.txt
 create mode 100644 Documentation/interconnect/interconnect.rst
 create mode 100644 drivers/interconnect/Kconfig
 create mode 100644 drivers/interconnect/Makefile
 create mode 100644 drivers/interconnect/core.c
 create mode 100644 drivers/interconnect/qcom/Kconfig
 create mode 100644 drivers/interconnect/qcom/Makefile
 create mode 100644 drivers/interconnect/qcom/msm8916.c
 create mode 100644 drivers/interconnect/qcom/smd-rpm.c
 create mode 100644 drivers/interconnect/qcom/smd-rpm.h
 create mode 100644 include/dt-bindings/interconnect/qcom.h
 create mode 100644 include/linux/interconnect-provider.h
 create mode 100644 include/linux/interconnect.h

[PATCH v5 0/8] Introduce on-chip interconnect API

[Georgi Djakov]

Modern SoCs have multiple processors and various dedicated cores (video, gpu,
graphics, modem). These cores are talking to each other and can generate a
lot of data flowing through the on-chip interconnects. These interconnect
buses could form different topologies such as crossbar, point to point buses,
hierarchical buses or use the network-on-chip concept.

These buses have been sized usually to handle use cases with high data
throughput but it is not necessary all the time and consume a lot of power.
Furthermore, the priority between masters can vary depending on the running
use case like video playback or CPU intensive tasks.

Having an API to control the requirement of the system in terms of bandwidth
and QoS, so we can adapt the interconnect configuration to match those by
scaling the frequencies, setting link priority and tuning QoS parameters.
This configuration can be a static, one-time operation done at boot for some
platforms or a dynamic set of operations that happen at run-time.

This patchset introduce a new API to get the requirement and configure the
interconnect buses across the entire chipset to fit with the current demand.
The API is NOT for changing the performance of the endpoint devices, but only
the interconnect path in between them.

The API is using a consumer/provider-based model, where the providers are
the interconnect buses and the consumers could be various drivers.
The consumers request interconnect resources (path) to an endpoint and set
the desired constraints on this data flow path. The provider(s) receive
requests from consumers and aggregate these requests for all master-slave
pairs on that path. Then the providers configure each participating in the
topology node according to the requested data flow path, physical links and
constraints. The topology could be complicated and multi-tiered and is SoC
specific.

Below is a simplified diagram of a real-world SoC topology. The interconnect
providers are the NoCs.

+----------------+    +----------------+
| HW Accelerator |--->|      M NoC     |<---------------+
+----------------+    +----------------+                |
                        |      |                    +------------+
 +-----+  +-------------+      V       +------+     |            |
 | DDR |  |                +--------+  | PCIe |     |            |
 +-----+  |                | Slaves |  +------+     |            |
   ^ ^    |                +--------+     |         |   C NoC    |
   | |    V                               V         |            |
+------------------+   +------------------------+   |            |   +-----+
|                  |-->|                        |-->|            |-->| CPU |
|                  |-->|                        |<--|            |   +-----+
|     Mem NoC      |   |         S NoC          |   +------------+
|                  |<--|                        |---------+    |
|                  |<--|                        |<------+ |    |   +--------+
+------------------+   +------------------------+       | |    +-->| Slaves |
  ^  ^    ^    ^          ^                             | |        +--------+
  |  |    |    |          |                             | V
+------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
| CPUs |  |  | GPU |   | DSP |  | Masters |-->|       P NoC    |-->| Slaves |
+------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
          |
      +-------+
      | Modem |
      +-------+

TODO:
 * Create icc_set_extended() to handle parameters such as latency and other
   QoS values.
 * Convert from using global node identifiers to local per provider ids.
 * Cache the path between the nodes instead of walking the graph on each get().
 * Sync interconnect requests with the idle state of the device.

Changes since patchset v4 (https://lkml.org/lkml/2018/3/9/856)
* Simplified locking by using a single global mutex. (Evan)
* Changed the aggregation function interface.
* Implemented functions for node, link, provider removal. (Evan)
* Naming changes on variables and functions, removed redundant code. (Evan)
* Fixes and clarifications in the docs. (Matthias, Evan, Amit, Alexandre)
* Removed mandatory reg DT property, made interconnect-names optional. (Bjorn)
* Made interconnect-cells property required to align with other bindings. (Neil)
* Moved msm8916 specific bindings into a separate file and patch. (Bjorn)
* Use the names, instead of the hardcoded ids for topology. (Matthias)
* Init the node before creating the links. (Evan)
* Added icc_units_to_bps macro. (Amit)

Changes since patchset v3 (https://lkml.org/lkml/2017/9/8/544)
* Refactored the constraints aggregation.
* Use the IDR API.
* Split the provider and consumer bindings into separate patches and propose
  new bindings for consumers, which allows to specify the local source port.
* Adopted the icc_ prefix for API functions.
* Introduced separate API functions for creating interconnect nodes and links.
* Added DT lookup support in addition to platform data.
* Dropped the event tracing patch for now.
* Added a patch to provide summary via debugfs.
* Use macro for the list of topology definitions in the platform driver.
* Various minor changes.

Changes since patchset v2 (https://lkml.org/lkml/2017/7/20/825)
* Split the aggregation into per node and per provider. Cache the
  aggregated values.
* Various small refactorings and cleanups in the framework.
* Added a patch introducing basic tracepoint support for monitoring
  the time required to update the interconnect nodes.

Changes since patchset v1 (https://lkml.org/lkml/2017/6/27/890)
* Updates in the documentation.
* Changes in request aggregation, locking.
* Dropped the aggregate() callback and use the default as it currently
  sufficient for the single vendor driver. Will add it later when needed.
* Dropped the dt-bindings draft patch for now.

Changes since RFC v2 (https://lkml.org/lkml/2017/6/12/316)
* Converted documentation to rst format.
* Fixed an incorrect call to mutex_lock. Renamed max_bw to peak_bw.

Changes since RFC v1 (https://lkml.org/lkml/2017/5/15/605)
* Refactored code into shorter functions.
* Added a new aggregate() API function.
* Rearranged some structs to reduce padding bytes.

Changes since RFC v0 (https://lkml.org/lkml/2017/3/1/599)
* Removed DT support and added optional Patch 3 with new bindings proposal.
* Converted the topology into internal driver data.
* Made the framework modular.
* interconnect_get() now takes (src and dst ports as arguments).
* Removed public declarations of some structs.
* Now passing prev/next nodes to the vendor driver.
* Properly remove requests on _put().
* Added refcounting.
* Updated documentation.
* Changed struct interconnect_path to use array instead of linked list.

Georgi Djakov (8):
  interconnect: Add generic on-chip interconnect API
  dt-bindings: Introduce interconnect provider bindings
  interconnect: Add debugfs support
  interconnect: qcom: Add RPM communication
  dt-bindings: interconnect: Document qcom,msm8916 NoC bindings
  interconnect: qcom: Add msm8916 interconnect provider driver
  dt-bindings: Introduce interconnect consumers bindings
  interconnect: Allow endpoints translation via DT

 .../bindings/interconnect/interconnect.txt    |  60 ++
 .../bindings/interconnect/qcom-msm8916.txt    |  39 +
 .../bindings/interconnect/qcom-smd.txt        |  32 +
 Documentation/interconnect/interconnect.rst   |  96 +++
 drivers/Kconfig                               |   2 +
 drivers/Makefile                              |   1 +
 drivers/interconnect/Kconfig                  |  15 +
 drivers/interconnect/Makefile                 |   3 +
 drivers/interconnect/core.c                   | 726 ++++++++++++++++++
 drivers/interconnect/qcom/Kconfig             |  11 +
 drivers/interconnect/qcom/Makefile            |   4 +
 drivers/interconnect/qcom/msm8916.c           | 495 ++++++++++++
 drivers/interconnect/qcom/smd-rpm.c           |  90 +++
 drivers/interconnect/qcom/smd-rpm.h           |  15 +
 include/dt-bindings/interconnect/qcom.h       | 350 +++++++++
 include/linux/interconnect-provider.h         | 127 +++
 include/linux/interconnect.h                  |  49 ++
 17 files changed, 2115 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/interconnect/interconnect.txt
 create mode 100644 Documentation/devicetree/bindings/interconnect/qcom-msm8916.txt
 create mode 100644 Documentation/devicetree/bindings/interconnect/qcom-smd.txt
 create mode 100644 Documentation/interconnect/interconnect.rst
 create mode 100644 drivers/interconnect/Kconfig
 create mode 100644 drivers/interconnect/Makefile
 create mode 100644 drivers/interconnect/core.c
 create mode 100644 drivers/interconnect/qcom/Kconfig
 create mode 100644 drivers/interconnect/qcom/Makefile
 create mode 100644 drivers/interconnect/qcom/msm8916.c
 create mode 100644 drivers/interconnect/qcom/smd-rpm.c
 create mode 100644 drivers/interconnect/qcom/smd-rpm.h
 create mode 100644 include/dt-bindings/interconnect/qcom.h
 create mode 100644 include/linux/interconnect-provider.h
 create mode 100644 include/linux/interconnect.h

[PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Georgi Djakov]

This patch introduce a new API to get requirements and configure the
interconnect buses across the entire chipset to fit with the current
demand.

The API is using a consumer/provider-based model, where the providers are
the interconnect buses and the consumers could be various drivers.
The consumers request interconnect resources (path) between endpoints and
set the desired constraints on this data flow path. The providers receive
requests from consumers and aggregate these requests for all master-slave
pairs on that path. Then the providers configure each participating in the
topology node according to the requested data flow path, physical links and
constraints. The topology could be complicated and multi-tiered and is SoC
specific.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 Documentation/interconnect/interconnect.rst |  96 ++++
 drivers/Kconfig                             |   2 +
 drivers/Makefile                            |   1 +
 drivers/interconnect/Kconfig                |  10 +
 drivers/interconnect/Makefile               |   2 +
 drivers/interconnect/core.c                 | 586 ++++++++++++++++++++
 include/linux/interconnect-provider.h       | 127 +++++
 include/linux/interconnect.h                |  42 ++
 8 files changed, 866 insertions(+)
 create mode 100644 Documentation/interconnect/interconnect.rst
 create mode 100644 drivers/interconnect/Kconfig
 create mode 100644 drivers/interconnect/Makefile
 create mode 100644 drivers/interconnect/core.c
 create mode 100644 include/linux/interconnect-provider.h
 create mode 100644 include/linux/interconnect.h

diff --git a/Documentation/interconnect/interconnect.rst b/Documentation/interconnect/interconnect.rst
new file mode 100644
index 000000000000..a1ebd83ad0a1
--- /dev/null
+++ b/Documentation/interconnect/interconnect.rst
@@ -0,0 +1,96 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+=====================================
+GENERIC SYSTEM INTERCONNECT SUBSYSTEM
+=====================================
+
+Introduction
+------------
+
+This framework is designed to provide a standard kernel interface to control
+the settings of the interconnects on a SoC. These settings can be throughput,
+latency and priority between multiple interconnected devices or functional
+blocks. This can be controlled dynamically in order to save power or provide
+maximum performance.
+
+The interconnect bus is a hardware with configurable parameters, which can be
+set on a data path according to the requests received from various drivers.
+An example of interconnect buses are the interconnects between various
+components or functional blocks in chipsets. There can be multiple interconnects
+on a SoC that can be multi-tiered.
+
+Below is a simplified diagram of a real-world SoC interconnect bus topology.
+
+::
+
+ +----------------+    +----------------+
+ | HW Accelerator |--->|      M NoC     |<---------------+
+ +----------------+    +----------------+                |
+                         |      |                    +------------+
+  +-----+  +-------------+      V       +------+     |            |
+  | DDR |  |                +--------+  | PCIe |     |            |
+  +-----+  |                | Slaves |  +------+     |            |
+    ^ ^    |                +--------+     |         |   C NoC    |
+    | |    V                               V         |            |
+ +------------------+   +------------------------+   |            |   +-----+
+ |                  |-->|                        |-->|            |-->| CPU |
+ |                  |-->|                        |<--|            |   +-----+
+ |     Mem NoC      |   |         S NoC          |   +------------+
+ |                  |<--|                        |---------+    |
+ |                  |<--|                        |<------+ |    |   +--------+
+ +------------------+   +------------------------+       | |    +-->| Slaves |
+   ^  ^    ^    ^          ^                             | |        +--------+
+   |  |    |    |          |                             | V
+ +------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
+ | CPUs |  |  | GPU |   | DSP |  | Masters |-->|       P NoC    |-->| Slaves |
+ +------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
+           |
+       +-------+
+       | Modem |
+       +-------+
+
+Terminology
+-----------
+
+Interconnect provider is the software definition of the interconnect hardware.
+The interconnect providers on the above diagram are M NoC, S NoC, C NoC, P NoC
+and Mem NoC.
+
+Interconnect node is the software definition of the interconnect hardware
+port. Each interconnect provider consists of multiple interconnect nodes,
+which are connected to other SoC components including other interconnect
+providers. The point on the diagram where the CPUs connects to the memory is
+called an interconnect node, which belongs to the Mem NoC interconnect provider.
+
+Interconnect endpoints are the first or the last element of the path. Every
+endpoint is a node, but not every node is an endpoint.
+
+Interconnect path is everything between two endpoints including all the nodes
+that have to be traversed to reach from a source to destination node. It may
+include multiple master-slave pairs across several interconnect providers.
+
+Interconnect consumers are the entities which make use of the data paths exposed
+by the providers. The consumers send requests to providers requesting various
+throughput, latency and priority. Usually the consumers are device drivers, that
+send request based on their needs. An example for a consumer is a video decoder
+that supports various formats and image sizes.
+
+Interconnect providers
+----------------------
+
+Interconnect provider is an entity that implements methods to initialize and
+configure a interconnect bus hardware. The interconnect provider drivers should
+be registered with the interconnect provider core.
+
+The interconnect framework provider API functions are documented in
+.. kernel-doc:: include/linux/interconnect-provider.h
+
+Interconnect consumers
+----------------------
+
+Interconnect consumers are the clients which use the interconnect APIs to
+get paths between endpoints and set their bandwidth/latency/QoS requirements
+for these interconnect paths.
+
+The interconnect framework consumer API functions are documented in
+.. kernel-doc:: include/linux/interconnect.h
diff --git a/drivers/Kconfig b/drivers/Kconfig
index 95b9ccc08165..3ed6ede9d021 100644
--- a/drivers/Kconfig
+++ b/drivers/Kconfig
@@ -217,4 +217,6 @@ source "drivers/siox/Kconfig"
 
 source "drivers/slimbus/Kconfig"
 
+source "drivers/interconnect/Kconfig"
+
 endmenu
diff --git a/drivers/Makefile b/drivers/Makefile
index 24cd47014657..0cca95740d9b 100644
--- a/drivers/Makefile
+++ b/drivers/Makefile
@@ -185,3 +185,4 @@ obj-$(CONFIG_TEE)		+= tee/
 obj-$(CONFIG_MULTIPLEXER)	+= mux/
 obj-$(CONFIG_UNISYS_VISORBUS)	+= visorbus/
 obj-$(CONFIG_SIOX)		+= siox/
+obj-$(CONFIG_INTERCONNECT)	+= interconnect/
diff --git a/drivers/interconnect/Kconfig b/drivers/interconnect/Kconfig
new file mode 100644
index 000000000000..a261c7d41deb
--- /dev/null
+++ b/drivers/interconnect/Kconfig
@@ -0,0 +1,10 @@
+menuconfig INTERCONNECT
+	tristate "On-Chip Interconnect management support"
+	help
+	  Support for management of the on-chip interconnects.
+
+	  This framework is designed to provide a generic interface for
+	  managing the interconnects in a SoC.
+
+	  If unsure, say no.
+
diff --git a/drivers/interconnect/Makefile b/drivers/interconnect/Makefile
new file mode 100644
index 000000000000..97fca2e09d24
--- /dev/null
+++ b/drivers/interconnect/Makefile
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0
+obj-$(CONFIG_INTERCONNECT)		+= core.o
diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
new file mode 100644
index 000000000000..e7f96fc6722e
--- /dev/null
+++ b/drivers/interconnect/core.c
@@ -0,0 +1,586 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Interconnect framework core driver
+ *
+ * Copyright (c) 2018, Linaro Ltd.
+ * Author: Georgi Djakov <georgi.djakov@linaro.org>
+ */
+
+#include <linux/device.h>
+#include <linux/idr.h>
+#include <linux/init.h>
+#include <linux/interconnect.h>
+#include <linux/interconnect-provider.h>
+#include <linux/list.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/slab.h>
+
+static DEFINE_IDR(icc_idr);
+static LIST_HEAD(icc_provider_list);
+static DEFINE_MUTEX(icc_lock);
+
+/**
+ * struct icc_req - constraints that are attached to each node
+ *
+ * @req_node: entry in list of requests for the particular @node
+ * @node: the interconnect node to which this constraint applies
+ * @dev: reference to the device that sets the constraints
+ * @avg_bw: an integer describing the average bandwidth in kbps
+ * @peak_bw: an integer describing the peak bandwidth in kbps
+ */
+struct icc_req {
+	struct hlist_node req_node;
+	struct icc_node *node;
+	struct device *dev;
+	u32 avg_bw;
+	u32 peak_bw;
+};
+
+/**
+ * struct icc_path - interconnect path structure
+ * @num_nodes: number of hops (nodes)
+ * @reqs: array of the requests applicable to this path of nodes
+ */
+struct icc_path {
+	size_t num_nodes;
+	struct icc_req reqs[0];
+};
+
+static struct icc_node *node_find(const int id)
+{
+	struct icc_node *node;
+
+	mutex_lock(&icc_lock);
+	node = idr_find(&icc_idr, id);
+	mutex_unlock(&icc_lock);
+
+	return node;
+}
+
+static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
+{
+	struct icc_node *node = dst;
+	struct icc_path *path;
+	size_t i;
+
+	path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
+		       GFP_KERNEL);
+	if (!path)
+		return ERR_PTR(-ENOMEM);
+
+	path->num_nodes = num_nodes;
+
+	for (i = 0; i < num_nodes; i++) {
+		hlist_add_head(&path->reqs[i].req_node, &node->req_list);
+
+		path->reqs[i].node = node;
+		/* reference to previous node was saved during path traversal */
+		node = node->reverse;
+	}
+
+	return path;
+}
+
+static struct icc_path *path_find(struct device *dev, struct icc_node *src,
+				  struct icc_node *dst)
+{
+	struct icc_node *n, *node = NULL;
+	struct icc_provider *provider;
+	struct list_head traverse_list;
+	struct list_head edge_list;
+	struct list_head visited_list;
+	size_t i, depth = 0;
+	bool found = false;
+	int ret = -EPROBE_DEFER;
+
+	INIT_LIST_HEAD(&traverse_list);
+	INIT_LIST_HEAD(&edge_list);
+	INIT_LIST_HEAD(&visited_list);
+
+	list_add_tail(&src->search_list, &traverse_list);
+	src->reverse = NULL;
+
+	do {
+		list_for_each_entry_safe(node, n, &traverse_list, search_list) {
+			if (node == dst) {
+				found = true;
+				list_add(&node->search_list, &visited_list);
+				break;
+			}
+			for (i = 0; i < node->num_links; i++) {
+				struct icc_node *tmp = node->links[i];
+
+				if (!tmp) {
+					ret = -ENOENT;
+					goto out;
+				}
+
+				if (tmp->is_traversed)
+					continue;
+
+				tmp->is_traversed = true;
+				tmp->reverse = node;
+				list_add(&tmp->search_list, &edge_list);
+			}
+		}
+		if (found)
+			break;
+
+		list_splice_init(&traverse_list, &visited_list);
+		list_splice_init(&edge_list, &traverse_list);
+
+		/* count the hops away from the source */
+		depth++;
+
+	} while (!list_empty(&traverse_list));
+
+out:
+	/* reset the traversed state */
+	list_for_each_entry(provider, &icc_provider_list, provider_list) {
+		list_for_each_entry(n, &provider->nodes, node_list)
+			if (n->is_traversed)
+				n->is_traversed = false;
+	}
+
+	if (found) {
+		struct icc_path *path = path_allocate(dst, depth);
+
+		if (IS_ERR(path))
+			return path;
+
+		/* initialize the path */
+		for (i = 0; i < path->num_nodes; i++) {
+			node = path->reqs[i].node;
+			path->reqs[i].dev = dev;
+			node->provider->users++;
+		}
+		return path;
+	}
+
+	return ERR_PTR(ret);
+}
+
+/*
+ * We want the path to honor all bandwidth requests, so the average
+ * bandwidth requirements from each consumer are aggregated at each node
+ * and provider level. By default the average bandwidth is the sum of all
+ * averages and the peak will be the highest of all peak bandwidth requests.
+ */
+
+static int aggregate_requests(struct icc_node *node)
+{
+	struct icc_provider *p = node->provider;
+	struct icc_req *r;
+
+	node->avg_bw = 0;
+	node->peak_bw = 0;
+
+	hlist_for_each_entry(r, &node->req_list, req_node)
+		p->aggregate(node, r->avg_bw, r->peak_bw,
+			     &node->avg_bw, &node->peak_bw);
+
+	return 0;
+}
+
+static void aggregate_provider(struct icc_provider *p)
+{
+	struct icc_node *n;
+
+	p->avg_bw = 0;
+	p->peak_bw = 0;
+
+	list_for_each_entry(n, &p->nodes, node_list)
+		p->aggregate(n, n->avg_bw, n->peak_bw,
+			     &p->avg_bw, &p->peak_bw);
+}
+
+static int apply_constraints(struct icc_path *path)
+{
+	struct icc_node *next, *prev = NULL;
+	int ret = 0;
+	int i;
+
+	for (i = 0; i < path->num_nodes; i++, prev = next) {
+		struct icc_provider *p;
+
+		next = path->reqs[i].node;
+		/*
+		 * Both endpoints should be valid master-slave pairs of the
+		 * same interconnect provider that will be configured.
+		 */
+		if (!prev || next->provider != prev->provider)
+			continue;
+
+		p = next->provider;
+
+		aggregate_provider(p);
+
+		if (p->set) {
+			/* set the constraints */
+			ret = p->set(prev, next, p->avg_bw, p->peak_bw);
+		}
+
+		if (ret)
+			goto out;
+	}
+out:
+	return ret;
+}
+
+/**
+ * icc_set() - set constraints on an interconnect path between two endpoints
+ * @path: reference to the path returned by icc_get()
+ * @avg_bw: average bandwidth in kbps
+ * @peak_bw: peak bandwidth in kbps
+ *
+ * This function is used by an interconnect consumer to express its own needs
+ * in terms of bandwidth for a previously requested path between two endpoints.
+ * The requests are aggregated and each node is updated accordingly. The entire
+ * path is locked by a mutex to ensure that the set() is completed.
+ * The @path can be NULL when the "interconnects" DT properties is missing,
+ * which will mean that no constraints will be set.
+ *
+ * Returns 0 on success, or an appropriate error code otherwise.
+ */
+int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
+{
+	struct icc_node *node;
+	struct icc_provider *p;
+	size_t i;
+	int ret = 0;
+
+	if (!path)
+		return 0;
+
+	mutex_lock(&icc_lock);
+
+	for (i = 0; i < path->num_nodes; i++) {
+		node = path->reqs[i].node;
+		p = node->provider;
+
+		/* update the consumer request for this path */
+		path->reqs[i].avg_bw = avg_bw;
+		path->reqs[i].peak_bw = peak_bw;
+
+		/* aggregate requests for this node */
+		aggregate_requests(node);
+	}
+
+	ret = apply_constraints(path);
+	if (ret)
+		pr_err("interconnect: error applying constraints (%d)", ret);
+
+	mutex_unlock(&icc_lock);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(icc_set);
+
+/**
+ * icc_get() - return a handle for path between two endpoints
+ * @dev: the device requesting the path
+ * @src_id: source device port id
+ * @dst_id: destination device port id
+ *
+ * This function will search for a path between two endpoints and return an
+ * icc_path handle on success. Use icc_put() to release
+ * constraints when the they are not needed anymore.
+ *
+ * Return: icc_path pointer on success, or ERR_PTR() on error
+ */
+struct icc_path *icc_get(struct device *dev, const int src_id, const int dst_id)
+{
+	struct icc_node *src, *dst;
+	struct icc_path *path = ERR_PTR(-EPROBE_DEFER);
+
+	src = node_find(src_id);
+	if (!src) {
+		dev_err(dev, "%s: invalid src=%d\n", __func__, src_id);
+		goto out;
+	}
+
+	dst = node_find(dst_id);
+	if (!dst) {
+		dev_err(dev, "%s: invalid dst=%d\n", __func__, dst_id);
+		goto out;
+	}
+
+	mutex_lock(&icc_lock);
+	path = path_find(dev, src, dst);
+	mutex_unlock(&icc_lock);
+	if (IS_ERR(path)) {
+		dev_err(dev, "%s: invalid path=%ld\n", __func__, PTR_ERR(path));
+		goto out;
+	}
+
+out:
+	return path;
+}
+EXPORT_SYMBOL_GPL(icc_get);
+
+/**
+ * icc_put() - release the reference to the icc_path
+ * @path: interconnect path
+ *
+ * Use this function to release the constraints on a path when the path is
+ * no longer needed. The constraints will be re-aggregated.
+ */
+void icc_put(struct icc_path *path)
+{
+	struct icc_node *node;
+	size_t i;
+	int ret;
+
+	if (!path || WARN_ON_ONCE(IS_ERR(path)))
+		return;
+
+	ret = icc_set(path, 0, 0);
+	if (ret)
+		pr_err("%s: error (%d)\n", __func__, ret);
+
+	mutex_lock(&icc_lock);
+	for (i = 0; i < path->num_nodes; i++) {
+		node = path->reqs[i].node;
+		hlist_del(&path->reqs[i].req_node);
+
+		node->provider->users--;
+	}
+	mutex_unlock(&icc_lock);
+
+	kfree(path);
+}
+EXPORT_SYMBOL_GPL(icc_put);
+
+/**
+ * icc_node_create() - create a node
+ * @id: node id
+ *
+ * Return: icc_node pointer on success, or ERR_PTR() on error
+ */
+struct icc_node *icc_node_create(int id)
+{
+	struct icc_node *node;
+
+	/* check if node already exists */
+	node = node_find(id);
+	if (node)
+		goto out;
+
+	node = kzalloc(sizeof(*node), GFP_KERNEL);
+	if (!node) {
+		node = ERR_PTR(-ENOMEM);
+		goto out;
+	}
+
+	mutex_lock(&icc_lock);
+
+	id = idr_alloc(&icc_idr, node, id, id + 1, GFP_KERNEL);
+	if (WARN(id < 0, "couldn't get idr")) {
+		node = ERR_PTR(id);
+		goto out;
+	}
+
+	node->id = id;
+
+out:
+	mutex_unlock(&icc_lock);
+
+	return node;
+}
+EXPORT_SYMBOL_GPL(icc_node_create);
+
+/**
+ * icc_node_remove() - remove a node
+ * @id: node id
+ *
+ */
+void icc_node_remove(int id)
+{
+	struct icc_node *node;
+
+	node = node_find(id);
+	if (node) {
+		mutex_lock(&icc_lock);
+		idr_remove(&icc_idr, node->id);
+		mutex_unlock(&icc_lock);
+	}
+
+	kfree(node);
+}
+EXPORT_SYMBOL_GPL(icc_node_remove);
+
+/**
+ * icc_link_create() - create a link between two nodes
+ * @src_id: source node id
+ * @dst_id: destination node id
+ *
+ * Create a link between two nodes. The nodes might belong to different
+ * interconnect providers and the @dst_id node might not exist (if the
+ * provider driver has not probed yet). So just create the @dst_id node
+ * and when the actual provider driver is probed, the rest of the node
+ * data is filled.
+ *
+ * Return: 0 on success, or an error code otherwise
+ */
+int icc_link_create(struct icc_node *node, const int dst_id)
+{
+	struct icc_node *dst;
+	struct icc_node **new;
+	int ret = 0;
+
+	if (!node->provider)
+		return -EINVAL;
+
+	dst = node_find(dst_id);
+	if (!dst) {
+		dst = icc_node_create(dst_id);
+
+		if (IS_ERR(dst)) {
+			ret = PTR_ERR(dst);
+			goto out;
+		}
+	}
+
+	mutex_lock(&icc_lock);
+
+	new = krealloc(node->links,
+		       (node->num_links + 1) * sizeof(*node->links),
+		       GFP_KERNEL);
+	if (!new) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	node->links = new;
+	node->links[node->num_links++] = dst;
+
+out:
+	mutex_unlock(&icc_lock);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(icc_link_create);
+
+/**
+ * icc_link_remove() - remove a link between two nodes
+ * @src: pointer to source node
+ * @dst: pointer to destination node
+ *
+ * Return: 0 on success, or an error code otherwise
+ */
+int icc_link_remove(struct icc_node *src, struct icc_node *dst)
+{
+	struct icc_node **new;
+	int ret = 0;
+	int i, j;
+
+	if (IS_ERR_OR_NULL(src))
+		return PTR_ERR(src);
+
+	if (IS_ERR_OR_NULL(dst))
+		return PTR_ERR(dst);
+
+	mutex_lock(&icc_lock);
+
+	new = krealloc(src->links,
+		       (src->num_links - 1) * sizeof(*src->links),
+		       GFP_KERNEL);
+	if (!new) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	for (i = 0, j = 0; j < src->num_links; j++) {
+		if (src->links[j] == dst)
+			continue;
+
+		new[i++] = src->links[j];
+	}
+
+	src->links = new;
+	src->num_links--;
+
+out:
+	mutex_unlock(&icc_lock);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(icc_link_remove);
+
+/**
+ * icc_node_add() - add an interconnect node to interconnect provider
+ * @node: pointer to the interconnect node
+ * @provider: pointer to the interconnect provider
+ *
+ * Return: 0 on success, or an error code otherwise
+ */
+int icc_node_add(struct icc_node *node, struct icc_provider *provider)
+{
+	mutex_lock(&icc_lock);
+
+	node->provider = provider;
+	list_add(&node->node_list, &provider->nodes);
+
+	mutex_unlock(&icc_lock);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(icc_node_add);
+
+/**
+ * icc_provider_add() - add a new interconnect provider
+ * @icc_provider: the interconnect provider that will be added into topology
+ *
+ * Return: 0 on success, or an error code otherwise
+ */
+int icc_provider_add(struct icc_provider *provider)
+{
+	if (WARN_ON(!provider->set))
+		return -EINVAL;
+
+	mutex_init(&icc_lock);
+
+	INIT_LIST_HEAD(&provider->nodes);
+	list_add(&provider->provider_list, &icc_provider_list);
+
+	mutex_unlock(&icc_lock);
+
+	dev_dbg(provider->dev, "interconnect provider added to topology\n");
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(icc_provider_add);
+
+/**
+ * icc_provider_del() - delete previously added interconnect provider
+ * @icc_provider: the interconnect provider that will be removed from topology
+ *
+ * Return: 0 on success, or an error code otherwise
+ */
+int icc_provider_del(struct icc_provider *provider)
+{
+	mutex_lock(&icc_lock);
+	if (provider->users) {
+		pr_warn("interconnect provider still has %d users\n",
+			provider->users);
+		mutex_unlock(&icc_lock);
+		return -EBUSY;
+	}
+
+	if (!list_empty_careful(&provider->nodes)) {
+		pr_warn("interconnect provider still has nodes\n");
+		mutex_unlock(&icc_lock);
+		return -EEXIST;
+	}
+
+	list_del(&provider->provider_list);
+	mutex_unlock(&icc_lock);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(icc_provider_del);
+
+MODULE_AUTHOR("Georgi Djakov <georgi.djakov@linaro.org");
+MODULE_DESCRIPTION("Interconnect Driver Core");
+MODULE_LICENSE("GPL v2");
diff --git a/include/linux/interconnect-provider.h b/include/linux/interconnect-provider.h
new file mode 100644
index 000000000000..f4613c6dce4f
--- /dev/null
+++ b/include/linux/interconnect-provider.h
@@ -0,0 +1,127 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2018, Linaro Ltd.
+ * Author: Georgi Djakov <georgi.djakov@linaro.org>
+ */
+
+#ifndef _LINUX_INTERCONNECT_PROVIDER_H
+#define _LINUX_INTERCONNECT_PROVIDER_H
+
+#include <linux/interconnect.h>
+
+#define icc_units_to_bps(bw)  ((bw) * 1000ULL)
+
+struct icc_node;
+
+/**
+ * struct icc_provider - interconnect provider (controller) entity that might
+ * provide multiple interconnect controls
+ *
+ * @provider_list: list of the registered interconnect providers
+ * @nodes: internal list of the interconnect provider nodes
+ * @set: pointer to device specific set operation function
+ * @aggregate: pointer to device specific aggregate operation function
+ * @dev: the device this interconnect provider belongs to
+ * @users: count of active users
+ * @avg_bw: aggregated value of average bandwidth requests from all nodes
+ * @peak_bw: aggregated value of peak bandwidth requests from all nodes
+ * @data: pointer to private data
+ */
+struct icc_provider {
+	struct list_head	provider_list;
+	struct list_head	nodes;
+	int (*set)(struct icc_node *src, struct icc_node *dst,
+		   u32 avg_bw, u32 peak_bw);
+	int (*aggregate)(struct icc_node *node, u32 avg_bw, u32 peak_bw,
+			 u32 *agg_avg, u32 *agg_peak);
+	struct device		*dev;
+	int			users;
+	u32			avg_bw;
+	u32			peak_bw;
+	void			*data;
+};
+
+/**
+ * struct icc_node - entity that is part of the interconnect topology
+ *
+ * @id: platform specific node id
+ * @name: node name used in debugfs
+ * @links: a list of targets pointing to where we can go next when traversing
+ * @num_links: number of links to other interconnect nodes
+ * @provider: points to the interconnect provider of this node
+ * @node_list: the list entry in the parent provider's "nodes" list
+ * @search_list: list used when walking the nodes graph
+ * @reverse: pointer to previous node when walking the nodes graph
+ * @is_traversed: flag that is used when walking the nodes graph
+ * @req_list: a list of QoS constraint requests associated with this node
+ * @avg_bw: aggregated value of average bandwidth requests from all consumers
+ * @peak_bw: aggregated value of peak bandwidth requests from all consumers
+ * @data: pointer to private data
+ */
+struct icc_node {
+	int			id;
+	const char              *name;
+	struct icc_node		**links;
+	size_t			num_links;
+
+	struct icc_provider	*provider;
+	struct list_head	node_list;
+	struct list_head	orphan_list;
+	struct list_head	search_list;
+	struct icc_node		*reverse;
+	bool			is_traversed;
+	struct hlist_head	req_list;
+	u32			avg_bw;
+	u32			peak_bw;
+	void			*data;
+};
+
+#if IS_ENABLED(CONFIG_INTERCONNECT)
+
+struct icc_node *icc_node_create(int id);
+void icc_node_remove(int id);
+int icc_link_create(struct icc_node *node, const int dst_id);
+int icc_link_remove(struct icc_node *src, struct icc_node *dst);
+int icc_node_add(struct icc_node *node, struct icc_provider *provider);
+int icc_provider_add(struct icc_provider *provider);
+int icc_provider_del(struct icc_provider *provider);
+
+#else
+
+static inline struct icc_node *icc_node_create(int id)
+{
+	return ERR_PTR(-ENOTSUPP);
+}
+
+void icc_node_remove(int id)
+{
+}
+
+static inline int icc_link_create(struct icc_node *node, const int dst_id)
+{
+	return -ENOTSUPP;
+}
+
+int icc_link_remove(struct icc_node *src, struct icc_node *dst)
+{
+	return -ENOTSUPP;
+}
+
+int icc_node_add(struct icc_node *node, struct icc_provider *provider)
+{
+	return -ENOTSUPP;
+}
+
+static inline int icc_provider_add(struct icc_provider *provider)
+{
+	return -ENOTSUPP;
+}
+
+static inline int icc_provider_del(struct icc_provider *provider)
+{
+	return -ENOTSUPP;
+}
+
+#endif /* CONFIG_INTERCONNECT */
+
+#endif /* _LINUX_INTERCONNECT_PROVIDER_H */
diff --git a/include/linux/interconnect.h b/include/linux/interconnect.h
new file mode 100644
index 000000000000..593215371fd6
--- /dev/null
+++ b/include/linux/interconnect.h
@@ -0,0 +1,42 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2018, Linaro Ltd.
+ * Author: Georgi Djakov <georgi.djakov@linaro.org>
+ */
+
+#ifndef _LINUX_INTERCONNECT_H
+#define _LINUX_INTERCONNECT_H
+
+#include <linux/mutex.h>
+#include <linux/types.h>
+
+struct icc_path;
+struct device;
+
+#if IS_ENABLED(CONFIG_INTERCONNECT)
+
+struct icc_path *icc_get(struct device *dev, const int src_id,
+			 const int dst_id);
+void icc_put(struct icc_path *path);
+int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw);
+
+#else
+
+static inline struct icc_path *icc_get(struct device *dev, const int src_id,
+				       const int dst_id)
+{
+	return NULL;
+}
+
+static inline void icc_put(struct icc_path *path)
+{
+}
+
+static inline int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
+{
+	return 0;
+}
+
+#endif /* CONFIG_INTERCONNECT */
+
+#endif /* _LINUX_INTERCONNECT_H */

[PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Georgi Djakov]

This patch introduce a new API to get requirements and configure the
interconnect buses across the entire chipset to fit with the current
demand.

The API is using a consumer/provider-based model, where the providers are
the interconnect buses and the consumers could be various drivers.
The consumers request interconnect resources (path) between endpoints and
set the desired constraints on this data flow path. The providers receive
requests from consumers and aggregate these requests for all master-slave
pairs on that path. Then the providers configure each participating in the
topology node according to the requested data flow path, physical links and
constraints. The topology could be complicated and multi-tiered and is SoC
specific.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 Documentation/interconnect/interconnect.rst |  96 ++++
 drivers/Kconfig                             |   2 +
 drivers/Makefile                            |   1 +
 drivers/interconnect/Kconfig                |  10 +
 drivers/interconnect/Makefile               |   2 +
 drivers/interconnect/core.c                 | 586 ++++++++++++++++++++
 include/linux/interconnect-provider.h       | 127 +++++
 include/linux/interconnect.h                |  42 ++
 8 files changed, 866 insertions(+)
 create mode 100644 Documentation/interconnect/interconnect.rst
 create mode 100644 drivers/interconnect/Kconfig
 create mode 100644 drivers/interconnect/Makefile
 create mode 100644 drivers/interconnect/core.c
 create mode 100644 include/linux/interconnect-provider.h
 create mode 100644 include/linux/interconnect.h

diff --git a/Documentation/interconnect/interconnect.rst b/Documentation/interconnect/interconnect.rst
new file mode 100644
index 000000000000..a1ebd83ad0a1
--- /dev/null
+++ b/Documentation/interconnect/interconnect.rst
@@ -0,0 +1,96 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+=====================================
+GENERIC SYSTEM INTERCONNECT SUBSYSTEM
+=====================================
+
+Introduction
+------------
+
+This framework is designed to provide a standard kernel interface to control
+the settings of the interconnects on a SoC. These settings can be throughput,
+latency and priority between multiple interconnected devices or functional
+blocks. This can be controlled dynamically in order to save power or provide
+maximum performance.
+
+The interconnect bus is a hardware with configurable parameters, which can be
+set on a data path according to the requests received from various drivers.
+An example of interconnect buses are the interconnects between various
+components or functional blocks in chipsets. There can be multiple interconnects
+on a SoC that can be multi-tiered.
+
+Below is a simplified diagram of a real-world SoC interconnect bus topology.
+
+::
+
+ +----------------+    +----------------+
+ | HW Accelerator |--->|      M NoC     |<---------------+
+ +----------------+    +----------------+                |
+                         |      |                    +------------+
+  +-----+  +-------------+      V       +------+     |            |
+  | DDR |  |                +--------+  | PCIe |     |            |
+  +-----+  |                | Slaves |  +------+     |            |
+    ^ ^    |                +--------+     |         |   C NoC    |
+    | |    V                               V         |            |
+ +------------------+   +------------------------+   |            |   +-----+
+ |                  |-->|                        |-->|            |-->| CPU |
+ |                  |-->|                        |<--|            |   +-----+
+ |     Mem NoC      |   |         S NoC          |   +------------+
+ |                  |<--|                        |---------+    |
+ |                  |<--|                        |<------+ |    |   +--------+
+ +------------------+   +------------------------+       | |    +-->| Slaves |
+   ^  ^    ^    ^          ^                             | |        +--------+
+   |  |    |    |          |                             | V
+ +------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
+ | CPUs |  |  | GPU |   | DSP |  | Masters |-->|       P NoC    |-->| Slaves |
+ +------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
+           |
+       +-------+
+       | Modem |
+       +-------+
+
+Terminology
+-----------
+
+Interconnect provider is the software definition of the interconnect hardware.
+The interconnect providers on the above diagram are M NoC, S NoC, C NoC, P NoC
+and Mem NoC.
+
+Interconnect node is the software definition of the interconnect hardware
+port. Each interconnect provider consists of multiple interconnect nodes,
+which are connected to other SoC components including other interconnect
+providers. The point on the diagram where the CPUs connects to the memory is
+called an interconnect node, which belongs to the Mem NoC interconnect provider.
+
+Interconnect endpoints are the first or the last element of the path. Every
+endpoint is a node, but not every node is an endpoint.
+
+Interconnect path is everything between two endpoints including all the nodes
+that have to be traversed to reach from a source to destination node. It may
+include multiple master-slave pairs across several interconnect providers.
+
+Interconnect consumers are the entities which make use of the data paths exposed
+by the providers. The consumers send requests to providers requesting various
+throughput, latency and priority. Usually the consumers are device drivers, that
+send request based on their needs. An example for a consumer is a video decoder
+that supports various formats and image sizes.
+
+Interconnect providers
+----------------------
+
+Interconnect provider is an entity that implements methods to initialize and
+configure a interconnect bus hardware. The interconnect provider drivers should
+be registered with the interconnect provider core.
+
+The interconnect framework provider API functions are documented in
+.. kernel-doc:: include/linux/interconnect-provider.h
+
+Interconnect consumers
+----------------------
+
+Interconnect consumers are the clients which use the interconnect APIs to
+get paths between endpoints and set their bandwidth/latency/QoS requirements
+for these interconnect paths.
+
+The interconnect framework consumer API functions are documented in
+.. kernel-doc:: include/linux/interconnect.h
diff --git a/drivers/Kconfig b/drivers/Kconfig
index 95b9ccc08165..3ed6ede9d021 100644
--- a/drivers/Kconfig
+++ b/drivers/Kconfig
@@ -217,4 +217,6 @@ source "drivers/siox/Kconfig"
 
 source "drivers/slimbus/Kconfig"
 
+source "drivers/interconnect/Kconfig"
+
 endmenu
diff --git a/drivers/Makefile b/drivers/Makefile
index 24cd47014657..0cca95740d9b 100644
--- a/drivers/Makefile
+++ b/drivers/Makefile
@@ -185,3 +185,4 @@ obj-$(CONFIG_TEE)		+= tee/
 obj-$(CONFIG_MULTIPLEXER)	+= mux/
 obj-$(CONFIG_UNISYS_VISORBUS)	+= visorbus/
 obj-$(CONFIG_SIOX)		+= siox/
+obj-$(CONFIG_INTERCONNECT)	+= interconnect/
diff --git a/drivers/interconnect/Kconfig b/drivers/interconnect/Kconfig
new file mode 100644
index 000000000000..a261c7d41deb
--- /dev/null
+++ b/drivers/interconnect/Kconfig
@@ -0,0 +1,10 @@
+menuconfig INTERCONNECT
+	tristate "On-Chip Interconnect management support"
+	help
+	  Support for management of the on-chip interconnects.
+
+	  This framework is designed to provide a generic interface for
+	  managing the interconnects in a SoC.
+
+	  If unsure, say no.
+
diff --git a/drivers/interconnect/Makefile b/drivers/interconnect/Makefile
new file mode 100644
index 000000000000..97fca2e09d24
--- /dev/null
+++ b/drivers/interconnect/Makefile
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0
+obj-$(CONFIG_INTERCONNECT)		+= core.o
diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
new file mode 100644
index 000000000000..e7f96fc6722e
--- /dev/null
+++ b/drivers/interconnect/core.c
@@ -0,0 +1,586 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Interconnect framework core driver
+ *
+ * Copyright (c) 2018, Linaro Ltd.
+ * Author: Georgi Djakov <georgi.djakov@linaro.org>
+ */
+
+#include <linux/device.h>
+#include <linux/idr.h>
+#include <linux/init.h>
+#include <linux/interconnect.h>
+#include <linux/interconnect-provider.h>
+#include <linux/list.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/slab.h>
+
+static DEFINE_IDR(icc_idr);
+static LIST_HEAD(icc_provider_list);
+static DEFINE_MUTEX(icc_lock);
+
+/**
+ * struct icc_req - constraints that are attached to each node
+ *
+ * @req_node: entry in list of requests for the particular @node
+ * @node: the interconnect node to which this constraint applies
+ * @dev: reference to the device that sets the constraints
+ * @avg_bw: an integer describing the average bandwidth in kbps
+ * @peak_bw: an integer describing the peak bandwidth in kbps
+ */
+struct icc_req {
+	struct hlist_node req_node;
+	struct icc_node *node;
+	struct device *dev;
+	u32 avg_bw;
+	u32 peak_bw;
+};
+
+/**
+ * struct icc_path - interconnect path structure
+ * @num_nodes: number of hops (nodes)
+ * @reqs: array of the requests applicable to this path of nodes
+ */
+struct icc_path {
+	size_t num_nodes;
+	struct icc_req reqs[0];
+};
+
+static struct icc_node *node_find(const int id)
+{
+	struct icc_node *node;
+
+	mutex_lock(&icc_lock);
+	node = idr_find(&icc_idr, id);
+	mutex_unlock(&icc_lock);
+
+	return node;
+}
+
+static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
+{
+	struct icc_node *node = dst;
+	struct icc_path *path;
+	size_t i;
+
+	path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
+		       GFP_KERNEL);
+	if (!path)
+		return ERR_PTR(-ENOMEM);
+
+	path->num_nodes = num_nodes;
+
+	for (i = 0; i < num_nodes; i++) {
+		hlist_add_head(&path->reqs[i].req_node, &node->req_list);
+
+		path->reqs[i].node = node;
+		/* reference to previous node was saved during path traversal */
+		node = node->reverse;
+	}
+
+	return path;
+}
+
+static struct icc_path *path_find(struct device *dev, struct icc_node *src,
+				  struct icc_node *dst)
+{
+	struct icc_node *n, *node = NULL;
+	struct icc_provider *provider;
+	struct list_head traverse_list;
+	struct list_head edge_list;
+	struct list_head visited_list;
+	size_t i, depth = 0;
+	bool found = false;
+	int ret = -EPROBE_DEFER;
+
+	INIT_LIST_HEAD(&traverse_list);
+	INIT_LIST_HEAD(&edge_list);
+	INIT_LIST_HEAD(&visited_list);
+
+	list_add_tail(&src->search_list, &traverse_list);
+	src->reverse = NULL;
+
+	do {
+		list_for_each_entry_safe(node, n, &traverse_list, search_list) {
+			if (node == dst) {
+				found = true;
+				list_add(&node->search_list, &visited_list);
+				break;
+			}
+			for (i = 0; i < node->num_links; i++) {
+				struct icc_node *tmp = node->links[i];
+
+				if (!tmp) {
+					ret = -ENOENT;
+					goto out;
+				}
+
+				if (tmp->is_traversed)
+					continue;
+
+				tmp->is_traversed = true;
+				tmp->reverse = node;
+				list_add(&tmp->search_list, &edge_list);
+			}
+		}
+		if (found)
+			break;
+
+		list_splice_init(&traverse_list, &visited_list);
+		list_splice_init(&edge_list, &traverse_list);
+
+		/* count the hops away from the source */
+		depth++;
+
+	} while (!list_empty(&traverse_list));
+
+out:
+	/* reset the traversed state */
+	list_for_each_entry(provider, &icc_provider_list, provider_list) {
+		list_for_each_entry(n, &provider->nodes, node_list)
+			if (n->is_traversed)
+				n->is_traversed = false;
+	}
+
+	if (found) {
+		struct icc_path *path = path_allocate(dst, depth);
+
+		if (IS_ERR(path))
+			return path;
+
+		/* initialize the path */
+		for (i = 0; i < path->num_nodes; i++) {
+			node = path->reqs[i].node;
+			path->reqs[i].dev = dev;
+			node->provider->users++;
+		}
+		return path;
+	}
+
+	return ERR_PTR(ret);
+}
+
+/*
+ * We want the path to honor all bandwidth requests, so the average
+ * bandwidth requirements from each consumer are aggregated at each node
+ * and provider level. By default the average bandwidth is the sum of all
+ * averages and the peak will be the highest of all peak bandwidth requests.
+ */
+
+static int aggregate_requests(struct icc_node *node)
+{
+	struct icc_provider *p = node->provider;
+	struct icc_req *r;
+
+	node->avg_bw = 0;
+	node->peak_bw = 0;
+
+	hlist_for_each_entry(r, &node->req_list, req_node)
+		p->aggregate(node, r->avg_bw, r->peak_bw,
+			     &node->avg_bw, &node->peak_bw);
+
+	return 0;
+}
+
+static void aggregate_provider(struct icc_provider *p)
+{
+	struct icc_node *n;
+
+	p->avg_bw = 0;
+	p->peak_bw = 0;
+
+	list_for_each_entry(n, &p->nodes, node_list)
+		p->aggregate(n, n->avg_bw, n->peak_bw,
+			     &p->avg_bw, &p->peak_bw);
+}
+
+static int apply_constraints(struct icc_path *path)
+{
+	struct icc_node *next, *prev = NULL;
+	int ret = 0;
+	int i;
+
+	for (i = 0; i < path->num_nodes; i++, prev = next) {
+		struct icc_provider *p;
+
+		next = path->reqs[i].node;
+		/*
+		 * Both endpoints should be valid master-slave pairs of the
+		 * same interconnect provider that will be configured.
+		 */
+		if (!prev || next->provider != prev->provider)
+			continue;
+
+		p = next->provider;
+
+		aggregate_provider(p);
+
+		if (p->set) {
+			/* set the constraints */
+			ret = p->set(prev, next, p->avg_bw, p->peak_bw);
+		}
+
+		if (ret)
+			goto out;
+	}
+out:
+	return ret;
+}
+
+/**
+ * icc_set() - set constraints on an interconnect path between two endpoints
+ * @path: reference to the path returned by icc_get()
+ * @avg_bw: average bandwidth in kbps
+ * @peak_bw: peak bandwidth in kbps
+ *
+ * This function is used by an interconnect consumer to express its own needs
+ * in terms of bandwidth for a previously requested path between two endpoints.
+ * The requests are aggregated and each node is updated accordingly. The entire
+ * path is locked by a mutex to ensure that the set() is completed.
+ * The @path can be NULL when the "interconnects" DT properties is missing,
+ * which will mean that no constraints will be set.
+ *
+ * Returns 0 on success, or an appropriate error code otherwise.
+ */
+int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
+{
+	struct icc_node *node;
+	struct icc_provider *p;
+	size_t i;
+	int ret = 0;
+
+	if (!path)
+		return 0;
+
+	mutex_lock(&icc_lock);
+
+	for (i = 0; i < path->num_nodes; i++) {
+		node = path->reqs[i].node;
+		p = node->provider;
+
+		/* update the consumer request for this path */
+		path->reqs[i].avg_bw = avg_bw;
+		path->reqs[i].peak_bw = peak_bw;
+
+		/* aggregate requests for this node */
+		aggregate_requests(node);
+	}
+
+	ret = apply_constraints(path);
+	if (ret)
+		pr_err("interconnect: error applying constraints (%d)", ret);
+
+	mutex_unlock(&icc_lock);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(icc_set);
+
+/**
+ * icc_get() - return a handle for path between two endpoints
+ * @dev: the device requesting the path
+ * @src_id: source device port id
+ * @dst_id: destination device port id
+ *
+ * This function will search for a path between two endpoints and return an
+ * icc_path handle on success. Use icc_put() to release
+ * constraints when the they are not needed anymore.
+ *
+ * Return: icc_path pointer on success, or ERR_PTR() on error
+ */
+struct icc_path *icc_get(struct device *dev, const int src_id, const int dst_id)
+{
+	struct icc_node *src, *dst;
+	struct icc_path *path = ERR_PTR(-EPROBE_DEFER);
+
+	src = node_find(src_id);
+	if (!src) {
+		dev_err(dev, "%s: invalid src=%d\n", __func__, src_id);
+		goto out;
+	}
+
+	dst = node_find(dst_id);
+	if (!dst) {
+		dev_err(dev, "%s: invalid dst=%d\n", __func__, dst_id);
+		goto out;
+	}
+
+	mutex_lock(&icc_lock);
+	path = path_find(dev, src, dst);
+	mutex_unlock(&icc_lock);
+	if (IS_ERR(path)) {
+		dev_err(dev, "%s: invalid path=%ld\n", __func__, PTR_ERR(path));
+		goto out;
+	}
+
+out:
+	return path;
+}
+EXPORT_SYMBOL_GPL(icc_get);
+
+/**
+ * icc_put() - release the reference to the icc_path
+ * @path: interconnect path
+ *
+ * Use this function to release the constraints on a path when the path is
+ * no longer needed. The constraints will be re-aggregated.
+ */
+void icc_put(struct icc_path *path)
+{
+	struct icc_node *node;
+	size_t i;
+	int ret;
+
+	if (!path || WARN_ON_ONCE(IS_ERR(path)))
+		return;
+
+	ret = icc_set(path, 0, 0);
+	if (ret)
+		pr_err("%s: error (%d)\n", __func__, ret);
+
+	mutex_lock(&icc_lock);
+	for (i = 0; i < path->num_nodes; i++) {
+		node = path->reqs[i].node;
+		hlist_del(&path->reqs[i].req_node);
+
+		node->provider->users--;
+	}
+	mutex_unlock(&icc_lock);
+
+	kfree(path);
+}
+EXPORT_SYMBOL_GPL(icc_put);
+
+/**
+ * icc_node_create() - create a node
+ * @id: node id
+ *
+ * Return: icc_node pointer on success, or ERR_PTR() on error
+ */
+struct icc_node *icc_node_create(int id)
+{
+	struct icc_node *node;
+
+	/* check if node already exists */
+	node = node_find(id);
+	if (node)
+		goto out;
+
+	node = kzalloc(sizeof(*node), GFP_KERNEL);
+	if (!node) {
+		node = ERR_PTR(-ENOMEM);
+		goto out;
+	}
+
+	mutex_lock(&icc_lock);
+
+	id = idr_alloc(&icc_idr, node, id, id + 1, GFP_KERNEL);
+	if (WARN(id < 0, "couldn't get idr")) {
+		node = ERR_PTR(id);
+		goto out;
+	}
+
+	node->id = id;
+
+out:
+	mutex_unlock(&icc_lock);
+
+	return node;
+}
+EXPORT_SYMBOL_GPL(icc_node_create);
+
+/**
+ * icc_node_remove() - remove a node
+ * @id: node id
+ *
+ */
+void icc_node_remove(int id)
+{
+	struct icc_node *node;
+
+	node = node_find(id);
+	if (node) {
+		mutex_lock(&icc_lock);
+		idr_remove(&icc_idr, node->id);
+		mutex_unlock(&icc_lock);
+	}
+
+	kfree(node);
+}
+EXPORT_SYMBOL_GPL(icc_node_remove);
+
+/**
+ * icc_link_create() - create a link between two nodes
+ * @src_id: source node id
+ * @dst_id: destination node id
+ *
+ * Create a link between two nodes. The nodes might belong to different
+ * interconnect providers and the @dst_id node might not exist (if the
+ * provider driver has not probed yet). So just create the @dst_id node
+ * and when the actual provider driver is probed, the rest of the node
+ * data is filled.
+ *
+ * Return: 0 on success, or an error code otherwise
+ */
+int icc_link_create(struct icc_node *node, const int dst_id)
+{
+	struct icc_node *dst;
+	struct icc_node **new;
+	int ret = 0;
+
+	if (!node->provider)
+		return -EINVAL;
+
+	dst = node_find(dst_id);
+	if (!dst) {
+		dst = icc_node_create(dst_id);
+
+		if (IS_ERR(dst)) {
+			ret = PTR_ERR(dst);
+			goto out;
+		}
+	}
+
+	mutex_lock(&icc_lock);
+
+	new = krealloc(node->links,
+		       (node->num_links + 1) * sizeof(*node->links),
+		       GFP_KERNEL);
+	if (!new) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	node->links = new;
+	node->links[node->num_links++] = dst;
+
+out:
+	mutex_unlock(&icc_lock);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(icc_link_create);
+
+/**
+ * icc_link_remove() - remove a link between two nodes
+ * @src: pointer to source node
+ * @dst: pointer to destination node
+ *
+ * Return: 0 on success, or an error code otherwise
+ */
+int icc_link_remove(struct icc_node *src, struct icc_node *dst)
+{
+	struct icc_node **new;
+	int ret = 0;
+	int i, j;
+
+	if (IS_ERR_OR_NULL(src))
+		return PTR_ERR(src);
+
+	if (IS_ERR_OR_NULL(dst))
+		return PTR_ERR(dst);
+
+	mutex_lock(&icc_lock);
+
+	new = krealloc(src->links,
+		       (src->num_links - 1) * sizeof(*src->links),
+		       GFP_KERNEL);
+	if (!new) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	for (i = 0, j = 0; j < src->num_links; j++) {
+		if (src->links[j] == dst)
+			continue;
+
+		new[i++] = src->links[j];
+	}
+
+	src->links = new;
+	src->num_links--;
+
+out:
+	mutex_unlock(&icc_lock);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(icc_link_remove);
+
+/**
+ * icc_node_add() - add an interconnect node to interconnect provider
+ * @node: pointer to the interconnect node
+ * @provider: pointer to the interconnect provider
+ *
+ * Return: 0 on success, or an error code otherwise
+ */
+int icc_node_add(struct icc_node *node, struct icc_provider *provider)
+{
+	mutex_lock(&icc_lock);
+
+	node->provider = provider;
+	list_add(&node->node_list, &provider->nodes);
+
+	mutex_unlock(&icc_lock);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(icc_node_add);
+
+/**
+ * icc_provider_add() - add a new interconnect provider
+ * @icc_provider: the interconnect provider that will be added into topology
+ *
+ * Return: 0 on success, or an error code otherwise
+ */
+int icc_provider_add(struct icc_provider *provider)
+{
+	if (WARN_ON(!provider->set))
+		return -EINVAL;
+
+	mutex_init(&icc_lock);
+
+	INIT_LIST_HEAD(&provider->nodes);
+	list_add(&provider->provider_list, &icc_provider_list);
+
+	mutex_unlock(&icc_lock);
+
+	dev_dbg(provider->dev, "interconnect provider added to topology\n");
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(icc_provider_add);
+
+/**
+ * icc_provider_del() - delete previously added interconnect provider
+ * @icc_provider: the interconnect provider that will be removed from topology
+ *
+ * Return: 0 on success, or an error code otherwise
+ */
+int icc_provider_del(struct icc_provider *provider)
+{
+	mutex_lock(&icc_lock);
+	if (provider->users) {
+		pr_warn("interconnect provider still has %d users\n",
+			provider->users);
+		mutex_unlock(&icc_lock);
+		return -EBUSY;
+	}
+
+	if (!list_empty_careful(&provider->nodes)) {
+		pr_warn("interconnect provider still has nodes\n");
+		mutex_unlock(&icc_lock);
+		return -EEXIST;
+	}
+
+	list_del(&provider->provider_list);
+	mutex_unlock(&icc_lock);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(icc_provider_del);
+
+MODULE_AUTHOR("Georgi Djakov <georgi.djakov at linaro.org");
+MODULE_DESCRIPTION("Interconnect Driver Core");
+MODULE_LICENSE("GPL v2");
diff --git a/include/linux/interconnect-provider.h b/include/linux/interconnect-provider.h
new file mode 100644
index 000000000000..f4613c6dce4f
--- /dev/null
+++ b/include/linux/interconnect-provider.h
@@ -0,0 +1,127 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2018, Linaro Ltd.
+ * Author: Georgi Djakov <georgi.djakov@linaro.org>
+ */
+
+#ifndef _LINUX_INTERCONNECT_PROVIDER_H
+#define _LINUX_INTERCONNECT_PROVIDER_H
+
+#include <linux/interconnect.h>
+
+#define icc_units_to_bps(bw)  ((bw) * 1000ULL)
+
+struct icc_node;
+
+/**
+ * struct icc_provider - interconnect provider (controller) entity that might
+ * provide multiple interconnect controls
+ *
+ * @provider_list: list of the registered interconnect providers
+ * @nodes: internal list of the interconnect provider nodes
+ * @set: pointer to device specific set operation function
+ * @aggregate: pointer to device specific aggregate operation function
+ * @dev: the device this interconnect provider belongs to
+ * @users: count of active users
+ * @avg_bw: aggregated value of average bandwidth requests from all nodes
+ * @peak_bw: aggregated value of peak bandwidth requests from all nodes
+ * @data: pointer to private data
+ */
+struct icc_provider {
+	struct list_head	provider_list;
+	struct list_head	nodes;
+	int (*set)(struct icc_node *src, struct icc_node *dst,
+		   u32 avg_bw, u32 peak_bw);
+	int (*aggregate)(struct icc_node *node, u32 avg_bw, u32 peak_bw,
+			 u32 *agg_avg, u32 *agg_peak);
+	struct device		*dev;
+	int			users;
+	u32			avg_bw;
+	u32			peak_bw;
+	void			*data;
+};
+
+/**
+ * struct icc_node - entity that is part of the interconnect topology
+ *
+ * @id: platform specific node id
+ * @name: node name used in debugfs
+ * @links: a list of targets pointing to where we can go next when traversing
+ * @num_links: number of links to other interconnect nodes
+ * @provider: points to the interconnect provider of this node
+ * @node_list: the list entry in the parent provider's "nodes" list
+ * @search_list: list used when walking the nodes graph
+ * @reverse: pointer to previous node when walking the nodes graph
+ * @is_traversed: flag that is used when walking the nodes graph
+ * @req_list: a list of QoS constraint requests associated with this node
+ * @avg_bw: aggregated value of average bandwidth requests from all consumers
+ * @peak_bw: aggregated value of peak bandwidth requests from all consumers
+ * @data: pointer to private data
+ */
+struct icc_node {
+	int			id;
+	const char              *name;
+	struct icc_node		**links;
+	size_t			num_links;
+
+	struct icc_provider	*provider;
+	struct list_head	node_list;
+	struct list_head	orphan_list;
+	struct list_head	search_list;
+	struct icc_node		*reverse;
+	bool			is_traversed;
+	struct hlist_head	req_list;
+	u32			avg_bw;
+	u32			peak_bw;
+	void			*data;
+};
+
+#if IS_ENABLED(CONFIG_INTERCONNECT)
+
+struct icc_node *icc_node_create(int id);
+void icc_node_remove(int id);
+int icc_link_create(struct icc_node *node, const int dst_id);
+int icc_link_remove(struct icc_node *src, struct icc_node *dst);
+int icc_node_add(struct icc_node *node, struct icc_provider *provider);
+int icc_provider_add(struct icc_provider *provider);
+int icc_provider_del(struct icc_provider *provider);
+
+#else
+
+static inline struct icc_node *icc_node_create(int id)
+{
+	return ERR_PTR(-ENOTSUPP);
+}
+
+void icc_node_remove(int id)
+{
+}
+
+static inline int icc_link_create(struct icc_node *node, const int dst_id)
+{
+	return -ENOTSUPP;
+}
+
+int icc_link_remove(struct icc_node *src, struct icc_node *dst)
+{
+	return -ENOTSUPP;
+}
+
+int icc_node_add(struct icc_node *node, struct icc_provider *provider)
+{
+	return -ENOTSUPP;
+}
+
+static inline int icc_provider_add(struct icc_provider *provider)
+{
+	return -ENOTSUPP;
+}
+
+static inline int icc_provider_del(struct icc_provider *provider)
+{
+	return -ENOTSUPP;
+}
+
+#endif /* CONFIG_INTERCONNECT */
+
+#endif /* _LINUX_INTERCONNECT_PROVIDER_H */
diff --git a/include/linux/interconnect.h b/include/linux/interconnect.h
new file mode 100644
index 000000000000..593215371fd6
--- /dev/null
+++ b/include/linux/interconnect.h
@@ -0,0 +1,42 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2018, Linaro Ltd.
+ * Author: Georgi Djakov <georgi.djakov@linaro.org>
+ */
+
+#ifndef _LINUX_INTERCONNECT_H
+#define _LINUX_INTERCONNECT_H
+
+#include <linux/mutex.h>
+#include <linux/types.h>
+
+struct icc_path;
+struct device;
+
+#if IS_ENABLED(CONFIG_INTERCONNECT)
+
+struct icc_path *icc_get(struct device *dev, const int src_id,
+			 const int dst_id);
+void icc_put(struct icc_path *path);
+int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw);
+
+#else
+
+static inline struct icc_path *icc_get(struct device *dev, const int src_id,
+				       const int dst_id)
+{
+	return NULL;
+}
+
+static inline void icc_put(struct icc_path *path)
+{
+}
+
+static inline int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
+{
+	return 0;
+}
+
+#endif /* CONFIG_INTERCONNECT */
+
+#endif /* _LINUX_INTERCONNECT_H */

[PATCH v5 2/8] dt-bindings: Introduce interconnect provider bindings

[Georgi Djakov]

This binding is intended to represent the interconnect hardware present
in some of the modern SoCs. Currently it consists only of a binding for
the interconnect hardware devices (provider).

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 .../bindings/interconnect/interconnect.txt    | 33 +++++++++++++++++++
 1 file changed, 33 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/interconnect/interconnect.txt

diff --git a/Documentation/devicetree/bindings/interconnect/interconnect.txt b/Documentation/devicetree/bindings/interconnect/interconnect.txt
new file mode 100644
index 000000000000..6e2b2971b094
--- /dev/null
+++ b/Documentation/devicetree/bindings/interconnect/interconnect.txt
@@ -0,0 +1,33 @@
+Interconnect Provider Device Tree Bindings
+=========================================
+
+The purpose of this document is to define a common set of generic interconnect
+providers/consumers properties.
+
+
+= interconnect providers =
+
+The interconnect provider binding is intended to represent the interconnect
+controllers in the system. Each provider registers a set of interconnect
+nodes, which expose the interconnect related capabilities of the interconnect
+to consumer drivers. These capabilities can be throughput, latency, priority
+etc. The consumer drivers set constraints on interconnect path (or endpoints)
+depending on the use case. Interconnect providers can also be interconnect
+consumers, such as in the case where two network-on-chip fabrics interface
+directly
+
+Required properties:
+- compatible : contains the interconnect provider compatible string
+- #interconnect-cells : number of cells in a interconnect specifier needed to
+			encode the interconnect node id
+
+Example:
+
+		snoc: snoc@580000 {
+			compatible = "qcom,msm8916-snoc";
+			#interconnect-cells = <1>;
+			reg = <0x580000 0x14000>;
+			clock-names = "bus_clk", "bus_a_clk";
+			clocks = <&rpmcc RPM_SMD_SNOC_CLK>,
+				 <&rpmcc RPM_SMD_SNOC_A_CLK>;
+		};

[PATCH v5 2/8] dt-bindings: Introduce interconnect provider bindings

[Georgi Djakov]

This binding is intended to represent the interconnect hardware present
in some of the modern SoCs. Currently it consists only of a binding for
the interconnect hardware devices (provider).

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 .../bindings/interconnect/interconnect.txt    | 33 +++++++++++++++++++
 1 file changed, 33 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/interconnect/interconnect.txt

diff --git a/Documentation/devicetree/bindings/interconnect/interconnect.txt b/Documentation/devicetree/bindings/interconnect/interconnect.txt
new file mode 100644
index 000000000000..6e2b2971b094
--- /dev/null
+++ b/Documentation/devicetree/bindings/interconnect/interconnect.txt
@@ -0,0 +1,33 @@
+Interconnect Provider Device Tree Bindings
+=========================================
+
+The purpose of this document is to define a common set of generic interconnect
+providers/consumers properties.
+
+
+= interconnect providers =
+
+The interconnect provider binding is intended to represent the interconnect
+controllers in the system. Each provider registers a set of interconnect
+nodes, which expose the interconnect related capabilities of the interconnect
+to consumer drivers. These capabilities can be throughput, latency, priority
+etc. The consumer drivers set constraints on interconnect path (or endpoints)
+depending on the use case. Interconnect providers can also be interconnect
+consumers, such as in the case where two network-on-chip fabrics interface
+directly
+
+Required properties:
+- compatible : contains the interconnect provider compatible string
+- #interconnect-cells : number of cells in a interconnect specifier needed to
+			encode the interconnect node id
+
+Example:
+
+		snoc: snoc at 580000 {
+			compatible = "qcom,msm8916-snoc";
+			#interconnect-cells = <1>;
+			reg = <0x580000 0x14000>;
+			clock-names = "bus_clk", "bus_a_clk";
+			clocks = <&rpmcc RPM_SMD_SNOC_CLK>,
+				 <&rpmcc RPM_SMD_SNOC_A_CLK>;
+		};

[PATCH v5 3/8] interconnect: Add debugfs support

[Georgi Djakov]

Add a functionality to provide information about the current constraints
per each node and provider.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 drivers/interconnect/core.c | 78 +++++++++++++++++++++++++++++++++++++
 1 file changed, 78 insertions(+)

diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
index e7f96fc6722e..dafc001ce758 100644
--- a/drivers/interconnect/core.c
+++ b/drivers/interconnect/core.c
@@ -6,6 +6,7 @@
  * Author: Georgi Djakov <georgi.djakov@linaro.org>
  */
 
+#include <linux/debugfs.h>
 #include <linux/device.h>
 #include <linux/idr.h>
 #include <linux/init.h>
@@ -15,10 +16,12 @@
 #include <linux/module.h>
 #include <linux/mutex.h>
 #include <linux/slab.h>
+#include <linux/uaccess.h>
 
 static DEFINE_IDR(icc_idr);
 static LIST_HEAD(icc_provider_list);
 static DEFINE_MUTEX(icc_lock);
+static struct dentry *icc_debugfs_dir;
 
 /**
  * struct icc_req - constraints that are attached to each node
@@ -47,6 +50,81 @@ struct icc_path {
 	struct icc_req reqs[0];
 };
 
+#ifdef CONFIG_DEBUG_FS
+
+static void icc_summary_show_one(struct seq_file *s, struct icc_node *n)
+{
+	if (!n)
+		return;
+
+	seq_printf(s, "%-30s %12d %12d\n",
+		   n->name, n->avg_bw, n->peak_bw);
+}
+
+static int icc_summary_show(struct seq_file *s, void *data)
+{
+	struct icc_provider *provider;
+
+	seq_puts(s, " node                                   avg         peak\n");
+	seq_puts(s, "--------------------------------------------------------\n");
+
+	mutex_lock(&icc_lock);
+
+	list_for_each_entry(provider, &icc_provider_list, provider_list) {
+		struct icc_node *n;
+
+		list_for_each_entry(n, &provider->nodes, node_list) {
+			struct icc_req *r;
+
+			icc_summary_show_one(s, n);
+			hlist_for_each_entry(r, &n->req_list, req_node) {
+				if (!r->dev)
+					continue;
+
+				seq_printf(s, "    %-26s %12d %12d\n",
+					   dev_name(r->dev), r->avg_bw,
+					   r->peak_bw);
+			}
+		}
+	}
+
+	mutex_unlock(&icc_lock);
+
+	return 0;
+}
+
+static int icc_summary_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, icc_summary_show, inode->i_private);
+}
+
+static const struct file_operations icc_summary_fops = {
+	.open		= icc_summary_open,
+	.read		= seq_read,
+	.llseek		= seq_lseek,
+	.release	= single_release,
+};
+
+static int __init icc_debugfs_init(void)
+{
+	struct dentry *file;
+
+	icc_debugfs_dir = debugfs_create_dir("interconnect", NULL);
+	if (!icc_debugfs_dir) {
+		pr_err("interconnect: error creating debugfs directory\n");
+		return -ENODEV;
+	}
+
+	file = debugfs_create_file("interconnect_summary", 0444,
+				   icc_debugfs_dir, NULL, &icc_summary_fops);
+	if (!file)
+		return -ENODEV;
+
+	return 0;
+}
+late_initcall(icc_debugfs_init);
+#endif
+
 static struct icc_node *node_find(const int id)
 {
 	struct icc_node *node;

[PATCH v5 3/8] interconnect: Add debugfs support

[Georgi Djakov]

Add a functionality to provide information about the current constraints
per each node and provider.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 drivers/interconnect/core.c | 78 +++++++++++++++++++++++++++++++++++++
 1 file changed, 78 insertions(+)

diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
index e7f96fc6722e..dafc001ce758 100644
--- a/drivers/interconnect/core.c
+++ b/drivers/interconnect/core.c
@@ -6,6 +6,7 @@
  * Author: Georgi Djakov <georgi.djakov@linaro.org>
  */
 
+#include <linux/debugfs.h>
 #include <linux/device.h>
 #include <linux/idr.h>
 #include <linux/init.h>
@@ -15,10 +16,12 @@
 #include <linux/module.h>
 #include <linux/mutex.h>
 #include <linux/slab.h>
+#include <linux/uaccess.h>
 
 static DEFINE_IDR(icc_idr);
 static LIST_HEAD(icc_provider_list);
 static DEFINE_MUTEX(icc_lock);
+static struct dentry *icc_debugfs_dir;
 
 /**
  * struct icc_req - constraints that are attached to each node
@@ -47,6 +50,81 @@ struct icc_path {
 	struct icc_req reqs[0];
 };
 
+#ifdef CONFIG_DEBUG_FS
+
+static void icc_summary_show_one(struct seq_file *s, struct icc_node *n)
+{
+	if (!n)
+		return;
+
+	seq_printf(s, "%-30s %12d %12d\n",
+		   n->name, n->avg_bw, n->peak_bw);
+}
+
+static int icc_summary_show(struct seq_file *s, void *data)
+{
+	struct icc_provider *provider;
+
+	seq_puts(s, " node                                   avg         peak\n");
+	seq_puts(s, "--------------------------------------------------------\n");
+
+	mutex_lock(&icc_lock);
+
+	list_for_each_entry(provider, &icc_provider_list, provider_list) {
+		struct icc_node *n;
+
+		list_for_each_entry(n, &provider->nodes, node_list) {
+			struct icc_req *r;
+
+			icc_summary_show_one(s, n);
+			hlist_for_each_entry(r, &n->req_list, req_node) {
+				if (!r->dev)
+					continue;
+
+				seq_printf(s, "    %-26s %12d %12d\n",
+					   dev_name(r->dev), r->avg_bw,
+					   r->peak_bw);
+			}
+		}
+	}
+
+	mutex_unlock(&icc_lock);
+
+	return 0;
+}
+
+static int icc_summary_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, icc_summary_show, inode->i_private);
+}
+
+static const struct file_operations icc_summary_fops = {
+	.open		= icc_summary_open,
+	.read		= seq_read,
+	.llseek		= seq_lseek,
+	.release	= single_release,
+};
+
+static int __init icc_debugfs_init(void)
+{
+	struct dentry *file;
+
+	icc_debugfs_dir = debugfs_create_dir("interconnect", NULL);
+	if (!icc_debugfs_dir) {
+		pr_err("interconnect: error creating debugfs directory\n");
+		return -ENODEV;
+	}
+
+	file = debugfs_create_file("interconnect_summary", 0444,
+				   icc_debugfs_dir, NULL, &icc_summary_fops);
+	if (!file)
+		return -ENODEV;
+
+	return 0;
+}
+late_initcall(icc_debugfs_init);
+#endif
+
 static struct icc_node *node_find(const int id)
 {
 	struct icc_node *node;

[PATCH v5 4/8] interconnect: qcom: Add RPM communication

[Georgi Djakov]

On some Qualcomm SoCs, there is a remote processor, which controls some of
the Network-On-Chip interconnect resources. Other CPUs express their needs
by communicating with this processor. Add a driver to handle comminication
with this remote processor.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 .../bindings/interconnect/qcom-smd.txt        | 32 +++++++
 drivers/interconnect/qcom/Makefile            |  2 +
 drivers/interconnect/qcom/smd-rpm.c           | 90 +++++++++++++++++++
 drivers/interconnect/qcom/smd-rpm.h           | 15 ++++
 4 files changed, 139 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/interconnect/qcom-smd.txt
 create mode 100644 drivers/interconnect/qcom/Makefile
 create mode 100644 drivers/interconnect/qcom/smd-rpm.c
 create mode 100644 drivers/interconnect/qcom/smd-rpm.h

diff --git a/Documentation/devicetree/bindings/interconnect/qcom-smd.txt b/Documentation/devicetree/bindings/interconnect/qcom-smd.txt
new file mode 100644
index 000000000000..88a5aeb50935
--- /dev/null
+++ b/Documentation/devicetree/bindings/interconnect/qcom-smd.txt
@@ -0,0 +1,32 @@
+Qualcomm SMD-RPM interconnect driver binding
+------------------------------------------------
+The RPM (Resource Power Manager) is a dedicated hardware engine
+for managing the shared SoC resources in order to keep the lowest
+power profile. It communicates with other hardware subsystems via
+the shared memory driver (SMD) back-end and accepts requests for
+various resources.
+
+Required properties :
+- compatible : shall contain only one of the following:
+			"qcom,interconnect-smd-rpm"
+
+Example:
+	smd {
+		compatible = "qcom,smd";
+
+		rpm {
+			interrupts = <0 168 1>;
+			qcom,ipc = <&apcs 8 0>;
+			qcom,smd-edge = <15>;
+
+			rpm_requests {
+				compatible = "qcom,rpm-msm8916";
+				qcom,smd-channels = "rpm_requests";
+
+				interconnect-smd-rpm {
+					compatible = "qcom,interconnect-smd-rpm";
+				};
+
+			};
+		};
+	};
diff --git a/drivers/interconnect/qcom/Makefile b/drivers/interconnect/qcom/Makefile
new file mode 100644
index 000000000000..2a0c41db91c4
--- /dev/null
+++ b/drivers/interconnect/qcom/Makefile
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0
+obj-y += smd-rpm.o
diff --git a/drivers/interconnect/qcom/smd-rpm.c b/drivers/interconnect/qcom/smd-rpm.c
new file mode 100644
index 000000000000..0cf772f51642
--- /dev/null
+++ b/drivers/interconnect/qcom/smd-rpm.c
@@ -0,0 +1,90 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * RPM over SMD communication wrapper for interconects
+ *
+ * Copyright (C) 2018 Linaro Ltd
+ * Author: Georgi Djakov <georgi.djakov@linaro.org>
+ */
+
+#include <linux/interconnect-provider.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_platform.h>
+#include <linux/platform_device.h>
+#include <linux/soc/qcom/smd-rpm.h>
+#include "smd-rpm.h"
+
+#define	RPM_KEY_BW	0x00007762
+
+static struct qcom_icc_rpm {
+	struct qcom_smd_rpm *rpm;
+} icc_rpm_smd;
+
+struct icc_rpm_smd_req {
+	__le32 key;
+	__le32 nbytes;
+	__le32 value;
+};
+
+bool qcom_icc_rpm_smd_available(void)
+{
+	if (!icc_rpm_smd.rpm)
+		return false;
+
+	return true;
+}
+
+int qcom_icc_rpm_smd_send(int ctx, int rsc_type, int id, u32 val)
+{
+	struct icc_rpm_smd_req req = {
+		.key = cpu_to_le32(RPM_KEY_BW),
+		.nbytes = cpu_to_le32(sizeof(u32)),
+		.value = cpu_to_le32(val),
+	};
+
+	return qcom_rpm_smd_write(icc_rpm_smd.rpm, ctx, rsc_type, id, &req,
+				  sizeof(req));
+}
+EXPORT_SYMBOL(qcom_icc_rpm_smd_send);
+
+static int qcom_icc_rpm_smd_probe(struct platform_device *pdev)
+{
+	icc_rpm_smd.rpm = dev_get_drvdata(pdev->dev.parent);
+	if (!icc_rpm_smd.rpm) {
+		dev_err(&pdev->dev, "unable to retrieve handle to RPM\n");
+		return -ENODEV;
+	}
+
+	return 0;
+}
+
+static const struct of_device_id qcom_icc_rpm_smd_dt_match[] = {
+	{ .compatible = "qcom,interconnect-smd-rpm", },
+	{ },
+};
+
+MODULE_DEVICE_TABLE(of, qcom_interconnect_rpm_smd_dt_match);
+
+static struct platform_driver qcom_interconnect_rpm_smd_driver = {
+	.driver = {
+		.name		= "qcom-interconnect-smd-rpm",
+		.of_match_table	= qcom_icc_rpm_smd_dt_match,
+	},
+	.probe = qcom_icc_rpm_smd_probe,
+};
+
+static int __init rpm_smd_interconnect_init(void)
+{
+	return platform_driver_register(&qcom_interconnect_rpm_smd_driver);
+}
+subsys_initcall(rpm_smd_interconnect_init);
+
+static void __exit rpm_smd_interconnect_exit(void)
+{
+	platform_driver_unregister(&qcom_interconnect_rpm_smd_driver);
+}
+module_exit(rpm_smd_interconnect_exit)
+
+MODULE_AUTHOR("Georgi Djakov <georgi.djakov@linaro.org>");
+MODULE_DESCRIPTION("Qualcomm SMD RPM interconnect driver");
+MODULE_LICENSE("GPL v2");
diff --git a/drivers/interconnect/qcom/smd-rpm.h b/drivers/interconnect/qcom/smd-rpm.h
new file mode 100644
index 000000000000..0f4a3da31cf6
--- /dev/null
+++ b/drivers/interconnect/qcom/smd-rpm.h
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2018, Linaro Ltd.
+ * Author: Georgi Djakov <georgi.djakov@linaro.org>
+ */
+
+#ifndef __LINUX_INTERCONNECT_QCOM_RPM_H
+#define __LINUX_INTERCONNECT_QCOM_RPM_H
+
+#include <linux/soc/qcom/smd-rpm.h>
+
+bool qcom_icc_rpm_smd_available(void);
+int qcom_icc_rpm_smd_send(int ctx, int rsc_type, int id, u32 val);
+
+#endif

[PATCH v5 4/8] interconnect: qcom: Add RPM communication

[Georgi Djakov]

On some Qualcomm SoCs, there is a remote processor, which controls some of
the Network-On-Chip interconnect resources. Other CPUs express their needs
by communicating with this processor. Add a driver to handle comminication
with this remote processor.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 .../bindings/interconnect/qcom-smd.txt        | 32 +++++++
 drivers/interconnect/qcom/Makefile            |  2 +
 drivers/interconnect/qcom/smd-rpm.c           | 90 +++++++++++++++++++
 drivers/interconnect/qcom/smd-rpm.h           | 15 ++++
 4 files changed, 139 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/interconnect/qcom-smd.txt
 create mode 100644 drivers/interconnect/qcom/Makefile
 create mode 100644 drivers/interconnect/qcom/smd-rpm.c
 create mode 100644 drivers/interconnect/qcom/smd-rpm.h

diff --git a/Documentation/devicetree/bindings/interconnect/qcom-smd.txt b/Documentation/devicetree/bindings/interconnect/qcom-smd.txt
new file mode 100644
index 000000000000..88a5aeb50935
--- /dev/null
+++ b/Documentation/devicetree/bindings/interconnect/qcom-smd.txt
@@ -0,0 +1,32 @@
+Qualcomm SMD-RPM interconnect driver binding
+------------------------------------------------
+The RPM (Resource Power Manager) is a dedicated hardware engine
+for managing the shared SoC resources in order to keep the lowest
+power profile. It communicates with other hardware subsystems via
+the shared memory driver (SMD) back-end and accepts requests for
+various resources.
+
+Required properties :
+- compatible : shall contain only one of the following:
+			"qcom,interconnect-smd-rpm"
+
+Example:
+	smd {
+		compatible = "qcom,smd";
+
+		rpm {
+			interrupts = <0 168 1>;
+			qcom,ipc = <&apcs 8 0>;
+			qcom,smd-edge = <15>;
+
+			rpm_requests {
+				compatible = "qcom,rpm-msm8916";
+				qcom,smd-channels = "rpm_requests";
+
+				interconnect-smd-rpm {
+					compatible = "qcom,interconnect-smd-rpm";
+				};
+
+			};
+		};
+	};
diff --git a/drivers/interconnect/qcom/Makefile b/drivers/interconnect/qcom/Makefile
new file mode 100644
index 000000000000..2a0c41db91c4
--- /dev/null
+++ b/drivers/interconnect/qcom/Makefile
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0
+obj-y += smd-rpm.o
diff --git a/drivers/interconnect/qcom/smd-rpm.c b/drivers/interconnect/qcom/smd-rpm.c
new file mode 100644
index 000000000000..0cf772f51642
--- /dev/null
+++ b/drivers/interconnect/qcom/smd-rpm.c
@@ -0,0 +1,90 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * RPM over SMD communication wrapper for interconects
+ *
+ * Copyright (C) 2018 Linaro Ltd
+ * Author: Georgi Djakov <georgi.djakov@linaro.org>
+ */
+
+#include <linux/interconnect-provider.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_platform.h>
+#include <linux/platform_device.h>
+#include <linux/soc/qcom/smd-rpm.h>
+#include "smd-rpm.h"
+
+#define	RPM_KEY_BW	0x00007762
+
+static struct qcom_icc_rpm {
+	struct qcom_smd_rpm *rpm;
+} icc_rpm_smd;
+
+struct icc_rpm_smd_req {
+	__le32 key;
+	__le32 nbytes;
+	__le32 value;
+};
+
+bool qcom_icc_rpm_smd_available(void)
+{
+	if (!icc_rpm_smd.rpm)
+		return false;
+
+	return true;
+}
+
+int qcom_icc_rpm_smd_send(int ctx, int rsc_type, int id, u32 val)
+{
+	struct icc_rpm_smd_req req = {
+		.key = cpu_to_le32(RPM_KEY_BW),
+		.nbytes = cpu_to_le32(sizeof(u32)),
+		.value = cpu_to_le32(val),
+	};
+
+	return qcom_rpm_smd_write(icc_rpm_smd.rpm, ctx, rsc_type, id, &req,
+				  sizeof(req));
+}
+EXPORT_SYMBOL(qcom_icc_rpm_smd_send);
+
+static int qcom_icc_rpm_smd_probe(struct platform_device *pdev)
+{
+	icc_rpm_smd.rpm = dev_get_drvdata(pdev->dev.parent);
+	if (!icc_rpm_smd.rpm) {
+		dev_err(&pdev->dev, "unable to retrieve handle to RPM\n");
+		return -ENODEV;
+	}
+
+	return 0;
+}
+
+static const struct of_device_id qcom_icc_rpm_smd_dt_match[] = {
+	{ .compatible = "qcom,interconnect-smd-rpm", },
+	{ },
+};
+
+MODULE_DEVICE_TABLE(of, qcom_interconnect_rpm_smd_dt_match);
+
+static struct platform_driver qcom_interconnect_rpm_smd_driver = {
+	.driver = {
+		.name		= "qcom-interconnect-smd-rpm",
+		.of_match_table	= qcom_icc_rpm_smd_dt_match,
+	},
+	.probe = qcom_icc_rpm_smd_probe,
+};
+
+static int __init rpm_smd_interconnect_init(void)
+{
+	return platform_driver_register(&qcom_interconnect_rpm_smd_driver);
+}
+subsys_initcall(rpm_smd_interconnect_init);
+
+static void __exit rpm_smd_interconnect_exit(void)
+{
+	platform_driver_unregister(&qcom_interconnect_rpm_smd_driver);
+}
+module_exit(rpm_smd_interconnect_exit)
+
+MODULE_AUTHOR("Georgi Djakov <georgi.djakov@linaro.org>");
+MODULE_DESCRIPTION("Qualcomm SMD RPM interconnect driver");
+MODULE_LICENSE("GPL v2");
diff --git a/drivers/interconnect/qcom/smd-rpm.h b/drivers/interconnect/qcom/smd-rpm.h
new file mode 100644
index 000000000000..0f4a3da31cf6
--- /dev/null
+++ b/drivers/interconnect/qcom/smd-rpm.h
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2018, Linaro Ltd.
+ * Author: Georgi Djakov <georgi.djakov@linaro.org>
+ */
+
+#ifndef __LINUX_INTERCONNECT_QCOM_RPM_H
+#define __LINUX_INTERCONNECT_QCOM_RPM_H
+
+#include <linux/soc/qcom/smd-rpm.h>
+
+bool qcom_icc_rpm_smd_available(void);
+int qcom_icc_rpm_smd_send(int ctx, int rsc_type, int id, u32 val);
+
+#endif

[PATCH v5 5/8] dt-bindings: interconnect: Document qcom,msm8916 NoC bindings

[Georgi Djakov]

Document the device-tree bindings Network-On-Chip interconnect driver
for Qualcomm msm8916 platforms.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 .../bindings/interconnect/qcom-msm8916.txt    |  39 ++
 include/dt-bindings/interconnect/qcom.h       | 350 ++++++++++++++++++
 2 files changed, 389 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/interconnect/qcom-msm8916.txt
 create mode 100644 include/dt-bindings/interconnect/qcom.h

diff --git a/Documentation/devicetree/bindings/interconnect/qcom-msm8916.txt b/Documentation/devicetree/bindings/interconnect/qcom-msm8916.txt
new file mode 100644
index 000000000000..f309eaed3d19
--- /dev/null
+++ b/Documentation/devicetree/bindings/interconnect/qcom-msm8916.txt
@@ -0,0 +1,39 @@
+Qualcomm MSM8916 Network-On-Chip interconnect driver binding
+----------------------------------------------------
+
+Required properties :
+- compatible : shall contain only one of the following:
+			"qcom,msm8916-bimc"
+			"qcom,msm8916-pnoc"
+			"qcom,msm8916-snoc"
+- #interconnect-cells : should contain 1
+- reg : shall contain base register location and length
+
+Optional properties :
+clocks : list of phandles and specifiers to all interconnect bus clocks
+clock-names : clock names should include both "bus_clk" and "bus_a_clk"
+
+Examples:
+
+		snoc: snoc@580000 {
+			compatible = "qcom,msm8916-snoc";
+			#interconnect-cells = <1>;
+			reg = <0x580000 0x14000>;
+			clock-names = "bus_clk", "bus_a_clk";
+			clocks = <&rpmcc RPM_SMD_SNOC_CLK>, <&rpmcc RPM_SMD_SNOC_A_CLK>;
+		};
+		bimc: bimc@400000 {
+			compatible = "qcom,msm8916-bimc";
+			#interconnect-cells = <1>;
+			reg = <0x400000 0x62000>;
+			clock-names = "bus_clk", "bus_a_clk";
+			clocks = <&rpmcc RPM_SMD_BIMC_CLK>, <&rpmcc RPM_SMD_BIMC_A_CLK>;
+		};
+		pnoc: pnoc@500000 {
+			compatible = "qcom,msm8916-pnoc";
+			#interconnect-cells = <1>;
+			reg = <0x500000 0x11000>;
+			clock-names = "bus_clk", "bus_a_clk";
+			clocks = <&rpmcc RPM_SMD_PCNOC_CLK>, <&rpmcc RPM_SMD_PCNOC_A_CLK>;
+		};
+
diff --git a/include/dt-bindings/interconnect/qcom.h b/include/dt-bindings/interconnect/qcom.h
new file mode 100644
index 000000000000..2cd378d2f575
--- /dev/null
+++ b/include/dt-bindings/interconnect/qcom.h
@@ -0,0 +1,350 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Qualcomm interconnect IDs
+ *
+ * Copyright (c) 2018, Linaro Ltd.
+ * Author: Georgi Djakov <georgi.djakov@linaro.org>
+ */
+
+#ifndef __QCOM_INTERCONNECT_IDS_H
+#define __QCOM_INTERCONNECT_IDS_H
+
+#define FAB_BIMC 0
+#define FAB_SYS_NOC 1024
+#define FAB_MMSS_NOC 2048
+#define FAB_OCMEM_NOC 3072
+#define FAB_PERIPH_NOC 4096
+#define FAB_CONFIG_NOC 5120
+#define FAB_OCMEM_VNOC 6144
+
+#define MASTER_AMPSS_M0 1
+#define MASTER_AMPSS_M1 2
+#define APPSS_MASTER_FAB_MMSS 3
+#define APPSS_MASTER_FAB_SYSTEM 4
+#define SYSTEM_MASTER_FAB_APPSS 5
+#define MASTER_SPS 6
+#define MASTER_ADM_PORT0 7
+#define MASTER_ADM_PORT1 8
+#define SYSTEM_MASTER_ADM1_PORT0 9
+#define MASTER_ADM1_PORT1 10
+#define MASTER_LPASS_PROC 11
+#define MASTER_MSS_PROCI 12
+#define MASTER_MSS_PROCD 13
+#define MASTER_MSS_MDM_PORT0 14
+#define MASTER_LPASS 15
+#define SYSTEM_MASTER_CPSS_FPB 16
+#define SYSTEM_MASTER_SYSTEM_FPB 17
+#define SYSTEM_MASTER_MMSS_FPB 18
+#define MASTER_ADM1_CI 19
+#define MASTER_ADM0_CI 20
+#define MASTER_MSS_MDM_PORT1 21
+#define MASTER_MDP_PORT0 22
+#define MASTER_MDP_PORT1 23
+#define MMSS_MASTER_ADM1_PORT0 24
+#define MASTER_ROTATOR 25
+#define MASTER_GRAPHICS_3D 26
+#define MASTER_JPEG_DEC 27
+#define MASTER_GRAPHICS_2D_CORE0 28
+#define MASTER_VFE 29
+#define MASTER_VPE 30
+#define MASTER_JPEG_ENC 31
+#define MASTER_GRAPHICS_2D_CORE1 32
+#define MMSS_MASTER_APPS_FAB 33
+#define MASTER_HD_CODEC_PORT0 34
+#define MASTER_HD_CODEC_PORT1 35
+#define MASTER_SPDM 36
+#define MASTER_RPM 37
+#define MASTER_MSS 38
+#define MASTER_RIVA 39
+#define SYSTEM_MASTER_UNUSED_6 40
+#define MASTER_MSS_SW_PROC 41
+#define MASTER_MSS_FW_PROC 42
+#define MMSS_MASTER_UNUSED_2 43
+#define MASTER_GSS_NAV 44
+#define MASTER_PCIE 45
+#define MASTER_SATA 46
+#define MASTER_CRYPTO 47
+#define MASTER_VIDEO_CAP 48
+#define MASTER_GRAPHICS_3D_PORT1 49
+#define MASTER_VIDEO_ENC 50
+#define MASTER_VIDEO_DEC 51
+#define MASTER_LPASS_AHB 52
+#define MASTER_QDSS_BAM 53
+#define MASTER_SNOC_CFG 54
+#define MASTER_CRYPTO_CORE0 55
+#define MASTER_CRYPTO_CORE1 56
+#define MASTER_MSS_NAV 57
+#define MASTER_OCMEM_DMA 58
+#define MASTER_WCSS 59
+#define MASTER_QDSS_ETR 60
+#define MASTER_USB3 61
+#define MASTER_JPEG 62
+#define MASTER_VIDEO_P0 63
+#define MASTER_VIDEO_P1 64
+#define MASTER_MSS_PROC 65
+#define MASTER_JPEG_OCMEM 66
+#define MASTER_MDP_OCMEM 67
+#define MASTER_VIDEO_P0_OCMEM 68
+#define MASTER_VIDEO_P1_OCMEM 69
+#define MASTER_VFE_OCMEM 70
+#define MASTER_CNOC_ONOC_CFG 71
+#define MASTER_RPM_INST 72
+#define MASTER_RPM_DATA 73
+#define MASTER_RPM_SYS 74
+#define MASTER_DEHR 75
+#define MASTER_QDSS_DAP 76
+#define MASTER_TIC 77
+#define MASTER_SDCC_1 78
+#define MASTER_SDCC_3 79
+#define MASTER_SDCC_4 80
+#define MASTER_SDCC_2 81
+#define MASTER_TSIF 82
+#define MASTER_BAM_DMA 83
+#define MASTER_BLSP_2 84
+#define MASTER_USB_HSIC 85
+#define MASTER_BLSP_1 86
+#define MASTER_USB_HS 87
+#define MASTER_PNOC_CFG 88
+#define MASTER_V_OCMEM_GFX3D 89
+#define MASTER_IPA 90
+#define MASTER_QPIC 91
+#define MASTER_MDPE 92
+#define MASTER_USB_HS2 93
+#define MASTER_VPU 94
+#define MASTER_UFS 95
+#define MASTER_BCAST 96
+#define MASTER_CRYPTO_CORE2 97
+#define MASTER_EMAC 98
+#define MASTER_VPU_1 99
+#define MASTER_PCIE_1 100
+#define MASTER_USB3_1 101
+#define MASTER_CNOC_MNOC_MMSS_CFG 102
+#define MASTER_CNOC_MNOC_CFG 103
+#define MASTER_TCU_0 104
+#define MASTER_TCU_1 105
+#define MASTER_CPP 106
+#define MASTER_AUDIO 107
+
+#define SNOC_MM_INT_0 10000
+#define SNOC_MM_INT_1 10001
+#define SNOC_MM_INT_2 10002
+#define SNOC_MM_INT_BIMC 10003
+#define SNOC_INT_0 10004
+#define SNOC_INT_1 10005
+#define SNOC_INT_BIMC 10006
+#define SNOC_BIMC_0_MAS 10007
+#define SNOC_BIMC_1_MAS 10008
+#define SNOC_QDSS_INT 10009
+#define PNOC_SNOC_MAS 10010
+#define PNOC_SNOC_SLV 10011
+#define PNOC_INT_0 10012
+#define PNOC_INT_1 10013
+#define PNOC_M_0 10014
+#define PNOC_M_1 10015
+#define BIMC_SNOC_MAS 10016
+#define BIMC_SNOC_SLV 10017
+#define PNOC_SLV_0 10018
+#define PNOC_SLV_1 10019
+#define PNOC_SLV_2 10020
+#define PNOC_SLV_3 10021
+#define PNOC_SLV_4 10022
+#define PNOC_SLV_8 10023
+#define PNOC_SLV_9 10024
+#define SNOC_BIMC_0_SLV 10025
+#define SNOC_BIMC_1_SLV 10026
+#define MNOC_BIMC_MAS 10027
+#define MNOC_BIMC_SLV 10028
+#define BIMC_MNOC_MAS 10029
+#define BIMC_MNOC_SLV 10030
+#define SNOC_BIMC_MAS 10031
+#define SNOC_BIMC_SLV 10032
+#define CNOC_SNOC_MAS 10033
+#define CNOC_SNOC_SLV 10034
+#define SNOC_CNOC_MAS 10035
+#define SNOC_CNOC_SLV 10036
+#define OVNOC_SNOC_MAS 10037
+#define OVNOC_SNOC_SLV 10038
+#define SNOC_OVNOC_MAS 10039
+#define SNOC_OVNOC_SLV 10040
+#define SNOC_PNOC_MAS 10041
+#define SNOC_PNOC_SLV 10042
+#define BIMC_INT_APPS_EBI 10043
+#define BIMC_INT_APPS_SNOC 10044
+#define SNOC_BIMC_2_MAS 10045
+#define SNOC_BIMC_2_SLV 10046
+#define PNOC_SLV_5 10047
+#define PNOC_SLV_7 10048
+#define PNOC_INT_2 10049
+#define PNOC_INT_3 10050
+#define PNOC_INT_4 10051
+#define PNOC_INT_5 10052
+#define PNOC_INT_6 10053
+#define PNOC_INT_7 10054
+
+#define SLAVE_EBI_CH0 512
+#define SLAVE_EBI_CH1 513
+#define SLAVE_AMPSS_L2 514
+#define APPSS_SLAVE_FAB_MMSS 515
+#define APPSS_SLAVE_FAB_SYSTEM 516
+#define SYSTEM_SLAVE_FAB_APPS 517
+#define SLAVE_SPS 518
+#define SLAVE_SYSTEM_IMEM 519
+#define SLAVE_AMPSS 520
+#define SLAVE_MSS 521
+#define SLAVE_LPASS 522
+#define SYSTEM_SLAVE_CPSS_FPB 523
+#define SYSTEM_SLAVE_SYSTEM_FPB 524
+#define SYSTEM_SLAVE_MMSS_FPB 525
+#define SLAVE_CORESIGHT 526
+#define SLAVE_RIVA 527
+#define SLAVE_SMI 528
+#define MMSS_SLAVE_FAB_APPS 529
+#define MMSS_SLAVE_FAB_APPS_1 530
+#define SLAVE_MM_IMEM 531
+#define SLAVE_CRYPTO 532
+#define SLAVE_SPDM 533
+#define SLAVE_RPM 534
+#define SLAVE_RPM_MSG_RAM 535
+#define SLAVE_MPM 536
+#define SLAVE_PMIC1_SSBI1_A 537
+#define SLAVE_PMIC1_SSBI1_B 538
+#define SLAVE_PMIC1_SSBI1_C 539
+#define SLAVE_PMIC2_SSBI2_A 540
+#define SLAVE_PMIC2_SSBI2_B 541
+#define SLAVE_GSBI1_UART 542
+#define SLAVE_GSBI2_UART 543
+#define SLAVE_GSBI3_UART 544
+#define SLAVE_GSBI4_UART 545
+#define SLAVE_GSBI5_UART 546
+#define SLAVE_GSBI6_UART 547
+#define SLAVE_GSBI7_UART 548
+#define SLAVE_GSBI8_UART 549
+#define SLAVE_GSBI9_UART 550
+#define SLAVE_GSBI10_UART 551
+#define SLAVE_GSBI11_UART 552
+#define SLAVE_GSBI12_UART 553
+#define SLAVE_GSBI1_QUP 554
+#define SLAVE_GSBI2_QUP 555
+#define SLAVE_GSBI3_QUP 556
+#define SLAVE_GSBI4_QUP 557
+#define SLAVE_GSBI5_QUP 558
+#define SLAVE_GSBI6_QUP 559
+#define SLAVE_GSBI7_QUP 560
+#define SLAVE_GSBI8_QUP 561
+#define SLAVE_GSBI9_QUP 562
+#define SLAVE_GSBI10_QUP 563
+#define SLAVE_GSBI11_QUP 564
+#define SLAVE_GSBI12_QUP 565
+#define SLAVE_EBI2_NAND 566
+#define SLAVE_EBI2_CS0 567
+#define SLAVE_EBI2_CS1 568
+#define SLAVE_EBI2_CS2 569
+#define SLAVE_EBI2_CS3 570
+#define SLAVE_EBI2_CS4 571
+#define SLAVE_EBI2_CS5 572
+#define SLAVE_USB_FS1 573
+#define SLAVE_USB_FS2 574
+#define SLAVE_TSIF 575
+#define SLAVE_MSM_TSSC 576
+#define SLAVE_MSM_PDM 577
+#define SLAVE_MSM_DIMEM 578
+#define SLAVE_MSM_TCSR 579
+#define SLAVE_MSM_PRNG 580
+#define SLAVE_GSS 581
+#define SLAVE_SATA 582
+#define SLAVE_USB3 583
+#define SLAVE_WCSS 584
+#define SLAVE_OCIMEM 585
+#define SLAVE_SNOC_OCMEM 586
+#define SLAVE_SERVICE_SNOC 587
+#define SLAVE_QDSS_STM 588
+#define SLAVE_CAMERA_CFG 589
+#define SLAVE_DISPLAY_CFG 590
+#define SLAVE_OCMEM_CFG 591
+#define SLAVE_CPR_CFG 592
+#define SLAVE_CPR_XPU_CFG 593
+#define SLAVE_MISC_CFG 594
+#define SLAVE_MISC_XPU_CFG 595
+#define SLAVE_VENUS_CFG 596
+#define SLAVE_MISC_VENUS_CFG 597
+#define SLAVE_GRAPHICS_3D_CFG 598
+#define SLAVE_MMSS_CLK_CFG 599
+#define SLAVE_MMSS_CLK_XPU_CFG 600
+#define SLAVE_MNOC_MPU_CFG 601
+#define SLAVE_ONOC_MPU_CFG 602
+#define SLAVE_SERVICE_MNOC 603
+#define SLAVE_OCMEM 604
+#define SLAVE_SERVICE_ONOC 605
+#define SLAVE_SDCC_1 606
+#define SLAVE_SDCC_3 607
+#define SLAVE_SDCC_2 608
+#define SLAVE_SDCC_4 609
+#define SLAVE_BAM_DMA 610
+#define SLAVE_BLSP_2 611
+#define SLAVE_USB_HSIC 612
+#define SLAVE_BLSP_1 613
+#define SLAVE_USB_HS 614
+#define SLAVE_PDM 615
+#define SLAVE_PERIPH_APU_CFG 616
+#define SLAVE_PNOC_MPU_CFG 617
+#define SLAVE_PRNG 618
+#define SLAVE_SERVICE_PNOC 619
+#define SLAVE_CLK_CTL 620
+#define SLAVE_CNOC_MSS 621
+#define SLAVE_SECURITY 622
+#define SLAVE_TCSR 623
+#define SLAVE_TLMM 624
+#define SLAVE_CRYPTO_0_CFG 625
+#define SLAVE_CRYPTO_1_CFG 626
+#define SLAVE_IMEM_CFG 627
+#define SLAVE_MESSAGE_RAM 628
+#define SLAVE_BIMC_CFG 629
+#define SLAVE_BOOT_ROM 630
+#define SLAVE_CNOC_MNOC_MMSS_CFG 631
+#define SLAVE_PMIC_ARB 632
+#define SLAVE_SPDM_WRAPPER 633
+#define SLAVE_DEHR_CFG 634
+#define SLAVE_QDSS_CFG 635
+#define SLAVE_RBCPR_CFG 636
+#define SLAVE_RBCPR_QDSS_APU_CFG 637
+#define SLAVE_SNOC_MPU_CFG 638
+#define SLAVE_CNOC_ONOC_CFG 639
+#define SLAVE_CNOC_MNOC_CFG 640
+#define SLAVE_PNOC_CFG 641
+#define SLAVE_SNOC_CFG 642
+#define SLAVE_EBI1_DLL_CFG 643
+#define SLAVE_PHY_APU_CFG 644
+#define SLAVE_EBI1_PHY_CFG 645
+#define SLAVE_SERVICE_CNOC 646
+#define SLAVE_IPS_CFG 647
+#define SLAVE_QPIC 648
+#define SLAVE_DSI_CFG 649
+#define SLAVE_UFS_CFG 650
+#define SLAVE_RBCPR_CX_CFG 651
+#define SLAVE_RBCPR_MX_CFG 652
+#define SLAVE_PCIE_CFG 653
+#define SLAVE_USB_PHYS_CFG 654
+#define SLAVE_VIDEO_CAP_CFG 655
+#define SLAVE_AVSYNC_CFG 656
+#define SLAVE_CRYPTO_2_CFG 657
+#define SLAVE_VPU_CFG 658
+#define SLAVE_BCAST_CFG 659
+#define SLAVE_KLM_CFG 660
+#define SLAVE_GENI_IR_CFG 661
+#define SLAVE_OCMEM_GFX 662
+#define SLAVE_CATS_128 663
+#define SLAVE_OCMEM_64 664
+#define SLAVE_PCIE_0 665
+#define SLAVE_PCIE_1 666
+#define SLAVE_PCIE_0_CFG 667
+#define SLAVE_PCIE_1_CFG 668
+#define SLAVE_SRVC_MNOC 669
+#define SLAVE_USB_HS2 670
+#define SLAVE_AUDIO 671
+#define SLAVE_TCU 672
+#define SLAVE_APPSS 673
+#define SLAVE_PCIE_PARF 674
+#define SLAVE_USB3_PHY_CFG 675
+#define SLAVE_IPA_CFG 676
+
+#endif

[PATCH v5 5/8] dt-bindings: interconnect: Document qcom, msm8916 NoC bindings

[Georgi Djakov]

Document the device-tree bindings Network-On-Chip interconnect driver
for Qualcomm msm8916 platforms.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 .../bindings/interconnect/qcom-msm8916.txt    |  39 ++
 include/dt-bindings/interconnect/qcom.h       | 350 ++++++++++++++++++
 2 files changed, 389 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/interconnect/qcom-msm8916.txt
 create mode 100644 include/dt-bindings/interconnect/qcom.h

diff --git a/Documentation/devicetree/bindings/interconnect/qcom-msm8916.txt b/Documentation/devicetree/bindings/interconnect/qcom-msm8916.txt
new file mode 100644
index 000000000000..f309eaed3d19
--- /dev/null
+++ b/Documentation/devicetree/bindings/interconnect/qcom-msm8916.txt
@@ -0,0 +1,39 @@
+Qualcomm MSM8916 Network-On-Chip interconnect driver binding
+----------------------------------------------------
+
+Required properties :
+- compatible : shall contain only one of the following:
+			"qcom,msm8916-bimc"
+			"qcom,msm8916-pnoc"
+			"qcom,msm8916-snoc"
+- #interconnect-cells : should contain 1
+- reg : shall contain base register location and length
+
+Optional properties :
+clocks : list of phandles and specifiers to all interconnect bus clocks
+clock-names : clock names should include both "bus_clk" and "bus_a_clk"
+
+Examples:
+
+		snoc: snoc at 580000 {
+			compatible = "qcom,msm8916-snoc";
+			#interconnect-cells = <1>;
+			reg = <0x580000 0x14000>;
+			clock-names = "bus_clk", "bus_a_clk";
+			clocks = <&rpmcc RPM_SMD_SNOC_CLK>, <&rpmcc RPM_SMD_SNOC_A_CLK>;
+		};
+		bimc: bimc at 400000 {
+			compatible = "qcom,msm8916-bimc";
+			#interconnect-cells = <1>;
+			reg = <0x400000 0x62000>;
+			clock-names = "bus_clk", "bus_a_clk";
+			clocks = <&rpmcc RPM_SMD_BIMC_CLK>, <&rpmcc RPM_SMD_BIMC_A_CLK>;
+		};
+		pnoc: pnoc at 500000 {
+			compatible = "qcom,msm8916-pnoc";
+			#interconnect-cells = <1>;
+			reg = <0x500000 0x11000>;
+			clock-names = "bus_clk", "bus_a_clk";
+			clocks = <&rpmcc RPM_SMD_PCNOC_CLK>, <&rpmcc RPM_SMD_PCNOC_A_CLK>;
+		};
+
diff --git a/include/dt-bindings/interconnect/qcom.h b/include/dt-bindings/interconnect/qcom.h
new file mode 100644
index 000000000000..2cd378d2f575
--- /dev/null
+++ b/include/dt-bindings/interconnect/qcom.h
@@ -0,0 +1,350 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Qualcomm interconnect IDs
+ *
+ * Copyright (c) 2018, Linaro Ltd.
+ * Author: Georgi Djakov <georgi.djakov@linaro.org>
+ */
+
+#ifndef __QCOM_INTERCONNECT_IDS_H
+#define __QCOM_INTERCONNECT_IDS_H
+
+#define FAB_BIMC 0
+#define FAB_SYS_NOC 1024
+#define FAB_MMSS_NOC 2048
+#define FAB_OCMEM_NOC 3072
+#define FAB_PERIPH_NOC 4096
+#define FAB_CONFIG_NOC 5120
+#define FAB_OCMEM_VNOC 6144
+
+#define MASTER_AMPSS_M0 1
+#define MASTER_AMPSS_M1 2
+#define APPSS_MASTER_FAB_MMSS 3
+#define APPSS_MASTER_FAB_SYSTEM 4
+#define SYSTEM_MASTER_FAB_APPSS 5
+#define MASTER_SPS 6
+#define MASTER_ADM_PORT0 7
+#define MASTER_ADM_PORT1 8
+#define SYSTEM_MASTER_ADM1_PORT0 9
+#define MASTER_ADM1_PORT1 10
+#define MASTER_LPASS_PROC 11
+#define MASTER_MSS_PROCI 12
+#define MASTER_MSS_PROCD 13
+#define MASTER_MSS_MDM_PORT0 14
+#define MASTER_LPASS 15
+#define SYSTEM_MASTER_CPSS_FPB 16
+#define SYSTEM_MASTER_SYSTEM_FPB 17
+#define SYSTEM_MASTER_MMSS_FPB 18
+#define MASTER_ADM1_CI 19
+#define MASTER_ADM0_CI 20
+#define MASTER_MSS_MDM_PORT1 21
+#define MASTER_MDP_PORT0 22
+#define MASTER_MDP_PORT1 23
+#define MMSS_MASTER_ADM1_PORT0 24
+#define MASTER_ROTATOR 25
+#define MASTER_GRAPHICS_3D 26
+#define MASTER_JPEG_DEC 27
+#define MASTER_GRAPHICS_2D_CORE0 28
+#define MASTER_VFE 29
+#define MASTER_VPE 30
+#define MASTER_JPEG_ENC 31
+#define MASTER_GRAPHICS_2D_CORE1 32
+#define MMSS_MASTER_APPS_FAB 33
+#define MASTER_HD_CODEC_PORT0 34
+#define MASTER_HD_CODEC_PORT1 35
+#define MASTER_SPDM 36
+#define MASTER_RPM 37
+#define MASTER_MSS 38
+#define MASTER_RIVA 39
+#define SYSTEM_MASTER_UNUSED_6 40
+#define MASTER_MSS_SW_PROC 41
+#define MASTER_MSS_FW_PROC 42
+#define MMSS_MASTER_UNUSED_2 43
+#define MASTER_GSS_NAV 44
+#define MASTER_PCIE 45
+#define MASTER_SATA 46
+#define MASTER_CRYPTO 47
+#define MASTER_VIDEO_CAP 48
+#define MASTER_GRAPHICS_3D_PORT1 49
+#define MASTER_VIDEO_ENC 50
+#define MASTER_VIDEO_DEC 51
+#define MASTER_LPASS_AHB 52
+#define MASTER_QDSS_BAM 53
+#define MASTER_SNOC_CFG 54
+#define MASTER_CRYPTO_CORE0 55
+#define MASTER_CRYPTO_CORE1 56
+#define MASTER_MSS_NAV 57
+#define MASTER_OCMEM_DMA 58
+#define MASTER_WCSS 59
+#define MASTER_QDSS_ETR 60
+#define MASTER_USB3 61
+#define MASTER_JPEG 62
+#define MASTER_VIDEO_P0 63
+#define MASTER_VIDEO_P1 64
+#define MASTER_MSS_PROC 65
+#define MASTER_JPEG_OCMEM 66
+#define MASTER_MDP_OCMEM 67
+#define MASTER_VIDEO_P0_OCMEM 68
+#define MASTER_VIDEO_P1_OCMEM 69
+#define MASTER_VFE_OCMEM 70
+#define MASTER_CNOC_ONOC_CFG 71
+#define MASTER_RPM_INST 72
+#define MASTER_RPM_DATA 73
+#define MASTER_RPM_SYS 74
+#define MASTER_DEHR 75
+#define MASTER_QDSS_DAP 76
+#define MASTER_TIC 77
+#define MASTER_SDCC_1 78
+#define MASTER_SDCC_3 79
+#define MASTER_SDCC_4 80
+#define MASTER_SDCC_2 81
+#define MASTER_TSIF 82
+#define MASTER_BAM_DMA 83
+#define MASTER_BLSP_2 84
+#define MASTER_USB_HSIC 85
+#define MASTER_BLSP_1 86
+#define MASTER_USB_HS 87
+#define MASTER_PNOC_CFG 88
+#define MASTER_V_OCMEM_GFX3D 89
+#define MASTER_IPA 90
+#define MASTER_QPIC 91
+#define MASTER_MDPE 92
+#define MASTER_USB_HS2 93
+#define MASTER_VPU 94
+#define MASTER_UFS 95
+#define MASTER_BCAST 96
+#define MASTER_CRYPTO_CORE2 97
+#define MASTER_EMAC 98
+#define MASTER_VPU_1 99
+#define MASTER_PCIE_1 100
+#define MASTER_USB3_1 101
+#define MASTER_CNOC_MNOC_MMSS_CFG 102
+#define MASTER_CNOC_MNOC_CFG 103
+#define MASTER_TCU_0 104
+#define MASTER_TCU_1 105
+#define MASTER_CPP 106
+#define MASTER_AUDIO 107
+
+#define SNOC_MM_INT_0 10000
+#define SNOC_MM_INT_1 10001
+#define SNOC_MM_INT_2 10002
+#define SNOC_MM_INT_BIMC 10003
+#define SNOC_INT_0 10004
+#define SNOC_INT_1 10005
+#define SNOC_INT_BIMC 10006
+#define SNOC_BIMC_0_MAS 10007
+#define SNOC_BIMC_1_MAS 10008
+#define SNOC_QDSS_INT 10009
+#define PNOC_SNOC_MAS 10010
+#define PNOC_SNOC_SLV 10011
+#define PNOC_INT_0 10012
+#define PNOC_INT_1 10013
+#define PNOC_M_0 10014
+#define PNOC_M_1 10015
+#define BIMC_SNOC_MAS 10016
+#define BIMC_SNOC_SLV 10017
+#define PNOC_SLV_0 10018
+#define PNOC_SLV_1 10019
+#define PNOC_SLV_2 10020
+#define PNOC_SLV_3 10021
+#define PNOC_SLV_4 10022
+#define PNOC_SLV_8 10023
+#define PNOC_SLV_9 10024
+#define SNOC_BIMC_0_SLV 10025
+#define SNOC_BIMC_1_SLV 10026
+#define MNOC_BIMC_MAS 10027
+#define MNOC_BIMC_SLV 10028
+#define BIMC_MNOC_MAS 10029
+#define BIMC_MNOC_SLV 10030
+#define SNOC_BIMC_MAS 10031
+#define SNOC_BIMC_SLV 10032
+#define CNOC_SNOC_MAS 10033
+#define CNOC_SNOC_SLV 10034
+#define SNOC_CNOC_MAS 10035
+#define SNOC_CNOC_SLV 10036
+#define OVNOC_SNOC_MAS 10037
+#define OVNOC_SNOC_SLV 10038
+#define SNOC_OVNOC_MAS 10039
+#define SNOC_OVNOC_SLV 10040
+#define SNOC_PNOC_MAS 10041
+#define SNOC_PNOC_SLV 10042
+#define BIMC_INT_APPS_EBI 10043
+#define BIMC_INT_APPS_SNOC 10044
+#define SNOC_BIMC_2_MAS 10045
+#define SNOC_BIMC_2_SLV 10046
+#define PNOC_SLV_5 10047
+#define PNOC_SLV_7 10048
+#define PNOC_INT_2 10049
+#define PNOC_INT_3 10050
+#define PNOC_INT_4 10051
+#define PNOC_INT_5 10052
+#define PNOC_INT_6 10053
+#define PNOC_INT_7 10054
+
+#define SLAVE_EBI_CH0 512
+#define SLAVE_EBI_CH1 513
+#define SLAVE_AMPSS_L2 514
+#define APPSS_SLAVE_FAB_MMSS 515
+#define APPSS_SLAVE_FAB_SYSTEM 516
+#define SYSTEM_SLAVE_FAB_APPS 517
+#define SLAVE_SPS 518
+#define SLAVE_SYSTEM_IMEM 519
+#define SLAVE_AMPSS 520
+#define SLAVE_MSS 521
+#define SLAVE_LPASS 522
+#define SYSTEM_SLAVE_CPSS_FPB 523
+#define SYSTEM_SLAVE_SYSTEM_FPB 524
+#define SYSTEM_SLAVE_MMSS_FPB 525
+#define SLAVE_CORESIGHT 526
+#define SLAVE_RIVA 527
+#define SLAVE_SMI 528
+#define MMSS_SLAVE_FAB_APPS 529
+#define MMSS_SLAVE_FAB_APPS_1 530
+#define SLAVE_MM_IMEM 531
+#define SLAVE_CRYPTO 532
+#define SLAVE_SPDM 533
+#define SLAVE_RPM 534
+#define SLAVE_RPM_MSG_RAM 535
+#define SLAVE_MPM 536
+#define SLAVE_PMIC1_SSBI1_A 537
+#define SLAVE_PMIC1_SSBI1_B 538
+#define SLAVE_PMIC1_SSBI1_C 539
+#define SLAVE_PMIC2_SSBI2_A 540
+#define SLAVE_PMIC2_SSBI2_B 541
+#define SLAVE_GSBI1_UART 542
+#define SLAVE_GSBI2_UART 543
+#define SLAVE_GSBI3_UART 544
+#define SLAVE_GSBI4_UART 545
+#define SLAVE_GSBI5_UART 546
+#define SLAVE_GSBI6_UART 547
+#define SLAVE_GSBI7_UART 548
+#define SLAVE_GSBI8_UART 549
+#define SLAVE_GSBI9_UART 550
+#define SLAVE_GSBI10_UART 551
+#define SLAVE_GSBI11_UART 552
+#define SLAVE_GSBI12_UART 553
+#define SLAVE_GSBI1_QUP 554
+#define SLAVE_GSBI2_QUP 555
+#define SLAVE_GSBI3_QUP 556
+#define SLAVE_GSBI4_QUP 557
+#define SLAVE_GSBI5_QUP 558
+#define SLAVE_GSBI6_QUP 559
+#define SLAVE_GSBI7_QUP 560
+#define SLAVE_GSBI8_QUP 561
+#define SLAVE_GSBI9_QUP 562
+#define SLAVE_GSBI10_QUP 563
+#define SLAVE_GSBI11_QUP 564
+#define SLAVE_GSBI12_QUP 565
+#define SLAVE_EBI2_NAND 566
+#define SLAVE_EBI2_CS0 567
+#define SLAVE_EBI2_CS1 568
+#define SLAVE_EBI2_CS2 569
+#define SLAVE_EBI2_CS3 570
+#define SLAVE_EBI2_CS4 571
+#define SLAVE_EBI2_CS5 572
+#define SLAVE_USB_FS1 573
+#define SLAVE_USB_FS2 574
+#define SLAVE_TSIF 575
+#define SLAVE_MSM_TSSC 576
+#define SLAVE_MSM_PDM 577
+#define SLAVE_MSM_DIMEM 578
+#define SLAVE_MSM_TCSR 579
+#define SLAVE_MSM_PRNG 580
+#define SLAVE_GSS 581
+#define SLAVE_SATA 582
+#define SLAVE_USB3 583
+#define SLAVE_WCSS 584
+#define SLAVE_OCIMEM 585
+#define SLAVE_SNOC_OCMEM 586
+#define SLAVE_SERVICE_SNOC 587
+#define SLAVE_QDSS_STM 588
+#define SLAVE_CAMERA_CFG 589
+#define SLAVE_DISPLAY_CFG 590
+#define SLAVE_OCMEM_CFG 591
+#define SLAVE_CPR_CFG 592
+#define SLAVE_CPR_XPU_CFG 593
+#define SLAVE_MISC_CFG 594
+#define SLAVE_MISC_XPU_CFG 595
+#define SLAVE_VENUS_CFG 596
+#define SLAVE_MISC_VENUS_CFG 597
+#define SLAVE_GRAPHICS_3D_CFG 598
+#define SLAVE_MMSS_CLK_CFG 599
+#define SLAVE_MMSS_CLK_XPU_CFG 600
+#define SLAVE_MNOC_MPU_CFG 601
+#define SLAVE_ONOC_MPU_CFG 602
+#define SLAVE_SERVICE_MNOC 603
+#define SLAVE_OCMEM 604
+#define SLAVE_SERVICE_ONOC 605
+#define SLAVE_SDCC_1 606
+#define SLAVE_SDCC_3 607
+#define SLAVE_SDCC_2 608
+#define SLAVE_SDCC_4 609
+#define SLAVE_BAM_DMA 610
+#define SLAVE_BLSP_2 611
+#define SLAVE_USB_HSIC 612
+#define SLAVE_BLSP_1 613
+#define SLAVE_USB_HS 614
+#define SLAVE_PDM 615
+#define SLAVE_PERIPH_APU_CFG 616
+#define SLAVE_PNOC_MPU_CFG 617
+#define SLAVE_PRNG 618
+#define SLAVE_SERVICE_PNOC 619
+#define SLAVE_CLK_CTL 620
+#define SLAVE_CNOC_MSS 621
+#define SLAVE_SECURITY 622
+#define SLAVE_TCSR 623
+#define SLAVE_TLMM 624
+#define SLAVE_CRYPTO_0_CFG 625
+#define SLAVE_CRYPTO_1_CFG 626
+#define SLAVE_IMEM_CFG 627
+#define SLAVE_MESSAGE_RAM 628
+#define SLAVE_BIMC_CFG 629
+#define SLAVE_BOOT_ROM 630
+#define SLAVE_CNOC_MNOC_MMSS_CFG 631
+#define SLAVE_PMIC_ARB 632
+#define SLAVE_SPDM_WRAPPER 633
+#define SLAVE_DEHR_CFG 634
+#define SLAVE_QDSS_CFG 635
+#define SLAVE_RBCPR_CFG 636
+#define SLAVE_RBCPR_QDSS_APU_CFG 637
+#define SLAVE_SNOC_MPU_CFG 638
+#define SLAVE_CNOC_ONOC_CFG 639
+#define SLAVE_CNOC_MNOC_CFG 640
+#define SLAVE_PNOC_CFG 641
+#define SLAVE_SNOC_CFG 642
+#define SLAVE_EBI1_DLL_CFG 643
+#define SLAVE_PHY_APU_CFG 644
+#define SLAVE_EBI1_PHY_CFG 645
+#define SLAVE_SERVICE_CNOC 646
+#define SLAVE_IPS_CFG 647
+#define SLAVE_QPIC 648
+#define SLAVE_DSI_CFG 649
+#define SLAVE_UFS_CFG 650
+#define SLAVE_RBCPR_CX_CFG 651
+#define SLAVE_RBCPR_MX_CFG 652
+#define SLAVE_PCIE_CFG 653
+#define SLAVE_USB_PHYS_CFG 654
+#define SLAVE_VIDEO_CAP_CFG 655
+#define SLAVE_AVSYNC_CFG 656
+#define SLAVE_CRYPTO_2_CFG 657
+#define SLAVE_VPU_CFG 658
+#define SLAVE_BCAST_CFG 659
+#define SLAVE_KLM_CFG 660
+#define SLAVE_GENI_IR_CFG 661
+#define SLAVE_OCMEM_GFX 662
+#define SLAVE_CATS_128 663
+#define SLAVE_OCMEM_64 664
+#define SLAVE_PCIE_0 665
+#define SLAVE_PCIE_1 666
+#define SLAVE_PCIE_0_CFG 667
+#define SLAVE_PCIE_1_CFG 668
+#define SLAVE_SRVC_MNOC 669
+#define SLAVE_USB_HS2 670
+#define SLAVE_AUDIO 671
+#define SLAVE_TCU 672
+#define SLAVE_APPSS 673
+#define SLAVE_PCIE_PARF 674
+#define SLAVE_USB3_PHY_CFG 675
+#define SLAVE_IPA_CFG 676
+
+#endif

[PATCH v5 6/8] interconnect: qcom: Add msm8916 interconnect provider driver

[Georgi Djakov]

Add driver for the Qualcomm interconnect buses found in msm8916 based
platforms.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 drivers/interconnect/Kconfig        |   5 +
 drivers/interconnect/Makefile       |   1 +
 drivers/interconnect/qcom/Kconfig   |  11 +
 drivers/interconnect/qcom/Makefile  |   2 +
 drivers/interconnect/qcom/msm8916.c | 495 ++++++++++++++++++++++++++++
 5 files changed, 514 insertions(+)
 create mode 100644 drivers/interconnect/qcom/Kconfig
 create mode 100644 drivers/interconnect/qcom/msm8916.c

diff --git a/drivers/interconnect/Kconfig b/drivers/interconnect/Kconfig
index a261c7d41deb..07a8276fa35a 100644
--- a/drivers/interconnect/Kconfig
+++ b/drivers/interconnect/Kconfig
@@ -8,3 +8,8 @@ menuconfig INTERCONNECT
 
 	  If unsure, say no.
 
+if INTERCONNECT
+
+source "drivers/interconnect/qcom/Kconfig"
+
+endif
diff --git a/drivers/interconnect/Makefile b/drivers/interconnect/Makefile
index 97fca2e09d24..7944cbca0527 100644
--- a/drivers/interconnect/Makefile
+++ b/drivers/interconnect/Makefile
@@ -1,2 +1,3 @@
 # SPDX-License-Identifier: GPL-2.0
 obj-$(CONFIG_INTERCONNECT)		+= core.o
+obj-$(CONFIG_INTERCONNECT_QCOM)		+= qcom/
diff --git a/drivers/interconnect/qcom/Kconfig b/drivers/interconnect/qcom/Kconfig
new file mode 100644
index 000000000000..86465dc37bd4
--- /dev/null
+++ b/drivers/interconnect/qcom/Kconfig
@@ -0,0 +1,11 @@
+config INTERCONNECT_QCOM
+	bool "Qualcomm Network-on-Chip interconnect drivers"
+	depends on INTERCONNECT
+	depends on ARCH_QCOM || COMPILE_TEST
+	default y
+
+config INTERCONNECT_QCOM_MSM8916
+	tristate "Qualcomm MSM8916 interconnect driver"
+	depends on INTERCONNECT_QCOM
+	help
+	  This is a driver for the Qualcomm Network-on-Chip on msm8916-based platforms.
diff --git a/drivers/interconnect/qcom/Makefile b/drivers/interconnect/qcom/Makefile
index 2a0c41db91c4..e8b24c334c5d 100644
--- a/drivers/interconnect/qcom/Makefile
+++ b/drivers/interconnect/qcom/Makefile
@@ -1,2 +1,4 @@
 # SPDX-License-Identifier: GPL-2.0
 obj-y += smd-rpm.o
+
+obj-$(CONFIG_INTERCONNECT_QCOM_MSM8916) += msm8916.o
diff --git a/drivers/interconnect/qcom/msm8916.c b/drivers/interconnect/qcom/msm8916.c
new file mode 100644
index 000000000000..c34f18cdea33
--- /dev/null
+++ b/drivers/interconnect/qcom/msm8916.c
@@ -0,0 +1,495 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2018 Linaro Ltd
+ * Author: Georgi Djakov <georgi.djakov@linaro.org>
+ */
+
+#include <dt-bindings/interconnect/qcom.h>
+#include <linux/clk.h>
+#include <linux/device.h>
+#include <linux/io.h>
+#include <linux/interconnect-provider.h>
+#include <linux/module.h>
+#include <linux/of_device.h>
+#include <linux/of_platform.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+
+#include "smd-rpm.h"
+
+#define RPM_MASTER_FIELD_BW	0x00007762
+#define RPM_BUS_MASTER_REQ      0x73616d62
+#define RPM_BUS_SLAVE_REQ       0x766c7362
+
+#define to_qcom_provider(_provider) \
+	container_of(_provider, struct qcom_icc_provider, provider)
+
+enum qcom_qos_mode {
+	QCOM_QOS_MODE_BYPASS = 0,
+	QCOM_QOS_MODE_FIXED,
+	QCOM_QOS_MODE_MAX,
+};
+
+struct qcom_icc_provider {
+	struct icc_provider	provider;
+	void __iomem		*base;
+	struct clk		*bus_clk;
+	struct clk		*bus_a_clk;
+};
+
+#define MSM8916_MAX_LINKS	8
+
+/**
+ * struct qcom_icc_node - Qualcomm specific interconnect nodes
+ * @name: the node name used in debugfs
+ * @links: an array of nodes where we can go next while traversing
+ * @id: a unique node identifier
+ * @num_links: the total number of @links
+ * @port: the offset index into the masters QoS register space
+ * @buswidth: width of the interconnect between a node and the bus (bytes)
+ * @ap_owned: the AP CPU does the writing to QoS registers
+ * @rpm: reference to the RPM SMD driver
+ * @qos_mode: QoS mode for ap_owned resources
+ * @mas_rpm_id:	RPM id for devices that are bus masters
+ * @slv_rpm_id:	RPM id for devices that are bus slaves
+ * @rate: current bus clock rate in Hz
+ */
+struct qcom_icc_node {
+	unsigned char *name;
+	u16 links[MSM8916_MAX_LINKS];
+	u16 id;
+	u16 num_links;
+	u16 port;
+	u16 buswidth;
+	bool ap_owned;
+	struct qcom_smd_rpm *rpm;
+	enum qcom_qos_mode qos_mode;
+	int mas_rpm_id;
+	int slv_rpm_id;
+	u64 rate;
+};
+
+struct qcom_icc_desc {
+	struct qcom_icc_node **nodes;
+	size_t num_nodes;
+};
+
+#define DEFINE_QNODE(_name, _id, _port, _buswidth, _ap_owned,		\
+			_mas_rpm_id, _slv_rpm_id, _qos_mode,		\
+			_numlinks, ...)					\
+		static struct qcom_icc_node _name = {			\
+		.id = _id,						\
+		.name = #_name,						\
+		.port = _port,						\
+		.buswidth = _buswidth,					\
+		.qos_mode = _qos_mode,					\
+		.ap_owned = _ap_owned,					\
+		.mas_rpm_id = _mas_rpm_id,				\
+		.slv_rpm_id = _slv_rpm_id,				\
+		.num_links = _numlinks,					\
+		.links = { __VA_ARGS__ },				\
+	}
+
+DEFINE_QNODE(mas_video, MASTER_VIDEO_P0, 8, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_0, SNOC_MM_INT_2);
+DEFINE_QNODE(mas_jpeg, MASTER_JPEG, 6, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_0, SNOC_MM_INT_2);
+DEFINE_QNODE(mas_vfe, MASTER_VFE, 9, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_1, SNOC_MM_INT_2);
+DEFINE_QNODE(mas_mdp, MASTER_MDP_PORT0, 7, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_0, SNOC_MM_INT_2);
+DEFINE_QNODE(mas_qdss_bam, MASTER_QDSS_BAM, 11, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_QDSS_INT);
+DEFINE_QNODE(mas_snoc_cfg, MASTER_SNOC_CFG, 0, 16, 0, 20, -1, QCOM_QOS_MODE_BYPASS, 1, SNOC_QDSS_INT);
+DEFINE_QNODE(mas_qdss_etr, MASTER_QDSS_ETR, 10, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_QDSS_INT);
+DEFINE_QNODE(mm_int_0, SNOC_MM_INT_0, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_MM_INT_BIMC);
+DEFINE_QNODE(mm_int_1, SNOC_MM_INT_1, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_MM_INT_BIMC);
+DEFINE_QNODE(mm_int_2, SNOC_MM_INT_2, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_INT_0);
+DEFINE_QNODE(mm_int_bimc, SNOC_MM_INT_BIMC, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_1_MAS);
+DEFINE_QNODE(snoc_int_0, SNOC_INT_0, 0, 8, 0, 99, 130, QCOM_QOS_MODE_FIXED, 3, SLAVE_QDSS_STM, SLAVE_SYSTEM_IMEM, MNOC_BIMC_MAS);
+DEFINE_QNODE(snoc_int_1, SNOC_INT_1, 0, 8, 0, 100, 131, QCOM_QOS_MODE_FIXED, 3, SYSTEM_SLAVE_FAB_APPS, SLAVE_CATS_128, SLAVE_OCMEM_64);
+DEFINE_QNODE(snoc_int_bimc, SNOC_INT_BIMC, 0, 8, 0, 101, 132, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_0_MAS);
+DEFINE_QNODE(snoc_bimc_0_mas, SNOC_BIMC_0_MAS, 0, 8, 0, 3, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_0_SLV);
+DEFINE_QNODE(snoc_bimc_1_mas, SNOC_BIMC_1_MAS, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_1_SLV);
+DEFINE_QNODE(qdss_int, SNOC_QDSS_INT, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 2, SNOC_INT_0, SNOC_INT_BIMC);
+DEFINE_QNODE(bimc_snoc_slv, BIMC_SNOC_SLV, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 2, SNOC_INT_0, SNOC_INT_1);
+DEFINE_QNODE(snoc_pnoc_mas, MNOC_BIMC_MAS, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, MNOC_BIMC_SLV);
+DEFINE_QNODE(pnoc_snoc_slv, PNOC_SNOC_SLV, 0, 8, 0, -1, 45, QCOM_QOS_MODE_FIXED, 3, SNOC_INT_0, SNOC_INT_BIMC, SNOC_INT_1);
+DEFINE_QNODE(slv_srvc_snoc, SLAVE_SERVICE_SNOC, 0, 8, 0, -1, 29, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_qdss_stm, SLAVE_QDSS_STM, 0, 4, 0, -1, 30, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_imem, SLAVE_SYSTEM_IMEM, 0, 8, 0, -1, 26, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_apss, SYSTEM_SLAVE_FAB_APPS, 0, 4, 0, -1, 20, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_cats_0, SLAVE_CATS_128, 0, 16, 0, -1, 106, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_cats_1, SLAVE_OCMEM_64, 0, 8, 0, -1, 107, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(mas_apss, MASTER_AMPSS_M0, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
+DEFINE_QNODE(mas_tcu0, MASTER_TCU_0, 5, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
+DEFINE_QNODE(mas_tcu1, MASTER_TCU_1, 6, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
+DEFINE_QNODE(mas_gfx, MASTER_GRAPHICS_3D, 2, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
+DEFINE_QNODE(bimc_snoc_mas, BIMC_SNOC_MAS, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, BIMC_SNOC_SLV);
+DEFINE_QNODE(snoc_bimc_0_slv, SNOC_BIMC_0_SLV, 0, 8, 0, -1, 24, QCOM_QOS_MODE_FIXED, 1, SLAVE_EBI_CH0);
+DEFINE_QNODE(snoc_bimc_1_slv, SNOC_BIMC_1_SLV, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SLAVE_EBI_CH0);
+DEFINE_QNODE(slv_ebi_ch0, SLAVE_EBI_CH0, 0, 8, 0, -1, 0, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_apps_l2, SLAVE_AMPSS_L2, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(snoc_pnoc_slv, MNOC_BIMC_SLV, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_0);
+DEFINE_QNODE(pnoc_int_0, PNOC_INT_0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 8, PNOC_SNOC_MAS, PNOC_SLV_0, PNOC_SLV_1, PNOC_SLV_2, PNOC_SLV_3, PNOC_SLV_4, PNOC_SLV_8, PNOC_SLV_9);
+DEFINE_QNODE(pnoc_int_1, PNOC_INT_1, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_SNOC_MAS);
+DEFINE_QNODE(pnoc_m_0, PNOC_M_0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_0);
+DEFINE_QNODE(pnoc_m_1, PNOC_M_1, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_SNOC_MAS);
+DEFINE_QNODE(pnoc_s_0, PNOC_SLV_0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_CLK_CTL, SLAVE_TLMM, SLAVE_MSM_TCSR, SLAVE_SECURITY, SLAVE_MSS);
+DEFINE_QNODE(pnoc_s_1, PNOC_SLV_1, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_IMEM_CFG, SLAVE_CRYPTO_0_CFG, SLAVE_RPM_MSG_RAM, SLAVE_MSM_PDM, SLAVE_PRNG);
+DEFINE_QNODE(pnoc_s_2, PNOC_SLV_2, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_SPDM, SLAVE_BOOT_ROM, SLAVE_BIMC_CFG, SLAVE_PNOC_CFG, SLAVE_PMIC_ARB);
+DEFINE_QNODE(pnoc_s_3, PNOC_SLV_3, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_MPM, SLAVE_IPS_CFG, SLAVE_RBCPR_CFG, SLAVE_QDSS_CFG, SLAVE_DEHR_CFG);
+DEFINE_QNODE(pnoc_s_4, PNOC_SLV_4, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_VENUS_CFG, SLAVE_CAMERA_CFG, SLAVE_DISPLAY_CFG);
+DEFINE_QNODE(pnoc_s_8, PNOC_SLV_8, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_USB_HS, SLAVE_SDCC_1, SLAVE_BLSP_1);
+DEFINE_QNODE(pnoc_s_9, PNOC_SLV_9, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_SDCC_4, SLAVE_LPASS, SLAVE_GRAPHICS_3D_CFG);
+DEFINE_QNODE(slv_imem_cfg, SLAVE_IMEM_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_crypto_0_cfg, SLAVE_CRYPTO_0_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_msg_ram, SLAVE_RPM_MSG_RAM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_pdm, SLAVE_MSM_PDM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_prng, SLAVE_PRNG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_clk_ctl, SLAVE_CLK_CTL, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_mss, SLAVE_MSS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_tlmm, SLAVE_TLMM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_tcsr, SLAVE_MSM_TCSR, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_security, SLAVE_SECURITY, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_spdm, SLAVE_SPDM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_pnoc_cfg, SLAVE_PNOC_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_pmic_arb, SLAVE_PMIC_ARB, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_bimc_cfg, SLAVE_BIMC_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_boot_rom, SLAVE_BOOT_ROM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_mpm, SLAVE_MPM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_qdss_cfg, SLAVE_QDSS_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_rbcpr_cfg, SLAVE_RBCPR_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_snoc_cfg, SLAVE_IPS_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_dehr_cfg, SLAVE_DEHR_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_venus_cfg, SLAVE_VENUS_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_display_cfg, SLAVE_DISPLAY_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_camera_cfg, SLAVE_CAMERA_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_usb_hs, SLAVE_USB_HS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_sdcc_1, SLAVE_SDCC_1, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_blsp_1, SLAVE_BLSP_1, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_sdcc_2, SLAVE_SDCC_4, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_gfx_cfg, SLAVE_GRAPHICS_3D_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_audio, SLAVE_LPASS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(mas_blsp_1, MASTER_BLSP_1, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_1);
+DEFINE_QNODE(mas_spdm, MASTER_SPDM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_0);
+DEFINE_QNODE(mas_dehr, MASTER_DEHR, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_0);
+DEFINE_QNODE(mas_audio, MASTER_LPASS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_0);
+DEFINE_QNODE(mas_usb_hs, MASTER_USB_HS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_1);
+DEFINE_QNODE(mas_pnoc_crypto_0, MASTER_CRYPTO_CORE0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_1);
+DEFINE_QNODE(mas_pnoc_sdcc_1, MASTER_SDCC_1, 7, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_1);
+DEFINE_QNODE(mas_pnoc_sdcc_2, MASTER_SDCC_2, 8, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_1);
+DEFINE_QNODE(pnoc_snoc_mas, PNOC_SNOC_MAS, 0, 8, 0, 29, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_SNOC_SLV);
+
+static struct qcom_icc_node *msm8916_snoc_nodes[] = {
+	&mas_video,
+	&mas_jpeg,
+	&mas_vfe,
+	&mas_mdp,
+	&mas_qdss_bam,
+	&mas_snoc_cfg,
+	&mas_qdss_etr,
+	&mm_int_0,
+	&mm_int_1,
+	&mm_int_2,
+	&mm_int_bimc,
+	&snoc_int_0,
+	&snoc_int_1,
+	&snoc_int_bimc,
+	&snoc_bimc_0_mas,
+	&snoc_bimc_1_mas,
+	&qdss_int,
+	&bimc_snoc_slv,
+	&snoc_pnoc_mas,
+	&pnoc_snoc_slv,
+	&slv_srvc_snoc,
+	&slv_qdss_stm,
+	&slv_imem,
+	&slv_apss,
+	&slv_cats_0,
+	&slv_cats_1,
+};
+
+static struct qcom_icc_desc msm8916_snoc = {
+	.nodes = msm8916_snoc_nodes,
+	.num_nodes = ARRAY_SIZE(msm8916_snoc_nodes),
+};
+
+static struct qcom_icc_node *msm8916_bimc_nodes[] = {
+	&mas_apss,
+	&mas_tcu0,
+	&mas_tcu1,
+	&mas_gfx,
+	&bimc_snoc_mas,
+	&snoc_bimc_0_slv,
+	&snoc_bimc_1_slv,
+	&slv_ebi_ch0,
+	&slv_apps_l2,
+};
+
+static struct qcom_icc_desc msm8916_bimc = {
+	.nodes = msm8916_bimc_nodes,
+	.num_nodes = ARRAY_SIZE(msm8916_bimc_nodes),
+};
+
+static struct qcom_icc_node *msm8916_pnoc_nodes[] = {
+	&snoc_pnoc_slv,
+	&pnoc_int_0,
+	&pnoc_int_1,
+	&pnoc_m_0,
+	&pnoc_m_1,
+	&pnoc_s_0,
+	&pnoc_s_1,
+	&pnoc_s_2,
+	&pnoc_s_3,
+	&pnoc_s_4,
+	&pnoc_s_8,
+	&pnoc_s_9,
+	&slv_imem_cfg,
+	&slv_crypto_0_cfg,
+	&slv_msg_ram,
+	&slv_pdm,
+	&slv_prng,
+	&slv_clk_ctl,
+	&slv_mss,
+	&slv_tlmm,
+	&slv_tcsr,
+	&slv_security,
+	&slv_spdm,
+	&slv_pnoc_cfg,
+	&slv_pmic_arb,
+	&slv_bimc_cfg,
+	&slv_boot_rom,
+	&slv_mpm,
+	&slv_qdss_cfg,
+	&slv_rbcpr_cfg,
+	&slv_snoc_cfg,
+	&slv_dehr_cfg,
+	&slv_venus_cfg,
+	&slv_display_cfg,
+	&slv_camera_cfg,
+	&slv_usb_hs,
+	&slv_sdcc_1,
+	&slv_blsp_1,
+	&slv_sdcc_2,
+	&slv_gfx_cfg,
+	&slv_audio,
+	&mas_blsp_1,
+	&mas_spdm,
+	&mas_dehr,
+	&mas_audio,
+	&mas_usb_hs,
+	&mas_pnoc_crypto_0,
+	&mas_pnoc_sdcc_1,
+	&mas_pnoc_sdcc_2,
+	&pnoc_snoc_mas,
+};
+
+static struct qcom_icc_desc msm8916_pnoc = {
+	.nodes = msm8916_pnoc_nodes,
+	.num_nodes = ARRAY_SIZE(msm8916_pnoc_nodes),
+};
+
+static int qcom_icc_init(struct icc_node *node)
+{
+	struct qcom_icc_provider *qp = to_qcom_provider(node->provider);
+	int ret;
+
+	/* TODO: init qos and priority */
+
+	clk_set_rate(qp->bus_clk, INT_MAX);
+	ret = clk_prepare_enable(qp->bus_clk);
+	if (ret)
+		pr_info("%s: error enabling bus clk (%d)\n", __func__, ret);
+	clk_set_rate(qp->bus_a_clk, INT_MAX);
+	ret = clk_prepare_enable(qp->bus_a_clk);
+	if (ret)
+		pr_info("%s: error enabling bus_a clk (%d)\n", __func__, ret);
+
+	return 0;
+}
+
+static int qcom_icc_aggregate(struct icc_node *node, u32 avg_bw, u32 peak_bw,
+			      u32 *agg_avg, u32 *agg_peak)
+{
+	*agg_avg += avg_bw;
+	*agg_peak = max(*agg_peak, peak_bw);
+
+	return 0;
+}
+
+static int qcom_icc_set(struct icc_node *src, struct icc_node *dst,
+			u32 avg, u32 peak)
+{
+	struct qcom_icc_provider *qp;
+	struct qcom_icc_node *qn;
+	struct icc_node *node;
+	struct icc_provider *provider;
+	u64 avg_bw = icc_units_to_bps(avg);
+	u64 peak_bw = icc_units_to_bps(peak);
+	u64 rate = 0;
+	int ret = 0;
+
+	if (!src)
+		node = dst;
+	else
+		node = src;
+
+	qn = node->data;
+	provider = node->provider;
+	qp = to_qcom_provider(node->provider);
+
+	/* set bandwidth */
+	if (qn->ap_owned) {
+		/* TODO: set QoS */
+	} else {
+		/* send message to the RPM processor */
+		if (qn->mas_rpm_id != -1) {
+			ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
+						    RPM_BUS_MASTER_REQ,
+						    qn->mas_rpm_id,
+						    avg_bw);
+		}
+
+		if (qn->slv_rpm_id != -1) {
+			ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
+						    RPM_BUS_SLAVE_REQ,
+						    qn->slv_rpm_id,
+						    avg_bw);
+		}
+	}
+
+	rate = max(avg_bw, peak_bw);
+
+	do_div(rate, qn->buswidth);
+
+	if (qn->rate != rate) {
+		ret = clk_set_rate(qp->bus_clk, rate);
+		if (ret) {
+			pr_err("set clk rate %lld error %d\n", rate, ret);
+			return ret;
+		}
+
+		ret = clk_set_rate(qp->bus_a_clk, rate);
+		if (ret) {
+			pr_err("set clk rate %lld error %d\n", rate, ret);
+			return ret;
+		}
+
+		qn->rate = rate;
+	}
+
+	return ret;
+}
+
+static int qnoc_probe(struct platform_device *pdev)
+{
+	const struct qcom_icc_desc *desc;
+	struct qcom_icc_node **qnodes;
+	struct qcom_icc_provider *qp;
+	struct resource *res;
+	struct icc_provider *provider;
+	size_t num_nodes, i;
+	int ret;
+
+	/* wait for RPM */
+	if (!qcom_icc_rpm_smd_available())
+		return -EPROBE_DEFER;
+
+	desc = of_device_get_match_data(&pdev->dev);
+	if (!desc)
+		return -EINVAL;
+
+	qnodes = desc->nodes;
+	num_nodes = desc->num_nodes;
+
+	qp = devm_kzalloc(&pdev->dev, sizeof(*qp), GFP_KERNEL);
+	if (!qp)
+		return -ENOMEM;
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	qp->base = devm_ioremap_resource(&pdev->dev, res);
+	if (IS_ERR(qp->base))
+		return PTR_ERR(qp->base);
+
+	qp->bus_clk = devm_clk_get(&pdev->dev, "bus_clk");
+	if (IS_ERR(qp->bus_clk))
+		return PTR_ERR(qp->bus_clk);
+
+	qp->bus_a_clk = devm_clk_get(&pdev->dev, "bus_a_clk");
+	if (IS_ERR(qp->bus_a_clk))
+		return PTR_ERR(qp->bus_a_clk);
+
+	provider = &qp->provider;
+	provider->dev = &pdev->dev;
+	provider->set = &qcom_icc_set;
+	provider->aggregate = &qcom_icc_aggregate;
+	INIT_LIST_HEAD(&provider->nodes);
+	provider->data = qp;
+
+	ret = icc_provider_add(provider);
+	if (ret) {
+		dev_err(&pdev->dev, "error adding interconnect provider\n");
+		return ret;
+	}
+
+	for (i = 0; i < num_nodes; i++) {
+		struct icc_node *node;
+		int ret;
+		size_t j;
+
+		node = icc_node_create(qnodes[i]->id);
+		if (IS_ERR(node)) {
+			ret = PTR_ERR(node);
+			goto err;
+		}
+
+		node->name = qnodes[i]->name;
+		node->data = qnodes[i];
+		icc_node_add(node, provider);
+
+		dev_dbg(&pdev->dev, "registered node %s\n", node->name);
+
+		ret = qcom_icc_init(node);
+		if (ret)
+			dev_err(&pdev->dev, "%s init error (%d)\n", node->name,
+				ret);
+
+		/* populate links */
+		for (j = 0; j < qnodes[i]->num_links; j++)
+			if (qnodes[i]->links[j])
+				icc_link_create(node, qnodes[i]->links[j]);
+	}
+
+	platform_set_drvdata(pdev, provider);
+
+	return ret;
+err:
+	icc_provider_del(provider);
+	return ret;
+}
+
+static int qnoc_remove(struct platform_device *pdev)
+{
+	struct icc_provider *provider = platform_get_drvdata(pdev);
+
+	icc_provider_del(provider);
+
+	return 0;
+}
+
+static const struct of_device_id qnoc_of_match[] = {
+	{ .compatible = "qcom,msm8916-pnoc", .data = &msm8916_pnoc },
+	{ .compatible = "qcom,msm8916-snoc", .data = &msm8916_snoc },
+	{ .compatible = "qcom,msm8916-bimc", .data = &msm8916_bimc },
+	{ },
+};
+MODULE_DEVICE_TABLE(of, qnoc_of_match);
+
+static struct platform_driver qnoc_driver = {
+	.probe = qnoc_probe,
+	.remove = qnoc_remove,
+	.driver = {
+		.name = "qnoc-msm8916",
+		.of_match_table = qnoc_of_match,
+	},
+};
+module_platform_driver(qnoc_driver);
+MODULE_AUTHOR("Georgi Djakov <georgi.djakov@linaro.org>");
+MODULE_DESCRIPTION("Qualcomm msm8916 NoC driver");
+MODULE_LICENSE("GPL v2");

[PATCH v5 6/8] interconnect: qcom: Add msm8916 interconnect provider driver

[Georgi Djakov]

Add driver for the Qualcomm interconnect buses found in msm8916 based
platforms.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 drivers/interconnect/Kconfig        |   5 +
 drivers/interconnect/Makefile       |   1 +
 drivers/interconnect/qcom/Kconfig   |  11 +
 drivers/interconnect/qcom/Makefile  |   2 +
 drivers/interconnect/qcom/msm8916.c | 495 ++++++++++++++++++++++++++++
 5 files changed, 514 insertions(+)
 create mode 100644 drivers/interconnect/qcom/Kconfig
 create mode 100644 drivers/interconnect/qcom/msm8916.c

diff --git a/drivers/interconnect/Kconfig b/drivers/interconnect/Kconfig
index a261c7d41deb..07a8276fa35a 100644
--- a/drivers/interconnect/Kconfig
+++ b/drivers/interconnect/Kconfig
@@ -8,3 +8,8 @@ menuconfig INTERCONNECT
 
 	  If unsure, say no.
 
+if INTERCONNECT
+
+source "drivers/interconnect/qcom/Kconfig"
+
+endif
diff --git a/drivers/interconnect/Makefile b/drivers/interconnect/Makefile
index 97fca2e09d24..7944cbca0527 100644
--- a/drivers/interconnect/Makefile
+++ b/drivers/interconnect/Makefile
@@ -1,2 +1,3 @@
 # SPDX-License-Identifier: GPL-2.0
 obj-$(CONFIG_INTERCONNECT)		+= core.o
+obj-$(CONFIG_INTERCONNECT_QCOM)		+= qcom/
diff --git a/drivers/interconnect/qcom/Kconfig b/drivers/interconnect/qcom/Kconfig
new file mode 100644
index 000000000000..86465dc37bd4
--- /dev/null
+++ b/drivers/interconnect/qcom/Kconfig
@@ -0,0 +1,11 @@
+config INTERCONNECT_QCOM
+	bool "Qualcomm Network-on-Chip interconnect drivers"
+	depends on INTERCONNECT
+	depends on ARCH_QCOM || COMPILE_TEST
+	default y
+
+config INTERCONNECT_QCOM_MSM8916
+	tristate "Qualcomm MSM8916 interconnect driver"
+	depends on INTERCONNECT_QCOM
+	help
+	  This is a driver for the Qualcomm Network-on-Chip on msm8916-based platforms.
diff --git a/drivers/interconnect/qcom/Makefile b/drivers/interconnect/qcom/Makefile
index 2a0c41db91c4..e8b24c334c5d 100644
--- a/drivers/interconnect/qcom/Makefile
+++ b/drivers/interconnect/qcom/Makefile
@@ -1,2 +1,4 @@
 # SPDX-License-Identifier: GPL-2.0
 obj-y += smd-rpm.o
+
+obj-$(CONFIG_INTERCONNECT_QCOM_MSM8916) += msm8916.o
diff --git a/drivers/interconnect/qcom/msm8916.c b/drivers/interconnect/qcom/msm8916.c
new file mode 100644
index 000000000000..c34f18cdea33
--- /dev/null
+++ b/drivers/interconnect/qcom/msm8916.c
@@ -0,0 +1,495 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2018 Linaro Ltd
+ * Author: Georgi Djakov <georgi.djakov@linaro.org>
+ */
+
+#include <dt-bindings/interconnect/qcom.h>
+#include <linux/clk.h>
+#include <linux/device.h>
+#include <linux/io.h>
+#include <linux/interconnect-provider.h>
+#include <linux/module.h>
+#include <linux/of_device.h>
+#include <linux/of_platform.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+
+#include "smd-rpm.h"
+
+#define RPM_MASTER_FIELD_BW	0x00007762
+#define RPM_BUS_MASTER_REQ      0x73616d62
+#define RPM_BUS_SLAVE_REQ       0x766c7362
+
+#define to_qcom_provider(_provider) \
+	container_of(_provider, struct qcom_icc_provider, provider)
+
+enum qcom_qos_mode {
+	QCOM_QOS_MODE_BYPASS = 0,
+	QCOM_QOS_MODE_FIXED,
+	QCOM_QOS_MODE_MAX,
+};
+
+struct qcom_icc_provider {
+	struct icc_provider	provider;
+	void __iomem		*base;
+	struct clk		*bus_clk;
+	struct clk		*bus_a_clk;
+};
+
+#define MSM8916_MAX_LINKS	8
+
+/**
+ * struct qcom_icc_node - Qualcomm specific interconnect nodes
+ * @name: the node name used in debugfs
+ * @links: an array of nodes where we can go next while traversing
+ * @id: a unique node identifier
+ * @num_links: the total number of @links
+ * @port: the offset index into the masters QoS register space
+ * @buswidth: width of the interconnect between a node and the bus (bytes)
+ * @ap_owned: the AP CPU does the writing to QoS registers
+ * @rpm: reference to the RPM SMD driver
+ * @qos_mode: QoS mode for ap_owned resources
+ * @mas_rpm_id:	RPM id for devices that are bus masters
+ * @slv_rpm_id:	RPM id for devices that are bus slaves
+ * @rate: current bus clock rate in Hz
+ */
+struct qcom_icc_node {
+	unsigned char *name;
+	u16 links[MSM8916_MAX_LINKS];
+	u16 id;
+	u16 num_links;
+	u16 port;
+	u16 buswidth;
+	bool ap_owned;
+	struct qcom_smd_rpm *rpm;
+	enum qcom_qos_mode qos_mode;
+	int mas_rpm_id;
+	int slv_rpm_id;
+	u64 rate;
+};
+
+struct qcom_icc_desc {
+	struct qcom_icc_node **nodes;
+	size_t num_nodes;
+};
+
+#define DEFINE_QNODE(_name, _id, _port, _buswidth, _ap_owned,		\
+			_mas_rpm_id, _slv_rpm_id, _qos_mode,		\
+			_numlinks, ...)					\
+		static struct qcom_icc_node _name = {			\
+		.id = _id,						\
+		.name = #_name,						\
+		.port = _port,						\
+		.buswidth = _buswidth,					\
+		.qos_mode = _qos_mode,					\
+		.ap_owned = _ap_owned,					\
+		.mas_rpm_id = _mas_rpm_id,				\
+		.slv_rpm_id = _slv_rpm_id,				\
+		.num_links = _numlinks,					\
+		.links = { __VA_ARGS__ },				\
+	}
+
+DEFINE_QNODE(mas_video, MASTER_VIDEO_P0, 8, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_0, SNOC_MM_INT_2);
+DEFINE_QNODE(mas_jpeg, MASTER_JPEG, 6, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_0, SNOC_MM_INT_2);
+DEFINE_QNODE(mas_vfe, MASTER_VFE, 9, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_1, SNOC_MM_INT_2);
+DEFINE_QNODE(mas_mdp, MASTER_MDP_PORT0, 7, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_0, SNOC_MM_INT_2);
+DEFINE_QNODE(mas_qdss_bam, MASTER_QDSS_BAM, 11, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_QDSS_INT);
+DEFINE_QNODE(mas_snoc_cfg, MASTER_SNOC_CFG, 0, 16, 0, 20, -1, QCOM_QOS_MODE_BYPASS, 1, SNOC_QDSS_INT);
+DEFINE_QNODE(mas_qdss_etr, MASTER_QDSS_ETR, 10, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_QDSS_INT);
+DEFINE_QNODE(mm_int_0, SNOC_MM_INT_0, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_MM_INT_BIMC);
+DEFINE_QNODE(mm_int_1, SNOC_MM_INT_1, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_MM_INT_BIMC);
+DEFINE_QNODE(mm_int_2, SNOC_MM_INT_2, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_INT_0);
+DEFINE_QNODE(mm_int_bimc, SNOC_MM_INT_BIMC, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_1_MAS);
+DEFINE_QNODE(snoc_int_0, SNOC_INT_0, 0, 8, 0, 99, 130, QCOM_QOS_MODE_FIXED, 3, SLAVE_QDSS_STM, SLAVE_SYSTEM_IMEM, MNOC_BIMC_MAS);
+DEFINE_QNODE(snoc_int_1, SNOC_INT_1, 0, 8, 0, 100, 131, QCOM_QOS_MODE_FIXED, 3, SYSTEM_SLAVE_FAB_APPS, SLAVE_CATS_128, SLAVE_OCMEM_64);
+DEFINE_QNODE(snoc_int_bimc, SNOC_INT_BIMC, 0, 8, 0, 101, 132, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_0_MAS);
+DEFINE_QNODE(snoc_bimc_0_mas, SNOC_BIMC_0_MAS, 0, 8, 0, 3, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_0_SLV);
+DEFINE_QNODE(snoc_bimc_1_mas, SNOC_BIMC_1_MAS, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_1_SLV);
+DEFINE_QNODE(qdss_int, SNOC_QDSS_INT, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 2, SNOC_INT_0, SNOC_INT_BIMC);
+DEFINE_QNODE(bimc_snoc_slv, BIMC_SNOC_SLV, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 2, SNOC_INT_0, SNOC_INT_1);
+DEFINE_QNODE(snoc_pnoc_mas, MNOC_BIMC_MAS, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, MNOC_BIMC_SLV);
+DEFINE_QNODE(pnoc_snoc_slv, PNOC_SNOC_SLV, 0, 8, 0, -1, 45, QCOM_QOS_MODE_FIXED, 3, SNOC_INT_0, SNOC_INT_BIMC, SNOC_INT_1);
+DEFINE_QNODE(slv_srvc_snoc, SLAVE_SERVICE_SNOC, 0, 8, 0, -1, 29, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_qdss_stm, SLAVE_QDSS_STM, 0, 4, 0, -1, 30, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_imem, SLAVE_SYSTEM_IMEM, 0, 8, 0, -1, 26, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_apss, SYSTEM_SLAVE_FAB_APPS, 0, 4, 0, -1, 20, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_cats_0, SLAVE_CATS_128, 0, 16, 0, -1, 106, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_cats_1, SLAVE_OCMEM_64, 0, 8, 0, -1, 107, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(mas_apss, MASTER_AMPSS_M0, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
+DEFINE_QNODE(mas_tcu0, MASTER_TCU_0, 5, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
+DEFINE_QNODE(mas_tcu1, MASTER_TCU_1, 6, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
+DEFINE_QNODE(mas_gfx, MASTER_GRAPHICS_3D, 2, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
+DEFINE_QNODE(bimc_snoc_mas, BIMC_SNOC_MAS, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, BIMC_SNOC_SLV);
+DEFINE_QNODE(snoc_bimc_0_slv, SNOC_BIMC_0_SLV, 0, 8, 0, -1, 24, QCOM_QOS_MODE_FIXED, 1, SLAVE_EBI_CH0);
+DEFINE_QNODE(snoc_bimc_1_slv, SNOC_BIMC_1_SLV, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SLAVE_EBI_CH0);
+DEFINE_QNODE(slv_ebi_ch0, SLAVE_EBI_CH0, 0, 8, 0, -1, 0, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_apps_l2, SLAVE_AMPSS_L2, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(snoc_pnoc_slv, MNOC_BIMC_SLV, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_0);
+DEFINE_QNODE(pnoc_int_0, PNOC_INT_0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 8, PNOC_SNOC_MAS, PNOC_SLV_0, PNOC_SLV_1, PNOC_SLV_2, PNOC_SLV_3, PNOC_SLV_4, PNOC_SLV_8, PNOC_SLV_9);
+DEFINE_QNODE(pnoc_int_1, PNOC_INT_1, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_SNOC_MAS);
+DEFINE_QNODE(pnoc_m_0, PNOC_M_0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_0);
+DEFINE_QNODE(pnoc_m_1, PNOC_M_1, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_SNOC_MAS);
+DEFINE_QNODE(pnoc_s_0, PNOC_SLV_0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_CLK_CTL, SLAVE_TLMM, SLAVE_MSM_TCSR, SLAVE_SECURITY, SLAVE_MSS);
+DEFINE_QNODE(pnoc_s_1, PNOC_SLV_1, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_IMEM_CFG, SLAVE_CRYPTO_0_CFG, SLAVE_RPM_MSG_RAM, SLAVE_MSM_PDM, SLAVE_PRNG);
+DEFINE_QNODE(pnoc_s_2, PNOC_SLV_2, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_SPDM, SLAVE_BOOT_ROM, SLAVE_BIMC_CFG, SLAVE_PNOC_CFG, SLAVE_PMIC_ARB);
+DEFINE_QNODE(pnoc_s_3, PNOC_SLV_3, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_MPM, SLAVE_IPS_CFG, SLAVE_RBCPR_CFG, SLAVE_QDSS_CFG, SLAVE_DEHR_CFG);
+DEFINE_QNODE(pnoc_s_4, PNOC_SLV_4, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_VENUS_CFG, SLAVE_CAMERA_CFG, SLAVE_DISPLAY_CFG);
+DEFINE_QNODE(pnoc_s_8, PNOC_SLV_8, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_USB_HS, SLAVE_SDCC_1, SLAVE_BLSP_1);
+DEFINE_QNODE(pnoc_s_9, PNOC_SLV_9, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_SDCC_4, SLAVE_LPASS, SLAVE_GRAPHICS_3D_CFG);
+DEFINE_QNODE(slv_imem_cfg, SLAVE_IMEM_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_crypto_0_cfg, SLAVE_CRYPTO_0_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_msg_ram, SLAVE_RPM_MSG_RAM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_pdm, SLAVE_MSM_PDM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_prng, SLAVE_PRNG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_clk_ctl, SLAVE_CLK_CTL, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_mss, SLAVE_MSS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_tlmm, SLAVE_TLMM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_tcsr, SLAVE_MSM_TCSR, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_security, SLAVE_SECURITY, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_spdm, SLAVE_SPDM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_pnoc_cfg, SLAVE_PNOC_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_pmic_arb, SLAVE_PMIC_ARB, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_bimc_cfg, SLAVE_BIMC_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_boot_rom, SLAVE_BOOT_ROM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_mpm, SLAVE_MPM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_qdss_cfg, SLAVE_QDSS_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_rbcpr_cfg, SLAVE_RBCPR_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_snoc_cfg, SLAVE_IPS_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_dehr_cfg, SLAVE_DEHR_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_venus_cfg, SLAVE_VENUS_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_display_cfg, SLAVE_DISPLAY_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_camera_cfg, SLAVE_CAMERA_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_usb_hs, SLAVE_USB_HS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_sdcc_1, SLAVE_SDCC_1, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_blsp_1, SLAVE_BLSP_1, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_sdcc_2, SLAVE_SDCC_4, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_gfx_cfg, SLAVE_GRAPHICS_3D_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(slv_audio, SLAVE_LPASS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
+DEFINE_QNODE(mas_blsp_1, MASTER_BLSP_1, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_1);
+DEFINE_QNODE(mas_spdm, MASTER_SPDM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_0);
+DEFINE_QNODE(mas_dehr, MASTER_DEHR, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_0);
+DEFINE_QNODE(mas_audio, MASTER_LPASS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_0);
+DEFINE_QNODE(mas_usb_hs, MASTER_USB_HS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_1);
+DEFINE_QNODE(mas_pnoc_crypto_0, MASTER_CRYPTO_CORE0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_1);
+DEFINE_QNODE(mas_pnoc_sdcc_1, MASTER_SDCC_1, 7, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_1);
+DEFINE_QNODE(mas_pnoc_sdcc_2, MASTER_SDCC_2, 8, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_1);
+DEFINE_QNODE(pnoc_snoc_mas, PNOC_SNOC_MAS, 0, 8, 0, 29, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_SNOC_SLV);
+
+static struct qcom_icc_node *msm8916_snoc_nodes[] = {
+	&mas_video,
+	&mas_jpeg,
+	&mas_vfe,
+	&mas_mdp,
+	&mas_qdss_bam,
+	&mas_snoc_cfg,
+	&mas_qdss_etr,
+	&mm_int_0,
+	&mm_int_1,
+	&mm_int_2,
+	&mm_int_bimc,
+	&snoc_int_0,
+	&snoc_int_1,
+	&snoc_int_bimc,
+	&snoc_bimc_0_mas,
+	&snoc_bimc_1_mas,
+	&qdss_int,
+	&bimc_snoc_slv,
+	&snoc_pnoc_mas,
+	&pnoc_snoc_slv,
+	&slv_srvc_snoc,
+	&slv_qdss_stm,
+	&slv_imem,
+	&slv_apss,
+	&slv_cats_0,
+	&slv_cats_1,
+};
+
+static struct qcom_icc_desc msm8916_snoc = {
+	.nodes = msm8916_snoc_nodes,
+	.num_nodes = ARRAY_SIZE(msm8916_snoc_nodes),
+};
+
+static struct qcom_icc_node *msm8916_bimc_nodes[] = {
+	&mas_apss,
+	&mas_tcu0,
+	&mas_tcu1,
+	&mas_gfx,
+	&bimc_snoc_mas,
+	&snoc_bimc_0_slv,
+	&snoc_bimc_1_slv,
+	&slv_ebi_ch0,
+	&slv_apps_l2,
+};
+
+static struct qcom_icc_desc msm8916_bimc = {
+	.nodes = msm8916_bimc_nodes,
+	.num_nodes = ARRAY_SIZE(msm8916_bimc_nodes),
+};
+
+static struct qcom_icc_node *msm8916_pnoc_nodes[] = {
+	&snoc_pnoc_slv,
+	&pnoc_int_0,
+	&pnoc_int_1,
+	&pnoc_m_0,
+	&pnoc_m_1,
+	&pnoc_s_0,
+	&pnoc_s_1,
+	&pnoc_s_2,
+	&pnoc_s_3,
+	&pnoc_s_4,
+	&pnoc_s_8,
+	&pnoc_s_9,
+	&slv_imem_cfg,
+	&slv_crypto_0_cfg,
+	&slv_msg_ram,
+	&slv_pdm,
+	&slv_prng,
+	&slv_clk_ctl,
+	&slv_mss,
+	&slv_tlmm,
+	&slv_tcsr,
+	&slv_security,
+	&slv_spdm,
+	&slv_pnoc_cfg,
+	&slv_pmic_arb,
+	&slv_bimc_cfg,
+	&slv_boot_rom,
+	&slv_mpm,
+	&slv_qdss_cfg,
+	&slv_rbcpr_cfg,
+	&slv_snoc_cfg,
+	&slv_dehr_cfg,
+	&slv_venus_cfg,
+	&slv_display_cfg,
+	&slv_camera_cfg,
+	&slv_usb_hs,
+	&slv_sdcc_1,
+	&slv_blsp_1,
+	&slv_sdcc_2,
+	&slv_gfx_cfg,
+	&slv_audio,
+	&mas_blsp_1,
+	&mas_spdm,
+	&mas_dehr,
+	&mas_audio,
+	&mas_usb_hs,
+	&mas_pnoc_crypto_0,
+	&mas_pnoc_sdcc_1,
+	&mas_pnoc_sdcc_2,
+	&pnoc_snoc_mas,
+};
+
+static struct qcom_icc_desc msm8916_pnoc = {
+	.nodes = msm8916_pnoc_nodes,
+	.num_nodes = ARRAY_SIZE(msm8916_pnoc_nodes),
+};
+
+static int qcom_icc_init(struct icc_node *node)
+{
+	struct qcom_icc_provider *qp = to_qcom_provider(node->provider);
+	int ret;
+
+	/* TODO: init qos and priority */
+
+	clk_set_rate(qp->bus_clk, INT_MAX);
+	ret = clk_prepare_enable(qp->bus_clk);
+	if (ret)
+		pr_info("%s: error enabling bus clk (%d)\n", __func__, ret);
+	clk_set_rate(qp->bus_a_clk, INT_MAX);
+	ret = clk_prepare_enable(qp->bus_a_clk);
+	if (ret)
+		pr_info("%s: error enabling bus_a clk (%d)\n", __func__, ret);
+
+	return 0;
+}
+
+static int qcom_icc_aggregate(struct icc_node *node, u32 avg_bw, u32 peak_bw,
+			      u32 *agg_avg, u32 *agg_peak)
+{
+	*agg_avg += avg_bw;
+	*agg_peak = max(*agg_peak, peak_bw);
+
+	return 0;
+}
+
+static int qcom_icc_set(struct icc_node *src, struct icc_node *dst,
+			u32 avg, u32 peak)
+{
+	struct qcom_icc_provider *qp;
+	struct qcom_icc_node *qn;
+	struct icc_node *node;
+	struct icc_provider *provider;
+	u64 avg_bw = icc_units_to_bps(avg);
+	u64 peak_bw = icc_units_to_bps(peak);
+	u64 rate = 0;
+	int ret = 0;
+
+	if (!src)
+		node = dst;
+	else
+		node = src;
+
+	qn = node->data;
+	provider = node->provider;
+	qp = to_qcom_provider(node->provider);
+
+	/* set bandwidth */
+	if (qn->ap_owned) {
+		/* TODO: set QoS */
+	} else {
+		/* send message to the RPM processor */
+		if (qn->mas_rpm_id != -1) {
+			ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
+						    RPM_BUS_MASTER_REQ,
+						    qn->mas_rpm_id,
+						    avg_bw);
+		}
+
+		if (qn->slv_rpm_id != -1) {
+			ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
+						    RPM_BUS_SLAVE_REQ,
+						    qn->slv_rpm_id,
+						    avg_bw);
+		}
+	}
+
+	rate = max(avg_bw, peak_bw);
+
+	do_div(rate, qn->buswidth);
+
+	if (qn->rate != rate) {
+		ret = clk_set_rate(qp->bus_clk, rate);
+		if (ret) {
+			pr_err("set clk rate %lld error %d\n", rate, ret);
+			return ret;
+		}
+
+		ret = clk_set_rate(qp->bus_a_clk, rate);
+		if (ret) {
+			pr_err("set clk rate %lld error %d\n", rate, ret);
+			return ret;
+		}
+
+		qn->rate = rate;
+	}
+
+	return ret;
+}
+
+static int qnoc_probe(struct platform_device *pdev)
+{
+	const struct qcom_icc_desc *desc;
+	struct qcom_icc_node **qnodes;
+	struct qcom_icc_provider *qp;
+	struct resource *res;
+	struct icc_provider *provider;
+	size_t num_nodes, i;
+	int ret;
+
+	/* wait for RPM */
+	if (!qcom_icc_rpm_smd_available())
+		return -EPROBE_DEFER;
+
+	desc = of_device_get_match_data(&pdev->dev);
+	if (!desc)
+		return -EINVAL;
+
+	qnodes = desc->nodes;
+	num_nodes = desc->num_nodes;
+
+	qp = devm_kzalloc(&pdev->dev, sizeof(*qp), GFP_KERNEL);
+	if (!qp)
+		return -ENOMEM;
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	qp->base = devm_ioremap_resource(&pdev->dev, res);
+	if (IS_ERR(qp->base))
+		return PTR_ERR(qp->base);
+
+	qp->bus_clk = devm_clk_get(&pdev->dev, "bus_clk");
+	if (IS_ERR(qp->bus_clk))
+		return PTR_ERR(qp->bus_clk);
+
+	qp->bus_a_clk = devm_clk_get(&pdev->dev, "bus_a_clk");
+	if (IS_ERR(qp->bus_a_clk))
+		return PTR_ERR(qp->bus_a_clk);
+
+	provider = &qp->provider;
+	provider->dev = &pdev->dev;
+	provider->set = &qcom_icc_set;
+	provider->aggregate = &qcom_icc_aggregate;
+	INIT_LIST_HEAD(&provider->nodes);
+	provider->data = qp;
+
+	ret = icc_provider_add(provider);
+	if (ret) {
+		dev_err(&pdev->dev, "error adding interconnect provider\n");
+		return ret;
+	}
+
+	for (i = 0; i < num_nodes; i++) {
+		struct icc_node *node;
+		int ret;
+		size_t j;
+
+		node = icc_node_create(qnodes[i]->id);
+		if (IS_ERR(node)) {
+			ret = PTR_ERR(node);
+			goto err;
+		}
+
+		node->name = qnodes[i]->name;
+		node->data = qnodes[i];
+		icc_node_add(node, provider);
+
+		dev_dbg(&pdev->dev, "registered node %s\n", node->name);
+
+		ret = qcom_icc_init(node);
+		if (ret)
+			dev_err(&pdev->dev, "%s init error (%d)\n", node->name,
+				ret);
+
+		/* populate links */
+		for (j = 0; j < qnodes[i]->num_links; j++)
+			if (qnodes[i]->links[j])
+				icc_link_create(node, qnodes[i]->links[j]);
+	}
+
+	platform_set_drvdata(pdev, provider);
+
+	return ret;
+err:
+	icc_provider_del(provider);
+	return ret;
+}
+
+static int qnoc_remove(struct platform_device *pdev)
+{
+	struct icc_provider *provider = platform_get_drvdata(pdev);
+
+	icc_provider_del(provider);
+
+	return 0;
+}
+
+static const struct of_device_id qnoc_of_match[] = {
+	{ .compatible = "qcom,msm8916-pnoc", .data = &msm8916_pnoc },
+	{ .compatible = "qcom,msm8916-snoc", .data = &msm8916_snoc },
+	{ .compatible = "qcom,msm8916-bimc", .data = &msm8916_bimc },
+	{ },
+};
+MODULE_DEVICE_TABLE(of, qnoc_of_match);
+
+static struct platform_driver qnoc_driver = {
+	.probe = qnoc_probe,
+	.remove = qnoc_remove,
+	.driver = {
+		.name = "qnoc-msm8916",
+		.of_match_table = qnoc_of_match,
+	},
+};
+module_platform_driver(qnoc_driver);
+MODULE_AUTHOR("Georgi Djakov <georgi.djakov@linaro.org>");
+MODULE_DESCRIPTION("Qualcomm msm8916 NoC driver");
+MODULE_LICENSE("GPL v2");

[PATCH v5 7/8] dt-bindings: Introduce interconnect consumers bindings

[Georgi Djakov]

Add documentation for the interconnect consumer bindings, that will allow
to link a device node (consumer) to its interconnect controller hardware.

Tha aim is to enable drivers to request a framework API to configure an
interconnect path by providing their struct device pointer and a name.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 .../bindings/interconnect/interconnect.txt    | 27 +++++++++++++++++++
 1 file changed, 27 insertions(+)

diff --git a/Documentation/devicetree/bindings/interconnect/interconnect.txt b/Documentation/devicetree/bindings/interconnect/interconnect.txt
index 6e2b2971b094..0ad65dccbe8b 100644
--- a/Documentation/devicetree/bindings/interconnect/interconnect.txt
+++ b/Documentation/devicetree/bindings/interconnect/interconnect.txt
@@ -31,3 +31,30 @@ Example:
 			clocks = <&rpmcc RPM_SMD_SNOC_CLK>,
 				 <&rpmcc RPM_SMD_SNOC_A_CLK>;
 		};
+
+
+= interconnect consumers =
+
+The interconnect consumers are device nodes which consume the interconnect
+path(s) provided by the interconnect provider. There can be multiple
+interconnect providers on a SoC and the consumer may consume multiple paths
+from different providers depending on use case and the components it has to
+interact with.
+
+Required properties:
+interconnects : Pairs of phandles and interconnect provider specifier to denote
+	        the edge source and destination ports of the interconnect path.
+
+Optional properties:
+interconnect-names : List of interconnect path name strings sorted in the same
+		     order as the interconnects property. Consumers drivers will use
+		     interconnect-names to match interconnect paths with interconnect
+		     specifiers.
+
+Example:
+
+	sdhci@7864000 {
+		...
+		interconnects = <&pnoc MASTER_SDCC_1 &bimc SLAVE_EBI_CH0>;
+		interconnect-names = "ddr";
+	};

[PATCH v5 7/8] dt-bindings: Introduce interconnect consumers bindings

[Georgi Djakov]

Add documentation for the interconnect consumer bindings, that will allow
to link a device node (consumer) to its interconnect controller hardware.

Tha aim is to enable drivers to request a framework API to configure an
interconnect path by providing their struct device pointer and a name.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 .../bindings/interconnect/interconnect.txt    | 27 +++++++++++++++++++
 1 file changed, 27 insertions(+)

diff --git a/Documentation/devicetree/bindings/interconnect/interconnect.txt b/Documentation/devicetree/bindings/interconnect/interconnect.txt
index 6e2b2971b094..0ad65dccbe8b 100644
--- a/Documentation/devicetree/bindings/interconnect/interconnect.txt
+++ b/Documentation/devicetree/bindings/interconnect/interconnect.txt
@@ -31,3 +31,30 @@ Example:
 			clocks = <&rpmcc RPM_SMD_SNOC_CLK>,
 				 <&rpmcc RPM_SMD_SNOC_A_CLK>;
 		};
+
+
+= interconnect consumers =
+
+The interconnect consumers are device nodes which consume the interconnect
+path(s) provided by the interconnect provider. There can be multiple
+interconnect providers on a SoC and the consumer may consume multiple paths
+from different providers depending on use case and the components it has to
+interact with.
+
+Required properties:
+interconnects : Pairs of phandles and interconnect provider specifier to denote
+	        the edge source and destination ports of the interconnect path.
+
+Optional properties:
+interconnect-names : List of interconnect path name strings sorted in the same
+		     order as the interconnects property. Consumers drivers will use
+		     interconnect-names to match interconnect paths with interconnect
+		     specifiers.
+
+Example:
+
+	sdhci@7864000 {
+		...
+		interconnects = <&pnoc MASTER_SDCC_1 &bimc SLAVE_EBI_CH0>;
+		interconnect-names = "ddr";
+	};

[PATCH v5 7/8] dt-bindings: Introduce interconnect consumers bindings

[Georgi Djakov]

Add documentation for the interconnect consumer bindings, that will allow
to link a device node (consumer) to its interconnect controller hardware.

Tha aim is to enable drivers to request a framework API to configure an
interconnect path by providing their struct device pointer and a name.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 .../bindings/interconnect/interconnect.txt    | 27 +++++++++++++++++++
 1 file changed, 27 insertions(+)

diff --git a/Documentation/devicetree/bindings/interconnect/interconnect.txt b/Documentation/devicetree/bindings/interconnect/interconnect.txt
index 6e2b2971b094..0ad65dccbe8b 100644
--- a/Documentation/devicetree/bindings/interconnect/interconnect.txt
+++ b/Documentation/devicetree/bindings/interconnect/interconnect.txt
@@ -31,3 +31,30 @@ Example:
 			clocks = <&rpmcc RPM_SMD_SNOC_CLK>,
 				 <&rpmcc RPM_SMD_SNOC_A_CLK>;
 		};
+
+
+= interconnect consumers =
+
+The interconnect consumers are device nodes which consume the interconnect
+path(s) provided by the interconnect provider. There can be multiple
+interconnect providers on a SoC and the consumer may consume multiple paths
+from different providers depending on use case and the components it has to
+interact with.
+
+Required properties:
+interconnects : Pairs of phandles and interconnect provider specifier to denote
+	        the edge source and destination ports of the interconnect path.
+
+Optional properties:
+interconnect-names : List of interconnect path name strings sorted in the same
+		     order as the interconnects property. Consumers drivers will use
+		     interconnect-names to match interconnect paths with interconnect
+		     specifiers.
+
+Example:
+
+	sdhci at 7864000 {
+		...
+		interconnects = <&pnoc MASTER_SDCC_1 &bimc SLAVE_EBI_CH0>;
+		interconnect-names = "ddr";
+	};

[PATCH v5 8/8] interconnect: Allow endpoints translation via DT

[Georgi Djakov]

Currently we support only platform data for specifying the interconnect
endpoints. As now the endpoints are hard-coded into the consumer driver
this may lead to complications when a single driver is used by multiple
SoCs, which may have different interconnect topology.
To avoid cluttering the consumer drivers, introduce a translation function
to help us get the board specific interconnect data from device-tree.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 drivers/interconnect/core.c  | 62 ++++++++++++++++++++++++++++++++++++
 include/linux/interconnect.h |  7 ++++
 2 files changed, 69 insertions(+)

diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
index dafc001ce758..2f07807ca4b4 100644
--- a/drivers/interconnect/core.c
+++ b/drivers/interconnect/core.c
@@ -16,6 +16,7 @@
 #include <linux/module.h>
 #include <linux/mutex.h>
 #include <linux/slab.h>
+#include <linux/of.h>
 #include <linux/uaccess.h>
 
 static DEFINE_IDR(icc_idr);
@@ -306,6 +307,67 @@ static int apply_constraints(struct icc_path *path)
 	return ret;
 }
 
+struct icc_path *of_icc_get(struct device *dev, const char *name)
+{
+	struct device_node *np = NULL;
+	struct of_phandle_args src_args, dst_args;
+	u32 src_id, dst_id;
+	int idx = 0;
+	int ret;
+
+	if (!dev || !dev->of_node)
+		return ERR_PTR(-ENODEV);
+
+	np = dev->of_node;
+
+	/*
+	 * When the consumer DT node do not have "interconnects" property
+	 * return a NULL path to skip setting constraints.
+	 */
+	if (!of_find_property(np, "interconnects", NULL))
+		return NULL;
+
+	/*
+	 * We use a combination of phandle and specifier for endpoint. For now
+	 * lets support only global ids and extend this is the future if needed
+	 * without breaking DT compatibility.
+	 */
+	if (name) {
+		idx = of_property_match_string(np, "interconnect-names", name);
+		if (idx < 0)
+			return ERR_PTR(idx);
+	}
+
+	ret = of_parse_phandle_with_args(np, "interconnects",
+					 "#interconnect-cells", idx * 2,
+					 &src_args);
+	if (ret)
+		return ERR_PTR(ret);
+
+	of_node_put(src_args.np);
+
+	if (!src_args.args_count || src_args.args_count > 1)
+		return ERR_PTR(-EINVAL);
+
+	src_id = src_args.args[0];
+
+	ret = of_parse_phandle_with_args(np, "interconnects",
+					 "#interconnect-cells", idx * 2 + 1,
+					 &dst_args);
+	if (ret)
+		return ERR_PTR(ret);
+
+	of_node_put(dst_args.np);
+
+	if (!dst_args.args_count || dst_args.args_count > 1)
+		return ERR_PTR(-EINVAL);
+
+	dst_id = dst_args.args[0];
+
+	return icc_get(dev, src_id, dst_id);
+}
+EXPORT_SYMBOL_GPL(of_icc_get);
+
 /**
  * icc_set() - set constraints on an interconnect path between two endpoints
  * @path: reference to the path returned by icc_get()
diff --git a/include/linux/interconnect.h b/include/linux/interconnect.h
index 593215371fd6..ae6744da9bc2 100644
--- a/include/linux/interconnect.h
+++ b/include/linux/interconnect.h
@@ -17,6 +17,7 @@ struct device;
 
 struct icc_path *icc_get(struct device *dev, const int src_id,
 			 const int dst_id);
+struct icc_path *of_icc_get(struct device *dev, const char *name);
 void icc_put(struct icc_path *path);
 int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw);
 
@@ -28,6 +29,12 @@ static inline struct icc_path *icc_get(struct device *dev, const int src_id,
 	return NULL;
 }
 
+static inline struct icc_path *of_icc_get(struct device *dev,
+					  const char *name)
+{
+	return NULL;
+}
+
 static inline void icc_put(struct icc_path *path)
 {
 }

[PATCH v5 8/8] interconnect: Allow endpoints translation via DT

[Georgi Djakov]

Currently we support only platform data for specifying the interconnect
endpoints. As now the endpoints are hard-coded into the consumer driver
this may lead to complications when a single driver is used by multiple
SoCs, which may have different interconnect topology.
To avoid cluttering the consumer drivers, introduce a translation function
to help us get the board specific interconnect data from device-tree.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
---
 drivers/interconnect/core.c  | 62 ++++++++++++++++++++++++++++++++++++
 include/linux/interconnect.h |  7 ++++
 2 files changed, 69 insertions(+)

diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
index dafc001ce758..2f07807ca4b4 100644
--- a/drivers/interconnect/core.c
+++ b/drivers/interconnect/core.c
@@ -16,6 +16,7 @@
 #include <linux/module.h>
 #include <linux/mutex.h>
 #include <linux/slab.h>
+#include <linux/of.h>
 #include <linux/uaccess.h>
 
 static DEFINE_IDR(icc_idr);
@@ -306,6 +307,67 @@ static int apply_constraints(struct icc_path *path)
 	return ret;
 }
 
+struct icc_path *of_icc_get(struct device *dev, const char *name)
+{
+	struct device_node *np = NULL;
+	struct of_phandle_args src_args, dst_args;
+	u32 src_id, dst_id;
+	int idx = 0;
+	int ret;
+
+	if (!dev || !dev->of_node)
+		return ERR_PTR(-ENODEV);
+
+	np = dev->of_node;
+
+	/*
+	 * When the consumer DT node do not have "interconnects" property
+	 * return a NULL path to skip setting constraints.
+	 */
+	if (!of_find_property(np, "interconnects", NULL))
+		return NULL;
+
+	/*
+	 * We use a combination of phandle and specifier for endpoint. For now
+	 * lets support only global ids and extend this is the future if needed
+	 * without breaking DT compatibility.
+	 */
+	if (name) {
+		idx = of_property_match_string(np, "interconnect-names", name);
+		if (idx < 0)
+			return ERR_PTR(idx);
+	}
+
+	ret = of_parse_phandle_with_args(np, "interconnects",
+					 "#interconnect-cells", idx * 2,
+					 &src_args);
+	if (ret)
+		return ERR_PTR(ret);
+
+	of_node_put(src_args.np);
+
+	if (!src_args.args_count || src_args.args_count > 1)
+		return ERR_PTR(-EINVAL);
+
+	src_id = src_args.args[0];
+
+	ret = of_parse_phandle_with_args(np, "interconnects",
+					 "#interconnect-cells", idx * 2 + 1,
+					 &dst_args);
+	if (ret)
+		return ERR_PTR(ret);
+
+	of_node_put(dst_args.np);
+
+	if (!dst_args.args_count || dst_args.args_count > 1)
+		return ERR_PTR(-EINVAL);
+
+	dst_id = dst_args.args[0];
+
+	return icc_get(dev, src_id, dst_id);
+}
+EXPORT_SYMBOL_GPL(of_icc_get);
+
 /**
  * icc_set() - set constraints on an interconnect path between two endpoints
  * @path: reference to the path returned by icc_get()
diff --git a/include/linux/interconnect.h b/include/linux/interconnect.h
index 593215371fd6..ae6744da9bc2 100644
--- a/include/linux/interconnect.h
+++ b/include/linux/interconnect.h
@@ -17,6 +17,7 @@ struct device;
 
 struct icc_path *icc_get(struct device *dev, const int src_id,
 			 const int dst_id);
+struct icc_path *of_icc_get(struct device *dev, const char *name);
 void icc_put(struct icc_path *path);
 int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw);
 
@@ -28,6 +29,12 @@ static inline struct icc_path *icc_get(struct device *dev, const int src_id,
 	return NULL;
 }
 
+static inline struct icc_path *of_icc_get(struct device *dev,
+					  const char *name)
+{
+	return NULL;
+}
+
 static inline void icc_put(struct icc_path *path)
 {
 }

Re: [PATCH v5 4/8] interconnect: qcom: Add RPM communication

[Evan Green]

Hi Georgi,

On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> On some Qualcomm SoCs, there is a remote processor, which controls some of
> the Network-On-Chip interconnect resources. Other CPUs express their needs
> by communicating with this processor. Add a driver to handle comminication
> with this remote processor.
>
> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
> ---
>  .../bindings/interconnect/qcom-smd.txt        | 32 +++++++
>  drivers/interconnect/qcom/Makefile            |  2 +
>  drivers/interconnect/qcom/smd-rpm.c           | 90 +++++++++++++++++++
>  drivers/interconnect/qcom/smd-rpm.h           | 15 ++++
>  4 files changed, 139 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/interconnect/qcom-smd.txt
>  create mode 100644 drivers/interconnect/qcom/Makefile
>  create mode 100644 drivers/interconnect/qcom/smd-rpm.c
>  create mode 100644 drivers/interconnect/qcom/smd-rpm.h
>
> diff --git a/Documentation/devicetree/bindings/interconnect/qcom-smd.txt b/Documentation/devicetree/bindings/interconnect/qcom-smd.txt
> new file mode 100644
> index 000000000000..88a5aeb50935
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/interconnect/qcom-smd.txt
> @@ -0,0 +1,32 @@
> +Qualcomm SMD-RPM interconnect driver binding
> +------------------------------------------------
> +The RPM (Resource Power Manager) is a dedicated hardware engine
> +for managing the shared SoC resources in order to keep the lowest
> +power profile. It communicates with other hardware subsystems via
> +the shared memory driver (SMD) back-end and accepts requests for
> +various resources.
> +
> +Required properties :
> +- compatible : shall contain only one of the following:
> +                       "qcom,interconnect-smd-rpm"
> +
> +Example:
> +       smd {
> +               compatible = "qcom,smd";
> +
> +               rpm {
> +                       interrupts = <0 168 1>;
> +                       qcom,ipc = <&apcs 8 0>;
> +                       qcom,smd-edge = <15>;
> +
> +                       rpm_requests {
> +                               compatible = "qcom,rpm-msm8916";
> +                               qcom,smd-channels = "rpm_requests";
> +
> +                               interconnect-smd-rpm {
> +                                       compatible = "qcom,interconnect-smd-rpm";
> +                               };
> +
> +                       };
> +               };
> +       };
> diff --git a/drivers/interconnect/qcom/Makefile b/drivers/interconnect/qcom/Makefile
> new file mode 100644
> index 000000000000..2a0c41db91c4
> --- /dev/null
> +++ b/drivers/interconnect/qcom/Makefile
> @@ -0,0 +1,2 @@
> +# SPDX-License-Identifier: GPL-2.0
> +obj-y += smd-rpm.o

This new file has an unexpressed dependency on CONFIG_QCOM_SMD_RPM due
to qcom_rpm_smd_write. RPMh-based systems won't need this file, right?
Maybe this file should live behind some config?

> diff --git a/drivers/interconnect/qcom/smd-rpm.c b/drivers/interconnect/qcom/smd-rpm.c
> new file mode 100644
> index 000000000000..0cf772f51642
> --- /dev/null
> +++ b/drivers/interconnect/qcom/smd-rpm.c
> @@ -0,0 +1,90 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * RPM over SMD communication wrapper for interconects
> + *
> + * Copyright (C) 2018 Linaro Ltd
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#include <linux/interconnect-provider.h>
> +#include <linux/module.h>
> +#include <linux/of.h>
> +#include <linux/of_platform.h>
> +#include <linux/platform_device.h>
> +#include <linux/soc/qcom/smd-rpm.h>
> +#include "smd-rpm.h"
> +
> +#define        RPM_KEY_BW      0x00007762
> +
> +static struct qcom_icc_rpm {
> +       struct qcom_smd_rpm *rpm;
> +} icc_rpm_smd;
> +
> +struct icc_rpm_smd_req {
> +       __le32 key;
> +       __le32 nbytes;
> +       __le32 value;
> +};
> +
> +bool qcom_icc_rpm_smd_available(void)
> +{
> +       if (!icc_rpm_smd.rpm)
> +               return false;
> +
> +       return true;
> +}
> +
> +int qcom_icc_rpm_smd_send(int ctx, int rsc_type, int id, u32 val)
> +{
> +       struct icc_rpm_smd_req req = {
> +               .key = cpu_to_le32(RPM_KEY_BW),
> +               .nbytes = cpu_to_le32(sizeof(u32)),
> +               .value = cpu_to_le32(val),
> +       };
> +
> +       return qcom_rpm_smd_write(icc_rpm_smd.rpm, ctx, rsc_type, id, &req,
> +                                 sizeof(req));
> +}
> +EXPORT_SYMBOL(qcom_icc_rpm_smd_send);
> +
> +static int qcom_icc_rpm_smd_probe(struct platform_device *pdev)
> +{
> +       icc_rpm_smd.rpm = dev_get_drvdata(pdev->dev.parent);
> +       if (!icc_rpm_smd.rpm) {
> +               dev_err(&pdev->dev, "unable to retrieve handle to RPM\n");
> +               return -ENODEV;
> +       }
> +
> +       return 0;
> +}
> +
> +static const struct of_device_id qcom_icc_rpm_smd_dt_match[] = {
> +       { .compatible = "qcom,interconnect-smd-rpm", },
> +       { },
> +};
> +
> +MODULE_DEVICE_TABLE(of, qcom_interconnect_rpm_smd_dt_match);
> +
> +static struct platform_driver qcom_interconnect_rpm_smd_driver = {
> +       .driver = {
> +               .name           = "qcom-interconnect-smd-rpm",
> +               .of_match_table = qcom_icc_rpm_smd_dt_match,
> +       },
> +       .probe = qcom_icc_rpm_smd_probe,
> +};
> +
> +static int __init rpm_smd_interconnect_init(void)
> +{
> +       return platform_driver_register(&qcom_interconnect_rpm_smd_driver);
> +}
> +subsys_initcall(rpm_smd_interconnect_init);
> +
> +static void __exit rpm_smd_interconnect_exit(void)
> +{
> +       platform_driver_unregister(&qcom_interconnect_rpm_smd_driver);
> +}
> +module_exit(rpm_smd_interconnect_exit)
> +
> +MODULE_AUTHOR("Georgi Djakov <georgi.djakov@linaro.org>");
> +MODULE_DESCRIPTION("Qualcomm SMD RPM interconnect driver");
> +MODULE_LICENSE("GPL v2");
> diff --git a/drivers/interconnect/qcom/smd-rpm.h b/drivers/interconnect/qcom/smd-rpm.h
> new file mode 100644
> index 000000000000..0f4a3da31cf6
> --- /dev/null
> +++ b/drivers/interconnect/qcom/smd-rpm.h
> @@ -0,0 +1,15 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Copyright (c) 2018, Linaro Ltd.
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#ifndef __LINUX_INTERCONNECT_QCOM_RPM_H
> +#define __LINUX_INTERCONNECT_QCOM_RPM_H

This doesn't really reflect the path, should this be something different?

> +
> +#include <linux/soc/qcom/smd-rpm.h>
> +
> +bool qcom_icc_rpm_smd_available(void);
> +int qcom_icc_rpm_smd_send(int ctx, int rsc_type, int id, u32 val);
> +
> +#endif

[PATCH v5 4/8] interconnect: qcom: Add RPM communication

[Evan Green]

Hi Georgi,

On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> On some Qualcomm SoCs, there is a remote processor, which controls some of
> the Network-On-Chip interconnect resources. Other CPUs express their needs
> by communicating with this processor. Add a driver to handle comminication
> with this remote processor.
>
> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
> ---
>  .../bindings/interconnect/qcom-smd.txt        | 32 +++++++
>  drivers/interconnect/qcom/Makefile            |  2 +
>  drivers/interconnect/qcom/smd-rpm.c           | 90 +++++++++++++++++++
>  drivers/interconnect/qcom/smd-rpm.h           | 15 ++++
>  4 files changed, 139 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/interconnect/qcom-smd.txt
>  create mode 100644 drivers/interconnect/qcom/Makefile
>  create mode 100644 drivers/interconnect/qcom/smd-rpm.c
>  create mode 100644 drivers/interconnect/qcom/smd-rpm.h
>
> diff --git a/Documentation/devicetree/bindings/interconnect/qcom-smd.txt b/Documentation/devicetree/bindings/interconnect/qcom-smd.txt
> new file mode 100644
> index 000000000000..88a5aeb50935
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/interconnect/qcom-smd.txt
> @@ -0,0 +1,32 @@
> +Qualcomm SMD-RPM interconnect driver binding
> +------------------------------------------------
> +The RPM (Resource Power Manager) is a dedicated hardware engine
> +for managing the shared SoC resources in order to keep the lowest
> +power profile. It communicates with other hardware subsystems via
> +the shared memory driver (SMD) back-end and accepts requests for
> +various resources.
> +
> +Required properties :
> +- compatible : shall contain only one of the following:
> +                       "qcom,interconnect-smd-rpm"
> +
> +Example:
> +       smd {
> +               compatible = "qcom,smd";
> +
> +               rpm {
> +                       interrupts = <0 168 1>;
> +                       qcom,ipc = <&apcs 8 0>;
> +                       qcom,smd-edge = <15>;
> +
> +                       rpm_requests {
> +                               compatible = "qcom,rpm-msm8916";
> +                               qcom,smd-channels = "rpm_requests";
> +
> +                               interconnect-smd-rpm {
> +                                       compatible = "qcom,interconnect-smd-rpm";
> +                               };
> +
> +                       };
> +               };
> +       };
> diff --git a/drivers/interconnect/qcom/Makefile b/drivers/interconnect/qcom/Makefile
> new file mode 100644
> index 000000000000..2a0c41db91c4
> --- /dev/null
> +++ b/drivers/interconnect/qcom/Makefile
> @@ -0,0 +1,2 @@
> +# SPDX-License-Identifier: GPL-2.0
> +obj-y += smd-rpm.o

This new file has an unexpressed dependency on CONFIG_QCOM_SMD_RPM due
to qcom_rpm_smd_write. RPMh-based systems won't need this file, right?
Maybe this file should live behind some config?

> diff --git a/drivers/interconnect/qcom/smd-rpm.c b/drivers/interconnect/qcom/smd-rpm.c
> new file mode 100644
> index 000000000000..0cf772f51642
> --- /dev/null
> +++ b/drivers/interconnect/qcom/smd-rpm.c
> @@ -0,0 +1,90 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * RPM over SMD communication wrapper for interconects
> + *
> + * Copyright (C) 2018 Linaro Ltd
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#include <linux/interconnect-provider.h>
> +#include <linux/module.h>
> +#include <linux/of.h>
> +#include <linux/of_platform.h>
> +#include <linux/platform_device.h>
> +#include <linux/soc/qcom/smd-rpm.h>
> +#include "smd-rpm.h"
> +
> +#define        RPM_KEY_BW      0x00007762
> +
> +static struct qcom_icc_rpm {
> +       struct qcom_smd_rpm *rpm;
> +} icc_rpm_smd;
> +
> +struct icc_rpm_smd_req {
> +       __le32 key;
> +       __le32 nbytes;
> +       __le32 value;
> +};
> +
> +bool qcom_icc_rpm_smd_available(void)
> +{
> +       if (!icc_rpm_smd.rpm)
> +               return false;
> +
> +       return true;
> +}
> +
> +int qcom_icc_rpm_smd_send(int ctx, int rsc_type, int id, u32 val)
> +{
> +       struct icc_rpm_smd_req req = {
> +               .key = cpu_to_le32(RPM_KEY_BW),
> +               .nbytes = cpu_to_le32(sizeof(u32)),
> +               .value = cpu_to_le32(val),
> +       };
> +
> +       return qcom_rpm_smd_write(icc_rpm_smd.rpm, ctx, rsc_type, id, &req,
> +                                 sizeof(req));
> +}
> +EXPORT_SYMBOL(qcom_icc_rpm_smd_send);
> +
> +static int qcom_icc_rpm_smd_probe(struct platform_device *pdev)
> +{
> +       icc_rpm_smd.rpm = dev_get_drvdata(pdev->dev.parent);
> +       if (!icc_rpm_smd.rpm) {
> +               dev_err(&pdev->dev, "unable to retrieve handle to RPM\n");
> +               return -ENODEV;
> +       }
> +
> +       return 0;
> +}
> +
> +static const struct of_device_id qcom_icc_rpm_smd_dt_match[] = {
> +       { .compatible = "qcom,interconnect-smd-rpm", },
> +       { },
> +};
> +
> +MODULE_DEVICE_TABLE(of, qcom_interconnect_rpm_smd_dt_match);
> +
> +static struct platform_driver qcom_interconnect_rpm_smd_driver = {
> +       .driver = {
> +               .name           = "qcom-interconnect-smd-rpm",
> +               .of_match_table = qcom_icc_rpm_smd_dt_match,
> +       },
> +       .probe = qcom_icc_rpm_smd_probe,
> +};
> +
> +static int __init rpm_smd_interconnect_init(void)
> +{
> +       return platform_driver_register(&qcom_interconnect_rpm_smd_driver);
> +}
> +subsys_initcall(rpm_smd_interconnect_init);
> +
> +static void __exit rpm_smd_interconnect_exit(void)
> +{
> +       platform_driver_unregister(&qcom_interconnect_rpm_smd_driver);
> +}
> +module_exit(rpm_smd_interconnect_exit)
> +
> +MODULE_AUTHOR("Georgi Djakov <georgi.djakov@linaro.org>");
> +MODULE_DESCRIPTION("Qualcomm SMD RPM interconnect driver");
> +MODULE_LICENSE("GPL v2");
> diff --git a/drivers/interconnect/qcom/smd-rpm.h b/drivers/interconnect/qcom/smd-rpm.h
> new file mode 100644
> index 000000000000..0f4a3da31cf6
> --- /dev/null
> +++ b/drivers/interconnect/qcom/smd-rpm.h
> @@ -0,0 +1,15 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Copyright (c) 2018, Linaro Ltd.
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#ifndef __LINUX_INTERCONNECT_QCOM_RPM_H
> +#define __LINUX_INTERCONNECT_QCOM_RPM_H

This doesn't really reflect the path, should this be something different?

> +
> +#include <linux/soc/qcom/smd-rpm.h>
> +
> +bool qcom_icc_rpm_smd_available(void);
> +int qcom_icc_rpm_smd_send(int ctx, int rsc_type, int id, u32 val);
> +
> +#endif

Re: [PATCH v5 6/8] interconnect: qcom: Add msm8916 interconnect provider driver

[Evan Green]

On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> Add driver for the Qualcomm interconnect buses found in msm8916 based
> platforms.
>
> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
> ---
>  drivers/interconnect/Kconfig        |   5 +
>  drivers/interconnect/Makefile       |   1 +
>  drivers/interconnect/qcom/Kconfig   |  11 +
>  drivers/interconnect/qcom/Makefile  |   2 +
>  drivers/interconnect/qcom/msm8916.c | 495 ++++++++++++++++++++++++++++
>  5 files changed, 514 insertions(+)
>  create mode 100644 drivers/interconnect/qcom/Kconfig
>  create mode 100644 drivers/interconnect/qcom/msm8916.c
>
> diff --git a/drivers/interconnect/Kconfig b/drivers/interconnect/Kconfig
> index a261c7d41deb..07a8276fa35a 100644
> --- a/drivers/interconnect/Kconfig
> +++ b/drivers/interconnect/Kconfig
> @@ -8,3 +8,8 @@ menuconfig INTERCONNECT
>
>           If unsure, say no.
>
> +if INTERCONNECT
> +
> +source "drivers/interconnect/qcom/Kconfig"
> +
> +endif
> diff --git a/drivers/interconnect/Makefile b/drivers/interconnect/Makefile
> index 97fca2e09d24..7944cbca0527 100644
> --- a/drivers/interconnect/Makefile
> +++ b/drivers/interconnect/Makefile
> @@ -1,2 +1,3 @@
>  # SPDX-License-Identifier: GPL-2.0
>  obj-$(CONFIG_INTERCONNECT)             += core.o
> +obj-$(CONFIG_INTERCONNECT_QCOM)                += qcom/
> diff --git a/drivers/interconnect/qcom/Kconfig b/drivers/interconnect/qcom/Kconfig
> new file mode 100644
> index 000000000000..86465dc37bd4
> --- /dev/null
> +++ b/drivers/interconnect/qcom/Kconfig
> @@ -0,0 +1,11 @@
> +config INTERCONNECT_QCOM
> +       bool "Qualcomm Network-on-Chip interconnect drivers"
> +       depends on INTERCONNECT
> +       depends on ARCH_QCOM || COMPILE_TEST
> +       default y
> +
> +config INTERCONNECT_QCOM_MSM8916
> +       tristate "Qualcomm MSM8916 interconnect driver"
> +       depends on INTERCONNECT_QCOM
> +       help
> +         This is a driver for the Qualcomm Network-on-Chip on msm8916-based platforms.
> diff --git a/drivers/interconnect/qcom/Makefile b/drivers/interconnect/qcom/Makefile
> index 2a0c41db91c4..e8b24c334c5d 100644
> --- a/drivers/interconnect/qcom/Makefile
> +++ b/drivers/interconnect/qcom/Makefile
> @@ -1,2 +1,4 @@
>  # SPDX-License-Identifier: GPL-2.0
>  obj-y += smd-rpm.o
> +
> +obj-$(CONFIG_INTERCONNECT_QCOM_MSM8916) += msm8916.o
> diff --git a/drivers/interconnect/qcom/msm8916.c b/drivers/interconnect/qcom/msm8916.c
> new file mode 100644
> index 000000000000..c34f18cdea33
> --- /dev/null
> +++ b/drivers/interconnect/qcom/msm8916.c
> @@ -0,0 +1,495 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Copyright (C) 2018 Linaro Ltd
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#include <dt-bindings/interconnect/qcom.h>
> +#include <linux/clk.h>
> +#include <linux/device.h>
> +#include <linux/io.h>
> +#include <linux/interconnect-provider.h>

Nit: the previous two should be swapped for alphabetization.

> +#include <linux/module.h>
> +#include <linux/of_device.h>
> +#include <linux/of_platform.h>
> +#include <linux/platform_device.h>
> +#include <linux/slab.h>
> +
> +#include "smd-rpm.h"
> +
> +#define RPM_MASTER_FIELD_BW    0x00007762

This is unused, since it moved into that other file.

> +#define RPM_BUS_MASTER_REQ      0x73616d62
> +#define RPM_BUS_SLAVE_REQ       0x766c7362
> +
> +#define to_qcom_provider(_provider) \
> +       container_of(_provider, struct qcom_icc_provider, provider)
> +
> +enum qcom_qos_mode {
> +       QCOM_QOS_MODE_BYPASS = 0,
> +       QCOM_QOS_MODE_FIXED,
> +       QCOM_QOS_MODE_MAX,
> +};
> +
> +struct qcom_icc_provider {
> +       struct icc_provider     provider;
> +       void __iomem            *base;
> +       struct clk              *bus_clk;
> +       struct clk              *bus_a_clk;
> +};
> +
> +#define MSM8916_MAX_LINKS      8
> +
> +/**
> + * struct qcom_icc_node - Qualcomm specific interconnect nodes
> + * @name: the node name used in debugfs
> + * @links: an array of nodes where we can go next while traversing
> + * @id: a unique node identifier
> + * @num_links: the total number of @links
> + * @port: the offset index into the masters QoS register space
> + * @buswidth: width of the interconnect between a node and the bus (bytes)
> + * @ap_owned: the AP CPU does the writing to QoS registers
> + * @rpm: reference to the RPM SMD driver
> + * @qos_mode: QoS mode for ap_owned resources
> + * @mas_rpm_id:        RPM id for devices that are bus masters
> + * @slv_rpm_id:        RPM id for devices that are bus slaves
> + * @rate: current bus clock rate in Hz
> + */
> +struct qcom_icc_node {
> +       unsigned char *name;
> +       u16 links[MSM8916_MAX_LINKS];
> +       u16 id;
> +       u16 num_links;
> +       u16 port;
> +       u16 buswidth;
> +       bool ap_owned;
> +       struct qcom_smd_rpm *rpm;

This is unused.

> +       enum qcom_qos_mode qos_mode;
> +       int mas_rpm_id;
> +       int slv_rpm_id;
> +       u64 rate;
> +};
> +
> +struct qcom_icc_desc {
> +       struct qcom_icc_node **nodes;
> +       size_t num_nodes;
> +};
> +
> +#define DEFINE_QNODE(_name, _id, _port, _buswidth, _ap_owned,          \
> +                       _mas_rpm_id, _slv_rpm_id, _qos_mode,            \
> +                       _numlinks, ...)                                 \
> +               static struct qcom_icc_node _name = {                   \
> +               .id = _id,                                              \
> +               .name = #_name,                                         \
> +               .port = _port,                                          \
> +               .buswidth = _buswidth,                                  \
> +               .qos_mode = _qos_mode,                                  \
> +               .ap_owned = _ap_owned,                                  \
> +               .mas_rpm_id = _mas_rpm_id,                              \
> +               .slv_rpm_id = _slv_rpm_id,                              \
> +               .num_links = _numlinks,                                 \
> +               .links = { __VA_ARGS__ },                               \
> +       }
> +
> +DEFINE_QNODE(mas_video, MASTER_VIDEO_P0, 8, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_0, SNOC_MM_INT_2);
> +DEFINE_QNODE(mas_jpeg, MASTER_JPEG, 6, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_0, SNOC_MM_INT_2);
> +DEFINE_QNODE(mas_vfe, MASTER_VFE, 9, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_1, SNOC_MM_INT_2);
> +DEFINE_QNODE(mas_mdp, MASTER_MDP_PORT0, 7, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_0, SNOC_MM_INT_2);
> +DEFINE_QNODE(mas_qdss_bam, MASTER_QDSS_BAM, 11, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_QDSS_INT);
> +DEFINE_QNODE(mas_snoc_cfg, MASTER_SNOC_CFG, 0, 16, 0, 20, -1, QCOM_QOS_MODE_BYPASS, 1, SNOC_QDSS_INT);
> +DEFINE_QNODE(mas_qdss_etr, MASTER_QDSS_ETR, 10, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_QDSS_INT);
> +DEFINE_QNODE(mm_int_0, SNOC_MM_INT_0, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_MM_INT_BIMC);
> +DEFINE_QNODE(mm_int_1, SNOC_MM_INT_1, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_MM_INT_BIMC);
> +DEFINE_QNODE(mm_int_2, SNOC_MM_INT_2, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_INT_0);
> +DEFINE_QNODE(mm_int_bimc, SNOC_MM_INT_BIMC, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_1_MAS);
> +DEFINE_QNODE(snoc_int_0, SNOC_INT_0, 0, 8, 0, 99, 130, QCOM_QOS_MODE_FIXED, 3, SLAVE_QDSS_STM, SLAVE_SYSTEM_IMEM, MNOC_BIMC_MAS);
> +DEFINE_QNODE(snoc_int_1, SNOC_INT_1, 0, 8, 0, 100, 131, QCOM_QOS_MODE_FIXED, 3, SYSTEM_SLAVE_FAB_APPS, SLAVE_CATS_128, SLAVE_OCMEM_64);
> +DEFINE_QNODE(snoc_int_bimc, SNOC_INT_BIMC, 0, 8, 0, 101, 132, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_0_MAS);
> +DEFINE_QNODE(snoc_bimc_0_mas, SNOC_BIMC_0_MAS, 0, 8, 0, 3, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_0_SLV);
> +DEFINE_QNODE(snoc_bimc_1_mas, SNOC_BIMC_1_MAS, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_1_SLV);
> +DEFINE_QNODE(qdss_int, SNOC_QDSS_INT, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 2, SNOC_INT_0, SNOC_INT_BIMC);
> +DEFINE_QNODE(bimc_snoc_slv, BIMC_SNOC_SLV, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 2, SNOC_INT_0, SNOC_INT_1);
> +DEFINE_QNODE(snoc_pnoc_mas, MNOC_BIMC_MAS, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, MNOC_BIMC_SLV);
> +DEFINE_QNODE(pnoc_snoc_slv, PNOC_SNOC_SLV, 0, 8, 0, -1, 45, QCOM_QOS_MODE_FIXED, 3, SNOC_INT_0, SNOC_INT_BIMC, SNOC_INT_1);
> +DEFINE_QNODE(slv_srvc_snoc, SLAVE_SERVICE_SNOC, 0, 8, 0, -1, 29, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_qdss_stm, SLAVE_QDSS_STM, 0, 4, 0, -1, 30, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_imem, SLAVE_SYSTEM_IMEM, 0, 8, 0, -1, 26, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_apss, SYSTEM_SLAVE_FAB_APPS, 0, 4, 0, -1, 20, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_cats_0, SLAVE_CATS_128, 0, 16, 0, -1, 106, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_cats_1, SLAVE_OCMEM_64, 0, 8, 0, -1, 107, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(mas_apss, MASTER_AMPSS_M0, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
> +DEFINE_QNODE(mas_tcu0, MASTER_TCU_0, 5, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
> +DEFINE_QNODE(mas_tcu1, MASTER_TCU_1, 6, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
> +DEFINE_QNODE(mas_gfx, MASTER_GRAPHICS_3D, 2, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
> +DEFINE_QNODE(bimc_snoc_mas, BIMC_SNOC_MAS, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, BIMC_SNOC_SLV);
> +DEFINE_QNODE(snoc_bimc_0_slv, SNOC_BIMC_0_SLV, 0, 8, 0, -1, 24, QCOM_QOS_MODE_FIXED, 1, SLAVE_EBI_CH0);
> +DEFINE_QNODE(snoc_bimc_1_slv, SNOC_BIMC_1_SLV, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SLAVE_EBI_CH0);
> +DEFINE_QNODE(slv_ebi_ch0, SLAVE_EBI_CH0, 0, 8, 0, -1, 0, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_apps_l2, SLAVE_AMPSS_L2, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(snoc_pnoc_slv, MNOC_BIMC_SLV, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_0);
> +DEFINE_QNODE(pnoc_int_0, PNOC_INT_0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 8, PNOC_SNOC_MAS, PNOC_SLV_0, PNOC_SLV_1, PNOC_SLV_2, PNOC_SLV_3, PNOC_SLV_4, PNOC_SLV_8, PNOC_SLV_9);
> +DEFINE_QNODE(pnoc_int_1, PNOC_INT_1, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_SNOC_MAS);
> +DEFINE_QNODE(pnoc_m_0, PNOC_M_0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_0);
> +DEFINE_QNODE(pnoc_m_1, PNOC_M_1, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_SNOC_MAS);
> +DEFINE_QNODE(pnoc_s_0, PNOC_SLV_0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_CLK_CTL, SLAVE_TLMM, SLAVE_MSM_TCSR, SLAVE_SECURITY, SLAVE_MSS);
> +DEFINE_QNODE(pnoc_s_1, PNOC_SLV_1, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_IMEM_CFG, SLAVE_CRYPTO_0_CFG, SLAVE_RPM_MSG_RAM, SLAVE_MSM_PDM, SLAVE_PRNG);
> +DEFINE_QNODE(pnoc_s_2, PNOC_SLV_2, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_SPDM, SLAVE_BOOT_ROM, SLAVE_BIMC_CFG, SLAVE_PNOC_CFG, SLAVE_PMIC_ARB);
> +DEFINE_QNODE(pnoc_s_3, PNOC_SLV_3, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_MPM, SLAVE_IPS_CFG, SLAVE_RBCPR_CFG, SLAVE_QDSS_CFG, SLAVE_DEHR_CFG);
> +DEFINE_QNODE(pnoc_s_4, PNOC_SLV_4, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_VENUS_CFG, SLAVE_CAMERA_CFG, SLAVE_DISPLAY_CFG);
> +DEFINE_QNODE(pnoc_s_8, PNOC_SLV_8, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_USB_HS, SLAVE_SDCC_1, SLAVE_BLSP_1);
> +DEFINE_QNODE(pnoc_s_9, PNOC_SLV_9, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_SDCC_4, SLAVE_LPASS, SLAVE_GRAPHICS_3D_CFG);
> +DEFINE_QNODE(slv_imem_cfg, SLAVE_IMEM_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_crypto_0_cfg, SLAVE_CRYPTO_0_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_msg_ram, SLAVE_RPM_MSG_RAM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_pdm, SLAVE_MSM_PDM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_prng, SLAVE_PRNG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_clk_ctl, SLAVE_CLK_CTL, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_mss, SLAVE_MSS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_tlmm, SLAVE_TLMM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_tcsr, SLAVE_MSM_TCSR, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_security, SLAVE_SECURITY, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_spdm, SLAVE_SPDM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_pnoc_cfg, SLAVE_PNOC_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_pmic_arb, SLAVE_PMIC_ARB, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_bimc_cfg, SLAVE_BIMC_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_boot_rom, SLAVE_BOOT_ROM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_mpm, SLAVE_MPM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_qdss_cfg, SLAVE_QDSS_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_rbcpr_cfg, SLAVE_RBCPR_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_snoc_cfg, SLAVE_IPS_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_dehr_cfg, SLAVE_DEHR_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_venus_cfg, SLAVE_VENUS_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_display_cfg, SLAVE_DISPLAY_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_camera_cfg, SLAVE_CAMERA_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_usb_hs, SLAVE_USB_HS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_sdcc_1, SLAVE_SDCC_1, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_blsp_1, SLAVE_BLSP_1, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_sdcc_2, SLAVE_SDCC_4, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_gfx_cfg, SLAVE_GRAPHICS_3D_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_audio, SLAVE_LPASS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(mas_blsp_1, MASTER_BLSP_1, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_1);
> +DEFINE_QNODE(mas_spdm, MASTER_SPDM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_0);
> +DEFINE_QNODE(mas_dehr, MASTER_DEHR, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_0);
> +DEFINE_QNODE(mas_audio, MASTER_LPASS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_0);
> +DEFINE_QNODE(mas_usb_hs, MASTER_USB_HS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_1);
> +DEFINE_QNODE(mas_pnoc_crypto_0, MASTER_CRYPTO_CORE0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_1);
> +DEFINE_QNODE(mas_pnoc_sdcc_1, MASTER_SDCC_1, 7, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_1);
> +DEFINE_QNODE(mas_pnoc_sdcc_2, MASTER_SDCC_2, 8, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_1);
> +DEFINE_QNODE(pnoc_snoc_mas, PNOC_SNOC_MAS, 0, 8, 0, 29, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_SNOC_SLV);
> +
> +static struct qcom_icc_node *msm8916_snoc_nodes[] = {
> +       &mas_video,
> +       &mas_jpeg,
> +       &mas_vfe,
> +       &mas_mdp,
> +       &mas_qdss_bam,
> +       &mas_snoc_cfg,
> +       &mas_qdss_etr,
> +       &mm_int_0,
> +       &mm_int_1,
> +       &mm_int_2,
> +       &mm_int_bimc,
> +       &snoc_int_0,
> +       &snoc_int_1,
> +       &snoc_int_bimc,
> +       &snoc_bimc_0_mas,
> +       &snoc_bimc_1_mas,
> +       &qdss_int,
> +       &bimc_snoc_slv,
> +       &snoc_pnoc_mas,
> +       &pnoc_snoc_slv,
> +       &slv_srvc_snoc,
> +       &slv_qdss_stm,
> +       &slv_imem,
> +       &slv_apss,
> +       &slv_cats_0,
> +       &slv_cats_1,
> +};
> +
> +static struct qcom_icc_desc msm8916_snoc = {
> +       .nodes = msm8916_snoc_nodes,
> +       .num_nodes = ARRAY_SIZE(msm8916_snoc_nodes),
> +};
> +
> +static struct qcom_icc_node *msm8916_bimc_nodes[] = {
> +       &mas_apss,
> +       &mas_tcu0,
> +       &mas_tcu1,
> +       &mas_gfx,
> +       &bimc_snoc_mas,
> +       &snoc_bimc_0_slv,
> +       &snoc_bimc_1_slv,
> +       &slv_ebi_ch0,
> +       &slv_apps_l2,
> +};
> +
> +static struct qcom_icc_desc msm8916_bimc = {
> +       .nodes = msm8916_bimc_nodes,
> +       .num_nodes = ARRAY_SIZE(msm8916_bimc_nodes),
> +};
> +
> +static struct qcom_icc_node *msm8916_pnoc_nodes[] = {
> +       &snoc_pnoc_slv,
> +       &pnoc_int_0,
> +       &pnoc_int_1,
> +       &pnoc_m_0,
> +       &pnoc_m_1,
> +       &pnoc_s_0,
> +       &pnoc_s_1,
> +       &pnoc_s_2,
> +       &pnoc_s_3,
> +       &pnoc_s_4,
> +       &pnoc_s_8,
> +       &pnoc_s_9,
> +       &slv_imem_cfg,
> +       &slv_crypto_0_cfg,
> +       &slv_msg_ram,
> +       &slv_pdm,
> +       &slv_prng,
> +       &slv_clk_ctl,
> +       &slv_mss,
> +       &slv_tlmm,
> +       &slv_tcsr,
> +       &slv_security,
> +       &slv_spdm,
> +       &slv_pnoc_cfg,
> +       &slv_pmic_arb,
> +       &slv_bimc_cfg,
> +       &slv_boot_rom,
> +       &slv_mpm,
> +       &slv_qdss_cfg,
> +       &slv_rbcpr_cfg,
> +       &slv_snoc_cfg,
> +       &slv_dehr_cfg,
> +       &slv_venus_cfg,
> +       &slv_display_cfg,
> +       &slv_camera_cfg,
> +       &slv_usb_hs,
> +       &slv_sdcc_1,
> +       &slv_blsp_1,
> +       &slv_sdcc_2,
> +       &slv_gfx_cfg,
> +       &slv_audio,
> +       &mas_blsp_1,
> +       &mas_spdm,
> +       &mas_dehr,
> +       &mas_audio,
> +       &mas_usb_hs,
> +       &mas_pnoc_crypto_0,
> +       &mas_pnoc_sdcc_1,
> +       &mas_pnoc_sdcc_2,
> +       &pnoc_snoc_mas,
> +};
> +
> +static struct qcom_icc_desc msm8916_pnoc = {
> +       .nodes = msm8916_pnoc_nodes,
> +       .num_nodes = ARRAY_SIZE(msm8916_pnoc_nodes),
> +};
> +
> +static int qcom_icc_init(struct icc_node *node)
> +{
> +       struct qcom_icc_provider *qp = to_qcom_provider(node->provider);
> +       int ret;
> +
> +       /* TODO: init qos and priority */
> +
> +       clk_set_rate(qp->bus_clk, INT_MAX);

Vroom! What's the rationale here? I wonder if it might be better to
avoid touching the clocks initially, and expect that the boot loader
sets up a decent initial set of bus frequencies for consumers that
never enable bus scaling? Otherwise, I worry that this driver becomes
basically an essential driver for the platform solely because of this
line and the one below, when really it might not be.

> +       ret = clk_prepare_enable(qp->bus_clk);
> +       if (ret)
> +               pr_info("%s: error enabling bus clk (%d)\n", __func__, ret);
> +       clk_set_rate(qp->bus_a_clk, INT_MAX);
> +       ret = clk_prepare_enable(qp->bus_a_clk);
> +       if (ret)
> +               pr_info("%s: error enabling bus_a clk (%d)\n", __func__, ret);
> +
> +       return 0;
> +}
> +
> +static int qcom_icc_aggregate(struct icc_node *node, u32 avg_bw, u32 peak_bw,
> +                             u32 *agg_avg, u32 *agg_peak)
> +{
> +       *agg_avg += avg_bw;
> +       *agg_peak = max(*agg_peak, peak_bw);
> +
> +       return 0;
> +}
> +
> +static int qcom_icc_set(struct icc_node *src, struct icc_node *dst,
> +                       u32 avg, u32 peak)
> +{
> +       struct qcom_icc_provider *qp;
> +       struct qcom_icc_node *qn;
> +       struct icc_node *node;
> +       struct icc_provider *provider;
> +       u64 avg_bw = icc_units_to_bps(avg);
> +       u64 peak_bw = icc_units_to_bps(peak);
> +       u64 rate = 0;
> +       int ret = 0;
> +
> +       if (!src)
> +               node = dst;
> +       else
> +               node = src;
> +
> +       qn = node->data;
> +       provider = node->provider;
> +       qp = to_qcom_provider(node->provider);
> +
> +       /* set bandwidth */
> +       if (qn->ap_owned) {
> +               /* TODO: set QoS */
> +       } else {
> +               /* send message to the RPM processor */
> +               if (qn->mas_rpm_id != -1) {
> +                       ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
> +                                                   RPM_BUS_MASTER_REQ,
> +                                                   qn->mas_rpm_id,
> +                                                   avg_bw);

You don't check for failure here, and potentially clobber failure with
success. Unless only one of mas_rpm_id and slv_rpm_id is ever set. But
no, I see SNOC_INT_0/1 set both. Same for below.

> +               }
> +
> +               if (qn->slv_rpm_id != -1) {
> +                       ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
> +                                                   RPM_BUS_SLAVE_REQ,
> +                                                   qn->slv_rpm_id,
> +                                                   avg_bw);
> +               }
> +       }
> +
> +       rate = max(avg_bw, peak_bw);
> +
> +       do_div(rate, qn->buswidth);
> +
> +       if (qn->rate != rate) {
> +               ret = clk_set_rate(qp->bus_clk, rate);
> +               if (ret) {
> +                       pr_err("set clk rate %lld error %d\n", rate, ret);
> +                       return ret;
> +               }
> +
> +               ret = clk_set_rate(qp->bus_a_clk, rate);
> +               if (ret) {
> +                       pr_err("set clk rate %lld error %d\n", rate, ret);
> +                       return ret;
> +               }
> +
> +               qn->rate = rate;
> +       }
> +
> +       return ret;
> +}
> +
> +static int qnoc_probe(struct platform_device *pdev)
> +{
> +       const struct qcom_icc_desc *desc;
> +       struct qcom_icc_node **qnodes;
> +       struct qcom_icc_provider *qp;
> +       struct resource *res;
> +       struct icc_provider *provider;
> +       size_t num_nodes, i;
> +       int ret;
> +
> +       /* wait for RPM */
> +       if (!qcom_icc_rpm_smd_available())
> +               return -EPROBE_DEFER;
> +
> +       desc = of_device_get_match_data(&pdev->dev);
> +       if (!desc)
> +               return -EINVAL;
> +
> +       qnodes = desc->nodes;
> +       num_nodes = desc->num_nodes;
> +
> +       qp = devm_kzalloc(&pdev->dev, sizeof(*qp), GFP_KERNEL);
> +       if (!qp)
> +               return -ENOMEM;
> +
> +       res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> +       qp->base = devm_ioremap_resource(&pdev->dev, res);
> +       if (IS_ERR(qp->base))
> +               return PTR_ERR(qp->base);
> +
> +       qp->bus_clk = devm_clk_get(&pdev->dev, "bus_clk");
> +       if (IS_ERR(qp->bus_clk))
> +               return PTR_ERR(qp->bus_clk);
> +
> +       qp->bus_a_clk = devm_clk_get(&pdev->dev, "bus_a_clk");
> +       if (IS_ERR(qp->bus_a_clk))
> +               return PTR_ERR(qp->bus_a_clk);
> +
> +       provider = &qp->provider;
> +       provider->dev = &pdev->dev;
> +       provider->set = &qcom_icc_set;
> +       provider->aggregate = &qcom_icc_aggregate;
> +       INIT_LIST_HEAD(&provider->nodes);
> +       provider->data = qp;
> +
> +       ret = icc_provider_add(provider);
> +       if (ret) {
> +               dev_err(&pdev->dev, "error adding interconnect provider\n");
> +               return ret;
> +       }
> +
> +       for (i = 0; i < num_nodes; i++) {
> +               struct icc_node *node;
> +               int ret;
> +               size_t j;
> +
> +               node = icc_node_create(qnodes[i]->id);
> +               if (IS_ERR(node)) {
> +                       ret = PTR_ERR(node);
> +                       goto err;
> +               }
> +
> +               node->name = qnodes[i]->name;
> +               node->data = qnodes[i];
> +               icc_node_add(node, provider);
> +
> +               dev_dbg(&pdev->dev, "registered node %s\n", node->name);
> +
> +               ret = qcom_icc_init(node);
> +               if (ret)
> +                       dev_err(&pdev->dev, "%s init error (%d)\n", node->name,
> +                               ret);
> +
> +               /* populate links */
> +               for (j = 0; j < qnodes[i]->num_links; j++)
> +                       if (qnodes[i]->links[j])
> +                               icc_link_create(node, qnodes[i]->links[j]);
> +       }
> +
> +       platform_set_drvdata(pdev, provider);
> +
> +       return ret;
> +err:
> +       icc_provider_del(provider);
> +       return ret;
> +}
> +
> +static int qnoc_remove(struct platform_device *pdev)
> +{
> +       struct icc_provider *provider = platform_get_drvdata(pdev);
> +
> +       icc_provider_del(provider);

This function can fail, so you should probably pass the failure along.

> +
> +       return 0;
> +}
> +
> +static const struct of_device_id qnoc_of_match[] = {
> +       { .compatible = "qcom,msm8916-pnoc", .data = &msm8916_pnoc },
> +       { .compatible = "qcom,msm8916-snoc", .data = &msm8916_snoc },
> +       { .compatible = "qcom,msm8916-bimc", .data = &msm8916_bimc },
> +       { },
> +};
> +MODULE_DEVICE_TABLE(of, qnoc_of_match);
> +
> +static struct platform_driver qnoc_driver = {
> +       .probe = qnoc_probe,
> +       .remove = qnoc_remove,
> +       .driver = {
> +               .name = "qnoc-msm8916",
> +               .of_match_table = qnoc_of_match,
> +       },
> +};
> +module_platform_driver(qnoc_driver);
> +MODULE_AUTHOR("Georgi Djakov <georgi.djakov@linaro.org>");
> +MODULE_DESCRIPTION("Qualcomm msm8916 NoC driver");
> +MODULE_LICENSE("GPL v2");

[PATCH v5 6/8] interconnect: qcom: Add msm8916 interconnect provider driver

[Evan Green]

On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> Add driver for the Qualcomm interconnect buses found in msm8916 based
> platforms.
>
> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
> ---
>  drivers/interconnect/Kconfig        |   5 +
>  drivers/interconnect/Makefile       |   1 +
>  drivers/interconnect/qcom/Kconfig   |  11 +
>  drivers/interconnect/qcom/Makefile  |   2 +
>  drivers/interconnect/qcom/msm8916.c | 495 ++++++++++++++++++++++++++++
>  5 files changed, 514 insertions(+)
>  create mode 100644 drivers/interconnect/qcom/Kconfig
>  create mode 100644 drivers/interconnect/qcom/msm8916.c
>
> diff --git a/drivers/interconnect/Kconfig b/drivers/interconnect/Kconfig
> index a261c7d41deb..07a8276fa35a 100644
> --- a/drivers/interconnect/Kconfig
> +++ b/drivers/interconnect/Kconfig
> @@ -8,3 +8,8 @@ menuconfig INTERCONNECT
>
>           If unsure, say no.
>
> +if INTERCONNECT
> +
> +source "drivers/interconnect/qcom/Kconfig"
> +
> +endif
> diff --git a/drivers/interconnect/Makefile b/drivers/interconnect/Makefile
> index 97fca2e09d24..7944cbca0527 100644
> --- a/drivers/interconnect/Makefile
> +++ b/drivers/interconnect/Makefile
> @@ -1,2 +1,3 @@
>  # SPDX-License-Identifier: GPL-2.0
>  obj-$(CONFIG_INTERCONNECT)             += core.o
> +obj-$(CONFIG_INTERCONNECT_QCOM)                += qcom/
> diff --git a/drivers/interconnect/qcom/Kconfig b/drivers/interconnect/qcom/Kconfig
> new file mode 100644
> index 000000000000..86465dc37bd4
> --- /dev/null
> +++ b/drivers/interconnect/qcom/Kconfig
> @@ -0,0 +1,11 @@
> +config INTERCONNECT_QCOM
> +       bool "Qualcomm Network-on-Chip interconnect drivers"
> +       depends on INTERCONNECT
> +       depends on ARCH_QCOM || COMPILE_TEST
> +       default y
> +
> +config INTERCONNECT_QCOM_MSM8916
> +       tristate "Qualcomm MSM8916 interconnect driver"
> +       depends on INTERCONNECT_QCOM
> +       help
> +         This is a driver for the Qualcomm Network-on-Chip on msm8916-based platforms.
> diff --git a/drivers/interconnect/qcom/Makefile b/drivers/interconnect/qcom/Makefile
> index 2a0c41db91c4..e8b24c334c5d 100644
> --- a/drivers/interconnect/qcom/Makefile
> +++ b/drivers/interconnect/qcom/Makefile
> @@ -1,2 +1,4 @@
>  # SPDX-License-Identifier: GPL-2.0
>  obj-y += smd-rpm.o
> +
> +obj-$(CONFIG_INTERCONNECT_QCOM_MSM8916) += msm8916.o
> diff --git a/drivers/interconnect/qcom/msm8916.c b/drivers/interconnect/qcom/msm8916.c
> new file mode 100644
> index 000000000000..c34f18cdea33
> --- /dev/null
> +++ b/drivers/interconnect/qcom/msm8916.c
> @@ -0,0 +1,495 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Copyright (C) 2018 Linaro Ltd
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#include <dt-bindings/interconnect/qcom.h>
> +#include <linux/clk.h>
> +#include <linux/device.h>
> +#include <linux/io.h>
> +#include <linux/interconnect-provider.h>

Nit: the previous two should be swapped for alphabetization.

> +#include <linux/module.h>
> +#include <linux/of_device.h>
> +#include <linux/of_platform.h>
> +#include <linux/platform_device.h>
> +#include <linux/slab.h>
> +
> +#include "smd-rpm.h"
> +
> +#define RPM_MASTER_FIELD_BW    0x00007762

This is unused, since it moved into that other file.

> +#define RPM_BUS_MASTER_REQ      0x73616d62
> +#define RPM_BUS_SLAVE_REQ       0x766c7362
> +
> +#define to_qcom_provider(_provider) \
> +       container_of(_provider, struct qcom_icc_provider, provider)
> +
> +enum qcom_qos_mode {
> +       QCOM_QOS_MODE_BYPASS = 0,
> +       QCOM_QOS_MODE_FIXED,
> +       QCOM_QOS_MODE_MAX,
> +};
> +
> +struct qcom_icc_provider {
> +       struct icc_provider     provider;
> +       void __iomem            *base;
> +       struct clk              *bus_clk;
> +       struct clk              *bus_a_clk;
> +};
> +
> +#define MSM8916_MAX_LINKS      8
> +
> +/**
> + * struct qcom_icc_node - Qualcomm specific interconnect nodes
> + * @name: the node name used in debugfs
> + * @links: an array of nodes where we can go next while traversing
> + * @id: a unique node identifier
> + * @num_links: the total number of @links
> + * @port: the offset index into the masters QoS register space
> + * @buswidth: width of the interconnect between a node and the bus (bytes)
> + * @ap_owned: the AP CPU does the writing to QoS registers
> + * @rpm: reference to the RPM SMD driver
> + * @qos_mode: QoS mode for ap_owned resources
> + * @mas_rpm_id:        RPM id for devices that are bus masters
> + * @slv_rpm_id:        RPM id for devices that are bus slaves
> + * @rate: current bus clock rate in Hz
> + */
> +struct qcom_icc_node {
> +       unsigned char *name;
> +       u16 links[MSM8916_MAX_LINKS];
> +       u16 id;
> +       u16 num_links;
> +       u16 port;
> +       u16 buswidth;
> +       bool ap_owned;
> +       struct qcom_smd_rpm *rpm;

This is unused.

> +       enum qcom_qos_mode qos_mode;
> +       int mas_rpm_id;
> +       int slv_rpm_id;
> +       u64 rate;
> +};
> +
> +struct qcom_icc_desc {
> +       struct qcom_icc_node **nodes;
> +       size_t num_nodes;
> +};
> +
> +#define DEFINE_QNODE(_name, _id, _port, _buswidth, _ap_owned,          \
> +                       _mas_rpm_id, _slv_rpm_id, _qos_mode,            \
> +                       _numlinks, ...)                                 \
> +               static struct qcom_icc_node _name = {                   \
> +               .id = _id,                                              \
> +               .name = #_name,                                         \
> +               .port = _port,                                          \
> +               .buswidth = _buswidth,                                  \
> +               .qos_mode = _qos_mode,                                  \
> +               .ap_owned = _ap_owned,                                  \
> +               .mas_rpm_id = _mas_rpm_id,                              \
> +               .slv_rpm_id = _slv_rpm_id,                              \
> +               .num_links = _numlinks,                                 \
> +               .links = { __VA_ARGS__ },                               \
> +       }
> +
> +DEFINE_QNODE(mas_video, MASTER_VIDEO_P0, 8, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_0, SNOC_MM_INT_2);
> +DEFINE_QNODE(mas_jpeg, MASTER_JPEG, 6, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_0, SNOC_MM_INT_2);
> +DEFINE_QNODE(mas_vfe, MASTER_VFE, 9, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_1, SNOC_MM_INT_2);
> +DEFINE_QNODE(mas_mdp, MASTER_MDP_PORT0, 7, 16, 1, -1, -1, QCOM_QOS_MODE_BYPASS, 2, SNOC_MM_INT_0, SNOC_MM_INT_2);
> +DEFINE_QNODE(mas_qdss_bam, MASTER_QDSS_BAM, 11, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_QDSS_INT);
> +DEFINE_QNODE(mas_snoc_cfg, MASTER_SNOC_CFG, 0, 16, 0, 20, -1, QCOM_QOS_MODE_BYPASS, 1, SNOC_QDSS_INT);
> +DEFINE_QNODE(mas_qdss_etr, MASTER_QDSS_ETR, 10, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_QDSS_INT);
> +DEFINE_QNODE(mm_int_0, SNOC_MM_INT_0, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_MM_INT_BIMC);
> +DEFINE_QNODE(mm_int_1, SNOC_MM_INT_1, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_MM_INT_BIMC);
> +DEFINE_QNODE(mm_int_2, SNOC_MM_INT_2, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_INT_0);
> +DEFINE_QNODE(mm_int_bimc, SNOC_MM_INT_BIMC, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_1_MAS);
> +DEFINE_QNODE(snoc_int_0, SNOC_INT_0, 0, 8, 0, 99, 130, QCOM_QOS_MODE_FIXED, 3, SLAVE_QDSS_STM, SLAVE_SYSTEM_IMEM, MNOC_BIMC_MAS);
> +DEFINE_QNODE(snoc_int_1, SNOC_INT_1, 0, 8, 0, 100, 131, QCOM_QOS_MODE_FIXED, 3, SYSTEM_SLAVE_FAB_APPS, SLAVE_CATS_128, SLAVE_OCMEM_64);
> +DEFINE_QNODE(snoc_int_bimc, SNOC_INT_BIMC, 0, 8, 0, 101, 132, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_0_MAS);
> +DEFINE_QNODE(snoc_bimc_0_mas, SNOC_BIMC_0_MAS, 0, 8, 0, 3, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_0_SLV);
> +DEFINE_QNODE(snoc_bimc_1_mas, SNOC_BIMC_1_MAS, 0, 16, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SNOC_BIMC_1_SLV);
> +DEFINE_QNODE(qdss_int, SNOC_QDSS_INT, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 2, SNOC_INT_0, SNOC_INT_BIMC);
> +DEFINE_QNODE(bimc_snoc_slv, BIMC_SNOC_SLV, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 2, SNOC_INT_0, SNOC_INT_1);
> +DEFINE_QNODE(snoc_pnoc_mas, MNOC_BIMC_MAS, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, MNOC_BIMC_SLV);
> +DEFINE_QNODE(pnoc_snoc_slv, PNOC_SNOC_SLV, 0, 8, 0, -1, 45, QCOM_QOS_MODE_FIXED, 3, SNOC_INT_0, SNOC_INT_BIMC, SNOC_INT_1);
> +DEFINE_QNODE(slv_srvc_snoc, SLAVE_SERVICE_SNOC, 0, 8, 0, -1, 29, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_qdss_stm, SLAVE_QDSS_STM, 0, 4, 0, -1, 30, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_imem, SLAVE_SYSTEM_IMEM, 0, 8, 0, -1, 26, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_apss, SYSTEM_SLAVE_FAB_APPS, 0, 4, 0, -1, 20, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_cats_0, SLAVE_CATS_128, 0, 16, 0, -1, 106, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_cats_1, SLAVE_OCMEM_64, 0, 8, 0, -1, 107, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(mas_apss, MASTER_AMPSS_M0, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
> +DEFINE_QNODE(mas_tcu0, MASTER_TCU_0, 5, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
> +DEFINE_QNODE(mas_tcu1, MASTER_TCU_1, 6, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
> +DEFINE_QNODE(mas_gfx, MASTER_GRAPHICS_3D, 2, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_EBI_CH0, BIMC_SNOC_MAS, SLAVE_AMPSS_L2);
> +DEFINE_QNODE(bimc_snoc_mas, BIMC_SNOC_MAS, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, BIMC_SNOC_SLV);
> +DEFINE_QNODE(snoc_bimc_0_slv, SNOC_BIMC_0_SLV, 0, 8, 0, -1, 24, QCOM_QOS_MODE_FIXED, 1, SLAVE_EBI_CH0);
> +DEFINE_QNODE(snoc_bimc_1_slv, SNOC_BIMC_1_SLV, 0, 8, 1, -1, -1, QCOM_QOS_MODE_FIXED, 1, SLAVE_EBI_CH0);
> +DEFINE_QNODE(slv_ebi_ch0, SLAVE_EBI_CH0, 0, 8, 0, -1, 0, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_apps_l2, SLAVE_AMPSS_L2, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(snoc_pnoc_slv, MNOC_BIMC_SLV, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_0);
> +DEFINE_QNODE(pnoc_int_0, PNOC_INT_0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 8, PNOC_SNOC_MAS, PNOC_SLV_0, PNOC_SLV_1, PNOC_SLV_2, PNOC_SLV_3, PNOC_SLV_4, PNOC_SLV_8, PNOC_SLV_9);
> +DEFINE_QNODE(pnoc_int_1, PNOC_INT_1, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_SNOC_MAS);
> +DEFINE_QNODE(pnoc_m_0, PNOC_M_0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_0);
> +DEFINE_QNODE(pnoc_m_1, PNOC_M_1, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_SNOC_MAS);
> +DEFINE_QNODE(pnoc_s_0, PNOC_SLV_0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_CLK_CTL, SLAVE_TLMM, SLAVE_MSM_TCSR, SLAVE_SECURITY, SLAVE_MSS);
> +DEFINE_QNODE(pnoc_s_1, PNOC_SLV_1, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_IMEM_CFG, SLAVE_CRYPTO_0_CFG, SLAVE_RPM_MSG_RAM, SLAVE_MSM_PDM, SLAVE_PRNG);
> +DEFINE_QNODE(pnoc_s_2, PNOC_SLV_2, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_SPDM, SLAVE_BOOT_ROM, SLAVE_BIMC_CFG, SLAVE_PNOC_CFG, SLAVE_PMIC_ARB);
> +DEFINE_QNODE(pnoc_s_3, PNOC_SLV_3, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 5, SLAVE_MPM, SLAVE_IPS_CFG, SLAVE_RBCPR_CFG, SLAVE_QDSS_CFG, SLAVE_DEHR_CFG);
> +DEFINE_QNODE(pnoc_s_4, PNOC_SLV_4, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_VENUS_CFG, SLAVE_CAMERA_CFG, SLAVE_DISPLAY_CFG);
> +DEFINE_QNODE(pnoc_s_8, PNOC_SLV_8, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_USB_HS, SLAVE_SDCC_1, SLAVE_BLSP_1);
> +DEFINE_QNODE(pnoc_s_9, PNOC_SLV_9, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 3, SLAVE_SDCC_4, SLAVE_LPASS, SLAVE_GRAPHICS_3D_CFG);
> +DEFINE_QNODE(slv_imem_cfg, SLAVE_IMEM_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_crypto_0_cfg, SLAVE_CRYPTO_0_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_msg_ram, SLAVE_RPM_MSG_RAM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_pdm, SLAVE_MSM_PDM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_prng, SLAVE_PRNG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_clk_ctl, SLAVE_CLK_CTL, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_mss, SLAVE_MSS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_tlmm, SLAVE_TLMM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_tcsr, SLAVE_MSM_TCSR, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_security, SLAVE_SECURITY, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_spdm, SLAVE_SPDM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_pnoc_cfg, SLAVE_PNOC_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_pmic_arb, SLAVE_PMIC_ARB, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_bimc_cfg, SLAVE_BIMC_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_boot_rom, SLAVE_BOOT_ROM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_mpm, SLAVE_MPM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_qdss_cfg, SLAVE_QDSS_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_rbcpr_cfg, SLAVE_RBCPR_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_snoc_cfg, SLAVE_IPS_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_dehr_cfg, SLAVE_DEHR_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_venus_cfg, SLAVE_VENUS_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_display_cfg, SLAVE_DISPLAY_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_camera_cfg, SLAVE_CAMERA_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_usb_hs, SLAVE_USB_HS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_sdcc_1, SLAVE_SDCC_1, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_blsp_1, SLAVE_BLSP_1, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_sdcc_2, SLAVE_SDCC_4, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_gfx_cfg, SLAVE_GRAPHICS_3D_CFG, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(slv_audio, SLAVE_LPASS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 0, 0);
> +DEFINE_QNODE(mas_blsp_1, MASTER_BLSP_1, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_1);
> +DEFINE_QNODE(mas_spdm, MASTER_SPDM, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_0);
> +DEFINE_QNODE(mas_dehr, MASTER_DEHR, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_0);
> +DEFINE_QNODE(mas_audio, MASTER_LPASS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_0);
> +DEFINE_QNODE(mas_usb_hs, MASTER_USB_HS, 0, 4, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_M_1);
> +DEFINE_QNODE(mas_pnoc_crypto_0, MASTER_CRYPTO_CORE0, 0, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_1);
> +DEFINE_QNODE(mas_pnoc_sdcc_1, MASTER_SDCC_1, 7, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_1);
> +DEFINE_QNODE(mas_pnoc_sdcc_2, MASTER_SDCC_2, 8, 8, 0, -1, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_INT_1);
> +DEFINE_QNODE(pnoc_snoc_mas, PNOC_SNOC_MAS, 0, 8, 0, 29, -1, QCOM_QOS_MODE_FIXED, 1, PNOC_SNOC_SLV);
> +
> +static struct qcom_icc_node *msm8916_snoc_nodes[] = {
> +       &mas_video,
> +       &mas_jpeg,
> +       &mas_vfe,
> +       &mas_mdp,
> +       &mas_qdss_bam,
> +       &mas_snoc_cfg,
> +       &mas_qdss_etr,
> +       &mm_int_0,
> +       &mm_int_1,
> +       &mm_int_2,
> +       &mm_int_bimc,
> +       &snoc_int_0,
> +       &snoc_int_1,
> +       &snoc_int_bimc,
> +       &snoc_bimc_0_mas,
> +       &snoc_bimc_1_mas,
> +       &qdss_int,
> +       &bimc_snoc_slv,
> +       &snoc_pnoc_mas,
> +       &pnoc_snoc_slv,
> +       &slv_srvc_snoc,
> +       &slv_qdss_stm,
> +       &slv_imem,
> +       &slv_apss,
> +       &slv_cats_0,
> +       &slv_cats_1,
> +};
> +
> +static struct qcom_icc_desc msm8916_snoc = {
> +       .nodes = msm8916_snoc_nodes,
> +       .num_nodes = ARRAY_SIZE(msm8916_snoc_nodes),
> +};
> +
> +static struct qcom_icc_node *msm8916_bimc_nodes[] = {
> +       &mas_apss,
> +       &mas_tcu0,
> +       &mas_tcu1,
> +       &mas_gfx,
> +       &bimc_snoc_mas,
> +       &snoc_bimc_0_slv,
> +       &snoc_bimc_1_slv,
> +       &slv_ebi_ch0,
> +       &slv_apps_l2,
> +};
> +
> +static struct qcom_icc_desc msm8916_bimc = {
> +       .nodes = msm8916_bimc_nodes,
> +       .num_nodes = ARRAY_SIZE(msm8916_bimc_nodes),
> +};
> +
> +static struct qcom_icc_node *msm8916_pnoc_nodes[] = {
> +       &snoc_pnoc_slv,
> +       &pnoc_int_0,
> +       &pnoc_int_1,
> +       &pnoc_m_0,
> +       &pnoc_m_1,
> +       &pnoc_s_0,
> +       &pnoc_s_1,
> +       &pnoc_s_2,
> +       &pnoc_s_3,
> +       &pnoc_s_4,
> +       &pnoc_s_8,
> +       &pnoc_s_9,
> +       &slv_imem_cfg,
> +       &slv_crypto_0_cfg,
> +       &slv_msg_ram,
> +       &slv_pdm,
> +       &slv_prng,
> +       &slv_clk_ctl,
> +       &slv_mss,
> +       &slv_tlmm,
> +       &slv_tcsr,
> +       &slv_security,
> +       &slv_spdm,
> +       &slv_pnoc_cfg,
> +       &slv_pmic_arb,
> +       &slv_bimc_cfg,
> +       &slv_boot_rom,
> +       &slv_mpm,
> +       &slv_qdss_cfg,
> +       &slv_rbcpr_cfg,
> +       &slv_snoc_cfg,
> +       &slv_dehr_cfg,
> +       &slv_venus_cfg,
> +       &slv_display_cfg,
> +       &slv_camera_cfg,
> +       &slv_usb_hs,
> +       &slv_sdcc_1,
> +       &slv_blsp_1,
> +       &slv_sdcc_2,
> +       &slv_gfx_cfg,
> +       &slv_audio,
> +       &mas_blsp_1,
> +       &mas_spdm,
> +       &mas_dehr,
> +       &mas_audio,
> +       &mas_usb_hs,
> +       &mas_pnoc_crypto_0,
> +       &mas_pnoc_sdcc_1,
> +       &mas_pnoc_sdcc_2,
> +       &pnoc_snoc_mas,
> +};
> +
> +static struct qcom_icc_desc msm8916_pnoc = {
> +       .nodes = msm8916_pnoc_nodes,
> +       .num_nodes = ARRAY_SIZE(msm8916_pnoc_nodes),
> +};
> +
> +static int qcom_icc_init(struct icc_node *node)
> +{
> +       struct qcom_icc_provider *qp = to_qcom_provider(node->provider);
> +       int ret;
> +
> +       /* TODO: init qos and priority */
> +
> +       clk_set_rate(qp->bus_clk, INT_MAX);

Vroom! What's the rationale here? I wonder if it might be better to
avoid touching the clocks initially, and expect that the boot loader
sets up a decent initial set of bus frequencies for consumers that
never enable bus scaling? Otherwise, I worry that this driver becomes
basically an essential driver for the platform solely because of this
line and the one below, when really it might not be.

> +       ret = clk_prepare_enable(qp->bus_clk);
> +       if (ret)
> +               pr_info("%s: error enabling bus clk (%d)\n", __func__, ret);
> +       clk_set_rate(qp->bus_a_clk, INT_MAX);
> +       ret = clk_prepare_enable(qp->bus_a_clk);
> +       if (ret)
> +               pr_info("%s: error enabling bus_a clk (%d)\n", __func__, ret);
> +
> +       return 0;
> +}
> +
> +static int qcom_icc_aggregate(struct icc_node *node, u32 avg_bw, u32 peak_bw,
> +                             u32 *agg_avg, u32 *agg_peak)
> +{
> +       *agg_avg += avg_bw;
> +       *agg_peak = max(*agg_peak, peak_bw);
> +
> +       return 0;
> +}
> +
> +static int qcom_icc_set(struct icc_node *src, struct icc_node *dst,
> +                       u32 avg, u32 peak)
> +{
> +       struct qcom_icc_provider *qp;
> +       struct qcom_icc_node *qn;
> +       struct icc_node *node;
> +       struct icc_provider *provider;
> +       u64 avg_bw = icc_units_to_bps(avg);
> +       u64 peak_bw = icc_units_to_bps(peak);
> +       u64 rate = 0;
> +       int ret = 0;
> +
> +       if (!src)
> +               node = dst;
> +       else
> +               node = src;
> +
> +       qn = node->data;
> +       provider = node->provider;
> +       qp = to_qcom_provider(node->provider);
> +
> +       /* set bandwidth */
> +       if (qn->ap_owned) {
> +               /* TODO: set QoS */
> +       } else {
> +               /* send message to the RPM processor */
> +               if (qn->mas_rpm_id != -1) {
> +                       ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
> +                                                   RPM_BUS_MASTER_REQ,
> +                                                   qn->mas_rpm_id,
> +                                                   avg_bw);

You don't check for failure here, and potentially clobber failure with
success. Unless only one of mas_rpm_id and slv_rpm_id is ever set. But
no, I see SNOC_INT_0/1 set both. Same for below.

> +               }
> +
> +               if (qn->slv_rpm_id != -1) {
> +                       ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
> +                                                   RPM_BUS_SLAVE_REQ,
> +                                                   qn->slv_rpm_id,
> +                                                   avg_bw);
> +               }
> +       }
> +
> +       rate = max(avg_bw, peak_bw);
> +
> +       do_div(rate, qn->buswidth);
> +
> +       if (qn->rate != rate) {
> +               ret = clk_set_rate(qp->bus_clk, rate);
> +               if (ret) {
> +                       pr_err("set clk rate %lld error %d\n", rate, ret);
> +                       return ret;
> +               }
> +
> +               ret = clk_set_rate(qp->bus_a_clk, rate);
> +               if (ret) {
> +                       pr_err("set clk rate %lld error %d\n", rate, ret);
> +                       return ret;
> +               }
> +
> +               qn->rate = rate;
> +       }
> +
> +       return ret;
> +}
> +
> +static int qnoc_probe(struct platform_device *pdev)
> +{
> +       const struct qcom_icc_desc *desc;
> +       struct qcom_icc_node **qnodes;
> +       struct qcom_icc_provider *qp;
> +       struct resource *res;
> +       struct icc_provider *provider;
> +       size_t num_nodes, i;
> +       int ret;
> +
> +       /* wait for RPM */
> +       if (!qcom_icc_rpm_smd_available())
> +               return -EPROBE_DEFER;
> +
> +       desc = of_device_get_match_data(&pdev->dev);
> +       if (!desc)
> +               return -EINVAL;
> +
> +       qnodes = desc->nodes;
> +       num_nodes = desc->num_nodes;
> +
> +       qp = devm_kzalloc(&pdev->dev, sizeof(*qp), GFP_KERNEL);
> +       if (!qp)
> +               return -ENOMEM;
> +
> +       res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> +       qp->base = devm_ioremap_resource(&pdev->dev, res);
> +       if (IS_ERR(qp->base))
> +               return PTR_ERR(qp->base);
> +
> +       qp->bus_clk = devm_clk_get(&pdev->dev, "bus_clk");
> +       if (IS_ERR(qp->bus_clk))
> +               return PTR_ERR(qp->bus_clk);
> +
> +       qp->bus_a_clk = devm_clk_get(&pdev->dev, "bus_a_clk");
> +       if (IS_ERR(qp->bus_a_clk))
> +               return PTR_ERR(qp->bus_a_clk);
> +
> +       provider = &qp->provider;
> +       provider->dev = &pdev->dev;
> +       provider->set = &qcom_icc_set;
> +       provider->aggregate = &qcom_icc_aggregate;
> +       INIT_LIST_HEAD(&provider->nodes);
> +       provider->data = qp;
> +
> +       ret = icc_provider_add(provider);
> +       if (ret) {
> +               dev_err(&pdev->dev, "error adding interconnect provider\n");
> +               return ret;
> +       }
> +
> +       for (i = 0; i < num_nodes; i++) {
> +               struct icc_node *node;
> +               int ret;
> +               size_t j;
> +
> +               node = icc_node_create(qnodes[i]->id);
> +               if (IS_ERR(node)) {
> +                       ret = PTR_ERR(node);
> +                       goto err;
> +               }
> +
> +               node->name = qnodes[i]->name;
> +               node->data = qnodes[i];
> +               icc_node_add(node, provider);
> +
> +               dev_dbg(&pdev->dev, "registered node %s\n", node->name);
> +
> +               ret = qcom_icc_init(node);
> +               if (ret)
> +                       dev_err(&pdev->dev, "%s init error (%d)\n", node->name,
> +                               ret);
> +
> +               /* populate links */
> +               for (j = 0; j < qnodes[i]->num_links; j++)
> +                       if (qnodes[i]->links[j])
> +                               icc_link_create(node, qnodes[i]->links[j]);
> +       }
> +
> +       platform_set_drvdata(pdev, provider);
> +
> +       return ret;
> +err:
> +       icc_provider_del(provider);
> +       return ret;
> +}
> +
> +static int qnoc_remove(struct platform_device *pdev)
> +{
> +       struct icc_provider *provider = platform_get_drvdata(pdev);
> +
> +       icc_provider_del(provider);

This function can fail, so you should probably pass the failure along.

> +
> +       return 0;
> +}
> +
> +static const struct of_device_id qnoc_of_match[] = {
> +       { .compatible = "qcom,msm8916-pnoc", .data = &msm8916_pnoc },
> +       { .compatible = "qcom,msm8916-snoc", .data = &msm8916_snoc },
> +       { .compatible = "qcom,msm8916-bimc", .data = &msm8916_bimc },
> +       { },
> +};
> +MODULE_DEVICE_TABLE(of, qnoc_of_match);
> +
> +static struct platform_driver qnoc_driver = {
> +       .probe = qnoc_probe,
> +       .remove = qnoc_remove,
> +       .driver = {
> +               .name = "qnoc-msm8916",
> +               .of_match_table = qnoc_of_match,
> +       },
> +};
> +module_platform_driver(qnoc_driver);
> +MODULE_AUTHOR("Georgi Djakov <georgi.djakov@linaro.org>");
> +MODULE_DESCRIPTION("Qualcomm msm8916 NoC driver");
> +MODULE_LICENSE("GPL v2");

Re: [PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Evan Green]

Hi Georgi. Thanks for the new spin of this.

On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> This patch introduce a new API to get requirements and configure the
> interconnect buses across the entire chipset to fit with the current
> demand.
>
> The API is using a consumer/provider-based model, where the providers are
> the interconnect buses and the consumers could be various drivers.
> The consumers request interconnect resources (path) between endpoints and
> set the desired constraints on this data flow path. The providers receive
> requests from consumers and aggregate these requests for all master-slave
> pairs on that path. Then the providers configure each participating in the
> topology node according to the requested data flow path, physical links and
> constraints. The topology could be complicated and multi-tiered and is SoC
> specific.
>
> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
> ---
>  Documentation/interconnect/interconnect.rst |  96 ++++
>  drivers/Kconfig                             |   2 +
>  drivers/Makefile                            |   1 +
>  drivers/interconnect/Kconfig                |  10 +
>  drivers/interconnect/Makefile               |   2 +
>  drivers/interconnect/core.c                 | 586 ++++++++++++++++++++
>  include/linux/interconnect-provider.h       | 127 +++++
>  include/linux/interconnect.h                |  42 ++
>  8 files changed, 866 insertions(+)
>  create mode 100644 Documentation/interconnect/interconnect.rst
>  create mode 100644 drivers/interconnect/Kconfig
>  create mode 100644 drivers/interconnect/Makefile
>  create mode 100644 drivers/interconnect/core.c
>  create mode 100644 include/linux/interconnect-provider.h
>  create mode 100644 include/linux/interconnect.h
>
> diff --git a/Documentation/interconnect/interconnect.rst b/Documentation/interconnect/interconnect.rst
> new file mode 100644
> index 000000000000..a1ebd83ad0a1
> --- /dev/null
> +++ b/Documentation/interconnect/interconnect.rst
> @@ -0,0 +1,96 @@
> +.. SPDX-License-Identifier: GPL-2.0
> +
> +=====================================
> +GENERIC SYSTEM INTERCONNECT SUBSYSTEM
> +=====================================
> +
> +Introduction
> +------------
> +
> +This framework is designed to provide a standard kernel interface to control
> +the settings of the interconnects on a SoC. These settings can be throughput,
> +latency and priority between multiple interconnected devices or functional
> +blocks. This can be controlled dynamically in order to save power or provide
> +maximum performance.
> +
> +The interconnect bus is a hardware with configurable parameters, which can be
> +set on a data path according to the requests received from various drivers.
> +An example of interconnect buses are the interconnects between various
> +components or functional blocks in chipsets. There can be multiple interconnects
> +on a SoC that can be multi-tiered.
> +
> +Below is a simplified diagram of a real-world SoC interconnect bus topology.
> +
> +::
> +
> + +----------------+    +----------------+
> + | HW Accelerator |--->|      M NoC     |<---------------+
> + +----------------+    +----------------+                |
> +                         |      |                    +------------+
> +  +-----+  +-------------+      V       +------+     |            |
> +  | DDR |  |                +--------+  | PCIe |     |            |
> +  +-----+  |                | Slaves |  +------+     |            |
> +    ^ ^    |                +--------+     |         |   C NoC    |
> +    | |    V                               V         |            |
> + +------------------+   +------------------------+   |            |   +-----+
> + |                  |-->|                        |-->|            |-->| CPU |
> + |                  |-->|                        |<--|            |   +-----+
> + |     Mem NoC      |   |         S NoC          |   +------------+
> + |                  |<--|                        |---------+    |
> + |                  |<--|                        |<------+ |    |   +--------+
> + +------------------+   +------------------------+       | |    +-->| Slaves |
> +   ^  ^    ^    ^          ^                             | |        +--------+
> +   |  |    |    |          |                             | V
> + +------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
> + | CPUs |  |  | GPU |   | DSP |  | Masters |-->|       P NoC    |-->| Slaves |
> + +------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
> +           |
> +       +-------+
> +       | Modem |
> +       +-------+
> +
> +Terminology
> +-----------
> +
> +Interconnect provider is the software definition of the interconnect hardware.
> +The interconnect providers on the above diagram are M NoC, S NoC, C NoC, P NoC
> +and Mem NoC.
> +
> +Interconnect node is the software definition of the interconnect hardware
> +port. Each interconnect provider consists of multiple interconnect nodes,
> +which are connected to other SoC components including other interconnect
> +providers. The point on the diagram where the CPUs connects to the memory is
> +called an interconnect node, which belongs to the Mem NoC interconnect provider.
> +
> +Interconnect endpoints are the first or the last element of the path. Every
> +endpoint is a node, but not every node is an endpoint.
> +
> +Interconnect path is everything between two endpoints including all the nodes
> +that have to be traversed to reach from a source to destination node. It may
> +include multiple master-slave pairs across several interconnect providers.
> +
> +Interconnect consumers are the entities which make use of the data paths exposed
> +by the providers. The consumers send requests to providers requesting various
> +throughput, latency and priority. Usually the consumers are device drivers, that
> +send request based on their needs. An example for a consumer is a video decoder
> +that supports various formats and image sizes.
> +
> +Interconnect providers
> +----------------------
> +
> +Interconnect provider is an entity that implements methods to initialize and
> +configure a interconnect bus hardware. The interconnect provider drivers should
> +be registered with the interconnect provider core.
> +
> +The interconnect framework provider API functions are documented in
> +.. kernel-doc:: include/linux/interconnect-provider.h
> +
> +Interconnect consumers
> +----------------------
> +
> +Interconnect consumers are the clients which use the interconnect APIs to
> +get paths between endpoints and set their bandwidth/latency/QoS requirements
> +for these interconnect paths.
> +
> +The interconnect framework consumer API functions are documented in
> +.. kernel-doc:: include/linux/interconnect.h
> diff --git a/drivers/Kconfig b/drivers/Kconfig
> index 95b9ccc08165..3ed6ede9d021 100644
> --- a/drivers/Kconfig
> +++ b/drivers/Kconfig
> @@ -217,4 +217,6 @@ source "drivers/siox/Kconfig"
>
>  source "drivers/slimbus/Kconfig"
>
> +source "drivers/interconnect/Kconfig"
> +
>  endmenu
> diff --git a/drivers/Makefile b/drivers/Makefile
> index 24cd47014657..0cca95740d9b 100644
> --- a/drivers/Makefile
> +++ b/drivers/Makefile
> @@ -185,3 +185,4 @@ obj-$(CONFIG_TEE)           += tee/
>  obj-$(CONFIG_MULTIPLEXER)      += mux/
>  obj-$(CONFIG_UNISYS_VISORBUS)  += visorbus/
>  obj-$(CONFIG_SIOX)             += siox/
> +obj-$(CONFIG_INTERCONNECT)     += interconnect/
> diff --git a/drivers/interconnect/Kconfig b/drivers/interconnect/Kconfig
> new file mode 100644
> index 000000000000..a261c7d41deb
> --- /dev/null
> +++ b/drivers/interconnect/Kconfig
> @@ -0,0 +1,10 @@
> +menuconfig INTERCONNECT
> +       tristate "On-Chip Interconnect management support"
> +       help
> +         Support for management of the on-chip interconnects.
> +
> +         This framework is designed to provide a generic interface for
> +         managing the interconnects in a SoC.
> +
> +         If unsure, say no.
> +
> diff --git a/drivers/interconnect/Makefile b/drivers/interconnect/Makefile
> new file mode 100644
> index 000000000000..97fca2e09d24
> --- /dev/null
> +++ b/drivers/interconnect/Makefile
> @@ -0,0 +1,2 @@
> +# SPDX-License-Identifier: GPL-2.0
> +obj-$(CONFIG_INTERCONNECT)             += core.o
> diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
> new file mode 100644
> index 000000000000..e7f96fc6722e
> --- /dev/null
> +++ b/drivers/interconnect/core.c
> @@ -0,0 +1,586 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Interconnect framework core driver
> + *
> + * Copyright (c) 2018, Linaro Ltd.
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#include <linux/device.h>
> +#include <linux/idr.h>
> +#include <linux/init.h>
> +#include <linux/interconnect.h>
> +#include <linux/interconnect-provider.h>
> +#include <linux/list.h>
> +#include <linux/module.h>
> +#include <linux/mutex.h>
> +#include <linux/slab.h>
> +
> +static DEFINE_IDR(icc_idr);
> +static LIST_HEAD(icc_provider_list);
> +static DEFINE_MUTEX(icc_lock);
> +
> +/**
> + * struct icc_req - constraints that are attached to each node
> + *
> + * @req_node: entry in list of requests for the particular @node
> + * @node: the interconnect node to which this constraint applies
> + * @dev: reference to the device that sets the constraints
> + * @avg_bw: an integer describing the average bandwidth in kbps
> + * @peak_bw: an integer describing the peak bandwidth in kbps
> + */
> +struct icc_req {
> +       struct hlist_node req_node;
> +       struct icc_node *node;
> +       struct device *dev;
> +       u32 avg_bw;
> +       u32 peak_bw;
> +};
> +
> +/**
> + * struct icc_path - interconnect path structure
> + * @num_nodes: number of hops (nodes)
> + * @reqs: array of the requests applicable to this path of nodes
> + */
> +struct icc_path {
> +       size_t num_nodes;
> +       struct icc_req reqs[0];
> +};
> +
> +static struct icc_node *node_find(const int id)
> +{
> +       struct icc_node *node;
> +
> +       mutex_lock(&icc_lock);
> +       node = idr_find(&icc_idr, id);
> +       mutex_unlock(&icc_lock);

I wonder if this is too low of a level to be dealing with the lock. I
notice that everywhere you use this function, you afterwards
immediately grab the lock and do more stuff. Maybe this function
should have a comment saying it assumes the lock is already held, and
then you can grab the lock in the callers, since you're doing that
anyway.

> +
> +       return node;
> +}
> +
> +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
> +{
> +       struct icc_node *node = dst;
> +       struct icc_path *path;
> +       size_t i;
> +
> +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
> +                      GFP_KERNEL);
> +       if (!path)
> +               return ERR_PTR(-ENOMEM);
> +
> +       path->num_nodes = num_nodes;
> +
> +       for (i = 0; i < num_nodes; i++) {
> +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
> +
> +               path->reqs[i].node = node;
> +               /* reference to previous node was saved during path traversal */
> +               node = node->reverse;
> +       }
> +
> +       return path;
> +}
> +
> +static struct icc_path *path_find(struct device *dev, struct icc_node *src,
> +                                 struct icc_node *dst)
> +{

I personally prefer a comment somewhere indicating that this function
assumes icc_lock is already held. Not sure if that's conventional or
not.

> +       struct icc_node *n, *node = NULL;
> +       struct icc_provider *provider;
> +       struct list_head traverse_list;
> +       struct list_head edge_list;
> +       struct list_head visited_list;
> +       size_t i, depth = 0;
> +       bool found = false;
> +       int ret = -EPROBE_DEFER;
> +
> +       INIT_LIST_HEAD(&traverse_list);
> +       INIT_LIST_HEAD(&edge_list);
> +       INIT_LIST_HEAD(&visited_list);
> +
> +       list_add_tail(&src->search_list, &traverse_list);
> +       src->reverse = NULL;
> +
> +       do {
> +               list_for_each_entry_safe(node, n, &traverse_list, search_list) {
> +                       if (node == dst) {
> +                               found = true;
> +                               list_add(&node->search_list, &visited_list);
> +                               break;
> +                       }
> +                       for (i = 0; i < node->num_links; i++) {
> +                               struct icc_node *tmp = node->links[i];
> +
> +                               if (!tmp) {
> +                                       ret = -ENOENT;
> +                                       goto out;
> +                               }
> +
> +                               if (tmp->is_traversed)
> +                                       continue;
> +
> +                               tmp->is_traversed = true;
> +                               tmp->reverse = node;
> +                               list_add(&tmp->search_list, &edge_list);
> +                       }
> +               }
> +               if (found)
> +                       break;
> +
> +               list_splice_init(&traverse_list, &visited_list);
> +               list_splice_init(&edge_list, &traverse_list);
> +
> +               /* count the hops away from the source */
> +               depth++;
> +
> +       } while (!list_empty(&traverse_list));
> +
> +out:
> +       /* reset the traversed state */
> +       list_for_each_entry(provider, &icc_provider_list, provider_list) {
> +               list_for_each_entry(n, &provider->nodes, node_list)
> +                       if (n->is_traversed)
> +                               n->is_traversed = false;

Remove the conditional, just set is_traversed to false.

> +       }
> +
> +       if (found) {
> +               struct icc_path *path = path_allocate(dst, depth);

Is the path supposed to include the source? For instance, if the dst
were a neighbor, depth would be one, so only dst would be in the path.
It seems like it might be worthwhile to have the source in there too.

> +
> +               if (IS_ERR(path))
> +                       return path;
> +
> +               /* initialize the path */
> +               for (i = 0; i < path->num_nodes; i++) {
> +                       node = path->reqs[i].node;
> +                       path->reqs[i].dev = dev;
> +                       node->provider->users++;

Should this loop live inside path_allocate? I'm unsure, but maybe at
least path->reqs[i].dev = dev, since it feels like standard
initialization of the path.

> +               }
> +               return path;
> +       }
> +
> +       return ERR_PTR(ret);
> +}
> +
> +/*
> + * We want the path to honor all bandwidth requests, so the average
> + * bandwidth requirements from each consumer are aggregated at each node
> + * and provider level. By default the average bandwidth is the sum of all
> + * averages and the peak will be the highest of all peak bandwidth requests.
> + */
> +
> +static int aggregate_requests(struct icc_node *node)
> +{
> +       struct icc_provider *p = node->provider;
> +       struct icc_req *r;
> +
> +       node->avg_bw = 0;
> +       node->peak_bw = 0;
> +
> +       hlist_for_each_entry(r, &node->req_list, req_node)
> +               p->aggregate(node, r->avg_bw, r->peak_bw,
> +                            &node->avg_bw, &node->peak_bw);
> +
> +       return 0;
> +}
> +
> +static void aggregate_provider(struct icc_provider *p)
> +{
> +       struct icc_node *n;
> +
> +       p->avg_bw = 0;
> +       p->peak_bw = 0;
> +
> +       list_for_each_entry(n, &p->nodes, node_list)
> +               p->aggregate(n, n->avg_bw, n->peak_bw,
> +                            &p->avg_bw, &p->peak_bw);
> +}
> +
> +static int apply_constraints(struct icc_path *path)
> +{
> +       struct icc_node *next, *prev = NULL;
> +       int ret = 0;
> +       int i;
> +
> +       for (i = 0; i < path->num_nodes; i++, prev = next) {
> +               struct icc_provider *p;
> +
> +               next = path->reqs[i].node;
> +               /*
> +                * Both endpoints should be valid master-slave pairs of the
> +                * same interconnect provider that will be configured.
> +                */
> +               if (!prev || next->provider != prev->provider)
> +                       continue;
> +
> +               p = next->provider;
> +
> +               aggregate_provider(p);
> +
> +               if (p->set) {
> +                       /* set the constraints */
> +                       ret = p->set(prev, next, p->avg_bw, p->peak_bw);
> +               }
> +
> +               if (ret)
> +                       goto out;
> +       }
> +out:
> +       return ret;
> +}
> +
> +/**
> + * icc_set() - set constraints on an interconnect path between two endpoints
> + * @path: reference to the path returned by icc_get()
> + * @avg_bw: average bandwidth in kbps
> + * @peak_bw: peak bandwidth in kbps
> + *
> + * This function is used by an interconnect consumer to express its own needs
> + * in terms of bandwidth for a previously requested path between two endpoints.
> + * The requests are aggregated and each node is updated accordingly. The entire
> + * path is locked by a mutex to ensure that the set() is completed.
> + * The @path can be NULL when the "interconnects" DT properties is missing,
> + * which will mean that no constraints will be set.
> + *
> + * Returns 0 on success, or an appropriate error code otherwise.
> + */
> +int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
> +{
> +       struct icc_node *node;
> +       struct icc_provider *p;
> +       size_t i;
> +       int ret = 0;
> +
> +       if (!path)
> +               return 0;
> +
> +       mutex_lock(&icc_lock);
> +
> +       for (i = 0; i < path->num_nodes; i++) {
> +               node = path->reqs[i].node;
> +               p = node->provider;
> +
> +               /* update the consumer request for this path */
> +               path->reqs[i].avg_bw = avg_bw;
> +               path->reqs[i].peak_bw = peak_bw;
> +
> +               /* aggregate requests for this node */
> +               aggregate_requests(node);
> +       }
> +
> +       ret = apply_constraints(path);
> +       if (ret)
> +               pr_err("interconnect: error applying constraints (%d)", ret);
> +
> +       mutex_unlock(&icc_lock);
> +
> +       return ret;
> +}
> +EXPORT_SYMBOL_GPL(icc_set);
> +
> +/**
> + * icc_get() - return a handle for path between two endpoints
> + * @dev: the device requesting the path
> + * @src_id: source device port id
> + * @dst_id: destination device port id
> + *
> + * This function will search for a path between two endpoints and return an
> + * icc_path handle on success. Use icc_put() to release
> + * constraints when the they are not needed anymore.
> + *
> + * Return: icc_path pointer on success, or ERR_PTR() on error
> + */
> +struct icc_path *icc_get(struct device *dev, const int src_id, const int dst_id)
> +{
> +       struct icc_node *src, *dst;
> +       struct icc_path *path = ERR_PTR(-EPROBE_DEFER);
> +
> +       src = node_find(src_id);
> +       if (!src) {
> +               dev_err(dev, "%s: invalid src=%d\n", __func__, src_id);
> +               goto out;
> +       }
> +
> +       dst = node_find(dst_id);
> +       if (!dst) {
> +               dev_err(dev, "%s: invalid dst=%d\n", __func__, dst_id);
> +               goto out;
> +       }
> +
> +       mutex_lock(&icc_lock);
> +       path = path_find(dev, src, dst);
> +       mutex_unlock(&icc_lock);
> +       if (IS_ERR(path)) {
> +               dev_err(dev, "%s: invalid path=%ld\n", __func__, PTR_ERR(path));
> +               goto out;
> +       }
> +
> +out:
> +       return path;
> +}
> +EXPORT_SYMBOL_GPL(icc_get);
> +
> +/**
> + * icc_put() - release the reference to the icc_path
> + * @path: interconnect path
> + *
> + * Use this function to release the constraints on a path when the path is
> + * no longer needed. The constraints will be re-aggregated.
> + */
> +void icc_put(struct icc_path *path)
> +{
> +       struct icc_node *node;
> +       size_t i;
> +       int ret;
> +
> +       if (!path || WARN_ON_ONCE(IS_ERR(path)))

Why only once?

> +               return;
> +
> +       ret = icc_set(path, 0, 0);
> +       if (ret)
> +               pr_err("%s: error (%d)\n", __func__, ret);
> +
> +       mutex_lock(&icc_lock);
> +       for (i = 0; i < path->num_nodes; i++) {
> +               node = path->reqs[i].node;
> +               hlist_del(&path->reqs[i].req_node);
> +
> +               node->provider->users--;
> +       }
> +       mutex_unlock(&icc_lock);
> +
> +       kfree(path);
> +}
> +EXPORT_SYMBOL_GPL(icc_put);
> +
> +/**
> + * icc_node_create() - create a node
> + * @id: node id
> + *
> + * Return: icc_node pointer on success, or ERR_PTR() on error
> + */
> +struct icc_node *icc_node_create(int id)
> +{
> +       struct icc_node *node;
> +
> +       /* check if node already exists */
> +       node = node_find(id);
> +       if (node)
> +               goto out;
> +
> +       node = kzalloc(sizeof(*node), GFP_KERNEL);
> +       if (!node) {
> +               node = ERR_PTR(-ENOMEM);
> +               goto out;
> +       }
> +
> +       mutex_lock(&icc_lock);
> +
> +       id = idr_alloc(&icc_idr, node, id, id + 1, GFP_KERNEL);
> +       if (WARN(id < 0, "couldn't get idr")) {
> +               node = ERR_PTR(id);
> +               goto out;
> +       }
> +
> +       node->id = id;
> +
> +out:
> +       mutex_unlock(&icc_lock);
> +
> +       return node;
> +}
> +EXPORT_SYMBOL_GPL(icc_node_create);
> +
> +/**
> + * icc_node_remove() - remove a node
> + * @id: node id
> + *
> + */
> +void icc_node_remove(int id)
> +{
> +       struct icc_node *node;
> +
> +       node = node_find(id);
> +       if (node) {
> +               mutex_lock(&icc_lock);
> +               idr_remove(&icc_idr, node->id);

Should we throw a warning if there are any paths that go through this
node (ie req_list is non-empty)?

> +               mutex_unlock(&icc_lock);
> +       }
> +
> +       kfree(node);
> +}
> +EXPORT_SYMBOL_GPL(icc_node_remove);
> +
> +/**
> + * icc_link_create() - create a link between two nodes
> + * @src_id: source node id
> + * @dst_id: destination node id
> + *
> + * Create a link between two nodes. The nodes might belong to different
> + * interconnect providers and the @dst_id node might not exist (if the
> + * provider driver has not probed yet). So just create the @dst_id node
> + * and when the actual provider driver is probed, the rest of the node
> + * data is filled.
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_link_create(struct icc_node *node, const int dst_id)
> +{
> +       struct icc_node *dst;
> +       struct icc_node **new;
> +       int ret = 0;
> +
> +       if (!node->provider)
> +               return -EINVAL;
> +
> +       dst = node_find(dst_id);
> +       if (!dst) {
> +               dst = icc_node_create(dst_id);
> +
> +               if (IS_ERR(dst)) {
> +                       ret = PTR_ERR(dst);
> +                       goto out;
> +               }
> +       }
> +
> +       mutex_lock(&icc_lock);
> +
> +       new = krealloc(node->links,
> +                      (node->num_links + 1) * sizeof(*node->links),
> +                      GFP_KERNEL);
> +       if (!new) {
> +               ret = -ENOMEM;
> +               goto out;
> +       }
> +
> +       node->links = new;
> +       node->links[node->num_links++] = dst;
> +
> +out:
> +       mutex_unlock(&icc_lock);
> +
> +       return ret;
> +}
> +EXPORT_SYMBOL_GPL(icc_link_create);
> +
> +/**
> + * icc_link_remove() - remove a link between two nodes
> + * @src: pointer to source node
> + * @dst: pointer to destination node
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_link_remove(struct icc_node *src, struct icc_node *dst)
> +{
> +       struct icc_node **new;
> +       int ret = 0;
> +       int i, j;
> +
> +       if (IS_ERR_OR_NULL(src))
> +               return PTR_ERR(src);
> +
> +       if (IS_ERR_OR_NULL(dst))
> +               return PTR_ERR(dst);

I wonder if we should return a fixed error in these cases like
-EINVAL, rather than handing through whatever crazy value is in
src/dst.

> +
> +       mutex_lock(&icc_lock);
> +
> +       new = krealloc(src->links,
> +                      (src->num_links - 1) * sizeof(*src->links),
> +                      GFP_KERNEL);
> +       if (!new) {
> +               ret = -ENOMEM;
> +               goto out;
> +       }
> +
> +       for (i = 0, j = 0; j < src->num_links; j++) {
> +               if (src->links[j] == dst)
> +                       continue;
> +
> +               new[i++] = src->links[j];
> +       }
> +
> +       src->links = new;
> +       src->num_links--;

My understanding is that once you call realloc and it succeeds, you
must assume your old memory is gone and your new memory is only as big
as the new size you request. So you shouldn't call krealloc until
you've fixed the array up. Is the order of the links array important?
If not, you could just take the element at the end and stick it in the
slot that's being deleted. Then decrease the size and do your realloc.

> +
> +out:
> +       mutex_unlock(&icc_lock);
> +
> +       return ret;
> +}
> +EXPORT_SYMBOL_GPL(icc_link_remove);
> +
> +/**
> + * icc_node_add() - add an interconnect node to interconnect provider
> + * @node: pointer to the interconnect node
> + * @provider: pointer to the interconnect provider
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_node_add(struct icc_node *node, struct icc_provider *provider)
> +{
> +       mutex_lock(&icc_lock);
> +
> +       node->provider = provider;
> +       list_add(&node->node_list, &provider->nodes);
> +
> +       mutex_unlock(&icc_lock);
> +
> +       return 0;
> +}
> +EXPORT_SYMBOL_GPL(icc_node_add);
> +
> +/**
> + * icc_provider_add() - add a new interconnect provider
> + * @icc_provider: the interconnect provider that will be added into topology
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_provider_add(struct icc_provider *provider)
> +{
> +       if (WARN_ON(!provider->set))
> +               return -EINVAL;
> +
> +       mutex_init(&icc_lock);
> +
> +       INIT_LIST_HEAD(&provider->nodes);
> +       list_add(&provider->provider_list, &icc_provider_list);
> +
> +       mutex_unlock(&icc_lock);
> +
> +       dev_dbg(provider->dev, "interconnect provider added to topology\n");
> +
> +       return 0;
> +}
> +EXPORT_SYMBOL_GPL(icc_provider_add);
> +
> +/**
> + * icc_provider_del() - delete previously added interconnect provider
> + * @icc_provider: the interconnect provider that will be removed from topology
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_provider_del(struct icc_provider *provider)
> +{
> +       mutex_lock(&icc_lock);
> +       if (provider->users) {
> +               pr_warn("interconnect provider still has %d users\n",
> +                       provider->users);
> +               mutex_unlock(&icc_lock);
> +               return -EBUSY;
> +       }
> +
> +       if (!list_empty_careful(&provider->nodes)) {
> +               pr_warn("interconnect provider still has nodes\n");
> +               mutex_unlock(&icc_lock);
> +               return -EEXIST;

How come you're returning different error codes for these two cases?
The error in both cases is effectively "you failed to clean up after
yourself", so maybe EBUSY makes sense for both of them. The pr_warn
helps to differentiate between the two for debugging.

> +       }
> +
> +       list_del(&provider->provider_list);
> +       mutex_unlock(&icc_lock);
> +
> +       return 0;
> +}
> +EXPORT_SYMBOL_GPL(icc_provider_del);
> +
> +MODULE_AUTHOR("Georgi Djakov <georgi.djakov@linaro.org");
> +MODULE_DESCRIPTION("Interconnect Driver Core");
> +MODULE_LICENSE("GPL v2");
> diff --git a/include/linux/interconnect-provider.h b/include/linux/interconnect-provider.h
> new file mode 100644
> index 000000000000..f4613c6dce4f
> --- /dev/null
> +++ b/include/linux/interconnect-provider.h
> @@ -0,0 +1,127 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Copyright (c) 2018, Linaro Ltd.
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#ifndef _LINUX_INTERCONNECT_PROVIDER_H
> +#define _LINUX_INTERCONNECT_PROVIDER_H
> +
> +#include <linux/interconnect.h>
> +
> +#define icc_units_to_bps(bw)  ((bw) * 1000ULL)
> +
> +struct icc_node;
> +
> +/**
> + * struct icc_provider - interconnect provider (controller) entity that might
> + * provide multiple interconnect controls
> + *
> + * @provider_list: list of the registered interconnect providers
> + * @nodes: internal list of the interconnect provider nodes
> + * @set: pointer to device specific set operation function
> + * @aggregate: pointer to device specific aggregate operation function
> + * @dev: the device this interconnect provider belongs to
> + * @users: count of active users
> + * @avg_bw: aggregated value of average bandwidth requests from all nodes
> + * @peak_bw: aggregated value of peak bandwidth requests from all nodes
> + * @data: pointer to private data
> + */
> +struct icc_provider {
> +       struct list_head        provider_list;
> +       struct list_head        nodes;
> +       int (*set)(struct icc_node *src, struct icc_node *dst,
> +                  u32 avg_bw, u32 peak_bw);
> +       int (*aggregate)(struct icc_node *node, u32 avg_bw, u32 peak_bw,
> +                        u32 *agg_avg, u32 *agg_peak);
> +       struct device           *dev;
> +       int                     users;
> +       u32                     avg_bw;
> +       u32                     peak_bw;
> +       void                    *data;
> +};
> +
> +/**
> + * struct icc_node - entity that is part of the interconnect topology
> + *
> + * @id: platform specific node id
> + * @name: node name used in debugfs
> + * @links: a list of targets pointing to where we can go next when traversing
> + * @num_links: number of links to other interconnect nodes
> + * @provider: points to the interconnect provider of this node
> + * @node_list: the list entry in the parent provider's "nodes" list
> + * @search_list: list used when walking the nodes graph
> + * @reverse: pointer to previous node when walking the nodes graph
> + * @is_traversed: flag that is used when walking the nodes graph
> + * @req_list: a list of QoS constraint requests associated with this node
> + * @avg_bw: aggregated value of average bandwidth requests from all consumers
> + * @peak_bw: aggregated value of peak bandwidth requests from all consumers
> + * @data: pointer to private data
> + */
> +struct icc_node {
> +       int                     id;
> +       const char              *name;
> +       struct icc_node         **links;
> +       size_t                  num_links;
> +
> +       struct icc_provider     *provider;
> +       struct list_head        node_list;
> +       struct list_head        orphan_list;

Is this used?

> +       struct list_head        search_list;
> +       struct icc_node         *reverse;
> +       bool                    is_traversed;
> +       struct hlist_head       req_list;
> +       u32                     avg_bw;
> +       u32                     peak_bw;
> +       void                    *data;
> +};
> +
> +#if IS_ENABLED(CONFIG_INTERCONNECT)
> +
> +struct icc_node *icc_node_create(int id);
> +void icc_node_remove(int id);
> +int icc_link_create(struct icc_node *node, const int dst_id);
> +int icc_link_remove(struct icc_node *src, struct icc_node *dst);
> +int icc_node_add(struct icc_node *node, struct icc_provider *provider);
> +int icc_provider_add(struct icc_provider *provider);
> +int icc_provider_del(struct icc_provider *provider);
> +
> +#else
> +
> +static inline struct icc_node *icc_node_create(int id)
> +{
> +       return ERR_PTR(-ENOTSUPP);
> +}
> +
> +void icc_node_remove(int id)
> +{
> +}
> +
> +static inline int icc_link_create(struct icc_node *node, const int dst_id)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +int icc_link_remove(struct icc_node *src, struct icc_node *dst)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +int icc_node_add(struct icc_node *node, struct icc_provider *provider)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +static inline int icc_provider_add(struct icc_provider *provider)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +static inline int icc_provider_del(struct icc_provider *provider)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +#endif /* CONFIG_INTERCONNECT */
> +
> +#endif /* _LINUX_INTERCONNECT_PROVIDER_H */
> diff --git a/include/linux/interconnect.h b/include/linux/interconnect.h
> new file mode 100644
> index 000000000000..593215371fd6
> --- /dev/null
> +++ b/include/linux/interconnect.h
> @@ -0,0 +1,42 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Copyright (c) 2018, Linaro Ltd.
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#ifndef _LINUX_INTERCONNECT_H
> +#define _LINUX_INTERCONNECT_H
> +
> +#include <linux/mutex.h>
> +#include <linux/types.h>
> +
> +struct icc_path;
> +struct device;
> +
> +#if IS_ENABLED(CONFIG_INTERCONNECT)
> +
> +struct icc_path *icc_get(struct device *dev, const int src_id,
> +                        const int dst_id);
> +void icc_put(struct icc_path *path);
> +int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw);
> +
> +#else
> +
> +static inline struct icc_path *icc_get(struct device *dev, const int src_id,
> +                                      const int dst_id)
> +{
> +       return NULL;
> +}
> +
> +static inline void icc_put(struct icc_path *path)
> +{
> +}
> +
> +static inline int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
> +{
> +       return 0;

I was originally going to suggest that this should return a failure.
Then I talked myself out of it, saying that if the interconnect
framework is not compiled in, then clients should assume all their bus
needs are already met. I guess this is the correct assumption?

> +}
> +
> +#endif /* CONFIG_INTERCONNECT */
> +
> +#endif /* _LINUX_INTERCONNECT_H */

Re: [PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Evan Green]

Hi Georgi. Thanks for the new spin of this.

On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> This patch introduce a new API to get requirements and configure the
> interconnect buses across the entire chipset to fit with the current
> demand.
>
> The API is using a consumer/provider-based model, where the providers are
> the interconnect buses and the consumers could be various drivers.
> The consumers request interconnect resources (path) between endpoints and
> set the desired constraints on this data flow path. The providers receive
> requests from consumers and aggregate these requests for all master-slave
> pairs on that path. Then the providers configure each participating in the
> topology node according to the requested data flow path, physical links and
> constraints. The topology could be complicated and multi-tiered and is SoC
> specific.
>
> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
> ---
>  Documentation/interconnect/interconnect.rst |  96 ++++
>  drivers/Kconfig                             |   2 +
>  drivers/Makefile                            |   1 +
>  drivers/interconnect/Kconfig                |  10 +
>  drivers/interconnect/Makefile               |   2 +
>  drivers/interconnect/core.c                 | 586 ++++++++++++++++++++
>  include/linux/interconnect-provider.h       | 127 +++++
>  include/linux/interconnect.h                |  42 ++
>  8 files changed, 866 insertions(+)
>  create mode 100644 Documentation/interconnect/interconnect.rst
>  create mode 100644 drivers/interconnect/Kconfig
>  create mode 100644 drivers/interconnect/Makefile
>  create mode 100644 drivers/interconnect/core.c
>  create mode 100644 include/linux/interconnect-provider.h
>  create mode 100644 include/linux/interconnect.h
>
> diff --git a/Documentation/interconnect/interconnect.rst b/Documentation/interconnect/interconnect.rst
> new file mode 100644
> index 000000000000..a1ebd83ad0a1
> --- /dev/null
> +++ b/Documentation/interconnect/interconnect.rst
> @@ -0,0 +1,96 @@
> +.. SPDX-License-Identifier: GPL-2.0
> +
> +=====================================
> +GENERIC SYSTEM INTERCONNECT SUBSYSTEM
> +=====================================
> +
> +Introduction
> +------------
> +
> +This framework is designed to provide a standard kernel interface to control
> +the settings of the interconnects on a SoC. These settings can be throughput,
> +latency and priority between multiple interconnected devices or functional
> +blocks. This can be controlled dynamically in order to save power or provide
> +maximum performance.
> +
> +The interconnect bus is a hardware with configurable parameters, which can be
> +set on a data path according to the requests received from various drivers.
> +An example of interconnect buses are the interconnects between various
> +components or functional blocks in chipsets. There can be multiple interconnects
> +on a SoC that can be multi-tiered.
> +
> +Below is a simplified diagram of a real-world SoC interconnect bus topology.
> +
> +::
> +
> + +----------------+    +----------------+
> + | HW Accelerator |--->|      M NoC     |<---------------+
> + +----------------+    +----------------+                |
> +                         |      |                    +------------+
> +  +-----+  +-------------+      V       +------+     |            |
> +  | DDR |  |                +--------+  | PCIe |     |            |
> +  +-----+  |                | Slaves |  +------+     |            |
> +    ^ ^    |                +--------+     |         |   C NoC    |
> +    | |    V                               V         |            |
> + +------------------+   +------------------------+   |            |   +-----+
> + |                  |-->|                        |-->|            |-->| CPU |
> + |                  |-->|                        |<--|            |   +-----+
> + |     Mem NoC      |   |         S NoC          |   +------------+
> + |                  |<--|                        |---------+    |
> + |                  |<--|                        |<------+ |    |   +--------+
> + +------------------+   +------------------------+       | |    +-->| Slaves |
> +   ^  ^    ^    ^          ^                             | |        +--------+
> +   |  |    |    |          |                             | V
> + +------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
> + | CPUs |  |  | GPU |   | DSP |  | Masters |-->|       P NoC    |-->| Slaves |
> + +------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
> +           |
> +       +-------+
> +       | Modem |
> +       +-------+
> +
> +Terminology
> +-----------
> +
> +Interconnect provider is the software definition of the interconnect hardware.
> +The interconnect providers on the above diagram are M NoC, S NoC, C NoC, P NoC
> +and Mem NoC.
> +
> +Interconnect node is the software definition of the interconnect hardware
> +port. Each interconnect provider consists of multiple interconnect nodes,
> +which are connected to other SoC components including other interconnect
> +providers. The point on the diagram where the CPUs connects to the memory is
> +called an interconnect node, which belongs to the Mem NoC interconnect provider.
> +
> +Interconnect endpoints are the first or the last element of the path. Every
> +endpoint is a node, but not every node is an endpoint.
> +
> +Interconnect path is everything between two endpoints including all the nodes
> +that have to be traversed to reach from a source to destination node. It may
> +include multiple master-slave pairs across several interconnect providers.
> +
> +Interconnect consumers are the entities which make use of the data paths exposed
> +by the providers. The consumers send requests to providers requesting various
> +throughput, latency and priority. Usually the consumers are device drivers, that
> +send request based on their needs. An example for a consumer is a video decoder
> +that supports various formats and image sizes.
> +
> +Interconnect providers
> +----------------------
> +
> +Interconnect provider is an entity that implements methods to initialize and
> +configure a interconnect bus hardware. The interconnect provider drivers should
> +be registered with the interconnect provider core.
> +
> +The interconnect framework provider API functions are documented in
> +.. kernel-doc:: include/linux/interconnect-provider.h
> +
> +Interconnect consumers
> +----------------------
> +
> +Interconnect consumers are the clients which use the interconnect APIs to
> +get paths between endpoints and set their bandwidth/latency/QoS requirements
> +for these interconnect paths.
> +
> +The interconnect framework consumer API functions are documented in
> +.. kernel-doc:: include/linux/interconnect.h
> diff --git a/drivers/Kconfig b/drivers/Kconfig
> index 95b9ccc08165..3ed6ede9d021 100644
> --- a/drivers/Kconfig
> +++ b/drivers/Kconfig
> @@ -217,4 +217,6 @@ source "drivers/siox/Kconfig"
>
>  source "drivers/slimbus/Kconfig"
>
> +source "drivers/interconnect/Kconfig"
> +
>  endmenu
> diff --git a/drivers/Makefile b/drivers/Makefile
> index 24cd47014657..0cca95740d9b 100644
> --- a/drivers/Makefile
> +++ b/drivers/Makefile
> @@ -185,3 +185,4 @@ obj-$(CONFIG_TEE)           += tee/
>  obj-$(CONFIG_MULTIPLEXER)      += mux/
>  obj-$(CONFIG_UNISYS_VISORBUS)  += visorbus/
>  obj-$(CONFIG_SIOX)             += siox/
> +obj-$(CONFIG_INTERCONNECT)     += interconnect/
> diff --git a/drivers/interconnect/Kconfig b/drivers/interconnect/Kconfig
> new file mode 100644
> index 000000000000..a261c7d41deb
> --- /dev/null
> +++ b/drivers/interconnect/Kconfig
> @@ -0,0 +1,10 @@
> +menuconfig INTERCONNECT
> +       tristate "On-Chip Interconnect management support"
> +       help
> +         Support for management of the on-chip interconnects.
> +
> +         This framework is designed to provide a generic interface for
> +         managing the interconnects in a SoC.
> +
> +         If unsure, say no.
> +
> diff --git a/drivers/interconnect/Makefile b/drivers/interconnect/Makefile
> new file mode 100644
> index 000000000000..97fca2e09d24
> --- /dev/null
> +++ b/drivers/interconnect/Makefile
> @@ -0,0 +1,2 @@
> +# SPDX-License-Identifier: GPL-2.0
> +obj-$(CONFIG_INTERCONNECT)             += core.o
> diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
> new file mode 100644
> index 000000000000..e7f96fc6722e
> --- /dev/null
> +++ b/drivers/interconnect/core.c
> @@ -0,0 +1,586 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Interconnect framework core driver
> + *
> + * Copyright (c) 2018, Linaro Ltd.
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#include <linux/device.h>
> +#include <linux/idr.h>
> +#include <linux/init.h>
> +#include <linux/interconnect.h>
> +#include <linux/interconnect-provider.h>
> +#include <linux/list.h>
> +#include <linux/module.h>
> +#include <linux/mutex.h>
> +#include <linux/slab.h>
> +
> +static DEFINE_IDR(icc_idr);
> +static LIST_HEAD(icc_provider_list);
> +static DEFINE_MUTEX(icc_lock);
> +
> +/**
> + * struct icc_req - constraints that are attached to each node
> + *
> + * @req_node: entry in list of requests for the particular @node
> + * @node: the interconnect node to which this constraint applies
> + * @dev: reference to the device that sets the constraints
> + * @avg_bw: an integer describing the average bandwidth in kbps
> + * @peak_bw: an integer describing the peak bandwidth in kbps
> + */
> +struct icc_req {
> +       struct hlist_node req_node;
> +       struct icc_node *node;
> +       struct device *dev;
> +       u32 avg_bw;
> +       u32 peak_bw;
> +};
> +
> +/**
> + * struct icc_path - interconnect path structure
> + * @num_nodes: number of hops (nodes)
> + * @reqs: array of the requests applicable to this path of nodes
> + */
> +struct icc_path {
> +       size_t num_nodes;
> +       struct icc_req reqs[0];
> +};
> +
> +static struct icc_node *node_find(const int id)
> +{
> +       struct icc_node *node;
> +
> +       mutex_lock(&icc_lock);
> +       node = idr_find(&icc_idr, id);
> +       mutex_unlock(&icc_lock);

I wonder if this is too low of a level to be dealing with the lock. I
notice that everywhere you use this function, you afterwards
immediately grab the lock and do more stuff. Maybe this function
should have a comment saying it assumes the lock is already held, and
then you can grab the lock in the callers, since you're doing that
anyway.

> +
> +       return node;
> +}
> +
> +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
> +{
> +       struct icc_node *node = dst;
> +       struct icc_path *path;
> +       size_t i;
> +
> +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
> +                      GFP_KERNEL);
> +       if (!path)
> +               return ERR_PTR(-ENOMEM);
> +
> +       path->num_nodes = num_nodes;
> +
> +       for (i = 0; i < num_nodes; i++) {
> +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
> +
> +               path->reqs[i].node = node;
> +               /* reference to previous node was saved during path traversal */
> +               node = node->reverse;
> +       }
> +
> +       return path;
> +}
> +
> +static struct icc_path *path_find(struct device *dev, struct icc_node *src,
> +                                 struct icc_node *dst)
> +{

I personally prefer a comment somewhere indicating that this function
assumes icc_lock is already held. Not sure if that's conventional or
not.

> +       struct icc_node *n, *node = NULL;
> +       struct icc_provider *provider;
> +       struct list_head traverse_list;
> +       struct list_head edge_list;
> +       struct list_head visited_list;
> +       size_t i, depth = 0;
> +       bool found = false;
> +       int ret = -EPROBE_DEFER;
> +
> +       INIT_LIST_HEAD(&traverse_list);
> +       INIT_LIST_HEAD(&edge_list);
> +       INIT_LIST_HEAD(&visited_list);
> +
> +       list_add_tail(&src->search_list, &traverse_list);
> +       src->reverse = NULL;
> +
> +       do {
> +               list_for_each_entry_safe(node, n, &traverse_list, search_list) {
> +                       if (node == dst) {
> +                               found = true;
> +                               list_add(&node->search_list, &visited_list);
> +                               break;
> +                       }
> +                       for (i = 0; i < node->num_links; i++) {
> +                               struct icc_node *tmp = node->links[i];
> +
> +                               if (!tmp) {
> +                                       ret = -ENOENT;
> +                                       goto out;
> +                               }
> +
> +                               if (tmp->is_traversed)
> +                                       continue;
> +
> +                               tmp->is_traversed = true;
> +                               tmp->reverse = node;
> +                               list_add(&tmp->search_list, &edge_list);
> +                       }
> +               }
> +               if (found)
> +                       break;
> +
> +               list_splice_init(&traverse_list, &visited_list);
> +               list_splice_init(&edge_list, &traverse_list);
> +
> +               /* count the hops away from the source */
> +               depth++;
> +
> +       } while (!list_empty(&traverse_list));
> +
> +out:
> +       /* reset the traversed state */
> +       list_for_each_entry(provider, &icc_provider_list, provider_list) {
> +               list_for_each_entry(n, &provider->nodes, node_list)
> +                       if (n->is_traversed)
> +                               n->is_traversed = false;

Remove the conditional, just set is_traversed to false.

> +       }
> +
> +       if (found) {
> +               struct icc_path *path = path_allocate(dst, depth);

Is the path supposed to include the source? For instance, if the dst
were a neighbor, depth would be one, so only dst would be in the path.
It seems like it might be worthwhile to have the source in there too.

> +
> +               if (IS_ERR(path))
> +                       return path;
> +
> +               /* initialize the path */
> +               for (i = 0; i < path->num_nodes; i++) {
> +                       node = path->reqs[i].node;
> +                       path->reqs[i].dev = dev;
> +                       node->provider->users++;

Should this loop live inside path_allocate? I'm unsure, but maybe at
least path->reqs[i].dev = dev, since it feels like standard
initialization of the path.

> +               }
> +               return path;
> +       }
> +
> +       return ERR_PTR(ret);
> +}
> +
> +/*
> + * We want the path to honor all bandwidth requests, so the average
> + * bandwidth requirements from each consumer are aggregated at each node
> + * and provider level. By default the average bandwidth is the sum of all
> + * averages and the peak will be the highest of all peak bandwidth requests.
> + */
> +
> +static int aggregate_requests(struct icc_node *node)
> +{
> +       struct icc_provider *p = node->provider;
> +       struct icc_req *r;
> +
> +       node->avg_bw = 0;
> +       node->peak_bw = 0;
> +
> +       hlist_for_each_entry(r, &node->req_list, req_node)
> +               p->aggregate(node, r->avg_bw, r->peak_bw,
> +                            &node->avg_bw, &node->peak_bw);
> +
> +       return 0;
> +}
> +
> +static void aggregate_provider(struct icc_provider *p)
> +{
> +       struct icc_node *n;
> +
> +       p->avg_bw = 0;
> +       p->peak_bw = 0;
> +
> +       list_for_each_entry(n, &p->nodes, node_list)
> +               p->aggregate(n, n->avg_bw, n->peak_bw,
> +                            &p->avg_bw, &p->peak_bw);
> +}
> +
> +static int apply_constraints(struct icc_path *path)
> +{
> +       struct icc_node *next, *prev = NULL;
> +       int ret = 0;
> +       int i;
> +
> +       for (i = 0; i < path->num_nodes; i++, prev = next) {
> +               struct icc_provider *p;
> +
> +               next = path->reqs[i].node;
> +               /*
> +                * Both endpoints should be valid master-slave pairs of the
> +                * same interconnect provider that will be configured.
> +                */
> +               if (!prev || next->provider != prev->provider)
> +                       continue;
> +
> +               p = next->provider;
> +
> +               aggregate_provider(p);
> +
> +               if (p->set) {
> +                       /* set the constraints */
> +                       ret = p->set(prev, next, p->avg_bw, p->peak_bw);
> +               }
> +
> +               if (ret)
> +                       goto out;
> +       }
> +out:
> +       return ret;
> +}
> +
> +/**
> + * icc_set() - set constraints on an interconnect path between two endpoints
> + * @path: reference to the path returned by icc_get()
> + * @avg_bw: average bandwidth in kbps
> + * @peak_bw: peak bandwidth in kbps
> + *
> + * This function is used by an interconnect consumer to express its own needs
> + * in terms of bandwidth for a previously requested path between two endpoints.
> + * The requests are aggregated and each node is updated accordingly. The entire
> + * path is locked by a mutex to ensure that the set() is completed.
> + * The @path can be NULL when the "interconnects" DT properties is missing,
> + * which will mean that no constraints will be set.
> + *
> + * Returns 0 on success, or an appropriate error code otherwise.
> + */
> +int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
> +{
> +       struct icc_node *node;
> +       struct icc_provider *p;
> +       size_t i;
> +       int ret = 0;
> +
> +       if (!path)
> +               return 0;
> +
> +       mutex_lock(&icc_lock);
> +
> +       for (i = 0; i < path->num_nodes; i++) {
> +               node = path->reqs[i].node;
> +               p = node->provider;
> +
> +               /* update the consumer request for this path */
> +               path->reqs[i].avg_bw = avg_bw;
> +               path->reqs[i].peak_bw = peak_bw;
> +
> +               /* aggregate requests for this node */
> +               aggregate_requests(node);
> +       }
> +
> +       ret = apply_constraints(path);
> +       if (ret)
> +               pr_err("interconnect: error applying constraints (%d)", ret);
> +
> +       mutex_unlock(&icc_lock);
> +
> +       return ret;
> +}
> +EXPORT_SYMBOL_GPL(icc_set);
> +
> +/**
> + * icc_get() - return a handle for path between two endpoints
> + * @dev: the device requesting the path
> + * @src_id: source device port id
> + * @dst_id: destination device port id
> + *
> + * This function will search for a path between two endpoints and return an
> + * icc_path handle on success. Use icc_put() to release
> + * constraints when the they are not needed anymore.
> + *
> + * Return: icc_path pointer on success, or ERR_PTR() on error
> + */
> +struct icc_path *icc_get(struct device *dev, const int src_id, const int dst_id)
> +{
> +       struct icc_node *src, *dst;
> +       struct icc_path *path = ERR_PTR(-EPROBE_DEFER);
> +
> +       src = node_find(src_id);
> +       if (!src) {
> +               dev_err(dev, "%s: invalid src=%d\n", __func__, src_id);
> +               goto out;
> +       }
> +
> +       dst = node_find(dst_id);
> +       if (!dst) {
> +               dev_err(dev, "%s: invalid dst=%d\n", __func__, dst_id);
> +               goto out;
> +       }
> +
> +       mutex_lock(&icc_lock);
> +       path = path_find(dev, src, dst);
> +       mutex_unlock(&icc_lock);
> +       if (IS_ERR(path)) {
> +               dev_err(dev, "%s: invalid path=%ld\n", __func__, PTR_ERR(path));
> +               goto out;
> +       }
> +
> +out:
> +       return path;
> +}
> +EXPORT_SYMBOL_GPL(icc_get);
> +
> +/**
> + * icc_put() - release the reference to the icc_path
> + * @path: interconnect path
> + *
> + * Use this function to release the constraints on a path when the path is
> + * no longer needed. The constraints will be re-aggregated.
> + */
> +void icc_put(struct icc_path *path)
> +{
> +       struct icc_node *node;
> +       size_t i;
> +       int ret;
> +
> +       if (!path || WARN_ON_ONCE(IS_ERR(path)))

Why only once?

> +               return;
> +
> +       ret = icc_set(path, 0, 0);
> +       if (ret)
> +               pr_err("%s: error (%d)\n", __func__, ret);
> +
> +       mutex_lock(&icc_lock);
> +       for (i = 0; i < path->num_nodes; i++) {
> +               node = path->reqs[i].node;
> +               hlist_del(&path->reqs[i].req_node);
> +
> +               node->provider->users--;
> +       }
> +       mutex_unlock(&icc_lock);
> +
> +       kfree(path);
> +}
> +EXPORT_SYMBOL_GPL(icc_put);
> +
> +/**
> + * icc_node_create() - create a node
> + * @id: node id
> + *
> + * Return: icc_node pointer on success, or ERR_PTR() on error
> + */
> +struct icc_node *icc_node_create(int id)
> +{
> +       struct icc_node *node;
> +
> +       /* check if node already exists */
> +       node = node_find(id);
> +       if (node)
> +               goto out;
> +
> +       node = kzalloc(sizeof(*node), GFP_KERNEL);
> +       if (!node) {
> +               node = ERR_PTR(-ENOMEM);
> +               goto out;
> +       }
> +
> +       mutex_lock(&icc_lock);
> +
> +       id = idr_alloc(&icc_idr, node, id, id + 1, GFP_KERNEL);
> +       if (WARN(id < 0, "couldn't get idr")) {
> +               node = ERR_PTR(id);
> +               goto out;
> +       }
> +
> +       node->id = id;
> +
> +out:
> +       mutex_unlock(&icc_lock);
> +
> +       return node;
> +}
> +EXPORT_SYMBOL_GPL(icc_node_create);
> +
> +/**
> + * icc_node_remove() - remove a node
> + * @id: node id
> + *
> + */
> +void icc_node_remove(int id)
> +{
> +       struct icc_node *node;
> +
> +       node = node_find(id);
> +       if (node) {
> +               mutex_lock(&icc_lock);
> +               idr_remove(&icc_idr, node->id);

Should we throw a warning if there are any paths that go through this
node (ie req_list is non-empty)?

> +               mutex_unlock(&icc_lock);
> +       }
> +
> +       kfree(node);
> +}
> +EXPORT_SYMBOL_GPL(icc_node_remove);
> +
> +/**
> + * icc_link_create() - create a link between two nodes
> + * @src_id: source node id
> + * @dst_id: destination node id
> + *
> + * Create a link between two nodes. The nodes might belong to different
> + * interconnect providers and the @dst_id node might not exist (if the
> + * provider driver has not probed yet). So just create the @dst_id node
> + * and when the actual provider driver is probed, the rest of the node
> + * data is filled.
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_link_create(struct icc_node *node, const int dst_id)
> +{
> +       struct icc_node *dst;
> +       struct icc_node **new;
> +       int ret = 0;
> +
> +       if (!node->provider)
> +               return -EINVAL;
> +
> +       dst = node_find(dst_id);
> +       if (!dst) {
> +               dst = icc_node_create(dst_id);
> +
> +               if (IS_ERR(dst)) {
> +                       ret = PTR_ERR(dst);
> +                       goto out;
> +               }
> +       }
> +
> +       mutex_lock(&icc_lock);
> +
> +       new = krealloc(node->links,
> +                      (node->num_links + 1) * sizeof(*node->links),
> +                      GFP_KERNEL);
> +       if (!new) {
> +               ret = -ENOMEM;
> +               goto out;
> +       }
> +
> +       node->links = new;
> +       node->links[node->num_links++] = dst;
> +
> +out:
> +       mutex_unlock(&icc_lock);
> +
> +       return ret;
> +}
> +EXPORT_SYMBOL_GPL(icc_link_create);
> +
> +/**
> + * icc_link_remove() - remove a link between two nodes
> + * @src: pointer to source node
> + * @dst: pointer to destination node
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_link_remove(struct icc_node *src, struct icc_node *dst)
> +{
> +       struct icc_node **new;
> +       int ret = 0;
> +       int i, j;
> +
> +       if (IS_ERR_OR_NULL(src))
> +               return PTR_ERR(src);
> +
> +       if (IS_ERR_OR_NULL(dst))
> +               return PTR_ERR(dst);

I wonder if we should return a fixed error in these cases like
-EINVAL, rather than handing through whatever crazy value is in
src/dst.

> +
> +       mutex_lock(&icc_lock);
> +
> +       new = krealloc(src->links,
> +                      (src->num_links - 1) * sizeof(*src->links),
> +                      GFP_KERNEL);
> +       if (!new) {
> +               ret = -ENOMEM;
> +               goto out;
> +       }
> +
> +       for (i = 0, j = 0; j < src->num_links; j++) {
> +               if (src->links[j] == dst)
> +                       continue;
> +
> +               new[i++] = src->links[j];
> +       }
> +
> +       src->links = new;
> +       src->num_links--;

My understanding is that once you call realloc and it succeeds, you
must assume your old memory is gone and your new memory is only as big
as the new size you request. So you shouldn't call krealloc until
you've fixed the array up. Is the order of the links array important?
If not, you could just take the element at the end and stick it in the
slot that's being deleted. Then decrease the size and do your realloc.

> +
> +out:
> +       mutex_unlock(&icc_lock);
> +
> +       return ret;
> +}
> +EXPORT_SYMBOL_GPL(icc_link_remove);
> +
> +/**
> + * icc_node_add() - add an interconnect node to interconnect provider
> + * @node: pointer to the interconnect node
> + * @provider: pointer to the interconnect provider
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_node_add(struct icc_node *node, struct icc_provider *provider)
> +{
> +       mutex_lock(&icc_lock);
> +
> +       node->provider = provider;
> +       list_add(&node->node_list, &provider->nodes);
> +
> +       mutex_unlock(&icc_lock);
> +
> +       return 0;
> +}
> +EXPORT_SYMBOL_GPL(icc_node_add);
> +
> +/**
> + * icc_provider_add() - add a new interconnect provider
> + * @icc_provider: the interconnect provider that will be added into topology
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_provider_add(struct icc_provider *provider)
> +{
> +       if (WARN_ON(!provider->set))
> +               return -EINVAL;
> +
> +       mutex_init(&icc_lock);
> +
> +       INIT_LIST_HEAD(&provider->nodes);
> +       list_add(&provider->provider_list, &icc_provider_list);
> +
> +       mutex_unlock(&icc_lock);
> +
> +       dev_dbg(provider->dev, "interconnect provider added to topology\n");
> +
> +       return 0;
> +}
> +EXPORT_SYMBOL_GPL(icc_provider_add);
> +
> +/**
> + * icc_provider_del() - delete previously added interconnect provider
> + * @icc_provider: the interconnect provider that will be removed from topology
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_provider_del(struct icc_provider *provider)
> +{
> +       mutex_lock(&icc_lock);
> +       if (provider->users) {
> +               pr_warn("interconnect provider still has %d users\n",
> +                       provider->users);
> +               mutex_unlock(&icc_lock);
> +               return -EBUSY;
> +       }
> +
> +       if (!list_empty_careful(&provider->nodes)) {
> +               pr_warn("interconnect provider still has nodes\n");
> +               mutex_unlock(&icc_lock);
> +               return -EEXIST;

How come you're returning different error codes for these two cases?
The error in both cases is effectively "you failed to clean up after
yourself", so maybe EBUSY makes sense for both of them. The pr_warn
helps to differentiate between the two for debugging.

> +       }
> +
> +       list_del(&provider->provider_list);
> +       mutex_unlock(&icc_lock);
> +
> +       return 0;
> +}
> +EXPORT_SYMBOL_GPL(icc_provider_del);
> +
> +MODULE_AUTHOR("Georgi Djakov <georgi.djakov@linaro.org");
> +MODULE_DESCRIPTION("Interconnect Driver Core");
> +MODULE_LICENSE("GPL v2");
> diff --git a/include/linux/interconnect-provider.h b/include/linux/interconnect-provider.h
> new file mode 100644
> index 000000000000..f4613c6dce4f
> --- /dev/null
> +++ b/include/linux/interconnect-provider.h
> @@ -0,0 +1,127 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Copyright (c) 2018, Linaro Ltd.
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#ifndef _LINUX_INTERCONNECT_PROVIDER_H
> +#define _LINUX_INTERCONNECT_PROVIDER_H
> +
> +#include <linux/interconnect.h>
> +
> +#define icc_units_to_bps(bw)  ((bw) * 1000ULL)
> +
> +struct icc_node;
> +
> +/**
> + * struct icc_provider - interconnect provider (controller) entity that might
> + * provide multiple interconnect controls
> + *
> + * @provider_list: list of the registered interconnect providers
> + * @nodes: internal list of the interconnect provider nodes
> + * @set: pointer to device specific set operation function
> + * @aggregate: pointer to device specific aggregate operation function
> + * @dev: the device this interconnect provider belongs to
> + * @users: count of active users
> + * @avg_bw: aggregated value of average bandwidth requests from all nodes
> + * @peak_bw: aggregated value of peak bandwidth requests from all nodes
> + * @data: pointer to private data
> + */
> +struct icc_provider {
> +       struct list_head        provider_list;
> +       struct list_head        nodes;
> +       int (*set)(struct icc_node *src, struct icc_node *dst,
> +                  u32 avg_bw, u32 peak_bw);
> +       int (*aggregate)(struct icc_node *node, u32 avg_bw, u32 peak_bw,
> +                        u32 *agg_avg, u32 *agg_peak);
> +       struct device           *dev;
> +       int                     users;
> +       u32                     avg_bw;
> +       u32                     peak_bw;
> +       void                    *data;
> +};
> +
> +/**
> + * struct icc_node - entity that is part of the interconnect topology
> + *
> + * @id: platform specific node id
> + * @name: node name used in debugfs
> + * @links: a list of targets pointing to where we can go next when traversing
> + * @num_links: number of links to other interconnect nodes
> + * @provider: points to the interconnect provider of this node
> + * @node_list: the list entry in the parent provider's "nodes" list
> + * @search_list: list used when walking the nodes graph
> + * @reverse: pointer to previous node when walking the nodes graph
> + * @is_traversed: flag that is used when walking the nodes graph
> + * @req_list: a list of QoS constraint requests associated with this node
> + * @avg_bw: aggregated value of average bandwidth requests from all consumers
> + * @peak_bw: aggregated value of peak bandwidth requests from all consumers
> + * @data: pointer to private data
> + */
> +struct icc_node {
> +       int                     id;
> +       const char              *name;
> +       struct icc_node         **links;
> +       size_t                  num_links;
> +
> +       struct icc_provider     *provider;
> +       struct list_head        node_list;
> +       struct list_head        orphan_list;

Is this used?

> +       struct list_head        search_list;
> +       struct icc_node         *reverse;
> +       bool                    is_traversed;
> +       struct hlist_head       req_list;
> +       u32                     avg_bw;
> +       u32                     peak_bw;
> +       void                    *data;
> +};
> +
> +#if IS_ENABLED(CONFIG_INTERCONNECT)
> +
> +struct icc_node *icc_node_create(int id);
> +void icc_node_remove(int id);
> +int icc_link_create(struct icc_node *node, const int dst_id);
> +int icc_link_remove(struct icc_node *src, struct icc_node *dst);
> +int icc_node_add(struct icc_node *node, struct icc_provider *provider);
> +int icc_provider_add(struct icc_provider *provider);
> +int icc_provider_del(struct icc_provider *provider);
> +
> +#else
> +
> +static inline struct icc_node *icc_node_create(int id)
> +{
> +       return ERR_PTR(-ENOTSUPP);
> +}
> +
> +void icc_node_remove(int id)
> +{
> +}
> +
> +static inline int icc_link_create(struct icc_node *node, const int dst_id)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +int icc_link_remove(struct icc_node *src, struct icc_node *dst)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +int icc_node_add(struct icc_node *node, struct icc_provider *provider)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +static inline int icc_provider_add(struct icc_provider *provider)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +static inline int icc_provider_del(struct icc_provider *provider)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +#endif /* CONFIG_INTERCONNECT */
> +
> +#endif /* _LINUX_INTERCONNECT_PROVIDER_H */
> diff --git a/include/linux/interconnect.h b/include/linux/interconnect.h
> new file mode 100644
> index 000000000000..593215371fd6
> --- /dev/null
> +++ b/include/linux/interconnect.h
> @@ -0,0 +1,42 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Copyright (c) 2018, Linaro Ltd.
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#ifndef _LINUX_INTERCONNECT_H
> +#define _LINUX_INTERCONNECT_H
> +
> +#include <linux/mutex.h>
> +#include <linux/types.h>
> +
> +struct icc_path;
> +struct device;
> +
> +#if IS_ENABLED(CONFIG_INTERCONNECT)
> +
> +struct icc_path *icc_get(struct device *dev, const int src_id,
> +                        const int dst_id);
> +void icc_put(struct icc_path *path);
> +int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw);
> +
> +#else
> +
> +static inline struct icc_path *icc_get(struct device *dev, const int src_id,
> +                                      const int dst_id)
> +{
> +       return NULL;
> +}
> +
> +static inline void icc_put(struct icc_path *path)
> +{
> +}
> +
> +static inline int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
> +{
> +       return 0;

I was originally going to suggest that this should return a failure.
Then I talked myself out of it, saying that if the interconnect
framework is not compiled in, then clients should assume all their bus
needs are already met. I guess this is the correct assumption?

> +}
> +
> +#endif /* CONFIG_INTERCONNECT */
> +
> +#endif /* _LINUX_INTERCONNECT_H */

[PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Evan Green]

Hi Georgi. Thanks for the new spin of this.

On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> This patch introduce a new API to get requirements and configure the
> interconnect buses across the entire chipset to fit with the current
> demand.
>
> The API is using a consumer/provider-based model, where the providers are
> the interconnect buses and the consumers could be various drivers.
> The consumers request interconnect resources (path) between endpoints and
> set the desired constraints on this data flow path. The providers receive
> requests from consumers and aggregate these requests for all master-slave
> pairs on that path. Then the providers configure each participating in the
> topology node according to the requested data flow path, physical links and
> constraints. The topology could be complicated and multi-tiered and is SoC
> specific.
>
> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
> ---
>  Documentation/interconnect/interconnect.rst |  96 ++++
>  drivers/Kconfig                             |   2 +
>  drivers/Makefile                            |   1 +
>  drivers/interconnect/Kconfig                |  10 +
>  drivers/interconnect/Makefile               |   2 +
>  drivers/interconnect/core.c                 | 586 ++++++++++++++++++++
>  include/linux/interconnect-provider.h       | 127 +++++
>  include/linux/interconnect.h                |  42 ++
>  8 files changed, 866 insertions(+)
>  create mode 100644 Documentation/interconnect/interconnect.rst
>  create mode 100644 drivers/interconnect/Kconfig
>  create mode 100644 drivers/interconnect/Makefile
>  create mode 100644 drivers/interconnect/core.c
>  create mode 100644 include/linux/interconnect-provider.h
>  create mode 100644 include/linux/interconnect.h
>
> diff --git a/Documentation/interconnect/interconnect.rst b/Documentation/interconnect/interconnect.rst
> new file mode 100644
> index 000000000000..a1ebd83ad0a1
> --- /dev/null
> +++ b/Documentation/interconnect/interconnect.rst
> @@ -0,0 +1,96 @@
> +.. SPDX-License-Identifier: GPL-2.0
> +
> +=====================================
> +GENERIC SYSTEM INTERCONNECT SUBSYSTEM
> +=====================================
> +
> +Introduction
> +------------
> +
> +This framework is designed to provide a standard kernel interface to control
> +the settings of the interconnects on a SoC. These settings can be throughput,
> +latency and priority between multiple interconnected devices or functional
> +blocks. This can be controlled dynamically in order to save power or provide
> +maximum performance.
> +
> +The interconnect bus is a hardware with configurable parameters, which can be
> +set on a data path according to the requests received from various drivers.
> +An example of interconnect buses are the interconnects between various
> +components or functional blocks in chipsets. There can be multiple interconnects
> +on a SoC that can be multi-tiered.
> +
> +Below is a simplified diagram of a real-world SoC interconnect bus topology.
> +
> +::
> +
> + +----------------+    +----------------+
> + | HW Accelerator |--->|      M NoC     |<---------------+
> + +----------------+    +----------------+                |
> +                         |      |                    +------------+
> +  +-----+  +-------------+      V       +------+     |            |
> +  | DDR |  |                +--------+  | PCIe |     |            |
> +  +-----+  |                | Slaves |  +------+     |            |
> +    ^ ^    |                +--------+     |         |   C NoC    |
> +    | |    V                               V         |            |
> + +------------------+   +------------------------+   |            |   +-----+
> + |                  |-->|                        |-->|            |-->| CPU |
> + |                  |-->|                        |<--|            |   +-----+
> + |     Mem NoC      |   |         S NoC          |   +------------+
> + |                  |<--|                        |---------+    |
> + |                  |<--|                        |<------+ |    |   +--------+
> + +------------------+   +------------------------+       | |    +-->| Slaves |
> +   ^  ^    ^    ^          ^                             | |        +--------+
> +   |  |    |    |          |                             | V
> + +------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
> + | CPUs |  |  | GPU |   | DSP |  | Masters |-->|       P NoC    |-->| Slaves |
> + +------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
> +           |
> +       +-------+
> +       | Modem |
> +       +-------+
> +
> +Terminology
> +-----------
> +
> +Interconnect provider is the software definition of the interconnect hardware.
> +The interconnect providers on the above diagram are M NoC, S NoC, C NoC, P NoC
> +and Mem NoC.
> +
> +Interconnect node is the software definition of the interconnect hardware
> +port. Each interconnect provider consists of multiple interconnect nodes,
> +which are connected to other SoC components including other interconnect
> +providers. The point on the diagram where the CPUs connects to the memory is
> +called an interconnect node, which belongs to the Mem NoC interconnect provider.
> +
> +Interconnect endpoints are the first or the last element of the path. Every
> +endpoint is a node, but not every node is an endpoint.
> +
> +Interconnect path is everything between two endpoints including all the nodes
> +that have to be traversed to reach from a source to destination node. It may
> +include multiple master-slave pairs across several interconnect providers.
> +
> +Interconnect consumers are the entities which make use of the data paths exposed
> +by the providers. The consumers send requests to providers requesting various
> +throughput, latency and priority. Usually the consumers are device drivers, that
> +send request based on their needs. An example for a consumer is a video decoder
> +that supports various formats and image sizes.
> +
> +Interconnect providers
> +----------------------
> +
> +Interconnect provider is an entity that implements methods to initialize and
> +configure a interconnect bus hardware. The interconnect provider drivers should
> +be registered with the interconnect provider core.
> +
> +The interconnect framework provider API functions are documented in
> +.. kernel-doc:: include/linux/interconnect-provider.h
> +
> +Interconnect consumers
> +----------------------
> +
> +Interconnect consumers are the clients which use the interconnect APIs to
> +get paths between endpoints and set their bandwidth/latency/QoS requirements
> +for these interconnect paths.
> +
> +The interconnect framework consumer API functions are documented in
> +.. kernel-doc:: include/linux/interconnect.h
> diff --git a/drivers/Kconfig b/drivers/Kconfig
> index 95b9ccc08165..3ed6ede9d021 100644
> --- a/drivers/Kconfig
> +++ b/drivers/Kconfig
> @@ -217,4 +217,6 @@ source "drivers/siox/Kconfig"
>
>  source "drivers/slimbus/Kconfig"
>
> +source "drivers/interconnect/Kconfig"
> +
>  endmenu
> diff --git a/drivers/Makefile b/drivers/Makefile
> index 24cd47014657..0cca95740d9b 100644
> --- a/drivers/Makefile
> +++ b/drivers/Makefile
> @@ -185,3 +185,4 @@ obj-$(CONFIG_TEE)           += tee/
>  obj-$(CONFIG_MULTIPLEXER)      += mux/
>  obj-$(CONFIG_UNISYS_VISORBUS)  += visorbus/
>  obj-$(CONFIG_SIOX)             += siox/
> +obj-$(CONFIG_INTERCONNECT)     += interconnect/
> diff --git a/drivers/interconnect/Kconfig b/drivers/interconnect/Kconfig
> new file mode 100644
> index 000000000000..a261c7d41deb
> --- /dev/null
> +++ b/drivers/interconnect/Kconfig
> @@ -0,0 +1,10 @@
> +menuconfig INTERCONNECT
> +       tristate "On-Chip Interconnect management support"
> +       help
> +         Support for management of the on-chip interconnects.
> +
> +         This framework is designed to provide a generic interface for
> +         managing the interconnects in a SoC.
> +
> +         If unsure, say no.
> +
> diff --git a/drivers/interconnect/Makefile b/drivers/interconnect/Makefile
> new file mode 100644
> index 000000000000..97fca2e09d24
> --- /dev/null
> +++ b/drivers/interconnect/Makefile
> @@ -0,0 +1,2 @@
> +# SPDX-License-Identifier: GPL-2.0
> +obj-$(CONFIG_INTERCONNECT)             += core.o
> diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
> new file mode 100644
> index 000000000000..e7f96fc6722e
> --- /dev/null
> +++ b/drivers/interconnect/core.c
> @@ -0,0 +1,586 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Interconnect framework core driver
> + *
> + * Copyright (c) 2018, Linaro Ltd.
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#include <linux/device.h>
> +#include <linux/idr.h>
> +#include <linux/init.h>
> +#include <linux/interconnect.h>
> +#include <linux/interconnect-provider.h>
> +#include <linux/list.h>
> +#include <linux/module.h>
> +#include <linux/mutex.h>
> +#include <linux/slab.h>
> +
> +static DEFINE_IDR(icc_idr);
> +static LIST_HEAD(icc_provider_list);
> +static DEFINE_MUTEX(icc_lock);
> +
> +/**
> + * struct icc_req - constraints that are attached to each node
> + *
> + * @req_node: entry in list of requests for the particular @node
> + * @node: the interconnect node to which this constraint applies
> + * @dev: reference to the device that sets the constraints
> + * @avg_bw: an integer describing the average bandwidth in kbps
> + * @peak_bw: an integer describing the peak bandwidth in kbps
> + */
> +struct icc_req {
> +       struct hlist_node req_node;
> +       struct icc_node *node;
> +       struct device *dev;
> +       u32 avg_bw;
> +       u32 peak_bw;
> +};
> +
> +/**
> + * struct icc_path - interconnect path structure
> + * @num_nodes: number of hops (nodes)
> + * @reqs: array of the requests applicable to this path of nodes
> + */
> +struct icc_path {
> +       size_t num_nodes;
> +       struct icc_req reqs[0];
> +};
> +
> +static struct icc_node *node_find(const int id)
> +{
> +       struct icc_node *node;
> +
> +       mutex_lock(&icc_lock);
> +       node = idr_find(&icc_idr, id);
> +       mutex_unlock(&icc_lock);

I wonder if this is too low of a level to be dealing with the lock. I
notice that everywhere you use this function, you afterwards
immediately grab the lock and do more stuff. Maybe this function
should have a comment saying it assumes the lock is already held, and
then you can grab the lock in the callers, since you're doing that
anyway.

> +
> +       return node;
> +}
> +
> +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
> +{
> +       struct icc_node *node = dst;
> +       struct icc_path *path;
> +       size_t i;
> +
> +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
> +                      GFP_KERNEL);
> +       if (!path)
> +               return ERR_PTR(-ENOMEM);
> +
> +       path->num_nodes = num_nodes;
> +
> +       for (i = 0; i < num_nodes; i++) {
> +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
> +
> +               path->reqs[i].node = node;
> +               /* reference to previous node was saved during path traversal */
> +               node = node->reverse;
> +       }
> +
> +       return path;
> +}
> +
> +static struct icc_path *path_find(struct device *dev, struct icc_node *src,
> +                                 struct icc_node *dst)
> +{

I personally prefer a comment somewhere indicating that this function
assumes icc_lock is already held. Not sure if that's conventional or
not.

> +       struct icc_node *n, *node = NULL;
> +       struct icc_provider *provider;
> +       struct list_head traverse_list;
> +       struct list_head edge_list;
> +       struct list_head visited_list;
> +       size_t i, depth = 0;
> +       bool found = false;
> +       int ret = -EPROBE_DEFER;
> +
> +       INIT_LIST_HEAD(&traverse_list);
> +       INIT_LIST_HEAD(&edge_list);
> +       INIT_LIST_HEAD(&visited_list);
> +
> +       list_add_tail(&src->search_list, &traverse_list);
> +       src->reverse = NULL;
> +
> +       do {
> +               list_for_each_entry_safe(node, n, &traverse_list, search_list) {
> +                       if (node == dst) {
> +                               found = true;
> +                               list_add(&node->search_list, &visited_list);
> +                               break;
> +                       }
> +                       for (i = 0; i < node->num_links; i++) {
> +                               struct icc_node *tmp = node->links[i];
> +
> +                               if (!tmp) {
> +                                       ret = -ENOENT;
> +                                       goto out;
> +                               }
> +
> +                               if (tmp->is_traversed)
> +                                       continue;
> +
> +                               tmp->is_traversed = true;
> +                               tmp->reverse = node;
> +                               list_add(&tmp->search_list, &edge_list);
> +                       }
> +               }
> +               if (found)
> +                       break;
> +
> +               list_splice_init(&traverse_list, &visited_list);
> +               list_splice_init(&edge_list, &traverse_list);
> +
> +               /* count the hops away from the source */
> +               depth++;
> +
> +       } while (!list_empty(&traverse_list));
> +
> +out:
> +       /* reset the traversed state */
> +       list_for_each_entry(provider, &icc_provider_list, provider_list) {
> +               list_for_each_entry(n, &provider->nodes, node_list)
> +                       if (n->is_traversed)
> +                               n->is_traversed = false;

Remove the conditional, just set is_traversed to false.

> +       }
> +
> +       if (found) {
> +               struct icc_path *path = path_allocate(dst, depth);

Is the path supposed to include the source? For instance, if the dst
were a neighbor, depth would be one, so only dst would be in the path.
It seems like it might be worthwhile to have the source in there too.

> +
> +               if (IS_ERR(path))
> +                       return path;
> +
> +               /* initialize the path */
> +               for (i = 0; i < path->num_nodes; i++) {
> +                       node = path->reqs[i].node;
> +                       path->reqs[i].dev = dev;
> +                       node->provider->users++;

Should this loop live inside path_allocate? I'm unsure, but maybe at
least path->reqs[i].dev = dev, since it feels like standard
initialization of the path.

> +               }
> +               return path;
> +       }
> +
> +       return ERR_PTR(ret);
> +}
> +
> +/*
> + * We want the path to honor all bandwidth requests, so the average
> + * bandwidth requirements from each consumer are aggregated at each node
> + * and provider level. By default the average bandwidth is the sum of all
> + * averages and the peak will be the highest of all peak bandwidth requests.
> + */
> +
> +static int aggregate_requests(struct icc_node *node)
> +{
> +       struct icc_provider *p = node->provider;
> +       struct icc_req *r;
> +
> +       node->avg_bw = 0;
> +       node->peak_bw = 0;
> +
> +       hlist_for_each_entry(r, &node->req_list, req_node)
> +               p->aggregate(node, r->avg_bw, r->peak_bw,
> +                            &node->avg_bw, &node->peak_bw);
> +
> +       return 0;
> +}
> +
> +static void aggregate_provider(struct icc_provider *p)
> +{
> +       struct icc_node *n;
> +
> +       p->avg_bw = 0;
> +       p->peak_bw = 0;
> +
> +       list_for_each_entry(n, &p->nodes, node_list)
> +               p->aggregate(n, n->avg_bw, n->peak_bw,
> +                            &p->avg_bw, &p->peak_bw);
> +}
> +
> +static int apply_constraints(struct icc_path *path)
> +{
> +       struct icc_node *next, *prev = NULL;
> +       int ret = 0;
> +       int i;
> +
> +       for (i = 0; i < path->num_nodes; i++, prev = next) {
> +               struct icc_provider *p;
> +
> +               next = path->reqs[i].node;
> +               /*
> +                * Both endpoints should be valid master-slave pairs of the
> +                * same interconnect provider that will be configured.
> +                */
> +               if (!prev || next->provider != prev->provider)
> +                       continue;
> +
> +               p = next->provider;
> +
> +               aggregate_provider(p);
> +
> +               if (p->set) {
> +                       /* set the constraints */
> +                       ret = p->set(prev, next, p->avg_bw, p->peak_bw);
> +               }
> +
> +               if (ret)
> +                       goto out;
> +       }
> +out:
> +       return ret;
> +}
> +
> +/**
> + * icc_set() - set constraints on an interconnect path between two endpoints
> + * @path: reference to the path returned by icc_get()
> + * @avg_bw: average bandwidth in kbps
> + * @peak_bw: peak bandwidth in kbps
> + *
> + * This function is used by an interconnect consumer to express its own needs
> + * in terms of bandwidth for a previously requested path between two endpoints.
> + * The requests are aggregated and each node is updated accordingly. The entire
> + * path is locked by a mutex to ensure that the set() is completed.
> + * The @path can be NULL when the "interconnects" DT properties is missing,
> + * which will mean that no constraints will be set.
> + *
> + * Returns 0 on success, or an appropriate error code otherwise.
> + */
> +int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
> +{
> +       struct icc_node *node;
> +       struct icc_provider *p;
> +       size_t i;
> +       int ret = 0;
> +
> +       if (!path)
> +               return 0;
> +
> +       mutex_lock(&icc_lock);
> +
> +       for (i = 0; i < path->num_nodes; i++) {
> +               node = path->reqs[i].node;
> +               p = node->provider;
> +
> +               /* update the consumer request for this path */
> +               path->reqs[i].avg_bw = avg_bw;
> +               path->reqs[i].peak_bw = peak_bw;
> +
> +               /* aggregate requests for this node */
> +               aggregate_requests(node);
> +       }
> +
> +       ret = apply_constraints(path);
> +       if (ret)
> +               pr_err("interconnect: error applying constraints (%d)", ret);
> +
> +       mutex_unlock(&icc_lock);
> +
> +       return ret;
> +}
> +EXPORT_SYMBOL_GPL(icc_set);
> +
> +/**
> + * icc_get() - return a handle for path between two endpoints
> + * @dev: the device requesting the path
> + * @src_id: source device port id
> + * @dst_id: destination device port id
> + *
> + * This function will search for a path between two endpoints and return an
> + * icc_path handle on success. Use icc_put() to release
> + * constraints when the they are not needed anymore.
> + *
> + * Return: icc_path pointer on success, or ERR_PTR() on error
> + */
> +struct icc_path *icc_get(struct device *dev, const int src_id, const int dst_id)
> +{
> +       struct icc_node *src, *dst;
> +       struct icc_path *path = ERR_PTR(-EPROBE_DEFER);
> +
> +       src = node_find(src_id);
> +       if (!src) {
> +               dev_err(dev, "%s: invalid src=%d\n", __func__, src_id);
> +               goto out;
> +       }
> +
> +       dst = node_find(dst_id);
> +       if (!dst) {
> +               dev_err(dev, "%s: invalid dst=%d\n", __func__, dst_id);
> +               goto out;
> +       }
> +
> +       mutex_lock(&icc_lock);
> +       path = path_find(dev, src, dst);
> +       mutex_unlock(&icc_lock);
> +       if (IS_ERR(path)) {
> +               dev_err(dev, "%s: invalid path=%ld\n", __func__, PTR_ERR(path));
> +               goto out;
> +       }
> +
> +out:
> +       return path;
> +}
> +EXPORT_SYMBOL_GPL(icc_get);
> +
> +/**
> + * icc_put() - release the reference to the icc_path
> + * @path: interconnect path
> + *
> + * Use this function to release the constraints on a path when the path is
> + * no longer needed. The constraints will be re-aggregated.
> + */
> +void icc_put(struct icc_path *path)
> +{
> +       struct icc_node *node;
> +       size_t i;
> +       int ret;
> +
> +       if (!path || WARN_ON_ONCE(IS_ERR(path)))

Why only once?

> +               return;
> +
> +       ret = icc_set(path, 0, 0);
> +       if (ret)
> +               pr_err("%s: error (%d)\n", __func__, ret);
> +
> +       mutex_lock(&icc_lock);
> +       for (i = 0; i < path->num_nodes; i++) {
> +               node = path->reqs[i].node;
> +               hlist_del(&path->reqs[i].req_node);
> +
> +               node->provider->users--;
> +       }
> +       mutex_unlock(&icc_lock);
> +
> +       kfree(path);
> +}
> +EXPORT_SYMBOL_GPL(icc_put);
> +
> +/**
> + * icc_node_create() - create a node
> + * @id: node id
> + *
> + * Return: icc_node pointer on success, or ERR_PTR() on error
> + */
> +struct icc_node *icc_node_create(int id)
> +{
> +       struct icc_node *node;
> +
> +       /* check if node already exists */
> +       node = node_find(id);
> +       if (node)
> +               goto out;
> +
> +       node = kzalloc(sizeof(*node), GFP_KERNEL);
> +       if (!node) {
> +               node = ERR_PTR(-ENOMEM);
> +               goto out;
> +       }
> +
> +       mutex_lock(&icc_lock);
> +
> +       id = idr_alloc(&icc_idr, node, id, id + 1, GFP_KERNEL);
> +       if (WARN(id < 0, "couldn't get idr")) {
> +               node = ERR_PTR(id);
> +               goto out;
> +       }
> +
> +       node->id = id;
> +
> +out:
> +       mutex_unlock(&icc_lock);
> +
> +       return node;
> +}
> +EXPORT_SYMBOL_GPL(icc_node_create);
> +
> +/**
> + * icc_node_remove() - remove a node
> + * @id: node id
> + *
> + */
> +void icc_node_remove(int id)
> +{
> +       struct icc_node *node;
> +
> +       node = node_find(id);
> +       if (node) {
> +               mutex_lock(&icc_lock);
> +               idr_remove(&icc_idr, node->id);

Should we throw a warning if there are any paths that go through this
node (ie req_list is non-empty)?

> +               mutex_unlock(&icc_lock);
> +       }
> +
> +       kfree(node);
> +}
> +EXPORT_SYMBOL_GPL(icc_node_remove);
> +
> +/**
> + * icc_link_create() - create a link between two nodes
> + * @src_id: source node id
> + * @dst_id: destination node id
> + *
> + * Create a link between two nodes. The nodes might belong to different
> + * interconnect providers and the @dst_id node might not exist (if the
> + * provider driver has not probed yet). So just create the @dst_id node
> + * and when the actual provider driver is probed, the rest of the node
> + * data is filled.
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_link_create(struct icc_node *node, const int dst_id)
> +{
> +       struct icc_node *dst;
> +       struct icc_node **new;
> +       int ret = 0;
> +
> +       if (!node->provider)
> +               return -EINVAL;
> +
> +       dst = node_find(dst_id);
> +       if (!dst) {
> +               dst = icc_node_create(dst_id);
> +
> +               if (IS_ERR(dst)) {
> +                       ret = PTR_ERR(dst);
> +                       goto out;
> +               }
> +       }
> +
> +       mutex_lock(&icc_lock);
> +
> +       new = krealloc(node->links,
> +                      (node->num_links + 1) * sizeof(*node->links),
> +                      GFP_KERNEL);
> +       if (!new) {
> +               ret = -ENOMEM;
> +               goto out;
> +       }
> +
> +       node->links = new;
> +       node->links[node->num_links++] = dst;
> +
> +out:
> +       mutex_unlock(&icc_lock);
> +
> +       return ret;
> +}
> +EXPORT_SYMBOL_GPL(icc_link_create);
> +
> +/**
> + * icc_link_remove() - remove a link between two nodes
> + * @src: pointer to source node
> + * @dst: pointer to destination node
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_link_remove(struct icc_node *src, struct icc_node *dst)
> +{
> +       struct icc_node **new;
> +       int ret = 0;
> +       int i, j;
> +
> +       if (IS_ERR_OR_NULL(src))
> +               return PTR_ERR(src);
> +
> +       if (IS_ERR_OR_NULL(dst))
> +               return PTR_ERR(dst);

I wonder if we should return a fixed error in these cases like
-EINVAL, rather than handing through whatever crazy value is in
src/dst.

> +
> +       mutex_lock(&icc_lock);
> +
> +       new = krealloc(src->links,
> +                      (src->num_links - 1) * sizeof(*src->links),
> +                      GFP_KERNEL);
> +       if (!new) {
> +               ret = -ENOMEM;
> +               goto out;
> +       }
> +
> +       for (i = 0, j = 0; j < src->num_links; j++) {
> +               if (src->links[j] == dst)
> +                       continue;
> +
> +               new[i++] = src->links[j];
> +       }
> +
> +       src->links = new;
> +       src->num_links--;

My understanding is that once you call realloc and it succeeds, you
must assume your old memory is gone and your new memory is only as big
as the new size you request. So you shouldn't call krealloc until
you've fixed the array up. Is the order of the links array important?
If not, you could just take the element at the end and stick it in the
slot that's being deleted. Then decrease the size and do your realloc.

> +
> +out:
> +       mutex_unlock(&icc_lock);
> +
> +       return ret;
> +}
> +EXPORT_SYMBOL_GPL(icc_link_remove);
> +
> +/**
> + * icc_node_add() - add an interconnect node to interconnect provider
> + * @node: pointer to the interconnect node
> + * @provider: pointer to the interconnect provider
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_node_add(struct icc_node *node, struct icc_provider *provider)
> +{
> +       mutex_lock(&icc_lock);
> +
> +       node->provider = provider;
> +       list_add(&node->node_list, &provider->nodes);
> +
> +       mutex_unlock(&icc_lock);
> +
> +       return 0;
> +}
> +EXPORT_SYMBOL_GPL(icc_node_add);
> +
> +/**
> + * icc_provider_add() - add a new interconnect provider
> + * @icc_provider: the interconnect provider that will be added into topology
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_provider_add(struct icc_provider *provider)
> +{
> +       if (WARN_ON(!provider->set))
> +               return -EINVAL;
> +
> +       mutex_init(&icc_lock);
> +
> +       INIT_LIST_HEAD(&provider->nodes);
> +       list_add(&provider->provider_list, &icc_provider_list);
> +
> +       mutex_unlock(&icc_lock);
> +
> +       dev_dbg(provider->dev, "interconnect provider added to topology\n");
> +
> +       return 0;
> +}
> +EXPORT_SYMBOL_GPL(icc_provider_add);
> +
> +/**
> + * icc_provider_del() - delete previously added interconnect provider
> + * @icc_provider: the interconnect provider that will be removed from topology
> + *
> + * Return: 0 on success, or an error code otherwise
> + */
> +int icc_provider_del(struct icc_provider *provider)
> +{
> +       mutex_lock(&icc_lock);
> +       if (provider->users) {
> +               pr_warn("interconnect provider still has %d users\n",
> +                       provider->users);
> +               mutex_unlock(&icc_lock);
> +               return -EBUSY;
> +       }
> +
> +       if (!list_empty_careful(&provider->nodes)) {
> +               pr_warn("interconnect provider still has nodes\n");
> +               mutex_unlock(&icc_lock);
> +               return -EEXIST;

How come you're returning different error codes for these two cases?
The error in both cases is effectively "you failed to clean up after
yourself", so maybe EBUSY makes sense for both of them. The pr_warn
helps to differentiate between the two for debugging.

> +       }
> +
> +       list_del(&provider->provider_list);
> +       mutex_unlock(&icc_lock);
> +
> +       return 0;
> +}
> +EXPORT_SYMBOL_GPL(icc_provider_del);
> +
> +MODULE_AUTHOR("Georgi Djakov <georgi.djakov@linaro.org");
> +MODULE_DESCRIPTION("Interconnect Driver Core");
> +MODULE_LICENSE("GPL v2");
> diff --git a/include/linux/interconnect-provider.h b/include/linux/interconnect-provider.h
> new file mode 100644
> index 000000000000..f4613c6dce4f
> --- /dev/null
> +++ b/include/linux/interconnect-provider.h
> @@ -0,0 +1,127 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Copyright (c) 2018, Linaro Ltd.
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#ifndef _LINUX_INTERCONNECT_PROVIDER_H
> +#define _LINUX_INTERCONNECT_PROVIDER_H
> +
> +#include <linux/interconnect.h>
> +
> +#define icc_units_to_bps(bw)  ((bw) * 1000ULL)
> +
> +struct icc_node;
> +
> +/**
> + * struct icc_provider - interconnect provider (controller) entity that might
> + * provide multiple interconnect controls
> + *
> + * @provider_list: list of the registered interconnect providers
> + * @nodes: internal list of the interconnect provider nodes
> + * @set: pointer to device specific set operation function
> + * @aggregate: pointer to device specific aggregate operation function
> + * @dev: the device this interconnect provider belongs to
> + * @users: count of active users
> + * @avg_bw: aggregated value of average bandwidth requests from all nodes
> + * @peak_bw: aggregated value of peak bandwidth requests from all nodes
> + * @data: pointer to private data
> + */
> +struct icc_provider {
> +       struct list_head        provider_list;
> +       struct list_head        nodes;
> +       int (*set)(struct icc_node *src, struct icc_node *dst,
> +                  u32 avg_bw, u32 peak_bw);
> +       int (*aggregate)(struct icc_node *node, u32 avg_bw, u32 peak_bw,
> +                        u32 *agg_avg, u32 *agg_peak);
> +       struct device           *dev;
> +       int                     users;
> +       u32                     avg_bw;
> +       u32                     peak_bw;
> +       void                    *data;
> +};
> +
> +/**
> + * struct icc_node - entity that is part of the interconnect topology
> + *
> + * @id: platform specific node id
> + * @name: node name used in debugfs
> + * @links: a list of targets pointing to where we can go next when traversing
> + * @num_links: number of links to other interconnect nodes
> + * @provider: points to the interconnect provider of this node
> + * @node_list: the list entry in the parent provider's "nodes" list
> + * @search_list: list used when walking the nodes graph
> + * @reverse: pointer to previous node when walking the nodes graph
> + * @is_traversed: flag that is used when walking the nodes graph
> + * @req_list: a list of QoS constraint requests associated with this node
> + * @avg_bw: aggregated value of average bandwidth requests from all consumers
> + * @peak_bw: aggregated value of peak bandwidth requests from all consumers
> + * @data: pointer to private data
> + */
> +struct icc_node {
> +       int                     id;
> +       const char              *name;
> +       struct icc_node         **links;
> +       size_t                  num_links;
> +
> +       struct icc_provider     *provider;
> +       struct list_head        node_list;
> +       struct list_head        orphan_list;

Is this used?

> +       struct list_head        search_list;
> +       struct icc_node         *reverse;
> +       bool                    is_traversed;
> +       struct hlist_head       req_list;
> +       u32                     avg_bw;
> +       u32                     peak_bw;
> +       void                    *data;
> +};
> +
> +#if IS_ENABLED(CONFIG_INTERCONNECT)
> +
> +struct icc_node *icc_node_create(int id);
> +void icc_node_remove(int id);
> +int icc_link_create(struct icc_node *node, const int dst_id);
> +int icc_link_remove(struct icc_node *src, struct icc_node *dst);
> +int icc_node_add(struct icc_node *node, struct icc_provider *provider);
> +int icc_provider_add(struct icc_provider *provider);
> +int icc_provider_del(struct icc_provider *provider);
> +
> +#else
> +
> +static inline struct icc_node *icc_node_create(int id)
> +{
> +       return ERR_PTR(-ENOTSUPP);
> +}
> +
> +void icc_node_remove(int id)
> +{
> +}
> +
> +static inline int icc_link_create(struct icc_node *node, const int dst_id)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +int icc_link_remove(struct icc_node *src, struct icc_node *dst)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +int icc_node_add(struct icc_node *node, struct icc_provider *provider)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +static inline int icc_provider_add(struct icc_provider *provider)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +static inline int icc_provider_del(struct icc_provider *provider)
> +{
> +       return -ENOTSUPP;
> +}
> +
> +#endif /* CONFIG_INTERCONNECT */
> +
> +#endif /* _LINUX_INTERCONNECT_PROVIDER_H */
> diff --git a/include/linux/interconnect.h b/include/linux/interconnect.h
> new file mode 100644
> index 000000000000..593215371fd6
> --- /dev/null
> +++ b/include/linux/interconnect.h
> @@ -0,0 +1,42 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Copyright (c) 2018, Linaro Ltd.
> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> + */
> +
> +#ifndef _LINUX_INTERCONNECT_H
> +#define _LINUX_INTERCONNECT_H
> +
> +#include <linux/mutex.h>
> +#include <linux/types.h>
> +
> +struct icc_path;
> +struct device;
> +
> +#if IS_ENABLED(CONFIG_INTERCONNECT)
> +
> +struct icc_path *icc_get(struct device *dev, const int src_id,
> +                        const int dst_id);
> +void icc_put(struct icc_path *path);
> +int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw);
> +
> +#else
> +
> +static inline struct icc_path *icc_get(struct device *dev, const int src_id,
> +                                      const int dst_id)
> +{
> +       return NULL;
> +}
> +
> +static inline void icc_put(struct icc_path *path)
> +{
> +}
> +
> +static inline int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
> +{
> +       return 0;

I was originally going to suggest that this should return a failure.
Then I talked myself out of it, saying that if the interconnect
framework is not compiled in, then clients should assume all their bus
needs are already met. I guess this is the correct assumption?

> +}
> +
> +#endif /* CONFIG_INTERCONNECT */
> +
> +#endif /* _LINUX_INTERCONNECT_H */

Re: [PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Matthias Kaehlcke]

On Tue, Jun 26, 2018 at 01:57:21PM -0700, Evan Green wrote:
> Hi Georgi. Thanks for the new spin of this.
> 
> On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
> >
> > This patch introduce a new API to get requirements and configure the
> > interconnect buses across the entire chipset to fit with the current
> > demand.
> >
> > The API is using a consumer/provider-based model, where the providers are
> > the interconnect buses and the consumers could be various drivers.
> > The consumers request interconnect resources (path) between endpoints and
> > set the desired constraints on this data flow path. The providers receive
> > requests from consumers and aggregate these requests for all master-slave
> > pairs on that path. Then the providers configure each participating in the
> > topology node according to the requested data flow path, physical links and
> > constraints. The topology could be complicated and multi-tiered and is SoC
> > specific.
> >
> > Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
> > ---
> >  Documentation/interconnect/interconnect.rst |  96 ++++
> >  drivers/Kconfig                             |   2 +
> >  drivers/Makefile                            |   1 +
> >  drivers/interconnect/Kconfig                |  10 +
> >  drivers/interconnect/Makefile               |   2 +
> >  drivers/interconnect/core.c                 | 586 ++++++++++++++++++++
> >  include/linux/interconnect-provider.h       | 127 +++++
> >  include/linux/interconnect.h                |  42 ++
> >  8 files changed, 866 insertions(+)
> >  create mode 100644 Documentation/interconnect/interconnect.rst
> >  create mode 100644 drivers/interconnect/Kconfig
> >  create mode 100644 drivers/interconnect/Makefile
> >  create mode 100644 drivers/interconnect/core.c
> >  create mode 100644 include/linux/interconnect-provider.h
> >  create mode 100644 include/linux/interconnect.h
> >
> > ...
> >
> > diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
> > new file mode 100644
> > index 000000000000..e7f96fc6722e
> > --- /dev/null
> > +++ b/drivers/interconnect/core.c
> > @@ -0,0 +1,586 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * Interconnect framework core driver
> > + *
> > + * Copyright (c) 2018, Linaro Ltd.
> > + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> > + */
> > +
> > +#include <linux/device.h>
> > +#include <linux/idr.h>
> > +#include <linux/init.h>
> > +#include <linux/interconnect.h>
> > +#include <linux/interconnect-provider.h>
> > +#include <linux/list.h>
> > +#include <linux/module.h>
> > +#include <linux/mutex.h>
> > +#include <linux/slab.h>
> > +
> > +static DEFINE_IDR(icc_idr);
> > +static LIST_HEAD(icc_provider_list);
> > +static DEFINE_MUTEX(icc_lock);
> > +
> > +/**
> > + * struct icc_req - constraints that are attached to each node
> > + *
> > + * @req_node: entry in list of requests for the particular @node
> > + * @node: the interconnect node to which this constraint applies
> > + * @dev: reference to the device that sets the constraints
> > + * @avg_bw: an integer describing the average bandwidth in kbps
> > + * @peak_bw: an integer describing the peak bandwidth in kbps
> > + */
> > +struct icc_req {
> > +       struct hlist_node req_node;
> > +       struct icc_node *node;
> > +       struct device *dev;
> > +       u32 avg_bw;
> > +       u32 peak_bw;
> > +};
> > +
> > +/**
> > + * struct icc_path - interconnect path structure
> > + * @num_nodes: number of hops (nodes)
> > + * @reqs: array of the requests applicable to this path of nodes
> > + */
> > +struct icc_path {
> > +       size_t num_nodes;
> > +       struct icc_req reqs[0];
> > +};
> > +
> > +static struct icc_node *node_find(const int id)
> > +{
> > +       struct icc_node *node;
> > +
> > +       mutex_lock(&icc_lock);
> > +       node = idr_find(&icc_idr, id);
> > +       mutex_unlock(&icc_lock);
> 
> I wonder if this is too low of a level to be dealing with the lock. I
> notice that everywhere you use this function, you afterwards
> immediately grab the lock and do more stuff. Maybe this function
> should have a comment saying it assumes the lock is already held, and
> then you can grab the lock in the callers, since you're doing that
> anyway.

I think the canonical way to document the expectation would be:

WARN_ON(!mutex_is_locked(&icc_lock));

> > +
> > +       return node;
> > +}
> > +
> > +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
> > +{
> > +       struct icc_node *node = dst;
> > +       struct icc_path *path;
> > +       size_t i;
> > +
> > +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
> > +                      GFP_KERNEL);
> > +       if (!path)
> > +               return ERR_PTR(-ENOMEM);
> > +
> > +       path->num_nodes = num_nodes;
> > +
> > +       for (i = 0; i < num_nodes; i++) {
> > +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
> > +
> > +               path->reqs[i].node = node;
> > +               /* reference to previous node was saved during path traversal */
> > +               node = node->reverse;
> > +       }
> > +
> > +       return path;
> > +}
> > +
> > +static struct icc_path *path_find(struct device *dev, struct icc_node *src,
> > +                                 struct icc_node *dst)
> > +{
> 
> I personally prefer a comment somewhere indicating that this function
> assumes icc_lock is already held. Not sure if that's conventional or
> not.

Same as above.

[PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Matthias Kaehlcke]

On Tue, Jun 26, 2018 at 01:57:21PM -0700, Evan Green wrote:
> Hi Georgi. Thanks for the new spin of this.
> 
> On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
> >
> > This patch introduce a new API to get requirements and configure the
> > interconnect buses across the entire chipset to fit with the current
> > demand.
> >
> > The API is using a consumer/provider-based model, where the providers are
> > the interconnect buses and the consumers could be various drivers.
> > The consumers request interconnect resources (path) between endpoints and
> > set the desired constraints on this data flow path. The providers receive
> > requests from consumers and aggregate these requests for all master-slave
> > pairs on that path. Then the providers configure each participating in the
> > topology node according to the requested data flow path, physical links and
> > constraints. The topology could be complicated and multi-tiered and is SoC
> > specific.
> >
> > Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
> > ---
> >  Documentation/interconnect/interconnect.rst |  96 ++++
> >  drivers/Kconfig                             |   2 +
> >  drivers/Makefile                            |   1 +
> >  drivers/interconnect/Kconfig                |  10 +
> >  drivers/interconnect/Makefile               |   2 +
> >  drivers/interconnect/core.c                 | 586 ++++++++++++++++++++
> >  include/linux/interconnect-provider.h       | 127 +++++
> >  include/linux/interconnect.h                |  42 ++
> >  8 files changed, 866 insertions(+)
> >  create mode 100644 Documentation/interconnect/interconnect.rst
> >  create mode 100644 drivers/interconnect/Kconfig
> >  create mode 100644 drivers/interconnect/Makefile
> >  create mode 100644 drivers/interconnect/core.c
> >  create mode 100644 include/linux/interconnect-provider.h
> >  create mode 100644 include/linux/interconnect.h
> >
> > ...
> >
> > diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
> > new file mode 100644
> > index 000000000000..e7f96fc6722e
> > --- /dev/null
> > +++ b/drivers/interconnect/core.c
> > @@ -0,0 +1,586 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * Interconnect framework core driver
> > + *
> > + * Copyright (c) 2018, Linaro Ltd.
> > + * Author: Georgi Djakov <georgi.djakov@linaro.org>
> > + */
> > +
> > +#include <linux/device.h>
> > +#include <linux/idr.h>
> > +#include <linux/init.h>
> > +#include <linux/interconnect.h>
> > +#include <linux/interconnect-provider.h>
> > +#include <linux/list.h>
> > +#include <linux/module.h>
> > +#include <linux/mutex.h>
> > +#include <linux/slab.h>
> > +
> > +static DEFINE_IDR(icc_idr);
> > +static LIST_HEAD(icc_provider_list);
> > +static DEFINE_MUTEX(icc_lock);
> > +
> > +/**
> > + * struct icc_req - constraints that are attached to each node
> > + *
> > + * @req_node: entry in list of requests for the particular @node
> > + * @node: the interconnect node to which this constraint applies
> > + * @dev: reference to the device that sets the constraints
> > + * @avg_bw: an integer describing the average bandwidth in kbps
> > + * @peak_bw: an integer describing the peak bandwidth in kbps
> > + */
> > +struct icc_req {
> > +       struct hlist_node req_node;
> > +       struct icc_node *node;
> > +       struct device *dev;
> > +       u32 avg_bw;
> > +       u32 peak_bw;
> > +};
> > +
> > +/**
> > + * struct icc_path - interconnect path structure
> > + * @num_nodes: number of hops (nodes)
> > + * @reqs: array of the requests applicable to this path of nodes
> > + */
> > +struct icc_path {
> > +       size_t num_nodes;
> > +       struct icc_req reqs[0];
> > +};
> > +
> > +static struct icc_node *node_find(const int id)
> > +{
> > +       struct icc_node *node;
> > +
> > +       mutex_lock(&icc_lock);
> > +       node = idr_find(&icc_idr, id);
> > +       mutex_unlock(&icc_lock);
> 
> I wonder if this is too low of a level to be dealing with the lock. I
> notice that everywhere you use this function, you afterwards
> immediately grab the lock and do more stuff. Maybe this function
> should have a comment saying it assumes the lock is already held, and
> then you can grab the lock in the callers, since you're doing that
> anyway.

I think the canonical way to document the expectation would be:

WARN_ON(!mutex_is_locked(&icc_lock));

> > +
> > +       return node;
> > +}
> > +
> > +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
> > +{
> > +       struct icc_node *node = dst;
> > +       struct icc_path *path;
> > +       size_t i;
> > +
> > +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
> > +                      GFP_KERNEL);
> > +       if (!path)
> > +               return ERR_PTR(-ENOMEM);
> > +
> > +       path->num_nodes = num_nodes;
> > +
> > +       for (i = 0; i < num_nodes; i++) {
> > +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
> > +
> > +               path->reqs[i].node = node;
> > +               /* reference to previous node was saved during path traversal */
> > +               node = node->reverse;
> > +       }
> > +
> > +       return path;
> > +}
> > +
> > +static struct icc_path *path_find(struct device *dev, struct icc_node *src,
> > +                                 struct icc_node *dst)
> > +{
> 
> I personally prefer a comment somewhere indicating that this function
> assumes icc_lock is already held. Not sure if that's conventional or
> not.

Same as above.

Re: [PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Matthias Kaehlcke]

Hi Georgi,

On Wed, Jun 20, 2018 at 03:11:34PM +0300, Georgi Djakov wrote:
> This patch introduce a new API to get requirements and configure the

nit: s/introduce/introduces/

> interconnect buses across the entire chipset to fit with the current
> demand.
> 
> The API is using a consumer/provider-based model, where the providers are
> the interconnect buses and the consumers could be various drivers.
> The consumers request interconnect resources (path) between endpoints and
> set the desired constraints on this data flow path. The providers receive
> requests from consumers and aggregate these requests for all master-slave
> pairs on that path. Then the providers configure each participating in the
> topology node according to the requested data flow path, physical links and
> constraints. The topology could be complicated and multi-tiered and is SoC
> specific.
> 
> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
> ---
> diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
>
> ...
>
> +static struct icc_path *path_find(struct device *dev, struct icc_node *src,
> +				  struct icc_node *dst)
> +{
> +	struct icc_node *n, *node = NULL;
> +	struct icc_provider *provider;
> +	struct list_head traverse_list;
> +	struct list_head edge_list;
> +	struct list_head visited_list;
> +	size_t i, depth = 0;
> +	bool found = false;
> +	int ret = -EPROBE_DEFER;
> +
> +	INIT_LIST_HEAD(&traverse_list);
> +	INIT_LIST_HEAD(&edge_list);
> +	INIT_LIST_HEAD(&visited_list);
> +
> +	list_add_tail(&src->search_list, &traverse_list);
> +	src->reverse = NULL;
> +
> +	do {
> +		list_for_each_entry_safe(node, n, &traverse_list, search_list) {
> +			if (node == dst) {
> +				found = true;
> +				list_add(&node->search_list, &visited_list);
> +				break;
> +			}
> +			for (i = 0; i < node->num_links; i++) {
> +				struct icc_node *tmp = node->links[i];
> +
> +				if (!tmp) {
> +					ret = -ENOENT;
> +					goto out;
> +				}
> +
> +				if (tmp->is_traversed)
> +					continue;
> +
> +				tmp->is_traversed = true;
> +				tmp->reverse = node;
> +				list_add(&tmp->search_list, &edge_list);
> +			}
> +		}
> +		if (found)
> +			break;
> +
> +		list_splice_init(&traverse_list, &visited_list);
> +		list_splice_init(&edge_list, &traverse_list);
> +
> +		/* count the hops away from the source */
> +		depth++;
> +
> +	} while (!list_empty(&traverse_list));
> +
> +out:
> +	/* reset the traversed state */
> +	list_for_each_entry(provider, &icc_provider_list, provider_list) {
> +		list_for_each_entry(n, &provider->nodes, node_list)
> +			if (n->is_traversed)
> +				n->is_traversed = false;
> +	}
> +
> +	if (found) {
> +		struct icc_path *path = path_allocate(dst, depth);
> +
> +		if (IS_ERR(path))
> +			return path;
> +
> +		/* initialize the path */
> +		for (i = 0; i < path->num_nodes; i++) {
> +			node = path->reqs[i].node;
> +			path->reqs[i].dev = dev;
> +			node->provider->users++;

nit: doing the assignment of path->reqs[i].dev before assiging 'node'
or after incrementing the 'users' would slightly improve readability.

> +static int apply_constraints(struct icc_path *path)
> +{
> +	struct icc_node *next, *prev = NULL;
> +	int ret = 0;
> +	int i;
> +
> +	for (i = 0; i < path->num_nodes; i++, prev = next) {
> +		struct icc_provider *p;
> +
> +		next = path->reqs[i].node;
> +		/*
> +		 * Both endpoints should be valid master-slave pairs of the
> +		 * same interconnect provider that will be configured.
> +		 */
> +		if (!prev || next->provider != prev->provider)
> +			continue;
> +
> +		p = next->provider;
> +
> +		aggregate_provider(p);
> +
> +		if (p->set) {
> +			/* set the constraints */
> +			ret = p->set(prev, next, p->avg_bw, p->peak_bw);
> +		}

remove curly brackets

EDIT: actually the condition can be removed, icc_provider_add() fails
when p->set is NULL.

> +int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
> +{
> +	struct icc_node *node;
> +	struct icc_provider *p;
> +	size_t i;
> +	int ret = 0;

initialization is not necessary

> +struct icc_path *icc_get(struct device *dev, const int src_id, const int dst_id)
> +{
> +	struct icc_node *src, *dst;
> +	struct icc_path *path = ERR_PTR(-EPROBE_DEFER);
> +
> +	src = node_find(src_id);
> +	if (!src) {
> +		dev_err(dev, "%s: invalid src=%d\n", __func__, src_id);
> +		goto out;
> +	}
> +
> +	dst = node_find(dst_id);
> +	if (!dst) {
> +		dev_err(dev, "%s: invalid dst=%d\n", __func__, dst_id);
> +		goto out;
> +	}
> +
> +	mutex_lock(&icc_lock);
> +	path = path_find(dev, src, dst);
> +	mutex_unlock(&icc_lock);
> +	if (IS_ERR(path)) {
> +		dev_err(dev, "%s: invalid path=%ld\n", __func__, PTR_ERR(path));
> +		goto out;

this goto isn't really needed

> +struct icc_node *icc_node_create(int id)
> +{
> +	struct icc_node *node;
> +
> +	/* check if node already exists */
> +	node = node_find(id);
> +	if (node)
> +		goto out;
> +
> +	node = kzalloc(sizeof(*node), GFP_KERNEL);
> +	if (!node) {
> +		node = ERR_PTR(-ENOMEM);
> +		goto out;
> +	}
> +
> +	mutex_lock(&icc_lock);
> +
> +	id = idr_alloc(&icc_idr, node, id, id + 1, GFP_KERNEL);
> +	if (WARN(id < 0, "couldn't get idr")) {

kfree(node);

> +int icc_node_add(struct icc_node *node, struct icc_provider *provider)
> +{
> +	mutex_lock(&icc_lock);
> +
> +	node->provider = provider;
> +	list_add(&node->node_list, &provider->nodes);
> +
> +	mutex_unlock(&icc_lock);
> +
> +	return 0;
> +}

The function returns always 0. Should probably be void so callers
don't add pointless checks of the return value.

> +int icc_provider_add(struct icc_provider *provider)
> +{
> +	if (WARN_ON(!provider->set))
> +		return -EINVAL;
> +
> +	mutex_init(&icc_lock);

Shouldn't this be mutex_lock()?

> +int icc_provider_del(struct icc_provider *provider)
> +{
> +	mutex_lock(&icc_lock);
> +	if (provider->users) {
> +		pr_warn("interconnect provider still has %d users\n",
> +			provider->users);
> +		mutex_unlock(&icc_lock);
> +		return -EBUSY;
> +	}
> +
> +	if (!list_empty_careful(&provider->nodes)) {
> +		pr_warn("interconnect provider still has nodes\n");
> +		mutex_unlock(&icc_lock);
> +		return -EEXIST;
> +	}

Could this be just list_empty()? If I didn't miss something icc_lock
is held in all paths that change p->nodes (assuming that all changes
should be done through the interfaces in this file).

Actually this check will always fail if icc_node_add() was called for
this provider, it doesn't seem nodes are ever removed.

Cheers

Matthias

[PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Matthias Kaehlcke]

Hi Georgi,

On Wed, Jun 20, 2018 at 03:11:34PM +0300, Georgi Djakov wrote:
> This patch introduce a new API to get requirements and configure the

nit: s/introduce/introduces/

> interconnect buses across the entire chipset to fit with the current
> demand.
> 
> The API is using a consumer/provider-based model, where the providers are
> the interconnect buses and the consumers could be various drivers.
> The consumers request interconnect resources (path) between endpoints and
> set the desired constraints on this data flow path. The providers receive
> requests from consumers and aggregate these requests for all master-slave
> pairs on that path. Then the providers configure each participating in the
> topology node according to the requested data flow path, physical links and
> constraints. The topology could be complicated and multi-tiered and is SoC
> specific.
> 
> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
> ---
> diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
>
> ...
>
> +static struct icc_path *path_find(struct device *dev, struct icc_node *src,
> +				  struct icc_node *dst)
> +{
> +	struct icc_node *n, *node = NULL;
> +	struct icc_provider *provider;
> +	struct list_head traverse_list;
> +	struct list_head edge_list;
> +	struct list_head visited_list;
> +	size_t i, depth = 0;
> +	bool found = false;
> +	int ret = -EPROBE_DEFER;
> +
> +	INIT_LIST_HEAD(&traverse_list);
> +	INIT_LIST_HEAD(&edge_list);
> +	INIT_LIST_HEAD(&visited_list);
> +
> +	list_add_tail(&src->search_list, &traverse_list);
> +	src->reverse = NULL;
> +
> +	do {
> +		list_for_each_entry_safe(node, n, &traverse_list, search_list) {
> +			if (node == dst) {
> +				found = true;
> +				list_add(&node->search_list, &visited_list);
> +				break;
> +			}
> +			for (i = 0; i < node->num_links; i++) {
> +				struct icc_node *tmp = node->links[i];
> +
> +				if (!tmp) {
> +					ret = -ENOENT;
> +					goto out;
> +				}
> +
> +				if (tmp->is_traversed)
> +					continue;
> +
> +				tmp->is_traversed = true;
> +				tmp->reverse = node;
> +				list_add(&tmp->search_list, &edge_list);
> +			}
> +		}
> +		if (found)
> +			break;
> +
> +		list_splice_init(&traverse_list, &visited_list);
> +		list_splice_init(&edge_list, &traverse_list);
> +
> +		/* count the hops away from the source */
> +		depth++;
> +
> +	} while (!list_empty(&traverse_list));
> +
> +out:
> +	/* reset the traversed state */
> +	list_for_each_entry(provider, &icc_provider_list, provider_list) {
> +		list_for_each_entry(n, &provider->nodes, node_list)
> +			if (n->is_traversed)
> +				n->is_traversed = false;
> +	}
> +
> +	if (found) {
> +		struct icc_path *path = path_allocate(dst, depth);
> +
> +		if (IS_ERR(path))
> +			return path;
> +
> +		/* initialize the path */
> +		for (i = 0; i < path->num_nodes; i++) {
> +			node = path->reqs[i].node;
> +			path->reqs[i].dev = dev;
> +			node->provider->users++;

nit: doing the assignment of path->reqs[i].dev before assiging 'node'
or after incrementing the 'users' would slightly improve readability.

> +static int apply_constraints(struct icc_path *path)
> +{
> +	struct icc_node *next, *prev = NULL;
> +	int ret = 0;
> +	int i;
> +
> +	for (i = 0; i < path->num_nodes; i++, prev = next) {
> +		struct icc_provider *p;
> +
> +		next = path->reqs[i].node;
> +		/*
> +		 * Both endpoints should be valid master-slave pairs of the
> +		 * same interconnect provider that will be configured.
> +		 */
> +		if (!prev || next->provider != prev->provider)
> +			continue;
> +
> +		p = next->provider;
> +
> +		aggregate_provider(p);
> +
> +		if (p->set) {
> +			/* set the constraints */
> +			ret = p->set(prev, next, p->avg_bw, p->peak_bw);
> +		}

remove curly brackets

EDIT: actually the condition can be removed, icc_provider_add() fails
when p->set is NULL.

> +int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
> +{
> +	struct icc_node *node;
> +	struct icc_provider *p;
> +	size_t i;
> +	int ret = 0;

initialization is not necessary

> +struct icc_path *icc_get(struct device *dev, const int src_id, const int dst_id)
> +{
> +	struct icc_node *src, *dst;
> +	struct icc_path *path = ERR_PTR(-EPROBE_DEFER);
> +
> +	src = node_find(src_id);
> +	if (!src) {
> +		dev_err(dev, "%s: invalid src=%d\n", __func__, src_id);
> +		goto out;
> +	}
> +
> +	dst = node_find(dst_id);
> +	if (!dst) {
> +		dev_err(dev, "%s: invalid dst=%d\n", __func__, dst_id);
> +		goto out;
> +	}
> +
> +	mutex_lock(&icc_lock);
> +	path = path_find(dev, src, dst);
> +	mutex_unlock(&icc_lock);
> +	if (IS_ERR(path)) {
> +		dev_err(dev, "%s: invalid path=%ld\n", __func__, PTR_ERR(path));
> +		goto out;

this goto isn't really needed

> +struct icc_node *icc_node_create(int id)
> +{
> +	struct icc_node *node;
> +
> +	/* check if node already exists */
> +	node = node_find(id);
> +	if (node)
> +		goto out;
> +
> +	node = kzalloc(sizeof(*node), GFP_KERNEL);
> +	if (!node) {
> +		node = ERR_PTR(-ENOMEM);
> +		goto out;
> +	}
> +
> +	mutex_lock(&icc_lock);
> +
> +	id = idr_alloc(&icc_idr, node, id, id + 1, GFP_KERNEL);
> +	if (WARN(id < 0, "couldn't get idr")) {

kfree(node);

> +int icc_node_add(struct icc_node *node, struct icc_provider *provider)
> +{
> +	mutex_lock(&icc_lock);
> +
> +	node->provider = provider;
> +	list_add(&node->node_list, &provider->nodes);
> +
> +	mutex_unlock(&icc_lock);
> +
> +	return 0;
> +}

The function returns always 0. Should probably be void so callers
don't add pointless checks of the return value.

> +int icc_provider_add(struct icc_provider *provider)
> +{
> +	if (WARN_ON(!provider->set))
> +		return -EINVAL;
> +
> +	mutex_init(&icc_lock);

Shouldn't this be mutex_lock()?

> +int icc_provider_del(struct icc_provider *provider)
> +{
> +	mutex_lock(&icc_lock);
> +	if (provider->users) {
> +		pr_warn("interconnect provider still has %d users\n",
> +			provider->users);
> +		mutex_unlock(&icc_lock);
> +		return -EBUSY;
> +	}
> +
> +	if (!list_empty_careful(&provider->nodes)) {
> +		pr_warn("interconnect provider still has nodes\n");
> +		mutex_unlock(&icc_lock);
> +		return -EEXIST;
> +	}

Could this be just list_empty()? If I didn't miss something icc_lock
is held in all paths that change p->nodes (assuming that all changes
should be done through the interfaces in this file).

Actually this check will always fail if icc_node_add() was called for
this provider, it doesn't seem nodes are ever removed.

Cheers

Matthias

Re: [PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Rob Clark]

On Tue, Jun 26, 2018 at 4:57 PM, Evan Green <evgreen@chromium.org> wrote:
> Hi Georgi. Thanks for the new spin of this.
>
> On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>>
>> This patch introduce a new API to get requirements and configure the
>> interconnect buses across the entire chipset to fit with the current
>> demand.
>>
>> The API is using a consumer/provider-based model, where the providers are
>> the interconnect buses and the consumers could be various drivers.
>> The consumers request interconnect resources (path) between endpoints and
>> set the desired constraints on this data flow path. The providers receive
>> requests from consumers and aggregate these requests for all master-slave
>> pairs on that path. Then the providers configure each participating in the
>> topology node according to the requested data flow path, physical links and
>> constraints. The topology could be complicated and multi-tiered and is SoC
>> specific.
>>
>> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
>> ---
>>  Documentation/interconnect/interconnect.rst |  96 ++++
>>  drivers/Kconfig                             |   2 +
>>  drivers/Makefile                            |   1 +
>>  drivers/interconnect/Kconfig                |  10 +
>>  drivers/interconnect/Makefile               |   2 +
>>  drivers/interconnect/core.c                 | 586 ++++++++++++++++++++
>>  include/linux/interconnect-provider.h       | 127 +++++
>>  include/linux/interconnect.h                |  42 ++
>>  8 files changed, 866 insertions(+)
>>  create mode 100644 Documentation/interconnect/interconnect.rst
>>  create mode 100644 drivers/interconnect/Kconfig
>>  create mode 100644 drivers/interconnect/Makefile
>>  create mode 100644 drivers/interconnect/core.c
>>  create mode 100644 include/linux/interconnect-provider.h
>>  create mode 100644 include/linux/interconnect.h
>>
>> diff --git a/Documentation/interconnect/interconnect.rst b/Documentation/interconnect/interconnect.rst
>> new file mode 100644
>> index 000000000000..a1ebd83ad0a1
>> --- /dev/null
>> +++ b/Documentation/interconnect/interconnect.rst
>> @@ -0,0 +1,96 @@
>> +.. SPDX-License-Identifier: GPL-2.0
>> +
>> +=====================================
>> +GENERIC SYSTEM INTERCONNECT SUBSYSTEM
>> +=====================================
>> +
>> +Introduction
>> +------------
>> +
>> +This framework is designed to provide a standard kernel interface to control
>> +the settings of the interconnects on a SoC. These settings can be throughput,
>> +latency and priority between multiple interconnected devices or functional
>> +blocks. This can be controlled dynamically in order to save power or provide
>> +maximum performance.
>> +
>> +The interconnect bus is a hardware with configurable parameters, which can be
>> +set on a data path according to the requests received from various drivers.
>> +An example of interconnect buses are the interconnects between various
>> +components or functional blocks in chipsets. There can be multiple interconnects
>> +on a SoC that can be multi-tiered.
>> +
>> +Below is a simplified diagram of a real-world SoC interconnect bus topology.
>> +
>> +::
>> +
>> + +----------------+    +----------------+
>> + | HW Accelerator |--->|      M NoC     |<---------------+
>> + +----------------+    +----------------+                |
>> +                         |      |                    +------------+
>> +  +-----+  +-------------+      V       +------+     |            |
>> +  | DDR |  |                +--------+  | PCIe |     |            |
>> +  +-----+  |                | Slaves |  +------+     |            |
>> +    ^ ^    |                +--------+     |         |   C NoC    |
>> +    | |    V                               V         |            |
>> + +------------------+   +------------------------+   |            |   +-----+
>> + |                  |-->|                        |-->|            |-->| CPU |
>> + |                  |-->|                        |<--|            |   +-----+
>> + |     Mem NoC      |   |         S NoC          |   +------------+
>> + |                  |<--|                        |---------+    |
>> + |                  |<--|                        |<------+ |    |   +--------+
>> + +------------------+   +------------------------+       | |    +-->| Slaves |
>> +   ^  ^    ^    ^          ^                             | |        +--------+
>> +   |  |    |    |          |                             | V
>> + +------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
>> + | CPUs |  |  | GPU |   | DSP |  | Masters |-->|       P NoC    |-->| Slaves |
>> + +------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
>> +           |
>> +       +-------+
>> +       | Modem |
>> +       +-------+
>> +
>> +Terminology
>> +-----------
>> +
>> +Interconnect provider is the software definition of the interconnect hardware.
>> +The interconnect providers on the above diagram are M NoC, S NoC, C NoC, P NoC
>> +and Mem NoC.
>> +
>> +Interconnect node is the software definition of the interconnect hardware
>> +port. Each interconnect provider consists of multiple interconnect nodes,
>> +which are connected to other SoC components including other interconnect
>> +providers. The point on the diagram where the CPUs connects to the memory is
>> +called an interconnect node, which belongs to the Mem NoC interconnect provider.
>> +
>> +Interconnect endpoints are the first or the last element of the path. Every
>> +endpoint is a node, but not every node is an endpoint.
>> +
>> +Interconnect path is everything between two endpoints including all the nodes
>> +that have to be traversed to reach from a source to destination node. It may
>> +include multiple master-slave pairs across several interconnect providers.
>> +
>> +Interconnect consumers are the entities which make use of the data paths exposed
>> +by the providers. The consumers send requests to providers requesting various
>> +throughput, latency and priority. Usually the consumers are device drivers, that
>> +send request based on their needs. An example for a consumer is a video decoder
>> +that supports various formats and image sizes.
>> +
>> +Interconnect providers
>> +----------------------
>> +
>> +Interconnect provider is an entity that implements methods to initialize and
>> +configure a interconnect bus hardware. The interconnect provider drivers should
>> +be registered with the interconnect provider core.
>> +
>> +The interconnect framework provider API functions are documented in
>> +.. kernel-doc:: include/linux/interconnect-provider.h
>> +
>> +Interconnect consumers
>> +----------------------
>> +
>> +Interconnect consumers are the clients which use the interconnect APIs to
>> +get paths between endpoints and set their bandwidth/latency/QoS requirements
>> +for these interconnect paths.
>> +
>> +The interconnect framework consumer API functions are documented in
>> +.. kernel-doc:: include/linux/interconnect.h
>> diff --git a/drivers/Kconfig b/drivers/Kconfig
>> index 95b9ccc08165..3ed6ede9d021 100644
>> --- a/drivers/Kconfig
>> +++ b/drivers/Kconfig
>> @@ -217,4 +217,6 @@ source "drivers/siox/Kconfig"
>>
>>  source "drivers/slimbus/Kconfig"
>>
>> +source "drivers/interconnect/Kconfig"
>> +
>>  endmenu
>> diff --git a/drivers/Makefile b/drivers/Makefile
>> index 24cd47014657..0cca95740d9b 100644
>> --- a/drivers/Makefile
>> +++ b/drivers/Makefile
>> @@ -185,3 +185,4 @@ obj-$(CONFIG_TEE)           += tee/
>>  obj-$(CONFIG_MULTIPLEXER)      += mux/
>>  obj-$(CONFIG_UNISYS_VISORBUS)  += visorbus/
>>  obj-$(CONFIG_SIOX)             += siox/
>> +obj-$(CONFIG_INTERCONNECT)     += interconnect/
>> diff --git a/drivers/interconnect/Kconfig b/drivers/interconnect/Kconfig
>> new file mode 100644
>> index 000000000000..a261c7d41deb
>> --- /dev/null
>> +++ b/drivers/interconnect/Kconfig
>> @@ -0,0 +1,10 @@
>> +menuconfig INTERCONNECT
>> +       tristate "On-Chip Interconnect management support"
>> +       help
>> +         Support for management of the on-chip interconnects.
>> +
>> +         This framework is designed to provide a generic interface for
>> +         managing the interconnects in a SoC.
>> +
>> +         If unsure, say no.
>> +
>> diff --git a/drivers/interconnect/Makefile b/drivers/interconnect/Makefile
>> new file mode 100644
>> index 000000000000..97fca2e09d24
>> --- /dev/null
>> +++ b/drivers/interconnect/Makefile
>> @@ -0,0 +1,2 @@
>> +# SPDX-License-Identifier: GPL-2.0
>> +obj-$(CONFIG_INTERCONNECT)             += core.o
>> diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
>> new file mode 100644
>> index 000000000000..e7f96fc6722e
>> --- /dev/null
>> +++ b/drivers/interconnect/core.c
>> @@ -0,0 +1,586 @@
>> +// SPDX-License-Identifier: GPL-2.0
>> +/*
>> + * Interconnect framework core driver
>> + *
>> + * Copyright (c) 2018, Linaro Ltd.
>> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
>> + */
>> +
>> +#include <linux/device.h>
>> +#include <linux/idr.h>
>> +#include <linux/init.h>
>> +#include <linux/interconnect.h>
>> +#include <linux/interconnect-provider.h>
>> +#include <linux/list.h>
>> +#include <linux/module.h>
>> +#include <linux/mutex.h>
>> +#include <linux/slab.h>
>> +
>> +static DEFINE_IDR(icc_idr);
>> +static LIST_HEAD(icc_provider_list);
>> +static DEFINE_MUTEX(icc_lock);
>> +
>> +/**
>> + * struct icc_req - constraints that are attached to each node
>> + *
>> + * @req_node: entry in list of requests for the particular @node
>> + * @node: the interconnect node to which this constraint applies
>> + * @dev: reference to the device that sets the constraints
>> + * @avg_bw: an integer describing the average bandwidth in kbps
>> + * @peak_bw: an integer describing the peak bandwidth in kbps
>> + */
>> +struct icc_req {
>> +       struct hlist_node req_node;
>> +       struct icc_node *node;
>> +       struct device *dev;
>> +       u32 avg_bw;
>> +       u32 peak_bw;
>> +};
>> +
>> +/**
>> + * struct icc_path - interconnect path structure
>> + * @num_nodes: number of hops (nodes)
>> + * @reqs: array of the requests applicable to this path of nodes
>> + */
>> +struct icc_path {
>> +       size_t num_nodes;
>> +       struct icc_req reqs[0];
>> +};
>> +
>> +static struct icc_node *node_find(const int id)
>> +{
>> +       struct icc_node *node;
>> +
>> +       mutex_lock(&icc_lock);
>> +       node = idr_find(&icc_idr, id);
>> +       mutex_unlock(&icc_lock);
>
> I wonder if this is too low of a level to be dealing with the lock. I
> notice that everywhere you use this function, you afterwards
> immediately grab the lock and do more stuff. Maybe this function
> should have a comment saying it assumes the lock is already held, and
> then you can grab the lock in the callers, since you're doing that
> anyway.
>
>> +
>> +       return node;
>> +}
>> +
>> +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
>> +{
>> +       struct icc_node *node = dst;
>> +       struct icc_path *path;
>> +       size_t i;
>> +
>> +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
>> +                      GFP_KERNEL);
>> +       if (!path)
>> +               return ERR_PTR(-ENOMEM);
>> +
>> +       path->num_nodes = num_nodes;
>> +
>> +       for (i = 0; i < num_nodes; i++) {
>> +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
>> +
>> +               path->reqs[i].node = node;
>> +               /* reference to previous node was saved during path traversal */
>> +               node = node->reverse;
>> +       }
>> +
>> +       return path;
>> +}
>> +
>> +static struct icc_path *path_find(struct device *dev, struct icc_node *src,
>> +                                 struct icc_node *dst)
>> +{
>
> I personally prefer a comment somewhere indicating that this function
> assumes icc_lock is already held. Not sure if that's conventional or
> not.

drive-by-note: WARN_ON(!mutex_is_locked(...)) is way more useful than
a comment ;-)

splats as a form of documentation are hugely useful for things that
other drivers may (indirectly) call and (one should assume) mess up
the locking rules the first time they try

BR,
-R

>
>> +       struct icc_node *n, *node = NULL;
>> +       struct icc_provider *provider;
>> +       struct list_head traverse_list;
>> +       struct list_head edge_list;
>> +       struct list_head visited_list;
>> +       size_t i, depth = 0;
>> +       bool found = false;
>> +       int ret = -EPROBE_DEFER;
>> +
>> +       INIT_LIST_HEAD(&traverse_list);
>> +       INIT_LIST_HEAD(&edge_list);
>> +       INIT_LIST_HEAD(&visited_list);
>> +
>> +       list_add_tail(&src->search_list, &traverse_list);
>> +       src->reverse = NULL;
>> +
>> +       do {
>> +               list_for_each_entry_safe(node, n, &traverse_list, search_list) {
>> +                       if (node == dst) {
>> +                               found = true;
>> +                               list_add(&node->search_list, &visited_list);
>> +                               break;
>> +                       }
>> +                       for (i = 0; i < node->num_links; i++) {
>> +                               struct icc_node *tmp = node->links[i];
>> +
>> +                               if (!tmp) {
>> +                                       ret = -ENOENT;
>> +                                       goto out;
>> +                               }
>> +
>> +                               if (tmp->is_traversed)
>> +                                       continue;
>> +
>> +                               tmp->is_traversed = true;
>> +                               tmp->reverse = node;
>> +                               list_add(&tmp->search_list, &edge_list);
>> +                       }
>> +               }
>> +               if (found)
>> +                       break;
>> +
>> +               list_splice_init(&traverse_list, &visited_list);
>> +               list_splice_init(&edge_list, &traverse_list);
>> +
>> +               /* count the hops away from the source */
>> +               depth++;
>> +
>> +       } while (!list_empty(&traverse_list));
>> +
>> +out:
>> +       /* reset the traversed state */
>> +       list_for_each_entry(provider, &icc_provider_list, provider_list) {
>> +               list_for_each_entry(n, &provider->nodes, node_list)
>> +                       if (n->is_traversed)
>> +                               n->is_traversed = false;
>
> Remove the conditional, just set is_traversed to false.
>
>> +       }
>> +
>> +       if (found) {
>> +               struct icc_path *path = path_allocate(dst, depth);
>
> Is the path supposed to include the source? For instance, if the dst
> were a neighbor, depth would be one, so only dst would be in the path.
> It seems like it might be worthwhile to have the source in there too.
>
>> +
>> +               if (IS_ERR(path))
>> +                       return path;
>> +
>> +               /* initialize the path */
>> +               for (i = 0; i < path->num_nodes; i++) {
>> +                       node = path->reqs[i].node;
>> +                       path->reqs[i].dev = dev;
>> +                       node->provider->users++;
>
> Should this loop live inside path_allocate? I'm unsure, but maybe at
> least path->reqs[i].dev = dev, since it feels like standard
> initialization of the path.
>
>> +               }
>> +               return path;
>> +       }
>> +
>> +       return ERR_PTR(ret);
>> +}
>> +
>> +/*
>> + * We want the path to honor all bandwidth requests, so the average
>> + * bandwidth requirements from each consumer are aggregated at each node
>> + * and provider level. By default the average bandwidth is the sum of all
>> + * averages and the peak will be the highest of all peak bandwidth requests.
>> + */
>> +
>> +static int aggregate_requests(struct icc_node *node)
>> +{
>> +       struct icc_provider *p = node->provider;
>> +       struct icc_req *r;
>> +
>> +       node->avg_bw = 0;
>> +       node->peak_bw = 0;
>> +
>> +       hlist_for_each_entry(r, &node->req_list, req_node)
>> +               p->aggregate(node, r->avg_bw, r->peak_bw,
>> +                            &node->avg_bw, &node->peak_bw);
>> +
>> +       return 0;
>> +}
>> +
>> +static void aggregate_provider(struct icc_provider *p)
>> +{
>> +       struct icc_node *n;
>> +
>> +       p->avg_bw = 0;
>> +       p->peak_bw = 0;
>> +
>> +       list_for_each_entry(n, &p->nodes, node_list)
>> +               p->aggregate(n, n->avg_bw, n->peak_bw,
>> +                            &p->avg_bw, &p->peak_bw);
>> +}
>> +
>> +static int apply_constraints(struct icc_path *path)
>> +{
>> +       struct icc_node *next, *prev = NULL;
>> +       int ret = 0;
>> +       int i;
>> +
>> +       for (i = 0; i < path->num_nodes; i++, prev = next) {
>> +               struct icc_provider *p;
>> +
>> +               next = path->reqs[i].node;
>> +               /*
>> +                * Both endpoints should be valid master-slave pairs of the
>> +                * same interconnect provider that will be configured.
>> +                */
>> +               if (!prev || next->provider != prev->provider)
>> +                       continue;
>> +
>> +               p = next->provider;
>> +
>> +               aggregate_provider(p);
>> +
>> +               if (p->set) {
>> +                       /* set the constraints */
>> +                       ret = p->set(prev, next, p->avg_bw, p->peak_bw);
>> +               }
>> +
>> +               if (ret)
>> +                       goto out;
>> +       }
>> +out:
>> +       return ret;
>> +}
>> +
>> +/**
>> + * icc_set() - set constraints on an interconnect path between two endpoints
>> + * @path: reference to the path returned by icc_get()
>> + * @avg_bw: average bandwidth in kbps
>> + * @peak_bw: peak bandwidth in kbps
>> + *
>> + * This function is used by an interconnect consumer to express its own needs
>> + * in terms of bandwidth for a previously requested path between two endpoints.
>> + * The requests are aggregated and each node is updated accordingly. The entire
>> + * path is locked by a mutex to ensure that the set() is completed.
>> + * The @path can be NULL when the "interconnects" DT properties is missing,
>> + * which will mean that no constraints will be set.
>> + *
>> + * Returns 0 on success, or an appropriate error code otherwise.
>> + */
>> +int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
>> +{
>> +       struct icc_node *node;
>> +       struct icc_provider *p;
>> +       size_t i;
>> +       int ret = 0;
>> +
>> +       if (!path)
>> +               return 0;
>> +
>> +       mutex_lock(&icc_lock);
>> +
>> +       for (i = 0; i < path->num_nodes; i++) {
>> +               node = path->reqs[i].node;
>> +               p = node->provider;
>> +
>> +               /* update the consumer request for this path */
>> +               path->reqs[i].avg_bw = avg_bw;
>> +               path->reqs[i].peak_bw = peak_bw;
>> +
>> +               /* aggregate requests for this node */
>> +               aggregate_requests(node);
>> +       }
>> +
>> +       ret = apply_constraints(path);
>> +       if (ret)
>> +               pr_err("interconnect: error applying constraints (%d)", ret);
>> +
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return ret;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_set);
>> +
>> +/**
>> + * icc_get() - return a handle for path between two endpoints
>> + * @dev: the device requesting the path
>> + * @src_id: source device port id
>> + * @dst_id: destination device port id
>> + *
>> + * This function will search for a path between two endpoints and return an
>> + * icc_path handle on success. Use icc_put() to release
>> + * constraints when the they are not needed anymore.
>> + *
>> + * Return: icc_path pointer on success, or ERR_PTR() on error
>> + */
>> +struct icc_path *icc_get(struct device *dev, const int src_id, const int dst_id)
>> +{
>> +       struct icc_node *src, *dst;
>> +       struct icc_path *path = ERR_PTR(-EPROBE_DEFER);
>> +
>> +       src = node_find(src_id);
>> +       if (!src) {
>> +               dev_err(dev, "%s: invalid src=%d\n", __func__, src_id);
>> +               goto out;
>> +       }
>> +
>> +       dst = node_find(dst_id);
>> +       if (!dst) {
>> +               dev_err(dev, "%s: invalid dst=%d\n", __func__, dst_id);
>> +               goto out;
>> +       }
>> +
>> +       mutex_lock(&icc_lock);
>> +       path = path_find(dev, src, dst);
>> +       mutex_unlock(&icc_lock);
>> +       if (IS_ERR(path)) {
>> +               dev_err(dev, "%s: invalid path=%ld\n", __func__, PTR_ERR(path));
>> +               goto out;
>> +       }
>> +
>> +out:
>> +       return path;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_get);
>> +
>> +/**
>> + * icc_put() - release the reference to the icc_path
>> + * @path: interconnect path
>> + *
>> + * Use this function to release the constraints on a path when the path is
>> + * no longer needed. The constraints will be re-aggregated.
>> + */
>> +void icc_put(struct icc_path *path)
>> +{
>> +       struct icc_node *node;
>> +       size_t i;
>> +       int ret;
>> +
>> +       if (!path || WARN_ON_ONCE(IS_ERR(path)))
>
> Why only once?
>
>> +               return;
>> +
>> +       ret = icc_set(path, 0, 0);
>> +       if (ret)
>> +               pr_err("%s: error (%d)\n", __func__, ret);
>> +
>> +       mutex_lock(&icc_lock);
>> +       for (i = 0; i < path->num_nodes; i++) {
>> +               node = path->reqs[i].node;
>> +               hlist_del(&path->reqs[i].req_node);
>> +
>> +               node->provider->users--;
>> +       }
>> +       mutex_unlock(&icc_lock);
>> +
>> +       kfree(path);
>> +}
>> +EXPORT_SYMBOL_GPL(icc_put);
>> +
>> +/**
>> + * icc_node_create() - create a node
>> + * @id: node id
>> + *
>> + * Return: icc_node pointer on success, or ERR_PTR() on error
>> + */
>> +struct icc_node *icc_node_create(int id)
>> +{
>> +       struct icc_node *node;
>> +
>> +       /* check if node already exists */
>> +       node = node_find(id);
>> +       if (node)
>> +               goto out;
>> +
>> +       node = kzalloc(sizeof(*node), GFP_KERNEL);
>> +       if (!node) {
>> +               node = ERR_PTR(-ENOMEM);
>> +               goto out;
>> +       }
>> +
>> +       mutex_lock(&icc_lock);
>> +
>> +       id = idr_alloc(&icc_idr, node, id, id + 1, GFP_KERNEL);
>> +       if (WARN(id < 0, "couldn't get idr")) {
>> +               node = ERR_PTR(id);
>> +               goto out;
>> +       }
>> +
>> +       node->id = id;
>> +
>> +out:
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return node;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_node_create);
>> +
>> +/**
>> + * icc_node_remove() - remove a node
>> + * @id: node id
>> + *
>> + */
>> +void icc_node_remove(int id)
>> +{
>> +       struct icc_node *node;
>> +
>> +       node = node_find(id);
>> +       if (node) {
>> +               mutex_lock(&icc_lock);
>> +               idr_remove(&icc_idr, node->id);
>
> Should we throw a warning if there are any paths that go through this
> node (ie req_list is non-empty)?
>
>> +               mutex_unlock(&icc_lock);
>> +       }
>> +
>> +       kfree(node);
>> +}
>> +EXPORT_SYMBOL_GPL(icc_node_remove);
>> +
>> +/**
>> + * icc_link_create() - create a link between two nodes
>> + * @src_id: source node id
>> + * @dst_id: destination node id
>> + *
>> + * Create a link between two nodes. The nodes might belong to different
>> + * interconnect providers and the @dst_id node might not exist (if the
>> + * provider driver has not probed yet). So just create the @dst_id node
>> + * and when the actual provider driver is probed, the rest of the node
>> + * data is filled.
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_link_create(struct icc_node *node, const int dst_id)
>> +{
>> +       struct icc_node *dst;
>> +       struct icc_node **new;
>> +       int ret = 0;
>> +
>> +       if (!node->provider)
>> +               return -EINVAL;
>> +
>> +       dst = node_find(dst_id);
>> +       if (!dst) {
>> +               dst = icc_node_create(dst_id);
>> +
>> +               if (IS_ERR(dst)) {
>> +                       ret = PTR_ERR(dst);
>> +                       goto out;
>> +               }
>> +       }
>> +
>> +       mutex_lock(&icc_lock);
>> +
>> +       new = krealloc(node->links,
>> +                      (node->num_links + 1) * sizeof(*node->links),
>> +                      GFP_KERNEL);
>> +       if (!new) {
>> +               ret = -ENOMEM;
>> +               goto out;
>> +       }
>> +
>> +       node->links = new;
>> +       node->links[node->num_links++] = dst;
>> +
>> +out:
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return ret;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_link_create);
>> +
>> +/**
>> + * icc_link_remove() - remove a link between two nodes
>> + * @src: pointer to source node
>> + * @dst: pointer to destination node
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_link_remove(struct icc_node *src, struct icc_node *dst)
>> +{
>> +       struct icc_node **new;
>> +       int ret = 0;
>> +       int i, j;
>> +
>> +       if (IS_ERR_OR_NULL(src))
>> +               return PTR_ERR(src);
>> +
>> +       if (IS_ERR_OR_NULL(dst))
>> +               return PTR_ERR(dst);
>
> I wonder if we should return a fixed error in these cases like
> -EINVAL, rather than handing through whatever crazy value is in
> src/dst.
>
>> +
>> +       mutex_lock(&icc_lock);
>> +
>> +       new = krealloc(src->links,
>> +                      (src->num_links - 1) * sizeof(*src->links),
>> +                      GFP_KERNEL);
>> +       if (!new) {
>> +               ret = -ENOMEM;
>> +               goto out;
>> +       }
>> +
>> +       for (i = 0, j = 0; j < src->num_links; j++) {
>> +               if (src->links[j] == dst)
>> +                       continue;
>> +
>> +               new[i++] = src->links[j];
>> +       }
>> +
>> +       src->links = new;
>> +       src->num_links--;
>
> My understanding is that once you call realloc and it succeeds, you
> must assume your old memory is gone and your new memory is only as big
> as the new size you request. So you shouldn't call krealloc until
> you've fixed the array up. Is the order of the links array important?
> If not, you could just take the element at the end and stick it in the
> slot that's being deleted. Then decrease the size and do your realloc.
>
>> +
>> +out:
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return ret;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_link_remove);
>> +
>> +/**
>> + * icc_node_add() - add an interconnect node to interconnect provider
>> + * @node: pointer to the interconnect node
>> + * @provider: pointer to the interconnect provider
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_node_add(struct icc_node *node, struct icc_provider *provider)
>> +{
>> +       mutex_lock(&icc_lock);
>> +
>> +       node->provider = provider;
>> +       list_add(&node->node_list, &provider->nodes);
>> +
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return 0;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_node_add);
>> +
>> +/**
>> + * icc_provider_add() - add a new interconnect provider
>> + * @icc_provider: the interconnect provider that will be added into topology
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_provider_add(struct icc_provider *provider)
>> +{
>> +       if (WARN_ON(!provider->set))
>> +               return -EINVAL;
>> +
>> +       mutex_init(&icc_lock);
>> +
>> +       INIT_LIST_HEAD(&provider->nodes);
>> +       list_add(&provider->provider_list, &icc_provider_list);
>> +
>> +       mutex_unlock(&icc_lock);
>> +
>> +       dev_dbg(provider->dev, "interconnect provider added to topology\n");
>> +
>> +       return 0;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_provider_add);
>> +
>> +/**
>> + * icc_provider_del() - delete previously added interconnect provider
>> + * @icc_provider: the interconnect provider that will be removed from topology
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_provider_del(struct icc_provider *provider)
>> +{
>> +       mutex_lock(&icc_lock);
>> +       if (provider->users) {
>> +               pr_warn("interconnect provider still has %d users\n",
>> +                       provider->users);
>> +               mutex_unlock(&icc_lock);
>> +               return -EBUSY;
>> +       }
>> +
>> +       if (!list_empty_careful(&provider->nodes)) {
>> +               pr_warn("interconnect provider still has nodes\n");
>> +               mutex_unlock(&icc_lock);
>> +               return -EEXIST;
>
> How come you're returning different error codes for these two cases?
> The error in both cases is effectively "you failed to clean up after
> yourself", so maybe EBUSY makes sense for both of them. The pr_warn
> helps to differentiate between the two for debugging.
>
>> +       }
>> +
>> +       list_del(&provider->provider_list);
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return 0;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_provider_del);
>> +
>> +MODULE_AUTHOR("Georgi Djakov <georgi.djakov@linaro.org");
>> +MODULE_DESCRIPTION("Interconnect Driver Core");
>> +MODULE_LICENSE("GPL v2");
>> diff --git a/include/linux/interconnect-provider.h b/include/linux/interconnect-provider.h
>> new file mode 100644
>> index 000000000000..f4613c6dce4f
>> --- /dev/null
>> +++ b/include/linux/interconnect-provider.h
>> @@ -0,0 +1,127 @@
>> +/* SPDX-License-Identifier: GPL-2.0 */
>> +/*
>> + * Copyright (c) 2018, Linaro Ltd.
>> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
>> + */
>> +
>> +#ifndef _LINUX_INTERCONNECT_PROVIDER_H
>> +#define _LINUX_INTERCONNECT_PROVIDER_H
>> +
>> +#include <linux/interconnect.h>
>> +
>> +#define icc_units_to_bps(bw)  ((bw) * 1000ULL)
>> +
>> +struct icc_node;
>> +
>> +/**
>> + * struct icc_provider - interconnect provider (controller) entity that might
>> + * provide multiple interconnect controls
>> + *
>> + * @provider_list: list of the registered interconnect providers
>> + * @nodes: internal list of the interconnect provider nodes
>> + * @set: pointer to device specific set operation function
>> + * @aggregate: pointer to device specific aggregate operation function
>> + * @dev: the device this interconnect provider belongs to
>> + * @users: count of active users
>> + * @avg_bw: aggregated value of average bandwidth requests from all nodes
>> + * @peak_bw: aggregated value of peak bandwidth requests from all nodes
>> + * @data: pointer to private data
>> + */
>> +struct icc_provider {
>> +       struct list_head        provider_list;
>> +       struct list_head        nodes;
>> +       int (*set)(struct icc_node *src, struct icc_node *dst,
>> +                  u32 avg_bw, u32 peak_bw);
>> +       int (*aggregate)(struct icc_node *node, u32 avg_bw, u32 peak_bw,
>> +                        u32 *agg_avg, u32 *agg_peak);
>> +       struct device           *dev;
>> +       int                     users;
>> +       u32                     avg_bw;
>> +       u32                     peak_bw;
>> +       void                    *data;
>> +};
>> +
>> +/**
>> + * struct icc_node - entity that is part of the interconnect topology
>> + *
>> + * @id: platform specific node id
>> + * @name: node name used in debugfs
>> + * @links: a list of targets pointing to where we can go next when traversing
>> + * @num_links: number of links to other interconnect nodes
>> + * @provider: points to the interconnect provider of this node
>> + * @node_list: the list entry in the parent provider's "nodes" list
>> + * @search_list: list used when walking the nodes graph
>> + * @reverse: pointer to previous node when walking the nodes graph
>> + * @is_traversed: flag that is used when walking the nodes graph
>> + * @req_list: a list of QoS constraint requests associated with this node
>> + * @avg_bw: aggregated value of average bandwidth requests from all consumers
>> + * @peak_bw: aggregated value of peak bandwidth requests from all consumers
>> + * @data: pointer to private data
>> + */
>> +struct icc_node {
>> +       int                     id;
>> +       const char              *name;
>> +       struct icc_node         **links;
>> +       size_t                  num_links;
>> +
>> +       struct icc_provider     *provider;
>> +       struct list_head        node_list;
>> +       struct list_head        orphan_list;
>
> Is this used?
>
>> +       struct list_head        search_list;
>> +       struct icc_node         *reverse;
>> +       bool                    is_traversed;
>> +       struct hlist_head       req_list;
>> +       u32                     avg_bw;
>> +       u32                     peak_bw;
>> +       void                    *data;
>> +};
>> +
>> +#if IS_ENABLED(CONFIG_INTERCONNECT)
>> +
>> +struct icc_node *icc_node_create(int id);
>> +void icc_node_remove(int id);
>> +int icc_link_create(struct icc_node *node, const int dst_id);
>> +int icc_link_remove(struct icc_node *src, struct icc_node *dst);
>> +int icc_node_add(struct icc_node *node, struct icc_provider *provider);
>> +int icc_provider_add(struct icc_provider *provider);
>> +int icc_provider_del(struct icc_provider *provider);
>> +
>> +#else
>> +
>> +static inline struct icc_node *icc_node_create(int id)
>> +{
>> +       return ERR_PTR(-ENOTSUPP);
>> +}
>> +
>> +void icc_node_remove(int id)
>> +{
>> +}
>> +
>> +static inline int icc_link_create(struct icc_node *node, const int dst_id)
>> +{
>> +       return -ENOTSUPP;
>> +}
>> +
>> +int icc_link_remove(struct icc_node *src, struct icc_node *dst)
>> +{
>> +       return -ENOTSUPP;
>> +}
>> +
>> +int icc_node_add(struct icc_node *node, struct icc_provider *provider)
>> +{
>> +       return -ENOTSUPP;
>> +}
>> +
>> +static inline int icc_provider_add(struct icc_provider *provider)
>> +{
>> +       return -ENOTSUPP;
>> +}
>> +
>> +static inline int icc_provider_del(struct icc_provider *provider)
>> +{
>> +       return -ENOTSUPP;
>> +}
>> +
>> +#endif /* CONFIG_INTERCONNECT */
>> +
>> +#endif /* _LINUX_INTERCONNECT_PROVIDER_H */
>> diff --git a/include/linux/interconnect.h b/include/linux/interconnect.h
>> new file mode 100644
>> index 000000000000..593215371fd6
>> --- /dev/null
>> +++ b/include/linux/interconnect.h
>> @@ -0,0 +1,42 @@
>> +/* SPDX-License-Identifier: GPL-2.0 */
>> +/*
>> + * Copyright (c) 2018, Linaro Ltd.
>> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
>> + */
>> +
>> +#ifndef _LINUX_INTERCONNECT_H
>> +#define _LINUX_INTERCONNECT_H
>> +
>> +#include <linux/mutex.h>
>> +#include <linux/types.h>
>> +
>> +struct icc_path;
>> +struct device;
>> +
>> +#if IS_ENABLED(CONFIG_INTERCONNECT)
>> +
>> +struct icc_path *icc_get(struct device *dev, const int src_id,
>> +                        const int dst_id);
>> +void icc_put(struct icc_path *path);
>> +int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw);
>> +
>> +#else
>> +
>> +static inline struct icc_path *icc_get(struct device *dev, const int src_id,
>> +                                      const int dst_id)
>> +{
>> +       return NULL;
>> +}
>> +
>> +static inline void icc_put(struct icc_path *path)
>> +{
>> +}
>> +
>> +static inline int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
>> +{
>> +       return 0;
>
> I was originally going to suggest that this should return a failure.
> Then I talked myself out of it, saying that if the interconnect
> framework is not compiled in, then clients should assume all their bus
> needs are already met. I guess this is the correct assumption?
>
>> +}
>> +
>> +#endif /* CONFIG_INTERCONNECT */
>> +
>> +#endif /* _LINUX_INTERCONNECT_H */

[PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Rob Clark]

On Tue, Jun 26, 2018 at 4:57 PM, Evan Green <evgreen@chromium.org> wrote:
> Hi Georgi. Thanks for the new spin of this.
>
> On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>>
>> This patch introduce a new API to get requirements and configure the
>> interconnect buses across the entire chipset to fit with the current
>> demand.
>>
>> The API is using a consumer/provider-based model, where the providers are
>> the interconnect buses and the consumers could be various drivers.
>> The consumers request interconnect resources (path) between endpoints and
>> set the desired constraints on this data flow path. The providers receive
>> requests from consumers and aggregate these requests for all master-slave
>> pairs on that path. Then the providers configure each participating in the
>> topology node according to the requested data flow path, physical links and
>> constraints. The topology could be complicated and multi-tiered and is SoC
>> specific.
>>
>> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
>> ---
>>  Documentation/interconnect/interconnect.rst |  96 ++++
>>  drivers/Kconfig                             |   2 +
>>  drivers/Makefile                            |   1 +
>>  drivers/interconnect/Kconfig                |  10 +
>>  drivers/interconnect/Makefile               |   2 +
>>  drivers/interconnect/core.c                 | 586 ++++++++++++++++++++
>>  include/linux/interconnect-provider.h       | 127 +++++
>>  include/linux/interconnect.h                |  42 ++
>>  8 files changed, 866 insertions(+)
>>  create mode 100644 Documentation/interconnect/interconnect.rst
>>  create mode 100644 drivers/interconnect/Kconfig
>>  create mode 100644 drivers/interconnect/Makefile
>>  create mode 100644 drivers/interconnect/core.c
>>  create mode 100644 include/linux/interconnect-provider.h
>>  create mode 100644 include/linux/interconnect.h
>>
>> diff --git a/Documentation/interconnect/interconnect.rst b/Documentation/interconnect/interconnect.rst
>> new file mode 100644
>> index 000000000000..a1ebd83ad0a1
>> --- /dev/null
>> +++ b/Documentation/interconnect/interconnect.rst
>> @@ -0,0 +1,96 @@
>> +.. SPDX-License-Identifier: GPL-2.0
>> +
>> +=====================================
>> +GENERIC SYSTEM INTERCONNECT SUBSYSTEM
>> +=====================================
>> +
>> +Introduction
>> +------------
>> +
>> +This framework is designed to provide a standard kernel interface to control
>> +the settings of the interconnects on a SoC. These settings can be throughput,
>> +latency and priority between multiple interconnected devices or functional
>> +blocks. This can be controlled dynamically in order to save power or provide
>> +maximum performance.
>> +
>> +The interconnect bus is a hardware with configurable parameters, which can be
>> +set on a data path according to the requests received from various drivers.
>> +An example of interconnect buses are the interconnects between various
>> +components or functional blocks in chipsets. There can be multiple interconnects
>> +on a SoC that can be multi-tiered.
>> +
>> +Below is a simplified diagram of a real-world SoC interconnect bus topology.
>> +
>> +::
>> +
>> + +----------------+    +----------------+
>> + | HW Accelerator |--->|      M NoC     |<---------------+
>> + +----------------+    +----------------+                |
>> +                         |      |                    +------------+
>> +  +-----+  +-------------+      V       +------+     |            |
>> +  | DDR |  |                +--------+  | PCIe |     |            |
>> +  +-----+  |                | Slaves |  +------+     |            |
>> +    ^ ^    |                +--------+     |         |   C NoC    |
>> +    | |    V                               V         |            |
>> + +------------------+   +------------------------+   |            |   +-----+
>> + |                  |-->|                        |-->|            |-->| CPU |
>> + |                  |-->|                        |<--|            |   +-----+
>> + |     Mem NoC      |   |         S NoC          |   +------------+
>> + |                  |<--|                        |---------+    |
>> + |                  |<--|                        |<------+ |    |   +--------+
>> + +------------------+   +------------------------+       | |    +-->| Slaves |
>> +   ^  ^    ^    ^          ^                             | |        +--------+
>> +   |  |    |    |          |                             | V
>> + +------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
>> + | CPUs |  |  | GPU |   | DSP |  | Masters |-->|       P NoC    |-->| Slaves |
>> + +------+  |  +-----+   +-----+  +---------+   +----------------+   +--------+
>> +           |
>> +       +-------+
>> +       | Modem |
>> +       +-------+
>> +
>> +Terminology
>> +-----------
>> +
>> +Interconnect provider is the software definition of the interconnect hardware.
>> +The interconnect providers on the above diagram are M NoC, S NoC, C NoC, P NoC
>> +and Mem NoC.
>> +
>> +Interconnect node is the software definition of the interconnect hardware
>> +port. Each interconnect provider consists of multiple interconnect nodes,
>> +which are connected to other SoC components including other interconnect
>> +providers. The point on the diagram where the CPUs connects to the memory is
>> +called an interconnect node, which belongs to the Mem NoC interconnect provider.
>> +
>> +Interconnect endpoints are the first or the last element of the path. Every
>> +endpoint is a node, but not every node is an endpoint.
>> +
>> +Interconnect path is everything between two endpoints including all the nodes
>> +that have to be traversed to reach from a source to destination node. It may
>> +include multiple master-slave pairs across several interconnect providers.
>> +
>> +Interconnect consumers are the entities which make use of the data paths exposed
>> +by the providers. The consumers send requests to providers requesting various
>> +throughput, latency and priority. Usually the consumers are device drivers, that
>> +send request based on their needs. An example for a consumer is a video decoder
>> +that supports various formats and image sizes.
>> +
>> +Interconnect providers
>> +----------------------
>> +
>> +Interconnect provider is an entity that implements methods to initialize and
>> +configure a interconnect bus hardware. The interconnect provider drivers should
>> +be registered with the interconnect provider core.
>> +
>> +The interconnect framework provider API functions are documented in
>> +.. kernel-doc:: include/linux/interconnect-provider.h
>> +
>> +Interconnect consumers
>> +----------------------
>> +
>> +Interconnect consumers are the clients which use the interconnect APIs to
>> +get paths between endpoints and set their bandwidth/latency/QoS requirements
>> +for these interconnect paths.
>> +
>> +The interconnect framework consumer API functions are documented in
>> +.. kernel-doc:: include/linux/interconnect.h
>> diff --git a/drivers/Kconfig b/drivers/Kconfig
>> index 95b9ccc08165..3ed6ede9d021 100644
>> --- a/drivers/Kconfig
>> +++ b/drivers/Kconfig
>> @@ -217,4 +217,6 @@ source "drivers/siox/Kconfig"
>>
>>  source "drivers/slimbus/Kconfig"
>>
>> +source "drivers/interconnect/Kconfig"
>> +
>>  endmenu
>> diff --git a/drivers/Makefile b/drivers/Makefile
>> index 24cd47014657..0cca95740d9b 100644
>> --- a/drivers/Makefile
>> +++ b/drivers/Makefile
>> @@ -185,3 +185,4 @@ obj-$(CONFIG_TEE)           += tee/
>>  obj-$(CONFIG_MULTIPLEXER)      += mux/
>>  obj-$(CONFIG_UNISYS_VISORBUS)  += visorbus/
>>  obj-$(CONFIG_SIOX)             += siox/
>> +obj-$(CONFIG_INTERCONNECT)     += interconnect/
>> diff --git a/drivers/interconnect/Kconfig b/drivers/interconnect/Kconfig
>> new file mode 100644
>> index 000000000000..a261c7d41deb
>> --- /dev/null
>> +++ b/drivers/interconnect/Kconfig
>> @@ -0,0 +1,10 @@
>> +menuconfig INTERCONNECT
>> +       tristate "On-Chip Interconnect management support"
>> +       help
>> +         Support for management of the on-chip interconnects.
>> +
>> +         This framework is designed to provide a generic interface for
>> +         managing the interconnects in a SoC.
>> +
>> +         If unsure, say no.
>> +
>> diff --git a/drivers/interconnect/Makefile b/drivers/interconnect/Makefile
>> new file mode 100644
>> index 000000000000..97fca2e09d24
>> --- /dev/null
>> +++ b/drivers/interconnect/Makefile
>> @@ -0,0 +1,2 @@
>> +# SPDX-License-Identifier: GPL-2.0
>> +obj-$(CONFIG_INTERCONNECT)             += core.o
>> diff --git a/drivers/interconnect/core.c b/drivers/interconnect/core.c
>> new file mode 100644
>> index 000000000000..e7f96fc6722e
>> --- /dev/null
>> +++ b/drivers/interconnect/core.c
>> @@ -0,0 +1,586 @@
>> +// SPDX-License-Identifier: GPL-2.0
>> +/*
>> + * Interconnect framework core driver
>> + *
>> + * Copyright (c) 2018, Linaro Ltd.
>> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
>> + */
>> +
>> +#include <linux/device.h>
>> +#include <linux/idr.h>
>> +#include <linux/init.h>
>> +#include <linux/interconnect.h>
>> +#include <linux/interconnect-provider.h>
>> +#include <linux/list.h>
>> +#include <linux/module.h>
>> +#include <linux/mutex.h>
>> +#include <linux/slab.h>
>> +
>> +static DEFINE_IDR(icc_idr);
>> +static LIST_HEAD(icc_provider_list);
>> +static DEFINE_MUTEX(icc_lock);
>> +
>> +/**
>> + * struct icc_req - constraints that are attached to each node
>> + *
>> + * @req_node: entry in list of requests for the particular @node
>> + * @node: the interconnect node to which this constraint applies
>> + * @dev: reference to the device that sets the constraints
>> + * @avg_bw: an integer describing the average bandwidth in kbps
>> + * @peak_bw: an integer describing the peak bandwidth in kbps
>> + */
>> +struct icc_req {
>> +       struct hlist_node req_node;
>> +       struct icc_node *node;
>> +       struct device *dev;
>> +       u32 avg_bw;
>> +       u32 peak_bw;
>> +};
>> +
>> +/**
>> + * struct icc_path - interconnect path structure
>> + * @num_nodes: number of hops (nodes)
>> + * @reqs: array of the requests applicable to this path of nodes
>> + */
>> +struct icc_path {
>> +       size_t num_nodes;
>> +       struct icc_req reqs[0];
>> +};
>> +
>> +static struct icc_node *node_find(const int id)
>> +{
>> +       struct icc_node *node;
>> +
>> +       mutex_lock(&icc_lock);
>> +       node = idr_find(&icc_idr, id);
>> +       mutex_unlock(&icc_lock);
>
> I wonder if this is too low of a level to be dealing with the lock. I
> notice that everywhere you use this function, you afterwards
> immediately grab the lock and do more stuff. Maybe this function
> should have a comment saying it assumes the lock is already held, and
> then you can grab the lock in the callers, since you're doing that
> anyway.
>
>> +
>> +       return node;
>> +}
>> +
>> +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
>> +{
>> +       struct icc_node *node = dst;
>> +       struct icc_path *path;
>> +       size_t i;
>> +
>> +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
>> +                      GFP_KERNEL);
>> +       if (!path)
>> +               return ERR_PTR(-ENOMEM);
>> +
>> +       path->num_nodes = num_nodes;
>> +
>> +       for (i = 0; i < num_nodes; i++) {
>> +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
>> +
>> +               path->reqs[i].node = node;
>> +               /* reference to previous node was saved during path traversal */
>> +               node = node->reverse;
>> +       }
>> +
>> +       return path;
>> +}
>> +
>> +static struct icc_path *path_find(struct device *dev, struct icc_node *src,
>> +                                 struct icc_node *dst)
>> +{
>
> I personally prefer a comment somewhere indicating that this function
> assumes icc_lock is already held. Not sure if that's conventional or
> not.

drive-by-note: WARN_ON(!mutex_is_locked(...)) is way more useful than
a comment ;-)

splats as a form of documentation are hugely useful for things that
other drivers may (indirectly) call and (one should assume) mess up
the locking rules the first time they try

BR,
-R

>
>> +       struct icc_node *n, *node = NULL;
>> +       struct icc_provider *provider;
>> +       struct list_head traverse_list;
>> +       struct list_head edge_list;
>> +       struct list_head visited_list;
>> +       size_t i, depth = 0;
>> +       bool found = false;
>> +       int ret = -EPROBE_DEFER;
>> +
>> +       INIT_LIST_HEAD(&traverse_list);
>> +       INIT_LIST_HEAD(&edge_list);
>> +       INIT_LIST_HEAD(&visited_list);
>> +
>> +       list_add_tail(&src->search_list, &traverse_list);
>> +       src->reverse = NULL;
>> +
>> +       do {
>> +               list_for_each_entry_safe(node, n, &traverse_list, search_list) {
>> +                       if (node == dst) {
>> +                               found = true;
>> +                               list_add(&node->search_list, &visited_list);
>> +                               break;
>> +                       }
>> +                       for (i = 0; i < node->num_links; i++) {
>> +                               struct icc_node *tmp = node->links[i];
>> +
>> +                               if (!tmp) {
>> +                                       ret = -ENOENT;
>> +                                       goto out;
>> +                               }
>> +
>> +                               if (tmp->is_traversed)
>> +                                       continue;
>> +
>> +                               tmp->is_traversed = true;
>> +                               tmp->reverse = node;
>> +                               list_add(&tmp->search_list, &edge_list);
>> +                       }
>> +               }
>> +               if (found)
>> +                       break;
>> +
>> +               list_splice_init(&traverse_list, &visited_list);
>> +               list_splice_init(&edge_list, &traverse_list);
>> +
>> +               /* count the hops away from the source */
>> +               depth++;
>> +
>> +       } while (!list_empty(&traverse_list));
>> +
>> +out:
>> +       /* reset the traversed state */
>> +       list_for_each_entry(provider, &icc_provider_list, provider_list) {
>> +               list_for_each_entry(n, &provider->nodes, node_list)
>> +                       if (n->is_traversed)
>> +                               n->is_traversed = false;
>
> Remove the conditional, just set is_traversed to false.
>
>> +       }
>> +
>> +       if (found) {
>> +               struct icc_path *path = path_allocate(dst, depth);
>
> Is the path supposed to include the source? For instance, if the dst
> were a neighbor, depth would be one, so only dst would be in the path.
> It seems like it might be worthwhile to have the source in there too.
>
>> +
>> +               if (IS_ERR(path))
>> +                       return path;
>> +
>> +               /* initialize the path */
>> +               for (i = 0; i < path->num_nodes; i++) {
>> +                       node = path->reqs[i].node;
>> +                       path->reqs[i].dev = dev;
>> +                       node->provider->users++;
>
> Should this loop live inside path_allocate? I'm unsure, but maybe at
> least path->reqs[i].dev = dev, since it feels like standard
> initialization of the path.
>
>> +               }
>> +               return path;
>> +       }
>> +
>> +       return ERR_PTR(ret);
>> +}
>> +
>> +/*
>> + * We want the path to honor all bandwidth requests, so the average
>> + * bandwidth requirements from each consumer are aggregated at each node
>> + * and provider level. By default the average bandwidth is the sum of all
>> + * averages and the peak will be the highest of all peak bandwidth requests.
>> + */
>> +
>> +static int aggregate_requests(struct icc_node *node)
>> +{
>> +       struct icc_provider *p = node->provider;
>> +       struct icc_req *r;
>> +
>> +       node->avg_bw = 0;
>> +       node->peak_bw = 0;
>> +
>> +       hlist_for_each_entry(r, &node->req_list, req_node)
>> +               p->aggregate(node, r->avg_bw, r->peak_bw,
>> +                            &node->avg_bw, &node->peak_bw);
>> +
>> +       return 0;
>> +}
>> +
>> +static void aggregate_provider(struct icc_provider *p)
>> +{
>> +       struct icc_node *n;
>> +
>> +       p->avg_bw = 0;
>> +       p->peak_bw = 0;
>> +
>> +       list_for_each_entry(n, &p->nodes, node_list)
>> +               p->aggregate(n, n->avg_bw, n->peak_bw,
>> +                            &p->avg_bw, &p->peak_bw);
>> +}
>> +
>> +static int apply_constraints(struct icc_path *path)
>> +{
>> +       struct icc_node *next, *prev = NULL;
>> +       int ret = 0;
>> +       int i;
>> +
>> +       for (i = 0; i < path->num_nodes; i++, prev = next) {
>> +               struct icc_provider *p;
>> +
>> +               next = path->reqs[i].node;
>> +               /*
>> +                * Both endpoints should be valid master-slave pairs of the
>> +                * same interconnect provider that will be configured.
>> +                */
>> +               if (!prev || next->provider != prev->provider)
>> +                       continue;
>> +
>> +               p = next->provider;
>> +
>> +               aggregate_provider(p);
>> +
>> +               if (p->set) {
>> +                       /* set the constraints */
>> +                       ret = p->set(prev, next, p->avg_bw, p->peak_bw);
>> +               }
>> +
>> +               if (ret)
>> +                       goto out;
>> +       }
>> +out:
>> +       return ret;
>> +}
>> +
>> +/**
>> + * icc_set() - set constraints on an interconnect path between two endpoints
>> + * @path: reference to the path returned by icc_get()
>> + * @avg_bw: average bandwidth in kbps
>> + * @peak_bw: peak bandwidth in kbps
>> + *
>> + * This function is used by an interconnect consumer to express its own needs
>> + * in terms of bandwidth for a previously requested path between two endpoints.
>> + * The requests are aggregated and each node is updated accordingly. The entire
>> + * path is locked by a mutex to ensure that the set() is completed.
>> + * The @path can be NULL when the "interconnects" DT properties is missing,
>> + * which will mean that no constraints will be set.
>> + *
>> + * Returns 0 on success, or an appropriate error code otherwise.
>> + */
>> +int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
>> +{
>> +       struct icc_node *node;
>> +       struct icc_provider *p;
>> +       size_t i;
>> +       int ret = 0;
>> +
>> +       if (!path)
>> +               return 0;
>> +
>> +       mutex_lock(&icc_lock);
>> +
>> +       for (i = 0; i < path->num_nodes; i++) {
>> +               node = path->reqs[i].node;
>> +               p = node->provider;
>> +
>> +               /* update the consumer request for this path */
>> +               path->reqs[i].avg_bw = avg_bw;
>> +               path->reqs[i].peak_bw = peak_bw;
>> +
>> +               /* aggregate requests for this node */
>> +               aggregate_requests(node);
>> +       }
>> +
>> +       ret = apply_constraints(path);
>> +       if (ret)
>> +               pr_err("interconnect: error applying constraints (%d)", ret);
>> +
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return ret;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_set);
>> +
>> +/**
>> + * icc_get() - return a handle for path between two endpoints
>> + * @dev: the device requesting the path
>> + * @src_id: source device port id
>> + * @dst_id: destination device port id
>> + *
>> + * This function will search for a path between two endpoints and return an
>> + * icc_path handle on success. Use icc_put() to release
>> + * constraints when the they are not needed anymore.
>> + *
>> + * Return: icc_path pointer on success, or ERR_PTR() on error
>> + */
>> +struct icc_path *icc_get(struct device *dev, const int src_id, const int dst_id)
>> +{
>> +       struct icc_node *src, *dst;
>> +       struct icc_path *path = ERR_PTR(-EPROBE_DEFER);
>> +
>> +       src = node_find(src_id);
>> +       if (!src) {
>> +               dev_err(dev, "%s: invalid src=%d\n", __func__, src_id);
>> +               goto out;
>> +       }
>> +
>> +       dst = node_find(dst_id);
>> +       if (!dst) {
>> +               dev_err(dev, "%s: invalid dst=%d\n", __func__, dst_id);
>> +               goto out;
>> +       }
>> +
>> +       mutex_lock(&icc_lock);
>> +       path = path_find(dev, src, dst);
>> +       mutex_unlock(&icc_lock);
>> +       if (IS_ERR(path)) {
>> +               dev_err(dev, "%s: invalid path=%ld\n", __func__, PTR_ERR(path));
>> +               goto out;
>> +       }
>> +
>> +out:
>> +       return path;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_get);
>> +
>> +/**
>> + * icc_put() - release the reference to the icc_path
>> + * @path: interconnect path
>> + *
>> + * Use this function to release the constraints on a path when the path is
>> + * no longer needed. The constraints will be re-aggregated.
>> + */
>> +void icc_put(struct icc_path *path)
>> +{
>> +       struct icc_node *node;
>> +       size_t i;
>> +       int ret;
>> +
>> +       if (!path || WARN_ON_ONCE(IS_ERR(path)))
>
> Why only once?
>
>> +               return;
>> +
>> +       ret = icc_set(path, 0, 0);
>> +       if (ret)
>> +               pr_err("%s: error (%d)\n", __func__, ret);
>> +
>> +       mutex_lock(&icc_lock);
>> +       for (i = 0; i < path->num_nodes; i++) {
>> +               node = path->reqs[i].node;
>> +               hlist_del(&path->reqs[i].req_node);
>> +
>> +               node->provider->users--;
>> +       }
>> +       mutex_unlock(&icc_lock);
>> +
>> +       kfree(path);
>> +}
>> +EXPORT_SYMBOL_GPL(icc_put);
>> +
>> +/**
>> + * icc_node_create() - create a node
>> + * @id: node id
>> + *
>> + * Return: icc_node pointer on success, or ERR_PTR() on error
>> + */
>> +struct icc_node *icc_node_create(int id)
>> +{
>> +       struct icc_node *node;
>> +
>> +       /* check if node already exists */
>> +       node = node_find(id);
>> +       if (node)
>> +               goto out;
>> +
>> +       node = kzalloc(sizeof(*node), GFP_KERNEL);
>> +       if (!node) {
>> +               node = ERR_PTR(-ENOMEM);
>> +               goto out;
>> +       }
>> +
>> +       mutex_lock(&icc_lock);
>> +
>> +       id = idr_alloc(&icc_idr, node, id, id + 1, GFP_KERNEL);
>> +       if (WARN(id < 0, "couldn't get idr")) {
>> +               node = ERR_PTR(id);
>> +               goto out;
>> +       }
>> +
>> +       node->id = id;
>> +
>> +out:
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return node;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_node_create);
>> +
>> +/**
>> + * icc_node_remove() - remove a node
>> + * @id: node id
>> + *
>> + */
>> +void icc_node_remove(int id)
>> +{
>> +       struct icc_node *node;
>> +
>> +       node = node_find(id);
>> +       if (node) {
>> +               mutex_lock(&icc_lock);
>> +               idr_remove(&icc_idr, node->id);
>
> Should we throw a warning if there are any paths that go through this
> node (ie req_list is non-empty)?
>
>> +               mutex_unlock(&icc_lock);
>> +       }
>> +
>> +       kfree(node);
>> +}
>> +EXPORT_SYMBOL_GPL(icc_node_remove);
>> +
>> +/**
>> + * icc_link_create() - create a link between two nodes
>> + * @src_id: source node id
>> + * @dst_id: destination node id
>> + *
>> + * Create a link between two nodes. The nodes might belong to different
>> + * interconnect providers and the @dst_id node might not exist (if the
>> + * provider driver has not probed yet). So just create the @dst_id node
>> + * and when the actual provider driver is probed, the rest of the node
>> + * data is filled.
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_link_create(struct icc_node *node, const int dst_id)
>> +{
>> +       struct icc_node *dst;
>> +       struct icc_node **new;
>> +       int ret = 0;
>> +
>> +       if (!node->provider)
>> +               return -EINVAL;
>> +
>> +       dst = node_find(dst_id);
>> +       if (!dst) {
>> +               dst = icc_node_create(dst_id);
>> +
>> +               if (IS_ERR(dst)) {
>> +                       ret = PTR_ERR(dst);
>> +                       goto out;
>> +               }
>> +       }
>> +
>> +       mutex_lock(&icc_lock);
>> +
>> +       new = krealloc(node->links,
>> +                      (node->num_links + 1) * sizeof(*node->links),
>> +                      GFP_KERNEL);
>> +       if (!new) {
>> +               ret = -ENOMEM;
>> +               goto out;
>> +       }
>> +
>> +       node->links = new;
>> +       node->links[node->num_links++] = dst;
>> +
>> +out:
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return ret;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_link_create);
>> +
>> +/**
>> + * icc_link_remove() - remove a link between two nodes
>> + * @src: pointer to source node
>> + * @dst: pointer to destination node
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_link_remove(struct icc_node *src, struct icc_node *dst)
>> +{
>> +       struct icc_node **new;
>> +       int ret = 0;
>> +       int i, j;
>> +
>> +       if (IS_ERR_OR_NULL(src))
>> +               return PTR_ERR(src);
>> +
>> +       if (IS_ERR_OR_NULL(dst))
>> +               return PTR_ERR(dst);
>
> I wonder if we should return a fixed error in these cases like
> -EINVAL, rather than handing through whatever crazy value is in
> src/dst.
>
>> +
>> +       mutex_lock(&icc_lock);
>> +
>> +       new = krealloc(src->links,
>> +                      (src->num_links - 1) * sizeof(*src->links),
>> +                      GFP_KERNEL);
>> +       if (!new) {
>> +               ret = -ENOMEM;
>> +               goto out;
>> +       }
>> +
>> +       for (i = 0, j = 0; j < src->num_links; j++) {
>> +               if (src->links[j] == dst)
>> +                       continue;
>> +
>> +               new[i++] = src->links[j];
>> +       }
>> +
>> +       src->links = new;
>> +       src->num_links--;
>
> My understanding is that once you call realloc and it succeeds, you
> must assume your old memory is gone and your new memory is only as big
> as the new size you request. So you shouldn't call krealloc until
> you've fixed the array up. Is the order of the links array important?
> If not, you could just take the element at the end and stick it in the
> slot that's being deleted. Then decrease the size and do your realloc.
>
>> +
>> +out:
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return ret;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_link_remove);
>> +
>> +/**
>> + * icc_node_add() - add an interconnect node to interconnect provider
>> + * @node: pointer to the interconnect node
>> + * @provider: pointer to the interconnect provider
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_node_add(struct icc_node *node, struct icc_provider *provider)
>> +{
>> +       mutex_lock(&icc_lock);
>> +
>> +       node->provider = provider;
>> +       list_add(&node->node_list, &provider->nodes);
>> +
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return 0;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_node_add);
>> +
>> +/**
>> + * icc_provider_add() - add a new interconnect provider
>> + * @icc_provider: the interconnect provider that will be added into topology
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_provider_add(struct icc_provider *provider)
>> +{
>> +       if (WARN_ON(!provider->set))
>> +               return -EINVAL;
>> +
>> +       mutex_init(&icc_lock);
>> +
>> +       INIT_LIST_HEAD(&provider->nodes);
>> +       list_add(&provider->provider_list, &icc_provider_list);
>> +
>> +       mutex_unlock(&icc_lock);
>> +
>> +       dev_dbg(provider->dev, "interconnect provider added to topology\n");
>> +
>> +       return 0;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_provider_add);
>> +
>> +/**
>> + * icc_provider_del() - delete previously added interconnect provider
>> + * @icc_provider: the interconnect provider that will be removed from topology
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_provider_del(struct icc_provider *provider)
>> +{
>> +       mutex_lock(&icc_lock);
>> +       if (provider->users) {
>> +               pr_warn("interconnect provider still has %d users\n",
>> +                       provider->users);
>> +               mutex_unlock(&icc_lock);
>> +               return -EBUSY;
>> +       }
>> +
>> +       if (!list_empty_careful(&provider->nodes)) {
>> +               pr_warn("interconnect provider still has nodes\n");
>> +               mutex_unlock(&icc_lock);
>> +               return -EEXIST;
>
> How come you're returning different error codes for these two cases?
> The error in both cases is effectively "you failed to clean up after
> yourself", so maybe EBUSY makes sense for both of them. The pr_warn
> helps to differentiate between the two for debugging.
>
>> +       }
>> +
>> +       list_del(&provider->provider_list);
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return 0;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_provider_del);
>> +
>> +MODULE_AUTHOR("Georgi Djakov <georgi.djakov@linaro.org");
>> +MODULE_DESCRIPTION("Interconnect Driver Core");
>> +MODULE_LICENSE("GPL v2");
>> diff --git a/include/linux/interconnect-provider.h b/include/linux/interconnect-provider.h
>> new file mode 100644
>> index 000000000000..f4613c6dce4f
>> --- /dev/null
>> +++ b/include/linux/interconnect-provider.h
>> @@ -0,0 +1,127 @@
>> +/* SPDX-License-Identifier: GPL-2.0 */
>> +/*
>> + * Copyright (c) 2018, Linaro Ltd.
>> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
>> + */
>> +
>> +#ifndef _LINUX_INTERCONNECT_PROVIDER_H
>> +#define _LINUX_INTERCONNECT_PROVIDER_H
>> +
>> +#include <linux/interconnect.h>
>> +
>> +#define icc_units_to_bps(bw)  ((bw) * 1000ULL)
>> +
>> +struct icc_node;
>> +
>> +/**
>> + * struct icc_provider - interconnect provider (controller) entity that might
>> + * provide multiple interconnect controls
>> + *
>> + * @provider_list: list of the registered interconnect providers
>> + * @nodes: internal list of the interconnect provider nodes
>> + * @set: pointer to device specific set operation function
>> + * @aggregate: pointer to device specific aggregate operation function
>> + * @dev: the device this interconnect provider belongs to
>> + * @users: count of active users
>> + * @avg_bw: aggregated value of average bandwidth requests from all nodes
>> + * @peak_bw: aggregated value of peak bandwidth requests from all nodes
>> + * @data: pointer to private data
>> + */
>> +struct icc_provider {
>> +       struct list_head        provider_list;
>> +       struct list_head        nodes;
>> +       int (*set)(struct icc_node *src, struct icc_node *dst,
>> +                  u32 avg_bw, u32 peak_bw);
>> +       int (*aggregate)(struct icc_node *node, u32 avg_bw, u32 peak_bw,
>> +                        u32 *agg_avg, u32 *agg_peak);
>> +       struct device           *dev;
>> +       int                     users;
>> +       u32                     avg_bw;
>> +       u32                     peak_bw;
>> +       void                    *data;
>> +};
>> +
>> +/**
>> + * struct icc_node - entity that is part of the interconnect topology
>> + *
>> + * @id: platform specific node id
>> + * @name: node name used in debugfs
>> + * @links: a list of targets pointing to where we can go next when traversing
>> + * @num_links: number of links to other interconnect nodes
>> + * @provider: points to the interconnect provider of this node
>> + * @node_list: the list entry in the parent provider's "nodes" list
>> + * @search_list: list used when walking the nodes graph
>> + * @reverse: pointer to previous node when walking the nodes graph
>> + * @is_traversed: flag that is used when walking the nodes graph
>> + * @req_list: a list of QoS constraint requests associated with this node
>> + * @avg_bw: aggregated value of average bandwidth requests from all consumers
>> + * @peak_bw: aggregated value of peak bandwidth requests from all consumers
>> + * @data: pointer to private data
>> + */
>> +struct icc_node {
>> +       int                     id;
>> +       const char              *name;
>> +       struct icc_node         **links;
>> +       size_t                  num_links;
>> +
>> +       struct icc_provider     *provider;
>> +       struct list_head        node_list;
>> +       struct list_head        orphan_list;
>
> Is this used?
>
>> +       struct list_head        search_list;
>> +       struct icc_node         *reverse;
>> +       bool                    is_traversed;
>> +       struct hlist_head       req_list;
>> +       u32                     avg_bw;
>> +       u32                     peak_bw;
>> +       void                    *data;
>> +};
>> +
>> +#if IS_ENABLED(CONFIG_INTERCONNECT)
>> +
>> +struct icc_node *icc_node_create(int id);
>> +void icc_node_remove(int id);
>> +int icc_link_create(struct icc_node *node, const int dst_id);
>> +int icc_link_remove(struct icc_node *src, struct icc_node *dst);
>> +int icc_node_add(struct icc_node *node, struct icc_provider *provider);
>> +int icc_provider_add(struct icc_provider *provider);
>> +int icc_provider_del(struct icc_provider *provider);
>> +
>> +#else
>> +
>> +static inline struct icc_node *icc_node_create(int id)
>> +{
>> +       return ERR_PTR(-ENOTSUPP);
>> +}
>> +
>> +void icc_node_remove(int id)
>> +{
>> +}
>> +
>> +static inline int icc_link_create(struct icc_node *node, const int dst_id)
>> +{
>> +       return -ENOTSUPP;
>> +}
>> +
>> +int icc_link_remove(struct icc_node *src, struct icc_node *dst)
>> +{
>> +       return -ENOTSUPP;
>> +}
>> +
>> +int icc_node_add(struct icc_node *node, struct icc_provider *provider)
>> +{
>> +       return -ENOTSUPP;
>> +}
>> +
>> +static inline int icc_provider_add(struct icc_provider *provider)
>> +{
>> +       return -ENOTSUPP;
>> +}
>> +
>> +static inline int icc_provider_del(struct icc_provider *provider)
>> +{
>> +       return -ENOTSUPP;
>> +}
>> +
>> +#endif /* CONFIG_INTERCONNECT */
>> +
>> +#endif /* _LINUX_INTERCONNECT_PROVIDER_H */
>> diff --git a/include/linux/interconnect.h b/include/linux/interconnect.h
>> new file mode 100644
>> index 000000000000..593215371fd6
>> --- /dev/null
>> +++ b/include/linux/interconnect.h
>> @@ -0,0 +1,42 @@
>> +/* SPDX-License-Identifier: GPL-2.0 */
>> +/*
>> + * Copyright (c) 2018, Linaro Ltd.
>> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
>> + */
>> +
>> +#ifndef _LINUX_INTERCONNECT_H
>> +#define _LINUX_INTERCONNECT_H
>> +
>> +#include <linux/mutex.h>
>> +#include <linux/types.h>
>> +
>> +struct icc_path;
>> +struct device;
>> +
>> +#if IS_ENABLED(CONFIG_INTERCONNECT)
>> +
>> +struct icc_path *icc_get(struct device *dev, const int src_id,
>> +                        const int dst_id);
>> +void icc_put(struct icc_path *path);
>> +int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw);
>> +
>> +#else
>> +
>> +static inline struct icc_path *icc_get(struct device *dev, const int src_id,
>> +                                      const int dst_id)
>> +{
>> +       return NULL;
>> +}
>> +
>> +static inline void icc_put(struct icc_path *path)
>> +{
>> +}
>> +
>> +static inline int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
>> +{
>> +       return 0;
>
> I was originally going to suggest that this should return a failure.
> Then I talked myself out of it, saying that if the interconnect
> framework is not compiled in, then clients should assume all their bus
> needs are already met. I guess this is the correct assumption?
>
>> +}
>> +
>> +#endif /* CONFIG_INTERCONNECT */
>> +
>> +#endif /* _LINUX_INTERCONNECT_H */

Re: [PATCH v5 4/8] interconnect: qcom: Add RPM communication

[Matthias Kaehlcke]

On Wed, Jun 20, 2018 at 03:11:37PM +0300, Georgi Djakov wrote:
> On some Qualcomm SoCs, there is a remote processor, which controls some of
> the Network-On-Chip interconnect resources. Other CPUs express their needs
> by communicating with this processor. Add a driver to handle comminication

s/comminication/communication/

> with this remote processor.
> 
> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
> ---
>  .../bindings/interconnect/qcom-smd.txt        | 32 +++++++
>  drivers/interconnect/qcom/Makefile            |  2 +
>  drivers/interconnect/qcom/smd-rpm.c           | 90 +++++++++++++++++++
>  drivers/interconnect/qcom/smd-rpm.h           | 15 ++++
>  4 files changed, 139 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/interconnect/qcom-smd.txt
>  create mode 100644 drivers/interconnect/qcom/Makefile
>  create mode 100644 drivers/interconnect/qcom/smd-rpm.c
>  create mode 100644 drivers/interconnect/qcom/smd-rpm.h
> 
> diff --git a/Documentation/devicetree/bindings/interconnect/qcom-smd.txt b/Documentation/devicetree/bindings/interconnect/qcom-smd.txt
> new file mode 100644
> index 000000000000..88a5aeb50935
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/interconnect/qcom-smd.txt
> @@ -0,0 +1,32 @@
> +Qualcomm SMD-RPM interconnect driver binding
> +------------------------------------------------
> +The RPM (Resource Power Manager) is a dedicated hardware engine
> +for managing the shared SoC resources in order to keep the lowest
> +power profile. It communicates with other hardware subsystems via
> +the shared memory driver (SMD) back-end and accepts requests for
> +various resources.
> +
> +Required properties :
> +- compatible : shall contain only one of the following:
> +			"qcom,interconnect-smd-rpm"
> +
> +Example:
> +	smd {
> +		compatible = "qcom,smd";
> +
> +		rpm {
> +			interrupts = <0 168 1>;
> +			qcom,ipc = <&apcs 8 0>;
> +			qcom,smd-edge = <15>;
> +
> +			rpm_requests {
> +				compatible = "qcom,rpm-msm8916";
> +				qcom,smd-channels = "rpm_requests";
> +
> +				interconnect-smd-rpm {
> +					compatible = "qcom,interconnect-smd-rpm";
> +				};
> +
> +			};
> +		};
> +	};
> diff --git a/drivers/interconnect/qcom/Makefile b/drivers/interconnect/qcom/Makefile
> new file mode 100644
> index 000000000000..2a0c41db91c4
> --- /dev/null
> +++ b/drivers/interconnect/qcom/Makefile
> @@ -0,0 +1,2 @@
> +# SPDX-License-Identifier: GPL-2.0
> +obj-y += smd-rpm.o
> diff --git a/drivers/interconnect/qcom/smd-rpm.c b/drivers/interconnect/qcom/smd-rpm.c
> new file mode 100644
> index 000000000000..0cf772f51642
> --- /dev/null
> +++ b/drivers/interconnect/qcom/smd-rpm.c
> @@ -0,0 +1,90 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * RPM over SMD communication wrapper for interconects

s/interconects/interconnects/

[PATCH v5 4/8] interconnect: qcom: Add RPM communication

[Matthias Kaehlcke]

On Wed, Jun 20, 2018 at 03:11:37PM +0300, Georgi Djakov wrote:
> On some Qualcomm SoCs, there is a remote processor, which controls some of
> the Network-On-Chip interconnect resources. Other CPUs express their needs
> by communicating with this processor. Add a driver to handle comminication

s/comminication/communication/

> with this remote processor.
> 
> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
> ---
>  .../bindings/interconnect/qcom-smd.txt        | 32 +++++++
>  drivers/interconnect/qcom/Makefile            |  2 +
>  drivers/interconnect/qcom/smd-rpm.c           | 90 +++++++++++++++++++
>  drivers/interconnect/qcom/smd-rpm.h           | 15 ++++
>  4 files changed, 139 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/interconnect/qcom-smd.txt
>  create mode 100644 drivers/interconnect/qcom/Makefile
>  create mode 100644 drivers/interconnect/qcom/smd-rpm.c
>  create mode 100644 drivers/interconnect/qcom/smd-rpm.h
> 
> diff --git a/Documentation/devicetree/bindings/interconnect/qcom-smd.txt b/Documentation/devicetree/bindings/interconnect/qcom-smd.txt
> new file mode 100644
> index 000000000000..88a5aeb50935
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/interconnect/qcom-smd.txt
> @@ -0,0 +1,32 @@
> +Qualcomm SMD-RPM interconnect driver binding
> +------------------------------------------------
> +The RPM (Resource Power Manager) is a dedicated hardware engine
> +for managing the shared SoC resources in order to keep the lowest
> +power profile. It communicates with other hardware subsystems via
> +the shared memory driver (SMD) back-end and accepts requests for
> +various resources.
> +
> +Required properties :
> +- compatible : shall contain only one of the following:
> +			"qcom,interconnect-smd-rpm"
> +
> +Example:
> +	smd {
> +		compatible = "qcom,smd";
> +
> +		rpm {
> +			interrupts = <0 168 1>;
> +			qcom,ipc = <&apcs 8 0>;
> +			qcom,smd-edge = <15>;
> +
> +			rpm_requests {
> +				compatible = "qcom,rpm-msm8916";
> +				qcom,smd-channels = "rpm_requests";
> +
> +				interconnect-smd-rpm {
> +					compatible = "qcom,interconnect-smd-rpm";
> +				};
> +
> +			};
> +		};
> +	};
> diff --git a/drivers/interconnect/qcom/Makefile b/drivers/interconnect/qcom/Makefile
> new file mode 100644
> index 000000000000..2a0c41db91c4
> --- /dev/null
> +++ b/drivers/interconnect/qcom/Makefile
> @@ -0,0 +1,2 @@
> +# SPDX-License-Identifier: GPL-2.0
> +obj-y += smd-rpm.o
> diff --git a/drivers/interconnect/qcom/smd-rpm.c b/drivers/interconnect/qcom/smd-rpm.c
> new file mode 100644
> index 000000000000..0cf772f51642
> --- /dev/null
> +++ b/drivers/interconnect/qcom/smd-rpm.c
> @@ -0,0 +1,90 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * RPM over SMD communication wrapper for interconects

s/interconects/interconnects/

Re: [PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Vincent Guittot]

Hi Georgi

On Wed, 20 Jun 2018 at 14:11, Georgi Djakov <georgi.djakov@linaro.org> wrote:

[snip]

> +
> +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
> +{
> +       struct icc_node *node = dst;
> +       struct icc_path *path;
> +       size_t i;
> +
> +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),

Should be (num_nodes -1) * sizeof(*path->reqs) as there is already 1
icc_req in icc_path struct

> +                      GFP_KERNEL);
> +       if (!path)
> +               return ERR_PTR(-ENOMEM);
> +
> +       path->num_nodes = num_nodes;
> +
> +       for (i = 0; i < num_nodes; i++) {
> +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
> +
> +               path->reqs[i].node = node;
> +               /* reference to previous node was saved during path traversal */
> +               node = node->reverse;
> +       }
> +
> +       return path;
> +}
> +

[snip]

[PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Vincent Guittot]

Hi Georgi

On Wed, 20 Jun 2018 at 14:11, Georgi Djakov <georgi.djakov@linaro.org> wrote:

[snip]

> +
> +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
> +{
> +       struct icc_node *node = dst;
> +       struct icc_path *path;
> +       size_t i;
> +
> +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),

Should be (num_nodes -1) * sizeof(*path->reqs) as there is already 1
icc_req in icc_path struct

> +                      GFP_KERNEL);
> +       if (!path)
> +               return ERR_PTR(-ENOMEM);
> +
> +       path->num_nodes = num_nodes;
> +
> +       for (i = 0; i < num_nodes; i++) {
> +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
> +
> +               path->reqs[i].node = node;
> +               /* reference to previous node was saved during path traversal */
> +               node = node->reverse;
> +       }
> +
> +       return path;
> +}
> +

[snip]

Re: [PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Georgi Djakov]

Hi Evan,

Thanks for reviewing!

On 06/26/2018 11:57 PM, Evan Green wrote:
> Hi Georgi. Thanks for the new spin of this.
> 
> On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>>
>> This patch introduce a new API to get requirements and configure the
>> interconnect buses across the entire chipset to fit with the current
>> demand.
>>
>> The API is using a consumer/provider-based model, where the providers are
>> the interconnect buses and the consumers could be various drivers.
>> The consumers request interconnect resources (path) between endpoints and
>> set the desired constraints on this data flow path. The providers receive
>> requests from consumers and aggregate these requests for all master-slave
>> pairs on that path. Then the providers configure each participating in the
>> topology node according to the requested data flow path, physical links and
>> constraints. The topology could be complicated and multi-tiered and is SoC
>> specific.

[..]

>> +
>> +static struct icc_node *node_find(const int id)
>> +{
>> +       struct icc_node *node;
>> +
>> +       mutex_lock(&icc_lock);
>> +       node = idr_find(&icc_idr, id);
>> +       mutex_unlock(&icc_lock);
> 
> I wonder if this is too low of a level to be dealing with the lock. I
> notice that everywhere you use this function, you afterwards
> immediately grab the lock and do more stuff. Maybe this function
> should have a comment saying it assumes the lock is already held, and
> then you can grab the lock in the callers, since you're doing that
> anyway.

Ok, will try to do better next time.

>> +
>> +       return node;
>> +}
>> +
>> +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
>> +{
>> +       struct icc_node *node = dst;
>> +       struct icc_path *path;
>> +       size_t i;
>> +
>> +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
>> +                      GFP_KERNEL);
>> +       if (!path)
>> +               return ERR_PTR(-ENOMEM);
>> +
>> +       path->num_nodes = num_nodes;
>> +
>> +       for (i = 0; i < num_nodes; i++) {
>> +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
>> +
>> +               path->reqs[i].node = node;
>> +               /* reference to previous node was saved during path traversal */
>> +               node = node->reverse;
>> +       }
>> +
>> +       return path;
>> +}
>> +
>> +static struct icc_path *path_find(struct device *dev, struct icc_node *src,
>> +                                 struct icc_node *dst)
>> +{
> 
> I personally prefer a comment somewhere indicating that this function
> assumes icc_lock is already held. Not sure if that's conventional or
> not.
> 

Right, Rob and Matthias have also provided useful feedback on this! Thanks!

[..]

>> +       /* reset the traversed state */
>> +       list_for_each_entry(provider, &icc_provider_list, provider_list) {
>> +               list_for_each_entry(n, &provider->nodes, node_list)
>> +                       if (n->is_traversed)
>> +                               n->is_traversed = false;
> 
> Remove the conditional, just set is_traversed to false.
> 

Agree.

>> +       }
>> +
>> +       if (found) {
>> +               struct icc_path *path = path_allocate(dst, depth);
> 
> Is the path supposed to include the source? For instance, if the dst
> were a neighbor, depth would be one, so only dst would be in the path.
> It seems like it might be worthwhile to have the source in there too.

Agree that it would be logical to include the complete path. Will fix
the depth.

>> +
>> +               if (IS_ERR(path))
>> +                       return path;
>> +
>> +               /* initialize the path */
>> +               for (i = 0; i < path->num_nodes; i++) {
>> +                       node = path->reqs[i].node;
>> +                       path->reqs[i].dev = dev;
>> +                       node->provider->users++;
> 
> Should this loop live inside path_allocate? I'm unsure, but maybe at
> least path->reqs[i].dev = dev, since it feels like standard
> initialization of the path.
> 

Ok, will move it and also change the function name from path_allocate()
to path_init().

>> +
>> +/**
>> + * icc_put() - release the reference to the icc_path
>> + * @path: interconnect path
>> + *
>> + * Use this function to release the constraints on a path when the path is
>> + * no longer needed. The constraints will be re-aggregated.
>> + */
>> +void icc_put(struct icc_path *path)
>> +{
>> +       struct icc_node *node;
>> +       size_t i;
>> +       int ret;
>> +
>> +       if (!path || WARN_ON_ONCE(IS_ERR(path)))
> 
> Why only once?
> 

Will change to WARN_ON.

[..]

>> +void icc_node_remove(int id)
>> +{
>> +       struct icc_node *node;
>> +
>> +       node = node_find(id);
>> +       if (node) {
>> +               mutex_lock(&icc_lock);
>> +               idr_remove(&icc_idr, node->id);
> 
> Should we throw a warning if there are any paths that go through this
> node (ie req_list is non-empty)?
> 

Sounds good, will do it.
>> +               mutex_unlock(&icc_lock);
>> +       }
>> +
>> +       kfree(node);
>> +}
>> +EXPORT_SYMBOL_GPL(icc_node_remove);

[..]

>> +/**
>> + * icc_link_remove() - remove a link between two nodes
>> + * @src: pointer to source node
>> + * @dst: pointer to destination node
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_link_remove(struct icc_node *src, struct icc_node *dst)
>> +{
>> +       struct icc_node **new;
>> +       int ret = 0;
>> +       int i, j;
>> +
>> +       if (IS_ERR_OR_NULL(src))
>> +               return PTR_ERR(src);
>> +
>> +       if (IS_ERR_OR_NULL(dst))
>> +               return PTR_ERR(dst);
> 
> I wonder if we should return a fixed error in these cases like
> -EINVAL, rather than handing through whatever crazy value is in
> src/dst.

Ok, agree.

>> +
>> +       mutex_lock(&icc_lock);
>> +
>> +       new = krealloc(src->links,
>> +                      (src->num_links - 1) * sizeof(*src->links),
>> +                      GFP_KERNEL);
>> +       if (!new) {
>> +               ret = -ENOMEM;
>> +               goto out;
>> +       }
>> +
>> +       for (i = 0, j = 0; j < src->num_links; j++) {
>> +               if (src->links[j] == dst)
>> +                       continue;
>> +
>> +               new[i++] = src->links[j];
>> +       }
>> +
>> +       src->links = new;
>> +       src->num_links--;
> 
> My understanding is that once you call realloc and it succeeds, you
> must assume your old memory is gone and your new memory is only as big
> as the new size you request. So you shouldn't call krealloc until
> you've fixed the array up. Is the order of the links array important?
> If not, you could just take the element at the end and stick it in the
> slot that's being deleted. Then decrease the size and do your realloc.

Sorry, this was obviously untested. Your suggestions is good.

[..]

>> +
>> +/**
>> + * icc_provider_del() - delete previously added interconnect provider
>> + * @icc_provider: the interconnect provider that will be removed from topology
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_provider_del(struct icc_provider *provider)
>> +{
>> +       mutex_lock(&icc_lock);
>> +       if (provider->users) {
>> +               pr_warn("interconnect provider still has %d users\n",
>> +                       provider->users);
>> +               mutex_unlock(&icc_lock);
>> +               return -EBUSY;
>> +       }
>> +
>> +       if (!list_empty_careful(&provider->nodes)) {
>> +               pr_warn("interconnect provider still has nodes\n");
>> +               mutex_unlock(&icc_lock);
>> +               return -EEXIST;
> 
> How come you're returning different error codes for these two cases?
> The error in both cases is effectively "you failed to clean up after
> yourself", so maybe EBUSY makes sense for both of them. The pr_warn
> helps to differentiate between the two for debugging.

Ok.

>> +       }
>> +
>> +       list_del(&provider->provider_list);
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return 0;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_provider_del);
>> +

[..]

>> +struct icc_node {
>> +       int                     id;
>> +       const char              *name;
>> +       struct icc_node         **links;
>> +       size_t                  num_links;
>> +
>> +       struct icc_provider     *provider;
>> +       struct list_head        node_list;
>> +       struct list_head        orphan_list;
> 
> Is this used?

Ah, I thought I had already removed it!

>> +       struct list_head        search_list;
>> +       struct icc_node         *reverse;
>> +       bool                    is_traversed;
>> +       struct hlist_head       req_list;
>> +       u32                     avg_bw;
>> +       u32                     peak_bw;
>> +       void                    *data;
>> +};
>> +

[..]

>> +static inline int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
>> +{
>> +       return 0;
> 
> I was originally going to suggest that this should return a failure.
> Then I talked myself out of it, saying that if the interconnect
> framework is not compiled in, then clients should assume all their bus
> needs are already met. I guess this is the correct assumption?

Yes, exactly!

Thanks,
Georgi

Re: [PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Georgi Djakov]

Hi Evan,

Thanks for reviewing!

On 06/26/2018 11:57 PM, Evan Green wrote:
> Hi Georgi. Thanks for the new spin of this.
> 
> On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>>
>> This patch introduce a new API to get requirements and configure the
>> interconnect buses across the entire chipset to fit with the current
>> demand.
>>
>> The API is using a consumer/provider-based model, where the providers are
>> the interconnect buses and the consumers could be various drivers.
>> The consumers request interconnect resources (path) between endpoints and
>> set the desired constraints on this data flow path. The providers receive
>> requests from consumers and aggregate these requests for all master-slave
>> pairs on that path. Then the providers configure each participating in the
>> topology node according to the requested data flow path, physical links and
>> constraints. The topology could be complicated and multi-tiered and is SoC
>> specific.

[..]

>> +
>> +static struct icc_node *node_find(const int id)
>> +{
>> +       struct icc_node *node;
>> +
>> +       mutex_lock(&icc_lock);
>> +       node = idr_find(&icc_idr, id);
>> +       mutex_unlock(&icc_lock);
> 
> I wonder if this is too low of a level to be dealing with the lock. I
> notice that everywhere you use this function, you afterwards
> immediately grab the lock and do more stuff. Maybe this function
> should have a comment saying it assumes the lock is already held, and
> then you can grab the lock in the callers, since you're doing that
> anyway.

Ok, will try to do better next time.

>> +
>> +       return node;
>> +}
>> +
>> +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
>> +{
>> +       struct icc_node *node = dst;
>> +       struct icc_path *path;
>> +       size_t i;
>> +
>> +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
>> +                      GFP_KERNEL);
>> +       if (!path)
>> +               return ERR_PTR(-ENOMEM);
>> +
>> +       path->num_nodes = num_nodes;
>> +
>> +       for (i = 0; i < num_nodes; i++) {
>> +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
>> +
>> +               path->reqs[i].node = node;
>> +               /* reference to previous node was saved during path traversal */
>> +               node = node->reverse;
>> +       }
>> +
>> +       return path;
>> +}
>> +
>> +static struct icc_path *path_find(struct device *dev, struct icc_node *src,
>> +                                 struct icc_node *dst)
>> +{
> 
> I personally prefer a comment somewhere indicating that this function
> assumes icc_lock is already held. Not sure if that's conventional or
> not.
> 

Right, Rob and Matthias have also provided useful feedback on this! Thanks!

[..]

>> +       /* reset the traversed state */
>> +       list_for_each_entry(provider, &icc_provider_list, provider_list) {
>> +               list_for_each_entry(n, &provider->nodes, node_list)
>> +                       if (n->is_traversed)
>> +                               n->is_traversed = false;
> 
> Remove the conditional, just set is_traversed to false.
> 

Agree.

>> +       }
>> +
>> +       if (found) {
>> +               struct icc_path *path = path_allocate(dst, depth);
> 
> Is the path supposed to include the source? For instance, if the dst
> were a neighbor, depth would be one, so only dst would be in the path.
> It seems like it might be worthwhile to have the source in there too.

Agree that it would be logical to include the complete path. Will fix
the depth.

>> +
>> +               if (IS_ERR(path))
>> +                       return path;
>> +
>> +               /* initialize the path */
>> +               for (i = 0; i < path->num_nodes; i++) {
>> +                       node = path->reqs[i].node;
>> +                       path->reqs[i].dev = dev;
>> +                       node->provider->users++;
> 
> Should this loop live inside path_allocate? I'm unsure, but maybe at
> least path->reqs[i].dev = dev, since it feels like standard
> initialization of the path.
> 

Ok, will move it and also change the function name from path_allocate()
to path_init().

>> +
>> +/**
>> + * icc_put() - release the reference to the icc_path
>> + * @path: interconnect path
>> + *
>> + * Use this function to release the constraints on a path when the path is
>> + * no longer needed. The constraints will be re-aggregated.
>> + */
>> +void icc_put(struct icc_path *path)
>> +{
>> +       struct icc_node *node;
>> +       size_t i;
>> +       int ret;
>> +
>> +       if (!path || WARN_ON_ONCE(IS_ERR(path)))
> 
> Why only once?
> 

Will change to WARN_ON.

[..]

>> +void icc_node_remove(int id)
>> +{
>> +       struct icc_node *node;
>> +
>> +       node = node_find(id);
>> +       if (node) {
>> +               mutex_lock(&icc_lock);
>> +               idr_remove(&icc_idr, node->id);
> 
> Should we throw a warning if there are any paths that go through this
> node (ie req_list is non-empty)?
> 

Sounds good, will do it.
>> +               mutex_unlock(&icc_lock);
>> +       }
>> +
>> +       kfree(node);
>> +}
>> +EXPORT_SYMBOL_GPL(icc_node_remove);

[..]

>> +/**
>> + * icc_link_remove() - remove a link between two nodes
>> + * @src: pointer to source node
>> + * @dst: pointer to destination node
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_link_remove(struct icc_node *src, struct icc_node *dst)
>> +{
>> +       struct icc_node **new;
>> +       int ret = 0;
>> +       int i, j;
>> +
>> +       if (IS_ERR_OR_NULL(src))
>> +               return PTR_ERR(src);
>> +
>> +       if (IS_ERR_OR_NULL(dst))
>> +               return PTR_ERR(dst);
> 
> I wonder if we should return a fixed error in these cases like
> -EINVAL, rather than handing through whatever crazy value is in
> src/dst.

Ok, agree.

>> +
>> +       mutex_lock(&icc_lock);
>> +
>> +       new = krealloc(src->links,
>> +                      (src->num_links - 1) * sizeof(*src->links),
>> +                      GFP_KERNEL);
>> +       if (!new) {
>> +               ret = -ENOMEM;
>> +               goto out;
>> +       }
>> +
>> +       for (i = 0, j = 0; j < src->num_links; j++) {
>> +               if (src->links[j] == dst)
>> +                       continue;
>> +
>> +               new[i++] = src->links[j];
>> +       }
>> +
>> +       src->links = new;
>> +       src->num_links--;
> 
> My understanding is that once you call realloc and it succeeds, you
> must assume your old memory is gone and your new memory is only as big
> as the new size you request. So you shouldn't call krealloc until
> you've fixed the array up. Is the order of the links array important?
> If not, you could just take the element at the end and stick it in the
> slot that's being deleted. Then decrease the size and do your realloc.

Sorry, this was obviously untested. Your suggestions is good.

[..]

>> +
>> +/**
>> + * icc_provider_del() - delete previously added interconnect provider
>> + * @icc_provider: the interconnect provider that will be removed from topology
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_provider_del(struct icc_provider *provider)
>> +{
>> +       mutex_lock(&icc_lock);
>> +       if (provider->users) {
>> +               pr_warn("interconnect provider still has %d users\n",
>> +                       provider->users);
>> +               mutex_unlock(&icc_lock);
>> +               return -EBUSY;
>> +       }
>> +
>> +       if (!list_empty_careful(&provider->nodes)) {
>> +               pr_warn("interconnect provider still has nodes\n");
>> +               mutex_unlock(&icc_lock);
>> +               return -EEXIST;
> 
> How come you're returning different error codes for these two cases?
> The error in both cases is effectively "you failed to clean up after
> yourself", so maybe EBUSY makes sense for both of them. The pr_warn
> helps to differentiate between the two for debugging.

Ok.

>> +       }
>> +
>> +       list_del(&provider->provider_list);
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return 0;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_provider_del);
>> +

[..]

>> +struct icc_node {
>> +       int                     id;
>> +       const char              *name;
>> +       struct icc_node         **links;
>> +       size_t                  num_links;
>> +
>> +       struct icc_provider     *provider;
>> +       struct list_head        node_list;
>> +       struct list_head        orphan_list;
> 
> Is this used?

Ah, I thought I had already removed it!

>> +       struct list_head        search_list;
>> +       struct icc_node         *reverse;
>> +       bool                    is_traversed;
>> +       struct hlist_head       req_list;
>> +       u32                     avg_bw;
>> +       u32                     peak_bw;
>> +       void                    *data;
>> +};
>> +

[..]

>> +static inline int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
>> +{
>> +       return 0;
> 
> I was originally going to suggest that this should return a failure.
> Then I talked myself out of it, saying that if the interconnect
> framework is not compiled in, then clients should assume all their bus
> needs are already met. I guess this is the correct assumption?

Yes, exactly!

Thanks,
Georgi

[PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Georgi Djakov]

Hi Evan,

Thanks for reviewing!

On 06/26/2018 11:57 PM, Evan Green wrote:
> Hi Georgi. Thanks for the new spin of this.
> 
> On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>>
>> This patch introduce a new API to get requirements and configure the
>> interconnect buses across the entire chipset to fit with the current
>> demand.
>>
>> The API is using a consumer/provider-based model, where the providers are
>> the interconnect buses and the consumers could be various drivers.
>> The consumers request interconnect resources (path) between endpoints and
>> set the desired constraints on this data flow path. The providers receive
>> requests from consumers and aggregate these requests for all master-slave
>> pairs on that path. Then the providers configure each participating in the
>> topology node according to the requested data flow path, physical links and
>> constraints. The topology could be complicated and multi-tiered and is SoC
>> specific.

[..]

>> +
>> +static struct icc_node *node_find(const int id)
>> +{
>> +       struct icc_node *node;
>> +
>> +       mutex_lock(&icc_lock);
>> +       node = idr_find(&icc_idr, id);
>> +       mutex_unlock(&icc_lock);
> 
> I wonder if this is too low of a level to be dealing with the lock. I
> notice that everywhere you use this function, you afterwards
> immediately grab the lock and do more stuff. Maybe this function
> should have a comment saying it assumes the lock is already held, and
> then you can grab the lock in the callers, since you're doing that
> anyway.

Ok, will try to do better next time.

>> +
>> +       return node;
>> +}
>> +
>> +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
>> +{
>> +       struct icc_node *node = dst;
>> +       struct icc_path *path;
>> +       size_t i;
>> +
>> +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
>> +                      GFP_KERNEL);
>> +       if (!path)
>> +               return ERR_PTR(-ENOMEM);
>> +
>> +       path->num_nodes = num_nodes;
>> +
>> +       for (i = 0; i < num_nodes; i++) {
>> +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
>> +
>> +               path->reqs[i].node = node;
>> +               /* reference to previous node was saved during path traversal */
>> +               node = node->reverse;
>> +       }
>> +
>> +       return path;
>> +}
>> +
>> +static struct icc_path *path_find(struct device *dev, struct icc_node *src,
>> +                                 struct icc_node *dst)
>> +{
> 
> I personally prefer a comment somewhere indicating that this function
> assumes icc_lock is already held. Not sure if that's conventional or
> not.
> 

Right, Rob and Matthias have also provided useful feedback on this! Thanks!

[..]

>> +       /* reset the traversed state */
>> +       list_for_each_entry(provider, &icc_provider_list, provider_list) {
>> +               list_for_each_entry(n, &provider->nodes, node_list)
>> +                       if (n->is_traversed)
>> +                               n->is_traversed = false;
> 
> Remove the conditional, just set is_traversed to false.
> 

Agree.

>> +       }
>> +
>> +       if (found) {
>> +               struct icc_path *path = path_allocate(dst, depth);
> 
> Is the path supposed to include the source? For instance, if the dst
> were a neighbor, depth would be one, so only dst would be in the path.
> It seems like it might be worthwhile to have the source in there too.

Agree that it would be logical to include the complete path. Will fix
the depth.

>> +
>> +               if (IS_ERR(path))
>> +                       return path;
>> +
>> +               /* initialize the path */
>> +               for (i = 0; i < path->num_nodes; i++) {
>> +                       node = path->reqs[i].node;
>> +                       path->reqs[i].dev = dev;
>> +                       node->provider->users++;
> 
> Should this loop live inside path_allocate? I'm unsure, but maybe at
> least path->reqs[i].dev = dev, since it feels like standard
> initialization of the path.
> 

Ok, will move it and also change the function name from path_allocate()
to path_init().

>> +
>> +/**
>> + * icc_put() - release the reference to the icc_path
>> + * @path: interconnect path
>> + *
>> + * Use this function to release the constraints on a path when the path is
>> + * no longer needed. The constraints will be re-aggregated.
>> + */
>> +void icc_put(struct icc_path *path)
>> +{
>> +       struct icc_node *node;
>> +       size_t i;
>> +       int ret;
>> +
>> +       if (!path || WARN_ON_ONCE(IS_ERR(path)))
> 
> Why only once?
> 

Will change to WARN_ON.

[..]

>> +void icc_node_remove(int id)
>> +{
>> +       struct icc_node *node;
>> +
>> +       node = node_find(id);
>> +       if (node) {
>> +               mutex_lock(&icc_lock);
>> +               idr_remove(&icc_idr, node->id);
> 
> Should we throw a warning if there are any paths that go through this
> node (ie req_list is non-empty)?
> 

Sounds good, will do it.
>> +               mutex_unlock(&icc_lock);
>> +       }
>> +
>> +       kfree(node);
>> +}
>> +EXPORT_SYMBOL_GPL(icc_node_remove);

[..]

>> +/**
>> + * icc_link_remove() - remove a link between two nodes
>> + * @src: pointer to source node
>> + * @dst: pointer to destination node
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_link_remove(struct icc_node *src, struct icc_node *dst)
>> +{
>> +       struct icc_node **new;
>> +       int ret = 0;
>> +       int i, j;
>> +
>> +       if (IS_ERR_OR_NULL(src))
>> +               return PTR_ERR(src);
>> +
>> +       if (IS_ERR_OR_NULL(dst))
>> +               return PTR_ERR(dst);
> 
> I wonder if we should return a fixed error in these cases like
> -EINVAL, rather than handing through whatever crazy value is in
> src/dst.

Ok, agree.

>> +
>> +       mutex_lock(&icc_lock);
>> +
>> +       new = krealloc(src->links,
>> +                      (src->num_links - 1) * sizeof(*src->links),
>> +                      GFP_KERNEL);
>> +       if (!new) {
>> +               ret = -ENOMEM;
>> +               goto out;
>> +       }
>> +
>> +       for (i = 0, j = 0; j < src->num_links; j++) {
>> +               if (src->links[j] == dst)
>> +                       continue;
>> +
>> +               new[i++] = src->links[j];
>> +       }
>> +
>> +       src->links = new;
>> +       src->num_links--;
> 
> My understanding is that once you call realloc and it succeeds, you
> must assume your old memory is gone and your new memory is only as big
> as the new size you request. So you shouldn't call krealloc until
> you've fixed the array up. Is the order of the links array important?
> If not, you could just take the element at the end and stick it in the
> slot that's being deleted. Then decrease the size and do your realloc.

Sorry, this was obviously untested. Your suggestions is good.

[..]

>> +
>> +/**
>> + * icc_provider_del() - delete previously added interconnect provider
>> + * @icc_provider: the interconnect provider that will be removed from topology
>> + *
>> + * Return: 0 on success, or an error code otherwise
>> + */
>> +int icc_provider_del(struct icc_provider *provider)
>> +{
>> +       mutex_lock(&icc_lock);
>> +       if (provider->users) {
>> +               pr_warn("interconnect provider still has %d users\n",
>> +                       provider->users);
>> +               mutex_unlock(&icc_lock);
>> +               return -EBUSY;
>> +       }
>> +
>> +       if (!list_empty_careful(&provider->nodes)) {
>> +               pr_warn("interconnect provider still has nodes\n");
>> +               mutex_unlock(&icc_lock);
>> +               return -EEXIST;
> 
> How come you're returning different error codes for these two cases?
> The error in both cases is effectively "you failed to clean up after
> yourself", so maybe EBUSY makes sense for both of them. The pr_warn
> helps to differentiate between the two for debugging.

Ok.

>> +       }
>> +
>> +       list_del(&provider->provider_list);
>> +       mutex_unlock(&icc_lock);
>> +
>> +       return 0;
>> +}
>> +EXPORT_SYMBOL_GPL(icc_provider_del);
>> +

[..]

>> +struct icc_node {
>> +       int                     id;
>> +       const char              *name;
>> +       struct icc_node         **links;
>> +       size_t                  num_links;
>> +
>> +       struct icc_provider     *provider;
>> +       struct list_head        node_list;
>> +       struct list_head        orphan_list;
> 
> Is this used?

Ah, I thought I had already removed it!

>> +       struct list_head        search_list;
>> +       struct icc_node         *reverse;
>> +       bool                    is_traversed;
>> +       struct hlist_head       req_list;
>> +       u32                     avg_bw;
>> +       u32                     peak_bw;
>> +       void                    *data;
>> +};
>> +

[..]

>> +static inline int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
>> +{
>> +       return 0;
> 
> I was originally going to suggest that this should return a failure.
> Then I talked myself out of it, saying that if the interconnect
> framework is not compiled in, then clients should assume all their bus
> needs are already met. I guess this is the correct assumption?

Yes, exactly!

Thanks,
Georgi

Re: [PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Georgi Djakov]

Hi Matthias,

Thanks for reviewing!

On 06/27/2018 02:34 AM, Matthias Kaehlcke wrote:
> Hi Georgi,
> 
> On Wed, Jun 20, 2018 at 03:11:34PM +0300, Georgi Djakov wrote:
>> This patch introduce a new API to get requirements and configure the
> 
> nit: s/introduce/introduces/

Thanks!

[..]

>> +	if (found) {
>> +		struct icc_path *path = path_allocate(dst, depth);
>> +
>> +		if (IS_ERR(path))
>> +			return path;
>> +
>> +		/* initialize the path */
>> +		for (i = 0; i < path->num_nodes; i++) {
>> +			node = path->reqs[i].node;
>> +			path->reqs[i].dev = dev;
>> +			node->provider->users++;
> 
> nit: doing the assignment of path->reqs[i].dev before assiging 'node'
> or after incrementing the 'users' would slightly improve readability.

Ok, will re-factor this a bit.

>> +static int apply_constraints(struct icc_path *path)
>> +{
>> +	struct icc_node *next, *prev = NULL;
>> +	int ret = 0;
>> +	int i;
>> +
>> +	for (i = 0; i < path->num_nodes; i++, prev = next) {
>> +		struct icc_provider *p;
>> +
>> +		next = path->reqs[i].node;
>> +		/*
>> +		 * Both endpoints should be valid master-slave pairs of the
>> +		 * same interconnect provider that will be configured.
>> +		 */
>> +		if (!prev || next->provider != prev->provider)
>> +			continue;
>> +
>> +		p = next->provider;
>> +
>> +		aggregate_provider(p);
>> +
>> +		if (p->set) {
>> +			/* set the constraints */
>> +			ret = p->set(prev, next, p->avg_bw, p->peak_bw);
>> +		}
> 
> remove curly brackets
> 
> EDIT: actually the condition can be removed, icc_provider_add() fails
> when p->set is NULL.

Agree!

> 
>> +int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
>> +{
>> +	struct icc_node *node;
>> +	struct icc_provider *p;
>> +	size_t i;
>> +	int ret = 0;
> 
> initialization is not necessary
> 

Ok.

>> +struct icc_path *icc_get(struct device *dev, const int src_id, const int dst_id)
>> +{
>> +	struct icc_node *src, *dst;
>> +	struct icc_path *path = ERR_PTR(-EPROBE_DEFER);
>> +
>> +	src = node_find(src_id);
>> +	if (!src) {
>> +		dev_err(dev, "%s: invalid src=%d\n", __func__, src_id);
>> +		goto out;
>> +	}
>> +
>> +	dst = node_find(dst_id);
>> +	if (!dst) {
>> +		dev_err(dev, "%s: invalid dst=%d\n", __func__, dst_id);
>> +		goto out;
>> +	}
>> +
>> +	mutex_lock(&icc_lock);
>> +	path = path_find(dev, src, dst);
>> +	mutex_unlock(&icc_lock);
>> +	if (IS_ERR(path)) {
>> +		dev_err(dev, "%s: invalid path=%ld\n", __func__, PTR_ERR(path));
>> +		goto out;
> 
> this goto isn't really needed

Ok.

>> +struct icc_node *icc_node_create(int id)
>> +{
>> +	struct icc_node *node;
>> +
>> +	/* check if node already exists */
>> +	node = node_find(id);
>> +	if (node)
>> +		goto out;
>> +
>> +	node = kzalloc(sizeof(*node), GFP_KERNEL);
>> +	if (!node) {
>> +		node = ERR_PTR(-ENOMEM);
>> +		goto out;
>> +	}
>> +
>> +	mutex_lock(&icc_lock);
>> +
>> +	id = idr_alloc(&icc_idr, node, id, id + 1, GFP_KERNEL);
>> +	if (WARN(id < 0, "couldn't get idr")) {
> 
> kfree(node);

Thanks!

>> +int icc_node_add(struct icc_node *node, struct icc_provider *provider)
>> +{
>> +	mutex_lock(&icc_lock);
>> +
>> +	node->provider = provider;
>> +	list_add(&node->node_list, &provider->nodes);
>> +
>> +	mutex_unlock(&icc_lock);
>> +
>> +	return 0;
>> +}
> 
> The function returns always 0. Should probably be void so callers
> don't add pointless checks of the return value.

Agree, will change it!

>> +int icc_provider_add(struct icc_provider *provider)
>> +{
>> +	if (WARN_ON(!provider->set))
>> +		return -EINVAL;
>> +
>> +	mutex_init(&icc_lock);
> 
> Shouldn't this be mutex_lock()?

Yes, right!

>> +int icc_provider_del(struct icc_provider *provider)
>> +{
>> +	mutex_lock(&icc_lock);
>> +	if (provider->users) {
>> +		pr_warn("interconnect provider still has %d users\n",
>> +			provider->users);
>> +		mutex_unlock(&icc_lock);
>> +		return -EBUSY;
>> +	}
>> +
>> +	if (!list_empty_careful(&provider->nodes)) {
>> +		pr_warn("interconnect provider still has nodes\n");
>> +		mutex_unlock(&icc_lock);
>> +		return -EEXIST;
>> +	}
> 
> Could this be just list_empty()? If I didn't miss something icc_lock
> is held in all paths that change p->nodes (assuming that all changes
> should be done through the interfaces in this file).

It could be. Will update it. I just always want to be careful!

> Actually this check will always fail if icc_node_add() was called for
> this provider, it doesn't seem nodes are ever removed.
> 

The provider driver is responsible for the node removal.

Thanks,
Georgi

[PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Georgi Djakov]

Hi Matthias,

Thanks for reviewing!

On 06/27/2018 02:34 AM, Matthias Kaehlcke wrote:
> Hi Georgi,
> 
> On Wed, Jun 20, 2018 at 03:11:34PM +0300, Georgi Djakov wrote:
>> This patch introduce a new API to get requirements and configure the
> 
> nit: s/introduce/introduces/

Thanks!

[..]

>> +	if (found) {
>> +		struct icc_path *path = path_allocate(dst, depth);
>> +
>> +		if (IS_ERR(path))
>> +			return path;
>> +
>> +		/* initialize the path */
>> +		for (i = 0; i < path->num_nodes; i++) {
>> +			node = path->reqs[i].node;
>> +			path->reqs[i].dev = dev;
>> +			node->provider->users++;
> 
> nit: doing the assignment of path->reqs[i].dev before assiging 'node'
> or after incrementing the 'users' would slightly improve readability.

Ok, will re-factor this a bit.

>> +static int apply_constraints(struct icc_path *path)
>> +{
>> +	struct icc_node *next, *prev = NULL;
>> +	int ret = 0;
>> +	int i;
>> +
>> +	for (i = 0; i < path->num_nodes; i++, prev = next) {
>> +		struct icc_provider *p;
>> +
>> +		next = path->reqs[i].node;
>> +		/*
>> +		 * Both endpoints should be valid master-slave pairs of the
>> +		 * same interconnect provider that will be configured.
>> +		 */
>> +		if (!prev || next->provider != prev->provider)
>> +			continue;
>> +
>> +		p = next->provider;
>> +
>> +		aggregate_provider(p);
>> +
>> +		if (p->set) {
>> +			/* set the constraints */
>> +			ret = p->set(prev, next, p->avg_bw, p->peak_bw);
>> +		}
> 
> remove curly brackets
> 
> EDIT: actually the condition can be removed, icc_provider_add() fails
> when p->set is NULL.

Agree!

> 
>> +int icc_set(struct icc_path *path, u32 avg_bw, u32 peak_bw)
>> +{
>> +	struct icc_node *node;
>> +	struct icc_provider *p;
>> +	size_t i;
>> +	int ret = 0;
> 
> initialization is not necessary
> 

Ok.

>> +struct icc_path *icc_get(struct device *dev, const int src_id, const int dst_id)
>> +{
>> +	struct icc_node *src, *dst;
>> +	struct icc_path *path = ERR_PTR(-EPROBE_DEFER);
>> +
>> +	src = node_find(src_id);
>> +	if (!src) {
>> +		dev_err(dev, "%s: invalid src=%d\n", __func__, src_id);
>> +		goto out;
>> +	}
>> +
>> +	dst = node_find(dst_id);
>> +	if (!dst) {
>> +		dev_err(dev, "%s: invalid dst=%d\n", __func__, dst_id);
>> +		goto out;
>> +	}
>> +
>> +	mutex_lock(&icc_lock);
>> +	path = path_find(dev, src, dst);
>> +	mutex_unlock(&icc_lock);
>> +	if (IS_ERR(path)) {
>> +		dev_err(dev, "%s: invalid path=%ld\n", __func__, PTR_ERR(path));
>> +		goto out;
> 
> this goto isn't really needed

Ok.

>> +struct icc_node *icc_node_create(int id)
>> +{
>> +	struct icc_node *node;
>> +
>> +	/* check if node already exists */
>> +	node = node_find(id);
>> +	if (node)
>> +		goto out;
>> +
>> +	node = kzalloc(sizeof(*node), GFP_KERNEL);
>> +	if (!node) {
>> +		node = ERR_PTR(-ENOMEM);
>> +		goto out;
>> +	}
>> +
>> +	mutex_lock(&icc_lock);
>> +
>> +	id = idr_alloc(&icc_idr, node, id, id + 1, GFP_KERNEL);
>> +	if (WARN(id < 0, "couldn't get idr")) {
> 
> kfree(node);

Thanks!

>> +int icc_node_add(struct icc_node *node, struct icc_provider *provider)
>> +{
>> +	mutex_lock(&icc_lock);
>> +
>> +	node->provider = provider;
>> +	list_add(&node->node_list, &provider->nodes);
>> +
>> +	mutex_unlock(&icc_lock);
>> +
>> +	return 0;
>> +}
> 
> The function returns always 0. Should probably be void so callers
> don't add pointless checks of the return value.

Agree, will change it!

>> +int icc_provider_add(struct icc_provider *provider)
>> +{
>> +	if (WARN_ON(!provider->set))
>> +		return -EINVAL;
>> +
>> +	mutex_init(&icc_lock);
> 
> Shouldn't this be mutex_lock()?

Yes, right!

>> +int icc_provider_del(struct icc_provider *provider)
>> +{
>> +	mutex_lock(&icc_lock);
>> +	if (provider->users) {
>> +		pr_warn("interconnect provider still has %d users\n",
>> +			provider->users);
>> +		mutex_unlock(&icc_lock);
>> +		return -EBUSY;
>> +	}
>> +
>> +	if (!list_empty_careful(&provider->nodes)) {
>> +		pr_warn("interconnect provider still has nodes\n");
>> +		mutex_unlock(&icc_lock);
>> +		return -EEXIST;
>> +	}
> 
> Could this be just list_empty()? If I didn't miss something icc_lock
> is held in all paths that change p->nodes (assuming that all changes
> should be done through the interfaces in this file).

It could be. Will update it. I just always want to be careful!

> Actually this check will always fail if icc_node_add() was called for
> this provider, it doesn't seem nodes are ever removed.
> 

The provider driver is responsible for the node removal.

Thanks,
Georgi

Re: [PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Georgi Djakov]

Hi Vincent,

On 27.06.18 г. 9:19, Vincent Guittot wrote:
> Hi Georgi
> 
> On Wed, 20 Jun 2018 at 14:11, Georgi Djakov <georgi.djakov@linaro.org> wrote:
> 
> [snip]
> 
>> +
>> +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
>> +{
>> +       struct icc_node *node = dst;
>> +       struct icc_path *path;
>> +       size_t i;
>> +
>> +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
> 
> Should be (num_nodes -1) * sizeof(*path->reqs) as there is already 1
> icc_req in icc_path struct

reqs[] is a flexible array member and it's size is not included in
sizeof(*path)

Thanks,
Georgi

>> +                      GFP_KERNEL);
>> +       if (!path)
>> +               return ERR_PTR(-ENOMEM);
>> +
>> +       path->num_nodes = num_nodes;
>> +
>> +       for (i = 0; i < num_nodes; i++) {
>> +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
>> +
>> +               path->reqs[i].node = node;
>> +               /* reference to previous node was saved during path traversal */
>> +               node = node->reverse;
>> +       }
>> +
>> +       return path;
>> +}
>> +
> 
> [snip]
> 

[PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Georgi Djakov]

Hi Vincent,

On 27.06.18 ?. 9:19, Vincent Guittot wrote:
> Hi Georgi
> 
> On Wed, 20 Jun 2018 at 14:11, Georgi Djakov <georgi.djakov@linaro.org> wrote:
> 
> [snip]
> 
>> +
>> +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
>> +{
>> +       struct icc_node *node = dst;
>> +       struct icc_path *path;
>> +       size_t i;
>> +
>> +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
> 
> Should be (num_nodes -1) * sizeof(*path->reqs) as there is already 1
> icc_req in icc_path struct

reqs[] is a flexible array member and it's size is not included in
sizeof(*path)

Thanks,
Georgi

>> +                      GFP_KERNEL);
>> +       if (!path)
>> +               return ERR_PTR(-ENOMEM);
>> +
>> +       path->num_nodes = num_nodes;
>> +
>> +       for (i = 0; i < num_nodes; i++) {
>> +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
>> +
>> +               path->reqs[i].node = node;
>> +               /* reference to previous node was saved during path traversal */
>> +               node = node->reverse;
>> +       }
>> +
>> +       return path;
>> +}
>> +
> 
> [snip]
> 

Re: [PATCH v5 4/8] interconnect: qcom: Add RPM communication

[Georgi Djakov]

Hi Evan,

On 06/26/2018 11:47 PM, Evan Green wrote:
> Hi Georgi,
> 
> On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>>
>> On some Qualcomm SoCs, there is a remote processor, which controls some of
>> the Network-On-Chip interconnect resources. Other CPUs express their needs
>> by communicating with this processor. Add a driver to handle comminication
>> with this remote processor.
>>
>> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
>> ---

[..]

>> diff --git a/drivers/interconnect/qcom/Makefile b/drivers/interconnect/qcom/Makefile
>> new file mode 100644
>> index 000000000000..2a0c41db91c4
>> --- /dev/null
>> +++ b/drivers/interconnect/qcom/Makefile
>> @@ -0,0 +1,2 @@
>> +# SPDX-License-Identifier: GPL-2.0
>> +obj-y += smd-rpm.o
> 
> This new file has an unexpressed dependency on CONFIG_QCOM_SMD_RPM due
> to qcom_rpm_smd_write. RPMh-based systems won't need this file, right?
> Maybe this file should live behind some config?

Right, I just made this a separate config, as the the kbuild bot also
reported an issue with the modular build.

[..]

>> diff --git a/drivers/interconnect/qcom/smd-rpm.h b/drivers/interconnect/qcom/smd-rpm.h
>> new file mode 100644
>> index 000000000000..0f4a3da31cf6
>> --- /dev/null
>> +++ b/drivers/interconnect/qcom/smd-rpm.h
>> @@ -0,0 +1,15 @@
>> +/* SPDX-License-Identifier: GPL-2.0 */
>> +/*
>> + * Copyright (c) 2018, Linaro Ltd.
>> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
>> + */
>> +
>> +#ifndef __LINUX_INTERCONNECT_QCOM_RPM_H
>> +#define __LINUX_INTERCONNECT_QCOM_RPM_H
> 
> This doesn't really reflect the path, should this be something different?

Will update!

Thanks,
Georgi

[PATCH v5 4/8] interconnect: qcom: Add RPM communication

[Georgi Djakov]

Hi Evan,

On 06/26/2018 11:47 PM, Evan Green wrote:
> Hi Georgi,
> 
> On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>>
>> On some Qualcomm SoCs, there is a remote processor, which controls some of
>> the Network-On-Chip interconnect resources. Other CPUs express their needs
>> by communicating with this processor. Add a driver to handle comminication
>> with this remote processor.
>>
>> Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
>> ---

[..]

>> diff --git a/drivers/interconnect/qcom/Makefile b/drivers/interconnect/qcom/Makefile
>> new file mode 100644
>> index 000000000000..2a0c41db91c4
>> --- /dev/null
>> +++ b/drivers/interconnect/qcom/Makefile
>> @@ -0,0 +1,2 @@
>> +# SPDX-License-Identifier: GPL-2.0
>> +obj-y += smd-rpm.o
> 
> This new file has an unexpressed dependency on CONFIG_QCOM_SMD_RPM due
> to qcom_rpm_smd_write. RPMh-based systems won't need this file, right?
> Maybe this file should live behind some config?

Right, I just made this a separate config, as the the kbuild bot also
reported an issue with the modular build.

[..]

>> diff --git a/drivers/interconnect/qcom/smd-rpm.h b/drivers/interconnect/qcom/smd-rpm.h
>> new file mode 100644
>> index 000000000000..0f4a3da31cf6
>> --- /dev/null
>> +++ b/drivers/interconnect/qcom/smd-rpm.h
>> @@ -0,0 +1,15 @@
>> +/* SPDX-License-Identifier: GPL-2.0 */
>> +/*
>> + * Copyright (c) 2018, Linaro Ltd.
>> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
>> + */
>> +
>> +#ifndef __LINUX_INTERCONNECT_QCOM_RPM_H
>> +#define __LINUX_INTERCONNECT_QCOM_RPM_H
> 
> This doesn't really reflect the path, should this be something different?

Will update!

Thanks,
Georgi

Re: [PATCH v5 4/8] interconnect: qcom: Add RPM communication

[Georgi Djakov]

Hi Matthias,

On 06/27/2018 03:55 AM, Matthias Kaehlcke wrote:
> On Wed, Jun 20, 2018 at 03:11:37PM +0300, Georgi Djakov wrote:
>> On some Qualcomm SoCs, there is a remote processor, which controls some of
>> the Network-On-Chip interconnect resources. Other CPUs express their needs
>> by communicating with this processor. Add a driver to handle comminication
> 
> s/comminication/communication/
> 

[..]

>> +// SPDX-License-Identifier: GPL-2.0
>> +/*
>> + * RPM over SMD communication wrapper for interconects
> 
> s/interconects/interconnects/
> 

Thanks, will fix them!

BR,
Georgi

[PATCH v5 4/8] interconnect: qcom: Add RPM communication

[Georgi Djakov]

Hi Matthias,

On 06/27/2018 03:55 AM, Matthias Kaehlcke wrote:
> On Wed, Jun 20, 2018 at 03:11:37PM +0300, Georgi Djakov wrote:
>> On some Qualcomm SoCs, there is a remote processor, which controls some of
>> the Network-On-Chip interconnect resources. Other CPUs express their needs
>> by communicating with this processor. Add a driver to handle comminication
> 
> s/comminication/communication/
> 

[..]

>> +// SPDX-License-Identifier: GPL-2.0
>> +/*
>> + * RPM over SMD communication wrapper for interconects
> 
> s/interconects/interconnects/
> 

Thanks, will fix them!

BR,
Georgi

Re: [PATCH v5 6/8] interconnect: qcom: Add msm8916 interconnect provider driver

[Georgi Djakov]

Hi Evan,

On 06/26/2018 11:48 PM, Evan Green wrote:
> On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>>
>> Add driver for the Qualcomm interconnect buses found in msm8916 based
>> platforms.
>>

[..]

>> --- /dev/null
>> +++ b/drivers/interconnect/qcom/msm8916.c
>> @@ -0,0 +1,495 @@
>> +// SPDX-License-Identifier: GPL-2.0
>> +/*
>> + * Copyright (C) 2018 Linaro Ltd
>> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
>> + */
>> +
>> +#include <dt-bindings/interconnect/qcom.h>
>> +#include <linux/clk.h>
>> +#include <linux/device.h>
>> +#include <linux/io.h>
>> +#include <linux/interconnect-provider.h>
> 
> Nit: the previous two should be swapped for alphabetization.

Ok.

>> +#include <linux/module.h>
>> +#include <linux/of_device.h>
>> +#include <linux/of_platform.h>
>> +#include <linux/platform_device.h>
>> +#include <linux/slab.h>
>> +
>> +#include "smd-rpm.h"
>> +
>> +#define RPM_MASTER_FIELD_BW    0x00007762
> 
> This is unused, since it moved into that other file.

Ok.

[..]

>> +/**
>> + * struct qcom_icc_node - Qualcomm specific interconnect nodes
>> + * @name: the node name used in debugfs
>> + * @links: an array of nodes where we can go next while traversing
>> + * @id: a unique node identifier
>> + * @num_links: the total number of @links
>> + * @port: the offset index into the masters QoS register space
>> + * @buswidth: width of the interconnect between a node and the bus (bytes)
>> + * @ap_owned: the AP CPU does the writing to QoS registers
>> + * @rpm: reference to the RPM SMD driver
>> + * @qos_mode: QoS mode for ap_owned resources
>> + * @mas_rpm_id:        RPM id for devices that are bus masters
>> + * @slv_rpm_id:        RPM id for devices that are bus slaves
>> + * @rate: current bus clock rate in Hz
>> + */
>> +struct qcom_icc_node {
>> +       unsigned char *name;
>> +       u16 links[MSM8916_MAX_LINKS];
>> +       u16 id;
>> +       u16 num_links;
>> +       u16 port;
>> +       u16 buswidth;
>> +       bool ap_owned;
>> +       struct qcom_smd_rpm *rpm;
> 
> This is unused.

Ok, will remove.

>> +       enum qcom_qos_mode qos_mode;
>> +       int mas_rpm_id;
>> +       int slv_rpm_id;
>> +       u64 rate;
>> +};
>> +

[..]

>> +static int qcom_icc_init(struct icc_node *node)
>> +{
>> +       struct qcom_icc_provider *qp = to_qcom_provider(node->provider);
>> +       int ret;
>> +
>> +       /* TODO: init qos and priority */
>> +
>> +       clk_set_rate(qp->bus_clk, INT_MAX);
> 
> Vroom! What's the rationale here? I wonder if it might be better to
> avoid touching the clocks initially, and expect that the boot loader
> sets up a decent initial set of bus frequencies for consumers that
> never enable bus scaling? Otherwise, I worry that this driver becomes
> basically an essential driver for the platform solely because of this
> line and the one below, when really it might not be.

The idea is to run the interconnects at max rate until consumers start
sending requests, but i understand your worry and we can live without
this for now. The better solution would be to set maximum bandwidth and
remove it at late_init (after consumers are registered) or use this
patchset: https://lkml.org/lkml/2018/3/21/897
Actually I have some patches which add support for interconnects to keep
the bandwidth constraints active until consumers are registered. The
whole boot constraint thing adds complexity and introduces some
overhead, but hopefully can be optimized.

>> +       ret = clk_prepare_enable(qp->bus_clk);
>> +       if (ret)
>> +               pr_info("%s: error enabling bus clk (%d)\n", __func__, ret);
>> +       clk_set_rate(qp->bus_a_clk, INT_MAX);
>> +       ret = clk_prepare_enable(qp->bus_a_clk);
>> +       if (ret)
>> +               pr_info("%s: error enabling bus_a clk (%d)\n", __func__, ret);
>> +
>> +       return 0;
>> +}
>> +
>> +static int qcom_icc_aggregate(struct icc_node *node, u32 avg_bw, u32 peak_bw,
>> +                             u32 *agg_avg, u32 *agg_peak)
>> +{
>> +       *agg_avg += avg_bw;
>> +       *agg_peak = max(*agg_peak, peak_bw);
>> +
>> +       return 0;
>> +}
>> +
>> +static int qcom_icc_set(struct icc_node *src, struct icc_node *dst,
>> +                       u32 avg, u32 peak)
>> +{
>> +       struct qcom_icc_provider *qp;
>> +       struct qcom_icc_node *qn;
>> +       struct icc_node *node;
>> +       struct icc_provider *provider;
>> +       u64 avg_bw = icc_units_to_bps(avg);
>> +       u64 peak_bw = icc_units_to_bps(peak);
>> +       u64 rate = 0;
>> +       int ret = 0;
>> +
>> +       if (!src)
>> +               node = dst;
>> +       else
>> +               node = src;
>> +
>> +       qn = node->data;
>> +       provider = node->provider;
>> +       qp = to_qcom_provider(node->provider);
>> +
>> +       /* set bandwidth */
>> +       if (qn->ap_owned) {
>> +               /* TODO: set QoS */
>> +       } else {
>> +               /* send message to the RPM processor */
>> +               if (qn->mas_rpm_id != -1) {
>> +                       ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
>> +                                                   RPM_BUS_MASTER_REQ,
>> +                                                   qn->mas_rpm_id,
>> +                                                   avg_bw);
> 
> You don't check for failure here, and potentially clobber failure with
> success. Unless only one of mas_rpm_id and slv_rpm_id is ever set. But
> no, I see SNOC_INT_0/1 set both. Same for below.

Ok, will add error checking.

>> +               }
>> +
>> +               if (qn->slv_rpm_id != -1) {
>> +                       ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
>> +                                                   RPM_BUS_SLAVE_REQ,
>> +                                                   qn->slv_rpm_id,
>> +                                                   avg_bw);
>> +               }
>> +       }
>> +
>> +       rate = max(avg_bw, peak_bw);
>> +
>> +       do_div(rate, qn->buswidth);
>> +
>> +       if (qn->rate != rate) {
>> +               ret = clk_set_rate(qp->bus_clk, rate);
>> +               if (ret) {
>> +                       pr_err("set clk rate %lld error %d\n", rate, ret);
>> +                       return ret;
>> +               }
>> +
>> +               ret = clk_set_rate(qp->bus_a_clk, rate);
>> +               if (ret) {
>> +                       pr_err("set clk rate %lld error %d\n", rate, ret);
>> +                       return ret;
>> +               }
>> +
>> +               qn->rate = rate;
>> +       }
>> +
>> +       return ret;
>> +}
>> +
>> +static int qnoc_probe(struct platform_device *pdev)
>> +{
>> +       const struct qcom_icc_desc *desc;
>> +       struct qcom_icc_node **qnodes;
>> +       struct qcom_icc_provider *qp;
>> +       struct resource *res;
>> +       struct icc_provider *provider;
>> +       size_t num_nodes, i;
>> +       int ret;
>> +
>> +       /* wait for RPM */
>> +       if (!qcom_icc_rpm_smd_available())
>> +               return -EPROBE_DEFER;
>> +
>> +       desc = of_device_get_match_data(&pdev->dev);
>> +       if (!desc)
>> +               return -EINVAL;
>> +
>> +       qnodes = desc->nodes;
>> +       num_nodes = desc->num_nodes;
>> +
>> +       qp = devm_kzalloc(&pdev->dev, sizeof(*qp), GFP_KERNEL);
>> +       if (!qp)
>> +               return -ENOMEM;
>> +
>> +       res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
>> +       qp->base = devm_ioremap_resource(&pdev->dev, res);
>> +       if (IS_ERR(qp->base))
>> +               return PTR_ERR(qp->base);
>> +
>> +       qp->bus_clk = devm_clk_get(&pdev->dev, "bus_clk");
>> +       if (IS_ERR(qp->bus_clk))
>> +               return PTR_ERR(qp->bus_clk);
>> +
>> +       qp->bus_a_clk = devm_clk_get(&pdev->dev, "bus_a_clk");
>> +       if (IS_ERR(qp->bus_a_clk))
>> +               return PTR_ERR(qp->bus_a_clk);
>> +
>> +       provider = &qp->provider;
>> +       provider->dev = &pdev->dev;
>> +       provider->set = &qcom_icc_set;
>> +       provider->aggregate = &qcom_icc_aggregate;
>> +       INIT_LIST_HEAD(&provider->nodes);
>> +       provider->data = qp;
>> +
>> +       ret = icc_provider_add(provider);
>> +       if (ret) {
>> +               dev_err(&pdev->dev, "error adding interconnect provider\n");
>> +               return ret;
>> +       }
>> +
>> +       for (i = 0; i < num_nodes; i++) {
>> +               struct icc_node *node;
>> +               int ret;
>> +               size_t j;
>> +
>> +               node = icc_node_create(qnodes[i]->id);
>> +               if (IS_ERR(node)) {
>> +                       ret = PTR_ERR(node);
>> +                       goto err;
>> +               }
>> +
>> +               node->name = qnodes[i]->name;
>> +               node->data = qnodes[i];
>> +               icc_node_add(node, provider);
>> +
>> +               dev_dbg(&pdev->dev, "registered node %s\n", node->name);
>> +
>> +               ret = qcom_icc_init(node);
>> +               if (ret)
>> +                       dev_err(&pdev->dev, "%s init error (%d)\n", node->name,
>> +                               ret);
>> +
>> +               /* populate links */
>> +               for (j = 0; j < qnodes[i]->num_links; j++)
>> +                       if (qnodes[i]->links[j])
>> +                               icc_link_create(node, qnodes[i]->links[j]);
>> +       }
>> +
>> +       platform_set_drvdata(pdev, provider);
>> +
>> +       return ret;
>> +err:
>> +       icc_provider_del(provider);
>> +       return ret;
>> +}
>> +
>> +static int qnoc_remove(struct platform_device *pdev)
>> +{
>> +       struct icc_provider *provider = platform_get_drvdata(pdev);
>> +
>> +       icc_provider_del(provider);
> 
> This function can fail, so you should probably pass the failure along.
> 

Ok! Thank you!

BR,
Georgi

[PATCH v5 6/8] interconnect: qcom: Add msm8916 interconnect provider driver

[Georgi Djakov]

Hi Evan,

On 06/26/2018 11:48 PM, Evan Green wrote:
> On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>>
>> Add driver for the Qualcomm interconnect buses found in msm8916 based
>> platforms.
>>

[..]

>> --- /dev/null
>> +++ b/drivers/interconnect/qcom/msm8916.c
>> @@ -0,0 +1,495 @@
>> +// SPDX-License-Identifier: GPL-2.0
>> +/*
>> + * Copyright (C) 2018 Linaro Ltd
>> + * Author: Georgi Djakov <georgi.djakov@linaro.org>
>> + */
>> +
>> +#include <dt-bindings/interconnect/qcom.h>
>> +#include <linux/clk.h>
>> +#include <linux/device.h>
>> +#include <linux/io.h>
>> +#include <linux/interconnect-provider.h>
> 
> Nit: the previous two should be swapped for alphabetization.

Ok.

>> +#include <linux/module.h>
>> +#include <linux/of_device.h>
>> +#include <linux/of_platform.h>
>> +#include <linux/platform_device.h>
>> +#include <linux/slab.h>
>> +
>> +#include "smd-rpm.h"
>> +
>> +#define RPM_MASTER_FIELD_BW    0x00007762
> 
> This is unused, since it moved into that other file.

Ok.

[..]

>> +/**
>> + * struct qcom_icc_node - Qualcomm specific interconnect nodes
>> + * @name: the node name used in debugfs
>> + * @links: an array of nodes where we can go next while traversing
>> + * @id: a unique node identifier
>> + * @num_links: the total number of @links
>> + * @port: the offset index into the masters QoS register space
>> + * @buswidth: width of the interconnect between a node and the bus (bytes)
>> + * @ap_owned: the AP CPU does the writing to QoS registers
>> + * @rpm: reference to the RPM SMD driver
>> + * @qos_mode: QoS mode for ap_owned resources
>> + * @mas_rpm_id:        RPM id for devices that are bus masters
>> + * @slv_rpm_id:        RPM id for devices that are bus slaves
>> + * @rate: current bus clock rate in Hz
>> + */
>> +struct qcom_icc_node {
>> +       unsigned char *name;
>> +       u16 links[MSM8916_MAX_LINKS];
>> +       u16 id;
>> +       u16 num_links;
>> +       u16 port;
>> +       u16 buswidth;
>> +       bool ap_owned;
>> +       struct qcom_smd_rpm *rpm;
> 
> This is unused.

Ok, will remove.

>> +       enum qcom_qos_mode qos_mode;
>> +       int mas_rpm_id;
>> +       int slv_rpm_id;
>> +       u64 rate;
>> +};
>> +

[..]

>> +static int qcom_icc_init(struct icc_node *node)
>> +{
>> +       struct qcom_icc_provider *qp = to_qcom_provider(node->provider);
>> +       int ret;
>> +
>> +       /* TODO: init qos and priority */
>> +
>> +       clk_set_rate(qp->bus_clk, INT_MAX);
> 
> Vroom! What's the rationale here? I wonder if it might be better to
> avoid touching the clocks initially, and expect that the boot loader
> sets up a decent initial set of bus frequencies for consumers that
> never enable bus scaling? Otherwise, I worry that this driver becomes
> basically an essential driver for the platform solely because of this
> line and the one below, when really it might not be.

The idea is to run the interconnects at max rate until consumers start
sending requests, but i understand your worry and we can live without
this for now. The better solution would be to set maximum bandwidth and
remove it at late_init (after consumers are registered) or use this
patchset: https://lkml.org/lkml/2018/3/21/897
Actually I have some patches which add support for interconnects to keep
the bandwidth constraints active until consumers are registered. The
whole boot constraint thing adds complexity and introduces some
overhead, but hopefully can be optimized.

>> +       ret = clk_prepare_enable(qp->bus_clk);
>> +       if (ret)
>> +               pr_info("%s: error enabling bus clk (%d)\n", __func__, ret);
>> +       clk_set_rate(qp->bus_a_clk, INT_MAX);
>> +       ret = clk_prepare_enable(qp->bus_a_clk);
>> +       if (ret)
>> +               pr_info("%s: error enabling bus_a clk (%d)\n", __func__, ret);
>> +
>> +       return 0;
>> +}
>> +
>> +static int qcom_icc_aggregate(struct icc_node *node, u32 avg_bw, u32 peak_bw,
>> +                             u32 *agg_avg, u32 *agg_peak)
>> +{
>> +       *agg_avg += avg_bw;
>> +       *agg_peak = max(*agg_peak, peak_bw);
>> +
>> +       return 0;
>> +}
>> +
>> +static int qcom_icc_set(struct icc_node *src, struct icc_node *dst,
>> +                       u32 avg, u32 peak)
>> +{
>> +       struct qcom_icc_provider *qp;
>> +       struct qcom_icc_node *qn;
>> +       struct icc_node *node;
>> +       struct icc_provider *provider;
>> +       u64 avg_bw = icc_units_to_bps(avg);
>> +       u64 peak_bw = icc_units_to_bps(peak);
>> +       u64 rate = 0;
>> +       int ret = 0;
>> +
>> +       if (!src)
>> +               node = dst;
>> +       else
>> +               node = src;
>> +
>> +       qn = node->data;
>> +       provider = node->provider;
>> +       qp = to_qcom_provider(node->provider);
>> +
>> +       /* set bandwidth */
>> +       if (qn->ap_owned) {
>> +               /* TODO: set QoS */
>> +       } else {
>> +               /* send message to the RPM processor */
>> +               if (qn->mas_rpm_id != -1) {
>> +                       ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
>> +                                                   RPM_BUS_MASTER_REQ,
>> +                                                   qn->mas_rpm_id,
>> +                                                   avg_bw);
> 
> You don't check for failure here, and potentially clobber failure with
> success. Unless only one of mas_rpm_id and slv_rpm_id is ever set. But
> no, I see SNOC_INT_0/1 set both. Same for below.

Ok, will add error checking.

>> +               }
>> +
>> +               if (qn->slv_rpm_id != -1) {
>> +                       ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
>> +                                                   RPM_BUS_SLAVE_REQ,
>> +                                                   qn->slv_rpm_id,
>> +                                                   avg_bw);
>> +               }
>> +       }
>> +
>> +       rate = max(avg_bw, peak_bw);
>> +
>> +       do_div(rate, qn->buswidth);
>> +
>> +       if (qn->rate != rate) {
>> +               ret = clk_set_rate(qp->bus_clk, rate);
>> +               if (ret) {
>> +                       pr_err("set clk rate %lld error %d\n", rate, ret);
>> +                       return ret;
>> +               }
>> +
>> +               ret = clk_set_rate(qp->bus_a_clk, rate);
>> +               if (ret) {
>> +                       pr_err("set clk rate %lld error %d\n", rate, ret);
>> +                       return ret;
>> +               }
>> +
>> +               qn->rate = rate;
>> +       }
>> +
>> +       return ret;
>> +}
>> +
>> +static int qnoc_probe(struct platform_device *pdev)
>> +{
>> +       const struct qcom_icc_desc *desc;
>> +       struct qcom_icc_node **qnodes;
>> +       struct qcom_icc_provider *qp;
>> +       struct resource *res;
>> +       struct icc_provider *provider;
>> +       size_t num_nodes, i;
>> +       int ret;
>> +
>> +       /* wait for RPM */
>> +       if (!qcom_icc_rpm_smd_available())
>> +               return -EPROBE_DEFER;
>> +
>> +       desc = of_device_get_match_data(&pdev->dev);
>> +       if (!desc)
>> +               return -EINVAL;
>> +
>> +       qnodes = desc->nodes;
>> +       num_nodes = desc->num_nodes;
>> +
>> +       qp = devm_kzalloc(&pdev->dev, sizeof(*qp), GFP_KERNEL);
>> +       if (!qp)
>> +               return -ENOMEM;
>> +
>> +       res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
>> +       qp->base = devm_ioremap_resource(&pdev->dev, res);
>> +       if (IS_ERR(qp->base))
>> +               return PTR_ERR(qp->base);
>> +
>> +       qp->bus_clk = devm_clk_get(&pdev->dev, "bus_clk");
>> +       if (IS_ERR(qp->bus_clk))
>> +               return PTR_ERR(qp->bus_clk);
>> +
>> +       qp->bus_a_clk = devm_clk_get(&pdev->dev, "bus_a_clk");
>> +       if (IS_ERR(qp->bus_a_clk))
>> +               return PTR_ERR(qp->bus_a_clk);
>> +
>> +       provider = &qp->provider;
>> +       provider->dev = &pdev->dev;
>> +       provider->set = &qcom_icc_set;
>> +       provider->aggregate = &qcom_icc_aggregate;
>> +       INIT_LIST_HEAD(&provider->nodes);
>> +       provider->data = qp;
>> +
>> +       ret = icc_provider_add(provider);
>> +       if (ret) {
>> +               dev_err(&pdev->dev, "error adding interconnect provider\n");
>> +               return ret;
>> +       }
>> +
>> +       for (i = 0; i < num_nodes; i++) {
>> +               struct icc_node *node;
>> +               int ret;
>> +               size_t j;
>> +
>> +               node = icc_node_create(qnodes[i]->id);
>> +               if (IS_ERR(node)) {
>> +                       ret = PTR_ERR(node);
>> +                       goto err;
>> +               }
>> +
>> +               node->name = qnodes[i]->name;
>> +               node->data = qnodes[i];
>> +               icc_node_add(node, provider);
>> +
>> +               dev_dbg(&pdev->dev, "registered node %s\n", node->name);
>> +
>> +               ret = qcom_icc_init(node);
>> +               if (ret)
>> +                       dev_err(&pdev->dev, "%s init error (%d)\n", node->name,
>> +                               ret);
>> +
>> +               /* populate links */
>> +               for (j = 0; j < qnodes[i]->num_links; j++)
>> +                       if (qnodes[i]->links[j])
>> +                               icc_link_create(node, qnodes[i]->links[j]);
>> +       }
>> +
>> +       platform_set_drvdata(pdev, provider);
>> +
>> +       return ret;
>> +err:
>> +       icc_provider_del(provider);
>> +       return ret;
>> +}
>> +
>> +static int qnoc_remove(struct platform_device *pdev)
>> +{
>> +       struct icc_provider *provider = platform_get_drvdata(pdev);
>> +
>> +       icc_provider_del(provider);
> 
> This function can fail, so you should probably pass the failure along.
> 

Ok! Thank you!

BR,
Georgi

Re: [PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Vincent Guittot]

On Sun, 1 Jul 2018 at 13:09, Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> Hi Vincent,
>
> On 27.06.18 г. 9:19, Vincent Guittot wrote:
> > Hi Georgi
> >
> > On Wed, 20 Jun 2018 at 14:11, Georgi Djakov <georgi.djakov@linaro.org> wrote:
> >
> > [snip]
> >
> >> +
> >> +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
> >> +{
> >> +       struct icc_node *node = dst;
> >> +       struct icc_path *path;
> >> +       size_t i;
> >> +
> >> +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
> >
> > Should be (num_nodes -1) * sizeof(*path->reqs) as there is already 1
> > icc_req in icc_path struct
>
> reqs[] is a flexible array member and it's size is not included in
> sizeof(*path)

ah yes, my eyes saw "struct icc_req reqs[0];"  but my brain was
reading "struct icc_req reqs[1]"

Sorry for the noise

>
> Thanks,
> Georgi
>
> >> +                      GFP_KERNEL);
> >> +       if (!path)
> >> +               return ERR_PTR(-ENOMEM);
> >> +
> >> +       path->num_nodes = num_nodes;
> >> +
> >> +       for (i = 0; i < num_nodes; i++) {
> >> +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
> >> +
> >> +               path->reqs[i].node = node;
> >> +               /* reference to previous node was saved during path traversal */
> >> +               node = node->reverse;
> >> +       }
> >> +
> >> +       return path;
> >> +}
> >> +
> >
> > [snip]
> >
>

[PATCH v5 1/8] interconnect: Add generic on-chip interconnect API

[Vincent Guittot]

On Sun, 1 Jul 2018 at 13:09, Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> Hi Vincent,
>
> On 27.06.18 ?. 9:19, Vincent Guittot wrote:
> > Hi Georgi
> >
> > On Wed, 20 Jun 2018 at 14:11, Georgi Djakov <georgi.djakov@linaro.org> wrote:
> >
> > [snip]
> >
> >> +
> >> +static struct icc_path *path_allocate(struct icc_node *dst, ssize_t num_nodes)
> >> +{
> >> +       struct icc_node *node = dst;
> >> +       struct icc_path *path;
> >> +       size_t i;
> >> +
> >> +       path = kzalloc(sizeof(*path) + num_nodes * sizeof(*path->reqs),
> >
> > Should be (num_nodes -1) * sizeof(*path->reqs) as there is already 1
> > icc_req in icc_path struct
>
> reqs[] is a flexible array member and it's size is not included in
> sizeof(*path)

ah yes, my eyes saw "struct icc_req reqs[0];"  but my brain was
reading "struct icc_req reqs[1]"

Sorry for the noise

>
> Thanks,
> Georgi
>
> >> +                      GFP_KERNEL);
> >> +       if (!path)
> >> +               return ERR_PTR(-ENOMEM);
> >> +
> >> +       path->num_nodes = num_nodes;
> >> +
> >> +       for (i = 0; i < num_nodes; i++) {
> >> +               hlist_add_head(&path->reqs[i].req_node, &node->req_list);
> >> +
> >> +               path->reqs[i].node = node;
> >> +               /* reference to previous node was saved during path traversal */
> >> +               node = node->reverse;
> >> +       }
> >> +
> >> +       return path;
> >> +}
> >> +
> >
> > [snip]
> >
>

Re: [PATCH v5 6/8] interconnect: qcom: Add msm8916 interconnect provider driver

[Evan Green]

On Sun, Jul 1, 2018 at 5:12 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> Hi Evan,
>
> On 06/26/2018 11:48 PM, Evan Green wrote:
> > On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> >> +static int qcom_icc_init(struct icc_node *node)
> >> +{
> >> +       struct qcom_icc_provider *qp = to_qcom_provider(node->provider);
> >> +       int ret;
> >> +
> >> +       /* TODO: init qos and priority */
> >> +
> >> +       clk_set_rate(qp->bus_clk, INT_MAX);
> >
> > Vroom! What's the rationale here? I wonder if it might be better to
> > avoid touching the clocks initially, and expect that the boot loader
> > sets up a decent initial set of bus frequencies for consumers that
> > never enable bus scaling? Otherwise, I worry that this driver becomes
> > basically an essential driver for the platform solely because of this
> > line and the one below, when really it might not be.
>
> The idea is to run the interconnects at max rate until consumers start
> sending requests, but i understand your worry and we can live without
> this for now. The better solution would be to set maximum bandwidth and
> remove it at late_init (after consumers are registered) or use this
> patchset: https://lkml.org/lkml/2018/3/21/897
> Actually I have some patches which add support for interconnects to keep
> the bandwidth constraints active until consumers are registered. The
> whole boot constraint thing adds complexity and introduces some
> overhead, but hopefully can be optimized.

Ah, that makes sense. This is a trickier issue than I was thinking
before. On one hand, you don't want to shut off bandwidth to a device
that was set up correctly by the boot environment and should keep
working until the driver comes up, like LCD. But on the other hand, if
a driver fails to come up, or fails to ask for bus bandwidth, you're
now burning at max. And then there's the issue of whether or not this
should be a required or optional driver for platforms that support it
(it would be nice if the system booted even without this driver, but
maybe for others that's a non-goal). I agree this is shouldn't hold up
this initial set of framework patches, we can solve this in a future
set.

-Evan

Re: [PATCH v5 6/8] interconnect: qcom: Add msm8916 interconnect provider driver

[Evan Green]

On Sun, Jul 1, 2018 at 5:12 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> Hi Evan,
>
> On 06/26/2018 11:48 PM, Evan Green wrote:
> > On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> >> +static int qcom_icc_init(struct icc_node *node)
> >> +{
> >> +       struct qcom_icc_provider *qp = to_qcom_provider(node->provider);
> >> +       int ret;
> >> +
> >> +       /* TODO: init qos and priority */
> >> +
> >> +       clk_set_rate(qp->bus_clk, INT_MAX);
> >
> > Vroom! What's the rationale here? I wonder if it might be better to
> > avoid touching the clocks initially, and expect that the boot loader
> > sets up a decent initial set of bus frequencies for consumers that
> > never enable bus scaling? Otherwise, I worry that this driver becomes
> > basically an essential driver for the platform solely because of this
> > line and the one below, when really it might not be.
>
> The idea is to run the interconnects at max rate until consumers start
> sending requests, but i understand your worry and we can live without
> this for now. The better solution would be to set maximum bandwidth and
> remove it at late_init (after consumers are registered) or use this
> patchset: https://lkml.org/lkml/2018/3/21/897
> Actually I have some patches which add support for interconnects to keep
> the bandwidth constraints active until consumers are registered. The
> whole boot constraint thing adds complexity and introduces some
> overhead, but hopefully can be optimized.

Ah, that makes sense. This is a trickier issue than I was thinking
before. On one hand, you don't want to shut off bandwidth to a device
that was set up correctly by the boot environment and should keep
working until the driver comes up, like LCD. But on the other hand, if
a driver fails to come up, or fails to ask for bus bandwidth, you're
now burning at max. And then there's the issue of whether or not this
should be a required or optional driver for platforms that support it
(it would be nice if the system booted even without this driver, but
maybe for others that's a non-goal). I agree this is shouldn't hold up
this initial set of framework patches, we can solve this in a future
set.

-Evan

[PATCH v5 6/8] interconnect: qcom: Add msm8916 interconnect provider driver

[Evan Green]

On Sun, Jul 1, 2018 at 5:12 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> Hi Evan,
>
> On 06/26/2018 11:48 PM, Evan Green wrote:
> > On Wed, Jun 20, 2018 at 5:11 AM Georgi Djakov <georgi.djakov@linaro.org> wrote:
>
> >> +static int qcom_icc_init(struct icc_node *node)
> >> +{
> >> +       struct qcom_icc_provider *qp = to_qcom_provider(node->provider);
> >> +       int ret;
> >> +
> >> +       /* TODO: init qos and priority */
> >> +
> >> +       clk_set_rate(qp->bus_clk, INT_MAX);
> >
> > Vroom! What's the rationale here? I wonder if it might be better to
> > avoid touching the clocks initially, and expect that the boot loader
> > sets up a decent initial set of bus frequencies for consumers that
> > never enable bus scaling? Otherwise, I worry that this driver becomes
> > basically an essential driver for the platform solely because of this
> > line and the one below, when really it might not be.
>
> The idea is to run the interconnects at max rate until consumers start
> sending requests, but i understand your worry and we can live without
> this for now. The better solution would be to set maximum bandwidth and
> remove it at late_init (after consumers are registered) or use this
> patchset: https://lkml.org/lkml/2018/3/21/897
> Actually I have some patches which add support for interconnects to keep
> the bandwidth constraints active until consumers are registered. The
> whole boot constraint thing adds complexity and introduces some
> overhead, but hopefully can be optimized.

Ah, that makes sense. This is a trickier issue than I was thinking
before. On one hand, you don't want to shut off bandwidth to a device
that was set up correctly by the boot environment and should keep
working until the driver comes up, like LCD. But on the other hand, if
a driver fails to come up, or fails to ask for bus bandwidth, you're
now burning at max. And then there's the issue of whether or not this
should be a required or optional driver for platforms that support it
(it would be nice if the system booted even without this driver, but
maybe for others that's a non-goal). I agree this is shouldn't hold up
this initial set of framework patches, we can solve this in a future
set.

-Evan