linux-stable.git/drivers/soundwire/intel.c, branch linux-5.14.y

soundwire: intel: fix potential race condition during power down

2021-09-18T11:43:54+00:00

[ Upstream commit ea6942dad4b2a7e1735aa0f10f3d0b04b847750f ]

The power down sequence sets the link_up flag as false outside of the
mutex_lock. This is potentially unsafe.

In additional the flow in that sequence can be improved by first
testing if the link was powered, setting the link_up flag as false and
proceeding with the power down. In case the CPA bits cannot be
cleared, we only flag an error since we cannot deal with interrupts
any longer.

Signed-off-by: Pierre-Louis Bossart 
Reviewed-by: Ranjani Sridharan 
Signed-off-by: Bard Liao 
Link: https://lore.kernel.org/r/20210818024954.16873-2-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul 
Signed-off-by: Sasha Levin

soundwire: intel: move to auxiliary bus

2021-06-14T04:42:26+00:00

Now that the auxiliary_bus exists, there's no reason to use platform
devices as children of a PCI device any longer.

This patch refactors the code by extending a basic auxiliary device
with Intel link-specific structures that need to be passed between
controller and link levels. This refactoring is much cleaner with no
need for cross-pointers between device and link structures.

Note that the auxiliary bus API has separate init and add steps, which
requires more attention in the error unwinding paths. The main loop
needs to deal with kfree() and auxiliary_device_uninit() for the
current iteration before jumping to the common label which releases
everything allocated in prior iterations.

Signed-off-by: Pierre-Louis Bossart 
Reviewed-by: Guennadi Liakhovetski 
Reviewed-by: Ranjani Sridharan 
Signed-off-by: Bard Liao 
Link: https://lore.kernel.org/r/20210511052132.28150-1-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul

soundwire: intel: remove useless readl

2021-03-30T06:21:43+00:00

Cppcheck complains:

drivers/soundwire/intel.c:564:15: style: Variable 'link_control' is
assigned a value that is never used. [unreadVariable]
 link_control = intel_readl(shim, SDW_SHIM_LCTL);

This looks like a leftover from a previous version, remove.

Signed-off-by: Pierre-Louis Bossart 
Reviewed-by: Guennadi Liakhovetski 
Reviewed-by: Rander Wang 
Reviewed-by: Kai Vehmanen 
Signed-off-by: Bard Liao 
Link: https://lore.kernel.org/r/20210302091122.13952-9-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul

soundwire: intel: add missing \n in dev_err()

2021-03-23T06:49:53+00:00

We fixed a lot of warnings in 2019 but the magic of copy-paste keeps
adding new ones...

Signed-off-by: Pierre-Louis Bossart 
Reviewed-by: Rander Wang 
Reviewed-by: Guennadi Liakhovetski 
Signed-off-by: Bard Liao 
Link: https://lore.kernel.org/r/20210323005855.20890-2-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul

soundwire: intel: add master quirks for bus clash and parity

2021-03-22T12:16:44+00:00

Now that we have declarations and bus support, add quirks for Intel
platforms.

Co-developed-by: Pierre-Louis Bossart 
Signed-off-by: Pierre-Louis Bossart 
Signed-off-by: Bard Liao 
Reviewed-by: Guennadi Liakhovetski 
Link: https://lore.kernel.org/r/20210302082720.12322-4-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul

soundwire: Intel: introduce DMI quirks for HP Spectre x360 Convertible

2021-03-22T12:03:10+00:00

HP Spectre x360 Convertible devices expose invalid _ADR fields in the
DSDT, which prevents codec drivers from probing. A possible solution
is to override the DSDT, but that's just too painful for users.

This patch suggests a simple DMI-based quirk to remap the existing
invalid ADR information into valid ones.

BugLink: https://github.com/thesofproject/linux/issues/2700
Co-developed-by: Bard Liao 
Signed-off-by: Bard Liao 
Signed-off-by: Pierre-Louis Bossart 
Reviewed-by: Rander Wang 
Reviewed-by: Guennadi Liakhovetski 
Link: https://lore.kernel.org/r/20210302075105.11515-3-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul

soundwire: intel: don't return error when clock stop failed

2021-01-19T14:51:21+00:00

dev->power.runtime_error will be set to the return value of the runtime
suspend callback function, and runtime resume function will return
-EINVAL if dev->power.runtime_error is not 0.

Somehow the codec rarely doesn't return an ACK to the clock prepare
command. If we stop the runtime suspend process and return error, we
will not be able to resume again. Likewise, if the codec lost sync and
did not rejoin, the resume operation will also fail. As a result, the
SoundWire bus can not be used anymore.

This patch suggests to finish the runtime suspend process even if we fail
to stop sdw bus clock. In the case where we do a hardware reset, the codecs
will be reconfigured completely. In the case where we use the regular clock
stop, the codecs keep their state and worst case will fall off the bus and
reattach.

The only drawback is that the power consumption may be higher and
device-initiated interrupts may be lost, but at least audio function can
still work after resume.

Signed-off-by: Bard Liao 
Reviewed-by: Rander Wang 
Reviewed-by: Kai Vehmanen 
Reviewed-by: Pierre-Louis Bossart 
Link: https://lore.kernel.org/r/20210114030248.9005-1-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul

soundwire: intel: Use kzalloc for allocating only one thing

2021-01-06T06:00:02+00:00

Use kzalloc rather than kcalloc(1,...)

The semantic patch that makes this change is as follows:
(http://coccinelle.lip6.fr/)

// 
@@
@@

- kcalloc(1,
+ kzalloc(
          ...)
// 

Signed-off-by: Zheng Yongjun 
Link: https://lore.kernel.org/r/20201229135012.23472-1-zhengyongjun3@huawei.com
Signed-off-by: Vinod Koul

soundwire: intel: fix another unused-function warning

2020-12-05T07:41:54+00:00

Without CONFIG_PM, there is another warning about an unused function:

drivers/soundwire/intel.c:530:12: error: 'intel_link_power_down' defined but not used [-Werror=unused-function]

After a previous fix, the driver already uses both an #ifdef and
a __maybe_unused annotation but still gets it wrong. Remove the
ifdef and instead use __maybe_unused consistently to avoid the
problem for good.

Fixes: f046b2334083 ("soundwire: intel: fix intel_suspend/resume defined but not used warning")
Fixes: ebf878eddbb4 ("soundwire: intel: add pm_runtime support")
Signed-off-by: Arnd Bergmann 
Reviewed-by: Pierre-Louis Bossart 
Link: https://lore.kernel.org/r/20201203230502.1480063-1-arnd@kernel.org
Signed-off-by: Vinod Koul

soundwire: intel: enable test modes

2020-09-23T09:59:30+00:00

This patch adds debugfs support to override the Master and Slave data
modes. The settings only take effect prior to a new stream being
prepared/enabled, or on resume.

The test mode can be set to verify data integrity and detect bus
clashes, but can only be used to test capture paths. In this case the
input generated by a Slave source port is replaced by a fixed or
cyclical patterns.

Signed-off-by: Pierre-Louis Bossart 
Reviewed-by: Guennadi Liakhovetski 
Reviewed-by: Rander Wang 
Signed-off-by: Bard Liao 
Link: https://lore.kernel.org/r/20200920193207.31241-3-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul