In MT8195 SOF design, both DSP and audio driver would access audio
registers. Before DSP accesses audio registers, audio power and clock
should be enabled. DSP will hang up if DSP access audio register but
audio power and clock are disabled. Therefore, we add audio pm runtime
active checking before accessing audio registers in SOF BE's callback
hw_params function to avoid this situation.
Signed-off-by: YC Hung <yc.hung@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Acked-by: Curtis Malainey <cujomalainey@chromium.org>
Link: https://lore.kernel.org/r/20221209031053.8444-1-yc.hung@mediatek.com
Signed-off-by: Mark Brown <broonie@kernel.org>
DAIs parameters should be cleared if there's no active stream. Before,
we implemented it in soc_pcm_hw_free() by detecting stream_active. If
the running stream is the last active stream, we're going to clear
parameters.
However it will cause DAIs parameters never be cleared if there're
more than one stream. For example, we have stream1 and stream2 about
to stop. stream2 executes soc_pcm_hw_free() before stream1 executes
soc_pcm_close(). At the moment, stream2 should clear DAIs parameters.
Since stream_active is not yet updated by stream1 in soc_pcm_close(),
stream2 will not clear DAIs parameters. In result both stream1 and
stream2 don't clear the parameters.
This patch moves DAIs parameters cleanup after stream_active is
updated.
Signed-off-by: Chancel Liu <chancel.liu@nxp.com>
Link: https://lore.kernel.org/r/20221128060950.3540845-1-chancel.liu@nxp.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Merge series from Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>:
One new JasperLake configuration, core refactoring between RT1316 and
RT1318 and a minor uninitialized variable corner case.
Merge series from Cezary Rojewski <cezary.rojewski@intel.com>:
Even though skylake-driver is going to be replaced by the avs-driver,
the goal is to keep it functional on all the configurations it supports
until its EOL. When comparing chrome trees against upstream
skylake-driver, couple fixes pop up that are not part of upstream
repository. These fixes are backed up by real bugs (issue trackers),
address real problems. There is no reason for them to stay in the
internal tree.
Patches 1-4 combined together address issue where the driver updates the
presumably static audio format descriptions coming from the topology
files through its "fixup" functions. As long as given audio format is
used by a single path, nothing collides and any updates are harmless.
However, when multiple paths e.g.: DMIC and HDMI1 utilize the same audio
format descriptor, any updates caused by the opening of the first path,
may cause the second to fail.
The 5th change from the set fixes driver hang sporadically occurring
during shutdown procedure. Once HDAudio links are powered down along
with the AudioDSP, the hang stops reproducing.
The last change helps survive in environments with limited/fragmented
memory. While the BDL is small already, other buffers can be allocated
using scatter-gather. This basically aligns the code with what the
avs-driver does.
Merge series from Cezary Rojewski <cezary.rojewski@intel.com>:
The patchset focuses on debug functionality for the avs-driver.
Two major blocks are covered here: data probing and AudioDSP firmware
logging. Both are configured and controlled through debugfs.
Data probing is a AudioDSP debug functionality which allows for
gathering the actual data that is being routed to or from a module.
Helps in debugging its processing capabilities - navigate to a specific
module which may have caused a glitch within a pipeline (set of modules
bound together).
First few allow for assigning compress stream to a HDAudio stream, what
is currently limited to pcm substreams only. These patches were already
present on this list and reviewed in the past [1].
The next few tidy existing debug-related code up so it's ready for
addition of new functionalities and make it clear which part of the avs
is debug related and which is not. These also simplify the existing
locking around the trace fifo.
Afterward, debug-related IPCs are defined along with stub soc-component
and compress DAI operations. Not much is done there as it's not a
standard PCM streaming scenario. Most code found in compress operations
is inherited from the HOST side of HDAudio streaming found in pcm.c
file of the driver.
Finally, a debugfs file operations are defined. These facilitate
connecting to DSP modules from which the data shall be gathered as well
as control and configuration of firmware logging. Additionally, entries
are added to allow for dumping snapshots of key memory windows.
sof_es8336_remove() calls cancel_delayed_work(). However, that
function does not wait until the work function finishes. This
means that the callback function may still be running after
the driver's remove function has finished, which would result
in a use-after-free.
Fix by calling cancel_delayed_work_sync(), which ensures that
the work is properly cancelled, no longer running, and unable
to re-schedule itself.
Fixes: 89cdb224f2 ("ASoC: sof_es8336: reduce pop noise on speaker")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Acked-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20221205143721.3988988-1-yangyingliang@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Merge series from V sujith kumar Reddy <Vsujithkumar.Reddy@amd.com>:
Fix an issue with starting the ACP DSP and support debug dumps to aid
maintainability.
Define debugfs subdirectory delegated for IPC communication with DSP.
Input format: uint,uint,(...) which are later translated into DWORDS
sequence and further into instances of struct of interest given the IPC
type.
For Extractor probes, following have been enabled:
- PROBE_POINT_ADD (echo <..> probe_points)
- PROBE_POINT_REMOVE (echo <..> probe_points_remove)
- PROBE_POINT_INFO (cat probe_points)
Signed-off-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://lore.kernel.org/r/20221202152841.672536-14-cezary.rojewski@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Data probing is a cAVS firmware functionality that allows for data
extraction and injection directly from or to DMA stream. To support it,
new functions and types are added. These facilitate communication
with the firmware.
Total of eight IPCs:
- probe module initialization and cleanup
- addition and removal of probe points
- addition and removal of injection DMAs
- dumping list of currently connected probe points or enlisted DMAs
Signed-off-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://lore.kernel.org/r/20221202152841.672536-10-cezary.rojewski@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Debug-related fields and log-dumping are useful when debugfs is enabled.
Define them under CONFIG_DEBUG_FS and provide stubs when the config is
disabled so that the code that makes use of these needs not to be
complicated unnecessarily.
Members that are duplicated by this patch will be removed by the follow
up changes.
Signed-off-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://lore.kernel.org/r/20221202152841.672536-7-cezary.rojewski@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
