In the existing IPC support, the reply to each IPC message is handled in
an IRQ thread. The assumption is that the IRQ thread is scheduled without
significant delays.
On an experimental (iow, buggy) kernel, the IRQ thread dealing with the
reply to the last IPC message before powering-down the DSP can be delayed
by several seconds. The IRQ thread will proceed with register accesses
after the DSP is powered-down which results in a kernel crash.
While the bug which causes the delay is not in the audio stack, we must
handle such cases with defensive programming to avoid such crashes.
Call synchronize_irq() before proceeding to power down the DSP to make
sure that no irq thread is pending execution.
Closes: https://github.com/thesofproject/linux/issues/4608
Reviewed-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20231012191850.147140-3-pierre-louis.bossart@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is ignored (apart
from emitting a warning) and this typically results in resource leaks.
To improve here there is a quest to make the remove callback return
void. In the first step of this quest all drivers are converted to
.remove_new(), which already returns void. Eventually after all drivers
are converted, .remove_new() will be renamed to .remove().
The SOF platform drivers all use either sof_of_remove() or
sof_acpi_remove() which both return zero unconditionally. Change these
functions to return void and the drivers to use .remove_new(). There is
no semantical change.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Daniel Baluta <daniel.baluta@nxp.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/20231009155945.285537-1-u.kleine-koenig@pengutronix.de
Signed-off-by: Mark Brown <broonie@kernel.org>
If, for any reason, the open-coded arithmetic causes a wraparound,
the protection that `struct_size()` adds against potential integer
overflows is defeated. Fix this by hardening call to `struct_size()`
with `size_add()`.
Fixes: f9efae9549 ("ASoC: SOF: ipc4-topology: Add support for base config extension")
Signed-off-by: "Gustavo A. R. Silva" <gustavoars@kernel.org>
Reviewed-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/r/ZQSr15AYJpDpipg6@work
Signed-off-by: Mark Brown <broonie@kernel.org>
Setting ACP ACLK as clock source when ACP enters D0 state causing
firmware load failure as mentioned in below scenario.
- Load snd_sof_amd_rembrandt
- Play or Record audio
- Stop audio
- Unload snd_sof_amd_rembrandt
- Reload snd_sof_amd_rembrandt
If acp_clkmux_sel register field is set, then clock source will be
set to ACP ACLK when ACP enters D0 state.
During stream stop, if there is no active stream is running then
acp firmware will set the ACP ACLK value to zero.
When driver is reloaded and clock source is selected as ACP ACLK,
as ACP ACLK is programmed to zero, firmware loading will fail.
For RMB platform, remove the clock mux selection field so that
ACP will use internal clock source when ACP enters D0 state.
Fixes: 41cb85bc4b ("ASoC: SOF: amd: Add support for Rembrandt plaform.")
Reported-by: coolstar <coolstarorganization@gmail.com>
Closes: https://github.com/thesofproject/sof/issues/8137
Signed-off-by: Vijendar Mukunda <Vijendar.Mukunda@amd.com>
Link: https://lore.kernel.org/r/20230927071412.2416250-1-Vijendar.Mukunda@amd.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Merge series from Peter Ujfalusi <peter.ujfalusi@linux.intel.com>:
When a firmware crashes it creats a panic information into a telemetry
slot. The panic format is defined by Zephyr, includes stack and
additional information to help to identify the reason for the crash.
Part of the firmware exception handling the firmware also sends an
EXCEPTION_CAUGHT notification.
This series implements the kernel side handling of the exception: print
information into the kernel log export the whole telemetry slot to user
space for tools extract additional information from the panic dump.
Merge series from Peter Ujfalusi <peter.ujfalusi@linux.intel.com>:
Currently IPC4 has no notion of a switch or enum type of control which
is a generic concept in ALSA.
The generic support for these control types will be as follows:
- large config is used to send the channel-value par array
- param_id of a SWITCH type is 200
- param_id of an ENUM type is 201
Each module need to support a switch or/and enum must handle these
universal param_ids. The message payload is described by struct
sof_ipc4_control_msg_payload.
Currently IPC4 has no notion of a switch or enum type of control which is
a generic concept in ALSA.
The generic support for these control types will be as follows:
- large config is used to send the channel-value par array
- param_id of a SWITCH type is 200
- param_id of an ENUM type is 201
Each module need to support a switch or/and enum must handle these
universal param_ids.
The message payload is described by struct sof_ipc4_control_msg_payload.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20230919103115.30783-2-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Merge series from Peter Ujfalusi <peter.ujfalusi@linux.intel.com>:
If the copier supports a single format on the DAI side we should fixup the BE
dailink to use this single format.
Merge series from Peter Ujfalusi <peter.ujfalusi@linux.intel.com>:
This series implements support for split library loading to comply with the HDA
DMA programming sequence recommendation, which is:
1. DSP side DMA programming and GEN bit set to 1
2. host side DMA programming and RUN bit set to 1
The SOF support for library loading is based on this sequence, backwards
compatibility with older reference firmware is supported (where only the
LOAD_LIBRARY message is supported).
There is a certain sequence needs to be followed when configuring the HDA
DMA in host and DSP.
The firmware provides a way to handle this two stage sequencing by
splitting the library loading into two stage:
1st stage: LOAD_LIBRARY_PREPARE message
the lib_id is 0, used to configure the DMA on DSP side
2nd stage: LOAD_LIBRARY message
both dma_id and lib_id is valid, used for the actual transfer of
the library
In case a firmware without support for this two stage loading is used then
the second stage message will trigger the loading and the first stage will
return with error, which is ignored by the kernel.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Reviewed-by: Chao Song <chao.song@linux.intel.com>
Link: https://lore.kernel.org/r/20230915114018.1701-5-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
On Intel platforms there is a strict order requirement for the DMA
programming:
DSP side configures the buffer and sets the GEN bit
Host side sets the RUN bit.
In order to follow this flow, a new global message type has been added to
prepare the DSP side of the DMA:
host sends LOAD_LIBRARY_PREPARE with the dma_id
DSP side sets its buffer and sets the GEN bit
Host sets the RUN bit
Host sends LOAD_LIBRARY with dma_id and lib_id
DSP receives the library data.
It is up to the platform code to use the new prepare stage message and how
to handle the reply to it from the firmware, which can indicate that the
message type is not supported/handled.
In this case the kernel should proceed to the LOAD_LIBRARY stage assuming
a single stage library loading:
host sends LOAD_LIBRARY_PREPARE with the dma_id
DSP replies that the message type is not supported/handled
Host acknowledges the return code and sets the RUN bit
Host sends LOAD_LIBRARY with dma_id and lib_id
DSP receives the library data.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Reviewed-by: Chao Song <chao.song@linux.intel.com>
Link: https://lore.kernel.org/r/20230915114018.1701-4-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
When a copier exposes a single format, we can fixup the BE dailink with
that format. This is helpful when some codec have format restrictions and
e.g. don't support a 32-bit format. In that case, the copier output
formats mirror that restriction in the topology file.
An alternate solution was suggested earlier using a dedicated topology
token. When specified, the token would be used to fix-up the dailink. The
main reason why this solution was chosen is that there is a risk of a
disconnect between token definition and copier format. With a single piece
of information as suggested in this patch, there are fewer risks of a bad
configuration.
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Link: https://lore.kernel.org/r/20230915093507.7242-3-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
