ALSA: hda: Skip event processing for unregistered codecs
When codec is unbound but not yet removed, in the eyes of
snd_hdac_bus_process_unsol_events() it is still a valid target to
delegate work to. Such behaviour may lead to use-after-free errors.
Address by verifying if codec is actually registered.
Stefan Binding [Thu, 30 Jun 2022 00:23:35 +0000 (01:23 +0100)]
ALSA: hda: cs35l41: Add module parameter to control firmware load
By default, the driver will automatically load DSP firmware
for the amps, if available. Adding this option allows the
autoload to be optional, which allows for different configurations.
Stefan Binding [Thu, 30 Jun 2022 00:23:34 +0000 (01:23 +0100)]
ALSA: hda: cs35l41: Support Firmware switching and reloading
This is required to support CS35L41 calibration.
By default, speaker protection firmware will be loaded, if
available. However, different firmware is required to run
the calibration sequence, so it is necessary to add support
to be able to unload, switch and reload firmware.
This patch adds 2 ALSA Controls for each amp:
"DSP1 Firmware Load"
"DSP1 Firmware Type"
"DSP1 Firmware Load" can be used to unload and
load the firmware.
"DSP1 Firmware Type" can be used to switch the
target firmware to be loaded by "DSP1 Firmware Load"
Since loading firmware can add new ALSA controls, it is
necessary to ensure the firmware loading is run asynchronously
from the ALSA control itself to prevent deadlocks.
Note: When switching between firmwares, an ALSA control is
only added if it has not previously existed. If it had existed
previously, it will be re-enabled instead.
The config sequences for running with and without firmware and DSP
are different. The original behavior assumed that we would only
run without DSP only in the case where firmware load failed.
This meant the non-firmware sequence was written with the assumtion
that various registers would be set to their default value.
However, to support the ability to unload the firmware, the
non-firmware register sequence must be updated to update all
required registers, including values that would be defaulted,
in case the firmware sequence, which could have already run,
has changed their value.
Stefan Binding [Thu, 30 Jun 2022 00:23:30 +0000 (01:23 +0100)]
ALSA: hda: cs35l41: Support Hibernation during Suspend
CS35L41 supports hibernation during suspend when using
DSP firmware.
When the driver suspends it will hibernate the part, if
firmware is running, and resume will wake from hibernation.
CS35L41 driver will suspend/resume when requested by
hda driver.
Note that suspend/resume and hibernation is only supported
when firmware is running.
Merge tag 'asoc-v5.20' of https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into for-next
ASoC: Updates for v5.20
This is a big release thus far and there will probably be more changes
to come, it's a combination of a larger than usual crop of new drivers
and some subsysetm wide cleanups from Charles rather than anything
structural. The SOF and Intel DSP code both also continue to be very
actively developed.
- Restructing of the set_fmt() callbacks to be specified in terms of
the device rather than with semantics depending on if the device is
supposed to be a CODEC or SoC, making things clearer in situations
like CODEC to CODEC links.
- Clean up of the way we flag which DAI naming scheme we use to reflect
the progress that's been made modernising things.
- Merge of more of the Intel AVS driver stack, including some board
integrations.
- New version 4 mechanism for communication with SOF DSPs.
- Suppoort for dynamically selecting the PLL to use at runtime on i.MX
platforms.
- Improvements for CODEC to CODEC support in the generic cards.
- Support for AMD Jadeite and various machines, Intel MetorLake DSPs,
Mediatek MT8186 DSPs and MT6366, nVidia Tegra MDDRC, OPE and PEQ, NXP
TFA9890, Qualcomm SDM845, WCD9335 and WAS883x, and Texas Instruments
TAS2780.
ASoC: rsnd: care default case on rsnd_ssiu_busif_err_irq_ctrl()
Before, ssiu.c didn't care SSI5-8, thus,
commit 7bedfc26a4d190b ("ASoC: rsnd: care default case on
rsnd_ssiu_busif_err_status_clear()") cares it for status clear.
But we should care it for error irq handling, too.
This patch cares it.
Reported-by: Nguyen Bao Nguyen <nguyen.nguyen.yj@renesas.com> Reported-by: Nishiyama Kunihiko <kunihiko.nishiyama.dn@renesas.com> Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com> Link: https://lore.kernel.org/r/871quocio1.wl-kuninori.morimoto.gx@renesas.com Signed-off-by: Mark Brown <broonie@kernel.org>
The value of "*ppos" was supposed to be zero but it is
sizeof(ipc4_msg->header_u64) so it will copy the data into the middle of
the "ipc4_msg->data_ptr" buffer instead of to the start. The second
problem is "buffer" should be "buffer + sizeof(ipc4_msg->header_u64)".
This function is used for fuzz testing so the data is normally random
and this bug likely does not affect anyone very much.
In this context, it's simpler and more appropriate to use copy_from_user()
instead of simple_write_to_buffer() so I have re-written the function.
Fixes: 73bffe328996 ("ASoC: SOF: ipc-msg-injector: Add support for IPC4 messages") Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Link: https://lore.kernel.org/r/Ysg1tB2FKLnRMsel@kili Signed-off-by: Mark Brown <broonie@kernel.org>
Mark Brown [Wed, 13 Jul 2022 13:04:51 +0000 (14:04 +0100)]
ASoC: rockchip-i2s: Undo BCLK pinctrl changes
The version of the BCLK pinctrl management changes that made it into
v5.19 has caused problems on some systems due to overly strict DT
requirements but attempts to fix it have caused further breakage on
other platforms. Just drop the changes for this release, we already
have a better version queued for -next.
Fixes: cd489a4912f1 ("ASoC: rockchip: i2s: Fix NULL pointer dereference when pinctrl is not found") Fixes: 26dcc16b6d47 ("ASoC: rockchip: i2s: switch BCLK to GPIO") Signed-off-by: Mark Brown <broonie@kernel.org> Link: https://lore.kernel.org/r/20220713130451.31481-1-broonie@kernel.org Signed-off-by: Mark Brown <broonie@kernel.org>
Stefan Binding [Thu, 30 Jun 2022 00:23:28 +0000 (01:23 +0100)]
ALSA: hda: cs35l41: Support multiple load paths for firmware
To be able to support different firmwares and tuning
for different models, the driver needs to be able to
load a different firmware and coefficient file based
on its Subsystem ID.
The driver attempts to load the firmware in the
following order:
Stefan Binding [Thu, 30 Jun 2022 00:23:26 +0000 (01:23 +0100)]
ALSA: hda: cs35l41: Save Subsystem ID inside CS35L41 Driver
The Subsystem ID is read from the HDA driver, and will
be used by the CS35L41 driver to be able to uniquely
identify the laptop, which is required to be able to
define firmware to be used by specific models.
Vitaly Rodionov [Thu, 30 Jun 2022 00:23:25 +0000 (01:23 +0100)]
ALSA: hda: cs35l41: Add initial DSP support and firmware loading
This patch adds support for the CS35L41 DSP.
The DSP allows for extra features, such as running
speaker protection algorithms and hibernations.
To utilize these features, the driver must load
firmware into the DSP, as well as various tuning
files which allow for customization for specific
models.
Stefan Binding [Thu, 30 Jun 2022 00:23:23 +0000 (01:23 +0100)]
ALSA: hda: hda_cs_dsp_ctl: Add apis to write the controls directly
DSP controls are exposed as ALSA controls, however,
some of these controls are required to be accessed by
the driver. Add apis which allow read/write of these
controls. The write api will also notify the ALSA control
on value change.
Stefan Binding [Thu, 30 Jun 2022 00:23:22 +0000 (01:23 +0100)]
ALSA: hda: hda_cs_dsp_ctl: Add Library to support CS_DSP ALSA controls
The cs35l41 part contains a DSP which is able to run firmware.
The cs_dsp library can be used to control the DSP.
These controls can be exposed to userspace using ALSA controls.
This library adds apis to be able to interface between
cs_dsp and hda drivers and expose the relevant controls as
ALSA controls.
[ Note: the dependency of CONFIG_SND_HDA_CS_DSP_CONTROLS Kconfig is
corrected. Also, this Kconfig isn't enabled now but will be
actually enabled in a later patch -- tiwai ]
ALSA: usb-audio: Use atomic_try_cmpxchg in ep_state_update
Use atomic_try_cmpxchg instead of atomic_cmpxchg (*ptr, old, new) == old in
ep_state_update. x86 CMPXCHG instruction returns success in ZF flag,
so this change saves a compare after cmpxchg (and related move instruction
in front of cmpxchg).
When a pipeline is split into FE and BE parts, the BE pipeline may need to
be triggered separately in the BE trigger op. So add the trigger callback
in the link_res ops that will be invoked during BE DAI trigger.
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com> Reviewed-by: Rander Wang <rander.wang@intel.com> Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Acked-by: Vinod Koul <vkoul@kernel.org> Link: https://lore.kernel.org/r/20220708061312.25878-2-yung-chuan.liao@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org>
Liang He [Wed, 13 Jul 2022 10:20:13 +0000 (18:20 +0800)]
ASoC: mt6359: Fix refcount leak bug
In mt6359_parse_dt() and mt6359_accdet_parse_dt(), we should call
of_node_put() for the reference returned by of_get_child_by_name()
which has increased the refcount.
Fixes: e50f23f44fc1 ("ASoC: mt6359: fix failed to parse DT properties") Fixes: ad52183e6765 ("ASoC: mediatek: mt6359: add MT6359 accdet jack driver") Signed-off-by: Liang He <windhl@126.com> Link: https://lore.kernel.org/r/20220713102013.367336-1-windhl@126.com Signed-off-by: Mark Brown <broonie@kernel.org>
Randy Dunlap [Tue, 12 Jul 2022 18:33:48 +0000 (11:33 -0700)]
ASoC: amd: fix Jadeite kconfig warning and build errors
Since SND_SOC_ES8316 has a hard dependency on I2C and since 'select'
does not follow any dependency chains, SND_SOC_AMD_ST_ES8336_MACH
also needs to have a hard dependency on I2C.
Fixes a kconfig warning and subsequent build errors:
sound/soc/codecs/es8316.c:866:1: warning: data definition has no type or storage class
866 | module_i2c_driver(es8316_i2c_driver);
sound/soc/codecs/es8316.c:866:1: error: type defaults to ‘int’ in declaration of ‘module_i2c_driver’ [-Werror=implicit-int]
sound/soc/codecs/es8316.c:866:1: warning: parameter names (without types) in function declaration
sound/soc/codecs/es8316.c:857:26: warning: ‘es8316_i2c_driver’ defined but not used [-Wunused-variable]
857 | static struct i2c_driver es8316_i2c_driver = {
Fixes: 0ba418d5dd1e ("ASoC: amd: enable machine driver build for Jadeite platform") Signed-off-by: Randy Dunlap <rdunlap@infradead.org> Cc: Vijendar Mukunda <Vijendar.Mukunda@amd.com> Cc: Mark Brown <broonie@kernel.org> Cc: alsa-devel@alsa-project.org Cc: Jaroslav Kysela <perex@perex.cz> Cc: Takashi Iwai <tiwai@suse.com> Link: https://lore.kernel.org/r/20220712183348.31046-1-rdunlap@infradead.org Signed-off-by: Mark Brown <broonie@kernel.org>
Liang He [Wed, 13 Jul 2022 07:12:00 +0000 (15:12 +0800)]
ASoc: audio-graph-card2: Fix refcount leak bug in __graph_get_type()
We should call of_node_put() for the reference before its replacement
as it returned by of_get_parent() which has increased the refcount.
Besides, we should also call of_node_put() before return.
Dan Carpenter [Tue, 12 Jul 2022 14:42:12 +0000 (17:42 +0300)]
ASoC: amd: Fix error pointer dereference
The "gpio_pa" pointer is an error pointer, there is no need to try
put it. Calling gpiod_put() on it will lead to an error pointer
dereference.
Fixes: f714af7b4c8a ("ASoC: amd: add Machine driver for Jadeite platform") Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Link: https://lore.kernel.org/r/Ys2IRPHWGIwuVs21@kili Signed-off-by: Mark Brown <broonie@kernel.org>
ALSA: core: Fix missing return value comments for kernel docs
Each kernel doc comment expects the definition of the return value in
a proper format. This patch adds or fixes the missing entries for the
remaining ALSA core API functions.
ALSA: memalloc: Fix missing return value comments for kernel docs
Each kernel doc comment expects the definition of the return value in
a proper format. This patch adds or fixes the missing entries for
memory allocation helpers.
Each kernel doc comment expects the definition of the return value and
the summary for each struct / enum in a proper format. This patch
adds or fixes the missing entries for compress-offload API.
ALSA: dmaengine: Fix missing return value comments for kernel docs
Each kernel doc comment expects the definition of the return value in
a proper format. This patch adds or fixes the missing entries for PCM
dmaengine API.
ALSA: compress: Enable kernel doc markers for some functions
The exported functions snd_compress_new() and snd_compr_stop_error()
had already kernel-doc-style comments but they were not processed as
they weren't marked properly. Let's enable them.
This patch also fixes the missing argument id for snd_compress_new
comments, too.
Mark Brown [Tue, 12 Jul 2022 18:21:45 +0000 (19:21 +0100)]
ASoC: SOF: Intel: hda: Correct Firmware State Register use
Merge series from Peter Ujfalusi <peter.ujfalusi@linux.intel.com>:
The FSR (Firmware State Register) holds the ROM state information, it does not
contain error information.
The FSR itself is a bit more complicated as well as the state depends on the
module currently in use.
The error code from ROM or the status code from the firmware is located at the
next register.
Fix the handling of the FSR in order to provide usable and human readable (in
most cases) report on the status and error.
Mark Brown [Tue, 12 Jul 2022 18:21:42 +0000 (19:21 +0100)]
ASoC: SOF: Intel: Do not process IPC reply before firmware boot
Merge series from Peter Ujfalusi <peter.ujfalusi@linux.intel.com>:
By mistake a developer managed to create a 'corrupted' IPC4 firmware image which
loaded fine to the DSP and after boot it sent an IPC reply before we would have
received the FW_READY message.
It turned out that the image was an IPC3 firmware and the IPC reply was the IPC3
FW_READY notification message which got understood as an IPC4 reply message due
to the difference between the two IPC mechanism.
This caused a NULL pointer dereference since the reply memory will be allocated
after the FW_READY message.
To make sure this will not bite again, skip any spurious reply messages before
the FW_READY.
Mark Brown [Tue, 12 Jul 2022 18:21:41 +0000 (19:21 +0100)]
ASoC: SOF: Intel: hda-dai: No need to decouple host/link DMA twice
Merge series from Peter Ujfalusi <peter.ujfalusi@linux.intel.com>:
There is no need to decouple a decoupled stream twice.
Keep the decoupling in hda_link_stream_assign() only as it is going to be
executed in all cases.
Drop the outdated comment from hda_link_dma_hw_params() as well since the code
has changed around it.
Peter Ujfalusi [Tue, 12 Jul 2022 13:01:03 +0000 (16:01 +0300)]
ASoC: SOF: ipc3-topology: Prevent double freeing of ipc_control_data via load_bytes
We have sanity checks for byte controls and if any of the fail the locally
allocated scontrol->ipc_control_data is freed up, but not set to NULL.
On a rollback path of the error the higher level code will also try to free
the scontrol->ipc_control_data which will eventually going to lead to
memory corruption as double freeing memory is not a good thing.
Fixes: 806f4f0b3d81 ("ASoC: SOF: topology: Make control parsing IPC agnostic") Reported-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com> Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com> Reviewed-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Link: https://lore.kernel.org/r/20220712130103.31514-1-peter.ujfalusi@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org>
Mark Brown [Tue, 12 Jul 2022 15:10:05 +0000 (16:10 +0100)]
ASoC: SOF: Intel: hda: Use cold/purge boot after firmware crash
Merge series from Peter Ujfalusi <peter.ujfalusi@linux.intel.com>:
In case of a firmware crash we force the DSP to be powered down and rebooted.
To make sure that the next boot is going to be clean, force the boot process to
skip the IMR booting and re-download the firmware.
Daniel Baluta [Tue, 12 Jul 2022 14:15:30 +0000 (17:15 +0300)]
ASoC: SOF: compress: Prevent current kernel running with older FW
After introducing extended parameters we need to forbid older firmware
versions to run with the current and future kernel versions.
Although in theory the communication protocol will still work the
semantics at application level are undefined. So, prevent this by
disallowing older firmwares to run with newer kernels.
Signed-off-by: Daniel Baluta <daniel.baluta@nxp.com> Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com> Link: https://lore.kernel.org/r/20220712141531.14599-4-daniel.baluta@oss.nxp.com Signed-off-by: Mark Brown <broonie@kernel.org>
Daniel Baluta [Tue, 12 Jul 2022 14:15:29 +0000 (17:15 +0300)]
ASoC: SOF: Copy compress parameters into extended data
Allocate memory at the end of sof_ipc_stream_params to store
snd_compr_params in order to be sent them to SOF firmware.
This will help firmware correctly configure codecs parameters.
Notice, that we use 2 bytes from the reserved pool in order to store
the extended data length. This is compatible with older FWs where
there was no extended data.
Signed-off-by: Daniel Baluta <daniel.baluta@nxp.com> Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com> Link: https://lore.kernel.org/r/20220712141531.14599-3-daniel.baluta@oss.nxp.com Signed-off-by: Mark Brown <broonie@kernel.org>
Peter Ujfalusi [Tue, 12 Jul 2022 13:16:20 +0000 (16:16 +0300)]
ASoC: SOF: Intel: hda-dai: Do snd_hdac_ext_stream_decouple() only once
Call snd_hdac_ext_stream_decouple_locked() unconditionally in
hda_link_stream_assign(), the snd_hdac_ext_stream_decouple_locked() have
internal checks to avoid re-configuring.
There is no need to call snd_hdac_ext_stream_decouple() via
hda_link_dma_params() as the stream must have been set to decoupled when
it got assigned (even if it used local condition to call
snd_hdac_ext_stream_decouple_locked()).
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com> Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by: Rander Wang <rander.wang@intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Link: https://lore.kernel.org/r/20220712131620.13365-3-peter.ujfalusi@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org>
Peter Ujfalusi [Tue, 12 Jul 2022 13:16:19 +0000 (16:16 +0300)]
ASoC: SOF: Intel: hda-dai: Drop misleading comment regarding dma_data
The comment in hda_link_dma_hw_params() is no longer valid as the dma_data
is set to NULL at system suspend as well.
Instead of rewording the comment to state the obvious: try to take the
hext_stream from the dma_data and if it is not set then assign a new one
and store it as dma_data.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com> Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by: Rander Wang <rander.wang@intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Link: https://lore.kernel.org/r/20220712131620.13365-2-peter.ujfalusi@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org>
Peter Ujfalusi [Tue, 12 Jul 2022 12:57:32 +0000 (15:57 +0300)]
ASoC: SOF: Intel: hda: Correct the ROM/FW state reporting code
The FSR (Firmware State Register) can be found at offset 0 in the SRAM and
it is holding information about the state of the ROM/FW.
In case of a boot failure it can be used to get the state where the boot
process got stuck, it does not itself contains error codes as such.
The error code (or the firmware state information) is stored in the next
soft register at offset 0x4.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com> Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by: Rander Wang <rander.wang@intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Link: https://lore.kernel.org/r/20220712125734.30512-2-peter.ujfalusi@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org>
ASoC: soc-pcm: demote warnings on non-atomic BE connection
When an FE, typically non-atomic, is connected to an atomic BE, we
force the BE as non-atomic. There's no reason to throw a warning, this
is a perfectly fine configuration and a conversion that's required
by-design.
This removes the unconditional warnings such as
[ 12.054213] iDisp1: dpcm_be_connect: FE is nonatomic but BE is not, forcing BE as nonatomic
[ 12.074693] iDisp2: dpcm_be_connect: FE is nonatomic but BE is not, forcing BE as nonatomic
[ 12.096612] iDisp3: dpcm_be_connect: FE is nonatomic but BE is not, forcing BE as nonatomic
[ 12.118637] iDisp4: dpcm_be_connect: FE is nonatomic but BE is not, forcing BE as nonatomic
[ 12.140660] dmic01: dpcm_be_connect: FE is nonatomic but BE is not, forcing BE as nonatomic
[ 12.147521] dmic16k: dpcm_be_connect: FE is nonatomic but BE is not, forcing BE as nonatomic
and demotes them to dev_dbg(), as suggested in review comments.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Link: https://lore.kernel.org/r/20220708200641.26923-1-pierre-louis.bossart@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org>
Peter Ujfalusi [Tue, 12 Jul 2022 12:09:35 +0000 (15:09 +0300)]
ASoC: SOF: Intel: hda: Introduce skip_imr_boot flag
Use a dedicated flag instead of directly checking the
sdev->system_suspend_target to decide if we need to skip IMR boot due to
too deep sleep state where the memory used for IMR booting will not retain
its content.
The skip_imr_boot flag will be set true during suspend if the target state
is deeper than S3 and reset back to false on successful boot to re-enable
IMR booting in shallower sleep states.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com> Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Link: https://lore.kernel.org/r/20220712120936.28072-2-peter.ujfalusi@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org>
Peter Ujfalusi [Tue, 12 Jul 2022 12:23:57 +0000 (15:23 +0300)]
ASoC: SOF: Intel: mtl: Do not process IPC reply before firmware boot
It is not yet clear, but it is possible to create a firmware so broken
that it will send a reply message before a FW_READY message (it is not
yet clear if FW_READY will arrive later).
Since the reply_data is allocated only after the FW_READY message, this
will lead to a NULL pointer dereference if not filtered out.
Reported-by: Kai Vehmanen <kai.vehmanen@linux.intel.com> Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Link: https://lore.kernel.org/r/20220712122357.31282-4-peter.ujfalusi@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org>
Peter Ujfalusi [Tue, 12 Jul 2022 12:23:56 +0000 (15:23 +0300)]
ASoC: SOF: Intel: hda-ipc: Do not process IPC reply before firmware boot
It is not yet clear, but it is possible to create a firmware so broken
that it will send a reply message before a FW_READY message (it is not
yet clear if FW_READY will arrive later).
Since the reply_data is allocated only after the FW_READY message, this
will lead to a NULL pointer dereference if not filtered out.
The issue was reported with IPC4 firmware but the same condition is present
for IPC3.
Reported-by: Kai Vehmanen <kai.vehmanen@linux.intel.com> Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Link: https://lore.kernel.org/r/20220712122357.31282-3-peter.ujfalusi@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org>
Peter Ujfalusi [Tue, 12 Jul 2022 12:23:55 +0000 (15:23 +0300)]
ASoC: SOF: Intel: cnl: Do not process IPC reply before firmware boot
It is not yet clear, but it is possible to create a firmware so broken
that it will send a reply message before a FW_READY message (it is not
yet clear if FW_READY will arrive later).
Since the reply_data is allocated only after the FW_READY message, this
will lead to a NULL pointer dereference if not filtered out.
The issue was reported with IPC4 firmware but the same condition is present
for IPC3.
Reported-by: Kai Vehmanen <kai.vehmanen@linux.intel.com> Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Link: https://lore.kernel.org/r/20220712122357.31282-2-peter.ujfalusi@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org>
The comments are inconsistent and point to the wrong driver name.
The initially named i2c-multi-instantiate it was renamed to the
serial-multi-instantiate exactly due to support of the platforms
with multiple CS35L41 codecs.
Mark Brown [Mon, 11 Jul 2022 21:13:49 +0000 (22:13 +0100)]
ASoC: Intel: avs: Updates and cleanups
Merge series from Cezary Rojewski <cezary.rojewski@intel.com>:
Series consists of loosely connected patches and does not concentrate on
one specific subject.
First, as generic HDAudio codec driver is now part of ASoC, avs-driver
core is updated to register missing ext_bus operations. This completes
driver's core implementation.
The next change adds the last missing piece for port descriptions coming
from topology in formatted string format e.g.: ssp%d have full
effect. To do that, the port value needs to be provided to respective
copier configuration.
Third change relaxes core transition timings so that scenarios where
modules are interfering with each other while being on separate cores
are not occasionally causing trouble.
All other changes are addressing warnings, cleaning things up a little
and protecting driver from invalid firmware behavior - while not
expected in release binaries, does not hurt to add them.
Mark Brown [Mon, 11 Jul 2022 16:00:54 +0000 (17:00 +0100)]
ASoC: codecs: Series of fixes for realtek codecs used on RVPs
Merge series from Amadeusz Sławiński <amadeuszx.slawinski@linux.intel.com>:
Our tests platforms do use realtek codecs, while implementing avs driver
and machine boards for it, we identified some problems with those codec
drivers.
Mark Brown [Mon, 11 Jul 2022 15:18:15 +0000 (16:18 +0100)]
ADD legacy audio driver support for rembrandt
Merge series from V sujith kumar Reddy <Vsujithkumar.Reddy@amd.com>:
Add Generic driver to support multiple platform,
ADD HS control instance for Rembrandt platform.
Add nau8825,max98560 and rt5682s,rt1019 combination support for legacy
platform.
We should only have an error when enforcing strict mapping between
kernel and firmware versions. In all other cases, there is no reason
to throw a warning.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Link: https://lore.kernel.org/r/20220708200719.26961-1-pierre-louis.bossart@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org>
On our RVP platforms using rt298 with combojack we've seen issues with
controls being in incorrect state after suspend/resume cycle. This is
caused by codec driver not setting pins to correct state and causing
codec suspend method to not be called. Which on resume caused codec
registers to be in undefined state. Fix this by setting pins correctly
in jack detect function.
Above problem is caused by the fact that when jack == NULL code doesn't
reach rt298_jack_detect() function which sets pins. Alternatively
problem could be fixed by just moving rt298_jack_detect, but as rt298
codec is similar to rt286, align the code by setting pins explicitly.
Alexandru Elisei [Mon, 11 Jul 2022 13:05:22 +0000 (14:05 +0100)]
ASoC: rockchip: i2s: Fix NULL pointer dereference when pinctrl is not found
Commit 26dcc16b6d47 ("ASoC: rockchip: i2s: switch BCLK to GPIO") switched
BCLK to GPIO functions when probing the i2s bus interface, but missed
adding a check for when devm_pinctrl_get() returns an error. This can lead
to the following NULL pointer dereference on a rockpro64-v2 if there are no
"pinctrl" properties in the i2s device tree node.
Check that i2s->pinctrl is valid before attempting to search for the
bclk_on and bclk_off pinctrl states.
Old code resets SIE for up to 8 streams using byte accessor, but
register is laid out in following way:
31 GIE
30 CIE
29:x Reserved
x-1:0 SIE
If there is more than 8 streams, some of them may and up with enabled
interrupts. To fix this just clear whole INTCTL register when disabling
interrupts.
ALSA: hda: Fix page fault in snd_hda_codec_shutdown()
If early probe of HDAudio bus driver fails e.g.: due to missing
firmware file, snd_hda_codec_shutdown() ends in manipulating
uninitialized codec->pcm_list_head causing page fault.
Iinitialization of HDAudio codec in ASoC is split in two:
- snd_hda_codec_device_init()
- snd_hda_codec_device_new()
snd_hda_codec_device_init() is called during probe_codecs() by HDAudio
bus driver while snd_hda_codec_device_new() is called by
codec-component's ->probe(). The second call will not happen until all
components required by related sound card are present within the ASoC
framework. With firmware failing to load during the PCI's deferred
initialization i.e.: probe_work(), no platform components are ever
registered. HDAudio codec enumeration is done at that point though, so
the codec components became registered to ASoC framework, calling
snd_hda_codec_device_init() in the process.
Now, during platform reboot snd_hda_codec_shutdown() is called for every
codec found on the HDAudio bus causing oops if any of them has not
completed both of their initialization steps. Relocating field
initialization fixes the issue.
ALSA: hda: Fix put_device() inconsistency in error path
AVS HDAudio bus driver does not tie with codec drivers tighly. Codec
device and its respective driver cleanup procedures are split and may
not occur one after the other. Device cleanup is performed only on
snd_hdac_ext_bus_device_remove() i.e. it's the bus driver's
responsibility. If codec component probing fails, put_device() found in
snd_hda_codec_device_new() may lead to page fault. Relocate it to
snd_hda_codec_new() to address the problem on ASoC side while keeping
status quo for snd_hda_intel.
ALSA: hda: Make device usage_count consistent across subsequent probing
AVS HDAudio bus driver does not tie with codec drivers tighly and
snd_hda_codec_device_new() can be called after codec's module reload. In
such case, rpm is forbidden and invoking pm_runtime_forbid()
unconditionally causes device's usage_count to become unbalanced. This
is later caught by WARN_ON() found in sound/soc/hda.c. Detect such
circumstance and bump the usage_count instead.
ALSA: hda: Fix null-ptr-deref when i915 fails and hdmi is denylisted
If snd_hda_hdmi_codec module is denylisted and any event causes i915
enumeration to fail, is_likely_hdmi_codec() ends in null-ptr-deref.
As snd_soc_hda is an ASoC-based driver, its initialization is delayed
until all the necessary components appear in the system - allowing
actual sound card to enumerate. snd_hda_codec_configure() gets called by
the avs-driver core during probe_codecs() but the
snd_hda_codec_device_new(), necessary to complete codecs initialization,
happens only when codec-component of hda sound card is being probed.
Denylisting snd_hda_codec_hdmi module causes snd_hda_codec_configure()
to reach: codec_bind_generic() -> is_likely_hdmi_codec() which makes use
of ->wcaps and at this point the it isn't initialized yet - again,
requires completion of snd_hda_codec_device_new().
Mark Brown [Fri, 8 Jul 2022 20:46:29 +0000 (21:46 +0100)]
ASoC: audio-graph-card2.c: make Codec2Codec settings optional
Merge series from Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>:
Audio Graph Card2 has Codec2Codec support, but I noticed
- Current Codec2Codec setting value is not correct
because it is using of_get_property().
- simple-card-utils has default Codec2Codec settings
and it is overwriting Card2 settings
- This default settings works for non Codec2Codec case
(= DPCM::BE case) too.
Libin Yang [Fri, 8 Jul 2022 20:05:16 +0000 (15:05 -0500)]
ASoC: SOF: ipc4-topology: free memories allocated in sof_ipc4_get_audio_fmt
Free the memories allocated in sof_ipc4_get_audio_fmt in error handling
and ipc_free()
Fixes: da3d3d4d7e53 ("ASoC: SOF: ipc4-topology: Add support for parsing AIF_IN/AIF_OUT widgets") Fixes: 56a22c21c176 ("ASoC: SOF: ipc4-topology: Add support for parsing DAI_IN/DAI_OUT widgets") Fixes: 7c117fe181f0 ("ASoC: SOF: ipc4-topology: Add support for parsing and preparing pga widgets") Fixes: e63ddc5b09a7 ("ASoC: SOF: ipc4-topology: Add support for parsing mixer widgets") Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com> Signed-off-by: Libin Yang <libin.yang@intel.com> Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Link: https://lore.kernel.org/r/20220708200516.26853-3-pierre-louis.bossart@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org>
Libin Yang [Fri, 8 Jul 2022 20:05:15 +0000 (15:05 -0500)]
ASoC: SOF: ipc4-topology: check dai->private in ipc_free()
Set the swidget->private or dai->private to NULL after kfree in the error
handling in ipc_setup(). The private needs to be set NULL because if
ipc_setup() returns error, ipc_free() will be called later. ipc_free()
will judge the private is NULL or not to do the clearing.
For dai widget, dai->private is allocated and set in dai widget
ipc_setup(). So we need to check dai->private is NULL or not
in the ipc_free().
Fixes: da3d3d4d7e53 ("ASoC: SOF: ipc4-topology: Add support for parsing AIF_IN/AIF_OUT widgets") Fixes: 56a22c21c176 ("ASoC: SOF: ipc4-topology: Add support for parsing DAI_IN/DAI_OUT widgets") Fixes: 7c117fe181f0 ("ASoC: SOF: ipc4-topology: Add support for parsing and preparing pga widgets") Fixes: e63ddc5b09a7 ("ASoC: SOF: ipc4-topology: Add support for parsing mixer widgets") Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com> Signed-off-by: Libin Yang <libin.yang@intel.com> Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Link: https://lore.kernel.org/r/20220708200516.26853-2-pierre-louis.bossart@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org>
Module unloading operation performs memory unmapping and the weight of
the opration does not different from any other standard IPC. There is no
dependency on secondary task like in module loading scenario where
larger message timeout is recommended.
ASoC: Intel: avs: Use helper function to set up DMA
dma_set_mask() and dma_set_coherent_mask() can be performed with one
call to dma_set_mask_and_coherent(), which slightly reduces amount of
code on our side.
Apparently it is possible for code to allocate large buffers which may
cause warnings as reported in [1]. This was fixed for HDA, SOF and
skylake in patchset [2], fix it also for avs driver.
To allow for driver's filesystem interfaces e.g.: debugfs, to be touched
even when the device is asleep, mark IPC-channel as blocked when the
device is suspended. This causes any invocation of said interfaces that
do not toggle PM themselves to gracefully fail with "Operation not
permitted" message.
ASoC: Intel: avs: Copy only as many RX bytes as necessary
There is no need to copy number of bytes specified by IPC message caller
if DSP firmware returned lower number. In consequence, LARGE_CONFIG_GET
handler is simplified.
To avoid any false positives when checking CPA after setting SPA, do a
short wait. For stall operation, give HW more time to propagate the
change before moving on.
ASoC: Intel: avs: Assign I2S gateway when parsing topology
For formatted port - ssp%d - descriptions to have an effect, copier
module templates need to be updated with specified port value. This
value is later propagated to the firmware when module instances are
being instantiated.