The ASoC core has now been changed to default to the non-legacy DAI
naming, as such drivers using the new scheme no longer need to specify
the non_legacy_dai_naming flag.
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20220623125250.2355471-38-ckeepax@opensource.cirrus.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Change the legacy DAI naming flag from opting in to the new scheme
(non_legacy_dai_naming), to opting out of it (legacy_dai_naming).
These drivers appear to be on the CPU side of the DAI link and
currently uses the legacy naming, so add the new flag.
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20220623125250.2355471-17-ckeepax@opensource.cirrus.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Merge series from Charles Keepax <ckeepax@opensource.cirrus.com>:
Currently the set_fmt callback always passes clock provider/consumer
with respect to the CODEC. This made sense when the framework was
directly broken down into platforms and CODECs. However, as things
are now broken down into components which can be connected as either
the CPU or CODEC side of a DAI link it simplifies things if each
side of the link is just told if it is provider or consumer of the
clocks. Making this change allows us to remove one of the last parts
of the ASoC core that needs to know if a driver is a CODEC driver,
where it flips the clock format specifier if a CODEC driver is used on
the CPU side of a DAI link, as well as just being conceptually more
consistent with componentisation.
The basic idea of this patch chain is to change the set_fmt callback
from specifying if the CODEC is provider/consumer into directly
specifying if the component is provider/consumer. To do this we add
some new defines, and then to preserve bisectability, the migration is
done by adding a new callback, converting over all existing CPU side
drivers, converting the core, and then finally reverting back to the
old callback.
Converting the platform drivers makes sense as the existing defines
are from the perspective of the CODEC and there are more CODEC drivers
than platform drivers.
Obviously a fair amount of this patch chain I was only able to build
test, so any testing that can be done would be greatly appreciated.
cppcheck complains about invalid NULL dereferences but there's indeed
no need to initialize a loop variable.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com>
Link: https://lore.kernel.org/r/20220520211719.607543-17-pierre-louis.bossart@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Now the core has been migrated across to the new direct clock
specification we can move the drivers back to the normal set_fmt
callback.
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20220519154318.2153729-42-ckeepax@opensource.cirrus.com
Signed-off-by: Mark Brown <broonie@kernel.org>
As part of updating the core to directly tell drivers if they are clock
provider or consumer update these CPU side drivers to use the new direct
callback.
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20220519154318.2153729-15-ckeepax@opensource.cirrus.com
Signed-off-by: Mark Brown <broonie@kernel.org>
This commit e138233e56 causes the
following system crash when using audio on G12A/G12B & SM1 systems:
BUG: sleeping function called from invalid context at kernel/locking/mutex.c:282
in_atomic(): 1, irqs_disabled(): 128, non_block: 0, pid: 0, name: swapper/0
preempt_count: 10001, expected: 0
RCU nest depth: 0, expected: 0
Preemption disabled at:
schedule_preempt_disabled+0x20/0x2c
mutex_lock+0x24/0x60
_snd_pcm_stream_lock_irqsave+0x20/0x3c
snd_pcm_period_elapsed+0x24/0xa4
axg_fifo_pcm_irq_block+0x64/0xdc
__handle_irq_event_percpu+0x104/0x264
handle_irq_event+0x48/0xb4
...
start_kernel+0x3f0/0x484
__primary_switched+0xc0/0xc8
Revert this commit until the crash is fixed.
Fixes: e138233e56 ("ASoC: meson: axg-card: make links nonatomic")
Reported-by: Dmitry Shmidt <dimitrysh@google.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Acked-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20220421155725.2589089-2-narmstrong@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
This reverts commit bf5e4887ee because
the following and required commit e138233e56
causes the following system crash when using audio:
BUG: sleeping function called from invalid context at kernel/locking/mutex.c:282
Fixes: bf5e4887ee ("ASoC: meson: axg-tdm-interface: manage formatters in trigger")
Reported-by: Dmitry Shmidt <dimitrysh@google.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Acked-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20220421155725.2589089-1-narmstrong@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The G12A tohdmi has a custom put() operation which returns 0 when the value
of the mux changes, meaning that events are not generated for userspace.
Change to return 1 in this case, the function returns early in the case
where there is no change.
Signed-off-by: Mark Brown <broonie@kernel.org>
Reviewed-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20220421123803.292063-4-broonie@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
The AIU CODEC has a custom put() operation which returns 0 when the value
of the mux changes, meaning that events are not generated for userspace.
Change to return 1 in this case, the function returns early in the case
where there is no change.
Signed-off-by: Mark Brown <broonie@kernel.org>
Reviewed-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20220421123803.292063-3-broonie@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
The AIU ACODEC has a custom put() operation which returns 0 when the value
of the mux changes, meaning that events are not generated for userspace.
Change to return 1 in this case, the function returns early in the case
where there is no change.
Signed-off-by: Mark Brown <broonie@kernel.org>
Reviewed-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20220421123803.292063-2-broonie@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
On a S905W-based system I get the following error:
debugfs: Directory 'c1105400.audio-controller' with parent 'P230-Q200' already present!
Turned out that multiple components having the same name triggers this
error in soc_init_component_debugfs(). With the patch the error is gone
and that's the debugfs entries.
/sys/kernel/debug/asoc/P230-Q200/acodec:c1105400.audio-controller
/sys/kernel/debug/asoc/P230-Q200/hdmi:c1105400.audio-controller
/sys/kernel/debug/asoc/P230-Q200/cpu:c1105400.audio-controller
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Link: https://lore.kernel.org/r/38053baf-c33b-7fdf-7593-99b22153a9c0@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The out-of-tree vendor driver uses the following approach to set the
AIU_I2S_MISC register:
1) write AIU_MEM_I2S_START_PTR and AIU_MEM_I2S_RD_PTR
2) configure AIU_I2S_MUTE_SWAP[15:0]
3) write AIU_MEM_I2S_END_PTR
4) set AIU_I2S_MISC[2] to 1 (documented as: "put I2S interface in hold
mode")
5) set AIU_I2S_MISC[4] to 1 (depending on the driver revision it always
stays at 1 while for older drivers this bit is unset in step 4)
6) set AIU_I2S_MISC[2] to 0
7) write AIU_MEM_I2S_MASKS
8) toggle AIU_MEM_I2S_CONTROL[0]
9) toggle AIU_MEM_I2S_BUF_CNTL[0]
Move setting the AIU_I2S_MISC[2] bit to aiu_fifo_i2s_hw_params() so it
resembles the flow in the vendor kernel more closely. While here also
configure AIU_I2S_MISC[4] (documented as: "force each audio data to
left or right according to the bit attached with the audio data")
similar to how the vendor driver does this. This fixes the infamous and
long-standing "machine gun noise" issue (a buffer underrun issue).
Fixes: 6ae9ca9ce9 ("ASoC: meson: aiu: add i2s and spdif support")
Reported-by: Christian Hewitt <christianshewitt@gmail.com>
Reported-by: Geraldo Nascimento <geraldogabriel@gmail.com>
Tested-by: Christian Hewitt <christianshewitt@gmail.com>
Tested-by: Geraldo Nascimento <geraldogabriel@gmail.com>
Acked-by: Jerome Brunet <jbrunet@baylibre.com>
Cc: stable@vger.kernel.org
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Link: https://lore.kernel.org/r/20211206210804.2512999-3-martin.blumenstingl@googlemail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The FIFO registers which take an DMA-able address are only 32-bit wide
on AIU. Add dma_coerce_mask_and_coherent() to make the DMA core aware of
this limitation.
Fixes: 6ae9ca9ce9 ("ASoC: meson: aiu: add i2s and spdif support")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Link: https://lore.kernel.org/r/20211206210804.2512999-2-martin.blumenstingl@googlemail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
So far, the formatters have been reset/enabled using the .prepare()
callback. This was done in this callback because walking the formatters use
a mutex so it could not be done in .trigger(), which is atomic by default.
It turns out there is a problem on capture path of the AXG series.
The FIFO may get out of sync with the TDM decoder if the IP are not enabled
in a specific order. The FIFO must be enabled before the formatter starts
producing data. IOW, we must deal with FE before the BE. The .prepare()
callback is called on the BEs before the FE so it is not OK for the AXG.
The .trigger() callback order can be configured, and it deals with the FE
before the BEs by default. To solve our problem, we just need to start and
stop the formatters from the .trigger() callback. It is OK do so now that
the links have been made 'nonatomic' in the card driver.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20211020114217.133153-3-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Non atomic operations need to be performed in the trigger callback
of the TDM interfaces. Those are BEs but what matters is the nonatomic
flag of the FE in the DPCM context. Just set nonatomic for everything so,
at least, it is clear.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20211020114217.133153-2-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Implement driver_name to provide an alternative to card_name for userspace
configuration of Amlogic audio cards.
Suggested-by: Matthias Reichl <hias@horus.com>
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Acked-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20211017160028.23318-1-christianshewitt@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Do nothing if format was zero at snd_soc_runtime_set_dai_fmt().
soc-core.c can be more simple code by this patch.
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/87ee8jt7d3.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Mark Brown <broonie@kernel.org>
There is a spelling mistake in a dev_err error message. Fix it.
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Link: https://lore.kernel.org/r/20210924231242.144692-1-colin.king@canonical.com
Signed-off-by: Mark Brown <broonie@kernel.org>
This patch switch to use snd_soc_daifmt_parse_format/clock_provider() from
snd_soc_of_parse_daifmt().
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Reviewed-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/8735tlw9cy.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Mark Brown <broonie@kernel.org>
This adds support for the TOACODEC found in Amlogic SM1 SoCs.
The bits are shifted for more selection of clock sources, so this only
maps the same support for G12A to the SM1 bits.
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://lore.kernel.org/r/20210511074829.4110036-3-narmstrong@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Switch usage to regmap field for bits handled by the g12a_toacodec_mux_put_enum()
function to avoid uselesss code duplication when adding SM1 variant support.
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://lore.kernel.org/r/20210511074829.4110036-2-narmstrong@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Previous fifo depth patch was only tested on axg, not g12 or sm1.
Of course, while adding hw_params dai callback for the axg, I forgot to do
the same for g12 and sm1, leaving the depth unset and breaking playback on
these SoCs.
Add hw_params callback to the g12 dai_ops to fix the problem.
Fixes: 6f68accaa8 ("ASoC: meson: axg-frddr: set fifo depth according to the period")
Reported-by: Christian Hewitt <christianshewitt@gmail.com>
Tested-by: Christian Hewitt <christianshewitt@gmail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20210412132256.89920-1-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
On the AXG family, the fifo irq is not necessary for the HW to operate.
It is just used to notify that a period has elapsed. If userpace does not
care for these wakeups (such as pipewire), we are just wasting CPU cycles.
Add support for NO_PERIOD_WAKEUP and disable irq when they are no needed.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20210407145914.311479-1-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
When the period is small, using all the FRDDR fifo depth increases the
latency of the playback because the following device won't start pulling
data until the fifo reaches the depth set. We can adjust this depth so trim
it down for small periods.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20210407145714.311138-1-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
cppcheck complains about potential null pointer dereference but it's
rather an unnecessary assignment to NULL before walking through a
list.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20210326215927.936377-7-pierre-louis.bossart@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
cppcheck complains about potential null pointer dereference but it's
rather an unnecessary assignment to NULL before walking through a
list.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20210326215927.936377-6-pierre-louis.bossart@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The pointer to of_phandle_args passed to snd_soc_get_dai_name() and
of_xlate_dai_name() implementations is not modified. Since it is being
used only to translate passed OF node to a DAI name, it should not be
modified, so mark it as const for correctness and safer code.
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Link: https://lore.kernel.org/r/20210221153024.453583-1-krzk@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Using a managed buffer will pre-allocate the buffer using
snd_pcm_lib_preallocate_pages() and automatically free it when the PCM is
destroyed.
In addition it will call snd_pcm_lib_malloc_pages() before the driver's
hw_params() callback and snd_pcm_lib_free_pages() after the driver's
hw_free() callback.
This slightly reduces the boilerplate code of the driver.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Reviewed-by: Jerome Brunet <jbrunet@baylibre.com>
Tested-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20201218154544.25513-1-lars@metafoo.de
Signed-off-by: Mark Brown <broonie@kernel.org>
When the axg-tdm-interface was introduced, the backend DAI was marked as an
endpoint when DPCM was walking the DAPM graph to find a its BE.
It is no longer the case since this
commit 8dd26dff00 ("ASoC: dapm: Fix handling of custom_stop_condition on DAPM graph walks")
Because of this, when DPCM finds a BE it does everything it needs on the
DAIs but it won't power up the widgets between the FE and the BE if there
is no actual endpoint after the BE.
On meson-axg HWs, the loopback is a special DAI of the tdm-interface BE.
It is only linked to the dummy codec since there no actual HW after it.
>From the DAPM perspective, the DAI has no endpoint. Because of this, the TDM
decoder, which is a widget between the FE and BE is not powered up.
>From the user perspective, everything seems fine but no data is produced.
Connecting the Loopback DAI to a dummy DAPM endpoint solves the problem.
Fixes: 8dd26dff00 ("ASoC: dapm: Fix handling of custom_stop_condition on DAPM graph walks")
Cc: Charles Keepax <ckeepax@opensource.cirrus.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20201217150812.3247405-1-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The signal captured on from tdm decoder of the AXG SoC is incorrect. It
appears amplified. The skew offset of the decoder is wrong.
Setting the skew offset to 3, like the g12 and sm1 SoCs, solves and gives
correct data.
Fixes: 13a22e6a98 ("ASoC: meson: add tdm input driver")
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20201217150834.3247526-1-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The driver can be compile tested with !CONFIG_OF making certain data
unused:
sound/soc/meson/t9015.c:315:34: warning: ‘t9015_ids’ defined but not used [-Wunused-const-variable=]
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Link: https://lore.kernel.org/r/20201125164452.89239-9-krzk@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
When compiled with CONFIG_HAVE_CLK, the kernel need to get provider for the
clock API. This is usually selected by the platform and the sound drivers
should not really care about this. However COMPILE_TEST is special and the
platform required may not have been selected, leading to this type of
error:
> aiu-encoder-spdif.c:(.text+0x3a0): undefined reference to `clk_set_parent'
Since we need a sane provider of the API with COMPILE_TEST, depends on
COMMON_CLK.
Fixes: 6dc4fa179f ("ASoC: meson: add axg fifo base driver")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20201116172423.546855-1-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
On g12 and following platforms, The first channel of record with more than
2 channels ends being placed randomly on an even channel of the output.
On these SoCs, a bit was added to force the first channel to be placed at
the beginning of the output. Apparently the behavior if the bit is not set
is not easily predictable. According to the documentation, this bit is not
present on the axg series.
Set the bit on g12 and fix the problem.
Fixes: a3c23a8ad4 ("ASoC: meson: axg-toddr: add g12a support")
Reported-by: Nicolas Belin <nbelin@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20200828151438.350974-1-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Use the new common snd_soc_of_parse_aux_devs() helper function
to parse auxiliary devices from the device tree. The new helper
is just a copy of meson_card_add_aux_devices() so there is no
functional change.
Signed-off-by: Stephan Gerhold <stephan@gerhold.net>
Link: https://lore.kernel.org/r/20200801100257.22658-3-stephan@gerhold.net
Signed-off-by: Mark Brown <broonie@kernel.org>
Commit b73287f0b0 ("ASoC: soc-pcm: dpcm: fix playback/capture checks")
changed the meaning of dpcm_playback/dpcm_capture and now requires the
CPU DAI BE to aligned with those flags.
This broke all Amlogic cards with uni-directional backends (All gx and
most axg cards).
While I'm still confused as to how this change is an improvement, those
cards can't remain broken forever. Hopefully, next time an API change is
done like that, all the users will be updated as part of the change, and
not left to fend for themselves.
Fixes: b73287f0b0 ("ASoC: soc-pcm: dpcm: fix playback/capture checks")
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20200731120603.2243261-1-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Standard dai format property don't need the "amlogic," prefix.
There nothing amlogic specific about them. Just remove it.
Fixes: 435857e015 ("ASoC: meson: align axg card driver with DT bindings documentation")
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20200729154456.1983396-5-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
After carefully checking, it appears that both tdmout and tdmin require the
rising edge of the sclk they get to be synchronized with the frame sync
event (which should be a rising edge of lrclk).
TDMIN was improperly set before this patch. Remove the sclk_invert quirk
which is no longer needed and fix the sclk phase.
Fixes: 1a11d88f49 ("ASoC: meson: add tdm formatter base driver")
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20200729154456.1983396-4-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Shaomin Deng