Hi,
The series is inspired by the effort to standardize TI's arm64 dtsi files to keep
all nodes in 'okay' state and let the board dts files disable not needed
peripherals (and not properly configured):
https://lore.kernel.org/lkml/20201104224356.18040-1-nm@ti.com/
In the unlikely (or likely?) event when the dts misses to disable the McASP node
which is not configured we currenly and luckily just manage to not crash as we
had fixup code in place for legacy pdata boots.
This however prints out a message which does not really help to identify the
issue.
This series will reduce some of the noise during boot (first patch) then
adds the needed changes to handle the variations of 'okay':
A - have all required DT properties for audio
B - gpiochip is enabled
A && !B -> everything is OK for audio, no gpiochip registered
A && B -> everything is OK for audio, gpiochip is registered
!A && B -> audio is not OK, gpiochip is registered. dev_dbg() for audio and do
not register SOC DAI and PCM
!A && !B -> audio is not OK, no gpiochip. dev_err() and fail the probe
Regards,
Peter
---
Peter Ujfalusi (4):
ASoC: ti: davinci-mcasp: Use platform_get_irq_byname_optional
ASoC: ti: davinci-mcasp: Remove legacy dma_request parsing
ASoC: ti: davinci-mcasp: Simplify the configuration parameter handling
ASoC: ti: davinci-mcasp: Handle missing required DT properties
sound/soc/ti/davinci-mcasp.c | 294 ++++++++++++++---------------------
1 file changed, 119 insertions(+), 175 deletions(-)
--
Peter
Texas Instruments Finland Oy, Porkkalankatu 22, 00180 Helsinki.
Y-tunnus/Business ID: 0615521-4. Kotipaikka/Domicile: Helsinki
platform_get_irq() would print error message internally, so dev_err()
after platform_get_irq() is not needed
Signed-off-by: Shengjiu Wang <shengjiu.wang@nxp.com>
Acked-by: Nicolin Chen <nicoleotsuka@gmail.com>
Link: https://lore.kernel.org/r/1604715643-29507-1-git-send-email-shengjiu.wang@nxp.com
Signed-off-by: Mark Brown <broonie@kernel.org>
McASP needs three required properties to be usable for audio:
op-mode, tdm-slots and the serial-dir array.
Instead of probing the driver even without the needed information we should
make sure that all the parameters are provided for operation.
The fact that McASP can act as a GPIO controller for it's pins complicates
this a bit, but as a general rule we can:
- we fail the probe if McASP is not configured to be used as gpiochip
- we will not register the DAI (and PCM) if gpiochip is defined
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Link: https://lore.kernel.org/r/20201106072551.689-5-peter.ujfalusi@ti.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Replace the davinci_mcasp_set_pdata_from_of() function which returned a
pdata pointer with davinci_mcasp_get_config() to return an actual error
code and handle all pdata validation and private mcasp struct setup in
there.
Drop the unused ram-size-playback and sram-size-capture query from DT at
the same time.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Link: https://lore.kernel.org/r/20201106072551.689-4-peter.ujfalusi@ti.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The legacy dma_request (which was holding the DMA request number) is no
longer in use for a long time.
All legacy platforms has been converted to dma_slave_map.
Remove it along with the DT parsing to get tx_dma_channel and
rx_dma_channel.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Link: https://lore.kernel.org/r/20201106072551.689-3-peter.ujfalusi@ti.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Depending on the integration of McASP either the 'common' or the
'rx' and 'tx' or only the 'tx' interrupt number is valid, provided.
By switching to platform_get_irq_byname_optional() we can clean up the
bootlog from messages like:
davinci-mcasp 2ba0000.mcasp: IRQ common not found
The irq number == 0 is not valid, fix the check at the same time.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Link: https://lore.kernel.org/r/20201106072551.689-2-peter.ujfalusi@ti.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The AUD2HTX is a digital module that provides a bridge between
the Audio Subsystem and the HDMI RTX Subsystem. This module
includes intermediate storage to queue SDMA transactions prior
to being synchronized and passed to the HDMI RTX Subsystem over
the Audio Link.
The AUD2HTX contains a DMA request routed to the SDMA module.
This DMA request is controlled based on the watermark level in
the 32-entry sample buffer.
Signed-off-by: Shengjiu Wang <shengjiu.wang@nxp.com>
Acked-by: Nicolin Chen <nicoleotsuka@gmail.com>
Link: https://lore.kernel.org/r/1604281947-26874-2-git-send-email-shengjiu.wang@nxp.com
Signed-off-by: Mark Brown <broonie@kernel.org>
ecause clk_prepare_enable() and clk_disable_unprepare() already checked
NULL clock parameter, so the additional checks are unnecessary, just
remove them.
Signed-off-by: Xu Wang <vulab@iscas.ac.cn>
Link: https://lore.kernel.org/r/20201029090104.29552-1-vulab@iscas.ac.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
Because clk_disable_unprepare() already checked NULL clock parameter,
so the additional check is unnecessary, just remove it.
Signed-off-by: Xu Wang <vulab@iscas.ac.cn>
Link: https://lore.kernel.org/r/20201029084137.28771-1-vulab@iscas.ac.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
This patch fixes below warning when module is compiled with W=1 C=1
sound/soc/qcom/qdsp6/q6afe-clocks.c:122:18: warning: symbol 'q6afe_clks'
was not declared. Should it be static?
Fixes: 520a1c396d ("ASoC: q6afe-clocks: add q6afe clock controller")
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Link: https://lore.kernel.org/r/20201105114435.22860-1-srinivas.kandagatla@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
soc-pcm's dpcm_fe_dai_do_trigger() supported DRAIN commnad up to kernel
v5.4 where explicit switch(cmd) has been introduced which takes into
account all SNDRV_PCM_TRIGGER_xxx but SNDRV_PCM_TRIGGER_DRAIN. Update
switch statement to reactive support for it.
As DRAIN is somewhat unique by lacking negative/stop counterpart, bring
behaviour of dpcm_fe_dai_do_trigger() for said command back to its
pre-v5.4 state by adding it to START/RESUME/PAUSE_RELEASE group.
Fixes: acbf27746e ("ASoC: pcm: update FE/BE trigger order based on the command")
Signed-off-by: Cezary Rojewski <cezary.rojewski@intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Link: https://lore.kernel.org/r/20201026100129.8216-1-cezary.rojewski@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Fix the following sparse warnings:
./mt8192-dai-i2s.c:2040:5: warning: symbol 'mt8192_dai_i2s_get_share' was not declared. Should it be static?
./mt8192-dai-i2s.c:2060:5: warning: symbol 'mt8192_dai_i2s_set_priv' was not declared. Should it be static?
./mt8192-afe-gpio.c:15:16: warning: symbol 'aud_pinctrl' was not declared. Should it be static?
./mt8192-afe-pcm.c:70:5: warning: symbol 'mt8192_get_memif_pbuf_size' was not declared. Should it be static?
./mt8192-afe-pcm.c:2137:39: warning: symbol 'mt8192_afe_component' was not declared. Should it be static?
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Zou Wei <zou_wei@huawei.com>
Acked-by: Jiaxin Yu <jiaxin.yu@mediatek.com>
Link: https://lore.kernel.org/r/1604579287-25251-1-git-send-email-zou_wei@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
A build of arm64 allmodconfig with next-20201105 fails with the error:
ERROR: modpost: "mt8192_afe_gpio_request" undefined!
ERROR: modpost: "mt8192_afe_gpio_init" undefined!
Export the symbols so that mt8192-mt6359-rt1015-rt5682.ko finds it.
Signed-off-by: Sudip Mukherjee <sudipm.mukherjee@gmail.com>
Link: https://lore.kernel.org/r/20201105124747.18383-1-sudipm.mukherjee@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
This patch fixes below warning when module is compiled with W=1 C=1
lpass-cpu.c:677:22: warning: symbol 'lpass_hdmi_regmap_config'
was not declared. Should it be static?
Fixes: 7cb37b7bd0 ("ASoC: qcom: Add support for lpass hdmi driver")
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Link: https://lore.kernel.org/r/20201105114100.18647-1-srinivas.kandagatla@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Almost all other allocations in ASoC API are resource managed, the only
exception is soc-topology.c.
This patchset clean ups few unnecessary functions in preparation for
change and then changes to devm_ functions for allocation.
Amadeusz Sławiński (6):
ASoC: topology: Remove unused functions from topology API
ASoC: topology: Remove multistep topology loading
ASoC: topology: Unify all device references
ASoC: topology: Change allocations to resource managed
ASoC: topology: Remove empty functions
ASoC: topology: Simplify remove_widget function
include/sound/soc-topology.h | 13 +-
sound/soc/intel/skylake/skl-topology.c | 9 +-
sound/soc/soc-topology.c | 302 +++++--------------------
sound/soc/sof/pcm.c | 2 +-
sound/soc/sof/topology.c | 4 +-
5 files changed, 57 insertions(+), 273 deletions(-)
--
2.25.1
Add TDM slot callback function to support TDM configuration
Signed-off-by: Shuming Fan <shumingf@realtek.com>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20201104092005.2227-1-shumingf@realtek.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Now that enum and mixer kcontrols are freed by resource management
framework, removing kcontrol becomes one function call, so simplify code
in remove_widget.
Reviewed-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Signed-off-by: Amadeusz Sławiński <amadeuszx.slawinski@linux.intel.com>
Reviewed-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://lore.kernel.org/r/20201030145427.3497990-7-amadeuszx.slawinski@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
After changing memory management to resource managed one, some of the
functions became no ops, remove them as they are no longer needed.
Reviewed-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Signed-off-by: Amadeusz Sławiński <amadeuszx.slawinski@linux.intel.com>
Reviewed-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://lore.kernel.org/r/20201030145427.3497990-6-amadeuszx.slawinski@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
In order for topology to be resource managed, change all allocations to
be resource managed:
k*alloc -> devm_k*alloc
kstrdup -> devm_kstrdup
Exceptions where non resource managed allocation is left is
soc_tplg_dapm_widget_create(), as it uses pointer to memory locally and
frees it up after use, as well as soc_tplg_dapm_graph_elems_load(),
which has temporary pointer to table of routes.
After conversion all redundant calls in error and clean up paths were
removed.
Also removed some variables which become unneeded when there is no calls
using them.
Reviewed-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Signed-off-by: Amadeusz Sławiński <amadeuszx.slawinski@linux.intel.com>
Reviewed-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://lore.kernel.org/r/20201030145427.3497990-5-amadeuszx.slawinski@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
In few places tplg->comp->dev is used, while everywhere else tplg->dev
is being used. Unify those uses towards tplg->dev, as it is being set to
comp->dev during initialization anyway.
Reviewed-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Signed-off-by: Amadeusz Sławiński <amadeuszx.slawinski@linux.intel.com>
Reviewed-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://lore.kernel.org/r/20201030145427.3497990-4-amadeuszx.slawinski@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
In theory topology can be loaded in multiple steps by providing index to
snd_soc_tplg_component_load, however, from usability point of view it
doesn't make sense, as can be seen from all current users loading
topology in one go. Remove the unnecessary parameter.
Reviewed-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Signed-off-by: Amadeusz Sławiński <amadeuszx.slawinski@linux.intel.com>
Reviewed-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://lore.kernel.org/r/20201030145427.3497990-3-amadeuszx.slawinski@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Topology API exposes snd_soc_tplg_widget_remove and
snd_soc_tplg_widget_remove_all, but both are nowhere used. All current
users load and unload topology as a whole. As following commits
introduce resource managed memory, remove them to simplify code and
reduce maintenance burden.
Reviewed-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Signed-off-by: Amadeusz Sławiński <amadeuszx.slawinski@linux.intel.com>
Reviewed-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://lore.kernel.org/r/20201030145427.3497990-2-amadeuszx.slawinski@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
This code is correct, but it should be indented one more tab.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Link: https://lore.kernel.org/r/20201103101853.GD1127762@mwanda
Signed-off-by: Mark Brown <broonie@kernel.org>
davinci_mcasp_get_dt_params alway return zero, and its return value
could be ignored by the caller. So make it 'void' type to avoid the
check its return value.
Fixes: 764958f2b5 ("ASoC: ti: davinci-mcasp: Support for auxclk-fs-ratio")
Signed-off-by: Zhang Qilong <zhangqilong3@huawei.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Link: https://lore.kernel.org/r/20201102103428.32678-1-zhangqilong3@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Besides the fact that muxclk is optional, muxclk can be set using
assigned-clocks, removing the need to set it in driver. The warning is
thus unneeded, so we can transform it in a debug print, eventually to just
reflect that muxclk was not set by the driver.
Signed-off-by: Codrin Ciubotariu <codrin.ciubotariu@microchip.com>
Link: https://lore.kernel.org/r/20201103100554.1307190-1-codrin.ciubotariu@microchip.com
Signed-off-by: Mark Brown <broonie@kernel.org>
A semicolon is not needed after a switch statement.
Signed-off-by: Tom Rix <trix@redhat.com>
Acked-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://lore.kernel.org/r/20201101171943.2305030-1-trix@redhat.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Because clk_disable_unprepare() already checked NULL clock parameter,
so the additional check is unnecessary, just remove it.
Signed-off-by: Xu Wang <vulab@iscas.ac.cn>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Link: https://lore.kernel.org/r/20201029082513.28233-1-vulab@iscas.ac.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
First patch adds tdm 4-slot 100fs DAI setting to avoid jitter of using
64fs on CML boards. Second patch is a DMI quirk for HP Dooly.
Changes since v1:
-Add comment on Dooly's DMI quirk
Brent Lu (2):
ASoC: intel: sof_rt5682: Add support for cml_rt1015_rt5682
ASoC: intel: sof_rt5682: Add quirk for Dooly
sound/soc/intel/boards/sof_rt5682.c | 65 +++++++++++++++++--
.../intel/common/soc-acpi-intel-cml-match.c | 13 ++++
2 files changed, 73 insertions(+), 5 deletions(-)
--
2.17.1
This DMI product family string of this board is "Google_Hatch" so the
DMI quirk will take place. However, this board is using rt1015 speaker
amp instead of max98357a specified in the quirk. Therefore, we need an
new DMI quirk for this board.
Signed-off-by: Brent Lu <brent.lu@intel.com>
Acked-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20201030170559.20370-3-brent.lu@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
This patch adds the driver data and updates quirk info for cml with
rt1015 speaker amp and rt5682 headset codec. Due to different mclk
frequency on JSL and CML, we need to use 4 slot TDM 100fs to avoid
the SSP m/n counter.
Signed-off-by: Brent Lu <brent.lu@intel.com>
Acked-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20201030170559.20370-2-brent.lu@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Checkpatch script produces warning:
WARNING: function definition argument 'const struct sun4i_i2s *'
should also have an identifier name.
Let's fix this by adding identifier name to get_bclk_parent_rate()
and set_fmt() callback definition.
Acked-by: Maxime Ripard <mripard@kernel.org>
Signed-off-by: Clément Péron <peron.clem@gmail.com>
Link: https://lore.kernel.org/r/20201030144648.397824-10-peron.clem@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Because SUN4I_I2S_FIFO_CTRL_REG is volatile, writes done while the
regmap is cache-only are ignored. To work around this, move the
configuration to a callback that runs while the ASoC core has a
runtime PM reference to the device.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Chen-Yu Tsai <wens@csie.org>
Acked-by: Maxime Ripard <mripard@kernel.org>
Signed-off-by: Clément Péron <peron.clem@gmail.com>
Link: https://lore.kernel.org/r/20201030144648.397824-9-peron.clem@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The FIFO TX reg is volatile and sun8i i2s register
mapping is different from sun4i.
Even if in this case it's doesn't create an issue,
Avoid setting some regs that are undefined in sun8i.
Acked-by: Maxime Ripard <mripard@kernel.org>
Reviewed-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Clément Péron <peron.clem@gmail.com>
Link: https://lore.kernel.org/r/20201030144648.397824-8-peron.clem@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Extend the functionality of the driver to include support of 20 and
24 bits per sample.
Signed-off-by: Marcus Cooper <codekipper@gmail.com>
Acked-by: Maxime Ripard <mripard@kernel.org>
Reviewed-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Clément Péron <peron.clem@gmail.com>
Link: https://lore.kernel.org/r/20201030144648.397824-7-peron.clem@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
On the newer SoCs such as the H3 and A64 this is set by default
to transfer a 0 after each sample in each slot. However the A10
and A20 SoCs that this driver was developed on had a default
setting where it padded the audio gain with zeros.
This isn't a problem while we have only support for 16bit audio
but with larger sample resolution rates in the pipeline then SEXT
bits should be cleared so that they also pad at the LSB. Without
this the audio gets distorted.
Set sign extend sample for all the sunxi generations even if they
are not affected. This will keep consistency and avoid relying on
default.
Signed-off-by: Marcus Cooper <codekipper@gmail.com>
Reviewed-by: Chen-Yu Tsai <wens@csie.org>
Acked-by: Maxime Ripard <mripard@kernel.org>
Signed-off-by: Clément Péron <peron.clem@gmail.com>
Link: https://lore.kernel.org/r/20201030144648.397824-6-peron.clem@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
We are actually using a complex formula to just return a bunch of
simple values. Also this formula is wrong for sun4i when calling
get_wss() the function return 4 instead of 3.
Replace this with a simpler switch case.
Also drop the i2s params which is unused and return a simple int as
returning an error code could be out of range for an s8 and there is
no optim to return a s8 here.
Fixes: 619c15f7fa ("ASoC: sun4i-i2s: Change SR and WSS computation")
Reviewed-by: Chen-Yu Tsai <wens@csie.org>
Acked-by: Maxime Ripard <mripard@kernel.org>
Signed-off-by: Clément Péron <peron.clem@gmail.com>
Link: https://lore.kernel.org/r/20201030144648.397824-5-peron.clem@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Mark Brown
Shengjiu Wang
Cheng-Yi Chiang
Jack Yu
Peter Ujfalusi
Xu Wang
Srinivas Kandagatla
Cezary Rojewski
Zou Wei
Sudip Mukherjee
Shuming Fan
Jiaxin Yu
Amadeusz Sławiński
Tom Rix
Dan Carpenter
Zhang Qilong
Codrin Ciubotariu
Brent Lu
Clément Péron
Samuel Holland
Marcus Cooper