Some tools use the snd_pcm_info_get_name() to try to identify PCMs or for
other purposes.
Currently it is left empty with the dmaengine-pcm, in this case copy the
pcm->id string as pcm->name.
For example IGT is using this to find the HDMI PCM for testing audio on it.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reported-by: Arthur She <arthur.she@linaro.org>
Link: https://lore.kernel.org/r/20190906055524.7393-1-peter.ujfalusi@ti.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Currently the driver uses snd_soc_rtdcom_lookup() in it's mcbsp_start
function to try to stop/restart the DMA as the initial XSYNCERR workaround
need to be done before the DMA is armed.
There are couple of things wrong with this:
- the driver crashes with NULL pointer dereference as the
component->driver->ops is actually NULL
- the driver should not use snd_soc_rtdcom_lookup() in the first place
- Fiddling with DMA is never a good thing
Move the workaround handling to .prepare which is called before the DMA is
armed, so it complies with the requirements.
Reported-by (usage of snd_soc_rtdcom_lookup): Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Link: https://lore.kernel.org/r/20190830103841.25128-3-peter.ujfalusi@ti.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Append fallthrough statement to fix warning reported during compilation.
Fixes: b80d19c166 ("ASoC: Intel: Restore Baytrail ADSP streams only when ADSP was in reset")
Signed-off-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://lore.kernel.org/r/20190828095102.15737-1-cezary.rojewski@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The test case is
arecord -Dhw:0 -d 10 -f S16_LE -r 48000 -c 2 temp.wav &
aplay -Dhw:0 -d 30 -f S16_LE -r 48000 -c 2 test.wav
There will be error after end of arecord:
aplay: pcm_write:2051: write error: Input/output error
Capture and Playback work in parallel in master mode, one
substream stops, the other substream is impacted, the
reason is that clock is disabled wrongly.
The clock's reference count is not increased when second
substream starts, the hw_param() function returns in the
beginning because first substream is enabled, then in end
of first substream, the hw_free() disables the clock.
This patch is to move the clock enablement to the place
before checking of the device enablement in hw_param().
Signed-off-by: Shengjiu Wang <shengjiu.wang@nxp.com>
Link: https://lore.kernel.org/r/1567012817-12625-1-git-send-email-shengjiu.wang@nxp.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Mt8183 tdm out support S16_LE/S24_LE/S32_LE formats. When output S32_LE,
we need set hd_align so that memif can output MSB 24bits. When output
S24_LE, we need reset hd_align so that memif can output LSB 24bits.
Signed-off-by: Jiaxin Yu <jiaxin.yu@mediatek.com>
Link: https://lore.kernel.org/r/1566622726-27113-1-git-send-email-jiaxin.yu@mediatek.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The calibration process at booting will reset registers and bypass cache
to make sure the calibration is done.
We add mutex protection to avoid unexpected settings while
the registration process and calibration are interleaved.
Signed-off-by: Shuming Fan <shumingf@realtek.com>
Link: https://lore.kernel.org/r/20190826090052.1875-1-shumingf@realtek.com
Signed-off-by: Mark Brown <broonie@kernel.org>
This patch fixes inverted Left-Right channel of headphone mixer
volume by wrong shift_left, shift_right values.
Signed-off-by: Katsuhiro Suzuki <katsuhiro@katsuster.net>
Reviewed-by: Daniel Drake <drake@endlessm.com>
Link: https://lore.kernel.org/r/20190826153900.25969-2-katsuhiro@katsuster.net
Signed-off-by: Mark Brown <broonie@kernel.org>
We use snd_soc_dapm_new_control_unlocked for topology and have local
declaration, instead declare it properly in header like already declared
snd_soc_dapm_new_control.
Signed-off-by: Amadeusz Sławiński <amadeuszx.slawinski@intel.com>
Link: https://lore.kernel.org/r/20190827141712.21015-4-amadeuszx.slawinski@linux.intel.com
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
If ipc->ops.reply_msg_match is NULL, we may end up using uninitialized
mask value.
reported by smatch:
sound/soc/intel/common/sst-ipc.c:266 sst_ipc_reply_find_msg() error: uninitialized symbol 'mask'.
Signed-off-by: Amadeusz Sławiński <amadeuszx.slawinski@intel.com>
Link: https://lore.kernel.org/r/20190827141712.21015-3-amadeuszx.slawinski@linux.intel.com
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
Since hw_free() can be called multiple times and not just after a stop
trigger command, we should check whether the RX or TX ready interrupt was
truly enabled previously. For this, we assure that the condition of the
wait event is always true, except when RX/TX interrupts are enabled.
Fixes: 7e0cdf545a55 ("ASoC: mchp-i2s-mcc: add driver for I2SC Multi-Channel Controller")
Signed-off-by: Codrin Ciubotariu <codrin.ciubotariu@microchip.com>
Link: https://lore.kernel.org/r/20190820162411.24836-3-codrin.ciubotariu@microchip.com
Signed-off-by: Mark Brown <broonie@kernel.org>
If hw_free() gets called after hw_params(), GCLK remains prepared,
preventing further use of it. This patch fixes this by unpreparing the
clock in hw_free() or if hw_params() gets an error.
Fixes: 7e0cdf545a55 ("ASoC: mchp-i2s-mcc: add driver for I2SC Multi-Channel Controller")
Signed-off-by: Codrin Ciubotariu <codrin.ciubotariu@microchip.com>
Link: https://lore.kernel.org/r/20190820162411.24836-2-codrin.ciubotariu@microchip.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Every single baytrail chromebook sets PMC to 0, as can be seeing
below by searching through coreboot source code:
$ grep -rl "PMC_PLT_CLK\[0\]" .
./rambi/variants/glimmer/devicetree.cb
./rambi/variants/clapper/devicetree.cb
./rambi/variants/swanky/devicetree.cb
./rambi/variants/enguarde/devicetree.cb
./rambi/variants/winky/devicetree.cb
./rambi/variants/kip/devicetree.cb
./rambi/variants/squawks/devicetree.cb
./rambi/variants/orco/devicetree.cb
./rambi/variants/ninja/devicetree.cb
./rambi/variants/heli/devicetree.cb
./rambi/variants/sumo/devicetree.cb
./rambi/variants/banjo/devicetree.cb
./rambi/variants/candy/devicetree.cb
./rambi/variants/gnawty/devicetree.cb
./rambi/variants/rambi/devicetree.cb
./rambi/variants/quawks/devicetree.cb
Plus, Cyan (only non-baytrail chromebook with max98090) also needs
this patch for audio to work.
Thus, this commit adds all the missing devices to bsw_max98090 quirk
table, implemented by commit a182ecd380 ("ASoC: intel:
cht_bsw_max98090_ti: Add quirk for boards using pmc_plt_clk_0").
Signed-off-by: Daniel Stuart <daniel.stuart14@gmail.com>
Link: https://lore.kernel.org/r/20190815171300.30126-1-daniel.stuart14@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Use the devm variant to get the formatter reset so it is properly freed
on device removal
Fixes: 751bd5db52 ("ASoC: meson: axg-tdm-formatter: add reset")
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20190820123413.22249-1-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
When transitioning to supend state, uniphier_aio_dai_suspend() is called
and asserts reset lines and disables clocks.
However, if there are two or more DAIs, uniphier_aio_dai_suspend() are
called multiple times, and double reset assersion will cause.
This patch defines the counter that has the number of DAIs at first, and
whenever uniphier_aio_dai_suspend() are called, it decrements the
counter. And only if the counter is zero, it asserts reset lines and
disables clocks.
In the same way, uniphier_aio_dai_resume() are called, it increments the
counter after deasserting reset lines and enabling clocks.
Fixes: 139a342002 ("ASoC: uniphier: add support for UniPhier AIO CPU DAI driver")
Signed-off-by: Kunihiko Hayashi <hayashi.kunihiko@socionext.com>
Link: https://lore.kernel.org/r/1566281764-14059-1-git-send-email-hayashi.kunihiko@socionext.com
Signed-off-by: Mark Brown <broonie@kernel.org>
For i.MX6 SoloX, there is a mode of the SoC to shutdown all power
source of modules during system suspend and resume procedure.
Thus, AUDMUX needs to save all the values of registers before the
system suspend and restore them after the system resume.
Signed-off-by: Shengjiu Wang <shengjiu.wang@nxp.com>
Link: https://lore.kernel.org/r/1565931794-7218-1-git-send-email-shengjiu.wang@nxp.com
Signed-off-by: Mark Brown <broonie@kernel.org>
sound/soc/codecs/wm8737.c:112:29: warning:
high_3d defined but not used [-Wunused-const-variable=]
'high_3d' should be used for 3D High Cut-off.
Reported-by: Hulk Robot <hulkci@huawei.com>
Fixes: 2a9ae13a26 ("ASoC: Add initial WM8737 driver")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Acked-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20190815091920.64480-1-yuehaibing@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
sound/soc/codecs/cs4349.c:358:32: warning:
cs4349_runtime_pm defined but not used [-Wunused-const-variable=]
cs4349_runtime_pm ops already defined, it seems
we should enable it.
Reported-by: Hulk Robot <hulkci@huawei.com>
Fixes: e40da86 ("ASoC: cs4349: Add support for Cirrus Logic CS4349")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Link: https://lore.kernel.org/r/20190815090157.70036-1-yuehaibing@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
From stress testing of arecord, we found that period size
greater than ~900 will bring pl330 to DYING state and
can not recover within 100 iterations.
The result is that arecord will stuck and get I/O error,
and issue can not be recovered until reboot.
This issue does not happen when period size is small.
Set constraint of period size to 240 to prevent such issue.
With the constraint, there will be no issue after 2000 iterations.
We can revert this patch once the root cause is found
in rockchip's pl330 implementation.
Signed-off-by: Cheng-Yi Chiang <cychiang@chromium.org>
Link: https://lore.kernel.org/r/20190813074430.191791-1-cychiang@chromium.org
Signed-off-by: Mark Brown <broonie@kernel.org>
We need add DAPM MIC endpoint widget "SoC DMIC" and route, to enable
DMIC PCM DAPM support.
Signed-off-by: Keyon Jie <yang.jie@linux.intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20190809232236.21182-3-pierre-louis.bossart@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
hdac_hdmi_present_sense() calls the audio component to get ELD update,
then it reports the jack status change and updates DAPM graph
accordingly. This works when it's called from the normal code paths.
However, it may lead to a dead lock when it's called from the audio
component notifier. Namely, the DAPM update involves with the runtime
PM, and it eventually calls again the audio component get_power()
ops. Since i915 driver already takes a mutex around the audio
component ops calls, we'll eventually get the mutex doubly.
As a workaround, in this patch, only the jack state is updated in the
code path from hdac_hdmi_eld_notify_cb(), and the DAPM update is
deferred to a work so that it's processed in another context.
Reported-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Link: https://lore.kernel.org/r/20190809151531.24359-1-tiwai@suse.de
Signed-off-by: Mark Brown <broonie@kernel.org>
There is a race between hda codec device removing and the
jack-detecting work, which will lead to a page fault issue as the
latter work is accessing codec device which could be already removed.
Here add the cancellation of jack-detecting work before codecs are actually
removed to avoid the race and fix the issue.
Bug: https://github.com/thesofproject/linux/issues/1067
Signed-off-by: Keyon Jie <yang.jie@linux.intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20190807145030.26117-1-pierre-louis.bossart@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
On some platforms, sound card registration fails when a HDMI
monitor is not connected. This is caused by a recent commit
that switched the order in which the HDA controller and the
i915 are initialized. Initializing the i915 before initializing
the HDA controller fixes the problem.
Fixes: be1b577d01 ("ASoC: SOF: Intel: hda: fix the hda init chip"
Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Link: https://lore.kernel.org/r/20190806221958.19180-1-ranjani.sridharan@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
ADG is using clk_get_rate() under atomic context, thus, we might
have scheduling issue.
To avoid this issue, we need to get/keep clk rate under
non atomic context.
We need to handle ADG as special device at Renesas Sound driver.
From SW point of view, we want to impletent it as
rsnd_mod_ops :: prepare, but it makes code just complicate.
To avoid complicated code/patch, this patch adds new clk_rate[] array,
and keep clk IN rate when rsnd_adg_clk_enable() was called.
Reported-by: Leon Kong <Leon.KONG@cn.bosch.com>
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Tested-by: Leon Kong <Leon.KONG@cn.bosch.com>
Link: https://lore.kernel.org/r/87v9vb0xkp.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Mark Brown <broonie@kernel.org>
We shouldn't assume CPU physical address we get from page_to_phys()
is same as DMA address we get from dma_alloc_coherent(). On x86_64,
we won't run into any problem with the assumption when dma_ops is
nommu_dma_ops. However, DMA address is IOVA when IOMMU is enabled.
And it's most likely different from CPU physical address when AMD
IOMMU is not in passthrough mode.
This patch fixes page faults when IOMMU is enabled.
Signed-off-by: Vijendar Mukunda <vijendar.mukunda@amd.com>
Link: https://lore.kernel.org/r/1564753899-17124-2-git-send-email-Vijendar.Mukunda@amd.com
Signed-off-by: Mark Brown <broonie@kernel.org>
AMD platform device acp3x_rv_i2s created by parent PCI device
driver. Pass struct device of the parent to
snd_pcm_lib_preallocate_pages() so dma_alloc_coherent() can use
correct dma_ops. Otherwise, it will use default dma_ops which
is nommu_dma_ops on x86_64 even when IOMMU is enabled and
set to non passthrough mode.
Signed-off-by: Vijendar Mukunda <vijendar.mukunda@amd.com>
Link: https://lore.kernel.org/r/1564753899-17124-1-git-send-email-Vijendar.Mukunda@amd.com
Signed-off-by: Mark Brown <broonie@kernel.org>
88200 and 96000 sampling rate was not enabled on driver, so can't be played.
The error information:
max98373 3-0031：rate 96000 not supported
max98373 3-0031：ASoC: can't set max98373-aif1 hw params: -22
Signed-off-by: fengchunguo <chunguo.feng@amlogic.com>
Link: https://lore.kernel.org/r/20190731074156.5620-1-chunguo.feng@amlogic.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The A64 audio codec uses the original I2S block but the SR and
WSS computation currently assigned is for the newer block.
Fixes: 619c15f7fa (ASoC: sun4i-i2s: Change SR and WSS computation)
Signed-off-by: Marcus Cooper <codekipper@gmail.com>
Link: https://lore.kernel.org/r/20190729152130.27955-1-codekipper@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The slot_width is a property for the bus while the constraint for
SNDRV_PCM_HW_PARAM_SAMPLE_BITS is for the in memory format.
Applying slot_width constraint to sample_bits works most of the time, but
it will blacklist valid formats in some cases.
With slot_width 24 we can support S24_3LE and S24_LE formats as they both
look the same on the bus, but a a 24 constraint on sample_bits would not
allow S24_LE as it is stored in 32bits in memory.
Implement a simple hw_rule function to allow all formats which require less
or equal number of bits on the bus as slot_width (if configured).
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Link: https://lore.kernel.org/r/20190726064244.3762-2-peter.ujfalusi@ti.com
Signed-off-by: Mark Brown <broonie@kernel.org>
This reverts commit db51707b9c.
Revert "ASoC: rockchip: i2s: Support mono capture"
Previous discussion in
https://patchwork.kernel.org/patch/10147153/
explains the issue of the patch.
While device is configured as 1-ch, hardware is still
generating a 2-ch stream.
When user space reads the data and assumes it is a 1-ch stream,
the rate will be slower by 2x.
Revert the change so 1-ch is not supported.
User space can selectively take one channel data out of two channel
if 1-ch is preferred.
Currently, both channels record identical data.
Signed-off-by: Cheng-Yi Chiang <cychiang@chromium.org>
Link: https://lore.kernel.org/r/20190726044202.26866-1-cychiang@chromium.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Fix some typo to have the filaname given in a comment match the real name
of the file.
Some 'acpi' have erroneously been written 'apci'
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/20190725053523.16542-1-christophe.jaillet@wanadoo.fr
Signed-off-by: Mark Brown <broonie@kernel.org>
When running McASP as master capture alone will not record any audio unless
a parallel playback stream is running. As soon as the playback stops the
captured data is going to be silent again.
In McASP master mode we need to set the PDIR for the clock pins and fix
the mcasp_set_axr_pdir() to skip the bits in the PDIR registers above
AMUTE.
This went unnoticed as most of the boards uses McASP as slave and neither
of these issues are visible (audible) in those setups.
Fixes: ca3d943334 ("ASoC: davinci-mcasp: Update PDIR (pin direction) register handling")
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Link: https://lore.kernel.org/r/20190725083423.7321-1-peter.ujfalusi@ti.com
Signed-off-by: Mark Brown <broonie@kernel.org>
If the DAI format setup fails, there is no valid communication format
between CPU and CODEC, so fail card instantiation, rather than continue
with a card that will most likely not function properly.
Signed-off-by: Ricard Wanderlof <ricardw@axis.com>
Link: https://lore.kernel.org/r/alpine.DEB.2.20.1907241132350.6338@lnxricardw1.se.axis.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Downgrade "nothing to do in IRQ thread" message from error to a debug
message in the IPC interrupt handler thread.
The spurious wake-up can happen if a HDA stream interrupt is
raised while the IPC interrupt thread is running. IPC functionality
is not impacted by this condition, so debug is a more appropriate
trace level.
Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20190722141402.7194-21-pierre-louis.bossart@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
apq8016_sbc_parse_of() sets up multiple DAI links, depending on the
number of nodes in the device tree. However, at the moment
CPU and platform components are only allocated for the first link.
This causes an oops when more than one link is defined:
Internal error: Oops: 96000044 [#1] SMP
CPU: 0 PID: 1015 Comm: kworker/0:2 Not tainted 5.3.0-rc1 #4
Call trace:
apq8016_sbc_platform_probe+0x1a8/0x3f0
platform_drv_probe+0x50/0xa0
...
Move the allocation inside the loop to ensure that each link is
properly initialized.
Fixes: 98b232ca9e ("ASoC: qcom: apq8016_sbc: use modern dai_link style")
Signed-off-by: Stephan Gerhold <stephan@gerhold.net>
Acked-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/20190722130352.95874-1-stephan@gerhold.net
Signed-off-by: Mark Brown <broonie@kernel.org>
In snd_soc_dapm_new_control_unlocked(), a kernel buffer is allocated in
dapm_cnew_widget() to hold the new dapm widget. Then, different actions are
taken according to the id of the widget, i.e., 'w->id'. If any failure
occurs during this process, snd_soc_dapm_new_control_unlocked() should be
terminated by going to the 'request_failed' label. However, the allocated
kernel buffer is not freed on this code path, leading to a memory leak bug.
To fix the above issue, free the buffer before returning from
snd_soc_dapm_new_control_unlocked() through the 'request_failed' label.
Signed-off-by: Wenwen Wang <wenwen@cs.uga.edu>
Link: https://lore.kernel.org/r/1563803864-2809-1-git-send-email-wang6495@umn.edu
Signed-off-by: Mark Brown <broonie@kernel.org>
The TS3A227E says that the headset keypress detection needs the MICBIAS
power in order to report the key events to ensure proper operation
The headset keypress detection needs the MICBIAS power in order to report
the key events all the time as long as MIC is present. So MICBIAS pin
is forced on when a MICROPHONE is detected.
On Veyron Minnie I observed that if the MICBIAS power is not present and
the key press detection is activated (just because it is enabled when you
insert a headset), it randomly reports a keypress on insert.
E.g. (KEY_PLAYPAUSE)
Event: (SW_HEADPHONE_INSERT), value 1
Event: (SW_MICROPHONE_INSERT), value 1
Event: -------------- SYN_REPORT ------------
Event: (KEY_PLAYPAUSE), value 1
Userspace thinks that KEY_PLAYPAUSE is pressed and produces the annoying
effect that the media player starts a play/pause loop.
Note that, although most of the time the key reported is the one
associated with BTN_0, not always this is true. On my tests I also saw
different keys reported
Signed-off-by: Enric Balletbo i Serra <enric.balletbo@collabora.com>
Link: https://lore.kernel.org/r/20190719173929.24065-1-enric.balletbo@collabora.com
Signed-off-by: Mark Brown <broonie@kernel.org>
When sample rate of TX is different with sample rate of RX in
async mode, the MFreq selection will be wrong.
For example, sysclk = 24.576MHz, TX rate = 96000Hz, RX rate = 48000Hz.
Then ratio of TX = 256, ratio of RX = 512, For MFreq is shared by TX
and RX instance, the correct value of MFreq is 2 for both TX and RX.
But original method will cause MFreq = 0 for TX, MFreq = 2 for RX.
If TX is started after RX, RX will be impacted, RX work abnormal with
MFreq = 0.
This patch is to select proper MFreq value according to TX rate and
RX rate.
Fixes: 0c516b4ff8 ("ASoC: cs42xx8: Add codec driver support for CS42448/CS42888")
Signed-off-by: Shengjiu Wang <shengjiu.wang@nxp.com>
Link: https://lore.kernel.org/r/20190716094547.46787-1-shengjiu.wang@nxp.com
Signed-off-by: Mark Brown <broonie@kernel.org>
DPCM uses snd_soc_dapm_dai_get_connected_widgets to build a
list of the widgets connected to a specific front end DAI so it
can search through this list for available back end DAIs. The
custom_stop_condition was added to is_connected_ep to facilitate this
list not containing more widgets than is necessary. Doing so both
speeds up the DPCM handling as less widgets need to be searched and
avoids issues with CODEC to CODEC links as these would be confused
with back end DAIs if they appeared in the list of available widgets.
custom_stop_condition was implemented by aborting the graph walk
when the condition is triggered, however there is an issue with this
approach. Whilst walking the graph is_connected_ep should update the
endpoints cache on each widget, if the walk is aborted the number
of attached end points is unknown for that sub-graph. When the stop
condition triggered, the original patch ignored the triggering widget
and returned zero connected end points; a later patch updated this
to set the triggering widget's cache to 1 and return that. Both of
these approaches result in inaccurate values being stored in various
end point caches as the values propagate back through the graph,
which can result in later issues with widgets powering/not powering
unexpectedly.
As the original goal was to reduce the size of the widget list passed
to the DPCM code, the simplest solution is to limit the functionality
of the custom_stop_condition to the widget list. This means the rest
of the graph will still be processed resulting in correct end point
caches, but only widgets up to the stop condition will be added to the
returned widget list.
Fixes: 6742064aef ("ASoC: dapm: support user-defined stop condition in dai_get_connected_widgets")
Fixes: 5fdd022c20 ("ASoC: dpcm: play nice with CODEC<->CODEC links")
Fixes: 09464974ea ("ASoC: dapm: Fix to return correct path list in is_connected_ep.")
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20190718084333.15598-1-ckeepax@opensource.cirrus.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The cpu_dai variable is still being used after the of_node_put() call,
which may result in double-free:
of_node_put(cpu_dai); ---> released here
ret = devm_snd_soc_register_card(dev, card);
if (ret < 0) {
...
goto err_put_clk_i2s; --> jump to err_put_clk_i2s
...
err_put_clk_i2s:
clk_put(priv->clk_i2s_bus);
err_put_sclk:
clk_put(priv->sclk_i2s);
err_put_cpu_dai:
of_node_put(cpu_dai); --> double-free here
Fixes: d832d2b246 ("ASoC: samsung: odroid: Fix of_node refcount unbalance")
Signed-off-by: Wen Yang <wen.yang99@zte.com.cn>
Cc: Krzysztof Kozlowski <krzk@kernel.org>
Cc: Sangbeom Kim <sbkim73@samsung.com>
Cc: Sylwester Nawrocki <s.nawrocki@samsung.com>
Cc: Liam Girdwood <lgirdwood@gmail.com>
Cc: Mark Brown <broonie@kernel.org>
Cc: Jaroslav Kysela <perex@perex.cz>
Cc: Takashi Iwai <tiwai@suse.com>
Cc: alsa-devel@alsa-project.org
Cc: linux-kernel@vger.kernel.org
Link: https://lore.kernel.org/r/1562989575-33785-3-git-send-email-wen.yang99@zte.com.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
