| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
In snd_hdac_bus_init_chip(), we enable interrupt before
snd_hdac_bus_init_cmd_io() initializing dma buffers. If irq has
been acquired and irq handler uses the dma buffer, kernel may crash
when interrupt comes in.
Fix the problem by postponing enabling irq after dma buffer
initialization. And warn once on null dma buffer pointer during the
initialization.
Reviewed-by: Takashi Iwai <[email protected]>
Signed-off-by: Yu Zhao <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
This reverts commit 12eeeb4f4733bbc4481d01df35933fc15beb8b19.
The patch doesn't fix accessing memory with null pointer in
skl_interrupt().
There are two problems: 1) skl_init_chip() is called twice, before
and after dma buffer is allocate. The first call sets bus->chip_init
which prevents the second from initializing bus->corb.buf and
rirb.buf from bus->rb.area. 2) snd_hdac_bus_init_chip() enables
interrupt before snd_hdac_bus_init_cmd_io() initializing dma buffers.
There is a small window which skl_interrupt() can be called if irq
has been acquired. If so, it crashes when using null dma buffer
pointers.
Will fix the problems in the following patches. Also attaching the
crash for future reference.
[ 16.949148] general protection fault: 0000 [#1] PREEMPT SMP KASAN PTI
<snipped>
[ 16.950903] Call Trace:
[ 16.950906] <IRQ>
[ 16.950918] skl_interrupt+0x19e/0x2d6 [snd_soc_skl]
[ 16.950926] ? dma_supported+0xb5/0xb5 [snd_soc_skl]
[ 16.950933] __handle_irq_event_percpu+0x27a/0x6c8
[ 16.950937] ? __irq_wake_thread+0x1d1/0x1d1
[ 16.950942] ? __do_softirq+0x57a/0x69e
[ 16.950944] handle_irq_event_percpu+0x95/0x1ba
[ 16.950948] ? _raw_spin_unlock+0x65/0xdc
[ 16.950951] ? __handle_irq_event_percpu+0x6c8/0x6c8
[ 16.950953] ? _raw_spin_unlock+0x65/0xdc
[ 16.950957] ? time_cpufreq_notifier+0x483/0x483
[ 16.950959] handle_irq_event+0x89/0x123
[ 16.950962] handle_fasteoi_irq+0x16f/0x425
[ 16.950965] handle_irq+0x1fe/0x28e
[ 16.950969] do_IRQ+0x6e/0x12e
[ 16.950972] common_interrupt+0x7a/0x7a
[ 16.950974] </IRQ>
<snipped>
[ 16.951031] RIP: snd_hdac_bus_update_rirb+0x19b/0x4cf [snd_hda_core] RSP: ffff88015c807c08
[ 16.951036] ---[ end trace 58bf9ece1775bc92 ]---
Fixes: 2eeeb4f4733b ("ASoC: Intel: Skylake: Acquire irq after RIRB allocation")
Signed-off-by: Yu Zhao <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
It is strange if it has "dai" but doesn't have "dai->driver".
And more over "dai->driver->xxx" is used everywhere without
"dai->driver" pointer NULL checking.
It got Oops already if "dai->driver" was NULL.
Let's remove un-needed "dai->driver" NULL check.
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Current behaviour of ASoC core w.r.t to component removal is that it
unregisters dependent sound card totally. There is no support to
rebind the card if the component comes back.
Typical use case is DSP restart or kernel modules itself.
With this patch, core now maintains list of cards that are unbind due to
any of its depended components are removed and card not unregistered yet.
This list is cleared when the card is rebind successfully or when the
card is unregistered from machine driver.
This list of unbind cards are tried to bind once again after every new
component is successfully added, giving a fair chance for card bind
to be successful.
Signed-off-by: Srinivas Kandagatla <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
platform_driver_register will set the .owner field. So it is safe
to remove the redundant assignment.
The issue is detected with the help of Coccinelle.
Signed-off-by: zhong jiang <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
This patch adds SNDRV_PCM_INFO_INTERLEAVED into PCM hardware info.
Signed-off-by: Katsuhiro Suzuki <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
ALSA SoC is counting card->dai_link_list user,
but no-one is using it.
Let's remove it.
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Nested code is not readable.
This patch avoid it on soc_remove_dai().
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
rt5668_i2c_remove() is empty, and no longer needed.
Let's remove it.
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
In functions snd_soc_get_volsw_sx() or snd_soc_put_volsw_sx(),
if the result of (min + max) is negative, then fls() returns
signed integer with value as 32. This leads to signed integer
overflow as complete operation is considered as signed integer.
UBSAN: Undefined behaviour in sound/soc/soc-ops.c:382:50
signed integer overflow:
-2147483648 - 1 cannot be represented in type 'int'
Call trace:
[<ffffff852f746fe4>] __dump_stack lib/dump_stack.c:15 [inline]
[<ffffff852f746fe4>] dump_stack+0xec/0x158 lib/dump_stack.c:51
[<ffffff852f7b5f3c>] ubsan_epilogue+0x18/0x50 lib/ubsan.c:164
[<ffffff852f7b6840>] handle_overflow+0xf8/0x130 lib/ubsan.c:195
[<ffffff852f7b68f0>] __ubsan_handle_sub_overflow+0x34/0x44 lib/ubsan.c:211
[<ffffff85307971a0>] snd_soc_get_volsw_sx+0x1a8/0x1f8 sound/soc/soc-ops.c:382
Typecast the operation to unsigned int to fix the issue.
Signed-off-by: Rohit kumar <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
|
|
Switch to the new ac97 bus support in sound/ac97 instead of the legacy
snd_ac97 one.
Signed-off-by: Robert Jarzmik <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
If capture and playback are started on different channel (I2S/BT)
there is a possibilty that channel information passed from machine driver
is overwritten before the configuration is done in dma driver.
Example:
113.597588: cz_max_startup: ---playback sets BT channel
113.597694: cz_dmic1_startup: ---capture sets I2S channel
113.597979: acp_dma_hw_params: ---configures capture for I2S channel
113.598114: acp_dma_hw_params: ---configures playback for I2S channel
This is fixed by having 2 separate instance for playback and capture.
Signed-off-by: Akshu Agrawal <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
The MMTLR bit is in the CS4265_SPDIF_CTL2 register at address 0x12 bit 0
and not at address 0x0 bit 1. Fix this.
Signed-off-by: Sébastien Szymanski <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
Cc: [email protected]
|
|
snd_soc_dapm_new_dai may return an error pointer and currently this
isn't checked for in dapm_connect_dai_link_widgets. Add code to check
the return value and not add routes in that case.
Fixes: 778ff5bb8689 ("ASoC: dapm: Move connection of CODEC to CODEC DAIs")
Reported-by: Dan Carpenter <[email protected]>
Signed-off-by: Charles Keepax <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
kmemdup has implemented the function that kzalloc() + memcpy() will
do. and we prefer to kmemdup rather than the open coded implementation.
Signed-off-by: zhong jiang <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
kmemdup has implemented the function that kzalloc() + memcpy() will
do. and we prefer to use the kmemdup rather than the open coded implementation.
Signed-off-by: zhong jiang <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Now we have devm_snd_soc_register_component().
Let's use it instead of snd_soc_register_component().
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Now we have devm_snd_soc_register_component().
Let's use it instead of snd_soc_register_component().
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Now we have devm_snd_soc_register_component().
Let's use it instead of snd_soc_register_component().
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Now we have devm_snd_soc_register_component().
Let's use it instead of snd_soc_register_component().
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Now we have devm_snd_soc_register_component().
Let's use it instead of snd_soc_register_component().
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Now we have devm_snd_soc_register_component().
Let's use it instead of snd_soc_register_component().
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Now we have devm_snd_soc_register_component().
Let's use it instead of snd_soc_register_component().
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Now we have devm_snd_soc_register_component().
Let's use it instead of snd_soc_register_component().
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Now we have devm_snd_soc_register_component().
Let's use it instead of snd_soc_register_component().
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
HW register descriptions says:
"DMA Channel Reset...Software must confirm that this bit is
cleared before reprogramming any of the channel configuration registers."
There could be cases where dma stop errored out leaving dma channel
in reset state. We need to ensure that before the start of another dma,
channel is out of the reset state.
Signed-off-by: Akshu Agrawal <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Smatch complains that if both "psubstream" and "csubstream" are NULL
then "ret" is uninitialized. That probably can't happen, but it's
cleaner to just return zero anyway so let's do that.
Signed-off-by: Dan Carpenter <[email protected]>
Acked-by: Srinivas Kandagatla <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Smatch complains that these variables could be uninitialized. The first
one in snd_soc_dai_link_event() is probably a false positive, because
probably we know the lists are not empty. I would normally ignore the
warning, but GCC complains here as well so I just silenced the warning.
The "ret" in snd_soc_dapm_new_dai() does need to be initialized or it
leads to a bogus dereference in the caller.
Fixes: 3bbf5d34fd4a ("ASoC: dapm: Move error handling to snd_soc_dapm_new_control_unlocked")
Signed-off-by: Dan Carpenter <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Commit ca917f9fe1a0fab added use of usleep_range() but not
the corresponding "include <linux/delay.h>". The result is
with Chrome OS won't build because warnings are forced
to be errors:
mnt/host/source/src/third_party/kernel/v4.4/sound/soc/codecs/max98373.c:734:2: error: implicit declaration of function 'usleep_range' [-Werror,-Wimplicit-function-declaration]
usleep_range(10000, 11000);
^
Including delay.h "fixes" this.
Signed-off-by: Grant Grundler <[email protected]>
Reviewed-by: Benson Leung <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Playback of 44.1Khz contents with HDMI plugged returns
"Invalid pipe config" because HDMI paths in the FW
topology are configured to operate at 48Khz.
This patch filters out sampling rates not supported
at hdac_hdmi_create_dais() to let user space SRC
to do the converting.
Signed-off-by: Yong Zhi <[email protected]>
Reviewed-by: Pierre-Louis Bossart <[email protected]>
Reviewed-by: Takashi Iwai <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Commit 4a75aae17b2a ("ASoC: dapm: Add support for multi-CODEC
CODEC to CODEC links") adds loops that iterate over multiple
CODECs in snd_soc_dai_link_event. This also introduced a compiler
warning for a potentially uninitialised variable in the case
no CODECs are present. This should never be the case as the
DAI link must by definition contain at least 1 CODEC however
probably best to avoid the compiler warning by initialising ret
to zero.
Signed-off-by: Charles Keepax <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
This patch adds the device tree binding of the r8a77990 SoC.
Signed-off-by: Hiroyuki Yokoyama <[email protected]>
Signed-off-by: Yoshihiro Kaneko <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
In the case where lo_vag <= SGTL5000_LINE_OUT_GND_BASE, lo_vag
is set to zero and later vol_quot is computed by dividing by
lo_vag causing a division by zero error. Fix this by avoiding
a zero division and set vol_quot to zero in this specific case
so that the lowest setting for i is correctly set.
Signed-off-by: Colin Ian King <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
rsnd driver sometimes want to know which address is used when debugging.
But it will indicate "(____ptrval____)" if it used "%p" on dev_dbg().
Let's use "%pa" or "%px" for it.
Signed-off-by: Kuninori Morimoto <[email protected]>
Tested-by: Hiroyuki Yokoyama <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Current rsnd driver will fallback to PIO mode if it can't get DMA
handler. But, DMA might return -EPROBE_DEFER when probe timing.
This driver always fallback to PIO mode especially from
commit ac6bbf0cdf4206c ("iommu: Remove IOMMU_OF_DECLARE") because
of this reason.
The DMA driver will be probed later, but sound driver might be
probed as PIO mode in such case. This patch fixup this issue.
Then, -EPROBE_DEFER is not error. Thus, let's don't indicate error
message in such case.
And it needs to call rsnd_adg_remove() individually if probe failed,
because it registers clk which should be unregister.
Maybe PIO fallback feature itself is not needed,
but let's keep it so far.
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
ADG has buffer over flow bug if DT has more than 3 clock-frequency.
This patch fixup this issue, and uses first 2 values.
clock-frequency = <x y>; /* this is OK */
clock-frequency = <x y z>; /* this is NG */
Signed-off-by: Kuninori Morimoto <[email protected]>
Tested-by: Hiroyuki Yokoyama <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
commit 8c9d75033340 ("ASoC: rsnd: ssiu: Support BUSIF
other than BUSIF0") added new SSIU registers.
But it is using white-space for it.
This patch fixup it to use tab.
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
The !port check is redundant as it being performed in the following
check. Remove it.
Signed-off-by: Colin Ian King <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Larger CODECs may contain many several hundred widgets and which set of
parameters is selected only needs to be recorded on a per DAI basis. As
such move the selected CODEC to CODEC link params to be stored in the
runtime rather than the DAPM widget, to save some memory.
Signed-off-by: Charles Keepax <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Currently multi-CODEC is not supported on CODEC to CODEC links.
There are common applications where this would be useful, such
as connecting two mono amplifiers to an audio CODEC. Adding
support simply requires an update of snd_soc_dai_link_event
to loop over the attached CODEC DAIs.
Signed-off-by: Charles Keepax <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Currently, snd_soc_dapm_connect_dai_link_widgets connects up the routes
representing normal DAIs, however CODEC to CODEC links are hooked up
through separate infrastructure in soc_link_dai_widgets. Improve the
consistency of the code by using snd_soc_dapm_connect_dai_link for both
types of DAIs.
Signed-off-by: Charles Keepax <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Move the function snd_soc_dapm_new_control to be next to
snd_soc_dapm_new_controls and add some kernel doc for it.
Signed-off-by: Charles Keepax <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Currently DAPM has a lot of similar code to handle errors from
snd_soc_dapm_new_control_unlocked, and much of this code does
not really accurately reflect what the function returns.
Firstly, most places will check for a return value of
-EPROBE_DEFER and silence any error messages in that case. The
one notable exception here being dapm_kcontrol_data_alloc
which does currently print any error messages in the case
of snd_soc_dapm_new_control_unlocked returning NULL or an
error. Additionally the error prints being silenced in these
case are redundant as snd_soc_dapm_new_control_unlocked can
only return -EPROBE_DEFER or NULL when failing.
Secondly, most places will treat a return value of NULL as
an -ENOMEM. This is not correct either since any error except
EPROBE_DEFER will cause a return value of NULL from
snd_soc_dapm_new_control_unlocked.
Centralise this handling and the error messages within
snd_soc_dapm_new_control_unlocked and update the callers
to simply check IS_ERR and return. Note that this update is
slightly simpler in the case of dapm_kcontrol_data_alloc where
that is fairly close to the handling that was already in place.
Signed-off-by: Charles Keepax <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Since I'm leaving from Socionext, I'll unable to access specification
documents of this hardware (these are not public). So change the
state to orphan until someone will maintain this driver.
Signed-off-by: Katsuhiro Suzuki <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
ALSA SoC snd_soc_pcm_runtime has snd_soc_dai array for codec_dai.
To be more readable code, this patch adds
new for_each_rtd_codec_dai() macro, and replace existing code to it.
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
ALSA SoC snd_soc_dai_link has snd_soc_dai_link_component array
for codecs.
To be more readable code, this patch adds
new for_each_link_codecs() macro, and replace existing code to it.
Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
Similar to short circuit detection, when the ADC/DAC is saturated and
overflows poor audio quality can result and should be reported to the
user. This device support Automatic Dynamic Range Compression (DRC)
to reduce this but it is not enabled currently in this driver.
Signed-off-by: Andrew F. Davis <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
When dapm_power_widgets() is called, the dapm_pre_sequence_async() and
dapm_post_sequence_async() functions are scheduled for all DAPM contexts
(apart from the card DAPM context) regardless of whether the DAPM
context is already in the desired state. The overhead of this is not
insignificant and the more DAPM contexts there are the more overhead
there is.
For example, on the Tegra124 Jetson TK1, when profiling the time taken
to execute the dapm_power_widgets() the following times were observed.
Times for function dapm_power_widgets() are (us):
Min 23, Ave 190, Max 434, Count 39
Here 'Count' is the number of times that dapm_power_widgets() has been
called. Please note that the above time were measured using ktime_get()
to log the time on entry and exit from dapm_power_widgets(). So it
should be noted that these times may not be purely the time take to
execute this function if it is preempted. However, after applying this
patch and measuring the time taken to execute dapm_power_widgets() again
a significant improvement is seen as shown below.
Times for function dapm_power_widgets() are (us):
Min 4, Ave 16, Max 82, Count 39
Therefore, optimise the dapm_power_widgets() function by only scheduling
the dapm_pre/post_sequence_async() work if the DAPM context is not in
the desired state.
Signed-off-by: Jon Hunter <[email protected]>
Reviewed-by: Charles Keepax <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|
|
DT binding documentation for this new ASoC driver.
Signed-off-by: Codrin Ciubotariu <[email protected]>
Signed-off-by: Mark Brown <[email protected]>
|