linux-kernel - Re: [PATCH 1/6] ASoC: codecs: wsa884x: remove mute_unmute_on

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <5f15fc5b-375a-1e75-b32d-0d543fa1014d@marek.ca>
Date: Mon, 24 Nov 2025 09:55:56 -0500
From: Jonathan Marek <jonathan@...ek.ca>
To: Srinivas Kandagatla <srinivas.kandagatla@....qualcomm.com>,
 linux-arm-msm@...r.kernel.org
Cc: Srinivas Kandagatla <srini@...nel.org>,
 Liam Girdwood <lgirdwood@...il.com>, Mark Brown <broonie@...nel.org>,
 Jaroslav Kysela <perex@...ex.cz>, Takashi Iwai <tiwai@...e.com>,
 Krzysztof Kozlowski <krzk@...nel.org>,
 Neil Armstrong <neil.armstrong@...aro.org>,
 "open list:QCOM AUDIO (ASoC) DRIVERS" <linux-sound@...r.kernel.org>,
 open list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 1/6] ASoC: codecs: wsa884x: remove mute_unmute_on_trigger

On 11/24/25 9:08 AM, Srinivas Kandagatla wrote:
> 
> 
> On 11/24/25 6:45 AM, Jonathan Marek wrote:
>> trigger is atomic (non-schedulable), and soundwire register writes are not
>> safe to run in an atomic context. (bus is locked with a mutex, and qcom
>> driver's callback can also sleep if the FIFO is full).
>>
> Thanks Jonathan for the patch,
> 
> We have nonatomic=1 flag set for all the Qualcomm sound cards, Did you
> hit any schedule while atomic bug?
> 

Right, I missed that. I'm using a different driver which does not set 
nonatomic. But this driver to not need nonatomic - 
mute_unmute_on_trigger is a hack, if there is a timing requirement - 
then it needs to be explicit, the different timing with this flag is not 
reliable).

> 
> 
> In-fact this change has helped suppress most of the click and pop noises
> on laptops, specially with wsa codecs as they accumulate static if the
> ports are kept open without sending any data.
> 

28b0b18d5346 is important to fix the click and pop noises. But the 
useful part is the rest of the commit, not the mute_unmute_on_trigger 
flag. As long as the mute_stream() happens while the soundwire stream is 
enabled (between sdw_enable_stream and sdw_disable_stream), there should 
be no pop click.

AFAIK the pop/click is because of PDM: zeros (soundwire stream off) 
represent the minimum (negative maximum) amplitude, and the soundwire 
stream needs to be enabled to output a zero amplitude (alternating 
ones/zeros). Turning on the amp while the soundwire stream is not 
enabled will cause jumps between the minimum and zero amplitude.

> --srini
> 
> 
>> The important part of fixing the click/pop issue was removing the PA_EN
>> writes from the dapm events, AFAICT this flag doesn't help anyway.
>>
>> Fixes: 28b0b18d5346 ("ASoC: codec: wsa884x: make use of new mute_unmute_on_trigger flag")
>> Signed-off-by: Jonathan Marek <jonathan@...ek.ca>
>> ---
>>   sound/soc/codecs/wsa884x.c | 1 -
>>   1 file changed, 1 deletion(-)
>>
>> diff --git a/sound/soc/codecs/wsa884x.c b/sound/soc/codecs/wsa884x.c
>> index 2484d4b8e2d94..0218dfc13bc77 100644
>> --- a/sound/soc/codecs/wsa884x.c
>> +++ b/sound/soc/codecs/wsa884x.c
>> @@ -1840,7 +1840,6 @@ static const struct snd_soc_dai_ops wsa884x_dai_ops = {
>>   	.hw_free = wsa884x_hw_free,
>>   	.mute_stream = wsa884x_mute_stream,
>>   	.set_stream = wsa884x_set_stream,
>> -	.mute_unmute_on_trigger = true,
>>   };
>>   
>>   static struct snd_soc_dai_driver wsa884x_dais[] = {
>