linux-kernel - Re: [PATCH v2 00/11] Fix support for System Suspend for CS35L41 HDA

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: <d2e86701-0377-c4ae-2718-efdee5d3677d@opensource.cirrus.com>
Date:   Fri, 1 Sep 2023 17:57:01 +0100
From:   Stefan Binding <sbinding@...nsource.cirrus.com>
To:     Waldek Andrukiewicz <waldek.social@...me>,
        Takashi Iwai <tiwai@...e.de>
CC:     Mark Brown <broonie@...nel.org>, Jaroslav Kysela <perex@...ex.cz>,
        "Takashi Iwai" <tiwai@...e.com>, <alsa-devel@...a-project.org>,
        <linux-kernel@...r.kernel.org>, <patches@...nsource.cirrus.com>
Subject: Re: [PATCH v2 00/11] Fix support for System Suspend for CS35L41 HDA


On 01/09/2023 17:48, Waldek Andrukiewicz wrote:
> On 28.07.23 15:55, Stefan Binding wrote:
>> On 28/07/2023 12:10, Takashi Iwai wrote:
>>> On Mon, 24 Jul 2023 11:00:38 +0200,
>>> Takashi Iwai wrote:
>>>> On Fri, 21 Jul 2023 17:18:05 +0200,
>>>> Stefan Binding wrote:
>>>>> There have been a couple of customer reports of intermittant issues
>>>>> after
>>>>> system resume, where sometimes the DSP firmware stops responding.
>>>>> Investigations into this issue show that there is a race between
>>>>> receiving
>>>>> a prepare from the HDA core, and the firmware reload which is
>>>>> started by
>>>>> the system resume. This can causes the Global Enable on the CS35L41
>>>>> to be
>>>>> enabled during the firmware load, which can sometimes cause issues
>>>>> in the
>>>>> DSP.
>>>>>
>>>>> The existing system resume behaviour also did not resume the audio, if
>>>>> audio was previously playing when it was suspended.
>>>>> In addition, during investigation, it was found there were additional
>>>>> problems in the System Resume sequence, as well as the Playback
>>>>> sequence
>>>>> with External Boost, where the driver does not correctly follow its
>>>>> enable sequence for this mode. This can cause additional issues
>>>>> such as
>>>>> pops and clicks.
>>>>>
>>>>> This chain intends to correct the sequences for playback and system
>>>>> suspend/resume so that the driver: obeys the external boost enable
>>>>> sequence;
>>>>> resumes audio on system resume; and avoids the race condition on
>>>>> firmware
>>>>> load and playback during system resume.
>>>>>
>>>>> Changes since v1:
>>>>> - Split patch 1 into 2 separate patches
>>>>> - Combine Patches 6 and 9
>>>>>
>>>>> Stefan Binding (11):
>>>>>     ALSA: cs35l41: Use mbox command to enable speaker output for
>>>>> external
>>>>>       boost
>>>>>     ALSA: cs35l41: Poll for Power Up/Down rather than waiting a fixed
>>>>>       delay
>>>>>     ALSA: hda: cs35l41: Check mailbox status of pause command after
>>>>>       firmware load
>>>>>     ALSA: hda: cs35l41: Ensure we correctly re-sync regmap before
>>>>> system
>>>>>       suspending.
>>>>>     ALSA: hda: cs35l41: Ensure we pass up any errors during system
>>>>>       suspend.
>>>>>     ALSA: hda: cs35l41: Move Play and Pause into separate functions
>>>>>     ALSA: hda: hda_component: Add pre and post playback hooks to
>>>>>       hda_component
>>>>>     ALSA: hda: cs35l41: Use pre and post playback hooks
>>>>>     ALSA: hda: cs35l41: Rework System Suspend to ensure correct call
>>>>>       separation
>>>>>     ALSA: hda: cs35l41: Add device_link between HDA and cs35l41_hda
>>>>>     ALSA: hda: cs35l41: Ensure amp is only unmuted during playback
>>>> Applied all patches now to for-next branch.
>>> It seems that this patch set causes occasional freeze at suspend:
>>>     https://bugzilla.suse.com/show_bug.cgi?id=1213745
>>>
>>> Could you take a look?
>>>
>>>
>>> thanks,
>>>
>>> Takashi
>> Hi Takashi,
>>
>> The initial bug report shows one of the original issues that this patch
>> chain was trying to fix.
>>  From what I can tell from the second issue, something has caused the
>> CS35L41 to stop responding,
>> which in turn caused the system suspend call to fail, and the error is
>> passed up. Since system suspend
>> failed, there was no corresponding system resume, which means the
>> CS35L41 was stuck broken.
>>
>> I'm not sure what was meant by "freeze" and "overheating" in the bug
>> report, since the log seems to
>> indicate the laptop is still responsive, even if audio is broken.
>>
>> There is some oddity in the log, since one of the errors that was
>> printed should only be printed when
>> the CS35L41 is using External Boost, but I think this laptop is supposed
>> to use Internal Boost.
>> We'll investigate further.
>>
>> Thanks,
>>
>> Stefan Binding
>
> Hello,
>
> I would like to inform you that (one of) those patches is probably
> breaking sound on Lenovo Legion 7 16ACHg6. They were applied in Manjaro
> Linux kernel here:
>
> https://gitlab.manjaro.org/packages/core/linux64/-/commit/742e66f525170fe02dec42e47aedf53d3dc85195
>
> and when I install this kernel, there is no sound anymore. There is
> nothing more than those patches in this commit. A kernel compiled from
> the previous commit works fine.
>
> Sound used to work fine from 5.17 if I remember correctly.
>
> I hope I replied correctly to this thread, if not, apologies.
>
> Best Regards,
>
> Waldek Andrukiewicz
>
>
Hi,

This is a different issue that we have been able to reproduce.
It only affects that particular laptop, and is unrelated to the other issue.
We are currently working on a fix for this.

Thanks,

Stefan