linux-kernel - Re: [PATCH] ASoC: soc-pcm: fix fe and be race when accessing substream->runtime

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]

Date:   Thu, 29 Sep 2022 18:36:38 +0200
From:   Eugeniu Rosca <erosca@...adit-jv.com>
To:     Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>
CC:     Eugeniu Rosca <erosca@...adit-jv.com>,
        Yanmin Zhang <yanmin_zhang@...ux.intel.com>,
        <alsa-devel@...a-project.org>,
        Eugeniu Rosca <roscaeugeniu@...il.com>,
        Jiada Wang <jiada_wang@...tor.com>,
        <linux-kernel@...r.kernel.org>,
        Cezary Rojewski <cezary.rojewski@...el.com>,
        Zhang Yanmin <yanmin.zhang@...el.com>,
        Takashi Iwai <tiwai@...e.com>,
        Ramesh Babu <ramesh.babu@...el.com>,
        Liam Girdwood <lgirdwood@...il.com>,
        Dean Jenkins <Dean_Jenkins@...tor.com>,
        Mark Brown <broonie@...nel.org>,
        Ramesh Babu B <ramesh.babu.b@...el.com>,
        xiao jin <jin.xiao@...el.com>
Subject: Re: [PATCH] ASoC: soc-pcm: fix fe and be race when accessing
 substream->runtime

Hi Pierre,

On Mi, Sep 28, 2022 at 10:36:32 +0200, Pierre-Louis Bossart wrote:
> 
> 
> On 9/27/22 14:30, Eugeniu Rosca wrote:
> > Hi Pierre,
> > 
> > On Di, Sep 27, 2022 at 09:51:46 +0200, Pierre-Louis Bossart wrote:
> >> On 9/26/22 18:35, Eugeniu Rosca wrote:
> >>> From: xiao jin <jin.xiao@...el.com>
> >>>
> >>> After start of fe and be, fe might go to close without triggering
> >>> STOP, and substream->runtime is freed. However, be is still at
> >>> START state and its substream->runtime still points to the
> >>> freed runtime.
> >>>
> >>> Later on, FE is opened/started again, and triggers STOP.
> >>> snd_pcm_do_stop => dpcm_fe_dai_trigger
> >>>                 => dpcm_fe_dai_do_trigger
> >>>                 => dpcm_be_dai_trigger
> >>>                 => dpcm_do_trigger
> >>>                 => soc_pcm_trigger
> >>>                 => skl_platform_pcm_trigger
> >>> skl_platform_pcm_trigger accesses the freed old runtime data and
> >>> kernel panic.
> >>>
> >>> The patch fixes it by assigning be_substream->runtime in
> >>> dpcm_be_dai_startup when be's state is START.
> >>
> >> Can I ask on which kernel this patch was validated and on what platform?
> > 
> > As shared with Czarek, 
> > this patch was originally extracted from the out-of-tree Intel Apollo
> > Lake v4.1 KNL releases, hence it was validated on Intel ref.boards.
> > 
> > No re-testing/re-validation has been performed on the latest vanilla.
> 
> There's no way to predict how a patch for a kernel 4.1 - released 7
> years ago - would behave with a new kernel. If it's not tested it cannot
> be merged.

No disagreement here :)

> 
> > One of the goals behind submitting the patch is getting in touch
> > with the original authors, as well as the members of alsa-devel,
> > to assess if the patch is still relevant.
> 
> The only thing we could do is have more clarity on the test case and try
> to reproduce it.

Agreed. As soon as a test-case pops up where the patch makes a
difference in the runtime behavior, you will hear back from us.

> 
> >>
> >> We've done a lot of work since last year on DPCM states, 
> > 
> > Could you please feedback if the work on the DPCM states is
> > pre- or post-v5.10?
> 
> It doesn't matter for this discussion on the upstream kernel. But yes
> it's post v5.10.

Thanks. This is helpful in the downstream context.

> 
> > 
> >> and I wonder
> >> the problem mentioned above actually exists on recent kernels.
> >>
> >> Specifically, if the FE is closed, I don't get how the BE is not closed
> >> as well. And if this problem is found on a recent kernel, then it should
> >> be seen in the AVS driver as well, no?
> > 
> > It is totally conceivable (if not very likely) that the mainline
> > advancements in the sound subsystem make this patch obsolete.
> > 
> > I would be happy if that's the final outcome of our discussion
> > (since this will allow dropping the patch in our downstream kernel).

Best Regards
Eugeniu