linux-kernel - Re: [PATCH 2/2] ASoC: max98090: fix lockdep warning

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 for Android: free password hash cracker in your pocket

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

Message-ID: <CA+Px+wXkFE5b_8bLz7-c95TvEdqHGD5s-XKRYMVr40xQkqTWxQ@mail.gmail.com>
Date:   Thu, 9 Jan 2020 13:36:22 +0800
From:   Tzung-Bi Shih <tzungbi@...gle.com>
To:     Marek Szyprowski <m.szyprowski@...sung.com>
Cc:     ALSA development <alsa-devel@...a-project.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Linux Samsung SOC <linux-samsung-soc@...r.kernel.org>,
        Mark Brown <broonie@...nel.org>,
        Sylwester Nawrocki <s.nawrocki@...sung.com>,
        Dylan Reid <dgreid@...gle.com>,
        Jimmy Cheng-Yi Chiang <cychiang@...gle.com>,
        Krzysztof Kozlowski <krzk@...nel.org>,
        Bartlomiej Zolnierkiewicz <b.zolnierkie@...sung.com>
Subject: Re: [PATCH 2/2] ASoC: max98090: fix lockdep warning

On Wed, Jan 8, 2020 at 7:50 PM Marek Szyprowski
<m.szyprowski@...sung.com> wrote:
> Fix this by introducing a separate mutex only for serializing the SHDN
> hardware register related operations.

This fix makes less sense to me.  We used dapm_mutex intentionally
because: both DAPM and userspace mixer control would change SHDN bit
at the same time.

> This fixes the following lockdep warning observed on Exynos4412-based
> Odroid U3 board:
> ======================================================
> WARNING: possible circular locking dependency detected
> 5.5.0-rc5-next-20200107 #166 Not tainted
> ------------------------------------------------------
> alsactl/1104 is trying to acquire lock:
> ed0d50f4 (&card->dapm_mutex){+.+.}, at: max98090_shdn_save+0x1c/0x28
>
> but task is already holding lock:
> edb4b49c (&card->controls_rwsem){++++}, at: snd_ctl_ioctl+0xcc/0xbb8
>
> which lock already depends on the new lock.
>
> the existing dependency chain (in reverse order) is:
>
> -> #1 (&card->controls_rwsem){++++}:
>        snd_ctl_add_replace+0x3c/0x84
>        dapm_create_or_share_kcontrol+0x24c/0x2e0
>        snd_soc_dapm_new_widgets+0x308/0x594
>        snd_soc_bind_card+0x80c/0xad4
>        devm_snd_soc_register_card+0x34/0x6c
>        odroid_audio_probe+0x288/0x34c
>        platform_drv_probe+0x6c/0xa4
>        really_probe+0x200/0x490
>        driver_probe_device+0x78/0x1f8
>        bus_for_each_drv+0x74/0xb8
>        __device_attach+0xd4/0x16c
>        bus_probe_device+0x88/0x90
>        deferred_probe_work_func+0x3c/0xd0
>        process_one_work+0x22c/0x7c4
>        worker_thread+0x44/0x524
>        kthread+0x130/0x164
>        ret_from_fork+0x14/0x20
>        0x0
>
> -> #0 (&card->dapm_mutex){+.+.}:
>        lock_acquire+0xe8/0x270
>        __mutex_lock+0x9c/0xb18
>        mutex_lock_nested+0x1c/0x24
>        max98090_shdn_save+0x1c/0x28
>        max98090_put_enum_double+0x20/0x40
>        snd_ctl_ioctl+0x190/0xbb8
>        ksys_ioctl+0x470/0xaf8
>        ret_fast_syscall+0x0/0x28
>        0xbefaa564

As the stack trace suggested: when the card was still registering,
alsactl can see the mixer control and try to change it.

We have a discussion on an older thread
(https://mailman.alsa-project.org/pipermail/alsa-devel/2019-December/160454.html).
It is weird: userspace should not see things (e.g. no controlC0) until
snd_card_register( ).

I would like to spend some time to find the root cause.  It would be a
little challenging though (I have no real runtime to test...).