linux-kernel - Re: [cgroup] 3c87862ca1: WARNING:at_kernel/softirq.c:#__local_bh_enable

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, 9 Jun 2022 16:55:49 +0200
From:   Michal Koutný <mkoutny@...e.com>
To:     Tadeusz Struk <tadeusz.struk@...aro.org>
Cc:     kernel test robot <oliver.sang@...el.com>,
        0day robot <lkp@...el.com>, Tejun Heo <tj@...nel.org>,
        Zefan Li <lizefan.x@...edance.com>,
        Johannes Weiner <hannes@...xchg.org>,
        Christian Brauner <brauner@...nel.org>,
        Alexei Starovoitov <ast@...nel.org>,
        Daniel Borkmann <daniel@...earbox.net>,
        Andrii Nakryiko <andrii@...nel.org>,
        Martin KaFai Lau <kafai@...com>,
        Song Liu <songliubraving@...com>, Yonghong Song <yhs@...com>,
        John Fastabend <john.fastabend@...il.com>,
        KP Singh <kpsingh@...nel.org>,
        LKML <linux-kernel@...r.kernel.org>, cgroups@...r.kernel.org,
        lkp@...ts.01.org, netdev@...r.kernel.org, bpf@...r.kernel.org,
        stable@...r.kernel.org,
        syzbot+e42ae441c3b10acf9e9d@...kaller.appspotmail.com
Subject: Re: [cgroup] 3c87862ca1:
 WARNING:at_kernel/softirq.c:#__local_bh_enable_ip

Hello Tadeusz.

On Thu, Jun 09, 2022 at 07:30:41AM -0700, Tadeusz Struk <tadeusz.struk@...aro.org> wrote:
> Are you interested in fixing this at syzbot issue all?

The (original) syzbot report is conditioned by allocation failure that's
unlikely under normal conditions (AFAIU). Hence I don't treat it extra
high urgent.
OTOH, it's interesting and it points to some disparity worth fixing --
so I try helping (as time permits, so far I can only run the reproducers
via the syzbot).

> Do you have any more feedback on this?

Ad the patch v2 with spinlock per css -- that looks like an overkill to
me, I didn't look deeper into it.

Ad the in-thread patch with ancestry css_get(), the ->parent ref:
  - is inc'd in init_and_link_css(),
  - is dec'd in css_free_rwork_fn()
and thanks to ->online_cnt offlining is ordered (child before parent).

Where does your patch dec these ancestry references? (Or why would they
be missing in the first place?)

Thanks for digging into this!
Michal