Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 19 Jan 2023 17:25:53 -0500
From:   Johannes Weiner <hannes@cmpxchg.org>
To:     Suren Baghdasaryan <surenb@google.com>
Cc:     Munehisa Kamata <kamatam@amazon.com>, Tejun Heo <tj@kernel.org>,
        ebiggers@kernel.org, hdanton@sina.com,
        linux-kernel@vger.kernel.org, linux-mm@kvack.org, mengcc@amazon.com
Subject: Re: another use-after-free in ep_remove_wait_queue()
Message-ID: <Y8nDcY8EQDx+vSRv@cmpxchg.org>
References: <CAJuCfpEMEsSYcKakFiDK=QV+apW-2baLcUcw7uRyrmKkWVnR8A@mail.gmail.com>
 <20230113022555.2467724-1-kamatam@amazon.com>
 <CAJuCfpEH7kC=S8S_SRLW-X483kpaL4xdn5b35Ou08V7b56QdJA@mail.gmail.com>
 <CAJuCfpHV2-pnHd6U3paA1fO2gaYP1RTqAJwp_5QC7C2YR3JG5g@mail.gmail.com>
 <CAJuCfpFZ3B4530TgsSHqp5F_gwfrDujwRYewKReJru==MdEHQg@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CAJuCfpFZ3B4530TgsSHqp5F_gwfrDujwRYewKReJru==MdEHQg@mail.gmail.com>
Precedence: bulk

On Thu, Jan 19, 2023 at 01:01:42PM -0800, Suren Baghdasaryan wrote:
> I spent some more time digging into the details and this is what's
> happening. When we call rmdir to delete the cgroup with the pressure
> file being epoll'ed, roughly the following call chain happens in the
> context of the shell process:
> 
> do_rmdir
>   cgroup_rmdir
>     kernfs_drain_open_files
>       cgroup_file_release
>         cgroup_pressure_release
>           psi_trigger_destroy
> 
> Later on in the context of our reproducer, the last fput() is called
> causing wait queue removal:
> 
> fput
>   ep_eventpoll_release
>     ep_free
>       ep_remove_wait_queue
>         remove_wait_queue
> 
> By this time psi_trigger_destroy() already destroyed the trigger's
> waitqueue head and we hit UAF.
> I think the conceptual problem here (or maybe that's by design?) is
> that cgroup_file_release() is not really tied to the file's real
> lifetime (when the last fput() is issued). Otherwise fput() would call
> eventpoll_release() before f_op->release() and the order would be fine
> (we would remove the wait queue first in eventpoll_release() and then
> f_op->release() would cause trigger's destruction).
> Considering these findings, I think we can use the wake_up_pollfree()
> without contradicting the comment at
> https://elixir.bootlin.com/linux/latest/source/include/linux/wait.h#L253
> because indeed, cgroup_file_release() and therefore
> psi_trigger_destroy() are not tied to the file's lifetime.
> 
> I'm CC'ing Tejun to check if this makes sense to him and
> cgroup_file_release() is working as expected in this case.
> 
> Munehisha, if Tejun confirms this is all valid, could you please post
> a patch replacing wake_up_interruptible() with wake_up_pollfree()? We
> don't need to worry about wake_up_all() because we have a limitation
> of one trigger per file descriptor:

Solid analysis!

Indeed, wake_up_pollfree() should fix it.