linux-kernel - Re: [PATCH] io_uring: annotate sqd->thread access with data race in cancel path

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

Message-ID: <1d0a7d54-fcc7-495a-b9e7-be3344d21b6c@gmail.com>
Date: Sun, 12 Jan 2025 16:36:56 +0700
From: Bui Quang Minh <minhquangbui99@...il.com>
To: Pavel Begunkov <asml.silence@...il.com>, linux-kernel@...r.kernel.org
Cc: Jens Axboe <axboe@...nel.dk>, io-uring@...r.kernel.org,
 syzbot+3c750be01dab672c513d@...kaller.appspotmail.com,
 Li Zetao <lizetao1@...wei.com>
Subject: Re: [PATCH] io_uring: annotate sqd->thread access with data race in
 cancel path

On 1/12/25 08:21, Pavel Begunkov wrote:
> On 1/11/25 13:57, Bui Quang Minh wrote:
>> On 1/11/25 19:02, Pavel Begunkov wrote:
>>> On 1/11/25 10:59, Bui Quang Minh wrote:
>>>> The sqd->thread access in io_uring_cancel_generic is just for debug 
>>>> check
>>>> so we can safely ignore the data race.
>>>>
>>>> The sqd->thread access in io_uring_try_cancel_requests is to check 
>>>> if the
>>>> caller is the sq threadi with the check ctx->sq_data->thread == 
>>>> current. In
>>>> case this is called in a task other than the sq thread, we expect the
>>>> expression to be false. And in that case, the sq_data->thread read 
>>>> can race
>>>> with the NULL write in the sq thread termination. However, the race 
>>>> will
>>>> still make ctx->sq_data->thread == current be false, so we can safely
>>>> ignore the data race.
>>>>
>>>> Reported-by: syzbot+3c750be01dab672c513d@...kaller.appspotmail.com
>>>> Reported-by: Li Zetao <lizetao1@...wei.com>
>>>> Signed-off-by: Bui Quang Minh <minhquangbui99@...il.com>
>>>> ---
>>>>   io_uring/io_uring.c | 15 ++++++++++++---
>>>>   1 file changed, 12 insertions(+), 3 deletions(-)
>>>>
>>>> diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
>>>> index ff691f37462c..b1a116620ae1 100644
>>>> --- a/io_uring/io_uring.c
>>>> +++ b/io_uring/io_uring.c
>>>> @@ -3094,9 +3094,18 @@ static __cold bool 
>>>> io_uring_try_cancel_requests(struct io_ring_ctx *ctx,
>>>>           ret |= (cret != IO_WQ_CANCEL_NOTFOUND);
>>>>       }
>>>> -    /* SQPOLL thread does its own polling */
>>>> +    /*
>>>> +     * SQPOLL thread does its own polling
>>>> +     *
>>>> +     * We expect ctx->sq_data->thread == current to be false when
>>>> +     * this function is called on a task other than the sq thread.
>>>> +     * In that case, the sq_data->thread read can race with the
>>>> +     * NULL write in the sq thread termination. However, the race
>>>> +     * will still make ctx->sq_data->thread == current be false,
>>>> +     * so we can safely ignore the data race here.
>>>> +     */
>>>>       if ((!(ctx->flags & IORING_SETUP_SQPOLL) && cancel_all) ||
>>>> -        (ctx->sq_data && ctx->sq_data->thread == current)) {
>>>> +        (ctx->sq_data && data_race(ctx->sq_data->thread) == 
>>>> current)) {
>>>>           while (!wq_list_empty(&ctx->iopoll_list)) {
>>>>               io_iopoll_try_reap_events(ctx);
>>>>               ret = true;
>>>
>>> data_race() is a hammer we don't want to use to just silence warnings,
>>> it can hide real problems. The fact that it needs 6 lines of comments
>>> to explain is also not a good sign.
>>>
>>> Instead, you can pass a flag, i.e. io_uring_cancel_generic() will have
>>> non zero sqd IFF it's the SQPOLL task.
>>
>> At first, I think of using READ_ONCE here and WRITE_ONCE in the sq 
>> thread termination to avoid the data race. What do you think about 
>> this approach?
> 
> Same thing, that'd be complicating synchronisation when there
> shouldn't be any races in the first place. Having no races is
> easier than wrapping them into READ_ONCE and keeping in mind
> what that's even fine.

Okay, I'll send another patch with a new flag for the cancel path.

> Btw, the line you're changing doesn't even look right. SQPOLL
> clears sqd->task right before starting with cancellations, so
> sounds like it's mindlessly comparing NULL == current.

Hmm, I think it's correct but quite easy to get confused here. In the 
io_sq_thread, we explicitly call io_uring_cancel_generic before setting 
sqd->thread = NULL. The later io_uring_cancel_generic call in do_exit 
actually does nothing as we already set the task_struct->io_uring to 
NULL in the previous call.

Thanks,
Quang Minh.