Re: [syzbot] [io-uring?] KCSAN: data-race in io_req_task_cancel / io_wq_free_work

[Pavel Begunkov]

On 4/3/25 08:04, syzbot wrote:
> Hello,
>
> syzbot found the following issue on:
>
> HEAD commit:    92b71befc349 Merge tag 'objtool-urgent-2025-04-01' of git:..
> git tree:       upstream
> console output: https://syzkaller.appspot.com/x/log.txt?x=11195404580000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=a8120cade23cf14e
> dashboard link: https://syzkaller.appspot.com/bug?extid=903a2ad71fb3f1e47cf5
> compiler:       Debian clang version 15.0.6, GNU ld (GNU Binutils for Debian) 2.40
>
> Unfortunately, I don't have any reproducer for this issue yet.
>
> Downloadable assets:
> disk image: https://storage.googleapis.com/syzbot-assets/2e7df7bc2f52/disk-92b71bef.raw.xz
> vmlinux: https://storage.googleapis.com/syzbot-assets/be59123d5efb/vmlinux-92b71bef.xz
> kernel image: https://storage.googleapis.com/syzbot-assets/7c9eff86053e/bzImage-92b71bef.xz
>
> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> Reported-by: syzbot+903a2ad71fb3f1e47cf5@xxxxxxxxxxxxxxxxxxxxxxxxx

It's benign, the write is for unrelated bits and the flag used in
the read path is set before there could be any races, the compiler
would need to make up a value to anything bad to happen.

We can improve it a little bit though, it'll try that.

--
Pavel Begunkov