linux-kernel - Re: INFO: task hung in __io_uring_task

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>] [day] [month] [year] [list]

Message-ID: <CAGyP=7dQP_b066kgDjeUCs1RUNvkOwoxirHsubL2yG2ZyjwsqA@mail.gmail.com>
Date:   Mon, 4 Jan 2021 15:10:48 +0530
From:   Palash Oswal <oswalpalash@...il.com>
To:     Hillf Danton <hdanton@...a.com>
Cc:     axboe@...nel.dk, io-uring@...r.kernel.org,
        linux-kernel@...r.kernel.org, syzkaller-bugs@...glegroups.com,
        Pavel Begunkov <asml.silence@...il.com>
Subject: Re: INFO: task hung in __io_uring_task_cancel

On Mon, Jan 4, 2021 at 12:22 PM Hillf Danton <hdanton@...a.com> wrote:
> It is now updated.

Hello Hilf,

Thanks for the new diff. I tested by applying the diff on 5.10.4 with
the original reproducer, and the issue still persists.

root@...kaller:~# [  242.925799] INFO: task repro:416 blocked for more
than 120 seconds.
[  242.928095]       Not tainted 5.10.4+ #12
[  242.929034] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
disables this message.
[  242.930825] task:repro           state:D stack:    0 pid:  416
ppid:   415 flags:0x00000004
[  242.933404] Call Trace:
[  242.934365]  __schedule+0x28d/0x7e0
[  242.935199]  ? __percpu_counter_sum+0x75/0x90
[  242.936265]  schedule+0x4f/0xc0
[  242.937159]  __io_uring_task_cancel+0xc0/0xf0
[  242.938340]  ? wait_woken+0x80/0x80
[  242.939380]  bprm_execve+0x67/0x8a0
[  242.940163]  do_execveat_common+0x1d2/0x220
[  242.941090]  __x64_sys_execveat+0x5d/0x70
[  242.942056]  do_syscall_64+0x38/0x90
[  242.943088]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[  242.944511] RIP: 0033:0x7fd0b781e469
[  242.945422] RSP: 002b:00007fffda20e9c8 EFLAGS: 00000246 ORIG_RAX:
0000000000000142
[  242.947289] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007fd0b781e469
[  242.949031] RDX: 0000000000000000 RSI: 0000000020000180 RDI: 00000000ffffffff
[  242.950683] RBP: 00007fffda20e9e0 R08: 0000000000000000 R09: 00007fffda20e9e0
[  242.952450] R10: 0000000000000000 R11: 0000000000000246 R12: 0000556068200bf0
[  242.954045] R13: 00007fffda20eb00 R14: 0000000000000000 R15: 0000000000000000


linux git:(b1313fe517ca)  git diff
diff --git a/fs/io_uring.c b/fs/io_uring.c
index 0fcd065baa76..e0c5424e28b1 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -1867,8 +1867,7 @@ static void __io_free_req(struct io_kiocb *req)
     io_dismantle_req(req);

     percpu_counter_dec(&tctx->inflight);
-    if (atomic_read(&tctx->in_idle))
-        wake_up(&tctx->wait);
+    wake_up(&tctx->wait);
     put_task_struct(req->task);

     if (likely(!io_is_fallback_req(req)))
@@ -8853,12 +8852,11 @@ void __io_uring_task_cancel(void)
          * If we've seen completions, retry. This avoids a race where
          * a completion comes in before we did prepare_to_wait().
          */
-        if (inflight != tctx_inflight(tctx))
-            continue;
-        schedule();
+        if (inflight == tctx_inflight(tctx))
+            schedule();
+        finish_wait(&tctx->wait, &wait);
     } while (1);

-    finish_wait(&tctx->wait, &wait);
     atomic_dec(&tctx->in_idle);
 }