Re: [syzbot] WARNING in timerqueue_del (2)

[Thomas Gleixner]

On Sun, Aug 22 2021 at 19:45, syzbot wrote:
> HEAD commit:    3349d3625d62 Merge branch '40GbE' of git://git.kernel.org/..
> git tree:       net-next
> console output: https://syzkaller.appspot.com/x/log.txt?x=11282731300000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=a03b1e3ef878f6c1
> dashboard link: https://syzkaller.appspot.com/bug?extid=ae14beb9462a89054786
> compiler:       gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.1
>
> Unfortunately, I don't have any reproducer for this issue yet.
>
> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> Reported-by: syzbot+ae14beb9462a89054786@xxxxxxxxxxxxxxxxxxxxxxxxx
>
> ------------[ cut here ]------------
> WARNING: CPU: 1 PID: 3780 at lib/timerqueue.c:55 timerqueue_del+0xf2/0x140 lib/timerqueue.c:55
> Modules linked in:
> CPU: 1 PID: 3780 Comm: syz-executor.5 Not tainted 5.14.0-rc5-syzkaller #0
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
> RIP: 0010:timerqueue_del+0xf2/0x140 lib/timerqueue.c:55
> Code: 48 89 df e8 c0 7c ff ff 4c 89 e1 48 ba 00 00 00 00 00 fc ff df 48 c1 e9 03 80 3c 11 00 75 40 48 89 45 08 eb 82 e8 0e b8 82 fd <0f> 0b e9 4c ff ff ff 48 89 df e8 1f 51 c9 fd eb 93 4c 89 e7 e8 95
> RSP: 0000:ffffc9000101f370 EFLAGS: 00010046
> RAX: 0000000000040000 RBX: ffffe8ffffd3fce0 RCX: ffffc900145f4000
> RDX: 0000000000040000 RSI: ffffffff83f2f1a2 RDI: 0000000000000003
> RBP: ffff8880b9d42490 R08: ffffe8ffffd3fce0 R09: 0000000000000001
> R10: ffffffff83f2f0ec R11: 0000000000000000 R12: ffffe8ffffd3fce0
> R13: 0000000000000001 R14: ffff8880b9d423c0 R15: 0000000000000000
> FS:  00007f8847706700(0000) GS:ffff8880b9d00000(0000) knlGS:0000000000000000
> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 00007f333ef09f80 CR3: 0000000064822000 CR4: 00000000001506e0
> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> Call Trace:
>  __remove_hrtimer+0xa1/0x2a0 kernel/time/hrtimer.c:1014
>  remove_hrtimer+0x19f/0x410 kernel/time/hrtimer.c:1054
>  hrtimer_try_to_cancel kernel/time/hrtimer.c:1186 [inline]
>  hrtimer_try_to_cancel+0x102/0x1e0 kernel/time/hrtimer.c:1168
>  hrtimer_cancel+0x13/0x40 kernel/time/hrtimer.c:1295
>  napi_disable+0xc3/0x110 net/core/dev.c:6909

This smells badly of memory corruption.

remove_hrtimer() invokes __remove_hrtimer() only when hrtimer->state has
the HRTIMER_STATE_ENQUEUED bit set which in turn means that the hrtimer
is queued in the timerqueue. But timerqueue_del() sees an empty
hrtimer->node.

I double checked the hrtimer code whether there is a blind spot where
the timer state and the node state could get out of sync, but I could
not find one.

Thanks,

        tglx