On Thu, May 09, 2019 at 09:54:14AM -0700, Zubin Mithra wrote:
> Hello,
>
> Syzkaller has triggered a lockdep warning when fuzzing a 4.4 kernel with the following stacktrace.
>
> Call Trace:
> [<ffffffff81cb9fad>] __dump_stack lib/dump_stack.c:15 [inline]
> [<ffffffff81cb9fad>] dump_stack+0xc1/0x124 lib/dump_stack.c:51
> [<ffffffff813eceac>] print_circular_bug.cold.51+0x1bd/0x27d kernel/locking/lockdep.c:1226
> [<ffffffff81207f1a>] check_prev_add kernel/locking/lockdep.c:1853 [inline]
> [<ffffffff81207f1a>] check_prevs_add kernel/locking/lockdep.c:1958 [inline]
> [<ffffffff81207f1a>] validate_chain kernel/locking/lockdep.c:2144 [inline]
> [<ffffffff81207f1a>] __lock_acquire+0x38da/0x52a0 kernel/locking/lockdep.c:3213
> [<ffffffff8120b0be>] lock_acquire+0x15e/0x440 kernel/locking/lockdep.c:3592
> [<ffffffff82a53056>] __mutex_lock_common kernel/locking/mutex.c:624 [inline]
> [<ffffffff82a53056>] mutex_lock_nested+0xc6/0x10b0 kernel/locking/mutex.c:744
> [<ffffffff822e186c>] rtnl_lock+0x1c/0x20 net/core/rtnetlink.c:70
> [<ffffffff828ae743>] ipv6_sock_mc_close+0x113/0x350 net/ipv6/mcast.c:288
> [<ffffffff82875f06>] do_ipv6_setsockopt.isra.12+0xce6/0x2cc0 net/ipv6/ipv6_sockglue.c:202
> [<ffffffff82877f7c>] ipv6_setsockopt+0x9c/0x130 net/ipv6/ipv6_sockglue.c:905
> [<ffffffff828863af>] udpv6_setsockopt+0x4f/0x90 net/ipv6/udp.c:1436
> [<ffffffff82250fef>] sock_common_setsockopt+0x9f/0xe0 net/core/sock.c:2693
> [<ffffffff8224e223>] SYSC_setsockopt net/socket.c:1780 [inline]
> [<ffffffff8224e223>] SyS_setsockopt+0x163/0x250 net/socket.c:1759
> [<ffffffff82a5f267>] entry_SYSCALL_64_fastpath+0x1e/0xa0
>
> Could the following patch be applied in order to v4.4.y? This patch is present in
> linux-4.9.y.
> * 8651be8f14a1 ("ipv6: fix a potential deadlock in do_ipv6_setsockopt()")
>
> Tests run:
> * Chrome OS tryjobs
> * Syzkaller reproducer
Now queued up, thanks.
greg k-h
