general protection fault in __xfrm_policy_unlink

[syzbot]

Hello,

syzbot found the following crash on:

HEAD commit: c3c51ea3 x86: vdso: Fix leaky vdso linker with CC=clang.
git tree: android-4.4
console output: https://syzkaller.appspot.com/x/log.txt?x=176988b7800000
kernel config: https://syzkaller.appspot.com/x/.config?x=69a973bb5ca1350a
dashboard link: https://syzkaller.appspot.com/bug?extid=d70c7981e8661f32f54a
compiler: gcc (GCC) 8.0.1 20180413 (experimental)
syz repro: https://syzkaller.appspot.com/x/repro.syz?x=14ef360f800000
C reproducer: https://syzkaller.appspot.com/x/repro.c?x=164d302f800000

IMPORTANT: if you fix the bug, please add the following tag to the commit:
Reported-by: syzbot+d70c79...@syzkaller.appspotmail.com

kasan: CONFIG_KASAN_INLINE enabled
kasan: GPF could be caused by NULL-ptr deref or user memory accessgeneral
protection fault: 0000 [#1] PREEMPT SMP KASAN
Dumping ftrace buffer:
(ftrace buffer empty)
Modules linked in:
CPU: 0 PID: 3818 Comm: syz-executor231 Not tainted 4.4.132-gc3c51ea #47
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
Google 01/01/2011
task: ffff8801ca01b000 task.stack: ffff8801d9568000
RIP: 0010:[<ffffffff833be97a>] [<ffffffff833be97a>] __write_once_size
include/linux/compiler.h:247 [inline]
RIP: 0010:[<ffffffff833be97a>] [<ffffffff833be97a>] __hlist_del
include/linux/list.h:619 [inline]
RIP: 0010:[<ffffffff833be97a>] [<ffffffff833be97a>] hlist_del
include/linux/list.h:626 [inline]
RIP: 0010:[<ffffffff833be97a>] [<ffffffff833be97a>]
__xfrm_policy_unlink+0x1ca/0x420 net/xfrm/xfrm_policy.c:1280
RSP: 0018:ffff8801d956f910 EFLAGS: 00010246
RAX: dead000000000200 RBX: ffff8801c9f23490 RCX: 0000000000000000
RDX: 0000000000000000 RSI: dffffc0000000000 RDI: 0000000000000000
RBP: ffff8801d956f950 R08: 1ffff1003a7eecc2 R09: 0000000000000000
R10: 0000000000000001 R11: ffff8801ca01b000 R12: ffff8801c9f23318
R13: ffffffff84917640 R14: ffff8801c9f23320 R15: ffff8801c9f23300
FS: 0000000001a71880(0063) GS:ffff8801db200000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000dcf710 CR3: 00000001ce13b000 CR4: 00000000001606f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Stack:
ffff8801d3f76608 ffff8801c9f23308 0000000371c0bbea 0000000000000000
ffff8801d3ab2140 0000000000000000 ffff8801c9f23300 ffffffff84917640
ffff8801d956f9a0 ffffffff833d3f16 ffffffff00000003 ffffffff84918ff8
Call Trace:
[<ffffffff833d3f16>] xfrm_sk_policy_unlink net/xfrm/xfrm_policy.c:1296
[inline]
[<ffffffff833d3f16>] xfrm_sk_policy_insert+0x1c6/0x460
net/xfrm/xfrm_policy.c:1340
[<ffffffff833e229e>] xfrm_user_policy+0x36e/0x630
net/xfrm/xfrm_state.c:1884
[<ffffffff83223fec>] do_ip_setsockopt.isra.13+0x1d5c/0x2a10
net/ipv4/ip_sockglue.c:1142
[<ffffffff83224cda>] ip_setsockopt+0x3a/0xb0 net/ipv4/ip_sockglue.c:1220
[<ffffffff83242858>] tcp_setsockopt+0x88/0xe0 net/ipv4/tcp.c:2643
[<ffffffff82f2343a>] sock_common_setsockopt+0x9a/0xe0 net/core/sock.c:2659
[<ffffffff82f20356>] SYSC_setsockopt net/socket.c:1767 [inline]
[<ffffffff82f20356>] SyS_setsockopt+0x166/0x260 net/socket.c:1746
[<ffffffff838c0225>] entry_SYSCALL_64_fastpath+0x22/0x9e
Code: 00 00 fc ff df 4c 89 f6 48 c1 ee 03 80 3c 0e 00 0f 85 ba 01 00 00 48
be 00 00 00 00 00 fc ff df 49 8b 4f 20 48 89 cf 48 c1 ef 03 <80> 3c 37 00
0f 85 07 02 00 00 48 85 d2 48 89 11 48 89 4d c0 48
RIP [<ffffffff833be97a>] __write_once_size include/linux/compiler.h:247
[inline]
RIP [<ffffffff833be97a>] __hlist_del include/linux/list.h:619 [inline]
RIP [<ffffffff833be97a>] hlist_del include/linux/list.h:626 [inline]
RIP [<ffffffff833be97a>] __xfrm_policy_unlink+0x1ca/0x420
net/xfrm/xfrm_policy.c:1280
RSP <ffff8801d956f910>
---[ end trace 1ed42494295b4582 ]---


---
This bug is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzk...@googlegroups.com.

syzbot will keep track of this bug report. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
syzbot can test patches for this bug, for details see:
https://goo.gl/tpsmEJ#testing-patches