A kernel warning at net/netlink/af_netlink.c:797

A kernel warning at net/netlink/af_netlink.c:797

[Cong Wang]

Hi again, Patrick,

I also noticed the following kernel warning today, which is likely
caused by your netlink mmap patches, although I haven't tried to
revert/skip them to confirm.

I have CONFIG_NETLINK_MMAP=y in my .config. And line 797 is:

 797         WARN_ON(atomic_read(&sk->sk_rmem_alloc));  <=== THIS
 798         WARN_ON(atomic_read(&sk->sk_wmem_alloc));
 799         WARN_ON(nlk_sk(sk)->groups);



[   38.900152] ------------[ cut here ]------------
[   38.901996] WARNING: at net/netlink/af_netlink.c:797
netlink_sock_destruct+0xd8/0x115()
[   38.905460] Hardware name: Bochs
[   38.906396] Pid: 123, comm: kworker/u:1 Tainted: G      D W
3.9.0-rc5+ #847
[   38.908575] Call Trace:
[   38.909300]  [<ffffffff810503dd>] warn_slowpath_common+0x7e/0x98
[   38.910979]  [<ffffffff810504c2>] warn_slowpath_null+0x1a/0x1c
[   38.916594]  [<ffffffff817c3497>] netlink_sock_destruct+0xd8/0x115
[   38.918285]  [<ffffffff8175e1ae>] __sk_free+0x1d/0x178
[   38.919812]  [<ffffffff8175f4e1>] sk_free+0x26/0x2a
[   38.921340]  [<ffffffff8175f503>] sock_put+0x1e/0x22
[   38.922701]  [<ffffffff8175f6a4>] sk_release_kernel+0x4e/0x52
[   38.924304]  [<ffffffff817c4f5a>] netlink_kernel_release+0xe/0x10
[   38.926092]  [<ffffffff8177c209>] rtnetlink_net_exit+0x1b/0x2b
[   38.930048]  [<ffffffff817698bc>] ops_exit_list+0x27/0x50
[   38.931582]  [<ffffffff8176a127>] cleanup_net+0xe7/0x175
[   38.933486]  [<ffffffff8106a4f9>] process_one_work+0x251/0x415
[   38.935303]  [<ffffffff8106a42c>] ? process_one_work+0x184/0x415
[   38.936964]  [<ffffffff8106b175>] worker_thread+0x13e/0x1d2
[   38.939839]  [<ffffffff8106b037>] ? manage_workers+0xf1/0xf1
[   38.941488]  [<ffffffff81073145>] kthread+0xb1/0xb9
[   38.943141]  [<ffffffff8107e497>] ? finish_task_switch+0x3d/0xc9
[   38.944851]  [<ffffffff81073094>] ? freezing+0x17/0x17
[   38.946334]  [<ffffffff81981bac>] ret_from_fork+0x7c/0xb0
[   38.947867]  [<ffffffff81073094>] ? freezing+0x17/0x17
[   38.949313] ---[ end trace dcfd26e5be103cba ]---

Thanks.

Re: A kernel warning at net/netlink/af_netlink.c:797

[Patrick McHardy]

On Sun, Apr 21, 2013 at 05:21:46PM +0800, Cong Wang wrote:
> Hi again, Patrick,
> 
> I also noticed the following kernel warning today, which is likely
> caused by your netlink mmap patches, although I haven't tried to
> revert/skip them to confirm.
> 
> I have CONFIG_NETLINK_MMAP=y in my .config. And line 797 is:
> 
>  797         WARN_ON(atomic_read(&sk->sk_rmem_alloc));  <=== THIS
>  798         WARN_ON(atomic_read(&sk->sk_wmem_alloc));
>  799         WARN_ON(nlk_sk(sk)->groups);
> 
> 
> 
> [   38.900152] ------------[ cut here ]------------
> [   38.901996] WARNING: at net/netlink/af_netlink.c:797
> netlink_sock_destruct+0xd8/0x115()
> [   38.905460] Hardware name: Bochs
> [   38.906396] Pid: 123, comm: kworker/u:1 Tainted: G      D W
> 3.9.0-rc5+ #847
> [   38.908575] Call Trace:
> [   38.909300]  [<ffffffff810503dd>] warn_slowpath_common+0x7e/0x98
> [   38.910979]  [<ffffffff810504c2>] warn_slowpath_null+0x1a/0x1c
> [   38.916594]  [<ffffffff817c3497>] netlink_sock_destruct+0xd8/0x115
> [   38.918285]  [<ffffffff8175e1ae>] __sk_free+0x1d/0x178
> [   38.919812]  [<ffffffff8175f4e1>] sk_free+0x26/0x2a
> [   38.921340]  [<ffffffff8175f503>] sock_put+0x1e/0x22
> [   38.922701]  [<ffffffff8175f6a4>] sk_release_kernel+0x4e/0x52
> [   38.924304]  [<ffffffff817c4f5a>] netlink_kernel_release+0xe/0x10
> [   38.926092]  [<ffffffff8177c209>] rtnetlink_net_exit+0x1b/0x2b
> [   38.930048]  [<ffffffff817698bc>] ops_exit_list+0x27/0x50
> [   38.931582]  [<ffffffff8176a127>] cleanup_net+0xe7/0x175
> [   38.933486]  [<ffffffff8106a4f9>] process_one_work+0x251/0x415
> [   38.935303]  [<ffffffff8106a42c>] ? process_one_work+0x184/0x415
> [   38.936964]  [<ffffffff8106b175>] worker_thread+0x13e/0x1d2
> [   38.939839]  [<ffffffff8106b037>] ? manage_workers+0xf1/0xf1
> [   38.941488]  [<ffffffff81073145>] kthread+0xb1/0xb9
> [   38.943141]  [<ffffffff8107e497>] ? finish_task_switch+0x3d/0xc9
> [   38.944851]  [<ffffffff81073094>] ? freezing+0x17/0x17
> [   38.946334]  [<ffffffff81981bac>] ret_from_fork+0x7c/0xb0
> [   38.947867]  [<ffffffff81073094>] ? freezing+0x17/0x17
> [   38.949313] ---[ end trace dcfd26e5be103cba ]---
> 
> Thanks.

Can't figure out the reason right now, I'll try to reproduce this. Thanks.