[ovs-dev] [netlink v4+3] datapath: Append ODFPT_ACTIONS attribute as fragments.

Ben Pfaff blp at nicira.com
Mon Jan 24 19:24:04 UTC 2011


On Mon, Jan 24, 2011 at 11:18:40AM -0800, Ben Pfaff wrote:
> > I'm testing this patch now, I'll send a followup if changes are needed.
> 
> Changes were needed.  I've appended the incremental and then the full
> revised patch.
> 
> But there's still something wrong with my code here.  I get the
> following warning whenever a big flow gets dumped, so obviously the
> numbers don't add up somehow:
> 
> [  141.644000] ------------[ cut here ]------------
> [  141.696000] WARNING: at net/netlink/af_netlink.c:169 netlink_sock_destruct+0x61/0x97()
> [  141.800000] Hardware name: 
> [  141.832000] Modules linked in: openvswitch_mod
> [  141.884000] Pid: 602, comm: ovs-dpctl Tainted: G        W   2.6.36+ #20
> [  141.956000] Call Trace:
> [  141.984000]  [<c1029987>] warn_slowpath_common+0x65/0x7a
> [  142.052000]  [<c1232de7>] ? netlink_sock_destruct+0x61/0x97
> [  142.124000]  [<c10299ab>] warn_slowpath_null+0xf/0x13
> [  142.184000]  [<c1232de7>] netlink_sock_destruct+0x61/0x97
> [  142.248000]  [<c120b604>] __sk_free+0x13/0x11b
> [  142.300000]  [<c120b76e>] sk_free+0x16/0x18
> [  142.348000]  [<c1232d41>] sock_put+0x13/0x15
> [  142.400000]  [<c123425b>] netlink_release+0x1bf/0x1c9
> [  142.452000]  [<c1208b5a>] sock_release+0x14/0x5b
> [  142.516000]  [<c1208bbd>] sock_close+0x1c/0x20
> [  142.568000]  [<c10ac2e7>] fput+0xee/0x173
> [  142.612000]  [<c10a9c8c>] filp_close+0x51/0x5b
> [  142.660000]  [<c102b3ca>] put_files_struct+0xc3/0x17b
> [  142.724000]  [<c102b4b9>] exit_files+0x37/0x3c
> [  142.776000]  [<c102cd1a>] do_exit+0x1ef/0x5b5
> [  142.828000]  [<c10aaefa>] ? do_sync_write+0x0/0xc5
> [  142.888000]  [<c10ab4b5>] ? vfs_write+0x9f/0xde
> [  142.940000]  [<c102d142>] do_group_exit+0x62/0x85
> [  142.992000]  [<c102d178>] sys_exit_group+0x13/0x17
> [  143.052000]  [<c12cbc51>] syscall_call+0x7/0xb
> [  143.104000] ---[ end trace 4eaa2a86a8e2da26 ]---
> 
> Any ideas?

Oh, I think I see the problem.




More information about the dev mailing list