Re: Oops with racoon and linux-2.6.9-rc2-mm1

[Zilvinas Valinskas <zilvinas@gemtek.lt>]

Seems like the same patch is missing from linux 2.6.9-rc2-bk8 too.
As reported this solves the problem in mm tree ...

BR


On Thu, Sep 23, 2004 at 06:11:26AM +0200, Patrick McHardy wrote:
> Should be fixed by this patch.
> 
> Regards
> Patrick
> 
> Mathieu B?rard wrote:
> 
> >Hi,
> >
> >I get this Oops using racoon and linux 2.6.9-rc2-mm1,
> >though everything run smoothly with a 2.6.8 kernel.
> >
> >Here the oops (copied by hand):
> >
> >Warning: kfree_skb on hard IRQ 000000c0
> >scheduling while atomic: racoon/0xffffff00/1680
> >[<c027c7bb>] schedule+0x32b/0x480
> >[<c0114a4a>] sys_sched_yield+0x1a/0x20
> >[<c0153e3b>] coredump_wait+0x2b/0x90
> >[<c0153f6a>] do_coredump+0xca/0x1a1
> >[<c0132d1d>] buffered_rmqueue+0xdd/0x1b0
> >[<c0128290>] autoremove_wake_function+0x0/0x50
> >[<c011f365>] __dequeue_signal+0xe5/0x150
> >[<c011f3f3>] dequeue_signal+0x23/0x90
> >[<c0120b8d>] get_signal_to_deliver+0x24d/0x300
> >[<c0103c6f>] do_signal+0x8f/0x160
> >[<c015a2e0>] __pollwait+0x0/0xc0
> >[<c0227687>] sys_recv+0x37/0x40
> >[<c010618c>] do_IRQ+0xec/0x190
> >[<c0112a90>] do_page_fault+0x0/0x575
> >[<c0103d77>] do_notify_resume+0x37/0x3c
> >[<c0103f1a>] work_notifysig+0x13/0x15
> >Kernel panic - not syncing: Aiee, killing interrupt handler!
> >
> 

> # This is a BitKeeper generated diff -Nru style patch.
> #
> # ChangeSet
> #   2004/09/23 06:05:52+02:00 kaber@coreworks.de 
> #   [XFRM]: Fix unbalanced spin_unlock_bh
> #   
> #   Signed-off-by: Patrick McHardy <kaber@trash.net>
> # 
> # net/xfrm/xfrm_state.c
> #   2004/09/23 06:05:25+02:00 kaber@coreworks.de +0 -1
> #   [XFRM]: Fix unbalanced spin_unlock_bh
> #   
> #   Signed-off-by: Patrick McHardy <kaber@trash.net>
> # 
> diff -Nru a/net/xfrm/xfrm_state.c b/net/xfrm/xfrm_state.c
> --- a/net/xfrm/xfrm_state.c	2004-09-23 06:07:23 +02:00
> +++ b/net/xfrm/xfrm_state.c	2004-09-23 06:07:23 +02:00
> @@ -592,7 +592,6 @@
>  		list_for_each_entry(x, xfrm_state_bydst+i, bydst) {
>  			if (x->km.seq == seq) {
>  				xfrm_state_hold(x);
> -				spin_unlock_bh(&xfrm_state_lock);
>  				return x;
>  			}
>  		}