2.6.22-rc6: local_bh_enable warning

Having upgraded from 2.6.16 to 2.6.22-rc6, I'm now seeing
the following.

Looks like netfilter is calling local_bh_enable() with IRQs
disabled,
which would appear to be illegal.  Thankfully, this is a
warn-once
warning.

WARNING: at /home/rmk/git/linux-2.6-rmk/kernel/softirq.c:138
local_bh_enable()
[<c0021c70>] (dump_stack+0x0/0x14) from
[<c00358c0>] (local_bh_enable+0x50/0xac)
[<c0035870>] (local_bh_enable+0x0/0xac) from
[<bf0090cc>] (nf_nat_cleanup_conntrack+0x54/0x60
[nf_nat])
 r6:c1a9aa00 r5:c1a9aa00 r4:c15bad3c
[<bf009078>] (nf_nat_cleanup_conntrack+0x0/0x60
[nf_nat]) from [<c0146d50>]
(destroy_conntrack+0x6c/0x118)
 r4:c15bac58
[<c0146ce4>] (destroy_conntrack+0x0/0x118) from
[<c0144820>] (nf_conntrack_destroy+0x24/0x2c)
 r4:c05d2a80
[<c01447fc>] (nf_conntrack_destroy+0x0/0x2c) from
[<c012c05c>] (__kfree_skb+0xd0/0x100)
[<c012bf8c>] (__kfree_skb+0x0/0x100) from
[<c012c0d8>] (kfree_skb+0x4c/0x50)
 r5:c12a3800 r4:00000300
[<c012c08c>] (kfree_skb+0x0/0x50) from
[<bf03cbb0>] (el3_start_xmit+0xb8/0xd0 [3c589_cs])
[<bf03caf8>] (el3_start_xmit+0x0/0xd0 [3c589_cs]) from
[<c01324dc>] (dev_hard_start_xmit+0x1a8/0x244)
 r7:c12a3800 r6:c1a9aa00 r5:c1a9aa00 r4:c12a3800
[<c0132334>] (dev_hard_start_xmit+0x0/0x244) from
[<c013fcc0>] (__qdisc_run+0xb0/0x198)
 r8:00000000 r7:c12a382c r6:c1a9aa00 r5:00000000
r4:c12a3800
[<c013fc10>] (__qdisc_run+0x0/0x198) from
[<c01326dc>] (dev_queue_xmit+0x164/0x234)
 r7:c0ed2560 r6:00000000 r5:c1a9aa00 r4:c12a3800
[<c0132578>] (dev_queue_xmit+0x0/0x234) from
[<c01572d0>] (ip_output+0x2c0/0x324)
 r6:0000000e r5:c1a9aa00 r4:00000000
[<c0157010>] (ip_output+0x0/0x324) from
[<c0158b94>] (ip_push_pending_frames+0x438/0x53c)
 r7:c05d2a80 r6:00000040 r5:c02e8e40 r4:c11fdc10
[<c015875c>] (ip_push_pending_frames+0x0/0x53c) from
[<c0158fa0>] (ip_send_reply+0x1c8/0x22c)
[<c0158dd8>] (ip_send_reply+0x0/0x22c) from
[<c016dd2c>] (tcp_v4_send_reset+0x148/0x17c)
[<c016dbe4>] (tcp_v4_send_reset+0x0/0x17c) from
[<c016ecb4>] (tcp_v4_do_rcv+0x368/0x3a8)
 r7:22a7d38c r6:c0a71060 r5:c1a9a8c0 r4:c0a71060
[<c016e94c>] (tcp_v4_do_rcv+0x0/0x3a8) from
[<c016f2b0>] (tcp_v4_rcv+0x5bc/0x8c4)
[<c016ecf4>] (tcp_v4_rcv+0x0/0x8c4) from
[<c0152a30>] (ip_local_deliver+0x188/0x218)
[<c01528a8>] (ip_local_deliver+0x0/0x218) from
[<c0153374>] (ip_rcv+0x568/0x5c0)
 r6:c1a9a8c0 r5:c1a9a8c0 r4:c11fd420
[<c0152e0c>] (ip_rcv+0x0/0x5c0) from
[<c0132ca8>] (netif_receive_skb+0x268/0x29c)
 r7:c12a3800 r6:00000008 r5:c025d96c r4:c1a9a8c0
[<c0132a40>] (netif_receive_skb+0x0/0x29c) from
[<c0132d88>] (process_backlog+0xac/0x18c)
 r8:0002f9af r7:c0241ee8 r6:00000000 r5:c025d60c
r4:c12a3800
[<c0132cdc>] (process_backlog+0x0/0x18c) from
[<c0132ef8>] (net_rx_action+0x90/0x168)
[<c0132e68>] (net_rx_action+0x0/0x168) from
[<c0035750>] (__do_softirq+0x5c/0xc8)
 r8:000189c0 r7:c0257960 r6:0000000a r5:c02579b4
r4:00000001
[<c00356f4>] (__do_softirq+0x0/0xc8) from
[<c0035980>] (irq_exit+0x40/0x48)
 r7:00000000 r6:c025c280 r5:c0245950 r4:00000006
[<c0035940>] (irq_exit+0x0/0x48) from
[<c001d04c>] (asm_do_IRQ+0x4c/0x60)
[<c001d000>] (asm_do_IRQ+0x0/0x60) from
[<c001da28>] (__irq_svc+0x28/0x80)

-- 
Russell King

Russell King wrote:
> Having upgraded from 2.6.16 to 2.6.22-rc6, I'm now
seeing the following.
> 
> Looks like netfilter is calling local_bh_enable() with
IRQs disabled,
> which would appear to be illegal.  Thankfully, this is
a warn-once
> warning.
> 
> WARNING: at
/home/rmk/git/linux-2.6-rmk/kernel/softirq.c:138
local_bh_enable()
> [...]
> [<c01447fc>] (nf_conntrack_destroy+0x0/0x2c) from
[<c012c05c>] (__kfree_skb+0xd0/0x100)
> [<c012bf8c>] (__kfree_skb+0x0/0x100) from
[<c012c0d8>] (kfree_skb+0x4c/0x50)
>  r5:c12a3800 r4:00000300
> [<c012c08c>] (kfree_skb+0x0/0x50) from
[<bf03cbb0>] (el3_start_xmit+0xb8/0xd0 [3c589_cs])
> [<bf03caf8>] (el3_start_xmit+0x0/0xd0 [3c589_cs])
from [<c01324dc>] (dev_hard_start_xmit+0x1a8/0x244)
>  r7:c12a3800 r6:c1a9aa00 r5:c1a9aa00 r4:c12a3800
> [<c0132334>] (dev_hard_start_xmit+0x0/0x244) from
[<c013fcc0>] (__qdisc_run+0xb0/0x198)


Thats a bug in the 3c589_cs driver. Patch attached.

Signed-off-by: Patrick McHardy <kabertrash.net>

Jan,
I've updated sources according your review (I hope I
understood as you meant 
it), thanks. I've changed also to xt_RTPPROXY using xtables
and tested at 
last stable kernel 2.6.21.5. Rereview welcome.

Key question again, do you have a svn repository where
publish it as external 
pom?

Tomas


> On Jun 27 2007 20:32, Tomas Mandys wrote:
> >so I've finally "finished" work on
RTPPROXY module, it seems it works
> >now for kernel 2.6.17.8.
> >Could änybody review code? What this module should
support is described
> >in quoted text bellow and in relatively extensive
> >netfilter-rtpptoxy-howto.txt plus in manpages in
tar. I'm not sure if
> >devel list accepts attachments so here is also the
link
> >http://w
ww.2p.cz/tmp/netfilter-rtpproxy.tgz.
>
> (1) CRLF line endings -> LF.
strange, when it happend ???....
>
> (2) I'd suggest writing xtables now. Make it
xt_RTPPROXY.
>
> >#include <linux/module.h>
> >#include <linux/vmalloc.h>
> >#include <linux/skbuff.h>
> >#include <linux/ip.h>
> >#include <linux/netfilter_ipv4/ip_tables.h>
>
> <linux/netfilter/x_tables.h>
>
> >#include
<linux/netfilter_ipv4/ip_conntrack.h>
> >#include
<linux/netfilter_ipv4/ipt_RTPPROXY.h>
> >#include <linux/netfilter_ipv4/ip_nat.h>
>
> These files are probably outdated (use nf_* instead)
> .......

Patrick McHardy wrote:
> Russell King wrote:
>> Having upgraded from 2.6.16 to 2.6.22-rc6, I'm now
seeing the following.
>>
>> Looks like netfilter is calling local_bh_enable()
with IRQs disabled,
>> which would appear to be illegal.  Thankfully, this
is a warn-once
>> warning.
>>
>> WARNING: at
/home/rmk/git/linux-2.6-rmk/kernel/softirq.c:138
local_bh_enable()
>> [...]
>> [<c01447fc>] (nf_conntrack_destroy+0x0/0x2c)
from [<c012c05c>] (__kfree_skb+0xd0/0x100)
>> [<c012bf8c>] (__kfree_skb+0x0/0x100) from
[<c012c0d8>] (kfree_skb+0x4c/0x50)
>>  r5:c12a3800 r4:00000300
>> [<c012c08c>] (kfree_skb+0x0/0x50) from
[<bf03cbb0>] (el3_start_xmit+0xb8/0xd0 [3c589_cs])
>> [<bf03caf8>] (el3_start_xmit+0x0/0xd0
[3c589_cs]) from [<c01324dc>]
(dev_hard_start_xmit+0x1a8/0x244)
>>  r7:c12a3800 r6:c1a9aa00 r5:c1a9aa00 r4:c12a3800
>> [<c0132334>] (dev_hard_start_xmit+0x0/0x244)
from [<c013fcc0>] (__qdisc_run+0xb0/0x198)
> 
> 
> Thats a bug in the 3c589_cs driver. Patch attached.
> 
> Signed-off-by: Patrick McHardy <kabertrash.net>

You would post this JUST after I finished my patch applying
run, 
wouldn't you?  

Will apply in a few minutes, thanks 

	Jeff