On Fri, Oct 22, 2004 at 07:57:43AM +1000, Herbert Xu wrote:
> On Thu, Oct 21, 2004 at 09:52:24AM -0700, Phil Oester wrote:
> >
> > KERNEL: assertion (!tcp_get_pcount(&tp->packets_out) || !skb_
> > queue_empty(&sk->sk_write_queue)) failed at net/ipv4/tcp_input.c (2496)
>
> Perhaps the MTU is growing upwards? Does this patch help?
>
> Dave, we need to use the skb's MSS in trim_head as otherwise the
> counters will be screwed up.
This patch gets it a bit further, but still panics (divide by 0?):
divide error: 0000 [#1]
SMP
CPU: 0
EIP: 0060:[<c029e36c>] Not tainted VLI
EFLAGS: 00010246 (2.6.9)
EIP is at tcp_set_skb_tso_segs+0x2c/0x50
eax: 000000f3 ebx: d75f5b00 ecx: 00000000 edx: 00000000
esi: 00000001 edi: d6c22460 ebp: d6c22650 esp: c035decc
ds: 007b es: 007b ss: 0068
Process swapper (pid: 0, threadinfo=c035c000 task=c030eac0)
Stack: d75f5b00 c029e858 d75f5b00 d75f5b00 d6c22460 fffffff5 c029f3dd c02971a6
00000002 000005b4 d6c22460 d6c22650 d6c224c4 c035df58 c02a148d 00000002
00004100 00000080 f7c52260 c0393bec c038bda0 f710f130 00000010 c035df4c
Call Trace:
[<c029e858>] tcp_trim_head+0x88/0xc0
[<c029f3dd>] tcp_retransmit_skb+0x8d/0x320
[<c02971a6>] tcp_enter_loss+0x76/0x240
[<c02a148d>] tcp_retransmit_timer+0xfd/0x430
[<c02a1889>] tcp_write_timer+0xc9/0x100
[<c02a17c0>] tcp_write_timer+0x0/0x100
[<c011ee49>] run_timer_softirq+0xd9/0x170
[<c011b0f6>] __do_softirq+0xb6/0xd0
[<c011b13d>] do_softirq+0x2d/0x30
[<c010de35>] smp_apic_timer_interrupt+0x85/0xf0
[<c010492e>] apic_timer_interrupt+0x1a/0x20
[<c0101f40>] default_idle+0x0/0x40
[<c0101f6c>] default_idle+0x2c/0x40
[<c0101ffb>] cpu_idle+0x3b/0x50
[<c035ea66>] start_kernel+0x156/0x180
[<c035e530>] unknown_bootoption+0x0/0x180
|