Re: r194546 amd64: kernel panic in tcp_sack.c

Hi,

Kamigishi Rei wrote:
> Hello, hope you're having a nice day,
> 
> I've been testing my system mostly to check if ZFS in -current is 
> stable, however so far I've been getting kernel panics in every other 
> area except ZFS. This time it's tcp_sack.c according to the panic 
> message, inside [intr].

lol, handy!

> 
> fujibayashi_at_ameagari ~ % uname -a
> FreeBSD ameagari.fujibayashi.jp 8.0-CURRENT FreeBSD 8.0-CURRENT #1 
> r194546: Thu Jun 25 19:44:18 JST 2009     
> root_at_ameagari.fujibayashi.jp:/usr/src/sys/amd64/compile/Ameagari  amd64
> 
> panic: tcp_sack_globalholes >= 0
> cpuid = 0
> KDB: enter: panic
> [thread pid 12 tid 100005]
> Stopped at kdb_enter+0x3d: movq $0,0x682580(%rip)
> db> bt
> Tracing pid 12 tid 100005 td 0xffffff0001320000
> kdb_enter() at kdb_enter+0x3d
> panic() at panic+0x17b
> tcp_sackhole_remove() at tcp_sackhole_remove+0xc7
> tcp_free_sackholes() at tcp_free_sackholes+0x48
> tcp_timer_rexmt() at tcp_timer_rexmt+0xb3
> softclock() at softclock+0x291
> intr_event_execute_handlers() at intr_event_execute_handlers+0x68
> ithread_loop() at ithread_loop+0xb2
> fork_exit() at fork_exit+0x12a
> fork_trampoline() at fork_trampoline+0xe
> --- trap 0, rip = 0, rsp = 0xffffff8000026d40, rbp = 0 ---
> db >
> 

I'm not intimately familiar with our SACK implementation, and these 
things are often extremely painful to track down. First step: is the 
panic reproducible?

> No core saved - when I tried to get it to save the core, it just raised 
> fatal trap 12 (page fault) and got a general protection fault afterwards.
> Any ideas?

How did you try to get it to save the core? A dump would be very useful 
to have around.

Cheers,
Lawrence