zfs, crash, 7-CURRENT

[Thede Loder]

Hi Craig,
Sorry for the delay - couldn't get to writing it up until now.  Here's 
the backtrace:

Tracing pid 3 tid 100011 td 0xc3a19000
_mtx_lock_flags (1c, 0, c6cd24a7, 1d8, c6ba6ad4, ...) at 
_mtx_lock_flags+0x46
vdev_geom_io_intr(c6ba6ad4,0,c0a9b395,bc5,c6ba6ad4,...) at 
vdev_geom_io_intr+0x41
biodone (c6ba6ad4, c0ba4488,24c,c0a89c98,64,...) at biodone+0x99
g_io_schedule_up(c3a19000,0,c0a8b393,5d,0,...) at g_io_schedule_up+0xc7
g_up_procbody (0,dc6c4d38,c0a8eced,315,c3a5d2ac,...) at g_up_procbody+0x8d
fork_exit(c06fe7e0,0,dc6c4d38, at fork_exit+0xb8
fork_trampoline() at fork_trampoline+0x8
--- trap 0, eip = 0, esp = 0xdc6c4d70, ebp = 0 ----

If it's relevant, the backtrace is typically preceded by a number of 
UDMA ICRC errors on the console naming an LBA value (which varies). 
Then I'm given the debugging prompt.

Cheers,
Thede


Craig Rodrigues wrote:
> 
> 
> On Wed, Jul 18, 2007 at 09:46:22AM -0700, Thede Loder wrote:
>> 
>> Hi Craig,
>> 
>> Will do.  I'll re-produce the crash tonight.
>> 
>> Also, since savecore was already enabled, I can provide six crash dumps 
>> (all the same type of crash).  Would those be helpful?
>> 
>> Thede
> 
> 
> The actual corefiles are useful only for someone debugging
> on your actual system (they would need the kernel.debug file that
> accompanies the kernel vmcore file).  For posting info to the
> mailing list, a backtrace is the most useful info to post.
> 
> Thanks!
> -- 
> Craig Rodrigues        
> rodrigc at crodrigues.org
>