Re: vfs_hash_get panic

On 12/16/06, Rong-en Fan <grafan_at_gmail.com> wrote:
> I'm running amd64 SMP -current around Dec 14. It seems that I can reproduce
> this panic by 'mergemaster -iU' at /usr/src reliably.

I did nothing but mergemater may not 100% reproduce this.
However, it seems that buildworld causes panic at different point
each time. I always do 'fsck -fy' each time after panic. Now
the host is in ddb again.

> Fatal trap 12: page fault while in kernel mode
> cpuid = 0; apic id = 00
> fault virtual address   = 0x68
> fault code              = supervisor read data, page not present
> instruction pointer     = 0x8:0xffffffff802e0ce3
> stack pointer           = 0x10:0xffffffffa0019510
> frame pointer           = 0x10:0xffffffffa0019570
> code segment            = base 0x0, limit 0xfffff, type 0x1b
>                         = DPL 0, pres 1, long 1, def32 0, gran 1
> processor eflags        = interrupt enabled, resume, IOPL = 0
> current process         = 786 (mtree)
>
> Due to no remote console attached, the console ddb backtrace
> is available at flickr:
>
> http://flickr.com/photos/rafan/323030710/
>
> I put my kernel conf and kgdb output here:
>
> http://www.rafan.org/FreeBSD/panic/vfs_hash_get/KERNEL
> http://www.rafan.org/FreeBSD/panic/vfs_hash_get/kgdb.txt
>
> I saw there is another report on _at_hackers this Oct. But it seems
> the originator solves the problem by upgrading to latest RELENG_6.
>
> Any suggestions? I can provide more information if someone interested.
>
> Regards,
> Rong-En Fan
>