sparc64 kernel dump debug

[Andrew Belashov]

Maxim Konovalov wrote:
>>Maxim Konovalov wrote:
>>| Are there any ways to debug a kernel dump on sparc64?  On our SMP
>>| netra1440:
>>
>>I use gdb53 or gdb6 from ports.
>>
>>Try my package as temp workaround:
>>
>><http://www.orel.ru/~bel/packages/gdb-6_20040914.tbz>
>>
>>MD5 (gdb-6_20040914.tbz) = bfc1b21a9abe92f6550dfa8acd9769c7
>>SIZE = 801882
> 
> 
> Yes, it works, thank you!  Any ideas what's happened:
> 
> panic: trap: fast data access mmu miss
> cpuid = 0
> boot() called on cpu#0
> Uptime: 11d1h56m16s
> Dumping 2048 MB (1 chunks)
>   chunk at 0: 2147483648 bytes |\^H/\^H
> ---
> #0  doadump () at /usr/src/sys/kern/kern_shutdown.c:233
> 233             savectx(&dumppcb);
> doadump () at /usr/src/sys/kern/kern_shutdown.c:233
> 233             savectx(&dumppcb);
> (kgdb) bt f
> #0  doadump () at /usr/src/sys/kern/kern_shutdown.c:233
> No locals.
> #1  0x00000000c01193c0 in boot (howto=260)
>     at /usr/src/sys/kern/kern_shutdown.c:385
>         first_buf_printf = 1
> #2  0x00000000c0119ccc in panic (fmt=0xc02d1138 "trap: %s")
>     at /usr/src/sys/kern/kern_shutdown.c:541
>         _tid = (struct thread *) 0xfffff80039daea00
>         td = (struct thread *) 0xfffff80039daea00
>         bootopt = 260
>         newpanic = 1
>         buf = "trap: fast data access mmu miss", '\0' <repeats 224
> times>
> #3  0x00000000c022e8c4 in trap (tf=0xff9e51c0)
>     at /usr/src/sys/sparc64/sparc64/trap.c:370
>         td = (struct thread *) 0xfffff80039daea00
>         p = (struct proc *) 0xc033f398
>         sticks = 3223512960
>         error = -1070788296
>         sig = -6401600
> #4  0x00000000c0038fe0 in tl1_trap ()
> No symbol table info available.
> #5  0x00000000c0038fe0 in tl1_trap ()
> 
> 
> ?
> 

Hmm... Stack corrupted?

Please show tf[0], td[0], p[0] from frame #3.