On Mon Oct 08 2007 at 10:08:52 +0200, Juergen Hannken-Illjes wrote:
> > Here's the output from my crashes:
> > 
> >   [...]
> >   Starting amavisd.
> >   trap: TLB miss (load or instr. fetch) in kernel mode
> >   status=0xff03, cause=0x8, epc=0x801eee00, vaddr=0
> >   pid=0 cmd=system usp=0x0 ksp=0xcc6e7d50
> >   Stopped in pid 0.6 (system) at  netbsd:specificdata_getspecific:        lw      v
> >   0,0(a1)
> >   db> t
> >   specificdata_getspecific+0 (c9851060,920,2,8fd8cfd8) ra 8022bf24 sz 0
> >   fscow_run+44 (c9851060,920,2,8fd8cfd8) ra 80236618 sz 40
> >   spec_strategy+98 (c9851060,920,2,8fd8cfd8) ra 80230668 sz 32
> >   VOP_STRATEGY+2c (8fdc2dc0,8fd8c990,2,8fd8cfd8) ra 80198150 sz 40
> >   uvm_swap_io+120 (8fdc2dc0,8fd8c990,2,8fd8cfd8) ra 80195994 sz 56
> >   swapcluster_flush+70 (8fdc2dc0,8fd8c990,2,8fd8cfd8) ra 80195d7c sz 32
> >   uvm_pageout+3cc (8fdc2dc0,8fd8c990,2,8fd8cfd8) ra 802767c0 sz 176
> >   mips3_lwp_trampoline+20 (8fdc2dc0,8fd8c990,2,8fd8cfd8) ra 0 sz 24
> >   User-level: pid 0.6
> > 
> >     Juergen> show buf <1. arg of fscow_run>
> > 
> >   db> show buf 0xc9851060
> >     vp 0x2 lblkno 0x0 blkno 0x500000004 rawblkno 0x8a9c928000000006 dev 0x1
> >     error 3 flags 0x0
> >     bufsize 0xc98518c0 bcount 0x8ff51def resid 0x0
> >     data 0x2 saveaddr 0x4 dep 0x3
> >     iodone 0x0
> 
> This buffer dosn't look ok.  Could you try this diff:

The buffer is probably ok, but not located at that address.  The arguments
in the backtrace don't look sensible and I'm guessing they're not
correctly displayed on mips.

-- 
Antti Kantee <pooka@iki.fi>                     Of course he runs NetBSD
http://www.iki.fi/pooka/                          http://www.NetBSD.org/
    "la qualité la plus indispensable du cuisinier est l'exactitude"