I had the same problem in an x86 box, and was a RAM problem, I used to buy Kingston and everything goes right.

At the beggining this machine was a FreeBSD, and RAM problems was converted in DMA errors that were simultaneus converted into write garbage in the disk. I tried netbsd there, and when rebuilding the entire system I get some gcc segfaults and kernelpanics in pmap and vm functions.

pmap and vm are functions that directly access to the RAM. I changed to another RAM and everything goes right.

I think this would be the problem...but talking about macppc...I'm sure if this is the problem, or it's just a kernel bug.

On Tue, 11 Oct 2005 11:16:09 +0100
Rui Paulo <rpaulo@fnop.net> wrote:

> On 2005.10.11 00:58:15 -0700, John Klos wrote:
> | 
> | I am at a complete loss. I've been doing bulk builds on this for weeks to 
> | see if it's stable, and after I moved it to the new datacenter today, this 
> | is what happens when I was cvs'ing...
> 
> I have similar problems when I'm on some high load:
> 
> db> bt
> 0xd510ba40: at panic+0x19c
> 0xd510bad0: at trap+0xfc
> 0xd510bb50: kernel MCHK trap by uvm_pagedequeue+0x90: srr1=0x49030
>             r1=0xd510bc10 cr=0x44422044 xer=0x20000000 ctr=0x38138c
> 0xd510bc10: at ADBDevTable+0xa23768
> 0xd510bc20: at uvm_pagefree+0x1ec
> 0xd510bc50: at genfs_putpages+0x988
> 0xd510bdb0: at VOP_PUTPAGES+0x50
> 0xd510bdf0: at uvn_put+0xc
> 0xd510be00: at uvmpd_scan_inactive+0x440
> 0xd510bec0: at uvmpd_scan+0x7c
> 0xd510bf00: at uvm_pageout+0x12c
> 0xd510bf40: at cpu_switchto+0x44
> 0xd510bf50: at ADBDevTable+0x4b9a49fc
> 
> 
> 
> 		-- Rui Paulo
>