Re: 5000/25

[David Holland <dholland-mips%netbsd.org@localhost>]

On Mon, Feb 24, 2014 at 05:48:12AM +0000, Alastair Boyanich wrote:
 > Complains with 6.1.3 along the lines of (please excuse me re-typing
 > this off the vt220, I hope I didn't make any errors..):
 > [...]
 > segment? 0 start 00000000 size 01800000
 > phys segment: 0x18000000 @ 0
 > adding 0x293000 @ 0x56d000 to freelist 1
 > adding 0x1000000 @ 0x800000 to freelist 0
 > pmap_steal_memory: seg 0: 0x56d 0x56d 0x800 0x800
 > pid 0(system): trap: cpu0, address error (load or I-fetch) in kernel mode
 > status=0x80010, cause=0x30000010, epc=0x8000001c, vaddr=0xdeadbeef
 > tf=0x8056cce8 ksp=0x8056cd88 ra=0x800b0120 ppl=0
 > panic: trap
 > halted.

Have you tried -current? If the problem has been around since 1.6.2
this probably won't help though.

Getting deadbeef at 0x80000001c is very suspicious since that's in the
low-level TLB refill handler. However, it may be nothing other than a
null dereference early enough in boot that the pagetable isn't set up
yet.

My first suggestion would be to build a kernel with ddb in it and try
to get a stack trace. If that doesn't work, there's probably nothing
for it but loading up with printfs to try to figure out where it's
dying. :(

-- 
David A. Holland
dholland%netbsd.org@localhost