I thought there was a thread on cache errors previously, but I
didn't track it down quickly.   I thought it was fixed...

But it's back again:

panic: cache error @ EPC 0x80340524 ErrCtl 0x1 CacheErr 0xa0015f79
trap: TLB miss (load or instr. fetch) in kernel mode
status=0x2, cause=0x8008, epc=0x803465f8, vaddr=0x306359c4
pid=343 cmd=sshd usp=0x7fffd5c8 ksp=0xc8899cf0
Stopped in pid 343.1 (sshd) at  0x803465f8:     lw      v0,0(v1)
db> bt
8034651c+dc (80435548,306359c4,c8899db8,100) ra 80341c70 sz 0
80341c54+1c (80435548,306359c4,c8899db8,100) ra 0 sz 0
User-level: pid 343.1
db> reboot
syncing disks... 3 3 2 2 2 1 done
rebooting...

This is 100% repeatable on both Indy R5K and O2 R5K platforms.
The way I can induce it is to ssh in once, then ssh in a second
time.  Once sshd fires up the second time, the cache error
occurs.

I haven't tried R4XXX machines yet.  I'll do that later and report
back if it does NOT exist there.

On a more positive note, I have been beating upon my Indy R5K 
running a MAIN-branch build from two days ago... Aside from one
bizarre file-creation error reported earlier and this one, no 
other problems have been found.  I'm commencing beating up on
the O2 R5K now as well as my R4xxx INDY machines and an Indigo 2.

-scott