I've been trying every now and again a new kernel, and have not had one 
stable enough to compile "the world" since late June.  They all end up 
panicking after a few days or so under no load, or sooner under load.  This 
time I was doing a make build and was in the cleandir phase.  I telnet to the 
box and also have a minicom window up to catch the panic.

I'm running a Olivetti M700-10 with netbsd-1-5 from Friday night.  And this 
is what I got.

What should I be looking for when this happens?  Do others have this kind of 
problems?

Any help is appreciated.  Thanks,

....maki....

FYI the address it faulted at is probably in pica_clkintr
80346a80 T pica_clkintr
80346abc T rd94_iointr



panic: TLB out of universe: ksp was 0xc33edfe0
Stopped at      0x80346aac:     lw      ra,16(sp)
db> trace
80346a80+2c (4001,20,16df13,0) ra 8020006c sz 24
User-level: curproc NULL
db> reboot
803430d4+40 (0,0,801fff78,801ffdc0) ra 80207894 sz 24
80207844+50 (0,0,801fff78,801ffdc0) ra 8020739c sz 32
802070e4+2b8 (0,0,801fff78,801ffdc0) ra 80207648 sz 184
80207598+b0 (0,0,0,801ffdc0) ra 8020b26c sz 80
8020b190+dc (0,0,0,801ffdc0) ra 80337530 sz 32
80337364+1cc (0,0,0,801ffdc0) ra 8033cbb8 sz 48
trap+67c (4003,4924,0,801ffdc0) ra 803639f4 sz 80
80363910+e4 (4003,4924,0,801ffdc0) ra 0 sz 0
User-level: curproc NULL
syncing disks... sonic: receive descriptors exhausted
sonic: receive buffers exhausted
6 6 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 giving up
Printing vnodes for busy buffers
tag 1 type VBLK, usecount 1485, writecount 0, refcount 267,
        tag VT_UFS, ino 6503, on dev 0, 0 flags 0x0, effnlink 1, nlink 1
        mode 060640, owner 0, group 0, size 0 not locked
20-8-16, 1:8:33
20-8-16, 1:8:30
System restart.