In message <200504010317.j313Hqbm020644@guild.plethora.net>, Peter Seebach writ
es:
>I have a machine which is quite reliably crashing after being up for a while.
>The crashes take a fairly consistent time to manifest.  The motherboard is a
>SuperMicro X5DA8, which is a dual-xeon board.  I'm not running in SMP; I can't
>even finish building a kernel!
>
>Backtraces always look a little weird, involving memory.

Just a followup:  I can run Slackware on this machine for hours and hours,
beating on it heavily, with no problems.  I built a 2.0.2 kernel on another
box and it still acts up.

kernel: protection fault trap, code=0
Stopped in pid 43.1 (tar) at    netbsd:pool_prime_page+0x136:   movl    %ebx,0(eax)
db{7}> bt
pool_prime_page()
pool_get()
ffs_vget()
ffs_valloc()
ufs_makeinode()
ufs_create()
vn_open()
sys_open()
syscall_plain()

db{7}> sync
syncing disks... panic: TLB IPI rendezvous failed (mask 43)

... Any suggestions?  I am sort of assuming that the machine's ironclad
stability under Linux indicates functional hardware.

And no, the idea of a bug in NetBSD is not an April Fools' joke.

-s