>Number:         2511
>Category:       port-i386
>Synopsis:       system crash with vm_fault
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    gnats-admin (GNATS administrator)
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Jun  3 23:50:01 1996
>Last-Modified:
>Originator:     Michael Richardson
>Organization:
 Home: mcr@sandelman.ocunix.on.ca. PGP key available.
>Release:        NetBSD-1.1 release
>Environment:
System: NetBSD l.ox.ottawa.on.ca 1.1 NetBSD 1.1 (BUD) #0
Architecture: i386

>Description:
	During intensive disk activity, system crashes with (transcribed)
	vm_fault(0xf8759000,0,1,0) -> 1
	kernel: page fault trap, code=0
	Stopped at	ffs_sync+0x1a: cmpb  $0,0xd0(%eax)

	This hard disk and AHA controller was in a 386/33 before. The
system got dupped by putting an IDE controller and drive into the
386 and dump/restore'ing the SCSI drive. The SCSI drive was then
moved to a 486DX/2-50. I have turned all the caches off in hope of getting
something to stay up long enough to copy previously sup'ed sources
(from my NetBSD/sun3 system) to this machine. The 386, which is
bud.sandelman.ocunix.on.ca is alive and well with the same kernel,
and ran fine for months with that drive/controller. It is an AMI
bios (both of them, but different years). 
  I'm happy to try a 1.1B (-current) kernel, only the system doesn't
stay up long enough to compile one. I could build one on my 386, but
it is pathetically slow... (Perry, got one you can build me? I prefer
we0 at 0x300,IRQ 10)
  P.S: 'call 0' results in a sync on NetBSD/sun3, and of course there
is "machine halt" and "machine reboot" instead of the reset button.
Is there a nicer way to get out of DDB?


>How-To-Repeat:
	Heavy disk activity, e.g. compiling kernel.
>Fix:
>Audit-Trail:
>Unformatted: