I've got a weird problem with NetBSD/sparc, and I suspect something
weird in the VM subsystem.

I (co-)run a mud on one of my home machines.  Over the last couple of
days, I started trying to move this to sparkle, a SS1+ running
NetBSD/sparc.

I started getting weird crashes: according to gdb, pointers were taking
on impossible values, and never quite the same crash twice.  My first
thought was, flaky RAM: I'd just recently pulled 4x1M and replaced it
with new 4x4M.  Moving the RAM to another machine (a 386 that has
memory tests capable of identifying both of my known-bad SIMMs) turned
up no problems.

Then I tried it on another SS1+ running the same OS (built from the
very same source tree), whose hardware hasn't been touched in ages.
Very similar crash.

Then I tried it on a SS2 running NetBSD 1.2 (still at 1.2 because, as I
mentioned in another thread, more modern kernels can't seem to play CDs
in its CD drive).  Worked like a charm.

Thus, I conjecture there's something softwary wrong, probably with the
VM subsystem.  The source tree does have a number of patches, but none
to UVM.  (I have two patches in vm/, but - I just checked - nothing in
there is getting used; I am using uvm/.)

Any way I can help debug this?  If necessary I can probably even set up
serial-console access to one of the affected machines, if pk or someone
wants to ssh in to one of my machines for it, set it up with a
crash-and-burn disk that I don't mind wiping later....

					der Mouse

			       mouse@rodents.montreal.qc.ca
		     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B