>> with gdb the core shows that the exception was in the middle of a
>> vax CALLS instruction...
> hmm, calls is a weird thing to barf out on.

I just had a quick look at my VARM.  I can't see any way CALLS can
generate anything that turns into to SIGFPE.  The only exception
specifically listed for it is reserved operand, if the entry mask has
reserved bits set; and of course if the stack is trashed, there can be
memory errors while pushing stuff on the stack.

I note "in the middle of", which sounds as though perhaps PC points
partway through the instruction.  If so, then it could well be that
something else went wrong and it branched into the middle of the
instruction, got a "random" floating-point opcode, and fell over.

If that's not it, then the only other options that occur to me are bugs
in the cache code (if the VS4k60 has a cache), or the page-in code,
such that it tried to execute something other than what should have
been there, or of course hardware failures of various sorts (such as
the memory issues isildur speculated about).

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B