I just installed a new motherboard, which appears to work more reliably than
my old motherboard (for example, the cache no longer fails to snoop DMA
transfers), but it appears that something is causing my system to crash when
I try to start up X windows.  However, it crashes in vm_map_lookup_entry
(fatal page fault), which isn't a routine I would have thought overly
sensitive to marginal hardware, certainly not more than any other routine).
I *think* the display hardware is OK, though I have some doubts -- thanks to
the sheet metal disagreeing with the new motherboard even more than it did with
the original, the display card is impossible to seat exactly right; however,
when I fire up DOS, it works perfectly fine, and not even DOS will ignore
bus errors.  I think.  (Hmm, however, the DOS program might not map the display
card into memory, unlike the X driver.)

Does anyone have any clues for how to chase this down more precisely?  Is it
likely to be a problem that the display card is not in the bottom VLB slot?

The MB ran quite stably all day with a busy workload, without corrupting the
disks or any of the other antisocial behaviors I've come to expect from sad
motherboards, but X is currently the kiss of death.  (The MB is a "G486USP"
manufactured by Lord-only-knows-who, with a UMC82C491 chipset; 3 VLB slots;
AMD DX2-66 processor; 16MB of RAM.