[ Umm, I guess I should have explicitly stated that you should also run
the copied-down oops through ksymoops with the proper System.map file
for the kernel that oops'es - it's rather hard for me to see what is
actually going on otherwise ]
> -------------------------------------------------------------------------
> ... at vrituall address c0000001
Ok, this could definitely be a bug that is masked by the Pentium 4MB page
tables - I'll disable the 4M page tables in the 2.1.x tree (it _does_ show up
as a decrease in performance, but you have to know where to look in otder to
see it ;)
The only thing I can see from the panic-report without any symbolic
addresses is that it disassembles into
decl (%eax)
movb %dl,%dh
inl $0x0,%eax
addb %cl,0xf0b7(%ebx)
addb %al,-123(%ebx)
The first instruction looks fine (it would indeed result in a panic like the
one you have), but the rest looks dubious. Hard to tell - even a _small_
error in transcripting the "code" bytes can result in total garbage.
Linus