Bug in Bob Supnik's Emulator?

[Steven M. Schultz]

> Tim. (shoppa at triumf.ca)

> More evidence of a bug is that 'vi' doesn't work right under Bob
> Supnik's emulator, either.  At one point Steven Schultz made some
> private speculations to me about where the problem might be, but
> I've forgotten the details.  Is it possible that these two bugs
> are both due to FP emulation?  Does the 2.11BSD 'vi' even use
> the FP registers?

	To the best of my knowledge 'vi' does NOT use any FP at all (other than
	the usual 32 bit arithmetic that all programs do if they do any 'long'
	arithmetic).

	My speculation is that there's a MMU emulation bug somewhere.  'vi' is
	a overlaid split I/D program.  Overlays in 2.11BSD are done via
	'page flipping' (altering MMU registers).  Also 2.11 uses the 'expand
	downward' bit on the stack (as well as relying on MMR3 - i think that's
	the one - for instruction restart after growing the stack).  If there's
	a subtle gotcha in the MMU emulation that will cause problems 
	eventually.  2.11 is not alone in using the ED bit and instruction
	restart - if the problem is MMU related it could show up under other
	systems (V7).   It would be interesting to know if 'vi' encountered
	problems on V7 but V7 doesn't have usermode overlays so getting 'vi'
	to run would be very problematic.

	Steven


Received: (from major at localhost)
	by minnie.cs.adfa.oz.au (8.8.5/8.8.5) id KAA14110
	for pups-liszt; Fri, 3 Apr 1998 10:59:44 +1000 (EST)
X-Authentication-Warning: minnie.cs.adfa.oz.au: major set sender to owner-pups at minnie.cs.adfa.oz.au using -f