Re: XL1200 Blasts thru FEP

[Kaufman@RIVERSIDE.SCRC.Symbolics.COM (David H. Kaufman)]

    Date: Mon, 7 Jan 1991 11:42 EST
    From: AI.gadbois@MCC.COM (David Gadbois)

	Date: Fri, 4 Jan 91 13:40 CST
	From: ian@iexist.att.com

	I'm wondering if other XL1200 (or maybe Ivory in general) users are
	experiencing the phenomenon where the machine errors and blasts all
	the way thru the FEP, just as though the power had been cycled. The
	problem seems to be reproducable and related to some my code.
	Several (buggy) programs have manifested this behavior. It always
	happens in the same place (as far as I can tell given the effect). I
	may well be blowing the stack, but this sort of thing used to plop
	you in the debugger, where at least you had a clue as to where you
	were and what might be the problem.  I get no warning, just CRASH!

    I have seen it on an XL400 under 8.0.1.  It is caused (reproducibly) by
    passing a string argument to SI:HALT, as in

	(SI:HALT "Continue
    ")

This is a different bug from the one that Ian Bruce reported.  (This bug
occurs on all Ivory machines running certain releases of the IFEP - the
particular versions escape me at the moment - whereas Ian's bug only
occurs on XL1200s.)  This bug will be fixed in 8.1; as I mentioned in
other mail, Ian's bug is fixed in 8.0ECO#1.

    Such forms are used by Rich Cohen's World Maker code.  The crash is
    pretty spectacular.  Apparently, Lisp is passing a bogus string pointer
    to the Ivory FEP.  This used to work on the Ivories and still does on
    the 36xxs.  I have not heard back from customer service yet on their
    progress in fixing this annoying bug.

    --David Gadbois