Mail Archives: djgpp/2001/06/19/11:00:13
Graaagh the Mighty <invalid AT erehwon DOT invalid> wrote:
> On Tue, 19 Jun 2001 10:04:01 +0200, Tim Van Holder
> <tim DOT vanholder AT falconsoft DOT be> sat on a tribble, which squeaked:
>>This is all very interesting, but it would be a LOT more informative
>>and helpful if you posted the code that resulted in the broken
>>behaviour.
> It's about a thousand lines long and the bug is *very* unstable --
> almost any change whatsoever stops it reproducing.
A wild pointer, almost certainly. Such "Heisenbugs" are typical of
accessing uninitialized memory, esp. for writing through a pointer
that wasn't properly initialized to point to somewhere sensible,
before. Each time you recompile or even re-run the program, the bug
may toggle into and out of existance.
> I won't have a problem once I can get a decent traceback. Why is it
> not generating the "call frame traceback EIPs"? And why does it crash
> Windows? A protected mode task should be utterly unable to bring the
> OS down,
"Should": yes. But reality shows that "should" rules are far too
seldomly paid any respect. And 9x is a rather prominent example of
that, regrettably.
Protected mode means that the CPU offers hardware assistance that an
OS can use to enforce proper segregation between independant tasks.
But by no means does that imply that the OS *has* to do that.
--
Hans-Bernhard Broeker (broeker AT physik DOT rwth-aachen DOT de)
Even if all the snow were burnt, ashes would remain.
- Raw text -