[ODE] BoundsChecker

Shaul Kedem shaul_kedem at yahoo.com
Thu Jul 3 13:13:02 2003


Do you get it after the same amount of time ?


--- DjArcas <djarcas@hotmail.com> wrote:
> I'm pretty sure the QNAN problem will make the
> collisions expand to fit
> available memory.
> 
> My stack is already 0x1200000, increased to see if
> that solved it. 0x800000
> exhibits the same issue.
> 
> ----- Original Message ----- 
> From: "Shaul Kedem" <shaul_kedem@yahoo.com>
> To: "DjArcas" <djarcas@hotmail.com>; <ode@q12.org>
> Sent: Thursday, July 03, 2003 9:07 PM
> Subject: Re: [ODE] BoundsChecker
> 
> 
> > DjArcas,
> >  If it's stack size - try to increase it and see
> what
> > happens (in the linker options - try something
> like
> > 0x800000)
> >
> > Shaul
> >
> > --- DjArcas <djarcas@hotmail.com> wrote:
> > > Nope nope and I think so... and I don't have
> *nix :)
> > >
> > > The only thing it's warning me about is
> literally
> > > thousands of 'Unitialized
> > > memory copy in swapProblem', inside SolveLCP.
> After
> > > a few dozen physics
> > > updates (all my objects have gone to QNAN
> already),
> > > I get a crash inside
> > > dSolveLCP, where it tries to allocate more
> memory:
> > >
> > > dReal *d = (dReal*) ALLOCA (n*sizeof(dReal));
> > >
> > > My guess is that the QNAN problem is causing all
> the
> > > objects to collide
> > > inside eachother, causing me to run out of
> stack.
> > >
> > >
> > >
> > > I've uploaded the purify report here:
> > >
> > > http://www.sawkins.com/crash.txt
> > >
> > >  but, I'll be honest, large swathes of it is
> > > gibberish to me!
> > >
> > > ----- Original Message ----- 
> > >
> > > From: "Shaul Kedem" <shaul_kedem@yahoo.com>
> > > To: "DjArcas" <djarcas@hotmail.com>;
> <ode@q12.org>
> > > Sent: Thursday, July 03, 2003 8:44 PM
> > > Subject: Re: [ODE] BoundsChecker
> > >
> > >
> > > > DjArcas,
> > > >  I used purify for a while (not the newest
> one) -
> > > >  - do you use release compile?
> > > >  - do you use a propriety memory manager?
> > > >  - does your version of purify/boundschecker
> > > support
> > > > heap allocation debugging?
> > > >
> > > >  Check all of the above, if you'll use msvcrt
> > > > malloc/new , use a debug compile and your
> > > > purify/bounds checker supports heap over
> bounds
> > > > checking (and is enabled) you should find your
> > > > problem.
> > > >
> > > >  If that doesn't help try compiling on unix
> using
> > > > electric fence or some other library that does
> > > memory
> > > > checking (there are a couple out there) - that
> is
> > > > assuming you can compile on a *nix machine.
> > > >
> > > > Hope this helps,
> > > > Shaul
> > > >
> > > > --- DjArcas <djarcas@hotmail.com> wrote:
> > > > > Does anyone have any experience with
> > > BoundsChecker
> > > > > or Rational Purify?
> > > > >
> > > > > I've tried running them both on my
> simulation,
> > > to no
> > > > > avail - they detect nothing relevant until I
> get
> > > an
> > > > > access violation inside StepIslands.
> > > > >
> > > > > Please, someone, help!
> > > > >
> > > >
> > > >
> > > > __________________________________
> > > > Do you Yahoo!?
> > > > SBC Yahoo! DSL - Now only $29.95 per month!
> > > > http://sbc.yahoo.com
> > > >
> >
> >
> > __________________________________
> > Do you Yahoo!?
> > SBC Yahoo! DSL - Now only $29.95 per month!
> > http://sbc.yahoo.com
> >


__________________________________
Do you Yahoo!?
SBC Yahoo! DSL - Now only $29.95 per month!
http://sbc.yahoo.com