Re: [Plib-users] problem with fnt and release mode

[Scott S. <sjs...@um...>]

Well, it only causes problems across functions, i.e.:


int* g_pTheInt; // global variable.

void foo()
{
  *g_pTheInt = *g_pTheInt + 1;
}


and later...


int* a;

*a = 10;

g_pTheInt = a;

foo();

printf("%d", *a); // a should now be 11, but this may not be the case

You can turn set the VC++ compiler optimization to "assume aliasing across
function calls", which will fix the problem (effectively reloading the
registers from memory, I assume).


------------------
Scott Shumaker
sjs...@um...



----- Original Message -----
From: "Steve Baker" <sjb...@ai...>
To: <pli...@li...>
Sent: Wednesday, December 20, 2000 9:02 PM
Subject: Re: [Plib-users] problem with fnt and release mode


> Paul Bleisch wrote:
> >
> > More than likely it is either "global optimizations" or "assume
> > no aliasing".  These two cause the most problems.
>
> Yikes!
>
> Does "assume no aliasing" mean "assume no pointer aliasing" ??
>
> If so, then you'll want to turn off that sucker - bury it *DEEP*!
>
> int x = 6 ;
> int y ;
>
> int *a ;
> int *b ;
>
>   a = & x ;
>   b = & x ;
>
> ...then later on...
>
>   *a = 1234 ;
>   *b = 0 ;
>   y = *a ;
>   printf ( "%d\n", y ) ;
>
> ...will probably print 1234 with a compiler that assumes there
> is no pointer aliasing.
>
> It's hard to imagine any piece of software of any complexity
> surviving that kind of punishment.
>
> --
> Steve Baker   HomeEmail: <sjb...@ai...>
>               WorkEmail: <sj...@li...>
>               HomePage : http://web2.airmail.net/sjbaker1
>               Projects : http://plib.sourceforge.net
>                          http://tuxaqfh.sourceforge.net
>                          http://tuxkart.sourceforge.net
>                          http://prettypoly.sourceforge.net
>
>
>
> _______________________________________________
> plib-users mailing list
> pli...@li...
> http://lists.sourceforge.net/mailman/listinfo/plib-users