[llvm-commits] CVS: llvm-test/MultiSource/Benchmarks/Olden/voronoi/newvor.c

[Chris Lattner]

On Apr 26, 2007, at 2:42 PM, Andrew Lenharth wrote:

> I need it because it isn't defined by stdlib.h on linux.  I need it
> because it is defaulting to:
> int memalign(...), lacking any other definition.  It happens to work
> in most usage this way, but it isn't correct.

Easy solution.  In that source file, unconditionally define  
MEMALIGN_IS_NOT_AVAILABLE

-Chris

> Andrew
>
> On 4/26/07, Jeff Cohen <jeffc at jolt-lang.org> wrote:
>> Chris Lattner wrote:
>> >
>> > On Apr 26, 2007, at 2:23 PM, Jeff Cohen wrote:
>> >
>> >> Chris Lattner wrote:
>> >>>
>> >>> On Apr 26, 2007, at 1:11 PM, Jeff Cohen wrote:
>> >>>
>> >>>> Absolutely true.  This change will cause the test to fail on  
>> the BSDs,
>> >>>> which will complain that malloc.h has been replaced by stdlib.h.
>> >>>> Please
>> >>>> use a different test.
>> >>>
>> >>> Better yet, just unconditionally #include <stdlib.h> and  
>> eliminate
>> >>> <malloc.h>
>> >>>
>> >>> -Chris
>> >>
>> >> It already does include stdlib.h.  I assume it isn't enough on his
>> >> system.
>> >
>> > Andrew, can alpha drop the malloc.h #include?  If not, why not?
>> >
>> > We should remove all instances of malloc.h from llvm-test.
>> >
>> > -Chris
>>
>> A more interesting question is why he needed to add it for  
>> x86_64.  If
>> Linux doesn't need it for x86, it shouldn't need it for x86_64  
>> either.
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>