[llvm-commits] CVS: llvm/lib/System/Mutex.cpp

[Chris Lattner]

On Jan 20, 2007, at 2:20 PM, Reid Spencer wrote:

> I think it would be more appropriate to add a test for
> pthread_mutexattr_setpshared in the configure script. On Linux,
> PTHREAD_PROCESS_PRIVATE is both an enumerator and a #define. If its  
> only
> an enumerator on some system, that check would be bad. Furthermore, a
> pthread.h could define the macro while not defining the function  
> (which
> would be a bug, but it could happen).  The check for the function  
> is the
> safest.

It is also a macro on macos.  I believe that it is standard for these  
things to be macros to allow #ifdef conditionals.  Why don't we just  
do this for now, and switch to a full autoconf test if we find that  
to be insufficient?

-Chris

> Reid.
>
> Perhaps, but what if PTHREAD_PROCESS_PRIVATE is defined in a header  
> file
> and
> On Sat, 2007-01-20 at 14:13 -0800, Chris Lattner wrote:
>>> For PR808: http://llvm.org/PR808 :
>>> NetBSD also doesn't have pthread_mutexattr_setpshared
>>
>>>
>>> -#if !defined(__FreeBSD__) && !defined(__OpenBSD__)
>>> +#if !defined(__FreeBSD__) && !defined(__OpenBSD__) && !defined
>>> (__NetBSD__)
>>>      // Make it a process local mutex
>>>      errorcode = pthread_mutexattr_setpshared(&attr,
>>> PTHREAD_PROCESS_PRIVATE);
>>>  #endif
>>
>> Can't we just change this to:
>>
>> #ifdef PTHREAD_PROCESS_PRIVATE
>>
>> ?
>>
>> -Chris
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>