On Wed, 7 Apr 2010, Fredrik Kuivinen wrote:

> On Wed, Apr 7, 2010 at 17:08, Nicolas Pitre <nico@fluxnic.net> wrote:
> > On Wed, 7 Apr 2010, Fredrik Kuivinen wrote:
> >> diff --git a/thread-utils.c b/thread-utils.c
> >> index 4f9c829..3c8d817 100644
> >> --- a/thread-utils.c
> >> +++ b/thread-utils.c
> >> @@ -1,4 +1,5 @@
> >>  #include "cache.h"
> >> +#include <pthread.h>
> >
> > This will fail compilation on Windows surely?
> 
> I think it will work. We use "#include <pthread.h>" in builtin/grep.c,
> builtin/pack-objects.c, and preload-index.c already.

True indeed.

> >>  #if defined(hpux) || defined(__hpux) || defined(_hpux)
> >>  #  include <sys/pstat.h>
> >> @@ -43,3 +44,24 @@ int online_cpus(void)
> >>
> >>       return 1;
> >>  }
> >> +
> >> +int init_recursive_mutex(pthread_mutex_t *m)
> >> +{
> >> +#ifdef _WIN32
> >> +     /* The mutexes in the WIN32 pthreads emulation layer are
> >> +      * recursive, so we don't have to do anything extra here. */
> >> +     return pthread_mutex_init(m, NULL);
> >> +#else
> >> +     pthread_mutexattr_t a;
> >> +     int ret;
> >> +     if (pthread_mutexattr_init(&a))
> >> +             die("pthread_mutexattr_init failed: %s", strerror(errno));
> >> +
> >> +     if (pthread_mutexattr_settype(&a, PTHREAD_MUTEX_RECURSIVE))
> >> +             die("pthread_mutexattr_settype failed: %s", strerror(errno));
> >> +
> >> +     ret = pthread_mutex_init(m, &a);
> >> +     pthread_mutexattr_destroy(&a);
> >> +     return ret;
> >
> > Are you sure the pthread_mutexattr_t object can be destroyed even if the
> > mutex is still in use?  Is the attribute object "attached" to the mutex
> > or merely used as a template?
> 
> It is safe. See
> http://www.opengroup.org/onlinepubs/009695399/functions/pthread_mutexattr_init.html

OK.  ACK to your patch then.


Nicolas