On Sat, 2005-06-11 at 02:37 +1000, Adam Conrad wrote: > Lon Hohberger wrote: > > > > Similarly, the code is written to rely on the documented behavior. So, > > when it's not available, maybe the wrapper should return > > (pid_t)pthread_self() instead of -ENOSYS? > > I threw -ENOSYS in mostly as a sanity check. If the suite is meant to be > portable to other POSIX systems, you'll certainly want to do something > else if gettid() doesn't exist, since it's linux-specific. Well, linux-cluster is, um, rather linux-specific. ;) Ok, so, I'll add a #warn and return pthread_self() for now - since it also doesn't fail and returns a thread-specific identifier (which is the primary reason we used gettid()...). > The only aim > of the patch was to shoot for portability accross all Linux platforms > (many of which don't have _syscall magic macros, and word on the street is > that we may not have them on any arches in the future). That can only be a good thing. -- Lon -- Linux-cluster@xxxxxxxxxx http://www.redhat.com/mailman/listinfo/linux-cluster
