since you were saying in bug # 60711: "where they're assuming some behaviour of LinuxThreads rather than relying on POSIX semantics. The design of NPTL is radically different: All the internal opaque structures are different, it uses Thread Local Storage, etc. The only guarantees for compatability are those that POSIX offers."
all I'm doing to manipulate threads is done through these functions (on more than those instances, though): g_thread_init(NULL); pthread_attr_init(&recording_thread_attr); pthread_attr_setdetachstate(&recording_thread_attr, PTHREAD_CREATE_JOINABLE); pthread_cond_init(&recording_condition_unpaused, NULL); pthread_cond_wait(&recording_condition_unpaused, &recording_mutex); pthread_create(&recording_thread, &recording_thread_attr, (void *) do_record_thread, (void *) job); pthread_exit(NULL); pthread_join(recording_thread, NULL /* (void **) &status*/ ); pthread_mutex_init(&recording_mutex, NULL); pthread_mutex_lock(&recording_mutex); pthread_mutex_unlock(&recording_mutex); XInitThreads(); pthread_mutex_trylock(&mp) pthread_kill(tid, SIGUSR1); those don't constitue relying on opaque structures, do they?!? Also, wouldn't that break on a single CPU, too? -- get libpthread assertion on multicore CPU https://launchpad.net/bugs/60708 -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
