Avoid potential signal loss before sigsuspend in suspend_handler if TSan

Issue #236 (bdwgc).

As comment between sem_post() and sigsuspend() says GC_sig_thr_restart
signal should be masked at that point otherwise there could be a race.
Thus, this commit removes pthread_sigmask(SIG_UNBLOCK) call before
sem_post() one in GC_suspend_handler_inner.

* pthread_stop_world.c [!GC_OPENBSD_UTHREADS && !NACL
&& THREAD_SANITIZER] (GC_suspend_handler_inner): Remove "set" local
variable; do not call sigemptyset(), pthread_sigmask(SIG_UNBLOCK),
sigaddset().

pthread_stop_world.c

@@ -392,20 +392,6 @@ STATIC void GC_suspend_handler_inner(ptr_t dummy GC_ATTR_UNUSED,
     me -> backing_store_ptr = bs_lo + stack_size;
 # endif
 
-# ifdef THREAD_SANITIZER
-    /* TSan disables signals around signal handlers.  Without   */
-    /* a pthread_sigmask call, sigsuspend may block forever.    */
-    {
-      sigset_t set;
-      sigemptyset(&set);
-      GC_ASSERT(GC_sig_suspend != SIGNAL_UNSET);
-      GC_ASSERT(GC_sig_thr_restart != SIGNAL_UNSET);
-      sigaddset(&set, GC_sig_suspend);
-      sigaddset(&set, GC_sig_thr_restart);
-      if (pthread_sigmask(SIG_UNBLOCK, &set, NULL) != 0)
-        ABORT("pthread_sigmask failed in suspend handler");
-    }
-# endif
   /* Tell the thread that wants to stop the world that this     */
   /* thread has been stopped.  Note that sem_post() is          */
   /* the only async-signal-safe primitive in LinuxThreads.      */