hurd: fix sigstate locking
authorSamuel Thibault <samuel.thibault@ens-lyon.org>
Sat, 7 Feb 2015 21:16:19 +0000 (22:16 +0100)
committerSamuel Thibault <samuel.thibault@ens-lyon.org>
Sat, 7 Feb 2015 21:16:19 +0000 (22:16 +0100)
It looks like _hurd_thread_sigstate used to return with the sigstate
lock held long ago, but since that's no longer the case, don't unlock
something that isn't locked.

Note that it's unlikely this change fixes anything in practice since
its current implementation (on i386) makes this call a nop.

ChangeLog
hurd/hurd/signal.h

index 5081237..97849df 100644 (file)
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,8 @@
+2015-02-07  Richard Braun  <rbraun@sceen.net>
+
+       * hurd/hurd/signal.h (_hurd_critical_section_lock): Don't unlock
+       sigstate.
+
 2015-02-07  Samuel Thibault  <samuel.thibault@ens-lyon.org>
 
        * sysdeps/pthread/aio_misc.c [!AIO_PRIO_DELTA_MAX]: Do not check
index 4a744c1..e332242 100644 (file)
@@ -175,9 +175,8 @@ _hurd_critical_section_lock (void)
       /* The thread variable is unset; this must be the first time we've
         asked for it.  In this case, the critical section flag cannot
         possible already be set.  Look up our sigstate structure the slow
-        way; this locks the sigstate lock.  */
+        way.  */
       ss = *location = _hurd_thread_sigstate (__mach_thread_self ());
-      __spin_unlock (&ss->lock);
     }
 
   if (! __spin_try_lock (&ss->critical_section_lock))