tile: use atomic op to unlock pthread_spinlock_t

Atomic ops are issued directly from the core, rather than
potentially sitting in the write buffer, so can improve the
performance of other waiters.  In addition, if we didn't end
up pulling a copy of the cache line where the lock is into cache,
by using an atomic op we don't have to acquire the cache line
before we can unlock.

diff --git a/ports/ChangeLog.tile b/ports/ChangeLog.tile
index 99e95db..06f24c2 100644 (file)
--- a/ports/ChangeLog.tile
+++ b/ports/ChangeLog.tile
@@ -1,3 +1,7 @@
+2012-11-06  Chris Metcalf  <cmetcalf@tilera.com>
+
+       * sysdeps/tile/nptl/pthread_spin_unlock.c: New file.
+
 2012-11-05  Chris Metcalf  <cmetcalf@tilera.com>
 
        * sysdeps/tile/math_private.h: Provide additional no-op defines

diff --git a/ports/sysdeps/tile/nptl/pthread_spin_unlock.c b/ports/sysdeps/tile/nptl/pthread_spin_unlock.c
new file mode 100644 (file)
index 0000000..260f6fa
--- /dev/null
+++ b/ports/sysdeps/tile/nptl/pthread_spin_unlock.c
@@ -0,0 +1,33 @@
+/* pthread_spin_unlock -- unlock a spin lock.  Tile version.
+   Copyright (C) 2012 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library.  If not, see
+   <http://www.gnu.org/licenses/>.  */
+
+#include "pthreadP.h"
+#include <atomic.h>
+
+int
+pthread_spin_unlock (pthread_spinlock_t *lock)
+{
+#ifdef __tilegx__
+  /* Use exchange() to bypass the write buffer. */
+  atomic_exchange_rel (lock, 0);
+#else
+  atomic_full_barrier ();
+  *lock = 0;
+#endif
+  return 0;
+}