[COMMITTED PATCH] NaCl: Add NaCl-specific __lll_timedlock_wait.

[COMMITTED PATCH] NaCl: Add NaCl-specific __lll_timedlock_wait.

[Roland McGrath]

	* sysdeps/nacl/lll_timedlock_wait.c: New file.

diff --git a/sysdeps/nacl/lll_timedlock_wait.c b/sysdeps/nacl/lll_timedlock_wait.c
new file mode 100644
index 0000000..df951ee
--- /dev/null
+++ b/sysdeps/nacl/lll_timedlock_wait.c
@@ -0,0 +1,52 @@
+/* Timed low level locking for pthread library.  NaCl version.
+   Copyright (C) 2015 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 <assert.h>
+#include <atomic.h>
+#include <errno.h>
+#include <lowlevellock.h>
+#include <sys/time.h>
+
+
+/* This behaves the same as the generic version in nptl/.  It's simpler
+   because it doesn't need to convert an absolute timeout to a relative
+   one (and back again in the lll_futex_timed_wait macro).  */
+
+int
+__lll_timedlock_wait (int *futex, const struct timespec *abstime, int private)
+{
+  /* Reject invalid timeouts.  */
+  if (__glibc_unlikely (abstime->tv_nsec < 0)
+      || __glibc_unlikely (abstime->tv_nsec >= 1000000000))
+    return EINVAL;
+
+  /* Try locking.  */
+  int result = 0;
+  while (atomic_exchange_acq (futex, 2) != 0)
+    {
+      /* If *futex == 2, wait until woken or timeout.  */
+      result = __nacl_irt_futex.futex_wait_abs ((volatile int *) futex, 2,
+						abstime);
+      if (__glibc_likely (result == 0)
+	  || __glibc_likely (result == ETIMEDOUT))
+	break;
+      assert (result == EAGAIN);
+    }
+
+  return result;
+}

Re: [COMMITTED PATCH] NaCl: Add NaCl-specific __lll_timedlock_wait.

[Torvald Riegel]

On Tue, 2015-05-26 at 15:31 -0700, Roland McGrath wrote:
> +/* This behaves the same as the generic version in nptl/.  It's simpler
> +   because it doesn't need to convert an absolute timeout to a relative
> +   one (and back again in the lll_futex_timed_wait macro).  */

If you remember the new glibc-internal futex API I proposed, it provided
variants for both absolute and relative timeouts.  Would now be a good
time to pursue adding this API so that changes like this wouldn't need
to be NaCl-specific?

Re: [COMMITTED PATCH] NaCl: Add NaCl-specific __lll_timedlock_wait.

[Roland McGrath]

> On Tue, 2015-05-26 at 15:31 -0700, Roland McGrath wrote:
> > +/* This behaves the same as the generic version in nptl/.  It's simpler
> > +   because it doesn't need to convert an absolute timeout to a relative
> > +   one (and back again in the lll_futex_timed_wait macro).  */
> 
> If you remember the new glibc-internal futex API I proposed, it provided
> variants for both absolute and relative timeouts.  Would now be a good
> time to pursue adding this API so that changes like this wouldn't need
> to be NaCl-specific?

Certainly!  As you might surmise, I'm going through old email backlog now.
I hope to get caught up by the end of this week, and then be in a position
to interact helpfully on new reviews.  (I've seen your new futex API patch
but not read it yet, and will get through all the old backlog first.)


Thanks,
Roland