exception handling predicament
David Miller
davem@davemloft.net
Sun Aug 21 21:36:00 GMT 2011
From: Jakub Jelinek <jakub@redhat.com>
Date: Fri, 19 Aug 2011 17:11:55 +0200
> If aio_suspend.c is the only spot where a cancellation point happens in the
> same function as corresponding cleanups, then IMHO the best fix is just to
> make sure it is in a different function (with noinline attribute, not marked
> as nothrow), which will be the cancellation point.
I've verified the following fixes the test case failures reliably.
Ok to commit?
2011-08-20 David S. Miller <davem@firepit.lan>
* sysdeps/pthread/aio_suspend.c (do_aio_misc_wait): New function.
(aio_suspend): Call it to force an exception region around the
AIO_MISC_WAIT() invocation.
diff --git a/sysdeps/pthread/aio_suspend.c b/sysdeps/pthread/aio_suspend.c
index b85b16d..99f3a80 100644
--- a/sysdeps/pthread/aio_suspend.c
+++ b/sysdeps/pthread/aio_suspend.c
@@ -92,6 +92,18 @@ cleanup (void *arg)
pthread_mutex_unlock (&__aio_requests_mutex);
}
+#ifdef DONT_NEED_AIO_MISC_COND
+static int
+__attribute__ ((noinline))
+do_aio_misc_wait(int *cntr, const struct timespec *timeout)
+{
+ int result = 0;
+
+ AIO_MISC_WAIT(result, *cntr, timeout, 1);
+
+ return result;
+}
+#endif
int
aio_suspend (list, nent, timeout)
@@ -169,7 +181,7 @@ aio_suspend (list, nent, timeout)
pthread_cleanup_push (cleanup, &clparam);
#ifdef DONT_NEED_AIO_MISC_COND
- AIO_MISC_WAIT (result, cntr, timeout, 1);
+ result = do_aio_misc_wait(&cntr, timeout);
#else
if (timeout == NULL)
result = pthread_cond_wait (&cond, &__aio_requests_mutex);
More information about the Libc-alpha
mailing list