On 14/07/15 09:19, Jakub Jelinek wrote:
On Tue, Jul 14, 2015 at 09:09:01AM +0200, Sebastian Huber wrote:
I pasted the wrong version, since I wanted to check that I get a failure in
case generation % BAR_INCR != 0.
There is also config/posix/bar* implementation, which would need to be
checked too. I'd suggest first committing a patch with both sem and mutex
being destroyed (note, config/posix/bar* has two semaphores in it), and
perhaps if you feel strongly about it an incremental patch to avoid the
destroying/initialization of the barrier if it works even with
config/posix/bar*.
Jakub
If I destroy the barrier of the last team, and then initialize it later
in gomp_new_team()
+static inline struct gomp_team *
+get_last_team (unsigned nthreads)
+{
+ struct gomp_thread *thr = gomp_thread ();
+ if (thr->ts.team == NULL)
+ {
+ struct gomp_thread_pool *pool = thr->thread_pool;
+ if (pool != NULL)
+ {
+ struct gomp_team *last_team = pool->last_team;
+ if (last_team != NULL && last_team->nthreads == nthreads)
+ {
+ pool->last_team = NULL;
+ gomp_barrier_destroy (&last_team->barrier);
+ return last_team;
+ }
+ }
+ }
+ return NULL;
+}
then I get test suite failures. Is it safe to destroy the team barrier
here or do we circumvent the last team mechanism which is supposed to
delay the destruction?
Example failure:
WARNING: program timed out.
FAIL: libgomp.c/appendix-a/a.15.1.c execution test
--
Sebastian Huber, embedded brains GmbH
Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone : +49 89 189 47 41-16
Fax : +49 89 189 47 41-09
E-Mail : sebastian.hu...@embedded-brains.de
PGP : Public key available on request.
Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.