about summary refs log tree commit diff
path: root/nptl/pthread_rwlock_trywrlock.c
diff options
context:
space:
mode:
authorCarlos O'Donell <carlos@redhat.com>2019-01-21 22:50:12 -0500
committerCarlos O'Donell <carlos@redhat.com>2019-01-31 20:47:55 -0500
commit5fc9ed4c4058bfbdf51ad6e7aac7d209b580e8c4 (patch)
tree5787bb93e87e93b3191c1d1d8e1bf61f5a99d3c2 /nptl/pthread_rwlock_trywrlock.c
parent932329a5134665bfad601a8709da2602d943fc38 (diff)
downloadglibc-5fc9ed4c4058bfbdf51ad6e7aac7d209b580e8c4.tar.gz
glibc-5fc9ed4c4058bfbdf51ad6e7aac7d209b580e8c4.tar.xz
glibc-5fc9ed4c4058bfbdf51ad6e7aac7d209b580e8c4.zip
nptl: Fix pthread_rwlock_try*lock stalls (Bug 23844)
For a full analysis of both the pthread_rwlock_tryrdlock() stall
and the pthread_rwlock_trywrlock() stall see:
https://sourceware.org/bugzilla/show_bug.cgi?id=23844#c14

In the pthread_rwlock_trydlock() function we fail to inspect for
PTHREAD_RWLOCK_FUTEX_USED in __wrphase_futex and wake the waiting
readers.

In the pthread_rwlock_trywrlock() function we write 1 to
__wrphase_futex and loose the setting of the PTHREAD_RWLOCK_FUTEX_USED
bit, again failing to wake waiting readers during unlock.

The fix in the case of pthread_rwlock_trydlock() is to check for
PTHREAD_RWLOCK_FUTEX_USED and wake the readers.

The fix in the case of pthread_rwlock_trywrlock() is to only write
1 to __wrphase_futex if we installed the write phase, since all other
readers would be spinning waiting for this step.

We add two new tests, one exercises the stall for
pthread_rwlock_trywrlock() which is easy to exercise, and one exercises
the stall for pthread_rwlock_trydlock() which is harder to exercise.

The pthread_rwlock_trywrlock() test fails consistently without the fix,
and passes after. The pthread_rwlock_tryrdlock() test fails roughly
5-10% of the time without the fix, and passes all the time after.

Signed-off-by: Carlos O'Donell <carlos@redhat.com>
Signed-off-by: Torvald Riegel <triegel@redhat.com>
Signed-off-by: Rik Prohaska <prohaska7@gmail.com>
Co-authored-by: Torvald Riegel <triegel@redhat.com>
Co-authored-by: Rik Prohaska <prohaska7@gmail.com>
Diffstat (limited to 'nptl/pthread_rwlock_trywrlock.c')
-rw-r--r--nptl/pthread_rwlock_trywrlock.c9
1 files changed, 8 insertions, 1 deletions
diff --git a/nptl/pthread_rwlock_trywrlock.c b/nptl/pthread_rwlock_trywrlock.c
index fd37a71ce4..fae475cc70 100644
--- a/nptl/pthread_rwlock_trywrlock.c
+++ b/nptl/pthread_rwlock_trywrlock.c
@@ -46,8 +46,15 @@ __pthread_rwlock_trywrlock (pthread_rwlock_t *rwlock)
 	  &rwlock->__data.__readers, &r,
 	  r | PTHREAD_RWLOCK_WRPHASE | PTHREAD_RWLOCK_WRLOCKED))
 	{
+	  /* We have become the primary writer and we cannot have shared
+	     the PTHREAD_RWLOCK_FUTEX_USED flag with someone else, so we
+	     can simply enable blocking (see full wrlock code).  */
 	  atomic_store_relaxed (&rwlock->__data.__writers_futex, 1);
-	  atomic_store_relaxed (&rwlock->__data.__wrphase_futex, 1);
+	  /* If we started a write phase, we need to enable readers to
+	     wait.  If we did not, we must not change it because other threads
+	     may have set the PTHREAD_RWLOCK_FUTEX_USED in the meantime.  */
+	  if ((r & PTHREAD_RWLOCK_WRPHASE) == 0)
+	    atomic_store_relaxed (&rwlock->__data.__wrphase_futex, 1);
 	  atomic_store_relaxed (&rwlock->__data.__cur_writer,
 	      THREAD_GETMEM (THREAD_SELF, tid));
 	  return 0;