]> git-server-git.apps.pok.os.sepia.ceph.com Git - ceph.git/commit
common: do not unlock rwlock on destruction 4076/head
authorFederico Simoncelli <fsimonce@redhat.com>
Sat, 15 Nov 2014 14:14:04 +0000 (14:14 +0000)
committerLoic Dachary <ldachary@redhat.com>
Wed, 18 Mar 2015 22:36:47 +0000 (23:36 +0100)
commit499d94f573e78f6545bd5a12ade6b5ba5a75bbca
tree6945e38e490775895d47a5b4b5a2d838aabb40f4
parent1f58a0adc3b23785fac00083d721b62f6a4c44a1
common: do not unlock rwlock on destruction

According to pthread_rwlock_unlock(3p):

 Results are undefined if the read-write lock rwlock is not held
 by the calling thread.

and:

 https://sourceware.org/bugzilla/show_bug.cgi?id=17561

 Calling pthread_rwlock_unlock on an rwlock which is not locked
 is undefined.

calling pthread_rwlock_unlock on RWLock destruction could cause
an unknown behavior for two reasons:

- the lock is acquired by another thread (undefined)
- the lock is not acquired (undefined)

Moreover since glibc-2.20 calling pthread_rwlock_unlock on a
rwlock that is not locked results in a SIGILL that kills the
application.

This patch removes the pthread_rwlock_unlock call on destruction
and replaces it with an assertion to check that the RWLock is
not in use.

Any code that relied on the implicit release is now going to
break the assertion, e.g.:

 {
   RWLock l;
   l.get(for_write);
 } // implicit release, wrong.

Signed-off-by: Federico Simoncelli <fsimonce@redhat.com>
(cherry picked from commit cf2104d4d991361c53f6e2fea93b69de10cd654b)
src/common/RWLock.h