autofs-5.1.1 - fix rwlock unlock crash
From: Ian Kent <raven@themaw.net>
It should be the case that the only way that automount can exit
the kernel communication read loop is after reading a packet and
performing shutdown operations.
However there are reports of pthread_rwlock_unlock() being called
from the exit cleanup function when the lock it is trying to unlock
is not held.
But the call to the cleanup function is outside the communication
loop so call it from each of the loop break points instead so that
the expected locks must be held.
Signed-off-by: Ian Kent <raven@themaw.net>
---
CHANGELOG | 1 +
daemon/automount.c | 7 ++++---
2 files changed, 5 insertions(+), 3 deletions(-)
diff --git a/CHANGELOG b/CHANGELOG
index fa8231c..3bd7145 100644
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -10,6 +10,7 @@
- make find_dc_server() return a status.
- make find_server() return a status.
- fix return handling of do_reconnect() in ldap module.
+- fix rwlock unlock crash.
21/04/2015 autofs-5.1.1
=======================
diff --git a/daemon/automount.c b/daemon/automount.c
index c81b4af..6cba3db 100644
--- a/daemon/automount.c
+++ b/daemon/automount.c
@@ -1757,6 +1757,7 @@ void *handle_mounts(void *arg)
*/
if (ap->type == LKP_DIRECT) {
umount_autofs(ap, NULL, 1);
+ handle_mounts_cleanup(ap);
break;
}
@@ -1767,8 +1768,10 @@ void *handle_mounts(void *arg)
* occurs while we're trying to umount.
*/
ret = umount_autofs(ap, NULL, 1);
- if (!ret)
+ if (!ret) {
+ handle_mounts_cleanup(ap);
break;
+ }
/* Failed shutdown returns to ready */
warn(ap->logopt,
@@ -1789,8 +1792,6 @@ void *handle_mounts(void *arg)
}
}
- handle_mounts_cleanup(ap);
-
return NULL;
}