libglusterfs/fd: fixed fd_anonymous() leak

fd was getting leaked whenever there was a overlapping operations,
which caused memory leak, and process fd leaks, which made most of
the operations on NFS mount of a replicate volume not work. With
the fix, things are back to normal.

Change-Id: I2d2158b2972ba5dae270d6ff7b1a827403653c04
Signed-off-by: Amar Tumballi <amarts@redhat.com>
BUG: 787368
Reviewed-on: http://review.gluster.com/2892
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Vijay Bellur <vijay@gluster.com>

1 files changed, 6 insertions, 2 deletions

diff --git a/libglusterfs/src/fd.c b/libglusterfs/src/fd.c
index 60add299d..46b75cb10 100644
--- a/libglusterfs/src/fd.c
+++ b/libglusterfs/src/fd.c
@@ -721,6 +721,10 @@ __fd_anonymous (inode_t *inode)
 
         fd = __fd_lookup (inode, (uint64_t)-1);
 
+        /* if (fd); then we already have increased the refcount in
+           __fd_lookup(), so no need of one more fd_ref().
+           if (!fd); then both create and bind wont bump up the ref
+           count, so we have to call fd_ref() after bind. */
         if (!fd) {
                 fd = __fd_create (inode, (uint64_t)-1);
 
@@ -728,9 +732,9 @@ __fd_anonymous (inode_t *inode)
                         return NULL;
 
                 __fd_bind (fd);
-        }
 
-        __fd_ref (fd);
+                __fd_ref (fd);
+        }
 
         return fd;
 }