ocfs2: use OCFS2_INODE_SKIP_ORPHAN_DIR in ocfs2_mknod error path

Mark the inode with flag OCFS2_INODE_SKIP_ORPHAN_DIR in ocfs2_mknod, so we
can kill the inode in case of error.

[ Fixed up comment style -Mark ]

Signed-off-by: Li Dongyang <lidongyang@novell.com>
Signed-off-by: Mark Fasheh <mfasheh@suse.com>

diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
index b66e488558253e994ad21ec7bd747b134e97a29a..8ff035eabfd8d5a262f0e332ed2b0dbf1369417f 100644 (file)
--- a/fs/ocfs2/namei.c
+++ b/fs/ocfs2/namei.c
@@ -445,11 +445,6 @@ leave:
 
        ocfs2_free_dir_lookup_result(&lookup);
 
-       if ((status < 0) && inode) {
-               clear_nlink(inode);
-               iput(inode);
-       }
-
        if (inode_ac)
                ocfs2_free_alloc_context(inode_ac);
 
@@ -459,6 +454,17 @@ leave:
        if (meta_ac)
                ocfs2_free_alloc_context(meta_ac);
 
+       /*
+        * We should call iput after the i_mutex of the bitmap been
+        * unlocked in ocfs2_free_alloc_context, or the
+        * ocfs2_delete_inode will mutex_lock again.
+        */
+       if ((status < 0) && inode) {
+               OCFS2_I(inode)->ip_flags |= OCFS2_INODE_SKIP_ORPHAN_DIR;
+               clear_nlink(inode);
+               iput(inode);
+       }
+
        mlog_exit(status);
 
        return status;