ocfs2: deal with wraparounds of i_nlink in ocfs2_rename()
authorAl Viro <viro@zeniv.linux.org.uk>
Mon, 13 Feb 2012 02:00:05 +0000 (21:00 -0500)
committerAl Viro <viro@zeniv.linux.org.uk>
Tue, 14 Feb 2012 01:45:39 +0000 (20:45 -0500)
unfortunately, nlink_t may be smaller than 32 bits and ->i_nlink
on ocfs2 can grow up to 0xffffffff; storing it in nlink_t variable
will lose upper bits on such architectures.  Needs to be made u32,
until we get kernel-side nlink_t uniformly 32bit...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
fs/ocfs2/namei.c

index be244692550db3e18cf05987ff577717e1f33173..a9856e3eaaf09753b4921d56ccdfb172db5cad7b 100644 (file)
@@ -1053,7 +1053,7 @@ static int ocfs2_rename(struct inode *old_dir,
        handle_t *handle = NULL;
        struct buffer_head *old_dir_bh = NULL;
        struct buffer_head *new_dir_bh = NULL;
-       nlink_t old_dir_nlink = old_dir->i_nlink;
+       u32 old_dir_nlink = old_dir->i_nlink;
        struct ocfs2_dinode *old_di;
        struct ocfs2_dir_lookup_result old_inode_dot_dot_res = { NULL, };
        struct ocfs2_dir_lookup_result target_lookup_res = { NULL, };