ext4: Don't update ctime for non-extent-mapped inodes

The VFS handles updating ctime, so we don't need to update the inode's
ctime in ext4_splace_branch() to update the direct or indirect blocks.
This was harmless when we did this in ext3, but in ext4, thanks to
delayed allocation, updating the ctime in ext4_splice_branch() can
cause the ctime to mysteriously jump when the blocks are finally
allocated.

Thanks to Björn Steinbrink for pointing out this problem on the git
mailing list.

Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>

diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 8d0908afbd5bfaadb8c18fbe0383659b6d50d5e1..7c17ae275af4c0a0910bf5e7af7bc243199889de 100644 (file)
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -856,10 +856,6 @@ static int ext4_splice_branch(handle_t *handle, struct inode *inode,
        }
 
        /* We are done with atomic stuff, now do the rest of housekeeping */
-
-       inode->i_ctime = ext4_current_time(inode);
-       ext4_mark_inode_dirty(handle, inode);
-
        /* had we spliced it onto indirect block? */
        if (where->bh) {
                /*
@@ -878,8 +874,8 @@ static int ext4_splice_branch(handle_t *handle, struct inode *inode,
        } else {
                /*
                 * OK, we spliced it into the inode itself on a direct block.
-                * Inode was dirtied above.
                 */
+               ext4_mark_inode_dirty(handle, inode);
                jbd_debug(5, "splicing direct\n");
        }
        return err;