writeback: fix comment, use helper function
authorAndrew Morton <akpm@linux-foundation.org>
Wed, 17 Oct 2007 06:30:35 +0000 (23:30 -0700)
committerLinus Torvalds <torvalds@woody.linux-foundation.org>
Wed, 17 Oct 2007 15:43:02 +0000 (08:43 -0700)
There's a comment in there which claims that the inode is left on s_io
if nfs chickened out of writing some data.

But that's not been true for three years.
9290280ced13c85689adeffa587e9a53bd3a5873 fixed a livelock by moving these
inodes back onto s_dirty.  Fix the comment.

In the second leg of the `if', use redirty_tail() rather than open-coding it.

Add weaselly comment indicating lack of confidence in the code and lack of the
fortitude which would be needed to fiddle with it.

Cc: Mike Waychison <mikew@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/fs-writeback.c

index eb8dc1f227756995dcc928c18713b6bb12b744f9..a3d7a829137dc3739bbf81921da244fdc067f048 100644 (file)
@@ -201,7 +201,6 @@ __sync_single_inode(struct inode *inode, struct writeback_control *wbc)
 {
        unsigned dirty;
        struct address_space *mapping = inode->i_mapping;
-       struct super_block *sb = inode->i_sb;
        int wait = wbc->sync_mode == WB_SYNC_ALL;
        int ret;
 
@@ -237,7 +236,16 @@ __sync_single_inode(struct inode *inode, struct writeback_control *wbc)
                        /*
                         * We didn't write back all the pages.  nfs_writepages()
                         * sometimes bales out without doing anything. Redirty
-                        * the inode.  It is still on sb->s_io.
+                        * the inode.  It is moved from s_io onto s_dirty.
+                        */
+                       /*
+                        * akpm: if the caller was the kupdate function we put
+                        * this inode at the head of s_dirty so it gets first
+                        * consideration.  Otherwise, move it to the tail, for
+                        * the reasons described there.  I'm not really sure
+                        * how much sense this makes.  Presumably I had a good
+                        * reasons for doing it this way, and I'd rather not
+                        * muck with it at present.
                         */
                        if (wbc->for_kupdate) {
                                /*
@@ -257,8 +265,7 @@ __sync_single_inode(struct inode *inode, struct writeback_control *wbc)
                                 * all the other files.
                                 */
                                inode->i_state |= I_DIRTY_PAGES;
-                               inode->dirtied_when = jiffies;
-                               list_move(&inode->i_list, &sb->s_dirty);
+                               redirty_tail(inode);
                        }
                } else if (inode->i_state & I_DIRTY) {
                        /*