xfs: remove the unlock argument to xfs_buf_delwri_queue
authorChristoph Hellwig <hch@infradead.org>
Tue, 23 Aug 2011 08:28:04 +0000 (08:28 +0000)
committerAlex Elder <aelder@sgi.com>
Wed, 12 Oct 2011 02:14:59 +0000 (21:14 -0500)
We can just unlock the buffer in the caller, and the decrement of b_hold
would also be needed in the !unlock, we just never hit that case currently
given that the caller handles that case.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
fs/xfs/xfs_buf.c

index 2e71a26..04689db 100644 (file)
@@ -43,7 +43,7 @@
 
 static kmem_zone_t *xfs_buf_zone;
 STATIC int xfsbufd(void *);
-STATIC void xfs_buf_delwri_queue(xfs_buf_t *, int);
+STATIC void xfs_buf_delwri_queue(xfs_buf_t *);
 
 static struct workqueue_struct *xfslogd_workqueue;
 struct workqueue_struct *xfsdatad_workqueue;
@@ -940,7 +940,7 @@ xfs_buf_unlock(
        if ((bp->b_flags & (XBF_DELWRI|_XBF_DELWRI_Q)) == XBF_DELWRI) {
                atomic_inc(&bp->b_hold);
                bp->b_flags |= XBF_ASYNC;
-               xfs_buf_delwri_queue(bp, 0);
+               xfs_buf_delwri_queue(bp);
        }
 
        XB_CLEAR_OWNER(bp);
@@ -1049,7 +1049,8 @@ xfs_bdwrite(
        bp->b_flags &= ~XBF_READ;
        bp->b_flags |= (XBF_DELWRI | XBF_ASYNC);
 
-       xfs_buf_delwri_queue(bp, 1);
+       xfs_buf_delwri_queue(bp);
+       xfs_buf_unlock(bp);
 }
 
 /*
@@ -1562,8 +1563,7 @@ error:
  */
 STATIC void
 xfs_buf_delwri_queue(
-       xfs_buf_t               *bp,
-       int                     unlock)
+       xfs_buf_t               *bp)
 {
        struct list_head        *dwq = &bp->b_target->bt_delwrite_queue;
        spinlock_t              *dwlk = &bp->b_target->bt_delwrite_lock;
@@ -1576,8 +1576,7 @@ xfs_buf_delwri_queue(
        /* If already in the queue, dequeue and place at tail */
        if (!list_empty(&bp->b_list)) {
                ASSERT(bp->b_flags & _XBF_DELWRI_Q);
-               if (unlock)
-                       atomic_dec(&bp->b_hold);
+               atomic_dec(&bp->b_hold);
                list_del(&bp->b_list);
        }
 
@@ -1590,9 +1589,6 @@ xfs_buf_delwri_queue(
        list_add_tail(&bp->b_list, dwq);
        bp->b_queuetime = jiffies;
        spin_unlock(dwlk);
-
-       if (unlock)
-               xfs_buf_unlock(bp);
 }
 
 void