diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index ee0e981..67d5b2c 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -816,75 +816,6 @@
 	return 0;
 }
 
-/*
- * XFS AIL push thread support
- */
-void
-xfsaild_wakeup(
-	struct xfs_ail		*ailp,
-	xfs_lsn_t		threshold_lsn)
-{
-	/* only ever move the target forwards */
-	if (XFS_LSN_CMP(threshold_lsn, ailp->xa_target) > 0) {
-		ailp->xa_target = threshold_lsn;
-		wake_up_process(ailp->xa_task);
-	}
-}
-
-STATIC int
-xfsaild(
-	void	*data)
-{
-	struct xfs_ail	*ailp = data;
-	xfs_lsn_t	last_pushed_lsn = 0;
-	long		tout = 0; /* milliseconds */
-
-	while (!kthread_should_stop()) {
-		/*
-		 * for short sleeps indicating congestion, don't allow us to
-		 * get woken early. Otherwise all we do is bang on the AIL lock
-		 * without making progress.
-		 */
-		if (tout && tout <= 20)
-			__set_current_state(TASK_KILLABLE);
-		else
-			__set_current_state(TASK_INTERRUPTIBLE);
-		schedule_timeout(tout ?
-				 msecs_to_jiffies(tout) : MAX_SCHEDULE_TIMEOUT);
-
-		/* swsusp */
-		try_to_freeze();
-
-		ASSERT(ailp->xa_mount->m_log);
-		if (XFS_FORCED_SHUTDOWN(ailp->xa_mount))
-			continue;
-
-		tout = xfsaild_push(ailp, &last_pushed_lsn);
-	}
-
-	return 0;
-}	/* xfsaild */
-
-int
-xfsaild_start(
-	struct xfs_ail	*ailp)
-{
-	ailp->xa_target = 0;
-	ailp->xa_task = kthread_run(xfsaild, ailp, "xfsaild/%s",
-				    ailp->xa_mount->m_fsname);
-	if (IS_ERR(ailp->xa_task))
-		return -PTR_ERR(ailp->xa_task);
-	return 0;
-}
-
-void
-xfsaild_stop(
-	struct xfs_ail	*ailp)
-{
-	kthread_stop(ailp->xa_task);
-}
-
-
 /* Catch misguided souls that try to use this interface on XFS */
 STATIC struct inode *
 xfs_fs_alloc_inode(
@@ -1786,6 +1717,38 @@
 }
 
 STATIC int __init
+xfs_init_workqueues(void)
+{
+	/*
+	 * max_active is set to 8 to give enough concurency to allow
+	 * multiple work operations on each CPU to run. This allows multiple
+	 * filesystems to be running sync work concurrently, and scales with
+	 * the number of CPUs in the system.
+	 */
+	xfs_syncd_wq = alloc_workqueue("xfssyncd", WQ_CPU_INTENSIVE, 8);
+	if (!xfs_syncd_wq)
+		goto out;
+
+	xfs_ail_wq = alloc_workqueue("xfsail", WQ_CPU_INTENSIVE, 8);
+	if (!xfs_ail_wq)
+		goto out_destroy_syncd;
+
+	return 0;
+
+out_destroy_syncd:
+	destroy_workqueue(xfs_syncd_wq);
+out:
+	return -ENOMEM;
+}
+
+STATIC void __exit
+xfs_destroy_workqueues(void)
+{
+	destroy_workqueue(xfs_ail_wq);
+	destroy_workqueue(xfs_syncd_wq);
+}
+
+STATIC int __init
 init_xfs_fs(void)
 {
 	int			error;
@@ -1800,10 +1763,14 @@
 	if (error)
 		goto out;
 
-	error = xfs_mru_cache_init();
+	error = xfs_init_workqueues();
 	if (error)
 		goto out_destroy_zones;
 
+	error = xfs_mru_cache_init();
+	if (error)
+		goto out_destroy_wq;
+
 	error = xfs_filestream_init();
 	if (error)
 		goto out_mru_cache_uninit;
@@ -1820,27 +1787,17 @@
 	if (error)
 		goto out_cleanup_procfs;
 
-	/*
-	 * max_active is set to 8 to give enough concurency to allow
-	 * multiple work operations on each CPU to run. This allows multiple
-	 * filesystems to be running sync work concurrently, and scales with
-	 * the number of CPUs in the system.
-	 */
-	xfs_syncd_wq = alloc_workqueue("xfssyncd", WQ_CPU_INTENSIVE, 8);
-	if (!xfs_syncd_wq) {
-		error = -ENOMEM;
+	error = xfs_init_workqueues();
+	if (error)
 		goto out_sysctl_unregister;
-	}
 
 	vfs_initquota();
 
 	error = register_filesystem(&xfs_fs_type);
 	if (error)
-		goto out_destroy_xfs_syncd;
+		goto out_sysctl_unregister;
 	return 0;
 
- out_destroy_xfs_syncd:
-	destroy_workqueue(xfs_syncd_wq);
  out_sysctl_unregister:
 	xfs_sysctl_unregister();
  out_cleanup_procfs:
@@ -1851,6 +1808,8 @@
 	xfs_filestream_uninit();
  out_mru_cache_uninit:
 	xfs_mru_cache_uninit();
+ out_destroy_wq:
+	xfs_destroy_workqueues();
  out_destroy_zones:
 	xfs_destroy_zones();
  out:
@@ -1862,12 +1821,12 @@
 {
 	vfs_exitquota();
 	unregister_filesystem(&xfs_fs_type);
-	destroy_workqueue(xfs_syncd_wq);
 	xfs_sysctl_unregister();
 	xfs_cleanup_procfs();
 	xfs_buf_terminate();
 	xfs_filestream_uninit();
 	xfs_mru_cache_uninit();
+	xfs_destroy_workqueues();
 	xfs_destroy_zones();
 }
 
diff --git a/fs/xfs/xfs_trans_ail.c b/fs/xfs/xfs_trans_ail.c
index 12aff95..cb3aeac 100644
--- a/fs/xfs/xfs_trans_ail.c
+++ b/fs/xfs/xfs_trans_ail.c
@@ -28,6 +28,8 @@
 #include "xfs_trans_priv.h"
 #include "xfs_error.h"
 
+struct workqueue_struct	*xfs_ail_wq;	/* AIL workqueue */
+
 STATIC void xfs_ail_splice(struct xfs_ail *, struct list_head *, xfs_lsn_t);
 STATIC void xfs_ail_delete(struct xfs_ail *, xfs_log_item_t *);
 STATIC xfs_log_item_t * xfs_ail_min(struct xfs_ail *);
@@ -69,36 +71,6 @@
 }
 
 /*
- * xfs_trans_push_ail
- *
- * This routine is called to move the tail of the AIL forward.  It does this by
- * trying to flush items in the AIL whose lsns are below the given
- * threshold_lsn.
- *
- * the push is run asynchronously in a separate thread, so we return the tail
- * of the log right now instead of the tail after the push. This means we will
- * either continue right away, or we will sleep waiting on the async thread to
- * do its work.
- *
- * We do this unlocked - we only need to know whether there is anything in the
- * AIL at the time we are called. We don't need to access the contents of
- * any of the objects, so the lock is not needed.
- */
-void
-xfs_trans_ail_push(
-	struct xfs_ail	*ailp,
-	xfs_lsn_t	threshold_lsn)
-{
-	xfs_log_item_t	*lip;
-
-	lip = xfs_ail_min(ailp);
-	if (lip && !XFS_FORCED_SHUTDOWN(ailp->xa_mount)) {
-		if (XFS_LSN_CMP(threshold_lsn, ailp->xa_target) > 0)
-			xfsaild_wakeup(ailp, threshold_lsn);
-	}
-}
-
-/*
  * AIL traversal cursor initialisation.
  *
  * The cursor keeps track of where our current traversal is up
@@ -236,16 +208,16 @@
 }
 
 /*
- * xfsaild_push does the work of pushing on the AIL.  Returning a timeout of
- * zero indicates that the caller should sleep until woken.
+ * xfs_ail_worker does the work of pushing on the AIL. It will requeue itself
+ * to run at a later time if there is more work to do to complete the push.
  */
-long
-xfsaild_push(
-	struct xfs_ail	*ailp,
-	xfs_lsn_t	*last_lsn)
+STATIC void
+xfs_ail_worker(
+	struct work_struct *work)
 {
-	long		tout = 0;
-	xfs_lsn_t	last_pushed_lsn = *last_lsn;
+	struct xfs_ail	*ailp = container_of(to_delayed_work(work),
+					struct xfs_ail, xa_work);
+	long		tout;
 	xfs_lsn_t	target =  ailp->xa_target;
 	xfs_lsn_t	lsn;
 	xfs_log_item_t	*lip;
@@ -256,15 +228,15 @@
 
 	spin_lock(&ailp->xa_lock);
 	xfs_trans_ail_cursor_init(ailp, cur);
-	lip = xfs_trans_ail_cursor_first(ailp, cur, *last_lsn);
+	lip = xfs_trans_ail_cursor_first(ailp, cur, ailp->xa_last_pushed_lsn);
 	if (!lip || XFS_FORCED_SHUTDOWN(mp)) {
 		/*
 		 * AIL is empty or our push has reached the end.
 		 */
 		xfs_trans_ail_cursor_done(ailp, cur);
 		spin_unlock(&ailp->xa_lock);
-		*last_lsn = 0;
-		return tout;
+		ailp->xa_last_pushed_lsn = 0;
+		return;
 	}
 
 	XFS_STATS_INC(xs_push_ail);
@@ -301,13 +273,13 @@
 		case XFS_ITEM_SUCCESS:
 			XFS_STATS_INC(xs_push_ail_success);
 			IOP_PUSH(lip);
-			last_pushed_lsn = lsn;
+			ailp->xa_last_pushed_lsn = lsn;
 			break;
 
 		case XFS_ITEM_PUSHBUF:
 			XFS_STATS_INC(xs_push_ail_pushbuf);
 			IOP_PUSHBUF(lip);
-			last_pushed_lsn = lsn;
+			ailp->xa_last_pushed_lsn = lsn;
 			push_xfsbufd = 1;
 			break;
 
@@ -319,7 +291,7 @@
 
 		case XFS_ITEM_LOCKED:
 			XFS_STATS_INC(xs_push_ail_locked);
-			last_pushed_lsn = lsn;
+			ailp->xa_last_pushed_lsn = lsn;
 			stuck++;
 			break;
 
@@ -374,9 +346,23 @@
 		wake_up_process(mp->m_ddev_targp->bt_task);
 	}
 
+	/* assume we have more work to do in a short while */
+	tout = 10;
 	if (!count) {
 		/* We're past our target or empty, so idle */
-		last_pushed_lsn = 0;
+		ailp->xa_last_pushed_lsn = 0;
+
+		/*
+		 * Check for an updated push target before clearing the
+		 * XFS_AIL_PUSHING_BIT. If the target changed, we've got more
+		 * work to do. Wait a bit longer before starting that work.
+		 */
+		smp_rmb();
+		if (ailp->xa_target == target) {
+			clear_bit(XFS_AIL_PUSHING_BIT, &ailp->xa_flags);
+			return;
+		}
+		tout = 50;
 	} else if (XFS_LSN_CMP(lsn, target) >= 0) {
 		/*
 		 * We reached the target so wait a bit longer for I/O to
@@ -384,7 +370,7 @@
 		 * start the next scan from the start of the AIL.
 		 */
 		tout = 50;
-		last_pushed_lsn = 0;
+		ailp->xa_last_pushed_lsn = 0;
 	} else if ((stuck * 100) / count > 90) {
 		/*
 		 * Either there is a lot of contention on the AIL or we
@@ -396,14 +382,48 @@
 		 * continuing from where we were.
 		 */
 		tout = 20;
-	} else {
-		/* more to do, but wait a short while before continuing */
-		tout = 10;
 	}
-	*last_lsn = last_pushed_lsn;
-	return tout;
+
+	/* There is more to do, requeue us.  */
+	queue_delayed_work(xfs_syncd_wq, &ailp->xa_work,
+					msecs_to_jiffies(tout));
 }
 
+/*
+ * This routine is called to move the tail of the AIL forward.  It does this by
+ * trying to flush items in the AIL whose lsns are below the given
+ * threshold_lsn.
+ *
+ * The push is run asynchronously in a workqueue, which means the caller needs
+ * to handle waiting on the async flush for space to become available.
+ * We don't want to interrupt any push that is in progress, hence we only queue
+ * work if we set the pushing bit approriately.
+ *
+ * We do this unlocked - we only need to know whether there is anything in the
+ * AIL at the time we are called. We don't need to access the contents of
+ * any of the objects, so the lock is not needed.
+ */
+void
+xfs_trans_ail_push(
+	struct xfs_ail	*ailp,
+	xfs_lsn_t	threshold_lsn)
+{
+	xfs_log_item_t	*lip;
+
+	lip = xfs_ail_min(ailp);
+	if (!lip || XFS_FORCED_SHUTDOWN(ailp->xa_mount) ||
+	    XFS_LSN_CMP(threshold_lsn, ailp->xa_target) <= 0)
+		return;
+
+	/*
+	 * Ensure that the new target is noticed in push code before it clears
+	 * the XFS_AIL_PUSHING_BIT.
+	 */
+	smp_wmb();
+	ailp->xa_target = threshold_lsn;
+	if (!test_and_set_bit(XFS_AIL_PUSHING_BIT, &ailp->xa_flags))
+		queue_delayed_work(xfs_syncd_wq, &ailp->xa_work, 0);
+}
 
 /*
  * This is to be called when an item is unlocked that may have
@@ -615,7 +635,6 @@
 	xfs_mount_t	*mp)
 {
 	struct xfs_ail	*ailp;
-	int		error;
 
 	ailp = kmem_zalloc(sizeof(struct xfs_ail), KM_MAYFAIL);
 	if (!ailp)
@@ -624,15 +643,9 @@
 	ailp->xa_mount = mp;
 	INIT_LIST_HEAD(&ailp->xa_ail);
 	spin_lock_init(&ailp->xa_lock);
-	error = xfsaild_start(ailp);
-	if (error)
-		goto out_free_ailp;
+	INIT_DELAYED_WORK(&ailp->xa_work, xfs_ail_worker);
 	mp->m_ail = ailp;
 	return 0;
-
-out_free_ailp:
-	kmem_free(ailp);
-	return error;
 }
 
 void
@@ -641,7 +654,7 @@
 {
 	struct xfs_ail	*ailp = mp->m_ail;
 
-	xfsaild_stop(ailp);
+	cancel_delayed_work_sync(&ailp->xa_work);
 	kmem_free(ailp);
 }
 
diff --git a/fs/xfs/xfs_trans_priv.h b/fs/xfs/xfs_trans_priv.h
index 35162c2..6ebd322 100644
--- a/fs/xfs/xfs_trans_priv.h
+++ b/fs/xfs/xfs_trans_priv.h
@@ -65,16 +65,22 @@
 struct xfs_ail {
 	struct xfs_mount	*xa_mount;
 	struct list_head	xa_ail;
-	uint			xa_gen;
-	struct task_struct	*xa_task;
 	xfs_lsn_t		xa_target;
 	struct xfs_ail_cursor	xa_cursors;
 	spinlock_t		xa_lock;
+	struct delayed_work	xa_work;
+	xfs_lsn_t		xa_last_pushed_lsn;
+	unsigned long		xa_flags;
 };
 
+#define XFS_AIL_PUSHING_BIT	0
+
 /*
  * From xfs_trans_ail.c
  */
+
+extern struct workqueue_struct	*xfs_ail_wq;	/* AIL workqueue */
+
 void	xfs_trans_ail_update_bulk(struct xfs_ail *ailp,
 				struct xfs_log_item **log_items, int nr_items,
 				xfs_lsn_t lsn) __releases(ailp->xa_lock);
@@ -112,11 +118,6 @@
 void			xfs_trans_ail_cursor_done(struct xfs_ail *ailp,
 					struct xfs_ail_cursor *cur);
 
-long	xfsaild_push(struct xfs_ail *, xfs_lsn_t *);
-void	xfsaild_wakeup(struct xfs_ail *, xfs_lsn_t);
-int	xfsaild_start(struct xfs_ail *);
-void	xfsaild_stop(struct xfs_ail *);
-
 #if BITS_PER_LONG != 64
 static inline void
 xfs_trans_ail_copy_lsn(