xfs: revert to using a kthread for AIL pushing

Currently we have a few issues with the way the workqueue code is used to
implement AIL pushing:

 - it accidentally uses the same workqueue as the syncer action, and thus
   can be prevented from running if there are enough sync actions active
   in the system.
 - it doesn't use the HIGHPRI flag to queue at the head of the queue of
   work items

At this point I'm not confident enough in getting all the workqueue flags and
tweaks right to provide a perfectly reliable execution context for AIL
pushing, which is the most important piece in XFS to make forward progress
when the log fills.

Revert back to use a kthread per filesystem which fixes all the above issues
at the cost of having a task struct and stack around for each mounted
filesystem.  In addition this also gives us much better ways to diagnose
any issues involving hung AIL pushing and removes a small amount of code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reported-by: Stefan Priebe <s.priebe@profihost.ag>
Tested-by: Stefan Priebe <s.priebe@profihost.ag>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>

1 files changed, 41 insertions, 32 deletions

diff --git a/fs/xfs/xfs_trans_ail.c b/fs/xfs/xfs_trans_ail.c
index e4cd9180d33a..3a1e7ca54c2d 100644
--- a/fs/xfs/xfs_trans_ail.c
+++ b/fs/xfs/xfs_trans_ail.c
@@ -28,8 +28,6 @@
 #include "xfs_trans_priv.h"
 #include "xfs_error.h"
 
-struct workqueue_struct *xfs_ail_wq;    /* AIL workqueue */
-
 #ifdef DEBUG
 /*
  * Check that the list is sorted as it should be.
@@ -356,16 +354,10 @@ xfs_ail_delete(
         xfs_trans_ail_cursor_clear(ailp, lip);
 }
 
-/*
- * 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.
- */
-STATIC void
-xfs_ail_worker(
-        struct work_struct      *work)
+static long
+xfsaild_push(
+        struct xfs_ail          *ailp)
 {
-        struct xfs_ail          *ailp = container_of(to_delayed_work(work),
-                                        struct xfs_ail, xa_work);
         xfs_mount_t             *mp = ailp->xa_mount;
         struct xfs_ail_cursor   cur;
         xfs_log_item_t          *lip;
@@ -505,20 +497,6 @@ out_done:
                 /* We're past our target or empty, so idle */
                 ailp->xa_last_pushed_lsn = 0;
 
-                /*
-                 * We clear the XFS_AIL_PUSHING_BIT first before checking
-                 * whether the target has changed. If the target has changed,
-                 * this pushes the requeue race directly onto the result of the
-                 * atomic test/set bit, so we are guaranteed that either the
-                 * the pusher that changed the target or ourselves will requeue
-                 * the work (but not both).
-                 */
-                clear_bit(XFS_AIL_PUSHING_BIT, &ailp->xa_flags);
-                smp_rmb();
-                if (XFS_LSN_CMP(ailp->xa_target, target) == 0 ||
-                    test_and_set_bit(XFS_AIL_PUSHING_BIT, &ailp->xa_flags))
-                        return;
-
                 tout = 50;
         } else if (XFS_LSN_CMP(lsn, target) >= 0) {
                 /*
@@ -541,9 +519,30 @@ out_done:
                 tout = 20;
         }
 
-        /* There is more to do, requeue us.  */
-        queue_delayed_work(xfs_syncd_wq, &ailp->xa_work,
-                                        msecs_to_jiffies(tout));
+        return tout;
+}
+
+static int
+xfsaild(
+        void            *data)
+{
+        struct xfs_ail  *ailp = data;
+        long            tout = 0;       /* milliseconds */
+
+        while (!kthread_should_stop()) {
+                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);
+
+                try_to_freeze();
+
+                tout = xfsaild_push(ailp);
+        }
+
+        return 0;
 }
 
 /*
@@ -578,8 +577,9 @@ xfs_ail_push(
          */
         smp_wmb();
         xfs_trans_ail_copy_lsn(ailp, &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);
+        smp_wmb();
+
+        wake_up_process(ailp->xa_task);
 }
 
 /*
@@ -817,9 +817,18 @@ xfs_trans_ail_init(
         INIT_LIST_HEAD(&ailp->xa_ail);
         INIT_LIST_HEAD(&ailp->xa_cursors);
         spin_lock_init(&ailp->xa_lock);
-        INIT_DELAYED_WORK(&ailp->xa_work, xfs_ail_worker);
+
+        ailp->xa_task = kthread_run(xfsaild, ailp, "xfsaild/%s",
+                        ailp->xa_mount->m_fsname);
+        if (IS_ERR(ailp->xa_task))
+                goto out_free_ailp;
+
         mp->m_ail = ailp;
         return 0;
+
+out_free_ailp:
+        kmem_free(ailp);
+        return ENOMEM;
 }
 
 void
@@ -828,6 +837,6 @@ xfs_trans_ail_destroy(
 {
         struct xfs_ail  *ailp = mp->m_ail;
 
-        cancel_delayed_work_sync(&ailp->xa_work);
+        kthread_stop(ailp->xa_task);
         kmem_free(ailp);
 }