Another set of vdev queue optimizations.

Switch FIFO queues (SYNC/TRIM) and active queue of vdev queue from
time-sorted AVL-trees to simple lists.  AVL-trees are too expensive
for such a simple task.  To change I/O priority without searching
through the trees, add io_queue_state field to struct zio.

To not check number of queued I/Os for each priority add vq_cqueued
bitmap to struct vdev_queue.  Update it when adding/removing I/Os.
Make vq_cactive a separate array instead of struct vdev_queue_class
member.  Together those allow to avoid lots of cache misses when
looking for work in vdev_queue_class_to_issue().

Introduce deadline of ~0.5s for LBA-sorted queues.  Before this I
saw some I/Os waiting in a queue for up to 8 seconds and possibly
more due to starvation.  With this change I no longer see it.  I
had to slightly more complicate the comparison function, but since
it uses all the same cache lines the difference is minimal.  For a
sequential I/Os the new code in vdev_queue_io_to_issue() actually
often uses more simple avl_first(), falling back to avl_find() and
avl_nearest() only when needed.

Arrange members in struct zio to access only one cache line when
searching through vdev queues.  While there, remove io_alloc_node,
reusing the io_queue_node instead.  Those two are never used same
time.

Remove zfs_vdev_aggregate_trim parameter.  It was disabled for 4
years since implemented, while still wasted time maintaining the
offset-sorted tree of TRIM requests.  Just remove the tree.

Remove locking from txg_all_lists_empty().  It is racy by design,
while 2 pair of locks/unlocks take noticeable time under the vdev
queue lock.

With these changes in my tests with volblocksize=4KB I measure vdev
queue lock spin time reduction by 50% on read and 75% on write.

Signed-off-by:	Alexander Motin <[email protected]>
Sponsored by:	iXsystems, Inc.

include/sys/vdev.h

@@ -170,8 +170,9 @@ extern zio_t *vdev_queue_io(zio_t *zio);
 extern void vdev_queue_io_done(zio_t *zio);
 extern void vdev_queue_change_io_priority(zio_t *zio, zio_priority_t priority);
 
-extern int vdev_queue_length(vdev_t *vd);
+extern uint32_t vdev_queue_length(vdev_t *vd);
 extern uint64_t vdev_queue_last_offset(vdev_t *vd);
+extern uint64_t vdev_queue_class_length(vdev_t *vq, zio_priority_t p);
 
 extern void vdev_config_dirty(vdev_t *vd);
 extern void vdev_config_clean(vdev_t *vd);

include/sys/vdev_impl.h

@@ -149,27 +149,24 @@ struct vdev_cache {
 	kmutex_t	vc_lock;
 };
 
-typedef struct vdev_queue_class {
-	uint32_t	vqc_active;
-
-	/*
-	 * Sorted by offset or timestamp, depending on if the queue is
-	 * LBA-ordered vs FIFO.
-	 */
-	avl_tree_t	vqc_queued_tree;
+typedef union vdev_queue_class {
+	list_t		vqc_list;
+	avl_tree_t	vqc_tree;
 } vdev_queue_class_t;
 
 struct vdev_queue {
 	vdev_t		*vq_vdev;
 	vdev_queue_class_t vq_class[ZIO_PRIORITY_NUM_QUEUEABLE];
-	avl_tree_t	vq_active_tree;
 	avl_tree_t	vq_read_offset_tree;
 	avl_tree_t	vq_write_offset_tree;
-	avl_tree_t	vq_trim_offset_tree;
 	uint64_t	vq_last_offset;
 	zio_priority_t	vq_last_prio;	/* Last sent I/O priority. */
+	uint32_t	vq_cqueued;	/* Classes with queued I/Os. */
+	uint32_t	vq_cactive[ZIO_PRIORITY_NUM_QUEUEABLE];
+	uint32_t	vq_active;	/* Number of active I/Os. */
 	uint32_t	vq_ia_active;	/* Active interactive I/Os. */
 	uint32_t	vq_nia_credit;	/* Non-interactive I/Os credit. */
+	list_t		vq_active_list;	/* List of active I/Os. */
 	hrtime_t	vq_io_complete_ts; /* time last i/o completed */
 	hrtime_t	vq_io_delta_ts;
 	zio_t		vq_io_search; /* used as local for stack reduction */

include/sys/zio.h

@@ -437,6 +437,12 @@ typedef struct zio_link {
 	list_node_t	zl_child_node;
 } zio_link_t;
 
+enum zio_qstate {
+	ZIO_QS_NONE = 0,
+	ZIO_QS_QUEUED,
+	ZIO_QS_ACTIVE,
+};
+
 struct zio {
 	/* Core information about this I/O */
 	zbookmark_phys_t	io_bookmark;
@@ -481,16 +487,19 @@ struct zio {
 	const zio_vsd_ops_t *io_vsd_ops;
 	metaslab_class_t *io_metaslab_class;	/* dva throttle class */
 
+	enum zio_qstate	io_queue_state;	/* vdev queue state */
+	union {
+		list_node_t l;
+		avl_node_t a;
+	} io_queue_node ____cacheline_aligned;	/* allocator and vdev queues */
+	avl_node_t	io_offset_node;	/* vdev offset queues */
 	uint64_t	io_offset;
 	hrtime_t	io_timestamp;	/* submitted at */
 	hrtime_t	io_queued_timestamp;
 	hrtime_t	io_target_timestamp;
 	hrtime_t	io_delta;	/* vdev queue service delta */
 	hrtime_t	io_delay;	/* Device access time (disk or */
 					/* file). */
-	avl_node_t	io_queue_node;
-	avl_node_t	io_offset_node;
-	avl_node_t	io_alloc_node;
 	zio_alloc_list_t 	io_alloc_list;
 
 	/* Internal pipeline state */

man/man4/zfs.4

@@ -2016,12 +2016,6 @@ Historical statistics for this many latest TXGs will be available in
 Flush dirty data to disk at least every this many seconds (maximum TXG
 duration).
 .
-.It Sy zfs_vdev_aggregate_trim Ns = Ns Sy 0 Ns | Ns 1 Pq uint
-Allow TRIM I/O operations to be aggregated.
-This is normally not helpful because the extents to be trimmed
-will have been already been aggregated by the metaslab.
-This option is provided for debugging and performance analysis.
-.
 .It Sy zfs_vdev_aggregation_limit Ns = Ns Sy 1048576 Ns B Po 1 MiB Pc Pq uint
 Max vdev I/O aggregation size.
 .

module/zfs/spa_misc.c

@@ -730,7 +730,7 @@ spa_add(const char *name, nvlist_t *config, const char *altroot)
 		mutex_init(&spa->spa_allocs[i].spaa_lock, NULL, MUTEX_DEFAULT,
 		    NULL);
 		avl_create(&spa->spa_allocs[i].spaa_tree, zio_bookmark_compare,
-		    sizeof (zio_t), offsetof(zio_t, io_alloc_node));
+		    sizeof (zio_t), offsetof(zio_t, io_queue_node.a));
 	}
 	avl_create(&spa->spa_metaslabs_by_flushed, metaslab_sort_by_flushed,
 	    sizeof (metaslab_t), offsetof(metaslab_t, ms_spa_txg_node));

module/zfs/txg.c

@@ -895,15 +895,10 @@ txg_list_destroy(txg_list_t *tl)
 boolean_t
 txg_all_lists_empty(txg_list_t *tl)
 {
-	mutex_enter(&tl->tl_lock);
-	for (int i = 0; i < TXG_SIZE; i++) {
-		if (!txg_list_empty_impl(tl, i)) {
-			mutex_exit(&tl->tl_lock);
-			return (B_FALSE);
-		}
-	}
-	mutex_exit(&tl->tl_lock);
-	return (B_TRUE);
+	boolean_t res = B_TRUE;
+	for (int i = 0; i < TXG_SIZE; i++)
+		res &= (tl->tl_head[i] == NULL);
+	return (res);
 }
 
 /*

module/zfs/vdev.c

@@ -4613,11 +4613,9 @@ vdev_get_stats_ex_impl(vdev_t *vd, vdev_stat_t *vs, vdev_stat_ex_t *vsx)
 
 		memcpy(vsx, &vd->vdev_stat_ex, sizeof (vd->vdev_stat_ex));
 
-		for (t = 0; t < ARRAY_SIZE(vd->vdev_queue.vq_class); t++) {
-			vsx->vsx_active_queue[t] =
-			    vd->vdev_queue.vq_class[t].vqc_active;
-			vsx->vsx_pend_queue[t] = avl_numnodes(
-			    &vd->vdev_queue.vq_class[t].vqc_queued_tree);
+		for (t = 0; t < ZIO_PRIORITY_NUM_QUEUEABLE; t++) {
+			vsx->vsx_active_queue[t] = vd->vdev_queue.vq_cactive[t];
+			vsx->vsx_pend_queue[t] = vdev_queue_class_length(vd, t);
 		}
 	}
 }
@@ -5475,20 +5473,20 @@ vdev_deadman(vdev_t *vd, const char *tag)
 		vdev_queue_t *vq = &vd->vdev_queue;
 
 		mutex_enter(&vq->vq_lock);
-		if (avl_numnodes(&vq->vq_active_tree) > 0) {
+		if (vq->vq_active > 0) {
 			spa_t *spa = vd->vdev_spa;
 			zio_t *fio;
 			uint64_t delta;
 
-			zfs_dbgmsg("slow vdev: %s has %lu active IOs",
-			    vd->vdev_path, avl_numnodes(&vq->vq_active_tree));
+			zfs_dbgmsg("slow vdev: %s has %u active IOs",
+			    vd->vdev_path, vq->vq_active);
 
 			/*
 			 * Look at the head of all the pending queues,
 			 * if any I/O has been outstanding for longer than
 			 * the spa_deadman_synctime invoke the deadman logic.
 			 */
-			fio = avl_first(&vq->vq_active_tree);
+			fio = list_head(&vq->vq_active_list);
 			delta = gethrtime() - fio->io_timestamp;
 			if (delta > spa_deadman_synctime(spa))
 				zio_deadman(fio, tag);