X-Git-Url: http://ftp.safe.ca/?a=blobdiff_plain;f=drivers%2Fmd%2Fdm-mpath.c;h=826bce7343b3697e5abdfd4c62e760250f6ecf3a;hb=587611d6e4d5c0fb5e9492cb06d9054744d69536;hp=9bf3460c5540b9fdbefbacaed15788f7a6868fb0;hpb=6000a368cd8e6da1caf101411bdb494cd6fb8b09;p=safe%2Fjmp%2Flinux-2.6

diff --git a/drivers/md/dm-mpath.c b/drivers/md/dm-mpath.c
index 9bf3460..826bce7 100644
--- a/drivers/md/dm-mpath.c
+++ b/drivers/md/dm-mpath.c
@@ -5,10 +5,9 @@
  * This file is released under the GPL.
  */
 
-#include "dm.h"
+#include <linux/device-mapper.h>
+
 #include "dm-path-selector.h"
-#include "dm-bio-list.h"
-#include "dm-bio-record.h"
 #include "dm-uevent.h"
 
 #include <linux/ctype.h>
@@ -35,6 +34,7 @@ struct pgpath {
 
 	struct dm_path path;
 	struct work_struct deactivate_path;
+	struct work_struct activate_path;
 };
 
 #define path_to_pgpath(__pgp) container_of((__pgp), struct pgpath, path)
@@ -64,12 +64,12 @@ struct multipath {
 	spinlock_t lock;
 
 	const char *hw_handler_name;
-	struct work_struct activate_path;
-	struct pgpath *pgpath_to_activate;
+	char *hw_handler_params;
 	unsigned nr_priority_groups;
 	struct list_head priority_groups;
 	unsigned pg_init_required;	/* pg_init needs calling? */
 	unsigned pg_init_in_progress;	/* Only one pg_init allowed at once */
+	wait_queue_head_t pg_init_wait;	/* Wait for pg_init completion */
 
 	unsigned nr_valid_paths;	/* Total number of usable paths */
 	struct pgpath *current_pgpath;
@@ -84,7 +84,7 @@ struct multipath {
 	unsigned pg_init_count;		/* Number of times pg_init called */
 
 	struct work_struct process_queued_ios;
-	struct bio_list queued_ios;
+	struct list_head queued_ios;
 	unsigned queue_size;
 
 	struct work_struct trigger_event;
@@ -94,6 +94,8 @@ struct multipath {
 	 * can resubmit bios on error.
 	 */
 	mempool_t *mpio_pool;
+
+	struct mutex work_mutex;
 };
 
 /*
@@ -101,7 +103,7 @@ struct multipath {
  */
 struct dm_mpath_io {
 	struct pgpath *pgpath;
-	struct dm_bio_details details;
+	size_t nr_bytes;
 };
 
 typedef int (*action_fn) (struct pgpath *pgpath);
@@ -128,6 +130,7 @@ static struct pgpath *alloc_pgpath(void)
 	if (pgpath) {
 		pgpath->is_active = 1;
 		INIT_WORK(&pgpath->deactivate_path, deactivate_path);
+		INIT_WORK(&pgpath->activate_path, activate_path);
 	}
 
 	return pgpath;
@@ -160,7 +163,6 @@ static struct priority_group *alloc_priority_group(void)
 
 static void free_pgpaths(struct list_head *pgpaths, struct dm_target *ti)
 {
-	unsigned long flags;
 	struct pgpath *pgpath, *tmp;
 	struct multipath *m = ti->private;
 
@@ -169,10 +171,6 @@ static void free_pgpaths(struct list_head *pgpaths, struct dm_target *ti)
 		if (m->hw_handler_name)
 			scsi_dh_detach(bdev_get_queue(pgpath->path.dev->bdev));
 		dm_put_device(ti, pgpath->path.dev);
-		spin_lock_irqsave(&m->lock, flags);
-		if (m->pgpath_to_activate == pgpath)
-			m->pgpath_to_activate = NULL;
-		spin_unlock_irqrestore(&m->lock, flags);
 		free_pgpath(pgpath);
 	}
 }
@@ -198,11 +196,13 @@ static struct multipath *alloc_multipath(struct dm_target *ti)
 	m = kzalloc(sizeof(*m), GFP_KERNEL);
 	if (m) {
 		INIT_LIST_HEAD(&m->priority_groups);
+		INIT_LIST_HEAD(&m->queued_ios);
 		spin_lock_init(&m->lock);
 		m->queue_io = 1;
 		INIT_WORK(&m->process_queued_ios, process_queued_ios);
 		INIT_WORK(&m->trigger_event, trigger_event);
-		INIT_WORK(&m->activate_path, activate_path);
+		init_waitqueue_head(&m->pg_init_wait);
+		mutex_init(&m->work_mutex);
 		m->mpio_pool = mempool_create_slab_pool(MIN_IOS, _mpio_cache);
 		if (!m->mpio_pool) {
 			kfree(m);
@@ -225,6 +225,7 @@ static void free_multipath(struct multipath *m)
 	}
 
 	kfree(m->hw_handler_name);
+	kfree(m->hw_handler_params);
 	mempool_destroy(m->mpio_pool);
 	kfree(m);
 }
@@ -234,6 +235,21 @@ static void free_multipath(struct multipath *m)
  * Path selection
  *-----------------------------------------------*/
 
+static void __pg_init_all_paths(struct multipath *m)
+{
+	struct pgpath *pgpath;
+
+	m->pg_init_count++;
+	m->pg_init_required = 0;
+	list_for_each_entry(pgpath, &m->current_pg->pgpaths, list) {
+		/* Skip failed paths */
+		if (!pgpath->is_active)
+			continue;
+		if (queue_work(kmpath_handlerd, &pgpath->activate_path))
+			m->pg_init_in_progress++;
+	}
+}
+
 static void __switch_pg(struct multipath *m, struct pgpath *pgpath)
 {
 	m->current_pg = pgpath->pg;
@@ -250,11 +266,12 @@ static void __switch_pg(struct multipath *m, struct pgpath *pgpath)
 	m->pg_init_count = 0;
 }
 
-static int __choose_path_in_pg(struct multipath *m, struct priority_group *pg)
+static int __choose_path_in_pg(struct multipath *m, struct priority_group *pg,
+			       size_t nr_bytes)
 {
 	struct dm_path *path;
 
-	path = pg->ps.type->select_path(&pg->ps, &m->repeat_count);
+	path = pg->ps.type->select_path(&pg->ps, &m->repeat_count, nr_bytes);
 	if (!path)
 		return -ENXIO;
 
@@ -266,7 +283,7 @@ static int __choose_path_in_pg(struct multipath *m, struct priority_group *pg)
 	return 0;
 }
 
-static void __choose_pgpath(struct multipath *m)
+static void __choose_pgpath(struct multipath *m, size_t nr_bytes)
 {
 	struct priority_group *pg;
 	unsigned bypassed = 1;
@@ -278,12 +295,12 @@ static void __choose_pgpath(struct multipath *m)
 	if (m->next_pg) {
 		pg = m->next_pg;
 		m->next_pg = NULL;
-		if (!__choose_path_in_pg(m, pg))
+		if (!__choose_path_in_pg(m, pg, nr_bytes))
 			return;
 	}
 
 	/* Don't change PG until it has no remaining paths */
-	if (m->current_pg && !__choose_path_in_pg(m, m->current_pg))
+	if (m->current_pg && !__choose_path_in_pg(m, m->current_pg, nr_bytes))
 		return;
 
 	/*
@@ -295,7 +312,7 @@ static void __choose_pgpath(struct multipath *m)
 		list_for_each_entry(pg, &m->priority_groups, list) {
 			if (pg->bypassed == bypassed)
 				continue;
-			if (!__choose_path_in_pg(m, pg))
+			if (!__choose_path_in_pg(m, pg, nr_bytes))
 				return;
 		}
 	} while (bypassed--);
@@ -322,19 +339,21 @@ static int __must_push_back(struct multipath *m)
 		dm_noflush_suspending(m->ti));
 }
 
-static int map_io(struct multipath *m, struct bio *bio,
+static int map_io(struct multipath *m, struct request *clone,
 		  struct dm_mpath_io *mpio, unsigned was_queued)
 {
 	int r = DM_MAPIO_REMAPPED;
+	size_t nr_bytes = blk_rq_bytes(clone);
 	unsigned long flags;
 	struct pgpath *pgpath;
+	struct block_device *bdev;
 
 	spin_lock_irqsave(&m->lock, flags);
 
 	/* Do we need to select a new pgpath? */
 	if (!m->current_pgpath ||
 	    (!m->queue_io && (m->repeat_count && --m->repeat_count == 0)))
-		__choose_pgpath(m);
+		__choose_pgpath(m, nr_bytes);
 
 	pgpath = m->current_pgpath;
 
@@ -344,21 +363,28 @@ static int map_io(struct multipath *m, struct bio *bio,
 	if ((pgpath && m->queue_io) ||
 	    (!pgpath && m->queue_if_no_path)) {
 		/* Queue for the daemon to resubmit */
-		bio_list_add(&m->queued_ios, bio);
+		list_add_tail(&clone->queuelist, &m->queued_ios);
 		m->queue_size++;
 		if ((m->pg_init_required && !m->pg_init_in_progress) ||
 		    !m->queue_io)
 			queue_work(kmultipathd, &m->process_queued_ios);
 		pgpath = NULL;
 		r = DM_MAPIO_SUBMITTED;
-	} else if (pgpath)
-		bio->bi_bdev = pgpath->path.dev->bdev;
-	else if (__must_push_back(m))
+	} else if (pgpath) {
+		bdev = pgpath->path.dev->bdev;
+		clone->q = bdev_get_queue(bdev);
+		clone->rq_disk = bdev->bd_disk;
+	} else if (__must_push_back(m))
 		r = DM_MAPIO_REQUEUE;
 	else
 		r = -EIO;	/* Failed */
 
 	mpio->pgpath = pgpath;
+	mpio->nr_bytes = nr_bytes;
+
+	if (r == DM_MAPIO_REMAPPED && pgpath->pg->ps.type->start_io)
+		pgpath->pg->ps.type->start_io(&pgpath->pg->ps, &pgpath->path,
+					      nr_bytes);
 
 	spin_unlock_irqrestore(&m->lock, flags);
 
@@ -396,30 +422,31 @@ static void dispatch_queued_ios(struct multipath *m)
 {
 	int r;
 	unsigned long flags;
-	struct bio *bio = NULL, *next;
 	struct dm_mpath_io *mpio;
 	union map_info *info;
+	struct request *clone, *n;
+	LIST_HEAD(cl);
 
 	spin_lock_irqsave(&m->lock, flags);
-	bio = bio_list_get(&m->queued_ios);
+	list_splice_init(&m->queued_ios, &cl);
 	spin_unlock_irqrestore(&m->lock, flags);
 
-	while (bio) {
-		next = bio->bi_next;
-		bio->bi_next = NULL;
+	list_for_each_entry_safe(clone, n, &cl, queuelist) {
+		list_del_init(&clone->queuelist);
 
-		info = dm_get_mapinfo(bio);
+		info = dm_get_rq_mapinfo(clone);
 		mpio = info->ptr;
 
-		r = map_io(m, bio, mpio, 1);
-		if (r < 0)
-			bio_endio(bio, r);
-		else if (r == DM_MAPIO_REMAPPED)
-			generic_make_request(bio);
-		else if (r == DM_MAPIO_REQUEUE)
-			bio_endio(bio, -EIO);
-
-		bio = next;
+		r = map_io(m, clone, mpio, 1);
+		if (r < 0) {
+			mempool_free(mpio, m->mpio_pool);
+			dm_kill_unmapped_request(clone, r);
+		} else if (r == DM_MAPIO_REMAPPED)
+			dm_dispatch_request(clone);
+		else if (r == DM_MAPIO_REQUEUE) {
+			mempool_free(mpio, m->mpio_pool);
+			dm_requeue_unmapped_request(clone);
+		}
 	}
 }
 
@@ -428,7 +455,7 @@ static void process_queued_ios(struct work_struct *work)
 	struct multipath *m =
 		container_of(work, struct multipath, process_queued_ios);
 	struct pgpath *pgpath = NULL;
-	unsigned init_required = 0, must_queue = 1;
+	unsigned must_queue = 1;
 	unsigned long flags;
 
 	spin_lock_irqsave(&m->lock, flags);
@@ -437,28 +464,19 @@ static void process_queued_ios(struct work_struct *work)
 		goto out;
 
 	if (!m->current_pgpath)
-		__choose_pgpath(m);
+		__choose_pgpath(m, 0);
 
 	pgpath = m->current_pgpath;
-	m->pgpath_to_activate = m->current_pgpath;
 
 	if ((pgpath && !m->queue_io) ||
 	    (!pgpath && !m->queue_if_no_path))
 		must_queue = 0;
 
-	if (m->pg_init_required && !m->pg_init_in_progress) {
-		m->pg_init_count++;
-		m->pg_init_required = 0;
-		m->pg_init_in_progress = 1;
-		init_required = 1;
-	}
+	if (m->pg_init_required && !m->pg_init_in_progress && pgpath)
+		__pg_init_all_paths(m);
 
 out:
 	spin_unlock_irqrestore(&m->lock, flags);
-
-	if (init_required)
-		queue_work(kmpath_handlerd, &m->activate_path);
-
 	if (!must_queue)
 		dispatch_queued_ios(m);
 }
@@ -553,6 +571,12 @@ static int parse_path_selector(struct arg_set *as, struct priority_group *pg,
 		return -EINVAL;
 	}
 
+	if (ps_argc > as->argc) {
+		dm_put_path_selector(pst);
+		ti->error = "not enough arguments for path selector";
+		return -EINVAL;
+	}
+
 	r = pst->create(&pg->ps, ps_argc, as->argv);
 	if (r) {
 		dm_put_path_selector(pst);
@@ -583,20 +607,42 @@ static struct pgpath *parse_path(struct arg_set *as, struct path_selector *ps,
 	if (!p)
 		return ERR_PTR(-ENOMEM);
 
-	r = dm_get_device(ti, shift(as), ti->begin, ti->len,
-			  dm_table_get_mode(ti->table), &p->path.dev);
+	r = dm_get_device(ti, shift(as), dm_table_get_mode(ti->table),
+			  &p->path.dev);
 	if (r) {
 		ti->error = "error getting device";
 		goto bad;
 	}
 
 	if (m->hw_handler_name) {
-		r = scsi_dh_attach(bdev_get_queue(p->path.dev->bdev),
-				   m->hw_handler_name);
+		struct request_queue *q = bdev_get_queue(p->path.dev->bdev);
+
+		r = scsi_dh_attach(q, m->hw_handler_name);
+		if (r == -EBUSY) {
+			/*
+			 * Already attached to different hw_handler,
+			 * try to reattach with correct one.
+			 */
+			scsi_dh_detach(q);
+			r = scsi_dh_attach(q, m->hw_handler_name);
+		}
+
 		if (r < 0) {
+			ti->error = "error attaching hardware handler";
 			dm_put_device(ti, p->path.dev);
 			goto bad;
 		}
+
+		if (m->hw_handler_params) {
+			r = scsi_dh_set_params(q, m->hw_handler_params);
+			if (r < 0) {
+				ti->error = "unable to set hardware "
+							"handler parameters";
+				scsi_dh_detach(q);
+				dm_put_device(ti, p->path.dev);
+				goto bad;
+			}
+		}
 	}
 
 	r = ps->type->add_path(ps, &p->path, as->argc, as->argv, &ti->error);
@@ -687,6 +733,7 @@ static struct priority_group *parse_priority_group(struct arg_set *as,
 static int parse_hw_handler(struct arg_set *as, struct multipath *m)
 {
 	unsigned hw_argc;
+	int ret;
 	struct dm_target *ti = m->ti;
 
 	static struct param _params[] = {
@@ -699,17 +746,42 @@ static int parse_hw_handler(struct arg_set *as, struct multipath *m)
 	if (!hw_argc)
 		return 0;
 
+	if (hw_argc > as->argc) {
+		ti->error = "not enough arguments for hardware handler";
+		return -EINVAL;
+	}
+
 	m->hw_handler_name = kstrdup(shift(as), GFP_KERNEL);
 	request_module("scsi_dh_%s", m->hw_handler_name);
 	if (scsi_dh_handler_exist(m->hw_handler_name) == 0) {
 		ti->error = "unknown hardware handler type";
-		kfree(m->hw_handler_name);
-		m->hw_handler_name = NULL;
-		return -EINVAL;
+		ret = -EINVAL;
+		goto fail;
+	}
+
+	if (hw_argc > 1) {
+		char *p;
+		int i, j, len = 4;
+
+		for (i = 0; i <= hw_argc - 2; i++)
+			len += strlen(as->argv[i]) + 1;
+		p = m->hw_handler_params = kzalloc(len, GFP_KERNEL);
+		if (!p) {
+			ti->error = "memory allocation failed";
+			ret = -ENOMEM;
+			goto fail;
+		}
+		j = sprintf(p, "%d", hw_argc - 1);
+		for (i = 0, p+=j+1; i <= hw_argc - 2; i++, p+=j+1)
+			j = sprintf(p, "%s", as->argv[i]);
 	}
 	consume(as, hw_argc - 1);
 
 	return 0;
+fail:
+	kfree(m->hw_handler_name);
+	m->hw_handler_name = NULL;
+	return ret;
 }
 
 static int parse_features(struct arg_set *as, struct multipath *m)
@@ -819,6 +891,8 @@ static int multipath_ctr(struct dm_target *ti, unsigned int argc,
 		goto bad;
 	}
 
+	ti->num_flush_requests = 1;
+
 	return 0;
 
  bad:
@@ -826,31 +900,65 @@ static int multipath_ctr(struct dm_target *ti, unsigned int argc,
 	return r;
 }
 
-static void multipath_dtr(struct dm_target *ti)
+static void multipath_wait_for_pg_init_completion(struct multipath *m)
 {
-	struct multipath *m = (struct multipath *) ti->private;
+	DECLARE_WAITQUEUE(wait, current);
+	unsigned long flags;
+
+	add_wait_queue(&m->pg_init_wait, &wait);
 
+	while (1) {
+		set_current_state(TASK_UNINTERRUPTIBLE);
+
+		spin_lock_irqsave(&m->lock, flags);
+		if (!m->pg_init_in_progress) {
+			spin_unlock_irqrestore(&m->lock, flags);
+			break;
+		}
+		spin_unlock_irqrestore(&m->lock, flags);
+
+		io_schedule();
+	}
+	set_current_state(TASK_RUNNING);
+
+	remove_wait_queue(&m->pg_init_wait, &wait);
+}
+
+static void flush_multipath_work(struct multipath *m)
+{
 	flush_workqueue(kmpath_handlerd);
+	multipath_wait_for_pg_init_completion(m);
 	flush_workqueue(kmultipathd);
+	flush_scheduled_work();
+}
+
+static void multipath_dtr(struct dm_target *ti)
+{
+	struct multipath *m = ti->private;
+
+	flush_multipath_work(m);
 	free_multipath(m);
 }
 
 /*
- * Map bios, recording original fields for later in case we have to resubmit
+ * Map cloned requests
  */
-static int multipath_map(struct dm_target *ti, struct bio *bio,
+static int multipath_map(struct dm_target *ti, struct request *clone,
 			 union map_info *map_context)
 {
 	int r;
 	struct dm_mpath_io *mpio;
 	struct multipath *m = (struct multipath *) ti->private;
 
-	mpio = mempool_alloc(m->mpio_pool, GFP_NOIO);
-	dm_bio_record(&mpio->details, bio);
+	mpio = mempool_alloc(m->mpio_pool, GFP_ATOMIC);
+	if (!mpio)
+		/* ENOMEM, requeue */
+		return DM_MAPIO_REQUEUE;
+	memset(mpio, 0, sizeof(*mpio));
 
 	map_context->ptr = mpio;
-	bio->bi_rw |= (1 << BIO_RW_FAILFAST_TRANSPORT);
-	r = map_io(m, bio, mpio, 0);
+	clone->cmd_flags |= REQ_FAILFAST_TRANSPORT;
+	r = map_io(m, clone, mpio, 0);
 	if (r < 0 || r == DM_MAPIO_REQUEUE)
 		mempool_free(mpio, m->mpio_pool);
 
@@ -884,7 +992,7 @@ static int fail_path(struct pgpath *pgpath)
 	dm_path_uevent(DM_UEVENT_PATH_FAILED, m->ti,
 		      pgpath->path.dev->name, m->nr_valid_paths);
 
-	queue_work(kmultipathd, &m->trigger_event);
+	schedule_work(&m->trigger_event);
 	queue_work(kmultipathd, &pgpath->deactivate_path);
 
 out:
@@ -920,14 +1028,18 @@ static int reinstate_path(struct pgpath *pgpath)
 
 	pgpath->is_active = 1;
 
-	m->current_pgpath = NULL;
-	if (!m->nr_valid_paths++ && m->queue_size)
+	if (!m->nr_valid_paths++ && m->queue_size) {
+		m->current_pgpath = NULL;
 		queue_work(kmultipathd, &m->process_queued_ios);
+	} else if (m->hw_handler_name && (m->current_pg == pgpath->pg)) {
+		if (queue_work(kmpath_handlerd, &pgpath->activate_path))
+			m->pg_init_in_progress++;
+	}
 
 	dm_path_uevent(DM_UEVENT_PATH_REINSTATED, m->ti,
 		      pgpath->path.dev->name, m->nr_valid_paths);
 
-	queue_work(kmultipathd, &m->trigger_event);
+	schedule_work(&m->trigger_event);
 
 out:
 	spin_unlock_irqrestore(&m->lock, flags);
@@ -971,7 +1083,7 @@ static void bypass_pg(struct multipath *m, struct priority_group *pg,
 
 	spin_unlock_irqrestore(&m->lock, flags);
 
-	queue_work(kmultipathd, &m->trigger_event);
+	schedule_work(&m->trigger_event);
 }
 
 /*
@@ -1001,7 +1113,7 @@ static int switch_pg_num(struct multipath *m, const char *pgstr)
 	}
 	spin_unlock_irqrestore(&m->lock, flags);
 
-	queue_work(kmultipathd, &m->trigger_event);
+	schedule_work(&m->trigger_event);
 	return 0;
 }
 
@@ -1049,9 +1161,9 @@ static int pg_init_limit_reached(struct multipath *m, struct pgpath *pgpath)
 	return limit_reached;
 }
 
-static void pg_init_done(struct dm_path *path, int errors)
+static void pg_init_done(void *data, int errors)
 {
-	struct pgpath *pgpath = path_to_pgpath(path);
+	struct pgpath *pgpath = data;
 	struct priority_group *pg = pgpath->pg;
 	struct multipath *m = pg->m;
 	unsigned long flags;
@@ -1065,8 +1177,8 @@ static void pg_init_done(struct dm_path *path, int errors)
 			errors = 0;
 			break;
 		}
-		DMERR("Cannot failover device because scsi_dh_%s was not "
-		      "loaded.", m->hw_handler_name);
+		DMERR("Could not failover the device: Handler scsi_dh_%s "
+		      "Error %d.", m->hw_handler_name, errors);
 		/*
 		 * Fail path for now, so we do not ping pong
 		 */
@@ -1098,87 +1210,79 @@ static void pg_init_done(struct dm_path *path, int errors)
 
 	spin_lock_irqsave(&m->lock, flags);
 	if (errors) {
-		DMERR("Could not failover device. Error %d.", errors);
-		m->current_pgpath = NULL;
-		m->current_pg = NULL;
-	} else if (!m->pg_init_required) {
-		m->queue_io = 0;
+		if (pgpath == m->current_pgpath) {
+			DMERR("Could not failover device. Error %d.", errors);
+			m->current_pgpath = NULL;
+			m->current_pg = NULL;
+		}
+	} else if (!m->pg_init_required)
 		pg->bypassed = 0;
-	}
 
-	m->pg_init_in_progress = 0;
+	if (--m->pg_init_in_progress)
+		/* Activations of other paths are still on going */
+		goto out;
+
+	if (!m->pg_init_required)
+		m->queue_io = 0;
+
 	queue_work(kmultipathd, &m->process_queued_ios);
+
+	/*
+	 * Wake up any thread waiting to suspend.
+	 */
+	wake_up(&m->pg_init_wait);
+
+out:
 	spin_unlock_irqrestore(&m->lock, flags);
 }
 
 static void activate_path(struct work_struct *work)
 {
-	int ret;
-	struct multipath *m =
-		container_of(work, struct multipath, activate_path);
-	struct dm_path *path;
-	unsigned long flags;
+	struct pgpath *pgpath =
+		container_of(work, struct pgpath, activate_path);
 
-	spin_lock_irqsave(&m->lock, flags);
-	path = &m->pgpath_to_activate->path;
-	m->pgpath_to_activate = NULL;
-	spin_unlock_irqrestore(&m->lock, flags);
-	if (!path)
-		return;
-	ret = scsi_dh_activate(bdev_get_queue(path->dev->bdev));
-	pg_init_done(path, ret);
+	scsi_dh_activate(bdev_get_queue(pgpath->path.dev->bdev),
+				pg_init_done, pgpath);
 }
 
 /*
  * end_io handling
  */
-static int do_end_io(struct multipath *m, struct bio *bio,
+static int do_end_io(struct multipath *m, struct request *clone,
 		     int error, struct dm_mpath_io *mpio)
 {
+	/*
+	 * We don't queue any clone request inside the multipath target
+	 * during end I/O handling, since those clone requests don't have
+	 * bio clones.  If we queue them inside the multipath target,
+	 * we need to make bio clones, that requires memory allocation.
+	 * (See drivers/md/dm.c:end_clone_bio() about why the clone requests
+	 *  don't have bio clones.)
+	 * Instead of queueing the clone request here, we queue the original
+	 * request into dm core, which will remake a clone request and
+	 * clone bios for it and resubmit it later.
+	 */
+	int r = DM_ENDIO_REQUEUE;
 	unsigned long flags;
 
-	if (!error)
+	if (!error && !clone->errors)
 		return 0;	/* I/O complete */
 
-	if ((error == -EWOULDBLOCK) && bio_rw_ahead(bio))
-		return error;
-
 	if (error == -EOPNOTSUPP)
 		return error;
 
-	spin_lock_irqsave(&m->lock, flags);
-	if (!m->nr_valid_paths) {
-		if (__must_push_back(m)) {
-			spin_unlock_irqrestore(&m->lock, flags);
-			return DM_ENDIO_REQUEUE;
-		} else if (!m->queue_if_no_path) {
-			spin_unlock_irqrestore(&m->lock, flags);
-			return -EIO;
-		} else {
-			spin_unlock_irqrestore(&m->lock, flags);
-			goto requeue;
-		}
-	}
-	spin_unlock_irqrestore(&m->lock, flags);
-
 	if (mpio->pgpath)
 		fail_path(mpio->pgpath);
 
-      requeue:
-	dm_bio_restore(&mpio->details, bio);
-
-	/* queue for the daemon to resubmit or fail */
 	spin_lock_irqsave(&m->lock, flags);
-	bio_list_add(&m->queued_ios, bio);
-	m->queue_size++;
-	if (!m->queue_io)
-		queue_work(kmultipathd, &m->process_queued_ios);
+	if (!m->nr_valid_paths && !m->queue_if_no_path && !__must_push_back(m))
+		r = -EIO;
 	spin_unlock_irqrestore(&m->lock, flags);
 
-	return DM_ENDIO_INCOMPLETE;	/* io not complete */
+	return r;
 }
 
-static int multipath_end_io(struct dm_target *ti, struct bio *bio,
+static int multipath_end_io(struct dm_target *ti, struct request *clone,
 			    int error, union map_info *map_context)
 {
 	struct multipath *m = ti->private;
@@ -1187,14 +1291,13 @@ static int multipath_end_io(struct dm_target *ti, struct bio *bio,
 	struct path_selector *ps;
 	int r;
 
-	r  = do_end_io(m, bio, error, mpio);
+	r  = do_end_io(m, clone, error, mpio);
 	if (pgpath) {
 		ps = &pgpath->pg->ps;
 		if (ps->type->end_io)
-			ps->type->end_io(ps, &pgpath->path);
+			ps->type->end_io(ps, &pgpath->path, mpio->nr_bytes);
 	}
-	if (r != DM_ENDIO_INCOMPLETE)
-		mempool_free(mpio, m->mpio_pool);
+	mempool_free(mpio, m->mpio_pool);
 
 	return r;
 }
@@ -1212,6 +1315,15 @@ static void multipath_presuspend(struct dm_target *ti)
 	queue_if_no_path(m, 0, 1);
 }
 
+static void multipath_postsuspend(struct dm_target *ti)
+{
+	struct multipath *m = ti->private;
+
+	mutex_lock(&m->work_mutex);
+	flush_multipath_work(m);
+	mutex_unlock(&m->work_mutex);
+}
+
 /*
  * Restore the queue_if_no_path setting.
  */
@@ -1348,75 +1460,84 @@ static int multipath_status(struct dm_target *ti, status_type_t type,
 
 static int multipath_message(struct dm_target *ti, unsigned argc, char **argv)
 {
-	int r;
+	int r = -EINVAL;
 	struct dm_dev *dev;
 	struct multipath *m = (struct multipath *) ti->private;
 	action_fn action;
 
+	mutex_lock(&m->work_mutex);
+
+	if (dm_suspended(ti)) {
+		r = -EBUSY;
+		goto out;
+	}
+
 	if (argc == 1) {
-		if (!strnicmp(argv[0], MESG_STR("queue_if_no_path")))
-			return queue_if_no_path(m, 1, 0);
-		else if (!strnicmp(argv[0], MESG_STR("fail_if_no_path")))
-			return queue_if_no_path(m, 0, 0);
+		if (!strnicmp(argv[0], MESG_STR("queue_if_no_path"))) {
+			r = queue_if_no_path(m, 1, 0);
+			goto out;
+		} else if (!strnicmp(argv[0], MESG_STR("fail_if_no_path"))) {
+			r = queue_if_no_path(m, 0, 0);
+			goto out;
+		}
 	}
 
-	if (argc != 2)
-		goto error;
+	if (argc != 2) {
+		DMWARN("Unrecognised multipath message received.");
+		goto out;
+	}
 
-	if (!strnicmp(argv[0], MESG_STR("disable_group")))
-		return bypass_pg_num(m, argv[1], 1);
-	else if (!strnicmp(argv[0], MESG_STR("enable_group")))
-		return bypass_pg_num(m, argv[1], 0);
-	else if (!strnicmp(argv[0], MESG_STR("switch_group")))
-		return switch_pg_num(m, argv[1]);
-	else if (!strnicmp(argv[0], MESG_STR("reinstate_path")))
+	if (!strnicmp(argv[0], MESG_STR("disable_group"))) {
+		r = bypass_pg_num(m, argv[1], 1);
+		goto out;
+	} else if (!strnicmp(argv[0], MESG_STR("enable_group"))) {
+		r = bypass_pg_num(m, argv[1], 0);
+		goto out;
+	} else if (!strnicmp(argv[0], MESG_STR("switch_group"))) {
+		r = switch_pg_num(m, argv[1]);
+		goto out;
+	} else if (!strnicmp(argv[0], MESG_STR("reinstate_path")))
 		action = reinstate_path;
 	else if (!strnicmp(argv[0], MESG_STR("fail_path")))
 		action = fail_path;
-	else
-		goto error;
+	else {
+		DMWARN("Unrecognised multipath message received.");
+		goto out;
+	}
 
-	r = dm_get_device(ti, argv[1], ti->begin, ti->len,
-			  dm_table_get_mode(ti->table), &dev);
+	r = dm_get_device(ti, argv[1], dm_table_get_mode(ti->table), &dev);
 	if (r) {
 		DMWARN("message: error getting device %s",
 		       argv[1]);
-		return -EINVAL;
+		goto out;
 	}
 
 	r = action_dev(m, dev, action);
 
 	dm_put_device(ti, dev);
 
+out:
+	mutex_unlock(&m->work_mutex);
 	return r;
-
-error:
-	DMWARN("Unrecognised multipath message received.");
-	return -EINVAL;
 }
 
-static int multipath_ioctl(struct dm_target *ti, struct inode *inode,
-			   struct file *filp, unsigned int cmd,
+static int multipath_ioctl(struct dm_target *ti, unsigned int cmd,
 			   unsigned long arg)
 {
 	struct multipath *m = (struct multipath *) ti->private;
 	struct block_device *bdev = NULL;
+	fmode_t mode = 0;
 	unsigned long flags;
-	struct file fake_file = {};
-	struct dentry fake_dentry = {};
 	int r = 0;
 
-	fake_file.f_path.dentry = &fake_dentry;
-
 	spin_lock_irqsave(&m->lock, flags);
 
 	if (!m->current_pgpath)
-		__choose_pgpath(m);
+		__choose_pgpath(m, 0);
 
 	if (m->current_pgpath) {
 		bdev = m->current_pgpath->path.dev->bdev;
-		fake_dentry.d_inode = bdev->bd_inode;
-		fake_file.f_mode = m->current_pgpath->path.dev->mode;
+		mode = m->current_pgpath->path.dev->mode;
 	}
 
 	if (m->queue_io)
@@ -1426,8 +1547,96 @@ static int multipath_ioctl(struct dm_target *ti, struct inode *inode,
 
 	spin_unlock_irqrestore(&m->lock, flags);
 
-	return r ? : blkdev_driver_ioctl(bdev->bd_inode, &fake_file,
-					 bdev->bd_disk, cmd, arg);
+	return r ? : __blkdev_driver_ioctl(bdev, mode, cmd, arg);
+}
+
+static int multipath_iterate_devices(struct dm_target *ti,
+				     iterate_devices_callout_fn fn, void *data)
+{
+	struct multipath *m = ti->private;
+	struct priority_group *pg;
+	struct pgpath *p;
+	int ret = 0;
+
+	list_for_each_entry(pg, &m->priority_groups, list) {
+		list_for_each_entry(p, &pg->pgpaths, list) {
+			ret = fn(ti, p->path.dev, ti->begin, ti->len, data);
+			if (ret)
+				goto out;
+		}
+	}
+
+out:
+	return ret;
+}
+
+static int __pgpath_busy(struct pgpath *pgpath)
+{
+	struct request_queue *q = bdev_get_queue(pgpath->path.dev->bdev);
+
+	return dm_underlying_device_busy(q);
+}
+
+/*
+ * We return "busy", only when we can map I/Os but underlying devices
+ * are busy (so even if we map I/Os now, the I/Os will wait on
+ * the underlying queue).
+ * In other words, if we want to kill I/Os or queue them inside us
+ * due to map unavailability, we don't return "busy".  Otherwise,
+ * dm core won't give us the I/Os and we can't do what we want.
+ */
+static int multipath_busy(struct dm_target *ti)
+{
+	int busy = 0, has_active = 0;
+	struct multipath *m = ti->private;
+	struct priority_group *pg;
+	struct pgpath *pgpath;
+	unsigned long flags;
+
+	spin_lock_irqsave(&m->lock, flags);
+
+	/* Guess which priority_group will be used at next mapping time */
+	if (unlikely(!m->current_pgpath && m->next_pg))
+		pg = m->next_pg;
+	else if (likely(m->current_pg))
+		pg = m->current_pg;
+	else
+		/*
+		 * We don't know which pg will be used at next mapping time.
+		 * We don't call __choose_pgpath() here to avoid to trigger
+		 * pg_init just by busy checking.
+		 * So we don't know whether underlying devices we will be using
+		 * at next mapping time are busy or not. Just try mapping.
+		 */
+		goto out;
+
+	/*
+	 * If there is one non-busy active path at least, the path selector
+	 * will be able to select it. So we consider such a pg as not busy.
+	 */
+	busy = 1;
+	list_for_each_entry(pgpath, &pg->pgpaths, list)
+		if (pgpath->is_active) {
+			has_active = 1;
+
+			if (!__pgpath_busy(pgpath)) {
+				busy = 0;
+				break;
+			}
+		}
+
+	if (!has_active)
+		/*
+		 * No active path in this pg, so this pg won't be used and
+		 * the current_pg will be changed at next mapping time.
+		 * We need to try mapping to determine it.
+		 */
+		busy = 0;
+
+out:
+	spin_unlock_irqrestore(&m->lock, flags);
+
+	return busy;
 }
 
 /*-----------------------------------------------------------------
@@ -1435,17 +1644,20 @@ static int multipath_ioctl(struct dm_target *ti, struct inode *inode,
  *---------------------------------------------------------------*/
 static struct target_type multipath_target = {
 	.name = "multipath",
-	.version = {1, 0, 5},
+	.version = {1, 1, 1},
 	.module = THIS_MODULE,
 	.ctr = multipath_ctr,
 	.dtr = multipath_dtr,
-	.map = multipath_map,
-	.end_io = multipath_end_io,
+	.map_rq = multipath_map,
+	.rq_end_io = multipath_end_io,
 	.presuspend = multipath_presuspend,
+	.postsuspend = multipath_postsuspend,
 	.resume = multipath_resume,
 	.status = multipath_status,
 	.message = multipath_message,
 	.ioctl  = multipath_ioctl,
+	.iterate_devices = multipath_iterate_devices,
+	.busy = multipath_busy,
 };
 
 static int __init dm_multipath_init(void)
@@ -1496,14 +1708,10 @@ static int __init dm_multipath_init(void)
 
 static void __exit dm_multipath_exit(void)
 {
-	int r;
-
 	destroy_workqueue(kmpath_handlerd);
 	destroy_workqueue(kmultipathd);
 
-	r = dm_unregister_target(&multipath_target);
-	if (r < 0)
-		DMERR("target unregister failed %d", r);
+	dm_unregister_target(&multipath_target);
 	kmem_cache_destroy(_mpio_cache);
 }