Merge branch 'quilt/device-mapper'

Conflicts:
	drivers/md/dm.c

11 files changed, 349 insertions, 127 deletions

diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
index 3bdbb6115702..a8aab9cf26b9 100644
--- a/drivers/md/dm-crypt.c
+++ b/drivers/md/dm-crypt.c
@@ -1168,12 +1168,11 @@ static int crypt_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 	if (ivmode && cc->iv_gen_ops) {
 		if (ivopts)
 			*(ivopts - 1) = ':';
-		cc->iv_mode = kmalloc(strlen(ivmode) + 1, GFP_KERNEL);
+		cc->iv_mode = kstrdup(ivmode, GFP_KERNEL);
 		if (!cc->iv_mode) {
 			ti->error = "Error kmallocing iv_mode string";
 			goto bad_ivmode_string;
 		}
-		strcpy(cc->iv_mode, ivmode);
 	} else
 		cc->iv_mode = NULL;
 
diff --git a/drivers/md/dm-exception-store.c b/drivers/md/dm-exception-store.c
index 2b7907b6dd09..0bdb201c2c2a 100644
--- a/drivers/md/dm-exception-store.c
+++ b/drivers/md/dm-exception-store.c
@@ -173,7 +173,9 @@ int dm_exception_store_set_chunk_size(struct dm_exception_store *store,
 
 	/* Validate the chunk size against the device block size */
 	if (chunk_size %
-	    (bdev_logical_block_size(dm_snap_cow(store->snap)->bdev) >> 9)) {
+	    (bdev_logical_block_size(dm_snap_cow(store->snap)->bdev) >> 9) ||
+	    chunk_size %
+	    (bdev_logical_block_size(dm_snap_origin(store->snap)->bdev) >> 9)) {
 		*error = "Chunk size is not a multiple of device blocksize";
 		return -EINVAL;
 	}
diff --git a/drivers/md/dm-exception-store.h b/drivers/md/dm-exception-store.h
index e8dfa06af3ba..0b2536247cf5 100644
--- a/drivers/md/dm-exception-store.h
+++ b/drivers/md/dm-exception-store.h
@@ -126,8 +126,9 @@ struct dm_exception_store {
 };
 
 /*
- * Obtain the cow device used by a given snapshot.
+ * Obtain the origin or cow device used by a given snapshot.
  */
+struct dm_dev *dm_snap_origin(struct dm_snapshot *snap);
 struct dm_dev *dm_snap_cow(struct dm_snapshot *snap);
 
 /*
diff --git a/drivers/md/dm-ioctl.c b/drivers/md/dm-ioctl.c
index d7500e1c26f2..b96a908db0e7 100644
--- a/drivers/md/dm-ioctl.c
+++ b/drivers/md/dm-ioctl.c
@@ -285,19 +285,20 @@ retry:
 	up_write(&_hash_lock);
 }
 
-static int dm_hash_rename(uint32_t cookie, uint32_t *flags, const char *old,
-			  const char *new)
+static struct mapped_device *dm_hash_rename(struct dm_ioctl *param,
+					    const char *new)
 {
 	char *new_name, *old_name;
 	struct hash_cell *hc;
 	struct dm_table *table;
+	struct mapped_device *md;
 
 	/*
 	 * duplicate new.
 	 */
 	new_name = kstrdup(new, GFP_KERNEL);
 	if (!new_name)
-		return -ENOMEM;
+		return ERR_PTR(-ENOMEM);
 
 	down_write(&_hash_lock);
 
@@ -306,24 +307,24 @@ static int dm_hash_rename(uint32_t cookie, uint32_t *flags, const char *old,
 	 */
 	hc = __get_name_cell(new);
 	if (hc) {
-		DMWARN("asked to rename to an already existing name %s -> %s",
-		       old, new);
+		DMWARN("asked to rename to an already-existing name %s -> %s",
+		       param->name, new);
 		dm_put(hc->md);
 		up_write(&_hash_lock);
 		kfree(new_name);
-		return -EBUSY;
+		return ERR_PTR(-EBUSY);
 	}
 
 	/*
 	 * Is there such a device as 'old' ?
 	 */
-	hc = __get_name_cell(old);
+	hc = __get_name_cell(param->name);
 	if (!hc) {
-		DMWARN("asked to rename a non existent device %s -> %s",
-		       old, new);
+		DMWARN("asked to rename a non-existent device %s -> %s",
+		       param->name, new);
 		up_write(&_hash_lock);
 		kfree(new_name);
-		return -ENXIO;
+		return ERR_PTR(-ENXIO);
 	}
 
 	/*
@@ -345,13 +346,14 @@ static int dm_hash_rename(uint32_t cookie, uint32_t *flags, const char *old,
 		dm_table_put(table);
 	}
 
-	if (!dm_kobject_uevent(hc->md, KOBJ_CHANGE, cookie))
-		*flags |= DM_UEVENT_GENERATED_FLAG;
+	if (!dm_kobject_uevent(hc->md, KOBJ_CHANGE, param->event_nr))
+		param->flags |= DM_UEVENT_GENERATED_FLAG;
 
-	dm_put(hc->md);
+	md = hc->md;
 	up_write(&_hash_lock);
 	kfree(old_name);
-	return 0;
+
+	return md;
 }
 
 /*-----------------------------------------------------------------
@@ -573,7 +575,7 @@ static struct dm_table *dm_get_live_or_inactive_table(struct mapped_device *md,
  * Fills in a dm_ioctl structure, ready for sending back to
  * userland.
  */
-static int __dev_status(struct mapped_device *md, struct dm_ioctl *param)
+static void __dev_status(struct mapped_device *md, struct dm_ioctl *param)
 {
 	struct gendisk *disk = dm_disk(md);
 	struct dm_table *table;
@@ -617,8 +619,6 @@ static int __dev_status(struct mapped_device *md, struct dm_ioctl *param)
 			dm_table_put(table);
 		}
 	}
-
-	return 0;
 }
 
 static int dev_create(struct dm_ioctl *param, size_t param_size)
@@ -638,14 +638,14 @@ static int dev_create(struct dm_ioctl *param, size_t param_size)
 		return r;
 
 	r = dm_hash_insert(param->name, *param->uuid ? param->uuid : NULL, md);
-	if (r) {
-		dm_put(md);
-		return r;
-	}
+	if (r)
+		goto out;
 
 	param->flags &= ~DM_INACTIVE_PRESENT_FLAG;
 
-	r = __dev_status(md, param);
+	__dev_status(md, param);
+
+out:
 	dm_put(md);
 
 	return r;
@@ -762,6 +762,7 @@ static int dev_rename(struct dm_ioctl *param, size_t param_size)
 {
 	int r;
 	char *new_name = (char *) param + param->data_start;
+	struct mapped_device *md;
 
 	if (new_name < param->data ||
 	    invalid_str(new_name, (void *) param + param_size) ||
@@ -774,10 +775,14 @@ static int dev_rename(struct dm_ioctl *param, size_t param_size)
 	if (r)
 		return r;
 
-	param->data_size = 0;
+	md = dm_hash_rename(param, new_name);
+	if (IS_ERR(md))
+		return PTR_ERR(md);
+
+	__dev_status(md, param);
+	dm_put(md);
 
-	return dm_hash_rename(param->event_nr, &param->flags, param->name,
-			      new_name);
+	return 0;
 }
 
 static int dev_set_geometry(struct dm_ioctl *param, size_t param_size)
@@ -818,8 +823,6 @@ static int dev_set_geometry(struct dm_ioctl *param, size_t param_size)
 	geometry.start = indata[3];
 
 	r = dm_set_geometry(md, &geometry);
-	if (!r)
-		r = __dev_status(md, param);
 
 	param->data_size = 0;
 
@@ -843,13 +846,17 @@ static int do_suspend(struct dm_ioctl *param)
 	if (param->flags & DM_NOFLUSH_FLAG)
 		suspend_flags |= DM_SUSPEND_NOFLUSH_FLAG;
 
-	if (!dm_suspended_md(md))
+	if (!dm_suspended_md(md)) {
 		r = dm_suspend(md, suspend_flags);
+		if (r)
+			goto out;
+	}
 
-	if (!r)
-		r = __dev_status(md, param);
+	__dev_status(md, param);
 
+out:
 	dm_put(md);
+
 	return r;
 }
 
@@ -911,7 +918,7 @@ static int do_resume(struct dm_ioctl *param)
 		dm_table_destroy(old_map);
 
 	if (!r)
-		r = __dev_status(md, param);
+		__dev_status(md, param);
 
 	dm_put(md);
 	return r;
@@ -935,16 +942,16 @@ static int dev_suspend(struct dm_ioctl *param, size_t param_size)
  */
 static int dev_status(struct dm_ioctl *param, size_t param_size)
 {
-	int r;
 	struct mapped_device *md;
 
 	md = find_device(param);
 	if (!md)
 		return -ENXIO;
 
-	r = __dev_status(md, param);
+	__dev_status(md, param);
 	dm_put(md);
-	return r;
+
+	return 0;
 }
 
 /*
@@ -1019,7 +1026,7 @@ static void retrieve_status(struct dm_table *table,
  */
 static int dev_wait(struct dm_ioctl *param, size_t param_size)
 {
-	int r;
+	int r = 0;
 	struct mapped_device *md;
 	struct dm_table *table;
 
@@ -1040,9 +1047,7 @@ static int dev_wait(struct dm_ioctl *param, size_t param_size)
 	 * changed to trigger the event, so we may as well tell
 	 * him and save an ioctl.
 	 */
-	r = __dev_status(md, param);
-	if (r)
-		goto out;
+	__dev_status(md, param);
 
 	table = dm_get_live_or_inactive_table(md, param);
 	if (table) {
@@ -1050,8 +1055,9 @@ static int dev_wait(struct dm_ioctl *param, size_t param_size)
 		dm_table_put(table);
 	}
 
- out:
+out:
 	dm_put(md);
+
 	return r;
 }
 
@@ -1170,6 +1176,30 @@ static int table_load(struct dm_ioctl *param, size_t param_size)
 		goto out;
 	}
 
+	/* Protect md->type and md->queue against concurrent table loads. */
+	dm_lock_md_type(md);
+	if (dm_get_md_type(md) == DM_TYPE_NONE)
+		/* Initial table load: acquire type of table. */
+		dm_set_md_type(md, dm_table_get_type(t));
+	else if (dm_get_md_type(md) != dm_table_get_type(t)) {
+		DMWARN("can't change device type after initial table load.");
+		dm_table_destroy(t);
+		dm_unlock_md_type(md);
+		r = -EINVAL;
+		goto out;
+	}
+
+	/* setup md->queue to reflect md's type (may block) */
+	r = dm_setup_md_queue(md);
+	if (r) {
+		DMWARN("unable to set up device queue for new table.");
+		dm_table_destroy(t);
+		dm_unlock_md_type(md);
+		goto out;
+	}
+	dm_unlock_md_type(md);
+
+	/* stage inactive table */
 	down_write(&_hash_lock);
 	hc = dm_get_mdptr(md);
 	if (!hc || hc->md != md) {
@@ -1186,7 +1216,7 @@ static int table_load(struct dm_ioctl *param, size_t param_size)
 	up_write(&_hash_lock);
 
 	param->flags |= DM_INACTIVE_PRESENT_FLAG;
-	r = __dev_status(md, param);
+	__dev_status(md, param);
 
 out:
 	dm_put(md);
@@ -1196,7 +1226,6 @@ out:
 
 static int table_clear(struct dm_ioctl *param, size_t param_size)
 {
-	int r;
 	struct hash_cell *hc;
 	struct mapped_device *md;
 
@@ -1216,11 +1245,12 @@ static int table_clear(struct dm_ioctl *param, size_t param_size)
 
 	param->flags &= ~DM_INACTIVE_PRESENT_FLAG;
 
-	r = __dev_status(hc->md, param);
+	__dev_status(hc->md, param);
 	md = hc->md;
 	up_write(&_hash_lock);
 	dm_put(md);
-	return r;
+
+	return 0;
 }
 
 /*
@@ -1265,7 +1295,6 @@ static void retrieve_deps(struct dm_table *table,
 
 static int table_deps(struct dm_ioctl *param, size_t param_size)
 {
-	int r = 0;
 	struct mapped_device *md;
 	struct dm_table *table;
 
@@ -1273,9 +1302,7 @@ static int table_deps(struct dm_ioctl *param, size_t param_size)
 	if (!md)
 		return -ENXIO;
 
-	r = __dev_status(md, param);
-	if (r)
-		goto out;
+	__dev_status(md, param);
 
 	table = dm_get_live_or_inactive_table(md, param);
 	if (table) {
@@ -1283,9 +1310,9 @@ static int table_deps(struct dm_ioctl *param, size_t param_size)
 		dm_table_put(table);
 	}
 
- out:
 	dm_put(md);
-	return r;
+
+	return 0;
 }
 
 /*
@@ -1294,7 +1321,6 @@ static int table_deps(struct dm_ioctl *param, size_t param_size)
  */
 static int table_status(struct dm_ioctl *param, size_t param_size)
 {
-	int r;
 	struct mapped_device *md;
 	struct dm_table *table;
 
@@ -1302,9 +1328,7 @@ static int table_status(struct dm_ioctl *param, size_t param_size)
 	if (!md)
 		return -ENXIO;
 
-	r = __dev_status(md, param);
-	if (r)
-		goto out;
+	__dev_status(md, param);
 
 	table = dm_get_live_or_inactive_table(md, param);
 	if (table) {
@@ -1312,9 +1336,9 @@ static int table_status(struct dm_ioctl *param, size_t param_size)
 		dm_table_put(table);
 	}
 
-out:
 	dm_put(md);
-	return r;
+
+	return 0;
 }
 
 /*
@@ -1333,10 +1357,6 @@ static int target_message(struct dm_ioctl *param, size_t param_size)
 	if (!md)
 		return -ENXIO;
 
-	r = __dev_status(md, param);
-	if (r)
-		goto out;
-
 	if (tmsg < (struct dm_target_msg *) param->data ||
 	    invalid_str(tmsg->message, (void *) param + param_size)) {
 		DMWARN("Invalid target message parameters.");
@@ -1593,6 +1613,7 @@ static long dm_compat_ctl_ioctl(struct file *file, uint command, ulong u)
 #endif
 
 static const struct file_operations _ctl_fops = {
+	.open = nonseekable_open,
 	.unlocked_ioctl	 = dm_ctl_ioctl,
 	.compat_ioctl = dm_compat_ctl_ioctl,
 	.owner	 = THIS_MODULE,
diff --git a/drivers/md/dm-linear.c b/drivers/md/dm-linear.c
index 9200dbf2391a..7071f173a29c 100644
--- a/drivers/md/dm-linear.c
+++ b/drivers/md/dm-linear.c
@@ -152,6 +152,7 @@ static struct target_type linear_target = {
 	.ioctl  = linear_ioctl,
 	.merge  = linear_merge,
 	.iterate_devices = linear_iterate_devices,
+	.features = DM_TARGET_SUPPORTS_DISCARDS,
 };
 
 int __init dm_linear_init(void)
diff --git a/drivers/md/dm-mpath.c b/drivers/md/dm-mpath.c
index 826bce7343b3..da2223aa727f 100644
--- a/drivers/md/dm-mpath.c
+++ b/drivers/md/dm-mpath.c
@@ -706,6 +706,7 @@ static struct priority_group *parse_priority_group(struct arg_set *as,
 
 		if (as->argc < nr_params) {
 			ti->error = "not enough path parameters";
+			r = -EINVAL;
 			goto bad;
 		}
 
diff --git a/drivers/md/dm-snap-persistent.c b/drivers/md/dm-snap-persistent.c
index c097d8a4823d..cc2bdb83f9ad 100644
--- a/drivers/md/dm-snap-persistent.c
+++ b/drivers/md/dm-snap-persistent.c
@@ -266,7 +266,7 @@ static int chunk_io(struct pstore *ps, void *area, chunk_t chunk, int rw,
  */
 static chunk_t area_location(struct pstore *ps, chunk_t area)
 {
-	return 1 + ((ps->exceptions_per_area + 1) * area);
+	return NUM_SNAPSHOT_HDR_CHUNKS + ((ps->exceptions_per_area + 1) * area);
 }
 
 /*
@@ -780,8 +780,8 @@ static int persistent_commit_merge(struct dm_exception_store *store,
 	 * ps->current_area does not get reduced by prepare_merge() until
 	 * after commit_merge() has removed the nr_merged previous exceptions.
 	 */
-	ps->next_free = (area_location(ps, ps->current_area) - 1) +
-			(ps->current_committed + 1) + NUM_SNAPSHOT_HDR_CHUNKS;
+	ps->next_free = area_location(ps, ps->current_area) +
+			ps->current_committed + 1;
 
 	return 0;
 }
diff --git a/drivers/md/dm-snap.c b/drivers/md/dm-snap.c
index 54853773510c..96feada5e761 100644
--- a/drivers/md/dm-snap.c
+++ b/drivers/md/dm-snap.c
@@ -148,6 +148,12 @@ struct dm_snapshot {
 #define RUNNING_MERGE          0
 #define SHUTDOWN_MERGE         1
 
+struct dm_dev *dm_snap_origin(struct dm_snapshot *s)
+{
+	return s->origin;
+}
+EXPORT_SYMBOL(dm_snap_origin);
+
 struct dm_dev *dm_snap_cow(struct dm_snapshot *s)
 {
 	return s->cow;
@@ -1065,10 +1071,6 @@ static int snapshot_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 		origin_mode = FMODE_WRITE;
 	}
 
-	origin_path = argv[0];
-	argv++;
-	argc--;
-
 	s = kmalloc(sizeof(*s), GFP_KERNEL);
 	if (!s) {
 		ti->error = "Cannot allocate snapshot context private "
@@ -1077,6 +1079,16 @@ static int snapshot_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 		goto bad;
 	}
 
+	origin_path = argv[0];
+	argv++;
+	argc--;
+
+	r = dm_get_device(ti, origin_path, origin_mode, &s->origin);
+	if (r) {
+		ti->error = "Cannot get origin device";
+		goto bad_origin;
+	}
+
 	cow_path = argv[0];
 	argv++;
 	argc--;
@@ -1097,12 +1109,6 @@ static int snapshot_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 	argv += args_used;
 	argc -= args_used;
 
-	r = dm_get_device(ti, origin_path, origin_mode, &s->origin);
-	if (r) {
-		ti->error = "Cannot get origin device";
-		goto bad_origin;
-	}
-
 	s->ti = ti;
 	s->valid = 1;
 	s->active = 0;
@@ -1212,15 +1218,15 @@ bad_kcopyd:
 	dm_exception_table_exit(&s->complete, exception_cache);
 
 bad_hash_tables:
-	dm_put_device(ti, s->origin);
-
-bad_origin:
 	dm_exception_store_destroy(s->store);
 
 bad_store:
 	dm_put_device(ti, s->cow);
 
 bad_cow:
+	dm_put_device(ti, s->origin);
+
+bad_origin:
 	kfree(s);
 
 bad:
@@ -1314,12 +1320,12 @@ static void snapshot_dtr(struct dm_target *ti)
 
 	mempool_destroy(s->pending_pool);
 
-	dm_put_device(ti, s->origin);
-
 	dm_exception_store_destroy(s->store);
 
 	dm_put_device(ti, s->cow);
 
+	dm_put_device(ti, s->origin);
+
 	kfree(s);
 }
 
@@ -1899,8 +1905,14 @@ static int snapshot_iterate_devices(struct dm_target *ti,
 				    iterate_devices_callout_fn fn, void *data)
 {
 	struct dm_snapshot *snap = ti->private;
+	int r;
+
+	r = fn(ti, snap->origin, 0, ti->len, data);
+
+	if (!r)
+		r = fn(ti, snap->cow, 0, get_dev_size(snap->cow->bdev), data);
 
-	return fn(ti, snap->origin, 0, ti->len, data);
+	return r;
 }
 
 
@@ -2159,6 +2171,21 @@ static int origin_status(struct dm_target *ti, status_type_t type, char *result,
 	return 0;
 }
 
+static int origin_merge(struct dm_target *ti, struct bvec_merge_data *bvm,
+			struct bio_vec *biovec, int max_size)
+{
+	struct dm_dev *dev = ti->private;
+	struct request_queue *q = bdev_get_queue(dev->bdev);
+
+	if (!q->merge_bvec_fn)
+		return max_size;
+
+	bvm->bi_bdev = dev->bdev;
+	bvm->bi_sector = bvm->bi_sector;
+
+	return min(max_size, q->merge_bvec_fn(q, bvm, biovec));
+}
+
 static int origin_iterate_devices(struct dm_target *ti,
 				  iterate_devices_callout_fn fn, void *data)
 {
@@ -2176,6 +2203,7 @@ static struct target_type origin_target = {
 	.map     = origin_map,
 	.resume  = origin_resume,
 	.status  = origin_status,
+	.merge	 = origin_merge,
 	.iterate_devices = origin_iterate_devices,
 };
 
diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index 9924ea23032d..3f6d6b4ef60a 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -54,6 +54,8 @@ struct dm_table {
 	sector_t *highs;
 	struct dm_target *targets;
 
+	unsigned discards_supported:1;
+
 	/*
 	 * Indicates the rw permissions for the new logical
 	 * device.  This should be a combination of FMODE_READ
@@ -203,6 +205,7 @@ int dm_table_create(struct dm_table **result, fmode_t mode,
 
 	INIT_LIST_HEAD(&t->devices);
 	atomic_set(&t->holders, 0);
+	t->discards_supported = 1;
 
 	if (!num_targets)
 		num_targets = KEYS_PER_NODE;
@@ -770,6 +773,9 @@ int dm_table_add_target(struct dm_table *t, const char *type,
 
 	t->highs[t->num_targets++] = tgt->begin + tgt->len - 1;
 
+	if (!(tgt->type->features & DM_TARGET_SUPPORTS_DISCARDS))
+		t->discards_supported = 0;
+
 	return 0;
 
  bad:
@@ -1086,6 +1092,11 @@ void dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
 	else
 		queue_flag_set_unlocked(QUEUE_FLAG_CLUSTER, q);
 
+	if (!dm_table_supports_discards(t))
+		queue_flag_clear_unlocked(QUEUE_FLAG_DISCARD, q);
+	else
+		queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, q);
+
 	dm_table_set_integrity(t);
 
 	/*
@@ -1232,6 +1243,39 @@ struct mapped_device *dm_table_get_md(struct dm_table *t)
 	return t->md;
 }
 
+static int device_discard_capable(struct dm_target *ti, struct dm_dev *dev,
+				  sector_t start, sector_t len, void *data)
+{
+	struct request_queue *q = bdev_get_queue(dev->bdev);
+
+	return q && blk_queue_discard(q);
+}
+
+bool dm_table_supports_discards(struct dm_table *t)
+{
+	struct dm_target *ti;
+	unsigned i = 0;
+
+	if (!t->discards_supported)
+		return 0;
+
+	/*
+	 * Ensure that at least one underlying device supports discards.
+	 * t->devices includes internal dm devices such as mirror logs
+	 * so we need to use iterate_devices here, which targets
+	 * supporting discard must provide.
+	 */
+	while (i < dm_table_get_num_targets(t)) {
+		ti = dm_table_get_target(t, i++);
+
+		if (ti->type->iterate_devices &&
+		    ti->type->iterate_devices(ti, device_discard_capable, NULL))
+			return 1;
+	}
+
+	return 0;
+}
+
 EXPORT_SYMBOL(dm_vcalloc);
 EXPORT_SYMBOL(dm_get_device);
 EXPORT_SYMBOL(dm_put_device);
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index a3f21dc02bd8..1ca2b39a3a99 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -124,6 +124,10 @@ struct mapped_device {
 	unsigned long flags;
 
 	struct request_queue *queue;
+	unsigned type;
+	/* Protect queue and type against concurrent access. */
+	struct mutex type_lock;
+
 	struct gendisk *disk;
 	char name[16];
 
@@ -638,8 +642,14 @@ static void dec_pending(struct dm_io *io, int error)
 			 * There can be just one barrier request so we use
 			 * a per-device variable for error reporting.
 			 * Note that you can't touch the bio after end_io_acct
+			 *
+			 * We ignore -EOPNOTSUPP for empty flush reported by
+			 * underlying devices. We assume that if the device
+			 * doesn't support empty barriers, it doesn't need
+			 * cache flushing commands.
 			 */
-			if (!md->barrier_error && io_error != -EOPNOTSUPP)
+			if (!md->barrier_error &&
+			    !(bio_empty_barrier(bio) && io_error == -EOPNOTSUPP))
 				md->barrier_error = io_error;
 			end_io_acct(io);
 			free_io(md, io);
@@ -1201,6 +1211,53 @@ static int __clone_and_map_empty_barrier(struct clone_info *ci)
 	return 0;
 }
 
+/*
+ * Perform all io with a single clone.
+ */
+static void __clone_and_map_simple(struct clone_info *ci, struct dm_target *ti)
+{
+	struct bio *clone, *bio = ci->bio;
+	struct dm_target_io *tio;
+
+	tio = alloc_tio(ci, ti);
+	clone = clone_bio(bio, ci->sector, ci->idx,
+			  bio->bi_vcnt - ci->idx, ci->sector_count,
+			  ci->md->bs);
+	__map_bio(ti, clone, tio);
+	ci->sector_count = 0;
+}
+
+static int __clone_and_map_discard(struct clone_info *ci)
+{
+	struct dm_target *ti;
+	sector_t max;
+
+	ti = dm_table_find_target(ci->map, ci->sector);
+	if (!dm_target_is_valid(ti))
+		return -EIO;
+
+	/*
+	 * Even though the device advertised discard support,
+	 * reconfiguration might have changed that since the
+	 * check was performed.
+	 */
+
+	if (!(ti->type->features & DM_TARGET_SUPPORTS_DISCARDS))
+		return -EOPNOTSUPP;
+
+	max = max_io_len(ci->md, ci->sector, ti);
+
+	if (ci->sector_count > max)
+		/*
+		 * FIXME: Handle a discard that spans two or more targets.
+		 */
+		return -EOPNOTSUPP;
+
+	__clone_and_map_simple(ci, ti);
+
+	return 0;
+}
+
 static int __clone_and_map(struct clone_info *ci)
 {
 	struct bio *clone, *bio = ci->bio;
@@ -1211,27 +1268,21 @@ static int __clone_and_map(struct clone_info *ci)
 	if (unlikely(bio_empty_barrier(bio)))
 		return __clone_and_map_empty_barrier(ci);
 
+	if (unlikely(bio->bi_rw & REQ_DISCARD))
+		return __clone_and_map_discard(ci);
+
 	ti = dm_table_find_target(ci->map, ci->sector);
 	if (!dm_target_is_valid(ti))
 		return -EIO;
 
 	max = max_io_len(ci->md, ci->sector, ti);
 
-	/*
-	 * Allocate a target io object.
-	 */
-	tio = alloc_tio(ci, ti);
-
 	if (ci->sector_count <= max) {
 		/*
 		 * Optimise for the simple case where we can do all of
 		 * the remaining io with a single clone.
 		 */
-		clone = clone_bio(bio, ci->sector, ci->idx,
-				  bio->bi_vcnt - ci->idx, ci->sector_count,
-				  ci->md->bs);
-		__map_bio(ti, clone, tio);
-		ci->sector_count = 0;
+		__clone_and_map_simple(ci, ti);
 
 	} else if (to_sector(bio->bi_io_vec[ci->idx].bv_len) <= max) {
 		/*
@@ -1252,6 +1303,7 @@ static int __clone_and_map(struct clone_info *ci)
 			len += bv_len;
 		}
 
+		tio = alloc_tio(ci, ti);
 		clone = clone_bio(bio, ci->sector, ci->idx, i - ci->idx, len,
 				  ci->md->bs);
 		__map_bio(ti, clone, tio);
@@ -1275,12 +1327,11 @@ static int __clone_and_map(struct clone_info *ci)
 					return -EIO;
 
 				max = max_io_len(ci->md, ci->sector, ti);
-
-				tio = alloc_tio(ci, ti);
 			}
 
 			len = min(remaining, max);
 
+			tio = alloc_tio(ci, ti);
 			clone = split_bvec(bio, ci->sector, ci->idx,
 					   bv->bv_offset + offset, len,
 					   ci->md->bs);
@@ -1845,6 +1896,28 @@ static const struct block_device_operations dm_blk_dops;
 static void dm_wq_work(struct work_struct *work);
 static void dm_rq_barrier_work(struct work_struct *work);
 
+static void dm_init_md_queue(struct mapped_device *md)
+{
+	/*
+	 * Request-based dm devices cannot be stacked on top of bio-based dm
+	 * devices.  The type of this dm device has not been decided yet.
+	 * The type is decided at the first table loading time.
+	 * To prevent problematic device stacking, clear the queue flag
+	 * for request stacking support until then.
+	 *
+	 * This queue is new, so no concurrency on the queue_flags.
+	 */
+	queue_flag_clear_unlocked(QUEUE_FLAG_STACKABLE, md->queue);
+
+	md->queue->queuedata = md;
+	md->queue->backing_dev_info.congested_fn = dm_any_congested;
+	md->queue->backing_dev_info.congested_data = md;
+	blk_queue_make_request(md->queue, dm_request);
+	blk_queue_bounce_limit(md->queue, BLK_BOUNCE_ANY);
+	md->queue->unplug_fn = dm_unplug_all;
+	blk_queue_merge_bvec(md->queue, dm_merge_bvec);
+}
+
 /*
  * Allocate and initialise a blank device with a given minor.
  */
@@ -1870,8 +1943,10 @@ static struct mapped_device *alloc_dev(int minor)
 	if (r < 0)
 		goto bad_minor;
 
+	md->type = DM_TYPE_NONE;
 	init_rwsem(&md->io_lock);
 	mutex_init(&md->suspend_lock);
+	mutex_init(&md->type_lock);
 	spin_lock_init(&md->deferred_lock);
 	spin_lock_init(&md->barrier_error_lock);
 	rwlock_init(&md->map_lock);
@@ -1882,33 +1957,11 @@ static struct mapped_device *alloc_dev(int minor)
 	INIT_LIST_HEAD(&md->uevent_list);
 	spin_lock_init(&md->uevent_lock);
 
-	md->queue = blk_init_queue(dm_request_fn, NULL);
+	md->queue = blk_alloc_queue(GFP_KERNEL);
 	if (!md->queue)
 		goto bad_queue;
 
-	/*
-	 * Request-based dm devices cannot be stacked on top of bio-based dm
-	 * devices.  The type of this dm device has not been decided yet,
-	 * although we initialized the queue using blk_init_queue().
-	 * The type is decided at the first table loading time.
-	 * To prevent problematic device stacking, clear the queue flag
-	 * for request stacking support until then.
-	 *
-	 * This queue is new, so no concurrency on the queue_flags.
-	 */
-	queue_flag_clear_unlocked(QUEUE_FLAG_STACKABLE, md->queue);
-	md->saved_make_request_fn = md->queue->make_request_fn;
-	md->queue->queuedata = md;
-	md->queue->backing_dev_info.congested_fn = dm_any_congested;
-	md->queue->backing_dev_info.congested_data = md;
-	blk_queue_make_request(md->queue, dm_request);
-	blk_queue_bounce_limit(md->queue, BLK_BOUNCE_ANY);
-	md->queue->unplug_fn = dm_unplug_all;
-	blk_queue_merge_bvec(md->queue, dm_merge_bvec);
-	blk_queue_softirq_done(md->queue, dm_softirq_done);
-	blk_queue_prep_rq(md->queue, dm_prep_fn);
-	blk_queue_lld_busy(md->queue, dm_lld_busy);
-	blk_queue_ordered(md->queue, QUEUE_ORDERED_DRAIN_FLUSH);
+	dm_init_md_queue(md);
 
 	md->disk = alloc_disk(1);
 	if (!md->disk)
@@ -2123,6 +2176,72 @@ int dm_create(int minor, struct mapped_device **result)
 	return 0;
 }
 
+/*
+ * Functions to manage md->type.
+ * All are required to hold md->type_lock.
+ */
+void dm_lock_md_type(struct mapped_device *md)
+{
+	mutex_lock(&md->type_lock);
+}
+
+void dm_unlock_md_type(struct mapped_device *md)
+{
+	mutex_unlock(&md->type_lock);
+}
+
+void dm_set_md_type(struct mapped_device *md, unsigned type)
+{
+	md->type = type;
+}
+
+unsigned dm_get_md_type(struct mapped_device *md)
+{
+	return md->type;
+}
+
+/*
+ * Fully initialize a request-based queue (->elevator, ->request_fn, etc).
+ */
+static int dm_init_request_based_queue(struct mapped_device *md)
+{
+	struct request_queue *q = NULL;
+
+	if (md->queue->elevator)
+		return 1;
+
+	/* Fully initialize the queue */
+	q = blk_init_allocated_queue(md->queue, dm_request_fn, NULL);
+	if (!q)
+		return 0;
+
+	md->queue = q;
+	md->saved_make_request_fn = md->queue->make_request_fn;
+	dm_init_md_queue(md);
+	blk_queue_softirq_done(md->queue, dm_softirq_done);
+	blk_queue_prep_rq(md->queue, dm_prep_fn);
+	blk_queue_lld_busy(md->queue, dm_lld_busy);
+	blk_queue_ordered(md->queue, QUEUE_ORDERED_DRAIN_FLUSH);
+
+	elv_register_queue(md->queue);
+
+	return 1;
+}
+
+/*
+ * Setup the DM device's queue based on md's type
+ */
+int dm_setup_md_queue(struct mapped_device *md)
+{
+	if ((dm_get_md_type(md) == DM_TYPE_REQUEST_BASED) &&
+	    !dm_init_request_based_queue(md)) {
+		DMWARN("Cannot initialize queue for request-based mapped device");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
 static struct mapped_device *dm_find_md(dev_t dev)
 {
 	struct mapped_device *md;
@@ -2253,7 +2372,12 @@ static void process_barrier(struct mapped_device *md, struct bio *bio)
 
 	if (!bio_empty_barrier(bio)) {
 		__split_and_process_bio(md, bio);
-		dm_flush(md);
+		/*
+		 * If the request isn't supported, don't waste time with
+		 * the second flush.
+		 */
+		if (md->barrier_error != -EOPNOTSUPP)
+			dm_flush(md);
 	}
 
 	if (md->barrier_error != DM_ENDIO_REQUEUE)
@@ -2398,13 +2522,6 @@ struct dm_table *dm_swap_table(struct mapped_device *md, struct dm_table *table)
 		goto out;
 	}
 
-	/* cannot change the device type, once a table is bound */
-	if (md->map &&
-	    (dm_table_get_type(md->map) != dm_table_get_type(table))) {
-		DMWARN("can't change the device type after a table is bound");
-		goto out;
-	}
-
 	map = __bind(md, table, &limits);
 
 out:
diff --git a/drivers/md/dm.h b/drivers/md/dm.h
index bad1724d4869..032ce2f1ed35 100644
--- a/drivers/md/dm.h
+++ b/drivers/md/dm.h
@@ -62,10 +62,18 @@ int dm_table_any_busy_target(struct dm_table *t);
 int dm_table_set_type(struct dm_table *t);
 unsigned dm_table_get_type(struct dm_table *t);
 bool dm_table_request_based(struct dm_table *t);
+bool dm_table_supports_discards(struct dm_table *t);
 int dm_table_alloc_md_mempools(struct dm_table *t);
 void dm_table_free_md_mempools(struct dm_table *t);
 struct dm_md_mempools *dm_table_get_md_mempools(struct dm_table *t);
 
+void dm_lock_md_type(struct mapped_device *md);
+void dm_unlock_md_type(struct mapped_device *md);
+void dm_set_md_type(struct mapped_device *md, unsigned type);
+unsigned dm_get_md_type(struct mapped_device *md);
+
+int dm_setup_md_queue(struct mapped_device *md);
+
 /*
  * To check the return value from dm_table_find_target().
  */