Merge tag 'dm-4.4-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm

Pull device mapper updates from Mike Snitzer:
"Smaller set of DM changes for this merge. I've based these changes on
Jens' for-4.4/reservations branch because the associated DM changes
required it.

- Revert a dm-multipath change that caused a regression for
unprivledged users (e.g. kvm guests) that issued ioctls when a
multipath device had no available paths.

- Include Christoph's refactoring of DM's ioctl handling and add
support for passing through persistent reservations with DM
multipath.

- All other changes are very simple cleanups"

* tag 'dm-4.4-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm:
dm switch: simplify conditional in alloc_region_table()
dm delay: document that offsets are specified in sectors
dm delay: capitalize the start of an delay_ctr() error message
dm delay: Use DM_MAPIO macros instead of open-coded equivalents
dm linear: remove redundant target name from error messages
dm persistent data: eliminate unnecessary return values
dm: eliminate unused "bioset" process for each bio-based DM device
dm: convert ffs to __ffs
dm: drop NULL test before kmem_cache_destroy() and mempool_destroy()
dm: add support for passing through persistent reservations
dm: refactor ioctl handling
Revert "dm mpath: fix stalls when handling invalid ioctls"
dm: initialize non-blk-mq queue data before queue is used

Documentation/device-mapper/delay.txt

···
     <device> <offset> <delay> [<write_device> <write_offset> <write_delay>]
 
 With separate write parameters, the first set is only used for reads.
+Offsets are specified in sectors.
 Delays are specified in milliseconds.
 
 Example scripts

drivers/md/dm-bufio.c

···
 
 	c->bdev = bdev;
 	c->block_size = block_size;
-	c->sectors_per_block_bits = ffs(block_size) - 1 - SECTOR_SHIFT;
-	c->pages_per_block_bits = (ffs(block_size) - 1 >= PAGE_SHIFT) ?
-				  ffs(block_size) - 1 - PAGE_SHIFT : 0;
-	c->blocks_per_page_bits = (ffs(block_size) - 1 < PAGE_SHIFT ?
-				  PAGE_SHIFT - (ffs(block_size) - 1) : 0);
+	c->sectors_per_block_bits = __ffs(block_size) - SECTOR_SHIFT;
+	c->pages_per_block_bits = (__ffs(block_size) >= PAGE_SHIFT) ?
+				  __ffs(block_size) - PAGE_SHIFT : 0;
+	c->blocks_per_page_bits = (__ffs(block_size) < PAGE_SHIFT ?
+				  PAGE_SHIFT - __ffs(block_size) : 0);
 
 	c->aux_size = aux_size;
 	c->alloc_callback = alloc_callback;
···
 	cancel_delayed_work_sync(&dm_bufio_work);
 	destroy_workqueue(dm_bufio_wq);
 
-	for (i = 0; i < ARRAY_SIZE(dm_bufio_caches); i++) {
-		struct kmem_cache *kc = dm_bufio_caches[i];
-
-		if (kc)
-			kmem_cache_destroy(kc);
-	}
+	for (i = 0; i < ARRAY_SIZE(dm_bufio_caches); i++)
+		kmem_cache_destroy(dm_bufio_caches[i]);
 
 	for (i = 0; i < ARRAY_SIZE(dm_bufio_cache_names); i++)
 		kfree(dm_bufio_cache_names[i]);

drivers/md/dm-cache-metadata.c

···
 		}
 	}
 
-	return dm_bm_unlock(b);
+	dm_bm_unlock(b);
+
+	return 0;
 }
 
 static void __setup_mapping_info(struct dm_cache_metadata *cmd)
···
 	dm_disk_bitset_init(cmd->tm, &cmd->discard_info);
 	sb_flags = le32_to_cpu(disk_super->flags);
 	cmd->clean_when_opened = test_bit(CLEAN_SHUTDOWN, &sb_flags);
-	return dm_bm_unlock(sblock);
+	dm_bm_unlock(sblock);
+
+	return 0;
 
 bad:
 	dm_bm_unlock(sblock);

drivers/md/dm-cache-policy-cleaner.c

···
 static int alloc_hash(struct hash *hash, unsigned elts)
 {
 	hash->nr_buckets = next_power(elts >> 4, 16);
-	hash->hash_bits = ffs(hash->nr_buckets) - 1;
+	hash->hash_bits = __ffs(hash->nr_buckets);
 	hash->table = vzalloc(sizeof(*hash->table) * hash->nr_buckets);
 
 	return hash->table ? 0 : -ENOMEM;

drivers/md/dm-cache-policy-mq.c

···
 	mq->generation_period = max((unsigned) from_cblock(cache_size), 1024U);
 
 	mq->nr_buckets = next_power(from_cblock(cache_size) / 2, 16);
-	mq->hash_bits = ffs(mq->nr_buckets) - 1;
+	mq->hash_bits = __ffs(mq->nr_buckets);
 	mq->table = vzalloc(sizeof(*mq->table) * mq->nr_buckets);
 	if (!mq->table)
 		goto bad_alloc_table;

drivers/md/dm-cache-policy-smq.c

···
 
 	ht->es = es;
 	nr_buckets = roundup_pow_of_two(max(nr_entries / 4u, 16u));
-	ht->hash_bits = ffs(nr_buckets) - 1;
+	ht->hash_bits = __ffs(nr_buckets);
 
 	ht->buckets = vmalloc(sizeof(*ht->buckets) * nr_buckets);
 	if (!ht->buckets)

drivers/md/dm-cache-target.c

···
 {
 	unsigned i;
 
-	if (cache->migration_pool)
-		mempool_destroy(cache->migration_pool);
+	mempool_destroy(cache->migration_pool);
 
 	if (cache->all_io_ds)
 		dm_deferred_set_destroy(cache->all_io_ds);

drivers/md/dm-crypt.c

···
 	if (cc->bs)
 		bioset_free(cc->bs);
 
-	if (cc->page_pool)
-		mempool_destroy(cc->page_pool);
-	if (cc->req_pool)
-		mempool_destroy(cc->req_pool);
+	mempool_destroy(cc->page_pool);
+	mempool_destroy(cc->req_pool);
 
 	if (cc->iv_gen_ops && cc->iv_gen_ops->dtr)
 		cc->iv_gen_ops->dtr(cc);

drivers/md/dm-delay.c

···
  *    <device> <offset> <delay> [<write_device> <write_offset> <write_delay>]
  *
  * With separate write parameters, the first set is only used for reads.
+ * Offsets are specified in sectors.
  * Delays are specified in milliseconds.
  */
 static int delay_ctr(struct dm_target *ti, unsigned int argc, char **argv)
···
 	int ret;
 
 	if (argc != 3 && argc != 6) {
-		ti->error = "requires exactly 3 or 6 arguments";
+		ti->error = "Requires exactly 3 or 6 arguments";
 		return -EINVAL;
 	}
 
···
 	unsigned long expires = 0;
 
 	if (!delay || !atomic_read(&dc->may_delay))
-		return 1;
+		return DM_MAPIO_REMAPPED;
 
 	delayed = dm_per_bio_data(bio, sizeof(struct dm_delay_info));
 
···
 
 	queue_timeout(dc, expires);
 
-	return 0;
+	return DM_MAPIO_SUBMITTED;
 }
 
 static void delay_presuspend(struct dm_target *ti)

drivers/md/dm-era-target.c

···
 		}
 	}
 
-	return dm_bm_unlock(b);
+	dm_bm_unlock(b);
+
+	return 0;
 }
 
 /*----------------------------------------------------------------*/
···
 	md->metadata_snap = le64_to_cpu(disk->metadata_snap);
 	md->archived_writesets = true;
 
-	return dm_bm_unlock(sblock);
+	dm_bm_unlock(sblock);
+
+	return 0;
 
 bad:
 	dm_bm_unlock(sblock);
···
 
 	md->metadata_snap = dm_block_location(clone);
 
-	r = dm_tm_unlock(md->tm, clone);
-	if (r) {
-		DMERR("%s: couldn't unlock clone", __func__);
-		md->metadata_snap = SUPERBLOCK_LOCATION;
-		return r;
-	}
+	dm_tm_unlock(md->tm, clone);
 
 	return 0;
 }

drivers/md/dm-exception-store.c

···
 
 	store->chunk_size = chunk_size;
 	store->chunk_mask = chunk_size - 1;
-	store->chunk_shift = ffs(chunk_size) - 1;
+	store->chunk_shift = __ffs(chunk_size);
 
 	return 0;
 }

drivers/md/dm-flakey.c

···
 	}
 }
 
-static int flakey_ioctl(struct dm_target *ti, unsigned int cmd, unsigned long arg)
+static int flakey_prepare_ioctl(struct dm_target *ti,
+		struct block_device **bdev, fmode_t *mode)
 {
 	struct flakey_c *fc = ti->private;
-	struct dm_dev *dev = fc->dev;
-	int r = 0;
+
+	*bdev = fc->dev->bdev;
 
 	/*
 	 * Only pass ioctls through if the device sizes match exactly.
 	 */
 	if (fc->start ||
-	    ti->len != i_size_read(dev->bdev->bd_inode) >> SECTOR_SHIFT)
-		r = scsi_verify_blk_ioctl(NULL, cmd);
-
-	return r ? : __blkdev_driver_ioctl(dev->bdev, dev->mode, cmd, arg);
+	    ti->len != i_size_read((*bdev)->bd_inode) >> SECTOR_SHIFT)
+		return 1;
+	return 0;
 }
 
 static int flakey_iterate_devices(struct dm_target *ti, iterate_devices_callout_fn fn, void *data)
···
 	.map    = flakey_map,
 	.end_io = flakey_end_io,
 	.status = flakey_status,
-	.ioctl	= flakey_ioctl,
+	.prepare_ioctl = flakey_prepare_ioctl,
 	.iterate_devices = flakey_iterate_devices,
 };
 

drivers/md/dm-io.c

···
 	return client;
 
    bad:
-	if (client->pool)
-		mempool_destroy(client->pool);
+	mempool_destroy(client->pool);
 	kfree(client);
 	return ERR_PTR(-ENOMEM);
 }

drivers/md/dm-linear.c

···
 
 	lc = kmalloc(sizeof(*lc), GFP_KERNEL);
 	if (lc == NULL) {
-		ti->error = "dm-linear: Cannot allocate linear context";
+		ti->error = "Cannot allocate linear context";
 		return -ENOMEM;
 	}
 
 	ret = -EINVAL;
 	if (sscanf(argv[1], "%llu%c", &tmp, &dummy) != 1) {
-		ti->error = "dm-linear: Invalid device sector";
+		ti->error = "Invalid device sector";
 		goto bad;
 	}
 	lc->start = tmp;
 
 	ret = dm_get_device(ti, argv[0], dm_table_get_mode(ti->table), &lc->dev);
 	if (ret) {
-		ti->error = "dm-linear: Device lookup failed";
+		ti->error = "Device lookup failed";
 		goto bad;
 	}
 
···
 	}
 }
 
-static int linear_ioctl(struct dm_target *ti, unsigned int cmd,
-			unsigned long arg)
+static int linear_prepare_ioctl(struct dm_target *ti,
+		struct block_device **bdev, fmode_t *mode)
 {
 	struct linear_c *lc = (struct linear_c *) ti->private;
 	struct dm_dev *dev = lc->dev;
-	int r = 0;
+
+	*bdev = dev->bdev;
 
 	/*
 	 * Only pass ioctls through if the device sizes match exactly.
 	 */
 	if (lc->start ||
 	    ti->len != i_size_read(dev->bdev->bd_inode) >> SECTOR_SHIFT)
-		r = scsi_verify_blk_ioctl(NULL, cmd);
-
-	return r ? : __blkdev_driver_ioctl(dev->bdev, dev->mode, cmd, arg);
+		return 1;
+	return 0;
 }
 
 static int linear_iterate_devices(struct dm_target *ti,
···
 	.dtr    = linear_dtr,
 	.map    = linear_map,
 	.status = linear_status,
-	.ioctl  = linear_ioctl,
+	.prepare_ioctl = linear_prepare_ioctl,
 	.iterate_devices = linear_iterate_devices,
 };
 

drivers/md/dm-log-userspace-base.c

···
 out:
 	kfree(devices_rdata);
 	if (r) {
-		if (lc->flush_entry_pool)
-			mempool_destroy(lc->flush_entry_pool);
+		mempool_destroy(lc->flush_entry_pool);
 		kfree(lc);
 		kfree(ctr_str);
 	} else {

drivers/md/dm-log-writes.c

···
 	}
 }
 
-static int log_writes_ioctl(struct dm_target *ti, unsigned int cmd,
-			    unsigned long arg)
+static int log_writes_prepare_ioctl(struct dm_target *ti,
+		struct block_device **bdev, fmode_t *mode)
 {
 	struct log_writes_c *lc = ti->private;
 	struct dm_dev *dev = lc->dev;
-	int r = 0;
 
+	*bdev = dev->bdev;
 	/*
 	 * Only pass ioctls through if the device sizes match exactly.
 	 */
 	if (ti->len != i_size_read(dev->bdev->bd_inode) >> SECTOR_SHIFT)
-		r = scsi_verify_blk_ioctl(NULL, cmd);
-
-	return r ? : __blkdev_driver_ioctl(dev->bdev, dev->mode, cmd, arg);
+		return 1;
+	return 0;
 }
 
 static int log_writes_iterate_devices(struct dm_target *ti,
···
 	.map    = log_writes_map,
 	.end_io = normal_end_io,
 	.status = log_writes_status,
-	.ioctl	= log_writes_ioctl,
+	.prepare_ioctl = log_writes_prepare_ioctl,
 	.message = log_writes_message,
 	.iterate_devices = log_writes_iterate_devices,
 	.io_hints = log_writes_io_hints,

drivers/md/dm-mpath.c

···
 	return r;
 }
 
-static int multipath_ioctl(struct dm_target *ti, unsigned int cmd,
-			   unsigned long arg)
+static int multipath_prepare_ioctl(struct dm_target *ti,
+		struct block_device **bdev, fmode_t *mode)
 {
 	struct multipath *m = ti->private;
 	struct pgpath *pgpath;
-	struct block_device *bdev;
-	fmode_t mode;
 	unsigned long flags;
 	int r;
 
-	bdev = NULL;
-	mode = 0;
 	r = 0;
 
 	spin_lock_irqsave(&m->lock, flags);
···
 	pgpath = m->current_pgpath;
 
 	if (pgpath) {
-		bdev = pgpath->path.dev->bdev;
-		mode = pgpath->path.dev->mode;
+		*bdev = pgpath->path.dev->bdev;
+		*mode = pgpath->path.dev->mode;
 	}
 
 	if ((pgpath && m->queue_io) || (!pgpath && m->queue_if_no_path))
 		r = -ENOTCONN;
-	else if (!bdev)
+	else if (!*bdev)
 		r = -EIO;
 
 	spin_unlock_irqrestore(&m->lock, flags);
-
-	/*
-	 * Only pass ioctls through if the device sizes match exactly.
-	 */
-	if (!bdev || ti->len != i_size_read(bdev->bd_inode) >> SECTOR_SHIFT) {
-		int err = scsi_verify_blk_ioctl(NULL, cmd);
-		if (err)
-			r = err;
-	}
 
 	if (r == -ENOTCONN && !fatal_signal_pending(current)) {
 		spin_lock_irqsave(&m->lock, flags);
···
 		dm_table_run_md_queue_async(m->ti->table);
 	}
 
-	return r ? : __blkdev_driver_ioctl(bdev, mode, cmd, arg);
+	/*
+	 * Only pass ioctls through if the device sizes match exactly.
+	 */
+	if (!r && ti->len != i_size_read((*bdev)->bd_inode) >> SECTOR_SHIFT)
+		return 1;
+	return r;
 }
 
 static int multipath_iterate_devices(struct dm_target *ti,
···
  *---------------------------------------------------------------*/
 static struct target_type multipath_target = {
 	.name = "multipath",
-	.version = {1, 9, 0},
+	.version = {1, 10, 0},
 	.module = THIS_MODULE,
 	.ctr = multipath_ctr,
 	.dtr = multipath_dtr,
···
 	.resume = multipath_resume,
 	.status = multipath_status,
 	.message = multipath_message,
-	.ioctl  = multipath_ioctl,
+	.prepare_ioctl = multipath_prepare_ioctl,
 	.iterate_devices = multipath_iterate_devices,
 	.busy = multipath_busy,
 };

drivers/md/dm-region-hash.c

···
 	rh->max_recovery = max_recovery;
 	rh->log = log;
 	rh->region_size = region_size;
-	rh->region_shift = ffs(region_size) - 1;
+	rh->region_shift = __ffs(region_size);
 	rwlock_init(&rh->hash_lock);
 	rh->mask = nr_buckets - 1;
 	rh->nr_buckets = nr_buckets;
···
 	if (rh->log)
 		dm_dirty_log_destroy(rh->log);
 
-	if (rh->region_pool)
-		mempool_destroy(rh->region_pool);
-
+	mempool_destroy(rh->region_pool);
 	vfree(rh->buckets);
 	kfree(rh);
 }

drivers/md/dm-snap-persistent.c

···
 		    bdev_logical_block_size(dm_snap_cow(ps->store->snap)->
 					    bdev) >> 9);
 		ps->store->chunk_mask = ps->store->chunk_size - 1;
-		ps->store->chunk_shift = ffs(ps->store->chunk_size) - 1;
+		ps->store->chunk_shift = __ffs(ps->store->chunk_size);
 		chunk_size_supplied = 0;
 	}
 

drivers/md/dm-switch.c

···
 	if (sector_div(nr_regions, sctx->region_size))
 		nr_regions++;
 
-	sctx->nr_regions = nr_regions;
-	if (sctx->nr_regions != nr_regions || sctx->nr_regions >= ULONG_MAX) {
+	if (nr_regions >= ULONG_MAX) {
 		ti->error = "Region table too large";
 		return -EINVAL;
 	}
+	sctx->nr_regions = nr_regions;
 
 	nr_slots = nr_regions;
 	if (sector_div(nr_slots, sctx->region_entries_per_slot))
···
  *
  * Passthrough all ioctls to the path for sector 0
  */
-static int switch_ioctl(struct dm_target *ti, unsigned cmd,
-			unsigned long arg)
+static int switch_prepare_ioctl(struct dm_target *ti,
+		struct block_device **bdev, fmode_t *mode)
 {
 	struct switch_ctx *sctx = ti->private;
-	struct block_device *bdev;
-	fmode_t mode;
 	unsigned path_nr;
-	int r = 0;
 
 	path_nr = switch_get_path_nr(sctx, 0);
 
-	bdev = sctx->path_list[path_nr].dmdev->bdev;
-	mode = sctx->path_list[path_nr].dmdev->mode;
+	*bdev = sctx->path_list[path_nr].dmdev->bdev;
+	*mode = sctx->path_list[path_nr].dmdev->mode;
 
 	/*
 	 * Only pass ioctls through if the device sizes match exactly.
 	 */
-	if (ti->len + sctx->path_list[path_nr].start != i_size_read(bdev->bd_inode) >> SECTOR_SHIFT)
-		r = scsi_verify_blk_ioctl(NULL, cmd);
-
-	return r ? : __blkdev_driver_ioctl(bdev, mode, cmd, arg);
+	if (ti->len + sctx->path_list[path_nr].start !=
+	    i_size_read((*bdev)->bd_inode) >> SECTOR_SHIFT)
+		return 1;
+	return 0;
 }
 
 static int switch_iterate_devices(struct dm_target *ti,
···
 	.map = switch_map,
 	.message = switch_message,
 	.status = switch_status,
-	.ioctl = switch_ioctl,
+	.prepare_ioctl = switch_prepare_ioctl,
 	.iterate_devices = switch_iterate_devices,
 };
 

drivers/md/dm-thin-metadata.c

···
 		}
 	}
 
-	return dm_bm_unlock(b);
+	dm_bm_unlock(b);
+
+	return 0;
 }
 
 static void __setup_btree_details(struct dm_pool_metadata *pmd)
···
 	}
 
 	__setup_btree_details(pmd);
-	return dm_bm_unlock(sblock);
+	dm_bm_unlock(sblock);
+
+	return 0;
 
 bad_cleanup_data_sm:
 	dm_sm_destroy(pmd->data_sm);
···
 	dm_btree_del(&pmd->details_info, le64_to_cpu(disk_super->device_details_root));
 	dm_sm_dec_block(pmd->metadata_sm, held_root);
 
-	return dm_tm_unlock(pmd->tm, copy);
+	dm_tm_unlock(pmd->tm, copy);
+
+	return 0;
 }
 
 int dm_pool_release_metadata_snap(struct dm_pool_metadata *pmd)
···
 	disk_super = dm_block_data(sblock);
 	*result = le64_to_cpu(disk_super->held_root);
 
-	return dm_bm_unlock(sblock);
+	dm_bm_unlock(sblock);
+
+	return 0;
 }
 
 int dm_pool_get_metadata_snap(struct dm_pool_metadata *pmd,

drivers/md/dm-verity.c

···
 	}
 }
 
-static int verity_ioctl(struct dm_target *ti, unsigned cmd,
-			unsigned long arg)
+static int verity_prepare_ioctl(struct dm_target *ti,
+		struct block_device **bdev, fmode_t *mode)
 {
 	struct dm_verity *v = ti->private;
-	int r = 0;
+
+	*bdev = v->data_dev->bdev;
 
 	if (v->data_start ||
 	    ti->len != i_size_read(v->data_dev->bdev->bd_inode) >> SECTOR_SHIFT)
-		r = scsi_verify_blk_ioctl(NULL, cmd);
-
-	return r ? : __blkdev_driver_ioctl(v->data_dev->bdev, v->data_dev->mode,
-				     cmd, arg);
+		return 1;
+	return 0;
 }
 
 static int verity_iterate_devices(struct dm_target *ti,
···
 	.dtr		= verity_dtr,
 	.map		= verity_map,
 	.status		= verity_status,
-	.ioctl		= verity_ioctl,
+	.prepare_ioctl	= verity_prepare_ioctl,
 	.iterate_devices = verity_iterate_devices,
 	.io_hints	= verity_io_hints,
 };

drivers/md/dm.c

···
 #include <linux/ktime.h>
 #include <linux/elevator.h> /* for rq_end_sector() */
 #include <linux/blk-mq.h>
+#include <linux/pr.h>
 
 #include <trace/events/block.h>
 
···
 	return dm_get_geometry(md, geo);
 }
 
-static int dm_blk_ioctl(struct block_device *bdev, fmode_t mode,
-			unsigned int cmd, unsigned long arg)
+static int dm_get_live_table_for_ioctl(struct mapped_device *md,
+		struct dm_target **tgt, struct block_device **bdev,
+		fmode_t *mode, int *srcu_idx)
 {
-	struct mapped_device *md = bdev->bd_disk->private_data;
-	int srcu_idx;
 	struct dm_table *map;
-	struct dm_target *tgt;
-	int r = -ENOTTY;
+	int r;
 
 retry:
-	map = dm_get_live_table(md, &srcu_idx);
-
+	r = -ENOTTY;
+	map = dm_get_live_table(md, srcu_idx);
 	if (!map || !dm_table_get_size(map))
 		goto out;
 
···
 	if (dm_table_get_num_targets(map) != 1)
 		goto out;
 
-	tgt = dm_table_get_target(map, 0);
-	if (!tgt->type->ioctl)
+	*tgt = dm_table_get_target(map, 0);
+
+	if (!(*tgt)->type->prepare_ioctl)
 		goto out;
 
 	if (dm_suspended_md(md)) {
···
 		goto out;
 	}
 
-	r = tgt->type->ioctl(tgt, cmd, arg);
+	r = (*tgt)->type->prepare_ioctl(*tgt, bdev, mode);
+	if (r < 0)
+		goto out;
+
+	return r;
 
 out:
-	dm_put_live_table(md, srcu_idx);
-
+	dm_put_live_table(md, *srcu_idx);
 	if (r == -ENOTCONN) {
 		msleep(10);
 		goto retry;
 	}
+	return r;
+}
 
+static int dm_blk_ioctl(struct block_device *bdev, fmode_t mode,
+			unsigned int cmd, unsigned long arg)
+{
+	struct mapped_device *md = bdev->bd_disk->private_data;
+	struct dm_target *tgt;
+	int srcu_idx, r;
+
+	r = dm_get_live_table_for_ioctl(md, &tgt, &bdev, &mode, &srcu_idx);
+	if (r < 0)
+		return r;
+
+	if (r > 0) {
+		/*
+		 * Target determined this ioctl is being issued against
+		 * a logical partition of the parent bdev; so extra
+		 * validation is needed.
+		 */
+		r = scsi_verify_blk_ioctl(NULL, cmd);
+		if (r)
+			goto out;
+	}
+
+	r =  __blkdev_driver_ioctl(bdev, mode, cmd, arg);
+out:
+	dm_put_live_table(md, srcu_idx);
 	return r;
 }
 
···
 
 	map = dm_get_live_table(md, &srcu_idx);
 
-	blk_queue_split(q, &bio, q->bio_split);
-
 	generic_start_io_acct(rw, bio_sectors(bio), &dm_disk(md)->part0);
 
 	/* if we're suspended, we have to queue this io for later */
···
 	 * This queue is new, so no concurrency on the queue_flags.
 	 */
 	queue_flag_clear_unlocked(QUEUE_FLAG_STACKABLE, md->queue);
+
+	/*
+	 * Initialize data that will only be used by a non-blk-mq DM queue
+	 * - must do so here (in alloc_dev callchain) before queue is used
+	 */
+	md->queue->queuedata = md;
+	md->queue->backing_dev_info.congested_data = md;
 }
 
 static void dm_init_old_md_queue(struct mapped_device *md)
···
 	/*
 	 * Initialize aspects of queue that aren't relevant for blk-mq
 	 */
-	md->queue->queuedata = md;
 	md->queue->backing_dev_info.congested_fn = dm_any_congested;
-	md->queue->backing_dev_info.congested_data = md;
-
 	blk_queue_bounce_limit(md->queue, BLK_BOUNCE_ANY);
 }
 
···
 		destroy_workqueue(md->wq);
 	if (md->kworker_task)
 		kthread_stop(md->kworker_task);
-	if (md->io_pool)
-		mempool_destroy(md->io_pool);
-	if (md->rq_pool)
-		mempool_destroy(md->rq_pool);
+	mempool_destroy(md->io_pool);
+	mempool_destroy(md->rq_pool);
 	if (md->bs)
 		bioset_free(md->bs);
 
···
 	case DM_TYPE_BIO_BASED:
 		dm_init_old_md_queue(md);
 		blk_queue_make_request(md->queue, dm_make_request);
+		/*
+		 * DM handles splitting bios as needed.  Free the bio_split bioset
+		 * since it won't be used (saves 1 process per bio-based DM device).
+		 */
+		bioset_free(md->queue->bio_split);
+		md->queue->bio_split = NULL;
 		break;
 	}
 
···
 	if (!pools)
 		return;
 
-	if (pools->io_pool)
-		mempool_destroy(pools->io_pool);
-
-	if (pools->rq_pool)
-		mempool_destroy(pools->rq_pool);
+	mempool_destroy(pools->io_pool);
+	mempool_destroy(pools->rq_pool);
 
 	if (pools->bs)
 		bioset_free(pools->bs);
···
 	kfree(pools);
 }
 
+static int dm_pr_register(struct block_device *bdev, u64 old_key, u64 new_key,
+		u32 flags)
+{
+	struct mapped_device *md = bdev->bd_disk->private_data;
+	const struct pr_ops *ops;
+	struct dm_target *tgt;
+	fmode_t mode;
+	int srcu_idx, r;
+
+	r = dm_get_live_table_for_ioctl(md, &tgt, &bdev, &mode, &srcu_idx);
+	if (r < 0)
+		return r;
+
+	ops = bdev->bd_disk->fops->pr_ops;
+	if (ops && ops->pr_register)
+		r = ops->pr_register(bdev, old_key, new_key, flags);
+	else
+		r = -EOPNOTSUPP;
+
+	dm_put_live_table(md, srcu_idx);
+	return r;
+}
+
+static int dm_pr_reserve(struct block_device *bdev, u64 key, enum pr_type type,
+		u32 flags)
+{
+	struct mapped_device *md = bdev->bd_disk->private_data;
+	const struct pr_ops *ops;
+	struct dm_target *tgt;
+	fmode_t mode;
+	int srcu_idx, r;
+
+	r = dm_get_live_table_for_ioctl(md, &tgt, &bdev, &mode, &srcu_idx);
+	if (r < 0)
+		return r;
+
+	ops = bdev->bd_disk->fops->pr_ops;
+	if (ops && ops->pr_reserve)
+		r = ops->pr_reserve(bdev, key, type, flags);
+	else
+		r = -EOPNOTSUPP;
+
+	dm_put_live_table(md, srcu_idx);
+	return r;
+}
+
+static int dm_pr_release(struct block_device *bdev, u64 key, enum pr_type type)
+{
+	struct mapped_device *md = bdev->bd_disk->private_data;
+	const struct pr_ops *ops;
+	struct dm_target *tgt;
+	fmode_t mode;
+	int srcu_idx, r;
+
+	r = dm_get_live_table_for_ioctl(md, &tgt, &bdev, &mode, &srcu_idx);
+	if (r < 0)
+		return r;
+
+	ops = bdev->bd_disk->fops->pr_ops;
+	if (ops && ops->pr_release)
+		r = ops->pr_release(bdev, key, type);
+	else
+		r = -EOPNOTSUPP;
+
+	dm_put_live_table(md, srcu_idx);
+	return r;
+}
+
+static int dm_pr_preempt(struct block_device *bdev, u64 old_key, u64 new_key,
+		enum pr_type type, bool abort)
+{
+	struct mapped_device *md = bdev->bd_disk->private_data;
+	const struct pr_ops *ops;
+	struct dm_target *tgt;
+	fmode_t mode;
+	int srcu_idx, r;
+
+	r = dm_get_live_table_for_ioctl(md, &tgt, &bdev, &mode, &srcu_idx);
+	if (r < 0)
+		return r;
+
+	ops = bdev->bd_disk->fops->pr_ops;
+	if (ops && ops->pr_preempt)
+		r = ops->pr_preempt(bdev, old_key, new_key, type, abort);
+	else
+		r = -EOPNOTSUPP;
+
+	dm_put_live_table(md, srcu_idx);
+	return r;
+}
+
+static int dm_pr_clear(struct block_device *bdev, u64 key)
+{
+	struct mapped_device *md = bdev->bd_disk->private_data;
+	const struct pr_ops *ops;
+	struct dm_target *tgt;
+	fmode_t mode;
+	int srcu_idx, r;
+
+	r = dm_get_live_table_for_ioctl(md, &tgt, &bdev, &mode, &srcu_idx);
+	if (r < 0)
+		return r;
+
+	ops = bdev->bd_disk->fops->pr_ops;
+	if (ops && ops->pr_clear)
+		r = ops->pr_clear(bdev, key);
+	else
+		r = -EOPNOTSUPP;
+
+	dm_put_live_table(md, srcu_idx);
+	return r;
+}
+
+static const struct pr_ops dm_pr_ops = {
+	.pr_register	= dm_pr_register,
+	.pr_reserve	= dm_pr_reserve,
+	.pr_release	= dm_pr_release,
+	.pr_preempt	= dm_pr_preempt,
+	.pr_clear	= dm_pr_clear,
+};
+
 static const struct block_device_operations dm_blk_dops = {
 	.open = dm_blk_open,
 	.release = dm_blk_close,
 	.ioctl = dm_blk_ioctl,
 	.getgeo = dm_blk_getgeo,
+	.pr_ops = &dm_pr_ops,
 	.owner = THIS_MODULE
 };
 

drivers/md/persistent-data/dm-array.c

···
 /*
  * Unlocks an array block.
  */
-static int unlock_ablock(struct dm_array_info *info, struct dm_block *block)
+static void unlock_ablock(struct dm_array_info *info, struct dm_block *block)
 {
-	return dm_tm_unlock(info->btree_info.tm, block);
+	dm_tm_unlock(info->btree_info.tm, block);
 }
 
 /*----------------------------------------------------------------*/

drivers/md/persistent-data/dm-block-manager.c

···
 }
 EXPORT_SYMBOL_GPL(dm_bm_write_lock_zero);
 
-int dm_bm_unlock(struct dm_block *b)
+void dm_bm_unlock(struct dm_block *b)
 {
 	struct buffer_aux *aux;
 	aux = dm_bufio_get_aux_data(to_buffer(b));
···
 		bl_up_read(&aux->lock);
 
 	dm_bufio_release(to_buffer(b));
-
-	return 0;
 }
 EXPORT_SYMBOL_GPL(dm_bm_unlock);
 

drivers/md/persistent-data/dm-block-manager.h

···
 			  struct dm_block_validator *v,
 			  struct dm_block **result);
 
-int dm_bm_unlock(struct dm_block *b);
+void dm_bm_unlock(struct dm_block *b);
 
 /*
  * It's a common idiom to have a superblock that should be committed last.

drivers/md/persistent-data/dm-btree-internal.h

···
 		  struct dm_btree_value_type *vt);
 
 int new_block(struct dm_btree_info *info, struct dm_block **result);
-int unlock_block(struct dm_btree_info *info, struct dm_block *b);
+void unlock_block(struct dm_btree_info *info, struct dm_block *b);
 
 /*
  * Spines keep track of the rolling locks.  There are 2 variants, read-only

drivers/md/persistent-data/dm-btree-remove.c

···
 	return 0;
 }
 
-static int exit_child(struct dm_btree_info *info, struct child *c)
+static void exit_child(struct dm_btree_info *info, struct child *c)
 {
-	return dm_tm_unlock(info->tm, c->block);
+	dm_tm_unlock(info->tm, c->block);
 }
 
 static void shift(struct btree_node *left, struct btree_node *right, int count)
···
 
 	__rebalance2(info, parent, &left, &right);
 
-	r = exit_child(info, &left);
-	if (r) {
-		exit_child(info, &right);
-		return r;
-	}
+	exit_child(info, &left);
+	exit_child(info, &right);
 
-	return exit_child(info, &right);
+	return 0;
 }
 
 /*
···
 
 	__rebalance3(info, parent, &left, &center, &right);
 
-	r = exit_child(info, &left);
-	if (r) {
-		exit_child(info, &center);
-		exit_child(info, &right);
-		return r;
-	}
-
-	r = exit_child(info, &center);
-	if (r) {
-		exit_child(info, &right);
-		return r;
-	}
-
-	r = exit_child(info, &right);
-	if (r)
-		return r;
+	exit_child(info, &left);
+	exit_child(info, &center);
+	exit_child(info, &right);
 
 	return 0;
 }
···
 
 		memcpy(n, dm_block_data(child),
 		       dm_bm_block_size(dm_tm_get_bm(info->tm)));
-		r = dm_tm_unlock(info->tm, child);
-		if (r)
-			return r;
+		dm_tm_unlock(info->tm, child);
 
 		dm_tm_dec(info->tm, dm_block_location(child));
 		return 0;

drivers/md/persistent-data/dm-btree-spine.c

···
 	return dm_tm_new_block(info->tm, &btree_node_validator, result);
 }
 
-int unlock_block(struct dm_btree_info *info, struct dm_block *b)
+void unlock_block(struct dm_btree_info *info, struct dm_block *b)
 {
-	return dm_tm_unlock(info->tm, b);
+	dm_tm_unlock(info->tm, b);
 }
 
 /*----------------------------------------------------------------*/
···
 	int r = 0, i;
 
 	for (i = 0; i < s->count; i++) {
-		int r2 = unlock_block(s->info, s->nodes[i]);
-		if (r2 < 0)
-			r = r2;
+		unlock_block(s->info, s->nodes[i]);
 	}
 
 	return r;
···
 	int r;
 
 	if (s->count == 2) {
-		r = unlock_block(s->info, s->nodes[0]);
-		if (r < 0)
-			return r;
+		unlock_block(s->info, s->nodes[0]);
 		s->nodes[0] = s->nodes[1];
 		s->count--;
 	}
···
 	int r = 0, i;
 
 	for (i = 0; i < s->count; i++) {
-		int r2 = unlock_block(s->info, s->nodes[i]);
-		if (r2 < 0)
-			r = r2;
+		unlock_block(s->info, s->nodes[i]);
 	}
 
 	return r;
···
 	int r;
 
 	if (s->count == 2) {
-		r = unlock_block(s->info, s->nodes[0]);
-		if (r < 0)
-			return r;
+		unlock_block(s->info, s->nodes[0]);
 		s->nodes[0] = s->nodes[1];
 		s->count--;
 	}

drivers/md/persistent-data/dm-btree.c

···
 	n->header.value_size = cpu_to_le32(info->value_type.size);
 
 	*root = dm_block_location(b);
-	return unlock_block(info, b);
+	unlock_block(info, b);
+
+	return 0;
 }
 EXPORT_SYMBOL_GPL(dm_btree_empty);
 

drivers/md/persistent-data/dm-space-map-common.c

···
 
 		idx.blocknr = cpu_to_le64(dm_block_location(b));
 
-		r = dm_tm_unlock(ll->tm, b);
-		if (r < 0)
-			return r;
+		dm_tm_unlock(ll->tm, b);
 
 		idx.nr_free = cpu_to_le32(ll->entries_per_block);
 		idx.none_free_before = 0;
···
 
 	*result = sm_lookup_bitmap(dm_bitmap_data(blk), b);
 
-	return dm_tm_unlock(ll->tm, blk);
+	dm_tm_unlock(ll->tm, blk);
+
+	return 0;
 }
 
 static int sm_ll_lookup_big_ref_count(struct ll_disk *ll, dm_block_t b,
···
 			return r;
 		}
 
-		r = dm_tm_unlock(ll->tm, blk);
-		if (r < 0)
-			return r;
+		dm_tm_unlock(ll->tm, blk);
 
 		*result = i * ll->entries_per_block + (dm_block_t) position;
 		return 0;
···
 	if (ref_count <= 2) {
 		sm_set_bitmap(bm_le, bit, ref_count);
 
-		r = dm_tm_unlock(ll->tm, nb);
-		if (r < 0)
-			return r;
+		dm_tm_unlock(ll->tm, nb);
 
 		if (old > 2) {
 			r = dm_btree_remove(&ll->ref_count_info,
···
 		__le32 le_rc = cpu_to_le32(ref_count);
 
 		sm_set_bitmap(bm_le, bit, 3);
-		r = dm_tm_unlock(ll->tm, nb);
-		if (r < 0)
-			return r;
+		dm_tm_unlock(ll->tm, nb);
 
 		__dm_bless_for_disk(&le_rc);
 		r = dm_btree_insert(&ll->ref_count_info, ll->ref_count_root,
···
 	memcpy(dm_block_data(b), &ll->mi_le, sizeof(ll->mi_le));
 	ll->bitmap_root = dm_block_location(b);
 
-	return dm_tm_unlock(ll->tm, b);
+	dm_tm_unlock(ll->tm, b);
+
+	return 0;
 }
 
 static int metadata_ll_open(struct ll_disk *ll)
···
 		return r;
 
 	memcpy(&ll->mi_le, dm_block_data(block), sizeof(ll->mi_le));
-	return dm_tm_unlock(ll->tm, block);
+	dm_tm_unlock(ll->tm, block);
+
+	return 0;
 }
 
 static dm_block_t metadata_ll_max_entries(struct ll_disk *ll)
···
 	memcpy(dm_block_data(b), &ll->mi_le, sizeof(ll->mi_le));
 	ll->bitmap_root = dm_block_location(b);
 
-	return dm_tm_unlock(ll->tm, b);
+	dm_tm_unlock(ll->tm, b);
+
+	return 0;
 }
 
 int sm_ll_new_metadata(struct ll_disk *ll, struct dm_transaction_manager *tm)

drivers/md/persistent-data/dm-transaction-manager.c

···
 }
 EXPORT_SYMBOL_GPL(dm_tm_read_lock);
 
-int dm_tm_unlock(struct dm_transaction_manager *tm, struct dm_block *b)
+void dm_tm_unlock(struct dm_transaction_manager *tm, struct dm_block *b)
 {
-	return dm_bm_unlock(b);
+	dm_bm_unlock(b);
 }
 EXPORT_SYMBOL_GPL(dm_tm_unlock);
 

drivers/md/persistent-data/dm-transaction-manager.h

···
 		    struct dm_block_validator *v,
 		    struct dm_block **result);
 
-int dm_tm_unlock(struct dm_transaction_manager *tm, struct dm_block *b);
+void dm_tm_unlock(struct dm_transaction_manager *tm, struct dm_block *b);
 
 /*
  * Functions for altering the reference count of a block directly.

include/linux/device-mapper.h

···
 
 typedef int (*dm_message_fn) (struct dm_target *ti, unsigned argc, char **argv);
 
-typedef int (*dm_ioctl_fn) (struct dm_target *ti, unsigned int cmd,
-			    unsigned long arg);
+typedef int (*dm_prepare_ioctl_fn) (struct dm_target *ti,
+			    struct block_device **bdev, fmode_t *mode);
 
 /*
  * These iteration functions are typically used to check (and combine)
···
 	dm_resume_fn resume;
 	dm_status_fn status;
 	dm_message_fn message;
-	dm_ioctl_fn ioctl;
+	dm_prepare_ioctl_fn prepare_ioctl;
 	dm_busy_fn busy;
 	dm_iterate_devices_fn iterate_devices;
 	dm_io_hints_fn io_hints;

include/uapi/linux/dm-ioctl.h

···
 #define DM_DEV_SET_GEOMETRY	_IOWR(DM_IOCTL, DM_DEV_SET_GEOMETRY_CMD, struct dm_ioctl)
 
 #define DM_VERSION_MAJOR	4
-#define DM_VERSION_MINOR	33
+#define DM_VERSION_MINOR	34
 #define DM_VERSION_PATCHLEVEL	0
-#define DM_VERSION_EXTRA	"-ioctl (2015-8-18)"
+#define DM_VERSION_EXTRA	"-ioctl (2015-10-28)"
 
 /* Status bits */
 #define DM_READONLY_FLAG	(1 << 0) /* In/Out */