Merge tag 'dm-4.6-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm

Pull device mapper updates from Mike Snitzer:

 - Most attention this cycle went to optimizing blk-mq request-based DM
   (dm-mq) that is used exclussively by DM multipath:

     - A stable fix for dm-mq that eliminates excessive context
       switching offers the biggest performance improvement (for both
       IOPs and throughput).

     - But more work is needed, during the next cycle, to reduce
       spinlock contention in DM multipath on large NUMA systems.

 - A stable fix for a NULL pointer seen when DM stats is enabled on a DM
   multipath device that must requeue an IO due to path failure.

 - A stable fix for DM snapshot to disallow the COW and origin devices
   from being identical.  This amounts to graceful failure in the face
   of userspace error because these devices shouldn't ever be identical.

 - Stable fixes for DM cache and DM thin provisioning to address crashes
   seen if/when their respective metadata device experiences failures
   that cause the transition to 'fail_io' mode.

 - The DM cache 'mq' policy is now an alias for the 'smq' policy.  The
   'smq' policy proved to be consistently better than 'mq'.  As such
   'mq', with all its complex user-facing tunables, has been eliminated.

 - Improve DM thin provisioning to consistently return -ENOSPC once the
   thin-pool's data volume is out of space.

 - Improve DM core to properly handle error propagation if
   bio_integrity_clone() fails in clone_bio().

 - Other small cleanups and improvements to DM core.

* tag 'dm-4.6-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm: (41 commits)
  dm: fix rq_end_stats() NULL pointer in dm_requeue_original_request()
  dm thin: consistently return -ENOSPC if pool has run out of data space
  dm cache: bump the target version
  dm cache: make sure every metadata function checks fail_io
  dm: add missing newline between DM_DEBUG_BLOCK_STACK_TRACING and DM_BUFIO
  dm cache policy smq: clarify that mq registration failure was for 'mq'
  dm: return error if bio_integrity_clone() fails in clone_bio()
  dm thin metadata: don't issue prefetches if a transaction abort has failed
  dm snapshot: disallow the COW and origin devices from being identical
  dm cache: make the 'mq' policy an alias for 'smq'
  dm: drop unnecessary assignment of md->queue
  dm: reorder 'struct mapped_device' members to fix alignment and holes
  dm: remove dummy definition of 'struct dm_table'
  dm: add 'dm_numa_node' module parameter
  dm thin metadata: remove needless newline from subtree_dec() DMERR message
  dm mpath: cleanup reinstate_path() et al based on code review
  dm mpath: remove __pgpath_busy forward declaration, rename to pgpath_busy
  dm mpath: switch from 'unsigned' to 'bool' for flags where appropriate
  dm round robin: use percpu 'repeat_count' and 'current_path'
  dm path selector: remove 'repeat_count' return from .select_path hook
  ...

Documentation/device-mapper/cache-policies.txt

@@ -28,51 +28,16 @@ Overview of supplied cache replacement policies
 multiqueue (mq)
 ---------------
 
-This policy has been deprecated in favor of the smq policy (see below).
+This policy is now an alias for smq (see below).
 
-The multiqueue policy has three sets of 16 queues: one set for entries
-waiting for the cache and another two for those in the cache (a set for
-clean entries and a set for dirty entries).
+The following tunables are accepted, but have no effect:
 
-Cache entries in the queues are aged based on logical time. Entry into
-the cache is based on variable thresholds and queue selection is based
-on hit count on entry. The policy aims to take different cache miss
-costs into account and to adjust to varying load patterns automatically.
-
-Message and constructor argument pairs are:
 	'sequential_threshold <#nr_sequential_ios>'
 	'random_threshold <#nr_random_ios>'
 	'read_promote_adjustment <value>'
 	'write_promote_adjustment <value>'
 	'discard_promote_adjustment <value>'
 
-The sequential threshold indicates the number of contiguous I/Os
-required before a stream is treated as sequential.  Once a stream is
-considered sequential it will bypass the cache.  The random threshold
-is the number of intervening non-contiguous I/Os that must be seen
-before the stream is treated as random again.
-
-The sequential and random thresholds default to 512 and 4 respectively.
-
-Large, sequential I/Os are probably better left on the origin device
-since spindles tend to have good sequential I/O bandwidth.  The
-io_tracker counts contiguous I/Os to try to spot when the I/O is in one
-of these sequential modes.  But there are use-cases for wanting to
-promote sequential blocks to the cache (e.g. fast application startup).
-If sequential threshold is set to 0 the sequential I/O detection is
-disabled and sequential I/O will no longer implicitly bypass the cache.
-Setting the random threshold to 0 does _not_ disable the random I/O
-stream detection.
-
-Internally the mq policy determines a promotion threshold.  If the hit
-count of a block not in the cache goes above this threshold it gets
-promoted to the cache.  The read, write and discard promote adjustment
-tunables allow you to tweak the promotion threshold by adding a small
-value based on the io type.  They default to 4, 8 and 1 respectively.
-If you're trying to quickly warm a new cache device you may wish to
-reduce these to encourage promotion.  Remember to switch them back to
-their defaults after the cache fills though.
-
 Stochastic multiqueue (smq)
 ---------------------------
 

block/blk-core.c

@@ -2198,7 +2198,7 @@ int blk_insert_cloned_request(struct request_queue *q, struct request *rq)
 	if (q->mq_ops) {
 		if (blk_queue_io_stat(q))
 			blk_account_io_start(rq, true);
-		blk_mq_insert_request(rq, false, true, true);
+		blk_mq_insert_request(rq, false, true, false);
 		return 0;
 	}
 

drivers/md/Kconfig

@@ -249,6 +249,7 @@ config DM_DEBUG_BLOCK_STACK_TRACING
 	 block manager locking used by thin provisioning and caching.
 
 	 If unsure, say N.
+
 config DM_BIO_PRISON
        tristate
        depends on BLK_DEV_DM
@@ -304,16 +305,6 @@ config DM_CACHE
          algorithms used to select which blocks are promoted, demoted,
          cleaned etc.  It supports writeback and writethrough modes.
 
-config DM_CACHE_MQ
-       tristate "MQ Cache Policy (EXPERIMENTAL)"
-       depends on DM_CACHE
-       default y
-       ---help---
-         A cache policy that uses a multiqueue ordered by recent hit
-         count to select which blocks should be promoted and demoted.
-         This is meant to be a general purpose policy.  It prioritises
-         reads over writes.
-
 config DM_CACHE_SMQ
        tristate "Stochastic MQ Cache Policy (EXPERIMENTAL)"
        depends on DM_CACHE

drivers/md/Makefile

@@ -12,7 +12,6 @@ dm-log-userspace-y \
 		+= dm-log-userspace-base.o dm-log-userspace-transfer.o
 dm-thin-pool-y	+= dm-thin.o dm-thin-metadata.o
 dm-cache-y	+= dm-cache-target.o dm-cache-metadata.o dm-cache-policy.o
-dm-cache-mq-y   += dm-cache-policy-mq.o
 dm-cache-smq-y   += dm-cache-policy-smq.o
 dm-cache-cleaner-y += dm-cache-policy-cleaner.o
 dm-era-y	+= dm-era-target.o
@@ -55,7 +54,6 @@ obj-$(CONFIG_DM_RAID)	+= dm-raid.o
 obj-$(CONFIG_DM_THIN_PROVISIONING)	+= dm-thin-pool.o
 obj-$(CONFIG_DM_VERITY)		+= dm-verity.o
 obj-$(CONFIG_DM_CACHE)		+= dm-cache.o
-obj-$(CONFIG_DM_CACHE_MQ)	+= dm-cache-mq.o
 obj-$(CONFIG_DM_CACHE_SMQ)	+= dm-cache-smq.o
 obj-$(CONFIG_DM_CACHE_CLEANER)	+= dm-cache-cleaner.o
 obj-$(CONFIG_DM_ERA)		+= dm-era.o

drivers/md/dm-cache-metadata.c

@@ -867,19 +867,40 @@ static int blocks_are_unmapped_or_clean(struct dm_cache_metadata *cmd,
 	return 0;
 }
 
-#define WRITE_LOCK(cmd) \
-	if (cmd->fail_io || dm_bm_is_read_only(cmd->bm)) \
+#define WRITE_LOCK(cmd)	\
+	down_write(&cmd->root_lock); \
+	if (cmd->fail_io || dm_bm_is_read_only(cmd->bm)) { \
+		up_write(&cmd->root_lock); \
 		return -EINVAL; \
-	down_write(&cmd->root_lock)
+	}
 
 #define WRITE_LOCK_VOID(cmd) \
-	if (cmd->fail_io || dm_bm_is_read_only(cmd->bm)) \
+	down_write(&cmd->root_lock); \
+	if (cmd->fail_io || dm_bm_is_read_only(cmd->bm)) { \
+		up_write(&cmd->root_lock); \
 		return; \
-	down_write(&cmd->root_lock)
+	}
 
 #define WRITE_UNLOCK(cmd) \
 	up_write(&cmd->root_lock)
 
+#define READ_LOCK(cmd) \
+	down_read(&cmd->root_lock); \
+	if (cmd->fail_io || dm_bm_is_read_only(cmd->bm)) { \
+		up_read(&cmd->root_lock); \
+		return -EINVAL; \
+	}
+
+#define READ_LOCK_VOID(cmd)	\
+	down_read(&cmd->root_lock); \
+	if (cmd->fail_io || dm_bm_is_read_only(cmd->bm)) { \
+		up_read(&cmd->root_lock); \
+		return; \
+	}
+
+#define READ_UNLOCK(cmd) \
+	up_read(&cmd->root_lock)
+
 int dm_cache_resize(struct dm_cache_metadata *cmd, dm_cblock_t new_cache_size)
 {
 	int r;
@@ -1015,22 +1036,20 @@ int dm_cache_load_discards(struct dm_cache_metadata *cmd,
 {
 	int r;
 
-	down_read(&cmd->root_lock);
+	READ_LOCK(cmd);
 	r = __load_discards(cmd, fn, context);
-	up_read(&cmd->root_lock);
+	READ_UNLOCK(cmd);
 
 	return r;
 }
 
-dm_cblock_t dm_cache_size(struct dm_cache_metadata *cmd)
+int dm_cache_size(struct dm_cache_metadata *cmd, dm_cblock_t *result)
 {
-	dm_cblock_t r;
+	READ_LOCK(cmd);
+	*result = cmd->cache_blocks;
+	READ_UNLOCK(cmd);
 
-	down_read(&cmd->root_lock);
-	r = cmd->cache_blocks;
-	up_read(&cmd->root_lock);
-
-	return r;
+	return 0;
 }
 
 static int __remove(struct dm_cache_metadata *cmd, dm_cblock_t cblock)
@@ -1188,9 +1207,9 @@ int dm_cache_load_mappings(struct dm_cache_metadata *cmd,
 {
 	int r;
 
-	down_read(&cmd->root_lock);
+	READ_LOCK(cmd);
 	r = __load_mappings(cmd, policy, fn, context);
-	up_read(&cmd->root_lock);
+	READ_UNLOCK(cmd);
 
 	return r;
 }
@@ -1215,18 +1234,18 @@ static int __dump_mappings(struct dm_cache_metadata *cmd)
 
 void dm_cache_dump(struct dm_cache_metadata *cmd)
 {
-	down_read(&cmd->root_lock);
+	READ_LOCK_VOID(cmd);
 	__dump_mappings(cmd);
-	up_read(&cmd->root_lock);
+	READ_UNLOCK(cmd);
 }
 
 int dm_cache_changed_this_transaction(struct dm_cache_metadata *cmd)
 {
 	int r;
 
-	down_read(&cmd->root_lock);
+	READ_LOCK(cmd);
 	r = cmd->changed;
-	up_read(&cmd->root_lock);
+	READ_UNLOCK(cmd);
 
 	return r;
 }
@@ -1276,9 +1295,9 @@ int dm_cache_set_dirty(struct dm_cache_metadata *cmd,
 void dm_cache_metadata_get_stats(struct dm_cache_metadata *cmd,
 				 struct dm_cache_statistics *stats)
 {
-	down_read(&cmd->root_lock);
+	READ_LOCK_VOID(cmd);
 	*stats = cmd->stats;
-	up_read(&cmd->root_lock);
+	READ_UNLOCK(cmd);
 }
 
 void dm_cache_metadata_set_stats(struct dm_cache_metadata *cmd,
@@ -1312,9 +1331,9 @@ int dm_cache_get_free_metadata_block_count(struct dm_cache_metadata *cmd,
 {
 	int r = -EINVAL;
 
-	down_read(&cmd->root_lock);
+	READ_LOCK(cmd);
 	r = dm_sm_get_nr_free(cmd->metadata_sm, result);
-	up_read(&cmd->root_lock);
+	READ_UNLOCK(cmd);
 
 	return r;
 }
@@ -1324,9 +1343,9 @@ int dm_cache_get_metadata_dev_size(struct dm_cache_metadata *cmd,
 {
 	int r = -EINVAL;
 
-	down_read(&cmd->root_lock);
+	READ_LOCK(cmd);
 	r = dm_sm_get_nr_blocks(cmd->metadata_sm, result);
-	up_read(&cmd->root_lock);
+	READ_UNLOCK(cmd);
 
 	return r;
 }
@@ -1417,7 +1436,13 @@ int dm_cache_write_hints(struct dm_cache_metadata *cmd, struct dm_cache_policy *
 
 int dm_cache_metadata_all_clean(struct dm_cache_metadata *cmd, bool *result)
 {
-	return blocks_are_unmapped_or_clean(cmd, 0, cmd->cache_blocks, result);
+	int r;
+
+	READ_LOCK(cmd);
+	r = blocks_are_unmapped_or_clean(cmd, 0, cmd->cache_blocks, result);
+	READ_UNLOCK(cmd);
+
+	return r;
 }
 
 void dm_cache_metadata_set_read_only(struct dm_cache_metadata *cmd)
@@ -1440,10 +1465,7 @@ int dm_cache_metadata_set_needs_check(struct dm_cache_metadata *cmd)
 	struct dm_block *sblock;
 	struct cache_disk_superblock *disk_super;
 
-	/*
-	 * We ignore fail_io for this function.
-	 */
-	down_write(&cmd->root_lock);
+	WRITE_LOCK(cmd);
 	set_bit(NEEDS_CHECK, &cmd->flags);
 
 	r = superblock_lock(cmd, &sblock);
@@ -1458,19 +1480,17 @@ int dm_cache_metadata_set_needs_check(struct dm_cache_metadata *cmd)
 	dm_bm_unlock(sblock);
 
 out:
-	up_write(&cmd->root_lock);
+	WRITE_UNLOCK(cmd);
 	return r;
 }
 
-bool dm_cache_metadata_needs_check(struct dm_cache_metadata *cmd)
+int dm_cache_metadata_needs_check(struct dm_cache_metadata *cmd, bool *result)
 {
-	bool needs_check;
+	READ_LOCK(cmd);
+	*result = !!test_bit(NEEDS_CHECK, &cmd->flags);
+	READ_UNLOCK(cmd);
 
-	down_read(&cmd->root_lock);
-	needs_check = !!test_bit(NEEDS_CHECK, &cmd->flags);
-	up_read(&cmd->root_lock);
-
-	return needs_check;
+	return 0;
 }
 
 int dm_cache_metadata_abort(struct dm_cache_metadata *cmd)

drivers/md/dm-cache-metadata.h

@@ -66,7 +66,7 @@ void dm_cache_metadata_close(struct dm_cache_metadata *cmd);
  * origin blocks to map to.
  */
 int dm_cache_resize(struct dm_cache_metadata *cmd, dm_cblock_t new_cache_size);
-dm_cblock_t dm_cache_size(struct dm_cache_metadata *cmd);
+int dm_cache_size(struct dm_cache_metadata *cmd, dm_cblock_t *result);
 
 int dm_cache_discard_bitset_resize(struct dm_cache_metadata *cmd,
 				   sector_t discard_block_size,
@@ -137,7 +137,7 @@ int dm_cache_write_hints(struct dm_cache_metadata *cmd, struct dm_cache_policy *
  */
 int dm_cache_metadata_all_clean(struct dm_cache_metadata *cmd, bool *result);
 
-bool dm_cache_metadata_needs_check(struct dm_cache_metadata *cmd);
+int dm_cache_metadata_needs_check(struct dm_cache_metadata *cmd, bool *result);
 int dm_cache_metadata_set_needs_check(struct dm_cache_metadata *cmd);
 void dm_cache_metadata_set_read_only(struct dm_cache_metadata *cmd);
 void dm_cache_metadata_set_read_write(struct dm_cache_metadata *cmd);

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

@@ -1,1473 +0,0 @@
-/*
- * Copyright (C) 2012 Red Hat. All rights reserved.
- *
- * This file is released under the GPL.
- */
-
-#include "dm-cache-policy.h"
-#include "dm.h"
-
-#include <linux/hash.h>
-#include <linux/jiffies.h>
-#include <linux/module.h>
-#include <linux/mutex.h>
-#include <linux/slab.h>
-#include <linux/vmalloc.h>
-
-#define DM_MSG_PREFIX "cache-policy-mq"
-
-static struct kmem_cache *mq_entry_cache;
-
-/*----------------------------------------------------------------*/
-
-static unsigned next_power(unsigned n, unsigned min)
-{
-	return roundup_pow_of_two(max(n, min));
-}
-
-/*----------------------------------------------------------------*/
-
-/*
- * Large, sequential ios are probably better left on the origin device since
- * spindles tend to have good bandwidth.
- *
- * The io_tracker tries to spot when the io is in one of these sequential
- * modes.
- *
- * Two thresholds to switch between random and sequential io mode are defaulting
- * as follows and can be adjusted via the constructor and message interfaces.
- */
-#define RANDOM_THRESHOLD_DEFAULT 4
-#define SEQUENTIAL_THRESHOLD_DEFAULT 512
-
-enum io_pattern {
-	PATTERN_SEQUENTIAL,
-	PATTERN_RANDOM
-};
-
-struct io_tracker {
-	enum io_pattern pattern;
-
-	unsigned nr_seq_samples;
-	unsigned nr_rand_samples;
-	unsigned thresholds[2];
-
-	dm_oblock_t last_end_oblock;
-};
-
-static void iot_init(struct io_tracker *t,
-		     int sequential_threshold, int random_threshold)
-{
-	t->pattern = PATTERN_RANDOM;
-	t->nr_seq_samples = 0;
-	t->nr_rand_samples = 0;
-	t->last_end_oblock = 0;
-	t->thresholds[PATTERN_RANDOM] = random_threshold;
-	t->thresholds[PATTERN_SEQUENTIAL] = sequential_threshold;
-}
-
-static enum io_pattern iot_pattern(struct io_tracker *t)
-{
-	return t->pattern;
-}
-
-static void iot_update_stats(struct io_tracker *t, struct bio *bio)
-{
-	if (bio->bi_iter.bi_sector == from_oblock(t->last_end_oblock) + 1)
-		t->nr_seq_samples++;
-	else {
-		/*
-		 * Just one non-sequential IO is enough to reset the
-		 * counters.
-		 */
-		if (t->nr_seq_samples) {
-			t->nr_seq_samples = 0;
-			t->nr_rand_samples = 0;
-		}
-
-		t->nr_rand_samples++;
-	}
-
-	t->last_end_oblock = to_oblock(bio_end_sector(bio) - 1);
-}
-
-static void iot_check_for_pattern_switch(struct io_tracker *t)
-{
-	switch (t->pattern) {
-	case PATTERN_SEQUENTIAL:
-		if (t->nr_rand_samples >= t->thresholds[PATTERN_RANDOM]) {
-			t->pattern = PATTERN_RANDOM;
-			t->nr_seq_samples = t->nr_rand_samples = 0;
-		}
-		break;
-
-	case PATTERN_RANDOM:
-		if (t->nr_seq_samples >= t->thresholds[PATTERN_SEQUENTIAL]) {
-			t->pattern = PATTERN_SEQUENTIAL;
-			t->nr_seq_samples = t->nr_rand_samples = 0;
-		}
-		break;
-	}
-}
-
-static void iot_examine_bio(struct io_tracker *t, struct bio *bio)
-{
-	iot_update_stats(t, bio);
-	iot_check_for_pattern_switch(t);
-}
-
-/*----------------------------------------------------------------*/
-
-
-/*
- * This queue is divided up into different levels.  Allowing us to push
- * entries to the back of any of the levels.  Think of it as a partially
- * sorted queue.
- */
-#define NR_QUEUE_LEVELS 16u
-#define NR_SENTINELS NR_QUEUE_LEVELS * 3
-
-#define WRITEBACK_PERIOD HZ
-
-struct queue {
-	unsigned nr_elts;
-	bool current_writeback_sentinels;
-	unsigned long next_writeback;
-	struct list_head qs[NR_QUEUE_LEVELS];
-	struct list_head sentinels[NR_SENTINELS];
-};
-
-static void queue_init(struct queue *q)
-{
-	unsigned i;
-
-	q->nr_elts = 0;
-	q->current_writeback_sentinels = false;
-	q->next_writeback = 0;
-	for (i = 0; i < NR_QUEUE_LEVELS; i++) {
-		INIT_LIST_HEAD(q->qs + i);
-		INIT_LIST_HEAD(q->sentinels + i);
-		INIT_LIST_HEAD(q->sentinels + NR_QUEUE_LEVELS + i);
-		INIT_LIST_HEAD(q->sentinels + (2 * NR_QUEUE_LEVELS) + i);
-	}
-}
-
-static unsigned queue_size(struct queue *q)
-{
-	return q->nr_elts;
-}
-
-static bool queue_empty(struct queue *q)
-{
-	return q->nr_elts == 0;
-}
-
-/*
- * Insert an entry to the back of the given level.
- */
-static void queue_push(struct queue *q, unsigned level, struct list_head *elt)
-{
-	q->nr_elts++;
-	list_add_tail(elt, q->qs + level);
-}
-
-static void queue_remove(struct queue *q, struct list_head *elt)
-{
-	q->nr_elts--;
-	list_del(elt);
-}
-
-static bool is_sentinel(struct queue *q, struct list_head *h)
-{
-	return (h >= q->sentinels) && (h < (q->sentinels + NR_SENTINELS));
-}
-
-/*
- * Gives us the oldest entry of the lowest popoulated level.  If the first
- * level is emptied then we shift down one level.
- */
-static struct list_head *queue_peek(struct queue *q)
-{
-	unsigned level;
-	struct list_head *h;
-
-	for (level = 0; level < NR_QUEUE_LEVELS; level++)
-		list_for_each(h, q->qs + level)
-			if (!is_sentinel(q, h))
-				return h;
-
-	return NULL;
-}
-
-static struct list_head *queue_pop(struct queue *q)
-{
-	struct list_head *r = queue_peek(q);
-
-	if (r) {
-		q->nr_elts--;
-		list_del(r);
-	}
-
-	return r;
-}
-
-/*
- * Pops an entry from a level that is not past a sentinel.
- */
-static struct list_head *queue_pop_old(struct queue *q)
-{
-	unsigned level;
-	struct list_head *h;
-
-	for (level = 0; level < NR_QUEUE_LEVELS; level++)
-		list_for_each(h, q->qs + level) {
-			if (is_sentinel(q, h))
-				break;
-
-			q->nr_elts--;
-			list_del(h);
-			return h;
-		}
-
-	return NULL;
-}
-
-static struct list_head *list_pop(struct list_head *lh)
-{
-	struct list_head *r = lh->next;
-
-	BUG_ON(!r);
-	list_del_init(r);
-
-	return r;
-}
-
-static struct list_head *writeback_sentinel(struct queue *q, unsigned level)
-{
-	if (q->current_writeback_sentinels)
-		return q->sentinels + NR_QUEUE_LEVELS + level;
-	else
-		return q->sentinels + 2 * NR_QUEUE_LEVELS + level;
-}
-
-static void queue_update_writeback_sentinels(struct queue *q)
-{
-	unsigned i;
-	struct list_head *h;
-
-	if (time_after(jiffies, q->next_writeback)) {
-		for (i = 0; i < NR_QUEUE_LEVELS; i++) {
-			h = writeback_sentinel(q, i);
-			list_del(h);
-			list_add_tail(h, q->qs + i);
-		}
-
-		q->next_writeback = jiffies + WRITEBACK_PERIOD;
-		q->current_writeback_sentinels = !q->current_writeback_sentinels;
-	}
-}
-
-/*
- * Sometimes we want to iterate through entries that have been pushed since
- * a certain event.  We use sentinel entries on the queues to delimit these
- * 'tick' events.
- */
-static void queue_tick(struct queue *q)
-{
-	unsigned i;
-
-	for (i = 0; i < NR_QUEUE_LEVELS; i++) {
-		list_del(q->sentinels + i);
-		list_add_tail(q->sentinels + i, q->qs + i);
-	}
-}
-
-typedef void (*iter_fn)(struct list_head *, void *);
-static void queue_iterate_tick(struct queue *q, iter_fn fn, void *context)
-{
-	unsigned i;
-	struct list_head *h;
-
-	for (i = 0; i < NR_QUEUE_LEVELS; i++) {
-		list_for_each_prev(h, q->qs + i) {
-			if (is_sentinel(q, h))
-				break;
-
-			fn(h, context);
-		}
-	}
-}
-
-/*----------------------------------------------------------------*/
-
-/*
- * Describes a cache entry.  Used in both the cache and the pre_cache.
- */
-struct entry {
-	struct hlist_node hlist;
-	struct list_head list;
-	dm_oblock_t oblock;
-
-	/*
-	 * FIXME: pack these better
-	 */
-	bool dirty:1;
-	unsigned hit_count;
-};
-
-/*
- * Rather than storing the cblock in an entry, we allocate all entries in
- * an array, and infer the cblock from the entry position.
- *
- * Free entries are linked together into a list.
- */
-struct entry_pool {
-	struct entry *entries, *entries_end;
-	struct list_head free;
-	unsigned nr_allocated;
-};
-
-static int epool_init(struct entry_pool *ep, unsigned nr_entries)
-{
-	unsigned i;
-
-	ep->entries = vzalloc(sizeof(struct entry) * nr_entries);
-	if (!ep->entries)
-		return -ENOMEM;
-
-	ep->entries_end = ep->entries + nr_entries;
-
-	INIT_LIST_HEAD(&ep->free);
-	for (i = 0; i < nr_entries; i++)
-		list_add(&ep->entries[i].list, &ep->free);
-
-	ep->nr_allocated = 0;
-
-	return 0;
-}
-
-static void epool_exit(struct entry_pool *ep)
-{
-	vfree(ep->entries);
-}
-
-static struct entry *alloc_entry(struct entry_pool *ep)
-{
-	struct entry *e;
-
-	if (list_empty(&ep->free))
-		return NULL;
-
-	e = list_entry(list_pop(&ep->free), struct entry, list);
-	INIT_LIST_HEAD(&e->list);
-	INIT_HLIST_NODE(&e->hlist);
-	ep->nr_allocated++;
-
-	return e;
-}
-
-/*
- * This assumes the cblock hasn't already been allocated.
- */
-static struct entry *alloc_particular_entry(struct entry_pool *ep, dm_cblock_t cblock)
-{
-	struct entry *e = ep->entries + from_cblock(cblock);
-
-	list_del_init(&e->list);
-	INIT_HLIST_NODE(&e->hlist);
-	ep->nr_allocated++;
-
-	return e;
-}
-
-static void free_entry(struct entry_pool *ep, struct entry *e)
-{
-	BUG_ON(!ep->nr_allocated);
-	ep->nr_allocated--;
-	INIT_HLIST_NODE(&e->hlist);
-	list_add(&e->list, &ep->free);
-}
-
-/*
- * Returns NULL if the entry is free.
- */
-static struct entry *epool_find(struct entry_pool *ep, dm_cblock_t cblock)
-{
-	struct entry *e = ep->entries + from_cblock(cblock);
-	return !hlist_unhashed(&e->hlist) ? e : NULL;
-}
-
-static bool epool_empty(struct entry_pool *ep)
-{
-	return list_empty(&ep->free);
-}
-
-static bool in_pool(struct entry_pool *ep, struct entry *e)
-{
-	return e >= ep->entries && e < ep->entries_end;
-}
-
-static dm_cblock_t infer_cblock(struct entry_pool *ep, struct entry *e)
-{
-	return to_cblock(e - ep->entries);
-}
-
-/*----------------------------------------------------------------*/
-
-struct mq_policy {
-	struct dm_cache_policy policy;
-
-	/* protects everything */
-	struct mutex lock;
-	dm_cblock_t cache_size;
-	struct io_tracker tracker;
-
-	/*
-	 * Entries come from two pools, one of pre-cache entries, and one
-	 * for the cache proper.
-	 */
-	struct entry_pool pre_cache_pool;
-	struct entry_pool cache_pool;
-
-	/*
-	 * We maintain three queues of entries.  The cache proper,
-	 * consisting of a clean and dirty queue, contains the currently
-	 * active mappings.  Whereas the pre_cache tracks blocks that
-	 * are being hit frequently and potential candidates for promotion
-	 * to the cache.
-	 */
-	struct queue pre_cache;
-	struct queue cache_clean;
-	struct queue cache_dirty;
-
-	/*
-	 * Keeps track of time, incremented by the core.  We use this to
-	 * avoid attributing multiple hits within the same tick.
-	 *
-	 * Access to tick_protected should be done with the spin lock held.
-	 * It's copied to tick at the start of the map function (within the
-	 * mutex).
-	 */
-	spinlock_t tick_lock;
-	unsigned tick_protected;
-	unsigned tick;
-
-	/*
-	 * A count of the number of times the map function has been called
-	 * and found an entry in the pre_cache or cache.  Currently used to
-	 * calculate the generation.
-	 */
-	unsigned hit_count;
-
-	/*
-	 * A generation is a longish period that is used to trigger some
-	 * book keeping effects.  eg, decrementing hit counts on entries.
-	 * This is needed to allow the cache to evolve as io patterns
-	 * change.
-	 */
-	unsigned generation;
-	unsigned generation_period; /* in lookups (will probably change) */
-
-	unsigned discard_promote_adjustment;
-	unsigned read_promote_adjustment;
-	unsigned write_promote_adjustment;
-
-	/*
-	 * The hash table allows us to quickly find an entry by origin
-	 * block.  Both pre_cache and cache entries are in here.
-	 */
-	unsigned nr_buckets;
-	dm_block_t hash_bits;
-	struct hlist_head *table;
-};
-
-#define DEFAULT_DISCARD_PROMOTE_ADJUSTMENT 1
-#define DEFAULT_READ_PROMOTE_ADJUSTMENT 4
-#define DEFAULT_WRITE_PROMOTE_ADJUSTMENT 8
-#define DISCOURAGE_DEMOTING_DIRTY_THRESHOLD 128
-
-/*----------------------------------------------------------------*/
-
-/*
- * Simple hash table implementation.  Should replace with the standard hash
- * table that's making its way upstream.
- */
-static void hash_insert(struct mq_policy *mq, struct entry *e)
-{
-	unsigned h = hash_64(from_oblock(e->oblock), mq->hash_bits);
-
-	hlist_add_head(&e->hlist, mq->table + h);
-}
-
-static struct entry *hash_lookup(struct mq_policy *mq, dm_oblock_t oblock)
-{
-	unsigned h = hash_64(from_oblock(oblock), mq->hash_bits);
-	struct hlist_head *bucket = mq->table + h;
-	struct entry *e;
-
-	hlist_for_each_entry(e, bucket, hlist)
-		if (e->oblock == oblock) {
-			hlist_del(&e->hlist);
-			hlist_add_head(&e->hlist, bucket);
-			return e;
-		}
-
-	return NULL;
-}
-
-static void hash_remove(struct entry *e)
-{
-	hlist_del(&e->hlist);
-}
-
-/*----------------------------------------------------------------*/
-
-static bool any_free_cblocks(struct mq_policy *mq)
-{
-	return !epool_empty(&mq->cache_pool);
-}
-
-static bool any_clean_cblocks(struct mq_policy *mq)
-{
-	return !queue_empty(&mq->cache_clean);
-}
-
-/*----------------------------------------------------------------*/
-
-/*
- * Now we get to the meat of the policy.  This section deals with deciding
- * when to to add entries to the pre_cache and cache, and move between
- * them.
- */
-
-/*
- * The queue level is based on the log2 of the hit count.
- */
-static unsigned queue_level(struct entry *e)
-{
-	return min((unsigned) ilog2(e->hit_count), NR_QUEUE_LEVELS - 1u);
-}
-
-static bool in_cache(struct mq_policy *mq, struct entry *e)
-{
-	return in_pool(&mq->cache_pool, e);
-}
-
-/*
- * Inserts the entry into the pre_cache or the cache.  Ensures the cache
- * block is marked as allocated if necc.  Inserts into the hash table.
- * Sets the tick which records when the entry was last moved about.
- */
-static void push(struct mq_policy *mq, struct entry *e)
-{
-	hash_insert(mq, e);
-
-	if (in_cache(mq, e))
-		queue_push(e->dirty ? &mq->cache_dirty : &mq->cache_clean,
-			   queue_level(e), &e->list);
-	else
-		queue_push(&mq->pre_cache, queue_level(e), &e->list);
-}
-
-/*
- * Removes an entry from pre_cache or cache.  Removes from the hash table.
- */
-static void del(struct mq_policy *mq, struct entry *e)
-{
-	if (in_cache(mq, e))
-		queue_remove(e->dirty ? &mq->cache_dirty : &mq->cache_clean, &e->list);
-	else
-		queue_remove(&mq->pre_cache, &e->list);
-
-	hash_remove(e);
-}
-
-/*
- * Like del, except it removes the first entry in the queue (ie. the least
- * recently used).
- */
-static struct entry *pop(struct mq_policy *mq, struct queue *q)
-{
-	struct entry *e;
-	struct list_head *h = queue_pop(q);
-
-	if (!h)
-		return NULL;
-
-	e = container_of(h, struct entry, list);
-	hash_remove(e);
-
-	return e;
-}
-
-static struct entry *pop_old(struct mq_policy *mq, struct queue *q)
-{
-	struct entry *e;
-	struct list_head *h = queue_pop_old(q);
-
-	if (!h)
-		return NULL;
-
-	e = container_of(h, struct entry, list);
-	hash_remove(e);
-
-	return e;
-}
-
-static struct entry *peek(struct queue *q)
-{
-	struct list_head *h = queue_peek(q);
-	return h ? container_of(h, struct entry, list) : NULL;
-}
-
-/*
- * The promotion threshold is adjusted every generation.  As are the counts
- * of the entries.
- *
- * At the moment the threshold is taken by averaging the hit counts of some
- * of the entries in the cache (the first 20 entries across all levels in
- * ascending order, giving preference to the clean entries at each level).
- *
- * We can be much cleverer than this though.  For example, each promotion
- * could bump up the threshold helping to prevent churn.  Much more to do
- * here.
- */
-
-#define MAX_TO_AVERAGE 20
-
-static void check_generation(struct mq_policy *mq)
-{
-	unsigned total = 0, nr = 0, count = 0, level;
-	struct list_head *head;
-	struct entry *e;
-
-	if ((mq->hit_count >= mq->generation_period) && (epool_empty(&mq->cache_pool))) {
-		mq->hit_count = 0;
-		mq->generation++;
-
-		for (level = 0; level < NR_QUEUE_LEVELS && count < MAX_TO_AVERAGE; level++) {
-			head = mq->cache_clean.qs + level;
-			list_for_each_entry(e, head, list) {
-				nr++;
-				total += e->hit_count;
-
-				if (++count >= MAX_TO_AVERAGE)
-					break;
-			}
-
-			head = mq->cache_dirty.qs + level;
-			list_for_each_entry(e, head, list) {
-				nr++;
-				total += e->hit_count;
-
-				if (++count >= MAX_TO_AVERAGE)
-					break;
-			}
-		}
-	}
-}
-
-/*
- * Whenever we use an entry we bump up it's hit counter, and push it to the
- * back to it's current level.
- */
-static void requeue(struct mq_policy *mq, struct entry *e)
-{
-	check_generation(mq);
-	del(mq, e);
-	push(mq, e);
-}
-
-/*
- * Demote the least recently used entry from the cache to the pre_cache.
- * Returns the new cache entry to use, and the old origin block it was
- * mapped to.
- *
- * We drop the hit count on the demoted entry back to 1 to stop it bouncing
- * straight back into the cache if it's subsequently hit.  There are
- * various options here, and more experimentation would be good:
- *
- * - just forget about the demoted entry completely (ie. don't insert it
-     into the pre_cache).
- * - divide the hit count rather that setting to some hard coded value.
- * - set the hit count to a hard coded value other than 1, eg, is it better
- *   if it goes in at level 2?
- */
-static int demote_cblock(struct mq_policy *mq,
-			 struct policy_locker *locker, dm_oblock_t *oblock)
-{
-	struct entry *demoted = peek(&mq->cache_clean);
-
-	if (!demoted)
-		/*
-		 * We could get a block from mq->cache_dirty, but that
-		 * would add extra latency to the triggering bio as it
-		 * waits for the writeback.  Better to not promote this
-		 * time and hope there's a clean block next time this block
-		 * is hit.
-		 */
-		return -ENOSPC;
-
-	if (locker->fn(locker, demoted->oblock))
-		/*
-		 * We couldn't lock the demoted block.
-		 */
-		return -EBUSY;
-
-	del(mq, demoted);
-	*oblock = demoted->oblock;
-	free_entry(&mq->cache_pool, demoted);
-
-	/*
-	 * We used to put the demoted block into the pre-cache, but I think
-	 * it's simpler to just let it work it's way up from zero again.
-	 * Stops blocks flickering in and out of the cache.
-	 */
-
-	return 0;
-}
-
-/*
- * Entries in the pre_cache whose hit count passes the promotion
- * threshold move to the cache proper.  Working out the correct
- * value for the promotion_threshold is crucial to this policy.
- */
-static unsigned promote_threshold(struct mq_policy *mq)
-{
-	struct entry *e;
-
-	if (any_free_cblocks(mq))
-		return 0;
-
-	e = peek(&mq->cache_clean);
-	if (e)
-		return e->hit_count;
-
-	e = peek(&mq->cache_dirty);
-	if (e)
-		return e->hit_count + DISCOURAGE_DEMOTING_DIRTY_THRESHOLD;
-
-	/* This should never happen */
-	return 0;
-}
-
-/*
- * We modify the basic promotion_threshold depending on the specific io.
- *
- * If the origin block has been discarded then there's no cost to copy it
- * to the cache.
- *
- * We bias towards reads, since they can be demoted at no cost if they
- * haven't been dirtied.
- */
-static unsigned adjusted_promote_threshold(struct mq_policy *mq,
-					   bool discarded_oblock, int data_dir)
-{
-	if (data_dir == READ)
-		return promote_threshold(mq) + mq->read_promote_adjustment;
-
-	if (discarded_oblock && (any_free_cblocks(mq) || any_clean_cblocks(mq))) {
-		/*
-		 * We don't need to do any copying at all, so give this a
-		 * very low threshold.
-		 */
-		return mq->discard_promote_adjustment;
-	}
-
-	return promote_threshold(mq) + mq->write_promote_adjustment;
-}
-
-static bool should_promote(struct mq_policy *mq, struct entry *e,
-			   bool discarded_oblock, int data_dir)
-{
-	return e->hit_count >=
-		adjusted_promote_threshold(mq, discarded_oblock, data_dir);
-}
-
-static int cache_entry_found(struct mq_policy *mq,
-			     struct entry *e,
-			     struct policy_result *result)
-{
-	requeue(mq, e);
-
-	if (in_cache(mq, e)) {
-		result->op = POLICY_HIT;
-		result->cblock = infer_cblock(&mq->cache_pool, e);
-	}
-
-	return 0;
-}
-
-/*
- * Moves an entry from the pre_cache to the cache.  The main work is
- * finding which cache block to use.
- */
-static int pre_cache_to_cache(struct mq_policy *mq, struct entry *e,
-			      struct policy_locker *locker,
-			      struct policy_result *result)
-{
-	int r;
-	struct entry *new_e;
-
-	/* Ensure there's a free cblock in the cache */
-	if (epool_empty(&mq->cache_pool)) {
-		result->op = POLICY_REPLACE;
-		r = demote_cblock(mq, locker, &result->old_oblock);
-		if (r) {
-			result->op = POLICY_MISS;
-			return 0;
-		}
-
-	} else
-		result->op = POLICY_NEW;
-
-	new_e = alloc_entry(&mq->cache_pool);
-	BUG_ON(!new_e);
-
-	new_e->oblock = e->oblock;
-	new_e->dirty = false;
-	new_e->hit_count = e->hit_count;
-
-	del(mq, e);
-	free_entry(&mq->pre_cache_pool, e);
-	push(mq, new_e);
-
-	result->cblock = infer_cblock(&mq->cache_pool, new_e);
-
-	return 0;
-}
-
-static int pre_cache_entry_found(struct mq_policy *mq, struct entry *e,
-				 bool can_migrate, bool discarded_oblock,
-				 int data_dir, struct policy_locker *locker,
-				 struct policy_result *result)
-{
-	int r = 0;
-
-	if (!should_promote(mq, e, discarded_oblock, data_dir)) {
-		requeue(mq, e);
-		result->op = POLICY_MISS;
-
-	} else if (!can_migrate)
-		r = -EWOULDBLOCK;
-
-	else {
-		requeue(mq, e);
-		r = pre_cache_to_cache(mq, e, locker, result);
-	}
-
-	return r;
-}
-
-static void insert_in_pre_cache(struct mq_policy *mq,
-				dm_oblock_t oblock)
-{
-	struct entry *e = alloc_entry(&mq->pre_cache_pool);
-
-	if (!e)
-		/*
-		 * There's no spare entry structure, so we grab the least
-		 * used one from the pre_cache.
-		 */
-		e = pop(mq, &mq->pre_cache);
-
-	if (unlikely(!e)) {
-		DMWARN("couldn't pop from pre cache");
-		return;
-	}
-
-	e->dirty = false;
-	e->oblock = oblock;
-	e->hit_count = 1;
-	push(mq, e);
-}
-
-static void insert_in_cache(struct mq_policy *mq, dm_oblock_t oblock,
-			    struct policy_locker *locker,
-			    struct policy_result *result)
-{
-	int r;
-	struct entry *e;
-
-	if (epool_empty(&mq->cache_pool)) {
-		result->op = POLICY_REPLACE;
-		r = demote_cblock(mq, locker, &result->old_oblock);
-		if (unlikely(r)) {
-			result->op = POLICY_MISS;
-			insert_in_pre_cache(mq, oblock);
-			return;
-		}
-
-		/*
-		 * This will always succeed, since we've just demoted.
-		 */
-		e = alloc_entry(&mq->cache_pool);
-		BUG_ON(!e);
-
-	} else {
-		e = alloc_entry(&mq->cache_pool);
-		result->op = POLICY_NEW;
-	}
-
-	e->oblock = oblock;
-	e->dirty = false;
-	e->hit_count = 1;
-	push(mq, e);
-
-	result->cblock = infer_cblock(&mq->cache_pool, e);
-}
-
-static int no_entry_found(struct mq_policy *mq, dm_oblock_t oblock,
-			  bool can_migrate, bool discarded_oblock,
-			  int data_dir, struct policy_locker *locker,
-			  struct policy_result *result)
-{
-	if (adjusted_promote_threshold(mq, discarded_oblock, data_dir) <= 1) {
-		if (can_migrate)
-			insert_in_cache(mq, oblock, locker, result);
-		else
-			return -EWOULDBLOCK;
-	} else {
-		insert_in_pre_cache(mq, oblock);
-		result->op = POLICY_MISS;
-	}
-
-	return 0;
-}
-
-/*
- * Looks the oblock up in the hash table, then decides whether to put in
- * pre_cache, or cache etc.
- */
-static int map(struct mq_policy *mq, dm_oblock_t oblock,
-	       bool can_migrate, bool discarded_oblock,
-	       int data_dir, struct policy_locker *locker,
-	       struct policy_result *result)
-{
-	int r = 0;
-	struct entry *e = hash_lookup(mq, oblock);
-
-	if (e && in_cache(mq, e))
-		r = cache_entry_found(mq, e, result);
-
-	else if (mq->tracker.thresholds[PATTERN_SEQUENTIAL] &&
-		 iot_pattern(&mq->tracker) == PATTERN_SEQUENTIAL)
-		result->op = POLICY_MISS;
-
-	else if (e)
-		r = pre_cache_entry_found(mq, e, can_migrate, discarded_oblock,
-					  data_dir, locker, result);
-
-	else
-		r = no_entry_found(mq, oblock, can_migrate, discarded_oblock,
-				   data_dir, locker, result);
-
-	if (r == -EWOULDBLOCK)
-		result->op = POLICY_MISS;
-
-	return r;
-}
-
-/*----------------------------------------------------------------*/
-
-/*
- * Public interface, via the policy struct.  See dm-cache-policy.h for a
- * description of these.
- */
-
-static struct mq_policy *to_mq_policy(struct dm_cache_policy *p)
-{
-	return container_of(p, struct mq_policy, policy);
-}
-
-static void mq_destroy(struct dm_cache_policy *p)
-{
-	struct mq_policy *mq = to_mq_policy(p);
-
-	vfree(mq->table);
-	epool_exit(&mq->cache_pool);
-	epool_exit(&mq->pre_cache_pool);
-	kfree(mq);
-}
-
-static void update_pre_cache_hits(struct list_head *h, void *context)
-{
-	struct entry *e = container_of(h, struct entry, list);
-	e->hit_count++;
-}
-
-static void update_cache_hits(struct list_head *h, void *context)
-{
-	struct mq_policy *mq = context;
-	struct entry *e = container_of(h, struct entry, list);
-	e->hit_count++;
-	mq->hit_count++;
-}
-
-static void copy_tick(struct mq_policy *mq)
-{
-	unsigned long flags, tick;
-
-	spin_lock_irqsave(&mq->tick_lock, flags);
-	tick = mq->tick_protected;
-	if (tick != mq->tick) {
-		queue_iterate_tick(&mq->pre_cache, update_pre_cache_hits, mq);
-		queue_iterate_tick(&mq->cache_dirty, update_cache_hits, mq);
-		queue_iterate_tick(&mq->cache_clean, update_cache_hits, mq);
-		mq->tick = tick;
-	}
-
-	queue_tick(&mq->pre_cache);
-	queue_tick(&mq->cache_dirty);
-	queue_tick(&mq->cache_clean);
-	queue_update_writeback_sentinels(&mq->cache_dirty);
-	spin_unlock_irqrestore(&mq->tick_lock, flags);
-}
-
-static int mq_map(struct dm_cache_policy *p, dm_oblock_t oblock,
-		  bool can_block, bool can_migrate, bool discarded_oblock,
-		  struct bio *bio, struct policy_locker *locker,
-		  struct policy_result *result)
-{
-	int r;
-	struct mq_policy *mq = to_mq_policy(p);
-
-	result->op = POLICY_MISS;
-
-	if (can_block)
-		mutex_lock(&mq->lock);
-	else if (!mutex_trylock(&mq->lock))
-		return -EWOULDBLOCK;
-
-	copy_tick(mq);
-
-	iot_examine_bio(&mq->tracker, bio);
-	r = map(mq, oblock, can_migrate, discarded_oblock,
-		bio_data_dir(bio), locker, result);
-
-	mutex_unlock(&mq->lock);
-
-	return r;
-}
-
-static int mq_lookup(struct dm_cache_policy *p, dm_oblock_t oblock, dm_cblock_t *cblock)
-{
-	int r;
-	struct mq_policy *mq = to_mq_policy(p);
-	struct entry *e;
-
-	if (!mutex_trylock(&mq->lock))
-		return -EWOULDBLOCK;
-
-	e = hash_lookup(mq, oblock);
-	if (e && in_cache(mq, e)) {
-		*cblock = infer_cblock(&mq->cache_pool, e);
-		r = 0;
-	} else
-		r = -ENOENT;
-
-	mutex_unlock(&mq->lock);
-
-	return r;
-}
-
-static void __mq_set_clear_dirty(struct mq_policy *mq, dm_oblock_t oblock, bool set)
-{
-	struct entry *e;
-
-	e = hash_lookup(mq, oblock);
-	BUG_ON(!e || !in_cache(mq, e));
-
-	del(mq, e);
-	e->dirty = set;
-	push(mq, e);
-}
-
-static void mq_set_dirty(struct dm_cache_policy *p, dm_oblock_t oblock)
-{
-	struct mq_policy *mq = to_mq_policy(p);
-
-	mutex_lock(&mq->lock);
-	__mq_set_clear_dirty(mq, oblock, true);
-	mutex_unlock(&mq->lock);
-}
-
-static void mq_clear_dirty(struct dm_cache_policy *p, dm_oblock_t oblock)
-{
-	struct mq_policy *mq = to_mq_policy(p);
-
-	mutex_lock(&mq->lock);
-	__mq_set_clear_dirty(mq, oblock, false);
-	mutex_unlock(&mq->lock);
-}
-
-static int mq_load_mapping(struct dm_cache_policy *p,
-			   dm_oblock_t oblock, dm_cblock_t cblock,
-			   uint32_t hint, bool hint_valid)
-{
-	struct mq_policy *mq = to_mq_policy(p);
-	struct entry *e;
-
-	e = alloc_particular_entry(&mq->cache_pool, cblock);
-	e->oblock = oblock;
-	e->dirty = false;	/* this gets corrected in a minute */
-	e->hit_count = hint_valid ? hint : 1;
-	push(mq, e);
-
-	return 0;
-}
-
-static int mq_save_hints(struct mq_policy *mq, struct queue *q,
-			 policy_walk_fn fn, void *context)
-{
-	int r;
-	unsigned level;
-	struct list_head *h;
-	struct entry *e;
-
-	for (level = 0; level < NR_QUEUE_LEVELS; level++)
-		list_for_each(h, q->qs + level) {
-			if (is_sentinel(q, h))
-				continue;
-
-			e = container_of(h, struct entry, list);
-			r = fn(context, infer_cblock(&mq->cache_pool, e),
-			       e->oblock, e->hit_count);
-			if (r)
-				return r;
-		}
-
-	return 0;
-}
-
-static int mq_walk_mappings(struct dm_cache_policy *p, policy_walk_fn fn,
-			    void *context)
-{
-	struct mq_policy *mq = to_mq_policy(p);
-	int r = 0;
-
-	mutex_lock(&mq->lock);
-
-	r = mq_save_hints(mq, &mq->cache_clean, fn, context);
-	if (!r)
-		r = mq_save_hints(mq, &mq->cache_dirty, fn, context);
-
-	mutex_unlock(&mq->lock);
-
-	return r;
-}
-
-static void __remove_mapping(struct mq_policy *mq, dm_oblock_t oblock)
-{
-	struct entry *e;
-
-	e = hash_lookup(mq, oblock);
-	BUG_ON(!e || !in_cache(mq, e));
-
-	del(mq, e);
-	free_entry(&mq->cache_pool, e);
-}
-
-static void mq_remove_mapping(struct dm_cache_policy *p, dm_oblock_t oblock)
-{
-	struct mq_policy *mq = to_mq_policy(p);
-
-	mutex_lock(&mq->lock);
-	__remove_mapping(mq, oblock);
-	mutex_unlock(&mq->lock);
-}
-
-static int __remove_cblock(struct mq_policy *mq, dm_cblock_t cblock)
-{
-	struct entry *e = epool_find(&mq->cache_pool, cblock);
-
-	if (!e)
-		return -ENODATA;
-
-	del(mq, e);
-	free_entry(&mq->cache_pool, e);
-
-	return 0;
-}
-
-static int mq_remove_cblock(struct dm_cache_policy *p, dm_cblock_t cblock)
-{
-	int r;
-	struct mq_policy *mq = to_mq_policy(p);
-
-	mutex_lock(&mq->lock);
-	r = __remove_cblock(mq, cblock);
-	mutex_unlock(&mq->lock);
-
-	return r;
-}
-
-#define CLEAN_TARGET_PERCENTAGE 25
-
-static bool clean_target_met(struct mq_policy *mq)
-{
-	/*
-	 * Cache entries may not be populated.  So we're cannot rely on the
-	 * size of the clean queue.
-	 */
-	unsigned nr_clean = from_cblock(mq->cache_size) - queue_size(&mq->cache_dirty);
-	unsigned target = from_cblock(mq->cache_size) * CLEAN_TARGET_PERCENTAGE / 100;
-
-	return nr_clean >= target;
-}
-
-static int __mq_writeback_work(struct mq_policy *mq, dm_oblock_t *oblock,
-			      dm_cblock_t *cblock)
-{
-	struct entry *e = pop_old(mq, &mq->cache_dirty);
-
-	if (!e && !clean_target_met(mq))
-		e = pop(mq, &mq->cache_dirty);
-
-	if (!e)
-		return -ENODATA;
-
-	*oblock = e->oblock;
-	*cblock = infer_cblock(&mq->cache_pool, e);
-	e->dirty = false;
-	push(mq, e);
-
-	return 0;
-}
-
-static int mq_writeback_work(struct dm_cache_policy *p, dm_oblock_t *oblock,
-			     dm_cblock_t *cblock, bool critical_only)
-{
-	int r;
-	struct mq_policy *mq = to_mq_policy(p);
-
-	mutex_lock(&mq->lock);
-	r = __mq_writeback_work(mq, oblock, cblock);
-	mutex_unlock(&mq->lock);
-
-	return r;
-}
-
-static void __force_mapping(struct mq_policy *mq,
-			    dm_oblock_t current_oblock, dm_oblock_t new_oblock)
-{
-	struct entry *e = hash_lookup(mq, current_oblock);
-
-	if (e && in_cache(mq, e)) {
-		del(mq, e);
-		e->oblock = new_oblock;
-		e->dirty = true;
-		push(mq, e);
-	}
-}
-
-static void mq_force_mapping(struct dm_cache_policy *p,
-			     dm_oblock_t current_oblock, dm_oblock_t new_oblock)
-{
-	struct mq_policy *mq = to_mq_policy(p);
-
-	mutex_lock(&mq->lock);
-	__force_mapping(mq, current_oblock, new_oblock);
-	mutex_unlock(&mq->lock);
-}
-
-static dm_cblock_t mq_residency(struct dm_cache_policy *p)
-{
-	dm_cblock_t r;
-	struct mq_policy *mq = to_mq_policy(p);
-
-	mutex_lock(&mq->lock);
-	r = to_cblock(mq->cache_pool.nr_allocated);
-	mutex_unlock(&mq->lock);
-
-	return r;
-}
-
-static void mq_tick(struct dm_cache_policy *p, bool can_block)
-{
-	struct mq_policy *mq = to_mq_policy(p);
-	unsigned long flags;
-
-	spin_lock_irqsave(&mq->tick_lock, flags);
-	mq->tick_protected++;
-	spin_unlock_irqrestore(&mq->tick_lock, flags);
-
-	if (can_block) {
-		mutex_lock(&mq->lock);
-		copy_tick(mq);
-		mutex_unlock(&mq->lock);
-	}
-}
-
-static int mq_set_config_value(struct dm_cache_policy *p,
-			       const char *key, const char *value)
-{
-	struct mq_policy *mq = to_mq_policy(p);
-	unsigned long tmp;
-
-	if (kstrtoul(value, 10, &tmp))
-		return -EINVAL;
-
-	if (!strcasecmp(key, "random_threshold")) {
-		mq->tracker.thresholds[PATTERN_RANDOM] = tmp;
-
-	} else if (!strcasecmp(key, "sequential_threshold")) {
-		mq->tracker.thresholds[PATTERN_SEQUENTIAL] = tmp;
-
-	} else if (!strcasecmp(key, "discard_promote_adjustment"))
-		mq->discard_promote_adjustment = tmp;
-
-	else if (!strcasecmp(key, "read_promote_adjustment"))
-		mq->read_promote_adjustment = tmp;
-
-	else if (!strcasecmp(key, "write_promote_adjustment"))
-		mq->write_promote_adjustment = tmp;
-
-	else
-		return -EINVAL;
-
-	return 0;
-}
-
-static int mq_emit_config_values(struct dm_cache_policy *p, char *result,
-				 unsigned maxlen, ssize_t *sz_ptr)
-{
-	ssize_t sz = *sz_ptr;
-	struct mq_policy *mq = to_mq_policy(p);
-
-	DMEMIT("10 random_threshold %u "
-	       "sequential_threshold %u "
-	       "discard_promote_adjustment %u "
-	       "read_promote_adjustment %u "
-	       "write_promote_adjustment %u ",
-	       mq->tracker.thresholds[PATTERN_RANDOM],
-	       mq->tracker.thresholds[PATTERN_SEQUENTIAL],
-	       mq->discard_promote_adjustment,
-	       mq->read_promote_adjustment,
-	       mq->write_promote_adjustment);
-
-	*sz_ptr = sz;
-	return 0;
-}
-
-/* Init the policy plugin interface function pointers. */
-static void init_policy_functions(struct mq_policy *mq)
-{
-	mq->policy.destroy = mq_destroy;
-	mq->policy.map = mq_map;
-	mq->policy.lookup = mq_lookup;
-	mq->policy.set_dirty = mq_set_dirty;
-	mq->policy.clear_dirty = mq_clear_dirty;
-	mq->policy.load_mapping = mq_load_mapping;
-	mq->policy.walk_mappings = mq_walk_mappings;
-	mq->policy.remove_mapping = mq_remove_mapping;
-	mq->policy.remove_cblock = mq_remove_cblock;
-	mq->policy.writeback_work = mq_writeback_work;
-	mq->policy.force_mapping = mq_force_mapping;
-	mq->policy.residency = mq_residency;
-	mq->policy.tick = mq_tick;
-	mq->policy.emit_config_values = mq_emit_config_values;
-	mq->policy.set_config_value = mq_set_config_value;
-}
-
-static struct dm_cache_policy *mq_create(dm_cblock_t cache_size,
-					 sector_t origin_size,
-					 sector_t cache_block_size)
-{
-	struct mq_policy *mq = kzalloc(sizeof(*mq), GFP_KERNEL);
-
-	if (!mq)
-		return NULL;
-
-	init_policy_functions(mq);
-	iot_init(&mq->tracker, SEQUENTIAL_THRESHOLD_DEFAULT, RANDOM_THRESHOLD_DEFAULT);
-	mq->cache_size = cache_size;
-
-	if (epool_init(&mq->pre_cache_pool, from_cblock(cache_size))) {
-		DMERR("couldn't initialize pool of pre-cache entries");
-		goto bad_pre_cache_init;
-	}
-
-	if (epool_init(&mq->cache_pool, from_cblock(cache_size))) {
-		DMERR("couldn't initialize pool of cache entries");
-		goto bad_cache_init;
-	}
-
-	mq->tick_protected = 0;
-	mq->tick = 0;
-	mq->hit_count = 0;
-	mq->generation = 0;
-	mq->discard_promote_adjustment = DEFAULT_DISCARD_PROMOTE_ADJUSTMENT;
-	mq->read_promote_adjustment = DEFAULT_READ_PROMOTE_ADJUSTMENT;
-	mq->write_promote_adjustment = DEFAULT_WRITE_PROMOTE_ADJUSTMENT;
-	mutex_init(&mq->lock);
-	spin_lock_init(&mq->tick_lock);
-
-	queue_init(&mq->pre_cache);
-	queue_init(&mq->cache_clean);
-	queue_init(&mq->cache_dirty);
-
-	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);
-	mq->table = vzalloc(sizeof(*mq->table) * mq->nr_buckets);
-	if (!mq->table)
-		goto bad_alloc_table;
-
-	return &mq->policy;
-
-bad_alloc_table:
-	epool_exit(&mq->cache_pool);
-bad_cache_init:
-	epool_exit(&mq->pre_cache_pool);
-bad_pre_cache_init:
-	kfree(mq);
-
-	return NULL;
-}
-
-/*----------------------------------------------------------------*/
-
-static struct dm_cache_policy_type mq_policy_type = {
-	.name = "mq",
-	.version = {1, 4, 0},
-	.hint_size = 4,
-	.owner = THIS_MODULE,
-	.create = mq_create
-};
-
-static int __init mq_init(void)
-{
-	int r;
-
-	mq_entry_cache = kmem_cache_create("dm_mq_policy_cache_entry",
-					   sizeof(struct entry),
-					   __alignof__(struct entry),
-					   0, NULL);
-	if (!mq_entry_cache)
-		return -ENOMEM;
-
-	r = dm_cache_policy_register(&mq_policy_type);
-	if (r) {
-		DMERR("register failed %d", r);
-		kmem_cache_destroy(mq_entry_cache);
-		return -ENOMEM;
-	}
-
-	return 0;
-}
-
-static void __exit mq_exit(void)
-{
-	dm_cache_policy_unregister(&mq_policy_type);
-
-	kmem_cache_destroy(mq_entry_cache);
-}
-
-module_init(mq_init);
-module_exit(mq_exit);
-
-MODULE_AUTHOR("Joe Thornber <dm-devel@redhat.com>");
-MODULE_LICENSE("GPL");
-MODULE_DESCRIPTION("mq cache policy");

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

@@ -1567,8 +1567,48 @@ static void smq_tick(struct dm_cache_policy *p, bool can_block)
 	spin_unlock_irqrestore(&mq->lock, flags);
 }
 
+/*
+ * smq has no config values, but the old mq policy did.  To avoid breaking
+ * software we continue to accept these configurables for the mq policy,
+ * but they have no effect.
+ */
+static int mq_set_config_value(struct dm_cache_policy *p,
+			       const char *key, const char *value)
+{
+	unsigned long tmp;
+
+	if (kstrtoul(value, 10, &tmp))
+		return -EINVAL;
+
+	if (!strcasecmp(key, "random_threshold") ||
+	    !strcasecmp(key, "sequential_threshold") ||
+	    !strcasecmp(key, "discard_promote_adjustment") ||
+	    !strcasecmp(key, "read_promote_adjustment") ||
+	    !strcasecmp(key, "write_promote_adjustment")) {
+		DMWARN("tunable '%s' no longer has any effect, mq policy is now an alias for smq", key);
+		return 0;
+	}
+
+	return -EINVAL;
+}
+
+static int mq_emit_config_values(struct dm_cache_policy *p, char *result,
+				 unsigned maxlen, ssize_t *sz_ptr)
+{
+	ssize_t sz = *sz_ptr;
+
+	DMEMIT("10 random_threshold 0 "
+	       "sequential_threshold 0 "
+	       "discard_promote_adjustment 0 "
+	       "read_promote_adjustment 0 "
+	       "write_promote_adjustment 0 ");
+
+	*sz_ptr = sz;
+	return 0;
+}
+
 /* Init the policy plugin interface function pointers. */
-static void init_policy_functions(struct smq_policy *mq)
+static void init_policy_functions(struct smq_policy *mq, bool mimic_mq)
 {
 	mq->policy.destroy = smq_destroy;
 	mq->policy.map = smq_map;
@@ -1583,6 +1623,11 @@ static void init_policy_functions(struct smq_policy *mq)
 	mq->policy.force_mapping = smq_force_mapping;
 	mq->policy.residency = smq_residency;
 	mq->policy.tick = smq_tick;
+
+	if (mimic_mq) {
+		mq->policy.set_config_value = mq_set_config_value;
+		mq->policy.emit_config_values = mq_emit_config_values;
+	}
 }
 
 static bool too_many_hotspot_blocks(sector_t origin_size,
@@ -1606,9 +1651,10 @@ static void calc_hotspot_params(sector_t origin_size,
 		*hotspot_block_size /= 2u;
 }
 
-static struct dm_cache_policy *smq_create(dm_cblock_t cache_size,
-					  sector_t origin_size,
-					  sector_t cache_block_size)
+static struct dm_cache_policy *__smq_create(dm_cblock_t cache_size,
+					    sector_t origin_size,
+					    sector_t cache_block_size,
+					    bool mimic_mq)
 {
 	unsigned i;
 	unsigned nr_sentinels_per_queue = 2u * NR_CACHE_LEVELS;
@@ -1618,7 +1664,7 @@ static struct dm_cache_policy *smq_create(dm_cblock_t cache_size,
 	if (!mq)
 		return NULL;
 
-	init_policy_functions(mq);
+	init_policy_functions(mq, mimic_mq);
 	mq->cache_size = cache_size;
 	mq->cache_block_size = cache_block_size;
 
@@ -1706,19 +1752,41 @@ bad_pool_init:
 	return NULL;
 }
 
+static struct dm_cache_policy *smq_create(dm_cblock_t cache_size,
+					  sector_t origin_size,
+					  sector_t cache_block_size)
+{
+	return __smq_create(cache_size, origin_size, cache_block_size, false);
+}
+
+static struct dm_cache_policy *mq_create(dm_cblock_t cache_size,
+					 sector_t origin_size,
+					 sector_t cache_block_size)
+{
+	return __smq_create(cache_size, origin_size, cache_block_size, true);
+}
+
 /*----------------------------------------------------------------*/
 
 static struct dm_cache_policy_type smq_policy_type = {
 	.name = "smq",
-	.version = {1, 0, 0},
+	.version = {1, 5, 0},
 	.hint_size = 4,
 	.owner = THIS_MODULE,
 	.create = smq_create
 };
 
+static struct dm_cache_policy_type mq_policy_type = {
+	.name = "mq",
+	.version = {1, 5, 0},
+	.hint_size = 4,
+	.owner = THIS_MODULE,
+	.create = mq_create,
+};
+
 static struct dm_cache_policy_type default_policy_type = {
 	.name = "default",
-	.version = {1, 4, 0},
+	.version = {1, 5, 0},
 	.hint_size = 4,
 	.owner = THIS_MODULE,
 	.create = smq_create,
@@ -1735,9 +1803,17 @@ static int __init smq_init(void)
 		return -ENOMEM;
 	}
 
+	r = dm_cache_policy_register(&mq_policy_type);
+	if (r) {
+		DMERR("register failed (as mq) %d", r);
+		dm_cache_policy_unregister(&smq_policy_type);
+		return -ENOMEM;
+	}
+
 	r = dm_cache_policy_register(&default_policy_type);
 	if (r) {
 		DMERR("register failed (as default) %d", r);
+		dm_cache_policy_unregister(&mq_policy_type);
 		dm_cache_policy_unregister(&smq_policy_type);
 		return -ENOMEM;
 	}
@@ -1748,6 +1824,7 @@ static int __init smq_init(void)
 static void __exit smq_exit(void)
 {
 	dm_cache_policy_unregister(&smq_policy_type);
+	dm_cache_policy_unregister(&mq_policy_type);
 	dm_cache_policy_unregister(&default_policy_type);
 }
 
@@ -1759,3 +1836,4 @@ MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("smq cache policy");
 
 MODULE_ALIAS("dm-cache-default");
+MODULE_ALIAS("dm-cache-mq");

drivers/md/dm-cache-target.c

@@ -984,9 +984,14 @@ static void notify_mode_switch(struct cache *cache, enum cache_metadata_mode mod
 
 static void set_cache_mode(struct cache *cache, enum cache_metadata_mode new_mode)
 {
-	bool needs_check = dm_cache_metadata_needs_check(cache->cmd);
+	bool needs_check;
 	enum cache_metadata_mode old_mode = get_cache_mode(cache);
 
+	if (dm_cache_metadata_needs_check(cache->cmd, &needs_check)) {
+		DMERR("unable to read needs_check flag, setting failure mode");
+		new_mode = CM_FAIL;
+	}
+
 	if (new_mode == CM_WRITE && needs_check) {
 		DMERR("%s: unable to switch cache to write mode until repaired.",
 		      cache_device_name(cache));
@@ -2771,7 +2776,7 @@ static int cache_create(struct cache_args *ca, struct cache **result)
 	ti->split_discard_bios = false;
 
 	cache->features = ca->features;
-	ti->per_bio_data_size = get_per_bio_data_size(cache);
+	ti->per_io_data_size = get_per_bio_data_size(cache);
 
 	cache->callbacks.congested_fn = cache_is_congested;
 	dm_table_add_target_callbacks(ti->table, &cache->callbacks);
@@ -3510,6 +3515,7 @@ static void cache_status(struct dm_target *ti, status_type_t type,
 	char buf[BDEVNAME_SIZE];
 	struct cache *cache = ti->private;
 	dm_cblock_t residency;
+	bool needs_check;
 
 	switch (type) {
 	case STATUSTYPE_INFO:
@@ -3583,7 +3589,9 @@ static void cache_status(struct dm_target *ti, status_type_t type,
 		else
 			DMEMIT("rw ");
 
-		if (dm_cache_metadata_needs_check(cache->cmd))
+		r = dm_cache_metadata_needs_check(cache->cmd, &needs_check);
+
+		if (r || needs_check)
 			DMEMIT("needs_check ");
 		else
 			DMEMIT("- ");
@@ -3806,7 +3814,7 @@ static void cache_io_hints(struct dm_target *ti, struct queue_limits *limits)
 
 static struct target_type cache_target = {
 	.name = "cache",
-	.version = {1, 8, 0},
+	.version = {1, 9, 0},
 	.module = THIS_MODULE,
 	.ctr = cache_ctr,
 	.dtr = cache_dtr,

drivers/md/dm-crypt.c

@@ -1788,7 +1788,7 @@ static int crypt_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 		goto bad;
 	}
 
-	cc->per_bio_data_size = ti->per_bio_data_size =
+	cc->per_bio_data_size = ti->per_io_data_size =
 		ALIGN(sizeof(struct dm_crypt_io) + cc->dmreq_start +
 		      sizeof(struct dm_crypt_request) + iv_size_padding + cc->iv_size,
 		      ARCH_KMALLOC_MINALIGN);

drivers/md/dm-delay.c

@@ -204,7 +204,7 @@ out:
 
 	ti->num_flush_bios = 1;
 	ti->num_discard_bios = 1;
-	ti->per_bio_data_size = sizeof(struct dm_delay_info);
+	ti->per_io_data_size = sizeof(struct dm_delay_info);
 	ti->private = dc;
 	return 0;
 

drivers/md/dm-flakey.c

@@ -220,7 +220,7 @@ static int flakey_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 
 	ti->num_flush_bios = 1;
 	ti->num_discard_bios = 1;
-	ti->per_bio_data_size = sizeof(struct per_bio_data);
+	ti->per_io_data_size = sizeof(struct per_bio_data);
 	ti->private = fc;
 	return 0;
 

drivers/md/dm-ioctl.c

@@ -1291,7 +1291,8 @@ static int table_load(struct dm_ioctl *param, size_t param_size)
 
 	immutable_target_type = dm_get_immutable_target_type(md);
 	if (immutable_target_type &&
-	    (immutable_target_type != dm_table_get_immutable_target_type(t))) {
+	    (immutable_target_type != dm_table_get_immutable_target_type(t)) &&
+	    !dm_table_get_wildcard_target(t)) {
 		DMWARN("can't replace immutable target type %s",
 		       immutable_target_type->name);
 		r = -EINVAL;
@@ -1303,7 +1304,7 @@ static int table_load(struct dm_ioctl *param, size_t param_size)
 		dm_set_md_type(md, dm_table_get_type(t));
 
 		/* setup md->queue to reflect md's type (may block) */
-		r = dm_setup_md_queue(md);
+		r = dm_setup_md_queue(md, t);
 		if (r) {
 			DMWARN("unable to set up device queue for new table.");
 			goto err_unlock_md_type;

drivers/md/dm-log-writes.c

@@ -475,7 +475,7 @@ static int log_writes_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 	ti->flush_supported = true;
 	ti->num_discard_bios = 1;
 	ti->discards_supported = true;
-	ti->per_bio_data_size = sizeof(struct per_bio_data);
+	ti->per_io_data_size = sizeof(struct per_bio_data);
 	ti->private = lc;
 	return 0;
 

drivers/md/dm-mpath.c

@@ -23,6 +23,7 @@
 #include <linux/delay.h>
 #include <scsi/scsi_dh.h>
 #include <linux/atomic.h>
+#include <linux/blk-mq.h>
 
 #define DM_MSG_PREFIX "multipath"
 #define DM_PG_INIT_DELAY_MSECS 2000
@@ -33,11 +34,12 @@ struct pgpath {
 	struct list_head list;
 
 	struct priority_group *pg;	/* Owning PG */
-	unsigned is_active;		/* Path status */
 	unsigned fail_count;		/* Cumulative failure count */
 
 	struct dm_path path;
 	struct delayed_work activate_path;
+
+	bool is_active:1;		/* Path status */
 };
 
 #define path_to_pgpath(__pgp) container_of((__pgp), struct pgpath, path)
@@ -53,10 +55,10 @@ struct priority_group {
 	struct path_selector ps;
 
 	unsigned pg_num;		/* Reference number */
-	unsigned bypassed;		/* Temporarily bypass this PG? */
-
 	unsigned nr_pgpaths;		/* Number of paths in PG */
 	struct list_head pgpaths;
+
+	bool bypassed:1;		/* Temporarily bypass this PG? */
 };
 
 /* Multipath context */
@@ -74,21 +76,20 @@ struct multipath {
 
 	wait_queue_head_t pg_init_wait;	/* Wait for pg_init completion */
 
-	unsigned pg_init_required;	/* pg_init needs calling? */
 	unsigned pg_init_in_progress;	/* Only one pg_init allowed at once */
-	unsigned pg_init_delay_retry;	/* Delay pg_init retry? */
 
 	unsigned nr_valid_paths;	/* Total number of usable paths */
 	struct pgpath *current_pgpath;
 	struct priority_group *current_pg;
 	struct priority_group *next_pg;	/* Switch to this PG if set */
-	unsigned repeat_count;		/* I/Os left before calling PS again */
 
-	unsigned queue_io:1;		/* Must we queue all I/O? */
-	unsigned queue_if_no_path:1;	/* Queue I/O if last path fails? */
-	unsigned saved_queue_if_no_path:1; /* Saved state during suspension */
-	unsigned retain_attached_hw_handler:1; /* If there's already a hw_handler present, don't change it. */
-	unsigned pg_init_disabled:1;	/* pg_init is not currently allowed */
+	bool queue_io:1;		/* Must we queue all I/O? */
+	bool queue_if_no_path:1;	/* Queue I/O if last path fails? */
+	bool saved_queue_if_no_path:1;	/* Saved state during suspension */
+	bool retain_attached_hw_handler:1; /* If there's already a hw_handler present, don't change it. */
+	bool pg_init_disabled:1;	/* pg_init is not currently allowed */
+	bool pg_init_required:1;	/* pg_init needs calling? */
+	bool pg_init_delay_retry:1;	/* Delay pg_init retry? */
 
 	unsigned pg_init_retries;	/* Number of times to retry pg_init */
 	unsigned pg_init_count;		/* Number of times pg_init called */
@@ -120,7 +121,6 @@ static struct kmem_cache *_mpio_cache;
 static struct workqueue_struct *kmultipathd, *kmpath_handlerd;
 static void trigger_event(struct work_struct *work);
 static void activate_path(struct work_struct *work);
-static int __pgpath_busy(struct pgpath *pgpath);
 
 
 /*-----------------------------------------------
@@ -132,7 +132,7 @@ static struct pgpath *alloc_pgpath(void)
 	struct pgpath *pgpath = kzalloc(sizeof(*pgpath), GFP_KERNEL);
 
 	if (pgpath) {
-		pgpath->is_active = 1;
+		pgpath->is_active = true;
 		INIT_DELAYED_WORK(&pgpath->activate_path, activate_path);
 	}
 
@@ -181,25 +181,31 @@ static void free_priority_group(struct priority_group *pg,
 	kfree(pg);
 }
 
-static struct multipath *alloc_multipath(struct dm_target *ti)
+static struct multipath *alloc_multipath(struct dm_target *ti, bool use_blk_mq)
 {
 	struct multipath *m;
-	unsigned min_ios = dm_get_reserved_rq_based_ios();
 
 	m = kzalloc(sizeof(*m), GFP_KERNEL);
 	if (m) {
 		INIT_LIST_HEAD(&m->priority_groups);
 		spin_lock_init(&m->lock);
-		m->queue_io = 1;
+		m->queue_io = true;
 		m->pg_init_delay_msecs = DM_PG_INIT_DELAY_DEFAULT;
 		INIT_WORK(&m->trigger_event, trigger_event);
 		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);
-			return NULL;
+
+		m->mpio_pool = NULL;
+		if (!use_blk_mq) {
+			unsigned min_ios = dm_get_reserved_rq_based_ios();
+
+			m->mpio_pool = mempool_create_slab_pool(min_ios, _mpio_cache);
+			if (!m->mpio_pool) {
+				kfree(m);
+				return NULL;
+			}
 		}
+
 		m->ti = ti;
 		ti->private = m;
 	}
@@ -222,26 +228,41 @@ static void free_multipath(struct multipath *m)
 	kfree(m);
 }
 
-static int set_mapinfo(struct multipath *m, union map_info *info)
+static struct dm_mpath_io *get_mpio(union map_info *info)
+{
+	return info->ptr;
+}
+
+static struct dm_mpath_io *set_mpio(struct multipath *m, union map_info *info)
 {
 	struct dm_mpath_io *mpio;
 
+	if (!m->mpio_pool) {
+		/* Use blk-mq pdu memory requested via per_io_data_size */
+		mpio = get_mpio(info);
+		memset(mpio, 0, sizeof(*mpio));
+		return mpio;
+	}
+
 	mpio = mempool_alloc(m->mpio_pool, GFP_ATOMIC);
 	if (!mpio)
-		return -ENOMEM;
+		return NULL;
 
 	memset(mpio, 0, sizeof(*mpio));
 	info->ptr = mpio;
 
-	return 0;
+	return mpio;
 }
 
-static void clear_mapinfo(struct multipath *m, union map_info *info)
+static void clear_request_fn_mpio(struct multipath *m, union map_info *info)
 {
-	struct dm_mpath_io *mpio = info->ptr;
+	/* Only needed for non blk-mq (.request_fn) multipath */
+	if (m->mpio_pool) {
+		struct dm_mpath_io *mpio = info->ptr;
 
-	info->ptr = NULL;
-	mempool_free(mpio, m->mpio_pool);
+		info->ptr = NULL;
+		mempool_free(mpio, m->mpio_pool);
+	}
 }
 
 /*-----------------------------------------------
@@ -257,7 +278,7 @@ static int __pg_init_all_paths(struct multipath *m)
 		return 0;
 
 	m->pg_init_count++;
-	m->pg_init_required = 0;
+	m->pg_init_required = false;
 
 	/* Check here to reset pg_init_required */
 	if (!m->current_pg)
@@ -283,11 +304,11 @@ static void __switch_pg(struct multipath *m, struct pgpath *pgpath)
 
 	/* Must we initialise the PG first, and queue I/O till it's ready? */
 	if (m->hw_handler_name) {
-		m->pg_init_required = 1;
-		m->queue_io = 1;
+		m->pg_init_required = true;
+		m->queue_io = true;
 	} else {
-		m->pg_init_required = 0;
-		m->queue_io = 0;
+		m->pg_init_required = false;
+		m->queue_io = false;
 	}
 
 	m->pg_init_count = 0;
@@ -298,7 +319,7 @@ static int __choose_path_in_pg(struct multipath *m, struct priority_group *pg,
 {
 	struct dm_path *path;
 
-	path = pg->ps.type->select_path(&pg->ps, &m->repeat_count, nr_bytes);
+	path = pg->ps.type->select_path(&pg->ps, nr_bytes);
 	if (!path)
 		return -ENXIO;
 
@@ -313,10 +334,10 @@ static int __choose_path_in_pg(struct multipath *m, struct priority_group *pg,
 static void __choose_pgpath(struct multipath *m, size_t nr_bytes)
 {
 	struct priority_group *pg;
-	unsigned bypassed = 1;
+	bool bypassed = true;
 
 	if (!m->nr_valid_paths) {
-		m->queue_io = 0;
+		m->queue_io = false;
 		goto failed;
 	}
 
@@ -344,7 +365,7 @@ static void __choose_pgpath(struct multipath *m, size_t nr_bytes)
 				continue;
 			if (!__choose_path_in_pg(m, pg, nr_bytes)) {
 				if (!bypassed)
-					m->pg_init_delay_retry = 1;
+					m->pg_init_delay_retry = true;
 				return;
 			}
 		}
@@ -380,7 +401,7 @@ static int __multipath_map(struct dm_target *ti, struct request *clone,
 			   union map_info *map_context,
 			   struct request *rq, struct request **__clone)
 {
-	struct multipath *m = (struct multipath *) ti->private;
+	struct multipath *m = ti->private;
 	int r = DM_MAPIO_REQUEUE;
 	size_t nr_bytes = clone ? blk_rq_bytes(clone) : blk_rq_bytes(rq);
 	struct pgpath *pgpath;
@@ -390,8 +411,7 @@ static int __multipath_map(struct dm_target *ti, struct request *clone,
 	spin_lock_irq(&m->lock);
 
 	/* Do we need to select a new pgpath? */
-	if (!m->current_pgpath ||
-	    (!m->queue_io && (m->repeat_count && --m->repeat_count == 0)))
+	if (!m->current_pgpath || !m->queue_io)
 		__choose_pgpath(m, nr_bytes);
 
 	pgpath = m->current_pgpath;
@@ -405,11 +425,11 @@ static int __multipath_map(struct dm_target *ti, struct request *clone,
 		goto out_unlock;
 	}
 
-	if (set_mapinfo(m, map_context) < 0)
+	mpio = set_mpio(m, map_context);
+	if (!mpio)
 		/* ENOMEM, requeue */
 		goto out_unlock;
 
-	mpio = map_context->ptr;
 	mpio->pgpath = pgpath;
 	mpio->nr_bytes = nr_bytes;
 
@@ -418,17 +438,24 @@ static int __multipath_map(struct dm_target *ti, struct request *clone,
 	spin_unlock_irq(&m->lock);
 
 	if (clone) {
-		/* Old request-based interface: allocated clone is passed in */
+		/*
+		 * Old request-based interface: allocated clone is passed in.
+		 * Used by: .request_fn stacked on .request_fn path(s).
+		 */
 		clone->q = bdev_get_queue(bdev);
 		clone->rq_disk = bdev->bd_disk;
 		clone->cmd_flags |= REQ_FAILFAST_TRANSPORT;
 	} else {
-		/* blk-mq request-based interface */
-		*__clone = blk_get_request(bdev_get_queue(bdev),
-					   rq_data_dir(rq), GFP_ATOMIC);
+		/*
+		 * blk-mq request-based interface; used by both:
+		 * .request_fn stacked on blk-mq path(s) and
+		 * blk-mq stacked on blk-mq path(s).
+		 */
+		*__clone = blk_mq_alloc_request(bdev_get_queue(bdev),
+						rq_data_dir(rq), BLK_MQ_REQ_NOWAIT);
 		if (IS_ERR(*__clone)) {
 			/* ENOMEM, requeue */
-			clear_mapinfo(m, map_context);
+			clear_request_fn_mpio(m, map_context);
 			return r;
 		}
 		(*__clone)->bio = (*__clone)->biotail = NULL;
@@ -463,14 +490,14 @@ static int multipath_clone_and_map(struct dm_target *ti, struct request *rq,
 
 static void multipath_release_clone(struct request *clone)
 {
-	blk_put_request(clone);
+	blk_mq_free_request(clone);
 }
 
 /*
  * If we run out of usable paths, should we queue I/O or error it?
  */
-static int queue_if_no_path(struct multipath *m, unsigned queue_if_no_path,
-			    unsigned save_old_value)
+static int queue_if_no_path(struct multipath *m, bool queue_if_no_path,
+			    bool save_old_value)
 {
 	unsigned long flags;
 
@@ -776,12 +803,12 @@ static int parse_features(struct dm_arg_set *as, struct multipath *m)
 		argc--;
 
 		if (!strcasecmp(arg_name, "queue_if_no_path")) {
-			r = queue_if_no_path(m, 1, 0);
+			r = queue_if_no_path(m, true, false);
 			continue;
 		}
 
 		if (!strcasecmp(arg_name, "retain_attached_hw_handler")) {
-			m->retain_attached_hw_handler = 1;
+			m->retain_attached_hw_handler = true;
 			continue;
 		}
 
@@ -820,11 +847,12 @@ static int multipath_ctr(struct dm_target *ti, unsigned int argc,
 	struct dm_arg_set as;
 	unsigned pg_count = 0;
 	unsigned next_pg_num;
+	bool use_blk_mq = dm_use_blk_mq(dm_table_get_md(ti->table));
 
 	as.argc = argc;
 	as.argv = argv;
 
-	m = alloc_multipath(ti);
+	m = alloc_multipath(ti, use_blk_mq);
 	if (!m) {
 		ti->error = "can't allocate multipath";
 		return -EINVAL;
@@ -880,6 +908,8 @@ static int multipath_ctr(struct dm_target *ti, unsigned int argc,
 	ti->num_flush_bios = 1;
 	ti->num_discard_bios = 1;
 	ti->num_write_same_bios = 1;
+	if (use_blk_mq)
+		ti->per_io_data_size = sizeof(struct dm_mpath_io);
 
 	return 0;
 
@@ -917,7 +947,7 @@ static void flush_multipath_work(struct multipath *m)
 	unsigned long flags;
 
 	spin_lock_irqsave(&m->lock, flags);
-	m->pg_init_disabled = 1;
+	m->pg_init_disabled = true;
 	spin_unlock_irqrestore(&m->lock, flags);
 
 	flush_workqueue(kmpath_handlerd);
@@ -926,7 +956,7 @@ static void flush_multipath_work(struct multipath *m)
 	flush_work(&m->trigger_event);
 
 	spin_lock_irqsave(&m->lock, flags);
-	m->pg_init_disabled = 0;
+	m->pg_init_disabled = false;
 	spin_unlock_irqrestore(&m->lock, flags);
 }
 
@@ -954,7 +984,7 @@ static int fail_path(struct pgpath *pgpath)
 	DMWARN("Failing path %s.", pgpath->path.dev->name);
 
 	pgpath->pg->ps.type->fail_path(&pgpath->pg->ps, &pgpath->path);
-	pgpath->is_active = 0;
+	pgpath->is_active = false;
 	pgpath->fail_count++;
 
 	m->nr_valid_paths--;
@@ -987,18 +1017,13 @@ static int reinstate_path(struct pgpath *pgpath)
 	if (pgpath->is_active)
 		goto out;
 
-	if (!pgpath->pg->ps.type->reinstate_path) {
-		DMWARN("Reinstate path not supported by path selector %s",
-		       pgpath->pg->ps.type->name);
-		r = -EINVAL;
-		goto out;
-	}
+	DMWARN("Reinstating path %s.", pgpath->path.dev->name);
 
 	r = pgpath->pg->ps.type->reinstate_path(&pgpath->pg->ps, &pgpath->path);
 	if (r)
 		goto out;
 
-	pgpath->is_active = 1;
+	pgpath->is_active = true;
 
 	if (!m->nr_valid_paths++) {
 		m->current_pgpath = NULL;
@@ -1045,7 +1070,7 @@ static int action_dev(struct multipath *m, struct dm_dev *dev,
  * Temporarily try to avoid having to use the specified PG
  */
 static void bypass_pg(struct multipath *m, struct priority_group *pg,
-		      int bypassed)
+		      bool bypassed)
 {
 	unsigned long flags;
 
@@ -1078,7 +1103,7 @@ static int switch_pg_num(struct multipath *m, const char *pgstr)
 
 	spin_lock_irqsave(&m->lock, flags);
 	list_for_each_entry(pg, &m->priority_groups, list) {
-		pg->bypassed = 0;
+		pg->bypassed = false;
 		if (--pgnum)
 			continue;
 
@@ -1096,7 +1121,7 @@ static int switch_pg_num(struct multipath *m, const char *pgstr)
  * Set/clear bypassed status of a PG.
  * PGs are numbered upwards from 1 in the order they were declared.
  */
-static int bypass_pg_num(struct multipath *m, const char *pgstr, int bypassed)
+static int bypass_pg_num(struct multipath *m, const char *pgstr, bool bypassed)
 {
 	struct priority_group *pg;
 	unsigned pgnum;
@@ -1120,17 +1145,17 @@ static int bypass_pg_num(struct multipath *m, const char *pgstr, int bypassed)
 /*
  * Should we retry pg_init immediately?
  */
-static int pg_init_limit_reached(struct multipath *m, struct pgpath *pgpath)
+static bool pg_init_limit_reached(struct multipath *m, struct pgpath *pgpath)
 {
 	unsigned long flags;
-	int limit_reached = 0;
+	bool limit_reached = false;
 
 	spin_lock_irqsave(&m->lock, flags);
 
 	if (m->pg_init_count <= m->pg_init_retries && !m->pg_init_disabled)
-		m->pg_init_required = 1;
+		m->pg_init_required = true;
 	else
-		limit_reached = 1;
+		limit_reached = true;
 
 	spin_unlock_irqrestore(&m->lock, flags);
 
@@ -1143,7 +1168,7 @@ static void pg_init_done(void *data, int errors)
 	struct priority_group *pg = pgpath->pg;
 	struct multipath *m = pg->m;
 	unsigned long flags;
-	unsigned delay_retry = 0;
+	bool delay_retry = false;
 
 	/* device or driver problems */
 	switch (errors) {
@@ -1166,7 +1191,7 @@ static void pg_init_done(void *data, int errors)
 		 * Probably doing something like FW upgrade on the
 		 * controller so try the other pg.
 		 */
-		bypass_pg(m, pg, 1);
+		bypass_pg(m, pg, true);
 		break;
 	case SCSI_DH_RETRY:
 		/* Wait before retrying. */
@@ -1177,6 +1202,7 @@ static void pg_init_done(void *data, int errors)
 			fail_path(pgpath);
 		errors = 0;
 		break;
+	case SCSI_DH_DEV_OFFLINED:
 	default:
 		/*
 		 * We probably do not want to fail the path for a device
@@ -1194,7 +1220,7 @@ static void pg_init_done(void *data, int errors)
 			m->current_pg = NULL;
 		}
 	} else if (!m->pg_init_required)
-		pg->bypassed = 0;
+		pg->bypassed = false;
 
 	if (--m->pg_init_in_progress)
 		/* Activations of other paths are still on going */
@@ -1205,7 +1231,7 @@ static void pg_init_done(void *data, int errors)
 		if (__pg_init_all_paths(m))
 			goto out;
 	}
-	m->queue_io = 0;
+	m->queue_io = false;
 
 	/*
 	 * Wake up any thread waiting to suspend.
@@ -1291,21 +1317,21 @@ static int multipath_end_io(struct dm_target *ti, struct request *clone,
 			    int error, union map_info *map_context)
 {
 	struct multipath *m = ti->private;
-	struct dm_mpath_io *mpio = map_context->ptr;
+	struct dm_mpath_io *mpio = get_mpio(map_context);
 	struct pgpath *pgpath;
 	struct path_selector *ps;
 	int r;
 
 	BUG_ON(!mpio);
 
-	r  = do_end_io(m, clone, error, mpio);
+	r = do_end_io(m, clone, error, mpio);
 	pgpath = mpio->pgpath;
 	if (pgpath) {
 		ps = &pgpath->pg->ps;
 		if (ps->type->end_io)
 			ps->type->end_io(ps, &pgpath->path, mpio->nr_bytes);
 	}
-	clear_mapinfo(m, map_context);
+	clear_request_fn_mpio(m, map_context);
 
 	return r;
 }
@@ -1318,9 +1344,9 @@ static int multipath_end_io(struct dm_target *ti, struct request *clone,
  */
 static void multipath_presuspend(struct dm_target *ti)
 {
-	struct multipath *m = (struct multipath *) ti->private;
+	struct multipath *m = ti->private;
 
-	queue_if_no_path(m, 0, 1);
+	queue_if_no_path(m, false, true);
 }
 
 static void multipath_postsuspend(struct dm_target *ti)
@@ -1337,7 +1363,7 @@ static void multipath_postsuspend(struct dm_target *ti)
  */
 static void multipath_resume(struct dm_target *ti)
 {
-	struct multipath *m = (struct multipath *) ti->private;
+	struct multipath *m = ti->private;
 	unsigned long flags;
 
 	spin_lock_irqsave(&m->lock, flags);
@@ -1366,7 +1392,7 @@ static void multipath_status(struct dm_target *ti, status_type_t type,
 {
 	int sz = 0;
 	unsigned long flags;
-	struct multipath *m = (struct multipath *) ti->private;
+	struct multipath *m = ti->private;
 	struct priority_group *pg;
 	struct pgpath *p;
 	unsigned pg_num;
@@ -1474,7 +1500,7 @@ static int multipath_message(struct dm_target *ti, unsigned argc, char **argv)
 {
 	int r = -EINVAL;
 	struct dm_dev *dev;
-	struct multipath *m = (struct multipath *) ti->private;
+	struct multipath *m = ti->private;
 	action_fn action;
 
 	mutex_lock(&m->work_mutex);
@@ -1486,10 +1512,10 @@ static int multipath_message(struct dm_target *ti, unsigned argc, char **argv)
 
 	if (argc == 1) {
 		if (!strcasecmp(argv[0], "queue_if_no_path")) {
-			r = queue_if_no_path(m, 1, 0);
+			r = queue_if_no_path(m, true, false);
 			goto out;
 		} else if (!strcasecmp(argv[0], "fail_if_no_path")) {
-			r = queue_if_no_path(m, 0, 0);
+			r = queue_if_no_path(m, false, false);
 			goto out;
 		}
 	}
@@ -1500,10 +1526,10 @@ static int multipath_message(struct dm_target *ti, unsigned argc, char **argv)
 	}
 
 	if (!strcasecmp(argv[0], "disable_group")) {
-		r = bypass_pg_num(m, argv[1], 1);
+		r = bypass_pg_num(m, argv[1], true);
 		goto out;
 	} else if (!strcasecmp(argv[0], "enable_group")) {
-		r = bypass_pg_num(m, argv[1], 0);
+		r = bypass_pg_num(m, argv[1], false);
 		goto out;
 	} else if (!strcasecmp(argv[0], "switch_group")) {
 		r = switch_pg_num(m, argv[1]);
@@ -1604,7 +1630,7 @@ out:
 	return ret;
 }
 
-static int __pgpath_busy(struct pgpath *pgpath)
+static int pgpath_busy(struct pgpath *pgpath)
 {
 	struct request_queue *q = bdev_get_queue(pgpath->path.dev->bdev);
 
@@ -1621,7 +1647,7 @@ static int __pgpath_busy(struct pgpath *pgpath)
  */
 static int multipath_busy(struct dm_target *ti)
 {
-	int busy = 0, has_active = 0;
+	bool busy = false, has_active = false;
 	struct multipath *m = ti->private;
 	struct priority_group *pg;
 	struct pgpath *pgpath;
@@ -1632,7 +1658,7 @@ static int multipath_busy(struct dm_target *ti)
 	/* pg_init in progress or no paths available */
 	if (m->pg_init_in_progress ||
 	    (!m->nr_valid_paths && m->queue_if_no_path)) {
-		busy = 1;
+		busy = true;
 		goto out;
 	}
 	/* Guess which priority_group will be used at next mapping time */
@@ -1654,13 +1680,12 @@ static int multipath_busy(struct dm_target *ti)
 	 * 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;
+	busy = true;
 	list_for_each_entry(pgpath, &pg->pgpaths, list)
 		if (pgpath->is_active) {
-			has_active = 1;
-
-			if (!__pgpath_busy(pgpath)) {
-				busy = 0;
+			has_active = true;
+			if (!pgpath_busy(pgpath)) {
+				busy = false;
 				break;
 			}
 		}
@@ -1671,7 +1696,7 @@ static int multipath_busy(struct dm_target *ti)
 		 * the current_pg will be changed at next mapping time.
 		 * We need to try mapping to determine it.
 		 */
-		busy = 0;
+		busy = false;
 
 out:
 	spin_unlock_irqrestore(&m->lock, flags);
@@ -1684,7 +1709,8 @@ out:
  *---------------------------------------------------------------*/
 static struct target_type multipath_target = {
 	.name = "multipath",
-	.version = {1, 10, 0},
+	.version = {1, 11, 0},
+	.features = DM_TARGET_SINGLETON | DM_TARGET_IMMUTABLE,
 	.module = THIS_MODULE,
 	.ctr = multipath_ctr,
 	.dtr = multipath_dtr,

drivers/md/dm-path-selector.h

@@ -50,13 +50,8 @@ struct path_selector_type {
 	/*
 	 * Chooses a path for this io, if no paths are available then
 	 * NULL will be returned.
-	 *
-	 * repeat_count is the number of times to use the path before
-	 * calling the function again.  0 means don't call it again unless
-	 * the path fails.
 	 */
 	struct dm_path *(*select_path) (struct path_selector *ps,
-					unsigned *repeat_count,
 					size_t nr_bytes);
 
 	/*

drivers/md/dm-queue-length.c

@@ -23,12 +23,13 @@
 #include <linux/atomic.h>
 
 #define DM_MSG_PREFIX	"multipath queue-length"
-#define QL_MIN_IO	128
-#define QL_VERSION	"0.1.0"
+#define QL_MIN_IO	1
+#define QL_VERSION	"0.2.0"
 
 struct selector {
 	struct list_head	valid_paths;
 	struct list_head	failed_paths;
+	spinlock_t lock;
 };
 
 struct path_info {
@@ -45,6 +46,7 @@ static struct selector *alloc_selector(void)
 	if (s) {
 		INIT_LIST_HEAD(&s->valid_paths);
 		INIT_LIST_HEAD(&s->failed_paths);
+		spin_lock_init(&s->lock);
 	}
 
 	return s;
@@ -113,6 +115,7 @@ static int ql_add_path(struct path_selector *ps, struct dm_path *path,
 	struct path_info *pi;
 	unsigned repeat_count = QL_MIN_IO;
 	char dummy;
+	unsigned long flags;
 
 	/*
 	 * Arguments: [<repeat_count>]
@@ -129,6 +132,11 @@ static int ql_add_path(struct path_selector *ps, struct dm_path *path,
 		return -EINVAL;
 	}
 
+	if (repeat_count > 1) {
+		DMWARN_LIMIT("repeat_count > 1 is deprecated, using 1 instead");
+		repeat_count = 1;
+	}
+
 	/* Allocate the path information structure */
 	pi = kmalloc(sizeof(*pi), GFP_KERNEL);
 	if (!pi) {
@@ -142,7 +150,9 @@ static int ql_add_path(struct path_selector *ps, struct dm_path *path,
 
 	path->pscontext = pi;
 
+	spin_lock_irqsave(&s->lock, flags);
 	list_add_tail(&pi->list, &s->valid_paths);
+	spin_unlock_irqrestore(&s->lock, flags);
 
 	return 0;
 }
@@ -151,16 +161,22 @@ static void ql_fail_path(struct path_selector *ps, struct dm_path *path)
 {
 	struct selector *s = ps->context;
 	struct path_info *pi = path->pscontext;
+	unsigned long flags;
 
+	spin_lock_irqsave(&s->lock, flags);
 	list_move(&pi->list, &s->failed_paths);
+	spin_unlock_irqrestore(&s->lock, flags);
 }
 
 static int ql_reinstate_path(struct path_selector *ps, struct dm_path *path)
 {
 	struct selector *s = ps->context;
 	struct path_info *pi = path->pscontext;
+	unsigned long flags;
 
+	spin_lock_irqsave(&s->lock, flags);
 	list_move_tail(&pi->list, &s->valid_paths);
+	spin_unlock_irqrestore(&s->lock, flags);
 
 	return 0;
 }
@@ -168,14 +184,16 @@ static int ql_reinstate_path(struct path_selector *ps, struct dm_path *path)
 /*
  * Select a path having the minimum number of in-flight I/Os
  */
-static struct dm_path *ql_select_path(struct path_selector *ps,
-				      unsigned *repeat_count, size_t nr_bytes)
+static struct dm_path *ql_select_path(struct path_selector *ps, size_t nr_bytes)
 {
 	struct selector *s = ps->context;
 	struct path_info *pi = NULL, *best = NULL;
+	struct dm_path *ret = NULL;
+	unsigned long flags;
 
+	spin_lock_irqsave(&s->lock, flags);
 	if (list_empty(&s->valid_paths))
-		return NULL;
+		goto out;
 
 	/* Change preferred (first in list) path to evenly balance. */
 	list_move_tail(s->valid_paths.next, &s->valid_paths);
@@ -190,11 +208,12 @@ static struct dm_path *ql_select_path(struct path_selector *ps,
 	}
 
 	if (!best)
-		return NULL;
-
-	*repeat_count = best->repeat_count;
+		goto out;
 
-	return best->path;
+	ret = best->path;
+out:
+	spin_unlock_irqrestore(&s->lock, flags);
+	return ret;
 }
 
 static int ql_start_io(struct path_selector *ps, struct dm_path *path,

drivers/md/dm-raid1.c

@@ -1121,7 +1121,7 @@ static int mirror_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 
 	ti->num_flush_bios = 1;
 	ti->num_discard_bios = 1;
-	ti->per_bio_data_size = sizeof(struct dm_raid1_bio_record);
+	ti->per_io_data_size = sizeof(struct dm_raid1_bio_record);
 	ti->discard_zeroes_data_unsupported = true;
 
 	ms->kmirrord_wq = alloc_workqueue("kmirrord", WQ_MEM_RECLAIM, 0);

drivers/md/dm-round-robin.c

@@ -17,6 +17,8 @@
 #include <linux/module.h>
 
 #define DM_MSG_PREFIX "multipath round-robin"
+#define RR_MIN_IO     1000
+#define RR_VERSION    "1.1.0"
 
 /*-----------------------------------------------------------------
  * Path-handling code, paths are held in lists
@@ -41,23 +43,48 @@ static void free_paths(struct list_head *paths)
  * Round-robin selector
  *---------------------------------------------------------------*/
 
-#define RR_MIN_IO		1000
-
 struct selector {
 	struct list_head valid_paths;
 	struct list_head invalid_paths;
+	spinlock_t lock;
+	struct dm_path * __percpu *current_path;
+	struct percpu_counter repeat_count;
 };
 
+static void set_percpu_current_path(struct selector *s, struct dm_path *path)
+{
+	int cpu;
+
+	for_each_possible_cpu(cpu)
+		*per_cpu_ptr(s->current_path, cpu) = path;
+}
+
 static struct selector *alloc_selector(void)
 {
 	struct selector *s = kmalloc(sizeof(*s), GFP_KERNEL);
 
-	if (s) {
-		INIT_LIST_HEAD(&s->valid_paths);
-		INIT_LIST_HEAD(&s->invalid_paths);
-	}
+	if (!s)
+		return NULL;
+
+	INIT_LIST_HEAD(&s->valid_paths);
+	INIT_LIST_HEAD(&s->invalid_paths);
+	spin_lock_init(&s->lock);
+
+	s->current_path = alloc_percpu(struct dm_path *);
+	if (!s->current_path)
+		goto out_current_path;
+	set_percpu_current_path(s, NULL);
+
+	if (percpu_counter_init(&s->repeat_count, 0, GFP_KERNEL))
+		goto out_repeat_count;
 
 	return s;
+
+out_repeat_count:
+	free_percpu(s->current_path);
+out_current_path:
+	kfree(s);
+	return NULL;;
 }
 
 static int rr_create(struct path_selector *ps, unsigned argc, char **argv)
@@ -74,10 +101,12 @@ static int rr_create(struct path_selector *ps, unsigned argc, char **argv)
 
 static void rr_destroy(struct path_selector *ps)
 {
-	struct selector *s = (struct selector *) ps->context;
+	struct selector *s = ps->context;
 
 	free_paths(&s->valid_paths);
 	free_paths(&s->invalid_paths);
+	free_percpu(s->current_path);
+	percpu_counter_destroy(&s->repeat_count);
 	kfree(s);
 	ps->context = NULL;
 }
@@ -111,10 +140,11 @@ static int rr_status(struct path_selector *ps, struct dm_path *path,
 static int rr_add_path(struct path_selector *ps, struct dm_path *path,
 		       int argc, char **argv, char **error)
 {
-	struct selector *s = (struct selector *) ps->context;
+	struct selector *s = ps->context;
 	struct path_info *pi;
 	unsigned repeat_count = RR_MIN_IO;
 	char dummy;
+	unsigned long flags;
 
 	if (argc > 1) {
 		*error = "round-robin ps: incorrect number of arguments";
@@ -139,42 +169,65 @@ static int rr_add_path(struct path_selector *ps, struct dm_path *path,
 
 	path->pscontext = pi;
 
+	spin_lock_irqsave(&s->lock, flags);
 	list_add_tail(&pi->list, &s->valid_paths);
+	spin_unlock_irqrestore(&s->lock, flags);
 
 	return 0;
 }
 
 static void rr_fail_path(struct path_selector *ps, struct dm_path *p)
 {
-	struct selector *s = (struct selector *) ps->context;
+	unsigned long flags;
+	struct selector *s = ps->context;
 	struct path_info *pi = p->pscontext;
 
+	spin_lock_irqsave(&s->lock, flags);
+	if (p == *this_cpu_ptr(s->current_path))
+		set_percpu_current_path(s, NULL);
+
 	list_move(&pi->list, &s->invalid_paths);
+	spin_unlock_irqrestore(&s->lock, flags);
 }
 
 static int rr_reinstate_path(struct path_selector *ps, struct dm_path *p)
 {
-	struct selector *s = (struct selector *) ps->context;
+	unsigned long flags;
+	struct selector *s = ps->context;
 	struct path_info *pi = p->pscontext;
 
+	spin_lock_irqsave(&s->lock, flags);
 	list_move(&pi->list, &s->valid_paths);
+	spin_unlock_irqrestore(&s->lock, flags);
 
 	return 0;
 }
 
-static struct dm_path *rr_select_path(struct path_selector *ps,
-				      unsigned *repeat_count, size_t nr_bytes)
+static struct dm_path *rr_select_path(struct path_selector *ps, size_t nr_bytes)
 {
-	struct selector *s = (struct selector *) ps->context;
+	unsigned long flags;
+	struct selector *s = ps->context;
 	struct path_info *pi = NULL;
+	struct dm_path *current_path = NULL;
+
+	current_path = *this_cpu_ptr(s->current_path);
+	if (current_path) {
+		percpu_counter_dec(&s->repeat_count);
+		if (percpu_counter_read_positive(&s->repeat_count) > 0)
+			return current_path;
+	}
 
+	spin_lock_irqsave(&s->lock, flags);
 	if (!list_empty(&s->valid_paths)) {
 		pi = list_entry(s->valid_paths.next, struct path_info, list);
 		list_move_tail(&pi->list, &s->valid_paths);
-		*repeat_count = pi->repeat_count;
+		percpu_counter_set(&s->repeat_count, pi->repeat_count);
+		set_percpu_current_path(s, pi->path);
+		current_path = pi->path;
 	}
+	spin_unlock_irqrestore(&s->lock, flags);
 
-	return pi ? pi->path : NULL;
+	return current_path;
 }
 
 static struct path_selector_type rr_ps = {
@@ -198,7 +251,7 @@ static int __init dm_rr_init(void)
 	if (r < 0)
 		DMERR("register failed %d", r);
 
-	DMINFO("version 1.0.0 loaded");
+	DMINFO("version " RR_VERSION " loaded");
 
 	return r;
 }

drivers/md/dm-service-time.c

@@ -19,11 +19,12 @@
 #define ST_MAX_RELATIVE_THROUGHPUT	100
 #define ST_MAX_RELATIVE_THROUGHPUT_SHIFT	7
 #define ST_MAX_INFLIGHT_SIZE	((size_t)-1 >> ST_MAX_RELATIVE_THROUGHPUT_SHIFT)
-#define ST_VERSION	"0.2.0"
+#define ST_VERSION	"0.3.0"
 
 struct selector {
 	struct list_head valid_paths;
 	struct list_head failed_paths;
+	spinlock_t lock;
 };
 
 struct path_info {
@@ -41,6 +42,7 @@ static struct selector *alloc_selector(void)
 	if (s) {
 		INIT_LIST_HEAD(&s->valid_paths);
 		INIT_LIST_HEAD(&s->failed_paths);
+		spin_lock_init(&s->lock);
 	}
 
 	return s;
@@ -111,6 +113,7 @@ static int st_add_path(struct path_selector *ps, struct dm_path *path,
 	unsigned repeat_count = ST_MIN_IO;
 	unsigned relative_throughput = 1;
 	char dummy;
+	unsigned long flags;
 
 	/*
 	 * Arguments: [<repeat_count> [<relative_throughput>]]
@@ -134,6 +137,11 @@ static int st_add_path(struct path_selector *ps, struct dm_path *path,
 		return -EINVAL;
 	}
 
+	if (repeat_count > 1) {
+		DMWARN_LIMIT("repeat_count > 1 is deprecated, using 1 instead");
+		repeat_count = 1;
+	}
+
 	if ((argc == 2) &&
 	    (sscanf(argv[1], "%u%c", &relative_throughput, &dummy) != 1 ||
 	     relative_throughput > ST_MAX_RELATIVE_THROUGHPUT)) {
@@ -155,7 +163,9 @@ static int st_add_path(struct path_selector *ps, struct dm_path *path,
 
 	path->pscontext = pi;
 
+	spin_lock_irqsave(&s->lock, flags);
 	list_add_tail(&pi->list, &s->valid_paths);
+	spin_unlock_irqrestore(&s->lock, flags);
 
 	return 0;
 }
@@ -164,16 +174,22 @@ static void st_fail_path(struct path_selector *ps, struct dm_path *path)
 {
 	struct selector *s = ps->context;
 	struct path_info *pi = path->pscontext;
+	unsigned long flags;
 
+	spin_lock_irqsave(&s->lock, flags);
 	list_move(&pi->list, &s->failed_paths);
+	spin_unlock_irqrestore(&s->lock, flags);
 }
 
 static int st_reinstate_path(struct path_selector *ps, struct dm_path *path)
 {
 	struct selector *s = ps->context;
 	struct path_info *pi = path->pscontext;
+	unsigned long flags;
 
+	spin_lock_irqsave(&s->lock, flags);
 	list_move_tail(&pi->list, &s->valid_paths);
+	spin_unlock_irqrestore(&s->lock, flags);
 
 	return 0;
 }
@@ -255,14 +271,16 @@ static int st_compare_load(struct path_info *pi1, struct path_info *pi2,
 	return pi2->relative_throughput - pi1->relative_throughput;
 }
 
-static struct dm_path *st_select_path(struct path_selector *ps,
-				      unsigned *repeat_count, size_t nr_bytes)
+static struct dm_path *st_select_path(struct path_selector *ps, size_t nr_bytes)
 {
 	struct selector *s = ps->context;
 	struct path_info *pi = NULL, *best = NULL;
+	struct dm_path *ret = NULL;
+	unsigned long flags;
 
+	spin_lock_irqsave(&s->lock, flags);
 	if (list_empty(&s->valid_paths))
-		return NULL;
+		goto out;
 
 	/* Change preferred (first in list) path to evenly balance. */
 	list_move_tail(s->valid_paths.next, &s->valid_paths);
@@ -272,11 +290,12 @@ static struct dm_path *st_select_path(struct path_selector *ps,
 			best = pi;
 
 	if (!best)
-		return NULL;
-
-	*repeat_count = best->repeat_count;
+		goto out;
 
-	return best->path;
+	ret = best->path;
+out:
+	spin_unlock_irqrestore(&s->lock, flags);
+	return ret;
 }
 
 static int st_start_io(struct path_selector *ps, struct dm_path *path,

drivers/md/dm-snap.c

@@ -1105,6 +1105,7 @@ static int snapshot_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 	int i;
 	int r = -EINVAL;
 	char *origin_path, *cow_path;
+	dev_t origin_dev, cow_dev;
 	unsigned args_used, num_flush_bios = 1;
 	fmode_t origin_mode = FMODE_READ;
 
@@ -1135,11 +1136,19 @@ static int snapshot_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 		ti->error = "Cannot get origin device";
 		goto bad_origin;
 	}
+	origin_dev = s->origin->bdev->bd_dev;
 
 	cow_path = argv[0];
 	argv++;
 	argc--;
 
+	cow_dev = dm_get_dev_t(cow_path);
+	if (cow_dev && cow_dev == origin_dev) {
+		ti->error = "COW device cannot be the same as origin device";
+		r = -EINVAL;
+		goto bad_cow;
+	}
+
 	r = dm_get_device(ti, cow_path, dm_table_get_mode(ti->table), &s->cow);
 	if (r) {
 		ti->error = "Cannot get COW device";
@@ -1201,7 +1210,7 @@ static int snapshot_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 
 	ti->private = s;
 	ti->num_flush_bios = num_flush_bios;
-	ti->per_bio_data_size = sizeof(struct dm_snap_tracked_chunk);
+	ti->per_io_data_size = sizeof(struct dm_snap_tracked_chunk);
 
 	/* Add snapshot to the list of snapshots for this origin */
 	/* Exceptions aren't triggered till snapshot_resume() is called */

drivers/md/dm-table.c

@@ -364,6 +364,26 @@ static int upgrade_mode(struct dm_dev_internal *dd, fmode_t new_mode,
 	return 0;
 }
 
+/*
+ * Convert the path to a device
+ */
+dev_t dm_get_dev_t(const char *path)
+{
+	dev_t uninitialized_var(dev);
+	struct block_device *bdev;
+
+	bdev = lookup_bdev(path);
+	if (IS_ERR(bdev))
+		dev = name_to_dev_t(path);
+	else {
+		dev = bdev->bd_dev;
+		bdput(bdev);
+	}
+
+	return dev;
+}
+EXPORT_SYMBOL_GPL(dm_get_dev_t);
+
 /*
  * Add a device to the list, or just increment the usage count if
  * it's already present.
@@ -372,23 +392,15 @@ int dm_get_device(struct dm_target *ti, const char *path, fmode_t mode,
 		  struct dm_dev **result)
 {
 	int r;
-	dev_t uninitialized_var(dev);
+	dev_t dev;
 	struct dm_dev_internal *dd;
 	struct dm_table *t = ti->table;
-	struct block_device *bdev;
 
 	BUG_ON(!t);
 
-	/* convert the path to a device */
-	bdev = lookup_bdev(path);
-	if (IS_ERR(bdev)) {
-		dev = name_to_dev_t(path);
-		if (!dev)
-			return -ENODEV;
-	} else {
-		dev = bdev->bd_dev;
-		bdput(bdev);
-	}
+	dev = dm_get_dev_t(path);
+	if (!dev)
+		return -ENODEV;
 
 	dd = find_device(&t->devices, dev);
 	if (!dd) {
@@ -920,6 +932,30 @@ struct target_type *dm_table_get_immutable_target_type(struct dm_table *t)
 	return t->immutable_target_type;
 }
 
+struct dm_target *dm_table_get_immutable_target(struct dm_table *t)
+{
+	/* Immutable target is implicitly a singleton */
+	if (t->num_targets > 1 ||
+	    !dm_target_is_immutable(t->targets[0].type))
+		return NULL;
+
+	return t->targets;
+}
+
+struct dm_target *dm_table_get_wildcard_target(struct dm_table *t)
+{
+	struct dm_target *uninitialized_var(ti);
+	unsigned i = 0;
+
+	while (i < dm_table_get_num_targets(t)) {
+		ti = dm_table_get_target(t, i++);
+		if (dm_target_is_wildcard(ti->type))
+			return ti;
+	}
+
+	return NULL;
+}
+
 bool dm_table_request_based(struct dm_table *t)
 {
 	return __table_type_request_based(dm_table_get_type(t));
@@ -933,7 +969,7 @@ bool dm_table_mq_request_based(struct dm_table *t)
 static int dm_table_alloc_md_mempools(struct dm_table *t, struct mapped_device *md)
 {
 	unsigned type = dm_table_get_type(t);
-	unsigned per_bio_data_size = 0;
+	unsigned per_io_data_size = 0;
 	struct dm_target *tgt;
 	unsigned i;
 
@@ -945,10 +981,10 @@ static int dm_table_alloc_md_mempools(struct dm_table *t, struct mapped_device *
 	if (type == DM_TYPE_BIO_BASED)
 		for (i = 0; i < t->num_targets; i++) {
 			tgt = t->targets + i;
-			per_bio_data_size = max(per_bio_data_size, tgt->per_bio_data_size);
+			per_io_data_size = max(per_io_data_size, tgt->per_io_data_size);
 		}
 
-	t->mempools = dm_alloc_md_mempools(md, type, t->integrity_supported, per_bio_data_size);
+	t->mempools = dm_alloc_md_mempools(md, type, t->integrity_supported, per_io_data_size);
 	if (!t->mempools)
 		return -ENOMEM;
 

drivers/md/dm-target.c

@@ -150,7 +150,8 @@ static void io_err_release_clone_rq(struct request *clone)
 
 static struct target_type error_target = {
 	.name = "error",
-	.version = {1, 3, 0},
+	.version = {1, 4, 0},
+	.features = DM_TARGET_WILDCARD,
 	.ctr  = io_err_ctr,
 	.dtr  = io_err_dtr,
 	.map  = io_err_map,

drivers/md/dm-thin-metadata.c

@@ -344,7 +344,7 @@ static void subtree_dec(void *context, const void *value)
 	memcpy(&root_le, value, sizeof(root_le));
 	root = le64_to_cpu(root_le);
 	if (dm_btree_del(info, root))
-		DMERR("btree delete failed\n");
+		DMERR("btree delete failed");
 }
 
 static int subtree_equal(void *context, const void *value1_le, const void *value2_le)
@@ -1981,5 +1981,8 @@ bool dm_pool_metadata_needs_check(struct dm_pool_metadata *pmd)
 
 void dm_pool_issue_prefetches(struct dm_pool_metadata *pmd)
 {
-	dm_tm_issue_prefetches(pmd->tm);
+	down_read(&pmd->root_lock);
+	if (!pmd->fail_io)
+		dm_tm_issue_prefetches(pmd->tm);
+	up_read(&pmd->root_lock);
 }

drivers/md/dm-thin.c

@@ -235,6 +235,7 @@ struct pool {
 	struct pool_features pf;
 	bool low_water_triggered:1;	/* A dm event has been sent */
 	bool suspended:1;
+	bool out_of_data_space:1;
 
 	struct dm_bio_prison *prison;
 	struct dm_kcopyd_client *copier;
@@ -461,9 +462,16 @@ static void cell_error_with_code(struct pool *pool,
 	dm_bio_prison_free_cell(pool->prison, cell);
 }
 
+static int get_pool_io_error_code(struct pool *pool)
+{
+	return pool->out_of_data_space ? -ENOSPC : -EIO;
+}
+
 static void cell_error(struct pool *pool, struct dm_bio_prison_cell *cell)
 {
-	cell_error_with_code(pool, cell, -EIO);
+	int error = get_pool_io_error_code(pool);
+
+	cell_error_with_code(pool, cell, error);
 }
 
 static void cell_success(struct pool *pool, struct dm_bio_prison_cell *cell)
@@ -622,7 +630,9 @@ static void error_retry_list_with_code(struct pool *pool, int error)
 
 static void error_retry_list(struct pool *pool)
 {
-	return error_retry_list_with_code(pool, -EIO);
+	int error = get_pool_io_error_code(pool);
+
+	return error_retry_list_with_code(pool, error);
 }
 
 /*
@@ -2419,6 +2429,7 @@ static void set_pool_mode(struct pool *pool, enum pool_mode new_mode)
 		 */
 		if (old_mode != new_mode)
 			notify_of_pool_mode_change_to_oods(pool);
+		pool->out_of_data_space = true;
 		pool->process_bio = process_bio_read_only;
 		pool->process_discard = process_discard_bio;
 		pool->process_cell = process_cell_read_only;
@@ -2432,6 +2443,7 @@ static void set_pool_mode(struct pool *pool, enum pool_mode new_mode)
 	case PM_WRITE:
 		if (old_mode != new_mode)
 			notify_of_pool_mode_change(pool, "write");
+		pool->out_of_data_space = false;
 		pool->pf.error_if_no_space = pt->requested_pf.error_if_no_space;
 		dm_pool_metadata_read_write(pool->pmd);
 		pool->process_bio = process_bio;
@@ -2832,6 +2844,7 @@ static struct pool *pool_create(struct mapped_device *pool_md,
 	INIT_LIST_HEAD(&pool->active_thins);
 	pool->low_water_triggered = false;
 	pool->suspended = true;
+	pool->out_of_data_space = false;
 
 	pool->shared_read_ds = dm_deferred_set_create();
 	if (!pool->shared_read_ds) {
@@ -3886,7 +3899,7 @@ static struct target_type pool_target = {
 	.name = "thin-pool",
 	.features = DM_TARGET_SINGLETON | DM_TARGET_ALWAYS_WRITEABLE |
 		    DM_TARGET_IMMUTABLE,
-	.version = {1, 17, 0},
+	.version = {1, 18, 0},
 	.module = THIS_MODULE,
 	.ctr = pool_ctr,
 	.dtr = pool_dtr,
@@ -4037,7 +4050,7 @@ static int thin_ctr(struct dm_target *ti, unsigned argc, char **argv)
 
 	ti->num_flush_bios = 1;
 	ti->flush_supported = true;
-	ti->per_bio_data_size = sizeof(struct dm_thin_endio_hook);
+	ti->per_io_data_size = sizeof(struct dm_thin_endio_hook);
 
 	/* In case the pool supports discards, pass them on. */
 	ti->discard_zeroes_data_unsupported = true;
@@ -4260,7 +4273,7 @@ static void thin_io_hints(struct dm_target *ti, struct queue_limits *limits)
 
 static struct target_type thin_target = {
 	.name = "thin",
-	.version = {1, 17, 0},
+	.version = {1, 18, 0},
 	.module	= THIS_MODULE,
 	.ctr = thin_ctr,
 	.dtr = thin_dtr,

drivers/md/dm-verity-fec.c

@@ -812,7 +812,7 @@ int verity_fec_ctr(struct dm_verity *v)
 	}
 
 	/* Reserve space for our per-bio data */
-	ti->per_bio_data_size += sizeof(struct dm_verity_fec_io);
+	ti->per_io_data_size += sizeof(struct dm_verity_fec_io);
 
 	return 0;
 }

drivers/md/dm-verity-target.c

@@ -354,7 +354,7 @@ int verity_for_bv_block(struct dm_verity *v, struct dm_verity_io *io,
 				       size_t len))
 {
 	unsigned todo = 1 << v->data_dev_block_bits;
-	struct bio *bio = dm_bio_from_per_bio_data(io, v->ti->per_bio_data_size);
+	struct bio *bio = dm_bio_from_per_bio_data(io, v->ti->per_io_data_size);
 
 	do {
 		int r;
@@ -460,7 +460,7 @@ static int verity_verify_io(struct dm_verity_io *io)
 static void verity_finish_io(struct dm_verity_io *io, int error)
 {
 	struct dm_verity *v = io->v;
-	struct bio *bio = dm_bio_from_per_bio_data(io, v->ti->per_bio_data_size);
+	struct bio *bio = dm_bio_from_per_bio_data(io, v->ti->per_io_data_size);
 
 	bio->bi_end_io = io->orig_bi_end_io;
 	bio->bi_error = error;
@@ -574,7 +574,7 @@ static int verity_map(struct dm_target *ti, struct bio *bio)
 	if (bio_data_dir(bio) == WRITE)
 		return -EIO;
 
-	io = dm_per_bio_data(bio, ti->per_bio_data_size);
+	io = dm_per_bio_data(bio, ti->per_io_data_size);
 	io->v = v;
 	io->orig_bi_end_io = bio->bi_end_io;
 	io->block = bio->bi_iter.bi_sector >> (v->data_dev_block_bits - SECTOR_SHIFT);
@@ -1036,15 +1036,15 @@ static int verity_ctr(struct dm_target *ti, unsigned argc, char **argv)
 		goto bad;
 	}
 
-	ti->per_bio_data_size = sizeof(struct dm_verity_io) +
+	ti->per_io_data_size = sizeof(struct dm_verity_io) +
 				v->shash_descsize + v->digest_size * 2;
 
 	r = verity_fec_ctr(v);
 	if (r)
 		goto bad;
 
-	ti->per_bio_data_size = roundup(ti->per_bio_data_size,
-					__alignof__(struct dm_verity_io));
+	ti->per_io_data_size = roundup(ti->per_io_data_size,
+				       __alignof__(struct dm_verity_io));
 
 	return 0;
 

drivers/md/dm.c

@@ -106,14 +106,6 @@ struct dm_rq_clone_bio_info {
 	struct bio clone;
 };
 
-union map_info *dm_get_rq_mapinfo(struct request *rq)
-{
-	if (rq && rq->end_io_data)
-		return &((struct dm_rq_target_io *)rq->end_io_data)->info;
-	return NULL;
-}
-EXPORT_SYMBOL_GPL(dm_get_rq_mapinfo);
-
 #define MINOR_ALLOCED ((void *)-1)
 
 /*
@@ -128,29 +120,19 @@ EXPORT_SYMBOL_GPL(dm_get_rq_mapinfo);
 #define DMF_DEFERRED_REMOVE 6
 #define DMF_SUSPENDED_INTERNALLY 7
 
-/*
- * A dummy definition to make RCU happy.
- * struct dm_table should never be dereferenced in this file.
- */
-struct dm_table {
-	int undefined__;
-};
-
 /*
  * Work processed by per-device workqueue.
  */
 struct mapped_device {
 	struct srcu_struct io_barrier;
 	struct mutex suspend_lock;
-	atomic_t holders;
-	atomic_t open_count;
 
 	/*
-	 * The current mapping.
+	 * The current mapping (struct dm_table *).
 	 * Use dm_get_live_table{_fast} or take suspend_lock for
 	 * dereference.
 	 */
-	struct dm_table __rcu *map;
+	void __rcu *map;
 
 	struct list_head table_devices;
 	struct mutex table_devices_lock;
@@ -158,10 +140,16 @@ struct mapped_device {
 	unsigned long flags;
 
 	struct request_queue *queue;
+	int numa_node_id;
+
 	unsigned type;
 	/* Protect queue and type against concurrent access. */
 	struct mutex type_lock;
 
+	atomic_t holders;
+	atomic_t open_count;
+
+	struct dm_target *immutable_target;
 	struct target_type *immutable_target_type;
 
 	struct gendisk *disk;
@@ -175,8 +163,20 @@ struct mapped_device {
 	atomic_t pending[2];
 	wait_queue_head_t wait;
 	struct work_struct work;
-	struct bio_list deferred;
 	spinlock_t deferred_lock;
+	struct bio_list deferred;
+
+	/*
+	 * Event handling.
+	 */
+	wait_queue_head_t eventq;
+	atomic_t event_nr;
+	atomic_t uevent_seq;
+	struct list_head uevent_list;
+	spinlock_t uevent_lock; /* Protect access to uevent_list */
+
+	/* the number of internal suspends */
+	unsigned internal_suspend_count;
 
 	/*
 	 * Processing queue (flush)
@@ -191,33 +191,22 @@ struct mapped_device {
 
 	struct bio_set *bs;
 
-	/*
-	 * Event handling.
-	 */
-	atomic_t event_nr;
-	wait_queue_head_t eventq;
-	atomic_t uevent_seq;
-	struct list_head uevent_list;
-	spinlock_t uevent_lock; /* Protect access to uevent_list */
-
 	/*
 	 * freeze/thaw support require holding onto a super block
 	 */
 	struct super_block *frozen_sb;
-	struct block_device *bdev;
 
 	/* forced geometry settings */
 	struct hd_geometry geometry;
 
+	struct block_device *bdev;
+
 	/* kobject and completion */
 	struct dm_kobject_holder kobj_holder;
 
 	/* zero-length flush that will be cloned and submitted to targets */
 	struct bio flush_bio;
 
-	/* the number of internal suspends */
-	unsigned internal_suspend_count;
-
 	struct dm_stats stats;
 
 	struct kthread_worker kworker;
@@ -230,8 +219,9 @@ struct mapped_device {
 	ktime_t last_rq_start_time;
 
 	/* for blk-mq request-based DM support */
-	struct blk_mq_tag_set tag_set;
-	bool use_blk_mq;
+	struct blk_mq_tag_set *tag_set;
+	bool use_blk_mq:1;
+	bool init_tio_pdu:1;
 };
 
 #ifdef CONFIG_DM_MQ_DEFAULT
@@ -240,10 +230,19 @@ static bool use_blk_mq = true;
 static bool use_blk_mq = false;
 #endif
 
+#define DM_MQ_NR_HW_QUEUES 1
+#define DM_MQ_QUEUE_DEPTH 2048
+#define DM_NUMA_NODE NUMA_NO_NODE
+
+static unsigned dm_mq_nr_hw_queues = DM_MQ_NR_HW_QUEUES;
+static unsigned dm_mq_queue_depth = DM_MQ_QUEUE_DEPTH;
+static int dm_numa_node = DM_NUMA_NODE;
+
 bool dm_use_blk_mq(struct mapped_device *md)
 {
 	return md->use_blk_mq;
 }
+EXPORT_SYMBOL_GPL(dm_use_blk_mq);
 
 /*
  * For mempools pre-allocation at the table loading time.
@@ -277,6 +276,27 @@ static unsigned reserved_bio_based_ios = RESERVED_BIO_BASED_IOS;
  */
 static unsigned reserved_rq_based_ios = RESERVED_REQUEST_BASED_IOS;
 
+static int __dm_get_module_param_int(int *module_param, int min, int max)
+{
+	int param = ACCESS_ONCE(*module_param);
+	int modified_param = 0;
+	bool modified = true;
+
+	if (param < min)
+		modified_param = min;
+	else if (param > max)
+		modified_param = max;
+	else
+		modified = false;
+
+	if (modified) {
+		(void)cmpxchg(module_param, param, modified_param);
+		param = modified_param;
+	}
+
+	return param;
+}
+
 static unsigned __dm_get_module_param(unsigned *module_param,
 				      unsigned def, unsigned max)
 {
@@ -310,6 +330,23 @@ unsigned dm_get_reserved_rq_based_ios(void)
 }
 EXPORT_SYMBOL_GPL(dm_get_reserved_rq_based_ios);
 
+static unsigned dm_get_blk_mq_nr_hw_queues(void)
+{
+	return __dm_get_module_param(&dm_mq_nr_hw_queues, 1, 32);
+}
+
+static unsigned dm_get_blk_mq_queue_depth(void)
+{
+	return __dm_get_module_param(&dm_mq_queue_depth,
+				     DM_MQ_QUEUE_DEPTH, BLK_MQ_MAX_DEPTH);
+}
+
+static unsigned dm_get_numa_node(void)
+{
+	return __dm_get_module_param_int(&dm_numa_node,
+					 DM_NUMA_NODE, num_online_nodes() - 1);
+}
+
 static int __init local_init(void)
 {
 	int r = -ENOMEM;
@@ -323,7 +360,7 @@ static int __init local_init(void)
 	if (!_rq_tio_cache)
 		goto out_free_io_cache;
 
-	_rq_cache = kmem_cache_create("dm_clone_request", sizeof(struct request),
+	_rq_cache = kmem_cache_create("dm_old_clone_request", sizeof(struct request),
 				      __alignof__(struct request), 0, NULL);
 	if (!_rq_cache)
 		goto out_free_rq_tio_cache;
@@ -556,16 +593,17 @@ static int dm_blk_getgeo(struct block_device *bdev, struct hd_geometry *geo)
 	return dm_get_geometry(md, geo);
 }
 
-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)
+static int dm_grab_bdev_for_ioctl(struct mapped_device *md,
+				  struct block_device **bdev,
+				  fmode_t *mode)
 {
+	struct dm_target *tgt;
 	struct dm_table *map;
-	int r;
+	int srcu_idx, r;
 
 retry:
 	r = -ENOTTY;
-	map = dm_get_live_table(md, srcu_idx);
+	map = dm_get_live_table(md, &srcu_idx);
 	if (!map || !dm_table_get_size(map))
 		goto out;
 
@@ -573,9 +611,8 @@ retry:
 	if (dm_table_get_num_targets(map) != 1)
 		goto out;
 
-	*tgt = dm_table_get_target(map, 0);
-
-	if (!(*tgt)->type->prepare_ioctl)
+	tgt = dm_table_get_target(map, 0);
+	if (!tgt->type->prepare_ioctl)
 		goto out;
 
 	if (dm_suspended_md(md)) {
@@ -583,14 +620,16 @@ retry:
 		goto out;
 	}
 
-	r = (*tgt)->type->prepare_ioctl(*tgt, bdev, mode);
+	r = tgt->type->prepare_ioctl(tgt, bdev, mode);
 	if (r < 0)
 		goto out;
 
+	bdgrab(*bdev);
+	dm_put_live_table(md, srcu_idx);
 	return r;
 
 out:
-	dm_put_live_table(md, *srcu_idx);
+	dm_put_live_table(md, srcu_idx);
 	if (r == -ENOTCONN && !fatal_signal_pending(current)) {
 		msleep(10);
 		goto retry;
@@ -602,11 +641,9 @@ 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;
-	struct block_device *tgt_bdev = NULL;
-	int srcu_idx, r;
+	int r;
 
-	r = dm_get_live_table_for_ioctl(md, &tgt, &tgt_bdev, &mode, &srcu_idx);
+	r = dm_grab_bdev_for_ioctl(md, &bdev, &mode);
 	if (r < 0)
 		return r;
 
@@ -621,9 +658,9 @@ static int dm_blk_ioctl(struct block_device *bdev, fmode_t mode,
 			goto out;
 	}
 
-	r =  __blkdev_driver_ioctl(tgt_bdev, mode, cmd, arg);
+	r =  __blkdev_driver_ioctl(bdev, mode, cmd, arg);
 out:
-	dm_put_live_table(md, srcu_idx);
+	bdput(bdev);
 	return r;
 }
 
@@ -642,24 +679,24 @@ static void free_tio(struct mapped_device *md, struct dm_target_io *tio)
 	bio_put(&tio->clone);
 }
 
-static struct dm_rq_target_io *alloc_rq_tio(struct mapped_device *md,
-					    gfp_t gfp_mask)
+static struct dm_rq_target_io *alloc_old_rq_tio(struct mapped_device *md,
+						gfp_t gfp_mask)
 {
 	return mempool_alloc(md->io_pool, gfp_mask);
 }
 
-static void free_rq_tio(struct dm_rq_target_io *tio)
+static void free_old_rq_tio(struct dm_rq_target_io *tio)
 {
 	mempool_free(tio, tio->md->io_pool);
 }
 
-static struct request *alloc_clone_request(struct mapped_device *md,
-					   gfp_t gfp_mask)
+static struct request *alloc_old_clone_request(struct mapped_device *md,
+					       gfp_t gfp_mask)
 {
 	return mempool_alloc(md->rq_pool, gfp_mask);
 }
 
-static void free_clone_request(struct mapped_device *md, struct request *rq)
+static void free_old_clone_request(struct mapped_device *md, struct request *rq)
 {
 	mempool_free(rq, md->rq_pool);
 }
@@ -827,7 +864,7 @@ int dm_get_table_device(struct mapped_device *md, dev_t dev, fmode_t mode,
 	mutex_lock(&md->table_devices_lock);
 	td = find_table_device(&md->table_devices, dev, mode);
 	if (!td) {
-		td = kmalloc(sizeof(*td), GFP_KERNEL);
+		td = kmalloc_node(sizeof(*td), GFP_KERNEL, md->numa_node_id);
 		if (!td) {
 			mutex_unlock(&md->table_devices_lock);
 			return -ENOMEM;
@@ -1109,12 +1146,8 @@ static void rq_completed(struct mapped_device *md, int rw, bool run_queue)
 	 * back into ->request_fn() could deadlock attempting to grab the
 	 * queue lock again.
 	 */
-	if (run_queue) {
-		if (md->queue->mq_ops)
-			blk_mq_run_hw_queues(md->queue, true);
-		else
-			blk_run_queue_async(md->queue);
-	}
+	if (!md->queue->mq_ops && run_queue)
+		blk_run_queue_async(md->queue);
 
 	/*
 	 * dm_put() must be at the end of this function. See the comment above
@@ -1134,15 +1167,10 @@ static void free_rq_clone(struct request *clone)
 		tio->ti->type->release_clone_rq(clone);
 	else if (!md->queue->mq_ops)
 		/* request_fn queue stacked on request_fn queue(s) */
-		free_clone_request(md, clone);
-	/*
-	 * NOTE: for the blk-mq queue stacked on request_fn queue(s) case:
-	 * no need to call free_clone_request() because we leverage blk-mq by
-	 * allocating the clone at the end of the blk-mq pdu (see: clone_rq)
-	 */
+		free_old_clone_request(md, clone);
 
 	if (!md->queue->mq_ops)
-		free_rq_tio(tio);
+		free_old_rq_tio(tio);
 }
 
 /*
@@ -1192,13 +1220,13 @@ static void dm_unprep_request(struct request *rq)
 	if (clone)
 		free_rq_clone(clone);
 	else if (!tio->md->queue->mq_ops)
-		free_rq_tio(tio);
+		free_old_rq_tio(tio);
 }
 
 /*
  * Requeue the original request of a clone.
  */
-static void old_requeue_request(struct request *rq)
+static void dm_old_requeue_request(struct request *rq)
 {
 	struct request_queue *q = rq->q;
 	unsigned long flags;
@@ -1209,45 +1237,57 @@ static void old_requeue_request(struct request *rq)
 	spin_unlock_irqrestore(q->queue_lock, flags);
 }
 
+static void dm_mq_requeue_request(struct request *rq)
+{
+	struct request_queue *q = rq->q;
+	unsigned long flags;
+
+	blk_mq_requeue_request(rq);
+	spin_lock_irqsave(q->queue_lock, flags);
+	if (!blk_queue_stopped(q))
+		blk_mq_kick_requeue_list(q);
+	spin_unlock_irqrestore(q->queue_lock, flags);
+}
+
 static void dm_requeue_original_request(struct mapped_device *md,
 					struct request *rq)
 {
 	int rw = rq_data_dir(rq);
 
+	rq_end_stats(md, rq);
 	dm_unprep_request(rq);
 
-	rq_end_stats(md, rq);
 	if (!rq->q->mq_ops)
-		old_requeue_request(rq);
-	else {
-		blk_mq_requeue_request(rq);
-		blk_mq_kick_requeue_list(rq->q);
-	}
+		dm_old_requeue_request(rq);
+	else
+		dm_mq_requeue_request(rq);
 
 	rq_completed(md, rw, false);
 }
 
-static void old_stop_queue(struct request_queue *q)
+static void dm_old_stop_queue(struct request_queue *q)
 {
 	unsigned long flags;
 
-	if (blk_queue_stopped(q))
+	spin_lock_irqsave(q->queue_lock, flags);
+	if (blk_queue_stopped(q)) {
+		spin_unlock_irqrestore(q->queue_lock, flags);
 		return;
+	}
 
-	spin_lock_irqsave(q->queue_lock, flags);
 	blk_stop_queue(q);
 	spin_unlock_irqrestore(q->queue_lock, flags);
 }
 
-static void stop_queue(struct request_queue *q)
+static void dm_stop_queue(struct request_queue *q)
 {
 	if (!q->mq_ops)
-		old_stop_queue(q);
+		dm_old_stop_queue(q);
 	else
 		blk_mq_stop_hw_queues(q);
 }
 
-static void old_start_queue(struct request_queue *q)
+static void dm_old_start_queue(struct request_queue *q)
 {
 	unsigned long flags;
 
@@ -1257,12 +1297,14 @@ static void old_start_queue(struct request_queue *q)
 	spin_unlock_irqrestore(q->queue_lock, flags);
 }
 
-static void start_queue(struct request_queue *q)
+static void dm_start_queue(struct request_queue *q)
 {
 	if (!q->mq_ops)
-		old_start_queue(q);
-	else
+		dm_old_start_queue(q);
+	else {
 		blk_mq_start_stopped_hw_queues(q, true);
+		blk_mq_kick_requeue_list(q);
+	}
 }
 
 static void dm_done(struct request *clone, int error, bool mapped)
@@ -1313,7 +1355,7 @@ static void dm_softirq_done(struct request *rq)
 		if (!rq->q->mq_ops) {
 			blk_end_request_all(rq, tio->error);
 			rq_completed(tio->md, rw, false);
-			free_rq_tio(tio);
+			free_old_rq_tio(tio);
 		} else {
 			blk_mq_end_request(rq, tio->error);
 			rq_completed(tio->md, rw, false);
@@ -1336,7 +1378,10 @@ static void dm_complete_request(struct request *rq, int error)
 	struct dm_rq_target_io *tio = tio_from_request(rq);
 
 	tio->error = error;
-	blk_complete_request(rq);
+	if (!rq->q->mq_ops)
+		blk_complete_request(rq);
+	else
+		blk_mq_complete_request(rq, error);
 }
 
 /*
@@ -1352,7 +1397,7 @@ static void dm_kill_unmapped_request(struct request *rq, int error)
 }
 
 /*
- * Called with the clone's queue lock held (for non-blk-mq)
+ * Called with the clone's queue lock held (in the case of .request_fn)
  */
 static void end_clone_request(struct request *clone, int error)
 {
@@ -1522,21 +1567,26 @@ static void bio_setup_sector(struct bio *bio, sector_t sector, unsigned len)
 /*
  * Creates a bio that consists of range of complete bvecs.
  */
-static void clone_bio(struct dm_target_io *tio, struct bio *bio,
-		      sector_t sector, unsigned len)
+static int clone_bio(struct dm_target_io *tio, struct bio *bio,
+		     sector_t sector, unsigned len)
 {
 	struct bio *clone = &tio->clone;
 
 	__bio_clone_fast(clone, bio);
 
-	if (bio_integrity(bio))
-		bio_integrity_clone(clone, bio, GFP_NOIO);
+	if (bio_integrity(bio)) {
+		int r = bio_integrity_clone(clone, bio, GFP_NOIO);
+		if (r < 0)
+			return r;
+	}
 
 	bio_advance(clone, to_bytes(sector - clone->bi_iter.bi_sector));
 	clone->bi_iter.bi_size = to_bytes(len);
 
 	if (bio_integrity(bio))
 		bio_integrity_trim(clone, 0, len);
+
+	return 0;
 }
 
 static struct dm_target_io *alloc_tio(struct clone_info *ci,
@@ -1593,13 +1643,14 @@ static int __send_empty_flush(struct clone_info *ci)
 	return 0;
 }
 
-static void __clone_and_map_data_bio(struct clone_info *ci, struct dm_target *ti,
+static int __clone_and_map_data_bio(struct clone_info *ci, struct dm_target *ti,
 				     sector_t sector, unsigned *len)
 {
 	struct bio *bio = ci->bio;
 	struct dm_target_io *tio;
 	unsigned target_bio_nr;
 	unsigned num_target_bios = 1;
+	int r = 0;
 
 	/*
 	 * Does the target want to receive duplicate copies of the bio?
@@ -1610,9 +1661,13 @@ static void __clone_and_map_data_bio(struct clone_info *ci, struct dm_target *ti
 	for (target_bio_nr = 0; target_bio_nr < num_target_bios; target_bio_nr++) {
 		tio = alloc_tio(ci, ti, target_bio_nr);
 		tio->len_ptr = len;
-		clone_bio(tio, bio, sector, *len);
+		r = clone_bio(tio, bio, sector, *len);
+		if (r < 0)
+			break;
 		__map_bio(tio);
 	}
+
+	return r;
 }
 
 typedef unsigned (*get_num_bios_fn)(struct dm_target *ti);
@@ -1689,6 +1744,7 @@ static int __split_and_process_non_flush(struct clone_info *ci)
 	struct bio *bio = ci->bio;
 	struct dm_target *ti;
 	unsigned len;
+	int r;
 
 	if (unlikely(bio->bi_rw & REQ_DISCARD))
 		return __send_discard(ci);
@@ -1701,7 +1757,9 @@ static int __split_and_process_non_flush(struct clone_info *ci)
 
 	len = min_t(sector_t, max_io_len(ci->sector, ti), ci->sector_count);
 
-	__clone_and_map_data_bio(ci, ti, ci->sector, &len);
+	r = __clone_and_map_data_bio(ci, ti, ci->sector, &len);
+	if (r < 0)
+		return r;
 
 	ci->sector += len;
 	ci->sector_count -= len;
@@ -1839,28 +1897,22 @@ static int setup_clone(struct request *clone, struct request *rq,
 	return 0;
 }
 
-static struct request *clone_rq(struct request *rq, struct mapped_device *md,
-				struct dm_rq_target_io *tio, gfp_t gfp_mask)
+static struct request *clone_old_rq(struct request *rq, struct mapped_device *md,
+				    struct dm_rq_target_io *tio, gfp_t gfp_mask)
 {
 	/*
-	 * Do not allocate a clone if tio->clone was already set
-	 * (see: dm_mq_queue_rq).
+	 * Create clone for use with .request_fn request_queue
 	 */
-	bool alloc_clone = !tio->clone;
 	struct request *clone;
 
-	if (alloc_clone) {
-		clone = alloc_clone_request(md, gfp_mask);
-		if (!clone)
-			return NULL;
-	} else
-		clone = tio->clone;
+	clone = alloc_old_clone_request(md, gfp_mask);
+	if (!clone)
+		return NULL;
 
 	blk_rq_init(NULL, clone);
 	if (setup_clone(clone, rq, tio, gfp_mask)) {
 		/* -ENOMEM */
-		if (alloc_clone)
-			free_clone_request(md, clone);
+		free_old_clone_request(md, clone);
 		return NULL;
 	}
 
@@ -1877,29 +1929,40 @@ static void init_tio(struct dm_rq_target_io *tio, struct request *rq,
 	tio->clone = NULL;
 	tio->orig = rq;
 	tio->error = 0;
-	memset(&tio->info, 0, sizeof(tio->info));
+	/*
+	 * Avoid initializing info for blk-mq; it passes
+	 * target-specific data through info.ptr
+	 * (see: dm_mq_init_request)
+	 */
+	if (!md->init_tio_pdu)
+		memset(&tio->info, 0, sizeof(tio->info));
 	if (md->kworker_task)
 		init_kthread_work(&tio->work, map_tio_request);
 }
 
-static struct dm_rq_target_io *prep_tio(struct request *rq,
-					struct mapped_device *md, gfp_t gfp_mask)
+static struct dm_rq_target_io *dm_old_prep_tio(struct request *rq,
+					       struct mapped_device *md,
+					       gfp_t gfp_mask)
 {
 	struct dm_rq_target_io *tio;
 	int srcu_idx;
 	struct dm_table *table;
 
-	tio = alloc_rq_tio(md, gfp_mask);
+	tio = alloc_old_rq_tio(md, gfp_mask);
 	if (!tio)
 		return NULL;
 
 	init_tio(tio, rq, md);
 
 	table = dm_get_live_table(md, &srcu_idx);
+	/*
+	 * Must clone a request if this .request_fn DM device
+	 * is stacked on .request_fn device(s).
+	 */
 	if (!dm_table_mq_request_based(table)) {
-		if (!clone_rq(rq, md, tio, gfp_mask)) {
+		if (!clone_old_rq(rq, md, tio, gfp_mask)) {
 			dm_put_live_table(md, srcu_idx);
-			free_rq_tio(tio);
+			free_old_rq_tio(tio);
 			return NULL;
 		}
 	}
@@ -1911,7 +1974,7 @@ static struct dm_rq_target_io *prep_tio(struct request *rq,
 /*
  * Called with the queue lock held.
  */
-static int dm_prep_fn(struct request_queue *q, struct request *rq)
+static int dm_old_prep_fn(struct request_queue *q, struct request *rq)
 {
 	struct mapped_device *md = q->queuedata;
 	struct dm_rq_target_io *tio;
@@ -1921,7 +1984,7 @@ static int dm_prep_fn(struct request_queue *q, struct request *rq)
 		return BLKPREP_KILL;
 	}
 
-	tio = prep_tio(rq, md, GFP_ATOMIC);
+	tio = dm_old_prep_tio(rq, md, GFP_ATOMIC);
 	if (!tio)
 		return BLKPREP_DEFER;
 
@@ -2079,12 +2142,18 @@ static bool dm_request_peeked_before_merge_deadline(struct mapped_device *md)
 static void dm_request_fn(struct request_queue *q)
 {
 	struct mapped_device *md = q->queuedata;
-	int srcu_idx;
-	struct dm_table *map = dm_get_live_table(md, &srcu_idx);
-	struct dm_target *ti;
+	struct dm_target *ti = md->immutable_target;
 	struct request *rq;
 	struct dm_rq_target_io *tio;
-	sector_t pos;
+	sector_t pos = 0;
+
+	if (unlikely(!ti)) {
+		int srcu_idx;
+		struct dm_table *map = dm_get_live_table(md, &srcu_idx);
+
+		ti = dm_table_find_target(map, pos);
+		dm_put_live_table(md, srcu_idx);
+	}
 
 	/*
 	 * For suspend, check blk_queue_stopped() and increment
@@ -2095,33 +2164,21 @@ static void dm_request_fn(struct request_queue *q)
 	while (!blk_queue_stopped(q)) {
 		rq = blk_peek_request(q);
 		if (!rq)
-			goto out;
+			return;
 
 		/* always use block 0 to find the target for flushes for now */
 		pos = 0;
 		if (!(rq->cmd_flags & REQ_FLUSH))
 			pos = blk_rq_pos(rq);
 
-		ti = dm_table_find_target(map, pos);
-		if (!dm_target_is_valid(ti)) {
-			/*
-			 * Must perform setup, that rq_completed() requires,
-			 * before calling dm_kill_unmapped_request
-			 */
-			DMERR_LIMIT("request attempted access beyond the end of device");
-			dm_start_request(md, rq);
-			dm_kill_unmapped_request(rq, -EIO);
-			continue;
+		if ((dm_request_peeked_before_merge_deadline(md) &&
+		     md_in_flight(md) && rq->bio && rq->bio->bi_vcnt == 1 &&
+		     md->last_rq_pos == pos && md->last_rq_rw == rq_data_dir(rq)) ||
+		    (ti->type->busy && ti->type->busy(ti))) {
+			blk_delay_queue(q, HZ / 100);
+			return;
 		}
 
-		if (dm_request_peeked_before_merge_deadline(md) &&
-		    md_in_flight(md) && rq->bio && rq->bio->bi_vcnt == 1 &&
-		    md->last_rq_pos == pos && md->last_rq_rw == rq_data_dir(rq))
-			goto delay_and_out;
-
-		if (ti->type->busy && ti->type->busy(ti))
-			goto delay_and_out;
-
 		dm_start_request(md, rq);
 
 		tio = tio_from_request(rq);
@@ -2130,13 +2187,6 @@ static void dm_request_fn(struct request_queue *q)
 		queue_kthread_work(&md->kworker, &tio->work);
 		BUG_ON(!irqs_disabled());
 	}
-
-	goto out;
-
-delay_and_out:
-	blk_delay_queue(q, HZ / 100);
-out:
-	dm_put_live_table(md, srcu_idx);
 }
 
 static int dm_any_congested(void *congested_data, int bdi_bits)
@@ -2146,19 +2196,18 @@ static int dm_any_congested(void *congested_data, int bdi_bits)
 	struct dm_table *map;
 
 	if (!test_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags)) {
-		map = dm_get_live_table_fast(md);
-		if (map) {
+		if (dm_request_based(md)) {
 			/*
-			 * Request-based dm cares about only own queue for
-			 * the query about congestion status of request_queue
+			 * With request-based DM we only need to check the
+			 * top-level queue for congestion.
 			 */
-			if (dm_request_based(md))
-				r = md->queue->backing_dev_info.wb.state &
-				    bdi_bits;
-			else
+			r = md->queue->backing_dev_info.wb.state & bdi_bits;
+		} else {
+			map = dm_get_live_table_fast(md);
+			if (map)
 				r = dm_table_any_congested(map, bdi_bits);
+			dm_put_live_table_fast(md);
 		}
-		dm_put_live_table_fast(md);
 	}
 
 	return r;
@@ -2238,7 +2287,7 @@ static void dm_init_md_queue(struct mapped_device *md)
 	md->queue->backing_dev_info.congested_data = md;
 }
 
-static void dm_init_old_md_queue(struct mapped_device *md)
+static void dm_init_normal_md_queue(struct mapped_device *md)
 {
 	md->use_blk_mq = false;
 	dm_init_md_queue(md);
@@ -2285,10 +2334,11 @@ static void cleanup_mapped_device(struct mapped_device *md)
  */
 static struct mapped_device *alloc_dev(int minor)
 {
-	int r;
-	struct mapped_device *md = kzalloc(sizeof(*md), GFP_KERNEL);
+	int r, numa_node_id = dm_get_numa_node();
+	struct mapped_device *md;
 	void *old_md;
 
+	md = kzalloc_node(sizeof(*md), GFP_KERNEL, numa_node_id);
 	if (!md) {
 		DMWARN("unable to allocate device, out of memory.");
 		return NULL;
@@ -2309,7 +2359,9 @@ static struct mapped_device *alloc_dev(int minor)
 	if (r < 0)
 		goto bad_io_barrier;
 
+	md->numa_node_id = numa_node_id;
 	md->use_blk_mq = use_blk_mq;
+	md->init_tio_pdu = false;
 	md->type = DM_TYPE_NONE;
 	mutex_init(&md->suspend_lock);
 	mutex_init(&md->type_lock);
@@ -2323,13 +2375,13 @@ static struct mapped_device *alloc_dev(int minor)
 	INIT_LIST_HEAD(&md->table_devices);
 	spin_lock_init(&md->uevent_lock);
 
-	md->queue = blk_alloc_queue(GFP_KERNEL);
+	md->queue = blk_alloc_queue_node(GFP_KERNEL, numa_node_id);
 	if (!md->queue)
 		goto bad;
 
 	dm_init_md_queue(md);
 
-	md->disk = alloc_disk(1);
+	md->disk = alloc_disk_node(1, numa_node_id);
 	if (!md->disk)
 		goto bad;
 
@@ -2393,8 +2445,10 @@ static void free_dev(struct mapped_device *md)
 	unlock_fs(md);
 
 	cleanup_mapped_device(md);
-	if (md->use_blk_mq)
-		blk_mq_free_tag_set(&md->tag_set);
+	if (md->tag_set) {
+		blk_mq_free_tag_set(md->tag_set);
+		kfree(md->tag_set);
+	}
 
 	free_table_devices(&md->table_devices);
 	dm_stats_cleanup(&md->stats);
@@ -2502,13 +2556,20 @@ static struct dm_table *__bind(struct mapped_device *md, struct dm_table *t,
 	 * This must be done before setting the queue restrictions,
 	 * because request-based dm may be run just after the setting.
 	 */
-	if (dm_table_request_based(t))
-		stop_queue(q);
+	if (dm_table_request_based(t)) {
+		dm_stop_queue(q);
+		/*
+		 * Leverage the fact that request-based DM targets are
+		 * immutable singletons and establish md->immutable_target
+		 * - used to optimize both dm_request_fn and dm_mq_queue_rq
+		 */
+		md->immutable_target = dm_table_get_immutable_target(t);
+	}
 
 	__bind_mempools(md, t);
 
 	old_map = rcu_dereference_protected(md->map, lockdep_is_held(&md->suspend_lock));
-	rcu_assign_pointer(md->map, t);
+	rcu_assign_pointer(md->map, (void *)t);
 	md->immutable_target_type = dm_table_get_immutable_target_type(t);
 
 	dm_table_set_restrictions(t, q, limits);
@@ -2574,7 +2635,6 @@ void dm_set_md_type(struct mapped_device *md, unsigned type)
 
 unsigned dm_get_md_type(struct mapped_device *md)
 {
-	BUG_ON(!mutex_is_locked(&md->type_lock));
 	return md->type;
 }
 
@@ -2594,7 +2654,7 @@ struct queue_limits *dm_get_queue_limits(struct mapped_device *md)
 }
 EXPORT_SYMBOL_GPL(dm_get_queue_limits);
 
-static void init_rq_based_worker_thread(struct mapped_device *md)
+static void dm_old_init_rq_based_worker_thread(struct mapped_device *md)
 {
 	/* Initialize the request-based DM worker thread */
 	init_kthread_worker(&md->kworker);
@@ -2603,26 +2663,22 @@ static void init_rq_based_worker_thread(struct mapped_device *md)
 }
 
 /*
- * Fully initialize a request-based queue (->elevator, ->request_fn, etc).
+ * Fully initialize a .request_fn request-based queue.
  */
-static int dm_init_request_based_queue(struct mapped_device *md)
+static int dm_old_init_request_queue(struct mapped_device *md)
 {
-	struct request_queue *q = NULL;
-
 	/* Fully initialize the queue */
-	q = blk_init_allocated_queue(md->queue, dm_request_fn, NULL);
-	if (!q)
+	if (!blk_init_allocated_queue(md->queue, dm_request_fn, NULL))
 		return -EINVAL;
 
 	/* disable dm_request_fn's merge heuristic by default */
 	md->seq_rq_merge_deadline_usecs = 0;
 
-	md->queue = q;
-	dm_init_old_md_queue(md);
+	dm_init_normal_md_queue(md);
 	blk_queue_softirq_done(md->queue, dm_softirq_done);
-	blk_queue_prep_rq(md->queue, dm_prep_fn);
+	blk_queue_prep_rq(md->queue, dm_old_prep_fn);
 
-	init_rq_based_worker_thread(md);
+	dm_old_init_rq_based_worker_thread(md);
 
 	elv_register_queue(md->queue);
 
@@ -2642,6 +2698,11 @@ static int dm_mq_init_request(void *data, struct request *rq,
 	 */
 	tio->md = md;
 
+	if (md->init_tio_pdu) {
+		/* target-specific per-io data is immediately after the tio */
+		tio->info.ptr = tio + 1;
+	}
+
 	return 0;
 }
 
@@ -2651,28 +2712,15 @@ static int dm_mq_queue_rq(struct blk_mq_hw_ctx *hctx,
 	struct request *rq = bd->rq;
 	struct dm_rq_target_io *tio = blk_mq_rq_to_pdu(rq);
 	struct mapped_device *md = tio->md;
-	int srcu_idx;
-	struct dm_table *map = dm_get_live_table(md, &srcu_idx);
-	struct dm_target *ti;
-	sector_t pos;
+	struct dm_target *ti = md->immutable_target;
 
-	/* always use block 0 to find the target for flushes for now */
-	pos = 0;
-	if (!(rq->cmd_flags & REQ_FLUSH))
-		pos = blk_rq_pos(rq);
+	if (unlikely(!ti)) {
+		int srcu_idx;
+		struct dm_table *map = dm_get_live_table(md, &srcu_idx);
 
-	ti = dm_table_find_target(map, pos);
-	if (!dm_target_is_valid(ti)) {
+		ti = dm_table_find_target(map, 0);
 		dm_put_live_table(md, srcu_idx);
-		DMERR_LIMIT("request attempted access beyond the end of device");
-		/*
-		 * Must perform setup, that rq_completed() requires,
-		 * before returning BLK_MQ_RQ_QUEUE_ERROR
-		 */
-		dm_start_request(md, rq);
-		return BLK_MQ_RQ_QUEUE_ERROR;
 	}
-	dm_put_live_table(md, srcu_idx);
 
 	if (ti->type->busy && ti->type->busy(ti))
 		return BLK_MQ_RQ_QUEUE_BUSY;
@@ -2688,20 +2736,12 @@ static int dm_mq_queue_rq(struct blk_mq_hw_ctx *hctx,
 	 */
 	tio->ti = ti;
 
-	/* Clone the request if underlying devices aren't blk-mq */
-	if (dm_table_get_type(map) == DM_TYPE_REQUEST_BASED) {
-		/* clone request is allocated at the end of the pdu */
-		tio->clone = (void *)blk_mq_rq_to_pdu(rq) + sizeof(struct dm_rq_target_io);
-		(void) clone_rq(rq, md, tio, GFP_ATOMIC);
-		queue_kthread_work(&md->kworker, &tio->work);
-	} else {
-		/* Direct call is fine since .queue_rq allows allocations */
-		if (map_request(tio, rq, md) == DM_MAPIO_REQUEUE) {
-			/* Undo dm_start_request() before requeuing */
-			rq_end_stats(md, rq);
-			rq_completed(md, rq_data_dir(rq), false);
-			return BLK_MQ_RQ_QUEUE_BUSY;
-		}
+	/* Direct call is fine since .queue_rq allows allocations */
+	if (map_request(tio, rq, md) == DM_MAPIO_REQUEUE) {
+		/* Undo dm_start_request() before requeuing */
+		rq_end_stats(md, rq);
+		rq_completed(md, rq_data_dir(rq), false);
+		return BLK_MQ_RQ_QUEUE_BUSY;
 	}
 
 	return BLK_MQ_RQ_QUEUE_OK;
@@ -2714,47 +2754,56 @@ static struct blk_mq_ops dm_mq_ops = {
 	.init_request = dm_mq_init_request,
 };
 
-static int dm_init_request_based_blk_mq_queue(struct mapped_device *md)
+static int dm_mq_init_request_queue(struct mapped_device *md,
+				    struct dm_target *immutable_tgt)
 {
-	unsigned md_type = dm_get_md_type(md);
 	struct request_queue *q;
 	int err;
 
-	memset(&md->tag_set, 0, sizeof(md->tag_set));
-	md->tag_set.ops = &dm_mq_ops;
-	md->tag_set.queue_depth = BLKDEV_MAX_RQ;
-	md->tag_set.numa_node = NUMA_NO_NODE;
-	md->tag_set.flags = BLK_MQ_F_SHOULD_MERGE | BLK_MQ_F_SG_MERGE;
-	md->tag_set.nr_hw_queues = 1;
-	if (md_type == DM_TYPE_REQUEST_BASED) {
-		/* make the memory for non-blk-mq clone part of the pdu */
-		md->tag_set.cmd_size = sizeof(struct dm_rq_target_io) + sizeof(struct request);
-	} else
-		md->tag_set.cmd_size = sizeof(struct dm_rq_target_io);
-	md->tag_set.driver_data = md;
-
-	err = blk_mq_alloc_tag_set(&md->tag_set);
+	if (dm_get_md_type(md) == DM_TYPE_REQUEST_BASED) {
+		DMERR("request-based dm-mq may only be stacked on blk-mq device(s)");
+		return -EINVAL;
+	}
+
+	md->tag_set = kzalloc_node(sizeof(struct blk_mq_tag_set), GFP_KERNEL, md->numa_node_id);
+	if (!md->tag_set)
+		return -ENOMEM;
+
+	md->tag_set->ops = &dm_mq_ops;
+	md->tag_set->queue_depth = dm_get_blk_mq_queue_depth();
+	md->tag_set->numa_node = md->numa_node_id;
+	md->tag_set->flags = BLK_MQ_F_SHOULD_MERGE | BLK_MQ_F_SG_MERGE;
+	md->tag_set->nr_hw_queues = dm_get_blk_mq_nr_hw_queues();
+	md->tag_set->driver_data = md;
+
+	md->tag_set->cmd_size = sizeof(struct dm_rq_target_io);
+	if (immutable_tgt && immutable_tgt->per_io_data_size) {
+		/* any target-specific per-io data is immediately after the tio */
+		md->tag_set->cmd_size += immutable_tgt->per_io_data_size;
+		md->init_tio_pdu = true;
+	}
+
+	err = blk_mq_alloc_tag_set(md->tag_set);
 	if (err)
-		return err;
+		goto out_kfree_tag_set;
 
-	q = blk_mq_init_allocated_queue(&md->tag_set, md->queue);
+	q = blk_mq_init_allocated_queue(md->tag_set, md->queue);
 	if (IS_ERR(q)) {
 		err = PTR_ERR(q);
 		goto out_tag_set;
 	}
-	md->queue = q;
 	dm_init_md_queue(md);
 
 	/* backfill 'mq' sysfs registration normally done in blk_register_queue */
 	blk_mq_register_disk(md->disk);
 
-	if (md_type == DM_TYPE_REQUEST_BASED)
-		init_rq_based_worker_thread(md);
-
 	return 0;
 
 out_tag_set:
-	blk_mq_free_tag_set(&md->tag_set);
+	blk_mq_free_tag_set(md->tag_set);
+out_kfree_tag_set:
+	kfree(md->tag_set);
+
 	return err;
 }
 
@@ -2769,28 +2818,28 @@ static unsigned filter_md_type(unsigned type, struct mapped_device *md)
 /*
  * Setup the DM device's queue based on md's type
  */
-int dm_setup_md_queue(struct mapped_device *md)
+int dm_setup_md_queue(struct mapped_device *md, struct dm_table *t)
 {
 	int r;
 	unsigned md_type = filter_md_type(dm_get_md_type(md), md);
 
 	switch (md_type) {
 	case DM_TYPE_REQUEST_BASED:
-		r = dm_init_request_based_queue(md);
+		r = dm_old_init_request_queue(md);
 		if (r) {
-			DMWARN("Cannot initialize queue for request-based mapped device");
+			DMERR("Cannot initialize queue for request-based mapped device");
 			return r;
 		}
 		break;
 	case DM_TYPE_MQ_REQUEST_BASED:
-		r = dm_init_request_based_blk_mq_queue(md);
+		r = dm_mq_init_request_queue(md, dm_table_get_immutable_target(t));
 		if (r) {
-			DMWARN("Cannot initialize queue for request-based blk-mq mapped device");
+			DMERR("Cannot initialize queue for request-based dm-mq mapped device");
 			return r;
 		}
 		break;
 	case DM_TYPE_BIO_BASED:
-		dm_init_old_md_queue(md);
+		dm_init_normal_md_queue(md);
 		blk_queue_make_request(md->queue, dm_make_request);
 		/*
 		 * DM handles splitting bios as needed.  Free the bio_split bioset
@@ -3133,7 +3182,7 @@ static int __dm_suspend(struct mapped_device *md, struct dm_table *map,
 	 * dm defers requests to md->wq from md->queue.
 	 */
 	if (dm_request_based(md)) {
-		stop_queue(md->queue);
+		dm_stop_queue(md->queue);
 		if (md->kworker_task)
 			flush_kthread_worker(&md->kworker);
 	}
@@ -3157,7 +3206,7 @@ static int __dm_suspend(struct mapped_device *md, struct dm_table *map,
 		dm_queue_flush(md);
 
 		if (dm_request_based(md))
-			start_queue(md->queue);
+			dm_start_queue(md->queue);
 
 		unlock_fs(md);
 		dm_table_presuspend_undo_targets(map);
@@ -3236,7 +3285,7 @@ static int __dm_resume(struct mapped_device *md, struct dm_table *map)
 	 * Request-based dm is queueing the deferred I/Os in its request_queue.
 	 */
 	if (dm_request_based(md))
-		start_queue(md->queue);
+		dm_start_queue(md->queue);
 
 	unlock_fs(md);
 
@@ -3482,9 +3531,9 @@ int dm_noflush_suspending(struct dm_target *ti)
 EXPORT_SYMBOL_GPL(dm_noflush_suspending);
 
 struct dm_md_mempools *dm_alloc_md_mempools(struct mapped_device *md, unsigned type,
-					    unsigned integrity, unsigned per_bio_data_size)
+					    unsigned integrity, unsigned per_io_data_size)
 {
-	struct dm_md_mempools *pools = kzalloc(sizeof(*pools), GFP_KERNEL);
+	struct dm_md_mempools *pools = kzalloc_node(sizeof(*pools), GFP_KERNEL, md->numa_node_id);
 	struct kmem_cache *cachep = NULL;
 	unsigned int pool_size = 0;
 	unsigned int front_pad;
@@ -3498,7 +3547,7 @@ struct dm_md_mempools *dm_alloc_md_mempools(struct mapped_device *md, unsigned t
 	case DM_TYPE_BIO_BASED:
 		cachep = _io_cache;
 		pool_size = dm_get_reserved_bio_based_ios();
-		front_pad = roundup(per_bio_data_size, __alignof__(struct dm_target_io)) + offsetof(struct dm_target_io, clone);
+		front_pad = roundup(per_io_data_size, __alignof__(struct dm_target_io)) + offsetof(struct dm_target_io, clone);
 		break;
 	case DM_TYPE_REQUEST_BASED:
 		cachep = _rq_tio_cache;
@@ -3511,8 +3560,7 @@ struct dm_md_mempools *dm_alloc_md_mempools(struct mapped_device *md, unsigned t
 		if (!pool_size)
 			pool_size = dm_get_reserved_rq_based_ios();
 		front_pad = offsetof(struct dm_rq_clone_bio_info, clone);
-		/* per_bio_data_size is not used. See __bind_mempools(). */
-		WARN_ON(per_bio_data_size != 0);
+		/* per_io_data_size is used for blk-mq pdu at queue allocation */
 		break;
 	default:
 		BUG();
@@ -3554,15 +3602,14 @@ void dm_free_md_mempools(struct dm_md_mempools *pools)
 }
 
 static int dm_pr_register(struct block_device *bdev, u64 old_key, u64 new_key,
-		u32 flags)
+			  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;
+	int r;
 
-	r = dm_get_live_table_for_ioctl(md, &tgt, &bdev, &mode, &srcu_idx);
+	r = dm_grab_bdev_for_ioctl(md, &bdev, &mode);
 	if (r < 0)
 		return r;
 
@@ -3572,20 +3619,19 @@ static int dm_pr_register(struct block_device *bdev, u64 old_key, u64 new_key,
 	else
 		r = -EOPNOTSUPP;
 
-	dm_put_live_table(md, srcu_idx);
+	bdput(bdev);
 	return r;
 }
 
 static int dm_pr_reserve(struct block_device *bdev, u64 key, enum pr_type type,
-		u32 flags)
+			 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;
+	int r;
 
-	r = dm_get_live_table_for_ioctl(md, &tgt, &bdev, &mode, &srcu_idx);
+	r = dm_grab_bdev_for_ioctl(md, &bdev, &mode);
 	if (r < 0)
 		return r;
 
@@ -3595,7 +3641,7 @@ static int dm_pr_reserve(struct block_device *bdev, u64 key, enum pr_type type,
 	else
 		r = -EOPNOTSUPP;
 
-	dm_put_live_table(md, srcu_idx);
+	bdput(bdev);
 	return r;
 }
 
@@ -3603,11 +3649,10 @@ 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;
+	int r;
 
-	r = dm_get_live_table_for_ioctl(md, &tgt, &bdev, &mode, &srcu_idx);
+	r = dm_grab_bdev_for_ioctl(md, &bdev, &mode);
 	if (r < 0)
 		return r;
 
@@ -3617,20 +3662,19 @@ static int dm_pr_release(struct block_device *bdev, u64 key, enum pr_type type)
 	else
 		r = -EOPNOTSUPP;
 
-	dm_put_live_table(md, srcu_idx);
+	bdput(bdev);
 	return r;
 }
 
 static int dm_pr_preempt(struct block_device *bdev, u64 old_key, u64 new_key,
-		enum pr_type type, bool abort)
+			 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;
+	int r;
 
-	r = dm_get_live_table_for_ioctl(md, &tgt, &bdev, &mode, &srcu_idx);
+	r = dm_grab_bdev_for_ioctl(md, &bdev, &mode);
 	if (r < 0)
 		return r;
 
@@ -3640,7 +3684,7 @@ static int dm_pr_preempt(struct block_device *bdev, u64 old_key, u64 new_key,
 	else
 		r = -EOPNOTSUPP;
 
-	dm_put_live_table(md, srcu_idx);
+	bdput(bdev);
 	return r;
 }
 
@@ -3648,11 +3692,10 @@ 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;
+	int r;
 
-	r = dm_get_live_table_for_ioctl(md, &tgt, &bdev, &mode, &srcu_idx);
+	r = dm_grab_bdev_for_ioctl(md, &bdev, &mode);
 	if (r < 0)
 		return r;
 
@@ -3662,7 +3705,7 @@ static int dm_pr_clear(struct block_device *bdev, u64 key)
 	else
 		r = -EOPNOTSUPP;
 
-	dm_put_live_table(md, srcu_idx);
+	bdput(bdev);
 	return r;
 }
 
@@ -3701,6 +3744,15 @@ MODULE_PARM_DESC(reserved_rq_based_ios, "Reserved IOs in request-based mempools"
 module_param(use_blk_mq, bool, S_IRUGO | S_IWUSR);
 MODULE_PARM_DESC(use_blk_mq, "Use block multiqueue for request-based DM devices");
 
+module_param(dm_mq_nr_hw_queues, uint, S_IRUGO | S_IWUSR);
+MODULE_PARM_DESC(dm_mq_nr_hw_queues, "Number of hardware queues for request-based dm-mq devices");
+
+module_param(dm_mq_queue_depth, uint, S_IRUGO | S_IWUSR);
+MODULE_PARM_DESC(dm_mq_queue_depth, "Queue depth for request-based dm-mq devices");
+
+module_param(dm_numa_node, int, S_IRUGO | S_IWUSR);
+MODULE_PARM_DESC(dm_numa_node, "NUMA node for DM device memory allocations");
+
 MODULE_DESCRIPTION(DM_NAME " driver");
 MODULE_AUTHOR("Joe Thornber <dm-devel@redhat.com>");
 MODULE_LICENSE("GPL");

drivers/md/dm.h

@@ -73,6 +73,8 @@ int dm_table_resume_targets(struct dm_table *t);
 int dm_table_any_congested(struct dm_table *t, int bdi_bits);
 unsigned dm_table_get_type(struct dm_table *t);
 struct target_type *dm_table_get_immutable_target_type(struct dm_table *t);
+struct dm_target *dm_table_get_immutable_target(struct dm_table *t);
+struct dm_target *dm_table_get_wildcard_target(struct dm_table *t);
 bool dm_table_request_based(struct dm_table *t);
 bool dm_table_mq_request_based(struct dm_table *t);
 void dm_table_free_md_mempools(struct dm_table *t);
@@ -84,7 +86,7 @@ void dm_set_md_type(struct mapped_device *md, unsigned type);
 unsigned dm_get_md_type(struct mapped_device *md);
 struct target_type *dm_get_immutable_target_type(struct mapped_device *md);
 
-int dm_setup_md_queue(struct mapped_device *md);
+int dm_setup_md_queue(struct mapped_device *md, struct dm_table *t);
 
 /*
  * To check the return value from dm_table_find_target().

include/linux/device-mapper.h

@@ -124,6 +124,8 @@ struct dm_dev {
 	char name[16];
 };
 
+dev_t dm_get_dev_t(const char *path);
+
 /*
  * Constructors should call these functions to ensure destination devices
  * are opened/closed correctly.
@@ -189,6 +191,13 @@ struct target_type {
 #define DM_TARGET_IMMUTABLE		0x00000004
 #define dm_target_is_immutable(type)	((type)->features & DM_TARGET_IMMUTABLE)
 
+/*
+ * Indicates that a target may replace any target; even immutable targets.
+ * .map, .map_rq, .clone_and_map_rq and .release_clone_rq are all defined.
+ */
+#define DM_TARGET_WILDCARD		0x00000008
+#define dm_target_is_wildcard(type)	((type)->features & DM_TARGET_WILDCARD)
+
 /*
  * Some targets need to be sent the same WRITE bio severals times so
  * that they can send copies of it to different devices.  This function
@@ -231,10 +240,10 @@ struct dm_target {
 	unsigned num_write_same_bios;
 
 	/*
-	 * The minimum number of extra bytes allocated in each bio for the
-	 * target to use.  dm_per_bio_data returns the data location.
+	 * The minimum number of extra bytes allocated in each io for the
+	 * target to use.
 	 */
-	unsigned per_bio_data_size;
+	unsigned per_io_data_size;
 
 	/*
 	 * If defined, this function is called to find out how many