Merge tag 'for-5.18/dm-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm

Pull device mapper updates from Mike Snitzer:

- Significant refactoring and fixing of how DM core does bio-based IO
accounting with focus on fixing wildly inaccurate IO stats for
dm-crypt (and other DM targets that defer bio submission in their own
workqueues). End result is proper IO accounting, made possible by
targets being updated to use the new dm_submit_bio_remap() interface.

- Add hipri bio polling support (REQ_POLLED) to bio-based DM.

- Reduce dm_io and dm_target_io structs so that a single dm_io (which
contains dm_target_io and first clone bio) weighs in at 256 bytes.
For reference the bio struct is 128 bytes.

- Various other small cleanups, fixes or improvements in DM core and
targets.

- Update MAINTAINERS with my kernel.org email address to allow
distinction between my "upstream" and "Red" Hats.

* tag 'for-5.18/dm-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm: (46 commits)
dm: consolidate spinlocks in dm_io struct
dm: reduce size of dm_io and dm_target_io structs
dm: switch dm_target_io booleans over to proper flags
dm: switch dm_io booleans over to proper flags
dm: update email address in MAINTAINERS
dm: return void from __send_empty_flush
dm: factor out dm_io_complete
dm cache: use dm_submit_bio_remap
dm: simplify dm_sumbit_bio_remap interface
dm thin: use dm_submit_bio_remap
dm: add WARN_ON_ONCE to dm_submit_bio_remap
dm: support bio polling
block: add ->poll_bio to block_device_operations
dm mpath: use DMINFO instead of printk with KERN_INFO
dm: stop using bdevname
dm-zoned: remove the ->name field in struct dmz_dev
dm: remove unnecessary local variables in __bind
dm: requeue IO if mapping table not yet available
dm io: remove stale comment block for dm_io()
dm thin metadata: remove unused dm_thin_remove_block and __remove
...

MAINTAINERS

···
 
 DEVICE-MAPPER  (LVM)
 M:	Alasdair Kergon <[e-mail]>
-M:	Mike Snitzer <[e-mail]>
+M:	Mike Snitzer <[e-mail]>
 M:	[e-mail]
 L:	[e-mail]
 S:	Maintained

block/blk-core.c

···
  *
  * bio_list_on_stack[0] contains bios submitted by the current ->submit_bio.
  * bio_list_on_stack[1] contains bios that were submitted before the current
- *	->submit_bio_bio, but that haven't been processed yet.
+ *	->submit_bio, but that haven't been processed yet.
  */
 static void __submit_bio_noacct(struct bio *bio)
 {
···
 {
 	struct request_queue *q = bdev_get_queue(bio->bi_bdev);
 	blk_qc_t cookie = READ_ONCE(bio->bi_cookie);
-	int ret;
+	int ret = 0;
 
 	if (cookie == BLK_QC_T_NONE ||
 	    !test_bit(QUEUE_FLAG_POLL, &q->queue_flags))
···
 
 	if (blk_queue_enter(q, BLK_MQ_REQ_NOWAIT))
 		return 0;
-	if (WARN_ON_ONCE(!queue_is_mq(q)))
-		ret = 0;	/* not yet implemented, should not happen */
-	else
+	if (queue_is_mq(q)) {
 		ret = blk_mq_poll(q, cookie, iob, flags);
+	} else {
+		struct gendisk *disk = q->disk;
+
+		if (disk && disk->fops->poll_bio)
+			ret = disk->fops->poll_bio(bio, iob, flags);
+	}
 	blk_queue_exit(q);
 	return ret;
 }

block/genhd.c

···
 	struct device *ddev = disk_to_dev(disk);
 	int ret;
 
+	/* Only makes sense for bio-based to set ->poll_bio */
+	if (queue_is_mq(disk->queue) && disk->fops->poll_bio)
+		return -EINVAL;
+
 	/*
 	 * The disk queue should now be all set with enough information about
 	 * the device for the elevator code to pick an adequate default

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

···
 	 * This scheme reminds me of a graph of entropy vs probability of a
 	 * binary variable.
 	 */
-	static unsigned table[] = {1, 1, 1, 2, 4, 6, 7, 8, 7, 6, 4, 4, 3, 3, 2, 2, 1};
+	static const unsigned int table[] = {
+		1, 1, 1, 2, 4, 6, 7, 8, 7, 6, 4, 4, 3, 3, 2, 2, 1
+	};
 
 	unsigned hits = mq->cache_stats.hits;
 	unsigned misses = mq->cache_stats.misses;

drivers/md/dm-cache-target.c

···
 static void accounted_request(struct cache *cache, struct bio *bio)
 {
 	accounted_begin(cache, bio);
-	submit_bio_noacct(bio);
+	dm_submit_bio_remap(bio, NULL);
 }
 
 static void issue_op(struct bio *bio, void *context)
···
 	bool commit_needed;
 
 	if (map_bio(cache, bio, get_bio_block(cache, bio), &commit_needed) == DM_MAPIO_REMAPPED)
-		submit_bio_noacct(bio);
+		dm_submit_bio_remap(bio, NULL);
 
 	return commit_needed;
 }
···
 
 	if (cache->features.discard_passdown) {
 		remap_to_origin(cache, bio);
-		submit_bio_noacct(bio);
+		dm_submit_bio_remap(bio, NULL);
 	} else
 		bio_endio(bio);
 
···
 {
 	int r;
 	sector_t metadata_dev_size;
-	char b[BDEVNAME_SIZE];
 
 	if (!at_least_one_arg(as, error))
 		return -EINVAL;
···
 
 	metadata_dev_size = get_dev_size(ca->metadata_dev);
 	if (metadata_dev_size > DM_CACHE_METADATA_MAX_SECTORS_WARNING)
-		DMWARN("Metadata device %s is larger than %u sectors: excess space will not be used.",
-		       bdevname(ca->metadata_dev->bdev, b), THIN_METADATA_MAX_SECTORS);
+		DMWARN("Metadata device %pg is larger than %u sectors: excess space will not be used.",
+		       ca->metadata_dev->bdev, THIN_METADATA_MAX_SECTORS);
 
 	return 0;
 }
···
 
 	cache->ti = ca->ti;
 	ti->private = cache;
+	ti->accounts_remapped_io = true;
 	ti->num_flush_bios = 2;
 	ti->flush_supported = true;
 
···
 	struct block_device *origin_bdev = cache->origin_dev->bdev;
 	struct queue_limits *origin_limits = &bdev_get_queue(origin_bdev)->limits;
 	const char *reason = NULL;
-	char buf[BDEVNAME_SIZE];
 
 	if (!cache->features.discard_passdown)
 		return;
···
 		reason = "max discard sectors smaller than a block";
 
 	if (reason) {
-		DMWARN("Origin device (%s) %s: Disabling discard passdown.",
-		       bdevname(origin_bdev, buf), reason);
+		DMWARN("Origin device (%pg) %s: Disabling discard passdown.",
+		       origin_bdev, reason);
 		cache->features.discard_passdown = false;
 	}
 }

drivers/md/dm-clone-target.c

···
 {
 	int r;
 	sector_t metadata_dev_size;
-	char b[BDEVNAME_SIZE];
 
 	r = dm_get_device(clone->ti, dm_shift_arg(as), FMODE_READ | FMODE_WRITE,
 			  &clone->metadata_dev);
···
 
 	metadata_dev_size = get_dev_size(clone->metadata_dev);
 	if (metadata_dev_size > DM_CLONE_METADATA_MAX_SECTORS_WARNING)
-		DMWARN("Metadata device %s is larger than %u sectors: excess space will not be used.",
-		       bdevname(clone->metadata_dev->bdev, b), DM_CLONE_METADATA_MAX_SECTORS);
+		DMWARN("Metadata device %pg is larger than %u sectors: excess space will not be used.",
+		       clone->metadata_dev->bdev, DM_CLONE_METADATA_MAX_SECTORS);
 
 	return 0;
 }
···
 	struct block_device *dest_dev = clone->dest_dev->bdev;
 	struct queue_limits *dest_limits = &bdev_get_queue(dest_dev)->limits;
 	const char *reason = NULL;
-	char buf[BDEVNAME_SIZE];
 
 	if (!test_bit(DM_CLONE_DISCARD_PASSDOWN, &clone->flags))
 		return;
···
 		reason = "max discard sectors smaller than a region";
 
 	if (reason) {
-		DMWARN("Destination device (%s) %s: Disabling discard passdown.",
-		       bdevname(dest_dev, buf), reason);
+		DMWARN("Destination device (%pd) %s: Disabling discard passdown.",
+		       dest_dev, reason);
 		clear_bit(DM_CLONE_DISCARD_PASSDOWN, &clone->flags);
 	}
 }

drivers/md/dm-core.h

···
 	struct gendisk *disk;
 	struct dax_device *dax_dev;
 
+	wait_queue_head_t wait;
+	unsigned long __percpu *pending_io;
+
+	/* forced geometry settings */
+	struct hd_geometry geometry;
+
+	/*
+	 * Processing queue (flush)
+	 */
+	struct workqueue_struct *wq;
+
 	/*
 	 * A list of ios that arrived while we were suspended.
 	 */
 	struct work_struct work;
-	wait_queue_head_t wait;
 	spinlock_t deferred_lock;
 	struct bio_list deferred;
 
···
 	struct list_head uevent_list;
 	spinlock_t uevent_lock; /* Protect access to uevent_list */
 
+	/* for blk-mq request-based DM support */
+	bool init_tio_pdu:1;
+	struct blk_mq_tag_set *tag_set;
+
+	struct dm_stats stats;
+
 	/* the number of internal suspends */
 	unsigned internal_suspend_count;
+
+	int swap_bios;
+	struct semaphore swap_bios_semaphore;
+	struct mutex swap_bios_lock;
 
 	/*
 	 * io objects are allocated from here.
···
 	struct bio_set io_bs;
 	struct bio_set bs;
 
-	/*
-	 * Processing queue (flush)
-	 */
-	struct workqueue_struct *wq;
-
-	/* forced geometry settings */
-	struct hd_geometry geometry;
-
 	/* kobject and completion */
 	struct dm_kobject_holder kobj_holder;
-
-	int swap_bios;
-	struct semaphore swap_bios_semaphore;
-	struct mutex swap_bios_lock;
-
-	struct dm_stats stats;
-
-	/* for blk-mq request-based DM support */
-	struct blk_mq_tag_set *tag_set;
-	bool init_tio_pdu:1;
 
 	struct srcu_struct io_barrier;
 
···
 /*
  * One of these is allocated per clone bio.
  */
-#define DM_TIO_MAGIC 7282014
+#define DM_TIO_MAGIC 28714
 struct dm_target_io {
-	unsigned int magic;
+	unsigned short magic;
+	unsigned short flags;
+	unsigned int target_bio_nr;
 	struct dm_io *io;
 	struct dm_target *ti;
-	unsigned int target_bio_nr;
 	unsigned int *len_ptr;
-	bool inside_dm_io;
+	sector_t old_sector;
 	struct bio clone;
 };
+
+/*
+ * dm_target_io flags
+ */
+enum {
+	DM_TIO_INSIDE_DM_IO,
+	DM_TIO_IS_DUPLICATE_BIO
+};
+
+static inline bool dm_tio_flagged(struct dm_target_io *tio, unsigned int bit)
+{
+	return (tio->flags & (1U << bit)) != 0;
+}
+
+static inline void dm_tio_set_flag(struct dm_target_io *tio, unsigned int bit)
+{
+	tio->flags |= (1U << bit);
+}
 
 /*
  * One of these is allocated per original bio.
  * It contains the first clone used for that original.
  */
-#define DM_IO_MAGIC 5191977
+#define DM_IO_MAGIC 19577
 struct dm_io {
-	unsigned int magic;
-	struct mapped_device *md;
-	blk_status_t status;
+	unsigned short magic;
+	unsigned short flags;
 	atomic_t io_count;
+	struct mapped_device *md;
 	struct bio *orig_bio;
+	blk_status_t status;
+	spinlock_t lock;
 	unsigned long start_time;
-	spinlock_t endio_lock;
+	void *data;
+	struct hlist_node node;
+	struct task_struct *map_task;
 	struct dm_stats_aux stats_aux;
 	/* last member of dm_target_io is 'struct bio' */
 	struct dm_target_io tio;
 };
+
+/*
+ * dm_io flags
+ */
+enum {
+	DM_IO_START_ACCT,
+	DM_IO_ACCOUNTED
+};
+
+static inline bool dm_io_flagged(struct dm_io *io, unsigned int bit)
+{
+	return (io->flags & (1U << bit)) != 0;
+}
+
+static inline void dm_io_set_flag(struct dm_io *io, unsigned int bit)
+{
+	io->flags |= (1U << bit);
+}
 
 static inline void dm_io_inc_pending(struct dm_io *io)
 {

drivers/md/dm-crypt.c

···
 	crypt_dec_pending(io);
 }
 
+#define CRYPT_MAP_READ_GFP GFP_NOWAIT
+
 static int kcryptd_io_read(struct dm_crypt_io *io, gfp_t gfp)
 {
 	struct crypt_config *cc = io->cc;
···
 		return 1;
 	}
 
-	submit_bio_noacct(clone);
+	dm_submit_bio_remap(io->base_bio, clone);
 	return 0;
 }
 
···
 {
 	struct bio *clone = io->ctx.bio_out;
 
-	submit_bio_noacct(clone);
+	dm_submit_bio_remap(io->base_bio, clone);
 }
 
 #define crypt_io_from_node(node) rb_entry((node), struct dm_crypt_io, rb_node)
···
 
 	if ((likely(!async) && test_bit(DM_CRYPT_NO_OFFLOAD, &cc->flags)) ||
 	    test_bit(DM_CRYPT_NO_WRITE_WORKQUEUE, &cc->flags)) {
-		submit_bio_noacct(clone);
+		dm_submit_bio_remap(io->base_bio, clone);
 		return;
 	}
 
···
 
 static int get_key_size(char **key_string)
 {
-	return (*key_string[0] == ':') ? -EINVAL : strlen(*key_string) >> 1;
+	return (*key_string[0] == ':') ? -EINVAL : (int)(strlen(*key_string) >> 1);
 }
 
 #endif /* CONFIG_KEYS */
···
 
 	ti->num_flush_bios = 1;
 	ti->limit_swap_bios = true;
+	ti->accounts_remapped_io = true;
 
 	dm_audit_log_ctr(DM_MSG_PREFIX, ti, 1);
 	return 0;
···
 		io->ctx.r.req = (struct skcipher_request *)(io + 1);
 
 	if (bio_data_dir(io->base_bio) == READ) {
-		if (kcryptd_io_read(io, GFP_NOWAIT))
+		if (kcryptd_io_read(io, CRYPT_MAP_READ_GFP))
 			kcryptd_queue_read(io);
 	} else
 		kcryptd_queue_crypt(io);
···
 
 static struct target_type crypt_target = {
 	.name   = "crypt",
-	.version = {1, 23, 0},
+	.version = {1, 24, 0},
 	.module = THIS_MODULE,
 	.ctr    = crypt_ctr,
 	.dtr    = crypt_dtr,

drivers/md/dm-delay.c

···
 	while (bio) {
 		n = bio->bi_next;
 		bio->bi_next = NULL;
-		submit_bio_noacct(bio);
+		dm_submit_bio_remap(bio, NULL);
 		bio = n;
 	}
 }
···
 
 	ti->num_flush_bios = 1;
 	ti->num_discard_bios = 1;
+	ti->accounts_remapped_io = true;
 	ti->per_io_data_size = sizeof(struct dm_delay_info);
 	return 0;
 
···
 
 static struct target_type delay_target = {
 	.name	     = "delay",
-	.version     = {1, 2, 1},
+	.version     = {1, 3, 0},
 	.features    = DM_TARGET_PASSES_INTEGRITY,
 	.module      = THIS_MODULE,
 	.ctr	     = delay_ctr,

drivers/md/dm-ima.c

···
 		scnprintf(device_table_data, DM_IMA_DEVICE_BUF_LEN,
 			  "%sname=%s,uuid=%s;device_resume=no_data;",
 			  DM_IMA_VERSION_STR, dev_name, dev_uuid);
-		l += strlen(device_table_data);
+		l = strlen(device_table_data);
 
 	}
 
···
 		scnprintf(device_table_data, DM_IMA_DEVICE_BUF_LEN,
 			  "%sname=%s,uuid=%s;device_remove=no_data;",
 			  DM_IMA_VERSION_STR, dev_name, dev_uuid);
-		l += strlen(device_table_data);
+		l = strlen(device_table_data);
 	}
 
 	memcpy(device_table_data + l, remove_all_str, remove_all_len);
···
 		scnprintf(device_table_data, DM_IMA_DEVICE_BUF_LEN,
 			  "%sname=%s,uuid=%s;table_clear=no_data;",
 			   DM_IMA_VERSION_STR, dev_name, dev_uuid);
-		l += strlen(device_table_data);
+		l = strlen(device_table_data);
 	}
 
 	capacity_len = strlen(capacity_str);

drivers/md/dm-io.c

···
 	return 0;
 }
 
-/*
- * New collapsed (a)synchronous interface.
- *
- * If the IO is asynchronous (i.e. it has notify.fn), you must either unplug
- * the queue with blk_unplug() some time later or set REQ_SYNC in
- * io_req->bi_opf. If you fail to do one of these, the IO will be submitted to
- * the disk after q->unplug_delay, which defaults to 3ms in blk-settings.c.
- */
 int dm_io(struct dm_io_request *io_req, unsigned num_regions,
 	  struct dm_io_region *where, unsigned long *sync_error_bits)
 {

drivers/md/dm-ioctl.c

···
 #include <linux/dm-ioctl.h>
 #include <linux/hdreg.h>
 #include <linux/compat.h>
+#include <linux/nospec.h>
 
 #include <linux/uaccess.h>
 #include <linux/ima.h>
···
 	if (unlikely(cmd >= ARRAY_SIZE(_ioctls)))
 		return NULL;
 
+	cmd = array_index_nospec(cmd, ARRAY_SIZE(_ioctls));
 	*ioctl_flags = _ioctls[cmd].flags;
 	return _ioctls[cmd].fn;
 }

drivers/md/dm-mpath.c

···
 	if (m->hw_handler_name) {
 		r = scsi_dh_attach(q, m->hw_handler_name);
 		if (r == -EBUSY) {
-			char b[BDEVNAME_SIZE];
-
-			printk(KERN_INFO "dm-mpath: retaining handler on device %s\n",
-			       bdevname(bdev, b));
+			DMINFO("retaining handler on device %pg", bdev);
 			goto retain;
 		}
 		if (r < 0) {

drivers/md/dm-rq.c

···
 
 	if (unlikely(!ti)) {
 		int srcu_idx;
-		struct dm_table *map = dm_get_live_table(md, &srcu_idx);
+		struct dm_table *map;
 
+		map = dm_get_live_table(md, &srcu_idx);
+		if (unlikely(!map)) {
+			dm_put_live_table(md, srcu_idx);
+			return BLK_STS_RESOURCE;
+		}
 		ti = dm_table_find_target(map, 0);
 		dm_put_live_table(md, srcu_idx);
 	}

drivers/md/dm-stats.c

···
 
 	mutex_init(&stats->mutex);
 	INIT_LIST_HEAD(&stats->list);
+	stats->precise_timestamps = false;
 	stats->last = alloc_percpu(struct dm_stats_last_position);
 	for_each_possible_cpu(cpu) {
 		last = per_cpu_ptr(stats->last, cpu);
···
 	}
 	free_percpu(stats->last);
 	mutex_destroy(&stats->mutex);
+}
+
+static void dm_stats_recalc_precise_timestamps(struct dm_stats *stats)
+{
+	struct list_head *l;
+	struct dm_stat *tmp_s;
+	bool precise_timestamps = false;
+
+	list_for_each(l, &stats->list) {
+		tmp_s = container_of(l, struct dm_stat, list_entry);
+		if (tmp_s->stat_flags & STAT_PRECISE_TIMESTAMPS) {
+			precise_timestamps = true;
+			break;
+		}
+	}
+	stats->precise_timestamps = precise_timestamps;
 }
 
 static int dm_stats_create(struct dm_stats *stats, sector_t start, sector_t end,
···
 	}
 	ret_id = s->id;
 	list_add_tail_rcu(&s->list_entry, l);
+
+	dm_stats_recalc_precise_timestamps(stats);
+
 	mutex_unlock(&stats->mutex);
 
 	resume_callback(md);
···
 	}
 
 	list_del_rcu(&s->list_entry);
+
+	dm_stats_recalc_precise_timestamps(stats);
+
 	mutex_unlock(&stats->mutex);
 
 	/*
···
 
 void dm_stats_account_io(struct dm_stats *stats, unsigned long bi_rw,
 			 sector_t bi_sector, unsigned bi_sectors, bool end,
-			 unsigned long duration_jiffies,
+			 unsigned long start_time,
 			 struct dm_stats_aux *stats_aux)
 {
 	struct dm_stat *s;
 	sector_t end_sector;
 	struct dm_stats_last_position *last;
 	bool got_precise_time;
+	unsigned long duration_jiffies = 0;
 
 	if (unlikely(!bi_sectors))
 		return;
···
 				       ));
 		WRITE_ONCE(last->last_sector, end_sector);
 		WRITE_ONCE(last->last_rw, bi_rw);
-	}
+	} else
+		duration_jiffies = jiffies - start_time;
 
 	rcu_read_lock();
 
 	got_precise_time = false;
 	list_for_each_entry_rcu(s, &stats->list, list_entry) {
 		if (s->stat_flags & STAT_PRECISE_TIMESTAMPS && !got_precise_time) {
-			if (!end)
-				stats_aux->duration_ns = ktime_to_ns(ktime_get());
-			else
+			/* start (!end) duration_ns is set by DM core's alloc_io() */
+			if (end)
 				stats_aux->duration_ns = ktime_to_ns(ktime_get()) - stats_aux->duration_ns;
 			got_precise_time = true;
 		}

drivers/md/dm-stats.h

···
 	struct mutex mutex;
 	struct list_head list;	/* list of struct dm_stat */
 	struct dm_stats_last_position __percpu *last;
-	sector_t last_sector;
-	unsigned last_rw;
+	bool precise_timestamps;
 };
 
 struct dm_stats_aux {
···
 
 void dm_stats_account_io(struct dm_stats *stats, unsigned long bi_rw,
 			 sector_t bi_sector, unsigned bi_sectors, bool end,
-			 unsigned long duration_jiffies,
+			 unsigned long start_time,
 			 struct dm_stats_aux *aux);
 
 static inline bool dm_stats_used(struct dm_stats *st)
 {
 	return !list_empty(&st->list);
+}
+
+static inline void dm_stats_record_start(struct dm_stats *stats, struct dm_stats_aux *aux)
+{
+	if (unlikely(stats->precise_timestamps))
+		aux->duration_ns = ktime_to_ns(ktime_get());
 }
 
 #endif

drivers/md/dm-table.c

···
 	sector_t dev_size = bdev_nr_sectors(bdev);
 	unsigned short logical_block_size_sectors =
 		limits->logical_block_size >> SECTOR_SHIFT;
-	char b[BDEVNAME_SIZE];
 
 	if (!dev_size)
 		return 0;
 
 	if ((start >= dev_size) || (start + len > dev_size)) {
-		DMWARN("%s: %s too small for target: "
+		DMWARN("%s: %pg too small for target: "
 		       "start=%llu, len=%llu, dev_size=%llu",
-		       dm_device_name(ti->table->md), bdevname(bdev, b),
+		       dm_device_name(ti->table->md), bdev,
 		       (unsigned long long)start,
 		       (unsigned long long)len,
 		       (unsigned long long)dev_size);
···
 		unsigned int zone_sectors = bdev_zone_sectors(bdev);
 
 		if (start & (zone_sectors - 1)) {
-			DMWARN("%s: start=%llu not aligned to h/w zone size %u of %s",
+			DMWARN("%s: start=%llu not aligned to h/w zone size %u of %pg",
 			       dm_device_name(ti->table->md),
 			       (unsigned long long)start,
-			       zone_sectors, bdevname(bdev, b));
+			       zone_sectors, bdev);
 			return 1;
 		}
 
···
 		 * the sector range.
 		 */
 		if (len & (zone_sectors - 1)) {
-			DMWARN("%s: len=%llu not aligned to h/w zone size %u of %s",
+			DMWARN("%s: len=%llu not aligned to h/w zone size %u of %pg",
 			       dm_device_name(ti->table->md),
 			       (unsigned long long)len,
-			       zone_sectors, bdevname(bdev, b));
+			       zone_sectors, bdev);
 			return 1;
 		}
 	}
···
 
 	if (start & (logical_block_size_sectors - 1)) {
 		DMWARN("%s: start=%llu not aligned to h/w "
-		       "logical block size %u of %s",
+		       "logical block size %u of %pg",
 		       dm_device_name(ti->table->md),
 		       (unsigned long long)start,
-		       limits->logical_block_size, bdevname(bdev, b));
+		       limits->logical_block_size, bdev);
 		return 1;
 	}
 
 	if (len & (logical_block_size_sectors - 1)) {
 		DMWARN("%s: len=%llu not aligned to h/w "
-		       "logical block size %u of %s",
+		       "logical block size %u of %pg",
 		       dm_device_name(ti->table->md),
 		       (unsigned long long)len,
-		       limits->logical_block_size, bdevname(bdev, b));
+		       limits->logical_block_size, bdev);
 		return 1;
 	}
 
···
 	struct queue_limits *limits = data;
 	struct block_device *bdev = dev->bdev;
 	struct request_queue *q = bdev_get_queue(bdev);
-	char b[BDEVNAME_SIZE];
 
 	if (unlikely(!q)) {
-		DMWARN("%s: Cannot set limits for nonexistent device %s",
-		       dm_device_name(ti->table->md), bdevname(bdev, b));
+		DMWARN("%s: Cannot set limits for nonexistent device %pg",
+		       dm_device_name(ti->table->md), bdev);
 		return 0;
 	}
 
 	if (blk_stack_limits(limits, &q->limits,
 			get_start_sect(bdev) + start) < 0)
-		DMWARN("%s: adding target device %s caused an alignment inconsistency: "
+		DMWARN("%s: adding target device %pg caused an alignment inconsistency: "
 		       "physical_block_size=%u, logical_block_size=%u, "
 		       "alignment_offset=%u, start=%llu",
-		       dm_device_name(ti->table->md), bdevname(bdev, b),
+		       dm_device_name(ti->table->md), bdev,
 		       q->limits.physical_block_size,
 		       q->limits.logical_block_size,
 		       q->limits.alignment_offset,
···
 	return &t->targets[(KEYS_PER_NODE * n) + k];
 }
 
+static int device_not_poll_capable(struct dm_target *ti, struct dm_dev *dev,
+				   sector_t start, sector_t len, void *data)
+{
+	struct request_queue *q = bdev_get_queue(dev->bdev);
+
+	return !test_bit(QUEUE_FLAG_POLL, &q->queue_flags);
+}
+
 /*
  * type->iterate_devices() should be called when the sanity check needs to
  * iterate and check all underlying data devices. iterate_devices() will
···
 	(*num_devices)++;
 
 	return 0;
+}
+
+static int dm_table_supports_poll(struct dm_table *t)
+{
+	return !dm_table_any_dev_attr(t, device_not_poll_capable, NULL);
 }
 
 /*
···
 
 	dm_update_crypto_profile(q, t);
 	disk_update_readahead(t->md->disk);
+
+	/*
+	 * Check for request-based device is left to
+	 * dm_mq_init_request_queue()->blk_mq_init_allocated_queue().
+	 *
+	 * For bio-based device, only set QUEUE_FLAG_POLL when all
+	 * underlying devices supporting polling.
+	 */
+	if (__table_type_bio_based(t->type)) {
+		if (dm_table_supports_poll(t))
+			blk_queue_flag_set(QUEUE_FLAG_POLL, q);
+		else
+			blk_queue_flag_clear(QUEUE_FLAG_POLL, q);
+	}
 
 	return 0;
 }

drivers/md/dm-thin-metadata.c

···
 	return r;
 }
 
-static int __remove(struct dm_thin_device *td, dm_block_t block)
-{
-	int r;
-	struct dm_pool_metadata *pmd = td->pmd;
-	dm_block_t keys[2] = { td->id, block };
-
-	r = dm_btree_remove(&pmd->info, pmd->root, keys, &pmd->root);
-	if (r)
-		return r;
-
-	td->mapped_blocks--;
-	td->changed = true;
-
-	return 0;
-}
-
 static int __remove_range(struct dm_thin_device *td, dm_block_t begin, dm_block_t end)
 {
 	int r;
···
 	value = cpu_to_le64(mapping_root);
 	__dm_bless_for_disk(&value);
 	return dm_btree_insert(&pmd->tl_info, pmd->root, keys, &value, &pmd->root);
-}
-
-int dm_thin_remove_block(struct dm_thin_device *td, dm_block_t block)
-{
-	int r = -EINVAL;
-
-	pmd_write_lock(td->pmd);
-	if (!td->pmd->fail_io)
-		r = __remove(td, block);
-	pmd_write_unlock(td->pmd);
-
-	return r;
 }
 
 int dm_thin_remove_range(struct dm_thin_device *td,

drivers/md/dm-thin-metadata.h

···
 int dm_thin_insert_block(struct dm_thin_device *td, dm_block_t block,
 			 dm_block_t data_block);
 
-int dm_thin_remove_block(struct dm_thin_device *td, dm_block_t block);
 int dm_thin_remove_range(struct dm_thin_device *td,
 			 dm_block_t begin, dm_block_t end);
 

drivers/md/dm-thin.c

···
 
 static void throttle_work_update(struct throttle *t)
 {
-	if (!t->throttle_applied && jiffies > t->threshold) {
+	if (!t->throttle_applied && time_is_before_jiffies(t->threshold)) {
 		down_write(&t->lock);
 		t->throttle_applied = true;
 	}
···
 	struct pool *pool = tc->pool;
 
 	if (!bio_triggers_commit(tc, bio)) {
-		submit_bio_noacct(bio);
+		dm_submit_bio_remap(bio, NULL);
 		return;
 	}
 
···
 		if (bio->bi_opf & REQ_PREFLUSH)
 			bio_endio(bio);
 		else
-			submit_bio_noacct(bio);
+			dm_submit_bio_remap(bio, NULL);
 	}
 }
 
···
 	struct block_device *data_bdev = pt->data_dev->bdev;
 	struct queue_limits *data_limits = &bdev_get_queue(data_bdev)->limits;
 	const char *reason = NULL;
-	char buf[BDEVNAME_SIZE];
 
 	if (!pt->adjusted_pf.discard_passdown)
 		return;
···
 		reason = "max discard sectors smaller than a block";
 
 	if (reason) {
-		DMWARN("Data device (%s) %s: Disabling discard passdown.", bdevname(data_bdev, buf), reason);
+		DMWARN("Data device (%pg) %s: Disabling discard passdown.", data_bdev, reason);
 		pt->adjusted_pf.discard_passdown = false;
 	}
 }
···
 static void warn_if_metadata_device_too_big(struct block_device *bdev)
 {
 	sector_t metadata_dev_size = get_dev_size(bdev);
-	char buffer[BDEVNAME_SIZE];
 
 	if (metadata_dev_size > THIN_METADATA_MAX_SECTORS_WARNING)
-		DMWARN("Metadata device %s is larger than %u sectors: excess space will not be used.",
-		       bdevname(bdev, buffer), THIN_METADATA_MAX_SECTORS);
+		DMWARN("Metadata device %pg is larger than %u sectors: excess space will not be used.",
+		       bdev, THIN_METADATA_MAX_SECTORS);
 }
 
 static sector_t get_metadata_dev_size(struct block_device *bdev)
···
 
 	ti->num_flush_bios = 1;
 	ti->flush_supported = true;
+	ti->accounts_remapped_io = true;
 	ti->per_io_data_size = sizeof(struct dm_thin_endio_hook);
 
 	/* In case the pool supports discards, pass them on. */

drivers/md/dm-zoned-metadata.c

···
  */
 static int dmz_read_sb(struct dmz_metadata *zmd, struct dmz_sb *sb, int set)
 {
-	dmz_zmd_debug(zmd, "read superblock set %d dev %s block %llu",
-		      set, sb->dev->name, sb->block);
+	dmz_zmd_debug(zmd, "read superblock set %d dev %pg block %llu",
+		      set, sb->dev->bdev, sb->block);
 
 	return dmz_rdwr_block(sb->dev, REQ_OP_READ,
 			      sb->block, sb->mblk->page);

drivers/md/dm-zoned-target.c

···
 	}
 	dev->bdev = bdev;
 	dev->dev_idx = idx;
-	(void)bdevname(dev->bdev, dev->name);
 
 	dev->capacity = bdev_nr_sectors(bdev);
 	if (ti->begin) {

drivers/md/dm-zoned.h

···
 	struct dmz_metadata	*metadata;
 	struct dmz_reclaim	*reclaim;
 
-	char			name[BDEVNAME_SIZE];
 	uuid_t			uuid;
 
 	sector_t		capacity;
···
  * Message functions.
  */
 #define dmz_dev_info(dev, format, args...)	\
-	DMINFO("(%s): " format, (dev)->name, ## args)
+	DMINFO("(%pg): " format, (dev)->bdev, ## args)
 
 #define dmz_dev_err(dev, format, args...)	\
-	DMERR("(%s): " format, (dev)->name, ## args)
+	DMERR("(%pg): " format, (dev)->bdev, ## args)
 
 #define dmz_dev_warn(dev, format, args...)	\
-	DMWARN("(%s): " format, (dev)->name, ## args)
+	DMWARN("(%pg): " format, (dev)->bdev, ## args)
 
 #define dmz_dev_debug(dev, format, args...)	\
-	DMDEBUG("(%s): " format, (dev)->name, ## args)
+	DMDEBUG("(%pg): " format, (dev)->bdev, ## args)
 
 /*
  * Functions defined in dm-zoned-metadata.c

drivers/md/dm.c

···
 #define DM_COOKIE_ENV_VAR_NAME "DM_COOKIE"
 #define DM_COOKIE_LENGTH 24
 
+/*
+ * For REQ_POLLED fs bio, this flag is set if we link mapped underlying
+ * dm_io into one list, and reuse bio->bi_private as the list head. Before
+ * ending this fs bio, we will recover its ->bi_private.
+ */
+#define REQ_DM_POLL_LIST	REQ_DRV
+
 static const char *_name = DM_NAME;
 
 static unsigned int major = 0;
···
 	struct dm_io *io;
 	sector_t sector;
 	unsigned sector_count;
+	bool submit_as_polled;
 };
 
 #define DM_TARGET_IO_BIO_OFFSET (offsetof(struct dm_target_io, clone))
···
 
 void *dm_per_bio_data(struct bio *bio, size_t data_size)
 {
-	if (!clone_to_tio(bio)->inside_dm_io)
+	if (!dm_tio_flagged(clone_to_tio(bio), DM_TIO_INSIDE_DM_IO))
 		return (char *)bio - DM_TARGET_IO_BIO_OFFSET - data_size;
 	return (char *)bio - DM_IO_BIO_OFFSET - data_size;
 }
···
 }
 EXPORT_SYMBOL_GPL(dm_start_time_ns_from_clone);
 
-static void start_io_acct(struct dm_io *io)
+static bool bio_is_flush_with_data(struct bio *bio)
 {
-	struct mapped_device *md = io->md;
-	struct bio *bio = io->orig_bio;
-
-	bio_start_io_acct_time(bio, io->start_time);
-	if (unlikely(dm_stats_used(&md->stats)))
-		dm_stats_account_io(&md->stats, bio_data_dir(bio),
-				    bio->bi_iter.bi_sector, bio_sectors(bio),
-				    false, 0, &io->stats_aux);
+	return ((bio->bi_opf & REQ_PREFLUSH) && bio->bi_iter.bi_size);
 }
 
-static void end_io_acct(struct mapped_device *md, struct bio *bio,
-			unsigned long start_time, struct dm_stats_aux *stats_aux)
+static void dm_io_acct(bool end, struct mapped_device *md, struct bio *bio,
+		       unsigned long start_time, struct dm_stats_aux *stats_aux)
 {
-	unsigned long duration = jiffies - start_time;
+	bool is_flush_with_data;
+	unsigned int bi_size;
 
-	bio_end_io_acct(bio, start_time);
+	/* If REQ_PREFLUSH set save any payload but do not account it */
+	is_flush_with_data = bio_is_flush_with_data(bio);
+	if (is_flush_with_data) {
+		bi_size = bio->bi_iter.bi_size;
+		bio->bi_iter.bi_size = 0;
+	}
+
+	if (!end)
+		bio_start_io_acct_time(bio, start_time);
+	else
+		bio_end_io_acct(bio, start_time);
 
 	if (unlikely(dm_stats_used(&md->stats)))
 		dm_stats_account_io(&md->stats, bio_data_dir(bio),
 				    bio->bi_iter.bi_sector, bio_sectors(bio),
-				    true, duration, stats_aux);
+				    end, start_time, stats_aux);
 
-	/* nudge anyone waiting on suspend queue */
-	if (unlikely(wq_has_sleeper(&md->wait)))
-		wake_up(&md->wait);
+	/* Restore bio's payload so it does get accounted upon requeue */
+	if (is_flush_with_data)
+		bio->bi_iter.bi_size = bi_size;
+}
+
+static void __dm_start_io_acct(struct dm_io *io, struct bio *bio)
+{
+	dm_io_acct(false, io->md, bio, io->start_time, &io->stats_aux);
+}
+
+static void dm_start_io_acct(struct dm_io *io, struct bio *clone)
+{
+	/* Must account IO to DM device in terms of orig_bio */
+	struct bio *bio = io->orig_bio;
+
+	/*
+	 * Ensure IO accounting is only ever started once.
+	 * Expect no possibility for race unless DM_TIO_IS_DUPLICATE_BIO.
+	 */
+	if (!clone ||
+	    likely(!dm_tio_flagged(clone_to_tio(clone), DM_TIO_IS_DUPLICATE_BIO))) {
+		if (WARN_ON_ONCE(dm_io_flagged(io, DM_IO_ACCOUNTED)))
+			return;
+		dm_io_set_flag(io, DM_IO_ACCOUNTED);
+	} else {
+		unsigned long flags;
+		if (dm_io_flagged(io, DM_IO_ACCOUNTED))
+			return;
+		/* Can afford locking given DM_TIO_IS_DUPLICATE_BIO */
+		spin_lock_irqsave(&io->lock, flags);
+		dm_io_set_flag(io, DM_IO_ACCOUNTED);
+		spin_unlock_irqrestore(&io->lock, flags);
+	}
+
+	__dm_start_io_acct(io, bio);
+}
+
+static void dm_end_io_acct(struct dm_io *io, struct bio *bio)
+{
+	dm_io_acct(true, io->md, bio, io->start_time, &io->stats_aux);
 }
 
 static struct dm_io *alloc_io(struct mapped_device *md, struct bio *bio)
···
 	clone = bio_alloc_clone(bio->bi_bdev, bio, GFP_NOIO, &md->io_bs);
 
 	tio = clone_to_tio(clone);
-	tio->inside_dm_io = true;
+	tio->flags = 0;
+	dm_tio_set_flag(tio, DM_TIO_INSIDE_DM_IO);
 	tio->io = NULL;
 
 	io = container_of(tio, struct dm_io, tio);
 	io->magic = DM_IO_MAGIC;
 	io->status = 0;
 	atomic_set(&io->io_count, 1);
-	io->orig_bio = bio;
+	this_cpu_inc(*md->pending_io);
+	io->orig_bio = NULL;
 	io->md = md;
-	spin_lock_init(&io->endio_lock);
-
+	io->map_task = current;
+	spin_lock_init(&io->lock);
 	io->start_time = jiffies;
+	io->flags = 0;
+
+	dm_stats_record_start(&md->stats, &io->stats_aux);
 
 	return io;
 }
 
-static void free_io(struct mapped_device *md, struct dm_io *io)
+static void free_io(struct dm_io *io)
 {
 	bio_put(&io->tio.clone);
 }
···
 		unsigned target_bio_nr, unsigned *len, gfp_t gfp_mask)
 {
 	struct dm_target_io *tio;
+	struct bio *clone;
 
 	if (!ci->io->tio.io) {
 		/* the dm_target_io embedded in ci->io is available */
 		tio = &ci->io->tio;
+		/* alloc_io() already initialized embedded clone */
+		clone = &tio->clone;
 	} else {
-		struct bio *clone = bio_alloc_clone(ci->bio->bi_bdev, ci->bio,
-						    gfp_mask, &ci->io->md->bs);
+		clone = bio_alloc_clone(ci->bio->bi_bdev, ci->bio,
+					gfp_mask, &ci->io->md->bs);
 		if (!clone)
 			return NULL;
 
+		/* REQ_DM_POLL_LIST shouldn't be inherited */
+		clone->bi_opf &= ~REQ_DM_POLL_LIST;
+
 		tio = clone_to_tio(clone);
-		tio->inside_dm_io = false;
+		tio->flags = 0; /* also clears DM_TIO_INSIDE_DM_IO */
 	}
 
 	tio->magic = DM_TIO_MAGIC;
···
 	tio->ti = ti;
 	tio->target_bio_nr = target_bio_nr;
 	tio->len_ptr = len;
+	tio->old_sector = 0;
 
-	return &tio->clone;
+	if (len) {
+		clone->bi_iter.bi_size = to_bytes(*len);
+		if (bio_integrity(clone))
+			bio_integrity_trim(clone);
+	}
+
+	return clone;
 }
 
 static void free_tio(struct bio *clone)
 {
-	if (clone_to_tio(clone)->inside_dm_io)
+	if (dm_tio_flagged(clone_to_tio(clone), DM_TIO_INSIDE_DM_IO))
 		return;
 	bio_put(clone);
 }
···
 	return test_bit(DMF_NOFLUSH_SUSPENDING, &md->flags);
 }
 
+static void dm_io_complete(struct dm_io *io)
+{
+	blk_status_t io_error;
+	struct mapped_device *md = io->md;
+	struct bio *bio = io->orig_bio;
+
+	if (io->status == BLK_STS_DM_REQUEUE) {
+		unsigned long flags;
+		/*
+		 * Target requested pushing back the I/O.
+		 */
+		spin_lock_irqsave(&md->deferred_lock, flags);
+		if (__noflush_suspending(md) &&
+		    !WARN_ON_ONCE(dm_is_zone_write(md, bio))) {
+			/* NOTE early return due to BLK_STS_DM_REQUEUE below */
+			bio_list_add_head(&md->deferred, bio);
+		} else {
+			/*
+			 * noflush suspend was interrupted or this is
+			 * a write to a zoned target.
+			 */
+			io->status = BLK_STS_IOERR;
+		}
+		spin_unlock_irqrestore(&md->deferred_lock, flags);
+	}
+
+	io_error = io->status;
+	if (dm_io_flagged(io, DM_IO_ACCOUNTED))
+		dm_end_io_acct(io, bio);
+	else if (!io_error) {
+		/*
+		 * Must handle target that DM_MAPIO_SUBMITTED only to
+		 * then bio_endio() rather than dm_submit_bio_remap()
+		 */
+		__dm_start_io_acct(io, bio);
+		dm_end_io_acct(io, bio);
+	}
+	free_io(io);
+	smp_wmb();
+	this_cpu_dec(*md->pending_io);
+
+	/* nudge anyone waiting on suspend queue */
+	if (unlikely(wq_has_sleeper(&md->wait)))
+		wake_up(&md->wait);
+
+	if (io_error == BLK_STS_DM_REQUEUE) {
+		/*
+		 * Upper layer won't help us poll split bio, io->orig_bio
+		 * may only reflect a subset of the pre-split original,
+		 * so clear REQ_POLLED in case of requeue
+		 */
+		bio->bi_opf &= ~REQ_POLLED;
+		return;
+	}
+
+	if (bio_is_flush_with_data(bio)) {
+		/*
+		 * Preflush done for flush with data, reissue
+		 * without REQ_PREFLUSH.
+		 */
+		bio->bi_opf &= ~REQ_PREFLUSH;
+		queue_io(md, bio);
+	} else {
+		/* done with normal IO or empty flush */
+		if (io_error)
+			bio->bi_status = io_error;
+		bio_endio(bio);
+	}
+}
+
+static inline bool dm_tio_is_normal(struct dm_target_io *tio)
+{
+	return (dm_tio_flagged(tio, DM_TIO_INSIDE_DM_IO) &&
+		!dm_tio_flagged(tio, DM_TIO_IS_DUPLICATE_BIO));
+}
+
 /*
  * Decrements the number of outstanding ios that a bio has been
  * cloned into, completing the original io if necc.
  */
 void dm_io_dec_pending(struct dm_io *io, blk_status_t error)
 {
-	unsigned long flags;
-	blk_status_t io_error;
-	struct bio *bio;
-	struct mapped_device *md = io->md;
-	unsigned long start_time = 0;
-	struct dm_stats_aux stats_aux;
-
 	/* Push-back supersedes any I/O errors */
 	if (unlikely(error)) {
-		spin_lock_irqsave(&io->endio_lock, flags);
-		if (!(io->status == BLK_STS_DM_REQUEUE && __noflush_suspending(md)))
+		unsigned long flags;
+		spin_lock_irqsave(&io->lock, flags);
+		if (!(io->status == BLK_STS_DM_REQUEUE &&
+		      __noflush_suspending(io->md)))
 			io->status = error;
-		spin_unlock_irqrestore(&io->endio_lock, flags);
+		spin_unlock_irqrestore(&io->lock, flags);
 	}
 
-	if (atomic_dec_and_test(&io->io_count)) {
-		bio = io->orig_bio;
-		if (io->status == BLK_STS_DM_REQUEUE) {
-			/*
-			 * Target requested pushing back the I/O.
-			 */
-			spin_lock_irqsave(&md->deferred_lock, flags);
-			if (__noflush_suspending(md) &&
-			    !WARN_ON_ONCE(dm_is_zone_write(md, bio))) {
-				/* NOTE early return due to BLK_STS_DM_REQUEUE below */
-				bio_list_add_head(&md->deferred, bio);
-			} else {
-				/*
-				 * noflush suspend was interrupted or this is
-				 * a write to a zoned target.
-				 */
-				io->status = BLK_STS_IOERR;
-			}
-			spin_unlock_irqrestore(&md->deferred_lock, flags);
-		}
-
-		io_error = io->status;
-		start_time = io->start_time;
-		stats_aux = io->stats_aux;
-		free_io(md, io);
-		end_io_acct(md, bio, start_time, &stats_aux);
-
-		if (io_error == BLK_STS_DM_REQUEUE)
-			return;
-
-		if ((bio->bi_opf & REQ_PREFLUSH) && bio->bi_iter.bi_size) {
-			/*
-			 * Preflush done for flush with data, reissue
-			 * without REQ_PREFLUSH.
-			 */
-			bio->bi_opf &= ~REQ_PREFLUSH;
-			queue_io(md, bio);
-		} else {
-			/* done with normal IO or empty flush */
-			if (io_error)
-				bio->bi_status = io_error;
-			bio_endio(bio);
-		}
-	}
+	if (atomic_dec_and_test(&io->io_count))
+		dm_io_complete(io);
 }
 
 void disable_discard(struct mapped_device *md)
···
 /*
  * A target may call dm_accept_partial_bio only from the map routine.  It is
  * allowed for all bio types except REQ_PREFLUSH, REQ_OP_ZONE_* zone management
- * operations and REQ_OP_ZONE_APPEND (zone append writes).
+ * operations, REQ_OP_ZONE_APPEND (zone append writes) and any bio serviced by
+ * __send_duplicate_bios().
  *
  * dm_accept_partial_bio informs the dm that the target only wants to process
  * additional n_sectors sectors of the bio and the rest of the data should be
···
 	struct dm_target_io *tio = clone_to_tio(bio);
 	unsigned bi_size = bio->bi_iter.bi_size >> SECTOR_SHIFT;
 
-	BUG_ON(bio->bi_opf & REQ_PREFLUSH);
+	BUG_ON(dm_tio_flagged(tio, DM_TIO_IS_DUPLICATE_BIO));
 	BUG_ON(op_is_zone_mgmt(bio_op(bio)));
 	BUG_ON(bio_op(bio) == REQ_OP_ZONE_APPEND);
 	BUG_ON(bi_size > *tio->len_ptr);
···
 	bio->bi_iter.bi_size = n_sectors << SECTOR_SHIFT;
 }
 EXPORT_SYMBOL_GPL(dm_accept_partial_bio);
+
+static inline void __dm_submit_bio_remap(struct bio *clone,
+					 dev_t dev, sector_t old_sector)
+{
+	trace_block_bio_remap(clone, dev, old_sector);
+	submit_bio_noacct(clone);
+}
+
+/*
+ * @clone: clone bio that DM core passed to target's .map function
+ * @tgt_clone: clone of @clone bio that target needs submitted
+ *
+ * Targets should use this interface to submit bios they take
+ * ownership of when returning DM_MAPIO_SUBMITTED.
+ *
+ * Target should also enable ti->accounts_remapped_io
+ */
+void dm_submit_bio_remap(struct bio *clone, struct bio *tgt_clone)
+{
+	struct dm_target_io *tio = clone_to_tio(clone);
+	struct dm_io *io = tio->io;
+
+	WARN_ON_ONCE(!tio->ti->accounts_remapped_io);
+
+	/* establish bio that will get submitted */
+	if (!tgt_clone)
+		tgt_clone = clone;
+
+	/*
+	 * Account io->origin_bio to DM dev on behalf of target
+	 * that took ownership of IO with DM_MAPIO_SUBMITTED.
+	 */
+	if (io->map_task == current) {
+		/* Still in target's map function */
+		dm_io_set_flag(io, DM_IO_START_ACCT);
+	} else {
+		/*
+		 * Called by another thread, managed by DM target,
+		 * wait for dm_split_and_process_bio() to store
+		 * io->orig_bio
+		 */
+		while (unlikely(!smp_load_acquire(&io->orig_bio)))
+			msleep(1);
+		dm_start_io_acct(io, clone);
+	}
+
+	__dm_submit_bio_remap(tgt_clone, disk_devt(io->md->disk),
+			      tio->old_sector);
+}
+EXPORT_SYMBOL_GPL(dm_submit_bio_remap);
 
 static noinline void __set_swap_bios_limit(struct mapped_device *md, int latch)
 {
···
 {
 	struct dm_target_io *tio = clone_to_tio(clone);
 	int r;
-	sector_t sector;
 	struct dm_io *io = tio->io;
 	struct dm_target *ti = tio->ti;
 
 	clone->bi_end_io = clone_endio;
 
 	/*
-	 * Map the clone.  If r == 0 we don't need to do
-	 * anything, the target has assumed ownership of
-	 * this io.
+	 * Map the clone.
 	 */
 	dm_io_inc_pending(io);
-	sector = clone->bi_iter.bi_sector;
+	tio->old_sector = clone->bi_iter.bi_sector;
 
 	if (unlikely(swap_bios_limit(ti, clone))) {
 		struct mapped_device *md = io->md;
···
 
 	switch (r) {
 	case DM_MAPIO_SUBMITTED:
+		/* target has assumed ownership of this io */
+		if (!ti->accounts_remapped_io)
+			dm_io_set_flag(io, DM_IO_START_ACCT);
 		break;
 	case DM_MAPIO_REMAPPED:
-		/* the bio has been remapped so dispatch it */
-		trace_block_bio_remap(clone, bio_dev(io->orig_bio), sector);
-		submit_bio_noacct(clone);
+		/*
+		 * the bio has been remapped so dispatch it, but defer
+		 * dm_start_io_acct() until after possible bio_split().
+		 */
+		__dm_submit_bio_remap(clone, disk_devt(io->md->disk),
+				      tio->old_sector);
+		dm_io_set_flag(io, DM_IO_START_ACCT);
 		break;
 	case DM_MAPIO_KILL:
-		if (unlikely(swap_bios_limit(ti, clone))) {
-			struct mapped_device *md = io->md;
-			up(&md->swap_bios_semaphore);
-		}
-		free_tio(clone);
-		dm_io_dec_pending(io, BLK_STS_IOERR);
-		break;
 	case DM_MAPIO_REQUEUE:
-		if (unlikely(swap_bios_limit(ti, clone))) {
-			struct mapped_device *md = io->md;
-			up(&md->swap_bios_semaphore);
-		}
+		if (unlikely(swap_bios_limit(ti, clone)))
+			up(&io->md->swap_bios_semaphore);
 		free_tio(clone);
-		dm_io_dec_pending(io, BLK_STS_DM_REQUEUE);
+		if (r == DM_MAPIO_KILL)
+			dm_io_dec_pending(io, BLK_STS_IOERR);
+		else
+			dm_io_dec_pending(io, BLK_STS_DM_REQUEUE);
 		break;
 	default:
 		DMWARN("unimplemented target map return value: %d", r);
 		BUG();
 	}
-}
-
-static void bio_setup_sector(struct bio *bio, sector_t sector, unsigned len)
-{
-	bio->bi_iter.bi_sector = sector;
-	bio->bi_iter.bi_size = to_bytes(len);
-}
-
-/*
- * Creates a bio that consists of range of complete bvecs.
- */
-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, *clone;
-
-	clone = alloc_tio(ci, ti, 0, len, GFP_NOIO);
-	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);
-
-	__map_bio(clone);
-	return 0;
 }
 
 static void alloc_multiple_bios(struct bio_list *blist, struct clone_info *ci,
···
 		break;
 	case 1:
 		clone = alloc_tio(ci, ti, 0, len, GFP_NOIO);
-		if (len)
-			bio_setup_sector(clone, ci->sector, *len);
+		dm_tio_set_flag(clone_to_tio(clone), DM_TIO_IS_DUPLICATE_BIO);
 		__map_bio(clone);
 		break;
 	default:
 		alloc_multiple_bios(&blist, ci, ti, num_bios, len);
 		while ((clone = bio_list_pop(&blist))) {
-			if (len)
-				bio_setup_sector(clone, ci->sector, *len);
+			dm_tio_set_flag(clone_to_tio(clone), DM_TIO_IS_DUPLICATE_BIO);
 			__map_bio(clone);
 		}
 		break;
 	}
 }
 
-static int __send_empty_flush(struct clone_info *ci)
+static void __send_empty_flush(struct clone_info *ci)
 {
 	unsigned target_nr = 0;
 	struct dm_target *ti;
···
 	ci->bio = &flush_bio;
 	ci->sector_count = 0;
 
-	BUG_ON(bio_has_data(ci->bio));
 	while ((ti = dm_table_get_target(ci->map, target_nr++)))
 		__send_duplicate_bios(ci, ti, ti->num_flush_bios, NULL);
 
 	bio_uninit(ci->bio);
-	return 0;
 }
 
-static int __send_changing_extent_only(struct clone_info *ci, struct dm_target *ti,
-				       unsigned num_bios)
+static void __send_changing_extent_only(struct clone_info *ci, struct dm_target *ti,
+					unsigned num_bios)
 {
 	unsigned len;
-
-	/*
-	 * Even though the device advertised support for this type of
-	 * request, that does not mean every target supports it, and
-	 * reconfiguration might also have changed that since the
-	 * check was performed.
-	 */
-	if (!num_bios)
-		return -EOPNOTSUPP;
 
 	len = min_t(sector_t, ci->sector_count,
 		    max_io_len_target_boundary(ti, dm_target_offset(ti, ci->sector)));
 
-	__send_duplicate_bios(ci, ti, num_bios, &len);
-
+	/*
+	 * dm_accept_partial_bio cannot be used with duplicate bios,
+	 * so update clone_info cursor before __send_duplicate_bios().
+	 */
 	ci->sector += len;
 	ci->sector_count -= len;
 
-	return 0;
+	__send_duplicate_bios(ci, ti, num_bios, &len);
 }
 
 static bool is_abnormal_io(struct bio *bio)
···
 static bool __process_abnormal_io(struct clone_info *ci, struct dm_target *ti,
 				  int *result)
 {
-	struct bio *bio = ci->bio;
 	unsigned num_bios = 0;
 
-	switch (bio_op(bio)) {
+	switch (bio_op(ci->bio)) {
 	case REQ_OP_DISCARD:
 		num_bios = ti->num_discard_bios;
 		break;
···
 		return false;
 	}
 
-	*result = __send_changing_extent_only(ci, ti, num_bios);
+	/*
+	 * Even though the device advertised support for this type of
+	 * request, that does not mean every target supports it, and
+	 * reconfiguration might also have changed that since the
+	 * check was performed.
+	 */
+	if (!num_bios)
+		*result = -EOPNOTSUPP;
+	else {
+		__send_changing_extent_only(ci, ti, num_bios);
+		*result = 0;
+	}
 	return true;
+}
+
+/*
+ * Reuse ->bi_private as hlist head for storing all dm_io instances
+ * associated with this bio, and this bio's bi_private needs to be
+ * stored in dm_io->data before the reuse.
+ *
+ * bio->bi_private is owned by fs or upper layer, so block layer won't
+ * touch it after splitting. Meantime it won't be changed by anyone after
+ * bio is submitted. So this reuse is safe.
+ */
+static inline struct hlist_head *dm_get_bio_hlist_head(struct bio *bio)
+{
+	return (struct hlist_head *)&bio->bi_private;
+}
+
+static void dm_queue_poll_io(struct bio *bio, struct dm_io *io)
+{
+	struct hlist_head *head = dm_get_bio_hlist_head(bio);
+
+	if (!(bio->bi_opf & REQ_DM_POLL_LIST)) {
+		bio->bi_opf |= REQ_DM_POLL_LIST;
+		/*
+		 * Save .bi_private into dm_io, so that we can reuse
+		 * .bi_private as hlist head for storing dm_io list
+		 */
+		io->data = bio->bi_private;
+
+		INIT_HLIST_HEAD(head);
+
+		/* tell block layer to poll for completion */
+		bio->bi_cookie = ~BLK_QC_T_NONE;
+	} else {
+		/*
+		 * bio recursed due to split, reuse original poll list,
+		 * and save bio->bi_private too.
+		 */
+		io->data = hlist_entry(head->first, struct dm_io, node)->data;
+	}
+
+	hlist_add_head(&io->node, head);
 }
 
 /*
  * Select the correct strategy for processing a non-flush bio.
  */
-static int __split_and_process_non_flush(struct clone_info *ci)
+static int __split_and_process_bio(struct clone_info *ci)
 {
+	struct bio *clone;
 	struct dm_target *ti;
 	unsigned len;
 	int r;
···
 	if (__process_abnormal_io(ci, ti, &r))
 		return r;
 
-	len = min_t(sector_t, max_io_len(ti, ci->sector), ci->sector_count);
+	/*
+	 * Only support bio polling for normal IO, and the target io is
+	 * exactly inside the dm_io instance (verified in dm_poll_dm_io)
+	 */
+	ci->submit_as_polled = ci->bio->bi_opf & REQ_POLLED;
 
-	r = __clone_and_map_data_bio(ci, ti, ci->sector, &len);
-	if (r < 0)
-		return r;
+	len = min_t(sector_t, max_io_len(ti, ci->sector), ci->sector_count);
+	clone = alloc_tio(ci, ti, 0, &len, GFP_NOIO);
+	__map_bio(clone);
 
 	ci->sector += len;
 	ci->sector_count -= len;
···
 {
 	ci->map = map;
 	ci->io = alloc_io(md, bio);
+	ci->bio = bio;
+	ci->submit_as_polled = false;
 	ci->sector = bio->bi_iter.bi_sector;
+	ci->sector_count = bio_sectors(bio);
+
+	/* Shouldn't happen but sector_count was being set to 0 so... */
+	if (WARN_ON_ONCE(op_is_zone_mgmt(bio_op(bio)) && ci->sector_count))
+		ci->sector_count = 0;
 }
 
 /*
  * Entry point to split a bio into clones and submit them to the targets.
  */
-static void __split_and_process_bio(struct mapped_device *md,
-					struct dm_table *map, struct bio *bio)
+static void dm_split_and_process_bio(struct mapped_device *md,
+				     struct dm_table *map, struct bio *bio)
 {
 	struct clone_info ci;
+	struct bio *orig_bio = NULL;
 	int error = 0;
 
 	init_clone_info(&ci, md, map, bio);
 
 	if (bio->bi_opf & REQ_PREFLUSH) {
-		error = __send_empty_flush(&ci);
-		/* dm_io_dec_pending submits any data associated with flush */
-	} else if (op_is_zone_mgmt(bio_op(bio))) {
-		ci.bio = bio;
-		ci.sector_count = 0;
-		error = __split_and_process_non_flush(&ci);
-	} else {
-		ci.bio = bio;
-		ci.sector_count = bio_sectors(bio);
-		error = __split_and_process_non_flush(&ci);
-		if (ci.sector_count && !error) {
-			/*
-			 * Remainder must be passed to submit_bio_noacct()
-			 * so that it gets handled *after* bios already submitted
-			 * have been completely processed.
-			 * We take a clone of the original to store in
-			 * ci.io->orig_bio to be used by end_io_acct() and
-			 * for dec_pending to use for completion handling.
-			 */
-			struct bio *b = bio_split(bio, bio_sectors(bio) - ci.sector_count,
-						  GFP_NOIO, &md->queue->bio_split);
-			ci.io->orig_bio = b;
-
-			bio_chain(b, bio);
-			trace_block_split(b, bio->bi_iter.bi_sector);
-			submit_bio_noacct(bio);
-		}
+		__send_empty_flush(&ci);
+		/* dm_io_complete submits any data associated with flush */
+		goto out;
 	}
-	start_io_acct(ci.io);
 
-	/* drop the extra reference count */
-	dm_io_dec_pending(ci.io, errno_to_blk_status(error));
+	error = __split_and_process_bio(&ci);
+	ci.io->map_task = NULL;
+	if (error || !ci.sector_count)
+		goto out;
+
+	/*
+	 * Remainder must be passed to submit_bio_noacct() so it gets handled
+	 * *after* bios already submitted have been completely processed.
+	 * We take a clone of the original to store in ci.io->orig_bio to be
+	 * used by dm_end_io_acct() and for dm_io_complete() to use for
+	 * completion handling.
+	 */
+	orig_bio = bio_split(bio, bio_sectors(bio) - ci.sector_count,
+			     GFP_NOIO, &md->queue->bio_split);
+	bio_chain(orig_bio, bio);
+	trace_block_split(orig_bio, bio->bi_iter.bi_sector);
+	submit_bio_noacct(bio);
+out:
+	if (!orig_bio)
+		orig_bio = bio;
+	smp_store_release(&ci.io->orig_bio, orig_bio);
+	if (dm_io_flagged(ci.io, DM_IO_START_ACCT))
+		dm_start_io_acct(ci.io, NULL);
+
+	/*
+	 * Drop the extra reference count for non-POLLED bio, and hold one
+	 * reference for POLLED bio, which will be released in dm_poll_bio
+	 *
+	 * Add every dm_io instance into the hlist_head which is stored in
+	 * bio->bi_private, so that dm_poll_bio can poll them all.
+	 */
+	if (error || !ci.submit_as_polled)
+		dm_io_dec_pending(ci.io, errno_to_blk_status(error));
+	else
+		dm_queue_poll_io(bio, ci.io);
 }
 
 static void dm_submit_bio(struct bio *bio)
···
 	struct dm_table *map;
 
 	map = dm_get_live_table(md, &srcu_idx);
-	if (unlikely(!map)) {
-		DMERR_LIMIT("%s: mapping table unavailable, erroring io",
-			    dm_device_name(md));
-		bio_io_error(bio);
-		goto out;
-	}
 
-	/* If suspended, queue this IO for later */
-	if (unlikely(test_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags))) {
+	/* If suspended, or map not yet available, queue this IO for later */
+	if (unlikely(test_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags)) ||
+	    unlikely(!map)) {
 		if (bio->bi_opf & REQ_NOWAIT)
 			bio_wouldblock_error(bio);
 		else if (bio->bi_opf & REQ_RAHEAD)
···
 	if (is_abnormal_io(bio))
 		blk_queue_split(&bio);
 
-	__split_and_process_bio(md, map, bio);
+	dm_split_and_process_bio(md, map, bio);
 out:
 	dm_put_live_table(md, srcu_idx);
+}
+
+static bool dm_poll_dm_io(struct dm_io *io, struct io_comp_batch *iob,
+			  unsigned int flags)
+{
+	WARN_ON_ONCE(!dm_tio_is_normal(&io->tio));
+
+	/* don't poll if the mapped io is done */
+	if (atomic_read(&io->io_count) > 1)
+		bio_poll(&io->tio.clone, iob, flags);
+
+	/* bio_poll holds the last reference */
+	return atomic_read(&io->io_count) == 1;
+}
+
+static int dm_poll_bio(struct bio *bio, struct io_comp_batch *iob,
+		       unsigned int flags)
+{
+	struct hlist_head *head = dm_get_bio_hlist_head(bio);
+	struct hlist_head tmp = HLIST_HEAD_INIT;
+	struct hlist_node *next;
+	struct dm_io *io;
+
+	/* Only poll normal bio which was marked as REQ_DM_POLL_LIST */
+	if (!(bio->bi_opf & REQ_DM_POLL_LIST))
+		return 0;
+
+	WARN_ON_ONCE(hlist_empty(head));
+
+	hlist_move_list(head, &tmp);
+
+	/*
+	 * Restore .bi_private before possibly completing dm_io.
+	 *
+	 * bio_poll() is only possible once @bio has been completely
+	 * submitted via submit_bio_noacct()'s depth-first submission.
+	 * So there is no dm_queue_poll_io() race associated with
+	 * clearing REQ_DM_POLL_LIST here.
+	 */
+	bio->bi_opf &= ~REQ_DM_POLL_LIST;
+	bio->bi_private = hlist_entry(tmp.first, struct dm_io, node)->data;
+
+	hlist_for_each_entry_safe(io, next, &tmp, node) {
+		if (dm_poll_dm_io(io, iob, flags)) {
+			hlist_del_init(&io->node);
+			/*
+			 * clone_endio() has already occurred, so passing
+			 * error as 0 here doesn't override io->status
+			 */
+			dm_io_dec_pending(io, 0);
+		}
+	}
+
+	/* Not done? */
+	if (!hlist_empty(&tmp)) {
+		bio->bi_opf |= REQ_DM_POLL_LIST;
+		/* Reset bio->bi_private to dm_io list head */
+		hlist_move_list(&tmp, head);
+		return 0;
+	}
+	return 1;
 }
 
 /*-----------------------------------------------------------------
···
 		md->dax_dev = NULL;
 	}
 
+	dm_cleanup_zoned_dev(md);
 	if (md->disk) {
 		spin_lock(&_minor_lock);
 		md->disk->private_data = NULL;
···
 		blk_cleanup_disk(md->disk);
 	}
 
+	if (md->pending_io) {
+		free_percpu(md->pending_io);
+		md->pending_io = NULL;
+	}
+
 	cleanup_srcu_struct(&md->io_barrier);
 
 	mutex_destroy(&md->suspend_lock);
···
 	mutex_destroy(&md->swap_bios_lock);
 
 	dm_mq_cleanup_mapped_device(md);
-	dm_cleanup_zoned_dev(md);
 }
 
 /*
···
 
 	md->wq = alloc_workqueue("kdmflush/%s", WQ_MEM_RECLAIM, 0, md->name);
 	if (!md->wq)
+		goto bad;
+
+	md->pending_io = alloc_percpu(unsigned long);
+	if (!md->pending_io)
 		goto bad;
 
 	dm_stats_init(&md->stats);
···
 			       struct queue_limits *limits)
 {
 	struct dm_table *old_map;
-	struct request_queue *q = md->queue;
-	bool request_based = dm_table_request_based(t);
 	sector_t size;
 	int ret;
 
···
 
 	dm_table_event_callback(t, event_callback, md);
 
-	if (request_based) {
+	if (dm_table_request_based(t)) {
 		/*
 		 * Leverage the fact that request-based DM targets are
 		 * immutable singletons - used to optimize dm_mq_queue_rq.
···
 		goto out;
 	}
 
-	ret = dm_table_set_restrictions(t, q, limits);
+	ret = dm_table_set_restrictions(t, md->queue, limits);
 	if (ret) {
 		old_map = ERR_PTR(ret);
 		goto out;
···
 
 	if (old_map)
 		dm_sync_table(md);
-
 out:
 	return old_map;
 }
···
 }
 EXPORT_SYMBOL_GPL(dm_put);
 
-static bool md_in_flight_bios(struct mapped_device *md)
+static bool dm_in_flight_bios(struct mapped_device *md)
 {
 	int cpu;
-	struct block_device *part = dm_disk(md)->part0;
-	long sum = 0;
+	unsigned long sum = 0;
 
-	for_each_possible_cpu(cpu) {
-		sum += part_stat_local_read_cpu(part, in_flight[0], cpu);
-		sum += part_stat_local_read_cpu(part, in_flight[1], cpu);
-	}
+	for_each_possible_cpu(cpu)
+		sum += *per_cpu_ptr(md->pending_io, cpu);
 
 	return sum != 0;
 }
···
 	while (true) {
 		prepare_to_wait(&md->wait, &wait, task_state);
 
-		if (!md_in_flight_bios(md))
+		if (!dm_in_flight_bios(md))
 			break;
 
 		if (signal_pending_state(task_state, current)) {
···
 		io_schedule();
 	}
 	finish_wait(&md->wait, &wait);
+
+	smp_rmb();
 
 	return r;
 }
···
 	/*
 	 * Here we must make sure that no processes are submitting requests
 	 * to target drivers i.e. no one may be executing
-	 * __split_and_process_bio from dm_submit_bio.
+	 * dm_split_and_process_bio from dm_submit_bio.
 	 *
-	 * To get all processes out of __split_and_process_bio in dm_submit_bio,
+	 * To get all processes out of dm_split_and_process_bio in dm_submit_bio,
 	 * we take the write lock. To prevent any process from reentering
-	 * __split_and_process_bio from dm_submit_bio and quiesce the thread
+	 * dm_split_and_process_bio from dm_submit_bio and quiesce the thread
 	 * (dm_wq_work), we set DMF_BLOCK_IO_FOR_SUSPEND and call
 	 * flush_workqueue(md->wq).
 	 */
···
 
 static const struct block_device_operations dm_blk_dops = {
 	.submit_bio = dm_submit_bio,
+	.poll_bio = dm_poll_bio,
 	.open = dm_blk_open,
 	.release = dm_blk_close,
 	.ioctl = dm_blk_ioctl,

include/linux/blkdev.h

···
 
 struct block_device_operations {
 	void (*submit_bio)(struct bio *bio);
+	int (*poll_bio)(struct bio *bio, struct io_comp_batch *iob,
+			unsigned int flags);
 	int (*open) (struct block_device *, fmode_t);
 	void (*release) (struct gendisk *, fmode_t);
 	int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);

include/linux/device-mapper.h

···
 	bool limit_swap_bios:1;
 
 	/*
-	 * Set if this target implements a a zoned device and needs emulation of
+	 * Set if this target implements a zoned device and needs emulation of
 	 * zone append operations using regular writes.
 	 */
 	bool emulate_zone_append:1;
+
+	/*
+	 * Set if the target will submit IO using dm_submit_bio_remap()
+	 * after returning DM_MAPIO_SUBMITTED from its map function.
+	 */
+	bool accounts_remapped_io:1;
 };
 
 void *dm_per_bio_data(struct bio *bio, size_t data_size);
···
 int dm_post_suspending(struct dm_target *ti);
 int dm_noflush_suspending(struct dm_target *ti);
 void dm_accept_partial_bio(struct bio *bio, unsigned n_sectors);
+void dm_submit_bio_remap(struct bio *clone, struct bio *tgt_clone);
 union map_info *dm_get_rq_mapinfo(struct request *rq);
 
 #ifdef CONFIG_BLK_DEV_ZONED

include/uapi/linux/dm-ioctl.h

···
 #define DM_DEV_SET_GEOMETRY	_IOWR(DM_IOCTL, DM_DEV_SET_GEOMETRY_CMD, struct dm_ioctl)
 
 #define DM_VERSION_MAJOR	4
-#define DM_VERSION_MINOR	45
+#define DM_VERSION_MINOR	46
 #define DM_VERSION_PATCHLEVEL	0
-#define DM_VERSION_EXTRA	"-ioctl (2021-03-22)"
+#define DM_VERSION_EXTRA	"-ioctl (2022-02-22)"
 
 /* Status bits */
 #define DM_READONLY_FLAG	(1 << 0) /* In/Out */