Merge tag 'for-4.21/block-20181221' of git://git.kernel.dk/linux-block

Pull block updates from Jens Axboe:
 "This is the main pull request for block/storage for 4.21.

  Larger than usual, it was a busy round with lots of goodies queued up.
  Most notable is the removal of the old IO stack, which has been a long
  time coming. No new features for a while, everything coming in this
  week has all been fixes for things that were previously merged.

  This contains:

   - Use atomic counters instead of semaphores for mtip32xx (Arnd)

   - Cleanup of the mtip32xx request setup (Christoph)

   - Fix for circular locking dependency in loop (Jan, Tetsuo)

   - bcache (Coly, Guoju, Shenghui)
      * Optimizations for writeback caching
      * Various fixes and improvements

   - nvme (Chaitanya, Christoph, Sagi, Jay, me, Keith)
      * host and target support for NVMe over TCP
      * Error log page support
      * Support for separate read/write/poll queues
      * Much improved polling
      * discard OOM fallback
      * Tracepoint improvements

   - lightnvm (Hans, Hua, Igor, Matias, Javier)
      * Igor added packed metadata to pblk. Now drives without metadata
        per LBA can be used as well.
      * Fix from Geert on uninitialized value on chunk metadata reads.
      * Fixes from Hans and Javier to pblk recovery and write path.
      * Fix from Hua Su to fix a race condition in the pblk recovery
        code.
      * Scan optimization added to pblk recovery from Zhoujie.
      * Small geometry cleanup from me.

   - Conversion of the last few drivers that used the legacy path to
     blk-mq (me)

   - Removal of legacy IO path in SCSI (me, Christoph)

   - Removal of legacy IO stack and schedulers (me)

   - Support for much better polling, now without interrupts at all.
     blk-mq adds support for multiple queue maps, which enables us to
     have a map per type. This in turn enables nvme to have separate
     completion queues for polling, which can then be interrupt-less.
     Also means we're ready for async polled IO, which is hopefully
     coming in the next release.

   - Killing of (now) unused block exports (Christoph)

   - Unification of the blk-rq-qos and blk-wbt wait handling (Josef)

   - Support for zoned testing with null_blk (Masato)

   - sx8 conversion to per-host tag sets (Christoph)

   - IO priority improvements (Damien)

   - mq-deadline zoned fix (Damien)

   - Ref count blkcg series (Dennis)

   - Lots of blk-mq improvements and speedups (me)

   - sbitmap scalability improvements (me)

   - Make core inflight IO accounting per-cpu (Mikulas)

   - Export timeout setting in sysfs (Weiping)

   - Cleanup the direct issue path (Jianchao)

   - Export blk-wbt internals in block debugfs for easier debugging
     (Ming)

   - Lots of other fixes and improvements"

* tag 'for-4.21/block-20181221' of git://git.kernel.dk/linux-block: (364 commits)
  kyber: use sbitmap add_wait_queue/list_del wait helpers
  sbitmap: add helpers for add/del wait queue handling
  block: save irq state in blkg_lookup_create()
  dm: don't reuse bio for flushes
  nvme-pci: trace SQ status on completions
  nvme-rdma: implement polling queue map
  nvme-fabrics: allow user to pass in nr_poll_queues
  nvme-fabrics: allow nvmf_connect_io_queue to poll
  nvme-core: optionally poll sync commands
  block: make request_to_qc_t public
  nvme-tcp: fix spelling mistake "attepmpt" -> "attempt"
  nvme-tcp: fix endianess annotations
  nvmet-tcp: fix endianess annotations
  nvme-pci: refactor nvme_poll_irqdisable to make sparse happy
  nvme-pci: only set nr_maps to 2 if poll queues are supported
  nvmet: use a macro for default error location
  nvmet: fix comparison of a u16 with -1
  blk-mq: enable IO poll if .nr_queues of type poll > 0
  blk-mq: change blk_mq_queue_busy() to blk_mq_queue_inflight()
  blk-mq: skip zero-queue maps in blk_mq_map_swqueue
  ...

16 files changed, 231 insertions, 78 deletions

diff --git a/drivers/md/bcache/bcache.h b/drivers/md/bcache/bcache.h
index b61b83bbcfff..fdf75352e16a 100644
--- a/drivers/md/bcache/bcache.h
+++ b/drivers/md/bcache/bcache.h
@@ -627,6 +627,20 @@ struct cache_set {
 	struct bkey		gc_done;
 
 	/*
+	 * For automatical garbage collection after writeback completed, this
+	 * varialbe is used as bit fields,
+	 * - 0000 0001b (BCH_ENABLE_AUTO_GC): enable gc after writeback
+	 * - 0000 0010b (BCH_DO_AUTO_GC):     do gc after writeback
+	 * This is an optimization for following write request after writeback
+	 * finished, but read hit rate dropped due to clean data on cache is
+	 * discarded. Unless user explicitly sets it via sysfs, it won't be
+	 * enabled.
+	 */
+#define BCH_ENABLE_AUTO_GC	1
+#define BCH_DO_AUTO_GC		2
+	uint8_t			gc_after_writeback;
+
+	/*
 	 * The allocation code needs gc_mark in struct bucket to be correct, but
 	 * it's not while a gc is in progress. Protected by bucket_lock.
 	 */
@@ -658,7 +672,11 @@ struct cache_set {
 
 	/*
 	 * A btree node on disk could have too many bsets for an iterator to fit
-	 * on the stack - have to dynamically allocate them
+	 * on the stack - have to dynamically allocate them.
+	 * bch_cache_set_alloc() will make sure the pool can allocate iterators
+	 * equipped with enough room that can host
+	 *     (sb.bucket_size / sb.block_size)
+	 * btree_iter_sets, which is more than static MAX_BSETS.
 	 */
 	mempool_t		fill_iter;
 
diff --git a/drivers/md/bcache/btree.c b/drivers/md/bcache/btree.c
index 3f4211b5cd33..23cb1dc7296b 100644
--- a/drivers/md/bcache/btree.c
+++ b/drivers/md/bcache/btree.c
@@ -207,6 +207,11 @@ void bch_btree_node_read_done(struct btree *b)
 	struct bset *i = btree_bset_first(b);
 	struct btree_iter *iter;
 
+	/*
+	 * c->fill_iter can allocate an iterator with more memory space
+	 * than static MAX_BSETS.
+	 * See the comment arount cache_set->fill_iter.
+	 */
 	iter = mempool_alloc(&b->c->fill_iter, GFP_NOIO);
 	iter->size = b->c->sb.bucket_size / b->c->sb.block_size;
 	iter->used = 0;
diff --git a/drivers/md/bcache/btree.h b/drivers/md/bcache/btree.h
index a68d6c55783b..d1c72ef64edf 100644
--- a/drivers/md/bcache/btree.h
+++ b/drivers/md/bcache/btree.h
@@ -266,6 +266,24 @@ static inline void wake_up_gc(struct cache_set *c)
 	wake_up(&c->gc_wait);
 }
 
+static inline void force_wake_up_gc(struct cache_set *c)
+{
+	/*
+	 * Garbage collection thread only works when sectors_to_gc < 0,
+	 * calling wake_up_gc() won't start gc thread if sectors_to_gc is
+	 * not a nagetive value.
+	 * Therefore sectors_to_gc is set to -1 here, before waking up
+	 * gc thread by calling wake_up_gc(). Then gc_should_run() will
+	 * give a chance to permit gc thread to run. "Give a chance" means
+	 * before going into gc_should_run(), there is still possibility
+	 * that c->sectors_to_gc being set to other positive value. So
+	 * this routine won't 100% make sure gc thread will be woken up
+	 * to run.
+	 */
+	atomic_set(&c->sectors_to_gc, -1);
+	wake_up_gc(c);
+}
+
 #define MAP_DONE	0
 #define MAP_CONTINUE	1
 
diff --git a/drivers/md/bcache/debug.c b/drivers/md/bcache/debug.c
index 8f448b9c96a1..8b123be05254 100644
--- a/drivers/md/bcache/debug.c
+++ b/drivers/md/bcache/debug.c
@@ -249,8 +249,7 @@ void bch_debug_init_cache_set(struct cache_set *c)
 
 void bch_debug_exit(void)
 {
-	if (!IS_ERR_OR_NULL(bcache_debug))
-		debugfs_remove_recursive(bcache_debug);
+	debugfs_remove_recursive(bcache_debug);
 }
 
 void __init bch_debug_init(void)
diff --git a/drivers/md/bcache/journal.c b/drivers/md/bcache/journal.c
index 522c7426f3a0..b2fd412715b1 100644
--- a/drivers/md/bcache/journal.c
+++ b/drivers/md/bcache/journal.c
@@ -663,7 +663,7 @@ static void journal_write_unlocked(struct closure *cl)
 				 REQ_SYNC|REQ_META|REQ_PREFLUSH|REQ_FUA);
 		bch_bio_map(bio, w->data);
 
-		trace_bcache_journal_write(bio);
+		trace_bcache_journal_write(bio, w->data->keys);
 		bio_list_add(&list, bio);
 
 		SET_PTR_OFFSET(k, i, PTR_OFFSET(k, i) + sectors);
diff --git a/drivers/md/bcache/request.c b/drivers/md/bcache/request.c
index 3bf35914bb57..15070412a32e 100644
--- a/drivers/md/bcache/request.c
+++ b/drivers/md/bcache/request.c
@@ -311,11 +311,11 @@ err:
  * data is written it calls bch_journal, and after the keys have been added to
  * the next journal write they're inserted into the btree.
  *
- * It inserts the data in s->cache_bio; bi_sector is used for the key offset,
+ * It inserts the data in op->bio; bi_sector is used for the key offset,
  * and op->inode is used for the key inode.
  *
- * If s->bypass is true, instead of inserting the data it invalidates the
- * region of the cache represented by s->cache_bio and op->inode.
+ * If op->bypass is true, instead of inserting the data it invalidates the
+ * region of the cache represented by op->bio and op->inode.
  */
 void bch_data_insert(struct closure *cl)
 {
diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c
index 7bbd670a5a84..4dee119c3664 100644
--- a/drivers/md/bcache/super.c
+++ b/drivers/md/bcache/super.c
@@ -25,8 +25,8 @@
 #include <linux/reboot.h>
 #include <linux/sysfs.h>
 
-MODULE_LICENSE("GPL");
-MODULE_AUTHOR("Kent Overstreet <kent.overstreet@gmail.com>");
+unsigned int bch_cutoff_writeback;
+unsigned int bch_cutoff_writeback_sync;
 
 static const char bcache_magic[] = {
 	0xc6, 0x85, 0x73, 0xf6, 0x4e, 0x1a, 0x45, 0xca,
@@ -1510,8 +1510,7 @@ static void cache_set_free(struct closure *cl)
 	struct cache *ca;
 	unsigned int i;
 
-	if (!IS_ERR_OR_NULL(c->debug))
-		debugfs_remove(c->debug);
+	debugfs_remove(c->debug);
 
 	bch_open_buckets_free(c);
 	bch_btree_cache_free(c);
@@ -2424,6 +2423,32 @@ static void bcache_exit(void)
 	mutex_destroy(&bch_register_lock);
 }
 
+/* Check and fixup module parameters */
+static void check_module_parameters(void)
+{
+	if (bch_cutoff_writeback_sync == 0)
+		bch_cutoff_writeback_sync = CUTOFF_WRITEBACK_SYNC;
+	else if (bch_cutoff_writeback_sync > CUTOFF_WRITEBACK_SYNC_MAX) {
+		pr_warn("set bch_cutoff_writeback_sync (%u) to max value %u",
+			bch_cutoff_writeback_sync, CUTOFF_WRITEBACK_SYNC_MAX);
+		bch_cutoff_writeback_sync = CUTOFF_WRITEBACK_SYNC_MAX;
+	}
+
+	if (bch_cutoff_writeback == 0)
+		bch_cutoff_writeback = CUTOFF_WRITEBACK;
+	else if (bch_cutoff_writeback > CUTOFF_WRITEBACK_MAX) {
+		pr_warn("set bch_cutoff_writeback (%u) to max value %u",
+			bch_cutoff_writeback, CUTOFF_WRITEBACK_MAX);
+		bch_cutoff_writeback = CUTOFF_WRITEBACK_MAX;
+	}
+
+	if (bch_cutoff_writeback > bch_cutoff_writeback_sync) {
+		pr_warn("set bch_cutoff_writeback (%u) to %u",
+			bch_cutoff_writeback, bch_cutoff_writeback_sync);
+		bch_cutoff_writeback = bch_cutoff_writeback_sync;
+	}
+}
+
 static int __init bcache_init(void)
 {
 	static const struct attribute *files[] = {
@@ -2432,6 +2457,8 @@ static int __init bcache_init(void)
 		NULL
 	};
 
+	check_module_parameters();
+
 	mutex_init(&bch_register_lock);
 	init_waitqueue_head(&unregister_wait);
 	register_reboot_notifier(&reboot);
@@ -2468,5 +2495,18 @@ err:
 	return -ENOMEM;
 }
 
+/*
+ * Module hooks
+ */
 module_exit(bcache_exit);
 module_init(bcache_init);
+
+module_param(bch_cutoff_writeback, uint, 0);
+MODULE_PARM_DESC(bch_cutoff_writeback, "threshold to cutoff writeback");
+
+module_param(bch_cutoff_writeback_sync, uint, 0);
+MODULE_PARM_DESC(bch_cutoff_writeback_sync, "hard threshold to cutoff writeback");
+
+MODULE_DESCRIPTION("Bcache: a Linux block layer cache");
+MODULE_AUTHOR("Kent Overstreet <kent.overstreet@gmail.com>");
+MODULE_LICENSE("GPL");
diff --git a/drivers/md/bcache/sysfs.c b/drivers/md/bcache/sysfs.c
index 26f035a0c5b9..557a8a3270a1 100644
--- a/drivers/md/bcache/sysfs.c
+++ b/drivers/md/bcache/sysfs.c
@@ -16,7 +16,7 @@
 #include <linux/sort.h>
 #include <linux/sched/clock.h>
 
-/* Default is -1; we skip past it for struct cached_dev's cache mode */
+/* Default is 0 ("writethrough") */
 static const char * const bch_cache_modes[] = {
 	"writethrough",
 	"writeback",
@@ -25,7 +25,7 @@ static const char * const bch_cache_modes[] = {
 	NULL
 };
 
-/* Default is -1; we skip past it for stop_when_cache_set_failed */
+/* Default is 0 ("auto") */
 static const char * const bch_stop_on_failure_modes[] = {
 	"auto",
 	"always",
@@ -88,6 +88,8 @@ read_attribute(writeback_keys_done);
 read_attribute(writeback_keys_failed);
 read_attribute(io_errors);
 read_attribute(congested);
+read_attribute(cutoff_writeback);
+read_attribute(cutoff_writeback_sync);
 rw_attribute(congested_read_threshold_us);
 rw_attribute(congested_write_threshold_us);
 
@@ -128,6 +130,7 @@ rw_attribute(expensive_debug_checks);
 rw_attribute(cache_replacement_policy);
 rw_attribute(btree_shrinker_disabled);
 rw_attribute(copy_gc_enabled);
+rw_attribute(gc_after_writeback);
 rw_attribute(size);
 
 static ssize_t bch_snprint_string_list(char *buf,
@@ -264,7 +267,8 @@ STORE(__cached_dev)
 	d_strtoul(writeback_running);
 	d_strtoul(writeback_delay);
 
-	sysfs_strtoul_clamp(writeback_percent, dc->writeback_percent, 0, 40);
+	sysfs_strtoul_clamp(writeback_percent, dc->writeback_percent,
+			    0, bch_cutoff_writeback);
 
 	if (attr == &sysfs_writeback_rate) {
 		ssize_t ret;
@@ -384,8 +388,25 @@ STORE(bch_cached_dev)
 	mutex_lock(&bch_register_lock);
 	size = __cached_dev_store(kobj, attr, buf, size);
 
-	if (attr == &sysfs_writeback_running)
-		bch_writeback_queue(dc);
+	if (attr == &sysfs_writeback_running) {
+		/* dc->writeback_running changed in __cached_dev_store() */
+		if (IS_ERR_OR_NULL(dc->writeback_thread)) {
+			/*
+			 * reject setting it to 1 via sysfs if writeback
+			 * kthread is not created yet.
+			 */
+			if (dc->writeback_running) {
+				dc->writeback_running = false;
+				pr_err("%s: failed to run non-existent writeback thread",
+						dc->disk.disk->disk_name);
+			}
+		} else
+			/*
+			 * writeback kthread will check if dc->writeback_running
+			 * is true or false.
+			 */
+			bch_writeback_queue(dc);
+	}
 
 	if (attr == &sysfs_writeback_percent)
 		if (!test_and_set_bit(BCACHE_DEV_WB_RUNNING, &dc->disk.flags))
@@ -668,6 +689,9 @@ SHOW(__bch_cache_set)
 	sysfs_print(congested_write_threshold_us,
 		    c->congested_write_threshold_us);
 
+	sysfs_print(cutoff_writeback, bch_cutoff_writeback);
+	sysfs_print(cutoff_writeback_sync, bch_cutoff_writeback_sync);
+
 	sysfs_print(active_journal_entries,	fifo_used(&c->journal.pin));
 	sysfs_printf(verify,			"%i", c->verify);
 	sysfs_printf(key_merging_disabled,	"%i", c->key_merging_disabled);
@@ -676,6 +700,7 @@ SHOW(__bch_cache_set)
 	sysfs_printf(gc_always_rewrite,		"%i", c->gc_always_rewrite);
 	sysfs_printf(btree_shrinker_disabled,	"%i", c->shrinker_disabled);
 	sysfs_printf(copy_gc_enabled,		"%i", c->copy_gc_enabled);
+	sysfs_printf(gc_after_writeback,	"%i", c->gc_after_writeback);
 	sysfs_printf(io_disable,		"%i",
 		     test_bit(CACHE_SET_IO_DISABLE, &c->flags));
 
@@ -725,21 +750,8 @@ STORE(__bch_cache_set)
 		bch_cache_accounting_clear(&c->accounting);
 	}
 
-	if (attr == &sysfs_trigger_gc) {
-		/*
-		 * Garbage collection thread only works when sectors_to_gc < 0,
-		 * when users write to sysfs entry trigger_gc, most of time
-		 * they want to forcibly triger gargage collection. Here -1 is
-		 * set to c->sectors_to_gc, to make gc_should_run() give a
-		 * chance to permit gc thread to run. "give a chance" means
-		 * before going into gc_should_run(), there is still chance
-		 * that c->sectors_to_gc being set to other positive value. So
-		 * writing sysfs entry trigger_gc won't always make sure gc
-		 * thread takes effect.
-		 */
-		atomic_set(&c->sectors_to_gc, -1);
-		wake_up_gc(c);
-	}
+	if (attr == &sysfs_trigger_gc)
+		force_wake_up_gc(c);
 
 	if (attr == &sysfs_prune_cache) {
 		struct shrink_control sc;
@@ -789,6 +801,12 @@ STORE(__bch_cache_set)
 	sysfs_strtoul(gc_always_rewrite,	c->gc_always_rewrite);
 	sysfs_strtoul(btree_shrinker_disabled,	c->shrinker_disabled);
 	sysfs_strtoul(copy_gc_enabled,		c->copy_gc_enabled);
+	/*
+	 * write gc_after_writeback here may overwrite an already set
+	 * BCH_DO_AUTO_GC, it doesn't matter because this flag will be
+	 * set in next chance.
+	 */
+	sysfs_strtoul_clamp(gc_after_writeback, c->gc_after_writeback, 0, 1);
 
 	return size;
 }
@@ -869,7 +887,10 @@ static struct attribute *bch_cache_set_internal_files[] = {
 	&sysfs_gc_always_rewrite,
 	&sysfs_btree_shrinker_disabled,
 	&sysfs_copy_gc_enabled,
+	&sysfs_gc_after_writeback,
 	&sysfs_io_disable,
+	&sysfs_cutoff_writeback,
+	&sysfs_cutoff_writeback_sync,
 	NULL
 };
 KTYPE(bch_cache_set_internal);
diff --git a/drivers/md/bcache/writeback.c b/drivers/md/bcache/writeback.c
index 08c3a9f9676c..73f0efac2b9f 100644
--- a/drivers/md/bcache/writeback.c
+++ b/drivers/md/bcache/writeback.c
@@ -17,6 +17,15 @@
 #include <linux/sched/clock.h>
 #include <trace/events/bcache.h>
 
+static void update_gc_after_writeback(struct cache_set *c)
+{
+	if (c->gc_after_writeback != (BCH_ENABLE_AUTO_GC) ||
+	    c->gc_stats.in_use < BCH_AUTO_GC_DIRTY_THRESHOLD)
+		return;
+
+	c->gc_after_writeback |= BCH_DO_AUTO_GC;
+}
+
 /* Rate limiting */
 static uint64_t __calc_target_rate(struct cached_dev *dc)
 {
@@ -191,6 +200,7 @@ static void update_writeback_rate(struct work_struct *work)
 		if (!set_at_max_writeback_rate(c, dc)) {
 			down_read(&dc->writeback_lock);
 			__update_writeback_rate(dc);
+			update_gc_after_writeback(c);
 			up_read(&dc->writeback_lock);
 		}
 	}
@@ -689,6 +699,23 @@ static int bch_writeback_thread(void *arg)
 				up_write(&dc->writeback_lock);
 				break;
 			}
+
+			/*
+			 * When dirty data rate is high (e.g. 50%+), there might
+			 * be heavy buckets fragmentation after writeback
+			 * finished, which hurts following write performance.
+			 * If users really care about write performance they
+			 * may set BCH_ENABLE_AUTO_GC via sysfs, then when
+			 * BCH_DO_AUTO_GC is set, garbage collection thread
+			 * will be wake up here. After moving gc, the shrunk
+			 * btree and discarded free buckets SSD space may be
+			 * helpful for following write requests.
+			 */
+			if (c->gc_after_writeback ==
+			    (BCH_ENABLE_AUTO_GC|BCH_DO_AUTO_GC)) {
+				c->gc_after_writeback &= ~BCH_DO_AUTO_GC;
+				force_wake_up_gc(c);
+			}
 		}
 
 		up_write(&dc->writeback_lock);
@@ -777,7 +804,7 @@ void bch_cached_dev_writeback_init(struct cached_dev *dc)
 	bch_keybuf_init(&dc->writeback_keys);
 
 	dc->writeback_metadata		= true;
-	dc->writeback_running		= true;
+	dc->writeback_running		= false;
 	dc->writeback_percent		= 10;
 	dc->writeback_delay		= 30;
 	atomic_long_set(&dc->writeback_rate.rate, 1024);
@@ -805,6 +832,7 @@ int bch_cached_dev_writeback_start(struct cached_dev *dc)
 		cached_dev_put(dc);
 		return PTR_ERR(dc->writeback_thread);
 	}
+	dc->writeback_running = true;
 
 	WARN_ON(test_and_set_bit(BCACHE_DEV_WB_RUNNING, &dc->disk.flags));
 	schedule_delayed_work(&dc->writeback_rate_update,
diff --git a/drivers/md/bcache/writeback.h b/drivers/md/bcache/writeback.h
index d2b9fdbc8994..6a743d3bb338 100644
--- a/drivers/md/bcache/writeback.h
+++ b/drivers/md/bcache/writeback.h
@@ -5,12 +5,17 @@
 #define CUTOFF_WRITEBACK	40
 #define CUTOFF_WRITEBACK_SYNC	70
 
+#define CUTOFF_WRITEBACK_MAX		70
+#define CUTOFF_WRITEBACK_SYNC_MAX	90
+
 #define MAX_WRITEBACKS_IN_PASS  5
 #define MAX_WRITESIZE_IN_PASS   5000	/* *512b */
 
 #define WRITEBACK_RATE_UPDATE_SECS_MAX		60
 #define WRITEBACK_RATE_UPDATE_SECS_DEFAULT	5
 
+#define BCH_AUTO_GC_DIRTY_THRESHOLD	50
+
 /*
  * 14 (16384ths) is chosen here as something that each backing device
  * should be a reasonable fraction of the share, and not to blow up
@@ -53,6 +58,9 @@ static inline bool bcache_dev_stripe_dirty(struct cached_dev *dc,
 	}
 }
 
+extern unsigned int bch_cutoff_writeback;
+extern unsigned int bch_cutoff_writeback_sync;
+
 static inline bool should_writeback(struct cached_dev *dc, struct bio *bio,
 				    unsigned int cache_mode, bool would_skip)
 {
@@ -60,7 +68,7 @@ static inline bool should_writeback(struct cached_dev *dc, struct bio *bio,
 
 	if (cache_mode != CACHE_MODE_WRITEBACK ||
 	    test_bit(BCACHE_DEV_DETACHING, &dc->disk.flags) ||
-	    in_use > CUTOFF_WRITEBACK_SYNC)
+	    in_use > bch_cutoff_writeback_sync)
 		return false;
 
 	if (dc->partial_stripes_expensive &&
@@ -73,7 +81,7 @@ static inline bool should_writeback(struct cached_dev *dc, struct bio *bio,
 
 	return (op_is_sync(bio->bi_opf) ||
 		bio->bi_opf & (REQ_META|REQ_PRIO) ||
-		in_use <= CUTOFF_WRITEBACK);
+		in_use <= bch_cutoff_writeback);
 }
 
 static inline void bch_writeback_queue(struct cached_dev *dc)
diff --git a/drivers/md/dm-core.h b/drivers/md/dm-core.h
index 224d44503a06..95c6d86ab5e8 100644
--- a/drivers/md/dm-core.h
+++ b/drivers/md/dm-core.h
@@ -65,7 +65,6 @@ struct mapped_device {
 	 */
 	struct work_struct work;
 	wait_queue_head_t wait;
-	atomic_t pending[2];
 	spinlock_t deferred_lock;
 	struct bio_list deferred;
 
@@ -107,9 +106,6 @@ struct mapped_device {
 
 	struct block_device *bdev;
 
-	/* zero-length flush that will be cloned and submitted to targets */
-	struct bio flush_bio;
-
 	struct dm_stats stats;
 
 	/* for blk-mq request-based DM support */
@@ -119,7 +115,6 @@ struct mapped_device {
 	struct srcu_struct io_barrier;
 };
 
-int md_in_flight(struct mapped_device *md);
 void disable_write_same(struct mapped_device *md);
 void disable_write_zeroes(struct mapped_device *md);
 
diff --git a/drivers/md/dm-rq.c b/drivers/md/dm-rq.c
index 7cd36e4d1310..4e06be4f0a62 100644
--- a/drivers/md/dm-rq.c
+++ b/drivers/md/dm-rq.c
@@ -43,7 +43,7 @@ static unsigned dm_get_blk_mq_queue_depth(void)
 
 int dm_request_based(struct mapped_device *md)
 {
-	return queue_is_rq_based(md->queue);
+	return queue_is_mq(md->queue);
 }
 
 void dm_start_queue(struct request_queue *q)
@@ -130,10 +130,8 @@ static void rq_end_stats(struct mapped_device *md, struct request *orig)
  */
 static void rq_completed(struct mapped_device *md, int rw, bool run_queue)
 {
-	atomic_dec(&md->pending[rw]);
-
 	/* nudge anyone waiting on suspend queue */
-	if (!md_in_flight(md))
+	if (unlikely(waitqueue_active(&md->wait)))
 		wake_up(&md->wait);
 
 	/*
@@ -436,7 +434,6 @@ ssize_t dm_attr_rq_based_seq_io_merge_deadline_store(struct mapped_device *md,
 static void dm_start_request(struct mapped_device *md, struct request *orig)
 {
 	blk_mq_start_request(orig);
-	atomic_inc(&md->pending[rq_data_dir(orig)]);
 
 	if (unlikely(dm_stats_used(&md->stats))) {
 		struct dm_rq_target_io *tio = tio_from_request(orig);
diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index 9038c302d5c2..844f7d0f2ef8 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -919,12 +919,12 @@ static int device_is_rq_based(struct dm_target *ti, struct dm_dev *dev,
 	struct request_queue *q = bdev_get_queue(dev->bdev);
 	struct verify_rq_based_data *v = data;
 
-	if (q->mq_ops)
+	if (queue_is_mq(q))
 		v->mq_count++;
 	else
 		v->sq_count++;
 
-	return queue_is_rq_based(q);
+	return queue_is_mq(q);
 }
 
 static int dm_table_determine_type(struct dm_table *t)
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index 63a7c416b224..a4a06982ed91 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -646,26 +646,38 @@ static void free_tio(struct dm_target_io *tio)
 	bio_put(&tio->clone);
 }
 
-int md_in_flight(struct mapped_device *md)
+static bool md_in_flight_bios(struct mapped_device *md)
 {
-	return atomic_read(&md->pending[READ]) +
-	       atomic_read(&md->pending[WRITE]);
+	int cpu;
+	struct hd_struct *part = &dm_disk(md)->part0;
+	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);
+	}
+
+	return sum != 0;
+}
+
+static bool md_in_flight(struct mapped_device *md)
+{
+	if (queue_is_mq(md->queue))
+		return blk_mq_queue_inflight(md->queue);
+	else
+		return md_in_flight_bios(md);
 }
 
 static void start_io_acct(struct dm_io *io)
 {
 	struct mapped_device *md = io->md;
 	struct bio *bio = io->orig_bio;
-	int rw = bio_data_dir(bio);
 
 	io->start_time = jiffies;
 
 	generic_start_io_acct(md->queue, bio_op(bio), bio_sectors(bio),
 			      &dm_disk(md)->part0);
 
-	atomic_set(&dm_disk(md)->part0.in_flight[rw],
-		   atomic_inc_return(&md->pending[rw]));
-
 	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),
@@ -677,8 +689,6 @@ static void end_io_acct(struct dm_io *io)
 	struct mapped_device *md = io->md;
 	struct bio *bio = io->orig_bio;
 	unsigned long duration = jiffies - io->start_time;
-	int pending;
-	int rw = bio_data_dir(bio);
 
 	generic_end_io_acct(md->queue, bio_op(bio), &dm_disk(md)->part0,
 			    io->start_time);
@@ -688,16 +698,8 @@ static void end_io_acct(struct dm_io *io)
 				    bio->bi_iter.bi_sector, bio_sectors(bio),
 				    true, duration, &io->stats_aux);
 
-	/*
-	 * After this is decremented the bio must not be touched if it is
-	 * a flush.
-	 */
-	pending = atomic_dec_return(&md->pending[rw]);
-	atomic_set(&dm_disk(md)->part0.in_flight[rw], pending);
-	pending += atomic_read(&md->pending[rw^0x1]);
-
 	/* nudge anyone waiting on suspend queue */
-	if (!pending)
+	if (unlikely(waitqueue_active(&md->wait)))
 		wake_up(&md->wait);
 }
 
@@ -1417,10 +1419,21 @@ static int __send_empty_flush(struct clone_info *ci)
 	unsigned target_nr = 0;
 	struct dm_target *ti;
 
+	/*
+	 * Empty flush uses a statically initialized bio, as the base for
+	 * cloning.  However, blkg association requires that a bdev is
+	 * associated with a gendisk, which doesn't happen until the bdev is
+	 * opened.  So, blkg association is done at issue time of the flush
+	 * rather than when the device is created in alloc_dev().
+	 */
+	bio_set_dev(ci->bio, ci->io->md->bdev);
+
 	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_disassociate_blkg(ci->bio);
+
 	return 0;
 }
 
@@ -1598,7 +1611,16 @@ static blk_qc_t __split_and_process_bio(struct mapped_device *md,
 	init_clone_info(&ci, md, map, bio);
 
 	if (bio->bi_opf & REQ_PREFLUSH) {
-		ci.bio = &ci.io->md->flush_bio;
+		struct bio flush_bio;
+
+		/*
+		 * Use an on-stack bio for this, it's safe since we don't
+		 * need to reference it after submit. It's just used as
+		 * the basis for the clone(s).
+		 */
+		bio_init(&flush_bio, NULL, 0);
+		flush_bio.bi_opf = REQ_OP_WRITE | REQ_PREFLUSH | REQ_SYNC;
+		ci.bio = &flush_bio;
 		ci.sector_count = 0;
 		error = __send_empty_flush(&ci);
 		/* dec_pending submits any data associated with flush */
@@ -1654,7 +1676,16 @@ static blk_qc_t __process_bio(struct mapped_device *md,
 	init_clone_info(&ci, md, map, bio);
 
 	if (bio->bi_opf & REQ_PREFLUSH) {
-		ci.bio = &ci.io->md->flush_bio;
+		struct bio flush_bio;
+
+		/*
+		 * Use an on-stack bio for this, it's safe since we don't
+		 * need to reference it after submit. It's just used as
+		 * the basis for the clone(s).
+		 */
+		bio_init(&flush_bio, NULL, 0);
+		flush_bio.bi_opf = REQ_OP_WRITE | REQ_PREFLUSH | REQ_SYNC;
+		ci.bio = &flush_bio;
 		ci.sector_count = 0;
 		error = __send_empty_flush(&ci);
 		/* dec_pending submits any data associated with flush */
@@ -1898,7 +1929,7 @@ 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_node(GFP_KERNEL, numa_node_id, NULL);
+	md->queue = blk_alloc_queue_node(GFP_KERNEL, numa_node_id);
 	if (!md->queue)
 		goto bad;
 	md->queue->queuedata = md;
@@ -1908,8 +1939,6 @@ static struct mapped_device *alloc_dev(int minor)
 	if (!md->disk)
 		goto bad;
 
-	atomic_set(&md->pending[0], 0);
-	atomic_set(&md->pending[1], 0);
 	init_waitqueue_head(&md->wait);
 	INIT_WORK(&md->work, dm_wq_work);
 	init_waitqueue_head(&md->eventq);
@@ -1940,10 +1969,6 @@ static struct mapped_device *alloc_dev(int minor)
 	if (!md->bdev)
 		goto bad;
 
-	bio_init(&md->flush_bio, NULL, 0);
-	bio_set_dev(&md->flush_bio, md->bdev);
-	md->flush_bio.bi_opf = REQ_OP_WRITE | REQ_PREFLUSH | REQ_SYNC;
-
 	dm_stats_init(&md->stats);
 
 	/* Populate the mapping, nobody knows we exist yet */
diff --git a/drivers/md/md.c b/drivers/md/md.c
index fc488cb30a94..9a0a1e0934d5 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -334,7 +334,6 @@ static blk_qc_t md_make_request(struct request_queue *q, struct bio *bio)
 	const int sgrp = op_stat_group(bio_op(bio));
 	struct mddev *mddev = q->queuedata;
 	unsigned int sectors;
-	int cpu;
 
 	blk_queue_split(q, &bio);
 
@@ -359,9 +358,9 @@ static blk_qc_t md_make_request(struct request_queue *q, struct bio *bio)
 
 	md_handle_request(mddev, bio);
 
-	cpu = part_stat_lock();
-	part_stat_inc(cpu, &mddev->gendisk->part0, ios[sgrp]);
-	part_stat_add(cpu, &mddev->gendisk->part0, sectors[sgrp], sectors);
+	part_stat_lock();
+	part_stat_inc(&mddev->gendisk->part0, ios[sgrp]);
+	part_stat_add(&mddev->gendisk->part0, sectors[sgrp], sectors);
 	part_stat_unlock();
 
 	return BLK_QC_T_NONE;
diff --git a/drivers/md/raid0.c b/drivers/md/raid0.c
index ac1cffd2a09b..f3fb5bb8c82a 100644
--- a/drivers/md/raid0.c
+++ b/drivers/md/raid0.c
@@ -542,7 +542,7 @@ static void raid0_handle_discard(struct mddev *mddev, struct bio *bio)
 		    !discard_bio)
 			continue;
 		bio_chain(discard_bio, bio);
-		bio_clone_blkcg_association(discard_bio, bio);
+		bio_clone_blkg_association(discard_bio, bio);
 		if (mddev->gendisk)
 			trace_block_bio_remap(bdev_get_queue(rdev->bdev),
 				discard_bio, disk_devt(mddev->gendisk),