Merge remote-tracking branch 'jens/for-5.10/block' into dm-5.10

DM depends on these block 5.10 commits:

22ada802ede8 block: use lcm_not_zero() when stacking chunk_sectors
07d098e6bbad block: allow 'chunk_sectors' to be non-power-of-2
021a24460dc2 block: add QUEUE_FLAG_NOWAIT
6abc49468eea dm: add support for REQ_NOWAIT and enable it for linear target

Signed-off-by: Mike Snitzer <snitzer@redhat.com>

28 files changed, 2194 insertions, 1250 deletions

diff --git a/block/Kconfig b/block/Kconfig
index bbad5e8bbffe..a2297edfdde8 100644
--- a/block/Kconfig
+++ b/block/Kconfig
@@ -161,8 +161,6 @@ config BLK_WBT_MQ
 	depends on BLK_WBT
 	help
 	Enable writeback throttling by default on multiqueue devices.
-	Multiqueue currently doesn't have support for IO scheduling,
-	enabling this option is recommended.
 
 config BLK_DEBUG_FS
 	bool "Block layer debugging information in debugfs"
diff --git a/block/bfq-iosched.c b/block/bfq-iosched.c
index fa98470df3f0..9e81d1052091 100644
--- a/block/bfq-iosched.c
+++ b/block/bfq-iosched.c
@@ -4640,6 +4640,9 @@ static bool bfq_has_work(struct blk_mq_hw_ctx *hctx)
 {
 	struct bfq_data *bfqd = hctx->queue->elevator->elevator_data;
 
+	if (!atomic_read(&hctx->elevator_queued))
+		return false;
+
 	/*
 	 * Avoiding lock: a race on bfqd->busy_queues should cause at
 	 * most a call to dispatch for nothing
@@ -5554,6 +5557,7 @@ static void bfq_insert_requests(struct blk_mq_hw_ctx *hctx,
 		rq = list_first_entry(list, struct request, queuelist);
 		list_del_init(&rq->queuelist);
 		bfq_insert_request(hctx, rq, at_head);
+		atomic_inc(&hctx->elevator_queued);
 	}
 }
 
@@ -5921,6 +5925,7 @@ static void bfq_finish_requeue_request(struct request *rq)
 
 		bfq_completed_request(bfqq, bfqd);
 		bfq_finish_requeue_request_body(bfqq);
+		atomic_dec(&rq->mq_hctx->elevator_queued);
 
 		spin_unlock_irqrestore(&bfqd->lock, flags);
 	} else {
@@ -6360,8 +6365,8 @@ static void bfq_depth_updated(struct blk_mq_hw_ctx *hctx)
 	struct blk_mq_tags *tags = hctx->sched_tags;
 	unsigned int min_shallow;
 
-	min_shallow = bfq_update_depths(bfqd, &tags->bitmap_tags);
-	sbitmap_queue_min_shallow_depth(&tags->bitmap_tags, min_shallow);
+	min_shallow = bfq_update_depths(bfqd, tags->bitmap_tags);
+	sbitmap_queue_min_shallow_depth(tags->bitmap_tags, min_shallow);
 }
 
 static int bfq_init_hctx(struct blk_mq_hw_ctx *hctx, unsigned int index)
diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index c195365c9817..f9b55614d67d 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -119,6 +119,8 @@ static void blkg_async_bio_workfn(struct work_struct *work)
 					     async_bio_work);
 	struct bio_list bios = BIO_EMPTY_LIST;
 	struct bio *bio;
+	struct blk_plug plug;
+	bool need_plug = false;
 
 	/* as long as there are pending bios, @blkg can't go away */
 	spin_lock_bh(&blkg->async_bio_lock);
@@ -126,8 +128,15 @@ static void blkg_async_bio_workfn(struct work_struct *work)
 	bio_list_init(&blkg->async_bios);
 	spin_unlock_bh(&blkg->async_bio_lock);
 
+	/* start plug only when bio_list contains at least 2 bios */
+	if (bios.head && bios.head->bi_next) {
+		need_plug = true;
+		blk_start_plug(&plug);
+	}
 	while ((bio = bio_list_pop(&bios)))
 		submit_bio(bio);
+	if (need_plug)
+		blk_finish_plug(&plug);
 }
 
 /**
@@ -1613,16 +1622,24 @@ static void blkcg_scale_delay(struct blkcg_gq *blkg, u64 now)
 static void blkcg_maybe_throttle_blkg(struct blkcg_gq *blkg, bool use_memdelay)
 {
 	unsigned long pflags;
+	bool clamp;
 	u64 now = ktime_to_ns(ktime_get());
 	u64 exp;
 	u64 delay_nsec = 0;
 	int tok;
 
 	while (blkg->parent) {
-		if (atomic_read(&blkg->use_delay)) {
+		int use_delay = atomic_read(&blkg->use_delay);
+
+		if (use_delay) {
+			u64 this_delay;
+
 			blkcg_scale_delay(blkg, now);
-			delay_nsec = max_t(u64, delay_nsec,
-					   atomic64_read(&blkg->delay_nsec));
+			this_delay = atomic64_read(&blkg->delay_nsec);
+			if (this_delay > delay_nsec) {
+				delay_nsec = this_delay;
+				clamp = use_delay > 0;
+			}
 		}
 		blkg = blkg->parent;
 	}
@@ -1634,10 +1651,13 @@ static void blkcg_maybe_throttle_blkg(struct blkcg_gq *blkg, bool use_memdelay)
 	 * Let's not sleep for all eternity if we've amassed a huge delay.
 	 * Swapping or metadata IO can accumulate 10's of seconds worth of
 	 * delay, and we want userspace to be able to do _something_ so cap the
-	 * delays at 1 second.  If there's 10's of seconds worth of delay then
-	 * the tasks will be delayed for 1 second for every syscall.
+	 * delays at 0.25s. If there's 10's of seconds worth of delay then the
+	 * tasks will be delayed for 0.25 second for every syscall. If
+	 * blkcg_set_delay() was used as indicated by negative use_delay, the
+	 * caller is responsible for regulating the range.
 	 */
-	delay_nsec = min_t(u64, delay_nsec, 250 * NSEC_PER_MSEC);
+	if (clamp)
+		delay_nsec = min_t(u64, delay_nsec, 250 * NSEC_PER_MSEC);
 
 	if (use_memdelay)
 		psi_memstall_enter(&pflags);
diff --git a/block/blk-core.c b/block/blk-core.c
index 10c08ac50697..4884f1e7451b 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -116,8 +116,8 @@ void blk_rq_init(struct request_queue *q, struct request *rq)
 	rq->__sector = (sector_t) -1;
 	INIT_HLIST_NODE(&rq->hash);
 	RB_CLEAR_NODE(&rq->rb_node);
-	rq->tag = -1;
-	rq->internal_tag = -1;
+	rq->tag = BLK_MQ_NO_TAG;
+	rq->internal_tag = BLK_MQ_NO_TAG;
 	rq->start_time_ns = ktime_get_ns();
 	rq->part = NULL;
 	refcount_set(&rq->ref, 1);
@@ -538,11 +538,10 @@ struct request_queue *blk_alloc_queue(int node_id)
 	if (!q->stats)
 		goto fail_stats;
 
-	q->backing_dev_info->ra_pages = VM_READAHEAD_PAGES;
-	q->backing_dev_info->io_pages = VM_READAHEAD_PAGES;
-	q->backing_dev_info->capabilities = BDI_CAP_CGROUP_WRITEBACK;
 	q->node = node_id;
 
+	atomic_set(&q->nr_active_requests_shared_sbitmap, 0);
+
 	timer_setup(&q->backing_dev_info->laptop_mode_wb_timer,
 		    laptop_mode_timer_fn, 0);
 	timer_setup(&q->timeout, blk_rq_timed_out_timer, 0);
@@ -643,162 +642,6 @@ void blk_put_request(struct request *req)
 }
 EXPORT_SYMBOL(blk_put_request);
 
-static void blk_account_io_merge_bio(struct request *req)
-{
-	if (!blk_do_io_stat(req))
-		return;
-
-	part_stat_lock();
-	part_stat_inc(req->part, merges[op_stat_group(req_op(req))]);
-	part_stat_unlock();
-}
-
-bool bio_attempt_back_merge(struct request *req, struct bio *bio,
-		unsigned int nr_segs)
-{
-	const int ff = bio->bi_opf & REQ_FAILFAST_MASK;
-
-	if (!ll_back_merge_fn(req, bio, nr_segs))
-		return false;
-
-	trace_block_bio_backmerge(req->q, req, bio);
-	rq_qos_merge(req->q, req, bio);
-
-	if ((req->cmd_flags & REQ_FAILFAST_MASK) != ff)
-		blk_rq_set_mixed_merge(req);
-
-	req->biotail->bi_next = bio;
-	req->biotail = bio;
-	req->__data_len += bio->bi_iter.bi_size;
-
-	bio_crypt_free_ctx(bio);
-
-	blk_account_io_merge_bio(req);
-	return true;
-}
-
-bool bio_attempt_front_merge(struct request *req, struct bio *bio,
-		unsigned int nr_segs)
-{
-	const int ff = bio->bi_opf & REQ_FAILFAST_MASK;
-
-	if (!ll_front_merge_fn(req, bio, nr_segs))
-		return false;
-
-	trace_block_bio_frontmerge(req->q, req, bio);
-	rq_qos_merge(req->q, req, bio);
-
-	if ((req->cmd_flags & REQ_FAILFAST_MASK) != ff)
-		blk_rq_set_mixed_merge(req);
-
-	bio->bi_next = req->bio;
-	req->bio = bio;
-
-	req->__sector = bio->bi_iter.bi_sector;
-	req->__data_len += bio->bi_iter.bi_size;
-
-	bio_crypt_do_front_merge(req, bio);
-
-	blk_account_io_merge_bio(req);
-	return true;
-}
-
-bool bio_attempt_discard_merge(struct request_queue *q, struct request *req,
-		struct bio *bio)
-{
-	unsigned short segments = blk_rq_nr_discard_segments(req);
-
-	if (segments >= queue_max_discard_segments(q))
-		goto no_merge;
-	if (blk_rq_sectors(req) + bio_sectors(bio) >
-	    blk_rq_get_max_sectors(req, blk_rq_pos(req)))
-		goto no_merge;
-
-	rq_qos_merge(q, req, bio);
-
-	req->biotail->bi_next = bio;
-	req->biotail = bio;
-	req->__data_len += bio->bi_iter.bi_size;
-	req->nr_phys_segments = segments + 1;
-
-	blk_account_io_merge_bio(req);
-	return true;
-no_merge:
-	req_set_nomerge(q, req);
-	return false;
-}
-
-/**
- * blk_attempt_plug_merge - try to merge with %current's plugged list
- * @q: request_queue new bio is being queued at
- * @bio: new bio being queued
- * @nr_segs: number of segments in @bio
- * @same_queue_rq: pointer to &struct request that gets filled in when
- * another request associated with @q is found on the plug list
- * (optional, may be %NULL)
- *
- * Determine whether @bio being queued on @q can be merged with a request
- * on %current's plugged list.  Returns %true if merge was successful,
- * otherwise %false.
- *
- * Plugging coalesces IOs from the same issuer for the same purpose without
- * going through @q->queue_lock.  As such it's more of an issuing mechanism
- * than scheduling, and the request, while may have elvpriv data, is not
- * added on the elevator at this point.  In addition, we don't have
- * reliable access to the elevator outside queue lock.  Only check basic
- * merging parameters without querying the elevator.
- *
- * Caller must ensure !blk_queue_nomerges(q) beforehand.
- */
-bool blk_attempt_plug_merge(struct request_queue *q, struct bio *bio,
-		unsigned int nr_segs, struct request **same_queue_rq)
-{
-	struct blk_plug *plug;
-	struct request *rq;
-	struct list_head *plug_list;
-
-	plug = blk_mq_plug(q, bio);
-	if (!plug)
-		return false;
-
-	plug_list = &plug->mq_list;
-
-	list_for_each_entry_reverse(rq, plug_list, queuelist) {
-		bool merged = false;
-
-		if (rq->q == q && same_queue_rq) {
-			/*
-			 * Only blk-mq multiple hardware queues case checks the
-			 * rq in the same queue, there should be only one such
-			 * rq in a queue
-			 **/
-			*same_queue_rq = rq;
-		}
-
-		if (rq->q != q || !blk_rq_merge_ok(rq, bio))
-			continue;
-
-		switch (blk_try_merge(rq, bio)) {
-		case ELEVATOR_BACK_MERGE:
-			merged = bio_attempt_back_merge(rq, bio, nr_segs);
-			break;
-		case ELEVATOR_FRONT_MERGE:
-			merged = bio_attempt_front_merge(rq, bio, nr_segs);
-			break;
-		case ELEVATOR_DISCARD_MERGE:
-			merged = bio_attempt_discard_merge(q, rq, bio);
-			break;
-		default:
-			break;
-		}
-
-		if (merged)
-			return true;
-	}
-
-	return false;
-}
-
 static void handle_bad_sector(struct bio *bio, sector_t maxsector)
 {
 	char b[BDEVNAME_SIZE];
@@ -971,9 +814,9 @@ static noinline_for_stack bool submit_bio_checks(struct bio *bio)
 
 	/*
 	 * For a REQ_NOWAIT based request, return -EOPNOTSUPP
-	 * if queue is not a request based queue.
+	 * if queue does not support NOWAIT.
 	 */
-	if ((bio->bi_opf & REQ_NOWAIT) && !queue_is_mq(q))
+	if ((bio->bi_opf & REQ_NOWAIT) && !blk_queue_nowait(q))
 		goto not_supported;
 
 	if (should_fail_bio(bio))
@@ -1301,14 +1144,28 @@ EXPORT_SYMBOL(submit_bio);
  *    limits when retrying requests on other queues. Those requests need
  *    to be checked against the new queue limits again during dispatch.
  */
-static int blk_cloned_rq_check_limits(struct request_queue *q,
+static blk_status_t blk_cloned_rq_check_limits(struct request_queue *q,
 				      struct request *rq)
 {
-	if (blk_rq_sectors(rq) > blk_queue_get_max_sectors(q, req_op(rq))) {
+	unsigned int max_sectors = blk_queue_get_max_sectors(q, req_op(rq));
+
+	if (blk_rq_sectors(rq) > max_sectors) {
+		/*
+		 * SCSI device does not have a good way to return if
+		 * Write Same/Zero is actually supported. If a device rejects
+		 * a non-read/write command (discard, write same,etc.) the
+		 * low-level device driver will set the relevant queue limit to
+		 * 0 to prevent blk-lib from issuing more of the offending
+		 * operations. Commands queued prior to the queue limit being
+		 * reset need to be completed with BLK_STS_NOTSUPP to avoid I/O
+		 * errors being propagated to upper layers.
+		 */
+		if (max_sectors == 0)
+			return BLK_STS_NOTSUPP;
+
 		printk(KERN_ERR "%s: over max size limit. (%u > %u)\n",
-			__func__, blk_rq_sectors(rq),
-			blk_queue_get_max_sectors(q, req_op(rq)));
-		return -EIO;
+			__func__, blk_rq_sectors(rq), max_sectors);
+		return BLK_STS_IOERR;
 	}
 
 	/*
@@ -1321,10 +1178,10 @@ static int blk_cloned_rq_check_limits(struct request_queue *q,
 	if (rq->nr_phys_segments > queue_max_segments(q)) {
 		printk(KERN_ERR "%s: over max segments limit. (%hu > %hu)\n",
 			__func__, rq->nr_phys_segments, queue_max_segments(q));
-		return -EIO;
+		return BLK_STS_IOERR;
 	}
 
-	return 0;
+	return BLK_STS_OK;
 }
 
 /**
@@ -1334,8 +1191,11 @@ static int blk_cloned_rq_check_limits(struct request_queue *q,
  */
 blk_status_t blk_insert_cloned_request(struct request_queue *q, struct request *rq)
 {
-	if (blk_cloned_rq_check_limits(q, rq))
-		return BLK_STS_IOERR;
+	blk_status_t ret;
+
+	ret = blk_cloned_rq_check_limits(q, rq);
+	if (ret != BLK_STS_OK)
+		return ret;
 
 	if (rq->rq_disk &&
 	    should_fail_request(&rq->rq_disk->part0, blk_rq_bytes(rq)))
@@ -1461,10 +1321,9 @@ void blk_account_io_start(struct request *rq)
 	part_stat_unlock();
 }
 
-unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
-		unsigned int op)
+static unsigned long __part_start_io_acct(struct hd_struct *part,
+					  unsigned int sectors, unsigned int op)
 {
-	struct hd_struct *part = &disk->part0;
 	const int sgrp = op_stat_group(op);
 	unsigned long now = READ_ONCE(jiffies);
 
@@ -1477,12 +1336,26 @@ unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
 
 	return now;
 }
+
+unsigned long part_start_io_acct(struct gendisk *disk, struct hd_struct **part,
+				 struct bio *bio)
+{
+	*part = disk_map_sector_rcu(disk, bio->bi_iter.bi_sector);
+
+	return __part_start_io_acct(*part, bio_sectors(bio), bio_op(bio));
+}
+EXPORT_SYMBOL_GPL(part_start_io_acct);
+
+unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
+				 unsigned int op)
+{
+	return __part_start_io_acct(&disk->part0, sectors, op);
+}
 EXPORT_SYMBOL(disk_start_io_acct);
 
-void disk_end_io_acct(struct gendisk *disk, unsigned int op,
-		unsigned long start_time)
+static void __part_end_io_acct(struct hd_struct *part, unsigned int op,
+			       unsigned long start_time)
 {
-	struct hd_struct *part = &disk->part0;
 	const int sgrp = op_stat_group(op);
 	unsigned long now = READ_ONCE(jiffies);
 	unsigned long duration = now - start_time;
@@ -1493,6 +1366,20 @@ void disk_end_io_acct(struct gendisk *disk, unsigned int op,
 	part_stat_local_dec(part, in_flight[op_is_write(op)]);
 	part_stat_unlock();
 }
+
+void part_end_io_acct(struct hd_struct *part, struct bio *bio,
+		      unsigned long start_time)
+{
+	__part_end_io_acct(part, bio_op(bio), start_time);
+	hd_struct_put(part);
+}
+EXPORT_SYMBOL_GPL(part_end_io_acct);
+
+void disk_end_io_acct(struct gendisk *disk, unsigned int op,
+		      unsigned long start_time)
+{
+	__part_end_io_acct(&disk->part0, op, start_time);
+}
 EXPORT_SYMBOL(disk_end_io_acct);
 
 /*
diff --git a/block/blk-integrity.c b/block/blk-integrity.c
index c03705cbb9c9..2b36a8f9b813 100644
--- a/block/blk-integrity.c
+++ b/block/blk-integrity.c
@@ -408,7 +408,7 @@ void blk_integrity_register(struct gendisk *disk, struct blk_integrity *template
 	bi->tuple_size = template->tuple_size;
 	bi->tag_size = template->tag_size;
 
-	disk->queue->backing_dev_info->capabilities |= BDI_CAP_STABLE_WRITES;
+	blk_queue_flag_set(QUEUE_FLAG_STABLE_WRITES, disk->queue);
 
 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
 	if (disk->queue->ksm) {
@@ -428,7 +428,7 @@ EXPORT_SYMBOL(blk_integrity_register);
  */
 void blk_integrity_unregister(struct gendisk *disk)
 {
-	disk->queue->backing_dev_info->capabilities &= ~BDI_CAP_STABLE_WRITES;
+	blk_queue_flag_clear(QUEUE_FLAG_STABLE_WRITES, disk->queue);
 	memset(&disk->queue->integrity, 0, sizeof(struct blk_integrity));
 }
 EXPORT_SYMBOL(blk_integrity_unregister);
diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index d37b55db2409..b82649c1440b 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -68,7 +68,7 @@
  * gets 300/(100+300) or 75% share, and A0 and A1 equally splits the rest,
  * 12.5% each.  The distribution mechanism only cares about these flattened
  * shares.  They're called hweights (hierarchical weights) and always add
- * upto 1 (HWEIGHT_WHOLE).
+ * upto 1 (WEIGHT_ONE).
  *
  * A given cgroup's vtime runs slower in inverse proportion to its hweight.
  * For example, with 12.5% weight, A0's time runs 8 times slower (100/12.5)
@@ -179,6 +179,8 @@
 #include <linux/parser.h>
 #include <linux/sched/signal.h>
 #include <linux/blk-cgroup.h>
+#include <asm/local.h>
+#include <asm/local64.h>
 #include "blk-rq-qos.h"
 #include "blk-stat.h"
 #include "blk-wbt.h"
@@ -215,36 +217,21 @@ enum {
 	MAX_PERIOD		= USEC_PER_SEC,
 
 	/*
-	 * A cgroup's vtime can run 50% behind the device vtime, which
+	 * iocg->vtime is targeted at 50% behind the device vtime, which
 	 * serves as its IO credit buffer.  Surplus weight adjustment is
 	 * immediately canceled if the vtime margin runs below 10%.
 	 */
-	MARGIN_PCT		= 50,
-	INUSE_MARGIN_PCT	= 10,
+	MARGIN_MIN_PCT		= 10,
+	MARGIN_LOW_PCT		= 20,
+	MARGIN_TARGET_PCT	= 50,
 
-	/* Have some play in waitq timer operations */
-	WAITQ_TIMER_MARGIN_PCT	= 5,
+	INUSE_ADJ_STEP_PCT	= 25,
 
-	/*
-	 * vtime can wrap well within a reasonable uptime when vrate is
-	 * consistently raised.  Don't trust recorded cgroup vtime if the
-	 * period counter indicates that it's older than 5mins.
-	 */
-	VTIME_VALID_DUR		= 300 * USEC_PER_SEC,
-
-	/*
-	 * Remember the past three non-zero usages and use the max for
-	 * surplus calculation.  Three slots guarantee that we remember one
-	 * full period usage from the last active stretch even after
-	 * partial deactivation and re-activation periods.  Don't start
-	 * giving away weight before collecting two data points to prevent
-	 * hweight adjustments based on one partial activation period.
-	 */
-	NR_USAGE_SLOTS		= 3,
-	MIN_VALID_USAGES	= 2,
+	/* Have some play in timer operations */
+	TIMER_SLACK_PCT		= 1,
 
 	/* 1/64k is granular enough and can easily be handled w/ u32 */
-	HWEIGHT_WHOLE		= 1 << 16,
+	WEIGHT_ONE		= 1 << 16,
 
 	/*
 	 * As vtime is used to calculate the cost of each IO, it needs to
@@ -275,16 +262,37 @@ enum {
 	/* unbusy hysterisis */
 	UNBUSY_THR_PCT		= 75,
 
-	/* don't let cmds which take a very long time pin lagging for too long */
-	MAX_LAGGING_PERIODS	= 10,
-
 	/*
-	 * If usage% * 1.25 + 2% is lower than hweight% by more than 3%,
-	 * donate the surplus.
+	 * The effect of delay is indirect and non-linear and a huge amount of
+	 * future debt can accumulate abruptly while unthrottled. Linearly scale
+	 * up delay as debt is going up and then let it decay exponentially.
+	 * This gives us quick ramp ups while delay is accumulating and long
+	 * tails which can help reducing the frequency of debt explosions on
+	 * unthrottle. The parameters are experimentally determined.
+	 *
+	 * The delay mechanism provides adequate protection and behavior in many
+	 * cases. However, this is far from ideal and falls shorts on both
+	 * fronts. The debtors are often throttled too harshly costing a
+	 * significant level of fairness and possibly total work while the
+	 * protection against their impacts on the system can be choppy and
+	 * unreliable.
+	 *
+	 * The shortcoming primarily stems from the fact that, unlike for page
+	 * cache, the kernel doesn't have well-defined back-pressure propagation
+	 * mechanism and policies for anonymous memory. Fully addressing this
+	 * issue will likely require substantial improvements in the area.
 	 */
-	SURPLUS_SCALE_PCT	= 125,			/* * 125% */
-	SURPLUS_SCALE_ABS	= HWEIGHT_WHOLE / 50,	/* + 2% */
-	SURPLUS_MIN_ADJ_DELTA	= HWEIGHT_WHOLE / 33,	/* 3% */
+	MIN_DELAY_THR_PCT	= 500,
+	MAX_DELAY_THR_PCT	= 25000,
+	MIN_DELAY		= 250,
+	MAX_DELAY		= 250 * USEC_PER_MSEC,
+
+	/* halve debts if avg usage over 100ms is under 50% */
+	DFGV_USAGE_PCT		= 50,
+	DFGV_PERIOD		= 100 * USEC_PER_MSEC,
+
+	/* don't let cmds which take a very long time pin lagging for too long */
+	MAX_LAGGING_PERIODS	= 10,
 
 	/* switch iff the conditions are met for longer than this */
 	AUTOP_CYCLE_NSEC	= 10LLU * NSEC_PER_SEC,
@@ -372,9 +380,15 @@ struct ioc_params {
 	u32				too_slow_vrate_pct;
 };
 
+struct ioc_margins {
+	s64				min;
+	s64				low;
+	s64				target;
+};
+
 struct ioc_missed {
-	u32				nr_met;
-	u32				nr_missed;
+	local_t				nr_met;
+	local_t				nr_missed;
 	u32				last_met;
 	u32				last_missed;
 };
@@ -382,7 +396,7 @@ struct ioc_missed {
 struct ioc_pcpu_stat {
 	struct ioc_missed		missed[2];
 
-	u64				rq_wait_ns;
+	local64_t			rq_wait_ns;
 	u64				last_rq_wait_ns;
 };
 
@@ -393,8 +407,9 @@ struct ioc {
 	bool				enabled;
 
 	struct ioc_params		params;
+	struct ioc_margins		margins;
 	u32				period_us;
-	u32				margin_us;
+	u32				timer_slack_ns;
 	u64				vrate_min;
 	u64				vrate_max;
 
@@ -405,18 +420,24 @@ struct ioc {
 
 	enum ioc_running		running;
 	atomic64_t			vtime_rate;
+	u64				vtime_base_rate;
+	s64				vtime_err;
 
 	seqcount_spinlock_t		period_seqcount;
-	u32				period_at;	/* wallclock starttime */
+	u64				period_at;	/* wallclock starttime */
 	u64				period_at_vtime; /* vtime starttime */
 
 	atomic64_t			cur_period;	/* inc'd each period */
 	int				busy_level;	/* saturation history */
 
-	u64				inuse_margin_vtime;
 	bool				weights_updated;
 	atomic_t			hweight_gen;	/* for lazy hweights */
 
+	/* debt forgivness */
+	u64				dfgv_period_at;
+	u64				dfgv_period_rem;
+	u64				dfgv_usage_us_sum;
+
 	u64				autop_too_fast_at;
 	u64				autop_too_slow_at;
 	int				autop_idx;
@@ -424,6 +445,17 @@ struct ioc {
 	bool				user_cost_model:1;
 };
 
+struct iocg_pcpu_stat {
+	local64_t			abs_vusage;
+};
+
+struct iocg_stat {
+	u64				usage_us;
+	u64				wait_us;
+	u64				indebt_us;
+	u64				indelay_us;
+};
+
 /* per device-cgroup pair */
 struct ioc_gq {
 	struct blkg_policy_data		pd;
@@ -443,12 +475,17 @@ struct ioc_gq {
 	 *
 	 * `last_inuse` remembers `inuse` while an iocg is idle to persist
 	 * surplus adjustments.
+	 *
+	 * `inuse` may be adjusted dynamically during period. `saved_*` are used
+	 * to determine and track adjustments.
 	 */
 	u32				cfg_weight;
 	u32				weight;
 	u32				active;
 	u32				inuse;
+
 	u32				last_inuse;
+	s64				saved_margin;
 
 	sector_t			cursor;		/* to detect randio */
 
@@ -461,14 +498,14 @@ struct ioc_gq {
 	 * `vtime_done` is the same but progressed on completion rather
 	 * than issue.  The delta behind `vtime` represents the cost of
 	 * currently in-flight IOs.
-	 *
-	 * `last_vtime` is used to remember `vtime` at the end of the last
-	 * period to calculate utilization.
 	 */
 	atomic64_t			vtime;
 	atomic64_t			done_vtime;
 	u64				abs_vdebt;
-	u64				last_vtime;
+
+	/* current delay in effect and when it started */
+	u64				delay;
+	u64				delay_at;
 
 	/*
 	 * The period this iocg was last active in.  Used for deactivation
@@ -477,21 +514,35 @@ struct ioc_gq {
 	atomic64_t			active_period;
 	struct list_head		active_list;
 
-	/* see __propagate_active_weight() and current_hweight() for details */
+	/* see __propagate_weights() and current_hweight() for details */
 	u64				child_active_sum;
 	u64				child_inuse_sum;
+	u64				child_adjusted_sum;
 	int				hweight_gen;
 	u32				hweight_active;
 	u32				hweight_inuse;
-	bool				has_surplus;
+	u32				hweight_donating;
+	u32				hweight_after_donation;
+
+	struct list_head		walk_list;
+	struct list_head		surplus_list;
 
 	struct wait_queue_head		waitq;
 	struct hrtimer			waitq_timer;
-	struct hrtimer			delay_timer;
 
-	/* usage is recorded as fractions of HWEIGHT_WHOLE */
-	int				usage_idx;
-	u32				usages[NR_USAGE_SLOTS];
+	/* timestamp at the latest activation */
+	u64				activated_at;
+
+	/* statistics */
+	struct iocg_pcpu_stat __percpu	*pcpu_stat;
+	struct iocg_stat		local_stat;
+	struct iocg_stat		desc_stat;
+	struct iocg_stat		last_stat;
+	u64				last_stat_abs_vusage;
+	u64				usage_delta_us;
+	u64				wait_since;
+	u64				indebt_since;
+	u64				indelay_since;
 
 	/* this iocg's depth in the hierarchy and ancestors including self */
 	int				level;
@@ -506,7 +557,7 @@ struct ioc_cgrp {
 
 struct ioc_now {
 	u64				now_ns;
-	u32				now;
+	u64				now;
 	u64				vnow;
 	u64				vrate;
 };
@@ -656,7 +707,7 @@ static struct ioc_cgrp *blkcg_to_iocc(struct blkcg *blkcg)
  */
 static u64 abs_cost_to_cost(u64 abs_cost, u32 hw_inuse)
 {
-	return DIV64_U64_ROUND_UP(abs_cost * HWEIGHT_WHOLE, hw_inuse);
+	return DIV64_U64_ROUND_UP(abs_cost * WEIGHT_ONE, hw_inuse);
 }
 
 /*
@@ -664,18 +715,56 @@ static u64 abs_cost_to_cost(u64 abs_cost, u32 hw_inuse)
  */
 static u64 cost_to_abs_cost(u64 cost, u32 hw_inuse)
 {
-	return DIV64_U64_ROUND_UP(cost * hw_inuse, HWEIGHT_WHOLE);
+	return DIV64_U64_ROUND_UP(cost * hw_inuse, WEIGHT_ONE);
 }
 
-static void iocg_commit_bio(struct ioc_gq *iocg, struct bio *bio, u64 cost)
+static void iocg_commit_bio(struct ioc_gq *iocg, struct bio *bio,
+			    u64 abs_cost, u64 cost)
 {
+	struct iocg_pcpu_stat *gcs;
+
 	bio->bi_iocost_cost = cost;
 	atomic64_add(cost, &iocg->vtime);
+
+	gcs = get_cpu_ptr(iocg->pcpu_stat);
+	local64_add(abs_cost, &gcs->abs_vusage);
+	put_cpu_ptr(gcs);
+}
+
+static void iocg_lock(struct ioc_gq *iocg, bool lock_ioc, unsigned long *flags)
+{
+	if (lock_ioc) {
+		spin_lock_irqsave(&iocg->ioc->lock, *flags);
+		spin_lock(&iocg->waitq.lock);
+	} else {
+		spin_lock_irqsave(&iocg->waitq.lock, *flags);
+	}
+}
+
+static void iocg_unlock(struct ioc_gq *iocg, bool unlock_ioc, unsigned long *flags)
+{
+	if (unlock_ioc) {
+		spin_unlock(&iocg->waitq.lock);
+		spin_unlock_irqrestore(&iocg->ioc->lock, *flags);
+	} else {
+		spin_unlock_irqrestore(&iocg->waitq.lock, *flags);
+	}
 }
 
 #define CREATE_TRACE_POINTS
 #include <trace/events/iocost.h>
 
+static void ioc_refresh_margins(struct ioc *ioc)
+{
+	struct ioc_margins *margins = &ioc->margins;
+	u32 period_us = ioc->period_us;
+	u64 vrate = ioc->vtime_base_rate;
+
+	margins->min = (period_us * MARGIN_MIN_PCT / 100) * vrate;
+	margins->low = (period_us * MARGIN_LOW_PCT / 100) * vrate;
+	margins->target = (period_us * MARGIN_TARGET_PCT / 100) * vrate;
+}
+
 /* latency Qos params changed, update period_us and all the dependent params */
 static void ioc_refresh_period_us(struct ioc *ioc)
 {
@@ -709,9 +798,10 @@ static void ioc_refresh_period_us(struct ioc *ioc)
 
 	/* calculate dependent params */
 	ioc->period_us = period_us;
-	ioc->margin_us = period_us * MARGIN_PCT / 100;
-	ioc->inuse_margin_vtime = DIV64_U64_ROUND_UP(
-			period_us * VTIME_PER_USEC * INUSE_MARGIN_PCT, 100);
+	ioc->timer_slack_ns = div64_u64(
+		(u64)period_us * NSEC_PER_USEC * TIMER_SLACK_PCT,
+		100);
+	ioc_refresh_margins(ioc);
 }
 
 static int ioc_autop_idx(struct ioc *ioc)
@@ -738,8 +828,7 @@ static int ioc_autop_idx(