[PATCHSET v3 block/for-linus] IO cost model based work-conserving porportional controller

[PATCHSET v3 block/for-linus] IO cost model based work-conserving porportional controller

[Tejun Heo]

Changes from v2[2]:

* Fixed a divide-by-zero bug in current_hweight().

* pre_start_time and friends renamed to alloc_time and now has its own
  CONFIG option which is selected by IOCOST.

Changes from v1[1]:

* Prerequisite patchsets had cosmetic changes and merged.  Refreshed
  on top.

* Renamed from ioweight to iocost.  All source code and tools are
  updated accordingly.  Control knobs io.weight.qos and
  io.weight.cost_model are renamed to io.cost.qos and io.cost.model
  respectively.  This is a more fitting name which won't become a
  misnomer when, for example, cost based io.max is added.

* Various bug fixes and improvements.  A few bugs were discovered
  while testing against high-iops nvme device.  Auto parameter
  selection improved and verified across different classes of SSDs.

* Dropped bpf iocost support for now.

* Added coef generation script.

* Verified on high-iops nvme device.  Result is included below.

One challenge of controlling IO resources is the lack of trivially
observable cost metric.  This is distinguished from CPU and memory
where wallclock time and the number of bytes can serve as accurate
enough approximations.

Bandwidth and iops are the most commonly used metrics for IO devices
but depending on the type and specifics of the device, different IO
patterns easily lead to multiple orders of magnitude variations
rendering them useless for the purpose of IO capacity distribution.
While on-device time, with a lot of clutches, could serve as a useful
approximation for non-queued rotational devices, this is no longer
viable with modern devices, even the rotational ones.

While there is no cost metric we can trivially observe, it isn't a
complete mystery.  For example, on a rotational device, seek cost
dominates while a contiguous transfer contributes a smaller amount
proportional to the size.  If we can characterize at least the
relative costs of these different types of IOs, it should be possible
to implement a reasonable work-conserving proportional IO resource
distribution.

This patchset implements IO cost model based work-conserving
proportional controller.  It currently has a simple linear cost model
builtin where each IO is classified as sequential or random and given
a base cost accordingly and additional size-proportional cost is added
on top.  Each IO is given a cost based on the model and the controller
issues IOs for each cgroup according to their hierarchical weight.

By default, the controller adapts its overall IO rate so that it
doesn't build up buffer bloat in the request_queue layer, which
guarantees that the controller doesn't lose significant amount of
total work.  However, this may not provide sufficient differentiation
as the underlying device may have a deep queue and not be fair in how
the queued IOs are serviced.  The controller provides extra QoS
control knobs which allow tightening control feedback loop as
necessary.

For more details on the control mechanism, implementation and
interface, please refer to the comment at the top of
block/blk-iocost.c and Documentation/admin-guide/cgroup-v2.rst changes
in the "blkcg: implement blk-iocost" patch.

Here are some test results.  Each test run goes through the following
combinations with each combination running for a minute.  All tests
are performed against regular files on btrfs w/ deadline as the IO
scheduler.  Random IOs are direct w/ queue depth of 64.  Sequential
are normal buffered IOs.

        high priority (weight=500)      low priority (weight=100)

        Rand read                       None
        ditto                           Rand read
        ditto                           Seq  read
        ditto                           Rand write
        ditto                           Seq  write
        Seq  read                       None
        ditto                           Rand read
        ditto                           Seq  read
        ditto                           Rand write
        ditto                           Seq  write
        Rand write                      None
        ditto                           Rand read
        ditto                           Seq  read
        ditto                           Rand write
        ditto                           Seq  write
        Seq  write                      None
        ditto                           Rand read
        ditto                           Seq  read
        ditto                           Rand write
        ditto                           Seq  write

* 7200RPM SATA hard disk
  * No IO control
    https://photos.app.goo.gl/1KBHn7ykpC1LXRkB8
  * iocost, QoS: None
    https://photos.app.goo.gl/MLNQGxCtBQ8wAmjm7
  * iocost, QoS: rpct=95.00 rlat=40000 wpct=95.00 wlat=40000 min=25.00 max=200.00
    https://photos.app.goo.gl/XqXHm3Mkbm9w6Db46
* NCQ-blacklisted SATA SSD (QD==1)
  * No IO control
    https://photos.app.goo.gl/wCTXeu2uJ6LYL4pk8
  * iocost, QoS: None
    https://photos.app.goo.gl/T2HedKD2sywQgj7R9
  * iocost, QoS: rpct=95.00 rlat=20000 wpct=95.00 wlat=20000 min=50.00 max=200.00
    https://photos.app.goo.gl/urBTV8XQc1UqPJJw7
* SATA SSD (QD==32)
  * No IO control
    https://photos.app.goo.gl/TjEVykuVudSQcryh6
  * iocost, QoS: None
    https://photos.app.goo.gl/iyQBsky7bmM54Xiq7
  * iocost, QoS: rpct=95.00 rlat=10000 wpct=95.00 wlat=20000 min=50.00 max=400.00
    https://photos.app.goo.gl/q1a6URLDxPLMrnHy5
* NVME SSD (ran with 8 concurrent fio jobs to achieve saturation)
  * No IO control
    https://photos.app.goo.gl/S6xjEVTJzcfb3w1j7
  * iocost, QoS: None
    https://photos.app.goo.gl/SjQUUotJBAGr7vqz7
  * iocost, QoS: rpct=95.00 rlat=5000 wpct=95.00 wlat=5000 min=1.00 max=10000.00
    https://photos.app.goo.gl/RsaYBd2muX7CegoN7

Even without explicit QoS configuration, read-heavy scenarios can
obtain acceptable differentiation.  However, when write-heavy, the
deep buffering on the device side makes it difficult to maintain
control.  With QoS parameters set, the differentiation is acceptable
across all combinations.

The implementation comes with default cost model parameters which are
selected automatically which should provide acceptable behavior across
most common devices.  The parameters for hdd and consumer-grade SSDs
seem pretty robust.  The default parameter set and selection criteria
for highend SSDs might need further adjustments.

It is fairly easy to configure the QoS parameters and, if needed, cost
model coefficients.  We'll follow up with tooling and further
documentation.  Also, the last RFC patch in the series implements
support for bpf-based custom cost function.  Originally we thought
that we'd need per-device-type cost functions but the simple linear
model now seem good enough to cover all common device classes.  In
case custom cost functions become necessary, we can fully develop the
bpf based extension and also easily add different builtin cost models.

Andy Newell did the heavy lifting of analyzing IO workloads and device
characteristics, exploring various cost models, determining the
default model and parameters to use.

Josef Bacik implemented a prototype which explored the use of
different types of cost metrics including on-device time and Andy's
linear model.

This patchset is on top of the current block/for-next 53fc55c817c3
("Merge branch 'for-5.4/block' into for-next") and contains the
following 10 patches.

 0001-blkcg-pass-q-and-blkcg-into-blkcg_pol_alloc_pd_fn.patch
 0002-blkcg-make-cpd_init_fn-optional.patch
 0003-blkcg-separate-blkcg_conf_get_disk-out-of-blkg_conf_.patch
 0004-block-rq_qos-add-rq_qos_merge.patch
 0005-block-rq_qos-implement-rq_qos_ops-queue_depth_change.patch
 0006-blkcg-s-RQ_QOS_CGROUP-RQ_QOS_LATENCY.patch
 0007-blk-mq-add-optional-request-alloc_time_ns.patch
 0008-blkcg-implement-blk-iocost.patch
 0009-blkcg-add-tools-cgroup-iocost_monitor.py.patch
 0010-blkcg-add-tools-cgroup-iocost_coef_gen.py.patch

0001-0007 are prep patches.
0008 implements blk-iocost.
0009 adds monitoring script.
0010 adds linear cost model coefficient generation script.

The patchset is also available in the following git branch.

 git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup.git review-iow-v2

diffstat follows, Thanks.

 Documentation/admin-guide/cgroup-v2.rst |   97 +
 block/Kconfig                           |   13 
 block/Makefile                          |    1 
 block/bfq-cgroup.c                      |    5 
 block/blk-cgroup.c                      |   71 
 block/blk-core.c                        |    4 
 block/blk-iocost.c                      | 2395 ++++++++++++++++++++++++++++++++
 block/blk-iolatency.c                   |    8 
 block/blk-mq.c                          |   13 
 block/blk-rq-qos.c                      |   18 
 block/blk-rq-qos.h                      |   28 
 block/blk-settings.c                    |    2 
 block/blk-throttle.c                    |    6 
 block/blk-wbt.c                         |   18 
 block/blk-wbt.h                         |    4 
 include/linux/blk-cgroup.h              |    4 
 include/linux/blk_types.h               |    3 
 include/linux/blkdev.h                  |   13 
 include/trace/events/iocost.h           |  174 ++
 tools/cgroup/iocost_coef_gen.py         |  178 ++
 tools/cgroup/iocost_monitor.py          |  270 +++
 21 files changed, 3272 insertions(+), 53 deletions(-)

--
tejun

[1] http://lkml.kernel.org/r/20190614015620.1587672-1-tj@kernel.org
[2] http://lkml.kernel.org/r/20190710205128.1316483-1-tj@kernel.org

[PATCH 01/10] blkcg: pass @q and @blkcg into blkcg_pol_alloc_pd_fn()

[Tejun Heo]

Instead of @node, pass in @q and @blkcg so that the alloc function has
more context.  This doesn't cause any behavior change and will be used
by io.weight implementation.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/bfq-cgroup.c         | 5 +++--
 block/blk-cgroup.c         | 6 +++---
 block/blk-iolatency.c      | 6 ++++--
 block/blk-throttle.c       | 6 ++++--
 include/linux/blk-cgroup.h | 3 ++-
 5 files changed, 16 insertions(+), 10 deletions(-)

diff --git a/block/bfq-cgroup.c b/block/bfq-cgroup.c
index 0f6cd688924f..e6fb537b4bfc 100644
--- a/block/bfq-cgroup.c
+++ b/block/bfq-cgroup.c
@@ -501,11 +501,12 @@ static void bfq_cpd_free(struct blkcg_policy_data *cpd)
 	kfree(cpd_to_bfqgd(cpd));
 }
 
-static struct blkg_policy_data *bfq_pd_alloc(gfp_t gfp, int node)
+static struct blkg_policy_data *bfq_pd_alloc(gfp_t gfp, struct request_queue *q,
+					     struct blkcg *blkcg)
 {
 	struct bfq_group *bfqg;
 
-	bfqg = kzalloc_node(sizeof(*bfqg), gfp, node);
+	bfqg = kzalloc_node(sizeof(*bfqg), gfp, q->node);
 	if (!bfqg)
 		return NULL;
 
diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 55a7dc227dfb..6a82ca3fb5cf 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -175,7 +175,7 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct request_queue *q,
 			continue;
 
 		/* alloc per-policy data and attach it to blkg */
-		pd = pol->pd_alloc_fn(gfp_mask, q->node);
+		pd = pol->pd_alloc_fn(gfp_mask, q, blkcg);
 		if (!pd)
 			goto err_free;
 
@@ -1346,7 +1346,7 @@ int blkcg_activate_policy(struct request_queue *q,
 		blk_mq_freeze_queue(q);
 pd_prealloc:
 	if (!pd_prealloc) {
-		pd_prealloc = pol->pd_alloc_fn(GFP_KERNEL, q->node);
+		pd_prealloc = pol->pd_alloc_fn(GFP_KERNEL, q, &blkcg_root);
 		if (!pd_prealloc) {
 			ret = -ENOMEM;
 			goto out_bypass_end;
@@ -1362,7 +1362,7 @@ int blkcg_activate_policy(struct request_queue *q,
 		if (blkg->pd[pol->plid])
 			continue;
 
-		pd = pol->pd_alloc_fn(GFP_NOWAIT | __GFP_NOWARN, q->node);
+		pd = pol->pd_alloc_fn(GFP_NOWAIT | __GFP_NOWARN, q, &blkcg_root);
 		if (!pd)
 			swap(pd, pd_prealloc);
 		if (!pd) {
diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c
index 0fff7b56df0e..46fa6449f4bb 100644
--- a/block/blk-iolatency.c
+++ b/block/blk-iolatency.c
@@ -934,11 +934,13 @@ static size_t iolatency_pd_stat(struct blkg_policy_data *pd, char *buf,
 }
 
 
-static struct blkg_policy_data *iolatency_pd_alloc(gfp_t gfp, int node)
+static struct blkg_policy_data *iolatency_pd_alloc(gfp_t gfp,
+						   struct request_queue *q,
+						   struct blkcg *blkcg)
 {
 	struct iolatency_grp *iolat;
 
-	iolat = kzalloc_node(sizeof(*iolat), gfp, node);
+	iolat = kzalloc_node(sizeof(*iolat), gfp, q->node);
 	if (!iolat)
 		return NULL;
 	iolat->stats = __alloc_percpu_gfp(sizeof(struct latency_stat),
diff --git a/block/blk-throttle.c b/block/blk-throttle.c
index 8ab6c8153223..0445c998c377 100644
--- a/block/blk-throttle.c
+++ b/block/blk-throttle.c
@@ -478,12 +478,14 @@ static void throtl_service_queue_init(struct throtl_service_queue *sq)
 	timer_setup(&sq->pending_timer, throtl_pending_timer_fn, 0);
 }
 
-static struct blkg_policy_data *throtl_pd_alloc(gfp_t gfp, int node)
+static struct blkg_policy_data *throtl_pd_alloc(gfp_t gfp,
+						struct request_queue *q,
+						struct blkcg *blkcg)
 {
 	struct throtl_grp *tg;
 	int rw;
 
-	tg = kzalloc_node(sizeof(*tg), gfp, node);
+	tg = kzalloc_node(sizeof(*tg), gfp, q->node);
 	if (!tg)
 		return NULL;
 
diff --git a/include/linux/blk-cgroup.h b/include/linux/blk-cgroup.h
index 0bb79d858a13..261248e88eb1 100644
--- a/include/linux/blk-cgroup.h
+++ b/include/linux/blk-cgroup.h
@@ -149,7 +149,8 @@ typedef struct blkcg_policy_data *(blkcg_pol_alloc_cpd_fn)(gfp_t gfp);
 typedef void (blkcg_pol_init_cpd_fn)(struct blkcg_policy_data *cpd);
 typedef void (blkcg_pol_free_cpd_fn)(struct blkcg_policy_data *cpd);
 typedef void (blkcg_pol_bind_cpd_fn)(struct blkcg_policy_data *cpd);
-typedef struct blkg_policy_data *(blkcg_pol_alloc_pd_fn)(gfp_t gfp, int node);
+typedef struct blkg_policy_data *(blkcg_pol_alloc_pd_fn)(gfp_t gfp,
+				struct request_queue *q, struct blkcg *blkcg);
 typedef void (blkcg_pol_init_pd_fn)(struct blkg_policy_data *pd);
 typedef void (blkcg_pol_online_pd_fn)(struct blkg_policy_data *pd);
 typedef void (blkcg_pol_offline_pd_fn)(struct blkg_policy_data *pd);
-- 
2.17.1

[PATCH 02/10] blkcg: make ->cpd_init_fn() optional

[Tejun Heo]

For policies which can do enough initialization from ->cpd_alloc_fn(),
make ->cpd_init_fn() optional.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/blk-cgroup.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 6a82ca3fb5cf..78ccbdcfe723 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -1475,7 +1475,8 @@ int blkcg_policy_register(struct blkcg_policy *pol)
 			blkcg->cpd[pol->plid] = cpd;
 			cpd->blkcg = blkcg;
 			cpd->plid = pol->plid;
-			pol->cpd_init_fn(cpd);
+			if (pol->cpd_init_fn)
+				pol->cpd_init_fn(cpd);
 		}
 	}
 
-- 
2.17.1

[PATCH 03/10] blkcg: separate blkcg_conf_get_disk() out of blkg_conf_prep()

[Tejun Heo]

Separate out blkcg_conf_get_disk() so that it can be used by blkcg
policy interface file input parsers before the policy is actually
enabled.  This doesn't introduce any functional changes.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/blk-cgroup.c         | 62 ++++++++++++++++++++++++++------------
 include/linux/blk-cgroup.h |  1 +
 2 files changed, 44 insertions(+), 19 deletions(-)

diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 78ccbdcfe723..0e2619c1a422 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -753,6 +753,44 @@ static struct blkcg_gq *blkg_lookup_check(struct blkcg *blkcg,
 	return __blkg_lookup(blkcg, q, true /* update_hint */);
 }
 
+/**
+ * blkg_conf_prep - parse and prepare for per-blkg config update
+ * @inputp: input string pointer
+ *
+ * Parse the device node prefix part, MAJ:MIN, of per-blkg config update
+ * from @input and get and return the matching gendisk.  *@inputp is
+ * updated to point past the device node prefix.  Returns an ERR_PTR()
+ * value on error.
+ *
+ * Use this function iff blkg_conf_prep() can't be used for some reason.
+ */
+struct gendisk *blkcg_conf_get_disk(char **inputp)
+{
+	char *input = *inputp;
+	unsigned int major, minor;
+	struct gendisk *disk;
+	int key_len, part;
+
+	if (sscanf(input, "%u:%u%n", &major, &minor, &key_len) != 2)
+		return ERR_PTR(-EINVAL);
+
+	input += key_len;
+	if (!isspace(*input))
+		return ERR_PTR(-EINVAL);
+	input = skip_spaces(input);
+
+	disk = get_gendisk(MKDEV(major, minor), &part);
+	if (!disk)
+		return ERR_PTR(-ENODEV);
+	if (part) {
+		put_disk_and_module(disk);
+		return ERR_PTR(-ENODEV);
+	}
+
+	*inputp = input;
+	return disk;
+}
+
 /**
  * blkg_conf_prep - parse and prepare for per-blkg config update
  * @blkcg: target block cgroup
@@ -772,25 +810,11 @@ int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
 	struct gendisk *disk;
 	struct request_queue *q;
 	struct blkcg_gq *blkg;
-	unsigned int major, minor;
-	int key_len, part, ret;
-	char *body;
-
-	if (sscanf(input, "%u:%u%n", &major, &minor, &key_len) != 2)
-		return -EINVAL;
-
-	body = input + key_len;
-	if (!isspace(*body))
-		return -EINVAL;
-	body = skip_spaces(body);
+	int ret;
 
-	disk = get_gendisk(MKDEV(major, minor), &part);
-	if (!disk)
-		return -ENODEV;
-	if (part) {
-		ret = -ENODEV;
-		goto fail;
-	}
+	disk = blkcg_conf_get_disk(&input);
+	if (IS_ERR(disk))
+		return PTR_ERR(disk);
 
 	q = disk->queue;
 
@@ -856,7 +880,7 @@ int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
 success:
 	ctx->disk = disk;
 	ctx->blkg = blkg;
-	ctx->body = body;
+	ctx->body = input;
 	return 0;
 
 fail_unlock:
diff --git a/include/linux/blk-cgroup.h b/include/linux/blk-cgroup.h
index 261248e88eb1..bed9e43f9426 100644
--- a/include/linux/blk-cgroup.h
+++ b/include/linux/blk-cgroup.h
@@ -234,6 +234,7 @@ struct blkg_conf_ctx {
 	char				*body;
 };
 
+struct gendisk *blkcg_conf_get_disk(char **inputp);
 int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
 		   char *input, struct blkg_conf_ctx *ctx);
 void blkg_conf_finish(struct blkg_conf_ctx *ctx);
-- 
2.17.1

[PATCH 04/10] block/rq_qos: add rq_qos_merge()

[Tejun Heo]

Add a merge hook for rq_qos.  This will be used by io.weight.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/blk-core.c   | 4 ++++
 block/blk-rq-qos.c | 9 +++++++++
 block/blk-rq-qos.h | 9 +++++++++
 3 files changed, 22 insertions(+)

diff --git a/block/blk-core.c b/block/blk-core.c
index 77807a5d7f9e..875e8d105067 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -604,6 +604,7 @@ bool bio_attempt_back_merge(struct request *req, struct bio *bio,
 		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);
@@ -625,6 +626,7 @@ bool bio_attempt_front_merge(struct request *req, struct bio *bio,
 		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);
@@ -650,6 +652,8 @@ bool bio_attempt_discard_merge(struct request_queue *q, struct request *req,
 	    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;
diff --git a/block/blk-rq-qos.c b/block/blk-rq-qos.c
index 3954c0dc1443..f4eea78f5cc1 100644
--- a/block/blk-rq-qos.c
+++ b/block/blk-rq-qos.c
@@ -83,6 +83,15 @@ void __rq_qos_track(struct rq_qos *rqos, struct request *rq, struct bio *bio)
 	} while (rqos);
 }
 
+void __rq_qos_merge(struct rq_qos *rqos, struct request *rq, struct bio *bio)
+{
+	do {
+		if (rqos->ops->merge)
+			rqos->ops->merge(rqos, rq, bio);
+		rqos = rqos->next;
+	} while (rqos);
+}
+
 void __rq_qos_done_bio(struct rq_qos *rqos, struct bio *bio)
 {
 	do {
diff --git a/block/blk-rq-qos.h b/block/blk-rq-qos.h
index 2300e038b9fa..8e426a8505b6 100644
--- a/block/blk-rq-qos.h
+++ b/block/blk-rq-qos.h
@@ -35,6 +35,7 @@ struct rq_qos {
 struct rq_qos_ops {
 	void (*throttle)(struct rq_qos *, struct bio *);
 	void (*track)(struct rq_qos *, struct request *, struct bio *);
+	void (*merge)(struct rq_qos *, struct request *, struct bio *);
 	void (*issue)(struct rq_qos *, struct request *);
 	void (*requeue)(struct rq_qos *, struct request *);
 	void (*done)(struct rq_qos *, struct request *);
@@ -135,6 +136,7 @@ void __rq_qos_issue(struct rq_qos *rqos, struct request *rq);
 void __rq_qos_requeue(struct rq_qos *rqos, struct request *rq);
 void __rq_qos_throttle(struct rq_qos *rqos, struct bio *bio);
 void __rq_qos_track(struct rq_qos *rqos, struct request *rq, struct bio *bio);
+void __rq_qos_merge(struct rq_qos *rqos, struct request *rq, struct bio *bio);
 void __rq_qos_done_bio(struct rq_qos *rqos, struct bio *bio);
 
 static inline void rq_qos_cleanup(struct request_queue *q, struct bio *bio)
@@ -185,6 +187,13 @@ static inline void rq_qos_track(struct request_queue *q, struct request *rq,
 		__rq_qos_track(q->rq_qos, rq, bio);
 }
 
+static inline void rq_qos_merge(struct request_queue *q, struct request *rq,
+				struct bio *bio)
+{
+	if (q->rq_qos)
+		__rq_qos_merge(q->rq_qos, rq, bio);
+}
+
 void rq_qos_exit(struct request_queue *);
 
 #endif
-- 
2.17.1

[PATCH 05/10] block/rq_qos: implement rq_qos_ops->queue_depth_changed()

[Tejun Heo]

wbt already gets queue depth changed notification through
wbt_set_queue_depth().  Generalize it into
rq_qos_ops->queue_depth_changed() so that other rq_qos policies can
easily hook into the events too.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/blk-rq-qos.c   |  9 +++++++++
 block/blk-rq-qos.h   |  8 ++++++++
 block/blk-settings.c |  2 +-
 block/blk-wbt.c      | 18 ++++++++----------
 block/blk-wbt.h      |  4 ----
 5 files changed, 26 insertions(+), 15 deletions(-)

diff --git a/block/blk-rq-qos.c b/block/blk-rq-qos.c
index f4eea78f5cc1..61b635bc2a31 100644
--- a/block/blk-rq-qos.c
+++ b/block/blk-rq-qos.c
@@ -101,6 +101,15 @@ void __rq_qos_done_bio(struct rq_qos *rqos, struct bio *bio)
 	} while (rqos);
 }
 
+void __rq_qos_queue_depth_changed(struct rq_qos *rqos)
+{
+	do {
+		if (rqos->ops->queue_depth_changed)
+			rqos->ops->queue_depth_changed(rqos);
+		rqos = rqos->next;
+	} while (rqos);
+}
+
 /*
  * Return true, if we can't increase the depth further by scaling
  */
diff --git a/block/blk-rq-qos.h b/block/blk-rq-qos.h
index 8e426a8505b6..e15b6907b76d 100644
--- a/block/blk-rq-qos.h
+++ b/block/blk-rq-qos.h
@@ -41,6 +41,7 @@ struct rq_qos_ops {
 	void (*done)(struct rq_qos *, struct request *);
 	void (*done_bio)(struct rq_qos *, struct bio *);
 	void (*cleanup)(struct rq_qos *, struct bio *);
+	void (*queue_depth_changed)(struct rq_qos *);
 	void (*exit)(struct rq_qos *);
 	const struct blk_mq_debugfs_attr *debugfs_attrs;
 };
@@ -138,6 +139,7 @@ void __rq_qos_throttle(struct rq_qos *rqos, struct bio *bio);
 void __rq_qos_track(struct rq_qos *rqos, struct request *rq, struct bio *bio);
 void __rq_qos_merge(struct rq_qos *rqos, struct request *rq, struct bio *bio);
 void __rq_qos_done_bio(struct rq_qos *rqos, struct bio *bio);
+void __rq_qos_queue_depth_changed(struct rq_qos *rqos);
 
 static inline void rq_qos_cleanup(struct request_queue *q, struct bio *bio)
 {
@@ -194,6 +196,12 @@ static inline void rq_qos_merge(struct request_queue *q, struct request *rq,
 		__rq_qos_merge(q->rq_qos, rq, bio);
 }
 
+static inline void rq_qos_queue_depth_changed(struct request_queue *q)
+{
+	if (q->rq_qos)
+		__rq_qos_queue_depth_changed(q->rq_qos);
+}
+
 void rq_qos_exit(struct request_queue *);
 
 #endif
diff --git a/block/blk-settings.c b/block/blk-settings.c
index 2c1831207a8f..a058997b9cce 100644
--- a/block/blk-settings.c
+++ b/block/blk-settings.c
@@ -805,7 +805,7 @@ EXPORT_SYMBOL(blk_queue_update_dma_alignment);
 void blk_set_queue_depth(struct request_queue *q, unsigned int depth)
 {
 	q->queue_depth = depth;
-	wbt_set_queue_depth(q, depth);
+	rq_qos_queue_depth_changed(q);
 }
 EXPORT_SYMBOL(blk_set_queue_depth);
 
diff --git a/block/blk-wbt.c b/block/blk-wbt.c
index c4d3089e47f7..8af553a0ba00 100644
--- a/block/blk-wbt.c
+++ b/block/blk-wbt.c
@@ -629,15 +629,6 @@ static void wbt_requeue(struct rq_qos *rqos, struct request *rq)
 	}
 }
 
-void wbt_set_queue_depth(struct request_queue *q, unsigned int depth)
-{
-	struct rq_qos *rqos = wbt_rq_qos(q);
-	if (rqos) {
-		RQWB(rqos)->rq_depth.queue_depth = depth;
-		__wbt_update_limits(RQWB(rqos));
-	}
-}
-
 void wbt_set_write_cache(struct request_queue *q, bool write_cache_on)
 {
 	struct rq_qos *rqos = wbt_rq_qos(q);
@@ -689,6 +680,12 @@ static int wbt_data_dir(const struct request *rq)
 	return -1;
 }
 
+static void wbt_queue_depth_changed(struct rq_qos *rqos)
+{
+	RQWB(rqos)->rq_depth.queue_depth = blk_queue_depth(rqos->q);
+	__wbt_update_limits(RQWB(rqos));
+}
+
 static void wbt_exit(struct rq_qos *rqos)
 {
 	struct rq_wb *rwb = RQWB(rqos);
@@ -811,6 +808,7 @@ static struct rq_qos_ops wbt_rqos_ops = {
 	.requeue = wbt_requeue,
 	.done = wbt_done,
 	.cleanup = wbt_cleanup,
+	.queue_depth_changed = wbt_queue_depth_changed,
 	.exit = wbt_exit,
 #ifdef CONFIG_BLK_DEBUG_FS
 	.debugfs_attrs = wbt_debugfs_attrs,
@@ -853,7 +851,7 @@ int wbt_init(struct request_queue *q)
 
 	rwb->min_lat_nsec = wbt_default_latency_nsec(q);
 
-	wbt_set_queue_depth(q, blk_queue_depth(q));
+	wbt_queue_depth_changed(&rwb->rqos);
 	wbt_set_write_cache(q, test_bit(QUEUE_FLAG_WC, &q->queue_flags));
 
 	return 0;
diff --git a/block/blk-wbt.h b/block/blk-wbt.h
index f47218d5b3b2..8e4e37660971 100644
--- a/block/blk-wbt.h
+++ b/block/blk-wbt.h
@@ -95,7 +95,6 @@ void wbt_enable_default(struct request_queue *);
 u64 wbt_get_min_lat(struct request_queue *q);
 void wbt_set_min_lat(struct request_queue *q, u64 val);
 
-void wbt_set_queue_depth(struct request_queue *, unsigned int);
 void wbt_set_write_cache(struct request_queue *, bool);
 
 u64 wbt_default_latency_nsec(struct request_queue *);
@@ -118,9 +117,6 @@ static inline void wbt_disable_default(struct request_queue *q)
 static inline void wbt_enable_default(struct request_queue *q)
 {
 }
-static inline void wbt_set_queue_depth(struct request_queue *q, unsigned int depth)
-{
-}
 static inline void wbt_set_write_cache(struct request_queue *q, bool wc)
 {
 }
-- 
2.17.1

[PATCH 06/10] blkcg: s/RQ_QOS_CGROUP/RQ_QOS_LATENCY/

[Tejun Heo]

io.weight is gonna be another rq_qos cgroup mechanism.  Let's rename
RQ_QOS_CGROUP which is being used by io.latency to RQ_QOS_LATENCY in
preparation.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/blk-iolatency.c | 2 +-
 block/blk-rq-qos.h    | 8 ++++----
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c
index 46fa6449f4bb..c128d50cb410 100644
--- a/block/blk-iolatency.c
+++ b/block/blk-iolatency.c
@@ -725,7 +725,7 @@ int blk_iolatency_init(struct request_queue *q)
 		return -ENOMEM;
 
 	rqos = &blkiolat->rqos;
-	rqos->id = RQ_QOS_CGROUP;
+	rqos->id = RQ_QOS_LATENCY;
 	rqos->ops = &blkcg_iolatency_ops;
 	rqos->q = q;
 
diff --git a/block/blk-rq-qos.h b/block/blk-rq-qos.h
index e15b6907b76d..5f8b75826a98 100644
--- a/block/blk-rq-qos.h
+++ b/block/blk-rq-qos.h
@@ -14,7 +14,7 @@ struct blk_mq_debugfs_attr;
 
 enum rq_qos_id {
 	RQ_QOS_WBT,
-	RQ_QOS_CGROUP,
+	RQ_QOS_LATENCY,
 };
 
 struct rq_wait {
@@ -74,7 +74,7 @@ static inline struct rq_qos *wbt_rq_qos(struct request_queue *q)
 
 static inline struct rq_qos *blkcg_rq_qos(struct request_queue *q)
 {
-	return rq_qos_id(q, RQ_QOS_CGROUP);
+	return rq_qos_id(q, RQ_QOS_LATENCY);
 }
 
 static inline const char *rq_qos_id_to_name(enum rq_qos_id id)
@@ -82,8 +82,8 @@ static inline const char *rq_qos_id_to_name(enum rq_qos_id id)
 	switch (id) {
 	case RQ_QOS_WBT:
 		return "wbt";
-	case RQ_QOS_CGROUP:
-		return "cgroup";
+	case RQ_QOS_LATENCY:
+		return "latency";
 	}
 	return "unknown";
 }
-- 
2.17.1

[PATCH 07/10] blk-mq: add optional request->alloc_time_ns

[Tejun Heo]

There are currently two start time timestamps - start_time_ns and
io_start_time_ns.  The former marks the request allocation and and the
second issue-to-device time.  The planned io.weight controller needs
to measure the total time bios take to execute after it leaves rq_qos
including the time spent waiting for request to become available,
which can easily dominate on saturated devices.

This patch adds request->alloc_time_ns which records when the request
allocation attempt started.  As it isn't used for the usual stats,
make it optional behind CONFIG_BLK_RQ_ALLOC_TIME and
QUEUE_FLAG_RQ_ALLOC_TIME so that it can be compiled out when there are
no users and it's active only on queues which need it even when
compiled in.

v2: s/pre_start_time/alloc_time/ and add CONFIG_BLK_RQ_ALLOC_TIME
    gating as suggested by Jens.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/Kconfig          |  3 +++
 block/blk-mq.c         | 13 +++++++++++--
 include/linux/blkdev.h | 13 ++++++++++++-
 3 files changed, 26 insertions(+), 3 deletions(-)

diff --git a/block/Kconfig b/block/Kconfig
index 8b5f8e560eb4..1b62ad6d0e12 100644
--- a/block/Kconfig
+++ b/block/Kconfig
@@ -26,6 +26,9 @@ menuconfig BLOCK
 
 if BLOCK
 
+config BLK_RQ_ALLOC_TIME
+	bool
+
 config BLK_SCSI_REQUEST
 	bool
 
diff --git a/block/blk-mq.c b/block/blk-mq.c
index f6620a30752e..b622029b19ea 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -291,7 +291,7 @@ static inline bool blk_mq_need_time_stamp(struct request *rq)
 }
 
 static struct request *blk_mq_rq_ctx_init(struct blk_mq_alloc_data *data,
-		unsigned int tag, unsigned int op)
+		unsigned int tag, unsigned int op, u64 alloc_time_ns)
 {
 	struct blk_mq_tags *tags = blk_mq_tags_from_data(data);
 	struct request *rq = tags->static_rqs[tag];
@@ -325,6 +325,9 @@ static struct request *blk_mq_rq_ctx_init(struct blk_mq_alloc_data *data,
 	RB_CLEAR_NODE(&rq->rb_node);
 	rq->rq_disk = NULL;
 	rq->part = NULL;
+#ifdef CONFIG_BLK_RQ_ALLOC_TIME
+	rq->alloc_time_ns = alloc_time_ns;
+#endif
 	if (blk_mq_need_time_stamp(rq))
 		rq->start_time_ns = ktime_get_ns();
 	else
@@ -356,8 +359,14 @@ static struct request *blk_mq_get_request(struct request_queue *q,
 	struct request *rq;
 	unsigned int tag;
 	bool clear_ctx_on_error = false;
+	u64 alloc_time_ns = 0;
 
 	blk_queue_enter_live(q);
+
+	/* alloc_time includes depth and tag waits */
+	if (blk_queue_rq_alloc_time(q))
+		alloc_time_ns = ktime_get_ns();
+
 	data->q = q;
 	if (likely(!data->ctx)) {
 		data->ctx = blk_mq_get_ctx(q);
@@ -393,7 +402,7 @@ static struct request *blk_mq_get_request(struct request_queue *q,
 		return NULL;
 	}
 
-	rq = blk_mq_rq_ctx_init(data, tag, data->cmd_flags);
+	rq = blk_mq_rq_ctx_init(data, tag, data->cmd_flags, alloc_time_ns);
 	if (!op_is_flush(data->cmd_flags)) {
 		rq->elv.icq = NULL;
 		if (e && e->type->ops.prepare_request) {
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 1ac790178787..d0ad21e4771b 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -194,7 +194,11 @@ struct request {
 
 	struct gendisk *rq_disk;
 	struct hd_struct *part;
-	/* Time that I/O was submitted to the kernel. */
+#ifdef CONFIG_BLK_RQ_ALLOC_TIME
+	/* Time that the first bio started allocating this request. */
+	u64 alloc_time_ns;
+#endif
+	/* Time that this request was allocated for this IO. */
 	u64 start_time_ns;
 	/* Time that I/O was submitted to the device. */
 	u64 io_start_time_ns;
@@ -609,6 +613,7 @@ struct request_queue {
 #define QUEUE_FLAG_QUIESCED	24	/* queue has been quiesced */
 #define QUEUE_FLAG_PCI_P2PDMA	25	/* device supports PCI p2p requests */
 #define QUEUE_FLAG_ZONE_RESETALL 26	/* supports Zone Reset All */
+#define QUEUE_FLAG_RQ_ALLOC_TIME 27	/* record rq->alloc_time_ns */
 
 #define QUEUE_FLAG_MQ_DEFAULT	((1 << QUEUE_FLAG_IO_STAT) |		\
 				 (1 << QUEUE_FLAG_SAME_COMP))
@@ -637,6 +642,12 @@ bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
 	test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
 #define blk_queue_pci_p2pdma(q)	\
 	test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
+#ifdef CONFIG_BLK_RQ_ALLOC_TIME
+#define blk_queue_rq_alloc_time(q)	\
+	test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
+#else
+#define blk_queue_rq_alloc_time(q)	false
+#endif
 
 #define blk_noretry_request(rq) \
 	((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
-- 
2.17.1

[PATCH 08/10] blkcg: implement blk-iocost

[Tejun Heo]

This patchset implements IO cost model based work-conserving
proportional controller.

While io.latency provides the capability to comprehensively prioritize
and protect IOs depending on the cgroups, its protection is binary -
the lowest latency target cgroup which is suffering is protected at
the cost of all others.  In many use cases including stacking multiple
workload containers in a single system, it's necessary to distribute
IO capacity with better granularity.

One challenge of controlling IO resources is the lack of trivially
observable cost metric.  The most common metrics - bandwidth and iops
- can be off by orders of magnitude depending on the device type and
IO pattern.  However, the cost isn't a complete mystery.  Given
several key attributes, we can make fairly reliable predictions on how
expensive a given stream of IOs would be, at least compared to other
IO patterns.

The function which determines the cost of a given IO is the IO cost
model for the device.  This controller distributes IO capacity based
on the costs estimated by such model.  The more accurate the cost
model the better but the controller adapts based on IO completion
latency and as long as the relative costs across differents IO
patterns are consistent and sensible, it'll adapt to the actual
performance of the device.

Currently, the only implemented cost model is a simple linear one with
a few sets of default parameters for different classes of device.
This covers most common devices reasonably well.  All the
infrastructure to tune and add different cost models is already in
place and a later patch will also allow using bpf progs for cost
models.

Please see the top comment in blk-iocost.c and documentation for
more details.

v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix
    for a divide-by-zero bug in current_hweight() triggered by zero
    inuse_sum.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Andy Newell <newella@fb.com>
Cc: Josef Bacik <jbacik@fb.com>
Cc: Rik van Riel <riel@surriel.com>
---
 Documentation/admin-guide/cgroup-v2.rst |   94 +
 block/Kconfig                           |   10 +
 block/Makefile                          |    1 +
 block/blk-iocost.c                      | 2371 +++++++++++++++++++++++
 block/blk-rq-qos.h                      |    3 +
 include/linux/blk_types.h               |    3 +
 include/trace/events/iocost.h           |  174 ++
 7 files changed, 2656 insertions(+)
 create mode 100644 block/blk-iocost.c
 create mode 100644 include/trace/events/iocost.h

diff --git a/Documentation/admin-guide/cgroup-v2.rst b/Documentation/admin-guide/cgroup-v2.rst
index 3b29005aa981..1521c7e554f5 100644
--- a/Documentation/admin-guide/cgroup-v2.rst
+++ b/Documentation/admin-guide/cgroup-v2.rst
@@ -1435,6 +1435,100 @@ IO Interface Files
 	  8:16 rbytes=1459200 wbytes=314773504 rios=192 wios=353 dbytes=0 dios=0
 	  8:0 rbytes=90430464 wbytes=299008000 rios=8950 wios=1252 dbytes=50331648 dios=3021
 
+  io.cost.qos
+	A read-write nested-keyed file with exists only on the root
+	cgroup.
+
+	This file configures the Quality of Service of the IO cost
+	model based controller (CONFIG_BLK_CGROUP_IOCOST) which
+	currently implements "io.weight" proportional control.  Lines
+	are keyed by $MAJ:$MIN device numbers and not ordered.  The
+	line for a given device is populated on the first write for
+	the device on "io.cost.qos" or "io.cost.model".  The following
+	nested keys are defined.
+
+	  ======	=====================================
+	  enable	Weight-based control enable
+	  ctrl		"auto" or "user"
+	  rpct		Read latency percentile    [0, 100]
+	  rlat		Read latency threshold
+	  wpct		Write latency percentile   [0, 100]
+	  wlat		Write latency threshold
+	  min		Minimum scaling percentage [1, 10000]
+	  max		Maximum scaling percentage [1, 10000]
+	  ======	=====================================
+
+	The controller is disabled by default and can be enabled by
+	setting "enable" to 1.  "rpct" and "wpct" parameters default
+	to zero and the controller uses internal device saturation
+	state to adjust the overall IO rate between "min" and "max".
+
+	When a better control quality is needed, latency QoS
+	parameters can be configured.  For example::
+
+	  8:16 enable=1 ctrl=auto rpct=95.00 rlat=75000 wpct=95.00 wlat=150000 min=50.00 max=150.0
+
+	shows that on sdb, the controller is enabled, will consider
+	the device saturated if the 95th percentile of read completion
+	latencies is above 75ms or write 150ms, and adjust the overall
+	IO issue rate between 50% and 150% accordingly.
+
+	The lower the saturation point, the better the latency QoS at
+	the cost of aggregate bandwidth.  The narrower the allowed
+	adjustment range between "min" and "max", the more conformant
+	to the cost model the IO behavior.  Note that the IO issue
+	base rate may be far off from 100% and setting "min" and "max"
+	blindly can lead to a significant loss of device capacity or
+	control quality.  "min" and "max" are useful for regulating
+	devices which show wide temporary behavior changes - e.g. a
+	ssd which accepts writes at the line speed for a while and
+	then completely stalls for multiple seconds.
+
+	When "ctrl" is "auto", the parameters are controlled by the
+	kernel and may change automatically.  Setting "ctrl" to "user"
+	or setting any of the percentile and latency parameters puts
+	it into "user" mode and disables the automatic changes.  The
+	automatic mode can be restored by setting "ctrl" to "auto".
+
+  io.cost.model
+	A read-write nested-keyed file with exists only on the root
+	cgroup.
+
+	This file configures the cost model of the IO cost model based
+	controller (CONFIG_BLK_CGROUP_IOCOST) which currently
+	implements "io.weight" proportional control.  Lines are keyed
+	by $MAJ:$MIN device numbers and not ordered.  The line for a
+	given device is populated on the first write for the device on
+	"io.cost.qos" or "io.cost.model".  The following nested keys
+	are defined.
+
+	  =====		================================
+	  ctrl		"auto" or "user"
+	  model		The cost model in use - "linear"
+	  =====		================================
+
+	When "ctrl" is "auto", the kernel may change all parameters
+	dynamically.  When "ctrl" is set to "user" or any other
+	parameters are written to, "ctrl" become "user" and the
+	automatic changes are disabled.
+
+	When "model" is "linear", the following model parameters are
+	defined.
+
+	  =============	========================================
+	  [r|w]bps	The maximum sequential IO throughput
+	  [r|w]seqiops	The maximum 4k sequential IOs per second
+	  [r|w]randiops	The maximum 4k random IOs per second
+	  =============	========================================
+
+	From the above, the builtin linear model determines the base
+	costs of a sequential and random IO and the cost coefficient
+	for the IO size.  While simple, this model can cover most
+	common device classes acceptably.
+
+	The IO cost model isn't expected to be accurate in absolute
+	sense and is scaled to the device behavior dynamically.
+
   io.weight
 	A read-write flat-keyed file which exists on non-root cgroups.
 	The default is "default 100".
diff --git a/block/Kconfig b/block/Kconfig
index 1b62ad6d0e12..41c0917ce622 100644
--- a/block/Kconfig
+++ b/block/Kconfig
@@ -135,6 +135,16 @@ config BLK_CGROUP_IOLATENCY
 
 	Note, this is an experimental interface and could be changed someday.
 
+config BLK_CGROUP_IOCOST
+	bool "Enable support for cost model based cgroup IO controller"
+	depends on BLK_CGROUP=y
+	select BLK_RQ_ALLOC_TIME
+	---help---
+	Enabling this option enables the .weight interface for cost
+	model based proportional IO control.  The IO controller
+	distributes IO capacity between different groups based on
+	their share of the overall weight distribution.
+
 config BLK_WBT_MQ
 	bool "Multiqueue writeback throttling"
 	default y
diff --git a/block/Makefile b/block/Makefile
index eee1b4ceecf9..9ef57ace90d4 100644
--- a/block/Makefile
+++ b/block/Makefile
@@ -18,6 +18,7 @@ obj-$(CONFIG_BLK_DEV_BSGLIB)	+= bsg-lib.o
 obj-$(CONFIG_BLK_CGROUP)	+= blk-cgroup.o
 obj-$(CONFIG_BLK_DEV_THROTTLING)	+= blk-throttle.o
 obj-$(CONFIG_BLK_CGROUP_IOLATENCY)	+= blk-iolatency.o
+obj-$(CONFIG_BLK_CGROUP_IOCOST)	+= blk-iocost.o
 obj-$(CONFIG_MQ_IOSCHED_DEADLINE)	+= mq-deadline.o
 obj-$(CONFIG_MQ_IOSCHED_KYBER)	+= kyber-iosched.o
 bfq-y				:= bfq-iosched.o bfq-wf2q.o bfq-cgroup.o
diff --git a/block/blk-iocost.c b/block/blk-iocost.c
new file mode 100644
index 000000000000..680815620095
--- /dev/null
+++ b/block/blk-iocost.c
@@ -0,0 +1,2371 @@
+/* SPDX-License-Identifier: GPL-2.0
+ *
+ * IO cost model based controller.
+ *
+ * Copyright (C) 2019 Tejun Heo <tj@kernel.org>
+ * Copyright (C) 2019 Andy Newell <newella@fb.com>
+ * Copyright (C) 2019 Facebook
+ *
+ * One challenge of controlling IO resources is the lack of trivially
+ * observable cost metric.  This is distinguished from CPU and memory where
+ * wallclock time and the number of bytes can serve as accurate enough
+ * approximations.
+ *
+ * Bandwidth and iops are the most commonly used metrics for IO devices but
+ * depending on the type and specifics of the device, different IO patterns
+ * easily lead to multiple orders of magnitude variations rendering them
+ * useless for the purpose of IO capacity distribution.  While on-device
+ * time, with a lot of clutches, could serve as a useful approximation for
+ * non-queued rotational devices, this is no longer viable with modern
+ * devices, even the rotational ones.
+ *
+ * While there is no cost metric we can trivially observe, it isn't a
+ * complete mystery.  For example, on a rotational device, seek cost
+ * dominates while a contiguous transfer contributes a smaller amount
+ * proportional to the size.  If we can characterize at least the relative
+ * costs of these different types of IOs, it should be possible to
+ * implement a reasonable work-conserving proportional IO resource
+ * distribution.
+ *
+ * 1. IO Cost Model
+ *
+ * IO cost model estimates the cost of an IO given its basic parameters and
+ * history (e.g. the end sector of the last IO).  The cost is measured in
+ * device time.  If a given IO is estimated to cost 10ms, the device should
+ * be able to process ~100 of those IOs in a second.
+ *
+ * Currently, there's only one builtin cost model - linear.  Each IO is
+ * classified as sequential or random and given a base cost accordingly.
+ * On top of that, a size cost proportional to the length of the IO is
+ * added.  While simple, this model captures the operational
+ * characteristics of a wide varienty of devices well enough.  Default
+ * paramters for several different classes of devices are provided and the
+ * parameters can be configured from userspace via
+ * /sys/fs/cgroup/io.cost.model.
+ *
+ * If needed, tools/cgroup/iocost_coef_gen.py can be used to generate
+ * device-specific coefficients.
+ *
+ * 2. Control Strategy
+ *
+ * The device virtual time (vtime) is used as the primary control metric.
+ * The control strategy is composed of the following three parts.
+ *
+ * 2-1. Vtime Distribution
+ *
+ * When a cgroup becomes active in terms of IOs, its hierarchical share is
+ * calculated.  Please consider the following hierarchy where the numbers
+ * inside parentheses denote the configured weights.
+ *
+ *           root
+ *         /       \
+ *      A (w:100)  B (w:300)
+ *      /       \
+ *  A0 (w:100)  A1 (w:100)
+ *
+ * If B is idle and only A0 and A1 are actively issuing IOs, as the two are
+ * of equal weight, each gets 50% share.  If then B starts issuing IOs, B
+ * 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).
+ *
+ * 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)
+ * against the device vtime - an IO which takes 10ms on the underlying
+ * device is considered to take 80ms on A0.
+ *
+ * This constitutes the basis of IO capacity distribution.  Each cgroup's
+ * vtime is running at a rate determined by its hweight.  A cgroup tracks
+ * the vtime consumed by past IOs and can issue a new IO iff doing so
+ * wouldn't outrun the current device vtime.  Otherwise, the IO is
+ * suspended until the vtime has progressed enough to cover it.
+ *
+ * 2-2. Vrate Adjustment
+ *
+ * It's unrealistic to expect the cost model to be perfect.  There are too
+ * many devices and even on the same device the overall performance
+ * fluctuates depending on numerous factors such as IO mixture and device
+ * internal garbage collection.  The controller needs to adapt dynamically.
+ *
+ * This is achieved by adjusting the overall IO rate according to how busy
+ * the device is.  If the device becomes overloaded, we're sending down too
+ * many IOs and should generally slow down.  If there are waiting issuers
+ * but the device isn't saturated, we're issuing too few and should
+ * generally speed up.
+ *
+ * To slow down, we lower the vrate - the rate at which the device vtime
+ * passes compared to the wall clock.  For example, if the vtime is running
+ * at the vrate of 75%, all cgroups added up would only be able to issue
+ * 750ms worth of IOs per second, and vice-versa for speeding up.
+ *
+ * Device business is determined using two criteria - rq wait and
+ * completion latencies.
+ *
+ * When a device gets saturated, the on-device and then the request queues
+ * fill up and a bio which is ready to be issued has to wait for a request
+ * to become available.  When this delay becomes noticeable, it's a clear
+ * indication that the device is saturated and we lower the vrate.  This
+ * saturation signal is fairly conservative as it only triggers when both
+ * hardware and software queues are filled up, and is used as the default
+ * busy signal.
+ *
+ * As devices can have deep queues and be unfair in how the queued commands
+ * are executed, soley depending on rq wait may not result in satisfactory
+ * control quality.  For a better control quality, completion latency QoS
+ * parameters can be configured so that the device is considered saturated
+ * if N'th percentile completion latency rises above the set point.
+ *
+ * The completion latency requirements are a function of both the
+ * underlying device characteristics and the desired IO latency quality of
+ * service.  There is an inherent trade-off - the tighter the latency QoS,
+ * the higher the bandwidth lossage.  Latency QoS is disabled by default
+ * and can be set through /sys/fs/cgroup/io.cost.qos.
+ *
+ * 2-3. Work Conservation
+ *
+ * Imagine two cgroups A and B with equal weights.  A is issuing a small IO
+ * periodically while B is sending out enough parallel IOs to saturate the
+ * device on its own.  Let's say A's usage amounts to 100ms worth of IO
+ * cost per second, i.e., 10% of the device capacity.  The naive
+ * distribution of half and half would lead to 60% utilization of the
+ * device, a significant reduction in the total amount of work done
+ * compared to free-for-all competition.  This is too high a cost to pay
+ * for IO control.
+ *
+ * To conserve the total amount of work done, we keep track of how much
+ * each active cgroup is actually using and yield part of its weight if
+ * there are other cgroups which can make use of it.  In the above case,
+ * A's weight will be lowered so that it hovers above the actual usage and
+ * B would be able to use the rest.
+ *
+ * As we don't want to penalize a cgroup for donating its weight, the
+ * surplus weight adjustment factors in a margin and has an immediate
+ * snapback mechanism in case the cgroup needs more IO vtime for itself.
+ *
+ * Note that adjusting down surplus weights has the same effects as
+ * accelerating vtime for other cgroups and work conservation can also be
+ * implemented by adjusting vrate dynamically.  However, squaring who can
+ * donate and should take back how much requires hweight propagations
+ * anyway making it easier to implement and understand as a separate
+ * mechanism.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/timer.h>
+#include <linux/time64.h>
+#include <linux/parser.h>
+#include <linux/sched/signal.h>
+#include <linux/blk-cgroup.h>
+#include "blk-rq-qos.h"
+#include "blk-stat.h"
+#include "blk-wbt.h"
+
+#ifdef CONFIG_TRACEPOINTS
+
+/* copied from TRACE_CGROUP_PATH, see cgroup-internal.h */
+#define TRACE_IOCG_PATH_LEN 1024
+static DEFINE_SPINLOCK(trace_iocg_path_lock);
+static char trace_iocg_path[TRACE_IOCG_PATH_LEN];
+
+#define TRACE_IOCG_PATH(type, iocg, ...)					\
+	do {									\
+		unsigned long flags;						\
+		if (trace_iocost_##type##_enabled()) {				\
+			spin_lock_irqsave(&trace_iocg_path_lock, flags);	\
+			cgroup_path(iocg_to_blkg(iocg)->blkcg->css.cgroup,	\
+				    trace_iocg_path, TRACE_IOCG_PATH_LEN);	\
+			trace_iocost_##type(iocg, trace_iocg_path,		\
+					      ##__VA_ARGS__);			\
+			spin_unlock_irqrestore(&trace_iocg_path_lock, flags);	\
+		}								\
+	} while (0)
+
+#else	/* CONFIG_TRACE_POINTS */
+#define TRACE_IOCG_PATH(type, iocg, ...)	do { } while (0)
+#endif	/* CONFIG_TRACE_POINTS */
+
+enum {
+	MILLION			= 1000000,
+
+	/* timer period is calculated from latency requirements, bound it */
+	MIN_PERIOD		= USEC_PER_MSEC,
+	MAX_PERIOD		= USEC_PER_SEC,
+
+	/*
+	 * A cgroup's vtime can run 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,
+
+	/* Have some play in waitq timer operations */
+	WAITQ_TIMER_MARGIN_PCT	= 5,
+
+	/*
+	 * 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,
+
+	/* 1/64k is granular enough and can easily be handled w/ u32 */
+	HWEIGHT_WHOLE		= 1 << 16,
+
+	/*
+	 * As vtime is used to calculate the cost of each IO, it needs to
+	 * be fairly high precision.  For example, it should be able to
+	 * represent the cost of a single page worth of discard with
+	 * suffificient accuracy.  At the same time, it should be able to
+	 * represent reasonably long enough durations to be useful and
+	 * convenient during operation.
+	 *
+	 * 1s worth of vtime is 2^37.  This gives us both sub-nanosecond
+	 * granularity and days of wrap-around time even at extreme vrates.
+	 */
+	VTIME_PER_SEC_SHIFT	= 37,
+	VTIME_PER_SEC		= 1LLU << VTIME_PER_SEC_SHIFT,
+	VTIME_PER_USEC		= VTIME_PER_SEC / USEC_PER_SEC,
+
+	/* bound vrate adjustments within two orders of magnitude */
+	VRATE_MIN_PPM		= 10000,	/* 1% */
+	VRATE_MAX_PPM		= 100000000,	/* 10000% */
+
+	VRATE_MIN		= VTIME_PER_USEC * VRATE_MIN_PPM / MILLION,
+	VRATE_CLAMP_ADJ_PCT	= 4,
+
+	/* if IOs end up waiting for requests, issue less */
+	RQ_WAIT_BUSY_PCT	= 5,
+
+	/* 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.
+	 */
+	SURPLUS_SCALE_PCT	= 125,			/* * 125% */
+	SURPLUS_SCALE_ABS	= HWEIGHT_WHOLE / 50,	/* + 2% */
+	SURPLUS_MIN_ADJ_DELTA	= HWEIGHT_WHOLE / 33,	/* 3% */
+
+	/* switch iff the conditions are met for longer than this */
+	AUTOP_CYCLE_NSEC	= 10LLU * NSEC_PER_SEC,
+
+	/*
+	 * Count IO size in 4k pages.  The 12bit shift helps keeping
+	 * size-proportional components of cost calculation in closer
+	 * numbers of digits to per-IO cost components.
+	 */
+	IOC_PAGE_SHIFT		= 12,
+	IOC_PAGE_SIZE		= 1 << IOC_PAGE_SHIFT,
+	IOC_SECT_TO_PAGE_SHIFT	= IOC_PAGE_SHIFT - SECTOR_SHIFT,
+
+	/* if apart further than 16M, consider randio for linear model */
+	LCOEF_RANDIO_PAGES	= 4096,
+};
+
+enum ioc_running {
+	IOC_IDLE,
+	IOC_RUNNING,
+	IOC_STOP,
+};
+
+/* io.cost.qos controls including per-dev enable of the whole controller */
+enum {
+	QOS_ENABLE,
+	QOS_CTRL,
+	NR_QOS_CTRL_PARAMS,
+};
+
+/* io.cost.qos params */
+enum {
+	QOS_RPPM,
+	QOS_RLAT,
+	QOS_WPPM,
+	QOS_WLAT,
+	QOS_MIN,
+	QOS_MAX,
+	NR_QOS_PARAMS,
+};
+
+/* io.cost.model controls */
+enum {
+	COST_CTRL,
+	COST_MODEL,
+	NR_COST_CTRL_PARAMS,
+};
+
+/* builtin linear cost model coefficients */
+enum {
+	I_LCOEF_RBPS,
+	I_LCOEF_RSEQIOPS,
+	I_LCOEF_RRANDIOPS,
+	I_LCOEF_WBPS,
+	I_LCOEF_WSEQIOPS,
+	I_LCOEF_WRANDIOPS,
+	NR_I_LCOEFS,
+};
+
+enum {
+	LCOEF_RPAGE,
+	LCOEF_RSEQIO,
+	LCOEF_RRANDIO,
+	LCOEF_WPAGE,
+	LCOEF_WSEQIO,
+	LCOEF_WRANDIO,
+	NR_LCOEFS,
+};
+
+enum {
+	AUTOP_INVALID,
+	AUTOP_HDD,
+	AUTOP_SSD_QD1,
+	AUTOP_SSD_DFL,
+	AUTOP_SSD_FAST,
+};
+
+struct ioc_gq;
+
+struct ioc_params {
+	u32				qos[NR_QOS_PARAMS];
+	u64				i_lcoefs[NR_I_LCOEFS];
+	u64				lcoefs[NR_LCOEFS];
+	u32				too_fast_vrate_pct;
+	u32				too_slow_vrate_pct;
+};
+
+struct ioc_missed {
+	u32				nr_met;
+	u32				nr_missed;
+	u32				last_met;
+	u32				last_missed;
+};
+
+struct ioc_pcpu_stat {
+	struct ioc_missed		missed[2];
+
+	u64				rq_wait_ns;
+	u64				last_rq_wait_ns;
+};
+
+/* per device */
+struct ioc {
+	struct rq_qos			rqos;
+
+	bool				enabled;
+
+	struct ioc_params		params;
+	u32				period_us;
+	u32				margin_us;
+	u64				vrate_min;
+	u64				vrate_max;
+
+	spinlock_t			lock;
+	struct timer_list		timer;
+	struct list_head		active_iocgs;	/* active cgroups */
+	struct ioc_pcpu_stat __percpu	*pcpu_stat;
+
+	enum ioc_running		running;
+	atomic64_t			vtime_rate;
+
+	seqcount_t			period_seqcount;
+	u32				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 */
+
+	u64				autop_too_fast_at;
+	u64				autop_too_slow_at;
+	int				autop_idx;
+	bool				user_qos_params:1;
+	bool				user_cost_model:1;
+};
+
+/* per device-cgroup pair */
+struct ioc_gq {
+	struct blkg_policy_data		pd;
+	struct ioc			*ioc;
+
+	/*
+	 * A iocg can get its weight from two sources - an explicit
+	 * per-device-cgroup configuration or the default weight of the
+	 * cgroup.  `cfg_weight` is the explicit per-device-cgroup
+	 * configuration.  `weight` is the effective considering both
+	 * sources.
+	 *
+	 * When an idle cgroup becomes active its `active` goes from 0 to
+	 * `weight`.  `inuse` is the surplus adjusted active weight.
+	 * `active` and `inuse` are used to calculate `hweight_active` and
+	 * `hweight_inuse`.
+	 *
+	 * `last_inuse` remembers `inuse` while an iocg is idle to persist
+	 * surplus adjustments.
+	 */
+	u32				cfg_weight;
+	u32				weight;
+	u32				active;
+	u32				inuse;
+	u32				last_inuse;
+
+	sector_t			cursor;		/* to detect randio */
+
+	/*
+	 * `vtime` is this iocg's vtime cursor which progresses as IOs are
+	 * issued.  If lagging behind device vtime, the delta represents
+	 * the currently available IO budget.  If runnning ahead, the
+	 * overage.
+	 *
+	 * `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				last_vtime;
+
+	/*
+	 * The period this iocg was last active in.  Used for deactivation
+	 * and invalidating `vtime`.
+	 */
+	atomic64_t			active_period;
+	struct list_head		active_list;
+
+	/* see __propagate_active_weight() and current_hweight() for details */
+	u64				child_active_sum;
+	u64				child_inuse_sum;
+	int				hweight_gen;
+	u32				hweight_active;
+	u32				hweight_inuse;
+	bool				has_surplus;
+
+	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];
+
+	/* this iocg's depth in the hierarchy and ancestors including self */
+	int				level;
+	struct ioc_gq			*ancestors[];
+};
+
+/* per cgroup */
+struct ioc_cgrp {
+	struct blkcg_policy_data	cpd;
+	unsigned int			dfl_weight;
+};
+
+struct ioc_now {
+	u64				now_ns;
+	u32				now;
+	u64				vnow;
+	u64				vrate;
+};
+
+struct iocg_wait {
+	struct wait_queue_entry		wait;
+	struct bio			*bio;
+	u64				abs_cost;
+	bool				committed;
+};
+
+struct iocg_wake_ctx {
+	struct ioc_gq			*iocg;
+	u32				hw_inuse;
+	s64				vbudget;
+};
+
+static const struct ioc_params autop[] = {
+	[AUTOP_HDD] = {
+		.qos				= {
+			[QOS_RLAT]		=         50000, /* 50ms */
+			[QOS_WLAT]		=         50000,
+			[QOS_MIN]		= VRATE_MIN_PPM,
+			[QOS_MAX]		= VRATE_MAX_PPM,
+		},
+		.i_lcoefs			= {
+			[I_LCOEF_RBPS]		=     174019176,
+			[I_LCOEF_RSEQIOPS]	=         41708,
+			[I_LCOEF_RRANDIOPS]	=           370,
+			[I_LCOEF_WBPS]		=     178075866,
+			[I_LCOEF_WSEQIOPS]	=         42705,
+			[I_LCOEF_WRANDIOPS]	=           378,
+		},
+	},
+	[AUTOP_SSD_QD1] = {
+		.qos				= {
+			[QOS_RLAT]		=         25000, /* 25ms */
+			[QOS_WLAT]		=         25000,
+			[QOS_MIN]		= VRATE_MIN_PPM,
+			[QOS_MAX]		= VRATE_MAX_PPM,
+		},
+		.i_lcoefs			= {
+			[I_LCOEF_RBPS]		=     245855193,
+			[I_LCOEF_RSEQIOPS]	=         61575,
+			[I_LCOEF_RRANDIOPS]	=          6946,
+			[I_LCOEF_WBPS]		=     141365009,
+			[I_LCOEF_WSEQIOPS]	=         33716,
+			[I_LCOEF_WRANDIOPS]	=         26796,
+		},
+	},
+	[AUTOP_SSD_DFL] = {
+		.qos				= {
+			[QOS_RLAT]		=         25000, /* 25ms */
+			[QOS_WLAT]		=         25000,
+			[QOS_MIN]		= VRATE_MIN_PPM,
+			[QOS_MAX]		= VRATE_MAX_PPM,
+		},
+		.i_lcoefs			= {
+			[I_LCOEF_RBPS]		=     488636629,
+			[I_LCOEF_RSEQIOPS]	=          8932,
+			[I_LCOEF_RRANDIOPS]	=          8518,
+			[I_LCOEF_WBPS]		=     427891549,
+			[I_LCOEF_WSEQIOPS]	=         28755,
+			[I_LCOEF_WRANDIOPS]	=         21940,
+		},
+		.too_fast_vrate_pct		=           500,
+	},
+	[AUTOP_SSD_FAST] = {
+		.qos				= {
+			[QOS_RLAT]		=          5000, /* 5ms */
+			[QOS_WLAT]		=          5000,
+			[QOS_MIN]		= VRATE_MIN_PPM,
+			[QOS_MAX]		= VRATE_MAX_PPM,
+		},
+		.i_lcoefs			= {
+			[I_LCOEF_RBPS]		=    3102524156LLU,
+			[I_LCOEF_RSEQIOPS]	=        724816,
+			[I_LCOEF_RRANDIOPS]	=        778122,
+			[I_LCOEF_WBPS]		=    1742780862LLU,
+			[I_LCOEF_WSEQIOPS]	=        425702,
+			[I_LCOEF_WRANDIOPS]	=	 443193,
+		},
+		.too_slow_vrate_pct		=            10,
+	},
+};
+
+/*
+ * vrate adjust percentages indexed by ioc->busy_level.  We adjust up on
+ * vtime credit shortage and down on device saturation.
+ */
+static u32 vrate_adj_pct[] =
+	{ 0, 0, 0, 0,
+	  1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+	  2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
+	  4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8, 8, 8, 8, 16 };
+
+static struct blkcg_policy blkcg_policy_iocost;
+
+/* accessors and helpers */
+static struct ioc *rqos_to_ioc(struct rq_qos *rqos)
+{
+	return container_of(rqos, struct ioc, rqos);
+}
+
+static struct ioc *q_to_ioc(struct request_queue *q)
+{
+	return rqos_to_ioc(rq_qos_id(q, RQ_QOS_COST));
+}
+
+static const char *q_name(struct request_queue *q)
+{
+	if (test_bit(QUEUE_FLAG_REGISTERED, &q->queue_flags))
+		return kobject_name(q->kobj.parent);
+	else
+		return "<unknown>";
+}
+
+static const char __maybe_unused *ioc_name(struct ioc *ioc)
+{
+	return q_name(ioc->rqos.q);
+}
+
+static struct ioc_gq *pd_to_iocg(struct blkg_policy_data *pd)
+{
+	return pd ? container_of(pd, struct ioc_gq, pd) : NULL;
+}
+
+static struct ioc_gq *blkg_to_iocg(struct blkcg_gq *blkg)
+{
+	return pd_to_iocg(blkg_to_pd(blkg, &blkcg_policy_iocost));
+}
+
+static struct blkcg_gq *iocg_to_blkg(struct ioc_gq *iocg)
+{
+	return pd_to_blkg(&iocg->pd);
+}
+
+static struct ioc_cgrp *blkcg_to_iocc(struct blkcg *blkcg)
+{
+	return container_of(blkcg_to_cpd(blkcg, &blkcg_policy_iocost),
+			    struct ioc_cgrp, cpd);
+}
+
+/*
+ * Scale @abs_cost to the inverse of @hw_inuse.  The lower the hierarchical
+ * weight, the more expensive each IO.
+ */
+static u64 abs_cost_to_cost(u64 abs_cost, u32 hw_inuse)
+{
+	return DIV64_U64_ROUND_UP(abs_cost * HWEIGHT_WHOLE, hw_inuse);
+}
+
+static void iocg_commit_bio(struct ioc_gq *iocg, struct bio *bio, u64 cost)
+{
+	bio->bi_iocost_cost = cost;
+	atomic64_add(cost, &iocg->vtime);
+}
+
+#define CREATE_TRACE_POINTS
+#include <trace/events/iocost.h>
+
+/* latency Qos params changed, update period_us and all the dependent params */
+static void ioc_refresh_period_us(struct ioc *ioc)
+{
+	u32 ppm, lat, multi, period_us;
+
+	lockdep_assert_held(&ioc->lock);
+
+	/* pick the higher latency target */
+	if (ioc->params.qos[QOS_RLAT] >= ioc->params.qos[QOS_WLAT]) {
+		ppm = ioc->params.qos[QOS_RPPM];
+		lat = ioc->params.qos[QOS_RLAT];
+	} else {
+		ppm = ioc->params.qos[QOS_WPPM];
+		lat = ioc->params.qos[QOS_WLAT];
+	}
+
+	/*
+	 * We want the period to be long enough to contain a healthy number
+	 * of IOs while short enough for granular control.  Define it as a
+	 * multiple of the latency target.  Ideally, the multiplier should
+	 * be scaled according to the percentile so that it would nominally
+	 * contain a certain number of requests.  Let's be simpler and
+	 * scale it linearly so that it's 2x >= pct(90) and 10x at pct(50).
+	 */
+	if (ppm)
+		multi = max_t(u32, (MILLION - ppm) / 50000, 2);
+	else
+		multi = 2;
+	period_us = multi * lat;
+	period_us = clamp_t(u32, period_us, MIN_PERIOD, MAX_PERIOD);
+
+	/* 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);
+}
+
+static int ioc_autop_idx(struct ioc *ioc)
+{
+	int idx = ioc->autop_idx;
+	const struct ioc_params *p = &autop[idx];
+	u32 vrate_pct;
+	u64 now_ns;
+
+	/* rotational? */
+	if (!blk_queue_nonrot(ioc->rqos.q))
+		return AUTOP_HDD;
+
+	/* handle SATA SSDs w/ broken NCQ */
+	if (blk_queue_depth(ioc->rqos.q) == 1)
+		return AUTOP_SSD_QD1;
+
+	/* use one of the normal ssd sets */
+	if (idx < AUTOP_SSD_DFL)
+		return AUTOP_SSD_DFL;
+
+	/* if user is overriding anything, maintain what was there */
+	if (ioc->user_qos_params || ioc->user_cost_model)
+		return idx;
+
+	/* step up/down based on the vrate */
+	vrate_pct = div64_u64(atomic64_read(&ioc->vtime_rate) * 100,
+			      VTIME_PER_USEC);
+	now_ns = ktime_get_ns();
+
+	if (p->too_fast_vrate_pct && p->too_fast_vrate_pct <= vrate_pct) {
+		if (!ioc->autop_too_fast_at)
+			ioc->autop_too_fast_at = now_ns;
+		if (now_ns - ioc->autop_too_fast_at >= AUTOP_CYCLE_NSEC)
+			return idx + 1;
+	} else {
+		ioc->autop_too_fast_at = 0;
+	}
+
+	if (p->too_slow_vrate_pct && p->too_slow_vrate_pct >= vrate_pct) {
+		if (!ioc->autop_too_slow_at)
+			ioc->autop_too_slow_at = now_ns;
+		if (now_ns - ioc->autop_too_slow_at >= AUTOP_CYCLE_NSEC)
+			return idx - 1;
+	} else {
+		ioc->autop_too_slow_at = 0;
+	}
+
+	return idx;
+}
+
+/*
+ * Take the followings as input
+ *
+ *  @bps	maximum sequential throughput
+ *  @seqiops	maximum sequential 4k iops
+ *  @randiops	maximum random 4k iops
+ *
+ * and calculate the linear model cost coefficients.
+ *
+ *  *@page	per-page cost		1s / (@bps / 4096)
+ *  *@seqio	base cost of a seq IO	max((1s / @seqiops) - *@page, 0)
+ *  @randiops	base cost of a rand IO	max((1s / @randiops) - *@page, 0)
+ */
+static void calc_lcoefs(u64 bps, u64 seqiops, u64 randiops,
+			u64 *page, u64 *seqio, u64 *randio)
+{
+	u64 v;
+
+	*page = *seqio = *randio = 0;
+
+	if (bps)
+		*page = DIV64_U64_ROUND_UP(VTIME_PER_SEC,
+					   DIV_ROUND_UP_ULL(bps, IOC_PAGE_SIZE));
+
+	if (seqiops) {
+		v = DIV64_U64_ROUND_UP(VTIME_PER_SEC, seqiops);
+		if (v > *page)
+			*seqio = v - *page;
+	}
+
+	if (randiops) {
+		v = DIV64_U64_ROUND_UP(VTIME_PER_SEC, randiops);
+		if (v > *page)
+			*randio = v - *page;
+	}
+}
+
+static void ioc_refresh_lcoefs(struct ioc *ioc)
+{
+	u64 *u = ioc->params.i_lcoefs;
+	u64 *c = ioc->params.lcoefs;
+
+	calc_lcoefs(u[I_LCOEF_RBPS], u[I_LCOEF_RSEQIOPS], u[I_LCOEF_RRANDIOPS],
+		    &c[LCOEF_RPAGE], &c[LCOEF_RSEQIO], &c[LCOEF_RRANDIO]);
+	calc_lcoefs(u[I_LCOEF_WBPS], u[I_LCOEF_WSEQIOPS], u[I_LCOEF_WRANDIOPS],
+		    &c[LCOEF_WPAGE], &c[LCOEF_WSEQIO], &c[LCOEF_WRANDIO]);
+}
+
+static bool ioc_refresh_params(struct ioc *ioc, bool force)
+{
+	const struct ioc_params *p;
+	int idx;
+
+	lockdep_assert_held(&ioc->lock);
+
+	idx = ioc_autop_idx(ioc);
+	p = &autop[idx];
+
+	if (idx == ioc->autop_idx && !force)
+		return false;
+
+	if (idx != ioc->autop_idx)
+		atomic64_set(&ioc->vtime_rate, VTIME_PER_USEC);
+
+	ioc->autop_idx = idx;
+	ioc->autop_too_fast_at = 0;
+	ioc->autop_too_slow_at = 0;
+
+	if (!ioc->user_qos_params)
+		memcpy(ioc->params.qos, p->qos, sizeof(p->qos));
+	if (!ioc->user_cost_model)
+		memcpy(ioc->params.i_lcoefs, p->i_lcoefs, sizeof(p->i_lcoefs));
+
+	ioc_refresh_period_us(ioc);
+	ioc_refresh_lcoefs(ioc);
+
+	ioc->vrate_min = DIV64_U64_ROUND_UP((u64)ioc->params.qos[QOS_MIN] *
+					    VTIME_PER_USEC, MILLION);
+	ioc->vrate_max = div64_u64((u64)ioc->params.qos[QOS_MAX] *
+				   VTIME_PER_USEC, MILLION);
+
+	return true;
+}
+
+/* take a snapshot of the current [v]time and vrate */
+static void ioc_now(struct ioc *ioc, struct ioc_now *now)
+{
+	unsigned seq;
+
+	now->now_ns = ktime_get();
+	now->now = ktime_to_us(now->now_ns);
+	now->vrate = atomic64_read(&ioc->vtime_rate);
+
+	/*
+	 * The current vtime is
+	 *
+	 *   vtime at period start + (wallclock time since the start) * vrate
+	 *
+	 * As a consistent snapshot of `period_at_vtime` and `period_at` is
+	 * needed, they're seqcount protected.
+	 */
+	do {
+		seq = read_seqcount_begin(&ioc->period_seqcount);
+		now->vnow = ioc->period_at_vtime +
+			(now->now - ioc->period_at) * now->vrate;
+	} while (read_seqcount_retry(&ioc->period_seqcount, seq));
+}
+
+static void ioc_start_period(struct ioc *ioc, struct ioc_now *now)
+{
+	lockdep_assert_held(&ioc->lock);
+	WARN_ON_ONCE(ioc->running != IOC_RUNNING);
+
+	write_seqcount_begin(&ioc->period_seqcount);
+	ioc->period_at = now->now;
+	ioc->period_at_vtime = now->vnow;
+	write_seqcount_end(&ioc->period_seqcount);
+
+	ioc->timer.expires = jiffies + usecs_to_jiffies(ioc->period_us);
+	add_timer(&ioc->timer);
+}
+
+/*
+ * Update @iocg's `active` and `inuse` to @active and @inuse, update level
+ * weight sums and propagate upwards accordingly.
+ */
+static void __propagate_active_weight(struct ioc_gq *iocg, u32 active, u32 inuse)
+{
+	struct ioc *ioc = iocg->ioc;
+	int lvl;
+
+	lockdep_assert_held(&ioc->lock);
+
+	inuse = min(active, inuse);
+
+	for (lvl = iocg->level - 1; lvl >= 0; lvl--) {
+		struct ioc_gq *parent = iocg->ancestors[lvl];
+		struct ioc_gq *child = iocg->ancestors[lvl + 1];
+		u32 parent_active = 0, parent_inuse = 0;
+
+		/* update the level sums */
+		parent->child_active_sum += (s32)(active - child->active);
+		parent->child_inuse_sum += (s32)(inuse - child->inuse);
+		/* apply the udpates */
+		child->active = active;
+		child->inuse = inuse;
+
+		/*
+		 * The delta between inuse and active sums indicates that
+		 * that much of weight is being given away.  Parent's inuse
+		 * and active should reflect the ratio.
+		 */
+		if (parent->child_active_sum) {
+			parent_active = parent->weight;
+			parent_inuse = DIV64_U64_ROUND_UP(
+				parent_active * parent->child_inuse_sum,
+				parent->child_active_sum);
+		}
+
+		/* do we need to keep walking up? */
+		if (parent_active == parent->active &&
+		    parent_inuse == parent->inuse)
+			break;
+
+		active = parent_active;
+		inuse = parent_inuse;
+	}
+
+	ioc->weights_updated = true;
+}
+
+static void commit_active_weights(struct ioc *ioc)
+{
+	lockdep_assert_held(&ioc->lock);
+
+	if (ioc->weights_updated) {
+		/* paired with rmb in current_hweight(), see there */
+		smp_wmb();
+		atomic_inc(&ioc->hweight_gen);
+		ioc->weights_updated = false;
+	}
+}
+
+static void propagate_active_weight(struct ioc_gq *iocg, u32 active, u32 inuse)
+{
+	__propagate_active_weight(iocg, active, inuse);
+	commit_active_weights(iocg->ioc);
+}
+
+static void current_hweight(struct ioc_gq *iocg, u32 *hw_activep, u32 *hw_inusep)
+{
+	struct ioc *ioc = iocg->ioc;
+	int lvl;
+	u32 hwa, hwi;
+	int ioc_gen;
+
+	/* hot path - if uptodate, use cached */
+	ioc_gen = atomic_read(&ioc->hweight_gen);
+	if (ioc_gen == iocg->hweight_gen)
+		goto out;
+
+	/*
+	 * Paired with wmb in commit_active_weights().  If we saw the
+	 * updated hweight_gen, all the weight updates from
+	 * __propagate_active_weight() are visible too.
+	 *
+	 * We can race with weight updates during calculation and get it
+	 * wrong.  However, hweight_gen would have changed and a future
+	 * reader will recalculate and we're guaranteed to discard the
+	 * wrong result soon.
+	 */
+	smp_rmb();
+
+	hwa = hwi = HWEIGHT_WHOLE;
+	for (lvl = 0; lvl <= iocg->level - 1; lvl++) {
+		struct ioc_gq *parent = iocg->ancestors[lvl];
+		struct ioc_gq *child = iocg->ancestors[lvl + 1];
+		u32 active_sum = READ_ONCE(parent->child_active_sum);
+		u32 inuse_sum = READ_ONCE(parent->child_inuse_sum);
+		u32 active = READ_ONCE(child->active);
+		u32 inuse = READ_ONCE(child->inuse);
+
+		/* we can race with deactivations and either may read as zero */
+		if (!active_sum || !inuse_sum)
+			continue;
+
+		active_sum = max(active, active_sum);
+		hwa = hwa * active / active_sum;	/* max 16bits * 10000 */
+
+		inuse_sum = max(inuse, inuse_sum);
+		hwi = hwi * inuse / inuse_sum;		/* max 16bits * 10000 */
+	}
+
+	iocg->hweight_active = max_t(u32, hwa, 1);
+	iocg->hweight_inuse = max_t(u32, hwi, 1);
+	iocg->hweight_gen = ioc_gen;
+out:
+	if (hw_activep)
+		*hw_activep = iocg->hweight_active;
+	if (hw_inusep)
+		*hw_inusep = iocg->hweight_inuse;
+}
+
+static void weight_updated(struct ioc_gq *iocg)
+{
+	struct ioc *ioc = iocg->ioc;
+	struct blkcg_gq *blkg = iocg_to_blkg(iocg);
+	struct ioc_cgrp *iocc = blkcg_to_iocc(blkg->blkcg);
+	u32 weight;
+
+	lockdep_assert_held(&ioc->lock);
+
+	weight = iocg->cfg_weight ?: iocc->dfl_weight;
+	if (weight != iocg->weight && iocg->active)
+		propagate_active_weight(iocg, weight,
+			DIV64_U64_ROUND_UP(iocg->inuse * weight, iocg->weight));
+	iocg->weight = weight;
+}
+
+static bool iocg_activate(struct ioc_gq *iocg, struct ioc_now *now)
+{
+	struct ioc *ioc = iocg->ioc;
+	u64 last_period, cur_period, max_period_delta;
+	u64 vtime, vmargin, vmin;
+	int i;
+
+	/*
+	 * If seem to be already active, just update the stamp to tell the
+	 * timer that we're still active.  We don't mind occassional races.
+	 */
+	if (!list_empty(&iocg->active_list)) {
+		ioc_now(ioc, now);
+		cur_period = atomic64_read(&ioc->cur_period);
+		if (atomic64_read(&iocg->active_period) != cur_period)
+			atomic64_set(&iocg->active_period, cur_period);
+		return true;
+	}
+
+	/* racy check on internal node IOs, treat as root level IOs */
+	if (iocg->child_active_sum)
+		return false;
+
+	spin_lock_irq(&ioc->lock);
+
+	ioc_now(ioc, now);
+
+	/* update period */
+	cur_period = atomic64_read(&ioc->cur_period);
+	last_period = atomic64_read(&iocg->active_period);
+	atomic64_set(&iocg->active_period, cur_period);
+
+	/* already activated or breaking leaf-only constraint? */
+	for (i = iocg->level; i > 0; i--)
+		if (!list_empty(&iocg->active_list))
+			goto fail_unlock;
+	if (iocg->child_active_sum)
+		goto fail_unlock;
+
+	/*
+	 * vtime may wrap when vrate is raised substantially due to
+	 * underestimated IO costs.  Look at the period and ignore its
+	 * vtime if the iocg has been idle for too long.  Also, cap the
+	 * budget it can start with to the margin.
+	 */
+	max_period_delta = DIV64_U64_ROUND_UP(VTIME_VALID_DUR, ioc->period_us);
+	vtime = atomic64_read(&iocg->vtime);
+	vmargin = ioc->margin_us * now->vrate;
+	vmin = now->vnow - vmargin;
+
+	if (last_period + max_period_delta < cur_period ||
+	    time_before64(vtime, vmin)) {
+		atomic64_add(vmin - vtime, &iocg->vtime);
+		atomic64_add(vmin - vtime, &iocg->done_vtime);
+		vtime = vmin;
+	}
+
+	/*
+	 * Activate, propagate weight and start period timer if not
+	 * running.  Reset hweight_gen to avoid accidental match from
+	 * wrapping.
+	 */
+	iocg->hweight_gen = atomic_read(&ioc->hweight_gen) - 1;
+	list_add(&iocg->active_list, &ioc->active_iocgs);
+	propagate_active_weight(iocg, iocg->weight,
+				iocg->last_inuse ?: iocg->weight);
+
+	TRACE_IOCG_PATH(iocg_activate, iocg, now,
+			last_period, cur_period, vtime);
+
+	iocg->last_vtime = vtime;
+
+	if (ioc->running == IOC_IDLE) {
+		ioc->running = IOC_RUNNING;
+		ioc_start_period(ioc, now);
+	}
+
+	spin_unlock_irq(&ioc->lock);
+	return true;
+
+fail_unlock:
+	spin_unlock_irq(&ioc->lock);
+	return false;
+}
+
+static int iocg_wake_fn(struct wait_queue_entry *wq_entry, unsigned mode,
+			int flags, void *key)
+{
+	struct iocg_wait *wait = container_of(wq_entry, struct iocg_wait, wait);
+	struct iocg_wake_ctx *ctx = (struct iocg_wake_ctx *)key;
+	u64 cost = abs_cost_to_cost(wait->abs_cost, ctx->hw_inuse);
+
+	ctx->vbudget -= cost;
+
+	if (ctx->vbudget < 0)
+		return -1;
+
+	iocg_commit_bio(ctx->iocg, wait->bio, cost);
+
+	/*
+	 * autoremove_wake_function() removes the wait entry only when it
+	 * actually changed the task state.  We want the wait always
+	 * removed.  Remove explicitly and use default_wake_function().
+	 */
+	list_del_init(&wq_entry->entry);
+	wait->committed = true;
+
+	default_wake_function(wq_entry, mode, flags, key);
+	return 0;
+}
+
+static void iocg_kick_waitq(struct ioc_gq *iocg, struct ioc_now *now)
+{
+	struct ioc *ioc = iocg->ioc;
+	struct iocg_wake_ctx ctx = { .iocg = iocg };
+	u64 margin_ns = (u64)(ioc->period_us *
+			      WAITQ_TIMER_MARGIN_PCT / 100) * NSEC_PER_USEC;
+	u64 vshortage, expires, oexpires;
+
+	lockdep_assert_held(&iocg->waitq.lock);
+
+	/*
+	 * Wake up the ones which are due and see how much vtime we'll need
+	 * for the next one.
+	 */
+	current_hweight(iocg, NULL, &ctx.hw_inuse);
+	ctx.vbudget = now->vnow - atomic64_read(&iocg->vtime);
+	__wake_up_locked_key(&iocg->waitq, TASK_NORMAL, &ctx);
+	if (!waitqueue_active(&iocg->waitq))
+		return;
+	if (WARN_ON_ONCE(ctx.vbudget >= 0))
+		return;
+
+	/* determine next wakeup, add a quarter margin to guarantee chunking */
+	vshortage = -ctx.vbudget;
+	expires = now->now_ns +
+		DIV64_U64_ROUND_UP(vshortage, now->vrate) * NSEC_PER_USEC;
+	expires += margin_ns / 4;
+
+	/* if already active and close enough, don't bother */
+	oexpires = ktime_to_ns(hrtimer_get_softexpires(&iocg->waitq_timer));
+	if (hrtimer_is_queued(&iocg->waitq_timer) &&
+	    abs(oexpires - expires) <= margin_ns / 4)
+		return;
+
+	hrtimer_start_range_ns(&iocg->waitq_timer, ns_to_ktime(expires),
+			       margin_ns / 4, HRTIMER_MODE_ABS);
+}
+
+static enum hrtimer_restart iocg_waitq_timer_fn(struct hrtimer *timer)
+{
+	struct ioc_gq *iocg = container_of(timer, struct ioc_gq, waitq_timer);
+	struct ioc_now now;
+	unsigned long flags;
+
+	ioc_now(iocg->ioc, &now);
+
+	spin_lock_irqsave(&iocg->waitq.lock, flags);
+	iocg_kick_waitq(iocg, &now);
+	spin_unlock_irqrestore(&iocg->waitq.lock, flags);
+
+	return HRTIMER_NORESTART;
+}
+
+static void iocg_kick_delay(struct ioc_gq *iocg, struct ioc_now *now, u64 cost)
+{
+	struct ioc *ioc = iocg->ioc;
+	struct blkcg_gq *blkg = iocg_to_blkg(iocg);
+	u64 vtime = atomic64_read(&iocg->vtime);
+	u64 vmargin = ioc->margin_us * now->vrate;
+	u64 margin_ns = ioc->margin_us * NSEC_PER_USEC;
+	u64 expires, oexpires;
+
+	/* clear or maintain depending on the overage */
+	if (time_before_eq64(vtime, now->vnow)) {
+		blkcg_clear_delay(blkg);
+		return;
+	}
+	if (!atomic_read(&blkg->use_delay) &&
+	    time_before_eq64(vtime, now->vnow + vmargin))
+		return;
+
+	/* use delay */
+	if (cost) {
+		u64 cost_ns = DIV64_U64_ROUND_UP(cost * NSEC_PER_USEC,
+						 now->vrate);
+		blkcg_add_delay(blkg, now->now_ns, cost_ns);
+	}
+	blkcg_use_delay(blkg);
+
+	expires = now->now_ns + DIV64_U64_ROUND_UP(vtime - now->vnow,
+						   now->vrate) * NSEC_PER_USEC;
+
+	/* if already active and close enough, don't bother */
+	oexpires = ktime_to_ns(hrtimer_get_softexpires(&iocg->delay_timer));
+	if (hrtimer_is_queued(&iocg->delay_timer) &&
+	    abs(oexpires - expires) <= margin_ns / 4)
+		return;
+
+	hrtimer_start_range_ns(&iocg->delay_timer, ns_to_ktime(expires),
+			       margin_ns / 4, HRTIMER_MODE_ABS);
+}
+
+static enum hrtimer_restart iocg_delay_timer_fn(struct hrtimer *timer)
+{
+	struct ioc_gq *iocg = container_of(timer, struct ioc_gq, delay_timer);
+	struct ioc_now now;
+
+	ioc_now(iocg->ioc, &now);
+	iocg_kick_delay(iocg, &now, 0);
+
+	return HRTIMER_NORESTART;
+}
+
+static void ioc_lat_stat(struct ioc *ioc, u32 *missed_ppm_ar, u32 *rq_wait_pct_p)
+{
+	u32 nr_met[2] = { };
+	u32 nr_missed[2] = { };
+	u64 rq_wait_ns = 0;
+	int cpu, rw;
+
+	for_each_online_cpu(cpu) {
+		struct ioc_pcpu_stat *stat = per_cpu_ptr(ioc->pcpu_stat, cpu);
+		u64 this_rq_wait_ns;
+
+		for (rw = READ; rw <= WRITE; rw++) {
+			u32 this_met = READ_ONCE(stat->missed[rw].nr_met);
+			u32 this_missed = READ_ONCE(stat->missed[rw].nr_missed);
+
+			nr_met[rw] += this_met - stat->missed[rw].last_met;
+			nr_missed[rw] += this_missed - stat->missed[rw].last_missed;
+			stat->missed[rw].last_met = this_met;
+			stat->missed[rw].last_missed = this_missed;
+		}
+
+		this_rq_wait_ns = READ_ONCE(stat->rq_wait_ns);
+		rq_wait_ns += this_rq_wait_ns - stat->last_rq_wait_ns;
+		stat->last_rq_wait_ns = this_rq_wait_ns;
+	}
+
+	for (rw = READ; rw <= WRITE; rw++) {
+		if (nr_met[rw] + nr_missed[rw])
+			missed_ppm_ar[rw] =
+				DIV64_U64_ROUND_UP((u64)nr_missed[rw] * MILLION,
+						   nr_met[rw] + nr_missed[rw]);
+		else
+			missed_ppm_ar[rw] = 0;
+	}
+
+	*rq_wait_pct_p = div64_u64(rq_wait_ns * 100,
+				   ioc->period_us * NSEC_PER_USEC);
+}
+
+/* was iocg idle this period? */
+static bool iocg_is_idle(struct ioc_gq *iocg)
+{
+	struct ioc *ioc = iocg->ioc;
+
+	/* did something get issued this period? */
+	if (atomic64_read(&iocg->active_period) ==
+	    atomic64_read(&ioc->cur_period))
+		return false;
+
+	/* is something in flight? */
+	if (atomic64_read(&iocg->done_vtime) < atomic64_read(&iocg->vtime))
+		return false;
+
+	return true;
+}
+
+/* returns usage with margin added if surplus is large enough */
+static u32 surplus_adjusted_hweight_inuse(u32 usage, u32 hw_inuse)
+{
+	/* add margin */
+	usage = DIV_ROUND_UP(usage * SURPLUS_SCALE_PCT, 100);
+	usage += SURPLUS_SCALE_ABS;
+
+	/* don't bother if the surplus is too small */
+	if (usage + SURPLUS_MIN_ADJ_DELTA > hw_inuse)
+		return 0;
+
+	return usage;
+}
+
+static void ioc_timer_fn(struct timer_list *timer)
+{
+	struct ioc *ioc = container_of(timer, struct ioc, timer);
+	struct ioc_gq *iocg, *tiocg;
+	struct ioc_now now;
+	int nr_surpluses = 0, nr_shortages = 0, nr_lagging = 0;
+	u32 ppm_rthr = MILLION - ioc->params.qos[QOS_RPPM];
+	u32 ppm_wthr = MILLION - ioc->params.qos[QOS_WPPM];
+	u32 missed_ppm[2], rq_wait_pct;
+	u64 period_vtime;
+	int i;
+
+	/* how were the latencies during the period? */
+	ioc_lat_stat(ioc, missed_ppm, &rq_wait_pct);
+
+	/* take care of active iocgs */
+	spin_lock_irq(&ioc->lock);
+
+	ioc_now(ioc, &now);
+
+	period_vtime = now.vnow - ioc->period_at_vtime;
+	if (WARN_ON_ONCE(!period_vtime)) {
+		spin_unlock_irq(&ioc->lock);
+		return;
+	}
+
+	/*
+	 * Waiters determine the sleep durations based on the vrate they
+	 * saw at the time of sleep.  If vrate has increased, some waiters
+	 * could be sleeping for too long.  Wake up tardy waiters which
+	 * should have woken up in the last period and expire idle iocgs.
+	 */
+	list_for_each_entry_safe(iocg, tiocg, &ioc->active_iocgs, active_list) {
+		if (!waitqueue_active(&iocg->waitq) && !iocg_is_idle(iocg))
+			continue;
+
+		spin_lock(&iocg->waitq.lock);
+
+		if (waitqueue_active(&iocg->waitq)) {
+			/* might be oversleeping vtime / hweight changes, kick */
+			iocg_kick_waitq(iocg, &now);
+			iocg_kick_delay(iocg, &now, 0);
+		} else if (iocg_is_idle(iocg)) {
+			/* no waiter and idle, deactivate */
+			iocg->last_inuse = iocg->inuse;
+			__propagate_active_weight(iocg, 0, 0);
+			list_del_init(&iocg->active_list);
+		}
+
+		spin_unlock(&iocg->waitq.lock);
+	}
+	commit_active_weights(ioc);
+
+	/* calc usages and see whether some weights need to be moved around */
+	list_for_each_entry(iocg, &ioc->active_iocgs, active_list) {
+		u64 vdone, vtime, vusage, vmargin, vmin;
+		u32 hw_active, hw_inuse, usage;
+
+		/*
+		 * Collect unused and wind vtime closer to vnow to prevent
+		 * iocgs from accumulating a large amount of budget.
+		 */
+		vdone = atomic64_read(&iocg->done_vtime);
+		vtime = atomic64_read(&iocg->vtime);
+		current_hweight(iocg, &hw_active, &hw_inuse);
+
+		/*
+		 * Latency QoS detection doesn't account for IOs which are
+		 * in-flight for longer than a period.  Detect them by
+		 * comparing vdone against period start.  If lagging behind
+		 * IOs from past periods, don't increase vrate.
+		 */
+		if (!atomic_read(&iocg_to_blkg(iocg)->use_delay) &&
+		    time_after64(vtime, vdone) &&
+		    time_after64(vtime, now.vnow -
+				 MAX_LAGGING_PERIODS * period_vtime) &&
+		    time_before64(vdone, now.vnow - period_vtime))
+			nr_lagging++;
+
+		if (waitqueue_active(&iocg->waitq))
+			vusage = now.vnow - iocg->last_vtime;
+		else if (time_before64(iocg->last_vtime, vtime))
+			vusage = vtime - iocg->last_vtime;
+		else
+			vusage = 0;
+
+		iocg->last_vtime += vusage;
+		/*
+		 * Factor in in-flight vtime into vusage to avoid
+		 * high-latency completions appearing as idle.  This should
+		 * be done after the above ->last_time adjustment.
+		 */
+		vusage = max(vusage, vtime - vdone);
+
+		/* calculate hweight based usage ratio and record */
+		if (vusage) {
+			usage = DIV64_U64_ROUND_UP(vusage * hw_inuse,
+						   period_vtime);
+			iocg->usage_idx = (iocg->usage_idx + 1) % NR_USAGE_SLOTS;
+			iocg->usages[iocg->usage_idx] = usage;
+		} else {
+			usage = 0;
+		}
+
+		/* see whether there's surplus vtime */
+		vmargin = ioc->margin_us * now.vrate;
+		vmin = now.vnow - vmargin;
+
+		iocg->has_surplus = false;
+
+		if (!waitqueue_active(&iocg->waitq) &&
+		    time_before64(vtime, vmin)) {
+			u64 delta = vmin - vtime;
+
+			/* throw away surplus vtime */
+			atomic64_add(delta, &iocg->vtime);
+			atomic64_add(delta, &iocg->done_vtime);
+			iocg->last_vtime += delta;
+			/* if usage is sufficiently low, maybe it can donate */
+			if (surplus_adjusted_hweight_inuse(usage, hw_inuse)) {
+				iocg->has_surplus = true;
+				nr_surpluses++;
+			}
+		} else if (hw_inuse < hw_active) {
+			u32 new_hwi, new_inuse;
+
+			/* was donating but might need to take back some */
+			if (waitqueue_active(&iocg->waitq)) {
+				new_hwi = hw_active;
+			} else {
+				new_hwi = max(hw_inuse,
+					      usage * SURPLUS_SCALE_PCT / 100 +
+					      SURPLUS_SCALE_ABS);
+			}
+
+			new_inuse = div64_u64((u64)iocg->inuse * new_hwi,
+					      hw_inuse);
+			new_inuse = clamp_t(u32, new_inuse, 1, iocg->active);
+
+			if (new_inuse > iocg->inuse) {
+				TRACE_IOCG_PATH(inuse_takeback, iocg, &now,
+						iocg->inuse, new_inuse,
+						hw_inuse, new_hwi);
+				__propagate_active_weight(iocg, iocg->weight,
+							  new_inuse);
+			}
+		} else {
+			/* genuninely out of vtime */
+			nr_shortages++;
+		}
+	}
+
+	if (!nr_shortages || !nr_surpluses)
+		goto skip_surplus_transfers;
+
+	/* there are both shortages and surpluses, transfer surpluses */
+	list_for_each_entry(iocg, &ioc->active_iocgs, active_list) {
+		u32 usage, hw_active, hw_inuse, new_hwi, new_inuse;
+		int nr_valid = 0;
+
+		if (!iocg->has_surplus)
+			continue;
+
+		/* base the decision on max historical usage */
+		for (i = 0, usage = 0; i < NR_USAGE_SLOTS; i++) {
+			if (iocg->usages[i]) {
+				usage = max(usage, iocg->usages[i]);
+				nr_valid++;
+			}
+		}
+		if (nr_valid < MIN_VALID_USAGES)
+			continue;
+
+		current_hweight(iocg, &hw_active, &hw_inuse);
+		new_hwi = surplus_adjusted_hweight_inuse(usage, hw_inuse);
+		if (!new_hwi)
+			continue;
+
+		new_inuse = DIV64_U64_ROUND_UP((u64)iocg->inuse * new_hwi,
+					       hw_inuse);
+		if (new_inuse < iocg->inuse) {
+			TRACE_IOCG_PATH(inuse_giveaway, iocg, &now,
+					iocg->inuse, new_inuse,
+					hw_inuse, new_hwi);
+			__propagate_active_weight(iocg, iocg->weight, new_inuse);
+		}
+	}
+skip_surplus_transfers:
+	commit_active_weights(ioc);
+
+	/*
+	 * If q is getting clogged or we're missing too much, we're issuing
+	 * too much IO and should lower vtime rate.  If we're not missing
+	 * and experiencing shortages but not surpluses, we're too stingy
+	 * and should increase vtime rate.
+	 */
+	if (rq_wait_pct > RQ_WAIT_BUSY_PCT ||
+	    missed_ppm[READ] > ppm_rthr ||
+	    missed_ppm[WRITE] > ppm_wthr) {
+		ioc->busy_level = max(ioc->busy_level, 0);
+		ioc->busy_level++;
+	} else if (nr_lagging) {
+		ioc->busy_level = max(ioc->busy_level, 0);
+	} else if (nr_shortages && !nr_surpluses &&
+		   rq_wait_pct <= RQ_WAIT_BUSY_PCT * UNBUSY_THR_PCT / 100 &&
+		   missed_ppm[READ] <= ppm_rthr * UNBUSY_THR_PCT / 100 &&
+		   missed_ppm[WRITE] <= ppm_wthr * UNBUSY_THR_PCT / 100) {
+		ioc->busy_level = min(ioc->busy_level, 0);
+		ioc->busy_level--;
+	} else {
+		ioc->busy_level = 0;
+	}
+
+	ioc->busy_level = clamp(ioc->busy_level, -1000, 1000);
+
+	if (ioc->busy_level) {
+		u64 vrate = atomic64_read(&ioc->vtime_rate);
+		u64 vrate_min = ioc->vrate_min, vrate_max = ioc->vrate_max;
+
+		/* rq_wait signal is always reliable, ignore user vrate_min */
+		if (rq_wait_pct > RQ_WAIT_BUSY_PCT)
+			vrate_min = VRATE_MIN;
+
+		/*
+		 * If vrate is out of bounds, apply clamp gradually as the
+		 * bounds can change abruptly.  Otherwise, apply busy_level
+		 * based adjustment.
+		 */
+		if (vrate < vrate_min) {
+			vrate = div64_u64(vrate * (100 + VRATE_CLAMP_ADJ_PCT),
+					  100);
+			vrate = min(vrate, vrate_min);
+		} else if (vrate > vrate_max) {
+			vrate = div64_u64(vrate * (100 - VRATE_CLAMP_ADJ_PCT),
+					  100);
+			vrate = max(vrate, vrate_max);
+		} else {
+			int idx = min_t(int, abs(ioc->busy_level),
+					ARRAY_SIZE(vrate_adj_pct) - 1);
+			u32 adj_pct = vrate_adj_pct[idx];
+
+			if (ioc->busy_level > 0)
+				adj_pct = 100 - adj_pct;
+			else
+				adj_pct = 100 + adj_pct;
+
+			vrate = clamp(DIV64_U64_ROUND_UP(vrate * adj_pct, 100),
+				      vrate_min, vrate_max);
+		}
+
+		trace_iocost_ioc_vrate_adj(ioc, vrate, &missed_ppm, rq_wait_pct,
+					   nr_lagging, nr_shortages,
+					   nr_surpluses);
+
+		atomic64_set(&ioc->vtime_rate, vrate);
+		ioc->inuse_margin_vtime = DIV64_U64_ROUND_UP(
+			ioc->period_us * vrate * INUSE_MARGIN_PCT, 100);
+	}
+
+	ioc_refresh_params(ioc, false);
+
+	/*
+	 * This period is done.  Move onto the next one.  If nothing's
+	 * going on with the device, stop the timer.
+	 */
+	atomic64_inc(&ioc->cur_period);
+
+	if (ioc->running != IOC_STOP) {
+		if (!list_empty(&ioc->active_iocgs)) {
+			ioc_start_period(ioc, &now);
+		} else {
+			ioc->busy_level = 0;
+			ioc->running = IOC_IDLE;
+		}
+	}
+
+	spin_unlock_irq(&ioc->lock);
+}
+
+static void calc_vtime_cost_builtin(struct bio *bio, struct ioc_gq *iocg,
+				    bool is_merge, u64 *costp)
+{
+	struct ioc *ioc = iocg->ioc;
+	u64 coef_seqio, coef_randio, coef_page;
+	u64 pages = max_t(u64, bio_sectors(bio) >> IOC_SECT_TO_PAGE_SHIFT, 1);
+	u64 seek_pages = 0;
+	u64 cost = 0;
+
+	switch (bio_op(bio)) {
+	case REQ_OP_READ:
+		coef_seqio	= ioc->params.lcoefs[LCOEF_RSEQIO];
+		coef_randio	= ioc->params.lcoefs[LCOEF_RRANDIO];
+		coef_page	= ioc->params.lcoefs[LCOEF_RPAGE];
+		break;
+	case REQ_OP_WRITE:
+		coef_seqio	= ioc->params.lcoefs[LCOEF_WSEQIO];
+		coef_randio	= ioc->params.lcoefs[LCOEF_WRANDIO];
+		coef_page	= ioc->params.lcoefs[LCOEF_WPAGE];
+		break;
+	default:
+		goto out;
+	}
+
+	if (iocg->cursor) {
+		seek_pages = abs(bio->bi_iter.bi_sector - iocg->cursor);
+		seek_pages >>= IOC_SECT_TO_PAGE_SHIFT;
+	}
+
+	if (!is_merge) {
+		if (seek_pages > LCOEF_RANDIO_PAGES) {
+			cost += coef_randio;
+		} else {
+			cost += coef_seqio;
+		}
+	}
+	cost += pages * coef_page;
+out:
+	*costp = cost;
+}
+
+static u64 calc_vtime_cost(struct bio *bio, struct ioc_gq *iocg, bool is_merge)
+{
+	u64 cost;
+
+	calc_vtime_cost_builtin(bio, iocg, is_merge, &cost);
+	return cost;
+}
+
+static void ioc_rqos_throttle(struct rq_qos *rqos, struct bio *bio)
+{
+	struct blkcg_gq *blkg = bio->bi_blkg;
+	struct ioc *ioc = rqos_to_ioc(rqos);
+	struct ioc_gq *iocg = blkg_to_iocg(blkg);
+	struct ioc_now now;
+	struct iocg_wait wait;
+	u32 hw_active, hw_inuse;
+	u64 abs_cost, cost, vtime;
+
+	/* bypass IOs if disabled or for root cgroup */
+	if (!ioc->enabled || !iocg->level)
+		return;
+
+	/* always activate so that even 0 cost IOs get protected to some level */
+	if (!iocg_activate(iocg, &now))
+		return;
+
+	/* calculate the absolute vtime cost */
+	abs_cost = calc_vtime_cost(bio, iocg, false);
+	if (!abs_cost)
+		return;
+
+	iocg->cursor = bio_end_sector(bio);
+
+	vtime = atomic64_read(&iocg->vtime);
+	current_hweight(iocg, &hw_active, &hw_inuse);
+
+	if (hw_inuse < hw_active &&
+	    time_after_eq64(vtime + ioc->inuse_margin_vtime, now.vnow)) {
+		TRACE_IOCG_PATH(inuse_reset, iocg, &now,
+				iocg->inuse, iocg->weight, hw_inuse, hw_active);
+		spin_lock_irq(&ioc->lock);
+		propagate_active_weight(iocg, iocg->weight, iocg->weight);
+		spin_unlock_irq(&ioc->lock);
+		current_hweight(iocg, &hw_active, &hw_inuse);
+	}
+
+	cost = abs_cost_to_cost(abs_cost, hw_inuse);
+
+	/*
+	 * If no one's waiting and within budget, issue right away.  The
+	 * tests are racy but the races aren't systemic - we only miss once
+	 * in a while which is fine.
+	 */
+	if (!waitqueue_active(&iocg->waitq) &&
+	    time_before_eq64(vtime + cost, now.vnow)) {
+		iocg_commit_bio(iocg, bio, cost);
+		return;
+	}
+
+	if (bio_issue_as_root_blkg(bio) || fatal_signal_pending(current)) {
+		iocg_commit_bio(iocg, bio, cost);
+		iocg_kick_delay(iocg, &now, cost);
+		return;
+	}
+
+	/*
+	 * Append self to the waitq and schedule the wakeup timer if we're
+	 * the first waiter.  The timer duration is calculated based on the
+	 * current vrate.  vtime and hweight changes can make it too short
+	 * or too long.  Each wait entry records the absolute cost it's
+	 * waiting for to allow re-evaluation using a custom wait entry.
+	 *
+	 * If too short, the timer simply reschedules itself.  If too long,
+	 * the period timer will notice and trigger wakeups.
+	 *
+	 * All waiters are on iocg->waitq and the wait states are
+	 * synchronized using waitq.lock.
+	 */
+	spin_lock_irq(&iocg->waitq.lock);
+
+	/*
+	 * We activated above but w/o any synchronization.  Deactivation is
+	 * synchronized with waitq.lock and we won't get deactivated as
+	 * long as we're waiting, so we're good if we're activated here.
+	 * In the unlikely case that we are deactivated, just issue the IO.
+	 */
+	if (unlikely(list_empty(&iocg->active_list))) {
+		spin_unlock_irq(&iocg->waitq.lock);
+		iocg_commit_bio(iocg, bio, cost);
+		return;
+	}
+
+	init_waitqueue_func_entry(&wait.wait, iocg_wake_fn);
+	wait.wait.private = current;
+	wait.bio = bio;
+	wait.abs_cost = abs_cost;
+	wait.committed = false;	/* will be set true by waker */
+
+	__add_wait_queue_entry_tail(&iocg->waitq, &wait.wait);
+	iocg_kick_waitq(iocg, &now);
+
+	spin_unlock_irq(&iocg->waitq.lock);
+
+	while (true) {
+		set_current_state(TASK_UNINTERRUPTIBLE);
+		if (wait.committed)
+			break;
+		io_schedule();
+	}
+
+	/* waker already committed us, proceed */
+	finish_wait(&iocg->waitq, &wait.wait);
+}
+
+static void ioc_rqos_merge(struct rq_qos *rqos, struct request *rq,
+			   struct bio *bio)
+{
+	struct ioc_gq *iocg = blkg_to_iocg(bio->bi_blkg);
+	sector_t bio_end = bio_end_sector(bio);
+	u32 hw_inuse;
+	u64 abs_cost, cost;
+
+	/* add iff the existing request has cost assigned */
+	if (!rq->bio || !rq->bio->bi_iocost_cost)
+		return;
+
+	abs_cost = calc_vtime_cost(bio, iocg, true);
+	if (!abs_cost)
+		return;
+
+	/* update cursor if backmerging into the request at the cursor */
+	if (blk_rq_pos(rq) < bio_end &&
+	    blk_rq_pos(rq) + blk_rq_sectors(rq) == iocg->cursor)
+		iocg->cursor = bio_end;
+
+	current_hweight(iocg, NULL, &hw_inuse);
+	cost = div64_u64(abs_cost * HWEIGHT_WHOLE, hw_inuse);
+	bio->bi_iocost_cost = cost;
+
+	atomic64_add(cost, &iocg->vtime);
+}
+
+static void ioc_rqos_done_bio(struct rq_qos *rqos, struct bio *bio)
+{
+	struct ioc_gq *iocg = blkg_to_iocg(bio->bi_blkg);
+
+	if (iocg && bio->bi_iocost_cost)
+		atomic64_add(bio->bi_iocost_cost, &iocg->done_vtime);
+}
+
+static void ioc_rqos_done(struct rq_qos *rqos, struct request *rq)
+{
+	struct ioc *ioc = rqos_to_ioc(rqos);
+	u64 on_q_ns, rq_wait_ns;
+	int pidx, rw;
+
+	if (!ioc->enabled || !rq->alloc_time_ns || !rq->start_time_ns)
+		return;
+
+	switch (req_op(rq) & REQ_OP_MASK) {
+	case REQ_OP_READ:
+		pidx = QOS_RLAT;
+		rw = READ;
+		break;
+	case REQ_OP_WRITE:
+		pidx = QOS_WLAT;
+		rw = WRITE;
+		break;
+	default:
+		return;
+	}
+
+	on_q_ns = ktime_get_ns() - rq->alloc_time_ns;
+	rq_wait_ns = rq->start_time_ns - rq->alloc_time_ns;
+
+	if (on_q_ns <= ioc->params.qos[pidx] * NSEC_PER_USEC)
+		this_cpu_inc(ioc->pcpu_stat->missed[rw].nr_met);
+	else
+		this_cpu_inc(ioc->pcpu_stat->missed[rw].nr_missed);
+
+	this_cpu_add(ioc->pcpu_stat->rq_wait_ns, rq_wait_ns);
+}
+
+static void ioc_rqos_queue_depth_changed(struct rq_qos *rqos)
+{
+	struct ioc *ioc = rqos_to_ioc(rqos);
+
+	spin_lock_irq(&ioc->lock);
+	ioc_refresh_params(ioc, false);
+	spin_unlock_irq(&ioc->lock);
+}
+
+static void ioc_rqos_exit(struct rq_qos *rqos)
+{
+	struct ioc *ioc = rqos_to_ioc(rqos);
+
+	blkcg_deactivate_policy(rqos->q, &blkcg_policy_iocost);
+
+	spin_lock_irq(&ioc->lock);
+	ioc->running = IOC_STOP;
+	spin_unlock_irq(&ioc->lock);
+
+	del_timer_sync(&ioc->timer);
+	free_percpu(ioc->pcpu_stat);
+	kfree(ioc);
+}
+
+static struct rq_qos_ops ioc_rqos_ops = {
+	.throttle = ioc_rqos_throttle,
+	.merge = ioc_rqos_merge,
+	.done_bio = ioc_rqos_done_bio,
+	.done = ioc_rqos_done,
+	.queue_depth_changed = ioc_rqos_queue_depth_changed,
+	.exit = ioc_rqos_exit,
+};
+
+static int blk_iocost_init(struct request_queue *q)
+{
+	struct ioc *ioc;
+	struct rq_qos *rqos;
+	int ret;
+
+	ioc = kzalloc(sizeof(*ioc), GFP_KERNEL);
+	if (!ioc)
+		return -ENOMEM;
+
+	ioc->pcpu_stat = alloc_percpu(struct ioc_pcpu_stat);
+	if (!ioc->pcpu_stat) {
+		kfree(ioc);
+		return -ENOMEM;
+	}
+
+	rqos = &ioc->rqos;
+	rqos->id = RQ_QOS_COST;
+	rqos->ops = &ioc_rqos_ops;
+	rqos->q = q;
+
+	spin_lock_init(&ioc->lock);
+	timer_setup(&ioc->timer, ioc_timer_fn, 0);
+	INIT_LIST_HEAD(&ioc->active_iocgs);
+
+	ioc->running = IOC_IDLE;
+	atomic64_set(&ioc->vtime_rate, VTIME_PER_USEC);
+	seqcount_init(&ioc->period_seqcount);
+	ioc->period_at = ktime_to_us(ktime_get());
+	atomic64_set(&ioc->cur_period, 0);
+	atomic_set(&ioc->hweight_gen, 0);
+
+	spin_lock_irq(&ioc->lock);
+	ioc->autop_idx = AUTOP_INVALID;
+	ioc_refresh_params(ioc, true);
+	spin_unlock_irq(&ioc->lock);
+
+	rq_qos_add(q, rqos);
+	ret = blkcg_activate_policy(q, &blkcg_policy_iocost);
+	if (ret) {
+		rq_qos_del(q, rqos);
+		kfree(ioc);
+		return ret;
+	}
+	return 0;
+}
+
+static struct blkcg_policy_data *ioc_cpd_alloc(gfp_t gfp)
+{
+	struct ioc_cgrp *iocc;
+
+	iocc = kzalloc(sizeof(struct ioc_cgrp), gfp);
+	iocc->dfl_weight = CGROUP_WEIGHT_DFL;
+
+	return &iocc->cpd;
+}
+
+static void ioc_cpd_free(struct blkcg_policy_data *cpd)
+{
+	kfree(container_of(cpd, struct ioc_cgrp, cpd));
+}
+
+static struct blkg_policy_data *ioc_pd_alloc(gfp_t gfp, struct request_queue *q,
+					     struct blkcg *blkcg)
+{
+	int levels = blkcg->css.cgroup->level + 1;
+	struct ioc_gq *iocg;
+
+	iocg = kzalloc_node(sizeof(*iocg) + levels * sizeof(iocg->ancestors[0]),
+			    gfp, q->node);
+	if (!iocg)
+		return NULL;
+
+	return &iocg->pd;
+}
+
+static void ioc_pd_init(struct blkg_policy_data *pd)
+{
+	struct ioc_gq *iocg = pd_to_iocg(pd);
+	struct blkcg_gq *blkg = pd_to_blkg(&iocg->pd);
+	struct ioc *ioc = q_to_ioc(blkg->q);
+	struct ioc_now now;
+	struct blkcg_gq *tblkg;
+	unsigned long flags;
+
+	ioc_now(ioc, &now);
+
+	iocg->ioc = ioc;
+	atomic64_set(&iocg->vtime, now.vnow);
+	atomic64_set(&iocg->done_vtime, now.vnow);
+	atomic64_set(&iocg->active_period, atomic64_read(&ioc->cur_period));
+	INIT_LIST_HEAD(&iocg->active_list);
+	iocg->hweight_active = HWEIGHT_WHOLE;
+	iocg->hweight_inuse = HWEIGHT_WHOLE;
+
+	init_waitqueue_head(&iocg->waitq);
+	hrtimer_init(&iocg->waitq_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
+	iocg->waitq_timer.function = iocg_waitq_timer_fn;
+	hrtimer_init(&iocg->delay_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
+	iocg->delay_timer.function = iocg_delay_timer_fn;
+
+	iocg->level = blkg->blkcg->css.cgroup->level;
+
+	for (tblkg = blkg; tblkg; tblkg = tblkg->parent) {
+		struct ioc_gq *tiocg = blkg_to_iocg(tblkg);
+		iocg->ancestors[tiocg->level] = tiocg;
+	}
+
+	spin_lock_irqsave(&ioc->lock, flags);
+	weight_updated(iocg);
+	spin_unlock_irqrestore(&ioc->lock, flags);
+}
+
+static void ioc_pd_free(struct blkg_policy_data *pd)
+{
+	struct ioc_gq *iocg = pd_to_iocg(pd);
+	struct ioc *ioc = iocg->ioc;
+
+	if (ioc) {
+		hrtimer_cancel(&iocg->waitq_timer);
+		hrtimer_cancel(&iocg->delay_timer);
+
+		spin_lock(&ioc->lock);
+		if (!list_empty(&iocg->active_list)) {
+			propagate_active_weight(iocg, 0, 0);
+			list_del_init(&iocg->active_list);
+		}
+		spin_unlock(&ioc->lock);
+	}
+	kfree(iocg);
+}
+
+static u64 ioc_weight_prfill(struct seq_file *sf, struct blkg_policy_data *pd,
+			     int off)
+{
+	const char *dname = blkg_dev_name(pd->blkg);
+	struct ioc_gq *iocg = pd_to_iocg(pd);
+
+	if (dname && iocg->cfg_weight)
+		seq_printf(sf, "%s %u\n", dname, iocg->cfg_weight);
+	return 0;
+}
+
+
+static int ioc_weight_show(struct seq_file *sf, void *v)
+{
+	struct blkcg *blkcg = css_to_blkcg(seq_css(sf));
+	struct ioc_cgrp *iocc = blkcg_to_iocc(blkcg);
+
+	seq_printf(sf, "default %u\n", iocc->dfl_weight);
+	blkcg_print_blkgs(sf, blkcg, ioc_weight_prfill,
+			  &blkcg_policy_iocost, seq_cft(sf)->private, false);
+	return 0;
+}
+
+static ssize_t ioc_weight_write(struct kernfs_open_file *of, char *buf,
+				size_t nbytes, loff_t off)
+{
+	struct blkcg *blkcg = css_to_blkcg(of_css(of));
+	struct ioc_cgrp *iocc = blkcg_to_iocc(blkcg);
+	struct blkg_conf_ctx ctx;
+	struct ioc_gq *iocg;
+	u32 v;
+	int ret;
+
+	if (!strchr(buf, ':')) {
+		struct blkcg_gq *blkg;
+
+		if (!sscanf(buf, "default %u", &v) && !sscanf(buf, "%u", &v))
+			return -EINVAL;
+
+		if (v < CGROUP_WEIGHT_MIN || v > CGROUP_WEIGHT_MAX)
+			return -EINVAL;
+
+		spin_lock(&blkcg->lock);
+		iocc->dfl_weight = v;
+		hlist_for_each_entry(blkg, &blkcg->blkg_list, blkcg_node) {
+			struct ioc_gq *iocg = blkg_to_iocg(blkg);
+
+			if (iocg) {
+				spin_lock_irq(&iocg->ioc->lock);
+				weight_updated(iocg);
+				spin_unlock_irq(&iocg->ioc->lock);
+			}
+		}
+		spin_unlock(&blkcg->lock);
+
+		return nbytes;
+	}
+
+	ret = blkg_conf_prep(blkcg, &blkcg_policy_iocost, buf, &ctx);
+	if (ret)
+		return ret;
+
+	iocg = blkg_to_iocg(ctx.blkg);
+
+	if (!strncmp(ctx.body, "default", 7)) {
+		v = 0;
+	} else {
+		if (!sscanf(ctx.body, "%u", &v))
+			goto einval;
+		if (v < CGROUP_WEIGHT_MIN || v > CGROUP_WEIGHT_MAX)
+			goto einval;
+	}
+
+	spin_lock_irq(&iocg->ioc->lock);
+	iocg->cfg_weight = v;
+	weight_updated(iocg);
+	spin_unlock_irq(&iocg->ioc->lock);
+
+	blkg_conf_finish(&ctx);
+	return nbytes;
+
+einval:
+	blkg_conf_finish(&ctx);
+	return -EINVAL;
+}
+
+static u64 ioc_qos_prfill(struct seq_file *sf, struct blkg_policy_data *pd,
+			  int off)
+{
+	const char *dname = blkg_dev_name(pd->blkg);
+	struct ioc *ioc = pd_to_iocg(pd)->ioc;
+
+	if (!dname)
+		return 0;
+
+	seq_printf(sf, "%s enable=%d ctrl=%s rpct=%u.%02u rlat=%u wpct=%u.%02u wlat=%u min=%u.%02u max=%u.%02u\n",
+		   dname, ioc->enabled, ioc->user_qos_params ? "user" : "auto",
+		   ioc->params.qos[QOS_RPPM] / 10000,
+		   ioc->params.qos[QOS_RPPM] % 10000 / 100,
+		   ioc->params.qos[QOS_RLAT],
+		   ioc->params.qos[QOS_WPPM] / 10000,
+		   ioc->params.qos[QOS_WPPM] % 10000 / 100,
+		   ioc->params.qos[QOS_WLAT],
+		   ioc->params.qos[QOS_MIN] / 10000,
+		   ioc->params.qos[QOS_MIN] % 10000 / 100,
+		   ioc->params.qos[QOS_MAX] / 10000,
+		   ioc->params.qos[QOS_MAX] % 10000 / 100);
+	return 0;
+}
+
+static int ioc_qos_show(struct seq_file *sf, void *v)
+{
+	struct blkcg *blkcg = css_to_blkcg(seq_css(sf));
+
+	blkcg_print_blkgs(sf, blkcg, ioc_qos_prfill,
+			  &blkcg_policy_iocost, seq_cft(sf)->private, false);
+	return 0;
+}
+
+static const match_table_t qos_ctrl_tokens = {
+	{ QOS_ENABLE,		"enable=%u"	},
+	{ QOS_CTRL,		"ctrl=%s"	},
+	{ NR_QOS_CTRL_PARAMS,	NULL		},
+};
+
+static const match_table_t qos_tokens = {
+	{ QOS_RPPM,		"rpct=%s"	},
+	{ QOS_RLAT,		"rlat=%u"	},
+	{ QOS_WPPM,		"wpct=%s"	},
+	{ QOS_WLAT,		"wlat=%u"	},
+	{ QOS_MIN,		"min=%s"	},
+	{ QOS_MAX,		"max=%s"	},
+	{ NR_QOS_PARAMS,	NULL		},
+};
+
+static ssize_t ioc_qos_write(struct kernfs_open_file *of, char *input,
+			     size_t nbytes, loff_t off)
+{
+	struct gendisk *disk;
+	struct ioc *ioc;
+	u32 qos[NR_QOS_PARAMS];
+	bool enable, user;
+	char *p;
+	int ret;
+
+	disk = blkcg_conf_get_disk(&input);
+	if (IS_ERR(disk))
+		return PTR_ERR(disk);
+
+	ioc = q_to_ioc(disk->queue);
+	if (!ioc) {
+		ret = blk_iocost_init(disk->queue);
+		if (ret)
+			goto err;
+		ioc = q_to_ioc(disk->queue);
+	}
+
+	spin_lock_irq(&ioc->lock);
+	memcpy(qos, ioc->params.qos, sizeof(qos));
+	enable = ioc->enabled;
+	user = ioc->user_qos_params;
+	spin_unlock_irq(&ioc->lock);
+
+	while ((p = strsep(&input, " \t\n"))) {
+		substring_t args[MAX_OPT_ARGS];
+		char buf[32];
+		int tok;
+		s64 v;
+
+		if (!*p)
+			continue;
+
+		switch (match_token(p, qos_ctrl_tokens, args)) {
+		case QOS_ENABLE:
+			match_u64(&args[0], &v);
+			enable = v;
+			continue;
+		case QOS_CTRL:
+			match_strlcpy(buf, &args[0], sizeof(buf));
+			if (!strcmp(buf, "auto"))
+				user = false;
+			else if (!strcmp(buf, "user"))
+				user = true;
+			else
+				goto einval;
+			continue;
+		}
+
+		tok = match_token(p, qos_tokens, args);
+		switch (tok) {
+		case QOS_RPPM:
+		case QOS_WPPM:
+			if (match_strlcpy(buf, &args[0], sizeof(buf)) >=
+			    sizeof(buf))
+				goto einval;
+			if (cgroup_parse_float(buf, 2, &v))
+				goto einval;
+			if (v < 0 || v > 10000)
+				goto einval;
+			qos[tok] = v * 100;
+			break;
+		case QOS_RLAT:
+		case QOS_WLAT:
+			if (match_u64(&args[0], &v))
+				goto einval;
+			qos[tok] = v;
+			break;
+		case QOS_MIN:
+		case QOS_MAX:
+			if (match_strlcpy(buf, &args[0], sizeof(buf)) >=
+			    sizeof(buf))
+				goto einval;
+			if (cgroup_parse_float(buf, 2, &v))
+				goto einval;
+			if (v < 0)
+				goto einval;
+			qos[tok] = clamp_t(s64, v * 100,
+					   VRATE_MIN_PPM, VRATE_MAX_PPM);
+			break;
+		default:
+			goto einval;
+		}
+		user = true;
+	}
+
+	if (qos[QOS_MIN] > qos[QOS_MAX])
+		goto einval;
+
+	spin_lock_irq(&ioc->lock);
+
+	if (enable) {
+		blk_queue_flag_set(QUEUE_FLAG_RQ_ALLOC_TIME, ioc->rqos.q);
+		ioc->enabled = true;
+	} else {
+		blk_queue_flag_clear(QUEUE_FLAG_RQ_ALLOC_TIME, ioc->rqos.q);
+		ioc->enabled = false;
+	}
+
+	if (user) {
+		memcpy(ioc->params.qos, qos, sizeof(qos));
+		ioc->user_qos_params = true;
+	} else {
+		ioc->user_qos_params = false;
+	}
+
+	ioc_refresh_params(ioc, true);
+	spin_unlock_irq(&ioc->lock);
+
+	put_disk_and_module(disk);
+	return nbytes;
+einval:
+	ret = -EINVAL;
+err:
+	put_disk_and_module(disk);
+	return ret;
+}
+
+static u64 ioc_cost_model_prfill(struct seq_file *sf,
+				 struct blkg_policy_data *pd, int off)
+{
+	const char *dname = blkg_dev_name(pd->blkg);
+	struct ioc *ioc = pd_to_iocg(pd)->ioc;
+	u64 *u = ioc->params.i_lcoefs;
+
+	if (!dname)
+		return 0;
+
+	seq_printf(sf, "%s ctrl=%s model=linear "
+		   "rbps=%llu rseqiops=%llu rrandiops=%llu "
+		   "wbps=%llu wseqiops=%llu wrandiops=%llu\n",
+		   dname, ioc->user_cost_model ? "user" : "auto",
+		   u[I_LCOEF_RBPS], u[I_LCOEF_RSEQIOPS], u[I_LCOEF_RRANDIOPS],
+		   u[I_LCOEF_WBPS], u[I_LCOEF_WSEQIOPS], u[I_LCOEF_WRANDIOPS]);
+	return 0;
+}
+
+static int ioc_cost_model_show(struct seq_file *sf, void *v)
+{
+	struct blkcg *blkcg = css_to_blkcg(seq_css(sf));
+
+	blkcg_print_blkgs(sf, blkcg, ioc_cost_model_prfill,
+			  &blkcg_policy_iocost, seq_cft(sf)->private, false);
+	return 0;
+}
+
+static const match_table_t cost_ctrl_tokens = {
+	{ COST_CTRL,		"ctrl=%s"	},
+	{ COST_MODEL,		"model=%s"	},
+	{ NR_COST_CTRL_PARAMS,	NULL		},
+};
+
+static const match_table_t i_lcoef_tokens = {
+	{ I_LCOEF_RBPS,		"rbps=%u"	},
+	{ I_LCOEF_RSEQIOPS,	"rseqiops=%u"	},
+	{ I_LCOEF_RRANDIOPS,	"rrandiops=%u"	},
+	{ I_LCOEF_WBPS,		"wbps=%u"	},
+	{ I_LCOEF_WSEQIOPS,	"wseqiops=%u"	},
+	{ I_LCOEF_WRANDIOPS,	"wrandiops=%u"	},
+	{ NR_I_LCOEFS,		NULL		},
+};
+
+static ssize_t ioc_cost_model_write(struct kernfs_open_file *of, char *input,
+				    size_t nbytes, loff_t off)
+{
+	struct gendisk *disk;
+	struct ioc *ioc;
+	u64 u[NR_I_LCOEFS];
+	bool user;
+	char *p;
+	int ret;
+
+	disk = blkcg_conf_get_disk(&input);
+	if (IS_ERR(disk))
+		return PTR_ERR(disk);
+
+	ioc = q_to_ioc(disk->queue);
+	if (!ioc) {
+		ret = blk_iocost_init(disk->queue);
+		if (ret)
+			goto err;
+		ioc = q_to_ioc(disk->queue);
+	}
+
+	spin_lock_irq(&ioc->lock);
+	memcpy(u, ioc->params.i_lcoefs, sizeof(u));
+	user = ioc->user_cost_model;
+	spin_unlock_irq(&ioc->lock);
+
+	while ((p = strsep(&input, " \t\n"))) {
+		substring_t args[MAX_OPT_ARGS];
+		char buf[32];
+		int tok;
+		u64 v;
+
+		if (!*p)
+			continue;
+
+		switch (match_token(p, cost_ctrl_tokens, args)) {
+		case COST_CTRL:
+			match_strlcpy(buf, &args[0], sizeof(buf));
+			if (!strcmp(buf, "auto"))
+				user = false;
+			else if (!strcmp(buf, "user"))
+				user = true;
+			else
+				goto einval;
+			continue;
+		case COST_MODEL:
+			match_strlcpy(buf, &args[0], sizeof(buf));
+			if (strcmp(buf, "linear"))
+				goto einval;
+			continue;
+		}
+
+		tok = match_token(p, i_lcoef_tokens, args);
+		if (tok == NR_I_LCOEFS)
+			goto einval;
+		if (match_u64(&args[0], &v))
+			goto einval;
+		u[tok] = v;
+		user = true;
+	}
+
+	spin_lock_irq(&ioc->lock);
+	if (user) {
+		memcpy(ioc->params.i_lcoefs, u, sizeof(u));
+		ioc->user_cost_model = true;
+	} else {
+		ioc->user_cost_model = false;
+	}
+	ioc_refresh_params(ioc, true);
+	spin_unlock_irq(&ioc->lock);
+
+	put_disk_and_module(disk);
+	return nbytes;
+
+einval:
+	ret = -EINVAL;
+err:
+	put_disk_and_module(disk);
+	return ret;
+}
+
+static struct cftype ioc_files[] = {
+	{
+		.name = "weight",
+		.flags = CFTYPE_NOT_ON_ROOT,
+		.seq_show = ioc_weight_show,
+		.write = ioc_weight_write,
+	},
+	{
+		.name = "cost.qos",
+		.flags = CFTYPE_ONLY_ON_ROOT,
+		.seq_show = ioc_qos_show,
+		.write = ioc_qos_write,
+	},
+	{
+		.name = "cost.model",
+		.flags = CFTYPE_ONLY_ON_ROOT,
+		.seq_show = ioc_cost_model_show,
+		.write = ioc_cost_model_write,
+	},
+	{}
+};
+
+static struct blkcg_policy blkcg_policy_iocost = {
+	.dfl_cftypes	= ioc_files,
+	.cpd_alloc_fn	= ioc_cpd_alloc,
+	.cpd_free_fn	= ioc_cpd_free,
+	.pd_alloc_fn	= ioc_pd_alloc,
+	.pd_init_fn	= ioc_pd_init,
+	.pd_free_fn	= ioc_pd_free,
+};
+
+static int __init ioc_init(void)
+{
+	return blkcg_policy_register(&blkcg_policy_iocost);
+}
+
+static void __exit ioc_exit(void)
+{
+	return blkcg_policy_unregister(&blkcg_policy_iocost);
+}
+
+module_init(ioc_init);
+module_exit(ioc_exit);
diff --git a/block/blk-rq-qos.h b/block/blk-rq-qos.h
index 5f8b75826a98..08a09dbe0f4b 100644
--- a/block/blk-rq-qos.h
+++ b/block/blk-rq-qos.h
@@ -15,6 +15,7 @@ struct blk_mq_debugfs_attr;
 enum rq_qos_id {
 	RQ_QOS_WBT,
 	RQ_QOS_LATENCY,
+	RQ_QOS_COST,
 };
 
 struct rq_wait {
@@ -84,6 +85,8 @@ static inline const char *rq_qos_id_to_name(enum rq_qos_id id)
 		return "wbt";
 	case RQ_QOS_LATENCY:
 		return "latency";
+	case RQ_QOS_COST:
+		return "cost";
 	}
 	return "unknown";
 }
diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
index 75059c1d4244..6f7a8647cefb 100644
--- a/include/linux/blk_types.h
+++ b/include/linux/blk_types.h
@@ -169,6 +169,9 @@ struct bio {
 	 */
 	struct blkcg_gq		*bi_blkg;
 	struct bio_issue	bi_issue;
+#ifdef CONFIG_BLK_CGROUP_IOCOST
+	u64			bi_iocost_cost;
+#endif
 #endif
 	union {
 #if defined(CONFIG_BLK_DEV_INTEGRITY)
diff --git a/include/trace/events/iocost.h b/include/trace/events/iocost.h
new file mode 100644
index 000000000000..ec2217dd57ac
--- /dev/null
+++ b/include/trace/events/iocost.h
@@ -0,0 +1,174 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM iocost
+
+#if !defined(_TRACE_BLK_IOCOST_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_BLK_IOCOST_H
+
+#include <linux/tracepoint.h>
+
+TRACE_EVENT(iocost_iocg_activate,
+
+	TP_PROTO(struct ioc_gq *iocg, const char *path, struct ioc_now *now,
+		u64 last_period, u64 cur_period, u64 vtime),
+
+	TP_ARGS(iocg, path, now, last_period, cur_period, vtime),
+
+	TP_STRUCT__entry (
+		__string(devname, ioc_name(iocg->ioc))
+		__string(cgroup, path)
+		__field(u64, now)
+		__field(u64, vnow)
+		__field(u64, vrate)
+		__field(u64, last_period)
+		__field(u64, cur_period)
+		__field(u64, last_vtime)
+		__field(u64, vtime)
+		__field(u32, weight)
+		__field(u32, inuse)
+		__field(u64, hweight_active)
+		__field(u64, hweight_inuse)
+	),
+
+	TP_fast_assign(
+		__assign_str(devname, ioc_name(iocg->ioc));
+		__assign_str(cgroup, path);
+		__entry->now = now->now;
+		__entry->vnow = now->vnow;
+		__entry->vrate = now->vrate;
+		__entry->last_period = last_period;
+		__entry->cur_period = cur_period;
+		__entry->last_vtime = iocg->last_vtime;
+		__entry->vtime = vtime;
+		__entry->weight = iocg->weight;
+		__entry->inuse = iocg->inuse;
+		__entry->hweight_active = iocg->hweight_active;
+		__entry->hweight_inuse = iocg->hweight_inuse;
+	),
+
+	TP_printk("[%s:%s] now=%llu:%llu vrate=%llu "
+		  "period=%llu->%llu vtime=%llu->%llu "
+		  "weight=%u/%u hweight=%llu/%llu",
+		__get_str(devname), __get_str(cgroup),
+		__entry->now, __entry->vnow, __entry->vrate,
+		__entry->last_period, __entry->cur_period,
+		__entry->last_vtime, __entry->vtime,
+		__entry->inuse, __entry->weight,
+		__entry->hweight_inuse, __entry->hweight_active
+	)
+);
+
+DECLARE_EVENT_CLASS(iocg_inuse_update,
+
+	TP_PROTO(struct ioc_gq *iocg, const char *path, struct ioc_now *now,
+		u32 old_inuse, u32 new_inuse,
+		u64 old_hw_inuse, u64 new_hw_inuse),
+
+	TP_ARGS(iocg, path, now, old_inuse, new_inuse,
+		old_hw_inuse, new_hw_inuse),
+
+	TP_STRUCT__entry (
+		__string(devname, ioc_name(iocg->ioc))
+		__string(cgroup, path)
+		__field(u64, now)
+		__field(u32, old_inuse)
+		__field(u32, new_inuse)
+		__field(u64, old_hweight_inuse)
+		__field(u64, new_hweight_inuse)
+	),
+
+	TP_fast_assign(
+		__assign_str(devname, ioc_name(iocg->ioc));
+		__assign_str(cgroup, path);
+		__entry->now = now->now;
+		__entry->old_inuse = old_inuse;
+		__entry->new_inuse = new_inuse;
+		__entry->old_hweight_inuse = old_hw_inuse;
+		__entry->new_hweight_inuse = new_hw_inuse;
+	),
+
+	TP_printk("[%s:%s] now=%llu inuse=%u->%u hw_inuse=%llu->%llu",
+		__get_str(devname), __get_str(cgroup), __entry->now,
+		__entry->old_inuse, __entry->new_inuse,
+		__entry->old_hweight_inuse, __entry->new_hweight_inuse
+	)
+);
+
+DEFINE_EVENT(iocg_inuse_update, iocost_inuse_takeback,
+
+	TP_PROTO(struct ioc_gq *iocg, const char *path, struct ioc_now *now,
+		u32 old_inuse, u32 new_inuse,
+		u64 old_hw_inuse, u64 new_hw_inuse),
+
+	TP_ARGS(iocg, path, now, old_inuse, new_inuse,
+		old_hw_inuse, new_hw_inuse)
+);
+
+DEFINE_EVENT(iocg_inuse_update, iocost_inuse_giveaway,
+
+	TP_PROTO(struct ioc_gq *iocg, const char *path, struct ioc_now *now,
+		u32 old_inuse, u32 new_inuse,
+		u64 old_hw_inuse, u64 new_hw_inuse),
+
+	TP_ARGS(iocg, path, now, old_inuse, new_inuse,
+		old_hw_inuse, new_hw_inuse)
+);
+
+DEFINE_EVENT(iocg_inuse_update, iocost_inuse_reset,
+
+	TP_PROTO(struct ioc_gq *iocg, const char *path, struct ioc_now *now,
+		u32 old_inuse, u32 new_inuse,
+		u64 old_hw_inuse, u64 new_hw_inuse),
+
+	TP_ARGS(iocg, path, now, old_inuse, new_inuse,
+		old_hw_inuse, new_hw_inuse)
+);
+
+TRACE_EVENT(iocost_ioc_vrate_adj,
+
+	TP_PROTO(struct ioc *ioc, u64 new_vrate, u32 (*missed_ppm)[2],
+		u32 rq_wait_pct, int nr_lagging, int nr_shortages,
+		int nr_surpluses),
+
+	TP_ARGS(ioc, new_vrate, missed_ppm, rq_wait_pct, nr_lagging, nr_shortages,
+		nr_surpluses),
+
+	TP_STRUCT__entry (
+		__string(devname, ioc_name(ioc))
+		__field(u64, old_vrate)
+		__field(u64, new_vrate)
+		__field(int, busy_level)
+		__field(u32, read_missed_ppm)
+		__field(u32, write_missed_ppm)
+		__field(u32, rq_wait_pct)
+		__field(int, nr_lagging)
+		__field(int, nr_shortages)
+		__field(int, nr_surpluses)
+	),
+
+	TP_fast_assign(
+		__assign_str(devname, ioc_name(ioc));
+		__entry->old_vrate = atomic64_read(&ioc->vtime_rate);;
+		__entry->new_vrate = new_vrate;
+		__entry->busy_level = ioc->busy_level;
+		__entry->read_missed_ppm = (*missed_ppm)[READ];
+		__entry->write_missed_ppm = (*missed_ppm)[WRITE];
+		__entry->rq_wait_pct = rq_wait_pct;
+		__entry->nr_lagging = nr_lagging;
+		__entry->nr_shortages = nr_shortages;
+		__entry->nr_surpluses = nr_surpluses;
+	),
+
+	TP_printk("[%s] vrate=%llu->%llu busy=%d missed_ppm=%u:%u rq_wait_pct=%u lagging=%d shortages=%d surpluses=%d",
+		__get_str(devname), __entry->old_vrate, __entry->new_vrate,
+		__entry->busy_level,
+		__entry->read_missed_ppm, __entry->write_missed_ppm,
+		__entry->rq_wait_pct, __entry->nr_lagging, __entry->nr_shortages,
+		__entry->nr_surpluses
+	)
+);
+
+#endif /* _TRACE_BLK_IOCOST_H */
+
+/* This part must be outside protection */
+#include <trace/define_trace.h>
-- 
2.17.1

[PATCH] blkcg: fix missing free on error path of blk_iocost_init()

[Tejun Heo]

blk_iocost_init() forgot to free its percpu stat on the error path.
Fix it.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Hillf Danton <hdanton@sina.com>
---
 block/blk-iocost.c |    1 +
 1 file changed, 1 insertion(+)

diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index f04a4ed1cb45..9c8046ac5925 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -1876,6 +1876,7 @@ static int blk_iocost_init(struct request_queue *q)
 	ret = blkcg_activate_policy(q, &blkcg_policy_iocost);
 	if (ret) {
 		rq_qos_del(q, rqos);
+		free_percpu(ioc->pcpu_stat);
 		kfree(ioc);
 		return ret;
 	}

Re: [PATCH 08/10] blkcg: implement blk-iocost

[Michal Koutný]

[-- Attachment #1: Type: text/plain, Size: 1210 bytes --]

Hello.

On Wed, Aug 28, 2019 at 03:05:58PM -0700, Tejun Heo <tj@kernel.org> wrote:
> diff --git a/block/blk-iocost.c b/block/blk-iocost.c
> [...]
> +static struct cftype ioc_files[] = {
> +		.name = "weight",
> [...]
This adds the generic io.weight attribute. How will this compose with
the weight from IO schedulers? (AFAIK, only BFQ allows proportional
control as of now. +CC Paolo.)

I see this attributes are effectively per-cgroup per-device. Apparently,
one device should have only one weight across hierarchy. Would it make
sense to have io.bfq.weight and io.cost.weight with disjunctive devices?

(Alas, I have no idea how to make the users of io.weight happy, when
proportionality control mechanism seems orthogonal to the weight.
(Vector weights?))


> +		.name = "cost.qos",
> +		.flags = CFTYPE_ONLY_ON_ROOT,
> [...]
> +		.name = "cost.model",
> +		.flags = CFTYPE_ONLY_ON_ROOT,
I'm concerned that these aren't true cgroup attributes. The root cgroup
would act as container for global configuration options. Wouldn't these
values better fit as (configurable) attributes of the respective
devices?

Secondly, how is CFTYPE_ONLY_ON_ROOT supposed to be presented in cgroup
namespaces?

Thanks,
Michal

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [PATCH 08/10] blkcg: implement blk-iocost

[Tejun Heo]

Hello, Michal.

On Tue, Sep 10, 2019 at 02:55:14PM +0200, Michal Koutný wrote:
> This adds the generic io.weight attribute. How will this compose with
> the weight from IO schedulers? (AFAIK, only BFQ allows proportional
> control as of now. +CC Paolo.)

The two being enabled at the same time doesn't make sense, so we can
just switch over to bfq when bfq is selected as the iosched.  I asked
what Paolo wanted to do in terms of interface a couple times now but
didn't get an answer and he posted a patch which makes the two
controllers conflict, so....  Paolo, so it looks like you want to
rename all bfq files to drop the bfq prefix, right?  I can implement
the switching if so.

> I see this attributes are effectively per-cgroup per-device. Apparently,
> one device should have only one weight across hierarchy. Would it make
> sense to have io.bfq.weight and io.cost.weight with disjunctive devices?

It never makes sense to have both enabled, so I don't think that
interface makes sense.

> > +		.name = "cost.qos",
> > +		.flags = CFTYPE_ONLY_ON_ROOT,
> > [...]
> > +		.name = "cost.model",
> > +		.flags = CFTYPE_ONLY_ON_ROOT,
> I'm concerned that these aren't true cgroup attributes. The root cgroup
> would act as container for global configuration options. Wouldn't these
> values better fit as (configurable) attributes of the respective
> devices?

Initially, I put them under block device sysfs but it was too clumsy
with different config file formats and all.  I think it's better to
have global controller configs at the root cgroup.

> Secondly, how is CFTYPE_ONLY_ON_ROOT supposed to be presented in cgroup
> namespaces?

Not at all.  These are system-wide configs.  cgroup namespaces
shouldn't have anything which aren't in non-root cgroups.

Thanks.

-- 
tejun

Re: [PATCH 08/10] blkcg: implement blk-iocost

[Paolo Valente]

> Il giorno 10 set 2019, alle ore 18:08, Tejun Heo <tj@kernel.org> ha scritto:
> 
> Hello, Michal.
> 
> On Tue, Sep 10, 2019 at 02:55:14PM +0200, Michal Koutný wrote:
>> This adds the generic io.weight attribute. How will this compose with
>> the weight from IO schedulers? (AFAIK, only BFQ allows proportional
>> control as of now. +CC Paolo.)
> 
> The two being enabled at the same time doesn't make sense, so we can
> just switch over to bfq when bfq is selected as the iosched.  I asked
> what Paolo wanted to do in terms of interface a couple times now but
> didn't get an answer and he posted a patch which makes the two
> controllers conflict, so....  Paolo, so it looks like you want to
> rename all bfq files to drop the bfq prefix, right?

Yep, mainly because ... this is the solution you voted and you
yourself proposed [1] :)

[1] https://patchwork.kernel.org/patch/10988261/

>  I can implement
> the switching if so.
> 

That would be perfect.

Thanks,
Paolo

>> I see this attributes are effectively per-cgroup per-device. Apparently,
>> one device should have only one weight across hierarchy. Would it make
>> sense to have io.bfq.weight and io.cost.weight with disjunctive devices?
> 
> It never makes sense to have both enabled, so I don't think that
> interface makes sense.
> 
>>> +		.name = "cost.qos",
>>> +		.flags = CFTYPE_ONLY_ON_ROOT,
>>> [...]
>>> +		.name = "cost.model",
>>> +		.flags = CFTYPE_ONLY_ON_ROOT,
>> I'm concerned that these aren't true cgroup attributes. The root cgroup
>> would act as container for global configuration options. Wouldn't these
>> values better fit as (configurable) attributes of the respective
>> devices?
> 
> Initially, I put them under block device sysfs but it was too clumsy
> with different config file formats and all.  I think it's better to
> have global controller configs at the root cgroup.
> 
>> Secondly, how is CFTYPE_ONLY_ON_ROOT supposed to be presented in cgroup
>> namespaces?
> 
> Not at all.  These are system-wide configs.  cgroup namespaces
> shouldn't have anything which aren't in non-root cgroups.
> 
> Thanks.
> 
> -- 
> tejun

Re: [PATCH 08/10] blkcg: implement blk-iocost

[Tejun Heo]

Hello,

On Wed, Sep 11, 2019 at 10:18:53AM +0200, Paolo Valente wrote:
> > The two being enabled at the same time doesn't make sense, so we can
> > just switch over to bfq when bfq is selected as the iosched.  I asked
> > what Paolo wanted to do in terms of interface a couple times now but
> > didn't get an answer and he posted a patch which makes the two
> > controllers conflict, so....  Paolo, so it looks like you want to
> > rename all bfq files to drop the bfq prefix, right?
> 
> Yep, mainly because ... this is the solution you voted and you
> yourself proposed [1] :)
> 
> [1] https://patchwork.kernel.org/patch/10988261/

So, that was then.  Since then the interface change has been published
and userspace, at least some of them, already had to adjust.  Now, I
don't have any opinion on the matter and flipping again will cause
inconveniences to some subset of users.  It's your call.

> >  I can implement
> > the switching if so.
> 
> That would be perfect.

Whichever way it gets decided, this is easy enough.  I'll prep a
patch.

Thanks.

-- 
tejun

Re: [PATCH 08/10] blkcg: implement blk-iocost

[Tejun Heo]

Hey,

On Wed, Sep 11, 2019 at 07:16:30AM -0700, Tejun Heo wrote:
> > >  I can implement
> > > the switching if so.
> > 
> > That would be perfect.
> 
> Whichever way it gets decided, this is easy enough.  I'll prep a
> patch.

Here's the patch.

* It disables iocost when bfq iosched is selected.  iocost is only
  re-enabled when bfq module is unloaded.  While it can easily be made
  to flip when there's no bfq queue left, I think minimizing the
  number of switches is likely is better.

* It looks like building bfq as module has been broken for some time
  and I added symbol exports to make it work.  They should be
  separated into a separate patch.

* It doesn't actually rename any of the bfq interface files.  It just
  gets rid of potential conflict from iocost.

Please feel free to use / modify the patch however you see fit.

Thanks.

---
diff --git a/block/bfq-cgroup.c b/block/bfq-cgroup.c
index 86a607cf19a1..decda96770f4 100644
--- a/block/bfq-cgroup.c
+++ b/block/bfq-cgroup.c
@@ -1194,7 +1194,9 @@ struct bfq_group *bfq_create_group_hierarchy(struct bfq_data *bfqd, int node)
 }
 
 struct blkcg_policy blkcg_policy_bfq = {
+#ifndef CONFIG_BLK_CGROUP_IOCOST
 	.dfl_cftypes		= bfq_blkg_files,
+#endif
 	.legacy_cftypes		= bfq_blkcg_legacy_files,
 
 	.cpd_alloc_fn		= bfq_cpd_alloc,
diff --git a/block/bfq-iosched.c b/block/bfq-iosched.c
index b33be928d164..b4aaeef1fb87 100644
--- a/block/bfq-iosched.c
+++ b/block/bfq-iosched.c
@@ -6365,6 +6365,36 @@ static void bfq_init_root_group(struct bfq_group *root_group,
 	root_group->sched_data.bfq_class_idle_last_service = jiffies;
 }
 
+#if defined(CONFIG_BFQ_GROUP_IOSCHED) && defined(CONFIG_BLK_CGROUP_IOCOST)
+static bool bfq_enabled = false;
+
+static void bfq_enable(void)
+{
+	static DEFINE_MUTEX(bfq_enable_mutex);
+
+	mutex_lock(&bfq_enable_mutex);
+	if (!bfq_enabled) {
+		pr_info("bfq-iosched: Overriding iocost\n");
+		blkcg_policy_unregister(&blkcg_policy_iocost);
+		cgroup_add_dfl_cftypes(&io_cgrp_subsys, bfq_blkg_files);
+		bfq_enabled = true;
+	}
+	mutex_unlock(&bfq_enable_mutex);
+}
+
+static void __exit bfq_disable(void)
+{
+	if (bfq_enabled) {
+		pr_info("bfq-iosched: Restoring iocost\n");
+		cgroup_rm_cftypes(bfq_blkg_files);
+		blkcg_policy_register(&blkcg_policy_iocost);
+	}
+}
+#else
+static void bfq_enable(void) {}
+static void __exit bfq_disable(void) {}
+#endif
+
 static int bfq_init_queue(struct request_queue *q, struct elevator_type *e)
 {
 	struct bfq_data *bfqd;
@@ -6489,6 +6519,7 @@ static int bfq_init_queue(struct request_queue *q, struct elevator_type *e)
 	bfq_init_entity(&bfqd->oom_bfqq.entity, bfqd->root_group);
 
 	wbt_disable_default(q);
+	bfq_enable();
 	return 0;
 
 out_free:
@@ -6806,6 +6837,7 @@ static void __exit bfq_exit(void)
 	blkcg_policy_unregister(&blkcg_policy_bfq);
 #endif
 	bfq_slab_kill();
+	bfq_disable();
 }
 
 module_init(bfq_init);
diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 0e2619c1a422..b6f20be0fc78 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -900,6 +900,7 @@ int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
 	}
 	return ret;
 }
+EXPORT_SYMBOL_GPL(blkg_conf_prep);
 
 /**
  * blkg_conf_finish - finish up per-blkg config update
@@ -915,6 +916,7 @@ void blkg_conf_finish(struct blkg_conf_ctx *ctx)
 	rcu_read_unlock();
 	put_disk_and_module(ctx->disk);
 }
+EXPORT_SYMBOL_GPL(blkg_conf_finish);
 
 static int blkcg_print_stat(struct seq_file *sf, void *v)
 {
diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index 3b39deb8b9f8..1ef5b443c09a 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -605,8 +605,6 @@ static u32 vrate_adj_pct[] =
 	  2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
 	  4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8, 8, 8, 8, 16 };
 
-static struct blkcg_policy blkcg_policy_iocost;
-
 /* accessors and helpers */
 static struct ioc *rqos_to_ioc(struct rq_qos *rqos)
 {
@@ -2434,7 +2432,7 @@ static struct cftype ioc_files[] = {
 	{}
 };
 
-static struct blkcg_policy blkcg_policy_iocost = {
+struct blkcg_policy blkcg_policy_iocost = {
 	.dfl_cftypes	= ioc_files,
 	.cpd_alloc_fn	= ioc_cpd_alloc,
 	.cpd_free_fn	= ioc_cpd_free,
@@ -2442,6 +2440,7 @@ static struct blkcg_policy blkcg_policy_iocost = {
 	.pd_init_fn	= ioc_pd_init,
 	.pd_free_fn	= ioc_pd_free,
 };
+EXPORT_SYMBOL_GPL(blkcg_policy_iocost);
 
 static int __init ioc_init(void)
 {
diff --git a/include/linux/blk-cgroup.h b/include/linux/blk-cgroup.h
index bed9e43f9426..5669e3cfa1bc 100644
--- a/include/linux/blk-cgroup.h
+++ b/include/linux/blk-cgroup.h
@@ -815,6 +815,11 @@ static inline void blkcg_clear_delay(struct blkcg_gq *blkg)
 void blkcg_add_delay(struct blkcg_gq *blkg, u64 now, u64 delta);
 void blkcg_schedule_throttle(struct request_queue *q, bool use_memdelay);
 void blkcg_maybe_throttle_current(void);
+
+#ifdef CONFIG_BLK_CGROUP_IOCOST
+extern struct blkcg_policy blkcg_policy_iocost;
+#endif
+
 #else	/* CONFIG_BLK_CGROUP */
 
 struct blkcg {
diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
index 753afbca549f..ced23229f359 100644
--- a/kernel/cgroup/cgroup.c
+++ b/kernel/cgroup/cgroup.c
@@ -4059,6 +4059,7 @@ int cgroup_rm_cftypes(struct cftype *cfts)
 	mutex_unlock(&cgroup_mutex);
 	return ret;
 }
+EXPORT_SYMBOL_GPL(cgroup_rm_cftypes);
 
 /**
  * cgroup_add_cftypes - add an array of cftypes to a subsystem
@@ -4115,6 +4116,7 @@ int cgroup_add_dfl_cftypes(struct cgroup_subsys *ss, struct cftype *cfts)
 		cft->flags |= __CFTYPE_ONLY_ON_DFL;
 	return cgroup_add_cftypes(ss, cfts);
 }
+EXPORT_SYMBOL_GPL(cgroup_add_dfl_cftypes);
 
 /**
  * cgroup_add_legacy_cftypes - add an array of cftypes for legacy hierarchies

Re: [PATCH 08/10] blkcg: implement blk-iocost

[Paolo Valente]

> Il giorno 11 set 2019, alle ore 16:16, Tejun Heo <tj@kernel.org> ha scritto:
> 
> Hello,
> 
> On Wed, Sep 11, 2019 at 10:18:53AM +0200, Paolo Valente wrote:
>>> The two being enabled at the same time doesn't make sense, so we can
>>> just switch over to bfq when bfq is selected as the iosched.  I asked
>>> what Paolo wanted to do in terms of interface a couple times now but
>>> didn't get an answer and he posted a patch which makes the two
>>> controllers conflict, so....  Paolo, so it looks like you want to
>>> rename all bfq files to drop the bfq prefix, right?
>> 
>> Yep, mainly because ... this is the solution you voted and you
>> yourself proposed [1] :)
>> 
>> [1] https://patchwork.kernel.org/patch/10988261/
> 
> So, that was then.  Since then the interface change has been published
> and userspace, at least some of them, already had to adjust.  Now, I
> don't have any opinion on the matter and flipping again will cause
> inconveniences to some subset of users.  It's your call.
> 

Ok, then let's go for removing bfq prefix, as expected, e.g., by
systemd community [1].  A few days ago I reposted a patch removing the
prefix from all involved parameters [2], it should make all legacy
software work again.

[1] https://github.com/systemd/systemd/issues/7057
[2] https://lkml.org/lkml/2019/9/9/47

>>> I can implement
>>> the switching if so.
>> 
>> That would be perfect.
> 
> Whichever way it gets decided, this is easy enough.  I'll prep a
> patch.
> 

Thank you for that too,
Paolo

> Thanks.
> 
> -- 
> tejun

Re: [PATCH 08/10] blkcg: implement blk-iocost

[Michal Koutný]

[-- Attachment #1: Type: text/plain, Size: 1161 bytes --]

Hi (and apology for relatively late reply).

On Tue, Sep 10, 2019 at 09:08:55AM -0700, Tejun Heo <tj@kernel.org> wrote:
> I can implement the switching if so.
I see the "conflict" is solved by the switching.

> Initially, I put them under block device sysfs but it was too clumsy
> with different config file formats and all.
Do you have any more details on that? In the end, it all boils down to a
daemon/setup utility writing into the control files and it can use
whatever config files it decides, can't it?

> I think it's better to have global controller configs at the root
> cgroup.
I agree with the "global controller" configs, however, does it also hold
for "global controller per-device" configs? They seem closer to the
device than the controller. Potentially, the parameters could be used by
some other consumers in the future. (I'm not opposing the current form,
I just want to explore the space before an API is fixed.)


> Not at all.  These are system-wide configs.  cgroup namespaces
> shouldn't have anything which aren't in non-root cgroups.
Thanks, I understand the cgroup namespaces are not meant to be
transparent to their members.

Michal

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [PATCH 08/10] blkcg: implement blk-iocost

[Tejun Heo]

Hello,

On Thu, Oct 03, 2019 at 04:51:06PM +0200, Michal Koutný wrote:
> > Initially, I put them under block device sysfs but it was too clumsy
> > with different config file formats and all.
> Do you have any more details on that? In the end, it all boils down to a
> daemon/setup utility writing into the control files and it can use
> whatever config files it decides, can't it?

Yeah, I mean, we can make any interface work.  So, there are two
global knobs io.cost.model and io.cost.qos.  Of the tw, io.cost.model
is okay to move under block device but the qos file gets weird because
the content of the file is more resource control policies than device
properties.

Thanks.

-- 
tejun

Re: [PATCH 08/10] blkcg: implement blk-iocost

[Michal Koutný]

[-- Attachment #1: Type: text/plain, Size: 2050 bytes --]

On Thu, Oct 03, 2019 at 09:45:52AM -0700, Tejun Heo <tj@kernel.org> wrote:
> [...] but the qos file gets weird because the content of the file is
> more resource control policies than device properties.
I see two facets on this -- the semantics of the QoS controls and
storing controller parameters generally.

Because I'm not fully convinced using the root cgroup for the latter is
a good idea and I don't have a better one (what about
/sys/kernel/cgroup/?), I'd like to question the former to potentially
postpone finding the place for its parameters :-)


On Wed, Aug 28, 2019 at 03:05:58PM -0700, Tejun Heo <tj@kernel.org> wrote:
> [...]
> Please see the top comment in blk-iocost.c and documentation for
> more details.
I admit I did't grasp the explanations in the cgroup-v2.rst, perhaps
some of the explanations from blk-iocost.c would be useful there as
well.

IIUC, the controls are supposed to be abstracted and generic to express
high-level ideas and be independent of particular details.
Here a bunch of parameters is introduced whose tuning may become a
complex optimization task.

What is the metric that is the QoS controller striving to guarantee?
How does it differ from the io.latency policy?


> [...] 
> + * 2-2. Vrate Adjustment
> + * [...] When this delay becomes noticeable, it's a clear
> + * indication that the device is saturated and we lower the vrate.  This
> + * saturation signal is fairly conservative as it only triggers when both
> + * hardware and software queues are filled up, and is used as the default
> + * busy signal.
(The following paragraph is based only on naïve understanding of the
block layer.) So the device's vrate is lowered, causing its vtime
growing slower, i.e.  postponing issuing an IO later for all cgroups
accessing the device. But what's the purpose of this? If the queues fill
up, wouldn't be all naturally pushed back by the longer queue time
anyway? And wouldn't slowing down the device's vtime just cause queueing
elsewhere?

Thanks,
Michal

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [PATCH 08/10] blkcg: implement blk-iocost

[Tejun Heo]

Hello,

On Wed, Oct 09, 2019 at 05:36:29PM +0200, Michal Koutný wrote:
> Because I'm not fully convinced using the root cgroup for the latter is
> a good idea and I don't have a better one (what about
> /sys/kernel/cgroup/?), I'd like to question the former to potentially
> postpone finding the place for its parameters :-)

Yeah, I mean, I don't know.  If these params were useful outside
iocost controller itself, under /sys/block would be a better place but
it's kind tightly tied to vrate.  We likely can talk on the subject
for a really long time probalby because there's no clearly technically
better choice here, so...

> On Wed, Aug 28, 2019 at 03:05:58PM -0700, Tejun Heo <tj@kernel.org> wrote:
> > [...]
> > Please see the top comment in blk-iocost.c and documentation for
> > more details.
> I admit I did't grasp the explanations in the cgroup-v2.rst, perhaps
> some of the explanations from blk-iocost.c would be useful there as
> well.
> 
> IIUC, the controls are supposed to be abstracted and generic to express
> high-level ideas and be independent of particular details.
> Here a bunch of parameters is introduced whose tuning may become a
> complex optimization task.
> 
> What is the metric that is the QoS controller striving to guarantee?
> How does it differ from the io.latency policy?

Yeah, it's kinda unfortunate that it requires this many parameters but
at least my opinion is that that's reflecting the inherent
complexities of the underlying devices and how workloads interact with
them.  Andy knows and can explain this a lot better than me but here's
what's we're working on:

For the cost model, the plan is to build a database of model-specific
model parameters which are loaded during boot.  The cost model
parameters are pretty straight forward to determine, so hopefully this
won't be too difficult.

For QoS parameters, Andy is currently working on a method to determine
the set of parametesr which are at the edge of total work cliff -
ie. the point where tighetning QoS params further starts reducing the
total amount of work the device can do significantly.  This would be
the neutral parameters to use for a given device unless there are
overriding latency requirements, so it's likely that this can be part
of the model-specific parameter set.

We're currently deploying the controller to a lot of machines and
gathering data to verify model accuracies and controller behaviors.
It's working pretty well already and once the methods become more
mature, we'll upstream them (whichever projects they end up
belonging).


> > [...] 
> > + * 2-2. Vrate Adjustment
> > + * [...] When this delay becomes noticeable, it's a clear
> > + * indication that the device is saturated and we lower the vrate.  This
> > + * saturation signal is fairly conservative as it only triggers when both
> > + * hardware and software queues are filled up, and is used as the default
> > + * busy signal.
> (The following paragraph is based only on naïve understanding of the
> block layer.) So the device's vrate is lowered, causing its vtime
> growing slower, i.e.  postponing issuing an IO later for all cgroups
> accessing the device. But what's the purpose of this? If the queues fill
> up, wouldn't be all naturally pushed back by the longer queue time
> anyway? And wouldn't slowing down the device's vtime just cause queueing
> elsewhere?

Nothing can issue IOs indefinitely without some of them completing and
the total amount of work a workload can do is conjoined with the
completion latencies.  Most IO devices have queue depth which is at
some level reasonable given the performance characteritics of the
device; otherwise, the device would develop a really fat pipe in it
which can frustrate various use cases.  On top, block layer adds some
limited amount of queueing to avoid command bubbles (2x qd, usually),
so, while definitely not stringent in any way, the queueing is already
regulated so that things don't get too crazy.

Regulating based on qd may not be enough for latency sensitive
synchronous workloads; however, for a lot of workloads such as reading
file contents or copying them which have in-kernel windowing
mechanisms, it can provide a reasonable level of protection to keep
the effectiveness of the windowing mechanisms without sacrificing
noticeable level of total work.

Thanks.

-- 
tejun

Re: [PATCH 08/10] blkcg: implement blk-iocost

[Michal Koutný]

[-- Attachment #1: Type: text/plain, Size: 2053 bytes --]

Hello

(I realize it's likely late for the remark but I'd like to bring it up
anyway.)

On Mon, Oct 14, 2019 at 08:36:43AM -0700, Tejun Heo <tj@kernel.org> wrote:
> We likely can talk on the subject
> for a really long time probalby because there's no clearly technically
> better choice here, so...
I agree with you that functionally the two options are equal and also
from configuration POV they seem both sensible.

I checked where BFQ stores its per-device parameters and its under the
sysfs directory of given device's iosched directory. So from the user
perspective it'd be more consistent if all similar tunables resided
under that location.

(OTOH, I admit I'm not that familiar with block layer internals to
identify the overlap between IO scheduler and IO controller.)

> Yeah, it's kinda unfortunate that it requires this many parameters but
> at least my opinion is that that's reflecting the inherent
> complexities of the underlying devices and how workloads interact with
> them.
After I learnt about the existence of BFQ tunables, I'm no longer
concerned by the complexity of the parameter space.

Thanks for the explanations of QoS purpose.

> For QoS parameters, Andy is currently working on a method to determine
> the set of parametesr which are at the edge of total work cliff -
> ie. the point where tighetning QoS params further starts reducing the
> total amount of work the device can do significantly.
The QoS description in the Documentation/ describes the interpretation
of the individual parameters, however, this purpose and how it works was
not clear to be from that. I think the QoS policy would deserve similar
description in the Documentation/.

> Nothing can issue IOs indefinitely without some of them completing and
> the total amount of work a workload can do is conjoined with the
> completion latencies. [...]
I may reply to this point later. However, if that provably works, I'm
likely missing something in my understanding, so that'd be irrelevant.

Cheers,
Michal

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [PATCH 08/10] blkcg: implement blk-iocost

[Paolo Valente]

> Il giorno 1 nov 2019, alle ore 17:15, Michal Koutný <mkoutny@suse.com> ha scritto:
> 
> Hello
> 

Hi Michal,

> (I realize it's likely late for the remark but I'd like to bring it up
> anyway.)
> 
> On Mon, Oct 14, 2019 at 08:36:43AM -0700, Tejun Heo <tj@kernel.org> wrote:
>> We likely can talk on the subject
>> for a really long time probalby because there's no clearly technically
>> better choice here, so...
> I agree with you that functionally the two options are equal and also
> from configuration POV they seem both sensible.
> 
> I checked where BFQ stores its per-device parameters and its under the
> sysfs directory of given device's iosched directory. So from the user
> perspective it'd be more consistent if all similar tunables resided
> under that location.
> 
> (OTOH, I admit I'm not that familiar with block layer internals to
> identify the overlap between IO scheduler and IO controller.)
> 

If useful for you to know, BFQ parameters are not meant to changed
(apart from the low_latency tunable, if one wants full control on
weights).  Parameters are a testing aid, to use in case of an anomaly.
After solving the anomaly, default values should be used again.

Thanks,
Paolo

>> Yeah, it's kinda unfortunate that it requires this many parameters but
>> at least my opinion is that that's reflecting the inherent
>> complexities of the underlying devices and how workloads interact with
>> them.
> After I learnt about the existence of BFQ tunables, I'm no longer
> concerned by the complexity of the parameter space.
> 
> Thanks for the explanations of QoS purpose.
> 
>> For QoS parameters, Andy is currently working on a method to determine
>> the set of parametesr which are at the edge of total work cliff -
>> ie. the point where tighetning QoS params further starts reducing the
>> total amount of work the device can do significantly.
> The QoS description in the Documentation/ describes the interpretation
> of the individual parameters, however, this purpose and how it works was
> not clear to be from that. I think the QoS policy would deserve similar
> description in the Documentation/.
> 
>> Nothing can issue IOs indefinitely without some of them completing and
>> the total amount of work a workload can do is conjoined with the
>> completion latencies. [...]
> I may reply to this point later. However, if that provably works, I'm
> likely missing something in my understanding, so that'd be irrelevant.
> 
> Cheers,
> Michal

[PATCH 09/10] blkcg: add tools/cgroup/iocost_monitor.py

[Tejun Heo]

Instead of mucking with debugfs and ->pd_stat(), add drgn based
monitoring script.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Omar Sandoval <osandov@fb.com>
---
 block/blk-iocost.c             |  21 +++
 tools/cgroup/iocost_monitor.py | 270 +++++++++++++++++++++++++++++++++
 2 files changed, 291 insertions(+)
 create mode 100644 tools/cgroup/iocost_monitor.py

diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index 680815620095..3208d2fdc55e 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -149,6 +149,27 @@
  * donate and should take back how much requires hweight propagations
  * anyway making it easier to implement and understand as a separate
  * mechanism.
+ *
+ * 3. Monitoring
+ *
+ * Instead of debugfs or other clumsy monitoring mechanisms, this
+ * controller uses a drgn based monitoring script -
+ * tools/cgroup/iocost_monitor.py.  For details on drgn, please see
+ * https://github.com/osandov/drgn.  The ouput looks like the following.
+ *
+ *  sdb RUN   per=300ms cur_per=234.218:v203.695 busy= +1 vrate= 62.12%
+ *                 active      weight      hweight% inflt% del_ms usages%
+ *  test/a              *    50/   50  33.33/ 33.33  27.65  0*041 033:033:033
+ *  test/b              *   100/  100  66.67/ 66.67  17.56  0*000 066:079:077
+ *
+ * - per	: Timer period
+ * - cur_per	: Internal wall and device vtime clock
+ * - vrate	: Device virtual time rate against wall clock
+ * - weight	: Surplus-adjusted and configured weights
+ * - hweight	: Surplus-adjusted and configured hierarchical weights
+ * - inflt	: The percentage of in-flight IO cost at the end of last period
+ * - del_ms	: Deferred issuer delay induction level and duration
+ * - usages	: Usage history
  */
 
 #include <linux/kernel.h>
diff --git a/tools/cgroup/iocost_monitor.py b/tools/cgroup/iocost_monitor.py
new file mode 100644
index 000000000000..2c9445e966d8
--- /dev/null
+++ b/tools/cgroup/iocost_monitor.py
@@ -0,0 +1,270 @@
+#!/usr/bin/env drgn
+#
+# Copyright (C) 2019 Tejun Heo <tj@kernel.org>
+# Copyright (C) 2019 Facebook
+
+desc = """
+This is a drgn script to monitor the blk-iocost cgroup controller.
+See the comment at the top of block/blk-iocost.c for more details.
+For drgn, visit https://github.com/osandov/drgn.
+"""
+
+import sys
+import re
+import time
+import json
+
+import drgn
+from drgn import container_of
+from drgn.helpers.linux.list import list_for_each_entry,list_empty
+from drgn.helpers.linux.radixtree import radix_tree_for_each,radix_tree_lookup
+
+import argparse
+parser = argparse.ArgumentParser(description=desc,
+                                 formatter_class=argparse.RawTextHelpFormatter)
+parser.add_argument('devname', metavar='DEV',
+                    help='Target block device name (e.g. sda)')
+parser.add_argument('--cgroup', action='append', metavar='REGEX',
+                    help='Regex for target cgroups, ')
+parser.add_argument('--interval', '-i', metavar='SECONDS', type=float, default=1,
+                    help='Monitoring interval in seconds')
+parser.add_argument('--json', action='store_true',
+                    help='Output in json')
+args = parser.parse_args()
+
+def err(s):
+    print(s, file=sys.stderr, flush=True)
+    sys.exit(1)
+
+try:
+    blkcg_root = prog['blkcg_root']
+    plid = prog['blkcg_policy_iocost'].plid.value_()
+except:
+    err('The kernel does not have iocost enabled')
+
+IOC_RUNNING     = prog['IOC_RUNNING'].value_()
+NR_USAGE_SLOTS  = prog['NR_USAGE_SLOTS'].value_()
+HWEIGHT_WHOLE   = prog['HWEIGHT_WHOLE'].value_()
+VTIME_PER_SEC   = prog['VTIME_PER_SEC'].value_()
+VTIME_PER_USEC  = prog['VTIME_PER_USEC'].value_()
+AUTOP_SSD_FAST  = prog['AUTOP_SSD_FAST'].value_()
+AUTOP_SSD_DFL   = prog['AUTOP_SSD_DFL'].value_()
+AUTOP_SSD_QD1   = prog['AUTOP_SSD_QD1'].value_()
+AUTOP_HDD       = prog['AUTOP_HDD'].value_()
+
+autop_names = {
+    AUTOP_SSD_FAST:        'ssd_fast',
+    AUTOP_SSD_DFL:         'ssd_dfl',
+    AUTOP_SSD_QD1:         'ssd_qd1',
+    AUTOP_HDD:             'hdd',
+}
+
+class BlkgIterator:
+    def blkcg_name(blkcg):
+        return blkcg.css.cgroup.kn.name.string_().decode('utf-8')
+
+    def walk(self, blkcg, q_id, parent_path):
+        if not self.include_dying and \
+           not (blkcg.css.flags.value_() & prog['CSS_ONLINE'].value_()):
+            return
+
+        name = BlkgIterator.blkcg_name(blkcg)
+        path = parent_path + '/' + name if parent_path else name
+        blkg = drgn.Object(prog, 'struct blkcg_gq',
+                           address=radix_tree_lookup(blkcg.blkg_tree, q_id))
+        if not blkg.address_:
+            return
+
+        self.blkgs.append((path if path else '/', blkg))
+
+        for c in list_for_each_entry('struct blkcg',
+                                     blkcg.css.children.address_of_(), 'css.sibling'):
+            self.walk(c, q_id, path)
+
+    def __init__(self, root_blkcg, q_id, include_dying=False):
+        self.include_dying = include_dying
+        self.blkgs = []
+        self.walk(root_blkcg, q_id, '')
+
+    def __iter__(self):
+        return iter(self.blkgs)
+
+class IocStat:
+    def __init__(self, ioc):
+        global autop_names
+
+        self.enabled = ioc.enabled.value_()
+        self.running = ioc.running.value_() == IOC_RUNNING
+        self.period_ms = round(ioc.period_us.value_() / 1_000)
+        self.period_at = ioc.period_at.value_() / 1_000_000
+        self.vperiod_at = ioc.period_at_vtime.value_() / VTIME_PER_SEC
+        self.vrate_pct = ioc.vtime_rate.counter.value_() * 100 / VTIME_PER_USEC
+        self.busy_level = ioc.busy_level.value_()
+        self.autop_idx = ioc.autop_idx.value_()
+        self.user_cost_model = ioc.user_cost_model.value_()
+        self.user_qos_params = ioc.user_qos_params.value_()
+
+        if self.autop_idx in autop_names:
+            self.autop_name = autop_names[self.autop_idx]
+        else:
+            self.autop_name = '?'
+
+    def dict(self, now):
+        return { 'device'               : devname,
+                 'timestamp'            : now,
+                 'enabled'              : self.enabled,
+                 'running'              : self.running,
+                 'period_ms'            : self.period_ms,
+                 'period_at'            : self.period_at,
+                 'period_vtime_at'      : self.vperiod_at,
+                 'busy_level'           : self.busy_level,
+                 'vrate_pct'            : self.vrate_pct, }
+
+    def table_preamble_str(self):
+        state = ('RUN' if self.running else 'IDLE') if self.enabled else 'OFF'
+        output = f'{devname} {state:4} ' \
+                 f'per={self.period_ms}ms ' \
+                 f'cur_per={self.period_at:.3f}:v{self.vperiod_at:.3f} ' \
+                 f'busy={self.busy_level:+3} ' \
+                 f'vrate={self.vrate_pct:6.2f}% ' \
+                 f'params={self.autop_name}'
+        if self.user_cost_model or self.user_qos_params:
+            output += f'({"C" if self.user_cost_model else ""}{"Q" if self.user_qos_params else ""})'
+        return output
+
+    def table_header_str(self):
+        return f'{"":25} active {"weight":>9} {"hweight%":>13} {"inflt%":>6} ' \
+               f'{"del_ms":>6} {"usages%"}'
+
+class IocgStat:
+    def __init__(self, iocg):
+        ioc = iocg.ioc
+        blkg = iocg.pd.blkg
+
+        self.is_active = not list_empty(iocg.active_list.address_of_())
+        self.weight = iocg.weight.value_()
+        self.active = iocg.active.value_()
+        self.inuse = iocg.inuse.value_()
+        self.hwa_pct = iocg.hweight_active.value_() * 100 / HWEIGHT_WHOLE
+        self.hwi_pct = iocg.hweight_inuse.value_() * 100 / HWEIGHT_WHOLE
+
+        vdone = iocg.done_vtime.counter.value_()
+        vtime = iocg.vtime.counter.value_()
+        vrate = ioc.vtime_rate.counter.value_()
+        period_vtime = ioc.period_us.value_() * vrate
+        if period_vtime:
+            self.inflight_pct = (vtime - vdone) * 100 / period_vtime
+        else:
+            self.inflight_pct = 0
+
+        self.use_delay = min(blkg.use_delay.counter.value_(), 99)
+        self.delay_ms = min(round(blkg.delay_nsec.counter.value_() / 1_000_000), 999)
+
+        usage_idx = iocg.usage_idx.value_()
+        self.usages = []
+        self.usage = 0
+        for i in range(NR_USAGE_SLOTS):
+            usage = iocg.usages[(usage_idx + i) % NR_USAGE_SLOTS].value_()
+            upct = min(usage * 100 / HWEIGHT_WHOLE, 999)
+            self.usages.append(upct)
+            self.usage = max(self.usage, upct)
+
+    def dict(self, now, path):
+        out = { 'cgroup'                : path,
+                'timestamp'             : now,
+                'is_active'             : self.is_active,
+                'weight'                : self.weight,
+                'weight_active'         : self.active,
+                'weight_inuse'          : self.inuse,
+                'hweight_active_pct'    : self.hwa_pct,
+                'hweight_inuse_pct'     : self.hwi_pct,
+                'inflight_pct'          : self.inflight_pct,
+                'use_delay'             : self.use_delay,
+                'delay_ms'              : self.delay_ms,
+                'usage_pct'             : self.usage }
+        for i in range(len(self.usages)):
+            out[f'usage_pct_{i}'] = f'{self.usages[i]}'
+        return out
+
+    def table_row_str(self, path):
+        out = f'{path[-28:]:28} ' \
+              f'{"*" if self.is_active else " "} ' \
+              f'{self.inuse:5}/{self.active:5} ' \
+              f'{self.hwi_pct:6.2f}/{self.hwa_pct:6.2f} ' \
+              f'{self.inflight_pct:6.2f} ' \
+              f'{self.use_delay:2}*{self.delay_ms:03} '
+        for u in self.usages:
+            out += f'{round(u):03d}:'
+        out = out.rstrip(':')
+        return out
+
+# handle args
+table_fmt = not args.json
+interval = args.interval
+devname = args.devname
+
+if args.json:
+    table_fmt = False
+
+re_str = None
+if args.cgroup:
+    for r in args.cgroup:
+        if re_str is None:
+            re_str = r
+        else:
+            re_str += '|' + r
+
+filter_re = re.compile(re_str) if re_str else None
+
+# Locate the roots
+q_id = None
+root_iocg = None
+ioc = None
+
+for i, ptr in radix_tree_for_each(blkcg_root.blkg_tree):
+    blkg = drgn.Object(prog, 'struct blkcg_gq', address=ptr)
+    try:
+        if devname == blkg.q.kobj.parent.name.string_().decode('utf-8'):
+            q_id = blkg.q.id.value_()
+            if blkg.pd[plid]:
+                root_iocg = container_of(blkg.pd[plid], 'struct ioc_gq', 'pd')
+                ioc = root_iocg.ioc
+            break
+    except:
+        pass
+
+if ioc is None:
+    err(f'Could not find ioc for {devname}');
+
+# Keep printing
+while True:
+    now = time.time()
+    iocstat = IocStat(ioc)
+    output = ''
+
+    if table_fmt:
+        output += '\n' + iocstat.table_preamble_str()
+        output += '\n' + iocstat.table_header_str()
+    else:
+        output += json.dumps(iocstat.dict(now))
+
+    for path, blkg in BlkgIterator(blkcg_root, q_id):
+        if filter_re and not filter_re.match(path):
+            continue
+        if not blkg.pd[plid]:
+            continue
+
+        iocg = container_of(blkg.pd[plid], 'struct ioc_gq', 'pd')
+        iocg_stat = IocgStat(iocg)
+
+        if not filter_re and not iocg_stat.is_active:
+            continue
+
+        if table_fmt:
+            output += '\n' + iocg_stat.table_row_str(path)
+        else:
+            output += '\n' + json.dumps(iocg_stat.dict(now, path))
+
+    print(output)
+    sys.stdout.flush()
+    time.sleep(interval)
-- 
2.17.1

[PATCH 10/10] blkcg: add tools/cgroup/iocost_coef_gen.py

[Tejun Heo]

Add a script which can be used to generate device-specific iocost
linear model coefficients.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 Documentation/admin-guide/cgroup-v2.rst |   3 +
 block/blk-iocost.c                      |   3 +
 tools/cgroup/iocost_coef_gen.py         | 178 ++++++++++++++++++++++++
 3 files changed, 184 insertions(+)
 create mode 100644 tools/cgroup/iocost_coef_gen.py

diff --git a/Documentation/admin-guide/cgroup-v2.rst b/Documentation/admin-guide/cgroup-v2.rst
index 1521c7e554f5..3deacdc5e6d2 100644
--- a/Documentation/admin-guide/cgroup-v2.rst
+++ b/Documentation/admin-guide/cgroup-v2.rst
@@ -1529,6 +1529,9 @@ IO Interface Files
 	The IO cost model isn't expected to be accurate in absolute
 	sense and is scaled to the device behavior dynamically.
 
+	If needed, tools/cgroup/iocost_coef_gen.py can be used to
+	generate device-specific coefficients.
+
   io.weight
 	A read-write flat-keyed file which exists on non-root cgroups.
 	The default is "default 100".
diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index 3208d2fdc55e..f04a4ed1cb45 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -46,6 +46,9 @@
  * If needed, tools/cgroup/iocost_coef_gen.py can be used to generate
  * device-specific coefficients.
  *
+ * If needed, tools/cgroup/iocost_coef_gen.py can be used to generate
+ * device-specific coefficients.
+ *
  * 2. Control Strategy
  *
  * The device virtual time (vtime) is used as the primary control metric.
diff --git a/tools/cgroup/iocost_coef_gen.py b/tools/cgroup/iocost_coef_gen.py
new file mode 100644
index 000000000000..df17a2ae80e5
--- /dev/null
+++ b/tools/cgroup/iocost_coef_gen.py
@@ -0,0 +1,178 @@
+#!/usr/bin/env python3
+#
+# Copyright (C) 2019 Tejun Heo <tj@kernel.org>
+# Copyright (C) 2019 Andy Newell <newella@fb.com>
+# Copyright (C) 2019 Facebook
+
+desc = """
+Generate linear IO cost model coefficients used by the blk-iocost
+controller.  If the target raw testdev is specified, destructive tests
+are performed against the whole device; otherwise, on
+./iocost-coef-fio.testfile.  The result can be written directly to
+/sys/fs/cgroup/io.cost.model.
+
+On high performance devices, --numjobs > 1 is needed to achieve
+saturation.
+
+See Documentation/admin-guide/cgroup-v2.rst and block/blk-iocost.c
+for more details.
+"""
+
+import argparse
+import re
+import json
+import glob
+import os
+import sys
+import atexit
+import shutil
+import tempfile
+import subprocess
+
+parser = argparse.ArgumentParser(description=desc,
+                                 formatter_class=argparse.RawTextHelpFormatter)
+parser.add_argument('--testdev', metavar='DEV',
+                    help='Raw block device to use for testing, ignores --testfile-size')
+parser.add_argument('--testfile-size-gb', type=float, metavar='GIGABYTES', default=16,
+                    help='Testfile size in gigabytes (default: %(default)s)')
+parser.add_argument('--duration', type=int, metavar='SECONDS', default=120,
+                    help='Individual test run duration in seconds (default: %(default)s)')
+parser.add_argument('--seqio-block-mb', metavar='MEGABYTES', type=int, default=128,
+                    help='Sequential test block size in megabytes (default: %(default)s)')
+parser.add_argument('--seq-depth', type=int, metavar='DEPTH', default=64,
+                    help='Sequential test queue depth (default: %(default)s)')
+parser.add_argument('--rand-depth', type=int, metavar='DEPTH', default=64,
+                    help='Random test queue depth (default: %(default)s)')
+parser.add_argument('--numjobs', type=int, metavar='JOBS', default=1,
+                    help='Number of parallel fio jobs to run (default: %(default)s)')
+parser.add_argument('--quiet', action='store_true')
+parser.add_argument('--verbose', action='store_true')
+
+def info(msg):
+    if not args.quiet:
+        print(msg)
+
+def dbg(msg):
+    if args.verbose and not args.quiet:
+        print(msg)
+
+# determine ('DEVNAME', 'MAJ:MIN') for @path
+def dir_to_dev(path):
+    # find the block device the current directory is on
+    devname = subprocess.run(f'findmnt -nvo SOURCE -T{path}',
+                             stdout=subprocess.PIPE, shell=True).stdout
+    devname = os.path.basename(devname).decode('utf-8').strip()
+
+    # partition -> whole device
+    parents = glob.glob('/sys/block/*/' + devname)
+    if len(parents):
+        devname = os.path.basename(os.path.dirname(parents[0]))
+    rdev = os.stat(f'/dev/{devname}').st_rdev
+    return (devname, f'{os.major(rdev)}:{os.minor(rdev)}')
+
+def create_testfile(path, size):
+    global args
+
+    if os.path.isfile(path) and os.stat(path).st_size == size:
+        return
+
+    info(f'Creating testfile {path}')
+    subprocess.check_call(f'rm -f {path}', shell=True)
+    subprocess.check_call(f'touch {path}', shell=True)
+    subprocess.call(f'chattr +C {path}', shell=True)
+    subprocess.check_call(
+        f'pv -s {size} -pr /dev/urandom {"-q" if args.quiet else ""} | '
+        f'dd of={path} count={size} '
+        f'iflag=count_bytes,fullblock oflag=direct bs=16M status=none',
+        shell=True)
+
+def run_fio(testfile, duration, iotype, iodepth, blocksize, jobs):
+    global args
+
+    eta = 'never' if args.quiet else 'always'
+    outfile = tempfile.NamedTemporaryFile()
+    cmd = (f'fio --direct=1 --ioengine=libaio --name=coef '
+           f'--filename={testfile} --runtime={round(duration)} '
+           f'--readwrite={iotype} --iodepth={iodepth} --blocksize={blocksize} '
+           f'--eta={eta} --output-format json --output={outfile.name} '
+           f'--time_based --numjobs={jobs}')
+    if args.verbose:
+        dbg(f'Running {cmd}')
+    subprocess.check_call(cmd, shell=True)
+    with open(outfile.name, 'r') as f:
+        d = json.loads(f.read())
+    return sum(j['read']['bw_bytes'] + j['write']['bw_bytes'] for j in d['jobs'])
+
+def restore_elevator_nomerges():
+    global elevator_path, nomerges_path, elevator, nomerges
+
+    info(f'Restoring elevator to {elevator} and nomerges to {nomerges}')
+    with open(elevator_path, 'w') as f:
+        f.write(elevator)
+    with open(nomerges_path, 'w') as f:
+        f.write(nomerges)
+
+
+args = parser.parse_args()
+
+missing = False
+for cmd in [ 'findmnt', 'pv', 'dd', 'fio' ]:
+    if not shutil.which(cmd):
+        print(f'Required command "{cmd}" is missing', file=sys.stderr)
+        missing = True
+if missing:
+    sys.exit(1)
+
+if args.testdev:
+    devname = os.path.basename(args.testdev)
+    rdev = os.stat(f'/dev/{devname}').st_rdev
+    devno = f'{os.major(rdev)}:{os.minor(rdev)}'
+    testfile = f'/dev/{devname}'
+    info(f'Test target: {devname}({devno})')
+else:
+    devname, devno = dir_to_dev('.')
+    testfile = 'iocost-coef-fio.testfile'
+    testfile_size = int(args.testfile_size_gb * 2 ** 30)
+    create_testfile(testfile, testfile_size)
+    info(f'Test target: {testfile} on {devname}({devno})')
+
+elevator_path = f'/sys/block/{devname}/queue/scheduler'
+nomerges_path = f'/sys/block/{devname}/queue/nomerges'
+
+with open(elevator_path, 'r') as f:
+    elevator = re.sub(r'.*\[(.*)\].*', r'\1', f.read().strip())
+with open(nomerges_path, 'r') as f:
+    nomerges = f.read().strip()
+
+info(f'Temporarily disabling elevator and merges')
+atexit.register(restore_elevator_nomerges)
+with open(elevator_path, 'w') as f:
+    f.write('none')
+with open(nomerges_path, 'w') as f:
+    f.write('1')
+
+info('Determining rbps...')
+rbps = run_fio(testfile, args.duration, 'read',
+               1, args.seqio_block_mb * (2 ** 20), args.numjobs)
+info(f'\nrbps={rbps}, determining rseqiops...')
+rseqiops = round(run_fio(testfile, args.duration, 'read',
+                         args.seq_depth, 4096, args.numjobs) / 4096)
+info(f'\nrseqiops={rseqiops}, determining rrandiops...')
+rrandiops = round(run_fio(testfile, args.duration, 'randread',
+                          args.rand_depth, 4096, args.numjobs) / 4096)
+info(f'\nrrandiops={rrandiops}, determining wbps...')
+wbps = run_fio(testfile, args.duration, 'write',
+               1, args.seqio_block_mb * (2 ** 20), args.numjobs)
+info(f'\nwbps={wbps}, determining wseqiops...')
+wseqiops = round(run_fio(testfile, args.duration, 'write',
+                         args.seq_depth, 4096, args.numjobs) / 4096)
+info(f'\nwseqiops={wseqiops}, determining wrandiops...')
+wrandiops = round(run_fio(testfile, args.duration, 'randwrite',
+                          args.rand_depth, 4096, args.numjobs) / 4096)
+info(f'\nwrandiops={wrandiops}')
+restore_elevator_nomerges()
+atexit.unregister(restore_elevator_nomerges)
+info('')
+
+print(f'{devno} rbps={rbps} rseqiops={rseqiops} rrandiops={rrandiops} '
+      f'wbps={wbps} wseqiops={wseqiops} wrandiops={wrandiops}')
-- 
2.17.1

Re: [PATCHSET v3 block/for-linus] IO cost model based work-conserving porportional controller

[Jens Axboe]

On 8/28/19 4:05 PM, Tejun Heo wrote:
> Changes from v2[2]:
> 
> * Fixed a divide-by-zero bug in current_hweight().
> 
> * pre_start_time and friends renamed to alloc_time and now has its own
>    CONFIG option which is selected by IOCOST.

This looks fine to me now, I've queued it up for 5.4.

-- 
Jens Axboe

Re: [PATCH 07/10] blk-mq: add optional request->alloc_time_ns

[Tejun Heo]

Hello,

On Thu, Aug 29, 2019 at 04:22:48PM +0800, Hillf Danton wrote:
> >  static struct request *blk_mq_rq_ctx_init(struct blk_mq_alloc_data *data,
> > -		unsigned int tag, unsigned int op)
> > +		unsigned int tag, unsigned int op, u64 alloc_time_ns)
> >  {
> >  	struct blk_mq_tags *tags = blk_mq_tags_from_data(data);
> >  	struct request *rq = tags->static_rqs[tag];
> > @@ -325,6 +325,9 @@ static struct request *blk_mq_rq_ctx_init(struct blk_mq_alloc_data *data,
> >  	RB_CLEAR_NODE(&rq->rb_node);
> >  	rq->rq_disk = NULL;
> >  	rq->part = NULL;
> > +#ifdef CONFIG_BLK_RQ_ALLOC_TIME
> 
> Not only matches start_time, but everal hunks may collapse into one
> if checking alloc_time is lifted up.
> 
> 	if (blk_queue_rq_alloc_time(rq->q))
> 		rq->alloc_time_ns = ktime_get_ns();
> 	else
> 		rq->alloc_time_ns = 0;

Can you please elaborate?  Lifted up where?

Thanks.

-- 
tejun

Re: [PATCH 08/10] blkcg: implement blk-iocost

[Tejun Heo]

Hello,

On Thu, Aug 29, 2019 at 09:39:28PM +0800, Hillf Danton wrote:
> > +	on_q_ns = ktime_get_ns() - rq->alloc_time_ns;
> > +	rq_wait_ns = rq->start_time_ns - rq->alloc_time_ns;
> > +
> ehm... alloc_time makes no sense wrt start_time if this is their only
> use. What are you trying to measure with the two stamps if they are
> meaningfully apart?

Queue depth depletion.

> > +	rq_qos_add(q, rqos);
> > +	ret = blkcg_activate_policy(q, &blkcg_policy_iocost);
> > +	if (ret) {
> > +		rq_qos_del(q, rqos);
> 
> 		free_percpu(ioc->pcpu_stat);

Good catch.  Will send a follow up patch.

Thanks.

-- 
tejun

Re: [PATCHSET v3 block/for-linus] IO cost model based work-conserving porportional controller

[Paolo Valente]

Hi,
I see an important interface problem.  Userspace has been waiting for
io.weight to become eventually the file name for setting the weight
for the proportional-share policy [1,2].  If you use that name, how
will we solve this?

Thanks,
Paolo

[1] https://github.com/systemd/systemd/issues/7057#issuecomment-522747575
[2] https://github.com/systemd/systemd/pull/13335#issuecomment-523035303

> Il giorno 29 ago 2019, alle ore 00:05, Tejun Heo <tj@kernel.org> ha scritto:
> 
> Changes from v2[2]:
> 
> * Fixed a divide-by-zero bug in current_hweight().
> 
> * pre_start_time and friends renamed to alloc_time and now has its own
>  CONFIG option which is selected by IOCOST.
> 
> Changes from v1[1]:
> 
> * Prerequisite patchsets had cosmetic changes and merged.  Refreshed
>  on top.
> 
> * Renamed from ioweight to iocost.  All source code and tools are
>  updated accordingly.  Control knobs io.weight.qos and
>  io.weight.cost_model are renamed to io.cost.qos and io.cost.model
>  respectively.  This is a more fitting name which won't become a
>  misnomer when, for example, cost based io.max is added.
> 
> * Various bug fixes and improvements.  A few bugs were discovered
>  while testing against high-iops nvme device.  Auto parameter
>  selection improved and verified across different classes of SSDs.
> 
> * Dropped bpf iocost support for now.
> 
> * Added coef generation script.
> 
> * Verified on high-iops nvme device.  Result is included below.
> 
> One challenge of controlling IO resources is the lack of trivially
> observable cost metric.  This is distinguished from CPU and memory
> where wallclock time and the number of bytes can serve as accurate
> enough approximations.
> 
> Bandwidth and iops are the most commonly used metrics for IO devices
> but depending on the type and specifics of the device, different IO
> patterns easily lead to multiple orders of magnitude variations
> rendering them useless for the purpose of IO capacity distribution.
> While on-device time, with a lot of clutches, could serve as a useful
> approximation for non-queued rotational devices, this is no longer
> viable with modern devices, even the rotational ones.
> 
> While there is no cost metric we can trivially observe, it isn't a
> complete mystery.  For example, on a rotational device, seek cost
> dominates while a contiguous transfer contributes a smaller amount
> proportional to the size.  If we can characterize at least the
> relative costs of these different types of IOs, it should be possible
> to implement a reasonable work-conserving proportional IO resource
> distribution.
> 
> This patchset implements IO cost model based work-conserving
> proportional controller.  It currently has a simple linear cost model
> builtin where each IO is classified as sequential or random and given
> a base cost accordingly and additional size-proportional cost is added
> on top.  Each IO is given a cost based on the model and the controller
> issues IOs for each cgroup according to their hierarchical weight.
> 
> By default, the controller adapts its overall IO rate so that it
> doesn't build up buffer bloat in the request_queue layer, which
> guarantees that the controller doesn't lose significant amount of
> total work.  However, this may not provide sufficient differentiation
> as the underlying device may have a deep queue and not be fair in how
> the queued IOs are serviced.  The controller provides extra QoS
> control knobs which allow tightening control feedback loop as
> necessary.
> 
> For more details on the control mechanism, implementation and
> interface, please refer to the comment at the top of
> block/blk-iocost.c and Documentation/admin-guide/cgroup-v2.rst changes
> in the "blkcg: implement blk-iocost" patch.
> 
> Here are some test results.  Each test run goes through the following
> combinations with each combination running for a minute.  All tests
> are performed against regular files on btrfs w/ deadline as the IO
> scheduler.  Random IOs are direct w/ queue depth of 64.  Sequential
> are normal buffered IOs.
> 
>        high priority (weight=500)      low priority (weight=100)
> 
>        Rand read                       None
>        ditto                           Rand read
>        ditto                           Seq  read
>        ditto                           Rand write
>        ditto                           Seq  write
>        Seq  read                       None
>        ditto                           Rand read
>        ditto                           Seq  read
>        ditto                           Rand write
>        ditto                           Seq  write
>        Rand write                      None
>        ditto                           Rand read
>        ditto                           Seq  read
>        ditto                           Rand write
>        ditto                           Seq  write
>        Seq  write                      None
>        ditto                           Rand read
>        ditto                           Seq  read
>        ditto                           Rand write
>        ditto                           Seq  write
> 
> * 7200RPM SATA hard disk
>  * No IO control
>    https://photos.app.goo.gl/1KBHn7ykpC1LXRkB8
>  * iocost, QoS: None
>    https://photos.app.goo.gl/MLNQGxCtBQ8wAmjm7
>  * iocost, QoS: rpct=95.00 rlat=40000 wpct=95.00 wlat=40000 min=25.00 max=200.00
>    https://photos.app.goo.gl/XqXHm3Mkbm9w6Db46
> * NCQ-blacklisted SATA SSD (QD==1)
>  * No IO control
>    https://photos.app.goo.gl/wCTXeu2uJ6LYL4pk8
>  * iocost, QoS: None
>    https://photos.app.goo.gl/T2HedKD2sywQgj7R9
>  * iocost, QoS: rpct=95.00 rlat=20000 wpct=95.00 wlat=20000 min=50.00 max=200.00
>    https://photos.app.goo.gl/urBTV8XQc1UqPJJw7
> * SATA SSD (QD==32)
>  * No IO control
>    https://photos.app.goo.gl/TjEVykuVudSQcryh6
>  * iocost, QoS: None
>    https://photos.app.goo.gl/iyQBsky7bmM54Xiq7
>  * iocost, QoS: rpct=95.00 rlat=10000 wpct=95.00 wlat=20000 min=50.00 max=400.00
>    https://photos.app.goo.gl/q1a6URLDxPLMrnHy5
> * NVME SSD (ran with 8 concurrent fio jobs to achieve saturation)
>  * No IO control
>    https://photos.app.goo.gl/S6xjEVTJzcfb3w1j7
>  * iocost, QoS: None
>    https://photos.app.goo.gl/SjQUUotJBAGr7vqz7
>  * iocost, QoS: rpct=95.00 rlat=5000 wpct=95.00 wlat=5000 min=1.00 max=10000.00
>    https://photos.app.goo.gl/RsaYBd2muX7CegoN7
> 
> Even without explicit QoS configuration, read-heavy scenarios can
> obtain acceptable differentiation.  However, when write-heavy, the
> deep buffering on the device side makes it difficult to maintain
> control.  With QoS parameters set, the differentiation is acceptable
> across all combinations.
> 
> The implementation comes with default cost model parameters which are
> selected automatically which should provide acceptable behavior across
> most common devices.  The parameters for hdd and consumer-grade SSDs
> seem pretty robust.  The default parameter set and selection criteria
> for highend SSDs might need further adjustments.
> 
> It is fairly easy to configure the QoS parameters and, if needed, cost
> model coefficients.  We'll follow up with tooling and further
> documentation.  Also, the last RFC patch in the series implements
> support for bpf-based custom cost function.  Originally we thought
> that we'd need per-device-type cost functions but the simple linear
> model now seem good enough to cover all common device classes.  In
> case custom cost functions become necessary, we can fully develop the
> bpf based extension and also easily add different builtin cost models.
> 
> Andy Newell did the heavy lifting of analyzing IO workloads and device
> characteristics, exploring various cost models, determining the
> default model and parameters to use.
> 
> Josef Bacik implemented a prototype which explored the use of
> different types of cost metrics including on-device time and Andy's
> linear model.
> 
> This patchset is on top of the current block/for-next 53fc55c817c3
> ("Merge branch 'for-5.4/block' into for-next") and contains the
> following 10 patches.
> 
> 0001-blkcg-pass-q-and-blkcg-into-blkcg_pol_alloc_pd_fn.patch
> 0002-blkcg-make-cpd_init_fn-optional.patch
> 0003-blkcg-separate-blkcg_conf_get_disk-out-of-blkg_conf_.patch
> 0004-block-rq_qos-add-rq_qos_merge.patch
> 0005-block-rq_qos-implement-rq_qos_ops-queue_depth_change.patch
> 0006-blkcg-s-RQ_QOS_CGROUP-RQ_QOS_LATENCY.patch
> 0007-blk-mq-add-optional-request-alloc_time_ns.patch
> 0008-blkcg-implement-blk-iocost.patch
> 0009-blkcg-add-tools-cgroup-iocost_monitor.py.patch
> 0010-blkcg-add-tools-cgroup-iocost_coef_gen.py.patch
> 
> 0001-0007 are prep patches.
> 0008 implements blk-iocost.
> 0009 adds monitoring script.
> 0010 adds linear cost model coefficient generation script.
> 
> The patchset is also available in the following git branch.
> 
> git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup.git review-iow-v2
> 
> diffstat follows, Thanks.
> 
> Documentation/admin-guide/cgroup-v2.rst |   97 +
> block/Kconfig                           |   13 
> block/Makefile                          |    1 
> block/bfq-cgroup.c                      |    5 
> block/blk-cgroup.c                      |   71 
> block/blk-core.c                        |    4 
> block/blk-iocost.c                      | 2395 ++++++++++++++++++++++++++++++++
> block/blk-iolatency.c                   |    8 
> block/blk-mq.c                          |   13 
> block/blk-rq-qos.c                      |   18 
> block/blk-rq-qos.h                      |   28 
> block/blk-settings.c                    |    2 
> block/blk-throttle.c                    |    6 
> block/blk-wbt.c                         |   18 
> block/blk-wbt.h                         |    4 
> include/linux/blk-cgroup.h              |    4 
> include/linux/blk_types.h               |    3 
> include/linux/blkdev.h                  |   13 
> include/trace/events/iocost.h           |  174 ++
> tools/cgroup/iocost_coef_gen.py         |  178 ++
> tools/cgroup/iocost_monitor.py          |  270 +++
> 21 files changed, 3272 insertions(+), 53 deletions(-)
> 
> --
> tejun
> 
> [1] http://lkml.kernel.org/r/20190614015620.1587672-1-tj@kernel.org
> [2] http://lkml.kernel.org/r/20190710205128.1316483-1-tj@kernel.org
> 

Re: [PATCHSET v3 block/for-linus] IO cost model based work-conserving porportional controller

[Tejun Heo]

On Thu, Aug 29, 2019 at 05:54:38PM +0200, Paolo Valente wrote:
> I see an important interface problem.  Userspace has been waiting for
> io.weight to become eventually the file name for setting the weight
> for the proportional-share policy [1,2].  If you use that name, how
> will we solve this?

So, my plan is just disabling iocost if bfq is selected as the io
scheduler.  It makes no sense to use them together anyway.  What do
you wanna do about the existing interface files prefixed with bfq?
Just rename them?

Thanks.

-- 
tejun

[PATCHSET v2 block/for-linus] IO cost model based work-conserving porportional controller

[Tejun Heo]

Hello,

Changes from v1[1]:

* Prerequisite patchsets had cosmetic changes and merged.  Refreshed
  on top.

* Renamed from ioweight to iocost.  All source code and tools are
  updated accordingly.  Control knobs io.weight.qos and
  io.weight.cost_model are renamed to io.cost.qos and io.cost.model
  respectively.  This is a more fitting name which won't become a
  misnomer when, for example, cost based io.max is added.

* Various bug fixes and improvements.  A few bugs were discovered
  while testing against high-iops nvme device.  Auto parameter
  selection improved and verified across different classes of SSDs.

* Dropped bpf iocost support for now.

* Added coef generation script.

* Verified on high-iops nvme device.  Result is included below.

One challenge of controlling IO resources is the lack of trivially
observable cost metric.  This is distinguished from CPU and memory
where wallclock time and the number of bytes can serve as accurate
enough approximations.

Bandwidth and iops are the most commonly used metrics for IO devices
but depending on the type and specifics of the device, different IO
patterns easily lead to multiple orders of magnitude variations
rendering them useless for the purpose of IO capacity distribution.
While on-device time, with a lot of clutches, could serve as a useful
approximation for non-queued rotational devices, this is no longer
viable with modern devices, even the rotational ones.

While there is no cost metric we can trivially observe, it isn't a
complete mystery.  For example, on a rotational device, seek cost
dominates while a contiguous transfer contributes a smaller amount
proportional to the size.  If we can characterize at least the
relative costs of these different types of IOs, it should be possible
to implement a reasonable work-conserving proportional IO resource
distribution.

This patchset implements IO cost model based work-conserving
proportional controller.  It currently has a simple linear cost model
builtin where each IO is classified as sequential or random and given
a base cost accordingly and additional size-proportional cost is added
on top.  Each IO is given a cost based on the model and the controller
issues IOs for each cgroup according to their hierarchical weight.

By default, the controller adapts its overall IO rate so that it
doesn't build up buffer bloat in the request_queue layer, which
guarantees that the controller doesn't lose significant amount of
total work.  However, this may not provide sufficient differentiation
as the underlying device may have a deep queue and not be fair in how
the queued IOs are serviced.  The controller provides extra QoS
control knobs which allow tightening control feedback loop as
necessary.

For more details on the control mechanism, implementation and
interface, please refer to the comment at the top of
block/blk-iocost.c and Documentation/admin-guide/cgroup-v2.rst changes
in the "blkcg: implement blk-iocost" patch.

Here are some test results.  Each test run goes through the following
combinations with each combination running for a minute.  All tests
are performed against regular files on btrfs w/ deadline as the IO
scheduler.  Random IOs are direct w/ queue depth of 64.  Sequential
are normal buffered IOs.

	high priority (weight=500)	low priority (weight=100)

	Rand read			None
	ditto				Rand read
	ditto				Seq  read
	ditto				Rand write
	ditto				Seq  write
	Seq  read			None
	ditto				Rand read
	ditto				Seq  read
	ditto				Rand write
	ditto				Seq  write
	Rand write			None
	ditto				Rand read
	ditto				Seq  read
	ditto				Rand write
	ditto				Seq  write
	Seq  write			None
	ditto				Rand read
	ditto				Seq  read
	ditto				Rand write
	ditto				Seq  write

* 7200RPM SATA hard disk
  * No IO control
    https://photos.app.goo.gl/1KBHn7ykpC1LXRkB8
  * iocost, QoS: None
    https://photos.app.goo.gl/MLNQGxCtBQ8wAmjm7
  * iocost, QoS: rpct=95.00 rlat=40000 wpct=95.00 wlat=40000 min=25.00 max=200.00
    https://photos.app.goo.gl/XqXHm3Mkbm9w6Db46
* NCQ-blacklisted SATA SSD (QD==1)
  * No IO control
    https://photos.app.goo.gl/wCTXeu2uJ6LYL4pk8
  * iocost, QoS: None
    https://photos.app.goo.gl/T2HedKD2sywQgj7R9
  * iocost, QoS: rpct=95.00 rlat=20000 wpct=95.00 wlat=20000 min=50.00 max=200.00
    https://photos.app.goo.gl/urBTV8XQc1UqPJJw7
* SATA SSD (QD==32)
  * No IO control
    https://photos.app.goo.gl/TjEVykuVudSQcryh6
  * iocost, QoS: None
    https://photos.app.goo.gl/iyQBsky7bmM54Xiq7
  * iocost, QoS: rpct=95.00 rlat=10000 wpct=95.00 wlat=20000 min=50.00 max=400.00
    https://photos.app.goo.gl/q1a6URLDxPLMrnHy5
* NVME SSD (ran with 8 concurrent fio jobs to achieve saturation)
  * No IO control
    https://photos.app.goo.gl/S6xjEVTJzcfb3w1j7
  * iocost, QoS: None
    https://photos.app.goo.gl/SjQUUotJBAGr7vqz7
  * iocost, QoS: rpct=95.00 rlat=5000 wpct=95.00 wlat=5000 min=1.00 max=10000.00
    https://photos.app.goo.gl/RsaYBd2muX7CegoN7

Even without explicit QoS configuration, read-heavy scenarios can
obtain acceptable differentiation.  However, when write-heavy, the
deep buffering on the device side makes it difficult to maintain
control.  With QoS parameters set, the differentiation is acceptable
across all combinations.

The implementation comes with default cost model parameters which are
selected automatically which should provide acceptable behavior across
most common devices.  The parameters for hdd and consumer-grade SSDs
seem pretty robust.  The default parameter set and selection criteria
for highend SSDs might need further adjustments.

It is fairly easy to configure the QoS parameters and, if needed, cost
model coefficients.  We'll follow up with tooling and further
documentation.  Also, the last RFC patch in the series implements
support for bpf-based custom cost function.  Originally we thought
that we'd need per-device-type cost functions but the simple linear
model now seem good enough to cover all common device classes.  In
case custom cost functions become necessary, we can fully develop the
bpf based extension and also easily add different builtin cost models.

Andy Newell did the heavy lifting of analyzing IO workloads and device
characteristics, exploring various cost models, determining the
default model and parameters to use.

Josef Bacik implemented a prototype which explored the use of
different types of cost metrics including on-device time and Andy's
linear model.

This patchset is on top of block/for-linus 3a10f999ffd4
("blk-throttle: fix zero wait time for iops throttled group")
and contains the following ten patches.

 0001-blkcg-pass-q-and-blkcg-into-blkcg_pol_alloc_pd_fn.patch
 0002-blkcg-make-cpd_init_fn-optional.patch
 0003-blkcg-separate-blkcg_conf_get_disk-out-of-blkg_conf_.patch
 0004-block-rq_qos-add-rq_qos_merge.patch
 0005-block-rq_qos-implement-rq_qos_ops-queue_depth_change.patch
 0006-blkcg-s-RQ_QOS_CGROUP-RQ_QOS_LATENCY.patch
 0007-blk-mq-add-optional-request-pre_start_time_ns.patch
 0008-blkcg-implement-blk-iocost.patch
 0009-blkcg-add-tools-cgroup-iocost_monitor.py.patch
 0010-blkcg-add-tools-cgroup-iocost_coef_gen.py.patch

0001-0007 are prep patches.  0008 implements blk-iocost.  0009 adds
monitoring script.  0010 adds linear model coef generation script.

The patchset is also available in the following git branch.

 git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup.git review-iocost

diffstat follows, Thanks.

 Documentation/admin-guide/cgroup-v2.rst |   97 +
 block/Kconfig                           |    9 
 block/Makefile                          |    1 
 block/bfq-cgroup.c                      |    5 
 block/blk-cgroup.c                      |   71 
 block/blk-core.c                        |    4 
 block/blk-iocost.c                      | 2394 ++++++++++++++++++++++++++++++++
 block/blk-iolatency.c                   |    8 
 block/blk-mq.c                          |   11 
 block/blk-rq-qos.c                      |   18 
 block/blk-rq-qos.h                      |   28 
 block/blk-settings.c                    |    2 
 block/blk-throttle.c                    |    6 
 block/blk-wbt.c                         |   18 
 block/blk-wbt.h                         |    4 
 include/linux/blk-cgroup.h              |    4 
 include/linux/blk_types.h               |    3 
 include/linux/blkdev.h                  |    7 
 include/trace/events/iocost.h           |  174 ++
 tools/cgroup/iocost_coef_gen.py         |  178 ++
 tools/cgroup/iocost_monitor.py          |  270 +++
 21 files changed, 3259 insertions(+), 53 deletions(-)

--
tejun

[1] http://lkml.kernel.org/r/20190614015620.1587672-1-tj@kernel.org

[PATCH 08/10] blkcg: implement blk-iocost

[Tejun Heo]

This patchset implements IO cost model based work-conserving
proportional controller.

While io.latency provides the capability to comprehensively prioritize
and protect IOs depending on the cgroups, its protection is binary -
the lowest latency target cgroup which is suffering is protected at
the cost of all others.  In many use cases including stacking multiple
workload containers in a single system, it's necessary to distribute
IO capacity with better granularity.

One challenge of controlling IO resources is the lack of trivially
observable cost metric.  The most common metrics - bandwidth and iops
- can be off by orders of magnitude depending on the device type and
IO pattern.  However, the cost isn't a complete mystery.  Given
several key attributes, we can make fairly reliable predictions on how
expensive a given stream of IOs would be, at least compared to other
IO patterns.

The function which determines the cost of a given IO is the IO cost
model for the device.  This controller distributes IO capacity based
on the costs estimated by such model.  The more accurate the cost
model the better but the controller adapts based on IO completion
latency and as long as the relative costs across differents IO
patterns are consistent and sensible, it'll adapt to the actual
performance of the device.

Currently, the only implemented cost model is a simple linear one with
a few sets of default parameters for different classes of device.
This covers most common devices reasonably well.  All the
infrastructure to tune and add different cost models is already in
place and a later patch will also allow using bpf progs for cost
models.

Please see the top comment in blk-iocost.c and documentation for
more details.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Andy Newell <newella@fb.com>
Cc: Josef Bacik <jbacik@fb.com>
---
 Documentation/admin-guide/cgroup-v2.rst |   94 +
 block/Kconfig                           |    9 +
 block/Makefile                          |    1 +
 block/blk-iocost.c                      | 2370 +++++++++++++++++++++++
 block/blk-rq-qos.h                      |    3 +
 include/linux/blk_types.h               |    3 +
 include/trace/events/iocost.h           |  174 ++
 7 files changed, 2654 insertions(+)
 create mode 100644 block/blk-iocost.c
 create mode 100644 include/trace/events/iocost.h

diff --git a/Documentation/admin-guide/cgroup-v2.rst b/Documentation/admin-guide/cgroup-v2.rst
index 6223f485f7e1..aff812631c40 100644
--- a/Documentation/admin-guide/cgroup-v2.rst
+++ b/Documentation/admin-guide/cgroup-v2.rst
@@ -1425,6 +1425,100 @@ IO Interface Files
 	  8:16 rbytes=1459200 wbytes=314773504 rios=192 wios=353 dbytes=0 dios=0
 	  8:0 rbytes=90430464 wbytes=299008000 rios=8950 wios=1252 dbytes=50331648 dios=3021
 
+  io.cost.qos
+	A read-write nested-keyed file with exists only on the root
+	cgroup.
+
+	This file configures the Quality of Service of the IO cost
+	model based controller (CONFIG_BLK_CGROUP_IOCOST) which
+	currently implements "io.weight" proportional control.  Lines
+	are keyed by $MAJ:$MIN device numbers and not ordered.  The
+	line for a given device is populated on the first write for
+	the device on "io.cost.qos" or "io.cost.model".  The following
+	nested keys are defined.
+
+	  ======	=====================================
+	  enable	Weight-based control enable
+	  ctrl		"auto" or "user"
+	  rpct		Read latency percentile    [0, 100]
+	  rlat		Read latency threshold
+	  wpct		Write latency percentile   [0, 100]
+	  wlat		Write latency threshold
+	  min		Minimum scaling percentage [1, 10000]
+	  max		Maximum scaling percentage [1, 10000]
+	  ======	=====================================
+
+	The controller is disabled by default and can be enabled by
+	setting "enable" to 1.  "rpct" and "wpct" parameters default
+	to zero and the controller uses internal device saturation
+	state to adjust the overall IO rate between "min" and "max".
+
+	When a better control quality is needed, latency QoS
+	parameters can be configured.  For example::
+
+	  8:16 enable=1 ctrl=auto rpct=95.00 rlat=75000 wpct=95.00 wlat=150000 min=50.00 max=150.0
+
+	shows that on sdb, the controller is enabled, will consider
+	the device saturated if the 95th percentile of read completion
+	latencies is above 75ms or write 150ms, and adjust the overall
+	IO issue rate between 50% and 150% accordingly.
+
+	The lower the saturation point, the better the latency QoS at
+	the cost of aggregate bandwidth.  The narrower the allowed
+	adjustment range between "min" and "max", the more conformant
+	to the cost model the IO behavior.  Note that the IO issue
+	base rate may be far off from 100% and setting "min" and "max"
+	blindly can lead to a significant loss of device capacity or
+	control quality.  "min" and "max" are useful for regulating
+	devices which show wide temporary behavior changes - e.g. a
+	ssd which accepts writes at the line speed for a while and
+	then completely stalls for multiple seconds.
+
+	When "ctrl" is "auto", the parameters are controlled by the
+	kernel and may change automatically.  Setting "ctrl" to "user"
+	or setting any of the percentile and latency parameters puts
+	it into "user" mode and disables the automatic changes.  The
+	automatic mode can be restored by setting "ctrl" to "auto".
+
+  io.cost.model
+	A read-write nested-keyed file with exists only on the root
+	cgroup.
+
+	This file configures the cost model of the IO cost model based
+	controller (CONFIG_BLK_CGROUP_IOCOST) which currently
+	implements "io.weight" proportional control.  Lines are keyed
+	by $MAJ:$MIN device numbers and not ordered.  The line for a
+	given device is populated on the first write for the device on
+	"io.cost.qos" or "io.cost.model".  The following nested keys
+	are defined.
+
+	  =====		================================
+	  ctrl		"auto" or "user"
+	  model		The cost model in use - "linear"
+	  =====		================================
+
+	When "ctrl" is "auto", the kernel may change all parameters
+	dynamically.  When "ctrl" is set to "user" or any other
+	parameters are written to, "ctrl" become "user" and the
+	automatic changes are disabled.
+
+	When "model" is "linear", the following model parameters are
+	defined.
+
+	  =============	========================================
+	  [r|w]bps	The maximum sequential IO throughput
+	  [r|w]seqiops	The maximum 4k sequential IOs per second
+	  [r|w]randiops	The maximum 4k random IOs per second
+	  =============	========================================
+
+	From the above, the builtin linear model determines the base
+	costs of a sequential and random IO and the cost coefficient
+	for the IO size.  While simple, this model can cover most
+	common device classes acceptably.
+
+	The IO cost model isn't expected to be accurate in absolute
+	sense and is scaled to the device behavior dynamically.
+
   io.weight
 	A read-write flat-keyed file which exists on non-root cgroups.
 	The default is "default 100".
diff --git a/block/Kconfig b/block/Kconfig
index 56cb1695cd87..be1b82d94425 100644
--- a/block/Kconfig
+++ b/block/Kconfig
@@ -132,6 +132,15 @@ config BLK_CGROUP_IOLATENCY
 
 	Note, this is an experimental interface and could be changed someday.
 
+config BLK_CGROUP_IOCOST
+	bool "Enable support for cost model based cgroup IO controller"
+	depends on BLK_CGROUP=y
+	---help---
+	Enabling this option enables the .weight interface for cost
+	model based proportional IO control.  The IO controller
+	distributes IO capacity between different groups based on
+	their share of the overall weight distribution.
+
 config BLK_WBT_MQ
 	bool "Multiqueue writeback throttling"
 	default y
diff --git a/block/Makefile b/block/Makefile
index eee1b4ceecf9..9ef57ace90d4 100644
--- a/block/Makefile
+++ b/block/Makefile
@@ -18,6 +18,7 @@ obj-$(CONFIG_BLK_DEV_BSGLIB)	+= bsg-lib.o
 obj-$(CONFIG_BLK_CGROUP)	+= blk-cgroup.o
 obj-$(CONFIG_BLK_DEV_THROTTLING)	+= blk-throttle.o
 obj-$(CONFIG_BLK_CGROUP_IOLATENCY)	+= blk-iolatency.o
+obj-$(CONFIG_BLK_CGROUP_IOCOST)	+= blk-iocost.o
 obj-$(CONFIG_MQ_IOSCHED_DEADLINE)	+= mq-deadline.o
 obj-$(CONFIG_MQ_IOSCHED_KYBER)	+= kyber-iosched.o
 bfq-y				:= bfq-iosched.o bfq-wf2q.o bfq-cgroup.o
diff --git a/block/blk-iocost.c b/block/blk-iocost.c
new file mode 100644
index 000000000000..1f697e5729f6
--- /dev/null
+++ b/block/blk-iocost.c
@@ -0,0 +1,2370 @@
+/* SPDX-License-Identifier: GPL-2.0
+ *
+ * IO cost model based controller.
+ *
+ * Copyright (C) 2019 Tejun Heo <tj@kernel.org>
+ * Copyright (C) 2019 Andy Newell <newella@fb.com>
+ * Copyright (C) 2019 Facebook
+ *
+ * One challenge of controlling IO resources is the lack of trivially
+ * observable cost metric.  This is distinguished from CPU and memory where
+ * wallclock time and the number of bytes can serve as accurate enough
+ * approximations.
+ *
+ * Bandwidth and iops are the most commonly used metrics for IO devices but
+ * depending on the type and specifics of the device, different IO patterns
+ * easily lead to multiple orders of magnitude variations rendering them
+ * useless for the purpose of IO capacity distribution.  While on-device
+ * time, with a lot of clutches, could serve as a useful approximation for
+ * non-queued rotational devices, this is no longer viable with modern
+ * devices, even the rotational ones.
+ *
+ * While there is no cost metric we can trivially observe, it isn't a
+ * complete mystery.  For example, on a rotational device, seek cost
+ * dominates while a contiguous transfer contributes a smaller amount
+ * proportional to the size.  If we can characterize at least the relative
+ * costs of these different types of IOs, it should be possible to
+ * implement a reasonable work-conserving proportional IO resource
+ * distribution.
+ *
+ * 1. IO Cost Model
+ *
+ * IO cost model estimates the cost of an IO given its basic parameters and
+ * history (e.g. the end sector of the last IO).  The cost is measured in
+ * device time.  If a given IO is estimated to cost 10ms, the device should
+ * be able to process ~100 of those IOs in a second.
+ *
+ * Currently, there's only one builtin cost model - linear.  Each IO is
+ * classified as sequential or random and given a base cost accordingly.
+ * On top of that, a size cost proportional to the length of the IO is
+ * added.  While simple, this model captures the operational
+ * characteristics of a wide varienty of devices well enough.  Default
+ * paramters for several different classes of devices are provided and the
+ * parameters can be configured from userspace via
+ * /sys/fs/cgroup/io.cost.model.
+ *
+ * If needed, tools/cgroup/iocost_coef_gen.py can be used to generate
+ * device-specific coefficients.
+ *
+ * 2. Control Strategy
+ *
+ * The device virtual time (vtime) is used as the primary control metric.
+ * The control strategy is composed of the following three parts.
+ *
+ * 2-1. Vtime Distribution
+ *
+ * When a cgroup becomes active in terms of IOs, its hierarchical share is
+ * calculated.  Please consider the following hierarchy where the numbers
+ * inside parentheses denote the configured weights.
+ *
+ *           root
+ *         /       \
+ *      A (w:100)  B (w:300)
+ *      /       \
+ *  A0 (w:100)  A1 (w:100)
+ *
+ * If B is idle and only A0 and A1 are actively issuing IOs, as the two are
+ * of equal weight, each gets 50% share.  If then B starts issuing IOs, B
+ * 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).
+ *
+ * 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)
+ * against the device vtime - an IO which takes 10ms on the underlying
+ * device is considered to take 80ms on A0.
+ *
+ * This constitutes the basis of IO capacity distribution.  Each cgroup's
+ * vtime is running at a rate determined by its hweight.  A cgroup tracks
+ * the vtime consumed by past IOs and can issue a new IO iff doing so
+ * wouldn't outrun the current device vtime.  Otherwise, the IO is
+ * suspended until the vtime has progressed enough to cover it.
+ *
+ * 2-2. Vrate Adjustment
+ *
+ * It's unrealistic to expect the cost model to be perfect.  There are too
+ * many devices and even on the same device the overall performance
+ * fluctuates depending on numerous factors such as IO mixture and device
+ * internal garbage collection.  The controller needs to adapt dynamically.
+ *
+ * This is achieved by adjusting the overall IO rate according to how busy
+ * the device is.  If the device becomes overloaded, we're sending down too
+ * many IOs and should generally slow down.  If there are waiting issuers
+ * but the device isn't saturated, we're issuing too few and should
+ * generally speed up.
+ *
+ * To slow down, we lower the vrate - the rate at which the device vtime
+ * passes compared to the wall clock.  For example, if the vtime is running
+ * at the vrate of 75%, all cgroups added up would only be able to issue
+ * 750ms worth of IOs per second, and vice-versa for speeding up.
+ *
+ * Device business is determined using two criteria - rq wait and
+ * completion latencies.
+ *
+ * When a device gets saturated, the on-device and then the request queues
+ * fill up and a bio which is ready to be issued has to wait for a request
+ * to become available.  When this delay becomes noticeable, it's a clear
+ * indication that the device is saturated and we lower the vrate.  This
+ * saturation signal is fairly conservative as it only triggers when both
+ * hardware and software queues are filled up, and is used as the default
+ * busy signal.
+ *
+ * As devices can have deep queues and be unfair in how the queued commands
+ * are executed, soley depending on rq wait may not result in satisfactory
+ * control quality.  For a better control quality, completion latency QoS
+ * parameters can be configured so that the device is considered saturated
+ * if N'th percentile completion latency rises above the set point.
+ *
+ * The completion latency requirements are a function of both the
+ * underlying device characteristics and the desired IO latency quality of
+ * service.  There is an inherent trade-off - the tighter the latency QoS,
+ * the higher the bandwidth lossage.  Latency QoS is disabled by default
+ * and can be set through /sys/fs/cgroup/io.cost.qos.
+ *
+ * 2-3. Work Conservation
+ *
+ * Imagine two cgroups A and B with equal weights.  A is issuing a small IO
+ * periodically while B is sending out enough parallel IOs to saturate the
+ * device on its own.  Let's say A's usage amounts to 100ms worth of IO
+ * cost per second, i.e., 10% of the device capacity.  The naive
+ * distribution of half and half would lead to 60% utilization of the
+ * device, a significant reduction in the total amount of work done
+ * compared to free-for-all competition.  This is too high a cost to pay
+ * for IO control.
+ *
+ * To conserve the total amount of work done, we keep track of how much
+ * each active cgroup is actually using and yield part of its weight if
+ * there are other cgroups which can make use of it.  In the above case,
+ * A's weight will be lowered so that it hovers above the actual usage and
+ * B would be able to use the rest.
+ *
+ * As we don't want to penalize a cgroup for donating its weight, the
+ * surplus weight adjustment factors in a margin and has an immediate
+ * snapback mechanism in case the cgroup needs more IO vtime for itself.
+ *
+ * Note that adjusting down surplus weights has the same effects as
+ * accelerating vtime for other cgroups and work conservation can also be
+ * implemented by adjusting vrate dynamically.  However, squaring who can
+ * donate and should take back how much requires hweight propagations
+ * anyway making it easier to implement and understand as a separate
+ * mechanism.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/timer.h>
+#include <linux/time64.h>
+#include <linux/parser.h>
+#include <linux/sched/signal.h>
+#include <linux/blk-cgroup.h>
+#include "blk-rq-qos.h"
+#include "blk-stat.h"
+#include "blk-wbt.h"
+
+#ifdef CONFIG_TRACEPOINTS
+
+/* copied from TRACE_CGROUP_PATH, see cgroup-internal.h */
+#define TRACE_IOCG_PATH_LEN 1024
+static DEFINE_SPINLOCK(trace_iocg_path_lock);
+static char trace_iocg_path[TRACE_IOCG_PATH_LEN];
+
+#define TRACE_IOCG_PATH(type, iocg, ...)					\
+	do {									\
+		unsigned long flags;						\
+		if (trace_iocost_##type##_enabled()) {				\
+			spin_lock_irqsave(&trace_iocg_path_lock, flags);	\
+			cgroup_path(iocg_to_blkg(iocg)->blkcg->css.cgroup,	\
+				    trace_iocg_path, TRACE_IOCG_PATH_LEN);	\
+			trace_iocost_##type(iocg, trace_iocg_path,		\
+					      ##__VA_ARGS__);			\
+			spin_unlock_irqrestore(&trace_iocg_path_lock, flags);	\
+		}								\
+	} while (0)
+
+#else	/* CONFIG_TRACE_POINTS */
+#define TRACE_IOCG_PATH(type, iocg, ...)	do { } while (0)
+#endif	/* CONFIG_TRACE_POINTS */
+
+enum {
+	MILLION			= 1000000,
+
+	/* timer period is calculated from latency requirements, bound it */
+	MIN_PERIOD		= USEC_PER_MSEC,
+	MAX_PERIOD		= USEC_PER_SEC,
+
+	/*
+	 * A cgroup's vtime can run 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,
+
+	/* Have some play in waitq timer operations */
+	WAITQ_TIMER_MARGIN_PCT	= 5,
+
+	/*
+	 * 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,
+
+	/* 1/64k is granular enough and can easily be handled w/ u32 */
+	HWEIGHT_WHOLE		= 1 << 16,
+
+	/*
+	 * As vtime is used to calculate the cost of each IO, it needs to
+	 * be fairly high precision.  For example, it should be able to
+	 * represent the cost of a single page worth of discard with
+	 * suffificient accuracy.  At the same time, it should be able to
+	 * represent reasonably long enough durations to be useful and
+	 * convenient during operation.
+	 *
+	 * 1s worth of vtime is 2^37.  This gives us both sub-nanosecond
+	 * granularity and days of wrap-around time even at extreme vrates.
+	 */
+	VTIME_PER_SEC_SHIFT	= 37,
+	VTIME_PER_SEC		= 1LLU << VTIME_PER_SEC_SHIFT,
+	VTIME_PER_USEC		= VTIME_PER_SEC / USEC_PER_SEC,
+
+	/* bound vrate adjustments within two orders of magnitude */
+	VRATE_MIN_PPM		= 10000,	/* 1% */
+	VRATE_MAX_PPM		= 100000000,	/* 10000% */
+
+	VRATE_MIN		= VTIME_PER_USEC * VRATE_MIN_PPM / MILLION,
+	VRATE_CLAMP_ADJ_PCT	= 4,
+
+	/* if IOs end up waiting for requests, issue less */
+	RQ_WAIT_BUSY_PCT	= 5,
+
+	/* 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.
+	 */
+	SURPLUS_SCALE_PCT	= 125,			/* * 125% */
+	SURPLUS_SCALE_ABS	= HWEIGHT_WHOLE / 50,	/* + 2% */
+	SURPLUS_MIN_ADJ_DELTA	= HWEIGHT_WHOLE / 33,	/* 3% */
+
+	/* switch iff the conditions are met for longer than this */
+	AUTOP_CYCLE_NSEC	= 10LLU * NSEC_PER_SEC,
+
+	/*
+	 * Count IO size in 4k pages.  The 12bit shift helps keeping
+	 * size-proportional components of cost calculation in closer
+	 * numbers of digits to per-IO cost components.
+	 */
+	IOC_PAGE_SHIFT		= 12,
+	IOC_PAGE_SIZE		= 1 << IOC_PAGE_SHIFT,
+	IOC_SECT_TO_PAGE_SHIFT	= IOC_PAGE_SHIFT - SECTOR_SHIFT,
+
+	/* if apart further than 16M, consider randio for linear model */
+	LCOEF_RANDIO_PAGES	= 4096,
+};
+
+enum ioc_running {
+	IOC_IDLE,
+	IOC_RUNNING,
+	IOC_STOP,
+};
+
+/* io.cost.qos controls including per-dev enable of the whole controller */
+enum {
+	QOS_ENABLE,
+	QOS_CTRL,
+	NR_QOS_CTRL_PARAMS,
+};
+
+/* io.cost.qos params */
+enum {
+	QOS_RPPM,
+	QOS_RLAT,
+	QOS_WPPM,
+	QOS_WLAT,
+	QOS_MIN,
+	QOS_MAX,
+	NR_QOS_PARAMS,
+};
+
+/* io.cost.model controls */
+enum {
+	COST_CTRL,
+	COST_MODEL,
+	NR_COST_CTRL_PARAMS,
+};
+
+/* builtin linear cost model coefficients */
+enum {
+	I_LCOEF_RBPS,
+	I_LCOEF_RSEQIOPS,
+	I_LCOEF_RRANDIOPS,
+	I_LCOEF_WBPS,
+	I_LCOEF_WSEQIOPS,
+	I_LCOEF_WRANDIOPS,
+	NR_I_LCOEFS,
+};
+
+enum {
+	LCOEF_RPAGE,
+	LCOEF_RSEQIO,
+	LCOEF_RRANDIO,
+	LCOEF_WPAGE,
+	LCOEF_WSEQIO,
+	LCOEF_WRANDIO,
+	NR_LCOEFS,
+};
+
+enum {
+	AUTOP_INVALID,
+	AUTOP_HDD,
+	AUTOP_SSD_QD1,
+	AUTOP_SSD_DFL,
+	AUTOP_SSD_FAST,
+};
+
+struct ioc_gq;
+
+struct ioc_params {
+	u32				qos[NR_QOS_PARAMS];
+	u64				i_lcoefs[NR_I_LCOEFS];
+	u64				lcoefs[NR_LCOEFS];
+	u32				too_fast_vrate_pct;
+	u32				too_slow_vrate_pct;
+};
+
+struct ioc_missed {
+	u32				nr_met;
+	u32				nr_missed;
+	u32				last_met;
+	u32				last_missed;
+};
+
+struct ioc_pcpu_stat {
+	struct ioc_missed		missed[2];
+
+	u64				rq_wait_ns;
+	u64				last_rq_wait_ns;
+};
+
+/* per device */
+struct ioc {
+	struct rq_qos			rqos;
+
+	bool				enabled;
+
+	struct ioc_params		params;
+	u32				period_us;
+	u32				margin_us;
+	u64				vrate_min;
+	u64				vrate_max;
+
+	spinlock_t			lock;
+	struct timer_list		timer;
+	struct list_head		active_iocgs;	/* active cgroups */
+	struct ioc_pcpu_stat __percpu	*pcpu_stat;
+
+	enum ioc_running		running;
+	atomic64_t			vtime_rate;
+
+	seqcount_t			period_seqcount;
+	u32				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 */
+
+	u64				autop_too_fast_at;
+	u64				autop_too_slow_at;
+	int				autop_idx;
+	bool				user_qos_params:1;
+	bool				user_cost_model:1;
+};
+
+/* per device-cgroup pair */
+struct ioc_gq {
+	struct blkg_policy_data		pd;
+	struct ioc			*ioc;
+
+	/*
+	 * A iocg can get its weight from two sources - an explicit
+	 * per-device-cgroup configuration or the default weight of the
+	 * cgroup.  `cfg_weight` is the explicit per-device-cgroup
+	 * configuration.  `weight` is the effective considering both
+	 * sources.
+	 *
+	 * When an idle cgroup becomes active its `active` goes from 0 to
+	 * `weight`.  `inuse` is the surplus adjusted active weight.
+	 * `active` and `inuse` are used to calculate `hweight_active` and
+	 * `hweight_inuse`.
+	 *
+	 * `last_inuse` remembers `inuse` while an iocg is idle to persist
+	 * surplus adjustments.
+	 */
+	u32				cfg_weight;
+	u32				weight;
+	u32				active;
+	u32				inuse;
+	u32				last_inuse;
+
+	sector_t			cursor;		/* to detect randio */
+
+	/*
+	 * `vtime` is this iocg's vtime cursor which progresses as IOs are
+	 * issued.  If lagging behind device vtime, the delta represents
+	 * the currently available IO budget.  If runnning ahead, the
+	 * overage.
+	 *
+	 * `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				last_vtime;
+
+	/*
+	 * The period this iocg was last active in.  Used for deactivation
+	 * and invalidating `vtime`.
+	 */
+	atomic64_t			active_period;
+	struct list_head		active_list;
+
+	/* see __propagate_active_weight() and current_hweight() for details */
+	u64				child_active_sum;
+	u64				child_inuse_sum;
+	int				hweight_gen;
+	u32				hweight_active;
+	u32				hweight_inuse;
+	bool				has_surplus;
+
+	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];
+
+	/* this iocg's depth in the hierarchy and ancestors including self */
+	int				level;
+	struct ioc_gq			*ancestors[];
+};
+
+/* per cgroup */
+struct ioc_cgrp {
+	struct blkcg_policy_data	cpd;
+	unsigned int			dfl_weight;
+};
+
+struct ioc_now {
+	u64				now_ns;
+	u32				now;
+	u64				vnow;
+	u64				vrate;
+};
+
+struct iocg_wait {
+	struct wait_queue_entry		wait;
+	struct bio			*bio;
+	u64				abs_cost;
+	bool				committed;
+};
+
+struct iocg_wake_ctx {
+	struct ioc_gq			*iocg;
+	u32				hw_inuse;
+	s64				vbudget;
+};
+
+static const struct ioc_params autop[] = {
+	[AUTOP_HDD] = {
+		.qos				= {
+			[QOS_RLAT]		=         50000, /* 50ms */
+			[QOS_WLAT]		=         50000,
+			[QOS_MIN]		= VRATE_MIN_PPM,
+			[QOS_MAX]		= VRATE_MAX_PPM,
+		},
+		.i_lcoefs			= {
+			[I_LCOEF_RBPS]		=     174019176,
+			[I_LCOEF_RSEQIOPS]	=         41708,
+			[I_LCOEF_RRANDIOPS]	=           370,
+			[I_LCOEF_WBPS]		=     178075866,
+			[I_LCOEF_WSEQIOPS]	=         42705,
+			[I_LCOEF_WRANDIOPS]	=           378,
+		},
+	},
+	[AUTOP_SSD_QD1] = {
+		.qos				= {
+			[QOS_RLAT]		=         25000, /* 25ms */
+			[QOS_WLAT]		=         25000,
+			[QOS_MIN]		= VRATE_MIN_PPM,
+			[QOS_MAX]		= VRATE_MAX_PPM,
+		},
+		.i_lcoefs			= {
+			[I_LCOEF_RBPS]		=     245855193,
+			[I_LCOEF_RSEQIOPS]	=         61575,
+			[I_LCOEF_RRANDIOPS]	=          6946,
+			[I_LCOEF_WBPS]		=     141365009,
+			[I_LCOEF_WSEQIOPS]	=         33716,
+			[I_LCOEF_WRANDIOPS]	=         26796,
+		},
+	},
+	[AUTOP_SSD_DFL] = {
+		.qos				= {
+			[QOS_RLAT]		=         25000, /* 25ms */
+			[QOS_WLAT]		=         25000,
+			[QOS_MIN]		= VRATE_MIN_PPM,
+			[QOS_MAX]		= VRATE_MAX_PPM,
+		},
+		.i_lcoefs			= {
+			[I_LCOEF_RBPS]		=     488636629,
+			[I_LCOEF_RSEQIOPS]	=          8932,
+			[I_LCOEF_RRANDIOPS]	=          8518,
+			[I_LCOEF_WBPS]		=     427891549,
+			[I_LCOEF_WSEQIOPS]	=         28755,
+			[I_LCOEF_WRANDIOPS]	=         21940,
+		},
+		.too_fast_vrate_pct		=           500,
+	},
+	[AUTOP_SSD_FAST] = {
+		.qos				= {
+			[QOS_RLAT]		=          5000, /* 5ms */
+			[QOS_WLAT]		=          5000,
+			[QOS_MIN]		= VRATE_MIN_PPM,
+			[QOS_MAX]		= VRATE_MAX_PPM,
+		},
+		.i_lcoefs			= {
+			[I_LCOEF_RBPS]		=    3102524156LLU,
+			[I_LCOEF_RSEQIOPS]	=        724816,
+			[I_LCOEF_RRANDIOPS]	=        778122,
+			[I_LCOEF_WBPS]		=    1742780862LLU,
+			[I_LCOEF_WSEQIOPS]	=        425702,
+			[I_LCOEF_WRANDIOPS]	=	 443193,
+		},
+		.too_slow_vrate_pct		=            10,
+	},
+};
+
+/*
+ * vrate adjust percentages indexed by ioc->busy_level.  We adjust up on
+ * vtime credit shortage and down on device saturation.
+ */
+static u32 vrate_adj_pct[] =
+	{ 0, 0, 0, 0,
+	  1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+	  2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
+	  4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8, 8, 8, 8, 16 };
+
+static struct blkcg_policy blkcg_policy_iocost;
+
+/* accessors and helpers */
+static struct ioc *rqos_to_ioc(struct rq_qos *rqos)
+{
+	return container_of(rqos, struct ioc, rqos);
+}
+
+static struct ioc *q_to_ioc(struct request_queue *q)
+{
+	return rqos_to_ioc(rq_qos_id(q, RQ_QOS_COST));
+}
+
+static const char *q_name(struct request_queue *q)
+{
+	if (test_bit(QUEUE_FLAG_REGISTERED, &q->queue_flags))
+		return kobject_name(q->kobj.parent);
+	else
+		return "<unknown>";
+}
+
+static const char __maybe_unused *ioc_name(struct ioc *ioc)
+{
+	return q_name(ioc->rqos.q);
+}
+
+static struct ioc_gq *pd_to_iocg(struct blkg_policy_data *pd)
+{
+	return pd ? container_of(pd, struct ioc_gq, pd) : NULL;
+}
+
+static struct ioc_gq *blkg_to_iocg(struct blkcg_gq *blkg)
+{
+	return pd_to_iocg(blkg_to_pd(blkg, &blkcg_policy_iocost));
+}
+
+static struct blkcg_gq *iocg_to_blkg(struct ioc_gq *iocg)
+{
+	return pd_to_blkg(&iocg->pd);
+}
+
+static struct ioc_cgrp *blkcg_to_iocc(struct blkcg *blkcg)
+{
+	return container_of(blkcg_to_cpd(blkcg, &blkcg_policy_iocost),
+			    struct ioc_cgrp, cpd);
+}
+
+/*
+ * Scale @abs_cost to the inverse of @hw_inuse.  The lower the hierarchical
+ * weight, the more expensive each IO.
+ */
+static u64 abs_cost_to_cost(u64 abs_cost, u32 hw_inuse)
+{
+	return DIV64_U64_ROUND_UP(abs_cost * HWEIGHT_WHOLE, hw_inuse);
+}
+
+static void iocg_commit_bio(struct ioc_gq *iocg, struct bio *bio, u64 cost)
+{
+	bio->bi_iocost_cost = cost;
+	atomic64_add(cost, &iocg->vtime);
+}
+
+#define CREATE_TRACE_POINTS
+#include <trace/events/iocost.h>
+
+/* latency Qos params changed, update period_us and all the dependent params */
+static void ioc_refresh_period_us(struct ioc *ioc)
+{
+	u32 ppm, lat, multi, period_us;
+
+	lockdep_assert_held(&ioc->lock);
+
+	/* pick the higher latency target */
+	if (ioc->params.qos[QOS_RLAT] >= ioc->params.qos[QOS_WLAT]) {
+		ppm = ioc->params.qos[QOS_RPPM];
+		lat = ioc->params.qos[QOS_RLAT];
+	} else {
+		ppm = ioc->params.qos[QOS_WPPM];
+		lat = ioc->params.qos[QOS_WLAT];
+	}
+
+	/*
+	 * We want the period to be long enough to contain a healthy number
+	 * of IOs while short enough for granular control.  Define it as a
+	 * multiple of the latency target.  Ideally, the multiplier should
+	 * be scaled according to the percentile so that it would nominally
+	 * contain a certain number of requests.  Let's be simpler and
+	 * scale it linearly so that it's 2x >= pct(90) and 10x at pct(50).
+	 */
+	if (ppm)
+		multi = max_t(u32, (MILLION - ppm) / 50000, 2);
+	else
+		multi = 2;
+	period_us = multi * lat;
+	period_us = clamp_t(u32, period_us, MIN_PERIOD, MAX_PERIOD);
+
+	/* 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);
+}
+
+static int ioc_autop_idx(struct ioc *ioc)
+{
+	int idx = ioc->autop_idx;
+	const struct ioc_params *p = &autop[idx];
+	u32 vrate_pct;
+	u64 now_ns;
+
+	/* rotational? */
+	if (!blk_queue_nonrot(ioc->rqos.q))
+		return AUTOP_HDD;
+
+	/* handle SATA SSDs w/ broken NCQ */
+	if (blk_queue_depth(ioc->rqos.q) == 1)
+		return AUTOP_SSD_QD1;
+
+	/* use one of the normal ssd sets */
+	if (idx < AUTOP_SSD_DFL)
+		return AUTOP_SSD_DFL;
+
+	/* if user is overriding anything, maintain what was there */
+	if (ioc->user_qos_params || ioc->user_cost_model)
+		return idx;
+
+	/* step up/down based on the vrate */
+	vrate_pct = div64_u64(atomic64_read(&ioc->vtime_rate) * 100,
+			      VTIME_PER_USEC);
+	now_ns = ktime_get_ns();
+
+	if (p->too_fast_vrate_pct && p->too_fast_vrate_pct <= vrate_pct) {
+		if (!ioc->autop_too_fast_at)
+			ioc->autop_too_fast_at = now_ns;
+		if (now_ns - ioc->autop_too_fast_at >= AUTOP_CYCLE_NSEC)
+			return idx + 1;
+	} else {
+		ioc->autop_too_fast_at = 0;
+	}
+
+	if (p->too_slow_vrate_pct && p->too_slow_vrate_pct >= vrate_pct) {
+		if (!ioc->autop_too_slow_at)
+			ioc->autop_too_slow_at = now_ns;
+		if (now_ns - ioc->autop_too_slow_at >= AUTOP_CYCLE_NSEC)
+			return idx - 1;
+	} else {
+		ioc->autop_too_slow_at = 0;
+	}
+
+	return idx;
+}
+
+/*
+ * Take the followings as input
+ *
+ *  @bps	maximum sequential throughput
+ *  @seqiops	maximum sequential 4k iops
+ *  @randiops	maximum random 4k iops
+ *
+ * and calculate the linear model cost coefficients.
+ *
+ *  *@page	per-page cost		1s / (@bps / 4096)
+ *  *@seqio	base cost of a seq IO	max((1s / @seqiops) - *@page, 0)
+ *  @randiops	base cost of a rand IO	max((1s / @randiops) - *@page, 0)
+ */
+static void calc_lcoefs(u64 bps, u64 seqiops, u64 randiops,
+			u64 *page, u64 *seqio, u64 *randio)
+{
+	u64 v;
+
+	*page = *seqio = *randio = 0;
+
+	if (bps)
+		*page = DIV64_U64_ROUND_UP(VTIME_PER_SEC,
+					   DIV_ROUND_UP_ULL(bps, IOC_PAGE_SIZE));
+
+	if (seqiops) {
+		v = DIV64_U64_ROUND_UP(VTIME_PER_SEC, seqiops);
+		if (v > *page)
+			*seqio = v - *page;
+	}
+
+	if (randiops) {
+		v = DIV64_U64_ROUND_UP(VTIME_PER_SEC, randiops);
+		if (v > *page)
+			*randio = v - *page;
+	}
+}
+
+static void ioc_refresh_lcoefs(struct ioc *ioc)
+{
+	u64 *u = ioc->params.i_lcoefs;
+	u64 *c = ioc->params.lcoefs;
+
+	calc_lcoefs(u[I_LCOEF_RBPS], u[I_LCOEF_RSEQIOPS], u[I_LCOEF_RRANDIOPS],
+		    &c[LCOEF_RPAGE], &c[LCOEF_RSEQIO], &c[LCOEF_RRANDIO]);
+	calc_lcoefs(u[I_LCOEF_WBPS], u[I_LCOEF_WSEQIOPS], u[I_LCOEF_WRANDIOPS],
+		    &c[LCOEF_WPAGE], &c[LCOEF_WSEQIO], &c[LCOEF_WRANDIO]);
+}
+
+static bool ioc_refresh_params(struct ioc *ioc, bool force)
+{
+	const struct ioc_params *p;
+	int idx;
+
+	lockdep_assert_held(&ioc->lock);
+
+	idx = ioc_autop_idx(ioc);
+	p = &autop[idx];
+
+	if (idx == ioc->autop_idx && !force)
+		return false;
+
+	if (idx != ioc->autop_idx)
+		atomic64_set(&ioc->vtime_rate, VTIME_PER_USEC);
+
+	ioc->autop_idx = idx;
+	ioc->autop_too_fast_at = 0;
+	ioc->autop_too_slow_at = 0;
+
+	if (!ioc->user_qos_params)
+		memcpy(ioc->params.qos, p->qos, sizeof(p->qos));
+	if (!ioc->user_cost_model)
+		memcpy(ioc->params.i_lcoefs, p->i_lcoefs, sizeof(p->i_lcoefs));
+
+	ioc_refresh_period_us(ioc);
+	ioc_refresh_lcoefs(ioc);
+
+	ioc->vrate_min = DIV64_U64_ROUND_UP((u64)ioc->params.qos[QOS_MIN] *
+					    VTIME_PER_USEC, MILLION);
+	ioc->vrate_max = div64_u64((u64)ioc->params.qos[QOS_MAX] *
+				   VTIME_PER_USEC, MILLION);
+
+	return true;
+}
+
+/* take a snapshot of the current [v]time and vrate */
+static void ioc_now(struct ioc *ioc, struct ioc_now *now)
+{
+	unsigned seq;
+
+	now->now_ns = ktime_get();
+	now->now = ktime_to_us(now->now_ns);
+	now->vrate = atomic64_read(&ioc->vtime_rate);
+
+	/*
+	 * The current vtime is
+	 *
+	 *   vtime at period start + (wallclock time since the start) * vrate
+	 *
+	 * As a consistent snapshot of `period_at_vtime` and `period_at` is
+	 * needed, they're seqcount protected.
+	 */
+	do {
+		seq = read_seqcount_begin(&ioc->period_seqcount);
+		now->vnow = ioc->period_at_vtime +
+			(now->now - ioc->period_at) * now->vrate;
+	} while (read_seqcount_retry(&ioc->period_seqcount, seq));
+}
+
+static void ioc_start_period(struct ioc *ioc, struct ioc_now *now)
+{
+	lockdep_assert_held(&ioc->lock);
+	WARN_ON_ONCE(ioc->running != IOC_RUNNING);
+
+	write_seqcount_begin(&ioc->period_seqcount);
+	ioc->period_at = now->now;
+	ioc->period_at_vtime = now->vnow;
+	write_seqcount_end(&ioc->period_seqcount);
+
+	ioc->timer.expires = jiffies + usecs_to_jiffies(ioc->period_us);
+	add_timer(&ioc->timer);
+}
+
+/*
+ * Update @iocg's `active` and `inuse` to @active and @inuse, update level
+ * weight sums and propagate upwards accordingly.
+ */
+static void __propagate_active_weight(struct ioc_gq *iocg, u32 active, u32 inuse)
+{
+	struct ioc *ioc = iocg->ioc;
+	int lvl;
+
+	lockdep_assert_held(&ioc->lock);
+
+	inuse = min(active, inuse);
+
+	for (lvl = iocg->level - 1; lvl >= 0; lvl--) {
+		struct ioc_gq *parent = iocg->ancestors[lvl];
+		struct ioc_gq *child = iocg->ancestors[lvl + 1];
+		u32 parent_active = 0, parent_inuse = 0;
+
+		/* update the level sums */
+		parent->child_active_sum += (s32)(active - child->active);
+		parent->child_inuse_sum += (s32)(inuse - child->inuse);
+		/* apply the udpates */
+		child->active = active;
+		child->inuse = inuse;
+
+		/*
+		 * The delta between inuse and active sums indicates that
+		 * that much of weight is being given away.  Parent's inuse
+		 * and active should reflect the ratio.
+		 */
+		if (parent->child_active_sum) {
+			parent_active = parent->weight;
+			parent_inuse = DIV64_U64_ROUND_UP(
+				parent_active * parent->child_inuse_sum,
+				parent->child_active_sum);
+		}
+
+		/* do we need to keep walking up? */
+		if (parent_active == parent->active &&
+		    parent_inuse == parent->inuse)
+			break;
+
+		active = parent_active;
+		inuse = parent_inuse;
+	}
+
+	ioc->weights_updated = true;
+}
+
+static void commit_active_weights(struct ioc *ioc)
+{
+	lockdep_assert_held(&ioc->lock);
+
+	if (ioc->weights_updated) {
+		/* paired with rmb in current_hweight(), see there */
+		smp_wmb();
+		atomic_inc(&ioc->hweight_gen);
+		ioc->weights_updated = false;
+	}
+}
+
+static void propagate_active_weight(struct ioc_gq *iocg, u32 active, u32 inuse)
+{
+	__propagate_active_weight(iocg, active, inuse);
+	commit_active_weights(iocg->ioc);
+}
+
+static void current_hweight(struct ioc_gq *iocg, u32 *hw_activep, u32 *hw_inusep)
+{
+	struct ioc *ioc = iocg->ioc;
+	int lvl;
+	u32 hwa, hwi;
+	int ioc_gen;
+
+	/* hot path - if uptodate, use cached */
+	ioc_gen = atomic_read(&ioc->hweight_gen);
+	if (ioc_gen == iocg->hweight_gen)
+		goto out;
+
+	/*
+	 * Paired with wmb in commit_active_weights().  If we saw the
+	 * updated hweight_gen, all the weight updates from
+	 * __propagate_active_weight() are visible too.
+	 *
+	 * We can race with weight updates during calculation and get it
+	 * wrong.  However, hweight_gen would have changed and a future
+	 * reader will recalculate and we're guaranteed to discard the
+	 * wrong result soon.
+	 */
+	smp_rmb();
+
+	hwa = hwi = HWEIGHT_WHOLE;
+	for (lvl = 0; lvl <= iocg->level - 1; lvl++) {
+		struct ioc_gq *parent = iocg->ancestors[lvl];
+		struct ioc_gq *child = iocg->ancestors[lvl + 1];
+		u32 active_sum = READ_ONCE(parent->child_active_sum);
+		u32 inuse_sum = READ_ONCE(parent->child_inuse_sum);
+		u32 active = READ_ONCE(child->active);
+		u32 inuse = READ_ONCE(child->inuse);
+
+		if (!active_sum)
+			continue;
+
+		active_sum = max(active, active_sum);
+		hwa = hwa * active / active_sum;	/* max 16bits * 10000 */
+
+		inuse_sum = max(inuse, inuse_sum);
+		hwi = hwi * inuse / inuse_sum;		/* max 16bits * 10000 */
+	}
+
+	iocg->hweight_active = max_t(u32, hwa, 1);
+	iocg->hweight_inuse = max_t(u32, hwi, 1);
+	iocg->hweight_gen = ioc_gen;
+out:
+	if (hw_activep)
+		*hw_activep = iocg->hweight_active;
+	if (hw_inusep)
+		*hw_inusep = iocg->hweight_inuse;
+}
+
+static void weight_updated(struct ioc_gq *iocg)
+{
+	struct ioc *ioc = iocg->ioc;
+	struct blkcg_gq *blkg = iocg_to_blkg(iocg);
+	struct ioc_cgrp *iocc = blkcg_to_iocc(blkg->blkcg);
+	u32 weight;
+
+	lockdep_assert_held(&ioc->lock);
+
+	weight = iocg->cfg_weight ?: iocc->dfl_weight;
+	if (weight != iocg->weight && iocg->active)
+		propagate_active_weight(iocg, weight,
+			DIV64_U64_ROUND_UP(iocg->inuse * weight, iocg->weight));
+	iocg->weight = weight;
+}
+
+static bool iocg_activate(struct ioc_gq *iocg, struct ioc_now *now)
+{
+	struct ioc *ioc = iocg->ioc;
+	u64 last_period, cur_period, max_period_delta;
+	u64 vtime, vmargin, vmin;
+	int i;
+
+	/*
+	 * If seem to be already active, just update the stamp to tell the
+	 * timer that we're still active.  We don't mind occassional races.
+	 */
+	if (!list_empty(&iocg->active_list)) {
+		ioc_now(ioc, now);
+		cur_period = atomic64_read(&ioc->cur_period);
+		if (atomic64_read(&iocg->active_period) != cur_period)
+			atomic64_set(&iocg->active_period, cur_period);
+		return true;
+	}
+
+	/* racy check on internal node IOs, treat as root level IOs */
+	if (iocg->child_active_sum)
+		return false;
+
+	spin_lock_irq(&ioc->lock);
+
+	ioc_now(ioc, now);
+
+	/* update period */
+	cur_period = atomic64_read(&ioc->cur_period);
+	last_period = atomic64_read(&iocg->active_period);
+	atomic64_set(&iocg->active_period, cur_period);
+
+	/* already activated or breaking leaf-only constraint? */
+	for (i = iocg->level; i > 0; i--)
+		if (!list_empty(&iocg->active_list))
+			goto fail_unlock;
+	if (iocg->child_active_sum)
+		goto fail_unlock;
+
+	/*
+	 * vtime may wrap when vrate is raised substantially due to
+	 * underestimated IO costs.  Look at the period and ignore its
+	 * vtime if the iocg has been idle for too long.  Also, cap the
+	 * budget it can start with to the margin.
+	 */
+	max_period_delta = DIV64_U64_ROUND_UP(VTIME_VALID_DUR, ioc->period_us);
+	vtime = atomic64_read(&iocg->vtime);
+	vmargin = ioc->margin_us * now->vrate;
+	vmin = now->vnow - vmargin;
+
+	if (last_period + max_period_delta < cur_period ||
+	    time_before64(vtime, vmin)) {
+		atomic64_add(vmin - vtime, &iocg->vtime);
+		atomic64_add(vmin - vtime, &iocg->done_vtime);
+		vtime = vmin;
+	}
+
+	/*
+	 * Activate, propagate weight and start period timer if not
+	 * running.  Reset hweight_gen to avoid accidental match from
+	 * wrapping.
+	 */
+	iocg->hweight_gen = atomic_read(&ioc->hweight_gen) - 1;
+	list_add(&iocg->active_list, &ioc->active_iocgs);
+	propagate_active_weight(iocg, iocg->weight,
+				iocg->last_inuse ?: iocg->weight);
+
+	TRACE_IOCG_PATH(iocg_activate, iocg, now,
+			last_period, cur_period, vtime);
+
+	iocg->last_vtime = vtime;
+
+	if (ioc->running == IOC_IDLE) {
+		ioc->running = IOC_RUNNING;
+		ioc_start_period(ioc, now);
+	}
+
+	spin_unlock_irq(&ioc->lock);
+	return true;
+
+fail_unlock:
+	spin_unlock_irq(&ioc->lock);
+	return false;
+}
+
+static int iocg_wake_fn(struct wait_queue_entry *wq_entry, unsigned mode,
+			int flags, void *key)
+{
+	struct iocg_wait *wait = container_of(wq_entry, struct iocg_wait, wait);
+	struct iocg_wake_ctx *ctx = (struct iocg_wake_ctx *)key;
+	u64 cost = abs_cost_to_cost(wait->abs_cost, ctx->hw_inuse);
+
+	ctx->vbudget -= cost;
+
+	if (ctx->vbudget < 0)
+		return -1;
+
+	iocg_commit_bio(ctx->iocg, wait->bio, cost);
+
+	/*
+	 * autoremove_wake_function() removes the wait entry only when it
+	 * actually changed the task state.  We want the wait always
+	 * removed.  Remove explicitly and use default_wake_function().
+	 */
+	list_del_init(&wq_entry->entry);
+	wait->committed = true;
+
+	default_wake_function(wq_entry, mode, flags, key);
+	return 0;
+}
+
+static void iocg_kick_waitq(struct ioc_gq *iocg, struct ioc_now *now)
+{
+	struct ioc *ioc = iocg->ioc;
+	struct iocg_wake_ctx ctx = { .iocg = iocg };
+	u64 margin_ns = (u64)(ioc->period_us *
+			      WAITQ_TIMER_MARGIN_PCT / 100) * NSEC_PER_USEC;
+	u64 vshortage, expires, oexpires;
+
+	lockdep_assert_held(&iocg->waitq.lock);
+
+	/*
+	 * Wake up the ones which are due and see how much vtime we'll need
+	 * for the next one.
+	 */
+	current_hweight(iocg, NULL, &ctx.hw_inuse);
+	ctx.vbudget = now->vnow - atomic64_read(&iocg->vtime);
+	__wake_up_locked_key(&iocg->waitq, TASK_NORMAL, &ctx);
+	if (!waitqueue_active(&iocg->waitq))
+		return;
+	if (WARN_ON_ONCE(ctx.vbudget >= 0))
+		return;
+
+	/* determine next wakeup, add a quarter margin to guarantee chunking */
+	vshortage = -ctx.vbudget;
+	expires = now->now_ns +
+		DIV64_U64_ROUND_UP(vshortage, now->vrate) * NSEC_PER_USEC;
+	expires += margin_ns / 4;
+
+	/* if already active and close enough, don't bother */
+	oexpires = ktime_to_ns(hrtimer_get_softexpires(&iocg->waitq_timer));
+	if (hrtimer_is_queued(&iocg->waitq_timer) &&
+	    abs(oexpires - expires) <= margin_ns / 4)
+		return;
+
+	hrtimer_start_range_ns(&iocg->waitq_timer, ns_to_ktime(expires),
+			       margin_ns / 4, HRTIMER_MODE_ABS);
+}
+
+static enum hrtimer_restart iocg_waitq_timer_fn(struct hrtimer *timer)
+{
+	struct ioc_gq *iocg = container_of(timer, struct ioc_gq, waitq_timer);
+	struct ioc_now now;
+	unsigned long flags;
+
+	ioc_now(iocg->ioc, &now);
+
+	spin_lock_irqsave(&iocg->waitq.lock, flags);
+	iocg_kick_waitq(iocg, &now);
+	spin_unlock_irqrestore(&iocg->waitq.lock, flags);
+
+	return HRTIMER_NORESTART;
+}
+
+static void iocg_kick_delay(struct ioc_gq *iocg, struct ioc_now *now, u64 cost)
+{
+	struct ioc *ioc = iocg->ioc;
+	struct blkcg_gq *blkg = iocg_to_blkg(iocg);
+	u64 vtime = atomic64_read(&iocg->vtime);
+	u64 vmargin = ioc->margin_us * now->vrate;
+	u64 margin_ns = ioc->margin_us * NSEC_PER_USEC;
+	u64 expires, oexpires;
+
+	/* clear or maintain depending on the overage */
+	if (time_before_eq64(vtime, now->vnow)) {
+		blkcg_clear_delay(blkg);
+		return;
+	}
+	if (!atomic_read(&blkg->use_delay) &&
+	    time_before_eq64(vtime, now->vnow + vmargin))
+		return;
+
+	/* use delay */
+	if (cost) {
+		u64 cost_ns = DIV64_U64_ROUND_UP(cost * NSEC_PER_USEC,
+						 now->vrate);
+		blkcg_add_delay(blkg, now->now_ns, cost_ns);
+	}
+	blkcg_use_delay(blkg);
+
+	expires = now->now_ns + DIV64_U64_ROUND_UP(vtime - now->vnow,
+						   now->vrate) * NSEC_PER_USEC;
+
+	/* if already active and close enough, don't bother */
+	oexpires = ktime_to_ns(hrtimer_get_softexpires(&iocg->delay_timer));
+	if (hrtimer_is_queued(&iocg->delay_timer) &&
+	    abs(oexpires - expires) <= margin_ns / 4)
+		return;
+
+	hrtimer_start_range_ns(&iocg->delay_timer, ns_to_ktime(expires),
+			       margin_ns / 4, HRTIMER_MODE_ABS);
+}
+
+static enum hrtimer_restart iocg_delay_timer_fn(struct hrtimer *timer)
+{
+	struct ioc_gq *iocg = container_of(timer, struct ioc_gq, delay_timer);
+	struct ioc_now now;
+
+	ioc_now(iocg->ioc, &now);
+	iocg_kick_delay(iocg, &now, 0);
+
+	return HRTIMER_NORESTART;
+}
+
+static void ioc_lat_stat(struct ioc *ioc, u32 *missed_ppm_ar, u32 *rq_wait_pct_p)
+{
+	u32 nr_met[2] = { };
+	u32 nr_missed[2] = { };
+	u64 rq_wait_ns = 0;
+	int cpu, rw;
+
+	for_each_online_cpu(cpu) {
+		struct ioc_pcpu_stat *stat = per_cpu_ptr(ioc->pcpu_stat, cpu);
+		u64 this_rq_wait_ns;
+
+		for (rw = READ; rw <= WRITE; rw++) {
+			u32 this_met = READ_ONCE(stat->missed[rw].nr_met);
+			u32 this_missed = READ_ONCE(stat->missed[rw].nr_missed);
+
+			nr_met[rw] += this_met - stat->missed[rw].last_met;
+			nr_missed[rw] += this_missed - stat->missed[rw].last_missed;
+			stat->missed[rw].last_met = this_met;
+			stat->missed[rw].last_missed = this_missed;
+		}
+
+		this_rq_wait_ns = READ_ONCE(stat->rq_wait_ns);
+		rq_wait_ns += this_rq_wait_ns - stat->last_rq_wait_ns;
+		stat->last_rq_wait_ns = this_rq_wait_ns;
+	}
+
+	for (rw = READ; rw <= WRITE; rw++) {
+		if (nr_met[rw] + nr_missed[rw])
+			missed_ppm_ar[rw] =
+				DIV64_U64_ROUND_UP((u64)nr_missed[rw] * MILLION,
+						   nr_met[rw] + nr_missed[rw]);
+		else
+			missed_ppm_ar[rw] = 0;
+	}
+
+	*rq_wait_pct_p = div64_u64(rq_wait_ns * 100,
+				   ioc->period_us * NSEC_PER_USEC);
+}
+
+/* was iocg idle this period? */
+static bool iocg_is_idle(struct ioc_gq *iocg)
+{
+	struct ioc *ioc = iocg->ioc;
+
+	/* did something get issued this period? */
+	if (atomic64_read(&iocg->active_period) ==
+	    atomic64_read(&ioc->cur_period))
+		return false;
+
+	/* is something in flight? */
+	if (atomic64_read(&iocg->done_vtime) < atomic64_read(&iocg->vtime))
+		return false;
+
+	return true;
+}
+
+/* returns usage with margin added if surplus is large enough */
+static u32 surplus_adjusted_hweight_inuse(u32 usage, u32 hw_inuse)
+{
+	/* add margin */
+	usage = DIV_ROUND_UP(usage * SURPLUS_SCALE_PCT, 100);
+	usage += SURPLUS_SCALE_ABS;
+
+	/* don't bother if the surplus is too small */
+	if (usage + SURPLUS_MIN_ADJ_DELTA > hw_inuse)
+		return 0;
+
+	return usage;
+}
+
+static void ioc_timer_fn(struct timer_list *timer)
+{
+	struct ioc *ioc = container_of(timer, struct ioc, timer);
+	struct ioc_gq *iocg, *tiocg;
+	struct ioc_now now;
+	int nr_surpluses = 0, nr_shortages = 0, nr_lagging = 0;
+	u32 ppm_rthr = MILLION - ioc->params.qos[QOS_RPPM];
+	u32 ppm_wthr = MILLION - ioc->params.qos[QOS_WPPM];
+	u32 missed_ppm[2], rq_wait_pct;
+	u64 period_vtime;
+	int i;
+
+	/* how were the latencies during the period? */
+	ioc_lat_stat(ioc, missed_ppm, &rq_wait_pct);
+
+	/* take care of active iocgs */
+	spin_lock_irq(&ioc->lock);
+
+	ioc_now(ioc, &now);
+
+	period_vtime = now.vnow - ioc->period_at_vtime;
+	if (WARN_ON_ONCE(!period_vtime)) {
+		spin_unlock_irq(&ioc->lock);
+		return;
+	}
+
+	/*
+	 * Waiters determine the sleep durations based on the vrate they
+	 * saw at the time of sleep.  If vrate has increased, some waiters
+	 * could be sleeping for too long.  Wake up tardy waiters which
+	 * should have woken up in the last period and expire idle iocgs.
+	 */
+	list_for_each_entry_safe(iocg, tiocg, &ioc->active_iocgs, active_list) {
+		if (!waitqueue_active(&iocg->waitq) && !iocg_is_idle(iocg))
+			continue;
+
+		spin_lock(&iocg->waitq.lock);
+
+		if (waitqueue_active(&iocg->waitq)) {
+			/* might be oversleeping vtime / hweight changes, kick */
+			iocg_kick_waitq(iocg, &now);
+			iocg_kick_delay(iocg, &now, 0);
+		} else if (iocg_is_idle(iocg)) {
+			/* no waiter and idle, deactivate */
+			iocg->last_inuse = iocg->inuse;
+			__propagate_active_weight(iocg, 0, 0);
+			list_del_init(&iocg->active_list);
+		}
+
+		spin_unlock(&iocg->waitq.lock);
+	}
+	commit_active_weights(ioc);
+
+	/* calc usages and see whether some weights need to be moved around */
+	list_for_each_entry(iocg, &ioc->active_iocgs, active_list) {
+		u64 vdone, vtime, vusage, vmargin, vmin;
+		u32 hw_active, hw_inuse, usage;
+
+		/*
+		 * Collect unused and wind vtime closer to vnow to prevent
+		 * iocgs from accumulating a large amount of budget.
+		 */
+		vdone = atomic64_read(&iocg->done_vtime);
+		vtime = atomic64_read(&iocg->vtime);
+		current_hweight(iocg, &hw_active, &hw_inuse);
+
+		/*
+		 * Latency QoS detection doesn't account for IOs which are
+		 * in-flight for longer than a period.  Detect them by
+		 * comparing vdone against period start.  If lagging behind
+		 * IOs from past periods, don't increase vrate.
+		 */
+		if (!atomic_read(&iocg_to_blkg(iocg)->use_delay) &&
+		    time_after64(vtime, vdone) &&
+		    time_after64(vtime, now.vnow -
+				 MAX_LAGGING_PERIODS * period_vtime) &&
+		    time_before64(vdone, now.vnow - period_vtime))
+			nr_lagging++;
+
+		if (waitqueue_active(&iocg->waitq))
+			vusage = now.vnow - iocg->last_vtime;
+		else if (time_before64(iocg->last_vtime, vtime))
+			vusage = vtime - iocg->last_vtime;
+		else
+			vusage = 0;
+
+		iocg->last_vtime += vusage;
+		/*
+		 * Factor in in-flight vtime into vusage to avoid
+		 * high-latency completions appearing as idle.  This should
+		 * be done after the above ->last_time adjustment.
+		 */
+		vusage = max(vusage, vtime - vdone);
+
+		/* calculate hweight based usage ratio and record */
+		if (vusage) {
+			usage = DIV64_U64_ROUND_UP(vusage * hw_inuse,
+						   period_vtime);
+			iocg->usage_idx = (iocg->usage_idx + 1) % NR_USAGE_SLOTS;
+			iocg->usages[iocg->usage_idx] = usage;
+		} else {
+			usage = 0;
+		}
+
+		/* see whether there's surplus vtime */
+		vmargin = ioc->margin_us * now.vrate;
+		vmin = now.vnow - vmargin;
+
+		iocg->has_surplus = false;
+
+		if (!waitqueue_active(&iocg->waitq) &&
+		    time_before64(vtime, vmin)) {
+			u64 delta = vmin - vtime;
+
+			/* throw away surplus vtime */
+			atomic64_add(delta, &iocg->vtime);
+			atomic64_add(delta, &iocg->done_vtime);
+			iocg->last_vtime += delta;
+			/* if usage is sufficiently low, maybe it can donate */
+			if (surplus_adjusted_hweight_inuse(usage, hw_inuse)) {
+				iocg->has_surplus = true;
+				nr_surpluses++;
+			}
+		} else if (hw_inuse < hw_active) {
+			u32 new_hwi, new_inuse;
+
+			/* was donating but might need to take back some */
+			if (waitqueue_active(&iocg->waitq)) {
+				new_hwi = hw_active;
+			} else {
+				new_hwi = max(hw_inuse,
+					      usage * SURPLUS_SCALE_PCT / 100 +
+					      SURPLUS_SCALE_ABS);
+			}
+
+			new_inuse = div64_u64((u64)iocg->inuse * new_hwi,
+					      hw_inuse);
+			new_inuse = clamp_t(u32, new_inuse, 1, iocg->active);
+
+			if (new_inuse > iocg->inuse) {
+				TRACE_IOCG_PATH(inuse_takeback, iocg, &now,
+						iocg->inuse, new_inuse,
+						hw_inuse, new_hwi);
+				__propagate_active_weight(iocg, iocg->weight,
+							  new_inuse);
+			}
+		} else {
+			/* genuninely out of vtime */
+			nr_shortages++;
+		}
+	}
+
+	if (!nr_shortages || !nr_surpluses)
+		goto skip_surplus_transfers;
+
+	/* there are both shortages and surpluses, transfer surpluses */
+	list_for_each_entry(iocg, &ioc->active_iocgs, active_list) {
+		u32 usage, hw_active, hw_inuse, new_hwi, new_inuse;
+		int nr_valid = 0;
+
+		if (!iocg->has_surplus)
+			continue;
+
+		/* base the decision on max historical usage */
+		for (i = 0, usage = 0; i < NR_USAGE_SLOTS; i++) {
+			if (iocg->usages[i]) {
+				usage = max(usage, iocg->usages[i]);
+				nr_valid++;
+			}
+		}
+		if (nr_valid < MIN_VALID_USAGES)
+			continue;
+
+		current_hweight(iocg, &hw_active, &hw_inuse);
+		new_hwi = surplus_adjusted_hweight_inuse(usage, hw_inuse);
+		if (!new_hwi)
+			continue;
+
+		new_inuse = DIV64_U64_ROUND_UP((u64)iocg->inuse * new_hwi,
+					       hw_inuse);
+		if (new_inuse < iocg->inuse) {
+			TRACE_IOCG_PATH(inuse_giveaway, iocg, &now,
+					iocg->inuse, new_inuse,
+					hw_inuse, new_hwi);
+			__propagate_active_weight(iocg, iocg->weight, new_inuse);
+		}
+	}
+skip_surplus_transfers:
+	commit_active_weights(ioc);
+
+	/*
+	 * If q is getting clogged or we're missing too much, we're issuing
+	 * too much IO and should lower vtime rate.  If we're not missing
+	 * and experiencing shortages but not surpluses, we're too stingy
+	 * and should increase vtime rate.
+	 */
+	if (rq_wait_pct > RQ_WAIT_BUSY_PCT ||
+	    missed_ppm[READ] > ppm_rthr ||
+	    missed_ppm[WRITE] > ppm_wthr) {
+		ioc->busy_level = max(ioc->busy_level, 0);
+		ioc->busy_level++;
+	} else if (nr_lagging) {
+		ioc->busy_level = max(ioc->busy_level, 0);
+	} else if (nr_shortages && !nr_surpluses &&
+		   rq_wait_pct <= RQ_WAIT_BUSY_PCT * UNBUSY_THR_PCT / 100 &&
+		   missed_ppm[READ] <= ppm_rthr * UNBUSY_THR_PCT / 100 &&
+		   missed_ppm[WRITE] <= ppm_wthr * UNBUSY_THR_PCT / 100) {
+		ioc->busy_level = min(ioc->busy_level, 0);
+		ioc->busy_level--;
+	} else {
+		ioc->busy_level = 0;
+	}
+
+	ioc->busy_level = clamp(ioc->busy_level, -1000, 1000);
+
+	if (ioc->busy_level) {
+		u64 vrate = atomic64_read(&ioc->vtime_rate);
+		u64 vrate_min = ioc->vrate_min, vrate_max = ioc->vrate_max;
+
+		/* rq_wait signal is always reliable, ignore user vrate_min */
+		if (rq_wait_pct > RQ_WAIT_BUSY_PCT)
+			vrate_min = VRATE_MIN;
+
+		/*
+		 * If vrate is out of bounds, apply clamp gradually as the
+		 * bounds can change abruptly.  Otherwise, apply busy_level
+		 * based adjustment.
+		 */
+		if (vrate < vrate_min) {
+			vrate = div64_u64(vrate * (100 + VRATE_CLAMP_ADJ_PCT),
+					  100);
+			vrate = min(vrate, vrate_min);
+		} else if (vrate > vrate_max) {
+			vrate = div64_u64(vrate * (100 - VRATE_CLAMP_ADJ_PCT),
+					  100);
+			vrate = max(vrate, vrate_max);
+		} else {
+			int idx = min_t(int, abs(ioc->busy_level),
+					ARRAY_SIZE(vrate_adj_pct) - 1);
+			u32 adj_pct = vrate_adj_pct[idx];
+
+			if (ioc->busy_level > 0)
+				adj_pct = 100 - adj_pct;
+			else
+				adj_pct = 100 + adj_pct;
+
+			vrate = clamp(DIV64_U64_ROUND_UP(vrate * adj_pct, 100),
+				      vrate_min, vrate_max);
+		}
+
+		trace_iocost_ioc_vrate_adj(ioc, vrate, &missed_ppm, rq_wait_pct,
+					   nr_lagging, nr_shortages,
+					   nr_surpluses);
+
+		atomic64_set(&ioc->vtime_rate, vrate);
+		ioc->inuse_margin_vtime = DIV64_U64_ROUND_UP(
+			ioc->period_us * vrate * INUSE_MARGIN_PCT, 100);
+	}
+
+	ioc_refresh_params(ioc, false);
+
+	/*
+	 * This period is done.  Move onto the next one.  If nothing's
+	 * going on with the device, stop the timer.
+	 */
+	atomic64_inc(&ioc->cur_period);
+
+	if (ioc->running != IOC_STOP) {
+		if (!list_empty(&ioc->active_iocgs)) {
+			ioc_start_period(ioc, &now);
+		} else {
+			ioc->busy_level = 0;
+			ioc->running = IOC_IDLE;
+		}
+	}
+
+	spin_unlock_irq(&ioc->lock);
+}
+
+static void calc_vtime_cost_builtin(struct bio *bio, struct ioc_gq *iocg,
+				    bool is_merge, u64 *costp)
+{
+	struct ioc *ioc = iocg->ioc;
+	u64 coef_seqio, coef_randio, coef_page;
+	u64 pages = max_t(u64, bio_sectors(bio) >> IOC_SECT_TO_PAGE_SHIFT, 1);
+	u64 seek_pages = 0;
+	u64 cost = 0;
+
+	switch (bio_op(bio)) {
+	case REQ_OP_READ:
+		coef_seqio	= ioc->params.lcoefs[LCOEF_RSEQIO];
+		coef_randio	= ioc->params.lcoefs[LCOEF_RRANDIO];
+		coef_page	= ioc->params.lcoefs[LCOEF_RPAGE];
+		break;
+	case REQ_OP_WRITE:
+		coef_seqio	= ioc->params.lcoefs[LCOEF_WSEQIO];
+		coef_randio	= ioc->params.lcoefs[LCOEF_WRANDIO];
+		coef_page	= ioc->params.lcoefs[LCOEF_WPAGE];
+		break;
+	default:
+		goto out;
+	}
+
+	if (iocg->cursor) {
+		seek_pages = abs(bio->bi_iter.bi_sector - iocg->cursor);
+		seek_pages >>= IOC_SECT_TO_PAGE_SHIFT;
+	}
+
+	if (!is_merge) {
+		if (seek_pages > LCOEF_RANDIO_PAGES) {
+			cost += coef_randio;
+		} else {
+			cost += coef_seqio;
+		}
+	}
+	cost += pages * coef_page;
+out:
+	*costp = cost;
+}
+
+static u64 calc_vtime_cost(struct bio *bio, struct ioc_gq *iocg, bool is_merge)
+{
+	u64 cost;
+
+	calc_vtime_cost_builtin(bio, iocg, is_merge, &cost);
+	return cost;
+}
+
+static void ioc_rqos_throttle(struct rq_qos *rqos, struct bio *bio)
+{
+	struct blkcg_gq *blkg = bio->bi_blkg;
+	struct ioc *ioc = rqos_to_ioc(rqos);
+	struct ioc_gq *iocg = blkg_to_iocg(blkg);
+	struct ioc_now now;
+	struct iocg_wait wait;
+	u32 hw_active, hw_inuse;
+	u64 abs_cost, cost, vtime;
+
+	/* bypass IOs if disabled or for root cgroup */
+	if (!ioc->enabled || !iocg->level)
+		return;
+
+	/* always activate so that even 0 cost IOs get protected to some level */
+	if (!iocg_activate(iocg, &now))
+		return;
+
+	/* calculate the absolute vtime cost */
+	abs_cost = calc_vtime_cost(bio, iocg, false);
+	if (!abs_cost)
+		return;
+
+	iocg->cursor = bio_end_sector(bio);
+
+	vtime = atomic64_read(&iocg->vtime);
+	current_hweight(iocg, &hw_active, &hw_inuse);
+
+	if (hw_inuse < hw_active &&
+	    time_after_eq64(vtime + ioc->inuse_margin_vtime, now.vnow)) {
+		TRACE_IOCG_PATH(inuse_reset, iocg, &now,
+				iocg->inuse, iocg->weight, hw_inuse, hw_active);
+		spin_lock_irq(&ioc->lock);
+		propagate_active_weight(iocg, iocg->weight, iocg->weight);
+		spin_unlock_irq(&ioc->lock);
+		current_hweight(iocg, &hw_active, &hw_inuse);
+	}
+
+	cost = abs_cost_to_cost(abs_cost, hw_inuse);
+
+	/*
+	 * If no one's waiting and within budget, issue right away.  The
+	 * tests are racy but the races aren't systemic - we only miss once
+	 * in a while which is fine.
+	 */
+	if (!waitqueue_active(&iocg->waitq) &&
+	    time_before_eq64(vtime + cost, now.vnow)) {
+		iocg_commit_bio(iocg, bio, cost);
+		return;
+	}
+
+	if (bio_issue_as_root_blkg(bio) || fatal_signal_pending(current)) {
+		iocg_commit_bio(iocg, bio, cost);
+		iocg_kick_delay(iocg, &now, cost);
+		return;
+	}
+
+	/*
+	 * Append self to the waitq and schedule the wakeup timer if we're
+	 * the first waiter.  The timer duration is calculated based on the
+	 * current vrate.  vtime and hweight changes can make it too short
+	 * or too long.  Each wait entry records the absolute cost it's
+	 * waiting for to allow re-evaluation using a custom wait entry.
+	 *
+	 * If too short, the timer simply reschedules itself.  If too long,
+	 * the period timer will notice and trigger wakeups.
+	 *
+	 * All waiters are on iocg->waitq and the wait states are
+	 * synchronized using waitq.lock.
+	 */
+	spin_lock_irq(&iocg->waitq.lock);
+
+	/*
+	 * We activated above but w/o any synchronization.  Deactivation is
+	 * synchronized with waitq.lock and we won't get deactivated as
+	 * long as we're waiting, so we're good if we're activated here.
+	 * In the unlikely case that we are deactivated, just issue the IO.
+	 */
+	if (unlikely(list_empty(&iocg->active_list))) {
+		spin_unlock_irq(&iocg->waitq.lock);
+		iocg_commit_bio(iocg, bio, cost);
+		return;
+	}
+
+	init_waitqueue_func_entry(&wait.wait, iocg_wake_fn);
+	wait.wait.private = current;
+	wait.bio = bio;
+	wait.abs_cost = abs_cost;
+	wait.committed = false;	/* will be set true by waker */
+
+	__add_wait_queue_entry_tail(&iocg->waitq, &wait.wait);
+	iocg_kick_waitq(iocg, &now);
+
+	spin_unlock_irq(&iocg->waitq.lock);
+
+	while (true) {
+		set_current_state(TASK_UNINTERRUPTIBLE);
+		if (wait.committed)
+			break;
+		io_schedule();
+	}
+
+	/* waker already committed us, proceed */
+	finish_wait(&iocg->waitq, &wait.wait);
+}
+
+static void ioc_rqos_merge(struct rq_qos *rqos, struct request *rq,
+			   struct bio *bio)
+{
+	struct ioc_gq *iocg = blkg_to_iocg(bio->bi_blkg);
+	sector_t bio_end = bio_end_sector(bio);
+	u32 hw_inuse;
+	u64 abs_cost, cost;
+
+	/* add iff the existing request has cost assigned */
+	if (!rq->bio || !rq->bio->bi_iocost_cost)
+		return;
+
+	abs_cost = calc_vtime_cost(bio, iocg, true);
+	if (!abs_cost)
+		return;
+
+	/* update cursor if backmerging into the request at the cursor */
+	if (blk_rq_pos(rq) < bio_end &&
+	    blk_rq_pos(rq) + blk_rq_sectors(rq) == iocg->cursor)
+		iocg->cursor = bio_end;
+
+	current_hweight(iocg, NULL, &hw_inuse);
+	cost = div64_u64(abs_cost * HWEIGHT_WHOLE, hw_inuse);
+	bio->bi_iocost_cost = cost;
+
+	atomic64_add(cost, &iocg->vtime);
+}
+
+static void ioc_rqos_done_bio(struct rq_qos *rqos, struct bio *bio)
+{
+	struct ioc_gq *iocg = blkg_to_iocg(bio->bi_blkg);
+
+	if (iocg && bio->bi_iocost_cost)
+		atomic64_add(bio->bi_iocost_cost, &iocg->done_vtime);
+}
+
+static void ioc_rqos_done(struct rq_qos *rqos, struct request *rq)
+{
+	struct ioc *ioc = rqos_to_ioc(rqos);
+	u64 on_q_ns, rq_wait_ns;
+	int pidx, rw;
+
+	if (!ioc->enabled || !rq->pre_start_time_ns || !rq->start_time_ns)
+		return;
+
+	switch (req_op(rq) & REQ_OP_MASK) {
+	case REQ_OP_READ:
+		pidx = QOS_RLAT;
+		rw = READ;
+		break;
+	case REQ_OP_WRITE:
+		pidx = QOS_WLAT;
+		rw = WRITE;
+		break;
+	default:
+		return;
+	}
+
+	on_q_ns = ktime_get_ns() - rq->pre_start_time_ns;
+	rq_wait_ns = rq->start_time_ns - rq->pre_start_time_ns;
+
+	if (on_q_ns <= ioc->params.qos[pidx] * NSEC_PER_USEC)
+		this_cpu_inc(ioc->pcpu_stat->missed[rw].nr_met);
+	else
+		this_cpu_inc(ioc->pcpu_stat->missed[rw].nr_missed);
+
+	this_cpu_add(ioc->pcpu_stat->rq_wait_ns, rq_wait_ns);
+}
+
+static void ioc_rqos_queue_depth_changed(struct rq_qos *rqos)
+{
+	struct ioc *ioc = rqos_to_ioc(rqos);
+
+	spin_lock_irq(&ioc->lock);
+	ioc_refresh_params(ioc, false);
+	spin_unlock_irq(&ioc->lock);
+}
+
+static void ioc_rqos_exit(struct rq_qos *rqos)
+{
+	struct ioc *ioc = rqos_to_ioc(rqos);
+
+	blkcg_deactivate_policy(rqos->q, &blkcg_policy_iocost);
+
+	spin_lock_irq(&ioc->lock);
+	ioc->running = IOC_STOP;
+	spin_unlock_irq(&ioc->lock);
+
+	del_timer_sync(&ioc->timer);
+	free_percpu(ioc->pcpu_stat);
+	kfree(ioc);
+}
+
+static struct rq_qos_ops ioc_rqos_ops = {
+	.throttle = ioc_rqos_throttle,
+	.merge = ioc_rqos_merge,
+	.done_bio = ioc_rqos_done_bio,
+	.done = ioc_rqos_done,
+	.queue_depth_changed = ioc_rqos_queue_depth_changed,
+	.exit = ioc_rqos_exit,
+};
+
+static int blk_iocost_init(struct request_queue *q)
+{
+	struct ioc *ioc;
+	struct rq_qos *rqos;
+	int ret;
+
+	ioc = kzalloc(sizeof(*ioc), GFP_KERNEL);
+	if (!ioc)
+		return -ENOMEM;
+
+	ioc->pcpu_stat = alloc_percpu(struct ioc_pcpu_stat);
+	if (!ioc->pcpu_stat) {
+		kfree(ioc);
+		return -ENOMEM;
+	}
+
+	rqos = &ioc->rqos;
+	rqos->id = RQ_QOS_COST;
+	rqos->ops = &ioc_rqos_ops;
+	rqos->q = q;
+
+	spin_lock_init(&ioc->lock);
+	timer_setup(&ioc->timer, ioc_timer_fn, 0);
+	INIT_LIST_HEAD(&ioc->active_iocgs);
+
+	ioc->running = IOC_IDLE;
+	atomic64_set(&ioc->vtime_rate, VTIME_PER_USEC);
+	seqcount_init(&ioc->period_seqcount);
+	ioc->period_at = ktime_to_us(ktime_get());
+	atomic64_set(&ioc->cur_period, 0);
+	atomic_set(&ioc->hweight_gen, 0);
+
+	spin_lock_irq(&ioc->lock);
+	ioc->autop_idx = AUTOP_INVALID;
+	ioc_refresh_params(ioc, true);
+	spin_unlock_irq(&ioc->lock);
+
+	rq_qos_add(q, rqos);
+	ret = blkcg_activate_policy(q, &blkcg_policy_iocost);
+	if (ret) {
+		rq_qos_del(q, rqos);
+		kfree(ioc);
+		return ret;
+	}
+	return 0;
+}
+
+static struct blkcg_policy_data *ioc_cpd_alloc(gfp_t gfp)
+{
+	struct ioc_cgrp *iocc;
+
+	iocc = kzalloc(sizeof(struct ioc_cgrp), gfp);
+	iocc->dfl_weight = CGROUP_WEIGHT_DFL;
+
+	return &iocc->cpd;
+}
+
+static void ioc_cpd_free(struct blkcg_policy_data *cpd)
+{
+	kfree(container_of(cpd, struct ioc_cgrp, cpd));
+}
+
+static struct blkg_policy_data *ioc_pd_alloc(gfp_t gfp, struct request_queue *q,
+					     struct blkcg *blkcg)
+{
+	int levels = blkcg->css.cgroup->level + 1;
+	struct ioc_gq *iocg;
+
+	iocg = kzalloc_node(sizeof(*iocg) + levels * sizeof(iocg->ancestors[0]),
+			    gfp, q->node);
+	if (!iocg)
+		return NULL;
+
+	return &iocg->pd;
+}
+
+static void ioc_pd_init(struct blkg_policy_data *pd)
+{
+	struct ioc_gq *iocg = pd_to_iocg(pd);
+	struct blkcg_gq *blkg = pd_to_blkg(&iocg->pd);
+	struct ioc *ioc = q_to_ioc(blkg->q);
+	struct ioc_now now;
+	struct blkcg_gq *tblkg;
+	unsigned long flags;
+
+	ioc_now(ioc, &now);
+
+	iocg->ioc = ioc;
+	atomic64_set(&iocg->vtime, now.vnow);
+	atomic64_set(&iocg->done_vtime, now.vnow);
+	atomic64_set(&iocg->active_period, atomic64_read(&ioc->cur_period));
+	INIT_LIST_HEAD(&iocg->active_list);
+	iocg->hweight_active = HWEIGHT_WHOLE;
+	iocg->hweight_inuse = HWEIGHT_WHOLE;
+
+	init_waitqueue_head(&iocg->waitq);
+	hrtimer_init(&iocg->waitq_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
+	iocg->waitq_timer.function = iocg_waitq_timer_fn;
+	hrtimer_init(&iocg->delay_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
+	iocg->delay_timer.function = iocg_delay_timer_fn;
+
+	iocg->level = blkg->blkcg->css.cgroup->level;
+
+	for (tblkg = blkg; tblkg; tblkg = tblkg->parent) {
+		struct ioc_gq *tiocg = blkg_to_iocg(tblkg);
+		iocg->ancestors[tiocg->level] = tiocg;
+	}
+
+	spin_lock_irqsave(&ioc->lock, flags);
+	weight_updated(iocg);
+	spin_unlock_irqrestore(&ioc->lock, flags);
+}
+
+static void ioc_pd_free(struct blkg_policy_data *pd)
+{
+	struct ioc_gq *iocg = pd_to_iocg(pd);
+	struct ioc *ioc = iocg->ioc;
+
+	if (ioc) {
+		hrtimer_cancel(&iocg->waitq_timer);
+		hrtimer_cancel(&iocg->delay_timer);
+
+		spin_lock(&ioc->lock);
+		if (!list_empty(&iocg->active_list)) {
+			propagate_active_weight(iocg, 0, 0);
+			list_del_init(&iocg->active_list);
+		}
+		spin_unlock(&ioc->lock);
+	}
+	kfree(iocg);
+}
+
+static u64 ioc_weight_prfill(struct seq_file *sf, struct blkg_policy_data *pd,
+			     int off)
+{
+	const char *dname = blkg_dev_name(pd->blkg);
+	struct ioc_gq *iocg = pd_to_iocg(pd);
+
+	if (dname && iocg->cfg_weight)
+		seq_printf(sf, "%s %u\n", dname, iocg->cfg_weight);
+	return 0;
+}
+
+
+static int ioc_weight_show(struct seq_file *sf, void *v)
+{
+	struct blkcg *blkcg = css_to_blkcg(seq_css(sf));
+	struct ioc_cgrp *iocc = blkcg_to_iocc(blkcg);
+
+	seq_printf(sf, "default %u\n", iocc->dfl_weight);
+	blkcg_print_blkgs(sf, blkcg, ioc_weight_prfill,
+			  &blkcg_policy_iocost, seq_cft(sf)->private, false);
+	return 0;
+}
+
+static ssize_t ioc_weight_write(struct kernfs_open_file *of, char *buf,
+				size_t nbytes, loff_t off)
+{
+	struct blkcg *blkcg = css_to_blkcg(of_css(of));
+	struct ioc_cgrp *iocc = blkcg_to_iocc(blkcg);
+	struct blkg_conf_ctx ctx;
+	struct ioc_gq *iocg;
+	u32 v;
+	int ret;
+
+	if (!strchr(buf, ':')) {
+		struct blkcg_gq *blkg;
+
+		if (!sscanf(buf, "default %u", &v) && !sscanf(buf, "%u", &v))
+			return -EINVAL;
+
+		if (v < CGROUP_WEIGHT_MIN || v > CGROUP_WEIGHT_MAX)
+			return -EINVAL;
+
+		spin_lock(&blkcg->lock);
+		iocc->dfl_weight = v;
+		hlist_for_each_entry(blkg, &blkcg->blkg_list, blkcg_node) {
+			struct ioc_gq *iocg = blkg_to_iocg(blkg);
+
+			if (iocg) {
+				spin_lock_irq(&iocg->ioc->lock);
+				weight_updated(iocg);
+				spin_unlock_irq(&iocg->ioc->lock);
+			}
+		}
+		spin_unlock(&blkcg->lock);
+
+		return nbytes;
+	}
+
+	ret = blkg_conf_prep(blkcg, &blkcg_policy_iocost, buf, &ctx);
+	if (ret)
+		return ret;
+
+	iocg = blkg_to_iocg(ctx.blkg);
+
+	if (!strncmp(ctx.body, "default", 7)) {
+		v = 0;
+	} else {
+		if (!sscanf(ctx.body, "%u", &v))
+			goto einval;
+		if (v < CGROUP_WEIGHT_MIN || v > CGROUP_WEIGHT_MAX)
+			goto einval;
+	}
+
+	spin_lock_irq(&iocg->ioc->lock);
+	iocg->cfg_weight = v;
+	weight_updated(iocg);
+	spin_unlock_irq(&iocg->ioc->lock);
+
+	blkg_conf_finish(&ctx);
+	return nbytes;
+
+einval:
+	blkg_conf_finish(&ctx);
+	return -EINVAL;
+}
+
+static u64 ioc_qos_prfill(struct seq_file *sf, struct blkg_policy_data *pd,
+			  int off)
+{
+	const char *dname = blkg_dev_name(pd->blkg);
+	struct ioc *ioc = pd_to_iocg(pd)->ioc;
+
+	if (!dname)
+		return 0;
+
+	seq_printf(sf, "%s enable=%d ctrl=%s rpct=%u.%02u rlat=%u wpct=%u.%02u wlat=%u min=%u.%02u max=%u.%02u\n",
+		   dname, ioc->enabled, ioc->user_qos_params ? "user" : "auto",
+		   ioc->params.qos[QOS_RPPM] / 10000,
+		   ioc->params.qos[QOS_RPPM] % 10000 / 100,
+		   ioc->params.qos[QOS_RLAT],
+		   ioc->params.qos[QOS_WPPM] / 10000,
+		   ioc->params.qos[QOS_WPPM] % 10000 / 100,
+		   ioc->params.qos[QOS_WLAT],
+		   ioc->params.qos[QOS_MIN] / 10000,
+		   ioc->params.qos[QOS_MIN] % 10000 / 100,
+		   ioc->params.qos[QOS_MAX] / 10000,
+		   ioc->params.qos[QOS_MAX] % 10000 / 100);
+	return 0;
+}
+
+static int ioc_qos_show(struct seq_file *sf, void *v)
+{
+	struct blkcg *blkcg = css_to_blkcg(seq_css(sf));
+
+	blkcg_print_blkgs(sf, blkcg, ioc_qos_prfill,
+			  &blkcg_policy_iocost, seq_cft(sf)->private, false);
+	return 0;
+}
+
+static const match_table_t qos_ctrl_tokens = {
+	{ QOS_ENABLE,		"enable=%u"	},
+	{ QOS_CTRL,		"ctrl=%s"	},
+	{ NR_QOS_CTRL_PARAMS,	NULL		},
+};
+
+static const match_table_t qos_tokens = {
+	{ QOS_RPPM,		"rpct=%s"	},
+	{ QOS_RLAT,		"rlat=%u"	},
+	{ QOS_WPPM,		"wpct=%s"	},
+	{ QOS_WLAT,		"wlat=%u"	},
+	{ QOS_MIN,		"min=%s"	},
+	{ QOS_MAX,		"max=%s"	},
+	{ NR_QOS_PARAMS,	NULL		},
+};
+
+static ssize_t ioc_qos_write(struct kernfs_open_file *of, char *input,
+			     size_t nbytes, loff_t off)
+{
+	struct gendisk *disk;
+	struct ioc *ioc;
+	u32 qos[NR_QOS_PARAMS];
+	bool enable, user;
+	char *p;
+	int ret;
+
+	disk = blkcg_conf_get_disk(&input);
+	if (IS_ERR(disk))
+		return PTR_ERR(disk);
+
+	ioc = q_to_ioc(disk->queue);
+	if (!ioc) {
+		ret = blk_iocost_init(disk->queue);
+		if (ret)
+			goto err;
+		ioc = q_to_ioc(disk->queue);
+	}
+
+	spin_lock_irq(&ioc->lock);
+	memcpy(qos, ioc->params.qos, sizeof(qos));
+	enable = ioc->enabled;
+	user = ioc->user_qos_params;
+	spin_unlock_irq(&ioc->lock);
+
+	while ((p = strsep(&input, " \t\n"))) {
+		substring_t args[MAX_OPT_ARGS];
+		char buf[32];
+		int tok;
+		s64 v;
+
+		if (!*p)
+			continue;
+
+		switch (match_token(p, qos_ctrl_tokens, args)) {
+		case QOS_ENABLE:
+			match_u64(&args[0], &v);
+			enable = v;
+			continue;
+		case QOS_CTRL:
+			match_strlcpy(buf, &args[0], sizeof(buf));
+			if (!strcmp(buf, "auto"))
+				user = false;
+			else if (!strcmp(buf, "user"))
+				user = true;
+			else
+				goto einval;
+			continue;
+		}
+
+		tok = match_token(p, qos_tokens, args);
+		switch (tok) {
+		case QOS_RPPM:
+		case QOS_WPPM:
+			if (match_strlcpy(buf, &args[0], sizeof(buf)) >=
+			    sizeof(buf))
+				goto einval;
+			if (cgroup_parse_float(buf, 2, &v))
+				goto einval;
+			if (v < 0 || v > 10000)
+				goto einval;
+			qos[tok] = v * 100;
+			break;
+		case QOS_RLAT:
+		case QOS_WLAT:
+			if (match_u64(&args[0], &v))
+				goto einval;
+			qos[tok] = v;
+			break;
+		case QOS_MIN:
+		case QOS_MAX:
+			if (match_strlcpy(buf, &args[0], sizeof(buf)) >=
+			    sizeof(buf))
+				goto einval;
+			if (cgroup_parse_float(buf, 2, &v))
+				goto einval;
+			if (v < 0)
+				goto einval;
+			qos[tok] = clamp_t(s64, v * 100,
+					   VRATE_MIN_PPM, VRATE_MAX_PPM);
+			break;
+		default:
+			goto einval;
+		}
+		user = true;
+	}
+
+	if (qos[QOS_MIN] > qos[QOS_MAX])
+		goto einval;
+
+	spin_lock_irq(&ioc->lock);
+
+	if (enable) {
+		blk_queue_flag_set(QUEUE_FLAG_REC_PRESTART, ioc->rqos.q);
+		ioc->enabled = true;
+	} else {
+		blk_queue_flag_clear(QUEUE_FLAG_REC_PRESTART, ioc->rqos.q);
+		ioc->enabled = false;
+	}
+
+	if (user) {
+		memcpy(ioc->params.qos, qos, sizeof(qos));
+		ioc->user_qos_params = true;
+	} else {
+		ioc->user_qos_params = false;
+	}
+
+	ioc_refresh_params(ioc, true);
+	spin_unlock_irq(&ioc->lock);
+
+	put_disk_and_module(disk);
+	return nbytes;
+einval:
+	ret = -EINVAL;
+err:
+	put_disk_and_module(disk);
+	return ret;
+}
+
+static u64 ioc_cost_model_prfill(struct seq_file *sf,
+				 struct blkg_policy_data *pd, int off)
+{
+	const char *dname = blkg_dev_name(pd->blkg);
+	struct ioc *ioc = pd_to_iocg(pd)->ioc;
+	u64 *u = ioc->params.i_lcoefs;
+
+	if (!dname)
+		return 0;
+
+	seq_printf(sf, "%s ctrl=%s model=linear "
+		   "rbps=%llu rseqiops=%llu rrandiops=%llu "
+		   "wbps=%llu wseqiops=%llu wrandiops=%llu\n",
+		   dname, ioc->user_cost_model ? "user" : "auto",
+		   u[I_LCOEF_RBPS], u[I_LCOEF_RSEQIOPS], u[I_LCOEF_RRANDIOPS],
+		   u[I_LCOEF_WBPS], u[I_LCOEF_WSEQIOPS], u[I_LCOEF_WRANDIOPS]);
+	return 0;
+}
+
+static int ioc_cost_model_show(struct seq_file *sf, void *v)
+{
+	struct blkcg *blkcg = css_to_blkcg(seq_css(sf));
+
+	blkcg_print_blkgs(sf, blkcg, ioc_cost_model_prfill,
+			  &blkcg_policy_iocost, seq_cft(sf)->private, false);
+	return 0;
+}
+
+static const match_table_t cost_ctrl_tokens = {
+	{ COST_CTRL,		"ctrl=%s"	},
+	{ COST_MODEL,		"model=%s"	},
+	{ NR_COST_CTRL_PARAMS,	NULL		},
+};
+
+static const match_table_t i_lcoef_tokens = {
+	{ I_LCOEF_RBPS,		"rbps=%u"	},
+	{ I_LCOEF_RSEQIOPS,	"rseqiops=%u"	},
+	{ I_LCOEF_RRANDIOPS,	"rrandiops=%u"	},
+	{ I_LCOEF_WBPS,		"wbps=%u"	},
+	{ I_LCOEF_WSEQIOPS,	"wseqiops=%u"	},
+	{ I_LCOEF_WRANDIOPS,	"wrandiops=%u"	},
+	{ NR_I_LCOEFS,		NULL		},
+};
+
+static ssize_t ioc_cost_model_write(struct kernfs_open_file *of, char *input,
+				    size_t nbytes, loff_t off)
+{
+	struct gendisk *disk;
+	struct ioc *ioc;
+	u64 u[NR_I_LCOEFS];
+	bool user;
+	char *p;
+	int ret;
+
+	disk = blkcg_conf_get_disk(&input);
+	if (IS_ERR(disk))
+		return PTR_ERR(disk);
+
+	ioc = q_to_ioc(disk->queue);
+	if (!ioc) {
+		ret = blk_iocost_init(disk->queue);
+		if (ret)
+			goto err;
+		ioc = q_to_ioc(disk->queue);
+	}
+
+	spin_lock_irq(&ioc->lock);
+	memcpy(u, ioc->params.i_lcoefs, sizeof(u));
+	user = ioc->user_cost_model;
+	spin_unlock_irq(&ioc->lock);
+
+	while ((p = strsep(&input, " \t\n"))) {
+		substring_t args[MAX_OPT_ARGS];
+		char buf[32];
+		int tok;
+		u64 v;
+
+		if (!*p)
+			continue;
+
+		switch (match_token(p, cost_ctrl_tokens, args)) {
+		case COST_CTRL:
+			match_strlcpy(buf, &args[0], sizeof(buf));
+			if (!strcmp(buf, "auto"))
+				user = false;
+			else if (!strcmp(buf, "user"))
+				user = true;
+			else
+				goto einval;
+			continue;
+		case COST_MODEL:
+			match_strlcpy(buf, &args[0], sizeof(buf));
+			if (strcmp(buf, "linear"))
+				goto einval;
+			continue;
+		}
+
+		tok = match_token(p, i_lcoef_tokens, args);
+		if (tok == NR_I_LCOEFS)
+			goto einval;
+		if (match_u64(&args[0], &v))
+			goto einval;
+		u[tok] = v;
+		user = true;
+	}
+
+	spin_lock_irq(&ioc->lock);
+	if (user) {
+		memcpy(ioc->params.i_lcoefs, u, sizeof(u));
+		ioc->user_cost_model = true;
+	} else {
+		ioc->user_cost_model = false;
+	}
+	ioc_refresh_params(ioc, true);
+	spin_unlock_irq(&ioc->lock);
+
+	put_disk_and_module(disk);
+	return nbytes;
+
+einval:
+	ret = -EINVAL;
+err:
+	put_disk_and_module(disk);
+	return ret;
+}
+
+static struct cftype ioc_files[] = {
+	{
+		.name = "weight",
+		.flags = CFTYPE_NOT_ON_ROOT,
+		.seq_show = ioc_weight_show,
+		.write = ioc_weight_write,
+	},
+	{
+		.name = "cost.qos",
+		.flags = CFTYPE_ONLY_ON_ROOT,
+		.seq_show = ioc_qos_show,
+		.write = ioc_qos_write,
+	},
+	{
+		.name = "cost.model",
+		.flags = CFTYPE_ONLY_ON_ROOT,
+		.seq_show = ioc_cost_model_show,
+		.write = ioc_cost_model_write,
+	},
+	{}
+};
+
+static struct blkcg_policy blkcg_policy_iocost = {
+	.dfl_cftypes	= ioc_files,
+	.cpd_alloc_fn	= ioc_cpd_alloc,
+	.cpd_free_fn	= ioc_cpd_free,
+	.pd_alloc_fn	= ioc_pd_alloc,
+	.pd_init_fn	= ioc_pd_init,
+	.pd_free_fn	= ioc_pd_free,
+};
+
+static int __init ioc_init(void)
+{
+	return blkcg_policy_register(&blkcg_policy_iocost);
+}
+
+static void __exit ioc_exit(void)
+{
+	return blkcg_policy_unregister(&blkcg_policy_iocost);
+}
+
+module_init(ioc_init);
+module_exit(ioc_exit);
diff --git a/block/blk-rq-qos.h b/block/blk-rq-qos.h
index 5f8b75826a98..08a09dbe0f4b 100644
--- a/block/blk-rq-qos.h
+++ b/block/blk-rq-qos.h
@@ -15,6 +15,7 @@ struct blk_mq_debugfs_attr;
 enum rq_qos_id {
 	RQ_QOS_WBT,
 	RQ_QOS_LATENCY,
+	RQ_QOS_COST,
 };
 
 struct rq_wait {
@@ -84,6 +85,8 @@ static inline const char *rq_qos_id_to_name(enum rq_qos_id id)
 		return "wbt";
 	case RQ_QOS_LATENCY:
 		return "latency";
+	case RQ_QOS_COST:
+		return "cost";
 	}
 	return "unknown";
 }
diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
index feff3fe4467e..e53e84bf83a6 100644
--- a/include/linux/blk_types.h
+++ b/include/linux/blk_types.h
@@ -169,6 +169,9 @@ struct bio {
 	 */
 	struct blkcg_gq		*bi_blkg;
 	struct bio_issue	bi_issue;
+#ifdef CONFIG_BLK_CGROUP_IOCOST
+	u64			bi_iocost_cost;
+#endif
 #endif
 	union {
 #if defined(CONFIG_BLK_DEV_INTEGRITY)
diff --git a/include/trace/events/iocost.h b/include/trace/events/iocost.h
new file mode 100644
index 000000000000..ec2217dd57ac
--- /dev/null
+++ b/include/trace/events/iocost.h
@@ -0,0 +1,174 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM iocost
+
+#if !defined(_TRACE_BLK_IOCOST_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_BLK_IOCOST_H
+
+#include <linux/tracepoint.h>
+
+TRACE_EVENT(iocost_iocg_activate,
+
+	TP_PROTO(struct ioc_gq *iocg, const char *path, struct ioc_now *now,
+		u64 last_period, u64 cur_period, u64 vtime),
+
+	TP_ARGS(iocg, path, now, last_period, cur_period, vtime),
+
+	TP_STRUCT__entry (
+		__string(devname, ioc_name(iocg->ioc))
+		__string(cgroup, path)
+		__field(u64, now)
+		__field(u64, vnow)
+		__field(u64, vrate)
+		__field(u64, last_period)
+		__field(u64, cur_period)
+		__field(u64, last_vtime)
+		__field(u64, vtime)
+		__field(u32, weight)
+		__field(u32, inuse)
+		__field(u64, hweight_active)
+		__field(u64, hweight_inuse)
+	),
+
+	TP_fast_assign(
+		__assign_str(devname, ioc_name(iocg->ioc));
+		__assign_str(cgroup, path);
+		__entry->now = now->now;
+		__entry->vnow = now->vnow;
+		__entry->vrate = now->vrate;
+		__entry->last_period = last_period;
+		__entry->cur_period = cur_period;
+		__entry->last_vtime = iocg->last_vtime;
+		__entry->vtime = vtime;
+		__entry->weight = iocg->weight;
+		__entry->inuse = iocg->inuse;
+		__entry->hweight_active = iocg->hweight_active;
+		__entry->hweight_inuse = iocg->hweight_inuse;
+	),
+
+	TP_printk("[%s:%s] now=%llu:%llu vrate=%llu "
+		  "period=%llu->%llu vtime=%llu->%llu "
+		  "weight=%u/%u hweight=%llu/%llu",
+		__get_str(devname), __get_str(cgroup),
+		__entry->now, __entry->vnow, __entry->vrate,
+		__entry->last_period, __entry->cur_period,
+		__entry->last_vtime, __entry->vtime,
+		__entry->inuse, __entry->weight,
+		__entry->hweight_inuse, __entry->hweight_active
+	)
+);
+
+DECLARE_EVENT_CLASS(iocg_inuse_update,
+
+	TP_PROTO(struct ioc_gq *iocg, const char *path, struct ioc_now *now,
+		u32 old_inuse, u32 new_inuse,
+		u64 old_hw_inuse, u64 new_hw_inuse),
+
+	TP_ARGS(iocg, path, now, old_inuse, new_inuse,
+		old_hw_inuse, new_hw_inuse),
+
+	TP_STRUCT__entry (
+		__string(devname, ioc_name(iocg->ioc))
+		__string(cgroup, path)
+		__field(u64, now)
+		__field(u32, old_inuse)
+		__field(u32, new_inuse)
+		__field(u64, old_hweight_inuse)
+		__field(u64, new_hweight_inuse)
+	),
+
+	TP_fast_assign(
+		__assign_str(devname, ioc_name(iocg->ioc));
+		__assign_str(cgroup, path);
+		__entry->now = now->now;
+		__entry->old_inuse = old_inuse;
+		__entry->new_inuse = new_inuse;
+		__entry->old_hweight_inuse = old_hw_inuse;
+		__entry->new_hweight_inuse = new_hw_inuse;
+	),
+
+	TP_printk("[%s:%s] now=%llu inuse=%u->%u hw_inuse=%llu->%llu",
+		__get_str(devname), __get_str(cgroup), __entry->now,
+		__entry->old_inuse, __entry->new_inuse,
+		__entry->old_hweight_inuse, __entry->new_hweight_inuse
+	)
+);
+
+DEFINE_EVENT(iocg_inuse_update, iocost_inuse_takeback,
+
+	TP_PROTO(struct ioc_gq *iocg, const char *path, struct ioc_now *now,
+		u32 old_inuse, u32 new_inuse,
+		u64 old_hw_inuse, u64 new_hw_inuse),
+
+	TP_ARGS(iocg, path, now, old_inuse, new_inuse,
+		old_hw_inuse, new_hw_inuse)
+);
+
+DEFINE_EVENT(iocg_inuse_update, iocost_inuse_giveaway,
+
+	TP_PROTO(struct ioc_gq *iocg, const char *path, struct ioc_now *now,
+		u32 old_inuse, u32 new_inuse,
+		u64 old_hw_inuse, u64 new_hw_inuse),
+
+	TP_ARGS(iocg, path, now, old_inuse, new_inuse,
+		old_hw_inuse, new_hw_inuse)
+);
+
+DEFINE_EVENT(iocg_inuse_update, iocost_inuse_reset,
+
+	TP_PROTO(struct ioc_gq *iocg, const char *path, struct ioc_now *now,
+		u32 old_inuse, u32 new_inuse,
+		u64 old_hw_inuse, u64 new_hw_inuse),
+
+	TP_ARGS(iocg, path, now, old_inuse, new_inuse,
+		old_hw_inuse, new_hw_inuse)
+);
+
+TRACE_EVENT(iocost_ioc_vrate_adj,
+
+	TP_PROTO(struct ioc *ioc, u64 new_vrate, u32 (*missed_ppm)[2],
+		u32 rq_wait_pct, int nr_lagging, int nr_shortages,
+		int nr_surpluses),
+
+	TP_ARGS(ioc, new_vrate, missed_ppm, rq_wait_pct, nr_lagging, nr_shortages,
+		nr_surpluses),
+
+	TP_STRUCT__entry (
+		__string(devname, ioc_name(ioc))
+		__field(u64, old_vrate)
+		__field(u64, new_vrate)
+		__field(int, busy_level)
+		__field(u32, read_missed_ppm)
+		__field(u32, write_missed_ppm)
+		__field(u32, rq_wait_pct)
+		__field(int, nr_lagging)
+		__field(int, nr_shortages)
+		__field(int, nr_surpluses)
+	),
+
+	TP_fast_assign(
+		__assign_str(devname, ioc_name(ioc));
+		__entry->old_vrate = atomic64_read(&ioc->vtime_rate);;
+		__entry->new_vrate = new_vrate;
+		__entry->busy_level = ioc->busy_level;
+		__entry->read_missed_ppm = (*missed_ppm)[READ];
+		__entry->write_missed_ppm = (*missed_ppm)[WRITE];
+		__entry->rq_wait_pct = rq_wait_pct;
+		__entry->nr_lagging = nr_lagging;
+		__entry->nr_shortages = nr_shortages;
+		__entry->nr_surpluses = nr_surpluses;
+	),
+
+	TP_printk("[%s] vrate=%llu->%llu busy=%d missed_ppm=%u:%u rq_wait_pct=%u lagging=%d shortages=%d surpluses=%d",
+		__get_str(devname), __entry->old_vrate, __entry->new_vrate,
+		__entry->busy_level,
+		__entry->read_missed_ppm, __entry->write_missed_ppm,
+		__entry->rq_wait_pct, __entry->nr_lagging, __entry->nr_shortages,
+		__entry->nr_surpluses
+	)
+);
+
+#endif /* _TRACE_BLK_IOCOST_H */
+
+/* This part must be outside protection */
+#include <trace/define_trace.h>
-- 
2.17.1