switch blk-cgroup to work on gendisk v4

switch blk-cgroup to work on gendisk v4

[Christoph Hellwig]

Hi all,

blk-cgroup works on only on live disks and "file system" I/O from bios.
This all the information should be in the gendisk, and not the
request_queue that also exists for pure passthrough request based
devices.

Changes since v3:
 - improve a commit log
 - drop a change to not acquire a pointless disk reference

Changes since v2:
 - drop the patch to revert the async blkg freeing
 - simplify error unwinding when failing to allocate a blkg
 - add back the dead queue/disk check in blkg_alloc
 - split two patches to better document the changes
 - add another blk-wbt cleanup patch
 - rewrite a commit log
 - another typo fix

Changes since v1:
 - use the local disk variable in wbt_init instead of q->disk
 - various spelling fixes

Diffstat:
 block/bfq-cgroup.c        |   18 +--
 block/bfq-iosched.c       |    6 -
 block/blk-cgroup-rwstat.c |    2 
 block/blk-cgroup.c        |  226 +++++++++++++++++++++-------------------------
 block/blk-cgroup.h        |   38 +++----
 block/blk-iocost.c        |   40 +++-----
 block/blk-iolatency.c     |   41 +++-----
 block/blk-ioprio.c        |    6 -
 block/blk-mq-debugfs.c    |   10 --
 block/blk-rq-qos.c        |   67 +++++++++++++
 block/blk-rq-qos.h        |   66 +------------
 block/blk-settings.c      |    1 
 block/blk-stat.c          |    3 
 block/blk-sysfs.c         |    5 -
 block/blk-throttle.c      |   31 +++---
 block/blk-wbt.c           |  116 +++++++++++++++++++----
 block/blk-wbt.h           |   98 +------------------
 block/genhd.c             |   17 +--
 include/linux/blkdev.h    |   12 +-
 include/linux/sched.h     |    2 
 kernel/fork.c             |    2 
 mm/swapfile.c             |    2 
 22 files changed, 391 insertions(+), 418 deletions(-)

switch blk-cgroup to work on gendisk v4

[Christoph Hellwig]

Hi all,

blk-cgroup works on only on live disks and "file system" I/O from bios.
This all the information should be in the gendisk, and not the
request_queue that also exists for pure passthrough request based
devices.

Changes since v3:
 - improve a commit log
 - drop a change to not acquire a pointless disk reference

Changes since v2:
 - drop the patch to revert the async blkg freeing
 - simplify error unwinding when failing to allocate a blkg
 - add back the dead queue/disk check in blkg_alloc
 - split two patches to better document the changes
 - add another blk-wbt cleanup patch
 - rewrite a commit log
 - another typo fix

Changes since v1:
 - use the local disk variable in wbt_init instead of q->disk
 - various spelling fixes

Diffstat:
 block/bfq-cgroup.c        |   18 +--
 block/bfq-iosched.c       |    6 -
 block/blk-cgroup-rwstat.c |    2 
 block/blk-cgroup.c        |  226 +++++++++++++++++++++-------------------------
 block/blk-cgroup.h        |   38 +++----
 block/blk-iocost.c        |   40 +++-----
 block/blk-iolatency.c     |   41 +++-----
 block/blk-ioprio.c        |    6 -
 block/blk-mq-debugfs.c    |   10 --
 block/blk-rq-qos.c        |   67 +++++++++++++
 block/blk-rq-qos.h        |   66 +------------
 block/blk-settings.c      |    1 
 block/blk-stat.c          |    3 
 block/blk-sysfs.c         |    5 -
 block/blk-throttle.c      |   31 +++---
 block/blk-wbt.c           |  116 +++++++++++++++++++----
 block/blk-wbt.h           |   98 +------------------
 block/genhd.c             |   17 +--
 include/linux/blkdev.h    |   12 +-
 include/linux/sched.h     |    2 
 kernel/fork.c             |    2 
 mm/swapfile.c             |    2 
 22 files changed, 391 insertions(+), 418 deletions(-)

[PATCH 01/19] block: don't call blk_throtl_stat_add for non-READ/WRITE commands

[Christoph Hellwig]

blk_throtl_stat_add is called from blk_stat_add explicitly, unlike the
other stats that go through q->stats->callbacks.  To prepare for cgroup
data moving to the gendisk, ensure blk_throtl_stat_add is only called
for the plain READ and WRITE commands that it actually handles internally,
as blk_stat_add can also be called for passthrough commands on queues that
do not have a gendisk associated with them.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/blk-stat.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/block/blk-stat.c b/block/blk-stat.c
index 2ea01b5c1aca04..c6ca16abf911e2 100644
--- a/block/blk-stat.c
+++ b/block/blk-stat.c
@@ -58,7 +58,8 @@ void blk_stat_add(struct request *rq, u64 now)
 
 	value = (now >= rq->io_start_time_ns) ? now - rq->io_start_time_ns : 0;
 
-	blk_throtl_stat_add(rq, value);
+	if (req_op(rq) == REQ_OP_READ || req_op(rq) == REQ_OP_WRITE)
+		blk_throtl_stat_add(rq, value);
 
 	rcu_read_lock();
 	cpu = get_cpu();
-- 
2.39.0

[PATCH 01/19] block: don't call blk_throtl_stat_add for non-READ/WRITE commands

[Christoph Hellwig]

blk_throtl_stat_add is called from blk_stat_add explicitly, unlike the
other stats that go through q->stats->callbacks.  To prepare for cgroup
data moving to the gendisk, ensure blk_throtl_stat_add is only called
for the plain READ and WRITE commands that it actually handles internally,
as blk_stat_add can also be called for passthrough commands on queues that
do not have a gendisk associated with them.

Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
Reviewed-by: Andreas Herrmann <aherrmann-l3A5Bk7waGM@public.gmane.org>
Acked-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 block/blk-stat.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/block/blk-stat.c b/block/blk-stat.c
index 2ea01b5c1aca04..c6ca16abf911e2 100644
--- a/block/blk-stat.c
+++ b/block/blk-stat.c
@@ -58,7 +58,8 @@ void blk_stat_add(struct request *rq, u64 now)
 
 	value = (now >= rq->io_start_time_ns) ? now - rq->io_start_time_ns : 0;
 
-	blk_throtl_stat_add(rq, value);
+	if (req_op(rq) == REQ_OP_READ || req_op(rq) == REQ_OP_WRITE)
+		blk_throtl_stat_add(rq, value);
 
 	rcu_read_lock();
 	cpu = get_cpu();
-- 
2.39.0

[PATCH 02/19] blk-cgroup: delay blk-cgroup initialization until add_disk

[Christoph Hellwig]

There is no need to initialize the cgroup code before the disk is marked
live.  Moving the cgroup initialization earlier will help to have a
fully initialized struct device in the gendisk for the cgroup code to
use in the future.  Similarly tear the cgroup information down in
del_gendisk to be symmetric and because none of the cgroup tracking is
needed once non-passthrough I/O stops.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/genhd.c | 17 +++++++++--------
 1 file changed, 9 insertions(+), 8 deletions(-)

diff --git a/block/genhd.c b/block/genhd.c
index 23cf83b3331cde..705dec0800d62e 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -466,10 +466,14 @@ int __must_check device_add_disk(struct device *parent, struct gendisk *disk,
 	 */
 	pm_runtime_set_memalloc_noio(ddev, true);
 
-	ret = blk_integrity_add(disk);
+	ret = blkcg_init_disk(disk);
 	if (ret)
 		goto out_del_block_link;
 
+	ret = blk_integrity_add(disk);
+	if (ret)
+		goto out_blkcg_exit;
+
 	disk->part0->bd_holder_dir =
 		kobject_create_and_add("holders", &ddev->kobj);
 	if (!disk->part0->bd_holder_dir) {
@@ -534,6 +538,8 @@ int __must_check device_add_disk(struct device *parent, struct gendisk *disk,
 	kobject_put(disk->part0->bd_holder_dir);
 out_del_integrity:
 	blk_integrity_del(disk);
+out_blkcg_exit:
+	blkcg_exit_disk(disk);
 out_del_block_link:
 	if (!sysfs_deprecated)
 		sysfs_remove_link(block_depr, dev_name(ddev));
@@ -662,6 +668,8 @@ void del_gendisk(struct gendisk *disk)
 	rq_qos_exit(q);
 	blk_mq_unquiesce_queue(q);
 
+	blkcg_exit_disk(disk);
+
 	/*
 	 * If the disk does not own the queue, allow using passthrough requests
 	 * again.  Else leave the queue frozen to fail all I/O.
@@ -1171,8 +1179,6 @@ static void disk_release(struct device *dev)
 	    !test_bit(GD_ADDED, &disk->state))
 		blk_mq_exit_queue(disk->queue);
 
-	blkcg_exit_disk(disk);
-
 	bioset_exit(&disk->bio_split);
 
 	disk_release_events(disk);
@@ -1385,9 +1391,6 @@ struct gendisk *__alloc_disk_node(struct request_queue *q, int node_id,
 	if (xa_insert(&disk->part_tbl, 0, disk->part0, GFP_KERNEL))
 		goto out_destroy_part_tbl;
 
-	if (blkcg_init_disk(disk))
-		goto out_erase_part0;
-
 	rand_initialize_disk(disk);
 	disk_to_dev(disk)->class = &block_class;
 	disk_to_dev(disk)->type = &disk_type;
@@ -1400,8 +1403,6 @@ struct gendisk *__alloc_disk_node(struct request_queue *q, int node_id,
 #endif
 	return disk;
 
-out_erase_part0:
-	xa_erase(&disk->part_tbl, 0);
 out_destroy_part_tbl:
 	xa_destroy(&disk->part_tbl);
 	disk->part0->bd_disk = NULL;
-- 
2.39.0

[PATCH 02/19] blk-cgroup: delay blk-cgroup initialization until add_disk

[Christoph Hellwig]

There is no need to initialize the cgroup code before the disk is marked
live.  Moving the cgroup initialization earlier will help to have a
fully initialized struct device in the gendisk for the cgroup code to
use in the future.  Similarly tear the cgroup information down in
del_gendisk to be symmetric and because none of the cgroup tracking is
needed once non-passthrough I/O stops.

Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
Reviewed-by: Andreas Herrmann <aherrmann-l3A5Bk7waGM@public.gmane.org>
Acked-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 block/genhd.c | 17 +++++++++--------
 1 file changed, 9 insertions(+), 8 deletions(-)

diff --git a/block/genhd.c b/block/genhd.c
index 23cf83b3331cde..705dec0800d62e 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -466,10 +466,14 @@ int __must_check device_add_disk(struct device *parent, struct gendisk *disk,
 	 */
 	pm_runtime_set_memalloc_noio(ddev, true);
 
-	ret = blk_integrity_add(disk);
+	ret = blkcg_init_disk(disk);
 	if (ret)
 		goto out_del_block_link;
 
+	ret = blk_integrity_add(disk);
+	if (ret)
+		goto out_blkcg_exit;
+
 	disk->part0->bd_holder_dir =
 		kobject_create_and_add("holders", &ddev->kobj);
 	if (!disk->part0->bd_holder_dir) {
@@ -534,6 +538,8 @@ int __must_check device_add_disk(struct device *parent, struct gendisk *disk,
 	kobject_put(disk->part0->bd_holder_dir);
 out_del_integrity:
 	blk_integrity_del(disk);
+out_blkcg_exit:
+	blkcg_exit_disk(disk);
 out_del_block_link:
 	if (!sysfs_deprecated)
 		sysfs_remove_link(block_depr, dev_name(ddev));
@@ -662,6 +668,8 @@ void del_gendisk(struct gendisk *disk)
 	rq_qos_exit(q);
 	blk_mq_unquiesce_queue(q);
 
+	blkcg_exit_disk(disk);
+
 	/*
 	 * If the disk does not own the queue, allow using passthrough requests
 	 * again.  Else leave the queue frozen to fail all I/O.
@@ -1171,8 +1179,6 @@ static void disk_release(struct device *dev)
 	    !test_bit(GD_ADDED, &disk->state))
 		blk_mq_exit_queue(disk->queue);
 
-	blkcg_exit_disk(disk);
-
 	bioset_exit(&disk->bio_split);
 
 	disk_release_events(disk);
@@ -1385,9 +1391,6 @@ struct gendisk *__alloc_disk_node(struct request_queue *q, int node_id,
 	if (xa_insert(&disk->part_tbl, 0, disk->part0, GFP_KERNEL))
 		goto out_destroy_part_tbl;
 
-	if (blkcg_init_disk(disk))
-		goto out_erase_part0;
-
 	rand_initialize_disk(disk);
 	disk_to_dev(disk)->class = &block_class;
 	disk_to_dev(disk)->type = &disk_type;
@@ -1400,8 +1403,6 @@ struct gendisk *__alloc_disk_node(struct request_queue *q, int node_id,
 #endif
 	return disk;
 
-out_erase_part0:
-	xa_erase(&disk->part_tbl, 0);
 out_destroy_part_tbl:
 	xa_destroy(&disk->part_tbl);
 	disk->part0->bd_disk = NULL;
-- 
2.39.0

[PATCH 03/19] blk-cgroup: improve error unwinding in blkg_alloc

[Christoph Hellwig]

Unwind only the previous initialization steps that happened in blkg_alloc
using goto based unwinding.  This avoids the need for the !queue special
case in blkg_free and thus ensures that any blkg seens outside of
blkg_alloc is always fully constructed.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/blk-cgroup.c | 39 +++++++++++++++++++--------------------
 1 file changed, 19 insertions(+), 20 deletions(-)

diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index cb110fc51940aa..9df02a6d04d35a 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -128,22 +128,16 @@ static void blkg_free_workfn(struct work_struct *work)
 	 * blkcg_mutex is used to synchronize blkg_free_workfn() and
 	 * blkcg_deactivate_policy().
 	 */
-	if (q)
-		mutex_lock(&q->blkcg_mutex);
-
+	mutex_lock(&q->blkcg_mutex);
 	for (i = 0; i < BLKCG_MAX_POLS; i++)
 		if (blkg->pd[i])
 			blkcg_policy[i]->pd_free_fn(blkg->pd[i]);
-
 	if (blkg->parent)
 		blkg_put(blkg->parent);
+	list_del_init(&blkg->q_node);
+	mutex_unlock(&q->blkcg_mutex);
 
-	if (q) {
-		list_del_init(&blkg->q_node);
-		mutex_unlock(&q->blkcg_mutex);
-		blk_put_queue(q);
-	}
-
+	blk_put_queue(q);
 	free_percpu(blkg->iostat_cpu);
 	percpu_ref_exit(&blkg->refcnt);
 	kfree(blkg);
@@ -265,16 +259,13 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct gendisk *disk,
 	blkg = kzalloc_node(sizeof(*blkg), gfp_mask, disk->queue->node);
 	if (!blkg)
 		return NULL;
-
 	if (percpu_ref_init(&blkg->refcnt, blkg_release, 0, gfp_mask))
-		goto err_free;
-
+		goto out_free_blkg;
 	blkg->iostat_cpu = alloc_percpu_gfp(struct blkg_iostat_set, gfp_mask);
 	if (!blkg->iostat_cpu)
-		goto err_free;
-
+		goto out_exit_refcnt;
 	if (!blk_get_queue(disk->queue))
-		goto err_free;
+		goto out_free_iostat;
 
 	blkg->q = disk->queue;
 	INIT_LIST_HEAD(&blkg->q_node);
@@ -299,8 +290,7 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct gendisk *disk,
 		/* alloc per-policy data and attach it to blkg */
 		pd = pol->pd_alloc_fn(gfp_mask, disk->queue, blkcg);
 		if (!pd)
-			goto err_free;
-
+			goto out_free_pds;
 		blkg->pd[i] = pd;
 		pd->blkg = blkg;
 		pd->plid = i;
@@ -309,8 +299,17 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct gendisk *disk,
 
 	return blkg;
 
-err_free:
-	blkg_free(blkg);
+out_free_pds:
+	while (--i >= 0)
+		if (blkg->pd[i])
+			blkcg_policy[i]->pd_free_fn(blkg->pd[i]);
+	blk_put_queue(disk->queue);
+out_free_iostat:
+	free_percpu(blkg->iostat_cpu);
+out_exit_refcnt:
+	percpu_ref_exit(&blkg->refcnt);
+out_free_blkg:
+	kfree(blkg);
 	return NULL;
 }
 
-- 
2.39.0

[PATCH 03/19] blk-cgroup: improve error unwinding in blkg_alloc

[Christoph Hellwig]

Unwind only the previous initialization steps that happened in blkg_alloc
using goto based unwinding.  This avoids the need for the !queue special
case in blkg_free and thus ensures that any blkg seens outside of
blkg_alloc is always fully constructed.

Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
Acked-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 block/blk-cgroup.c | 39 +++++++++++++++++++--------------------
 1 file changed, 19 insertions(+), 20 deletions(-)

diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index cb110fc51940aa..9df02a6d04d35a 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -128,22 +128,16 @@ static void blkg_free_workfn(struct work_struct *work)
 	 * blkcg_mutex is used to synchronize blkg_free_workfn() and
 	 * blkcg_deactivate_policy().
 	 */
-	if (q)
-		mutex_lock(&q->blkcg_mutex);
-
+	mutex_lock(&q->blkcg_mutex);
 	for (i = 0; i < BLKCG_MAX_POLS; i++)
 		if (blkg->pd[i])
 			blkcg_policy[i]->pd_free_fn(blkg->pd[i]);
-
 	if (blkg->parent)
 		blkg_put(blkg->parent);
+	list_del_init(&blkg->q_node);
+	mutex_unlock(&q->blkcg_mutex);
 
-	if (q) {
-		list_del_init(&blkg->q_node);
-		mutex_unlock(&q->blkcg_mutex);
-		blk_put_queue(q);
-	}
-
+	blk_put_queue(q);
 	free_percpu(blkg->iostat_cpu);
 	percpu_ref_exit(&blkg->refcnt);
 	kfree(blkg);
@@ -265,16 +259,13 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct gendisk *disk,
 	blkg = kzalloc_node(sizeof(*blkg), gfp_mask, disk->queue->node);
 	if (!blkg)
 		return NULL;
-
 	if (percpu_ref_init(&blkg->refcnt, blkg_release, 0, gfp_mask))
-		goto err_free;
-
+		goto out_free_blkg;
 	blkg->iostat_cpu = alloc_percpu_gfp(struct blkg_iostat_set, gfp_mask);
 	if (!blkg->iostat_cpu)
-		goto err_free;
-
+		goto out_exit_refcnt;
 	if (!blk_get_queue(disk->queue))
-		goto err_free;
+		goto out_free_iostat;
 
 	blkg->q = disk->queue;
 	INIT_LIST_HEAD(&blkg->q_node);
@@ -299,8 +290,7 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct gendisk *disk,
 		/* alloc per-policy data and attach it to blkg */
 		pd = pol->pd_alloc_fn(gfp_mask, disk->queue, blkcg);
 		if (!pd)
-			goto err_free;
-
+			goto out_free_pds;
 		blkg->pd[i] = pd;
 		pd->blkg = blkg;
 		pd->plid = i;
@@ -309,8 +299,17 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct gendisk *disk,
 
 	return blkg;
 
-err_free:
-	blkg_free(blkg);
+out_free_pds:
+	while (--i >= 0)
+		if (blkg->pd[i])
+			blkcg_policy[i]->pd_free_fn(blkg->pd[i]);
+	blk_put_queue(disk->queue);
+out_free_iostat:
+	free_percpu(blkg->iostat_cpu);
+out_exit_refcnt:
+	percpu_ref_exit(&blkg->refcnt);
+out_free_blkg:
+	kfree(blkg);
 	return NULL;
 }
 
-- 
2.39.0

[PATCH 04/19] blk-cgroup: simplify blkg freeing from initialization failure paths

[Christoph Hellwig]

There is no need to delay freeing a blkg to a workqueue when freeing it
after an initialization failure.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/blk-cgroup.c | 27 +++++++--------------------
 1 file changed, 7 insertions(+), 20 deletions(-)

diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 9df02a6d04d35a..1038688568926e 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -114,10 +114,8 @@ static bool blkcg_policy_enabled(struct request_queue *q,
 	return pol && test_bit(pol->plid, q->blkcg_pols);
 }
 
-static void blkg_free_workfn(struct work_struct *work)
+static void blkg_free(struct blkcg_gq *blkg)
 {
-	struct blkcg_gq *blkg = container_of(work, struct blkcg_gq,
-					     free_work);
 	struct request_queue *q = blkg->q;
 	int i;
 
@@ -143,23 +141,9 @@ static void blkg_free_workfn(struct work_struct *work)
 	kfree(blkg);
 }
 
-/**
- * blkg_free - free a blkg
- * @blkg: blkg to free
- *
- * Free @blkg which may be partially allocated.
- */
-static void blkg_free(struct blkcg_gq *blkg)
+static void blkg_free_workfn(struct work_struct *work)
 {
-	if (!blkg)
-		return;
-
-	/*
-	 * Both ->pd_free_fn() and request queue's release handler may
-	 * sleep, so free us by scheduling one work func
-	 */
-	INIT_WORK(&blkg->free_work, blkg_free_workfn);
-	schedule_work(&blkg->free_work);
+	blkg_free(container_of(work, struct blkcg_gq, free_work));
 }
 
 static void __blkg_release(struct rcu_head *rcu)
@@ -170,7 +154,10 @@ static void __blkg_release(struct rcu_head *rcu)
 
 	/* release the blkcg and parent blkg refs this blkg has been holding */
 	css_put(&blkg->blkcg->css);
-	blkg_free(blkg);
+
+	/* ->pd_free_fn() may sleep, so free from a work queue */
+	INIT_WORK(&blkg->free_work, blkg_free_workfn);
+	schedule_work(&blkg->free_work);
 }
 
 /*
-- 
2.39.0

[PATCH 04/19] blk-cgroup: simplify blkg freeing from initialization failure paths

[Christoph Hellwig]

There is no need to delay freeing a blkg to a workqueue when freeing it
after an initialization failure.

Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
Acked-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 block/blk-cgroup.c | 27 +++++++--------------------
 1 file changed, 7 insertions(+), 20 deletions(-)

diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 9df02a6d04d35a..1038688568926e 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -114,10 +114,8 @@ static bool blkcg_policy_enabled(struct request_queue *q,
 	return pol && test_bit(pol->plid, q->blkcg_pols);
 }
 
-static void blkg_free_workfn(struct work_struct *work)
+static void blkg_free(struct blkcg_gq *blkg)
 {
-	struct blkcg_gq *blkg = container_of(work, struct blkcg_gq,
-					     free_work);
 	struct request_queue *q = blkg->q;
 	int i;
 
@@ -143,23 +141,9 @@ static void blkg_free_workfn(struct work_struct *work)
 	kfree(blkg);
 }
 
-/**
- * blkg_free - free a blkg
- * @blkg: blkg to free
- *
- * Free @blkg which may be partially allocated.
- */
-static void blkg_free(struct blkcg_gq *blkg)
+static void blkg_free_workfn(struct work_struct *work)
 {
-	if (!blkg)
-		return;
-
-	/*
-	 * Both ->pd_free_fn() and request queue's release handler may
-	 * sleep, so free us by scheduling one work func
-	 */
-	INIT_WORK(&blkg->free_work, blkg_free_workfn);
-	schedule_work(&blkg->free_work);
+	blkg_free(container_of(work, struct blkcg_gq, free_work));
 }
 
 static void __blkg_release(struct rcu_head *rcu)
@@ -170,7 +154,10 @@ static void __blkg_release(struct rcu_head *rcu)
 
 	/* release the blkcg and parent blkg refs this blkg has been holding */
 	css_put(&blkg->blkcg->css);
-	blkg_free(blkg);
+
+	/* ->pd_free_fn() may sleep, so free from a work queue */
+	INIT_WORK(&blkg->free_work, blkg_free_workfn);
+	schedule_work(&blkg->free_work);
 }
 
 /*
-- 
2.39.0

[PATCH 05/19] blk-cgroup: remove the !bdi->dev check in blkg_dev_name

[Christoph Hellwig]

bdi_dev_name already performs the same check.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/blk-cgroup.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 1038688568926e..0b3226cbf3f25d 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -572,7 +572,7 @@ static int blkcg_reset_stats(struct cgroup_subsys_state *css,
 
 const char *blkg_dev_name(struct blkcg_gq *blkg)
 {
-	if (!blkg->q->disk || !blkg->q->disk->bdi->dev)
+	if (!blkg->q->disk)
 		return NULL;
 	return bdi_dev_name(blkg->q->disk->bdi);
 }
-- 
2.39.0

[PATCH 05/19] blk-cgroup: remove the !bdi->dev check in blkg_dev_name

[Christoph Hellwig]

bdi_dev_name already performs the same check.

Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
Acked-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 block/blk-cgroup.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 1038688568926e..0b3226cbf3f25d 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -572,7 +572,7 @@ static int blkcg_reset_stats(struct cgroup_subsys_state *css,
 
 const char *blkg_dev_name(struct blkcg_gq *blkg)
 {
-	if (!blkg->q->disk || !blkg->q->disk->bdi->dev)
+	if (!blkg->q->disk)
 		return NULL;
 	return bdi_dev_name(blkg->q->disk->bdi);
 }
-- 
2.39.0

[PATCH 06/19] blk-cgroup: pin the gendisk in struct blkcg_gq

[Christoph Hellwig]

Currently each blkcg_gq holds a request_queue reference, which is what
is used in the policies.  But a lot of these interfaces will move over to
use a gendisk, so store a disk in struct blkcg_gq and hold a reference to
it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/bfq-cgroup.c        |  6 +++---
 block/blk-cgroup-rwstat.c |  2 +-
 block/blk-cgroup.c        | 35 +++++++++++++++++------------------
 block/blk-cgroup.h        | 11 +++++------
 block/blk-iocost.c        |  2 +-
 block/blk-iolatency.c     |  4 ++--
 block/blk-throttle.c      |  4 ++--
 7 files changed, 31 insertions(+), 33 deletions(-)

diff --git a/block/bfq-cgroup.c b/block/bfq-cgroup.c
index b42956ab555068..1d4a3f15049bda 100644
--- a/block/bfq-cgroup.c
+++ b/block/bfq-cgroup.c
@@ -405,7 +405,7 @@ static void bfqg_stats_xfer_dead(struct bfq_group *bfqg)
 
 	parent = bfqg_parent(bfqg);
 
-	lockdep_assert_held(&bfqg_to_blkg(bfqg)->q->queue_lock);
+	lockdep_assert_held(&bfqg_to_blkg(bfqg)->disk->queue->queue_lock);
 
 	if (unlikely(!parent))
 		return;
@@ -536,7 +536,7 @@ static void bfq_pd_init(struct blkg_policy_data *pd)
 {
 	struct blkcg_gq *blkg = pd_to_blkg(pd);
 	struct bfq_group *bfqg = blkg_to_bfqg(blkg);
-	struct bfq_data *bfqd = blkg->q->elevator->elevator_data;
+	struct bfq_data *bfqd = blkg->disk->queue->elevator->elevator_data;
 	struct bfq_entity *entity = &bfqg->entity;
 	struct bfq_group_data *d = blkcg_to_bfqgd(blkg->blkcg);
 
@@ -1201,7 +1201,7 @@ static u64 bfqg_prfill_stat_recursive(struct seq_file *sf,
 	struct cgroup_subsys_state *pos_css;
 	u64 sum = 0;
 
-	lockdep_assert_held(&blkg->q->queue_lock);
+	lockdep_assert_held(&blkg->disk->queue->queue_lock);
 
 	rcu_read_lock();
 	blkg_for_each_descendant_pre(pos_blkg, pos_css, blkg) {
diff --git a/block/blk-cgroup-rwstat.c b/block/blk-cgroup-rwstat.c
index 3304e841df7ce9..b8b8c82e667a3b 100644
--- a/block/blk-cgroup-rwstat.c
+++ b/block/blk-cgroup-rwstat.c
@@ -107,7 +107,7 @@ void blkg_rwstat_recursive_sum(struct blkcg_gq *blkg, struct blkcg_policy *pol,
 	struct cgroup_subsys_state *pos_css;
 	unsigned int i;
 
-	lockdep_assert_held(&blkg->q->queue_lock);
+	lockdep_assert_held(&blkg->disk->queue->queue_lock);
 
 	memset(sum, 0, sizeof(*sum));
 	rcu_read_lock();
diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 0b3226cbf3f25d..0e368387497d27 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -116,7 +116,6 @@ static bool blkcg_policy_enabled(struct request_queue *q,
 
 static void blkg_free(struct blkcg_gq *blkg)
 {
-	struct request_queue *q = blkg->q;
 	int i;
 
 	/*
@@ -126,16 +125,16 @@ static void blkg_free(struct blkcg_gq *blkg)
 	 * blkcg_mutex is used to synchronize blkg_free_workfn() and
 	 * blkcg_deactivate_policy().
 	 */
-	mutex_lock(&q->blkcg_mutex);
+	mutex_lock(&blkg->disk->queue->blkcg_mutex);
 	for (i = 0; i < BLKCG_MAX_POLS; i++)
 		if (blkg->pd[i])
 			blkcg_policy[i]->pd_free_fn(blkg->pd[i]);
 	if (blkg->parent)
 		blkg_put(blkg->parent);
 	list_del_init(&blkg->q_node);
-	mutex_unlock(&q->blkcg_mutex);
+	mutex_unlock(&blkg->disk->queue->blkcg_mutex);
 
-	blk_put_queue(q);
+	put_disk(blkg->disk);
 	free_percpu(blkg->iostat_cpu);
 	percpu_ref_exit(&blkg->refcnt);
 	kfree(blkg);
@@ -251,10 +250,12 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct gendisk *disk,
 	blkg->iostat_cpu = alloc_percpu_gfp(struct blkg_iostat_set, gfp_mask);
 	if (!blkg->iostat_cpu)
 		goto out_exit_refcnt;
-	if (!blk_get_queue(disk->queue))
+
+	if (test_bit(GD_DEAD, &disk->state))
 		goto out_free_iostat;
+	get_device(disk_to_dev(disk));
+	blkg->disk = disk;
 
-	blkg->q = disk->queue;
 	INIT_LIST_HEAD(&blkg->q_node);
 	spin_lock_init(&blkg->async_bio_lock);
 	bio_list_init(&blkg->async_bios);
@@ -290,7 +291,7 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct gendisk *disk,
 	while (--i >= 0)
 		if (blkg->pd[i])
 			blkcg_policy[i]->pd_free_fn(blkg->pd[i]);
-	blk_put_queue(disk->queue);
+	put_disk(blkg->disk);
 out_free_iostat:
 	free_percpu(blkg->iostat_cpu);
 out_exit_refcnt:
@@ -461,7 +462,7 @@ static void blkg_destroy(struct blkcg_gq *blkg)
 	struct blkcg *blkcg = blkg->blkcg;
 	int i;
 
-	lockdep_assert_held(&blkg->q->queue_lock);
+	lockdep_assert_held(&blkg->disk->queue->queue_lock);
 	lockdep_assert_held(&blkcg->lock);
 
 	/*
@@ -485,7 +486,7 @@ static void blkg_destroy(struct blkcg_gq *blkg)
 
 	blkg->online = false;
 
-	radix_tree_delete(&blkcg->blkg_tree, blkg->q->id);
+	radix_tree_delete(&blkcg->blkg_tree, blkg->disk->queue->id);
 	hlist_del_init_rcu(&blkg->blkcg_node);
 
 	/*
@@ -572,9 +573,7 @@ static int blkcg_reset_stats(struct cgroup_subsys_state *css,
 
 const char *blkg_dev_name(struct blkcg_gq *blkg)
 {
-	if (!blkg->q->disk)
-		return NULL;
-	return bdi_dev_name(blkg->q->disk->bdi);
+	return bdi_dev_name(blkg->disk->bdi);
 }
 
 /**
@@ -606,10 +605,10 @@ void blkcg_print_blkgs(struct seq_file *sf, struct blkcg *blkcg,
 
 	rcu_read_lock();
 	hlist_for_each_entry_rcu(blkg, &blkcg->blkg_list, blkcg_node) {
-		spin_lock_irq(&blkg->q->queue_lock);
-		if (blkcg_policy_enabled(blkg->q, pol))
+		spin_lock_irq(&blkg->disk->queue->queue_lock);
+		if (blkcg_policy_enabled(blkg->disk->queue, pol))
 			total += prfill(sf, blkg->pd[pol->plid], data);
-		spin_unlock_irq(&blkg->q->queue_lock);
+		spin_unlock_irq(&blkg->disk->queue->queue_lock);
 	}
 	rcu_read_unlock();
 
@@ -1033,9 +1032,9 @@ static int blkcg_print_stat(struct seq_file *sf, void *v)
 
 	rcu_read_lock();
 	hlist_for_each_entry_rcu(blkg, &blkcg->blkg_list, blkcg_node) {
-		spin_lock_irq(&blkg->q->queue_lock);
+		spin_lock_irq(&blkg->disk->queue->queue_lock);
 		blkcg_print_one_stat(blkg, sf);
-		spin_unlock_irq(&blkg->q->queue_lock);
+		spin_unlock_irq(&blkg->disk->queue->queue_lock);
 	}
 	rcu_read_unlock();
 	return 0;
@@ -1105,7 +1104,7 @@ static void blkcg_destroy_blkgs(struct blkcg *blkcg)
 	while (!hlist_empty(&blkcg->blkg_list)) {
 		struct blkcg_gq *blkg = hlist_entry(blkcg->blkg_list.first,
 						struct blkcg_gq, blkcg_node);
-		struct request_queue *q = blkg->q;
+		struct request_queue *q = blkg->disk->queue;
 
 		if (need_resched() || !spin_trylock(&q->queue_lock)) {
 			/*
diff --git a/block/blk-cgroup.h b/block/blk-cgroup.h
index b13ee84f358e85..996572a9a0b72e 100644
--- a/block/blk-cgroup.h
+++ b/block/blk-cgroup.h
@@ -53,8 +53,7 @@ struct blkg_iostat_set {
 
 /* association between a blk cgroup and a request queue */
 struct blkcg_gq {
-	/* Pointer to the associated request_queue */
-	struct request_queue		*q;
+	struct gendisk			*disk;
 	struct list_head		q_node;
 	struct hlist_node		blkcg_node;
 	struct blkcg			*blkcg;
@@ -255,11 +254,11 @@ static inline struct blkcg_gq *blkg_lookup(struct blkcg *blkcg,
 		return q->root_blkg;
 
 	blkg = rcu_dereference(blkcg->blkg_hint);
-	if (blkg && blkg->q == q)
+	if (blkg && blkg->disk->queue == q)
 		return blkg;
 
 	blkg = radix_tree_lookup(&blkcg->blkg_tree, q->id);
-	if (blkg && blkg->q != q)
+	if (blkg && blkg->disk->queue != q)
 		blkg = NULL;
 	return blkg;
 }
@@ -359,7 +358,7 @@ static inline void blkg_put(struct blkcg_gq *blkg)
 #define blkg_for_each_descendant_pre(d_blkg, pos_css, p_blkg)		\
 	css_for_each_descendant_pre((pos_css), &(p_blkg)->blkcg->css)	\
 		if (((d_blkg) = blkg_lookup(css_to_blkcg(pos_css),	\
-					    (p_blkg)->q)))
+					    (p_blkg)->disk->queue)))
 
 /**
  * blkg_for_each_descendant_post - post-order walk of a blkg's descendants
@@ -374,7 +373,7 @@ static inline void blkg_put(struct blkcg_gq *blkg)
 #define blkg_for_each_descendant_post(d_blkg, pos_css, p_blkg)		\
 	css_for_each_descendant_post((pos_css), &(p_blkg)->blkcg->css)	\
 		if (((d_blkg) = blkg_lookup(css_to_blkcg(pos_css),	\
-					    (p_blkg)->q)))
+					    (p_blkg)->disk->queue)))
 
 bool __blkcg_punt_bio_submit(struct bio *bio);
 
diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index 9694394ed883c4..a2b4e7146be5f1 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -2953,7 +2953,7 @@ 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 *ioc = q_to_ioc(blkg->disk->queue);
 	struct ioc_now now;
 	struct blkcg_gq *tblkg;
 	unsigned long flags;
diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c
index ecdc1074183625..b55eac2cf91944 100644
--- a/block/blk-iolatency.c
+++ b/block/blk-iolatency.c
@@ -974,12 +974,12 @@ static void iolatency_pd_init(struct blkg_policy_data *pd)
 {
 	struct iolatency_grp *iolat = pd_to_lat(pd);
 	struct blkcg_gq *blkg = lat_to_blkg(iolat);
-	struct rq_qos *rqos = blkcg_rq_qos(blkg->q);
+	struct rq_qos *rqos = blkcg_rq_qos(blkg->disk->queue);
 	struct blk_iolatency *blkiolat = BLKIOLATENCY(rqos);
 	u64 now = ktime_to_ns(ktime_get());
 	int cpu;
 
-	if (blk_queue_nonrot(blkg->q))
+	if (blk_queue_nonrot(blkg->disk->queue))
 		iolat->ssd = true;
 	else
 		iolat->ssd = false;
diff --git a/block/blk-throttle.c b/block/blk-throttle.c
index 6fb5a2f9e1eed5..f802d8f9099430 100644
--- a/block/blk-throttle.c
+++ b/block/blk-throttle.c
@@ -388,7 +388,7 @@ static void throtl_pd_init(struct blkg_policy_data *pd)
 {
 	struct throtl_grp *tg = pd_to_tg(pd);
 	struct blkcg_gq *blkg = tg_to_blkg(tg);
-	struct throtl_data *td = blkg->q->td;
+	struct throtl_data *td = blkg->disk->queue->td;
 	struct throtl_service_queue *sq = &tg->service_queue;
 
 	/*
@@ -1175,7 +1175,7 @@ static void throtl_pending_timer_fn(struct timer_list *t)
 
 	/* throtl_data may be gone, so figure out request queue by blkg */
 	if (tg)
-		q = tg->pd.blkg->q;
+		q = tg->pd.blkg->disk->queue;
 	else
 		q = td->queue;
 
-- 
2.39.0

[PATCH 07/19] blk-cgroup: store a gendisk to throttle in struct task_struct

[Christoph Hellwig]

Switch from a request_queue pointer and reference to a gendisk once
for the throttle information in struct task_struct.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
---
 block/blk-cgroup.c    | 32 +++++++++++++++-----------------
 include/linux/sched.h |  2 +-
 kernel/fork.c         |  2 +-
 mm/swapfile.c         |  2 +-
 4 files changed, 18 insertions(+), 20 deletions(-)

diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 0e368387497d27..168b2f803238f9 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -1362,9 +1362,9 @@ static void blkcg_bind(struct cgroup_subsys_state *root_css)
 
 static void blkcg_exit(struct task_struct *tsk)
 {
-	if (tsk->throttle_queue)
-		blk_put_queue(tsk->throttle_queue);
-	tsk->throttle_queue = NULL;
+	if (tsk->throttle_disk)
+		put_disk(tsk->throttle_disk);
+	tsk->throttle_disk = NULL;
 }
 
 struct cgroup_subsys io_cgrp_subsys = {
@@ -1815,29 +1815,29 @@ static void blkcg_maybe_throttle_blkg(struct blkcg_gq *blkg, bool use_memdelay)
  *
  * This is only called if we've been marked with set_notify_resume().  Obviously
  * we can be set_notify_resume() for reasons other than blkcg throttling, so we
- * check to see if current->throttle_queue is set and if not this doesn't do
+ * check to see if current->throttle_disk is set and if not this doesn't do
  * anything.  This should only ever be called by the resume code, it's not meant
  * to be called by people willy-nilly as it will actually do the work to
  * throttle the task if it is setup for throttling.
  */
 void blkcg_maybe_throttle_current(void)
 {
-	struct request_queue *q = current->throttle_queue;
+	struct gendisk *disk = current->throttle_disk;
 	struct blkcg *blkcg;
 	struct blkcg_gq *blkg;
 	bool use_memdelay = current->use_memdelay;
 
-	if (!q)
+	if (!disk)
 		return;
 
-	current->throttle_queue = NULL;
+	current->throttle_disk = NULL;
 	current->use_memdelay = false;
 
 	rcu_read_lock();
 	blkcg = css_to_blkcg(blkcg_css());
 	if (!blkcg)
 		goto out;
-	blkg = blkg_lookup(blkcg, q);
+	blkg = blkg_lookup(blkcg, disk->queue);
 	if (!blkg)
 		goto out;
 	if (!blkg_tryget(blkg))
@@ -1846,11 +1846,10 @@ void blkcg_maybe_throttle_current(void)
 
 	blkcg_maybe_throttle_blkg(blkg, use_memdelay);
 	blkg_put(blkg);
-	blk_put_queue(q);
+	put_disk(disk);
 	return;
 out:
 	rcu_read_unlock();
-	blk_put_queue(q);
 }
 
 /**
@@ -1872,18 +1871,17 @@ void blkcg_maybe_throttle_current(void)
  */
 void blkcg_schedule_throttle(struct gendisk *disk, bool use_memdelay)
 {
-	struct request_queue *q = disk->queue;
-
 	if (unlikely(current->flags & PF_KTHREAD))
 		return;
 
-	if (current->throttle_queue != q) {
-		if (!blk_get_queue(q))
+	if (current->throttle_disk != disk) {
+		if (test_bit(GD_DEAD, &disk->state))
 			return;
+		get_device(disk_to_dev(disk));
 
-		if (current->throttle_queue)
-			blk_put_queue(current->throttle_queue);
-		current->throttle_queue = q;
+		if (current->throttle_disk)
+			put_disk(current->throttle_disk);
+		current->throttle_disk = disk;
 	}
 
 	if (use_memdelay)
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 853d08f7562bda..6f6ce9ca709798 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1436,7 +1436,7 @@ struct task_struct {
 #endif
 
 #ifdef CONFIG_BLK_CGROUP
-	struct request_queue		*throttle_queue;
+	struct gendisk			*throttle_disk;
 #endif
 
 #ifdef CONFIG_UPROBES
diff --git a/kernel/fork.c b/kernel/fork.c
index 9f7fe354189785..d9c97704b7c9a4 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1044,7 +1044,7 @@ static struct task_struct *dup_task_struct(struct task_struct *orig, int node)
 #endif
 
 #ifdef CONFIG_BLK_CGROUP
-	tsk->throttle_queue = NULL;
+	tsk->throttle_disk = NULL;
 	tsk->use_memdelay = 0;
 #endif
 
diff --git a/mm/swapfile.c b/mm/swapfile.c
index 908a529bca12c9..3e0a742fb7bbff 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -3642,7 +3642,7 @@ void __cgroup_throttle_swaprate(struct page *page, gfp_t gfp_mask)
 	 * We've already scheduled a throttle, avoid taking the global swap
 	 * lock.
 	 */
-	if (current->throttle_queue)
+	if (current->throttle_disk)
 		return;
 
 	spin_lock(&swap_avail_lock);
-- 
2.39.0

[PATCH 07/19] blk-cgroup: store a gendisk to throttle in struct task_struct

[Christoph Hellwig]

Switch from a request_queue pointer and reference to a gendisk once
for the throttle information in struct task_struct.

Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
Reviewed-by: Andreas Herrmann <aherrmann-l3A5Bk7waGM@public.gmane.org>
---
 block/blk-cgroup.c    | 32 +++++++++++++++-----------------
 include/linux/sched.h |  2 +-
 kernel/fork.c         |  2 +-
 mm/swapfile.c         |  2 +-
 4 files changed, 18 insertions(+), 20 deletions(-)

diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 0e368387497d27..168b2f803238f9 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -1362,9 +1362,9 @@ static void blkcg_bind(struct cgroup_subsys_state *root_css)
 
 static void blkcg_exit(struct task_struct *tsk)
 {
-	if (tsk->throttle_queue)
-		blk_put_queue(tsk->throttle_queue);
-	tsk->throttle_queue = NULL;
+	if (tsk->throttle_disk)
+		put_disk(tsk->throttle_disk);
+	tsk->throttle_disk = NULL;
 }
 
 struct cgroup_subsys io_cgrp_subsys = {
@@ -1815,29 +1815,29 @@ static void blkcg_maybe_throttle_blkg(struct blkcg_gq *blkg, bool use_memdelay)
  *
  * This is only called if we've been marked with set_notify_resume().  Obviously
  * we can be set_notify_resume() for reasons other than blkcg throttling, so we
- * check to see if current->throttle_queue is set and if not this doesn't do
+ * check to see if current->throttle_disk is set and if not this doesn't do
  * anything.  This should only ever be called by the resume code, it's not meant
  * to be called by people willy-nilly as it will actually do the work to
  * throttle the task if it is setup for throttling.
  */
 void blkcg_maybe_throttle_current(void)
 {
-	struct request_queue *q = current->throttle_queue;
+	struct gendisk *disk = current->throttle_disk;
 	struct blkcg *blkcg;
 	struct blkcg_gq *blkg;
 	bool use_memdelay = current->use_memdelay;
 
-	if (!q)
+	if (!disk)
 		return;
 
-	current->throttle_queue = NULL;
+	current->throttle_disk = NULL;
 	current->use_memdelay = false;
 
 	rcu_read_lock();
 	blkcg = css_to_blkcg(blkcg_css());
 	if (!blkcg)
 		goto out;
-	blkg = blkg_lookup(blkcg, q);
+	blkg = blkg_lookup(blkcg, disk->queue);
 	if (!blkg)
 		goto out;
 	if (!blkg_tryget(blkg))
@@ -1846,11 +1846,10 @@ void blkcg_maybe_throttle_current(void)
 
 	blkcg_maybe_throttle_blkg(blkg, use_memdelay);
 	blkg_put(blkg);
-	blk_put_queue(q);
+	put_disk(disk);
 	return;
 out:
 	rcu_read_unlock();
-	blk_put_queue(q);
 }
 
 /**
@@ -1872,18 +1871,17 @@ void blkcg_maybe_throttle_current(void)
  */
 void blkcg_schedule_throttle(struct gendisk *disk, bool use_memdelay)
 {
-	struct request_queue *q = disk->queue;
-
 	if (unlikely(current->flags & PF_KTHREAD))
 		return;
 
-	if (current->throttle_queue != q) {
-		if (!blk_get_queue(q))
+	if (current->throttle_disk != disk) {
+		if (test_bit(GD_DEAD, &disk->state))
 			return;
+		get_device(disk_to_dev(disk));
 
-		if (current->throttle_queue)
-			blk_put_queue(current->throttle_queue);
-		current->throttle_queue = q;
+		if (current->throttle_disk)
+			put_disk(current->throttle_disk);
+		current->throttle_disk = disk;
 	}
 
 	if (use_memdelay)
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 853d08f7562bda..6f6ce9ca709798 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1436,7 +1436,7 @@ struct task_struct {
 #endif
 
 #ifdef CONFIG_BLK_CGROUP
-	struct request_queue		*throttle_queue;
+	struct gendisk			*throttle_disk;
 #endif
 
 #ifdef CONFIG_UPROBES
diff --git a/kernel/fork.c b/kernel/fork.c
index 9f7fe354189785..d9c97704b7c9a4 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1044,7 +1044,7 @@ static struct task_struct *dup_task_struct(struct task_struct *orig, int node)
 #endif
 
 #ifdef CONFIG_BLK_CGROUP
-	tsk->throttle_queue = NULL;
+	tsk->throttle_disk = NULL;
 	tsk->use_memdelay = 0;
 #endif
 
diff --git a/mm/swapfile.c b/mm/swapfile.c
index 908a529bca12c9..3e0a742fb7bbff 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -3642,7 +3642,7 @@ void __cgroup_throttle_swaprate(struct page *page, gfp_t gfp_mask)
 	 * We've already scheduled a throttle, avoid taking the global swap
 	 * lock.
 	 */
-	if (current->throttle_queue)
+	if (current->throttle_disk)
 		return;
 
 	spin_lock(&swap_avail_lock);
-- 
2.39.0

[PATCH 08/19] blk-wbt: pass a gendisk to wbt_{enable,disable}_default

[Christoph Hellwig]

Pass a gendisk to wbt_enable_default and wbt_disable_default to
prepare for phasing out usage of the request_queue in the blk-cgroup
code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/bfq-iosched.c | 4 ++--
 block/blk-iocost.c  | 4 ++--
 block/blk-sysfs.c   | 2 +-
 block/blk-wbt.c     | 7 ++++---
 block/blk-wbt.h     | 8 ++++----
 5 files changed, 13 insertions(+), 12 deletions(-)

diff --git a/block/bfq-iosched.c b/block/bfq-iosched.c
index 4705c4be90e721..5afa661fa2ea4c 100644
--- a/block/bfq-iosched.c
+++ b/block/bfq-iosched.c
@@ -7156,7 +7156,7 @@ static void bfq_exit_queue(struct elevator_queue *e)
 
 	blk_stat_disable_accounting(bfqd->queue);
 	clear_bit(ELEVATOR_FLAG_DISABLE_WBT, &e->flags);
-	wbt_enable_default(bfqd->queue);
+	wbt_enable_default(bfqd->queue->disk);
 
 	kfree(bfqd);
 }
@@ -7344,7 +7344,7 @@ static int bfq_init_queue(struct request_queue *q, struct elevator_type *e)
 	blk_queue_flag_set(QUEUE_FLAG_SQ_SCHED, q);
 
 	set_bit(ELEVATOR_FLAG_DISABLE_WBT, &eq->flags);
-	wbt_disable_default(q);
+	wbt_disable_default(q->disk);
 	blk_stat_enable_accounting(q);
 
 	return 0;
diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index a2b4e7146be5f1..dbb93f4f68d979 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -3285,11 +3285,11 @@ static ssize_t ioc_qos_write(struct kernfs_open_file *of, char *input,
 		blk_stat_enable_accounting(disk->queue);
 		blk_queue_flag_set(QUEUE_FLAG_RQ_ALLOC_TIME, disk->queue);
 		ioc->enabled = true;
-		wbt_disable_default(disk->queue);
+		wbt_disable_default(disk);
 	} else {
 		blk_queue_flag_clear(QUEUE_FLAG_RQ_ALLOC_TIME, disk->queue);
 		ioc->enabled = false;
-		wbt_enable_default(disk->queue);
+		wbt_enable_default(disk);
 	}
 
 	if (user) {
diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index 5486b6c57f6b8a..2074103865f45b 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -826,7 +826,7 @@ int blk_register_queue(struct gendisk *disk)
 		goto out_elv_unregister;
 
 	blk_queue_flag_set(QUEUE_FLAG_REGISTERED, q);
-	wbt_enable_default(q);
+	wbt_enable_default(disk);
 	blk_throtl_register(disk);
 
 	/* Now everything is ready and send out KOBJ_ADD uevent */
diff --git a/block/blk-wbt.c b/block/blk-wbt.c
index 68a774d7a7c9c0..8f9302134339c5 100644
--- a/block/blk-wbt.c
+++ b/block/blk-wbt.c
@@ -650,8 +650,9 @@ void wbt_set_write_cache(struct request_queue *q, bool write_cache_on)
 /*
  * Enable wbt if defaults are configured that way
  */
-void wbt_enable_default(struct request_queue *q)
+void wbt_enable_default(struct gendisk *disk)
 {
+	struct request_queue *q = disk->queue;
 	struct rq_qos *rqos;
 	bool disable_flag = q->elevator &&
 		    test_bit(ELEVATOR_FLAG_DISABLE_WBT, &q->elevator->flags);
@@ -718,9 +719,9 @@ static void wbt_exit(struct rq_qos *rqos)
 /*
  * Disable wbt, if enabled by default.
  */
-void wbt_disable_default(struct request_queue *q)
+void wbt_disable_default(struct gendisk *disk)
 {
-	struct rq_qos *rqos = wbt_rq_qos(q);
+	struct rq_qos *rqos = wbt_rq_qos(disk->queue);
 	struct rq_wb *rwb;
 	if (!rqos)
 		return;
diff --git a/block/blk-wbt.h b/block/blk-wbt.h
index e3ea6e7e290076..7ab1cba55c25f7 100644
--- a/block/blk-wbt.h
+++ b/block/blk-wbt.h
@@ -91,8 +91,8 @@ static inline unsigned int wbt_inflight(struct rq_wb *rwb)
 #ifdef CONFIG_BLK_WBT
 
 int wbt_init(struct request_queue *);
-void wbt_disable_default(struct request_queue *);
-void wbt_enable_default(struct request_queue *);
+void wbt_disable_default(struct gendisk *disk);
+void wbt_enable_default(struct gendisk *disk);
 
 u64 wbt_get_min_lat(struct request_queue *q);
 void wbt_set_min_lat(struct request_queue *q, u64 val);
@@ -108,10 +108,10 @@ static inline int wbt_init(struct request_queue *q)
 {
 	return -EINVAL;
 }
-static inline void wbt_disable_default(struct request_queue *q)
+static inline void wbt_disable_default(struct gendisk *disk)
 {
 }
-static inline void wbt_enable_default(struct request_queue *q)
+static inline void wbt_enable_default(struct gendisk *disk)
 {
 }
 static inline void wbt_set_write_cache(struct request_queue *q, bool wc)
-- 
2.39.0

[PATCH 08/19] blk-wbt: pass a gendisk to wbt_{enable,disable}_default

[Christoph Hellwig]

Pass a gendisk to wbt_enable_default and wbt_disable_default to
prepare for phasing out usage of the request_queue in the blk-cgroup
code.

Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
Reviewed-by: Andreas Herrmann <aherrmann-l3A5Bk7waGM@public.gmane.org>
Acked-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 block/bfq-iosched.c | 4 ++--
 block/blk-iocost.c  | 4 ++--
 block/blk-sysfs.c   | 2 +-
 block/blk-wbt.c     | 7 ++++---
 block/blk-wbt.h     | 8 ++++----
 5 files changed, 13 insertions(+), 12 deletions(-)

diff --git a/block/bfq-iosched.c b/block/bfq-iosched.c
index 4705c4be90e721..5afa661fa2ea4c 100644
--- a/block/bfq-iosched.c
+++ b/block/bfq-iosched.c
@@ -7156,7 +7156,7 @@ static void bfq_exit_queue(struct elevator_queue *e)
 
 	blk_stat_disable_accounting(bfqd->queue);
 	clear_bit(ELEVATOR_FLAG_DISABLE_WBT, &e->flags);
-	wbt_enable_default(bfqd->queue);
+	wbt_enable_default(bfqd->queue->disk);
 
 	kfree(bfqd);
 }
@@ -7344,7 +7344,7 @@ static int bfq_init_queue(struct request_queue *q, struct elevator_type *e)
 	blk_queue_flag_set(QUEUE_FLAG_SQ_SCHED, q);
 
 	set_bit(ELEVATOR_FLAG_DISABLE_WBT, &eq->flags);
-	wbt_disable_default(q);
+	wbt_disable_default(q->disk);
 	blk_stat_enable_accounting(q);
 
 	return 0;
diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index a2b4e7146be5f1..dbb93f4f68d979 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -3285,11 +3285,11 @@ static ssize_t ioc_qos_write(struct kernfs_open_file *of, char *input,
 		blk_stat_enable_accounting(disk->queue);
 		blk_queue_flag_set(QUEUE_FLAG_RQ_ALLOC_TIME, disk->queue);
 		ioc->enabled = true;
-		wbt_disable_default(disk->queue);
+		wbt_disable_default(disk);
 	} else {
 		blk_queue_flag_clear(QUEUE_FLAG_RQ_ALLOC_TIME, disk->queue);
 		ioc->enabled = false;
-		wbt_enable_default(disk->queue);
+		wbt_enable_default(disk);
 	}
 
 	if (user) {
diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index 5486b6c57f6b8a..2074103865f45b 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -826,7 +826,7 @@ int blk_register_queue(struct gendisk *disk)
 		goto out_elv_unregister;
 
 	blk_queue_flag_set(QUEUE_FLAG_REGISTERED, q);
-	wbt_enable_default(q);
+	wbt_enable_default(disk);
 	blk_throtl_register(disk);
 
 	/* Now everything is ready and send out KOBJ_ADD uevent */
diff --git a/block/blk-wbt.c b/block/blk-wbt.c
index 68a774d7a7c9c0..8f9302134339c5 100644
--- a/block/blk-wbt.c
+++ b/block/blk-wbt.c
@@ -650,8 +650,9 @@ void wbt_set_write_cache(struct request_queue *q, bool write_cache_on)
 /*
  * Enable wbt if defaults are configured that way
  */
-void wbt_enable_default(struct request_queue *q)
+void wbt_enable_default(struct gendisk *disk)
 {
+	struct request_queue *q = disk->queue;
 	struct rq_qos *rqos;
 	bool disable_flag = q->elevator &&
 		    test_bit(ELEVATOR_FLAG_DISABLE_WBT, &q->elevator->flags);
@@ -718,9 +719,9 @@ static void wbt_exit(struct rq_qos *rqos)
 /*
  * Disable wbt, if enabled by default.
  */
-void wbt_disable_default(struct request_queue *q)
+void wbt_disable_default(struct gendisk *disk)
 {
-	struct rq_qos *rqos = wbt_rq_qos(q);
+	struct rq_qos *rqos = wbt_rq_qos(disk->queue);
 	struct rq_wb *rwb;
 	if (!rqos)
 		return;
diff --git a/block/blk-wbt.h b/block/blk-wbt.h
index e3ea6e7e290076..7ab1cba55c25f7 100644
--- a/block/blk-wbt.h
+++ b/block/blk-wbt.h
@@ -91,8 +91,8 @@ static inline unsigned int wbt_inflight(struct rq_wb *rwb)
 #ifdef CONFIG_BLK_WBT
 
 int wbt_init(struct request_queue *);
-void wbt_disable_default(struct request_queue *);
-void wbt_enable_default(struct request_queue *);
+void wbt_disable_default(struct gendisk *disk);
+void wbt_enable_default(struct gendisk *disk);
 
 u64 wbt_get_min_lat(struct request_queue *q);
 void wbt_set_min_lat(struct request_queue *q, u64 val);
@@ -108,10 +108,10 @@ static inline int wbt_init(struct request_queue *q)
 {
 	return -EINVAL;
 }
-static inline void wbt_disable_default(struct request_queue *q)
+static inline void wbt_disable_default(struct gendisk *disk)
 {
 }
-static inline void wbt_enable_default(struct request_queue *q)
+static inline void wbt_enable_default(struct gendisk *disk)
 {
 }
 static inline void wbt_set_write_cache(struct request_queue *q, bool wc)
-- 
2.39.0

[PATCH 09/19] blk-wbt: pass a gendisk to wbt_init

[Christoph Hellwig]

Pass a gendisk to wbt_init to prepare for phasing out usage of the
request_queue in the blk-cgroup code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/blk-sysfs.c | 2 +-
 block/blk-wbt.c   | 5 +++--
 block/blk-wbt.h   | 4 ++--
 3 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index 2074103865f45b..c2adf640e5c816 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -500,7 +500,7 @@ static ssize_t queue_wb_lat_store(struct request_queue *q, const char *page,
 
 	rqos = wbt_rq_qos(q);
 	if (!rqos) {
-		ret = wbt_init(q);
+		ret = wbt_init(q->disk);
 		if (ret)
 			return ret;
 	}
diff --git a/block/blk-wbt.c b/block/blk-wbt.c
index 8f9302134339c5..542271fa99e8f7 100644
--- a/block/blk-wbt.c
+++ b/block/blk-wbt.c
@@ -671,7 +671,7 @@ void wbt_enable_default(struct gendisk *disk)
 		return;
 
 	if (queue_is_mq(q) && !disable_flag)
-		wbt_init(q);
+		wbt_init(disk);
 }
 EXPORT_SYMBOL_GPL(wbt_enable_default);
 
@@ -835,8 +835,9 @@ static struct rq_qos_ops wbt_rqos_ops = {
 #endif
 };
 
-int wbt_init(struct request_queue *q)
+int wbt_init(struct gendisk *disk)
 {
+	struct request_queue *q = disk->queue;
 	struct rq_wb *rwb;
 	int i;
 	int ret;
diff --git a/block/blk-wbt.h b/block/blk-wbt.h
index 7ab1cba55c25f7..b673da41a867d3 100644
--- a/block/blk-wbt.h
+++ b/block/blk-wbt.h
@@ -90,7 +90,7 @@ static inline unsigned int wbt_inflight(struct rq_wb *rwb)
 
 #ifdef CONFIG_BLK_WBT
 
-int wbt_init(struct request_queue *);
+int wbt_init(struct gendisk *disk);
 void wbt_disable_default(struct gendisk *disk);
 void wbt_enable_default(struct gendisk *disk);
 
@@ -104,7 +104,7 @@ u64 wbt_default_latency_nsec(struct request_queue *);
 
 #else
 
-static inline int wbt_init(struct request_queue *q)
+static inline int wbt_init(struct gendisk *disk)
 {
 	return -EINVAL;
 }
-- 
2.39.0

[PATCH 10/19] blk-wbt: move private information from blk-wbt.h to blk-wbt.c

[Christoph Hellwig]

A large part of blk-wbt.h is only used in blk-wbt.c, so move it there.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/blk-settings.c |  1 +
 block/blk-sysfs.c    |  1 +
 block/blk-wbt.c      | 77 +++++++++++++++++++++++++++++++++++++++
 block/blk-wbt.h      | 86 --------------------------------------------
 4 files changed, 79 insertions(+), 86 deletions(-)

diff --git a/block/blk-settings.c b/block/blk-settings.c
index 9c9713c9269cc4..896b4654ab00fa 100644
--- a/block/blk-settings.c
+++ b/block/blk-settings.c
@@ -16,6 +16,7 @@
 #include <linux/dma-mapping.h>
 
 #include "blk.h"
+#include "blk-rq-qos.h"
 #include "blk-wbt.h"
 
 void blk_queue_rq_timeout(struct request_queue *q, unsigned int timeout)
diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index c2adf640e5c816..d70ebecb534798 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -16,6 +16,7 @@
 #include "blk-mq.h"
 #include "blk-mq-debugfs.h"
 #include "blk-mq-sched.h"
+#include "blk-rq-qos.h"
 #include "blk-wbt.h"
 #include "blk-cgroup.h"
 #include "blk-throttle.h"
diff --git a/block/blk-wbt.c b/block/blk-wbt.c
index 542271fa99e8f7..58f41a98fda911 100644
--- a/block/blk-wbt.c
+++ b/block/blk-wbt.c
@@ -25,6 +25,7 @@
 #include <linux/backing-dev.h>
 #include <linux/swap.h>
 
+#include "blk-stat.h"
 #include "blk-wbt.h"
 #include "blk-rq-qos.h"
 #include "elevator.h"
@@ -32,6 +33,72 @@
 #define CREATE_TRACE_POINTS
 #include <trace/events/wbt.h>
 
+enum wbt_flags {
+	WBT_TRACKED		= 1,	/* write, tracked for throttling */
+	WBT_READ		= 2,	/* read */
+	WBT_KSWAPD		= 4,	/* write, from kswapd */
+	WBT_DISCARD		= 8,	/* discard */
+
+	WBT_NR_BITS		= 4,	/* number of bits */
+};
+
+enum {
+	WBT_RWQ_BG		= 0,
+	WBT_RWQ_KSWAPD,
+	WBT_RWQ_DISCARD,
+	WBT_NUM_RWQ,
+};
+
+/*
+ * If current state is WBT_STATE_ON/OFF_DEFAULT, it can be covered to any other
+ * state, if current state is WBT_STATE_ON/OFF_MANUAL, it can only be covered
+ * to WBT_STATE_OFF/ON_MANUAL.
+ */
+enum {
+	WBT_STATE_ON_DEFAULT	= 1,	/* on by default */
+	WBT_STATE_ON_MANUAL	= 2,	/* on manually by sysfs */
+	WBT_STATE_OFF_DEFAULT	= 3,	/* off by default */
+	WBT_STATE_OFF_MANUAL	= 4,	/* off manually by sysfs */
+};
+
+struct rq_wb {
+	/*
+	 * Settings that govern how we throttle
+	 */
+	unsigned int wb_background;		/* background writeback */
+	unsigned int wb_normal;			/* normal writeback */
+
+	short enable_state;			/* WBT_STATE_* */
+
+	/*
+	 * Number of consecutive periods where we don't have enough
+	 * information to make a firm scale up/down decision.
+	 */
+	unsigned int unknown_cnt;
+
+	u64 win_nsec;				/* default window size */
+	u64 cur_win_nsec;			/* current window size */
+
+	struct blk_stat_callback *cb;
+
+	u64 sync_issue;
+	void *sync_cookie;
+
+	unsigned int wc;
+
+	unsigned long last_issue;		/* last non-throttled issue */
+	unsigned long last_comp;		/* last non-throttled comp */
+	unsigned long min_lat_nsec;
+	struct rq_qos rqos;
+	struct rq_wait rq_wait[WBT_NUM_RWQ];
+	struct rq_depth rq_depth;
+};
+
+static inline struct rq_wb *RQWB(struct rq_qos *rqos)
+{
+	return container_of(rqos, struct rq_wb, rqos);
+}
+
 static inline void wbt_clear_state(struct request *rq)
 {
 	rq->wbt_flags = 0;
@@ -226,6 +293,16 @@ static u64 rwb_sync_issue_lat(struct rq_wb *rwb)
 	return now - issue;
 }
 
+static inline unsigned int wbt_inflight(struct rq_wb *rwb)
+{
+	unsigned int i, ret = 0;
+
+	for (i = 0; i < WBT_NUM_RWQ; i++)
+		ret += atomic_read(&rwb->rq_wait[i].inflight);
+
+	return ret;
+}
+
 enum {
 	LAT_OK = 1,
 	LAT_UNKNOWN,
diff --git a/block/blk-wbt.h b/block/blk-wbt.h
index b673da41a867d3..ba6cca5849a655 100644
--- a/block/blk-wbt.h
+++ b/block/blk-wbt.h
@@ -2,92 +2,6 @@
 #ifndef WB_THROTTLE_H
 #define WB_THROTTLE_H
 
-#include <linux/kernel.h>
-#include <linux/atomic.h>
-#include <linux/wait.h>
-#include <linux/timer.h>
-#include <linux/ktime.h>
-
-#include "blk-stat.h"
-#include "blk-rq-qos.h"
-
-enum wbt_flags {
-	WBT_TRACKED		= 1,	/* write, tracked for throttling */
-	WBT_READ		= 2,	/* read */
-	WBT_KSWAPD		= 4,	/* write, from kswapd */
-	WBT_DISCARD		= 8,	/* discard */
-
-	WBT_NR_BITS		= 4,	/* number of bits */
-};
-
-enum {
-	WBT_RWQ_BG		= 0,
-	WBT_RWQ_KSWAPD,
-	WBT_RWQ_DISCARD,
-	WBT_NUM_RWQ,
-};
-
-/*
- * If current state is WBT_STATE_ON/OFF_DEFAULT, it can be covered to any other
- * state, if current state is WBT_STATE_ON/OFF_MANUAL, it can only be covered
- * to WBT_STATE_OFF/ON_MANUAL.
- */
-enum {
-	WBT_STATE_ON_DEFAULT	= 1,	/* on by default */
-	WBT_STATE_ON_MANUAL	= 2,	/* on manually by sysfs */
-	WBT_STATE_OFF_DEFAULT	= 3,	/* off by default */
-	WBT_STATE_OFF_MANUAL	= 4,	/* off manually by sysfs */
-};
-
-struct rq_wb {
-	/*
-	 * Settings that govern how we throttle
-	 */
-	unsigned int wb_background;		/* background writeback */
-	unsigned int wb_normal;			/* normal writeback */
-
-	short enable_state;			/* WBT_STATE_* */
-
-	/*
-	 * Number of consecutive periods where we don't have enough
-	 * information to make a firm scale up/down decision.
-	 */
-	unsigned int unknown_cnt;
-
-	u64 win_nsec;				/* default window size */
-	u64 cur_win_nsec;			/* current window size */
-
-	struct blk_stat_callback *cb;
-
-	u64 sync_issue;
-	void *sync_cookie;
-
-	unsigned int wc;
-
-	unsigned long last_issue;		/* last non-throttled issue */
-	unsigned long last_comp;		/* last non-throttled comp */
-	unsigned long min_lat_nsec;
-	struct rq_qos rqos;
-	struct rq_wait rq_wait[WBT_NUM_RWQ];
-	struct rq_depth rq_depth;
-};
-
-static inline struct rq_wb *RQWB(struct rq_qos *rqos)
-{
-	return container_of(rqos, struct rq_wb, rqos);
-}
-
-static inline unsigned int wbt_inflight(struct rq_wb *rwb)
-{
-	unsigned int i, ret = 0;
-
-	for (i = 0; i < WBT_NUM_RWQ; i++)
-		ret += atomic_read(&rwb->rq_wait[i].inflight);
-
-	return ret;
-}
-
-
 #ifdef CONFIG_BLK_WBT
 
 int wbt_init(struct gendisk *disk);
-- 
2.39.0

[PATCH 10/19] blk-wbt: move private information from blk-wbt.h to blk-wbt.c

[Christoph Hellwig]

A large part of blk-wbt.h is only used in blk-wbt.c, so move it there.

Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
Acked-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 block/blk-settings.c |  1 +
 block/blk-sysfs.c    |  1 +
 block/blk-wbt.c      | 77 +++++++++++++++++++++++++++++++++++++++
 block/blk-wbt.h      | 86 --------------------------------------------
 4 files changed, 79 insertions(+), 86 deletions(-)

diff --git a/block/blk-settings.c b/block/blk-settings.c
index 9c9713c9269cc4..896b4654ab00fa 100644
--- a/block/blk-settings.c
+++ b/block/blk-settings.c
@@ -16,6 +16,7 @@
 #include <linux/dma-mapping.h>
 
 #include "blk.h"
+#include "blk-rq-qos.h"
 #include "blk-wbt.h"
 
 void blk_queue_rq_timeout(struct request_queue *q, unsigned int timeout)
diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index c2adf640e5c816..d70ebecb534798 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -16,6 +16,7 @@
 #include "blk-mq.h"
 #include "blk-mq-debugfs.h"
 #include "blk-mq-sched.h"
+#include "blk-rq-qos.h"
 #include "blk-wbt.h"
 #include "blk-cgroup.h"
 #include "blk-throttle.h"
diff --git a/block/blk-wbt.c b/block/blk-wbt.c
index 542271fa99e8f7..58f41a98fda911 100644
--- a/block/blk-wbt.c
+++ b/block/blk-wbt.c
@@ -25,6 +25,7 @@
 #include <linux/backing-dev.h>
 #include <linux/swap.h>
 
+#include "blk-stat.h"
 #include "blk-wbt.h"
 #include "blk-rq-qos.h"
 #include "elevator.h"
@@ -32,6 +33,72 @@
 #define CREATE_TRACE_POINTS
 #include <trace/events/wbt.h>
 
+enum wbt_flags {
+	WBT_TRACKED		= 1,	/* write, tracked for throttling */
+	WBT_READ		= 2,	/* read */
+	WBT_KSWAPD		= 4,	/* write, from kswapd */
+	WBT_DISCARD		= 8,	/* discard */
+
+	WBT_NR_BITS		= 4,	/* number of bits */
+};
+
+enum {
+	WBT_RWQ_BG		= 0,
+	WBT_RWQ_KSWAPD,
+	WBT_RWQ_DISCARD,
+	WBT_NUM_RWQ,
+};
+
+/*
+ * If current state is WBT_STATE_ON/OFF_DEFAULT, it can be covered to any other
+ * state, if current state is WBT_STATE_ON/OFF_MANUAL, it can only be covered
+ * to WBT_STATE_OFF/ON_MANUAL.
+ */
+enum {
+	WBT_STATE_ON_DEFAULT	= 1,	/* on by default */
+	WBT_STATE_ON_MANUAL	= 2,	/* on manually by sysfs */
+	WBT_STATE_OFF_DEFAULT	= 3,	/* off by default */
+	WBT_STATE_OFF_MANUAL	= 4,	/* off manually by sysfs */
+};
+
+struct rq_wb {
+	/*
+	 * Settings that govern how we throttle
+	 */
+	unsigned int wb_background;		/* background writeback */
+	unsigned int wb_normal;			/* normal writeback */
+
+	short enable_state;			/* WBT_STATE_* */
+
+	/*
+	 * Number of consecutive periods where we don't have enough
+	 * information to make a firm scale up/down decision.
+	 */
+	unsigned int unknown_cnt;
+
+	u64 win_nsec;				/* default window size */
+	u64 cur_win_nsec;			/* current window size */
+
+	struct blk_stat_callback *cb;
+
+	u64 sync_issue;
+	void *sync_cookie;
+
+	unsigned int wc;
+
+	unsigned long last_issue;		/* last non-throttled issue */
+	unsigned long last_comp;		/* last non-throttled comp */
+	unsigned long min_lat_nsec;
+	struct rq_qos rqos;
+	struct rq_wait rq_wait[WBT_NUM_RWQ];
+	struct rq_depth rq_depth;
+};
+
+static inline struct rq_wb *RQWB(struct rq_qos *rqos)
+{
+	return container_of(rqos, struct rq_wb, rqos);
+}
+
 static inline void wbt_clear_state(struct request *rq)
 {
 	rq->wbt_flags = 0;
@@ -226,6 +293,16 @@ static u64 rwb_sync_issue_lat(struct rq_wb *rwb)
 	return now - issue;
 }
 
+static inline unsigned int wbt_inflight(struct rq_wb *rwb)
+{
+	unsigned int i, ret = 0;
+
+	for (i = 0; i < WBT_NUM_RWQ; i++)
+		ret += atomic_read(&rwb->rq_wait[i].inflight);
+
+	return ret;
+}
+
 enum {
 	LAT_OK = 1,
 	LAT_UNKNOWN,
diff --git a/block/blk-wbt.h b/block/blk-wbt.h
index b673da41a867d3..ba6cca5849a655 100644
--- a/block/blk-wbt.h
+++ b/block/blk-wbt.h
@@ -2,92 +2,6 @@
 #ifndef WB_THROTTLE_H
 #define WB_THROTTLE_H
 
-#include <linux/kernel.h>
-#include <linux/atomic.h>
-#include <linux/wait.h>
-#include <linux/timer.h>
-#include <linux/ktime.h>
-
-#include "blk-stat.h"
-#include "blk-rq-qos.h"
-
-enum wbt_flags {
-	WBT_TRACKED		= 1,	/* write, tracked for throttling */
-	WBT_READ		= 2,	/* read */
-	WBT_KSWAPD		= 4,	/* write, from kswapd */
-	WBT_DISCARD		= 8,	/* discard */
-
-	WBT_NR_BITS		= 4,	/* number of bits */
-};
-
-enum {
-	WBT_RWQ_BG		= 0,
-	WBT_RWQ_KSWAPD,
-	WBT_RWQ_DISCARD,
-	WBT_NUM_RWQ,
-};
-
-/*
- * If current state is WBT_STATE_ON/OFF_DEFAULT, it can be covered to any other
- * state, if current state is WBT_STATE_ON/OFF_MANUAL, it can only be covered
- * to WBT_STATE_OFF/ON_MANUAL.
- */
-enum {
-	WBT_STATE_ON_DEFAULT	= 1,	/* on by default */
-	WBT_STATE_ON_MANUAL	= 2,	/* on manually by sysfs */
-	WBT_STATE_OFF_DEFAULT	= 3,	/* off by default */
-	WBT_STATE_OFF_MANUAL	= 4,	/* off manually by sysfs */
-};
-
-struct rq_wb {
-	/*
-	 * Settings that govern how we throttle
-	 */
-	unsigned int wb_background;		/* background writeback */
-	unsigned int wb_normal;			/* normal writeback */
-
-	short enable_state;			/* WBT_STATE_* */
-
-	/*
-	 * Number of consecutive periods where we don't have enough
-	 * information to make a firm scale up/down decision.
-	 */
-	unsigned int unknown_cnt;
-
-	u64 win_nsec;				/* default window size */
-	u64 cur_win_nsec;			/* current window size */
-
-	struct blk_stat_callback *cb;
-
-	u64 sync_issue;
-	void *sync_cookie;
-
-	unsigned int wc;
-
-	unsigned long last_issue;		/* last non-throttled issue */
-	unsigned long last_comp;		/* last non-throttled comp */
-	unsigned long min_lat_nsec;
-	struct rq_qos rqos;
-	struct rq_wait rq_wait[WBT_NUM_RWQ];
-	struct rq_depth rq_depth;
-};
-
-static inline struct rq_wb *RQWB(struct rq_qos *rqos)
-{
-	return container_of(rqos, struct rq_wb, rqos);
-}
-
-static inline unsigned int wbt_inflight(struct rq_wb *rwb)
-{
-	unsigned int i, ret = 0;
-
-	for (i = 0; i < WBT_NUM_RWQ; i++)
-		ret += atomic_read(&rwb->rq_wait[i].inflight);
-
-	return ret;
-}
-
-
 #ifdef CONFIG_BLK_WBT
 
 int wbt_init(struct gendisk *disk);
-- 
2.39.0

[PATCH 11/19] blk-wbt: open code wbt_queue_depth_changed in wbt_init

[Christoph Hellwig]

wbt_queue_depth_changed just updates a field and calls another function.
Open code it in wbt_init, so that the local queue variable can be used
instead of the one stored in the rq_qos.  This will allow delaying that
rq_qos->queue assignment in a subsequent patch.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/blk-wbt.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/block/blk-wbt.c b/block/blk-wbt.c
index 58f41a98fda911..119a43671089ea 100644
--- a/block/blk-wbt.c
+++ b/block/blk-wbt.c
@@ -941,8 +941,8 @@ int wbt_init(struct gendisk *disk)
 	rwb->wc = test_bit(QUEUE_FLAG_WC, &q->queue_flags);
 	rwb->rq_depth.default_depth = RWB_DEF_DEPTH;
 	rwb->min_lat_nsec = wbt_default_latency_nsec(q);
-
-	wbt_queue_depth_changed(&rwb->rqos);
+	rwb->rq_depth.queue_depth = blk_queue_depth(q);
+	wbt_update_limits(rwb);
 
 	/*
 	 * Assign rwb and add the stats callback.
-- 
2.39.0

[PATCH 12/19] blk-rq-qos: move rq_qos_add and rq_qos_del out of line

[Christoph Hellwig]

These two functions are rather larger and not in a fast path, so move
them out of line.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/blk-rq-qos.c | 60 +++++++++++++++++++++++++++++++++++++++++++++
 block/blk-rq-qos.h | 61 ++--------------------------------------------
 2 files changed, 62 insertions(+), 59 deletions(-)

diff --git a/block/blk-rq-qos.c b/block/blk-rq-qos.c
index 88f0fe7dcf5451..aae98dcb01ebe7 100644
--- a/block/blk-rq-qos.c
+++ b/block/blk-rq-qos.c
@@ -294,3 +294,63 @@ void rq_qos_exit(struct request_queue *q)
 		rqos->ops->exit(rqos);
 	}
 }
+
+int rq_qos_add(struct request_queue *q, struct rq_qos *rqos)
+{
+	/*
+	 * No IO can be in-flight when adding rqos, so freeze queue, which
+	 * is fine since we only support rq_qos for blk-mq queue.
+	 *
+	 * Reuse ->queue_lock for protecting against other concurrent
+	 * rq_qos adding/deleting
+	 */
+	blk_mq_freeze_queue(q);
+
+	spin_lock_irq(&q->queue_lock);
+	if (rq_qos_id(q, rqos->id))
+		goto ebusy;
+	rqos->next = q->rq_qos;
+	q->rq_qos = rqos;
+	spin_unlock_irq(&q->queue_lock);
+
+	blk_mq_unfreeze_queue(q);
+
+	if (rqos->ops->debugfs_attrs) {
+		mutex_lock(&q->debugfs_mutex);
+		blk_mq_debugfs_register_rqos(rqos);
+		mutex_unlock(&q->debugfs_mutex);
+	}
+
+	return 0;
+ebusy:
+	spin_unlock_irq(&q->queue_lock);
+	blk_mq_unfreeze_queue(q);
+	return -EBUSY;
+
+}
+
+void rq_qos_del(struct request_queue *q, struct rq_qos *rqos)
+{
+	struct rq_qos **cur;
+
+	/*
+	 * See comment in rq_qos_add() about freezing queue & using
+	 * ->queue_lock.
+	 */
+	blk_mq_freeze_queue(q);
+
+	spin_lock_irq(&q->queue_lock);
+	for (cur = &q->rq_qos; *cur; cur = &(*cur)->next) {
+		if (*cur == rqos) {
+			*cur = rqos->next;
+			break;
+		}
+	}
+	spin_unlock_irq(&q->queue_lock);
+
+	blk_mq_unfreeze_queue(q);
+
+	mutex_lock(&q->debugfs_mutex);
+	blk_mq_debugfs_unregister_rqos(rqos);
+	mutex_unlock(&q->debugfs_mutex);
+}
diff --git a/block/blk-rq-qos.h b/block/blk-rq-qos.h
index 1ef1f7d4bc3cbc..805eee8b031d00 100644
--- a/block/blk-rq-qos.h
+++ b/block/blk-rq-qos.h
@@ -85,65 +85,8 @@ static inline void rq_wait_init(struct rq_wait *rq_wait)
 	init_waitqueue_head(&rq_wait->wait);
 }
 
-static inline int rq_qos_add(struct request_queue *q, struct rq_qos *rqos)
-{
-	/*
-	 * No IO can be in-flight when adding rqos, so freeze queue, which
-	 * is fine since we only support rq_qos for blk-mq queue.
-	 *
-	 * Reuse ->queue_lock for protecting against other concurrent
-	 * rq_qos adding/deleting
-	 */
-	blk_mq_freeze_queue(q);
-
-	spin_lock_irq(&q->queue_lock);
-	if (rq_qos_id(q, rqos->id))
-		goto ebusy;
-	rqos->next = q->rq_qos;
-	q->rq_qos = rqos;
-	spin_unlock_irq(&q->queue_lock);
-
-	blk_mq_unfreeze_queue(q);
-
-	if (rqos->ops->debugfs_attrs) {
-		mutex_lock(&q->debugfs_mutex);
-		blk_mq_debugfs_register_rqos(rqos);
-		mutex_unlock(&q->debugfs_mutex);
-	}
-
-	return 0;
-ebusy:
-	spin_unlock_irq(&q->queue_lock);
-	blk_mq_unfreeze_queue(q);
-	return -EBUSY;
-
-}
-
-static inline void rq_qos_del(struct request_queue *q, struct rq_qos *rqos)
-{
-	struct rq_qos **cur;
-
-	/*
-	 * See comment in rq_qos_add() about freezing queue & using
-	 * ->queue_lock.
-	 */
-	blk_mq_freeze_queue(q);
-
-	spin_lock_irq(&q->queue_lock);
-	for (cur = &q->rq_qos; *cur; cur = &(*cur)->next) {
-		if (*cur == rqos) {
-			*cur = rqos->next;
-			break;
-		}
-	}
-	spin_unlock_irq(&q->queue_lock);
-
-	blk_mq_unfreeze_queue(q);
-
-	mutex_lock(&q->debugfs_mutex);
-	blk_mq_debugfs_unregister_rqos(rqos);
-	mutex_unlock(&q->debugfs_mutex);
-}
+int rq_qos_add(struct request_queue *q, struct rq_qos *rqos);
+void rq_qos_del(struct request_queue *q, struct rq_qos *rqos);
 
 typedef bool (acquire_inflight_cb_t)(struct rq_wait *rqw, void *private_data);
 typedef void (cleanup_cb_t)(struct rq_wait *rqw, void *private_data);
-- 
2.39.0

[PATCH 13/19] blk-rq-qos: make rq_qos_add and rq_qos_del more useful

[Christoph Hellwig]

Switch to passing a gendisk, and make rq_qos_add initialize all required
fields and drop the not required q argument from rq_qos_del.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/blk-iocost.c    | 13 +++----------
 block/blk-iolatency.c | 14 ++++----------
 block/blk-rq-qos.c    | 13 ++++++++++---
 block/blk-rq-qos.h    |  5 +++--
 block/blk-wbt.c       |  5 +----
 5 files changed, 21 insertions(+), 29 deletions(-)

diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index dbb93f4f68d979..5f28463cba0afe 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -2843,9 +2843,7 @@ static struct rq_qos_ops ioc_rqos_ops = {
 
 static int blk_iocost_init(struct gendisk *disk)
 {
-	struct request_queue *q = disk->queue;
 	struct ioc *ioc;
-	struct rq_qos *rqos;
 	int i, cpu, ret;
 
 	ioc = kzalloc(sizeof(*ioc), GFP_KERNEL);
@@ -2868,11 +2866,6 @@ static int blk_iocost_init(struct gendisk *disk)
 		local64_set(&ccs->rq_wait_ns, 0);
 	}
 
-	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);
@@ -2896,17 +2889,17 @@ static int blk_iocost_init(struct gendisk *disk)
 	 * called before policy activation completion, can't assume that the
 	 * target bio has an iocg associated and need to test for NULL iocg.
 	 */
-	ret = rq_qos_add(q, rqos);
+	ret = rq_qos_add(&ioc->rqos, disk, RQ_QOS_COST, &ioc_rqos_ops);
 	if (ret)
 		goto err_free_ioc;
 
-	ret = blkcg_activate_policy(q, &blkcg_policy_iocost);
+	ret = blkcg_activate_policy(disk->queue, &blkcg_policy_iocost);
 	if (ret)
 		goto err_del_qos;
 	return 0;
 
 err_del_qos:
-	rq_qos_del(q, rqos);
+	rq_qos_del(&ioc->rqos);
 err_free_ioc:
 	free_percpu(ioc->pcpu_stat);
 	kfree(ioc);
diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c
index b55eac2cf91944..1c394bd77aa0b4 100644
--- a/block/blk-iolatency.c
+++ b/block/blk-iolatency.c
@@ -757,24 +757,18 @@ static void blkiolatency_enable_work_fn(struct work_struct *work)
 
 int blk_iolatency_init(struct gendisk *disk)
 {
-	struct request_queue *q = disk->queue;
 	struct blk_iolatency *blkiolat;
-	struct rq_qos *rqos;
 	int ret;
 
 	blkiolat = kzalloc(sizeof(*blkiolat), GFP_KERNEL);
 	if (!blkiolat)
 		return -ENOMEM;
 
-	rqos = &blkiolat->rqos;
-	rqos->id = RQ_QOS_LATENCY;
-	rqos->ops = &blkcg_iolatency_ops;
-	rqos->q = q;
-
-	ret = rq_qos_add(q, rqos);
+	ret = rq_qos_add(&blkiolat->rqos, disk, RQ_QOS_LATENCY,
+			 &blkcg_iolatency_ops);
 	if (ret)
 		goto err_free;
-	ret = blkcg_activate_policy(q, &blkcg_policy_iolatency);
+	ret = blkcg_activate_policy(disk->queue, &blkcg_policy_iolatency);
 	if (ret)
 		goto err_qos_del;
 
@@ -784,7 +778,7 @@ int blk_iolatency_init(struct gendisk *disk)
 	return 0;
 
 err_qos_del:
-	rq_qos_del(q, rqos);
+	rq_qos_del(&blkiolat->rqos);
 err_free:
 	kfree(blkiolat);
 	return ret;
diff --git a/block/blk-rq-qos.c b/block/blk-rq-qos.c
index aae98dcb01ebe7..14bee1bd761362 100644
--- a/block/blk-rq-qos.c
+++ b/block/blk-rq-qos.c
@@ -295,8 +295,15 @@ void rq_qos_exit(struct request_queue *q)
 	}
 }
 
-int rq_qos_add(struct request_queue *q, struct rq_qos *rqos)
+int rq_qos_add(struct rq_qos *rqos, struct gendisk *disk, enum rq_qos_id id,
+		struct rq_qos_ops *ops)
 {
+	struct request_queue *q = disk->queue;
+
+	rqos->q = q;
+	rqos->id = id;
+	rqos->ops = ops;
+
 	/*
 	 * No IO can be in-flight when adding rqos, so freeze queue, which
 	 * is fine since we only support rq_qos for blk-mq queue.
@@ -326,11 +333,11 @@ int rq_qos_add(struct request_queue *q, struct rq_qos *rqos)
 	spin_unlock_irq(&q->queue_lock);
 	blk_mq_unfreeze_queue(q);
 	return -EBUSY;
-
 }
 
-void rq_qos_del(struct request_queue *q, struct rq_qos *rqos)
+void rq_qos_del(struct rq_qos *rqos)
 {
+	struct request_queue *q = rqos->q;
 	struct rq_qos **cur;
 
 	/*
diff --git a/block/blk-rq-qos.h b/block/blk-rq-qos.h
index 805eee8b031d00..22552785aa31ed 100644
--- a/block/blk-rq-qos.h
+++ b/block/blk-rq-qos.h
@@ -85,8 +85,9 @@ static inline void rq_wait_init(struct rq_wait *rq_wait)
 	init_waitqueue_head(&rq_wait->wait);
 }
 
-int rq_qos_add(struct request_queue *q, struct rq_qos *rqos);
-void rq_qos_del(struct request_queue *q, struct rq_qos *rqos);
+int rq_qos_add(struct rq_qos *rqos, struct gendisk *disk, enum rq_qos_id id,
+		struct rq_qos_ops *ops);
+void rq_qos_del(struct rq_qos *rqos);
 
 typedef bool (acquire_inflight_cb_t)(struct rq_wait *rqw, void *private_data);
 typedef void (cleanup_cb_t)(struct rq_wait *rqw, void *private_data);
diff --git a/block/blk-wbt.c b/block/blk-wbt.c
index 119a43671089ea..75565ae2775297 100644
--- a/block/blk-wbt.c
+++ b/block/blk-wbt.c
@@ -932,9 +932,6 @@ int wbt_init(struct gendisk *disk)
 	for (i = 0; i < WBT_NUM_RWQ; i++)
 		rq_wait_init(&rwb->rq_wait[i]);
 
-	rwb->rqos.id = RQ_QOS_WBT;
-	rwb->rqos.ops = &wbt_rqos_ops;
-	rwb->rqos.q = q;
 	rwb->last_comp = rwb->last_issue = jiffies;
 	rwb->win_nsec = RWB_WINDOW_NSEC;
 	rwb->enable_state = WBT_STATE_ON_DEFAULT;
@@ -947,7 +944,7 @@ int wbt_init(struct gendisk *disk)
 	/*
 	 * Assign rwb and add the stats callback.
 	 */
-	ret = rq_qos_add(q, &rwb->rqos);
+	ret = rq_qos_add(&rwb->rqos, disk, RQ_QOS_WBT, &wbt_rqos_ops);
 	if (ret)
 		goto err_free;
 
-- 
2.39.0

[PATCH 13/19] blk-rq-qos: make rq_qos_add and rq_qos_del more useful

[Christoph Hellwig]

Switch to passing a gendisk, and make rq_qos_add initialize all required
fields and drop the not required q argument from rq_qos_del.

Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
Reviewed-by: Andreas Herrmann <aherrmann-l3A5Bk7waGM@public.gmane.org>
Acked-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 block/blk-iocost.c    | 13 +++----------
 block/blk-iolatency.c | 14 ++++----------
 block/blk-rq-qos.c    | 13 ++++++++++---
 block/blk-rq-qos.h    |  5 +++--
 block/blk-wbt.c       |  5 +----
 5 files changed, 21 insertions(+), 29 deletions(-)

diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index dbb93f4f68d979..5f28463cba0afe 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -2843,9 +2843,7 @@ static struct rq_qos_ops ioc_rqos_ops = {
 
 static int blk_iocost_init(struct gendisk *disk)
 {
-	struct request_queue *q = disk->queue;
 	struct ioc *ioc;
-	struct rq_qos *rqos;
 	int i, cpu, ret;
 
 	ioc = kzalloc(sizeof(*ioc), GFP_KERNEL);
@@ -2868,11 +2866,6 @@ static int blk_iocost_init(struct gendisk *disk)
 		local64_set(&ccs->rq_wait_ns, 0);
 	}
 
-	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);
@@ -2896,17 +2889,17 @@ static int blk_iocost_init(struct gendisk *disk)
 	 * called before policy activation completion, can't assume that the
 	 * target bio has an iocg associated and need to test for NULL iocg.
 	 */
-	ret = rq_qos_add(q, rqos);
+	ret = rq_qos_add(&ioc->rqos, disk, RQ_QOS_COST, &ioc_rqos_ops);
 	if (ret)
 		goto err_free_ioc;
 
-	ret = blkcg_activate_policy(q, &blkcg_policy_iocost);
+	ret = blkcg_activate_policy(disk->queue, &blkcg_policy_iocost);
 	if (ret)
 		goto err_del_qos;
 	return 0;
 
 err_del_qos:
-	rq_qos_del(q, rqos);
+	rq_qos_del(&ioc->rqos);
 err_free_ioc:
 	free_percpu(ioc->pcpu_stat);
 	kfree(ioc);
diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c
index b55eac2cf91944..1c394bd77aa0b4 100644
--- a/block/blk-iolatency.c
+++ b/block/blk-iolatency.c
@@ -757,24 +757,18 @@ static void blkiolatency_enable_work_fn(struct work_struct *work)
 
 int blk_iolatency_init(struct gendisk *disk)
 {
-	struct request_queue *q = disk->queue;
 	struct blk_iolatency *blkiolat;
-	struct rq_qos *rqos;
 	int ret;
 
 	blkiolat = kzalloc(sizeof(*blkiolat), GFP_KERNEL);
 	if (!blkiolat)
 		return -ENOMEM;
 
-	rqos = &blkiolat->rqos;
-	rqos->id = RQ_QOS_LATENCY;
-	rqos->ops = &blkcg_iolatency_ops;
-	rqos->q = q;
-
-	ret = rq_qos_add(q, rqos);
+	ret = rq_qos_add(&blkiolat->rqos, disk, RQ_QOS_LATENCY,
+			 &blkcg_iolatency_ops);
 	if (ret)
 		goto err_free;
-	ret = blkcg_activate_policy(q, &blkcg_policy_iolatency);
+	ret = blkcg_activate_policy(disk->queue, &blkcg_policy_iolatency);
 	if (ret)
 		goto err_qos_del;
 
@@ -784,7 +778,7 @@ int blk_iolatency_init(struct gendisk *disk)
 	return 0;
 
 err_qos_del:
-	rq_qos_del(q, rqos);
+	rq_qos_del(&blkiolat->rqos);
 err_free:
 	kfree(blkiolat);
 	return ret;
diff --git a/block/blk-rq-qos.c b/block/blk-rq-qos.c
index aae98dcb01ebe7..14bee1bd761362 100644
--- a/block/blk-rq-qos.c
+++ b/block/blk-rq-qos.c
@@ -295,8 +295,15 @@ void rq_qos_exit(struct request_queue *q)
 	}
 }
 
-int rq_qos_add(struct request_queue *q, struct rq_qos *rqos)
+int rq_qos_add(struct rq_qos *rqos, struct gendisk *disk, enum rq_qos_id id,
+		struct rq_qos_ops *ops)
 {
+	struct request_queue *q = disk->queue;
+
+	rqos->q = q;
+	rqos->id = id;
+	rqos->ops = ops;
+
 	/*
 	 * No IO can be in-flight when adding rqos, so freeze queue, which
 	 * is fine since we only support rq_qos for blk-mq queue.
@@ -326,11 +333,11 @@ int rq_qos_add(struct request_queue *q, struct rq_qos *rqos)
 	spin_unlock_irq(&q->queue_lock);
 	blk_mq_unfreeze_queue(q);
 	return -EBUSY;
-
 }
 
-void rq_qos_del(struct request_queue *q, struct rq_qos *rqos)
+void rq_qos_del(struct rq_qos *rqos)
 {
+	struct request_queue *q = rqos->q;
 	struct rq_qos **cur;
 
 	/*
diff --git a/block/blk-rq-qos.h b/block/blk-rq-qos.h
index 805eee8b031d00..22552785aa31ed 100644
--- a/block/blk-rq-qos.h
+++ b/block/blk-rq-qos.h
@@ -85,8 +85,9 @@ static inline void rq_wait_init(struct rq_wait *rq_wait)
 	init_waitqueue_head(&rq_wait->wait);
 }
 
-int rq_qos_add(struct request_queue *q, struct rq_qos *rqos);
-void rq_qos_del(struct request_queue *q, struct rq_qos *rqos);
+int rq_qos_add(struct rq_qos *rqos, struct gendisk *disk, enum rq_qos_id id,
+		struct rq_qos_ops *ops);
+void rq_qos_del(struct rq_qos *rqos);
 
 typedef bool (acquire_inflight_cb_t)(struct rq_wait *rqw, void *private_data);
 typedef void (cleanup_cb_t)(struct rq_wait *rqw, void *private_data);
diff --git a/block/blk-wbt.c b/block/blk-wbt.c
index 119a43671089ea..75565ae2775297 100644
--- a/block/blk-wbt.c
+++ b/block/blk-wbt.c
@@ -932,9 +932,6 @@ int wbt_init(struct gendisk *disk)
 	for (i = 0; i < WBT_NUM_RWQ; i++)
 		rq_wait_init(&rwb->rq_wait[i]);
 
-	rwb->rqos.id = RQ_QOS_WBT;
-	rwb->rqos.ops = &wbt_rqos_ops;
-	rwb->rqos.q = q;
 	rwb->last_comp = rwb->last_issue = jiffies;
 	rwb->win_nsec = RWB_WINDOW_NSEC;
 	rwb->enable_state = WBT_STATE_ON_DEFAULT;
@@ -947,7 +944,7 @@ int wbt_init(struct gendisk *disk)
 	/*
 	 * Assign rwb and add the stats callback.
 	 */
-	ret = rq_qos_add(q, &rwb->rqos);
+	ret = rq_qos_add(&rwb->rqos, disk, RQ_QOS_WBT, &wbt_rqos_ops);
 	if (ret)
 		goto err_free;
 
-- 
2.39.0

[PATCH 14/19] blk-rq-qos: constify rq_qos_ops

[Christoph Hellwig]

These op vectors are constant, so mark them const.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/blk-iocost.c    | 2 +-
 block/blk-iolatency.c | 2 +-
 block/blk-rq-qos.c    | 2 +-
 block/blk-rq-qos.h    | 4 ++--
 block/blk-wbt.c       | 2 +-
 5 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index 5f28463cba0afe..6f1da7883905b3 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -2832,7 +2832,7 @@ static void ioc_rqos_exit(struct rq_qos *rqos)
 	kfree(ioc);
 }
 
-static struct rq_qos_ops ioc_rqos_ops = {
+static const struct rq_qos_ops ioc_rqos_ops = {
 	.throttle = ioc_rqos_throttle,
 	.merge = ioc_rqos_merge,
 	.done_bio = ioc_rqos_done_bio,
diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c
index 1c394bd77aa0b4..f6aeb3d3fdae59 100644
--- a/block/blk-iolatency.c
+++ b/block/blk-iolatency.c
@@ -650,7 +650,7 @@ static void blkcg_iolatency_exit(struct rq_qos *rqos)
 	kfree(blkiolat);
 }
 
-static struct rq_qos_ops blkcg_iolatency_ops = {
+static const struct rq_qos_ops blkcg_iolatency_ops = {
 	.throttle = blkcg_iolatency_throttle,
 	.done_bio = blkcg_iolatency_done_bio,
 	.exit = blkcg_iolatency_exit,
diff --git a/block/blk-rq-qos.c b/block/blk-rq-qos.c
index 14bee1bd761362..8e83734cfe8dbc 100644
--- a/block/blk-rq-qos.c
+++ b/block/blk-rq-qos.c
@@ -296,7 +296,7 @@ void rq_qos_exit(struct request_queue *q)
 }
 
 int rq_qos_add(struct rq_qos *rqos, struct gendisk *disk, enum rq_qos_id id,
-		struct rq_qos_ops *ops)
+		const struct rq_qos_ops *ops)
 {
 	struct request_queue *q = disk->queue;
 
diff --git a/block/blk-rq-qos.h b/block/blk-rq-qos.h
index 22552785aa31ed..2b7b668479f71a 100644
--- a/block/blk-rq-qos.h
+++ b/block/blk-rq-qos.h
@@ -25,7 +25,7 @@ struct rq_wait {
 };
 
 struct rq_qos {
-	struct rq_qos_ops *ops;
+	const struct rq_qos_ops *ops;
 	struct request_queue *q;
 	enum rq_qos_id id;
 	struct rq_qos *next;
@@ -86,7 +86,7 @@ static inline void rq_wait_init(struct rq_wait *rq_wait)
 }
 
 int rq_qos_add(struct rq_qos *rqos, struct gendisk *disk, enum rq_qos_id id,
-		struct rq_qos_ops *ops);
+		const struct rq_qos_ops *ops);
 void rq_qos_del(struct rq_qos *rqos);
 
 typedef bool (acquire_inflight_cb_t)(struct rq_wait *rqw, void *private_data);
diff --git a/block/blk-wbt.c b/block/blk-wbt.c
index 75565ae2775297..1a78d54c8152b0 100644
--- a/block/blk-wbt.c
+++ b/block/blk-wbt.c
@@ -898,7 +898,7 @@ static const struct blk_mq_debugfs_attr wbt_debugfs_attrs[] = {
 };
 #endif
 
-static struct rq_qos_ops wbt_rqos_ops = {
+static const struct rq_qos_ops wbt_rqos_ops = {
 	.throttle = wbt_wait,
 	.issue = wbt_issue,
 	.track = wbt_track,
-- 
2.39.0

[PATCH 14/19] blk-rq-qos: constify rq_qos_ops

[Christoph Hellwig]

These op vectors are constant, so mark them const.

Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
Reviewed-by: Andreas Herrmann <aherrmann-l3A5Bk7waGM@public.gmane.org>
Acked-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 block/blk-iocost.c    | 2 +-
 block/blk-iolatency.c | 2 +-
 block/blk-rq-qos.c    | 2 +-
 block/blk-rq-qos.h    | 4 ++--
 block/blk-wbt.c       | 2 +-
 5 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index 5f28463cba0afe..6f1da7883905b3 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -2832,7 +2832,7 @@ static void ioc_rqos_exit(struct rq_qos *rqos)
 	kfree(ioc);
 }
 
-static struct rq_qos_ops ioc_rqos_ops = {
+static const struct rq_qos_ops ioc_rqos_ops = {
 	.throttle = ioc_rqos_throttle,
 	.merge = ioc_rqos_merge,
 	.done_bio = ioc_rqos_done_bio,
diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c
index 1c394bd77aa0b4..f6aeb3d3fdae59 100644
--- a/block/blk-iolatency.c
+++ b/block/blk-iolatency.c
@@ -650,7 +650,7 @@ static void blkcg_iolatency_exit(struct rq_qos *rqos)
 	kfree(blkiolat);
 }
 
-static struct rq_qos_ops blkcg_iolatency_ops = {
+static const struct rq_qos_ops blkcg_iolatency_ops = {
 	.throttle = blkcg_iolatency_throttle,
 	.done_bio = blkcg_iolatency_done_bio,
 	.exit = blkcg_iolatency_exit,
diff --git a/block/blk-rq-qos.c b/block/blk-rq-qos.c
index 14bee1bd761362..8e83734cfe8dbc 100644
--- a/block/blk-rq-qos.c
+++ b/block/blk-rq-qos.c
@@ -296,7 +296,7 @@ void rq_qos_exit(struct request_queue *q)
 }
 
 int rq_qos_add(struct rq_qos *rqos, struct gendisk *disk, enum rq_qos_id id,
-		struct rq_qos_ops *ops)
+		const struct rq_qos_ops *ops)
 {
 	struct request_queue *q = disk->queue;
 
diff --git a/block/blk-rq-qos.h b/block/blk-rq-qos.h
index 22552785aa31ed..2b7b668479f71a 100644
--- a/block/blk-rq-qos.h
+++ b/block/blk-rq-qos.h
@@ -25,7 +25,7 @@ struct rq_wait {
 };
 
 struct rq_qos {
-	struct rq_qos_ops *ops;
+	const struct rq_qos_ops *ops;
 	struct request_queue *q;
 	enum rq_qos_id id;
 	struct rq_qos *next;
@@ -86,7 +86,7 @@ static inline void rq_wait_init(struct rq_wait *rq_wait)
 }
 
 int rq_qos_add(struct rq_qos *rqos, struct gendisk *disk, enum rq_qos_id id,
-		struct rq_qos_ops *ops);
+		const struct rq_qos_ops *ops);
 void rq_qos_del(struct rq_qos *rqos);
 
 typedef bool (acquire_inflight_cb_t)(struct rq_wait *rqw, void *private_data);
diff --git a/block/blk-wbt.c b/block/blk-wbt.c
index 75565ae2775297..1a78d54c8152b0 100644
--- a/block/blk-wbt.c
+++ b/block/blk-wbt.c
@@ -898,7 +898,7 @@ static const struct blk_mq_debugfs_attr wbt_debugfs_attrs[] = {
 };
 #endif
 
-static struct rq_qos_ops wbt_rqos_ops = {
+static const struct rq_qos_ops wbt_rqos_ops = {
 	.throttle = wbt_wait,
 	.issue = wbt_issue,
 	.track = wbt_track,
-- 
2.39.0

[PATCH 15/19] blk-rq-qos: store a gendisk instead of request_queue in struct rq_qos

[Christoph Hellwig]

This is what about half of the users already want, and it's only going to
grow more.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/blk-iocost.c     | 12 ++++++------
 block/blk-iolatency.c  | 14 +++++++-------
 block/blk-mq-debugfs.c | 10 ++++------
 block/blk-rq-qos.c     |  4 ++--
 block/blk-rq-qos.h     |  2 +-
 block/blk-wbt.c        | 16 +++++++---------
 6 files changed, 27 insertions(+), 31 deletions(-)

diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index 6f1da7883905b3..a2e9bf30039bcd 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -669,7 +669,7 @@ static struct ioc *q_to_ioc(struct request_queue *q)
 
 static const char __maybe_unused *ioc_name(struct ioc *ioc)
 {
-	struct gendisk *disk = ioc->rqos.q->disk;
+	struct gendisk *disk = ioc->rqos.disk;
 
 	if (!disk)
 		return "<unknown>";
@@ -808,11 +808,11 @@ static int ioc_autop_idx(struct ioc *ioc)
 	u64 now_ns;
 
 	/* rotational? */
-	if (!blk_queue_nonrot(ioc->rqos.q))
+	if (!blk_queue_nonrot(ioc->rqos.disk->queue))
 		return AUTOP_HDD;
 
 	/* handle SATA SSDs w/ broken NCQ */
-	if (blk_queue_depth(ioc->rqos.q) == 1)
+	if (blk_queue_depth(ioc->rqos.disk->queue) == 1)
 		return AUTOP_SSD_QD1;
 
 	/* use one of the normal ssd sets */
@@ -2649,7 +2649,7 @@ static void ioc_rqos_throttle(struct rq_qos *rqos, struct bio *bio)
 	if (use_debt) {
 		iocg_incur_debt(iocg, abs_cost, &now);
 		if (iocg_kick_delay(iocg, &now))
-			blkcg_schedule_throttle(rqos->q->disk,
+			blkcg_schedule_throttle(rqos->disk,
 					(bio->bi_opf & REQ_SWAP) == REQ_SWAP);
 		iocg_unlock(iocg, ioc_locked, &flags);
 		return;
@@ -2750,7 +2750,7 @@ static void ioc_rqos_merge(struct rq_qos *rqos, struct request *rq,
 	if (likely(!list_empty(&iocg->active_list))) {
 		iocg_incur_debt(iocg, abs_cost, &now);
 		if (iocg_kick_delay(iocg, &now))
-			blkcg_schedule_throttle(rqos->q->disk,
+			blkcg_schedule_throttle(rqos->disk,
 					(bio->bi_opf & REQ_SWAP) == REQ_SWAP);
 	} else {
 		iocg_commit_bio(iocg, bio, abs_cost, cost);
@@ -2821,7 +2821,7 @@ static void ioc_rqos_exit(struct rq_qos *rqos)
 {
 	struct ioc *ioc = rqos_to_ioc(rqos);
 
-	blkcg_deactivate_policy(rqos->q, &blkcg_policy_iocost);
+	blkcg_deactivate_policy(rqos->disk->queue, &blkcg_policy_iocost);
 
 	spin_lock_irq(&ioc->lock);
 	ioc->running = IOC_STOP;
diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c
index f6aeb3d3fdae59..8e1e43bbde6f0b 100644
--- a/block/blk-iolatency.c
+++ b/block/blk-iolatency.c
@@ -292,7 +292,7 @@ static void __blkcg_iolatency_throttle(struct rq_qos *rqos,
 	unsigned use_delay = atomic_read(&lat_to_blkg(iolat)->use_delay);
 
 	if (use_delay)
-		blkcg_schedule_throttle(rqos->q->disk, use_memdelay);
+		blkcg_schedule_throttle(rqos->disk, use_memdelay);
 
 	/*
 	 * To avoid priority inversions we want to just take a slot if we are
@@ -330,7 +330,7 @@ static void scale_cookie_change(struct blk_iolatency *blkiolat,
 				struct child_latency_info *lat_info,
 				bool up)
 {
-	unsigned long qd = blkiolat->rqos.q->nr_requests;
+	unsigned long qd = blkiolat->rqos.disk->queue->nr_requests;
 	unsigned long scale = scale_amount(qd, up);
 	unsigned long old = atomic_read(&lat_info->scale_cookie);
 	unsigned long max_scale = qd << 1;
@@ -372,7 +372,7 @@ static void scale_cookie_change(struct blk_iolatency *blkiolat,
  */
 static void scale_change(struct iolatency_grp *iolat, bool up)
 {
-	unsigned long qd = iolat->blkiolat->rqos.q->nr_requests;
+	unsigned long qd = iolat->blkiolat->rqos.disk->queue->nr_requests;
 	unsigned long scale = scale_amount(qd, up);
 	unsigned long old = iolat->max_depth;
 
@@ -646,7 +646,7 @@ static void blkcg_iolatency_exit(struct rq_qos *rqos)
 
 	timer_shutdown_sync(&blkiolat->timer);
 	flush_work(&blkiolat->enable_work);
-	blkcg_deactivate_policy(rqos->q, &blkcg_policy_iolatency);
+	blkcg_deactivate_policy(rqos->disk->queue, &blkcg_policy_iolatency);
 	kfree(blkiolat);
 }
 
@@ -665,7 +665,7 @@ static void blkiolatency_timer_fn(struct timer_list *t)
 
 	rcu_read_lock();
 	blkg_for_each_descendant_pre(blkg, pos_css,
-				     blkiolat->rqos.q->root_blkg) {
+				     blkiolat->rqos.disk->queue->root_blkg) {
 		struct iolatency_grp *iolat;
 		struct child_latency_info *lat_info;
 		unsigned long flags;
@@ -749,9 +749,9 @@ static void blkiolatency_enable_work_fn(struct work_struct *work)
 	 */
 	enabled = atomic_read(&blkiolat->enable_cnt);
 	if (enabled != blkiolat->enabled) {
-		blk_mq_freeze_queue(blkiolat->rqos.q);
+		blk_mq_freeze_queue(blkiolat->rqos.disk->queue);
 		blkiolat->enabled = enabled;
-		blk_mq_unfreeze_queue(blkiolat->rqos.q);
+		blk_mq_unfreeze_queue(blkiolat->rqos.disk->queue);
 	}
 }
 
diff --git a/block/blk-mq-debugfs.c b/block/blk-mq-debugfs.c
index bd942341b6382f..b01818f8e216e3 100644
--- a/block/blk-mq-debugfs.c
+++ b/block/blk-mq-debugfs.c
@@ -813,9 +813,9 @@ static const char *rq_qos_id_to_name(enum rq_qos_id id)
 
 void blk_mq_debugfs_unregister_rqos(struct rq_qos *rqos)
 {
-	lockdep_assert_held(&rqos->q->debugfs_mutex);
+	lockdep_assert_held(&rqos->disk->queue->debugfs_mutex);
 
-	if (!rqos->q->debugfs_dir)
+	if (!rqos->disk->queue->debugfs_dir)
 		return;
 	debugfs_remove_recursive(rqos->debugfs_dir);
 	rqos->debugfs_dir = NULL;
@@ -823,7 +823,7 @@ void blk_mq_debugfs_unregister_rqos(struct rq_qos *rqos)
 
 void blk_mq_debugfs_register_rqos(struct rq_qos *rqos)
 {
-	struct request_queue *q = rqos->q;
+	struct request_queue *q = rqos->disk->queue;
 	const char *dir_name = rq_qos_id_to_name(rqos->id);
 
 	lockdep_assert_held(&q->debugfs_mutex);
@@ -835,9 +835,7 @@ void blk_mq_debugfs_register_rqos(struct rq_qos *rqos)
 		q->rqos_debugfs_dir = debugfs_create_dir("rqos",
 							 q->debugfs_dir);
 
-	rqos->debugfs_dir = debugfs_create_dir(dir_name,
-					       rqos->q->rqos_debugfs_dir);
-
+	rqos->debugfs_dir = debugfs_create_dir(dir_name, q->rqos_debugfs_dir);
 	debugfs_create_files(rqos->debugfs_dir, rqos, rqos->ops->debugfs_attrs);
 }
 
diff --git a/block/blk-rq-qos.c b/block/blk-rq-qos.c
index 8e83734cfe8dbc..d8cc820a365e3a 100644
--- a/block/blk-rq-qos.c
+++ b/block/blk-rq-qos.c
@@ -300,7 +300,7 @@ int rq_qos_add(struct rq_qos *rqos, struct gendisk *disk, enum rq_qos_id id,
 {
 	struct request_queue *q = disk->queue;
 
-	rqos->q = q;
+	rqos->disk = disk;
 	rqos->id = id;
 	rqos->ops = ops;
 
@@ -337,7 +337,7 @@ int rq_qos_add(struct rq_qos *rqos, struct gendisk *disk, enum rq_qos_id id,
 
 void rq_qos_del(struct rq_qos *rqos)
 {
-	struct request_queue *q = rqos->q;
+	struct request_queue *q = rqos->disk->queue;
 	struct rq_qos **cur;
 
 	/*
diff --git a/block/blk-rq-qos.h b/block/blk-rq-qos.h
index 2b7b668479f71a..b02a1a3d33a89e 100644
--- a/block/blk-rq-qos.h
+++ b/block/blk-rq-qos.h
@@ -26,7 +26,7 @@ struct rq_wait {
 
 struct rq_qos {
 	const struct rq_qos_ops *ops;
-	struct request_queue *q;
+	struct gendisk *disk;
 	enum rq_qos_id id;
 	struct rq_qos *next;
 #ifdef CONFIG_BLK_DEBUG_FS
diff --git a/block/blk-wbt.c b/block/blk-wbt.c
index 1a78d54c8152b0..e49a4868453276 100644
--- a/block/blk-wbt.c
+++ b/block/blk-wbt.c
@@ -165,7 +165,7 @@ static void wb_timestamp(struct rq_wb *rwb, unsigned long *var)
  */
 static bool wb_recent_wait(struct rq_wb *rwb)
 {
-	struct bdi_writeback *wb = &rwb->rqos.q->disk->bdi->wb;
+	struct bdi_writeback *wb = &rwb->rqos.disk->bdi->wb;
 
 	return time_before(jiffies, wb->dirty_sleep + HZ);
 }
@@ -312,7 +312,7 @@ enum {
 
 static int latency_exceeded(struct rq_wb *rwb, struct blk_rq_stat *stat)
 {
-	struct backing_dev_info *bdi = rwb->rqos.q->disk->bdi;
+	struct backing_dev_info *bdi = rwb->rqos.disk->bdi;
 	struct rq_depth *rqd = &rwb->rq_depth;
 	u64 thislat;
 
@@ -365,7 +365,7 @@ static int latency_exceeded(struct rq_wb *rwb, struct blk_rq_stat *stat)
 
 static void rwb_trace_step(struct rq_wb *rwb, const char *msg)
 {
-	struct backing_dev_info *bdi = rwb->rqos.q->disk->bdi;
+	struct backing_dev_info *bdi = rwb->rqos.disk->bdi;
 	struct rq_depth *rqd = &rwb->rq_depth;
 
 	trace_wbt_step(bdi, msg, rqd->scale_step, rwb->cur_win_nsec,
@@ -435,13 +435,12 @@ static void wb_timer_fn(struct blk_stat_callback *cb)
 	unsigned int inflight = wbt_inflight(rwb);
 	int status;
 
-	if (!rwb->rqos.q->disk)
+	if (!rwb->rqos.disk)
 		return;
 
 	status = latency_exceeded(rwb, cb->stat);
 
-	trace_wbt_timer(rwb->rqos.q->disk->bdi, status, rqd->scale_step,
-			inflight);
+	trace_wbt_timer(rwb->rqos.disk->bdi, status, rqd->scale_step, inflight);
 
 	/*
 	 * If we exceeded the latency target, step down. If we did not,
@@ -779,16 +778,15 @@ static int wbt_data_dir(const struct request *rq)
 
 static void wbt_queue_depth_changed(struct rq_qos *rqos)
 {
-	RQWB(rqos)->rq_depth.queue_depth = blk_queue_depth(rqos->q);
+	RQWB(rqos)->rq_depth.queue_depth = blk_queue_depth(rqos->disk->queue);
 	wbt_update_limits(RQWB(rqos));
 }
 
 static void wbt_exit(struct rq_qos *rqos)
 {
 	struct rq_wb *rwb = RQWB(rqos);
-	struct request_queue *q = rqos->q;
 
-	blk_stat_remove_callback(q, rwb->cb);
+	blk_stat_remove_callback(rqos->disk->queue, rwb->cb);
 	blk_stat_free_callback(rwb->cb);
 	kfree(rwb);
 }
-- 
2.39.0

[PATCH 15/19] blk-rq-qos: store a gendisk instead of request_queue in struct rq_qos

[Christoph Hellwig]

This is what about half of the users already want, and it's only going to
grow more.

Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
Reviewed-by: Andreas Herrmann <aherrmann-l3A5Bk7waGM@public.gmane.org>
Acked-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 block/blk-iocost.c     | 12 ++++++------
 block/blk-iolatency.c  | 14 +++++++-------
 block/blk-mq-debugfs.c | 10 ++++------
 block/blk-rq-qos.c     |  4 ++--
 block/blk-rq-qos.h     |  2 +-
 block/blk-wbt.c        | 16 +++++++---------
 6 files changed, 27 insertions(+), 31 deletions(-)

diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index 6f1da7883905b3..a2e9bf30039bcd 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -669,7 +669,7 @@ static struct ioc *q_to_ioc(struct request_queue *q)
 
 static const char __maybe_unused *ioc_name(struct ioc *ioc)
 {
-	struct gendisk *disk = ioc->rqos.q->disk;
+	struct gendisk *disk = ioc->rqos.disk;
 
 	if (!disk)
 		return "<unknown>";
@@ -808,11 +808,11 @@ static int ioc_autop_idx(struct ioc *ioc)
 	u64 now_ns;
 
 	/* rotational? */
-	if (!blk_queue_nonrot(ioc->rqos.q))
+	if (!blk_queue_nonrot(ioc->rqos.disk->queue))
 		return AUTOP_HDD;
 
 	/* handle SATA SSDs w/ broken NCQ */
-	if (blk_queue_depth(ioc->rqos.q) == 1)
+	if (blk_queue_depth(ioc->rqos.disk->queue) == 1)
 		return AUTOP_SSD_QD1;
 
 	/* use one of the normal ssd sets */
@@ -2649,7 +2649,7 @@ static void ioc_rqos_throttle(struct rq_qos *rqos, struct bio *bio)
 	if (use_debt) {
 		iocg_incur_debt(iocg, abs_cost, &now);
 		if (iocg_kick_delay(iocg, &now))
-			blkcg_schedule_throttle(rqos->q->disk,
+			blkcg_schedule_throttle(rqos->disk,
 					(bio->bi_opf & REQ_SWAP) == REQ_SWAP);
 		iocg_unlock(iocg, ioc_locked, &flags);
 		return;
@@ -2750,7 +2750,7 @@ static void ioc_rqos_merge(struct rq_qos *rqos, struct request *rq,
 	if (likely(!list_empty(&iocg->active_list))) {
 		iocg_incur_debt(iocg, abs_cost, &now);
 		if (iocg_kick_delay(iocg, &now))
-			blkcg_schedule_throttle(rqos->q->disk,
+			blkcg_schedule_throttle(rqos->disk,
 					(bio->bi_opf & REQ_SWAP) == REQ_SWAP);
 	} else {
 		iocg_commit_bio(iocg, bio, abs_cost, cost);
@@ -2821,7 +2821,7 @@ static void ioc_rqos_exit(struct rq_qos *rqos)
 {
 	struct ioc *ioc = rqos_to_ioc(rqos);
 
-	blkcg_deactivate_policy(rqos->q, &blkcg_policy_iocost);
+	blkcg_deactivate_policy(rqos->disk->queue, &blkcg_policy_iocost);
 
 	spin_lock_irq(&ioc->lock);
 	ioc->running = IOC_STOP;
diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c
index f6aeb3d3fdae59..8e1e43bbde6f0b 100644
--- a/block/blk-iolatency.c
+++ b/block/blk-iolatency.c
@@ -292,7 +292,7 @@ static void __blkcg_iolatency_throttle(struct rq_qos *rqos,
 	unsigned use_delay = atomic_read(&lat_to_blkg(iolat)->use_delay);
 
 	if (use_delay)
-		blkcg_schedule_throttle(rqos->q->disk, use_memdelay);
+		blkcg_schedule_throttle(rqos->disk, use_memdelay);
 
 	/*
 	 * To avoid priority inversions we want to just take a slot if we are
@@ -330,7 +330,7 @@ static void scale_cookie_change(struct blk_iolatency *blkiolat,
 				struct child_latency_info *lat_info,
 				bool up)
 {
-	unsigned long qd = blkiolat->rqos.q->nr_requests;
+	unsigned long qd = blkiolat->rqos.disk->queue->nr_requests;
 	unsigned long scale = scale_amount(qd, up);
 	unsigned long old = atomic_read(&lat_info->scale_cookie);
 	unsigned long max_scale = qd << 1;
@@ -372,7 +372,7 @@ static void scale_cookie_change(struct blk_iolatency *blkiolat,
  */
 static void scale_change(struct iolatency_grp *iolat, bool up)
 {
-	unsigned long qd = iolat->blkiolat->rqos.q->nr_requests;
+	unsigned long qd = iolat->blkiolat->rqos.disk->queue->nr_requests;
 	unsigned long scale = scale_amount(qd, up);
 	unsigned long old = iolat->max_depth;
 
@@ -646,7 +646,7 @@ static void blkcg_iolatency_exit(struct rq_qos *rqos)
 
 	timer_shutdown_sync(&blkiolat->timer);
 	flush_work(&blkiolat->enable_work);
-	blkcg_deactivate_policy(rqos->q, &blkcg_policy_iolatency);
+	blkcg_deactivate_policy(rqos->disk->queue, &blkcg_policy_iolatency);
 	kfree(blkiolat);
 }
 
@@ -665,7 +665,7 @@ static void blkiolatency_timer_fn(struct timer_list *t)
 
 	rcu_read_lock();
 	blkg_for_each_descendant_pre(blkg, pos_css,
-				     blkiolat->rqos.q->root_blkg) {
+				     blkiolat->rqos.disk->queue->root_blkg) {
 		struct iolatency_grp *iolat;
 		struct child_latency_info *lat_info;
 		unsigned long flags;
@@ -749,9 +749,9 @@ static void blkiolatency_enable_work_fn(struct work_struct *work)
 	 */
 	enabled = atomic_read(&blkiolat->enable_cnt);
 	if (enabled != blkiolat->enabled) {
-		blk_mq_freeze_queue(blkiolat->rqos.q);
+		blk_mq_freeze_queue(blkiolat->rqos.disk->queue);
 		blkiolat->enabled = enabled;
-		blk_mq_unfreeze_queue(blkiolat->rqos.q);
+		blk_mq_unfreeze_queue(blkiolat->rqos.disk->queue);
 	}
 }
 
diff --git a/block/blk-mq-debugfs.c b/block/blk-mq-debugfs.c
index bd942341b6382f..b01818f8e216e3 100644
--- a/block/blk-mq-debugfs.c
+++ b/block/blk-mq-debugfs.c
@@ -813,9 +813,9 @@ static const char *rq_qos_id_to_name(enum rq_qos_id id)
 
 void blk_mq_debugfs_unregister_rqos(struct rq_qos *rqos)
 {
-	lockdep_assert_held(&rqos->q->debugfs_mutex);
+	lockdep_assert_held(&rqos->disk->queue->debugfs_mutex);
 
-	if (!rqos->q->debugfs_dir)
+	if (!rqos->disk->queue->debugfs_dir)
 		return;
 	debugfs_remove_recursive(rqos->debugfs_dir);
 	rqos->debugfs_dir = NULL;
@@ -823,7 +823,7 @@ void blk_mq_debugfs_unregister_rqos(struct rq_qos *rqos)
 
 void blk_mq_debugfs_register_rqos(struct rq_qos *rqos)
 {
-	struct request_queue *q = rqos->q;
+	struct request_queue *q = rqos->disk->queue;
 	const char *dir_name = rq_qos_id_to_name(rqos->id);
 
 	lockdep_assert_held(&q->debugfs_mutex);
@@ -835,9 +835,7 @@ void blk_mq_debugfs_register_rqos(struct rq_qos *rqos)
 		q->rqos_debugfs_dir = debugfs_create_dir("rqos",
 							 q->debugfs_dir);
 
-	rqos->debugfs_dir = debugfs_create_dir(dir_name,
-					       rqos->q->rqos_debugfs_dir);
-
+	rqos->debugfs_dir = debugfs_create_dir(dir_name, q->rqos_debugfs_dir);
 	debugfs_create_files(rqos->debugfs_dir, rqos, rqos->ops->debugfs_attrs);
 }
 
diff --git a/block/blk-rq-qos.c b/block/blk-rq-qos.c
index 8e83734cfe8dbc..d8cc820a365e3a 100644
--- a/block/blk-rq-qos.c
+++ b/block/blk-rq-qos.c
@@ -300,7 +300,7 @@ int rq_qos_add(struct rq_qos *rqos, struct gendisk *disk, enum rq_qos_id id,
 {
 	struct request_queue *q = disk->queue;
 
-	rqos->q = q;
+	rqos->disk = disk;
 	rqos->id = id;
 	rqos->ops = ops;
 
@@ -337,7 +337,7 @@ int rq_qos_add(struct rq_qos *rqos, struct gendisk *disk, enum rq_qos_id id,
 
 void rq_qos_del(struct rq_qos *rqos)
 {
-	struct request_queue *q = rqos->q;
+	struct request_queue *q = rqos->disk->queue;
 	struct rq_qos **cur;
 
 	/*
diff --git a/block/blk-rq-qos.h b/block/blk-rq-qos.h
index 2b7b668479f71a..b02a1a3d33a89e 100644
--- a/block/blk-rq-qos.h
+++ b/block/blk-rq-qos.h
@@ -26,7 +26,7 @@ struct rq_wait {
 
 struct rq_qos {
 	const struct rq_qos_ops *ops;
-	struct request_queue *q;
+	struct gendisk *disk;
 	enum rq_qos_id id;
 	struct rq_qos *next;
 #ifdef CONFIG_BLK_DEBUG_FS
diff --git a/block/blk-wbt.c b/block/blk-wbt.c
index 1a78d54c8152b0..e49a4868453276 100644
--- a/block/blk-wbt.c
+++ b/block/blk-wbt.c
@@ -165,7 +165,7 @@ static void wb_timestamp(struct rq_wb *rwb, unsigned long *var)
  */
 static bool wb_recent_wait(struct rq_wb *rwb)
 {
-	struct bdi_writeback *wb = &rwb->rqos.q->disk->bdi->wb;
+	struct bdi_writeback *wb = &rwb->rqos.disk->bdi->wb;
 
 	return time_before(jiffies, wb->dirty_sleep + HZ);
 }
@@ -312,7 +312,7 @@ enum {
 
 static int latency_exceeded(struct rq_wb *rwb, struct blk_rq_stat *stat)
 {
-	struct backing_dev_info *bdi = rwb->rqos.q->disk->bdi;
+	struct backing_dev_info *bdi = rwb->rqos.disk->bdi;
 	struct rq_depth *rqd = &rwb->rq_depth;
 	u64 thislat;
 
@@ -365,7 +365,7 @@ static int latency_exceeded(struct rq_wb *rwb, struct blk_rq_stat *stat)
 
 static void rwb_trace_step(struct rq_wb *rwb, const char *msg)
 {
-	struct backing_dev_info *bdi = rwb->rqos.q->disk->bdi;
+	struct backing_dev_info *bdi = rwb->rqos.disk->bdi;
 	struct rq_depth *rqd = &rwb->rq_depth;
 
 	trace_wbt_step(bdi, msg, rqd->scale_step, rwb->cur_win_nsec,
@@ -435,13 +435,12 @@ static void wb_timer_fn(struct blk_stat_callback *cb)
 	unsigned int inflight = wbt_inflight(rwb);
 	int status;
 
-	if (!rwb->rqos.q->disk)
+	if (!rwb->rqos.disk)
 		return;
 
 	status = latency_exceeded(rwb, cb->stat);
 
-	trace_wbt_timer(rwb->rqos.q->disk->bdi, status, rqd->scale_step,
-			inflight);
+	trace_wbt_timer(rwb->rqos.disk->bdi, status, rqd->scale_step, inflight);
 
 	/*
 	 * If we exceeded the latency target, step down. If we did not,
@@ -779,16 +778,15 @@ static int wbt_data_dir(const struct request *rq)
 
 static void wbt_queue_depth_changed(struct rq_qos *rqos)
 {
-	RQWB(rqos)->rq_depth.queue_depth = blk_queue_depth(rqos->q);
+	RQWB(rqos)->rq_depth.queue_depth = blk_queue_depth(rqos->disk->queue);
 	wbt_update_limits(RQWB(rqos));
 }
 
 static void wbt_exit(struct rq_qos *rqos)
 {
 	struct rq_wb *rwb = RQWB(rqos);
-	struct request_queue *q = rqos->q;
 
-	blk_stat_remove_callback(q, rwb->cb);
+	blk_stat_remove_callback(rqos->disk->queue, rwb->cb);
 	blk_stat_free_callback(rwb->cb);
 	kfree(rwb);
 }
-- 
2.39.0

[PATCH 16/19] blk-cgroup: pass a gendisk to blkcg_{de,}activate_policy

[Christoph Hellwig]

Prepare for storing the blkcg information in the gendisk instead of
the request_queue.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/bfq-cgroup.c    |  2 +-
 block/bfq-iosched.c   |  2 +-
 block/blk-cgroup.c    | 21 +++++++++++----------
 block/blk-cgroup.h    |  9 ++++-----
 block/blk-iocost.c    |  4 ++--
 block/blk-iolatency.c |  4 ++--
 block/blk-ioprio.c    |  4 ++--
 block/blk-throttle.c  |  4 ++--
 8 files changed, 25 insertions(+), 25 deletions(-)

diff --git a/block/bfq-cgroup.c b/block/bfq-cgroup.c
index 1d4a3f15049bda..032c14f0451a1d 100644
--- a/block/bfq-cgroup.c
+++ b/block/bfq-cgroup.c
@@ -1291,7 +1291,7 @@ struct bfq_group *bfq_create_group_hierarchy(struct bfq_data *bfqd, int node)
 {
 	int ret;
 
-	ret = blkcg_activate_policy(bfqd->queue, &blkcg_policy_bfq);
+	ret = blkcg_activate_policy(bfqd->queue->disk, &blkcg_policy_bfq);
 	if (ret)
 		return NULL;
 
diff --git a/block/bfq-iosched.c b/block/bfq-iosched.c
index 5afa661fa2ea4c..777dcab73c8e28 100644
--- a/block/bfq-iosched.c
+++ b/block/bfq-iosched.c
@@ -7146,7 +7146,7 @@ static void bfq_exit_queue(struct elevator_queue *e)
 	bfqg_and_blkg_put(bfqd->root_group);
 
 #ifdef CONFIG_BFQ_GROUP_IOSCHED
-	blkcg_deactivate_policy(bfqd->queue, &blkcg_policy_bfq);
+	blkcg_deactivate_policy(bfqd->queue->disk, &blkcg_policy_bfq);
 #else
 	spin_lock_irq(&bfqd->lock);
 	bfq_put_async_queues(bfqd, bfqd->root_group);
diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 168b2f803238f9..c20929bce8125c 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -1390,14 +1390,14 @@ struct cgroup_subsys io_cgrp_subsys = {
 EXPORT_SYMBOL_GPL(io_cgrp_subsys);
 
 /**
- * blkcg_activate_policy - activate a blkcg policy on a request_queue
- * @q: request_queue of interest
+ * blkcg_activate_policy - activate a blkcg policy on a gendisk
+ * @disk: gendisk of interest
  * @pol: blkcg policy to activate
  *
- * Activate @pol on @q.  Requires %GFP_KERNEL context.  @q goes through
+ * Activate @pol on @disk.  Requires %GFP_KERNEL context.  @disk goes through
  * bypass mode to populate its blkgs with policy_data for @pol.
  *
- * Activation happens with @q bypassed, so nobody would be accessing blkgs
+ * Activation happens with @disk bypassed, so nobody would be accessing blkgs
  * from IO path.  Update of each blkg is protected by both queue and blkcg
  * locks so that holding either lock and testing blkcg_policy_enabled() is
  * always enough for dereferencing policy data.
@@ -1405,9 +1405,9 @@ EXPORT_SYMBOL_GPL(io_cgrp_subsys);
  * The caller is responsible for synchronizing [de]activations and policy
  * [un]registerations.  Returns 0 on success, -errno on failure.
  */
-int blkcg_activate_policy(struct request_queue *q,
-			  const struct blkcg_policy *pol)
+int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol)
 {
+	struct request_queue *q = disk->queue;
 	struct blkg_policy_data *pd_prealloc = NULL;
 	struct blkcg_gq *blkg, *pinned_blkg = NULL;
 	int ret;
@@ -1508,16 +1508,17 @@ int blkcg_activate_policy(struct request_queue *q,
 EXPORT_SYMBOL_GPL(blkcg_activate_policy);
 
 /**
- * blkcg_deactivate_policy - deactivate a blkcg policy on a request_queue
- * @q: request_queue of interest
+ * blkcg_deactivate_policy - deactivate a blkcg policy on a gendisk
+ * @disk: gendisk of interest
  * @pol: blkcg policy to deactivate
  *
- * Deactivate @pol on @q.  Follows the same synchronization rules as
+ * Deactivate @pol on @disk.  Follows the same synchronization rules as
  * blkcg_activate_policy().
  */
-void blkcg_deactivate_policy(struct request_queue *q,
+void blkcg_deactivate_policy(struct gendisk *disk,
 			     const struct blkcg_policy *pol)
 {
+	struct request_queue *q = disk->queue;
 	struct blkcg_gq *blkg;
 
 	if (!blkcg_policy_enabled(q, pol))
diff --git a/block/blk-cgroup.h b/block/blk-cgroup.h
index 996572a9a0b72e..27068faa2cd086 100644
--- a/block/blk-cgroup.h
+++ b/block/blk-cgroup.h
@@ -194,9 +194,8 @@ void blkcg_exit_disk(struct gendisk *disk);
 /* Blkio controller policy registration */
 int blkcg_policy_register(struct blkcg_policy *pol);
 void blkcg_policy_unregister(struct blkcg_policy *pol);
-int blkcg_activate_policy(struct request_queue *q,
-			  const struct blkcg_policy *pol);
-void blkcg_deactivate_policy(struct request_queue *q,
+int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol);
+void blkcg_deactivate_policy(struct gendisk *disk,
 			     const struct blkcg_policy *pol);
 
 const char *blkg_dev_name(struct blkcg_gq *blkg);
@@ -495,9 +494,9 @@ static inline int blkcg_init_disk(struct gendisk *disk) { return 0; }
 static inline void blkcg_exit_disk(struct gendisk *disk) { }
 static inline int blkcg_policy_register(struct blkcg_policy *pol) { return 0; }
 static inline void blkcg_policy_unregister(struct blkcg_policy *pol) { }
-static inline int blkcg_activate_policy(struct request_queue *q,
+static inline int blkcg_activate_policy(struct gendisk *disk,
 					const struct blkcg_policy *pol) { return 0; }
-static inline void blkcg_deactivate_policy(struct request_queue *q,
+static inline void blkcg_deactivate_policy(struct gendisk *disk,
 					   const struct blkcg_policy *pol) { }
 
 static inline struct blkg_policy_data *blkg_to_pd(struct blkcg_gq *blkg,
diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index a2e9bf30039bcd..078b7770951962 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -2821,7 +2821,7 @@ static void ioc_rqos_exit(struct rq_qos *rqos)
 {
 	struct ioc *ioc = rqos_to_ioc(rqos);
 
-	blkcg_deactivate_policy(rqos->disk->queue, &blkcg_policy_iocost);
+	blkcg_deactivate_policy(rqos->disk, &blkcg_policy_iocost);
 
 	spin_lock_irq(&ioc->lock);
 	ioc->running = IOC_STOP;
@@ -2893,7 +2893,7 @@ static int blk_iocost_init(struct gendisk *disk)
 	if (ret)
 		goto err_free_ioc;
 
-	ret = blkcg_activate_policy(disk->queue, &blkcg_policy_iocost);
+	ret = blkcg_activate_policy(disk, &blkcg_policy_iocost);
 	if (ret)
 		goto err_del_qos;
 	return 0;
diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c
index 8e1e43bbde6f0b..39853fc5c2b02f 100644
--- a/block/blk-iolatency.c
+++ b/block/blk-iolatency.c
@@ -646,7 +646,7 @@ static void blkcg_iolatency_exit(struct rq_qos *rqos)
 
 	timer_shutdown_sync(&blkiolat->timer);
 	flush_work(&blkiolat->enable_work);
-	blkcg_deactivate_policy(rqos->disk->queue, &blkcg_policy_iolatency);
+	blkcg_deactivate_policy(rqos->disk, &blkcg_policy_iolatency);
 	kfree(blkiolat);
 }
 
@@ -768,7 +768,7 @@ int blk_iolatency_init(struct gendisk *disk)
 			 &blkcg_iolatency_ops);
 	if (ret)
 		goto err_free;
-	ret = blkcg_activate_policy(disk->queue, &blkcg_policy_iolatency);
+	ret = blkcg_activate_policy(disk, &blkcg_policy_iolatency);
 	if (ret)
 		goto err_qos_del;
 
diff --git a/block/blk-ioprio.c b/block/blk-ioprio.c
index 8bb6b8eba4cee8..8194826cc824bc 100644
--- a/block/blk-ioprio.c
+++ b/block/blk-ioprio.c
@@ -204,12 +204,12 @@ void blkcg_set_ioprio(struct bio *bio)
 
 void blk_ioprio_exit(struct gendisk *disk)
 {
-	blkcg_deactivate_policy(disk->queue, &ioprio_policy);
+	blkcg_deactivate_policy(disk, &ioprio_policy);
 }
 
 int blk_ioprio_init(struct gendisk *disk)
 {
-	return blkcg_activate_policy(disk->queue, &ioprio_policy);
+	return blkcg_activate_policy(disk, &ioprio_policy);
 }
 
 static int __init ioprio_init(void)
diff --git a/block/blk-throttle.c b/block/blk-throttle.c
index f802d8f9099430..efc0a9092c6942 100644
--- a/block/blk-throttle.c
+++ b/block/blk-throttle.c
@@ -2395,7 +2395,7 @@ int blk_throtl_init(struct gendisk *disk)
 	td->low_downgrade_time = jiffies;
 
 	/* activate policy */
-	ret = blkcg_activate_policy(q, &blkcg_policy_throtl);
+	ret = blkcg_activate_policy(disk, &blkcg_policy_throtl);
 	if (ret) {
 		free_percpu(td->latency_buckets[READ]);
 		free_percpu(td->latency_buckets[WRITE]);
@@ -2411,7 +2411,7 @@ void blk_throtl_exit(struct gendisk *disk)
 	BUG_ON(!q->td);
 	del_timer_sync(&q->td->service_queue.pending_timer);
 	throtl_shutdown_wq(q);
-	blkcg_deactivate_policy(q, &blkcg_policy_throtl);
+	blkcg_deactivate_policy(disk, &blkcg_policy_throtl);
 	free_percpu(q->td->latency_buckets[READ]);
 	free_percpu(q->td->latency_buckets[WRITE]);
 	kfree(q->td);
-- 
2.39.0

[PATCH 17/19] blk-cgroup: pass a gendisk to pd_alloc_fn

[Christoph Hellwig]

No need to the request_queue here, pass a gendisk and extract the
node ids from that.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/bfq-cgroup.c    |  6 +++---
 block/blk-cgroup.c    | 10 +++++-----
 block/blk-cgroup.h    |  4 ++--
 block/blk-iocost.c    |  7 ++++---
 block/blk-iolatency.c |  7 +++----
 block/blk-ioprio.c    |  2 +-
 block/blk-throttle.c  |  7 +++----
 7 files changed, 21 insertions(+), 22 deletions(-)

diff --git a/block/bfq-cgroup.c b/block/bfq-cgroup.c
index 032c14f0451a1d..37333c164ed458 100644
--- a/block/bfq-cgroup.c
+++ b/block/bfq-cgroup.c
@@ -513,12 +513,12 @@ 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, struct request_queue *q,
-					     struct blkcg *blkcg)
+static struct blkg_policy_data *bfq_pd_alloc(struct gendisk *disk,
+		struct blkcg *blkcg, gfp_t gfp)
 {
 	struct bfq_group *bfqg;
 
-	bfqg = kzalloc_node(sizeof(*bfqg), gfp, q->node);
+	bfqg = kzalloc_node(sizeof(*bfqg), gfp, disk->node_id);
 	if (!bfqg)
 		return NULL;
 
diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index c20929bce8125c..fc264b15588210 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -276,7 +276,7 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct gendisk *disk,
 			continue;
 
 		/* alloc per-policy data and attach it to blkg */
-		pd = pol->pd_alloc_fn(gfp_mask, disk->queue, blkcg);
+		pd = pol->pd_alloc_fn(disk, blkcg, gfp_mask);
 		if (!pd)
 			goto out_free_pds;
 		blkg->pd[i] = pd;
@@ -1432,8 +1432,8 @@ int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol)
 			pd = pd_prealloc;
 			pd_prealloc = NULL;
 		} else {
-			pd = pol->pd_alloc_fn(GFP_NOWAIT | __GFP_NOWARN, q,
-					      blkg->blkcg);
+			pd = pol->pd_alloc_fn(disk, blkg->blkcg,
+					      GFP_NOWAIT | __GFP_NOWARN);
 		}
 
 		if (!pd) {
@@ -1450,8 +1450,8 @@ int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol)
 
 			if (pd_prealloc)
 				pol->pd_free_fn(pd_prealloc);
-			pd_prealloc = pol->pd_alloc_fn(GFP_KERNEL, q,
-						       blkg->blkcg);
+			pd_prealloc = pol->pd_alloc_fn(disk, blkg->blkcg,
+						       GFP_KERNEL);
 			if (pd_prealloc)
 				goto retry;
 			else
diff --git a/block/blk-cgroup.h b/block/blk-cgroup.h
index 27068faa2cd086..3d9e42c519db86 100644
--- a/block/blk-cgroup.h
+++ b/block/blk-cgroup.h
@@ -154,8 +154,8 @@ 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,
-				struct request_queue *q, struct blkcg *blkcg);
+typedef struct blkg_policy_data *(blkcg_pol_alloc_pd_fn)(struct gendisk *disk,
+		struct blkcg *blkcg, gfp_t gfp);
 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);
diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index 078b7770951962..7a2dc9dc8e3ba0 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -2923,13 +2923,14 @@ 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)
+static struct blkg_policy_data *ioc_pd_alloc(struct gendisk *disk,
+		struct blkcg *blkcg, gfp_t gfp)
 {
 	int levels = blkcg->css.cgroup->level + 1;
 	struct ioc_gq *iocg;
 
-	iocg = kzalloc_node(struct_size(iocg, ancestors, levels), gfp, q->node);
+	iocg = kzalloc_node(struct_size(iocg, ancestors, levels), gfp,
+			    disk->node_id);
 	if (!iocg)
 		return NULL;
 
diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c
index 39853fc5c2b02f..bc0d217f5c1723 100644
--- a/block/blk-iolatency.c
+++ b/block/blk-iolatency.c
@@ -946,13 +946,12 @@ static void iolatency_pd_stat(struct blkg_policy_data *pd, struct seq_file *s)
 			iolat->max_depth, avg_lat, cur_win);
 }
 
-static struct blkg_policy_data *iolatency_pd_alloc(gfp_t gfp,
-						   struct request_queue *q,
-						   struct blkcg *blkcg)
+static struct blkg_policy_data *iolatency_pd_alloc(struct gendisk *disk,
+		struct blkcg *blkcg, gfp_t gfp)
 {
 	struct iolatency_grp *iolat;
 
-	iolat = kzalloc_node(sizeof(*iolat), gfp, q->node);
+	iolat = kzalloc_node(sizeof(*iolat), gfp, disk->node_id);
 	if (!iolat)
 		return NULL;
 	iolat->stats = __alloc_percpu_gfp(sizeof(struct latency_stat),
diff --git a/block/blk-ioprio.c b/block/blk-ioprio.c
index 8194826cc824bc..055529b9b92bab 100644
--- a/block/blk-ioprio.c
+++ b/block/blk-ioprio.c
@@ -116,7 +116,7 @@ static ssize_t ioprio_set_prio_policy(struct kernfs_open_file *of, char *buf,
 }
 
 static struct blkg_policy_data *
-ioprio_alloc_pd(gfp_t gfp, struct request_queue *q, struct blkcg *blkcg)
+ioprio_alloc_pd(struct gendisk *disk, struct blkcg *blkcg, gfp_t gfp)
 {
 	struct ioprio_blkg *ioprio_blkg;
 
diff --git a/block/blk-throttle.c b/block/blk-throttle.c
index efc0a9092c6942..74bb1e753ea09d 100644
--- a/block/blk-throttle.c
+++ b/block/blk-throttle.c
@@ -335,14 +335,13 @@ 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,
-						struct request_queue *q,
-						struct blkcg *blkcg)
+static struct blkg_policy_data *throtl_pd_alloc(struct gendisk *disk,
+		struct blkcg *blkcg, gfp_t gfp)
 {
 	struct throtl_grp *tg;
 	int rw;
 
-	tg = kzalloc_node(sizeof(*tg), gfp, q->node);
+	tg = kzalloc_node(sizeof(*tg), gfp, disk->node_id);
 	if (!tg)
 		return NULL;
 
-- 
2.39.0

[PATCH 17/19] blk-cgroup: pass a gendisk to pd_alloc_fn

[Christoph Hellwig]

No need to the request_queue here, pass a gendisk and extract the
node ids from that.

Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
Reviewed-by: Andreas Herrmann <aherrmann-l3A5Bk7waGM@public.gmane.org>
Acked-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 block/bfq-cgroup.c    |  6 +++---
 block/blk-cgroup.c    | 10 +++++-----
 block/blk-cgroup.h    |  4 ++--
 block/blk-iocost.c    |  7 ++++---
 block/blk-iolatency.c |  7 +++----
 block/blk-ioprio.c    |  2 +-
 block/blk-throttle.c  |  7 +++----
 7 files changed, 21 insertions(+), 22 deletions(-)

diff --git a/block/bfq-cgroup.c b/block/bfq-cgroup.c
index 032c14f0451a1d..37333c164ed458 100644
--- a/block/bfq-cgroup.c
+++ b/block/bfq-cgroup.c
@@ -513,12 +513,12 @@ 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, struct request_queue *q,
-					     struct blkcg *blkcg)
+static struct blkg_policy_data *bfq_pd_alloc(struct gendisk *disk,
+		struct blkcg *blkcg, gfp_t gfp)
 {
 	struct bfq_group *bfqg;
 
-	bfqg = kzalloc_node(sizeof(*bfqg), gfp, q->node);
+	bfqg = kzalloc_node(sizeof(*bfqg), gfp, disk->node_id);
 	if (!bfqg)
 		return NULL;
 
diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index c20929bce8125c..fc264b15588210 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -276,7 +276,7 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct gendisk *disk,
 			continue;
 
 		/* alloc per-policy data and attach it to blkg */
-		pd = pol->pd_alloc_fn(gfp_mask, disk->queue, blkcg);
+		pd = pol->pd_alloc_fn(disk, blkcg, gfp_mask);
 		if (!pd)
 			goto out_free_pds;
 		blkg->pd[i] = pd;
@@ -1432,8 +1432,8 @@ int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol)
 			pd = pd_prealloc;
 			pd_prealloc = NULL;
 		} else {
-			pd = pol->pd_alloc_fn(GFP_NOWAIT | __GFP_NOWARN, q,
-					      blkg->blkcg);
+			pd = pol->pd_alloc_fn(disk, blkg->blkcg,
+					      GFP_NOWAIT | __GFP_NOWARN);
 		}
 
 		if (!pd) {
@@ -1450,8 +1450,8 @@ int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol)
 
 			if (pd_prealloc)
 				pol->pd_free_fn(pd_prealloc);
-			pd_prealloc = pol->pd_alloc_fn(GFP_KERNEL, q,
-						       blkg->blkcg);
+			pd_prealloc = pol->pd_alloc_fn(disk, blkg->blkcg,
+						       GFP_KERNEL);
 			if (pd_prealloc)
 				goto retry;
 			else
diff --git a/block/blk-cgroup.h b/block/blk-cgroup.h
index 27068faa2cd086..3d9e42c519db86 100644
--- a/block/blk-cgroup.h
+++ b/block/blk-cgroup.h
@@ -154,8 +154,8 @@ 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,
-				struct request_queue *q, struct blkcg *blkcg);
+typedef struct blkg_policy_data *(blkcg_pol_alloc_pd_fn)(struct gendisk *disk,
+		struct blkcg *blkcg, gfp_t gfp);
 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);
diff --git a/block/blk-iocost.c b/block/blk-iocost.c
index 078b7770951962..7a2dc9dc8e3ba0 100644
--- a/block/blk-iocost.c
+++ b/block/blk-iocost.c
@@ -2923,13 +2923,14 @@ 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)
+static struct blkg_policy_data *ioc_pd_alloc(struct gendisk *disk,
+		struct blkcg *blkcg, gfp_t gfp)
 {
 	int levels = blkcg->css.cgroup->level + 1;
 	struct ioc_gq *iocg;
 
-	iocg = kzalloc_node(struct_size(iocg, ancestors, levels), gfp, q->node);
+	iocg = kzalloc_node(struct_size(iocg, ancestors, levels), gfp,
+			    disk->node_id);
 	if (!iocg)
 		return NULL;
 
diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c
index 39853fc5c2b02f..bc0d217f5c1723 100644
--- a/block/blk-iolatency.c
+++ b/block/blk-iolatency.c
@@ -946,13 +946,12 @@ static void iolatency_pd_stat(struct blkg_policy_data *pd, struct seq_file *s)
 			iolat->max_depth, avg_lat, cur_win);
 }
 
-static struct blkg_policy_data *iolatency_pd_alloc(gfp_t gfp,
-						   struct request_queue *q,
-						   struct blkcg *blkcg)
+static struct blkg_policy_data *iolatency_pd_alloc(struct gendisk *disk,
+		struct blkcg *blkcg, gfp_t gfp)
 {
 	struct iolatency_grp *iolat;
 
-	iolat = kzalloc_node(sizeof(*iolat), gfp, q->node);
+	iolat = kzalloc_node(sizeof(*iolat), gfp, disk->node_id);
 	if (!iolat)
 		return NULL;
 	iolat->stats = __alloc_percpu_gfp(sizeof(struct latency_stat),
diff --git a/block/blk-ioprio.c b/block/blk-ioprio.c
index 8194826cc824bc..055529b9b92bab 100644
--- a/block/blk-ioprio.c
+++ b/block/blk-ioprio.c
@@ -116,7 +116,7 @@ static ssize_t ioprio_set_prio_policy(struct kernfs_open_file *of, char *buf,
 }
 
 static struct blkg_policy_data *
-ioprio_alloc_pd(gfp_t gfp, struct request_queue *q, struct blkcg *blkcg)
+ioprio_alloc_pd(struct gendisk *disk, struct blkcg *blkcg, gfp_t gfp)
 {
 	struct ioprio_blkg *ioprio_blkg;
 
diff --git a/block/blk-throttle.c b/block/blk-throttle.c
index efc0a9092c6942..74bb1e753ea09d 100644
--- a/block/blk-throttle.c
+++ b/block/blk-throttle.c
@@ -335,14 +335,13 @@ 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,
-						struct request_queue *q,
-						struct blkcg *blkcg)
+static struct blkg_policy_data *throtl_pd_alloc(struct gendisk *disk,
+		struct blkcg *blkcg, gfp_t gfp)
 {
 	struct throtl_grp *tg;
 	int rw;
 
-	tg = kzalloc_node(sizeof(*tg), gfp, q->node);
+	tg = kzalloc_node(sizeof(*tg), gfp, disk->node_id);
 	if (!tg)
 		return NULL;
 
-- 
2.39.0

[PATCH 18/19] blk-cgroup: pass a gendisk to blkg_lookup

[Christoph Hellwig]

Pass a gendisk to blkg_lookup and use that to find the match as part
of phasing out usage of the request_queue in the blk-cgroup code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/blk-cgroup.c | 16 ++++++++--------
 block/blk-cgroup.h | 20 ++++++++++----------
 2 files changed, 18 insertions(+), 18 deletions(-)

diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index fc264b15588210..45a683e88bcac1 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -337,7 +337,7 @@ static struct blkcg_gq *blkg_create(struct blkcg *blkcg, struct gendisk *disk,
 
 	/* link parent */
 	if (blkcg_parent(blkcg)) {
-		blkg->parent = blkg_lookup(blkcg_parent(blkcg), disk->queue);
+		blkg->parent = blkg_lookup(blkcg_parent(blkcg), disk);
 		if (WARN_ON_ONCE(!blkg->parent)) {
 			ret = -ENODEV;
 			goto err_put_css;
@@ -409,12 +409,12 @@ static struct blkcg_gq *blkg_lookup_create(struct blkcg *blkcg,
 
 	WARN_ON_ONCE(!rcu_read_lock_held());
 
-	blkg = blkg_lookup(blkcg, q);
+	blkg = blkg_lookup(blkcg, disk);
 	if (blkg)
 		return blkg;
 
 	spin_lock_irqsave(&q->queue_lock, flags);
-	blkg = blkg_lookup(blkcg, q);
+	blkg = blkg_lookup(blkcg, disk);
 	if (blkg) {
 		if (blkcg != &blkcg_root &&
 		    blkg != rcu_dereference(blkcg->blkg_hint))
@@ -433,7 +433,7 @@ static struct blkcg_gq *blkg_lookup_create(struct blkcg *blkcg,
 		struct blkcg_gq *ret_blkg = q->root_blkg;
 
 		while (parent) {
-			blkg = blkg_lookup(parent, q);
+			blkg = blkg_lookup(parent, disk);
 			if (blkg) {
 				/* remember closest blkg */
 				ret_blkg = blkg;
@@ -719,7 +719,7 @@ int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
 		goto fail_unlock;
 	}
 
-	blkg = blkg_lookup(blkcg, q);
+	blkg = blkg_lookup(blkcg, disk);
 	if (blkg)
 		goto success;
 
@@ -733,7 +733,7 @@ int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
 		struct blkcg_gq *new_blkg;
 
 		parent = blkcg_parent(blkcg);
-		while (parent && !blkg_lookup(parent, q)) {
+		while (parent && !blkg_lookup(parent, disk)) {
 			pos = parent;
 			parent = blkcg_parent(parent);
 		}
@@ -763,7 +763,7 @@ int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
 			goto fail_preloaded;
 		}
 
-		blkg = blkg_lookup(pos, q);
+		blkg = blkg_lookup(pos, disk);
 		if (blkg) {
 			blkg_free(new_blkg);
 		} else {
@@ -1838,7 +1838,7 @@ void blkcg_maybe_throttle_current(void)
 	blkcg = css_to_blkcg(blkcg_css());
 	if (!blkcg)
 		goto out;
-	blkg = blkg_lookup(blkcg, disk->queue);
+	blkg = blkg_lookup(blkcg, disk);
 	if (!blkg)
 		goto out;
 	if (!blkg_tryget(blkg))
diff --git a/block/blk-cgroup.h b/block/blk-cgroup.h
index 3d9e42c519db86..151f24de253985 100644
--- a/block/blk-cgroup.h
+++ b/block/blk-cgroup.h
@@ -234,30 +234,30 @@ static inline bool bio_issue_as_root_blkg(struct bio *bio)
 }
 
 /**
- * blkg_lookup - lookup blkg for the specified blkcg - q pair
+ * blkg_lookup - lookup blkg for the specified blkcg - disk pair
  * @blkcg: blkcg of interest
- * @q: request_queue of interest
+ * @disk: gendisk of interest
  *
- * Lookup blkg for the @blkcg - @q pair.
+ * Lookup blkg for the @blkcg - @disk pair.
 
  * Must be called in a RCU critical section.
  */
 static inline struct blkcg_gq *blkg_lookup(struct blkcg *blkcg,
-					   struct request_queue *q)
+					   struct gendisk *disk)
 {
 	struct blkcg_gq *blkg;
 
 	WARN_ON_ONCE(!rcu_read_lock_held());
 
 	if (blkcg == &blkcg_root)
-		return q->root_blkg;
+		return disk->queue->root_blkg;
 
 	blkg = rcu_dereference(blkcg->blkg_hint);
-	if (blkg && blkg->disk->queue == q)
+	if (blkg && blkg->disk == disk)
 		return blkg;
 
-	blkg = radix_tree_lookup(&blkcg->blkg_tree, q->id);
-	if (blkg && blkg->disk->queue != q)
+	blkg = radix_tree_lookup(&blkcg->blkg_tree, disk->queue->id);
+	if (blkg && blkg->disk != disk)
 		blkg = NULL;
 	return blkg;
 }
@@ -357,7 +357,7 @@ static inline void blkg_put(struct blkcg_gq *blkg)
 #define blkg_for_each_descendant_pre(d_blkg, pos_css, p_blkg)		\
 	css_for_each_descendant_pre((pos_css), &(p_blkg)->blkcg->css)	\
 		if (((d_blkg) = blkg_lookup(css_to_blkcg(pos_css),	\
-					    (p_blkg)->disk->queue)))
+					    (p_blkg)->disk)))
 
 /**
  * blkg_for_each_descendant_post - post-order walk of a blkg's descendants
@@ -372,7 +372,7 @@ static inline void blkg_put(struct blkcg_gq *blkg)
 #define blkg_for_each_descendant_post(d_blkg, pos_css, p_blkg)		\
 	css_for_each_descendant_post((pos_css), &(p_blkg)->blkcg->css)	\
 		if (((d_blkg) = blkg_lookup(css_to_blkcg(pos_css),	\
-					    (p_blkg)->disk->queue)))
+					    (p_blkg)->disk)))
 
 bool __blkcg_punt_bio_submit(struct bio *bio);
 
-- 
2.39.0

[PATCH 19/19] blk-cgroup: move the cgroup information to struct gendisk

[Christoph Hellwig]

cgroup information only makes sense on a live gendisk that allows
file system I/O (which includes the raw block device).  So move over
the cgroup related members.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
---
 block/bfq-cgroup.c     |  4 +--
 block/blk-cgroup.c     | 66 +++++++++++++++++++++---------------------
 block/blk-cgroup.h     |  4 +--
 block/blk-iolatency.c  |  2 +-
 block/blk-throttle.c   | 16 ++++++----
 include/linux/blkdev.h | 12 ++++----
 6 files changed, 54 insertions(+), 50 deletions(-)

diff --git a/block/bfq-cgroup.c b/block/bfq-cgroup.c
index 37333c164ed458..4fdbbec71647f9 100644
--- a/block/bfq-cgroup.c
+++ b/block/bfq-cgroup.c
@@ -1001,7 +1001,7 @@ void bfq_end_wr_async(struct bfq_data *bfqd)
 {
 	struct blkcg_gq *blkg;
 
-	list_for_each_entry(blkg, &bfqd->queue->blkg_list, q_node) {
+	list_for_each_entry(blkg, &bfqd->queue->disk->blkg_list, entry) {
 		struct bfq_group *bfqg = blkg_to_bfqg(blkg);
 
 		bfq_end_wr_async_queues(bfqd, bfqg);
@@ -1295,7 +1295,7 @@ struct bfq_group *bfq_create_group_hierarchy(struct bfq_data *bfqd, int node)
 	if (ret)
 		return NULL;
 
-	return blkg_to_bfqg(bfqd->queue->root_blkg);
+	return blkg_to_bfqg(bfqd->queue->disk->root_blkg);
 }
 
 struct blkcg_policy blkcg_policy_bfq = {
diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 45a683e88bcac1..8faeca6022bea0 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -108,10 +108,10 @@ static struct cgroup_subsys_state *blkcg_css(void)
 	return task_css(current, io_cgrp_id);
 }
 
-static bool blkcg_policy_enabled(struct request_queue *q,
+static bool blkcg_policy_enabled(struct gendisk *disk,
 				 const struct blkcg_policy *pol)
 {
-	return pol && test_bit(pol->plid, q->blkcg_pols);
+	return pol && test_bit(pol->plid, disk->blkcg_pols);
 }
 
 static void blkg_free(struct blkcg_gq *blkg)
@@ -121,18 +121,18 @@ static void blkg_free(struct blkcg_gq *blkg)
 	/*
 	 * pd_free_fn() can also be called from blkcg_deactivate_policy(),
 	 * in order to make sure pd_free_fn() is called in order, the deletion
-	 * of the list blkg->q_node is delayed to here from blkg_destroy(), and
+	 * of the list blkg->entry is delayed to here from blkg_destroy(), and
 	 * blkcg_mutex is used to synchronize blkg_free_workfn() and
 	 * blkcg_deactivate_policy().
 	 */
-	mutex_lock(&blkg->disk->queue->blkcg_mutex);
+	mutex_lock(&blkg->disk->blkcg_mutex);
 	for (i = 0; i < BLKCG_MAX_POLS; i++)
 		if (blkg->pd[i])
 			blkcg_policy[i]->pd_free_fn(blkg->pd[i]);
 	if (blkg->parent)
 		blkg_put(blkg->parent);
-	list_del_init(&blkg->q_node);
-	mutex_unlock(&blkg->disk->queue->blkcg_mutex);
+	list_del_init(&blkg->entry);
+	mutex_unlock(&blkg->disk->blkcg_mutex);
 
 	put_disk(blkg->disk);
 	free_percpu(blkg->iostat_cpu);
@@ -256,7 +256,7 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct gendisk *disk,
 	get_device(disk_to_dev(disk));
 	blkg->disk = disk;
 
-	INIT_LIST_HEAD(&blkg->q_node);
+	INIT_LIST_HEAD(&blkg->entry);
 	spin_lock_init(&blkg->async_bio_lock);
 	bio_list_init(&blkg->async_bios);
 	INIT_WORK(&blkg->async_bio_work, blkg_async_bio_workfn);
@@ -272,7 +272,7 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct gendisk *disk,
 		struct blkcg_policy *pol = blkcg_policy[i];
 		struct blkg_policy_data *pd;
 
-		if (!blkcg_policy_enabled(disk->queue, pol))
+		if (!blkcg_policy_enabled(disk, pol))
 			continue;
 
 		/* alloc per-policy data and attach it to blkg */
@@ -358,7 +358,7 @@ static struct blkcg_gq *blkg_create(struct blkcg *blkcg, struct gendisk *disk,
 	ret = radix_tree_insert(&blkcg->blkg_tree, disk->queue->id, blkg);
 	if (likely(!ret)) {
 		hlist_add_head_rcu(&blkg->blkcg_node, &blkcg->blkg_list);
-		list_add(&blkg->q_node, &disk->queue->blkg_list);
+		list_add(&blkg->entry, &disk->blkg_list);
 
 		for (i = 0; i < BLKCG_MAX_POLS; i++) {
 			struct blkcg_policy *pol = blkcg_policy[i];
@@ -430,7 +430,7 @@ static struct blkcg_gq *blkg_lookup_create(struct blkcg *blkcg,
 	while (true) {
 		struct blkcg *pos = blkcg;
 		struct blkcg *parent = blkcg_parent(blkcg);
-		struct blkcg_gq *ret_blkg = q->root_blkg;
+		struct blkcg_gq *ret_blkg = disk->root_blkg;
 
 		while (parent) {
 			blkg = blkg_lookup(parent, disk);
@@ -512,7 +512,7 @@ static void blkg_destroy_all(struct gendisk *disk)
 
 restart:
 	spin_lock_irq(&q->queue_lock);
-	list_for_each_entry_safe(blkg, n, &q->blkg_list, q_node) {
+	list_for_each_entry_safe(blkg, n, &disk->blkg_list, entry) {
 		struct blkcg *blkcg = blkg->blkcg;
 
 		spin_lock(&blkcg->lock);
@@ -531,7 +531,7 @@ static void blkg_destroy_all(struct gendisk *disk)
 		}
 	}
 
-	q->root_blkg = NULL;
+	disk->root_blkg = NULL;
 	spin_unlock_irq(&q->queue_lock);
 }
 
@@ -606,7 +606,7 @@ void blkcg_print_blkgs(struct seq_file *sf, struct blkcg *blkcg,
 	rcu_read_lock();
 	hlist_for_each_entry_rcu(blkg, &blkcg->blkg_list, blkcg_node) {
 		spin_lock_irq(&blkg->disk->queue->queue_lock);
-		if (blkcg_policy_enabled(blkg->disk->queue, pol))
+		if (blkcg_policy_enabled(blkg->disk, pol))
 			total += prfill(sf, blkg->pd[pol->plid], data);
 		spin_unlock_irq(&blkg->disk->queue->queue_lock);
 	}
@@ -714,7 +714,7 @@ int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
 	rcu_read_lock();
 	spin_lock_irq(&q->queue_lock);
 
-	if (!blkcg_policy_enabled(q, pol)) {
+	if (!blkcg_policy_enabled(disk, pol)) {
 		ret = -EOPNOTSUPP;
 		goto fail_unlock;
 	}
@@ -757,7 +757,7 @@ int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
 		rcu_read_lock();
 		spin_lock_irq(&q->queue_lock);
 
-		if (!blkcg_policy_enabled(q, pol)) {
+		if (!blkcg_policy_enabled(disk, pol)) {
 			blkg_free(new_blkg);
 			ret = -EOPNOTSUPP;
 			goto fail_preloaded;
@@ -937,7 +937,7 @@ static void blkcg_fill_root_iostats(void)
 	class_dev_iter_init(&iter, &block_class, NULL, &disk_type);
 	while ((dev = class_dev_iter_next(&iter))) {
 		struct block_device *bdev = dev_to_bdev(dev);
-		struct blkcg_gq *blkg = bdev->bd_disk->queue->root_blkg;
+		struct blkcg_gq *blkg = bdev->bd_disk->root_blkg;
 		struct blkg_iostat tmp;
 		int cpu;
 		unsigned long flags;
@@ -1284,8 +1284,8 @@ int blkcg_init_disk(struct gendisk *disk)
 	bool preloaded;
 	int ret;
 
-	INIT_LIST_HEAD(&q->blkg_list);
-	mutex_init(&q->blkcg_mutex);
+	INIT_LIST_HEAD(&disk->blkg_list);
+	mutex_init(&disk->blkcg_mutex);
 
 	new_blkg = blkg_alloc(&blkcg_root, disk, GFP_KERNEL);
 	if (!new_blkg)
@@ -1299,7 +1299,7 @@ int blkcg_init_disk(struct gendisk *disk)
 	blkg = blkg_create(&blkcg_root, disk, new_blkg);
 	if (IS_ERR(blkg))
 		goto err_unlock;
-	q->root_blkg = blkg;
+	disk->root_blkg = blkg;
 	spin_unlock_irq(&q->queue_lock);
 
 	if (preloaded)
@@ -1412,7 +1412,7 @@ int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol)
 	struct blkcg_gq *blkg, *pinned_blkg = NULL;
 	int ret;
 
-	if (blkcg_policy_enabled(q, pol))
+	if (blkcg_policy_enabled(disk, pol))
 		return 0;
 
 	if (queue_is_mq(q))
@@ -1421,7 +1421,7 @@ int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol)
 	spin_lock_irq(&q->queue_lock);
 
 	/* blkg_list is pushed at the head, reverse walk to allocate parents first */
-	list_for_each_entry_reverse(blkg, &q->blkg_list, q_node) {
+	list_for_each_entry_reverse(blkg, &disk->blkg_list, entry) {
 		struct blkg_policy_data *pd;
 
 		if (blkg->pd[pol->plid])
@@ -1466,16 +1466,16 @@ int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol)
 
 	/* all allocated, init in the same order */
 	if (pol->pd_init_fn)
-		list_for_each_entry_reverse(blkg, &q->blkg_list, q_node)
+		list_for_each_entry_reverse(blkg, &disk->blkg_list, entry)
 			pol->pd_init_fn(blkg->pd[pol->plid]);
 
-	list_for_each_entry_reverse(blkg, &q->blkg_list, q_node) {
+	list_for_each_entry_reverse(blkg, &disk->blkg_list, entry) {
 		if (pol->pd_online_fn)
 			pol->pd_online_fn(blkg->pd[pol->plid]);
 		blkg->pd[pol->plid]->online = true;
 	}
 
-	__set_bit(pol->plid, q->blkcg_pols);
+	__set_bit(pol->plid, disk->blkcg_pols);
 	ret = 0;
 
 	spin_unlock_irq(&q->queue_lock);
@@ -1491,7 +1491,7 @@ int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol)
 enomem:
 	/* alloc failed, nothing's initialized yet, free everything */
 	spin_lock_irq(&q->queue_lock);
-	list_for_each_entry(blkg, &q->blkg_list, q_node) {
+	list_for_each_entry(blkg, &disk->blkg_list, entry) {
 		struct blkcg *blkcg = blkg->blkcg;
 
 		spin_lock(&blkcg->lock);
@@ -1521,18 +1521,18 @@ void blkcg_deactivate_policy(struct gendisk *disk,
 	struct request_queue *q = disk->queue;
 	struct blkcg_gq *blkg;
 
-	if (!blkcg_policy_enabled(q, pol))
+	if (!blkcg_policy_enabled(disk, pol))
 		return;
 
 	if (queue_is_mq(q))
 		blk_mq_freeze_queue(q);
 
-	mutex_lock(&q->blkcg_mutex);
+	mutex_lock(&disk->blkcg_mutex);
 	spin_lock_irq(&q->queue_lock);
 
-	__clear_bit(pol->plid, q->blkcg_pols);
+	__clear_bit(pol->plid, disk->blkcg_pols);
 
-	list_for_each_entry(blkg, &q->blkg_list, q_node) {
+	list_for_each_entry(blkg, &disk->blkg_list, entry) {
 		struct blkcg *blkcg = blkg->blkcg;
 
 		spin_lock(&blkcg->lock);
@@ -1546,7 +1546,7 @@ void blkcg_deactivate_policy(struct gendisk *disk,
 	}
 
 	spin_unlock_irq(&q->queue_lock);
-	mutex_unlock(&q->blkcg_mutex);
+	mutex_unlock(&disk->blkcg_mutex);
 
 	if (queue_is_mq(q))
 		blk_mq_unfreeze_queue(q);
@@ -1943,7 +1943,7 @@ static inline struct blkcg_gq *blkg_tryget_closest(struct bio *bio,
  * Associate @bio with the blkg found by combining the css's blkg and the
  * request_queue of the @bio.  An association failure is handled by walking up
  * the blkg tree.  Therefore, the blkg associated can be anything between @blkg
- * and q->root_blkg.  This situation only happens when a cgroup is dying and
+ * and disk->root_blkg.  This situation only happens when a cgroup is dying and
  * then the remaining bios will spill to the closest alive blkg.
  *
  * A reference will be taken on the blkg and will be released when @bio is
@@ -1958,8 +1958,8 @@ void bio_associate_blkg_from_css(struct bio *bio,
 	if (css && css->parent) {
 		bio->bi_blkg = blkg_tryget_closest(bio, css);
 	} else {
-		blkg_get(bdev_get_queue(bio->bi_bdev)->root_blkg);
-		bio->bi_blkg = bdev_get_queue(bio->bi_bdev)->root_blkg;
+		blkg_get(bio->bi_bdev->bd_disk->root_blkg);
+		bio->bi_blkg = bio->bi_bdev->bd_disk->root_blkg;
 	}
 }
 EXPORT_SYMBOL_GPL(bio_associate_blkg_from_css);
diff --git a/block/blk-cgroup.h b/block/blk-cgroup.h
index 151f24de253985..e442b406ca0da6 100644
--- a/block/blk-cgroup.h
+++ b/block/blk-cgroup.h
@@ -54,7 +54,7 @@ struct blkg_iostat_set {
 /* association between a blk cgroup and a request queue */
 struct blkcg_gq {
 	struct gendisk			*disk;
-	struct list_head		q_node;
+	struct list_head		entry;
 	struct hlist_node		blkcg_node;
 	struct blkcg			*blkcg;
 
@@ -250,7 +250,7 @@ static inline struct blkcg_gq *blkg_lookup(struct blkcg *blkcg,
 	WARN_ON_ONCE(!rcu_read_lock_held());
 
 	if (blkcg == &blkcg_root)
-		return disk->queue->root_blkg;
+		return disk->root_blkg;
 
 	blkg = rcu_dereference(blkcg->blkg_hint);
 	if (blkg && blkg->disk == disk)
diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c
index bc0d217f5c1723..5d5aa1e526b742 100644
--- a/block/blk-iolatency.c
+++ b/block/blk-iolatency.c
@@ -665,7 +665,7 @@ static void blkiolatency_timer_fn(struct timer_list *t)
 
 	rcu_read_lock();
 	blkg_for_each_descendant_pre(blkg, pos_css,
-				     blkiolat->rqos.disk->queue->root_blkg) {
+				     blkiolat->rqos.disk->root_blkg) {
 		struct iolatency_grp *iolat;
 		struct child_latency_info *lat_info;
 		unsigned long flags;
diff --git a/block/blk-throttle.c b/block/blk-throttle.c
index 74bb1e753ea09d..902203bdddb4b4 100644
--- a/block/blk-throttle.c
+++ b/block/blk-throttle.c
@@ -451,7 +451,8 @@ static void blk_throtl_update_limit_valid(struct throtl_data *td)
 	bool low_valid = false;
 
 	rcu_read_lock();
-	blkg_for_each_descendant_post(blkg, pos_css, td->queue->root_blkg) {
+	blkg_for_each_descendant_post(blkg, pos_css,
+			td->queue->disk->root_blkg) {
 		struct throtl_grp *tg = blkg_to_tg(blkg);
 
 		if (tg->bps[READ][LIMIT_LOW] || tg->bps[WRITE][LIMIT_LOW] ||
@@ -1180,7 +1181,7 @@ static void throtl_pending_timer_fn(struct timer_list *t)
 
 	spin_lock_irq(&q->queue_lock);
 
-	if (!q->root_blkg)
+	if (!q->disk->root_blkg)
 		goto out_unlock;
 
 	if (throtl_can_upgrade(td, NULL))
@@ -1322,7 +1323,8 @@ static void tg_conf_updated(struct throtl_grp *tg, bool global)
 	 * blk-throttle.
 	 */
 	blkg_for_each_descendant_pre(blkg, pos_css,
-			global ? tg->td->queue->root_blkg : tg_to_blkg(tg)) {
+			global ? tg->td->queue->disk->root_blkg :
+			tg_to_blkg(tg)) {
 		struct throtl_grp *this_tg = blkg_to_tg(blkg);
 		struct throtl_grp *parent_tg;
 
@@ -1717,7 +1719,7 @@ void blk_throtl_cancel_bios(struct gendisk *disk)
 	 * path need RCU protection and to prevent warning from lockdep.
 	 */
 	rcu_read_lock();
-	blkg_for_each_descendant_post(blkg, pos_css, q->root_blkg) {
+	blkg_for_each_descendant_post(blkg, pos_css, disk->root_blkg) {
 		struct throtl_grp *tg = blkg_to_tg(blkg);
 		struct throtl_service_queue *sq = &tg->service_queue;
 
@@ -1871,7 +1873,8 @@ static bool throtl_can_upgrade(struct throtl_data *td,
 		return false;
 
 	rcu_read_lock();
-	blkg_for_each_descendant_post(blkg, pos_css, td->queue->root_blkg) {
+	blkg_for_each_descendant_post(blkg, pos_css,
+			td->queue->disk->root_blkg) {
 		struct throtl_grp *tg = blkg_to_tg(blkg);
 
 		if (tg == this_tg)
@@ -1917,7 +1920,8 @@ static void throtl_upgrade_state(struct throtl_data *td)
 	td->low_upgrade_time = jiffies;
 	td->scale = 0;
 	rcu_read_lock();
-	blkg_for_each_descendant_post(blkg, pos_css, td->queue->root_blkg) {
+	blkg_for_each_descendant_post(blkg, pos_css,
+			td->queue->disk->root_blkg) {
 		struct throtl_grp *tg = blkg_to_tg(blkg);
 		struct throtl_service_queue *sq = &tg->service_queue;
 
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index b9637d63e6f024..79aec4ebadb9e0 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -163,6 +163,12 @@ struct gendisk {
 	struct timer_rand_state *random;
 	atomic_t sync_io;		/* RAID */
 	struct disk_events *ev;
+#ifdef CONFIG_BLK_CGROUP
+	DECLARE_BITMAP		(blkcg_pols, BLKCG_MAX_POLS);
+	struct blkcg_gq		*root_blkg;
+	struct list_head	blkg_list;
+	struct mutex		blkcg_mutex;
+#endif /* CONFIG_BLK_CGROUP */
 #ifdef  CONFIG_BLK_DEV_INTEGRITY
 	struct kobject integrity_kobj;
 #endif	/* CONFIG_BLK_DEV_INTEGRITY */
@@ -481,12 +487,6 @@ struct request_queue {
 	struct blk_mq_tags	*sched_shared_tags;
 
 	struct list_head	icq_list;
-#ifdef CONFIG_BLK_CGROUP
-	DECLARE_BITMAP		(blkcg_pols, BLKCG_MAX_POLS);
-	struct blkcg_gq		*root_blkg;
-	struct list_head	blkg_list;
-	struct mutex		blkcg_mutex;
-#endif
 
 	struct queue_limits	limits;
 
-- 
2.39.0

[PATCH 19/19] blk-cgroup: move the cgroup information to struct gendisk

[Christoph Hellwig]

cgroup information only makes sense on a live gendisk that allows
file system I/O (which includes the raw block device).  So move over
the cgroup related members.

Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
Reviewed-by: Andreas Herrmann <aherrmann-l3A5Bk7waGM@public.gmane.org>
Acked-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 block/bfq-cgroup.c     |  4 +--
 block/blk-cgroup.c     | 66 +++++++++++++++++++++---------------------
 block/blk-cgroup.h     |  4 +--
 block/blk-iolatency.c  |  2 +-
 block/blk-throttle.c   | 16 ++++++----
 include/linux/blkdev.h | 12 ++++----
 6 files changed, 54 insertions(+), 50 deletions(-)

diff --git a/block/bfq-cgroup.c b/block/bfq-cgroup.c
index 37333c164ed458..4fdbbec71647f9 100644
--- a/block/bfq-cgroup.c
+++ b/block/bfq-cgroup.c
@@ -1001,7 +1001,7 @@ void bfq_end_wr_async(struct bfq_data *bfqd)
 {
 	struct blkcg_gq *blkg;
 
-	list_for_each_entry(blkg, &bfqd->queue->blkg_list, q_node) {
+	list_for_each_entry(blkg, &bfqd->queue->disk->blkg_list, entry) {
 		struct bfq_group *bfqg = blkg_to_bfqg(blkg);
 
 		bfq_end_wr_async_queues(bfqd, bfqg);
@@ -1295,7 +1295,7 @@ struct bfq_group *bfq_create_group_hierarchy(struct bfq_data *bfqd, int node)
 	if (ret)
 		return NULL;
 
-	return blkg_to_bfqg(bfqd->queue->root_blkg);
+	return blkg_to_bfqg(bfqd->queue->disk->root_blkg);
 }
 
 struct blkcg_policy blkcg_policy_bfq = {
diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 45a683e88bcac1..8faeca6022bea0 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -108,10 +108,10 @@ static struct cgroup_subsys_state *blkcg_css(void)
 	return task_css(current, io_cgrp_id);
 }
 
-static bool blkcg_policy_enabled(struct request_queue *q,
+static bool blkcg_policy_enabled(struct gendisk *disk,
 				 const struct blkcg_policy *pol)
 {
-	return pol && test_bit(pol->plid, q->blkcg_pols);
+	return pol && test_bit(pol->plid, disk->blkcg_pols);
 }
 
 static void blkg_free(struct blkcg_gq *blkg)
@@ -121,18 +121,18 @@ static void blkg_free(struct blkcg_gq *blkg)
 	/*
 	 * pd_free_fn() can also be called from blkcg_deactivate_policy(),
 	 * in order to make sure pd_free_fn() is called in order, the deletion
-	 * of the list blkg->q_node is delayed to here from blkg_destroy(), and
+	 * of the list blkg->entry is delayed to here from blkg_destroy(), and
 	 * blkcg_mutex is used to synchronize blkg_free_workfn() and
 	 * blkcg_deactivate_policy().
 	 */
-	mutex_lock(&blkg->disk->queue->blkcg_mutex);
+	mutex_lock(&blkg->disk->blkcg_mutex);
 	for (i = 0; i < BLKCG_MAX_POLS; i++)
 		if (blkg->pd[i])
 			blkcg_policy[i]->pd_free_fn(blkg->pd[i]);
 	if (blkg->parent)
 		blkg_put(blkg->parent);
-	list_del_init(&blkg->q_node);
-	mutex_unlock(&blkg->disk->queue->blkcg_mutex);
+	list_del_init(&blkg->entry);
+	mutex_unlock(&blkg->disk->blkcg_mutex);
 
 	put_disk(blkg->disk);
 	free_percpu(blkg->iostat_cpu);
@@ -256,7 +256,7 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct gendisk *disk,
 	get_device(disk_to_dev(disk));
 	blkg->disk = disk;
 
-	INIT_LIST_HEAD(&blkg->q_node);
+	INIT_LIST_HEAD(&blkg->entry);
 	spin_lock_init(&blkg->async_bio_lock);
 	bio_list_init(&blkg->async_bios);
 	INIT_WORK(&blkg->async_bio_work, blkg_async_bio_workfn);
@@ -272,7 +272,7 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct gendisk *disk,
 		struct blkcg_policy *pol = blkcg_policy[i];
 		struct blkg_policy_data *pd;
 
-		if (!blkcg_policy_enabled(disk->queue, pol))
+		if (!blkcg_policy_enabled(disk, pol))
 			continue;
 
 		/* alloc per-policy data and attach it to blkg */
@@ -358,7 +358,7 @@ static struct blkcg_gq *blkg_create(struct blkcg *blkcg, struct gendisk *disk,
 	ret = radix_tree_insert(&blkcg->blkg_tree, disk->queue->id, blkg);
 	if (likely(!ret)) {
 		hlist_add_head_rcu(&blkg->blkcg_node, &blkcg->blkg_list);
-		list_add(&blkg->q_node, &disk->queue->blkg_list);
+		list_add(&blkg->entry, &disk->blkg_list);
 
 		for (i = 0; i < BLKCG_MAX_POLS; i++) {
 			struct blkcg_policy *pol = blkcg_policy[i];
@@ -430,7 +430,7 @@ static struct blkcg_gq *blkg_lookup_create(struct blkcg *blkcg,
 	while (true) {
 		struct blkcg *pos = blkcg;
 		struct blkcg *parent = blkcg_parent(blkcg);
-		struct blkcg_gq *ret_blkg = q->root_blkg;
+		struct blkcg_gq *ret_blkg = disk->root_blkg;
 
 		while (parent) {
 			blkg = blkg_lookup(parent, disk);
@@ -512,7 +512,7 @@ static void blkg_destroy_all(struct gendisk *disk)
 
 restart:
 	spin_lock_irq(&q->queue_lock);
-	list_for_each_entry_safe(blkg, n, &q->blkg_list, q_node) {
+	list_for_each_entry_safe(blkg, n, &disk->blkg_list, entry) {
 		struct blkcg *blkcg = blkg->blkcg;
 
 		spin_lock(&blkcg->lock);
@@ -531,7 +531,7 @@ static void blkg_destroy_all(struct gendisk *disk)
 		}
 	}
 
-	q->root_blkg = NULL;
+	disk->root_blkg = NULL;
 	spin_unlock_irq(&q->queue_lock);
 }
 
@@ -606,7 +606,7 @@ void blkcg_print_blkgs(struct seq_file *sf, struct blkcg *blkcg,
 	rcu_read_lock();
 	hlist_for_each_entry_rcu(blkg, &blkcg->blkg_list, blkcg_node) {
 		spin_lock_irq(&blkg->disk->queue->queue_lock);
-		if (blkcg_policy_enabled(blkg->disk->queue, pol))
+		if (blkcg_policy_enabled(blkg->disk, pol))
 			total += prfill(sf, blkg->pd[pol->plid], data);
 		spin_unlock_irq(&blkg->disk->queue->queue_lock);
 	}
@@ -714,7 +714,7 @@ int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
 	rcu_read_lock();
 	spin_lock_irq(&q->queue_lock);
 
-	if (!blkcg_policy_enabled(q, pol)) {
+	if (!blkcg_policy_enabled(disk, pol)) {
 		ret = -EOPNOTSUPP;
 		goto fail_unlock;
 	}
@@ -757,7 +757,7 @@ int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
 		rcu_read_lock();
 		spin_lock_irq(&q->queue_lock);
 
-		if (!blkcg_policy_enabled(q, pol)) {
+		if (!blkcg_policy_enabled(disk, pol)) {
 			blkg_free(new_blkg);
 			ret = -EOPNOTSUPP;
 			goto fail_preloaded;
@@ -937,7 +937,7 @@ static void blkcg_fill_root_iostats(void)
 	class_dev_iter_init(&iter, &block_class, NULL, &disk_type);
 	while ((dev = class_dev_iter_next(&iter))) {
 		struct block_device *bdev = dev_to_bdev(dev);
-		struct blkcg_gq *blkg = bdev->bd_disk->queue->root_blkg;
+		struct blkcg_gq *blkg = bdev->bd_disk->root_blkg;
 		struct blkg_iostat tmp;
 		int cpu;
 		unsigned long flags;
@@ -1284,8 +1284,8 @@ int blkcg_init_disk(struct gendisk *disk)
 	bool preloaded;
 	int ret;
 
-	INIT_LIST_HEAD(&q->blkg_list);
-	mutex_init(&q->blkcg_mutex);
+	INIT_LIST_HEAD(&disk->blkg_list);
+	mutex_init(&disk->blkcg_mutex);
 
 	new_blkg = blkg_alloc(&blkcg_root, disk, GFP_KERNEL);
 	if (!new_blkg)
@@ -1299,7 +1299,7 @@ int blkcg_init_disk(struct gendisk *disk)
 	blkg = blkg_create(&blkcg_root, disk, new_blkg);
 	if (IS_ERR(blkg))
 		goto err_unlock;
-	q->root_blkg = blkg;
+	disk->root_blkg = blkg;
 	spin_unlock_irq(&q->queue_lock);
 
 	if (preloaded)
@@ -1412,7 +1412,7 @@ int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol)
 	struct blkcg_gq *blkg, *pinned_blkg = NULL;
 	int ret;
 
-	if (blkcg_policy_enabled(q, pol))
+	if (blkcg_policy_enabled(disk, pol))
 		return 0;
 
 	if (queue_is_mq(q))
@@ -1421,7 +1421,7 @@ int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol)
 	spin_lock_irq(&q->queue_lock);
 
 	/* blkg_list is pushed at the head, reverse walk to allocate parents first */
-	list_for_each_entry_reverse(blkg, &q->blkg_list, q_node) {
+	list_for_each_entry_reverse(blkg, &disk->blkg_list, entry) {
 		struct blkg_policy_data *pd;
 
 		if (blkg->pd[pol->plid])
@@ -1466,16 +1466,16 @@ int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol)
 
 	/* all allocated, init in the same order */
 	if (pol->pd_init_fn)
-		list_for_each_entry_reverse(blkg, &q->blkg_list, q_node)
+		list_for_each_entry_reverse(blkg, &disk->blkg_list, entry)
 			pol->pd_init_fn(blkg->pd[pol->plid]);
 
-	list_for_each_entry_reverse(blkg, &q->blkg_list, q_node) {
+	list_for_each_entry_reverse(blkg, &disk->blkg_list, entry) {
 		if (pol->pd_online_fn)
 			pol->pd_online_fn(blkg->pd[pol->plid]);
 		blkg->pd[pol->plid]->online = true;
 	}
 
-	__set_bit(pol->plid, q->blkcg_pols);
+	__set_bit(pol->plid, disk->blkcg_pols);
 	ret = 0;
 
 	spin_unlock_irq(&q->queue_lock);
@@ -1491,7 +1491,7 @@ int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol)
 enomem:
 	/* alloc failed, nothing's initialized yet, free everything */
 	spin_lock_irq(&q->queue_lock);
-	list_for_each_entry(blkg, &q->blkg_list, q_node) {
+	list_for_each_entry(blkg, &disk->blkg_list, entry) {
 		struct blkcg *blkcg = blkg->blkcg;
 
 		spin_lock(&blkcg->lock);
@@ -1521,18 +1521,18 @@ void blkcg_deactivate_policy(struct gendisk *disk,
 	struct request_queue *q = disk->queue;
 	struct blkcg_gq *blkg;
 
-	if (!blkcg_policy_enabled(q, pol))
+	if (!blkcg_policy_enabled(disk, pol))
 		return;
 
 	if (queue_is_mq(q))
 		blk_mq_freeze_queue(q);
 
-	mutex_lock(&q->blkcg_mutex);
+	mutex_lock(&disk->blkcg_mutex);
 	spin_lock_irq(&q->queue_lock);
 
-	__clear_bit(pol->plid, q->blkcg_pols);
+	__clear_bit(pol->plid, disk->blkcg_pols);
 
-	list_for_each_entry(blkg, &q->blkg_list, q_node) {
+	list_for_each_entry(blkg, &disk->blkg_list, entry) {
 		struct blkcg *blkcg = blkg->blkcg;
 
 		spin_lock(&blkcg->lock);
@@ -1546,7 +1546,7 @@ void blkcg_deactivate_policy(struct gendisk *disk,
 	}
 
 	spin_unlock_irq(&q->queue_lock);
-	mutex_unlock(&q->blkcg_mutex);
+	mutex_unlock(&disk->blkcg_mutex);
 
 	if (queue_is_mq(q))
 		blk_mq_unfreeze_queue(q);
@@ -1943,7 +1943,7 @@ static inline struct blkcg_gq *blkg_tryget_closest(struct bio *bio,
  * Associate @bio with the blkg found by combining the css's blkg and the
  * request_queue of the @bio.  An association failure is handled by walking up
  * the blkg tree.  Therefore, the blkg associated can be anything between @blkg
- * and q->root_blkg.  This situation only happens when a cgroup is dying and
+ * and disk->root_blkg.  This situation only happens when a cgroup is dying and
  * then the remaining bios will spill to the closest alive blkg.
  *
  * A reference will be taken on the blkg and will be released when @bio is
@@ -1958,8 +1958,8 @@ void bio_associate_blkg_from_css(struct bio *bio,
 	if (css && css->parent) {
 		bio->bi_blkg = blkg_tryget_closest(bio, css);
 	} else {
-		blkg_get(bdev_get_queue(bio->bi_bdev)->root_blkg);
-		bio->bi_blkg = bdev_get_queue(bio->bi_bdev)->root_blkg;
+		blkg_get(bio->bi_bdev->bd_disk->root_blkg);
+		bio->bi_blkg = bio->bi_bdev->bd_disk->root_blkg;
 	}
 }
 EXPORT_SYMBOL_GPL(bio_associate_blkg_from_css);
diff --git a/block/blk-cgroup.h b/block/blk-cgroup.h
index 151f24de253985..e442b406ca0da6 100644
--- a/block/blk-cgroup.h
+++ b/block/blk-cgroup.h
@@ -54,7 +54,7 @@ struct blkg_iostat_set {
 /* association between a blk cgroup and a request queue */
 struct blkcg_gq {
 	struct gendisk			*disk;
-	struct list_head		q_node;
+	struct list_head		entry;
 	struct hlist_node		blkcg_node;
 	struct blkcg			*blkcg;
 
@@ -250,7 +250,7 @@ static inline struct blkcg_gq *blkg_lookup(struct blkcg *blkcg,
 	WARN_ON_ONCE(!rcu_read_lock_held());
 
 	if (blkcg == &blkcg_root)
-		return disk->queue->root_blkg;
+		return disk->root_blkg;
 
 	blkg = rcu_dereference(blkcg->blkg_hint);
 	if (blkg && blkg->disk == disk)
diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c
index bc0d217f5c1723..5d5aa1e526b742 100644
--- a/block/blk-iolatency.c
+++ b/block/blk-iolatency.c
@@ -665,7 +665,7 @@ static void blkiolatency_timer_fn(struct timer_list *t)
 
 	rcu_read_lock();
 	blkg_for_each_descendant_pre(blkg, pos_css,
-				     blkiolat->rqos.disk->queue->root_blkg) {
+				     blkiolat->rqos.disk->root_blkg) {
 		struct iolatency_grp *iolat;
 		struct child_latency_info *lat_info;
 		unsigned long flags;
diff --git a/block/blk-throttle.c b/block/blk-throttle.c
index 74bb1e753ea09d..902203bdddb4b4 100644
--- a/block/blk-throttle.c
+++ b/block/blk-throttle.c
@@ -451,7 +451,8 @@ static void blk_throtl_update_limit_valid(struct throtl_data *td)
 	bool low_valid = false;
 
 	rcu_read_lock();
-	blkg_for_each_descendant_post(blkg, pos_css, td->queue->root_blkg) {
+	blkg_for_each_descendant_post(blkg, pos_css,
+			td->queue->disk->root_blkg) {
 		struct throtl_grp *tg = blkg_to_tg(blkg);
 
 		if (tg->bps[READ][LIMIT_LOW] || tg->bps[WRITE][LIMIT_LOW] ||
@@ -1180,7 +1181,7 @@ static void throtl_pending_timer_fn(struct timer_list *t)
 
 	spin_lock_irq(&q->queue_lock);
 
-	if (!q->root_blkg)
+	if (!q->disk->root_blkg)
 		goto out_unlock;
 
 	if (throtl_can_upgrade(td, NULL))
@@ -1322,7 +1323,8 @@ static void tg_conf_updated(struct throtl_grp *tg, bool global)
 	 * blk-throttle.
 	 */
 	blkg_for_each_descendant_pre(blkg, pos_css,
-			global ? tg->td->queue->root_blkg : tg_to_blkg(tg)) {
+			global ? tg->td->queue->disk->root_blkg :
+			tg_to_blkg(tg)) {
 		struct throtl_grp *this_tg = blkg_to_tg(blkg);
 		struct throtl_grp *parent_tg;
 
@@ -1717,7 +1719,7 @@ void blk_throtl_cancel_bios(struct gendisk *disk)
 	 * path need RCU protection and to prevent warning from lockdep.
 	 */
 	rcu_read_lock();
-	blkg_for_each_descendant_post(blkg, pos_css, q->root_blkg) {
+	blkg_for_each_descendant_post(blkg, pos_css, disk->root_blkg) {
 		struct throtl_grp *tg = blkg_to_tg(blkg);
 		struct throtl_service_queue *sq = &tg->service_queue;
 
@@ -1871,7 +1873,8 @@ static bool throtl_can_upgrade(struct throtl_data *td,
 		return false;
 
 	rcu_read_lock();
-	blkg_for_each_descendant_post(blkg, pos_css, td->queue->root_blkg) {
+	blkg_for_each_descendant_post(blkg, pos_css,
+			td->queue->disk->root_blkg) {
 		struct throtl_grp *tg = blkg_to_tg(blkg);
 
 		if (tg == this_tg)
@@ -1917,7 +1920,8 @@ static void throtl_upgrade_state(struct throtl_data *td)
 	td->low_upgrade_time = jiffies;
 	td->scale = 0;
 	rcu_read_lock();
-	blkg_for_each_descendant_post(blkg, pos_css, td->queue->root_blkg) {
+	blkg_for_each_descendant_post(blkg, pos_css,
+			td->queue->disk->root_blkg) {
 		struct throtl_grp *tg = blkg_to_tg(blkg);
 		struct throtl_service_queue *sq = &tg->service_queue;
 
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index b9637d63e6f024..79aec4ebadb9e0 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -163,6 +163,12 @@ struct gendisk {
 	struct timer_rand_state *random;
 	atomic_t sync_io;		/* RAID */
 	struct disk_events *ev;
+#ifdef CONFIG_BLK_CGROUP
+	DECLARE_BITMAP		(blkcg_pols, BLKCG_MAX_POLS);
+	struct blkcg_gq		*root_blkg;
+	struct list_head	blkg_list;
+	struct mutex		blkcg_mutex;
+#endif /* CONFIG_BLK_CGROUP */
 #ifdef  CONFIG_BLK_DEV_INTEGRITY
 	struct kobject integrity_kobj;
 #endif	/* CONFIG_BLK_DEV_INTEGRITY */
@@ -481,12 +487,6 @@ struct request_queue {
 	struct blk_mq_tags	*sched_shared_tags;
 
 	struct list_head	icq_list;
-#ifdef CONFIG_BLK_CGROUP
-	DECLARE_BITMAP		(blkcg_pols, BLKCG_MAX_POLS);
-	struct blkcg_gq		*root_blkg;
-	struct list_head	blkg_list;
-	struct mutex		blkcg_mutex;
-#endif
 
 	struct queue_limits	limits;
 
-- 
2.39.0

Re: switch blk-cgroup to work on gendisk v4

[Jens Axboe]

On Fri, 03 Feb 2023 16:03:41 +0100, Christoph Hellwig wrote:
> blk-cgroup works on only on live disks and "file system" I/O from bios.
> This all the information should be in the gendisk, and not the
> request_queue that also exists for pure passthrough request based
> devices.
> 
> Changes since v3:
>  - improve a commit log
>  - drop a change to not acquire a pointless disk reference
> 
> [...]

Applied, thanks!

[01/19] block: don't call blk_throtl_stat_add for non-READ/WRITE commands
        commit: a886001c2da8dd02357d0d336ddb021903347f89
[02/19] blk-cgroup: delay blk-cgroup initialization until add_disk
        commit: 178fa7d49815ea8001f43ade37a22072829fd8ab
[03/19] blk-cgroup: improve error unwinding in blkg_alloc
        commit: 0b6f93bdf07e52620f725f721e547408e0d04c9d
[04/19] blk-cgroup: simplify blkg freeing from initialization failure paths
        commit: 27b642b07a4a5eb44dffa94a5171ce468bdc46f9
[05/19] blk-cgroup: remove the !bdi->dev check in blkg_dev_name
        commit: 180b04d450a7137270c12dbb6bebf1d5e6c0a6f2
[06/19] blk-cgroup: pin the gendisk in struct blkcg_gq
        commit: 84d7d462b16dd5f0bf7c7ca9254bf81db2c952a2
[07/19] blk-cgroup: store a gendisk to throttle in struct task_struct
        commit: f05837ed73d0c73e950b2d9f2612febb0d3d451e
[08/19] blk-wbt: pass a gendisk to wbt_{enable,disable}_default
        commit: 04aad37be1a88de6a1919996a615437ac74de479
[09/19] blk-wbt: pass a gendisk to wbt_init
        commit: 958f29654747a54f2272eb478e493eb97f492e06
[10/19] blk-wbt: move private information from blk-wbt.h to blk-wbt.c
        commit: 0bc65bd41dfd2f75b9f38812326d767db5cd0663
[11/19] blk-wbt: open code wbt_queue_depth_changed in wbt_init
        commit: 4e1d91ae876bd12f327340f11a16a1278985e7e1
[12/19] blk-rq-qos: move rq_qos_add and rq_qos_del out of line
        commit: b494f9c566ba5fe2cc8abe67fdeb0332c6b48d4b
[13/19] blk-rq-qos: make rq_qos_add and rq_qos_del more useful
        commit: ce57b558604e68277d31ca5ce49ec4579a8618c5
[14/19] blk-rq-qos: constify rq_qos_ops
        commit: 3963d84df7974b6687cb34bce3b9e0b2686f839c
[15/19] blk-rq-qos: store a gendisk instead of request_queue in struct rq_qos
        commit: ba91c849fa50dbc6519cf7808177b3a9b7f6bc97
[16/19] blk-cgroup: pass a gendisk to blkcg_{de,}activate_policy
        commit: 40e4996ec099a301083eb7e29095ebdfc31443da
[17/19] blk-cgroup: pass a gendisk to pd_alloc_fn
        commit: 0a0b4f79db2e6e745672aa3852cf5fdf7af14a0f
[18/19] blk-cgroup: pass a gendisk to blkg_lookup
        commit: 479664cee14d8452d3d76f8d0b7fccd0cbe4ed49
[19/19] blk-cgroup: move the cgroup information to struct gendisk
        commit: 3f13ab7c80fdb0ada86a8e3e818960bc1ccbaa59

Best regards,
-- 
Jens Axboe

Re: switch blk-cgroup to work on gendisk v4

[Jens Axboe]

On Fri, 03 Feb 2023 16:03:41 +0100, Christoph Hellwig wrote:
> blk-cgroup works on only on live disks and "file system" I/O from bios.
> This all the information should be in the gendisk, and not the
> request_queue that also exists for pure passthrough request based
> devices.
> 
> Changes since v3:
>  - improve a commit log
>  - drop a change to not acquire a pointless disk reference
> 
> [...]

Applied, thanks!

[01/19] block: don't call blk_throtl_stat_add for non-READ/WRITE commands
        commit: a886001c2da8dd02357d0d336ddb021903347f89
[02/19] blk-cgroup: delay blk-cgroup initialization until add_disk
        commit: 178fa7d49815ea8001f43ade37a22072829fd8ab
[03/19] blk-cgroup: improve error unwinding in blkg_alloc
        commit: 0b6f93bdf07e52620f725f721e547408e0d04c9d
[04/19] blk-cgroup: simplify blkg freeing from initialization failure paths
        commit: 27b642b07a4a5eb44dffa94a5171ce468bdc46f9
[05/19] blk-cgroup: remove the !bdi->dev check in blkg_dev_name
        commit: 180b04d450a7137270c12dbb6bebf1d5e6c0a6f2
[06/19] blk-cgroup: pin the gendisk in struct blkcg_gq
        commit: 84d7d462b16dd5f0bf7c7ca9254bf81db2c952a2
[07/19] blk-cgroup: store a gendisk to throttle in struct task_struct
        commit: f05837ed73d0c73e950b2d9f2612febb0d3d451e
[08/19] blk-wbt: pass a gendisk to wbt_{enable,disable}_default
        commit: 04aad37be1a88de6a1919996a615437ac74de479
[09/19] blk-wbt: pass a gendisk to wbt_init
        commit: 958f29654747a54f2272eb478e493eb97f492e06
[10/19] blk-wbt: move private information from blk-wbt.h to blk-wbt.c
        commit: 0bc65bd41dfd2f75b9f38812326d767db5cd0663
[11/19] blk-wbt: open code wbt_queue_depth_changed in wbt_init
        commit: 4e1d91ae876bd12f327340f11a16a1278985e7e1
[12/19] blk-rq-qos: move rq_qos_add and rq_qos_del out of line
        commit: b494f9c566ba5fe2cc8abe67fdeb0332c6b48d4b
[13/19] blk-rq-qos: make rq_qos_add and rq_qos_del more useful
        commit: ce57b558604e68277d31ca5ce49ec4579a8618c5
[14/19] blk-rq-qos: constify rq_qos_ops
        commit: 3963d84df7974b6687cb34bce3b9e0b2686f839c
[15/19] blk-rq-qos: store a gendisk instead of request_queue in struct rq_qos
        commit: ba91c849fa50dbc6519cf7808177b3a9b7f6bc97
[16/19] blk-cgroup: pass a gendisk to blkcg_{de,}activate_policy
        commit: 40e4996ec099a301083eb7e29095ebdfc31443da
[17/19] blk-cgroup: pass a gendisk to pd_alloc_fn
        commit: 0a0b4f79db2e6e745672aa3852cf5fdf7af14a0f
[18/19] blk-cgroup: pass a gendisk to blkg_lookup
        commit: 479664cee14d8452d3d76f8d0b7fccd0cbe4ed49
[19/19] blk-cgroup: move the cgroup information to struct gendisk
        commit: 3f13ab7c80fdb0ada86a8e3e818960bc1ccbaa59

Best regards,
-- 
Jens Axboe

switch blk-cgroup to work on gendisk v3

[Christoph Hellwig]

Hi all,

blk-cgroup works on only on live disks and "file system" I/O from bios.
This all the information should be in the gendisk, and not the
request_queue that also exists for pure passthrough request based
devices.

Changes since v2:
 - drop the patch to revert the async blkg freeing
 - simplify error unwinding when failing to allocate a blkg
 - add back the dead queue/disk check in blkg_alloc
 - split two patches to better document the changes
 - add another blk-wbt cleanup patch
 - rewrite a commit log
 - another typo fix

Changes since v1:
 - use the local disk variable in wbt_init instead of q->disk
 - various spelling fixes

Diffstat:
 block/bfq-cgroup.c        |   18 +--
 block/bfq-iosched.c       |    6 -
 block/blk-cgroup-rwstat.c |    2 
 block/blk-cgroup.c        |  231 +++++++++++++++++++++-------------------------
 block/blk-cgroup.h        |   38 +++----
 block/blk-iocost.c        |   40 +++----
 block/blk-iolatency.c     |   41 +++-----
 block/blk-ioprio.c        |    6 -
 block/blk-mq-debugfs.c    |   10 -
 block/blk-rq-qos.c        |   67 +++++++++++++
 block/blk-rq-qos.h        |   66 -------------
 block/blk-settings.c      |    1 
 block/blk-stat.c          |    3 
 block/blk-sysfs.c         |    5 
 block/blk-throttle.c      |   31 +++---
 block/blk-wbt.c           |  116 ++++++++++++++++++-----
 block/blk-wbt.h           |   98 +------------------
 block/genhd.c             |   17 +--
 include/linux/blkdev.h    |   12 +-
 include/linux/sched.h     |    2 
 kernel/fork.c             |    2 
 mm/swapfile.c             |    2 
 22 files changed, 395 insertions(+), 419 deletions(-)

[PATCH 05/19] blk-cgroup: remove the !bdi->dev check in blkg_dev_name

[Christoph Hellwig]

bdi_dev_name already performs the same check.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 block/blk-cgroup.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 1038688568926e..0b3226cbf3f25d 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -572,7 +572,7 @@ static int blkcg_reset_stats(struct cgroup_subsys_state *css,
 
 const char *blkg_dev_name(struct blkcg_gq *blkg)
 {
-	if (!blkg->q->disk || !blkg->q->disk->bdi->dev)
+	if (!blkg->q->disk)
 		return NULL;
 	return bdi_dev_name(blkg->q->disk->bdi);
 }
-- 
2.39.0