Skip to contentSkip to conversation

block: unify elevator changing and fix lockdep warning

PATCH Superseded Ming Lei 6 participants 71 messages Apr 24, 2025 – Apr 30, 2025

Subsystems (2)

v3 of 5 Superseded Latest is v5 →

lore
Share & tools
lore

Ming Lei

author
#0 lore raw

v3 cover block: unify elevator changing and fix lockdep warning

Hello Jens,

This patchset cleans up elevator change code, and unifying it via single helper, meantime moves kobject_add/del & debugfs register/unregister out of queue freezing & elevator_lock. This way fixes many lockdep warnings reported recently, especially since fs_reclaim is connected with freeze lock manually by commit ffa1e7ada456 ("block: Make request_queue lockdep splats show up earlier").

Thanks,
Ming

V3:
- replace srcu with rw_sem for avoiding race between add/del disk & elevator switch and updating nr_hw_queues (Nilay Shoff)

- add elv_update_nr_hw_queues() for elevator reattachment in case of updating nr_hw_queues, meantime keep elv_change_ctx as local structure (Christoph)

- replace ->elevator_lock with disk->rqos_state_mutex for covering wbt state change

- add new patch "block: use q->elevator with ->elevator_lock held in elv_iosched_show()"

- small cleanup & commit log improvement

V2:
- retry add/del disk when blk_mq_update_nr_hw_queues() is in-progress

- swap blk_mq_add_queue_tag_set() with blk_mq_map_swqueue() in blk_mq_init_allocated_queue() (Nilay Shroff)

- move ELEVATOR_FLAG_DISABLE_WBT to request queue's flags (Nilay Shoff)

- fix race because of delaying elevator unregister

- define flags of elv_change_ctx as bool (Christoph)

- improve comment and commit log (Christoph)

Ming Lei (20):
block: move blk_mq_add_queue_tag_set() after blk_mq_map_swqueue()
block: move ELEVATOR_FLAG_DISABLE_WBT a request queue flag
block: don't call freeze queue in elevator_switch() and
elevator_disable()
block: use q->elevator with ->elevator_lock held in elv_iosched_show()
block: add two helpers for registering/un-registering sched debugfs
block: move sched debugfs register into elvevator_register_queue
block: prevent adding/deleting disk during updating nr_hw_queues
block: don't allow to switch elevator if updating nr_hw_queues is
in-progress
block: simplify elevator reattachment for updating nr_hw_queues
block: move blk_unregister_queue() & device_del() after freeze wait
block: move queue freezing & elevator_lock into elevator_change()
block: add struct elv_change_ctx for unifying elevator change
block: unifying elevator change
block: pass elevator_queue to elv_register_queue & unregister_queue
block: fail to show/store elevator sysfs attribute if elevator is
dying
block: move elv_register[unregister]_queue out of elevator_lock
block: move debugfs/sysfs register out of freezing queue
block: remove several ->elevator_lock
block: move hctx cpuhp add/del out of queue freezing
block: move wbt_enable_default() out of queue freezing from sched

    ->exit()
 block/bfq-iosched.c    |   6 +-
 block/blk-mq-debugfs.c |  12 +-
 block/blk-mq-sched.c   |  41 +++---
 block/blk-mq.c         | 132 +++---------------
 block/blk-sysfs.c      |  24 ++--
 block/blk-wbt.c        |  13 +-
 block/blk.h            |   8 +-
 block/elevator.c       | 302 ++++++++++++++++++++++++++++-------------
 block/elevator.h       |   6 +-
 block/genhd.c          | 129 +++++++++++-------
 include/linux/blk-mq.h |   3 +
 include/linux/blkdev.h |   5 +
 12 files changed, 365 insertions(+), 316 deletions(-)
Signature · 1 line
2.47.0
author
#1 lore raw

PATCH v3 1/20 block: move blk_mq_add_queue_tag_set() after blk_mq_map_swqueue()

Move blk_mq_add_queue_tag_set() after blk_mq_map_swqueue(), and publish this request queue to tagset after everything is setup.

This way is safe because BLK_MQ_F_TAG_QUEUE_SHARED isn't used by blk_mq_map_swqueue(), and this flag is mainly checked in fast IO code path.

Prepare for removing ->elevator_lock from blk_mq_map_swqueue() which is supposed to be called when elevator switch can't be done.

---
 block/blk-mq.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
block/blk-mq.c +1 -1
index 554380bfd002..29cfc7ce2e0a 100644
@@ -4581,8 +4581,8 @@ int blk_mq_init_allocated_queue(struct blk_mq_tag_set *set,
 	q->nr_requests = set->queue_depth;
 
 	blk_mq_init_cpu_queues(q, set->nr_hw_queues);
-	blk_mq_add_queue_tag_set(set, q);
 	blk_mq_map_swqueue(q);
+	blk_mq_add_queue_tag_set(set, q);
 	return 0;
 
 err_hctxs:
Signature · 1 line
2.47.0
author
#2 lore raw

PATCH v3 2/20 block: move ELEVATOR_FLAG_DISABLE_WBT a request queue flag

In reply to Ming Lei

ELEVATOR_FLAG_DISABLE_WBT is only used by BFQ to disallow wbt when BFQ is in use. The flag is set in BFQ's init(), and cleared in BFQ's exit().

Making it as request queue flag, so that we can avoid to deal with elevator switch race. Also it isn't graceful to checking one scheduler flag in wbt_enable_default().

---
 block/bfq-iosched.c    | 4 ++--
 block/blk-mq-debugfs.c | 1 +
 block/blk-wbt.c        | 3 +--
 block/elevator.h       | 1 -
 include/linux/blkdev.h | 3 +++
 5 files changed, 7 insertions(+), 5 deletions(-)
block/bfq-iosched.c +2 -2
index abd80dc13562..cc6f59836dcd 100644
@@ -7210,7 +7210,7 @@ static void bfq_exit_queue(struct elevator_queue *e)
 #endif
 
 	blk_stat_disable_accounting(bfqd->queue);
-	clear_bit(ELEVATOR_FLAG_DISABLE_WBT, &e->flags);
+	blk_queue_flag_clear(QUEUE_FLAG_DISABLE_WBT_DEF, bfqd->queue);
 	wbt_enable_default(bfqd->queue->disk);
 
 	kfree(bfqd);
@@ -7397,7 +7397,7 @@ static int bfq_init_queue(struct request_queue *q, struct elevator_type *e)
 	/* We dispatch from request queue wide instead of hw queue */
 	blk_queue_flag_set(QUEUE_FLAG_SQ_SCHED, q);
 
-	set_bit(ELEVATOR_FLAG_DISABLE_WBT, &eq->flags);
+	blk_queue_flag_set(QUEUE_FLAG_DISABLE_WBT_DEF, q);
 	wbt_disable_default(q->disk);
 	blk_stat_enable_accounting(q);
 
block/blk-mq-debugfs.c +1 -0
index 3421b5521fe2..7710c409e432 100644
@@ -93,6 +93,7 @@ static const char *const blk_queue_flag_name[] = {
 	QUEUE_FLAG_NAME(RQ_ALLOC_TIME),
 	QUEUE_FLAG_NAME(HCTX_ACTIVE),
 	QUEUE_FLAG_NAME(SQ_SCHED),
+	QUEUE_FLAG_NAME(DISABLE_WBT_DEF),
 };
 #undef QUEUE_FLAG_NAME
 
block/blk-wbt.c +1 -2
index f1754d07f7e0..29cd2e33666f 100644
@@ -704,8 +704,7 @@ void wbt_enable_default(struct gendisk *disk)
 	struct rq_qos *rqos;
 	bool enable = IS_ENABLED(CONFIG_BLK_WBT_MQ);
 
-	if (q->elevator &&
-	    test_bit(ELEVATOR_FLAG_DISABLE_WBT, &q->elevator->flags))
+	if (blk_queue_disable_wbt(q))
 		enable = false;
 
 	/* Throttling already enabled? */
block/elevator.h +0 -1
index e4e44dfac503..e27af5492cdb 100644
@@ -121,7 +121,6 @@ struct elevator_queue
 };
 
 #define ELEVATOR_FLAG_REGISTERED	0
-#define ELEVATOR_FLAG_DISABLE_WBT	1
 
 /*
  * block elevator interface
include/linux/blkdev.h +3 -0
index 678dc38442bf..228b2e5ad493 100644
@@ -644,6 +644,7 @@ enum {
 	QUEUE_FLAG_RQ_ALLOC_TIME,	/* record rq->alloc_time_ns */
 	QUEUE_FLAG_HCTX_ACTIVE,		/* at least one blk-mq hctx is active */
 	QUEUE_FLAG_SQ_SCHED,		/* single queue style io dispatch */
+	QUEUE_FLAG_DISABLE_WBT_DEF,	/* for sched to disable/enable wbt */
 	QUEUE_FLAG_MAX
 };
 
@@ -679,6 +680,8 @@ void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
 #define blk_queue_sq_sched(q)	test_bit(QUEUE_FLAG_SQ_SCHED, &(q)->queue_flags)
 #define blk_queue_skip_tagset_quiesce(q) \
 	((q)->limits.features & BLK_FEAT_SKIP_TAGSET_QUIESCE)
+#define blk_queue_disable_wbt(q)	\
+	test_bit(QUEUE_FLAG_DISABLE_WBT_DEF, &(q)->queue_flags)
 
 extern void blk_set_pm_only(struct request_queue *q);
 extern void blk_clear_pm_only(struct request_queue *q);
Signature · 1 line
2.47.0
author
#4 lore raw

PATCH v3 3/20 block: don't call freeze queue in elevator_switch() and elevator_disable()

In reply to Ming Lei

Both elevator_switch() and elevator_disable() are only called from the two code paths, in which queue is guaranteed to be frozen.

So don't call freeze queue in the two functions, also add asserts for queue freeze.

---
 block/elevator.c | 9 ++-------
 1 file changed, 2 insertions(+), 7 deletions(-)
block/elevator.c +2 -7
index b4d08026b02c..5051a98dc08c 100644
@@ -615,12 +615,11 @@ void elevator_init_mq(struct request_queue *q)
  */
 int elevator_switch(struct request_queue *q, struct elevator_type *new_e)
 {
-	unsigned int memflags;
 	int ret;
 
+	WARN_ON_ONCE(q->mq_freeze_depth == 0);
 	lockdep_assert_held(&q->elevator_lock);
 
-	memflags = blk_mq_freeze_queue(q);
 	blk_mq_quiesce_queue(q);
 
 	if (q->elevator) {
@@ -641,7 +640,6 @@ int elevator_switch(struct request_queue *q, struct elevator_type *new_e)
 
 out_unfreeze:
 	blk_mq_unquiesce_queue(q);
-	blk_mq_unfreeze_queue(q, memflags);
 
 	if (ret) {
 		pr_warn("elv: switch to \"%s\" failed, falling back to \"none\"\n",
@@ -653,11 +651,9 @@ int elevator_switch(struct request_queue *q, struct elevator_type *new_e)
 
 void elevator_disable(struct request_queue *q)
 {
-	unsigned int memflags;
-
+	WARN_ON_ONCE(q->mq_freeze_depth == 0);
 	lockdep_assert_held(&q->elevator_lock);
 
-	memflags = blk_mq_freeze_queue(q);
 	blk_mq_quiesce_queue(q);
 
 	elv_unregister_queue(q);
@@ -668,7 +664,6 @@ void elevator_disable(struct request_queue *q)
 	blk_add_trace_msg(q, "elv switch: none");
 
 	blk_mq_unquiesce_queue(q);
-	blk_mq_unfreeze_queue(q, memflags);
 }
 
 /*
Signature · 1 line
2.47.0
author
#6 lore raw

PATCH v3 4/20 block: use q->elevator with ->elevator_lock held in elv_iosched_show()

In reply to Ming Lei

Use q->elevator with ->elevator_lock held in elv_iosched_show(), since the local cached elevator reference may become stale after getting ->elevator_lock.

---
 block/elevator.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
block/elevator.c +1 -2
index 5051a98dc08c..b32815594892 100644
@@ -739,7 +739,6 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 ssize_t elv_iosched_show(struct gendisk *disk, char *name)
 {
 	struct request_queue *q = disk->queue;
-	struct elevator_queue *eq = q->elevator;
 	struct elevator_type *cur = NULL, *e;
 	int len = 0;
 
@@ -748,7 +747,7 @@ ssize_t elv_iosched_show(struct gendisk *disk, char *name)
 		len += sprintf(name+len, "[none] ");
 	} else {
 		len += sprintf(name+len, "none ");
-		cur = eq->type;
+		cur = q->elevator->type;
 	}
 
 	spin_lock(&elv_list_lock);
Signature · 1 line
2.47.0
Reviewed-by on v3 4/20
#7 lore raw

Re: [PATCH V3 04/20] block: use q->elevator with ->elevator_lock held in elv_iosched_show()

Quoted 9 lines · Ming Lei

Use q->elevator with ->elevator_lock held in elv_iosched_show(), since the local cached elevator reference may become stale after getting ->elevator_lock.

---
  block/elevator.c | 3 +--
  1 file changed, 1 insertion(+), 2 deletions(-)

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
author
#10 lore raw

PATCH v3 5/20 block: add two helpers for registering/un-registering sched debugfs

In reply to Ming Lei

Add blk_mq_sched_reg_debugfs()/blk_mq_sched_unreg_debugfs() to clean up sched init/exit code a bit.

Register & unregister debugfs for sched & sched_hctx order is changed a bit, but it is safe because sched & sched_hctx is guaranteed to be ready when exporting via debugfs.

---
 block/blk-mq-sched.c | 45 +++++++++++++++++++++++++++++---------------
 1 file changed, 30 insertions(+), 15 deletions(-)
block/blk-mq-sched.c +30 -15
index 9b81771774ef..2abc5e0704e8 100644
@@ -434,6 +434,30 @@ static int blk_mq_init_sched_shared_tags(struct request_queue *queue)
 	return 0;
 }
 
+static void blk_mq_sched_reg_debugfs(struct request_queue *q)
+{
+	struct blk_mq_hw_ctx *hctx;
+	unsigned long i;
+
+	mutex_lock(&q->debugfs_mutex);
+	blk_mq_debugfs_register_sched(q);
+	queue_for_each_hw_ctx(q, hctx, i)
+		blk_mq_debugfs_register_sched_hctx(q, hctx);
+	mutex_unlock(&q->debugfs_mutex);
+}
+
+static void blk_mq_sched_unreg_debugfs(struct request_queue *q)
+{
+	struct blk_mq_hw_ctx *hctx;
+	unsigned long i;
+
+	mutex_lock(&q->debugfs_mutex);
+	queue_for_each_hw_ctx(q, hctx, i)
+		blk_mq_debugfs_unregister_sched_hctx(hctx);
+	blk_mq_debugfs_unregister_sched(q);
+	mutex_unlock(&q->debugfs_mutex);
+}
+
 /* caller must have a reference to @e, will grab another one if successful */
 int blk_mq_init_sched(struct request_queue *q, struct elevator_type *e)
 {
@@ -467,10 +491,6 @@ int blk_mq_init_sched(struct request_queue *q, struct elevator_type *e)
 	if (ret)
 		goto err_free_map_and_rqs;
 
-	mutex_lock(&q->debugfs_mutex);
-	blk_mq_debugfs_register_sched(q);
-	mutex_unlock(&q->debugfs_mutex);
-
 	queue_for_each_hw_ctx(q, hctx, i) {
 		if (e->ops.init_hctx) {
 			ret = e->ops.init_hctx(hctx, i);
@@ -482,11 +502,11 @@ int blk_mq_init_sched(struct request_queue *q, struct elevator_type *e)
 				return ret;
 			}
 		}
-		mutex_lock(&q->debugfs_mutex);
-		blk_mq_debugfs_register_sched_hctx(q, hctx);
-		mutex_unlock(&q->debugfs_mutex);
 	}
 
+	/* sched is initialized, it is ready to export it via debugfs */
+	blk_mq_sched_reg_debugfs(q);
+
 	return 0;
 
 err_free_map_and_rqs:
@@ -524,11 +544,10 @@ void blk_mq_exit_sched(struct request_queue *q, struct elevator_queue *e)
 	unsigned long i;
 	unsigned int flags = 0;
 
-	queue_for_each_hw_ctx(q, hctx, i) {
-		mutex_lock(&q->debugfs_mutex);
-		blk_mq_debugfs_unregister_sched_hctx(hctx);
-		mutex_unlock(&q->debugfs_mutex);
+	/* unexport via debugfs before exiting sched */
+	blk_mq_sched_unreg_debugfs(q);
 
+	queue_for_each_hw_ctx(q, hctx, i) {
 		if (e->type->ops.exit_hctx && hctx->sched_data) {
 			e->type->ops.exit_hctx(hctx, i);
 			hctx->sched_data = NULL;
@@ -536,10 +555,6 @@ void blk_mq_exit_sched(struct request_queue *q, struct elevator_queue *e)
 		flags = hctx->flags;
 	}
 
-	mutex_lock(&q->debugfs_mutex);
-	blk_mq_debugfs_unregister_sched(q);
-	mutex_unlock(&q->debugfs_mutex);
-
 	if (e->type->ops.exit_sched)
 		e->type->ops.exit_sched(e);
 	blk_mq_sched_tags_teardown(q, flags);
Signature · 1 line
2.47.0
author
#11 lore raw

PATCH v3 6/20 block: move sched debugfs register into elvevator_register_queue

In reply to Ming Lei

sched debugfs shares same lifetime with scheduler's kobject, and same lock(elevator lock), so move sched debugfs register/unregister into elevator_register_queue() and elevator_unregister_queue().

Then we needn't blk_mq_debugfs_register() for us to register sched debugfs any more.

---
 block/blk-mq-debugfs.c | 11 -----------
 block/blk-mq-sched.c   | 11 ++---------
 block/elevator.c       |  8 ++++++++
 block/elevator.h       |  3 +++
 4 files changed, 13 insertions(+), 20 deletions(-)
block/blk-mq-debugfs.c +0 -11
index 7710c409e432..2837a8ce8054 100644
@@ -625,20 +625,9 @@ void blk_mq_debugfs_register(struct request_queue *q)
 
 	debugfs_create_files(q->debugfs_dir, q, blk_mq_debugfs_queue_attrs);
 
-	/*
-	 * blk_mq_init_sched() attempted to do this already, but q->debugfs_dir
-	 * didn't exist yet (because we don't know what to name the directory
-	 * until the queue is registered to a gendisk).
-	 */
-	if (q->elevator && !q->sched_debugfs_dir)
-		blk_mq_debugfs_register_sched(q);
-
-	/* Similarly, blk_mq_init_hctx() couldn't do this previously. */
 	queue_for_each_hw_ctx(q, hctx, i) {
 		if (!hctx->debugfs_dir)
 			blk_mq_debugfs_register_hctx(q, hctx);
-		if (q->elevator && !hctx->sched_debugfs_dir)
-			blk_mq_debugfs_register_sched_hctx(q, hctx);
 	}
 
 	if (q->rq_qos) {
block/blk-mq-sched.c +2 -9
index 2abc5e0704e8..336a15ffecfa 100644
@@ -434,7 +434,7 @@ static int blk_mq_init_sched_shared_tags(struct request_queue *queue)
 	return 0;
 }
 
-static void blk_mq_sched_reg_debugfs(struct request_queue *q)
+void blk_mq_sched_reg_debugfs(struct request_queue *q)
 {
 	struct blk_mq_hw_ctx *hctx;
 	unsigned long i;
@@ -446,7 +446,7 @@ static void blk_mq_sched_reg_debugfs(struct request_queue *q)
 	mutex_unlock(&q->debugfs_mutex);
 }
 
-static void blk_mq_sched_unreg_debugfs(struct request_queue *q)
+void blk_mq_sched_unreg_debugfs(struct request_queue *q)
 {
 	struct blk_mq_hw_ctx *hctx;
 	unsigned long i;
@@ -503,10 +503,6 @@ int blk_mq_init_sched(struct request_queue *q, struct elevator_type *e)
 			}
 		}
 	}
-
-	/* sched is initialized, it is ready to export it via debugfs */
-	blk_mq_sched_reg_debugfs(q);
-
 	return 0;
 
 err_free_map_and_rqs:
@@ -544,9 +540,6 @@ void blk_mq_exit_sched(struct request_queue *q, struct elevator_queue *e)
 	unsigned long i;
 	unsigned int flags = 0;
 
-	/* unexport via debugfs before exiting sched */
-	blk_mq_sched_unreg_debugfs(q);
-
 	queue_for_each_hw_ctx(q, hctx, i) {
 		if (e->type->ops.exit_hctx && hctx->sched_data) {
 			e->type->ops.exit_hctx(hctx, i);
block/elevator.c +8 -0
index b32815594892..4400eb8fe54f 100644
@@ -472,6 +472,11 @@ int elv_register_queue(struct request_queue *q, bool uevent)
 		if (uevent)
 			kobject_uevent(&e->kobj, KOBJ_ADD);
 
+		/*
+		 * Sched is initialized, it is ready to export it via
+		 * debugfs
+		 */
+		blk_mq_sched_reg_debugfs(q);
 		set_bit(ELEVATOR_FLAG_REGISTERED, &e->flags);
 	}
 	return error;
@@ -486,6 +491,9 @@ void elv_unregister_queue(struct request_queue *q)
 	if (e && test_and_clear_bit(ELEVATOR_FLAG_REGISTERED, &e->flags)) {
 		kobject_uevent(&e->kobj, KOBJ_REMOVE);
 		kobject_del(&e->kobj);
+
+		/* unexport via debugfs before exiting sched */
+		blk_mq_sched_unreg_debugfs(q);
 	}
 }
 
block/elevator.h +3 -0
index e27af5492cdb..9198676644a9 100644
@@ -181,4 +181,7 @@ extern struct request *elv_rb_find(struct rb_root *, sector_t);
 #define rq_entry_fifo(ptr)	list_entry((ptr), struct request, queuelist)
 #define rq_fifo_clear(rq)	list_del_init(&(rq)->queuelist)
 
+void blk_mq_sched_reg_debugfs(struct request_queue *q);
+void blk_mq_sched_unreg_debugfs(struct request_queue *q);
+
 #endif /* _ELEVATOR_H */
Signature · 1 line
2.47.0
author
#12 lore raw

PATCH v3 7/20 block: prevent adding/deleting disk during updating nr_hw_queues

In reply to Ming Lei

Both adding/deleting disk code are reader of nr_hw_queues, so we can't allow them in-progress when updating nr_hw_queues, kernel panic and kasan has been reported in [1].

Prevent adding/deleting disk during updating nr_hw_queues by adding rw_semaphore to tagset, write lock is grabbed in blk_mq_update_nr_hw_queues(), and read lock is acquired when adding/deleting disk.

Also mark GFP_NOIO allocation scope for adding/deleting disk because blk_mq_update_nr_hw_queues() is part of some driver's error handler.

This way avoids lot of trouble.

---
 block/blk-mq.c         |  4 ++
 block/genhd.c          | 94 +++++++++++++++++++++++++++++++-----------
 include/linux/blk-mq.h |  3 ++
 3 files changed, 78 insertions(+), 23 deletions(-)
block/blk-mq.c +4 -0
index 29cfc7ce2e0a..1ed2d183f912 100644
@@ -4802,6 +4802,8 @@ int blk_mq_alloc_tag_set(struct blk_mq_tag_set *set)
 			goto out_free_srcu;
 	}
 
+	init_rwsem(&set->update_nr_hwq_sema);
+
 	ret = -ENOMEM;
 	set->tags = kcalloc_node(set->nr_hw_queues,
 				 sizeof(struct blk_mq_tags *), GFP_KERNEL,
@@ -5097,9 +5099,11 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
 
 void blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, int nr_hw_queues)
 {
+	down_write(&set->update_nr_hwq_sema);
 	mutex_lock(&set->tag_list_lock);
 	__blk_mq_update_nr_hw_queues(set, nr_hw_queues);
 	mutex_unlock(&set->tag_list_lock);
+	up_write(&set->update_nr_hwq_sema);
 }
 EXPORT_SYMBOL_GPL(blk_mq_update_nr_hw_queues);
 
block/genhd.c +71 -23

118 lines not on this page · raw message

include/linux/blk-mq.h +3 -0
index 8eb9b3310167..28bc03b2b0dc 100644
@@ -9,6 +9,7 @@
 #include <linux/prefetch.h>
 #include <linux/srcu.h>
 #include <linux/rw_hint.h>
+#include <linux/rwsem.h>
 
 struct blk_mq_tags;
 struct blk_flush_queue;
@@ -527,6 +528,8 @@ struct blk_mq_tag_set {
 	struct mutex		tag_list_lock;
 	struct list_head	tag_list;
 	struct srcu_struct	*srcu;
+
+	struct rw_semaphore	update_nr_hwq_sema;
 };
 
 /**
Signature · 1 line
2.47.0
Reviewed-by on v3 7/20
#13 lore raw

Re: [PATCH V3 07/20] block: prevent adding/deleting disk during updating nr_hw_queues

Quoted 17 lines · Ming Lei

Both adding/deleting disk code are reader of nr_hw_queues, so we can't allow them in-progress when updating nr_hw_queues, kernel panic and kasan has been reported in [1].

Prevent adding/deleting disk during updating nr_hw_queues by adding rw_semaphore to tagset, write lock is grabbed in blk_mq_update_nr_hw_queues(), and read lock is acquired when adding/deleting disk.

Also mark GFP_NOIO allocation scope for adding/deleting disk because blk_mq_update_nr_hw_queues() is part of some driver's error handler.

This way avoids lot of trouble.

Quoted 6 lines
---
  block/blk-mq.c         |  4 ++
  block/genhd.c          | 94 +++++++++++++++++++++++++++++++-----------
  include/linux/blk-mq.h |  3 ++
  3 files changed, 78 insertions(+), 23 deletions(-)

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
Reviewed-by on v3 7/20
#14 lore raw

Re: [PATCH V3 07/20] block: prevent adding/deleting disk during updating nr_hw_queues

In reply to Ming Lei
Quoted 11 lines · Ming Lei

Both adding/deleting disk code are reader of nr_hw_queues, so we can't allow them in-progress when updating nr_hw_queues, kernel panic and kasan has been reported in [1].

Prevent adding/deleting disk during updating nr_hw_queues by adding rw_semaphore to tagset, write lock is grabbed in blk_mq_update_nr_hw_queues(), and read lock is acquired when adding/deleting disk.

Also mark GFP_NOIO allocation scope for adding/deleting disk because blk_mq_update_nr_hw_queues() is part of some driver's error handler.

Quoted 6 lines

This way avoids lot of trouble.

Looks good to me:

Reviewed-by on v3 7/20
#15 lore raw

Re: [PATCH V3 07/20] block: prevent adding/deleting disk during updating nr_hw_queues

In reply to Ming Lei

block/blk-mq.c int blk_mq_alloc_tag_set(struct blk_mq_tag_set *set) · line 4805

Quoted 1 line · Ming Lei
block/blk-mq.c +1 -0
+	init_rwsem(&set->update_nr_hwq_sema);

Can you please call this update_nr_hwq_lock instead? _sema is a very unusual naming for a rw semaphore.

block/genhd.c int disk_scan_partitions(struct gendisk *disk, blk_mode_t mode) · line 404

Quoted 9 lines

* This function registers the partitioning information in @disk
* with the kernel. Also attach a fwnode to the disk device. */

block/genhd.c +3 -3
-int __must_check add_disk_fwnode(struct device *parent, struct gendisk *disk,
-				 const struct attribute_group **groups,
-				 struct fwnode_handle *fwnode)
+static int __add_disk_fwnode(struct device *parent, struct gendisk *disk,
+			     const struct attribute_group **groups,
+			     struct fwnode_handle *fwnode)

once this becomes internal there is no need for the _fwnode postfix and this can simply become __add_disk.

This probably wants a lockdep assert that te nr_hwq lock is held for the blk-mq case.

Also when you use two-tab indentation for the arguments you don't need to reformat for every little naming change.

Otherwise looks good:

#16 lore raw

Re: [PATCH V3 07/20] block: prevent adding/deleting disk during updating nr_hw_queues

In reply to Ming Lei
kernel test robot noticed the following build warnings:

Hi Ming,

kernel test robot noticed the following build warnings:

[auto build test WARNING on axboe-block/for-next]
[also build test WARNING on linus/master v6.15-rc3 next-20250424] [If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Ming-Lei/block-move-blk_mq_add_queue_tag_set-after-blk_mq_map_swqueue/20250424-232508
base:   https://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux-block.git for-next
patch link:    https://lore.kernel.org/r/20250424152148.1066220-8-ming.lei%40redhat.com
patch subject: [PATCH V3 07/20] block: prevent adding/deleting disk during updating nr_hw_queues
config: i386-buildonly-randconfig-004-20250425 (https://download.01.org/0day-ci/archive/20250426/[email protected]/config)
compiler: gcc-12 (Debian 12.2.0-14) 12.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250426/[email protected]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All warnings (new ones prefixed by >>):

>> block/genhd.c:406: warning: expecting prototype for add_disk_fwnode(). Prototype was for __add_disk_fwnode() instead

vim +406 block/genhd.c

9301fe73438499 block/genhd.c         Christoph Hellwig 2020-09-21  391  
^1da177e4c3f41 drivers/block/genhd.c Linus Torvalds    2005-04-16  392  /**
9dfd9ea93aeab5 block/genhd.c         Christian Marangi 2024-10-03  393   * add_disk_fwnode - add disk information to kernel list with fwnode
e63a46bef01ff3 block/genhd.c         Dan Williams      2016-06-15  394   * @parent: parent device for the disk
^1da177e4c3f41 drivers/block/genhd.c Linus Torvalds    2005-04-16  395   * @disk: per-device partitioning information
fef912bf860e8e block/genhd.c         Hannes Reinecke   2018-09-28  396   * @groups: Additional per-device sysfs groups
9dfd9ea93aeab5 block/genhd.c         Christian Marangi 2024-10-03  397   * @fwnode: attached disk fwnode
^1da177e4c3f41 drivers/block/genhd.c Linus Torvalds    2005-04-16  398   *
^1da177e4c3f41 drivers/block/genhd.c Linus Torvalds    2005-04-16  399   * This function registers the partitioning information in @disk
9dfd9ea93aeab5 block/genhd.c         Christian Marangi 2024-10-03  400   * with the kernel. Also attach a fwnode to the disk device.
^1da177e4c3f41 drivers/block/genhd.c Linus Torvalds    2005-04-16  401   */
54650b49b7412c block/genhd.c         Ming Lei          2025-04-24  402  static int __add_disk_fwnode(struct device *parent, struct gendisk *disk,
9dfd9ea93aeab5 block/genhd.c         Christian Marangi 2024-10-03  403  			     const struct attribute_group **groups,
9dfd9ea93aeab5 block/genhd.c         Christian Marangi 2024-10-03  404  			     struct fwnode_handle *fwnode)
d1254a8749711e block/genhd.c         Christoph Hellwig 2021-08-04  405  
^1da177e4c3f41 drivers/block/genhd.c Linus Torvalds    2005-04-16 @406  {
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  407  	struct device *ddev = disk_to_dev(disk);
7c3f828b522b07 block/genhd.c         Christoph Hellwig 2021-05-21  408  	int ret;
cf0ca9fe5dd9e3 block/genhd.c         Peter Zijlstra    2008-04-30  409  
3d9a9e9a77c5eb block/genhd.c         Ming Lei          2025-01-15  410  	if (WARN_ON_ONCE(bdev_nr_sectors(disk->part0) > BLK_DEV_MAX_SECTORS))
3d9a9e9a77c5eb block/genhd.c         Ming Lei          2025-01-15  411  		return -EINVAL;
3d9a9e9a77c5eb block/genhd.c         Ming Lei          2025-01-15  412  
6783811569aef2 block/genhd.c         Christoph Hellwig 2025-01-06  413  	if (queue_is_mq(disk->queue)) {
6783811569aef2 block/genhd.c         Christoph Hellwig 2025-01-06  414  		/*
6783811569aef2 block/genhd.c         Christoph Hellwig 2025-01-06  415  		 * ->submit_bio and ->poll_bio are bypassed for blk-mq drivers.
6783811569aef2 block/genhd.c         Christoph Hellwig 2025-01-06  416  		 */
6783811569aef2 block/genhd.c         Christoph Hellwig 2025-01-06  417  		if (disk->fops->submit_bio || disk->fops->poll_bio)
69fe0f29892077 block/genhd.c         Ming Lei          2022-03-04  418  			return -EINVAL;
69fe0f29892077 block/genhd.c         Ming Lei          2022-03-04  419  
737eb78e82d52d block/genhd.c         Damien Le Moal    2019-09-05  420  		/*
6783811569aef2 block/genhd.c         Christoph Hellwig 2025-01-06  421  		 * Initialize the I/O scheduler code and pick a default one if
6783811569aef2 block/genhd.c         Christoph Hellwig 2025-01-06  422  		 * needed.
737eb78e82d52d block/genhd.c         Damien Le Moal    2019-09-05  423  		 */
737eb78e82d52d block/genhd.c         Damien Le Moal    2019-09-05  424  		elevator_init_mq(disk->queue);
6783811569aef2 block/genhd.c         Christoph Hellwig 2025-01-06  425  	} else {
6783811569aef2 block/genhd.c         Christoph Hellwig 2025-01-06  426  		if (!disk->fops->submit_bio)
6783811569aef2 block/genhd.c         Christoph Hellwig 2025-01-06  427  			return -EINVAL;
ac2b6f9dee8f41 block/genhd.c         Al Viro           2024-04-12  428  		bdev_set_flag(disk->part0, BD_HAS_SUBMIT_BIO);
6783811569aef2 block/genhd.c         Christoph Hellwig 2025-01-06  429  	}
9f4107b07b17b5 block/genhd.c         Jens Axboe        2023-04-14  430  
7c3f828b522b07 block/genhd.c         Christoph Hellwig 2021-05-21  431  	/*
7c3f828b522b07 block/genhd.c         Christoph Hellwig 2021-05-21  432  	 * If the driver provides an explicit major number it also must provide
7c3f828b522b07 block/genhd.c         Christoph Hellwig 2021-05-21  433  	 * the number of minors numbers supported, and those will be used to
7c3f828b522b07 block/genhd.c         Christoph Hellwig 2021-05-21  434  	 * setup the gendisk.
7c3f828b522b07 block/genhd.c         Christoph Hellwig 2021-05-21  435  	 * Otherwise just allocate the device numbers for both the whole device
7c3f828b522b07 block/genhd.c         Christoph Hellwig 2021-05-21  436  	 * and all partitions from the extended dev_t space.
3e1a7ff8a0a7b9 block/genhd.c         Tejun Heo         2008-08-25  437  	 */
02341a08c9dec5 block/genhd.c         Yu Kuai           2022-10-22  438  	ret = -EINVAL;
7c3f828b522b07 block/genhd.c         Christoph Hellwig 2021-05-21  439  	if (disk->major) {
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  440  		if (WARN_ON(!disk->minors))
02341a08c9dec5 block/genhd.c         Yu Kuai           2022-10-22  441  			goto out_exit_elevator;
2e3c73fa0c419f block/genhd.c         Christoph Hellwig 2021-05-21  442  
2e3c73fa0c419f block/genhd.c         Christoph Hellwig 2021-05-21  443  		if (disk->minors > DISK_MAX_PARTS) {
2e3c73fa0c419f block/genhd.c         Christoph Hellwig 2021-05-21  444  			pr_err("block: can't allocate more than %d partitions\n",
2e3c73fa0c419f block/genhd.c         Christoph Hellwig 2021-05-21  445  				DISK_MAX_PARTS);
2e3c73fa0c419f block/genhd.c         Christoph Hellwig 2021-05-21  446  			disk->minors = DISK_MAX_PARTS;
2e3c73fa0c419f block/genhd.c         Christoph Hellwig 2021-05-21  447  		}
4c434392c47778 block/genhd.c         Li Nan            2023-12-19  448  		if (disk->first_minor > MINORMASK ||
4c434392c47778 block/genhd.c         Li Nan            2023-12-19  449  		    disk->minors > MINORMASK + 1 ||
4c434392c47778 block/genhd.c         Li Nan            2023-12-19  450  		    disk->first_minor + disk->minors > MINORMASK + 1)
02341a08c9dec5 block/genhd.c         Yu Kuai           2022-10-22  451  			goto out_exit_elevator;
7c3f828b522b07 block/genhd.c         Christoph Hellwig 2021-05-21  452  	} else {
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  453  		if (WARN_ON(disk->minors))
02341a08c9dec5 block/genhd.c         Yu Kuai           2022-10-22  454  			goto out_exit_elevator;
3e1a7ff8a0a7b9 block/genhd.c         Tejun Heo         2008-08-25  455  
7c3f828b522b07 block/genhd.c         Christoph Hellwig 2021-05-21  456  		ret = blk_alloc_ext_minor();
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  457  		if (ret < 0)
02341a08c9dec5 block/genhd.c         Yu Kuai           2022-10-22  458  			goto out_exit_elevator;
7c3f828b522b07 block/genhd.c         Christoph Hellwig 2021-05-21  459  		disk->major = BLOCK_EXT_MAJOR;
539711d7d6fe38 block/genhd.c         Christoph Hellwig 2021-08-24  460  		disk->first_minor = ret;
7c3f828b522b07 block/genhd.c         Christoph Hellwig 2021-05-21  461  	}
7c3f828b522b07 block/genhd.c         Christoph Hellwig 2021-05-21  462  
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  463  	/* delay uevents, until we scanned partition table */
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  464  	dev_set_uevent_suppress(ddev, 1);
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  465  
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  466  	ddev->parent = parent;
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  467  	ddev->groups = groups;
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  468  	dev_set_name(ddev, "%s", disk->disk_name);
9dfd9ea93aeab5 block/genhd.c         Christian Marangi 2024-10-03  469  	if (fwnode)
9dfd9ea93aeab5 block/genhd.c         Christian Marangi 2024-10-03  470  		device_set_node(ddev, fwnode);
8235b5c1e8c1c0 block/genhd.c         Christoph Hellwig 2021-08-18  471  	if (!(disk->flags & GENHD_FL_HIDDEN))
8235b5c1e8c1c0 block/genhd.c         Christoph Hellwig 2021-08-18  472  		ddev->devt = MKDEV(disk->major, disk->first_minor);
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  473  	ret = device_add(ddev);
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  474  	if (ret)
99d8690aae4b2f block/genhd.c         Christoph Hellwig 2021-12-21  475  		goto out_free_ext_minor;
99d8690aae4b2f block/genhd.c         Christoph Hellwig 2021-12-21  476  
99d8690aae4b2f block/genhd.c         Christoph Hellwig 2021-12-21  477  	ret = disk_alloc_events(disk);
99d8690aae4b2f block/genhd.c         Christoph Hellwig 2021-12-21  478  	if (ret)
99d8690aae4b2f block/genhd.c         Christoph Hellwig 2021-12-21  479  		goto out_device_del;
99d8690aae4b2f block/genhd.c         Christoph Hellwig 2021-12-21  480  
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  481  	ret = sysfs_create_link(block_depr, &ddev->kobj,
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  482  				kobject_name(&ddev->kobj));
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  483  	if (ret)
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  484  		goto out_device_del;
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  485  
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  486  	/*
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  487  	 * avoid probable deadlock caused by allocating memory with
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  488  	 * GFP_KERNEL in runtime_resume callback of its all ancestor
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  489  	 * devices
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  490  	 */
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  491  	pm_runtime_set_memalloc_noio(ddev, true);
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  492  
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  493  	disk->part0->bd_holder_dir =
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  494  		kobject_create_and_add("holders", &ddev->kobj);
fe7d064fa3faec block/genhd.c         Luis Chamberlain  2021-11-03  495  	if (!disk->part0->bd_holder_dir) {
fe7d064fa3faec block/genhd.c         Luis Chamberlain  2021-11-03  496  		ret = -ENOMEM;
ff53cd52d9bdbf block/genhd.c         Thomas Weißschuh  2023-03-18  497  		goto out_del_block_link;
fe7d064fa3faec block/genhd.c         Luis Chamberlain  2021-11-03  498  	}
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  499  	disk->slave_dir = kobject_create_and_add("slaves", &ddev->kobj);
fe7d064fa3faec block/genhd.c         Luis Chamberlain  2021-11-03  500  	if (!disk->slave_dir) {
fe7d064fa3faec block/genhd.c         Luis Chamberlain  2021-11-03  501  		ret = -ENOMEM;
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  502  		goto out_put_holder_dir;
fe7d064fa3faec block/genhd.c         Luis Chamberlain  2021-11-03  503  	}
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  504  
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  505  	ret = blk_register_queue(disk);
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  506  	if (ret)
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  507  		goto out_put_slave_dir;
75f4dca59694df block/genhd.c         Christoph Hellwig 2021-08-18  508  
9f18db572c97bc block/genhd.c         Christoph Hellwig 2021-11-22  509  	if (!(disk->flags & GENHD_FL_HIDDEN)) {
8235b5c1e8c1c0 block/genhd.c         Christoph Hellwig 2021-08-18  510  		ret = bdi_register(disk->bdi, "%u:%u",
8235b5c1e8c1c0 block/genhd.c         Christoph Hellwig 2021-08-18  511  				   disk->major, disk->first_minor);
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  512  		if (ret)
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  513  			goto out_unregister_queue;
8235b5c1e8c1c0 block/genhd.c         Christoph Hellwig 2021-08-18  514  		bdi_set_owner(disk->bdi, ddev);
9d5ee6767c8576 block/genhd.c         Christoph Hellwig 2021-08-18  515  		ret = sysfs_create_link(&ddev->kobj,
9d5ee6767c8576 block/genhd.c         Christoph Hellwig 2021-08-18  516  					&disk->bdi->dev->kobj, "bdi");
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  517  		if (ret)
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  518  			goto out_unregister_bdi;
8235b5c1e8c1c0 block/genhd.c         Christoph Hellwig 2021-08-18  519  
e5cfefa97bccf9 block/genhd.c         Yu Kuai           2023-02-17  520  		/* Make sure the first partition scan will be proceed */
140ce28dd3bee8 block/genhd.c         Christoph Hellwig 2024-05-02  521  		if (get_capacity(disk) && disk_has_partscan(disk))
e5cfefa97bccf9 block/genhd.c         Yu Kuai           2023-02-17  522  			set_bit(GD_NEED_PART_SCAN, &disk->state);
e5cfefa97bccf9 block/genhd.c         Yu Kuai           2023-02-17  523  
9d5ee6767c8576 block/genhd.c         Christoph Hellwig 2021-08-18  524  		bdev_add(disk->part0, ddev->devt);
e16e506ccd673a block/genhd.c         Christoph Hellwig 2021-11-22  525  		if (get_capacity(disk))
05bdb9965305bb block/genhd.c         Christoph Hellwig 2023-06-08  526  			disk_scan_partitions(disk, BLK_OPEN_READ);
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  527  
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  528  		/*
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  529  		 * Announce the disk and partitions after all partitions are
8235b5c1e8c1c0 block/genhd.c         Christoph Hellwig 2021-08-18  530  		 * created. (for hidden disks uevents remain suppressed forever)
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  531  		 */
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  532  		dev_set_uevent_suppress(ddev, 0);
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  533  		disk_uevent(disk, KOBJ_ADD);
a0a6314ae774f8 block/genhd.c         Christoph Hellwig 2022-10-10  534  	} else {
a0a6314ae774f8 block/genhd.c         Christoph Hellwig 2022-10-10  535  		/*
a0a6314ae774f8 block/genhd.c         Christoph Hellwig 2022-10-10  536  		 * Even if the block_device for a hidden gendisk is not
a0a6314ae774f8 block/genhd.c         Christoph Hellwig 2022-10-10  537  		 * registered, it needs to have a valid bd_dev so that the
a0a6314ae774f8 block/genhd.c         Christoph Hellwig 2022-10-10  538  		 * freeing of the dynamic major works.
a0a6314ae774f8 block/genhd.c         Christoph Hellwig 2022-10-10  539  		 */
a0a6314ae774f8 block/genhd.c         Christoph Hellwig 2022-10-10  540  		disk->part0->bd_dev = MKDEV(disk->major, disk->first_minor);
8ddcd653257c18 block/genhd.c         Christoph Hellwig 2017-11-02  541  	}
52b85909f85d06 block/genhd.c         Christoph Hellwig 2021-08-18  542  
73781b3b81e765 block/genhd.c         Christoph Hellwig 2024-06-26  543  	blk_apply_bdi_limits(disk->bdi, &disk->queue->limits);
77ea887e433ad8 block/genhd.c         Tejun Heo         2010-12-08  544  	disk_add_events(disk);
76792055c4c8b2 block/genhd.c         Christoph Hellwig 2022-02-15  545  	set_bit(GD_ADDED, &disk->state);
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  546  	return 0;
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  547  
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  548  out_unregister_bdi:
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  549  	if (!(disk->flags & GENHD_FL_HIDDEN))
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  550  		bdi_unregister(disk->bdi);
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  551  out_unregister_queue:
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  552  	blk_unregister_queue(disk);
fa81cbafbf5764 block/genhd.c         Chen Zhongjin     2022-10-29  553  	rq_qos_exit(disk->queue);
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  554  out_put_slave_dir:
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  555  	kobject_put(disk->slave_dir);
d90db3b1c8676b block/genhd.c         Christoph Hellwig 2022-11-15  556  	disk->slave_dir = NULL;
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  557  out_put_holder_dir:
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  558  	kobject_put(disk->part0->bd_holder_dir);
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  559  out_del_block_link:
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  560  	sysfs_remove_link(block_depr, dev_name(ddev));
5fa3d1a00c2d4b block/genhd.c         Li Nan            2023-12-11  561  	pm_runtime_set_memalloc_noio(ddev, false);
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  562  out_device_del:
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  563  	device_del(ddev);
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  564  out_free_ext_minor:
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  565  	if (disk->major == BLOCK_EXT_MAJOR)
83cbce9574462c block/genhd.c         Luis Chamberlain  2021-08-18  566  		blk_free_ext_minor(disk->first_minor);
02341a08c9dec5 block/genhd.c         Yu Kuai           2022-10-22  567  out_exit_elevator:
1bf70d08cc3b55 block/genhd.c         Nilay Shroff      2025-03-04  568  	if (disk->queue->elevator) {
1bf70d08cc3b55 block/genhd.c         Nilay Shroff      2025-03-04  569  		mutex_lock(&disk->queue->elevator_lock);
02341a08c9dec5 block/genhd.c         Yu Kuai           2022-10-22  570  		elevator_exit(disk->queue);
1bf70d08cc3b55 block/genhd.c         Nilay Shroff      2025-03-04  571  		mutex_unlock(&disk->queue->elevator_lock);
1bf70d08cc3b55 block/genhd.c         Nilay Shroff      2025-03-04  572  	}
278167fd2f8ffe block/genhd.c         Luis Chamberlain  2021-11-09  573  	return ret;
^1da177e4c3f41 drivers/block/genhd.c Linus Torvalds    2005-04-16  574  }
54650b49b7412c block/genhd.c         Ming Lei          2025-04-24  575  
Signature · 2 lines
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
author
#17 lore raw

PATCH v3 8/20 block: don't allow to switch elevator if updating nr_hw_queues is in-progress

In reply to Ming Lei

Elevator switch code is another nr_hw_queue reader in non-fast-IO code path, so it can't be done if updating nr_hw_queues is in-progress.

Take same approach with not allowing add/del disk when updating nr_hw_queues is in-progress, by grabbing read lock of set->update_nr_hwq_sema.

Take the nested variant for avoiding the following false positive splat[1], and this way is correct because:

- the read lock in elv_iosched_store() is not overlapped with the read lock in adding/deleting disk:

- kobject attribute is only available after the kobject is added and before it is deleted

-> #4 (&q->q_usage_counter(queue){++++}-{0:0}:
-> #3 (&q->limits_lock){+.+.}-{4:4}:
-> #2 (&disk->open_mutex){+.+.}-{4:4}:
-> #1 (&set->update_nr_hwq_lock){.+.+}-{4:4}:
-> #0 (kn->active#103){++++}-{0:0}:

---
 block/elevator.c | 3 +++
 1 file changed, 3 insertions(+)
block/elevator.c +3 -0
index 4400eb8fe54f..56da6ab7691a 100644
@@ -723,6 +723,7 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 	int ret;
 	unsigned int memflags;
 	struct request_queue *q = disk->queue;
+	struct blk_mq_tag_set *set = q->tag_set;
 
 	/*
 	 * If the attribute needs to load a module, do it before freezing the
@@ -734,6 +735,7 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 
 	elv_iosched_load_module(name);
 
+	down_read_nested(&set->update_nr_hwq_sema, 1);
 	memflags = blk_mq_freeze_queue(q);
 	mutex_lock(&q->elevator_lock);
 	ret = elevator_change(q, name);
@@ -741,6 +743,7 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 		ret = count;
 	mutex_unlock(&q->elevator_lock);
 	blk_mq_unfreeze_queue(q, memflags);
+	up_read(&set->update_nr_hwq_sema);
 	return ret;
 }
 
Signature · 1 line
2.47.0
Reviewed-by on v3 8/20
#18 lore raw

Re: [PATCH V3 08/20] block: don't allow to switch elevator if updating nr_hw_queues is in-progress

Quoted 22 lines · Ming Lei

Elevator switch code is another nr_hw_queue reader in non-fast-IO code path, so it can't be done if updating nr_hw_queues is in-progress.

Take same approach with not allowing add/del disk when updating nr_hw_queues is in-progress, by grabbing read lock of set->update_nr_hwq_sema.

Take the nested variant for avoiding the following false positive splat[1], and this way is correct because:

- the read lock in elv_iosched_store() is not overlapped with the read lock in adding/deleting disk:

- kobject attribute is only available after the kobject is added and before it is deleted

   -> #4 (&q->q_usage_counter(queue){++++}-{0:0}:
   -> #3 (&q->limits_lock){+.+.}-{4:4}:
   -> #2 (&disk->open_mutex){+.+.}-{4:4}:
   -> #1 (&set->update_nr_hwq_lock){.+.+}-{4:4}:
   -> #0 (kn->active#103){++++}-{0:0}:
Quoted 8 lines
---
  block/elevator.c | 3 +++
  1 file changed, 3 insertions(+)

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
#19 lore raw

Re: [PATCH V3 08/20] block: don't allow to switch elevator if updating nr_hw_queues is in-progress

In reply to Ming Lei
Quoted 43 lines · Ming Lei

Elevator switch code is another nr_hw_queue reader in non-fast-IO code path, so it can't be done if updating nr_hw_queues is in-progress.

Take same approach with not allowing add/del disk when updating nr_hw_queues is in-progress, by grabbing read lock of set->update_nr_hwq_sema.

Take the nested variant for avoiding the following false positive splat[1], and this way is correct because:

- the read lock in elv_iosched_store() is not overlapped with the read lock in adding/deleting disk:

- kobject attribute is only available after the kobject is added and before it is deleted

-> #4 (&q->q_usage_counter(queue){++++}-{0:0}:
-> #3 (&q->limits_lock){+.+.}-{4:4}:
-> #2 (&disk->open_mutex){+.+.}-{4:4}:
-> #1 (&set->update_nr_hwq_lock){.+.+}-{4:4}:
-> #0 (kn->active#103){++++}-{0:0}:

---
 block/elevator.c | 3 +++
 1 file changed, 3 insertions(+)
block/elevator.c +1 -0
index 4400eb8fe54f..56da6ab7691a 100644
@@ -723,6 +723,7 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 	int ret;
 	unsigned int memflags;
 	struct request_queue *q = disk->queue;
+	struct blk_mq_tag_set *set = q->tag_set;
 
 	/*
 	 * If the attribute needs to load a module, do it before freezing the

block/elevator.c ssize_t elv_iosched_store(struct gendisk *disk, const char *buf, · line 738

Quoted 4 lines
block/elevator.c +1 -0
@@ -734,6 +735,7 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 
 	elv_iosched_load_module(name);
 
+	down_read_nested(&set->update_nr_hwq_sema, 1);

Why do we need to add nested read lock here? The lockdep splat[1] which you reported earlier is possibly due to the same reader lock being acquired recursively in elv_iosched_store and then elevator_change?

On another note, if we suspect possible one-depth recursion for the same class of lock then then we should use SINGLE_DEPTH_NESTING (instead of using 1 here) for subclass. But still I am not clear why this lock needs nesting.

Thanks,
--Nilay

author
#20 lore raw

Re: [PATCH V3 08/20] block: don't allow to switch elevator if updating nr_hw_queues is in-progress

Quoted 47 lines · Nilay Shroff
Quoted 44 lines · Ming Lei

Elevator switch code is another nr_hw_queue reader in non-fast-IO code path, so it can't be done if updating nr_hw_queues is in-progress.

Take same approach with not allowing add/del disk when updating nr_hw_queues is in-progress, by grabbing read lock of set->update_nr_hwq_sema.

Take the nested variant for avoiding the following false positive splat[1], and this way is correct because:

- the read lock in elv_iosched_store() is not overlapped with the read lock in adding/deleting disk:

- kobject attribute is only available after the kobject is added and before it is deleted

-> #4 (&q->q_usage_counter(queue){++++}-{0:0}:
-> #3 (&q->limits_lock){+.+.}-{4:4}:
-> #2 (&disk->open_mutex){+.+.}-{4:4}:
-> #1 (&set->update_nr_hwq_lock){.+.+}-{4:4}:
-> #0 (kn->active#103){++++}-{0:0}:

---
 block/elevator.c | 3 +++
 1 file changed, 3 insertions(+)
block/elevator.c +1 -0
index 4400eb8fe54f..56da6ab7691a 100644
@@ -723,6 +723,7 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 	int ret;
 	unsigned int memflags;
 	struct request_queue *q = disk->queue;
+	struct blk_mq_tag_set *set = q->tag_set;
 
 	/*
 	 * If the attribute needs to load a module, do it before freezing the
@@ -734,6 +735,7 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 
Quoted 7 lines

block/elevator.c ssize_t elv_iosched_store(struct gendisk *disk, const char *buf, · line 738

Quoted 3 lines
block/elevator.c +1 -0
 	elv_iosched_load_module(name);
 
+	down_read_nested(&set->update_nr_hwq_sema, 1);

Why do we need to add nested read lock here? The lockdep splat[1] which you reported earlier is possibly due to the same reader lock being acquired recursively in elv_iosched_store and then elevator_change?

The splat isn't related with the nested read lock.

If you replace down_read_nested() with down_read(), the same splat can be triggered again when running blktests block/001.

Quoted 4 lines

On another note, if we suspect possible one-depth recursion for the same class of lock then then we should use SINGLE_DEPTH_NESTING (instead of using 1 here) for subclass. But still I am not clear why this lock needs nesting.

It is just one false positive, because elv_iosched_store() won't happen when adding disk.

Thanks,
Ming

#21 lore raw

Re: [PATCH V3 08/20] block: don't allow to switch elevator if updating nr_hw_queues is in-progress

Quoted 56 lines · Ming Lei
Quoted 47 lines · Nilay Shroff
Quoted 44 lines · Ming Lei

Elevator switch code is another nr_hw_queue reader in non-fast-IO code path, so it can't be done if updating nr_hw_queues is in-progress.

Take same approach with not allowing add/del disk when updating nr_hw_queues is in-progress, by grabbing read lock of set->update_nr_hwq_sema.

Take the nested variant for avoiding the following false positive splat[1], and this way is correct because:

- the read lock in elv_iosched_store() is not overlapped with the read lock in adding/deleting disk:

- kobject attribute is only available after the kobject is added and before it is deleted

-> #4 (&q->q_usage_counter(queue){++++}-{0:0}:
-> #3 (&q->limits_lock){+.+.}-{4:4}:
-> #2 (&disk->open_mutex){+.+.}-{4:4}:
-> #1 (&set->update_nr_hwq_lock){.+.+}-{4:4}:
-> #0 (kn->active#103){++++}-{0:0}:

---
 block/elevator.c | 3 +++
 1 file changed, 3 insertions(+)
block/elevator.c +1 -0
index 4400eb8fe54f..56da6ab7691a 100644
@@ -723,6 +723,7 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 	int ret;
 	unsigned int memflags;
 	struct request_queue *q = disk->queue;
+	struct blk_mq_tag_set *set = q->tag_set;
 
 	/*
 	 * If the attribute needs to load a module, do it before freezing the
@@ -734,6 +735,7 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 
Quoted 7 lines

block/elevator.c ssize_t elv_iosched_store(struct gendisk *disk, const char *buf, · line 738

Quoted 3 lines
block/elevator.c +1 -0
 	elv_iosched_load_module(name);
 
+	down_read_nested(&set->update_nr_hwq_sema, 1);

Why do we need to add nested read lock here? The lockdep splat[1] which you reported earlier is possibly due to the same reader lock being acquired recursively in elv_iosched_store and then elevator_change?

Quoted 5 lines

The splat isn't related with the nested read lock.

If you replace down_read_nested() with down_read(), the same splat can be triggered again when running blktests block/001.

I couldn't recreate it on my setup using above blktests.

Quoted 8 lines
Quoted 4 lines

On another note, if we suspect possible one-depth recursion for the same class of lock then then we should use SINGLE_DEPTH_NESTING (instead of using 1 here) for subclass. But still I am not clear why this lock needs nesting.

It is just one false positive, because elv_iosched_store() won't happen when adding disk.

Yes, but how could we avoid false positive? It's probably because of commit
ffa1e7ada456 ("block: Make request_queue lockdep splats show up earlier"). How about adding manual dependency of fs-reclaim on freeze-lock after we add the disk. Currently that manual dependency is added in blk_alloc_queue.

Thanks,
--Nilay

author
#22 lore raw

Re: [PATCH V3 08/20] block: don't allow to switch elevator if updating nr_hw_queues is in-progress

Quoted 59 lines · Nilay Shroff
Quoted 56 lines · Ming Lei
Quoted 54 lines · Nilay Shroff
Quoted 43 lines · Ming Lei

Elevator switch code is another nr_hw_queue reader in non-fast-IO code path, so it can't be done if updating nr_hw_queues is in-progress.

Take same approach with not allowing add/del disk when updating nr_hw_queues is in-progress, by grabbing read lock of set->update_nr_hwq_sema.

Take the nested variant for avoiding the following false positive splat[1], and this way is correct because:

- the read lock in elv_iosched_store() is not overlapped with the read lock in adding/deleting disk:

- kobject attribute is only available after the kobject is added and before it is deleted

-> #4 (&q->q_usage_counter(queue){++++}-{0:0}:
-> #3 (&q->limits_lock){+.+.}-{4:4}:
-> #2 (&disk->open_mutex){+.+.}-{4:4}:
-> #1 (&set->update_nr_hwq_lock){.+.+}-{4:4}:
-> #0 (kn->active#103){++++}-{0:0}:

---
 block/elevator.c | 3 +++
 1 file changed, 3 insertions(+)
block/elevator.c +1 -0
index 4400eb8fe54f..56da6ab7691a 100644
@@ -723,6 +723,7 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 	int ret;
 	unsigned int memflags;
 	struct request_queue *q = disk->queue;
+	struct blk_mq_tag_set *set = q->tag_set;
 
 	/*
 	 * If the attribute needs to load a module, do it before freezing the

block/elevator.c ssize_t elv_iosched_store(struct gendisk *disk, const char *buf, · line 738

Quoted 4 lines
block/elevator.c +1 -0
@@ -734,6 +735,7 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 
 	elv_iosched_load_module(name);
 
+	down_read_nested(&set->update_nr_hwq_sema, 1);

Why do we need to add nested read lock here? The lockdep splat[1] which you reported earlier is possibly due to the same reader lock being acquired recursively in elv_iosched_store and then elevator_change?

Quoted 6 lines
Quoted 5 lines

The splat isn't related with the nested read lock.

If you replace down_read_nested() with down_read(), the same splat can be triggered again when running blktests block/001.

I couldn't recreate it on my setup using above blktests.

It is reproduced in my test vm every time after killing the nested variant:

[   74.257200] ======================================================
[   74.259369] WARNING: possible circular locking dependency detected
[   74.260772] 6.15.0-rc3_ublk+ #547 Not tainted
[   74.261950] ------------------------------------------------------
[   74.263281] check/5077 is trying to acquire lock:
[   74.264492] ffff888105f1fd18 (kn->active#119){++++}-{0:0}, at: __kernfs_remove+0x213/0x680
[   74.266006]
               but task is already holding lock:
[   74.267998] ffff88828a661e20 (&q->q_usage_counter(queue)#14){++++}-{0:0}, at: del_gendisk+0xe5/0x180
[   74.269631]
               which lock already depends on the new lock.
[   74.272645]
               the existing dependency chain (in reverse order) is:
[   74.274804]
               -> #3 (&q->q_usage_counter(queue)#14){++++}-{0:0}:
[   74.277009]        blk_queue_enter+0x4c2/0x630
[   74.278218]        blk_mq_alloc_request+0x479/0xa00
[   74.279539]        scsi_execute_cmd+0x151/0xba0
[   74.281078]        sr_check_events+0x1bc/0xa40
[   74.283012]        cdrom_check_events+0x5c/0x120
[   74.284892]        disk_check_events+0xbe/0x390
[   74.286181]        disk_check_media_change+0xf1/0x220
[   74.287455]        sr_block_open+0xce/0x230
[   74.288528]        blkdev_get_whole+0x8d/0x200
[   74.289702]        bdev_open+0x614/0xc60
[   74.290882]        blkdev_open+0x1f6/0x360
[   74.292215]        do_dentry_open+0x491/0x1820
[   74.293309]        vfs_open+0x7a/0x440
[   74.294384]        path_openat+0x1b7e/0x2ce0
[   74.295507]        do_filp_open+0x1c5/0x450
[   74.296616]        do_sys_openat2+0xef/0x180
[   74.297667]        __x64_sys_openat+0x10e/0x210
[   74.298768]        do_syscall_64+0x92/0x180
[   74.299800]        entry_SYSCALL_64_after_hwframe+0x76/0x7e
[   74.300971]
               -> #2 (&disk->open_mutex){+.+.}-{4:4}:
[   74.302700]        __mutex_lock+0x19c/0x1990
[   74.303682]        bdev_open+0x6cd/0xc60
[   74.304613]        bdev_file_open_by_dev+0xc4/0x140
[   74.306008]        disk_scan_partitions+0x191/0x290
[   74.307716]        __add_disk_fwnode+0xd2a/0x1140
[   74.309394]        add_disk_fwnode+0x10e/0x220
[   74.311039]        nvme_alloc_ns+0x1833/0x2c30
[   74.312669]        nvme_scan_ns+0x5a0/0x6f0
[   74.314151]        async_run_entry_fn+0x94/0x540
[   74.315719]        process_one_work+0x86a/0x14a0
[   74.317287]        worker_thread+0x5bb/0xf90
[   74.318228]        kthread+0x371/0x720
[   74.319085]        ret_from_fork+0x31/0x70
[   74.319941]        ret_from_fork_asm+0x1a/0x30
[   74.320808]
               -> #1 (&set->update_nr_hwq_sema){.+.+}-{4:4}:
[   74.322311]        down_read+0x8e/0x470
[   74.323135]        elv_iosched_store+0x17a/0x210
[   74.324036]        queue_attr_store+0x234/0x340
[   74.324881]        kernfs_fop_write_iter+0x39b/0x5a0
[   74.325771]        vfs_write+0x5df/0xec0
[   74.326514]        ksys_write+0xff/0x200
[   74.327262]        do_syscall_64+0x92/0x180
[   74.328018]        entry_SYSCALL_64_after_hwframe+0x76/0x7e
[   74.328963]
               -> #0 (kn->active#119){++++}-{0:0}:
[   74.330433]        __lock_acquire+0x145f/0x2260
[   74.331329]        lock_acquire+0x163/0x300
[   74.332221]        kernfs_drain+0x39d/0x450
[   74.333002]        __kernfs_remove+0x213/0x680
[   74.333792]        kernfs_remove_by_name_ns+0xa2/0x100
[   74.334589]        remove_files+0x8d/0x1b0
[   74.335326]        sysfs_remove_group+0x7c/0x160
[   74.336118]        sysfs_remove_groups+0x55/0xb0
[   74.336869]        __kobject_del+0x7d/0x1d0
[   74.337637]        kobject_del+0x38/0x60
[   74.338340]        blk_unregister_queue+0x153/0x2c0
[   74.339125]        __del_gendisk+0x252/0x9d0
[   74.339959]        del_gendisk+0xe5/0x180
[   74.340756]        sr_remove+0x7b/0xd0
[   74.341429]        device_release_driver_internal+0x36d/0x520
[   74.342353]        bus_remove_device+0x1ef/0x3f0
[   74.343172]        device_del+0x3be/0x9b0
[   74.343951]        __scsi_remove_device+0x27f/0x340
[   74.344724]        sdev_store_delete+0x87/0x120
[   74.345508]        kernfs_fop_write_iter+0x39b/0x5a0
[   74.346287]        vfs_write+0x5df/0xec0
[   74.347170]        ksys_write+0xff/0x200
[   74.348312]        do_syscall_64+0x92/0x180
[   74.349519]        entry_SYSCALL_64_after_hwframe+0x76/0x7e
[   74.350797]
               other info that might help us debug this:
[   74.353554] Chain exists of:
                 kn->active#119 --> &disk->open_mutex --> &q->q_usage_counter(queue)#14
[   74.355535]  Possible unsafe locking scenario:
[   74.356650]        CPU0                    CPU1
[   74.357328]        ----                    ----
[   74.358026]   lock(&q->q_usage_counter(queue)#14);
[   74.358749]                                lock(&disk->open_mutex);
[   74.359561]                                lock(&q->q_usage_counter(queue)#14);
[   74.360488]   lock(kn->active#119);
[   74.361113]
                *** DEADLOCK ***
[   74.362574] 6 locks held by check/5077:
[   74.363193]  #0: ffff888114640420 (sb_writers#4){.+.+}-{0:0}, at: ksys_write+0xff/0x200
[   74.364274]  #1: ffff88829abb6088 (&of->mutex#2){+.+.}-{4:4}, at: kernfs_fop_write_iter+0x25b/0x5a0
[   74.365937]  #2: ffff8881176ca0e0 (&shost->scan_mutex){+.+.}-{4:4}, at: sdev_store_delete+0x7f/0x120
[   74.367643]  #3: ffff88828521c380 (&dev->mutex){....}-{4:4}, at: device_release_driver_internal+0x90/0x520
[   74.369464]  #4: ffff8881176ca380 (&set->update_nr_hwq_sema){.+.+}-{4:4}, at: del_gendisk+0xdd/0x180
[   74.370961]  #5: ffff88828a661e20 (&q->q_usage_counter(queue)#14){++++}-{0:0}, at: del_gendisk+0xe5/0x180
[   74.372050]
               stack backtrace:
[   74.373111] CPU: 10 UID: 0 PID: 5077 Comm: check Not tainted 6.15.0-rc3_ublk+ #547 PREEMPT(voluntary)
[   74.373116] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-1.fc39 04/01/2014
[   74.373118] Call Trace:
[   74.373121]  <TASK>
[   74.373123]  dump_stack_lvl+0x84/0xd0
[   74.373129]  print_circular_bug.cold+0x185/0x1d0
[   74.373134]  check_noncircular+0x14a/0x170
[   74.373140]  __lock_acquire+0x145f/0x2260
[   74.373145]  lock_acquire+0x163/0x300
[   74.373149]  ? __kernfs_remove+0x213/0x680
[   74.373155]  kernfs_drain+0x39d/0x450
[   74.373158]  ? __kernfs_remove+0x213/0x680
[   74.373161]  ? __pfx_kernfs_drain+0x10/0x10
[   74.373165]  ? find_held_lock+0x2b/0x80
[   74.373168]  ? kernfs_root+0xb0/0x1c0
[   74.373173]  __kernfs_remove+0x213/0x680
[   74.373176]  ? kernfs_find_ns+0x197/0x390
[   74.373183]  kernfs_remove_by_name_ns+0xa2/0x100
[   74.373186]  remove_files+0x8d/0x1b0
[   74.373191]  sysfs_remove_group+0x7c/0x160
[   74.373194]  sysfs_remove_groups+0x55/0xb0
[   74.373198]  __kobject_del+0x7d/0x1d0
[   74.373203]  kobject_del+0x38/0x60
[   74.373206]  blk_unregister_queue+0x153/0x2c0
[   74.373210]  __del_gendisk+0x252/0x9d0
[   74.373214]  ? down_read+0x1a7/0x470
[   74.373218]  ? __pfx___del_gendisk+0x10/0x10
[   74.373221]  ? __pfx_down_read+0x10/0x10
[   74.373224]  ? lockdep_hardirqs_on_prepare+0xdb/0x190
[   74.373227]  ? trace_hardirqs_on+0x18/0x150
[   74.373231]  del_gendisk+0xe5/0x180
[   74.373235]  sr_remove+0x7b/0xd0
[   74.373239]  device_release_driver_internal+0x36d/0x520
[   74.373243]  ? kobject_put+0x5e/0x4a0
[   74.373246]  bus_remove_device+0x1ef/0x3f0
[   74.373250]  device_del+0x3be/0x9b0
[   74.373254]  ? attribute_container_device_trigger+0x181/0x1f0
[   74.373257]  ? __pfx_device_del+0x10/0x10
[   74.373260]  ? __pfx_attribute_container_device_trigger+0x10/0x10
[   74.373264]  __scsi_remove_device+0x27f/0x340
[   74.373267]  sdev_store_delete+0x87/0x120
[   74.373270]  ? __pfx_sysfs_kf_write+0x10/0x10
[   74.373273]  kernfs_fop_write_iter+0x39b/0x5a0
[   74.373276]  ? __pfx_kernfs_fop_write_iter+0x10/0x10
[   74.373278]  vfs_write+0x5df/0xec0
[   74.373282]  ? trace_hardirqs_on+0x18/0x150
[   74.373285]  ? __pfx_vfs_write+0x10/0x10
[   74.373291]  ksys_write+0xff/0x200
[   74.373295]  ? __pfx_ksys_write+0x10/0x10
[   74.373298]  ? fput_close_sync+0xd6/0x160
[   74.373303]  do_syscall_64+0x92/0x180
[   74.373309]  ? trace_hardirqs_on_prepare+0x101/0x150
[   74.373313]  ? lockdep_hardirqs_on_prepare+0xdb/0x190
[   74.373317]  ? syscall_exit_to_user_mode+0x97/0x290
[   74.373322]  ? do_syscall_64+0x9f/0x180
[   74.373330]  ? fput_close+0xd6/0x160
[   74.373333]  ? __pfx_fput_close+0x10/0x10
[   74.373338]  ? filp_close+0x25/0x40
[   74.373341]  ? do_dup2+0x287/0x4f0
[   74.373346]  ? trace_hardirqs_on_prepare+0x101/0x150
[   74.373348]  ? lockdep_hardirqs_on_prepare+0xdb/0x190
[   74.373351]  ? syscall_exit_to_user_mode+0x97/0x290
[   74.373353]  ? do_syscall_64+0x9f/0x180
[   74.373357]  ? trace_hardirqs_on_prepare+0x101/0x150
[   74.373359]  ? lockdep_hardirqs_on_prepare+0xdb/0x190
[   74.373362]  ? syscall_exit_to_user_mode+0x97/0x290
[   74.373364]  ? do_syscall_64+0x9f/0x180
[   74.373368]  ? do_syscall_64+0x9f/0x180
[   74.373371]  ? trace_hardirqs_on_prepare+0x101/0x150
[   74.373373]  ? lockdep_hardirqs_on_prepare+0xdb/0x190
[   74.373376]  ? syscall_exit_to_user_mode+0x97/0x290
[   74.373379]  ? do_syscall_64+0x9f/0x180
[   74.373382]  ? do_syscall_64+0x9f/0x180
[   74.373385]  ? clear_bhb_loop+0x35/0x90
[   74.373388]  ? clear_bhb_loop+0x35/0x90
[   74.373390]  ? clear_bhb_loop+0x35/0x90
[   74.373393]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
[   74.373396] RIP: 0033:0x7fa3873a8756
[   74.373409] Code: 5d e8 41 8b 93 08 03 00 00 59 5e 48 83 f8 fc 75 19 83 e2 39 83 fa 08 75 11 e8 26 ff ff ff 66 0f 1f 44 00 00 48 8b 45 10 0f 05 <48> 8b 5d f8 c9 c3 0f 1f 40 00 f3 0f 1e fa 55 48 89 e5 48 83 ec 08
[   74.373412] RSP: 002b:00007ffede0285d0 EFLAGS: 00000202 ORIG_RAX: 0000000000000001
[   74.373416] RAX: ffffffffffffffda RBX: 0000000000000002 RCX: 00007fa3873a8756
[   74.373418] RDX: 0000000000000002 RSI: 000056557b3dc7d0 RDI: 0000000000000001
[   74.373420] RBP: 00007ffede0285f0 R08: 0000000000000000 R09: 0000000000000000
[   74.373421] R10: 0000000000000000 R11: 0000000000000202 R12: 0000000000000002
[   74.373423] R13: 000056557b3dc7d0 R14: 00007fa3875245c0 R15: 0000000000000000
[   74.373428]  </TASK>
Quoted 6 lines
Quoted 6 lines
Quoted 5 lines

On another note, if we suspect possible one-depth recursion for the same class of lock then then we should use SINGLE_DEPTH_NESTING (instead of using 1 here) for subclass. But still I am not clear why this lock needs nesting.

Quoted 6 lines
Quoted 3 lines

It is just one false positive, because elv_iosched_store() won't happen when adding disk.

Yes, but how could we avoid false positive? It's probably because of commit
ffa1e7ada456 ("block: Make request_queue lockdep splats show up earlier"). How about adding manual dependency of fs-reclaim on freeze-lock after we add the disk. Currently that manual dependency is added in blk_alloc_queue.

Please see the above trace, which isn't related with commit ffa1e7ada456, and the lock chain doesn't include 'fs_reclaim' at all.

commit ffa1e7ada456 isn't wrong too, which just helps us to expose deadlock risk early.

Thanks,
Ming

#23 lore raw

Re: [PATCH V3 08/20] block: don't allow to switch elevator if updating nr_hw_queues is in-progress

Quoted 107 lines · Ming Lei
Quoted 1 line

I couldn't recreate it on my setup using above blktests.

It is reproduced in my test vm every time after killing the nested variant:

[   74.257200] ======================================================
[   74.259369] WARNING: possible circular locking dependency detected
[   74.260772] 6.15.0-rc3_ublk+ #547 Not tainted
[   74.261950] ------------------------------------------------------
[   74.263281] check/5077 is trying to acquire lock:
[   74.264492] ffff888105f1fd18 (kn->active#119){++++}-{0:0}, at: __kernfs_remove+0x213/0x680
[   74.266006]
               but task is already holding lock:
[   74.267998] ffff88828a661e20 (&q->q_usage_counter(queue)#14){++++}-{0:0}, at: del_gendisk+0xe5/0x180
[   74.269631]
               which lock already depends on the new lock.
[   74.272645]
               the existing dependency chain (in reverse order) is:
[   74.274804]
               -> #3 (&q->q_usage_counter(queue)#14){++++}-{0:0}:
[   74.277009]        blk_queue_enter+0x4c2/0x630
[   74.278218]        blk_mq_alloc_request+0x479/0xa00
[   74.279539]        scsi_execute_cmd+0x151/0xba0
[   74.281078]        sr_check_events+0x1bc/0xa40
[   74.283012]        cdrom_check_events+0x5c/0x120
[   74.284892]        disk_check_events+0xbe/0x390
[   74.286181]        disk_check_media_change+0xf1/0x220
[   74.287455]        sr_block_open+0xce/0x230
[   74.288528]        blkdev_get_whole+0x8d/0x200
[   74.289702]        bdev_open+0x614/0xc60
[   74.290882]        blkdev_open+0x1f6/0x360
[   74.292215]        do_dentry_open+0x491/0x1820
[   74.293309]        vfs_open+0x7a/0x440
[   74.294384]        path_openat+0x1b7e/0x2ce0
[   74.295507]        do_filp_open+0x1c5/0x450
[   74.296616]        do_sys_openat2+0xef/0x180
[   74.297667]        __x64_sys_openat+0x10e/0x210
[   74.298768]        do_syscall_64+0x92/0x180
[   74.299800]        entry_SYSCALL_64_after_hwframe+0x76/0x7e
[   74.300971]
               -> #2 (&disk->open_mutex){+.+.}-{4:4}:
[   74.302700]        __mutex_lock+0x19c/0x1990
[   74.303682]        bdev_open+0x6cd/0xc60
[   74.304613]        bdev_file_open_by_dev+0xc4/0x140
[   74.306008]        disk_scan_partitions+0x191/0x290
[   74.307716]        __add_disk_fwnode+0xd2a/0x1140
[   74.309394]        add_disk_fwnode+0x10e/0x220
[   74.311039]        nvme_alloc_ns+0x1833/0x2c30
[   74.312669]        nvme_scan_ns+0x5a0/0x6f0
[   74.314151]        async_run_entry_fn+0x94/0x540
[   74.315719]        process_one_work+0x86a/0x14a0
[   74.317287]        worker_thread+0x5bb/0xf90
[   74.318228]        kthread+0x371/0x720
[   74.319085]        ret_from_fork+0x31/0x70
[   74.319941]        ret_from_fork_asm+0x1a/0x30
[   74.320808]
               -> #1 (&set->update_nr_hwq_sema){.+.+}-{4:4}:
[   74.322311]        down_read+0x8e/0x470
[   74.323135]        elv_iosched_store+0x17a/0x210
[   74.324036]        queue_attr_store+0x234/0x340
[   74.324881]        kernfs_fop_write_iter+0x39b/0x5a0
[   74.325771]        vfs_write+0x5df/0xec0
[   74.326514]        ksys_write+0xff/0x200
[   74.327262]        do_syscall_64+0x92/0x180
[   74.328018]        entry_SYSCALL_64_after_hwframe+0x76/0x7e
[   74.328963]
               -> #0 (kn->active#119){++++}-{0:0}:
[   74.330433]        __lock_acquire+0x145f/0x2260
[   74.331329]        lock_acquire+0x163/0x300
[   74.332221]        kernfs_drain+0x39d/0x450
[   74.333002]        __kernfs_remove+0x213/0x680
[   74.333792]        kernfs_remove_by_name_ns+0xa2/0x100
[   74.334589]        remove_files+0x8d/0x1b0
[   74.335326]        sysfs_remove_group+0x7c/0x160
[   74.336118]        sysfs_remove_groups+0x55/0xb0
[   74.336869]        __kobject_del+0x7d/0x1d0
[   74.337637]        kobject_del+0x38/0x60
[   74.338340]        blk_unregister_queue+0x153/0x2c0
[   74.339125]        __del_gendisk+0x252/0x9d0
[   74.339959]        del_gendisk+0xe5/0x180
[   74.340756]        sr_remove+0x7b/0xd0
[   74.341429]        device_release_driver_internal+0x36d/0x520
[   74.342353]        bus_remove_device+0x1ef/0x3f0
[   74.343172]        device_del+0x3be/0x9b0
[   74.343951]        __scsi_remove_device+0x27f/0x340
[   74.344724]        sdev_store_delete+0x87/0x120
[   74.345508]        kernfs_fop_write_iter+0x39b/0x5a0
[   74.346287]        vfs_write+0x5df/0xec0
[   74.347170]        ksys_write+0xff/0x200
[   74.348312]        do_syscall_64+0x92/0x180
[   74.349519]        entry_SYSCALL_64_after_hwframe+0x76/0x7e
[   74.350797]
               other info that might help us debug this:
[   74.353554] Chain exists of:
                 kn->active#119 --> &disk->open_mutex --> &q->q_usage_counter(queue)#14
[   74.355535]  Possible unsafe locking scenario:
[   74.356650]        CPU0                    CPU1
[   74.357328]        ----                    ----
[   74.358026]   lock(&q->q_usage_counter(queue)#14);
[   74.358749]                                lock(&disk->open_mutex);
[   74.359561]                                lock(&q->q_usage_counter(queue)#14);
[   74.360488]   lock(kn->active#119);
[   74.361113]
                *** DEADLOCK ***
Quoted 8 lines
[   74.362574] 6 locks held by check/5077:
[   74.363193]  #0: ffff888114640420 (sb_writers#4){.+.+}-{0:0}, at: ksys_write+0xff/0x200
[   74.364274]  #1: ffff88829abb6088 (&of->mutex#2){+.+.}-{4:4}, at: kernfs_fop_write_iter+0x25b/0x5a0
[   74.365937]  #2: ffff8881176ca0e0 (&shost->scan_mutex){+.+.}-{4:4}, at: sdev_store_delete+0x7f/0x120
[   74.367643]  #3: ffff88828521c380 (&dev->mutex){....}-{4:4}, at: device_release_driver_internal+0x90/0x520
[   74.369464]  #4: ffff8881176ca380 (&set->update_nr_hwq_sema){.+.+}-{4:4}, at: del_gendisk+0xdd/0x180
[   74.370961]  #5: ffff88828a661e20 (&q->q_usage_counter(queue)#14){++++}-{0:0}, at: del_gendisk+0xe5/0x180
[   74.372050]

This has baffled me as I don't understand how could read lock in elv_iosched_store (ruuning in context #1) depends on (same) read lock in add_disk_fwnode (running under another context #2) as both locks are represented by the same rw semaphore. As we see above both elv_iosched_store and add_disk_fwnode bot run under different contexts and so ideally they should be able to run concurrently acquiring the same read lock.

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 4 lines

On another note, if we suspect possible one-depth recursion for the same class of lock then then we should use SINGLE_DEPTH_NESTING (instead of using 1 here) for subclass. But still I am not clear why this lock needs nesting.

It is just one false positive, because elv_iosched_store() won't happen when adding disk.

Quoted 8 lines
Quoted 3 lines

Yes, but how could we avoid false positive? It's probably because of commit
ffa1e7ada456 ("block: Make request_queue lockdep splats show up earlier"). How about adding manual dependency of fs-reclaim on freeze-lock after we add the disk. Currently that manual dependency is added in blk_alloc_queue.

Please see the above trace, which isn't related with commit ffa1e7ada456, and the lock chain doesn't include 'fs_reclaim' at all.

commit ffa1e7ada456 isn't wrong too, which just helps us to expose deadlock risk early.

Yes I am not against this commit and now looking at the above splat I don't blame this commit.

Thanks,
--Nilay

author
#24 lore raw

Re: [PATCH V3 08/20] block: don't allow to switch elevator if updating nr_hw_queues is in-progress

Quoted 119 lines · Nilay Shroff
Quoted 107 lines · Ming Lei
Quoted 1 line

I couldn't recreate it on my setup using above blktests.

It is reproduced in my test vm every time after killing the nested variant:

[   74.257200] ======================================================
[   74.259369] WARNING: possible circular locking dependency detected
[   74.260772] 6.15.0-rc3_ublk+ #547 Not tainted
[   74.261950] ------------------------------------------------------
[   74.263281] check/5077 is trying to acquire lock:
[   74.264492] ffff888105f1fd18 (kn->active#119){++++}-{0:0}, at: __kernfs_remove+0x213/0x680
[   74.266006]
               but task is already holding lock:
[   74.267998] ffff88828a661e20 (&q->q_usage_counter(queue)#14){++++}-{0:0}, at: del_gendisk+0xe5/0x180
[   74.269631]
               which lock already depends on the new lock.
[   74.272645]
               the existing dependency chain (in reverse order) is:
[   74.274804]
               -> #3 (&q->q_usage_counter(queue)#14){++++}-{0:0}:
[   74.277009]        blk_queue_enter+0x4c2/0x630
[   74.278218]        blk_mq_alloc_request+0x479/0xa00
[   74.279539]        scsi_execute_cmd+0x151/0xba0
[   74.281078]        sr_check_events+0x1bc/0xa40
[   74.283012]        cdrom_check_events+0x5c/0x120
[   74.284892]        disk_check_events+0xbe/0x390
[   74.286181]        disk_check_media_change+0xf1/0x220
[   74.287455]        sr_block_open+0xce/0x230
[   74.288528]        blkdev_get_whole+0x8d/0x200
[   74.289702]        bdev_open+0x614/0xc60
[   74.290882]        blkdev_open+0x1f6/0x360
[   74.292215]        do_dentry_open+0x491/0x1820
[   74.293309]        vfs_open+0x7a/0x440
[   74.294384]        path_openat+0x1b7e/0x2ce0
[   74.295507]        do_filp_open+0x1c5/0x450
[   74.296616]        do_sys_openat2+0xef/0x180
[   74.297667]        __x64_sys_openat+0x10e/0x210
[   74.298768]        do_syscall_64+0x92/0x180
[   74.299800]        entry_SYSCALL_64_after_hwframe+0x76/0x7e
[   74.300971]
               -> #2 (&disk->open_mutex){+.+.}-{4:4}:
[   74.302700]        __mutex_lock+0x19c/0x1990
[   74.303682]        bdev_open+0x6cd/0xc60
[   74.304613]        bdev_file_open_by_dev+0xc4/0x140
[   74.306008]        disk_scan_partitions+0x191/0x290
[   74.307716]        __add_disk_fwnode+0xd2a/0x1140
[   74.309394]        add_disk_fwnode+0x10e/0x220
[   74.311039]        nvme_alloc_ns+0x1833/0x2c30
[   74.312669]        nvme_scan_ns+0x5a0/0x6f0
[   74.314151]        async_run_entry_fn+0x94/0x540
[   74.315719]        process_one_work+0x86a/0x14a0
[   74.317287]        worker_thread+0x5bb/0xf90
[   74.318228]        kthread+0x371/0x720
[   74.319085]        ret_from_fork+0x31/0x70
[   74.319941]        ret_from_fork_asm+0x1a/0x30
[   74.320808]
               -> #1 (&set->update_nr_hwq_sema){.+.+}-{4:4}:
[   74.322311]        down_read+0x8e/0x470
[   74.323135]        elv_iosched_store+0x17a/0x210
[   74.324036]        queue_attr_store+0x234/0x340
[   74.324881]        kernfs_fop_write_iter+0x39b/0x5a0
[   74.325771]        vfs_write+0x5df/0xec0
[   74.326514]        ksys_write+0xff/0x200
[   74.327262]        do_syscall_64+0x92/0x180
[   74.328018]        entry_SYSCALL_64_after_hwframe+0x76/0x7e
[   74.328963]
               -> #0 (kn->active#119){++++}-{0:0}:
[   74.330433]        __lock_acquire+0x145f/0x2260
[   74.331329]        lock_acquire+0x163/0x300
[   74.332221]        kernfs_drain+0x39d/0x450
[   74.333002]        __kernfs_remove+0x213/0x680
[   74.333792]        kernfs_remove_by_name_ns+0xa2/0x100
[   74.334589]        remove_files+0x8d/0x1b0
[   74.335326]        sysfs_remove_group+0x7c/0x160
[   74.336118]        sysfs_remove_groups+0x55/0xb0
[   74.336869]        __kobject_del+0x7d/0x1d0
[   74.337637]        kobject_del+0x38/0x60
[   74.338340]        blk_unregister_queue+0x153/0x2c0
[   74.339125]        __del_gendisk+0x252/0x9d0
[   74.339959]        del_gendisk+0xe5/0x180
[   74.340756]        sr_remove+0x7b/0xd0
[   74.341429]        device_release_driver_internal+0x36d/0x520
[   74.342353]        bus_remove_device+0x1ef/0x3f0
[   74.343172]        device_del+0x3be/0x9b0
[   74.343951]        __scsi_remove_device+0x27f/0x340
[   74.344724]        sdev_store_delete+0x87/0x120
[   74.345508]        kernfs_fop_write_iter+0x39b/0x5a0
[   74.346287]        vfs_write+0x5df/0xec0
[   74.347170]        ksys_write+0xff/0x200
[   74.348312]        do_syscall_64+0x92/0x180
[   74.349519]        entry_SYSCALL_64_after_hwframe+0x76/0x7e
[   74.350797]
               other info that might help us debug this:
[   74.353554] Chain exists of:
                 kn->active#119 --> &disk->open_mutex --> &q->q_usage_counter(queue)#14
[   74.355535]  Possible unsafe locking scenario:
[   74.356650]        CPU0                    CPU1
[   74.357328]        ----                    ----
[   74.358026]   lock(&q->q_usage_counter(queue)#14);
[   74.358749]                                lock(&disk->open_mutex);
[   74.359561]                                lock(&q->q_usage_counter(queue)#14);
[   74.360488]   lock(kn->active#119);
[   74.361113]
                *** DEADLOCK ***
Quoted 8 lines
[   74.362574] 6 locks held by check/5077:
[   74.363193]  #0: ffff888114640420 (sb_writers#4){.+.+}-{0:0}, at: ksys_write+0xff/0x200
[   74.364274]  #1: ffff88829abb6088 (&of->mutex#2){+.+.}-{4:4}, at: kernfs_fop_write_iter+0x25b/0x5a0
[   74.365937]  #2: ffff8881176ca0e0 (&shost->scan_mutex){+.+.}-{4:4}, at: sdev_store_delete+0x7f/0x120
[   74.367643]  #3: ffff88828521c380 (&dev->mutex){....}-{4:4}, at: device_release_driver_internal+0x90/0x520
[   74.369464]  #4: ffff8881176ca380 (&set->update_nr_hwq_sema){.+.+}-{4:4}, at: del_gendisk+0xdd/0x180
[   74.370961]  #5: ffff88828a661e20 (&q->q_usage_counter(queue)#14){++++}-{0:0}, at: del_gendisk+0xe5/0x180
[   74.372050]
Quoted 4 lines

This has baffled me as I don't understand how could read lock in elv_iosched_store (ruuning in context #1) depends on (same) read lock in add_disk_fwnode (running under another context #2) as both locks are represented by the same rw semaphore. As we see

That is why the read lock in elv_iosched_store() is annotated as nested(new lockdep map) for avoiding the false positive warning, because the two can't be grabbed at the same time.

Quoted 3 lines

above both elv_iosched_store and add_disk_fwnode bot run under different contexts and so ideally they should be able to run concurrently acquiring the same read lock.

In theory, it is yes, but reality is that scheduler store attribute isn't created until disk/queue kobject is added, then switching elevator can't happen until the kobject/attribute is exposed to userspace, that is why the nested annotation is correct.

Thanks,
Ming

author
#25 lore raw

PATCH v3 9/20 block: simplify elevator reattachment for updating nr_hw_queues

In reply to Ming Lei

In blk_mq_update_nr_hw_queues(), nr_hw_queues changes and elevator data depends on it, so elevator has to be reattached.

Now elevator switch isn't allowed during blk_mq_update_nr_hw_queues(), so we can simply call elevator_change() to reattach elevator sched tags after nr_hw_queues is updated.

Add elv_update_nr_hw_queues() for blk_mq_update_nr_hw_queues() to reattach elevator.

---
 block/blk-mq.c   | 90 +-----------------------------------------------
 block/blk.h      |  3 +-
 block/elevator.c | 32 +++++++++++++----
 3 files changed, 28 insertions(+), 97 deletions(-)
block/blk-mq.c +1 -89
index 1ed2d183f912..3afcddd21586 100644
@@ -4943,88 +4943,10 @@ int blk_mq_update_nr_requests(struct request_queue *q, unsigned int nr)
 	return ret;
 }
 
-/*
- * request_queue and elevator_type pair.
- * It is just used by __blk_mq_update_nr_hw_queues to cache
- * the elevator_type associated with a request_queue.
- */
-struct blk_mq_qe_pair {
-	struct list_head node;
-	struct request_queue *q;
-	struct elevator_type *type;
-};
-
-/*
- * Cache the elevator_type in qe pair list and switch the
- * io scheduler to 'none'
- */
-static bool blk_mq_elv_switch_none(struct list_head *head,
-		struct request_queue *q)
-{
-	struct blk_mq_qe_pair *qe;
-
-	qe = kmalloc(sizeof(*qe), GFP_NOIO | __GFP_NOWARN | __GFP_NORETRY);
-	if (!qe)
-		return false;
-
-	/* Accessing q->elevator needs protection from ->elevator_lock. */
-	mutex_lock(&q->elevator_lock);
-
-	if (!q->elevator) {
-		kfree(qe);
-		goto unlock;
-	}
-
-	INIT_LIST_HEAD(&qe->node);
-	qe->q = q;
-	qe->type = q->elevator->type;
-	/* keep a reference to the elevator module as we'll switch back */
-	__elevator_get(qe->type);
-	list_add(&qe->node, head);
-	elevator_disable(q);
-unlock:
-	mutex_unlock(&q->elevator_lock);
-
-	return true;
-}
-
-static struct blk_mq_qe_pair *blk_lookup_qe_pair(struct list_head *head,
-						struct request_queue *q)
-{
-	struct blk_mq_qe_pair *qe;
-
-	list_for_each_entry(qe, head, node)
-		if (qe->q == q)
-			return qe;
-
-	return NULL;
-}
-
-static void blk_mq_elv_switch_back(struct list_head *head,
-				  struct request_queue *q)
-{
-	struct blk_mq_qe_pair *qe;
-	struct elevator_type *t;
-
-	qe = blk_lookup_qe_pair(head, q);
-	if (!qe)
-		return;
-	t = qe->type;
-	list_del(&qe->node);
-	kfree(qe);
-
-	mutex_lock(&q->elevator_lock);
-	elevator_switch(q, t);
-	/* drop the reference acquired in blk_mq_elv_switch_none */
-	elevator_put(t);
-	mutex_unlock(&q->elevator_lock);
-}
-
 static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
 							int nr_hw_queues)
 {
 	struct request_queue *q;
-	LIST_HEAD(head);
 	int prev_nr_hw_queues = set->nr_hw_queues;
 	unsigned int memflags;
 	int i;
@@ -5042,15 +4964,6 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
 	list_for_each_entry(q, &set->tag_list, tag_set_list)
 		blk_mq_freeze_queue_nomemsave(q);
 
-	/*
-	 * Switch IO scheduler to 'none', cleaning up the data associated
-	 * with the previous scheduler. We will switch back once we are done
-	 * updating the new sw to hw queue mappings.
-	 */
-	list_for_each_entry(q, &set->tag_list, tag_set_list)
-		if (!blk_mq_elv_switch_none(&head, q))
-			goto switch_back;
-
 	list_for_each_entry(q, &set->tag_list, tag_set_list) {
 		blk_mq_debugfs_unregister_hctxs(q);
 		blk_mq_sysfs_unregister_hctxs(q);
@@ -5084,9 +4997,8 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
 		blk_mq_debugfs_register_hctxs(q);
 	}
 
-switch_back:
 	list_for_each_entry(q, &set->tag_list, tag_set_list)
-		blk_mq_elv_switch_back(&head, q);
+		elv_update_nr_hw_queues(q);
 
 	list_for_each_entry(q, &set->tag_list, tag_set_list)
 		blk_mq_unfreeze_queue_nomemrestore(q);
block/blk.h +1 -2
index 006e3be433d2..2969f4427996 100644
@@ -319,8 +319,7 @@ bool blk_bio_list_merge(struct request_queue *q, struct list_head *list,
 
 bool blk_insert_flush(struct request *rq);
 
-int elevator_switch(struct request_queue *q, struct elevator_type *new_e);
-void elevator_disable(struct request_queue *q);
+void elv_update_nr_hw_queues(struct request_queue *q);
 void elevator_exit(struct request_queue *q);
 int elv_register_queue(struct request_queue *q, bool uevent);
 void elv_unregister_queue(struct request_queue *q);
block/elevator.c +26 -6
index 56da6ab7691a..5705f7056516 100644
@@ -621,7 +621,7 @@ void elevator_init_mq(struct request_queue *q)
  * If switching fails, we are most likely running out of memory and not able
  * to restore the old io scheduler, so leaving the io scheduler being none.
  */
-int elevator_switch(struct request_queue *q, struct elevator_type *new_e)
+static int elevator_switch(struct request_queue *q, struct elevator_type *new_e)
 {
 	int ret;
 
@@ -657,7 +657,7 @@ int elevator_switch(struct request_queue *q, struct elevator_type *new_e)
 	return ret;
 }
 
-void elevator_disable(struct request_queue *q)
+static void elevator_disable(struct request_queue *q)
 {
 	WARN_ON_ONCE(q->mq_freeze_depth == 0);
 	lockdep_assert_held(&q->elevator_lock);
@@ -677,7 +677,8 @@ void elevator_disable(struct request_queue *q)
 /*
  * Switch this queue to the given IO scheduler.
  */
-static int elevator_change(struct request_queue *q, const char *elevator_name)
+static int __elevator_change(struct request_queue *q,
+			     const char *elevator_name)
 {
 	struct elevator_type *e;
 	int ret;
@@ -692,9 +693,6 @@ static int elevator_change(struct request_queue *q, const char *elevator_name)
 		return 0;
 	}
 
-	if (q->elevator && elevator_match(q->elevator->type, elevator_name))
-		return 0;
-
 	e = elevator_find_get(elevator_name);
 	if (!e)
 		return -EINVAL;
@@ -703,6 +701,28 @@ static int elevator_change(struct request_queue *q, const char *elevator_name)
 	return ret;
 }
 
+static int elevator_change(struct request_queue *q, const char *elevator_name)
+{
+	if (!q->elevator || !elevator_match(q->elevator->type, elevator_name))
+		return __elevator_change(q, elevator_name);
+	return 0;
+}
+
+/*
+ * The I/O scheduler depends on the number of hardware queues, this forces a
+ * reattachment when nr_hw_queues changes.
+ */
+void elv_update_nr_hw_queues(struct request_queue *q)
+{
+	const char *name = "none";
+
+	mutex_lock(&q->elevator_lock);
+	if (q->elevator && !blk_queue_dying(q))
+		name = q->elevator->type->elevator_name;
+	__elevator_change(q, name);
+	mutex_unlock(&q->elevator_lock);
+}
+
 static void elv_iosched_load_module(char *elevator_name)
 {
 	struct elevator_type *found;
Signature · 1 line
2.47.0
Reviewed-by on v3 9/20
#26 lore raw

Re: [PATCH V3 09/20] block: simplify elevator reattachment for updating nr_hw_queues

Quoted 10 lines · Ming Lei

In blk_mq_update_nr_hw_queues(), nr_hw_queues changes and elevator data depends on it, so elevator has to be reattached.

Now elevator switch isn't allowed during blk_mq_update_nr_hw_queues(), so we can simply call elevator_change() to reattach elevator sched tags after nr_hw_queues is updated.

Add elv_update_nr_hw_queues() for blk_mq_update_nr_hw_queues() to reattach elevator.

Quoted 9 lines
---
  block/blk-mq.c   | 90 +-----------------------------------------------
  block/blk.h      |  3 +-
  block/elevator.c | 32 +++++++++++++----
  3 files changed, 28 insertions(+), 97 deletions(-)

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
#27 lore raw

Re: [PATCH V3 09/20] block: simplify elevator reattachment for updating nr_hw_queues

In reply to Ming Lei

Review

Quoted 2 lines · Ming Lei
(quoted diff) +2 -0
+static int __elevator_change(struct request_queue *q,
+			     const char *elevator_name)

There's still not good reason for this helper.

I'd suggest you add the two first attached patches before this one (it'll need a bi of reabsing as all of them are after your entire sweries right now) and then fold the third one into this, which will give us less code and a cleaner interface.

From 44da16a97ef758d1d9fe7c54c8360388f585d0c5 Mon Sep 17 00:00:00 2001
From: Christoph Hellwig <[email protected]>
Date: Fri, 25 Apr 2025 07:01:39 -0700
Subject: block: look up the elevator type in elevator_switch

That makes the function nicely self-contained and can be used to avoid code duplication.

---
 block/elevator.c | 18 ++++++++----------
 1 file changed, 8 insertions(+), 10 deletions(-)
block/elevator.c +8 -10
index a637426da56d..773b8931d874 100644
@@ -591,14 +591,18 @@ static bool use_default_elevator(struct request_queue *q)
  * If switching fails, we are most likely running out of memory and not able
  * to restore the old io scheduler, so leaving the io scheduler being none.
  */
-static int elevator_switch(struct request_queue *q, struct elevator_type *new_e,
-			   struct elv_change_ctx *ctx)
+static int elevator_switch(struct request_queue *q, struct elv_change_ctx *ctx)
 {
+	struct elevator_type *new_e;
 	int ret;
 
 	WARN_ON_ONCE(q->mq_freeze_depth == 0);
 	lockdep_assert_held(&q->elevator_lock);
 
+	new_e = elevator_find_get(ctx->name);
+	if (!new_e)
+		return -EINVAL;
+
 	blk_mq_quiesce_queue(q);
 
 	if (q->elevator) {
@@ -621,6 +625,7 @@ static int elevator_switch(struct request_queue *q, struct elevator_type *new_e,
 			new_e->elevator_name);
 	}
 
+	elevator_put(new_e);
 	return ret;
 }
 
@@ -686,8 +691,6 @@ static int __elevator_change(struct request_queue *q,
 			     struct elv_change_ctx *ctx)
 {
 	const char *elevator_name = ctx->name;
-	struct elevator_type *e;
-	int ret;
 
 	lockdep_assert_held(&q->tag_set->update_nr_hwq_sema);
 
@@ -701,12 +704,7 @@ static int __elevator_change(struct request_queue *q,
 		return 0;
 	}
 
-	e = elevator_find_get(elevator_name);
-	if (!e)
-		return -EINVAL;
-	ret = elevator_switch(q, e, ctx);
-	elevator_put(e);
-	return ret;
+	return elevator_switch(q, ctx);
 }
 
 static int elevator_change(struct request_queue *q,
Signature · 1 line
2.47.2

From 1bfce1a308b9e46734ed56196b4e9fe31b5a0036 Mon Sep 17 00:00:00 2001
From: Christoph Hellwig <[email protected]>
Date: Fri, 25 Apr 2025 07:10:35 -0700
Subject: block: fold elevator_disable into elevator_switch

This removes duplicate code, and keeps the callers tidy.

---
 block/elevator.c | 59 ++++++++++++++++--------------------------------
 1 file changed, 20 insertions(+), 39 deletions(-)
block/elevator.c +20 -39

95 lines not on this page · raw message

Signature · 1 line
2.47.2

From dcda3f508e5f938cb27d4b743226ca4d8af75e28 Mon Sep 17 00:00:00 2001
From: Christoph Hellwig <[email protected]>
Date: Fri, 25 Apr 2025 07:18:42 -0700
Subject: block: remove __elevator_change

Not much of a point in sharing code between callers with very different expectations.

---
 block/elevator.c | 45 ++++++++++++++++++++-------------------------
 1 file changed, 20 insertions(+), 25 deletions(-)
block/elevator.c +20 -25

76 lines not on this page · raw message

Signature · 1 line
2.47.2
author
#28 lore raw

Re: [PATCH V3 09/20] block: simplify elevator reattachment for updating nr_hw_queues

Quoted 5 lines · Christoph Hellwig

Review

Quoted 2 lines · Ming Lei
(quoted diff) +2 -0
+static int __elevator_change(struct request_queue *q,
+			     const char *elevator_name)

There's still not good reason for this helper.

With the helper, we can avoid the 'force' reattachment flag.

Quoted 5 lines

I'd suggest you add the two first attached patches before this one (it'll need a bi of reabsing as all of them are after your entire sweries right now) and then fold the third one into this, which will give us less code and a cleaner interface.

I'd suggest to add your patches after this one, which changes elevator_switch() into local symbol.

Thanks,
Ming

author
#29 lore raw

PATCH v3 10/20 block: move blk_unregister_queue() & device_del() after freeze wait

In reply to Ming Lei

Move blk_unregister_queue() & device_del() after freeze wait, and prepare for unifying elevator switch.

This way is just fine, since bdev has been unhashed at the beginning of del_gendisk(), both blk_unregister_queue() & device_del() are dealing with kobject & debugfs thing only.

---
 block/genhd.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
block/genhd.c +3 -3
index 7f3ae3d23b26..53c5b1e9e1d2 100644
@@ -733,8 +733,6 @@ static void __del_gendisk(struct gendisk *disk)
 		bdi_unregister(disk->bdi);
 	}
 
-	blk_unregister_queue(disk);
-
 	kobject_put(disk->part0->bd_holder_dir);
 	kobject_put(disk->slave_dir);
 	disk->slave_dir = NULL;
@@ -743,10 +741,12 @@ static void __del_gendisk(struct gendisk *disk)
 	disk->part0->bd_stamp = 0;
 	sysfs_remove_link(block_depr, dev_name(disk_to_dev(disk)));
 	pm_runtime_set_memalloc_noio(disk_to_dev(disk), false);
-	device_del(disk_to_dev(disk));
 
 	blk_mq_freeze_queue_wait(q);
 
+	blk_unregister_queue(disk);
+	device_del(disk_to_dev(disk));
+
 	blk_throtl_cancel_bios(disk);
 
 	blk_sync_queue(q);
Signature · 1 line
2.47.0
Reviewed-by on v3 10/20
#30 lore raw

Re: [PATCH V3 10/20] block: move blk_unregister_queue() & device_del() after freeze wait

Quoted 12 lines · Ming Lei

Move blk_unregister_queue() & device_del() after freeze wait, and prepare for unifying elevator switch.

This way is just fine, since bdev has been unhashed at the beginning of del_gendisk(), both blk_unregister_queue() & device_del() are dealing with kobject & debugfs thing only.

---
  block/genhd.c | 6 +++---
  1 file changed, 3 insertions(+), 3 deletions(-)

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
Reviewed-by on v3 10/20
#31 lore raw

Re: [PATCH V3 10/20] block: move blk_unregister_queue() & device_del() after freeze wait

In reply to Ming Lei
Quoted 8 lines · Ming Lei

Move blk_unregister_queue() & device_del() after freeze wait, and prepare for unifying elevator switch.

This way is just fine, since bdev has been unhashed at the beginning of del_gendisk(), both blk_unregister_queue() & device_del() are dealing with kobject & debugfs thing only.

Looks good to me:

#32 lore raw

Re: [PATCH V3 10/20] block: move blk_unregister_queue() & device_del() after freeze wait

In reply to Ming Lei
Quoted 6 lines · Ming Lei

Move blk_unregister_queue() & device_del() after freeze wait, and prepare for unifying elevator switch.

This way is just fine, since bdev has been unhashed at the beginning of del_gendisk(), both blk_unregister_queue() & device_del() are dealing with kobject & debugfs thing only.

Can you please add a reference to the previous attempt here and that it was backed out again and why we think it's fine this time around?

author
#33 lore raw

Re: [PATCH V3 10/20] block: move blk_unregister_queue() & device_del() after freeze wait

Quoted 10 lines · Christoph Hellwig
Quoted 6 lines · Ming Lei

Move blk_unregister_queue() & device_del() after freeze wait, and prepare for unifying elevator switch.

This way is just fine, since bdev has been unhashed at the beginning of del_gendisk(), both blk_unregister_queue() & device_del() are dealing with kobject & debugfs thing only.

Can you please add a reference to the previous attempt here and that it was backed out again and why we think it's fine this time around?

Turns out this patch isn't necessary, we can avoid the small race easily between elevator switch and del_gendisk().

Will drop this patch in V4.

Thanks,
Ming

author
#34 lore raw

PATCH v3 11/20 block: move queue freezing & elevator_lock into elevator_change()

In reply to Ming Lei

Move queue freezing & elevator_lock into elevator_change(), and prepare for using elevator_change() for setting up & tearing down default elevator too.

Also add lockdep_assert_held() in __elevator_change() because either read or write lock is required for changing elevator.

---
 block/elevator.c | 18 +++++++++++-------
 1 file changed, 11 insertions(+), 7 deletions(-)
block/elevator.c +11 -7
index 5705f7056516..e70cd4b828a5 100644
@@ -683,6 +683,8 @@ static int __elevator_change(struct request_queue *q,
 	struct elevator_type *e;
 	int ret;
 
+	lockdep_assert_held(&q->tag_set->update_nr_hwq_sema);
+
 	/* Make sure queue is not in the middle of being removed */
 	if (!blk_queue_registered(q))
 		return -ENOENT;
@@ -703,9 +705,16 @@ static int __elevator_change(struct request_queue *q,
 
 static int elevator_change(struct request_queue *q, const char *elevator_name)
 {
+	unsigned int memflags;
+	int ret = 0;
+
+	memflags = blk_mq_freeze_queue(q);
+	mutex_lock(&q->elevator_lock);
 	if (!q->elevator || !elevator_match(q->elevator->type, elevator_name))
-		return __elevator_change(q, elevator_name);
-	return 0;
+		ret = __elevator_change(q, elevator_name);
+	mutex_unlock(&q->elevator_lock);
+	blk_mq_unfreeze_queue(q, memflags);
+	return ret;
 }
 
 /*
@@ -741,7 +750,6 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 	char elevator_name[ELV_NAME_MAX];
 	char *name;
 	int ret;
-	unsigned int memflags;
 	struct request_queue *q = disk->queue;
 	struct blk_mq_tag_set *set = q->tag_set;
 
@@ -756,13 +764,9 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 	elv_iosched_load_module(name);
 
 	down_read_nested(&set->update_nr_hwq_sema, 1);
-	memflags = blk_mq_freeze_queue(q);
-	mutex_lock(&q->elevator_lock);
 	ret = elevator_change(q, name);
 	if (!ret)
 		ret = count;
-	mutex_unlock(&q->elevator_lock);
-	blk_mq_unfreeze_queue(q, memflags);
 	up_read(&set->update_nr_hwq_sema);
 	return ret;
 }
Signature · 1 line
2.47.0
Reviewed-by on v3 11/20
#35 lore raw

Re: [PATCH V3 11/20] block: move queue freezing & elevator_lock into elevator_change()

Quoted 4 lines · Ming Lei

Move queue freezing & elevator_lock into elevator_change(), and prepare for using elevator_change() for setting up & tearing down default elevator too.

Quoted 8 lines

Also add lockdep_assert_held() in __elevator_change() because either read or write lock is required for changing elevator.

---
  block/elevator.c | 18 +++++++++++-------
  1 file changed, 11 insertions(+), 7 deletions(-)

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
Reviewed-by on v3 11/20
#36 lore raw

Re: [PATCH V3 11/20] block: move queue freezing & elevator_lock into elevator_change()

In reply to Ming Lei
Quoted 8 lines · Ming Lei

Move queue freezing & elevator_lock into elevator_change(), and prepare for using elevator_change() for setting up & tearing down default elevator too.

Also add lockdep_assert_held() in __elevator_change() because either read or write lock is required for changing elevator.

Looks good to me:

author
#38 lore raw

PATCH v3 12/20 block: add `struct elv_change_ctx` for unifying elevator change

In reply to Ming Lei

Add struct elv_change_ctx and prepare for unifying elevator change by [__]elevator_change(). With this way, any input & output parameter can be provided & observed in top helper.

This way helps to move kobject add/delete & debugfs register/unregister out of ->elevator_lock & freezing queue.

---
 block/elevator.c | 44 +++++++++++++++++++++++++++++---------------
 1 file changed, 29 insertions(+), 15 deletions(-)
block/elevator.c +29 -15
index e70cd4b828a5..4be318b0b4ef 100644
@@ -45,6 +45,12 @@
 #include "blk-wbt.h"
 #include "blk-cgroup.h"
 
+/* Holding context data for changing elevator */
+struct elv_change_ctx {
+	const char *name;
+	bool uevent;
+};
+
 static DEFINE_SPINLOCK(elv_list_lock);
 static LIST_HEAD(elv_list);
 
@@ -621,7 +627,8 @@ void elevator_init_mq(struct request_queue *q)
  * If switching fails, we are most likely running out of memory and not able
  * to restore the old io scheduler, so leaving the io scheduler being none.
  */
-static int elevator_switch(struct request_queue *q, struct elevator_type *new_e)
+static int elevator_switch(struct request_queue *q, struct elevator_type *new_e,
+			   struct elv_change_ctx *ctx)
 {
 	int ret;
 
@@ -639,7 +646,7 @@ static int elevator_switch(struct request_queue *q, struct elevator_type *new_e)
 	if (ret)
 		goto out_unfreeze;
 
-	ret = elv_register_queue(q, true);
+	ret = elv_register_queue(q, ctx->uevent);
 	if (ret) {
 		elevator_exit(q);
 		goto out_unfreeze;
@@ -678,8 +685,9 @@ static void elevator_disable(struct request_queue *q)
  * Switch this queue to the given IO scheduler.
  */
 static int __elevator_change(struct request_queue *q,
-			     const char *elevator_name)
+			     struct elv_change_ctx *ctx)
 {
+	const char *elevator_name = ctx->name;
 	struct elevator_type *e;
 	int ret;
 
@@ -698,20 +706,21 @@ static int __elevator_change(struct request_queue *q,
 	e = elevator_find_get(elevator_name);
 	if (!e)
 		return -EINVAL;
-	ret = elevator_switch(q, e);
+	ret = elevator_switch(q, e, ctx);
 	elevator_put(e);
 	return ret;
 }
 
-static int elevator_change(struct request_queue *q, const char *elevator_name)
+static int elevator_change(struct request_queue *q,
+			   struct elv_change_ctx *ctx)
 {
 	unsigned int memflags;
 	int ret = 0;
 
 	memflags = blk_mq_freeze_queue(q);
 	mutex_lock(&q->elevator_lock);
-	if (!q->elevator || !elevator_match(q->elevator->type, elevator_name))
-		ret = __elevator_change(q, elevator_name);
+	if (!q->elevator || !elevator_match(q->elevator->type, ctx->name))
+		ret = __elevator_change(q, ctx);
 	mutex_unlock(&q->elevator_lock);
 	blk_mq_unfreeze_queue(q, memflags);
 	return ret;
@@ -723,16 +732,19 @@ static int elevator_change(struct request_queue *q, const char *elevator_name)
  */
 void elv_update_nr_hw_queues(struct request_queue *q)
 {
-	const char *name = "none";
+	struct elv_change_ctx ctx = {
+		.name	= "none",
+		.uevent	= true,
+	};
 
 	mutex_lock(&q->elevator_lock);
 	if (q->elevator && !blk_queue_dying(q))
-		name = q->elevator->type->elevator_name;
-	__elevator_change(q, name);
+		ctx.name = q->elevator->type->elevator_name;
+	__elevator_change(q, &ctx);
 	mutex_unlock(&q->elevator_lock);
 }
 
-static void elv_iosched_load_module(char *elevator_name)
+static void elv_iosched_load_module(const char *elevator_name)
 {
 	struct elevator_type *found;
 
@@ -748,7 +760,9 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 			  size_t count)
 {
 	char elevator_name[ELV_NAME_MAX];
-	char *name;
+	struct elv_change_ctx ctx = {
+		.uevent = true,
+	};
 	int ret;
 	struct request_queue *q = disk->queue;
 	struct blk_mq_tag_set *set = q->tag_set;
@@ -759,12 +773,12 @@ ssize_t elv_iosched_store(struct gendisk *disk, const char *buf,
 	 * queue is the one for the device storing the module file.
 	 */
 	strscpy(elevator_name, buf, sizeof(elevator_name));
-	name = strstrip(elevator_name);
+	ctx.name = strstrip(elevator_name);
 
-	elv_iosched_load_module(name);
+	elv_iosched_load_module(ctx.name);
 
 	down_read_nested(&set->update_nr_hwq_sema, 1);
-	ret = elevator_change(q, name);
+	ret = elevator_change(q, &ctx);
 	if (!ret)
 		ret = count;
 	up_read(&set->update_nr_hwq_sema);
Signature · 1 line
2.47.0
Reviewed-by on v3 12/20
#39 lore raw

Re: [PATCH V3 12/20] block: add `struct elv_change_ctx` for unifying elevator change

Quoted 4 lines · Ming Lei

Add struct elv_change_ctx and prepare for unifying elevator change by [__]elevator_change(). With this way, any input & output parameter can be provided & observed in top helper.

Quoted 9 lines

This way helps to move kobject add/delete & debugfs register/unregister out of ->elevator_lock & freezing queue.

---
  block/elevator.c | 44 +++++++++++++++++++++++++++++---------------
  1 file changed, 29 insertions(+), 15 deletions(-)

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
#40 lore raw

Re: [PATCH V3 12/20] block: add `struct elv_change_ctx` for unifying elevator change

In reply to Ming Lei

block/elevator.c line 51

Quoted 3 lines · Ming Lei
block/elevator.c +3 -0
+struct elv_change_ctx {
+	const char *name;
+	bool uevent;

There's only one caller that wants to supress the uevents. So maybe invert the polarity so that it only has to be set in one place, which also documents how setting the initial scheduler is special a bit better.

block/elevator.c static int elevator_switch(struct request_queue *q, struct elevator_type *new_e) · line 649

Quoted 2 lines
block/elevator.c +1 -1
-	ret = elv_register_queue(q, true);
+	ret = elv_register_queue(q, ctx->uevent);

.. and pass the ctx on to elv_register_queue instead of converting paramter types. Although that might be woeth doing later when another argument derived from ctx gets passed as well.

author
#41 lore raw

Re: [PATCH V3 12/20] block: add `struct elv_change_ctx` for unifying elevator change

Quoted 9 lines · Christoph Hellwig

block/elevator.c line 51

Quoted 3 lines · Ming Lei
block/elevator.c +3 -0
+struct elv_change_ctx {
+	const char *name;
+	bool uevent;

There's only one caller that wants to supress the uevents. So maybe invert the polarity so that it only has to be set in one place, which also documents how setting the initial scheduler is special a bit better.

OK.

Quoted 7 lines

block/elevator.c static int elevator_switch(struct request_queue *q, struct elevator_type *new_e) · line 649

Quoted 3 lines
block/elevator.c +1 -1
-	ret = elv_register_queue(q, true);
+	ret = elv_register_queue(q, ctx->uevent);

.. and pass the ctx on to elv_register_queue instead of converting paramter types. Although that might be woeth doing later when another argument derived from ctx gets passed as well.

ctx isn't very useful for elv_register_queue(), which can't figure out the exact elevator queue to use from ctx since there are two(old, new).

Thanks,
Ming

author
#42 lore raw

PATCH v3 13/20 block: unifying elevator change

In reply to Ming Lei

elevator change is one well-define behavior:

- tear down current elevator if it exists

- setup new elevator

It is supposed to cover any case for changing elevator by single internal API, typically the following cases:

- setup default elevator in add_disk()

- switch to none in del_disk()

- reset elevator in blk_mq_update_nr_hw_queues()

- switch elevator in sysfs store elevator attribute

This patch uses elevator_change() to cover all above cases:

- every elevator switch is serialized with each other: add_disk/del_disk/ store elevator is serialized already, blk_mq_update_nr_hw_queues() uses srcu for syncing with the other three cases

- for both add_disk()/del_disk(), queue freeze works at atomic mode or has been froze, so the freeze in elevator_change() won't add extra delay

- struct elev_change_ctx instance holds any info for changing elevator

---
 block/blk-sysfs.c |  18 +++-----
 block/blk.h       |   5 +-
 block/elevator.c  | 114 +++++++++++++++++++++++++---------------------
 block/genhd.c     |  28 ++----------
 4 files changed, 75 insertions(+), 90 deletions(-)
block/blk-sysfs.c +6 -12
index 1f9b45b0b9ee..9aa05666f32a 100644
@@ -869,14 +869,8 @@ int blk_register_queue(struct gendisk *disk)
 	if (ret)
 		goto out_unregister_ia_ranges;
 
+	elevator_set_default(q);
 	mutex_lock(&q->elevator_lock);
-	if (q->elevator) {
-		ret = elv_register_queue(q, false);
-		if (ret) {
-			mutex_unlock(&q->elevator_lock);
-			goto out_crypto_sysfs_unregister;
-		}
-	}
 	wbt_enable_default(disk);
 	mutex_unlock(&q->elevator_lock);
 
@@ -902,8 +896,6 @@ int blk_register_queue(struct gendisk *disk)
 
 	return ret;
 
-out_crypto_sysfs_unregister:
-	blk_crypto_sysfs_unregister(disk);
 out_unregister_ia_ranges:
 	disk_unregister_independent_access_ranges(disk);
 out_debugfs_remove:
@@ -951,9 +943,11 @@ void blk_unregister_queue(struct gendisk *disk)
 		blk_mq_sysfs_unregister(disk);
 	blk_crypto_sysfs_unregister(disk);
 
-	mutex_lock(&q->elevator_lock);
-	elv_unregister_queue(q);
-	mutex_unlock(&q->elevator_lock);
+	if (q->elevator) {
+		blk_mq_quiesce_queue(q);
+		elevator_set_none(q);
+		blk_mq_unquiesce_queue(q);
+	}
 
 	mutex_lock(&q->sysfs_lock);
 	disk_unregister_independent_access_ranges(disk);
block/blk.h +2 -3
index 2969f4427996..6ba674d1ceba 100644
@@ -320,9 +320,8 @@ bool blk_bio_list_merge(struct request_queue *q, struct list_head *list,
 bool blk_insert_flush(struct request *rq);
 
 void elv_update_nr_hw_queues(struct request_queue *q);
-void elevator_exit(struct request_queue *q);
-int elv_register_queue(struct request_queue *q, bool uevent);
-void elv_unregister_queue(struct request_queue *q);
+void elevator_set_default(struct request_queue *q);
+void elevator_set_none(struct request_queue *q);
 
 ssize_t part_size_show(struct device *dev, struct device_attribute *attr,
 		char *buf);
block/elevator.c +62 -52

168 lines not on this page · raw message

block/genhd.c +5 -23
index 53c5b1e9e1d2..a50063c4c40f 100644
@@ -416,12 +416,6 @@ static int __add_disk_fwnode(struct device *parent, struct gendisk *disk,
 		 */
 		if (disk->fops->submit_bio || disk->fops->poll_bio)
 			return -EINVAL;
-
-		/*
-		 * Initialize the I/O scheduler code and pick a default one if
-		 * needed.
-		 */
-		elevator_init_mq(disk->queue);
 	} else {
 		if (!disk->fops->submit_bio)
 			return -EINVAL;
@@ -438,7 +432,7 @@ static int __add_disk_fwnode(struct device *parent, struct gendisk *disk,
 	ret = -EINVAL;
 	if (disk->major) {
 		if (WARN_ON(!disk->minors))
-			goto out_exit_elevator;
+			goto out;
 
 		if (disk->minors > DISK_MAX_PARTS) {
 			pr_err("block: can't allocate more than %d partitions\n",
@@ -448,14 +442,14 @@ static int __add_disk_fwnode(struct device *parent, struct gendisk *disk,
 		if (disk->first_minor > MINORMASK ||
 		    disk->minors > MINORMASK + 1 ||
 		    disk->first_minor + disk->minors > MINORMASK + 1)
-			goto out_exit_elevator;
+			goto out;
 	} else {
 		if (WARN_ON(disk->minors))
-			goto out_exit_elevator;
+			goto out;
 
 		ret = blk_alloc_ext_minor();
 		if (ret < 0)
-			goto out_exit_elevator;
+			goto out;
 		disk->major = BLOCK_EXT_MAJOR;
 		disk->first_minor = ret;
 	}
@@ -564,12 +558,7 @@ static int __add_disk_fwnode(struct device *parent, struct gendisk *disk,
 out_free_ext_minor:
 	if (disk->major == BLOCK_EXT_MAJOR)
 		blk_free_ext_minor(disk->first_minor);
-out_exit_elevator:
-	if (disk->queue->elevator) {
-		mutex_lock(&disk->queue->elevator_lock);
-		elevator_exit(disk->queue);
-		mutex_unlock(&disk->queue->elevator_lock);
-	}
+out:
 	return ret;
 }
 
@@ -755,14 +744,7 @@ static void __del_gendisk(struct gendisk *disk)
 	if (queue_is_mq(q))
 		blk_mq_cancel_work_sync(q);
 
-	blk_mq_quiesce_queue(q);
-	if (q->elevator) {
-		mutex_lock(&q->elevator_lock);
-		elevator_exit(q);
-		mutex_unlock(&q->elevator_lock);
-	}
 	rq_qos_exit(q);
-	blk_mq_unquiesce_queue(q);
 
 	/*
 	 * If the disk does not own the queue, allow using passthrough requests
Signature · 1 line
2.47.0
Reviewed-by on v3 13/20
#43 lore raw

Re: [PATCH V3 13/20] block: unifying elevator change

Quoted 29 lines · Ming Lei

elevator change is one well-define behavior:

- tear down current elevator if it exists

- setup new elevator

It is supposed to cover any case for changing elevator by single internal API, typically the following cases:

- setup default elevator in add_disk()

- switch to none in del_disk()

- reset elevator in blk_mq_update_nr_hw_queues()

- switch elevator in sysfs store elevator attribute

This patch uses elevator_change() to cover all above cases:

- every elevator switch is serialized with each other: add_disk/del_disk/ store elevator is serialized already, blk_mq_update_nr_hw_queues() uses srcu for syncing with the other three cases

- for both add_disk()/del_disk(), queue freeze works at atomic mode or has been froze, so the freeze in elevator_change() won't add extra delay

- struct elev_change_ctx instance holds any info for changing elevator

Quoted 8 lines
---
  block/blk-sysfs.c |  18 +++-----
  block/blk.h       |   5 +-
  block/elevator.c  | 114 +++++++++++++++++++++++++---------------------
  block/genhd.c     |  28 ++----------
  4 files changed, 75 insertions(+), 90 deletions(-)

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
#44 lore raw

Re: [PATCH V3 13/20] block: unifying elevator change

In reply to Ming Lei

block/genhd.c static int __add_disk_fwnode(struct device *parent, struct gendisk *disk, · line 572

Quoted 10 lines · Ming Lei
block/genhd.c +0 -6
@@ -564,12 +558,7 @@ static int __add_disk_fwnode(struct device *parent, struct gendisk *disk,
 out_free_ext_minor:
 	if (disk->major == BLOCK_EXT_MAJOR)
 		blk_free_ext_minor(disk->first_minor);
-out_exit_elevator:
-	if (disk->queue->elevator) {
-		mutex_lock(&disk->queue->elevator_lock);
-		elevator_exit(disk->queue);
-		mutex_unlock(&disk->queue->elevator_lock);
-	}

After restructuring __add_disk_fwnode() we may still need elevator_exit() in case when blk_register_queue is successful and so we should have set q->elevator to default. Now if after queue is registered successfully, for instance, bdi_register fails then in that case we still have to exit elevator to free its resources, isn't it?

Thanks,
--Nilay

author
#45 lore raw

Re: [PATCH V3 13/20] block: unifying elevator change

Quoted 19 lines · Nilay Shroff

block/genhd.c static int __add_disk_fwnode(struct device *parent, struct gendisk *disk, · line 572

Quoted 10 lines · Ming Lei
block/genhd.c +0 -6
@@ -564,12 +558,7 @@ static int __add_disk_fwnode(struct device *parent, struct gendisk *disk,
 out_free_ext_minor:
 	if (disk->major == BLOCK_EXT_MAJOR)
 		blk_free_ext_minor(disk->first_minor);
-out_exit_elevator:
-	if (disk->queue->elevator) {
-		mutex_lock(&disk->queue->elevator_lock);
-		elevator_exit(disk->queue);
-		mutex_unlock(&disk->queue->elevator_lock);
-	}

After restructuring __add_disk_fwnode() we may still need elevator_exit() in case when blk_register_queue is successful and so we should have set q->elevator to default. Now if after queue is registered successfully, for instance, bdi_register fails then in that case we still have to exit elevator to free its resources, isn't it?

If blk_register_queue is successful, it is blk_unregister_queue()'s responsibility for covering it, elevator_exit() isn't needed here.

Thanks,
Ming

#46 lore raw

Re: [PATCH V3 13/20] block: unifying elevator change

In reply to Ming Lei
Quoted 1 line · Ming Lei

elevator change is one well-define behavior:

Start the sentence captitalized.

block/elevator.c EXPORT_SYMBOL_GPL(elv_unregister); · line 578

Quoted 10 lines
block/elevator.c +4 -0
+static bool use_default_elevator(struct request_queue *q)
 {
 	if (q->tag_set->flags & BLK_MQ_F_NO_SCHED_BY_DEFAULT)
+		return false;
 
 	if (q->nr_hw_queues != 1 &&
 	    !blk_mq_is_shared_tags(q->tag_set->flags))
+		return false;
 
+	return true;

looking at the only caller I think this is better open coded there.

block/elevator.c static int __elevator_change(struct request_queue *q, · line 654

Quoted 6 lines
block/elevator.c +1 -1
@@ -694,7 +651,7 @@ static int __elevator_change(struct request_queue *q,
 	lockdep_assert_held(&q->tag_set->update_nr_hwq_sema);
 
 	/* Make sure queue is not in the middle of being removed */
-	if (!blk_queue_registered(q))
+	if (!ctx->init && !blk_queue_registered(q))

This init flag is confusingly also set by the teardown, so it is quite misnamed. But given that none of the locks held here protected the queue registered flag, why not just move the check out to elv_iosched_store and remove the entire need for the flag?

block/elevator.c static int elevator_change(struct request_queue *q, · line 708

Quoted 13 lines
block/elevator.c +13 -0
+/*
+ * Use the default elevator settings. If the chosen elevator initialization
+ * fails, fall back to the "none" elevator (no elevator).
+ */
+void elevator_set_default(struct request_queue *q)
+{
+	struct elv_change_ctx ctx = {
+		.init = true,
+	};
+	int err;
+
+	if (!queue_is_mq(q))
+		return;

Please keep this in the caller. Same for the set_none side.

block/elevator.c static int elevator_change(struct request_queue *q, · line 713

Quoted 5 lines
block/elevator.c +5 -0
+
+	ctx.name = use_default_elevator(q) ? "mq-deadline" : "none";
+	if (!q->elevator && !strcmp(ctx.name, "none"))
+		return;
+	err = elevator_change(q, &ctx);

q->elevator can't be set here. And the extra none strcmp is weird because we have the boolean information avaiable.

Just make this:

	struct elv_change_ctx ctx = {
		.name = "mq-deadline",
	};

	...

if (!(q->tag_set->flags & BLK_MQ_F_NO_SCHED_BY_DEFAULT) &&

	    (q->nr_hw_queues == 1 ||
	     blk_mq_is_shared_tags(q->tag_set->flags))) {
		err = elevator_change(q, &ctx);

block/elevator.c static int elevator_change(struct request_queue *q, · line 716

Quoted 3 lines
block/elevator.c +3 -0
+	if (err < 0)
+		pr_warn("\"%s\" set elevator failed %d, "
+			"falling back to \"none\"\n", ctx.name, err);

Please keep the more useful message from the current code.

author
#47 lore raw

PATCH v3 14/20 block: pass elevator_queue to elv_register_queue & unregister_queue

In reply to Ming Lei

Pass elevator_queue reference to elv_register_queue() & elv_unregister_queue().

No functional change, and prepare for moving the two out of elevator lock & freezing queue, when we need to store the old & new elevator queue in struct elv_change_ctx instance, then both two can co-exist for short while, so we have to pass the exact elevator_queue instance to elv_register_queue & unregister_queue.

---
 block/elevator.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)
block/elevator.c +8 -8
index 5ec66fbd4f24..dde8732b7a26 100644
@@ -459,9 +459,10 @@ static const struct kobj_type elv_ktype = {
 	.release	= elevator_release,
 };
 
-static int elv_register_queue(struct request_queue *q, bool uevent)
+static int elv_register_queue(struct request_queue *q,
+			      struct elevator_queue *e,
+			      bool uevent)
 {
-	struct elevator_queue *e = q->elevator;
 	int error;
 
 	lockdep_assert_held(&q->elevator_lock);
@@ -489,10 +490,9 @@ static int elv_register_queue(struct request_queue *q, bool uevent)
 	return error;
 }
 
-static void elv_unregister_queue(struct request_queue *q)
+static void elv_unregister_queue(struct request_queue *q,
+				 struct elevator_queue *e)
 {
-	struct elevator_queue *e = q->elevator;
-
 	lockdep_assert_held(&q->elevator_lock);
 
 	if (e && test_and_clear_bit(ELEVATOR_FLAG_REGISTERED, &e->flags)) {
@@ -595,7 +595,7 @@ static int elevator_switch(struct request_queue *q, struct elevator_type *new_e,
 	blk_mq_quiesce_queue(q);
 
 	if (q->elevator) {
-		elv_unregister_queue(q);
+		elv_unregister_queue(q, q->elevator);
 		elevator_exit(q);
 	}
 
@@ -603,7 +603,7 @@ static int elevator_switch(struct request_queue *q, struct elevator_type *new_e,
 	if (ret)
 		goto out_unfreeze;
 
-	ret = elv_register_queue(q, ctx->uevent);
+	ret = elv_register_queue(q, q->elevator, ctx->uevent);
 	if (ret) {
 		elevator_exit(q);
 		goto out_unfreeze;
@@ -628,7 +628,7 @@ static void elevator_disable(struct request_queue *q)
 
 	blk_mq_quiesce_queue(q);
 
-	elv_unregister_queue(q);
+	elv_unregister_queue(q, q->elevator);
 	elevator_exit(q);
 	blk_queue_flag_clear(QUEUE_FLAG_SQ_SCHED, q);
 	q->elevator = NULL;
Signature · 1 line
2.47.0
Reviewed-by on v3 14/20
#48 lore raw

Re: [PATCH V3 14/20] block: pass elevator_queue to elv_register_queue & unregister_queue

Quoted 8 lines · Ming Lei

Pass elevator_queue reference to elv_register_queue() & elv_unregister_queue().

No functional change, and prepare for moving the two out of elevator lock & freezing queue, when we need to store the old & new elevator queue in struct elv_change_ctx instance, then both two can co-exist for short while, so we have to pass the exact elevator_queue instance to elv_register_queue & unregister_queue.

Quoted 6 lines
---
  block/elevator.c | 16 ++++++++--------
  1 file changed, 8 insertions(+), 8 deletions(-)

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
author
#49 lore raw

PATCH v3 15/20 block: fail to show/store elevator sysfs attribute if elevator is dying

In reply to Ming Lei

Prepare for moving elv_register[unregister]_queue out of elevator_lock & queue freezing, so we may have to call elv_unregister_queue() after elevator ->exit() is called, then there is small window for user to call into ->show()/store(), and user-after-free can be caused.

Fail to show/store elevator sysfs attribute if elevator is dying by adding one new flag of ELEVATOR_FLAG_DYNG, which is protected by elevator ->sysfs_lock.

---
 block/blk-mq-sched.c |  1 +
 block/elevator.c     | 10 ++++++++--
 block/elevator.h     |  1 +
 3 files changed, 10 insertions(+), 2 deletions(-)
block/blk-mq-sched.c +1 -0
index 336a15ffecfa..55a0fd105147 100644
@@ -551,5 +551,6 @@ void blk_mq_exit_sched(struct request_queue *q, struct elevator_queue *e)
 	if (e->type->ops.exit_sched)
 		e->type->ops.exit_sched(e);
 	blk_mq_sched_tags_teardown(q, flags);
+	set_bit(ELEVATOR_FLAG_DYING, &q->elevator->flags);
 	q->elevator = NULL;
 }
block/elevator.c +8 -2
index dde8732b7a26..48f2f202af98 100644
@@ -426,7 +426,10 @@ elv_attr_show(struct kobject *kobj, struct attribute *attr, char *page)
 
 	e = container_of(kobj, struct elevator_queue, kobj);
 	mutex_lock(&e->sysfs_lock);
-	error = e->type ? entry->show(e, page) : -ENOENT;
+	if (test_bit(ELEVATOR_FLAG_DYING, &e->flags))
+		error = -ENODEV;
+	else
+		error = e->type ? entry->show(e, page) : -ENOENT;
 	mutex_unlock(&e->sysfs_lock);
 	return error;
 }
@@ -444,7 +447,10 @@ elv_attr_store(struct kobject *kobj, struct attribute *attr,
 
 	e = container_of(kobj, struct elevator_queue, kobj);
 	mutex_lock(&e->sysfs_lock);
-	error = e->type ? entry->store(e, page, length) : -ENOENT;
+	if (test_bit(ELEVATOR_FLAG_DYING, &e->flags))
+		error = -ENODEV;
+	else
+		error = e->type ? entry->store(e, page, length) : -ENOENT;
 	mutex_unlock(&e->sysfs_lock);
 	return error;
 }
block/elevator.h +1 -0
index 9198676644a9..76a90a1b7ed6 100644
@@ -121,6 +121,7 @@ struct elevator_queue
 };
 
 #define ELEVATOR_FLAG_REGISTERED	0
+#define ELEVATOR_FLAG_DYING		1
 
 /*
  * block elevator interface
Signature · 1 line
2.47.0
Reviewed-by on v3 15/20
#50 lore raw

Re: [PATCH V3 15/20] block: fail to show/store elevator sysfs attribute if elevator is dying

Quoted 9 lines · Ming Lei

Prepare for moving elv_register[unregister]_queue out of elevator_lock & queue freezing, so we may have to call elv_unregister_queue() after elevator ->exit() is called, then there is small window for user to call into ->show()/store(), and user-after-free can be caused.

Fail to show/store elevator sysfs attribute if elevator is dying by adding one new flag of ELEVATOR_FLAG_DYNG, which is protected by elevator ->sysfs_lock.

Quoted 8 lines
---
  block/blk-mq-sched.c |  1 +
  block/elevator.c     | 10 ++++++++--
  block/elevator.h     |  1 +
  3 files changed, 10 insertions(+), 2 deletions(-)

Not particularly happy about this, but I fear it's the best we can do for now.

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
#51 lore raw

Re: [PATCH V3 15/20] block: fail to show/store elevator sysfs attribute if elevator is dying

In reply to Ming Lei

block/elevator.c elv_attr_show(struct kobject *kobj, struct attribute *attr, char *page) · line 432

Quoted 4 lines · Ming Lei
block/elevator.c +4 -0
+	if (test_bit(ELEVATOR_FLAG_DYING, &e->flags))
+		error = -ENODEV;
+	else
+		error = e->type ? entry->show(e, page) : -ENOENT;

Weird style mix, I'd expand the check for ->type to a proper else if here as well. But how can e->type actually be NULL here to start with?

author
#52 lore raw

Re: [PATCH V3 15/20] block: fail to show/store elevator sysfs attribute if elevator is dying

Quoted 9 lines · Christoph Hellwig

block/elevator.c elv_attr_show(struct kobject *kobj, struct attribute *attr, char *page) · line 432

Quoted 4 lines · Ming Lei
block/elevator.c +4 -0
+	if (test_bit(ELEVATOR_FLAG_DYING, &e->flags))
+		error = -ENODEV;
+	else
+		error = e->type ? entry->show(e, page) : -ENOENT;

Weird style mix, I'd expand the check for ->type to a proper else if here as well. But how can e->type actually be NULL here to start with?

You are right, e->type can't be NULL, which is assigned in its allocation, never get cleared.

Thanks,
Ming

author
#53 lore raw

PATCH v3 16/20 block: move elv_register[unregister]_queue out of elevator_lock

In reply to Ming Lei

Move elv_register[unregister]_queue out of ->elevator_lock & queue freezing, so we can kill many lockdep warnings.

elv_register[unregister]_queue() is serialized, and just dealing with sysfs/ debugfs things, no need to be done with queue frozen.

With this change, elevator's ->exit() is called before calling elv_unregister_queue, then user may call into ->show()/store() of elevator's sysfs attributes, and we have covered this issue by adding ELEVATOR_FLAG_DYNG.

For blk-mq debugfs, hctx->sched_tags is always checked with ->elevator_lock by debugfs code, meantime hctx->sched_tags is updated with ->elevator_lock, so there isn't such issue.

---
 block/blk-mq.c   |  3 +-
 block/elevator.c | 80 ++++++++++++++++++++++++++++++++++++------------
 2 files changed, 62 insertions(+), 21 deletions(-)
block/blk-mq.c +1 -2
index 3afcddd21586..3ecbf73773ea 100644
@@ -4997,11 +4997,10 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
 		blk_mq_debugfs_register_hctxs(q);
 	}
 
+	/* elv_update_nr_hw_queues() unfreeze queue for us */
 	list_for_each_entry(q, &set->tag_list, tag_set_list)
 		elv_update_nr_hw_queues(q);
 
-	list_for_each_entry(q, &set->tag_list, tag_set_list)
-		blk_mq_unfreeze_queue_nomemrestore(q);
 	memalloc_noio_restore(memflags);
 
 	/* Free the excess tags when nr_hw_queues shrink. */
block/elevator.c +61 -19

172 lines not on this page · raw message

Signature · 1 line
2.47.0
Reviewed-by on v3 16/20
#54 lore raw

Re: [PATCH V3 16/20] block: move elv_register[unregister]_queue out of elevator_lock

Quoted 14 lines · Ming Lei

Move elv_register[unregister]_queue out of ->elevator_lock & queue freezing, so we can kill many lockdep warnings.

elv_register[unregister]_queue() is serialized, and just dealing with sysfs/ debugfs things, no need to be done with queue frozen.

With this change, elevator's ->exit() is called before calling elv_unregister_queue, then user may call into ->show()/store() of elevator's sysfs attributes, and we have covered this issue by adding ELEVATOR_FLAG_DYNG.

For blk-mq debugfs, hctx->sched_tags is always checked with ->elevator_lock by debugfs code, meantime hctx->sched_tags is updated with ->elevator_lock, so there isn't such issue.

Quoted 6 lines
---
  block/blk-mq.c   |  3 +-
  block/elevator.c | 80 ++++++++++++++++++++++++++++++++++++------------
  2 files changed, 62 insertions(+), 21 deletions(-)

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
Reviewed-by on v3 16/20
#55 lore raw

Re: [PATCH V3 16/20] block: move elv_register[unregister]_queue out of elevator_lock

In reply to Ming Lei
Quoted 10 lines · Ming Lei

Move elv_register[unregister]_queue out of ->elevator_lock & queue freezing, so we can kill many lockdep warnings.

elv_register[unregister]_queue() is serialized, and just dealing with sysfs/ debugfs things, no need to be done with queue frozen.

With this change, elevator's ->exit() is called before calling elv_unregister_queue, then user may call into ->show()/store() of elevator's sysfs attributes, and we have covered this issue by adding ELEVATOR_FLAG_DYNG.

Quoted 5 lines

For blk-mq debugfs, hctx->sched_tags is always checked with ->elevator_lock by debugfs code, meantime hctx->sched_tags is updated with ->elevator_lock, so there isn't such issue.

Looks good to me:

#56 lore raw

Re: [PATCH V3 16/20] block: move elv_register[unregister]_queue out of elevator_lock

In reply to Ming Lei

block/elevator.c static void elevator_release(struct kobject *kobj) · line 163

Quoted 2 lines · Ming Lei
block/elevator.c +1 -1
-static void elevator_exit(struct request_queue *q)
+static void __elevator_exit(struct request_queue *q)

Why does elevator_exit grow a spurious __ prefix?

author
#57 lore raw

PATCH v3 17/20 block: move debugfs/sysfs register out of freezing queue

In reply to Ming Lei

Move debugfs/sysfs register out of freezing queue in __blk_mq_update_nr_hw_queues(), so that the following lockdep dependency can be killed:

	#2 (&q->q_usage_counter(io)#16){++++}-{0:0}:
	#1 (fs_reclaim){+.+.}-{0:0}:
	#0 (&sb->s_type->i_mutex_key#3){+.+.}-{4:4}: //debugfs

And registering/un-registering debugfs/sysfs does not require queue to be frozen.

---
 block/blk-mq.c | 15 +++++++--------
 1 file changed, 7 insertions(+), 8 deletions(-)
block/blk-mq.c +7 -8
index 3ecbf73773ea..420150eb5a45 100644
@@ -4961,14 +4961,14 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
 		return;
 
 	memflags = memalloc_noio_save();
-	list_for_each_entry(q, &set->tag_list, tag_set_list)
-		blk_mq_freeze_queue_nomemsave(q);
-
 	list_for_each_entry(q, &set->tag_list, tag_set_list) {
 		blk_mq_debugfs_unregister_hctxs(q);
 		blk_mq_sysfs_unregister_hctxs(q);
 	}
 
+	list_for_each_entry(q, &set->tag_list, tag_set_list)
+		blk_mq_freeze_queue_nomemsave(q);
+
 	if (blk_mq_realloc_tag_set_tags(set, nr_hw_queues) < 0)
 		goto reregister;
 
@@ -4991,16 +4991,15 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
 		blk_mq_map_swqueue(q);
 	}
 
+	/* elv_update_nr_hw_queues() unfreeze queue for us */
+	list_for_each_entry(q, &set->tag_list, tag_set_list)
+		elv_update_nr_hw_queues(q);
+
 reregister:
 	list_for_each_entry(q, &set->tag_list, tag_set_list) {
 		blk_mq_sysfs_register_hctxs(q);
 		blk_mq_debugfs_register_hctxs(q);
 	}
-
-	/* elv_update_nr_hw_queues() unfreeze queue for us */
-	list_for_each_entry(q, &set->tag_list, tag_set_list)
-		elv_update_nr_hw_queues(q);
-
 	memalloc_noio_restore(memflags);
 
 	/* Free the excess tags when nr_hw_queues shrink. */
Signature · 1 line
2.47.0
Reviewed-by on v3 17/20
#58 lore raw

Re: [PATCH V3 17/20] block: move debugfs/sysfs register out of freezing queue

Quoted 8 lines · Ming Lei

Move debugfs/sysfs register out of freezing queue in __blk_mq_update_nr_hw_queues(), so that the following lockdep dependency can be killed:

	#2 (&q->q_usage_counter(io)#16){++++}-{0:0}:
	#1 (fs_reclaim){+.+.}-{0:0}:
	#0 (&sb->s_type->i_mutex_key#3){+.+.}-{4:4}: //debugfs
Quoted 9 lines

And registering/un-registering debugfs/sysfs does not require queue to be frozen.

---
  block/blk-mq.c | 15 +++++++--------
  1 file changed, 7 insertions(+), 8 deletions(-)

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
#59 lore raw

Re: [PATCH V3 17/20] block: move debugfs/sysfs register out of freezing queue

In reply to Ming Lei
Quoted 10 lines · Ming Lei

Move debugfs/sysfs register out of freezing queue in __blk_mq_update_nr_hw_queues(), so that the following lockdep dependency can be killed:

	#2 (&q->q_usage_counter(io)#16){++++}-{0:0}:
	#1 (fs_reclaim){+.+.}-{0:0}:
	#0 (&sb->s_type->i_mutex_key#3){+.+.}-{4:4}: //debugfs

And registering/un-registering debugfs/sysfs does not require queue to be frozen.

Please explain the why it's not required. And talking about lockdep dependencies is not very useful - it either is an actual lock ordering issue or a fale positive, but lockdep is just the messenger.

author
#60 lore raw

Re: [PATCH V3 17/20] block: move debugfs/sysfs register out of freezing queue

Quoted 9 lines · Christoph Hellwig
Quoted 8 lines · Ming Lei

Move debugfs/sysfs register out of freezing queue in __blk_mq_update_nr_hw_queues(), so that the following lockdep dependency can be killed:

	#2 (&q->q_usage_counter(io)#16){++++}-{0:0}:
	#1 (fs_reclaim){+.+.}-{0:0}:
	#0 (&sb->s_type->i_mutex_key#3){+.+.}-{4:4}: //debugfs
Quoted 6 lines
Quoted 2 lines

And registering/un-registering debugfs/sysfs does not require queue to be frozen.

Please explain the why it's not required. And talking about lockdep dependencies is not very useful - it either is an actual lock ordering issue or a fale positive, but lockdep is just the messenger.

Both kobject delete and debugfs unregister drains in-progress .show/.store and read, they do not need to freeze queue.

Thanks,
Ming

author
#61 lore raw

PATCH v3 18/20 block: remove several ->elevator_lock

In reply to Ming Lei

Both blk_mq_map_swqueue() and blk_mq_realloc_hw_ctxs() are called before the request queue is added to tagset list, so the two won't run concurrently with blk_mq_update_nr_hw_queues().

When the two functions are only called from queue initialization or blk_mq_update_nr_hw_queues(), elevator switch can't happen.

So remove these ->elevator_lock uses.

---
 block/blk-mq.c | 19 ++++---------------
 1 file changed, 4 insertions(+), 15 deletions(-)
block/blk-mq.c +4 -15
index 420150eb5a45..e08eda094ae7 100644
@@ -4112,8 +4112,6 @@ static void blk_mq_map_swqueue(struct request_queue *q)
 	struct blk_mq_ctx *ctx;
 	struct blk_mq_tag_set *set = q->tag_set;
 
-	mutex_lock(&q->elevator_lock);
-
 	queue_for_each_hw_ctx(q, hctx, i) {
 		cpumask_clear(hctx->cpumask);
 		hctx->nr_ctx = 0;
@@ -4218,8 +4216,6 @@ static void blk_mq_map_swqueue(struct request_queue *q)
 		hctx->next_cpu = blk_mq_first_mapped_cpu(hctx);
 		hctx->next_cpu_batch = BLK_MQ_CPU_WORK_BATCH;
 	}
-
-	mutex_unlock(&q->elevator_lock);
 }
 
 /*
@@ -4523,16 +4519,9 @@ static void __blk_mq_realloc_hw_ctxs(struct blk_mq_tag_set *set,
 }
 
 static void blk_mq_realloc_hw_ctxs(struct blk_mq_tag_set *set,
-				   struct request_queue *q, bool lock)
+				   struct request_queue *q)
 {
-	if (lock) {
-		/* protect against switching io scheduler  */
-		mutex_lock(&q->elevator_lock);
-		__blk_mq_realloc_hw_ctxs(set, q);
-		mutex_unlock(&q->elevator_lock);
-	} else {
-		__blk_mq_realloc_hw_ctxs(set, q);
-	}
+	__blk_mq_realloc_hw_ctxs(set, q);
 
 	/* unregister cpuhp callbacks for exited hctxs */
 	blk_mq_remove_hw_queues_cpuhp(q);
@@ -4564,7 +4553,7 @@ int blk_mq_init_allocated_queue(struct blk_mq_tag_set *set,
 
 	xa_init(&q->hctx_table);
 
-	blk_mq_realloc_hw_ctxs(set, q, false);
+	blk_mq_realloc_hw_ctxs(set, q);
 	if (!q->nr_hw_queues)
 		goto err_hctxs;
 
@@ -4975,7 +4964,7 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
 fallback:
 	blk_mq_update_queue_map(set);
 	list_for_each_entry(q, &set->tag_list, tag_set_list) {
-		blk_mq_realloc_hw_ctxs(set, q, true);
+		blk_mq_realloc_hw_ctxs(set, q);
 
 		if (q->nr_hw_queues != set->nr_hw_queues) {
 			int i = prev_nr_hw_queues;
Signature · 1 line
2.47.0
Reviewed-by on v3 18/20
#62 lore raw

Re: [PATCH V3 18/20] block: remove several ->elevator_lock

Quoted 7 lines · Ming Lei

Both blk_mq_map_swqueue() and blk_mq_realloc_hw_ctxs() are called before the request queue is added to tagset list, so the two won't run concurrently with blk_mq_update_nr_hw_queues().

When the two functions are only called from queue initialization or blk_mq_update_nr_hw_queues(), elevator switch can't happen.

Quoted 8 lines

So remove these ->elevator_lock uses.

---
  block/blk-mq.c | 19 ++++---------------
  1 file changed, 4 insertions(+), 15 deletions(-)

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
author
#64 lore raw

PATCH v3 19/20 block: move hctx cpuhp add/del out of queue freezing

In reply to Ming Lei

Move hctx cpuhp add/del out of queue freezing for not connecting freeze lock with cpuhp locks, then lockdep warning can be avoided.

This way is safe because both needn't queue to be frozen and scheduler switch isn't allowed.

---
 block/blk-mq.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
block/blk-mq.c +4 -1
index e08eda094ae7..add653f84797 100644
@@ -4964,7 +4964,7 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
 fallback:
 	blk_mq_update_queue_map(set);
 	list_for_each_entry(q, &set->tag_list, tag_set_list) {
-		blk_mq_realloc_hw_ctxs(set, q);
+		__blk_mq_realloc_hw_ctxs(set, q);
 
 		if (q->nr_hw_queues != set->nr_hw_queues) {
 			int i = prev_nr_hw_queues;
@@ -4988,6 +4988,9 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
 	list_for_each_entry(q, &set->tag_list, tag_set_list) {
 		blk_mq_sysfs_register_hctxs(q);
 		blk_mq_debugfs_register_hctxs(q);
+
+		blk_mq_remove_hw_queues_cpuhp(q);
+		blk_mq_add_hw_queues_cpuhp(q);
 	}
 	memalloc_noio_restore(memflags);
 
Signature · 1 line
2.47.0
Reviewed-by on v3 19/20
#65 lore raw

Re: [PATCH V3 19/20] block: move hctx cpuhp add/del out of queue freezing

Quoted 26 lines · Ming Lei

Move hctx cpuhp add/del out of queue freezing for not connecting freeze lock with cpuhp locks, then lockdep warning can be avoided.

This way is safe because both needn't queue to be frozen and scheduler switch isn't allowed.

---
  block/blk-mq.c | 5 ++++-
  1 file changed, 4 insertions(+), 1 deletion(-)
block/blk-mq.c +1 -1
index e08eda094ae7..add653f84797 100644
@@ -4964,7 +4964,7 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
  fallback:
  	blk_mq_update_queue_map(set);
  	list_for_each_entry(q, &set->tag_list, tag_set_list) {
-		blk_mq_realloc_hw_ctxs(set, q);
+		__blk_mq_realloc_hw_ctxs(set, q);
 
  		if (q->nr_hw_queues != set->nr_hw_queues) {
  			int i = prev_nr_hw_queues;

block/blk-mq.c static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, · line 4996

Quoted 9 lines
block/blk-mq.c +3 -0
@@ -4988,6 +4988,9 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
  	list_for_each_entry(q, &set->tag_list, tag_set_list) {
  		blk_mq_sysfs_register_hctxs(q);
  		blk_mq_debugfs_register_hctxs(q);
+
+		blk_mq_remove_hw_queues_cpuhp(q);
+		blk_mq_add_hw_queues_cpuhp(q);
  	}
  	memalloc_noio_restore(memflags);
 

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
author
#66 lore raw

PATCH v3 20/20 block: move wbt_enable_default() out of queue freezing from sched ->exit()

In reply to Ming Lei

scheduler's ->exit() is called with queue frozen and elevator lock is held, and wbt_enable_default() can't be called with queue frozen, otherwise the following lockdep warning is triggered:

	#6 (&q->rq_qos_mutex){+.+.}-{4:4}:
	#5 (&eq->sysfs_lock){+.+.}-{4:4}:
	#4 (&q->elevator_lock){+.+.}-{4:4}:
	#3 (&q->q_usage_counter(io)#3){++++}-{0:0}:
	#2 (fs_reclaim){+.+.}-{0:0}:
	#1 (&sb->s_type->i_mutex_key#3){+.+.}-{4:4}:
	#0 (&q->debugfs_mutex){+.+.}-{4:4}:

Fix the issue by moving wbt_enable_default() out of bfq's exit(), and call it from elevator_change_done().

Meantime add disk->rqos_state_mutex for covering wbt state change, which matches the purpose more than ->elevator_lock.

---
 block/bfq-iosched.c    |  2 +-
 block/blk-sysfs.c      |  6 ++----
 block/blk-wbt.c        | 10 ++++++++--
 block/elevator.c       |  5 +++++
 block/elevator.h       |  1 +
 block/genhd.c          |  1 +
 include/linux/blkdev.h |  2 ++
 7 files changed, 20 insertions(+), 7 deletions(-)
block/bfq-iosched.c +1 -1
index cc6f59836dcd..0cb1e9873aab 100644
@@ -7211,7 +7211,7 @@ static void bfq_exit_queue(struct elevator_queue *e)
 
 	blk_stat_disable_accounting(bfqd->queue);
 	blk_queue_flag_clear(QUEUE_FLAG_DISABLE_WBT_DEF, bfqd->queue);
-	wbt_enable_default(bfqd->queue->disk);
+	set_bit(ELEVATOR_FLAG_ENABLE_WBT_ON_EXIT, &e->flags);
 
 	kfree(bfqd);
 }
block/blk-sysfs.c +2 -4
index 9aa05666f32a..fafe9e9e97cc 100644
@@ -593,7 +593,7 @@ static ssize_t queue_wb_lat_store(struct gendisk *disk, const char *page,
 		return -EINVAL;
 
 	memflags = blk_mq_freeze_queue(q);
-	mutex_lock(&q->elevator_lock);
+	mutex_lock(&disk->rqos_state_mutex);
 
 	rqos = wbt_rq_qos(q);
 	if (!rqos) {
@@ -622,7 +622,7 @@ static ssize_t queue_wb_lat_store(struct gendisk *disk, const char *page,
 
 	blk_mq_unquiesce_queue(q);
 out:
-	mutex_unlock(&q->elevator_lock);
+	mutex_unlock(&disk->rqos_state_mutex);
 	blk_mq_unfreeze_queue(q, memflags);
 
 	return ret;
@@ -870,9 +870,7 @@ int blk_register_queue(struct gendisk *disk)
 		goto out_unregister_ia_ranges;
 
 	elevator_set_default(q);
-	mutex_lock(&q->elevator_lock);
 	wbt_enable_default(disk);
-	mutex_unlock(&q->elevator_lock);
 
 	blk_queue_flag_set(QUEUE_FLAG_REGISTERED, q);
 
block/blk-wbt.c +8 -2
index 29cd2e33666f..c8588bae1c1b 100644
@@ -704,6 +704,8 @@ void wbt_enable_default(struct gendisk *disk)
 	struct rq_qos *rqos;
 	bool enable = IS_ENABLED(CONFIG_BLK_WBT_MQ);
 
+	mutex_lock(&disk->rqos_state_mutex);
+
 	if (blk_queue_disable_wbt(q))
 		enable = false;
 
@@ -712,15 +714,17 @@ void wbt_enable_default(struct gendisk *disk)
 	if (rqos) {
 		if (enable && RQWB(rqos)->enable_state == WBT_STATE_OFF_DEFAULT)
 			RQWB(rqos)->enable_state = WBT_STATE_ON_DEFAULT;
-		return;
+		goto unlock;
 	}
 
 	/* Queue not registered? Maybe shutting down... */
 	if (!blk_queue_registered(q))
-		return;
+		goto unlock;
 
 	if (queue_is_mq(q) && enable)
 		wbt_init(disk);
+unlock:
+	mutex_unlock(&disk->rqos_state_mutex);
 }
 EXPORT_SYMBOL_GPL(wbt_enable_default);
 
@@ -773,11 +777,13 @@ void wbt_disable_default(struct gendisk *disk)
 	struct rq_wb *rwb;
 	if (!rqos)
 		return;
+	mutex_lock(&disk->rqos_state_mutex);
 	rwb = RQWB(rqos);
 	if (rwb->enable_state == WBT_STATE_ON_DEFAULT) {
 		blk_stat_deactivate(rwb->cb);
 		rwb->enable_state = WBT_STATE_OFF_DEFAULT;
 	}
+	mutex_unlock(&disk->rqos_state_mutex);
 }
 EXPORT_SYMBOL_GPL(wbt_disable_default);
 
block/elevator.c +5 -0
index aec8081a6be3..a637426da56d 100644
@@ -663,8 +663,13 @@ static int elevator_change_done(struct request_queue *q,
 	int ret = 0;
 
 	if (ctx->old) {
+		bool enable_wbt = test_bit(ELEVATOR_FLAG_ENABLE_WBT_ON_EXIT,
+				&ctx->old->flags);
+
 		elv_unregister_queue(q, ctx->old);
 		kobject_put(&ctx->old->kobj);
+		if (enable_wbt)
+			wbt_enable_default(q->disk);
 	}
 	if (ctx->new) {
 		ret = elv_register_queue(q, ctx->new, ctx->uevent);
block/elevator.h +1 -0
index 76a90a1b7ed6..a07ce773a38f 100644
@@ -122,6 +122,7 @@ struct elevator_queue
 
 #define ELEVATOR_FLAG_REGISTERED	0
 #define ELEVATOR_FLAG_DYING		1
+#define ELEVATOR_FLAG_ENABLE_WBT_ON_EXIT	2
 
 /*
  * block elevator interface
block/genhd.c +1 -0
index a50063c4c40f..e13be309552a 100644
@@ -1452,6 +1452,7 @@ struct gendisk *__alloc_disk_node(struct request_queue *q, int node_id,
 #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED
 	INIT_LIST_HEAD(&disk->slave_bdevs);
 #endif
+	mutex_init(&disk->rqos_state_mutex);
 	return disk;
 
 out_erase_part0:
include/linux/blkdev.h +2 -0
index 228b2e5ad493..281d56b910c2 100644
@@ -218,6 +218,8 @@ struct gendisk {
 	 * devices that do not have multiple independent access ranges.
 	 */
 	struct blk_independent_access_ranges *ia_ranges;
+
+	struct mutex rqos_state_mutex;	/* rqos state change mutex */
 };
 
 /**
Signature · 1 line
2.47.0
Reviewed-by on v3 20/20
#67 lore raw

Re: [PATCH V3 20/20] block: move wbt_enable_default() out of queue freezing from sched ->exit()

Quoted 12 lines · Ming Lei

scheduler's ->exit() is called with queue frozen and elevator lock is held, and wbt_enable_default() can't be called with queue frozen, otherwise the following lockdep warning is triggered:

	#6 (&q->rq_qos_mutex){+.+.}-{4:4}:
	#5 (&eq->sysfs_lock){+.+.}-{4:4}:
	#4 (&q->elevator_lock){+.+.}-{4:4}:
	#3 (&q->q_usage_counter(io)#3){++++}-{0:0}:
	#2 (fs_reclaim){+.+.}-{0:0}:
	#1 (&sb->s_type->i_mutex_key#3){+.+.}-{4:4}:
	#0 (&q->debugfs_mutex){+.+.}-{4:4}:
Quoted 7 lines

Fix the issue by moving wbt_enable_default() out of bfq's exit(), and call it from elevator_change_done().

Meantime add disk->rqos_state_mutex for covering wbt state change, which matches the purpose more than ->elevator_lock.

Looks good to me:

#68 lore raw

Re: [PATCH V3 20/20] block: move wbt_enable_default() out of queue freezing from sched ->exit()

In reply to Ming Lei
Quoted 12 lines · Ming Lei

scheduler's ->exit() is called with queue frozen and elevator lock is held, and wbt_enable_default() can't be called with queue frozen, otherwise the following lockdep warning is triggered:

	#6 (&q->rq_qos_mutex){+.+.}-{4:4}:
	#5 (&eq->sysfs_lock){+.+.}-{4:4}:
	#4 (&q->elevator_lock){+.+.}-{4:4}:
	#3 (&q->q_usage_counter(io)#3){++++}-{0:0}:
	#2 (fs_reclaim){+.+.}-{0:0}:
	#1 (&sb->s_type->i_mutex_key#3){+.+.}-{4:4}:
	#0 (&q->debugfs_mutex){+.+.}-{4:4}:
Quoted 7 lines

Fix the issue by moving wbt_enable_default() out of bfq's exit(), and call it from elevator_change_done().

Meantime add disk->rqos_state_mutex for covering wbt state change, which matches the purpose more than ->elevator_lock.

While testing this patch on my machine using blktests, I stumbled upon a lockdep splat shown below.(I could consistently recreate it):

run blktests block/005 at 2025-04-28 06:57:51

======================================================
WARNING: possible circular locking dependency detected
6.15.0-rc2+ #174 Not tainted
------------------------------------------------------
check/8088 is trying to acquire lock:
c0000000a0c03538 (&disk->rqos_state_mutex){+.+.}-{4:4}, at: wbt_disable_default+0x9c/0x118

but task is already holding lock:
c00000005b8f6c38 (&q->elevator_lock){+.+.}-{4:4}, at: elevator_change+0x94/0x214

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #3 (&q->elevator_lock){+.+.}-{4:4}:

       __mutex_lock+0x128/0xdd8
       elevator_change+0x94/0x214
       elv_iosched_store+0x14c/0x1f4
       queue_attr_store+0x194/0x1d0
       sysfs_kf_write+0xbc/0x110
       kernfs_fop_write_iter+0x264/0x384
       vfs_write+0x5b0/0x77c
       ksys_write+0xa0/0x180
       system_call_exception+0x1b0/0x4f0
       system_call_vectored_common+0x15c/0x2ec

-> #2 (&q->q_usage_counter(io)#23){++++}-{0:0}:

       blk_alloc_queue+0x46c/0x4bc
       blk_mq_alloc_queue+0xc0/0x160
       __blk_mq_alloc_disk+0x34/0x128
       nvme_alloc_ns+0x140/0x1804 [nvme_core]
       nvme_scan_ns+0x42c/0x564 [nvme_core]
       async_run_entry_fn+0x9c/0x30c
       process_one_work+0x514/0xd38
       worker_thread+0x390/0x6dc
       kthread+0x230/0x278
       start_kernel_thread+0x14/0x18

-> #1 (fs_reclaim){+.+.}-{0:0}:

       fs_reclaim_acquire+0x114/0x150
       __kmalloc_cache_noprof+0x70/0x5c0
       wbt_init+0x64/0x2fc
       wbt_enable_default+0x140/0x15c
       elevator_change_done+0x314/0x3a8
       elv_iosched_store+0x14c/0x1f4
       queue_attr_store+0x194/0x1d0
       sysfs_kf_write+0xbc/0x110
       kernfs_fop_write_iter+0x264/0x384
       vfs_write+0x5b0/0x77c
       ksys_write+0xa0/0x180
       system_call_exception+0x1b0/0x4f0
       system_call_vectored_common+0x15c/0x2ec

-> #0 (&disk->rqos_state_mutex){+.+.}-{4:4}:

       __lock_acquire+0x1b5c/0x29f8
       lock_acquire+0x23c/0x3f8
       __mutex_lock+0x128/0xdd8
       wbt_disable_default+0x9c/0x118
       bfq_init_queue+0x7b0/0x8c0
       blk_mq_init_sched+0x29c/0x3a8
       __elevator_change+0x3a4/0x8a4
       elevator_change+0x1a4/0x214
       elv_iosched_store+0x14c/0x1f4
       queue_attr_store+0x194/0x1d0
       sysfs_kf_write+0xbc/0x110
       kernfs_fop_write_iter+0x264/0x384
       vfs_write+0x5b0/0x77c
       ksys_write+0xa0/0x180
       system_call_exception+0x1b0/0x4f0
       system_call_vectored_common+0x15c/0x2ec

other info that might help us debug this:

Chain exists of:
  &disk->rqos_state_mutex --> &q->q_usage_counter(io)#23 --> &q->elevator_lock

 Possible unsafe locking scenario:
       CPU0                    CPU1
       ----                    ----
  lock(&q->elevator_lock);
                               lock(&q->q_usage_counter(io)#23);
                               lock(&q->elevator_lock);

lock(&disk->rqos_state_mutex);

*** DEADLOCK ***

7 locks held by check/8088:
#0: c0000000873f2400 (sb_writers#3){.+.+}-{0:0}, at: ksys_write+0xa0/0x180
#1: c00000008c10c088 (&of->mutex#2){+.+.}-{4:4}, at: kernfs_fop_write_iter+0x1e0/0x384
#2: c000000085239248 (kn->active#57){.+.+}-{0:0}, at: kernfs_fop_write_iter+0x1f8/0x384
#3: c0000000f801c190 (&set->update_nr_hwq_sema){.+.+}-{4:4}, at: elv_iosched_store+0x13c/0x1f4 #4: c00000005b8f6718 (&q->q_usage_counter(io)#23){++++}-{0:0}, at: blk_mq_freeze_queue_nomemsave+0x28/0x40 #5: c00000005b8f6750 (&q->q_usage_counter(queue)#21){+.+.}-{0:0}, at: blk_mq_freeze_queue_nomemsave+0x28/0x40 #6: c00000005b8f6c38 (&q->elevator_lock){+.+.}-{4:4}, at: elevator_change+0x94/0x214

stack backtrace:
CPU: 26 UID: 0 PID: 8088 Comm: check Kdump: loaded Not tainted 6.15.0-rc2+ #174 VOLUNTARY
Hardware name: IBM,9043-MRX POWER10 (architected) 0x800200 0xf000006 of:IBM,FW1060.00 (NM1060_028) hv:phyp pSeries
Call Trace:
[c0000000d7497240] [c0000000017b9888] dump_stack_lvl+0x100/0x184 (unreliable)
[c0000000d7497270] [c0000000002b546c] print_circular_bug+0x448/0x604
[c0000000d7497320] [c0000000002b5874] check_noncircular+0x24c/0x26c
[c0000000d74973f0] [c0000000002bbb78] __lock_acquire+0x1b5c/0x29f8
[c0000000d7497520] [c0000000002b915c] lock_acquire+0x23c/0x3f8
[c0000000d7497620] [c00000000181277c] __mutex_lock+0x128/0xdd8
[c0000000d7497780] [c000000000c73bf8] wbt_disable_default+0x9c/0x118
[c0000000d74977c0] [c000000000c4c2c0] bfq_init_queue+0x7b0/0x8c0
[c0000000d7497890] [c000000000bff634] blk_mq_init_sched+0x29c/0x3a8
[c0000000d7497910] [c000000000bc2a18] __elevator_change+0x3a4/0x8a4
[c0000000d74979b0] [c000000000bc30bc] elevator_change+0x1a4/0x214
[c0000000d7497a00] [c000000000bc427c] elv_iosched_store+0x14c/0x1f4
[c0000000d7497ae0] [c000000000bd07ec] queue_attr_store+0x194/0x1d0
[c0000000d7497c00] [c000000000a40f00] sysfs_kf_write+0xbc/0x110
[c0000000d7497c50] [c000000000a3cc4c] kernfs_fop_write_iter+0x264/0x384
[c0000000d7497cb0] [c0000000008bb9bc] vfs_write+0x5b0/0x77c
[c0000000d7497d90] [c0000000008bbf88] ksys_write+0xa0/0x180
[c0000000d7497df0] [c000000000039f70] system_call_exception+0x1b0/0x4f0
[c0000000d7497e50] [c00000000000cedc] system_call_vectored_common+0x15c/0x2ec
--- interrupt: 3000 at 0x7fffa413b034
NIP:  00007fffa413b034 LR: 00007fffa413b034 CTR: 0000000000000000
REGS: c0000000d7497e80 TRAP: 3000   Not tainted  (6.15.0-rc2+)
MSR:  800000000280f033 <SF,VEC,VSX,EE,PR,FP,ME,IR,DR,RI,LE>  CR: 44422408  XER: 00000000
IRQMASK: 0
GPR00: 0000000000000004 00007ffffd011260 000000010dfa7e00 0000000000000001
GPR04: 000000011c30b720 0000000000000004 0000000000000010 0000000000000001
GPR08: 0000000000000003 0000000000000000 0000000000000000 0000000000000000
GPR12: 0000000000000000 00007fffa43fab60 000000011c3adbc0 000000010dfa87b8
GPR16: 000000010dfa94d8 0000000020000000 0000000000000000 000000010deb9070
GPR20: 000000010df4beb8 00007ffffd011404 000000010df4f8a0 000000010dfa89bc
GPR24: 000000010dfa8a50 0000000000000000 000000011c30b720 0000000000000004
GPR28: 0000000000000004 00007fffa42418e0 000000011c30b720 0000000000000004

NIP [00007fffa413b034] 0x7fffa413b034
LR [00007fffa413b034] 0x7fffa413b034
--- interrupt: 3000

From the changes in this patch it appears that we have now got this new "disk->rqos_state_mutex" which we're holding while allocating the dynamic memory in wbt_init(). One way is to define GFP_NOIO scope while holding this mutex but then looking further it seems that we don't really need to keep holding this lock in wbt_enable_default() all the way till wbt_init returns. Probably, unlocking this mutex as soon as we're done updating the rqos state should be sufficient.

Similarly, in queue_wb_lat_store we may need to hold this mutex only while we invoke wbt_set_min_lat. And I just realized that in queue_wb_lat_show we need to replace q->elevator_lock with disk->rqos_state_mutex. I have attached the above change for your reference. You may address this in your next patch.

Thanks,
--Nilay

block/blk-sysfs.c +4 -4
index fafe9e9e97cc..0248dc9438c3 100644
@@ -560,7 +560,7 @@ static ssize_t queue_wb_lat_show(struct gendisk *disk, char *page)
 	ssize_t ret;
 	struct request_queue *q = disk->queue;
 
-	mutex_lock(&q->elevator_lock);
+	mutex_lock(&disk->rqos_state_mutex);
 	if (!wbt_rq_qos(q)) {
 		ret = -EINVAL;
 		goto out;
@@ -573,7 +573,7 @@ static ssize_t queue_wb_lat_show(struct gendisk *disk, char *page)
 
 	ret = sysfs_emit(page, "%llu\n", div_u64(wbt_get_min_lat(q), 1000));
 out:
-	mutex_unlock(&q->elevator_lock);
+	mutex_unlock(&disk->rqos_state_mutex);
 	return ret;
 }
 
@@ -593,7 +593,6 @@ static ssize_t queue_wb_lat_store(struct gendisk *disk, const char *page,
 		return -EINVAL;
 
 	memflags = blk_mq_freeze_queue(q);
-	mutex_lock(&disk->rqos_state_mutex);
 
 	rqos = wbt_rq_qos(q);
 	if (!rqos) {
@@ -618,11 +617,12 @@ static ssize_t queue_wb_lat_store(struct gendisk *disk, const char *page,
 	 */
 	blk_mq_quiesce_queue(q);
 
+	mutex_lock(&disk->rqos_state_mutex);
 	wbt_set_min_lat(q, val);
+	mutex_unlock(&disk->rqos_state_mutex);
 
 	blk_mq_unquiesce_queue(q);
 out:
-	mutex_unlock(&disk->rqos_state_mutex);
 	blk_mq_unfreeze_queue(q, memflags);
 
 	return ret;
block/blk-wbt.c +4 -4
index c8588bae1c1b..74ae7131ada9 100644
@@ -714,17 +714,17 @@ void wbt_enable_default(struct gendisk *disk)
 	if (rqos) {
 		if (enable && RQWB(rqos)->enable_state == WBT_STATE_OFF_DEFAULT)
 			RQWB(rqos)->enable_state = WBT_STATE_ON_DEFAULT;
-		goto unlock;
+		mutex_unlock(&disk->rqos_state_mutex);
+		return;
 	}
+	mutex_unlock(&disk->rqos_state_mutex);
 
 	/* Queue not registered? Maybe shutting down... */
 	if (!blk_queue_registered(q))
-		goto unlock;
+		return;
 
 	if (queue_is_mq(q) && enable)
 		wbt_init(disk);
-unlock:
-	mutex_unlock(&disk->rqos_state_mutex);
 }
 EXPORT_SYMBOL_GPL(wbt_enable_default);
 
#69 lore raw
In reply to Ming Lei

Am 24.04.25 um 17:21 schrieb Ming Lei:

Quoted 68 lines

Hello Jens,

This patchset cleans up elevator change code, and unifying it via single helper, meantime moves kobject_add/del & debugfs register/unregister out of queue freezing & elevator_lock. This way fixes many lockdep warnings reported recently, especially since fs_reclaim is connected with freeze lock manually by commit ffa1e7ada456 ("block: Make request_queue lockdep splats show up earlier").

Thanks,
Ming

V3:
- replace srcu with rw_sem for avoiding race between add/del disk & elevator switch and updating nr_hw_queues (Nilay Shoff)

- add elv_update_nr_hw_queues() for elevator reattachment in case of updating nr_hw_queues, meantime keep elv_change_ctx as local structure (Christoph)

- replace ->elevator_lock with disk->rqos_state_mutex for covering wbt state change

- add new patch "block: use q->elevator with ->elevator_lock held in elv_iosched_show()"

- small cleanup & commit log improvement

V2:
- retry add/del disk when blk_mq_update_nr_hw_queues() is in-progress

- swap blk_mq_add_queue_tag_set() with blk_mq_map_swqueue() in blk_mq_init_allocated_queue() (Nilay Shroff)

- move ELEVATOR_FLAG_DISABLE_WBT to request queue's flags (Nilay Shoff)

- fix race because of delaying elevator unregister

- define flags of elv_change_ctx as bool (Christoph)

- improve comment and commit log (Christoph)

Ming Lei (20):
block: move blk_mq_add_queue_tag_set() after blk_mq_map_swqueue()
block: move ELEVATOR_FLAG_DISABLE_WBT a request queue flag
block: don't call freeze queue in elevator_switch() and
elevator_disable()
block: use q->elevator with ->elevator_lock held in elv_iosched_show()
block: add two helpers for registering/un-registering sched debugfs
block: move sched debugfs register into elvevator_register_queue
block: prevent adding/deleting disk during updating nr_hw_queues
block: don't allow to switch elevator if updating nr_hw_queues is
in-progress
block: simplify elevator reattachment for updating nr_hw_queues
block: move blk_unregister_queue() & device_del() after freeze wait
block: move queue freezing & elevator_lock into elevator_change()
block: add struct elv_change_ctx for unifying elevator change
block: unifying elevator change
block: pass elevator_queue to elv_register_queue & unregister_queue
block: fail to show/store elevator sysfs attribute if elevator is
dying
block: move elv_register[unregister]_queue out of elevator_lock
block: move debugfs/sysfs register out of freezing queue
block: remove several ->elevator_lock
block: move hctx cpuhp add/del out of queue freezing
block: move wbt_enable_default() out of queue freezing from sched

    ->exit()
Quoted 13 lines
 block/bfq-iosched.c    |   6 +-
 block/blk-mq-debugfs.c |  12 +-
 block/blk-mq-sched.c   |  41 +++---
 block/blk-mq.c         | 132 +++---------------
 block/blk-sysfs.c      |  24 ++--
 block/blk-wbt.c        |  13 +-
 block/blk.h            |   8 +-
 block/elevator.c       | 302 ++++++++++++++++++++++++++++-------------
 block/elevator.h       |   6 +-
 block/genhd.c          | 129 +++++++++++-------
 include/linux/blk-mq.h |   3 +
 include/linux/blkdev.h |   5 +
 12 files changed, 365 insertions(+), 316 deletions(-)

Hi,
while testing the patchset on s390 I still get the following lockdep splat on each boot:

======================================================
 WARNING: possible circular locking dependency detected 6.15.0-rc4-gc2b4d8dcb3d2 #3 Not tainted
 ------------------------------------------------------
 (udev-worker)/1810 is trying to acquire lock:
 0000005fb84de3a8 (&q->elevator_lock){+.+.}-{4:4}, at: elevator_change+0x54/0x130

 but task is already holding lock:
 0000005fb84dde18 (&q->q_usage_counter(io)#34){++++}-{0:0}, at: blk_mq_freeze_queue_nomemsave+0x26/0x40

 which lock already depends on the new lock.

 the existing dependency chain (in reverse order) is:

 -> #3 (&q->q_usage_counter(io)#34){++++}-{0:0}:

        __lock_acquire+0x6da/0xcc0
        lock_acquire.part.0+0x10c/0x290
        lock_acquire+0xb0/0x1a0
        blk_alloc_queue+0x306/0x340
        blk_mq_alloc_queue+0x60/0xd0
        scsi_alloc_sdev+0x27c/0x3b0
        scsi_probe_and_add_lun+0x31a/0x480
        scsi_report_lun_scan+0x382/0x430
        __scsi_scan_target+0x11a/0x240
        scsi_scan_target+0xdc/0x100
        fc_scsi_scan_rport+0xc2/0xd0
        process_one_work+0x2a6/0x5d0
        worker_thread+0x220/0x410
        kthread+0x164/0x2d0
        __ret_from_fork+0x3c/0x60
        ret_from_fork+0xa/0x38

 -> #2 (fs_reclaim){+.+.}-{0:0}:

        __lock_acquire+0x6da/0xcc0
        lock_acquire.part.0+0x10c/0x290
        lock_acquire+0xb0/0x1a0
        __fs_reclaim_acquire+0x44/0x50
        fs_reclaim_acquire+0xba/0x100
        __kmalloc_noprof+0xae/0x5e0
        pcpu_alloc_chunk+0x30/0x170
        pcpu_create_chunk+0x22/0x130
        pcpu_alloc_noprof+0x842/0x970
        do_kmem_cache_create+0x1e0/0x4b0
        __kmem_cache_create_args+0x238/0x340
        register_ftrace_graph+0x438/0x460
        trace_selftest_startup_function_graph+0x62/0x260
        run_tracer_selftest+0x116/0x1b0
        register_tracer+0x192/0x260
        do_one_initcall+0x4a/0x180
        do_initcalls+0x146/0x170
        kernel_init_freeable+0x230/0x270
        kernel_init+0x2e/0x188
        __ret_from_fork+0x3c/0x60
        ret_from_fork+0xa/0x38

 -> #1 (pcpu_alloc_mutex){+.+.}-{4:4}:

        __lock_acquire+0x6da/0xcc0
        lock_acquire.part.0+0x10c/0x290
        lock_acquire+0xb0/0x1a0
        __mutex_lock+0xae/0xa20
        mutex_lock_killable_nested+0x32/0x40
        pcpu_alloc_noprof+0x6ea/0x970
        sbitmap_init_node+0x11a/0x230
        sbitmap_queue_init_node+0x3e/0x1c0
        blk_mq_init_tags+0xac/0x140
        blk_mq_alloc_map_and_rqs+0xb0/0x180
        blk_mq_init_sched+0xfc/0x200
        elevator_switch+0x74/0x260
        __elevator_change+0xde/0x150
        elevator_change+0xe4/0x130
        elevator_set_default+0x6c/0xc0
        blk_register_queue+0x188/0x210
        __add_disk_fwnode+0x202/0x420
        add_disk_fwnode+0x7a/0xb0
        sd_probe+0x22e/0x3f0
        really_probe+0x2ac/0x430
        driver_probe_device+0x3c/0xc0
        __device_attach_driver+0xc0/0x140
        bus_for_each_drv+0x90/0xe0
        __device_attach_async_helper+0x94/0xf0
        async_run_entry_fn+0x4a/0x180
        process_one_work+0x2a6/0x5d0
        worker_thread+0x220/0x410
        kthread+0x164/0x2d0
        __ret_from_fork+0x3c/0x60
        ret_from_fork+0xa/0x38

 -> #0 (&q->elevator_lock){+.+.}-{4:4}:

        check_prev_add+0x16c/0xff0
        validate_chain+0x734/0x9f0
        __lock_acquire+0x6da/0xcc0
        lock_acquire.part.0+0x10c/0x290
        lock_acquire+0xb0/0x1a0
        __mutex_lock+0xae/0xa20
        mutex_lock_nested+0x32/0x40
        elevator_change+0x54/0x130
        elv_iosched_store+0xec/0x140
        kernfs_fop_write_iter+0x166/0x230
        vfs_write+0x1ac/0x480
        ksys_write+0x7c/0x100
        __do_syscall+0x156/0x290
        system_call+0x74/0x98

        other info that might help us debug this:

 Chain exists of:

                 &q->elevator_lock --> fs_reclaim --> &q->q_usage_counter(io)#34

  Possible unsafe locking scenario:

        CPU0                    CPU1
        ----                    ---- lock(&q->q_usage_counter(io)#34);
                                lock(fs_reclaim); lock(&q->q_usage_counter(io)#34); lock(&q->elevator_lock);

                *** DEADLOCK ***

 6 locks held by (udev-worker)/1810:
  #0: 0000005f9d141450 (sb_writers#3){.+.+}-{0:0}, at: ksys_write+0x7c/0x100
  #1: 0000005f8ee55090 (&of->mutex#2){+.+.}-{4:4}, at: kernfs_fop_write_iter+0x12a/0x230
  #2: 0000005fd0be69a0 (kn->active#68){.+.+}-{0:0}, at: kernfs_fop_write_iter+0x136/0x230
  #3: 00000066feee93c0 (&set->update_nr_hwq_sema/1){.+.+}-{4:4}, at: elv_iosched_store+0xde/0x140 #4: 0000005fb84dde18 (&q->q_usage_counter(io)#34){++++}-{0:0}, at: blk_mq_freeze_queue_nomemsave+0x26/0x40 #5: 0000005fb84dde58 (&q->q_usage_counter(queue)#11){++++}-{0:0}, at: blk_mq_freeze_queue_nomemsave+0x26/0x40

               stack backtrace:
 CPU: 18 UID: 0 PID: 1810 Comm: (udev-worker) Not tainted 6.15.0-rc4-gc2b4d8dcb3d2 #3 PREEMPT
 Hardware name: IBM 9175 ME1 701 (LPAR)
 Call Trace:
  [<00000162f415409a>] dump_stack_lvl+0xa2/0xe8
  [<00000162f424cf9c>] print_circular_bug+0x18c/0x210
  [<00000162f424d198>] check_noncircular+0x178/0x190
  [<00000162f424e57c>] check_prev_add+0x16c/0xff0
  [<00000162f424fb34>] validate_chain+0x734/0x9f0
  [<00000162f4251f8a>] __lock_acquire+0x6da/0xcc0
  [<00000162f425267c>] lock_acquire.part.0+0x10c/0x290
  [<00000162f42528b0>] lock_acquire+0xb0/0x1a0
  [<00000162f525695e>] __mutex_lock+0xae/0xa20
  [<00000162f5257302>] mutex_lock_nested+0x32/0x40
  [<00000162f4c00a54>] elevator_change+0x54/0x130
  [<00000162f4c0136c>] elv_iosched_store+0xec/0x140
  [<00000162f46b6f66>] kernfs_fop_write_iter+0x166/0x230
  [<00000162f45cbe3c>] vfs_write+0x1ac/0x480
  [<00000162f45cc2ac>] ksys_write+0x7c/0x100
  [<00000162f524e956>] __do_syscall+0x156/0x290
  [<00000162f525ff94>] system_call+0x74/0x98

 INFO: lockdep is turned off.

The trigger is a udev rule changing the scheduler for the device.

thanks,
Stefan

author
#70 lore raw
Quoted 142 lines · Stefan Haberland

Am 24.04.25 um 17:21 schrieb Ming Lei:

Quoted 68 lines

Hello Jens,

This patchset cleans up elevator change code, and unifying it via single helper, meantime moves kobject_add/del & debugfs register/unregister out of queue freezing & elevator_lock. This way fixes many lockdep warnings reported recently, especially since fs_reclaim is connected with freeze lock manually by commit ffa1e7ada456 ("block: Make request_queue lockdep splats show up earlier").

Thanks,
Ming

V3:
- replace srcu with rw_sem for avoiding race between add/del disk & elevator switch and updating nr_hw_queues (Nilay Shoff)

- add elv_update_nr_hw_queues() for elevator reattachment in case of updating nr_hw_queues, meantime keep elv_change_ctx as local structure (Christoph)

- replace ->elevator_lock with disk->rqos_state_mutex for covering wbt state change

- add new patch "block: use q->elevator with ->elevator_lock held in elv_iosched_show()"

- small cleanup & commit log improvement

V2:
- retry add/del disk when blk_mq_update_nr_hw_queues() is in-progress

- swap blk_mq_add_queue_tag_set() with blk_mq_map_swqueue() in blk_mq_init_allocated_queue() (Nilay Shroff)

- move ELEVATOR_FLAG_DISABLE_WBT to request queue's flags (Nilay Shoff)

- fix race because of delaying elevator unregister

- define flags of elv_change_ctx as bool (Christoph)

- improve comment and commit log (Christoph)

Ming Lei (20):
block: move blk_mq_add_queue_tag_set() after blk_mq_map_swqueue()
block: move ELEVATOR_FLAG_DISABLE_WBT a request queue flag
block: don't call freeze queue in elevator_switch() and
elevator_disable()
block: use q->elevator with ->elevator_lock held in elv_iosched_show()
block: add two helpers for registering/un-registering sched debugfs
block: move sched debugfs register into elvevator_register_queue
block: prevent adding/deleting disk during updating nr_hw_queues
block: don't allow to switch elevator if updating nr_hw_queues is
in-progress
block: simplify elevator reattachment for updating nr_hw_queues
block: move blk_unregister_queue() & device_del() after freeze wait
block: move queue freezing & elevator_lock into elevator_change()
block: add struct elv_change_ctx for unifying elevator change
block: unifying elevator change
block: pass elevator_queue to elv_register_queue & unregister_queue
block: fail to show/store elevator sysfs attribute if elevator is
dying
block: move elv_register[unregister]_queue out of elevator_lock
block: move debugfs/sysfs register out of freezing queue
block: remove several ->elevator_lock
block: move hctx cpuhp add/del out of queue freezing
block: move wbt_enable_default() out of queue freezing from sched

    ->exit()
Quoted 13 lines
 block/bfq-iosched.c    |   6 +-
 block/blk-mq-debugfs.c |  12 +-
 block/blk-mq-sched.c   |  41 +++---
 block/blk-mq.c         | 132 +++---------------
 block/blk-sysfs.c      |  24 ++--
 block/blk-wbt.c        |  13 +-
 block/blk.h            |   8 +-
 block/elevator.c       | 302 ++++++++++++++++++++++++++++-------------
 block/elevator.h       |   6 +-
 block/genhd.c          | 129 +++++++++++-------
 include/linux/blk-mq.h |   3 +
 include/linux/blkdev.h |   5 +
 12 files changed, 365 insertions(+), 316 deletions(-)

Hi,
while testing the patchset on s390 I still get the following lockdep splat on each boot:

======================================================
 WARNING: possible circular locking dependency detected 6.15.0-rc4-gc2b4d8dcb3d2 #3 Not tainted
 ------------------------------------------------------
 (udev-worker)/1810 is trying to acquire lock:
 0000005fb84de3a8 (&q->elevator_lock){+.+.}-{4:4}, at: elevator_change+0x54/0x130

 but task is already holding lock:
 0000005fb84dde18 (&q->q_usage_counter(io)#34){++++}-{0:0}, at: blk_mq_freeze_queue_nomemsave+0x26/0x40

 which lock already depends on the new lock.

 the existing dependency chain (in reverse order) is:

 -> #3 (&q->q_usage_counter(io)#34){++++}-{0:0}:

        __lock_acquire+0x6da/0xcc0
        lock_acquire.part.0+0x10c/0x290
        lock_acquire+0xb0/0x1a0
        blk_alloc_queue+0x306/0x340
        blk_mq_alloc_queue+0x60/0xd0
        scsi_alloc_sdev+0x27c/0x3b0
        scsi_probe_and_add_lun+0x31a/0x480
        scsi_report_lun_scan+0x382/0x430
        __scsi_scan_target+0x11a/0x240
        scsi_scan_target+0xdc/0x100
        fc_scsi_scan_rport+0xc2/0xd0
        process_one_work+0x2a6/0x5d0
        worker_thread+0x220/0x410
        kthread+0x164/0x2d0
        __ret_from_fork+0x3c/0x60
        ret_from_fork+0xa/0x38

 -> #2 (fs_reclaim){+.+.}-{0:0}:

        __lock_acquire+0x6da/0xcc0
        lock_acquire.part.0+0x10c/0x290
        lock_acquire+0xb0/0x1a0
        __fs_reclaim_acquire+0x44/0x50
        fs_reclaim_acquire+0xba/0x100
        __kmalloc_noprof+0xae/0x5e0
        pcpu_alloc_chunk+0x30/0x170
        pcpu_create_chunk+0x22/0x130
        pcpu_alloc_noprof+0x842/0x970
        do_kmem_cache_create+0x1e0/0x4b0
        __kmem_cache_create_args+0x238/0x340
        register_ftrace_graph+0x438/0x460
        trace_selftest_startup_function_graph+0x62/0x260
        run_tracer_selftest+0x116/0x1b0
        register_tracer+0x192/0x260
        do_one_initcall+0x4a/0x180
        do_initcalls+0x146/0x170
        kernel_init_freeable+0x230/0x270
        kernel_init+0x2e/0x188
        __ret_from_fork+0x3c/0x60
        ret_from_fork+0xa/0x38
Quoted 7 lines

 -> #1 (pcpu_alloc_mutex){+.+.}-{4:4}:

        __lock_acquire+0x6da/0xcc0
        lock_acquire.part.0+0x10c/0x290
        lock_acquire+0xb0/0x1a0
        __mutex_lock+0xae/0xa20
        mutex_lock_killable_nested+0x32/0x40
        pcpu_alloc_noprof+0x6ea/0x970

It is one known dependency on percpu alloc lock, and we can't cover every one in single patchset, especially this patchset is becoming bigger.

And this percpu lock splat becomes much easier to address after the elevator patchset is merged.

We can move the elevator data allocation out of queue freeze, then the dependency can be cut.

Thanks
Ming