blk-mq: use NOIO context to prevent deadlock during debugfs creation

Creating debugfs entries can trigger fs reclaim, which can enter back
into the block layer request_queue. This can cause deadlock if the
queue is frozen.

Previously, a WARN_ON_ONCE check was used in debugfs_create_files()
to detect this condition, but it was racy since the queue can be frozen
from another context at any time.

Introduce blk_debugfs_lock()/blk_debugfs_unlock() helpers that combine
the debugfs_mutex with memalloc_noio_save()/restore() to prevent fs
reclaim from triggering block I/O. Also add blk_debugfs_lock_nomemsave()
and blk_debugfs_unlock_nomemrestore() variants for callers that don't
need NOIO protection (e.g., debugfs removal or read-only operations).

Replace all raw debugfs_mutex lock/unlock pairs with these helpers,
using the _nomemsave/_nomemrestore variants where appropriate.

Reported-by: Yi Zhang <yi.zhang@redhat.com>
Closes: https://lore.kernel.org/all/CAHj4cs9gNKEYAPagD9JADfO5UH+OiCr4P7OO2wjpfOYeM-RV=A@mail.gmail.com/
Reported-by: Shinichiro Kawasaki <shinichiro.kawasaki@wdc.com>
Closes: https://lore.kernel.org/all/aYWQR7CtYdk3K39g@shinmob/
Suggested-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Yu Kuai <yukuai@fnnas.com>
Reviewed-by: Nilay Shroff <nilay@linux.ibm.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

5 files changed, 50 insertions, 19 deletions

diff --git a/block/blk-mq-debugfs.c b/block/blk-mq-debugfs.c
index faeaa1fc86a7..28167c9baa55 100644
--- a/block/blk-mq-debugfs.c
+++ b/block/blk-mq-debugfs.c
@@ -614,11 +614,6 @@ static void debugfs_create_files(struct request_queue *q, struct dentry *parent,
 {
 	lockdep_assert_held(&q->debugfs_mutex);
 	/*
-	 * Creating new debugfs entries with queue freezed has the risk of
-	 * deadlock.
-	 */
-	WARN_ON_ONCE(q->mq_freeze_depth != 0);
-	/*
 	 * debugfs_mutex should not be nested under other locks that can be
 	 * grabbed while queue is frozen.
 	 */
@@ -693,12 +688,13 @@ void blk_mq_debugfs_unregister_hctx(struct blk_mq_hw_ctx *hctx)
 void blk_mq_debugfs_register_hctxs(struct request_queue *q)
 {
 	struct blk_mq_hw_ctx *hctx;
+	unsigned int memflags;
 	unsigned long i;
 
-	mutex_lock(&q->debugfs_mutex);
+	memflags = blk_debugfs_lock(q);
 	queue_for_each_hw_ctx(q, hctx, i)
 		blk_mq_debugfs_register_hctx(q, hctx);
-	mutex_unlock(&q->debugfs_mutex);
+	blk_debugfs_unlock(q, memflags);
 }
 
 void blk_mq_debugfs_unregister_hctxs(struct request_queue *q)
diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c
index e26898128a7e..97c3c8f45a9b 100644
--- a/block/blk-mq-sched.c
+++ b/block/blk-mq-sched.c
@@ -390,13 +390,14 @@ static void blk_mq_sched_tags_teardown(struct request_queue *q, unsigned int fla
 void blk_mq_sched_reg_debugfs(struct request_queue *q)
 {
 	struct blk_mq_hw_ctx *hctx;
+	unsigned int memflags;
 	unsigned long i;
 
-	mutex_lock(&q->debugfs_mutex);
+	memflags = blk_debugfs_lock(q);
 	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);
+	blk_debugfs_unlock(q, memflags);
 }
 
 void blk_mq_sched_unreg_debugfs(struct request_queue *q)
@@ -404,11 +405,11 @@ void blk_mq_sched_unreg_debugfs(struct request_queue *q)
 	struct blk_mq_hw_ctx *hctx;
 	unsigned long i;
 
-	mutex_lock(&q->debugfs_mutex);
+	blk_debugfs_lock_nomemsave(q);
 	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);
+	blk_debugfs_unlock_nomemrestore(q);
 }
 
 void blk_mq_free_sched_tags(struct elevator_tags *et,
diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index 003aa684e854..f3b1968c80ce 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -892,13 +892,13 @@ static void blk_debugfs_remove(struct gendisk *disk)
 {
 	struct request_queue *q = disk->queue;
 
-	mutex_lock(&q->debugfs_mutex);
+	blk_debugfs_lock_nomemsave(q);
 	blk_trace_shutdown(q);
 	debugfs_remove_recursive(q->debugfs_dir);
 	q->debugfs_dir = NULL;
 	q->sched_debugfs_dir = NULL;
 	q->rqos_debugfs_dir = NULL;
-	mutex_unlock(&q->debugfs_mutex);
+	blk_debugfs_unlock_nomemrestore(q);
 }
 
 /**
@@ -908,6 +908,7 @@ static void blk_debugfs_remove(struct gendisk *disk)
 int blk_register_queue(struct gendisk *disk)
 {
 	struct request_queue *q = disk->queue;
+	unsigned int memflags;
 	int ret;
 
 	ret = kobject_add(&disk->queue_kobj, &disk_to_dev(disk)->kobj, "queue");
@@ -921,11 +922,11 @@ int blk_register_queue(struct gendisk *disk)
 	}
 	mutex_lock(&q->sysfs_lock);
 
-	mutex_lock(&q->debugfs_mutex);
+	memflags = blk_debugfs_lock(q);
 	q->debugfs_dir = debugfs_create_dir(disk->disk_name, blk_debugfs_root);
 	if (queue_is_mq(q))
 		blk_mq_debugfs_register(q);
-	mutex_unlock(&q->debugfs_mutex);
+	blk_debugfs_unlock(q, memflags);
 
 	ret = disk_register_independent_access_ranges(disk);
 	if (ret)
diff --git a/block/blk-wbt.c b/block/blk-wbt.c
index 1415f2bf8611..6dba71e87387 100644
--- a/block/blk-wbt.c
+++ b/block/blk-wbt.c
@@ -776,6 +776,7 @@ void wbt_init_enable_default(struct gendisk *disk)
 {
 	struct request_queue *q = disk->queue;
 	struct rq_wb *rwb;
+	unsigned int memflags;
 
 	if (!__wbt_enable_default(disk))
 		return;
@@ -789,9 +790,9 @@ void wbt_init_enable_default(struct gendisk *disk)
 		return;
 	}
 
-	mutex_lock(&q->debugfs_mutex);
+	memflags = blk_debugfs_lock(q);
 	blk_mq_debugfs_register_rq_qos(q);
-	mutex_unlock(&q->debugfs_mutex);
+	blk_debugfs_unlock(q, memflags);
 }
 
 static u64 wbt_default_latency_nsec(struct request_queue *q)
@@ -1015,9 +1016,10 @@ int wbt_set_lat(struct gendisk *disk, s64 val)
 	blk_mq_unquiesce_queue(q);
 out:
 	blk_mq_unfreeze_queue(q, memflags);
-	mutex_lock(&q->debugfs_mutex);
+
+	memflags = blk_debugfs_lock(q);
 	blk_mq_debugfs_register_rq_qos(q);
-	mutex_unlock(&q->debugfs_mutex);
+	blk_debugfs_unlock(q, memflags);
 
 	return ret;
 }
diff --git a/block/blk.h b/block/blk.h
index a6b1de509733..f6053e9dd2aa 100644
--- a/block/blk.h
+++ b/block/blk.h
@@ -729,4 +729,35 @@ static inline void blk_unfreeze_release_lock(struct request_queue *q)
 }
 #endif
 
+/*
+ * debugfs directory and file creation can trigger fs reclaim, which can enter
+ * back into the block layer request_queue. This can cause deadlock if the
+ * queue is frozen. Use NOIO context together with debugfs_mutex to prevent fs
+ * reclaim from triggering block I/O.
+ */
+static inline void blk_debugfs_lock_nomemsave(struct request_queue *q)
+{
+	mutex_lock(&q->debugfs_mutex);
+}
+
+static inline void blk_debugfs_unlock_nomemrestore(struct request_queue *q)
+{
+	mutex_unlock(&q->debugfs_mutex);
+}
+
+static inline unsigned int __must_check blk_debugfs_lock(struct request_queue *q)
+{
+	unsigned int memflags = memalloc_noio_save();
+
+	blk_debugfs_lock_nomemsave(q);
+	return memflags;
+}
+
+static inline void blk_debugfs_unlock(struct request_queue *q,
+				      unsigned int memflags)
+{
+	blk_debugfs_unlock_nomemrestore(q);
+	memalloc_noio_restore(memflags);
+}
+
 #endif /* BLK_INTERNAL_H */