blk-mq: untangle debugfs and sysfs

Originally, I tied debugfs registration/unregistration together with
sysfs. There's no reason to do this, and it's getting in the way of
letting schedulers define their own debugfs attributes. Instead, tie the
debugfs registration to the lifetime of the structures themselves.

The saner lifetimes mean we can also get rid of the extra mq directory
and move everything one level up. I.e., nvme0n1/mq/hctx0/tags is now
just nvme0n1/hctx0/tags.
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>

block/blk-core.c

@@ -40,7 +40,6 @@
 
 #include "blk.h"
 #include "blk-mq.h"
-#include "blk-mq-debugfs.h"
 #include "blk-mq-sched.h"
 #include "blk-wbt.h"
 
@@ -562,13 +561,9 @@ void blk_cleanup_queue(struct request_queue *q)
 	 * prevent that q->request_fn() gets invoked after draining finished.
 	 */
 	blk_freeze_queue(q);
-	if (!q->mq_ops) {
 	spin_lock_irq(lock);
+	if (!q->mq_ops)
 		__blk_drain_queue(q, true);
-	} else {
-		blk_mq_debugfs_unregister_mq(q);
-		spin_lock_irq(lock);
-	}
 	queue_flag_set(QUEUE_FLAG_DEAD, q);
 	spin_unlock_irq(lock);
 

block/blk-mq-debugfs.c

@@ -687,18 +687,45 @@ static const struct blk_mq_debugfs_attr blk_mq_debugfs_ctx_attrs[] = {
 	{},
 };
 
+static bool debugfs_create_files(struct dentry *parent, void *data,
+				 const struct blk_mq_debugfs_attr *attr)
+{
+	d_inode(parent)->i_private = data;
+
+	for (; attr->name; attr++) {
+		if (!debugfs_create_file(attr->name, attr->mode, parent,
+					 (void *)attr, &blk_mq_debugfs_fops))
+			return false;
+	}
+	return true;
+}
+
 int blk_mq_debugfs_register(struct request_queue *q)
 {
+	struct blk_mq_hw_ctx *hctx;
+	int i;
+
 	if (!blk_debugfs_root)
 		return -ENOENT;
 
 	q->debugfs_dir = debugfs_create_dir(kobject_name(q->kobj.parent),
 					    blk_debugfs_root);
 	if (!q->debugfs_dir)
+		return -ENOMEM;
+
+	if (!debugfs_create_files(q->debugfs_dir, q,
+				  blk_mq_debugfs_queue_attrs))
 		goto err;
 
-	if (blk_mq_debugfs_register_mq(q))
+	/*
+	 * blk_mq_init_hctx() 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).
+	 */
+	queue_for_each_hw_ctx(q, hctx, i) {
+		if (!hctx->debugfs_dir && blk_mq_debugfs_register_hctx(q, hctx))
 			goto err;
+	}
 
 	return 0;
 
@@ -710,32 +737,17 @@ int blk_mq_debugfs_register(struct request_queue *q)
 void blk_mq_debugfs_unregister(struct request_queue *q)
 {
 	debugfs_remove_recursive(q->debugfs_dir);
-	q->mq_debugfs_dir = NULL;
 	q->debugfs_dir = NULL;
 }
 
-static bool debugfs_create_files(struct dentry *parent, void *data,
-				 const struct blk_mq_debugfs_attr *attr)
-{
-	d_inode(parent)->i_private = data;
-
-	for (; attr->name; attr++) {
-		if (!debugfs_create_file(attr->name, attr->mode, parent,
-					 (void *)attr, &blk_mq_debugfs_fops))
-			return false;
-	}
-	return true;
-}
-
-static int blk_mq_debugfs_register_ctx(struct request_queue *q,
-				       struct blk_mq_ctx *ctx,
-				       struct dentry *hctx_dir)
+static int blk_mq_debugfs_register_ctx(struct blk_mq_hw_ctx *hctx,
+				       struct blk_mq_ctx *ctx)
 {
 	struct dentry *ctx_dir;
 	char name[20];
 
 	snprintf(name, sizeof(name), "cpu%u", ctx->cpu);
-	ctx_dir = debugfs_create_dir(name, hctx_dir);
+	ctx_dir = debugfs_create_dir(name, hctx->debugfs_dir);
 	if (!ctx_dir)
 		return -ENOMEM;
 
@@ -745,59 +757,61 @@ static int blk_mq_debugfs_register_ctx(struct request_queue *q,
 	return 0;
 }
 
-static int blk_mq_debugfs_register_hctx(struct request_queue *q,
+int blk_mq_debugfs_register_hctx(struct request_queue *q,
 				 struct blk_mq_hw_ctx *hctx)
 {
 	struct blk_mq_ctx *ctx;
-	struct dentry *hctx_dir;
 	char name[20];
 	int i;
 
+	if (!q->debugfs_dir)
+		return -ENOENT;
+
 	snprintf(name, sizeof(name), "hctx%u", hctx->queue_num);
-	hctx_dir = debugfs_create_dir(name, q->mq_debugfs_dir);
-	if (!hctx_dir)
+	hctx->debugfs_dir = debugfs_create_dir(name, q->debugfs_dir);
+	if (!hctx->debugfs_dir)
 		return -ENOMEM;
 
-	if (!debugfs_create_files(hctx_dir, hctx, blk_mq_debugfs_hctx_attrs))
-		return -ENOMEM;
+	if (!debugfs_create_files(hctx->debugfs_dir, hctx,
+				  blk_mq_debugfs_hctx_attrs))
+		goto err;
 
 	hctx_for_each_ctx(hctx, ctx, i) {
-		if (blk_mq_debugfs_register_ctx(q, ctx, hctx_dir))
-			return -ENOMEM;
+		if (blk_mq_debugfs_register_ctx(hctx, ctx))
+			goto err;
 	}
 
 	return 0;
+
+err:
+	blk_mq_debugfs_unregister_hctx(hctx);
+	return -ENOMEM;
+}
+
+void blk_mq_debugfs_unregister_hctx(struct blk_mq_hw_ctx *hctx)
+{
+	debugfs_remove_recursive(hctx->debugfs_dir);
+	hctx->debugfs_dir = NULL;
 }
 
-int blk_mq_debugfs_register_mq(struct request_queue *q)
+int blk_mq_debugfs_register_hctxs(struct request_queue *q)
 {
 	struct blk_mq_hw_ctx *hctx;
 	int i;
 
-	if (!q->debugfs_dir)
-		return -ENOENT;
-
-	q->mq_debugfs_dir = debugfs_create_dir("mq", q->debugfs_dir);
-	if (!q->mq_debugfs_dir)
-		goto err;
-
-	if (!debugfs_create_files(q->mq_debugfs_dir, q, blk_mq_debugfs_queue_attrs))
-		goto err;
-
 	queue_for_each_hw_ctx(q, hctx, i) {
 		if (blk_mq_debugfs_register_hctx(q, hctx))
-			goto err;
+			return -ENOMEM;
 	}
 
 	return 0;
-
-err:
-	blk_mq_debugfs_unregister_mq(q);
-	return -ENOMEM;
 }
 
-void blk_mq_debugfs_unregister_mq(struct request_queue *q)
+void blk_mq_debugfs_unregister_hctxs(struct request_queue *q)
 {
-	debugfs_remove_recursive(q->mq_debugfs_dir);
-	q->mq_debugfs_dir = NULL;
+	struct blk_mq_hw_ctx *hctx;
+	int i;
+
+	queue_for_each_hw_ctx(q, hctx, i)
+		blk_mq_debugfs_unregister_hctx(hctx);
 }

block/blk-mq-debugfs.h

@@ -4,8 +4,11 @@
 #ifdef CONFIG_BLK_DEBUG_FS
 int blk_mq_debugfs_register(struct request_queue *q);
 void blk_mq_debugfs_unregister(struct request_queue *q);
-int blk_mq_debugfs_register_mq(struct request_queue *q);
-void blk_mq_debugfs_unregister_mq(struct request_queue *q);
+int blk_mq_debugfs_register_hctx(struct request_queue *q,
+				 struct blk_mq_hw_ctx *hctx);
+void blk_mq_debugfs_unregister_hctx(struct blk_mq_hw_ctx *hctx);
+int blk_mq_debugfs_register_hctxs(struct request_queue *q);
+void blk_mq_debugfs_unregister_hctxs(struct request_queue *q);
 #else
 static inline int blk_mq_debugfs_register(struct request_queue *q)
 {
@@ -16,12 +19,22 @@ static inline void blk_mq_debugfs_unregister(struct request_queue *q)
 {
 }
 
-static inline int blk_mq_debugfs_register_mq(struct request_queue *q)
+static inline int blk_mq_debugfs_register_hctx(struct request_queue *q,
+					       struct blk_mq_hw_ctx *hctx)
 {
 	return 0;
 }
 
-static inline void blk_mq_debugfs_unregister_mq(struct request_queue *q)
+static inline void blk_mq_debugfs_unregister_hctx(struct blk_mq_hw_ctx *hctx)
+{
+}
+
+static inline int blk_mq_debugfs_register_hctxs(struct request_queue *q)
+{
+	return 0;
+}
+
+static inline void blk_mq_debugfs_unregister_hctxs(struct request_queue *q)
 {
 }
 #endif

block/blk-mq-sysfs.c

@@ -11,7 +11,6 @@
 
 #include <linux/blk-mq.h>
 #include "blk-mq.h"
-#include "blk-mq-debugfs.h"
 #include "blk-mq-tag.h"
 
 static void blk_mq_sysfs_release(struct kobject *kobj)
@@ -259,8 +258,6 @@ static void __blk_mq_unregister_dev(struct device *dev, struct request_queue *q)
 	queue_for_each_hw_ctx(q, hctx, i)
 		blk_mq_unregister_hctx(hctx);
 
-	blk_mq_debugfs_unregister_mq(q);
-
 	kobject_uevent(&q->mq_kobj, KOBJ_REMOVE);
 	kobject_del(&q->mq_kobj);
 	kobject_put(&dev->kobj);
@@ -319,8 +316,6 @@ int __blk_mq_register_dev(struct device *dev, struct request_queue *q)
 
 	kobject_uevent(&q->mq_kobj, KOBJ_ADD);
 
-	blk_mq_debugfs_register(q);
-
 	queue_for_each_hw_ctx(q, hctx, i) {
 		ret = blk_mq_register_hctx(hctx);
 		if (ret)
@@ -336,8 +331,6 @@ int __blk_mq_register_dev(struct device *dev, struct request_queue *q)
 	while (--i >= 0)
 		blk_mq_unregister_hctx(q->queue_hw_ctx[i]);
 
-	blk_mq_debugfs_unregister_mq(q);
-
 	kobject_uevent(&q->mq_kobj, KOBJ_REMOVE);
 	kobject_del(&q->mq_kobj);
 	kobject_put(&dev->kobj);
@@ -365,8 +358,6 @@ void blk_mq_sysfs_unregister(struct request_queue *q)
 	if (!q->mq_sysfs_init_done)
 		goto unlock;
 
-	blk_mq_debugfs_unregister_mq(q);
-
 	queue_for_each_hw_ctx(q, hctx, i)
 		blk_mq_unregister_hctx(hctx);
 
@@ -383,8 +374,6 @@ int blk_mq_sysfs_register(struct request_queue *q)
 	if (!q->mq_sysfs_init_done)
 		goto unlock;
 
-	blk_mq_debugfs_register_mq(q);
-
 	queue_for_each_hw_ctx(q, hctx, i) {
 		ret = blk_mq_register_hctx(hctx);
 		if (ret)

block/blk-mq.c

@@ -31,6 +31,7 @@
 #include <linux/blk-mq.h>
 #include "blk.h"
 #include "blk-mq.h"
+#include "blk-mq-debugfs.h"
 #include "blk-mq-tag.h"
 #include "blk-stat.h"
 #include "blk-wbt.h"
@@ -1862,6 +1863,8 @@ static void blk_mq_exit_hctx(struct request_queue *q,
 		struct blk_mq_tag_set *set,
 		struct blk_mq_hw_ctx *hctx, unsigned int hctx_idx)
 {
+	blk_mq_debugfs_unregister_hctx(hctx);
+
 	blk_mq_tag_idle(hctx);
 
 	if (set->ops->exit_request)
@@ -1948,6 +1951,8 @@ static int blk_mq_init_hctx(struct request_queue *q,
 	if (hctx->flags & BLK_MQ_F_BLOCKING)
 		init_srcu_struct(&hctx->queue_rq_srcu);
 
+	blk_mq_debugfs_register_hctx(q, hctx);
+
 	return 0;
 
  free_fq:
@@ -2385,6 +2390,7 @@ static void blk_mq_queue_reinit(struct request_queue *q,
 {
 	WARN_ON_ONCE(!atomic_read(&q->mq_freeze_depth));
 
+	blk_mq_debugfs_unregister_hctxs(q);
 	blk_mq_sysfs_unregister(q);
 
 	/*
@@ -2396,6 +2402,7 @@ static void blk_mq_queue_reinit(struct request_queue *q,
 	blk_mq_map_swqueue(q, online_mask);
 
 	blk_mq_sysfs_register(q);
+	blk_mq_debugfs_register_hctxs(q);
 }
 
 /*

block/blk-sysfs.c

@@ -890,6 +890,8 @@ int blk_register_queue(struct gendisk *disk)
 	if (q->mq_ops)
 		__blk_mq_register_dev(dev, q);
 
+	blk_mq_debugfs_register(q);
+
 	kobject_uevent(&q->kobj, KOBJ_ADD);
 
 	wbt_enable_default(q);

include/linux/blk-mq.h

@@ -57,6 +57,10 @@ struct blk_mq_hw_ctx {
 	unsigned long		poll_considered;
 	unsigned long		poll_invoked;
 	unsigned long		poll_success;
+
+#ifdef CONFIG_BLK_DEBUG_FS
+	struct dentry		*debugfs_dir;
+#endif
 };
 
 struct blk_mq_tag_set {

include/linux/blkdev.h

@@ -579,7 +579,6 @@ struct request_queue {
 
 #ifdef CONFIG_BLK_DEBUG_FS
 	struct dentry		*debugfs_dir;
-	struct dentry		*mq_debugfs_dir;
 #endif
 
 	bool			mq_sysfs_init_done;