1 files changed, 157 insertions, 60 deletions

diff --git a/block/elevator.c b/block/elevator.c
index 2f17d66d0e61..bba10e83478a 100644
--- a/block/elevator.c
+++ b/block/elevator.c
@@ -83,8 +83,26 @@ bool elv_bio_merge_ok(struct request *rq, struct bio *bio)
 }
 EXPORT_SYMBOL(elv_bio_merge_ok);
 
-static bool elevator_match(const struct elevator_type *e, const char *name)
+static inline bool elv_support_features(unsigned int elv_features,
+					unsigned int required_features)
 {
+	return (required_features & elv_features) == required_features;
+}
+
+/**
+ * elevator_match - Test an elevator name and features
+ * @e: Scheduler to test
+ * @name: Elevator name to test
+ * @required_features: Features that the elevator must provide
+ *
+ * Return true is the elevator @e name matches @name and if @e provides all the
+ * the feratures spcified by @required_features.
+ */
+static bool elevator_match(const struct elevator_type *e, const char *name,
+			   unsigned int required_features)
+{
+	if (!elv_support_features(e->elevator_features, required_features))
+		return false;
 	if (!strcmp(e->elevator_name, name))
 		return true;
 	if (e->elevator_alias && !strcmp(e->elevator_alias, name))
@@ -93,15 +111,21 @@ static bool elevator_match(const struct elevator_type *e, const char *name)
 	return false;
 }
 
-/*
- * Return scheduler with name 'name'
+/**
+ * elevator_find - Find an elevator
+ * @name: Name of the elevator to find
+ * @required_features: Features that the elevator must provide
+ *
+ * Return the first registered scheduler with name @name and supporting the
+ * features @required_features and NULL otherwise.
  */
-static struct elevator_type *elevator_find(const char *name)
+static struct elevator_type *elevator_find(const char *name,
+					   unsigned int required_features)
 {
 	struct elevator_type *e;
 
 	list_for_each_entry(e, &elv_list, list) {
-		if (elevator_match(e, name))
+		if (elevator_match(e, name, required_features))
 			return e;
 	}
 
@@ -120,12 +144,12 @@ static struct elevator_type *elevator_get(struct request_queue *q,
 
 	spin_lock(&elv_list_lock);
 
-	e = elevator_find(name);
+	e = elevator_find(name, q->required_elevator_features);
 	if (!e && try_loading) {
 		spin_unlock(&elv_list_lock);
 		request_module("%s-iosched", name);
 		spin_lock(&elv_list_lock);
-		e = elevator_find(name);
+		e = elevator_find(name, q->required_elevator_features);
 	}
 
 	if (e && !try_module_get(e->elevator_owner))
@@ -135,20 +159,6 @@ static struct elevator_type *elevator_get(struct request_queue *q,
 	return e;
 }
 
-static char chosen_elevator[ELV_NAME_MAX];
-
-static int __init elevator_setup(char *str)
-{
-	/*
-	 * Be backwards-compatible with previous kernels, so users
-	 * won't get the wrong elevator.
-	 */
-	strncpy(chosen_elevator, str, sizeof(chosen_elevator) - 1);
-	return 1;
-}
-
-__setup("elevator=", elevator_setup);
-
 static struct kobj_type elv_ktype;
 
 struct elevator_queue *elevator_alloc(struct request_queue *q,
@@ -470,13 +480,16 @@ static struct kobj_type elv_ktype = {
 	.release	= elevator_release,
 };
 
-int elv_register_queue(struct request_queue *q)
+/*
+ * elv_register_queue is called from either blk_register_queue or
+ * elevator_switch, elevator switch is prevented from being happen
+ * in the two paths, so it is safe to not hold q->sysfs_lock.
+ */
+int elv_register_queue(struct request_queue *q, bool uevent)
 {
 	struct elevator_queue *e = q->elevator;
 	int error;
 
-	lockdep_assert_held(&q->sysfs_lock);
-
 	error = kobject_add(&e->kobj, &q->kobj, "%s", "iosched");
 	if (!error) {
 		struct elv_fs_entry *attr = e->type->elevator_attrs;
@@ -487,24 +500,34 @@ int elv_register_queue(struct request_queue *q)
 				attr++;
 			}
 		}
-		kobject_uevent(&e->kobj, KOBJ_ADD);
+		if (uevent)
+			kobject_uevent(&e->kobj, KOBJ_ADD);
+
+		mutex_lock(&q->sysfs_lock);
 		e->registered = 1;
+		mutex_unlock(&q->sysfs_lock);
 	}
 	return error;
 }
 
+/*
+ * elv_unregister_queue is called from either blk_unregister_queue or
+ * elevator_switch, elevator switch is prevented from being happen
+ * in the two paths, so it is safe to not hold q->sysfs_lock.
+ */
 void elv_unregister_queue(struct request_queue *q)
 {
-	lockdep_assert_held(&q->sysfs_lock);
-
 	if (q) {
 		struct elevator_queue *e = q->elevator;
 
 		kobject_uevent(&e->kobj, KOBJ_REMOVE);
 		kobject_del(&e->kobj);
+
+		mutex_lock(&q->sysfs_lock);
 		e->registered = 0;
 		/* Re-enable throttling in case elevator disabled it */
 		wbt_enable_default(q);
+		mutex_unlock(&q->sysfs_lock);
 	}
 }
 
@@ -526,7 +549,7 @@ int elv_register(struct elevator_type *e)
 
 	/* register, don't allow duplicate names */
 	spin_lock(&elv_list_lock);
-	if (elevator_find(e->elevator_name)) {
+	if (elevator_find(e->elevator_name, 0)) {
 		spin_unlock(&elv_list_lock);
 		kmem_cache_destroy(e->icq_cache);
 		return -EBUSY;
@@ -567,10 +590,32 @@ int elevator_switch_mq(struct request_queue *q,
 	lockdep_assert_held(&q->sysfs_lock);
 
 	if (q->elevator) {
-		if (q->elevator->registered)
+		if (q->elevator->registered) {
+			mutex_unlock(&q->sysfs_lock);
+
+			/*
+			 * Concurrent elevator switch can't happen becasue
+			 * sysfs write is always exclusively on same file.
+			 *
+			 * Also the elevator queue won't be freed after
+			 * sysfs_lock is released becasue kobject_del() in
+			 * blk_unregister_queue() waits for completion of
+			 * .store & .show on its attributes.
+			 */
 			elv_unregister_queue(q);
+
+			mutex_lock(&q->sysfs_lock);
+		}
 		ioc_clear_queue(q);
 		elevator_exit(q, q->elevator);
+
+		/*
+		 * sysfs_lock may be dropped, so re-check if queue is
+		 * unregistered. If yes, don't switch to new elevator
+		 * any more
+		 */
+		if (!blk_queue_registered(q))
+			return 0;
 	}
 
 	ret = blk_mq_init_sched(q, new_e);
@@ -578,7 +623,11 @@ int elevator_switch_mq(struct request_queue *q,
 		goto out;
 
 	if (new_e) {
-		ret = elv_register_queue(q);
+		mutex_unlock(&q->sysfs_lock);
+
+		ret = elv_register_queue(q, true);
+
+		mutex_lock(&q->sysfs_lock);
 		if (ret) {
 			elevator_exit(q, q->elevator);
 			goto out;
@@ -594,37 +643,89 @@ out:
 	return ret;
 }
 
+static inline bool elv_support_iosched(struct request_queue *q)
+{
+	if (q->tag_set && (q->tag_set->flags & BLK_MQ_F_NO_SCHED))
+		return false;
+	return true;
+}
+
 /*
- * For blk-mq devices, we default to using mq-deadline, if available, for single
- * queue devices.  If deadline isn't available OR we have multiple queues,
- * default to "none".
+ * For single queue devices, default to using mq-deadline. If we have multiple
+ * queues or mq-deadline is not available, default to "none".
  */
-int elevator_init_mq(struct request_queue *q)
+static struct elevator_type *elevator_get_default(struct request_queue *q)
+{
+	if (q->nr_hw_queues != 1)
+		return NULL;
+
+	return elevator_get(q, "mq-deadline", false);
+}
+
+/*
+ * Get the first elevator providing the features required by the request queue.
+ * Default to "none" if no matching elevator is found.
+ */
+static struct elevator_type *elevator_get_by_features(struct request_queue *q)
+{
+	struct elevator_type *e, *found = NULL;
+
+	spin_lock(&elv_list_lock);
+
+	list_for_each_entry(e, &elv_list, list) {
+		if (elv_support_features(e->elevator_features,
+					 q->required_elevator_features)) {
+			found = e;
+			break;
+		}
+	}
+
+	if (found && !try_module_get(found->elevator_owner))
+		found = NULL;
+
+	spin_unlock(&elv_list_lock);
+	return found;
+}
+
+/*
+ * For a device queue that has no required features, use the default elevator
+ * settings. Otherwise, use the first elevator available matching the required
+ * features. If no suitable elevator is find or if the chosen elevator
+ * initialization fails, fall back to the "none" elevator (no elevator).
+ */
+void elevator_init_mq(struct request_queue *q)
 {
 	struct elevator_type *e;
-	int err = 0;
+	int err;
 
-	if (q->nr_hw_queues != 1)
-		return 0;
+	if (!elv_support_iosched(q))
+		return;
+
+	WARN_ON_ONCE(test_bit(QUEUE_FLAG_REGISTERED, &q->queue_flags));
 
-	/*
-	 * q->sysfs_lock must be held to provide mutual exclusion between
-	 * elevator_switch() and here.
-	 */
-	mutex_lock(&q->sysfs_lock);
 	if (unlikely(q->elevator))
-		goto out_unlock;
+		return;
 
-	e = elevator_get(q, "mq-deadline", false);
+	if (!q->required_elevator_features)
+		e = elevator_get_default(q);
+	else
+		e = elevator_get_by_features(q);
 	if (!e)
-		goto out_unlock;
+		return;
+
+	blk_mq_freeze_queue(q);
+	blk_mq_quiesce_queue(q);
 
 	err = blk_mq_init_sched(q, e);
-	if (err)
+
+	blk_mq_unquiesce_queue(q);
+	blk_mq_unfreeze_queue(q);
+
+	if (err) {
+		pr_warn("\"%s\" elevator initialization failed, "
+			"falling back to \"none\"\n", e->elevator_name);
 		elevator_put(e);
-out_unlock:
-	mutex_unlock(&q->sysfs_lock);
-	return err;
+	}
 }
 
 
@@ -660,7 +761,7 @@ static int __elevator_change(struct request_queue *q, const char *name)
 	struct elevator_type *e;
 
 	/* Make sure queue is not in the middle of being removed */
-	if (!test_bit(QUEUE_FLAG_REGISTERED, &q->queue_flags))
+	if (!blk_queue_registered(q))
 		return -ENOENT;
 
 	/*
@@ -677,7 +778,8 @@ static int __elevator_change(struct request_queue *q, const char *name)
 	if (!e)
 		return -EINVAL;
 
-	if (q->elevator && elevator_match(q->elevator->type, elevator_name)) {
+	if (q->elevator &&
+	    elevator_match(q->elevator->type, elevator_name, 0)) {
 		elevator_put(e);
 		return 0;
 	}
@@ -685,13 +787,6 @@ static int __elevator_change(struct request_queue *q, const char *name)
 	return elevator_switch(q, e);
 }
 
-static inline bool elv_support_iosched(struct request_queue *q)
-{
-	if (q->tag_set && (q->tag_set->flags & BLK_MQ_F_NO_SCHED))
-		return false;
-	return true;
-}
-
 ssize_t elv_iosched_store(struct request_queue *q, const char *name,
 			  size_t count)
 {
@@ -724,11 +819,13 @@ ssize_t elv_iosched_show(struct request_queue *q, char *name)
 
 	spin_lock(&elv_list_lock);
 	list_for_each_entry(__e, &elv_list, list) {
-		if (elv && elevator_match(elv, __e->elevator_name)) {
+		if (elv && elevator_match(elv, __e->elevator_name, 0)) {
 			len += sprintf(name+len, "[%s] ", elv->elevator_name);
 			continue;
 		}
-		if (elv_support_iosched(q))
+		if (elv_support_iosched(q) &&
+		    elevator_match(__e, __e->elevator_name,
+				   q->required_elevator_features))
 			len += sprintf(name+len, "%s ", __e->elevator_name);
 	}
 	spin_unlock(&elv_list_lock);