cgroup: fix mount failure in a corner case

index 1c65f24..ae2b382 100644
@@ -1648,10 +1648,12 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 			 int flags, const char *unused_dev_name,
 			 void *data)
 {
+	struct cgroup_subsys *ss;
 	struct cgroup_root *root;
 	struct cgroup_sb_opts opts;
 	struct dentry *dentry;
 	int ret;
+	int i, j = -1;
 	bool new_sb;
 
 	/*

@@ -1677,6 +1679,23 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 		goto out_unlock;
 	}
 
+	/*
+	 * Destruction of cgroup root is asynchronous, so we may fail to
+	 * mount a cgroupfs if it immediately follows a umount. Let's wait
+	 * a little bit and retry.
+	 */
+	for_each_subsys(ss, i) {
+		if (!(opts.subsys_mask & (1 << i)))
+			continue;
+		if (!percpu_ref_tryget_live(&ss->root->cgrp.self.refcnt)) {
+			mutex_unlock(&cgroup_mutex);
+			msleep(10);
+			ret = restart_syscall

index f25a7c0..616c5c4 100644
@@ -210,6 +210,33 @@ void kernfs_kill_sb(struct super_block *sb)
 	kernfs_put(root_kn);
 }
 
+/**
+ * kernfs_pin_sb: try to pin the superblock associated with a kernfs_root
+ * @kernfs_root: the kernfs_root in question
+ * @ns: the namespace tag
+ *
+ * Pin the superblock so the superblock won't be destroyed in subsequent
+ * operations. Return NULL if there's no superblock associated to this
+ * kernfs_root, or -EINVAL if the superblock is being freed.
+ */
+struct super_block *kernfs_pin_sb(struct kernfs_root *root, const void *ns)
+{
+	struct kernfs_super_info *info;
+	struct super_block *sb = NULL;
+
+	mutex_lock(&kernfs_mutex);
+	list_for_each_entry(info, &root->supers, node) {
+		if (info->ns == ns) {
+			sb = info->sb;
+			if (!atomic_inc_not_zero(&info->sb->s_active))
+				sb = ERR_PTR(-EINVAL);
+			break;
+		}
+	}
+	mutex_unlock(&kernfs_mutex);
+	return sb;
+}
+
 void __init kernfs_init(void)
 {
 	kernfs_node_cache = kmem_cache_create("kernfs_node_cache",

index 589318b..9096296 100644
@@ -287,6 +287,7 @@ struct dentry *kernfs_mount_ns(struct file_system_type *fs_type, int flags,
 			       struct kernfs_root *root, bool *new_sb_created,
 			       const void *ns);
 void kernfs_kill_sb(struct super_block *sb);
+struct super_block *kernfs_pin_sb(struct kernfs_root *root, const void *ns);
 
 void kernfs_init(void);
 

index ae2b382..111b7c3 100644
@@ -1655,6 +1655,7 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 	int ret;
 	int i, j = -1;
 	bool new_sb;
+	struct super_block *sb = NULL;
 
 	/*
 	 * The first time anyone tries to mount a cgroup, enable the list

@@ -1737,14 +1738,18 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 
 		/*
 		 * A root's lifetime is governed by its root cgroup.
-		 * tryget_live failure indicate that the root is being
-		 * destroyed.  Wait for destruction to complete so that the
-		 * subsystems are free.  We can use wait_queue for the wait
-		 * but this path is super cold.  Let's just sleep for a bit
-		 * and retry.
+		 * pin_sb and tryget_live failure indicate that the root is
+		 * being destroyed.  Wait for destruction to complete so that
+		 * the subsystems are free.  We can use wait_queue for the
+		 * wait but this path is super cold.  Let's just sleep for
+		 * a bit and retry.
 		 */
-		if (!percpu_ref_tryget_live(&root->cgrp.self.refcnt)) {
+		sb = kernfs_pin_sb(root->kf_root, NULL);
+		if (IS_ERR(sb) ||
+		    !percpu_ref_tryget_live(&root->cgrp.self.refcnt)) {
 			mutex_unlock(&cgroup_mutex);
+			if (!IS_ERR_OR_NULL(sb))
+				deactivate_super(sb);
 			msleep(10);
 			ret = restart_syscall();
 			goto out_free;

@@ -1796,6 +1801,17 @@ out_free:
 	dentry = kernfs_mount(fs_type, flags, root->kf_root, &new_sb);
 	if (IS_ERR(dentry) || !new_sb)
 		cgroup_put(&root->cgrp);
+
+	if (sb) {
+		/*
+		 * On success kernfs_mount() returns with sb->s_umount held,
+		 * but kernfs_mount() also increases the superblock's refcnt,
+		 * so calling deactivate_super() to drop the refcnt we got when
+		 * looking up cgroup root won't acquire sb->s_umount again.
+		 */
+		WARN_ON(new_sb);
+		deactivate_super(sb);
+	}
 	return dentry;
 }
 

index 1c65f24..6826227 100644
@@ -1648,10 +1648,12 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 			 int flags, const char *unused_dev_name,
 			 void *data)
 {
+	struct cgroup_subsys *ss;
 	struct cgroup_root *root;
 	struct cgroup_sb_opts opts;
 	struct dentry *dentry;
 	int ret;
+	int i, j = -1;
 	bool new_sb;
 
 	/*

@@ -1677,6 +1679,25 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 		goto out_unlock;
 	}
 
+	/*
+	 * Destruction of cgroup root is asynchronous, so we may fail to
+	 * mount a cgroupfs if it immediately follows a umount. Let's wait
+	 * a little bit and retry.
+	 */
+	for_each_subsys(ss, i) {
+		if (!(opts.subsys_mask & (1 << i)) ||
+		    ss->root === &cgrp_dfl_root)
+			continue;
+
+		if (!percpu_ref_tryget_live(&ss->root->cgrp.self.refcnt)) {
+			mutex_unlock(&cgroup_mutex);
+			msleep(10);
+			ret = restart_syscall();
+			goto out_free;
+		}
+		j = i;
+	}
+
 	for_each_root(root) {
 		bool name_match = false;
 

@@ -1763,6 +1784,14 @@ out_free:
 	kfree(opts.release_agent);
 	kfree(opts.name);
 
+	for_each_subsys(ss, i) {
+		if (i > j)
+			break;
+		if (!(opts.subsys_mask & (1 << i)))
+			continue;
+		cgroup_put(&ss->root->cgrp);
+	}
+
 	if (ret)
 		return ERR_PTR(ret);
 

index 1c65f24..b94449f 100644
@@ -1648,10 +1648,12 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 			 int flags, const char *unused_dev_name,
 			 void *data)
 {
+	struct cgroup_subsys *ss;
 	struct cgroup_root *root;
 	struct cgroup_sb_opts opts;
 	struct dentry *dentry;
 	int ret;
+	int i, j = -1;
 	bool new_sb;
 
 	/*

@@ -1677,6 +1679,25 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 		goto out_unlock;
 	}
 
+	/*
+	 * Destruction of cgroup root is asynchronous, so we may fail to
+	 * mount a cgroupfs if it immediately follows a umount. Let's wait
+	 * a little bit and retry.
+	 */
+	for_each_subsys(ss, i) {
+		if (!(opts.subsys_mask & (1 << i)) ||
+		    ss->root == &cgrp_dfl_root)
+			continue;
+
+		if (!percpu_ref_tryget_live(&ss->root->cgrp.self.refcnt)) {
+			mutex_unlock(&cgroup_mutex);
+			msleep(10);
+			ret = restart_syscall();
+			goto out_free;
+		}
+		j = i;
+	}
+
 	for_each_root(root) {
 		bool name_match = false;
 

@@ -1763,6 +1784,14 @@ out_free:
 	kfree(opts.release_agent);
 	kfree(opts.name);
 
+	for_each_subsys(ss, i) {
+		if (i > j)
+			break;
+		if (!(opts.subsys_mask & (1 << i)))
+			continue;
+		cgroup_put(&ss->root->cgrp);
+	}
+
 	if (ret)
 		return ERR_PTR(ret);