cgroup: Make operations on the cgroup root_list RCU safe

BugLink: https://bugs.launchpad.net/bugs/2086242

[ Upstream commit d23b5c5 ]

At present, when we perform operations on the cgroup root_list, we must
hold the cgroup_mutex, which is a relatively heavyweight lock. In reality,
we can make operations on this list RCU-safe, eliminating the need to hold
the cgroup_mutex during traversal. Modifications to the list only occur in
the cgroup root setup and destroy paths, which should be infrequent in a
production environment. In contrast, traversal may occur frequently.
Therefore, making it RCU-safe would be beneficial.

Signed-off-by: Yafang Shao <[email protected]>
Signed-off-by: Tejun Heo <[email protected]>
Signed-off-by: Sasha Levin <[email protected]>
Signed-off-by: Koichiro Den <[email protected]>
Signed-off-by: Roxana Nicolescu <[email protected]>

Author: laoar

include/linux/cgroup-defs.h

@@ -516,6 +516,7 @@ struct cgroup_root {
 
 	/* A list running through the active hierarchies */
 	struct list_head root_list;
+	struct rcu_head rcu;
 
 	/* Hierarchy-specific flags */
 	unsigned int flags;

kernel/cgroup/cgroup-internal.h

@@ -172,7 +172,8 @@ extern struct list_head cgroup_roots;
 
 /* iterate across the hierarchies */
 #define for_each_root(root)						\
-	list_for_each_entry((root), &cgroup_roots, root_list)
+	list_for_each_entry_rcu((root), &cgroup_roots, root_list,	\
+				lockdep_is_held(&cgroup_mutex))
 
 /**
  * for_each_subsys - iterate all enabled cgroup subsystems

kernel/cgroup/cgroup.c

@@ -1332,7 +1332,7 @@ static void cgroup_exit_root_id(struct cgroup_root *root)
 
 void cgroup_free_root(struct cgroup_root *root)
 {
-	kfree(root);
+	kfree_rcu(root, rcu);
 }
 
 static void cgroup_destroy_root(struct cgroup_root *root)
@@ -1365,7 +1365,7 @@ static void cgroup_destroy_root(struct cgroup_root *root)
 	spin_unlock_irq(&css_set_lock);
 
 	if (!list_empty(&root->root_list)) {
-		list_del(&root->root_list);
+		list_del_rcu(&root->root_list);
 		cgroup_root_count--;
 	}
 
@@ -1411,7 +1411,6 @@ current_cgns_cgroup_from_root(struct cgroup_root *root)
 	}
 	rcu_read_unlock();
 
-	BUG_ON(!res);
 	return res;
 }
 
@@ -1421,7 +1420,6 @@ static struct cgroup *cset_cgroup_from_root(struct css_set *cset,
 {
 	struct cgroup *res = NULL;
 
-	lockdep_assert_held(&cgroup_mutex);
 	lockdep_assert_held(&css_set_lock);
 
 	if (cset == &init_css_set) {
@@ -1447,7 +1445,9 @@ static struct cgroup *cset_cgroup_from_root(struct css_set *cset,
 
 /*
  * Return the cgroup for "task" from the given hierarchy. Must be
- * called with cgroup_mutex and css_set_lock held.
+ * called with css_set_lock held to prevent task's groups from being modified.
+ * Must be called with either cgroup_mutex or rcu read lock to prevent the
+ * cgroup root from being destroyed.
  */
 struct cgroup *task_cgroup_from_root(struct task_struct *task,
 				     struct cgroup_root *root)
@@ -1986,7 +1986,7 @@ void init_cgroup_root(struct cgroup_fs_context *ctx)
 	struct cgroup_root *root = ctx->root;
 	struct cgroup *cgrp = &root->cgrp;
 
-	INIT_LIST_HEAD(&root->root_list);
+	INIT_LIST_HEAD_RCU(&root->root_list);
 	atomic_set(&root->nr_cgrps, 1);
 	cgrp->root = root;
 	init_cgroup_housekeeping(cgrp);
@@ -2068,7 +2068,7 @@ int cgroup_setup_root(struct cgroup_root *root, u16 ss_mask)
 	 * care of subsystems' refcounts, which are explicitly dropped in
 	 * the failure exit path.
 	 */
-	list_add(&root->root_list, &cgroup_roots);
+	list_add_rcu(&root->root_list, &cgroup_roots);
 	cgroup_root_count++;
 
 	/*