[Devel] [PATCH v8 6/9] ve/cgroup: unmark ve-root cgroups at container stop

[Valeriy Vdovin]

Signed-off-by: Valeriy Vdovin <valeriy.vdovin at virtuozzo.com>
---
 include/linux/cgroup.h |  1 +
 kernel/cgroup.c        | 48 ++++++++++++++++++++++++++++++++++++++++++++++++
 kernel/ve/ve.c         |  2 ++
 3 files changed, 51 insertions(+)

diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h
index 0a42f93..1bd0fe7 100644
--- a/include/linux/cgroup.h
+++ b/include/linux/cgroup.h
@@ -636,6 +636,7 @@ int cgroup_task_count(const struct cgroup *cgrp);
 
 #ifdef CONFIG_VE
 void cgroup_mark_ve_roots(struct ve_struct *ve);
+void cgroup_unmark_ve_roots(struct ve_struct *ve);
 #endif
 
 /*
diff --git a/kernel/cgroup.c b/kernel/cgroup.c
index 42acbb4..14987d0 100644
--- a/kernel/cgroup.c
+++ b/kernel/cgroup.c
@@ -635,6 +635,35 @@ static struct css_set *find_css_set(
 }
 
 /*
+ * Walk each cgroup link of a given css_set and find a cgroup that
+ * is the child of cgroupfs_root in argument.
+ */
+static struct cgroup *css_cgroup_from_root(struct css_set *css_set,
+					    struct cgroupfs_root *root)
+{
+	struct cgroup *res = NULL;
+
+	BUG_ON(!mutex_is_locked(&cgroup_mutex));
+	read_lock(&css_set_lock);
+
+	if (css_set == &init_css_set) {
+		res = &root->top_cgroup;
+	} else {
+		struct cg_cgroup_link *link;
+		list_for_each_entry(link, &css_set->cg_links, cg_link_list) {
+			struct cgroup *c = link->cgrp;
+			if (c->root == root) {
+				res = c;
+				break;
+			}
+		}
+	}
+	read_unlock(&css_set_lock);
+	BUG_ON(!res);
+	return res;
+}
+
+/*
  * Return the cgroup for "task" from the given hierarchy. Must be
  * called with cgroup_mutex held.
  */
@@ -4235,6 +4264,25 @@ void cgroup_mark_ve_roots(struct ve_struct *ve)
 	mutex_unlock(&cgroup_mutex);
 }
 
+void cgroup_unmark_ve_roots(struct ve_struct *ve)
+{
+	struct cgroup *cgrp;
+	struct cgroupfs_root *root;
+
+	mutex_lock(&cgroup_mutex);
+	for_each_active_root(root) {
+		cgrp = css_cgroup_from_root(ve->root_css_set, root);
+		BUG_ON(!rcu_dereference_protected(cgrp->ve_owner,
+				lockdep_is_held(&cgroup_mutex)));
+		rcu_assign_pointer(cgrp->ve_owner, NULL);
+		clear_bit(CGRP_VE_ROOT, &cgrp->flags);
+	}
+	mutex_unlock(&cgroup_mutex);
+	/* ve_owner == NULL will be visible */
+	synchronize_rcu();
+	flush_workqueue(ve->wq);
+}
+
 struct cgroup *cgroup_get_ve_root(struct cgroup *cgrp)
 {
 	struct cgroup *ve_root = NULL;
diff --git a/kernel/ve/ve.c b/kernel/ve/ve.c
index 539bba1..0ad1a55 100644
--- a/kernel/ve/ve.c
+++ b/kernel/ve/ve.c
@@ -618,6 +618,8 @@ void ve_exit_ns(struct pid_namespace *pid_ns)
 	if (!ve->ve_ns || ve->ve_ns->pid_ns != pid_ns)
 		return;
 
+	cgroup_unmark_ve_roots(ve);
+
 	ve_workqueue_stop(ve);
 
 	/*
-- 
1.8.3.1