]> git.karo-electronics.de Git - linux-beck.git/commitdiff
cgroup: Remove task_lock() from cgroup_post_fork()
authorFrederic Weisbecker <fweisbec@gmail.com>
Fri, 23 Dec 2011 03:25:23 +0000 (04:25 +0100)
committerTejun Heo <tj@kernel.org>
Tue, 27 Dec 2011 17:05:20 +0000 (09:05 -0800)
cgroup_post_fork() is protected between threadgroup_change_begin()
and threadgroup_change_end() against concurrent changes of the
child's css_set in cgroup_task_migrate(). Also the child can't
exit and call cgroup_exit() at this stage, this means it's css_set
can't be changed with init_css_set concurrently.

For these reasons, we don't need to hold task_lock() on the child
because it's css_set can only remain stable in this place.

Let's remove the lock there.

v2: Update comment to explain that we are safe against
cgroup_exit()

Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Containers <containers@lists.linux-foundation.org>
Cc: Cgroups <cgroups@vger.kernel.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Paul Menage <paul@paulmenage.org>
Cc: Mandeep Singh Baines <msb@chromium.org>
kernel/cgroup.c

index c6bd67b3fcf6e89ca8c04cb6c4939ef196252d64..548d8d4e86d03503ea335d627db55ea3277a347f 100644 (file)
@@ -4595,10 +4595,19 @@ void cgroup_post_fork(struct task_struct *child)
 {
        if (use_task_css_set_links) {
                write_lock(&css_set_lock);
-               task_lock(child);
-               if (list_empty(&child->cg_list))
+               if (list_empty(&child->cg_list)) {
+                       /*
+                        * It's safe to use child->cgroups without task_lock()
+                        * here because we are protected through
+                        * threadgroup_change_begin() against concurrent
+                        * css_set change in cgroup_task_migrate(). Also
+                        * the task can't exit at that point until
+                        * wake_up_new_task() is called, so we are protected
+                        * against cgroup_exit() setting child->cgroup to
+                        * init_css_set.
+                        */
                        list_add(&child->cg_list, &child->cgroups->tasks);
-               task_unlock(child);
+               }
                write_unlock(&css_set_lock);
        }
 }