diff options
| author | Tejun Heo <tj@kernel.org> | 2012-10-18 20:40:30 -0400 |
|---|---|---|
| committer | Tejun Heo <tj@kernel.org> | 2012-10-19 17:09:35 -0400 |
| commit | d87838321124061f6c935069d97f37010fa417e6 (patch) | |
| tree | 825e291d0ad7bdc7e7eb68e4bb829121de2586bc /kernel | |
| parent | 9bb71308b8133d643648776243e4d5599b1c193d (diff) | |
Revert "cgroup: Remove task_lock() from cgroup_post_fork()"
This reverts commit 7e3aa30ac8c904a706518b725c451bb486daaae9.
The commit incorrectly assumed that fork path always performed
threadgroup_change_begin/end() and depended on that for
synchronization against task exit and cgroup migration paths instead
of explicitly grabbing task_lock().
threadgroup_change is not locked when forking a new process (as
opposed to a new thread in the same process) and even if it were it
wouldn't be effective as different processes use different threadgroup
locks.
Revert the incorrect optimization.
Signed-off-by: Tejun Heo <tj@kernel.org>
LKML-Reference: <20121008020000.GB2575@localhost>
Acked-by: Li Zefan <lizefan@huawei.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: stable@vger.kernel.org
Diffstat (limited to 'kernel')
| -rw-r--r-- | kernel/cgroup.c | 15 |
1 files changed, 3 insertions, 12 deletions
diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 75aec12c78a0..f24f724620dd 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c | |||
| @@ -4883,19 +4883,10 @@ void cgroup_post_fork(struct task_struct *child) | |||
| 4883 | */ | 4883 | */ |
| 4884 | if (use_task_css_set_links) { | 4884 | if (use_task_css_set_links) { |
| 4885 | write_lock(&css_set_lock); | 4885 | write_lock(&css_set_lock); |
| 4886 | if (list_empty(&child->cg_list)) { | 4886 | task_lock(child); |
| 4887 | /* | 4887 | if (list_empty(&child->cg_list)) |
| 4888 | * It's safe to use child->cgroups without task_lock() | ||
| 4889 | * here because we are protected through | ||
| 4890 | * threadgroup_change_begin() against concurrent | ||
| 4891 | * css_set change in cgroup_task_migrate(). Also | ||
| 4892 | * the task can't exit at that point until | ||
| 4893 | * wake_up_new_task() is called, so we are protected | ||
| 4894 | * against cgroup_exit() setting child->cgroup to | ||
| 4895 | * init_css_set. | ||
| 4896 | */ | ||
| 4897 | list_add(&child->cg_list, &child->cgroups->tasks); | 4888 | list_add(&child->cg_list, &child->cgroups->tasks); |
| 4898 | } | 4889 | task_unlock(child); |
| 4899 | write_unlock(&css_set_lock); | 4890 | write_unlock(&css_set_lock); |
| 4900 | } | 4891 | } |
| 4901 | } | 4892 | } |
