2021-05-10 18:47:49

by Roman Gushchin

Subject: [PATCH] cgroup: inline cgroup_task_freeze()

After the introduction of the cgroup.kill there is only one call site
of cgroup_task_freeze() left: cgroup_exit(). cgroup_task_freeze() is
currently taking rcu_read_lock() to read task's cgroup flags, but
because it's always called with css_set_lock locked, the rcu protection
is excessive.

Simplify the code by inlining cgroup_task_freeze().

Signed-off-by: Roman Gushchin <[email protected]>
---
include/linux/cgroup.h | 18 ------------------
kernel/cgroup/cgroup.c | 3 ++-
2 files changed, 2 insertions(+), 19 deletions(-)

diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h
index 4f2f79de083e..a72764287cb5 100644
--- a/include/linux/cgroup.h
+++ b/include/linux/cgroup.h
@@ -906,20 +906,6 @@ void cgroup_freeze(struct cgroup *cgrp, bool freeze);
void cgroup_freezer_migrate_task(struct task_struct *task, struct cgroup *src,
struct cgroup *dst);

-static inline bool cgroup_task_freeze(struct task_struct *task)
-{
- bool ret;
-
- if (task->flags & PF_KTHREAD)
- return false;
-
- rcu_read_lock();
- ret = test_bit(CGRP_FREEZE, &task_dfl_cgroup(task)->flags);
- rcu_read_unlock();
-
- return ret;
-}
-
static inline bool cgroup_task_frozen(struct task_struct *task)
{
return task->frozen;
@@ -929,10 +915,6 @@ static inline bool cgroup_task_frozen(struct task_struct *task)

static inline void cgroup_enter_frozen(void) { }
static inline void cgroup_leave_frozen(bool always_leave) { }
-static inline bool cgroup_task_freeze(struct task_struct *task)
-{
- return false;
-}
static inline bool cgroup_task_frozen(struct task_struct *task)
{
return false;
diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
index e640fc78d731..29be28e903ab 100644
--- a/kernel/cgroup/cgroup.c
+++ b/kernel/cgroup/cgroup.c
@@ -6267,7 +6267,8 @@ void cgroup_exit(struct task_struct *tsk)
cset->nr_tasks--;

WARN_ON_ONCE(cgroup_task_frozen(tsk));
- if (unlikely(cgroup_task_freeze(tsk)))
+ if (unlikely(!(task->flags & PF_KTHREAD) &&
+ test_bit(CGRP_FREEZE, &task_dfl_cgroup(task)->flags)))
cgroup_update_frozen(task_dfl_cgroup(tsk));

spin_unlock_irq(&css_set_lock);
--
2.30.2

2021-05-10 18:57:27

by Tejun Heo

Subject: Re: [PATCH] cgroup: inline cgroup_task_freeze()

On Mon, May 10, 2021 at 11:45:56AM -0700, Roman Gushchin wrote:
> After the introduction of the cgroup.kill there is only one call site
> of cgroup_task_freeze() left: cgroup_exit(). cgroup_task_freeze() is
> currently taking rcu_read_lock() to read task's cgroup flags, but
> because it's always called with css_set_lock locked, the rcu protection
> is excessive.
>
> Simplify the code by inlining cgroup_task_freeze().
>
> Signed-off-by: Roman Gushchin <[email protected]>

Applied to cgroup/for-5.14.

Thanks.

--
tejun