[PATCH 5.10 3/4] sched/cpuacct: Optimize away RCU read lock

[ovidiu.panait@windriver.com]

From: Chengming Zhou <zhouchengming@bytedance.com>

commit dc6e0818bc9a0336d9accf3ea35d146d72aa7a18 upstream.

Since cpuacct_charge() is called from the scheduler update_curr(),
we must already have rq lock held, then the RCU read lock can
be optimized away.

And do the same thing in it's wrapper cgroup_account_cputime(),
but we can't use lockdep_assert_rq_held() there, which defined
in kernel/sched/sched.h.

Suggested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Chengming Zhou <zhouchengming@bytedance.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220220051426.5274-2-zhouchengming@bytedance.com
[OP: adjusted lockdep_assert_rq_held() -> lockdep_assert_held()]
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
---
 include/linux/cgroup.h | 2 --
 kernel/sched/cpuacct.c | 4 +---
 2 files changed, 1 insertion(+), 5 deletions(-)

diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h
index 959b370733f0..7653f5418950 100644
--- a/include/linux/cgroup.h
+++ b/include/linux/cgroup.h
@@ -779,11 +779,9 @@ static inline void cgroup_account_cputime(struct task_struct *task,
 
 	cpuacct_charge(task, delta_exec);
 
-	rcu_read_lock();
 	cgrp = task_dfl_cgroup(task);
 	if (cgroup_parent(cgrp))
 		__cgroup_account_cputime(cgrp, delta_exec);
-	rcu_read_unlock();
 }
 
 static inline void cgroup_account_cputime_field(struct task_struct *task,
diff --git a/kernel/sched/cpuacct.c b/kernel/sched/cpuacct.c
index 3c59c541dd31..8ee298321d78 100644
--- a/kernel/sched/cpuacct.c
+++ b/kernel/sched/cpuacct.c
@@ -331,12 +331,10 @@ void cpuacct_charge(struct task_struct *tsk, u64 cputime)
 	unsigned int cpu = task_cpu(tsk);
 	struct cpuacct *ca;
 
-	rcu_read_lock();
+	lockdep_assert_held(&cpu_rq(cpu)->lock);
 
 	for (ca = task_ca(tsk); ca; ca = parent_ca(ca))
 		*per_cpu_ptr(ca->cpuusage, cpu) += cputime;
-
-	rcu_read_unlock();
 }
 
 /*
-- 
2.31.1