linux-kernel - Re: [PATCH v2] sched/rt: Account execution time for cgroup and thread group if rt entity is task

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Wed, 25 Oct 2023 12:19:36 +0200
From:   Peter Zijlstra <peterz@...radead.org>
To:     Yajun Deng <yajun.deng@...ux.dev>
Cc:     mingo@...hat.com, juri.lelli@...hat.com,
        vincent.guittot@...aro.org, dietmar.eggemann@....com,
        rostedt@...dmis.org, bsegall@...gle.com, mgorman@...e.de,
        bristot@...hat.com, vschneid@...hat.com,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] sched/rt: Account execution time for cgroup and
 thread group if rt entity is task

On Mon, Oct 23, 2023 at 04:09:54PM +0800, Yajun Deng wrote:
> The rt entity can be a task group. Like the fair scheduler class, we don't
> need to account execution time for cgroup and thread group if the rt
> entity isn't a task.

Why not? I mean, this Changelog would be ever so much better if it
actually explained something instead of said: do as fair does.


>  kernel/sched/rt.c    |  4 ----
>  kernel/sched/sched.h | 13 +++++++++++--
>  2 files changed, 11 insertions(+), 6 deletions(-)
> 
> diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
> index 6aaf0a3d6081..ccded5670b61 100644
> --- a/kernel/sched/rt.c
> +++ b/kernel/sched/rt.c
> @@ -165,8 +165,6 @@ static void destroy_rt_bandwidth(struct rt_bandwidth *rt_b)
>  	hrtimer_cancel(&rt_b->rt_period_timer);
>  }
>  
> -#define rt_entity_is_task(rt_se) (!(rt_se)->my_q)
> -
>  static inline struct task_struct *rt_task_of(struct sched_rt_entity *rt_se)
>  {
>  #ifdef CONFIG_SCHED_DEBUG
> @@ -283,8 +281,6 @@ int alloc_rt_sched_group(struct task_group *tg, struct task_group *parent)
>  
>  #else /* CONFIG_RT_GROUP_SCHED */
>  
> -#define rt_entity_is_task(rt_se) (1)
> -
>  static inline struct task_struct *rt_task_of(struct sched_rt_entity *rt_se)
>  {
>  	return container_of(rt_se, struct task_struct, rt);
> diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
> index 65cad0e5729e..95d696e23a89 100644
> --- a/kernel/sched/sched.h
> +++ b/kernel/sched/sched.h
> @@ -779,6 +779,12 @@ static inline long se_runnable(struct sched_entity *se)
>  }
>  #endif
>  
> +#ifdef CONFIG_RT_GROUP_SCHED
> +#define rt_entity_is_task(rt_se) (!(rt_se)->my_q)
> +#else
> +#define rt_entity_is_task(rt_se) (1)
> +#endif
> +
>  #ifdef CONFIG_SMP
>  /*
>   * XXX we want to get rid of these helpers and use the full load resolution.
> @@ -3266,9 +3272,12 @@ static inline void update_current_exec_runtime(struct task_struct *curr,
>  						u64 now, u64 delta_exec)
>  {
>  	curr->se.sum_exec_runtime += delta_exec;
> -	account_group_exec_runtime(curr, delta_exec);
> -
>  	curr->se.exec_start = now;
> +
> +	if (curr->sched_class == &rt_sched_class && !rt_entity_is_task(&curr->rt))
> +		return;
> +
> +	account_group_exec_runtime(curr, delta_exec);
>  	cgroup_account_cputime(curr, delta_exec);
>  }


*groan*

what's wrong with something like the below?

---
diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
index fb1996a674db..36f30b236900 100644
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -1327,7 +1327,7 @@ static void update_curr_dl(struct rq *rq)
 
 	trace_sched_stat_runtime(curr, delta_exec, 0);
 
-	update_current_exec_runtime(curr, now, delta_exec);
+	update_current_exec_runtime(curr, now, delta_exec, true);
 
 	if (dl_entity_is_special(dl_se))
 		return;
diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
index 3b627ab586fb..7d6b21b80150 100644
--- a/kernel/sched/rt.c
+++ b/kernel/sched/rt.c
@@ -1014,7 +1014,7 @@ static void update_curr_rt(struct rq *rq)
 
 	trace_sched_stat_runtime(curr, delta_exec, 0);
 
-	update_current_exec_runtime(curr, now, delta_exec);
+	update_current_exec_runtime(curr, now, delta_exec, rt_entity_is_task(rt_se));
 
 	if (!rt_bandwidth_enabled())
 		return;
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 41d760df458f..ce114a935af0 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -3272,13 +3272,16 @@ extern void sched_dynamic_update(int mode);
 #endif
 
 static inline void update_current_exec_runtime(struct task_struct *curr,
-						u64 now, u64 delta_exec)
+						u64 now, u64 delta_exec,
+						bool cgroup)
 {
 	curr->se.sum_exec_runtime += delta_exec;
-	account_group_exec_runtime(curr, delta_exec);
-
 	curr->se.exec_start = now;
-	cgroup_account_cputime(curr, delta_exec);
+
+	if (cgroup) {
+		account_group_exec_runtime(curr, delta_exec);
+		cgroup_account_cputime(curr, delta_exec);
+	}
 }
 
 #ifdef CONFIG_SCHED_MM_CID
diff --git a/kernel/sched/stop_task.c b/kernel/sched/stop_task.c
index 6cf7304e6449..1bec2af7ce8d 100644
--- a/kernel/sched/stop_task.c
+++ b/kernel/sched/stop_task.c
@@ -81,7 +81,7 @@ static void put_prev_task_stop(struct rq *rq, struct task_struct *prev)
 	schedstat_set(curr->stats.exec_max,
 		      max(curr->stats.exec_max, delta_exec));
 
-	update_current_exec_runtime(curr, now, delta_exec);
+	update_current_exec_runtime(curr, now, delta_exec, true);
 }
 
 /*