linux-kernel - Re: [PATCH] Reduce rq lock contention in load

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

Date:   Thu, 8 Dec 2022 17:07:52 +0800
From:   Abel Wu <wuyun.abel@...edance.com>
To:     chenying <chenying.kernel@...edance.com>, mingo@...hat.com,
        Peter Zijlstra <peterz@...radead.org>,
        Juri Lelli <juri.lelli@...hat.com>,
        Vincent Guittot <vincent.guittot@...aro.org>,
        Dietmar Eggemann <dietmar.eggemann@....com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Benjamin Segall <bsegall@...gle.com>
Cc:     linux-kernel <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] Reduce rq lock contention in load_balance()

Hi Ying,

On 11/24/22 5:07 PM, chenying wrote:
> ...
> diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
> index a4a20046e586..384690bda8c3 100644
> --- a/kernel/sched/sched.h
> +++ b/kernel/sched/sched.h
> @@ -954,6 +954,7 @@ struct balance_callback {
>   struct rq {
>       /* runqueue lock: */
>       raw_spinlock_t        __lock;
> +    raw_spinlock_t          lb_lock;

Do we really need a new lock for doing this? I may suggest the
following:

diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 87522c3de7b2..30d84e066a9a 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -1048,6 +1048,7 @@ struct rq {

         struct balance_callback *balance_callback;

+       unsigned char           balancing;
         unsigned char           nohz_idle_balance;
         unsigned char           idle_balance;

and skip in-balancing runqueues early when find_busiest_queue().

Thanks,
	Abel

> 
>       /*
>        * nr_running and cpu_load should be in the same cacheline because