linux-kernel - Re: [PATCH RFC] sched: Micro optimization in pick_next_task() and in check_preempt

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]

Message-ID: <11d755e9-e4f8-dd9e-30b0-45aebe260b2f@virtuozzo.com>
Date:   Thu, 19 Dec 2019 17:46:37 +0300
From:   Kirill Tkhai <ktkhai@...tuozzo.com>
To:     Steven Rostedt <rostedt@...dmis.org>
Cc:     Peter Zijlstra <peterz@...radead.org>, mingo@...hat.com,
        juri.lelli@...hat.com, vincent.guittot@...aro.org,
        dietmar.eggemann@....com, bsegall@...gle.com, mgorman@...e.de,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH RFC] sched: Micro optimization in pick_next_task() and in
 check_preempt_curr()

On 19.12.2019 17:43, Steven Rostedt wrote:
> On Thu, 19 Dec 2019 17:25:40 +0300
> Kirill Tkhai <ktkhai@...tuozzo.com> wrote:
> 
>> (17:19:25) nbjoerg: but it is not guarenteed behavior
>> (17:19:50) nbjoerg: if for some strange reason you really need to enforce relative orders of global objects, put them in consecutively named sections
> 
> Which appears to work. I tried this patch on top of yours:
> 
> Not sure how this does with locality though.

Hm, I'm not sure, but AFAIR some (all?) sections are aligned at 4K.
Will this bring holes (4K-sizeof(struct sched_class)) in address
space?


> -- Steve
> 
> diff --git a/include/asm-generic/vmlinux.lds.h b/include/asm-generic/vmlinux.lds.h
> index e00f41aa8ec4..ff12a422ff19 100644
> --- a/include/asm-generic/vmlinux.lds.h
> +++ b/include/asm-generic/vmlinux.lds.h
> @@ -108,6 +108,13 @@
>  #define SBSS_MAIN .sbss
>  #endif
>  
> +#define SCHED_DATA				\
> +	*(__idle_sched_class)			\
> +	*(__fair_sched_class)			\
> +	*(__rt_sched_class)			\
> +	*(__dl_sched_class)			\
> +	*(__stop_sched_class)
> +
>  /*
>   * Align to a 32 byte boundary equal to the
>   * alignment gcc 4.5 uses for a struct
> @@ -308,6 +315,7 @@
>  #define DATA_DATA							\
>  	*(.xiptext)							\
>  	*(DATA_MAIN)							\
> +	SCHED_DATA							\
>  	*(.ref.data)							\
>  	*(.data..shared_aligned) /* percpu related */			\
>  	MEM_KEEP(init.data*)						\
> diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
> index 43323f875cb9..5abdbe569f93 100644
> --- a/kernel/sched/deadline.c
> +++ b/kernel/sched/deadline.c
> @@ -2428,7 +2428,8 @@ static void prio_changed_dl(struct rq *rq, struct task_struct *p,
>  	}
>  }
>  
> -const struct sched_class dl_sched_class = {
> +const struct sched_class dl_sched_class
> +	__attribute__((section("__dl_sched_class"))) = {
>  	.next			= &rt_sched_class,
>  	.enqueue_task		= enqueue_task_dl,
>  	.dequeue_task		= dequeue_task_dl,
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index 08a233e97a01..e745fe0e0cd3 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -10745,7 +10745,8 @@ static unsigned int get_rr_interval_fair(struct rq *rq, struct task_struct *task
>  /*
>   * All the scheduling class methods:
>   */
> -const struct sched_class fair_sched_class = {
> +const struct sched_class fair_sched_class
> +	__attribute__((section("__fair_sched_class"))) = {
>  	.next			= &idle_sched_class,
>  	.enqueue_task		= enqueue_task_fair,
>  	.dequeue_task		= dequeue_task_fair,
> diff --git a/kernel/sched/idle.c b/kernel/sched/idle.c
> index ffa959e91227..700a9c826f0e 100644
> --- a/kernel/sched/idle.c
> +++ b/kernel/sched/idle.c
> @@ -454,7 +454,8 @@ static void update_curr_idle(struct rq *rq)
>  /*
>   * Simple, special scheduling class for the per-CPU idle tasks:
>   */
> -const struct sched_class idle_sched_class = {
> +const struct sched_class idle_sched_class
> +	__attribute__((section("__idle_sched_class"))) = {
>  	/* .next is NULL */
>  	/* no enqueue/yield_task for idle tasks */
>  
> diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
> index e591d40fd645..5d3f9bcddaeb 100644
> --- a/kernel/sched/rt.c
> +++ b/kernel/sched/rt.c
> @@ -2354,7 +2354,8 @@ static unsigned int get_rr_interval_rt(struct rq *rq, struct task_struct *task)
>  		return 0;
>  }
>  
> -const struct sched_class rt_sched_class = {
> +const struct sched_class rt_sched_class
> +	__attribute__((section("__rt_sched_class"))) = {
>  	.next			= &fair_sched_class,
>  	.enqueue_task		= enqueue_task_rt,
>  	.dequeue_task		= dequeue_task_rt,
> diff --git a/kernel/sched/stop_task.c b/kernel/sched/stop_task.c
> index 4c9e9975684f..03bc7530ff75 100644
> --- a/kernel/sched/stop_task.c
> +++ b/kernel/sched/stop_task.c
> @@ -115,7 +115,8 @@ static void update_curr_stop(struct rq *rq)
>  /*
>   * Simple, special scheduling class for the per-CPU stop tasks:
>   */
> -const struct sched_class stop_sched_class = {
> +const struct sched_class stop_sched_class
> +	__attribute__((section("__stop_sched_class"))) = {
>  	.next			= &dl_sched_class,
>  
>  	.enqueue_task		= enqueue_task_stop,
>