[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20201124114042.GJ3306@suse.de>
Date: Tue, 24 Nov 2020 11:40:42 +0000
From: Mel Gorman <mgorman@...e.de>
To: Yafang Shao <laoar.shao@...il.com>
Cc: mingo@...hat.com, peterz@...radead.org, juri.lelli@...hat.com,
vincent.guittot@...aro.org, dietmar.eggemann@....com,
rostedt@...dmis.org, bsegall@...gle.com, bristot@...hat.com,
linux-kernel@...r.kernel.org, linux-rt-users@...r.kernel.org
Subject: Re: [RFC PATCH v2 3/5] sched: make schedstats helper independent of
cfs_rq
On Mon, Nov 23, 2020 at 08:58:06PM +0800, Yafang Shao wrote:
> The 'cfs_rq' in these helpers
> update_stats_{wait_start, wait_end, enqueue_sleeper} is only used to get
> the rq_clock, so we can pass the rq directly. Then these helpers can be
> used by all sched class after being moved into stats.h.
>
> After that change, the size of vmlinux is increased around 824Bytes.
> w/o this patch, with this patch
> Size of vmlinux: 78443832 78444656
>
> Signed-off-by: Yafang Shao <laoar.shao@...il.com>
The inline helpers are quite large. When I was suggesting that the overhead
was minimal, what I expected what that the inline functions would be a
schedstat_enabled() followed by a real function call. It would introduce
a small additional overhead when schedstats are enabled but avoid vmlinux
growing too large
e.g.
static inline void
update_stats_wait_start(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
if (!schedstat_enabled())
return;
__update_stats_wait_start(cfs_rq, se);
}
where __update_stats_wait_start then lives in kernel/sched/stats.c
--
Mel Gorman
SUSE Labs
Powered by blists - more mailing lists