Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1760582Ab3EALuu (ORCPT ); Wed, 1 May 2013 07:50:50 -0400 Received: from mail-la0-f48.google.com ([209.85.215.48]:38854 "EHLO mail-la0-f48.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751422Ab3EALuo (ORCPT ); Wed, 1 May 2013 07:50:44 -0400 MIME-Version: 1.0 In-Reply-To: <20130501110047.GD28253@dyad.programming.kicks-ass.net> References: <1367291838-5490-1-git-send-email-kosaki.motohiro@gmail.com> <1367291838-5490-10-git-send-email-kosaki.motohiro@gmail.com> <20130501110047.GD28253@dyad.programming.kicks-ass.net> From: Paul Turner Date: Wed, 1 May 2013 04:50:12 -0700 Message-ID: Subject: Re: [PATCH 09/10] sched: task_sched_runtime introduce micro optimization To: Peter Zijlstra Cc: KOSAKI Motohiro , LKML , Olivier Langlois , Martin Schwidefsky , Steven Rostedt , David Miller , Thomas Gleixner , Frederic Weisbecker , Ingo Molnar , KOSAKI Motohiro Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2205 Lines: 66 On Wed, May 1, 2013 at 4:00 AM, Peter Zijlstra wrote: > On Mon, Apr 29, 2013 at 11:17:17PM -0400, kosaki.motohiro@gmail.com wrote: >> From: KOSAKI Motohiro >> >> rq lock in task_sched_runtime() is necessary for two reasons. 1) >> accessing se.sum_exec_runtime is inatomic on 32bit and 2) >> do_task_delta_exec() require it. >> >> And then, 64bit can avoid holds rq lock when add_delta is false. >> >> Signed-off-by: KOSAKI Motohiro >> --- >> kernel/sched/core.c | 6 ++++++ >> 1 files changed, 6 insertions(+), 0 deletions(-) >> >> diff --git a/kernel/sched/core.c b/kernel/sched/core.c >> index b817e6d..24ba1c6 100644 >> --- a/kernel/sched/core.c >> +++ b/kernel/sched/core.c >> @@ -2657,6 +2657,12 @@ unsigned long long task_sched_runtime(struct task_struct *p, bool add_delta) >> struct rq *rq; >> u64 ns = 0; >> >> + /* Micro optimization. */ > > Instead of the above; how about something like: > > /* 64-bit doesn't need locks to atomically read a 64bit value */ > >> +#ifdef CONFIG_64BIT >> + if (!add_delta) >> + return p->se.sum_exec_runtime; >> +#endif Stronger: +#ifdef CONFIG_64BIT + if (!p->on_cpu) + return p->se.sum_exec_runtime; +#endif [ Or !p->on_cpu || !add_delta ]. We can take the racy read versus p->on_cpu since: If we race with it leaving cpu: we take lock, we're correct If we race with it entering cpu: unaccounted time ---> 0, this is indistinguishable from the read occurring a few cycles earlier. >> >> rq = task_rq_lock(p, &flags); >> ns = p->se.sum_exec_runtime; >> if (add_delta) >> -- >> 1.7.1 >> > -- > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > Please read the FAQ at http://www.tux.org/lkml/ -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/