From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S965033AbeBMNhi (ORCPT ); Tue, 13 Feb 2018 08:37:38 -0500 Received: from outbound-smtp02.blacknight.com ([81.17.249.8]:43307 "EHLO outbound-smtp02.blacknight.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S964846AbeBMNhc (ORCPT ); Tue, 13 Feb 2018 08:37:32 -0500 From: Mel Gorman To: Peter Zijlstra Cc: Ingo Molnar , Mike Galbraith , Matt Fleming , Giovanni Gherdovich , LKML , Mel Gorman Subject: [PATCH 4/6] sched/fair: Do not migrate due to a sync wakeup on exit Date: Tue, 13 Feb 2018 13:37:28 +0000 Message-Id: <20180213133730.24064-5-mgorman@techsingularity.net> X-Mailer: git-send-email 2.15.1 In-Reply-To: <20180213133730.24064-1-mgorman@techsingularity.net> References: <20180213133730.24064-1-mgorman@techsingularity.net> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org From: Peter Zijlstra When a task exits, it notifies the parent that it has exited. This is a sync wakeup and the exiting task may pull the parent towards the wakers CPU. For simple workloads like using a shell, it was observed that the shell is pulled across nodes by exiting processes. This is daft as the parent may be long-lived and properly placed. This patch special cases a sync wakeup on exit to avoid pulling tasks across nodes. Testing on a range of workloads and machines showed very little differences in performance although there was a small 3% boost on some machines running a shellscript intensive workload (git regression test suite). Signed-off-by: Mel Gorman --- kernel/sched/fair.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 28c8d9c91955..e5bbcbefd01b 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -6350,7 +6350,7 @@ select_task_rq_fair(struct task_struct *p, int prev_cpu, int sd_flag, int wake_f int cpu = smp_processor_id(); int new_cpu = prev_cpu; int want_affine = 0; - int sync = wake_flags & WF_SYNC; + int sync = (wake_flags & WF_SYNC) && !(current->flags & PF_EXITING); if (sd_flag & SD_BALANCE_WAKE) { record_wakee(p); -- 2.15.1