From patchwork Tue Feb 6 19:23:07 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Vincent Guittot X-Patchwork-Id: 127080 Delivered-To: patch@linaro.org Received: by 10.46.124.24 with SMTP id x24csp3208466ljc; Tue, 6 Feb 2018 11:23:31 -0800 (PST) X-Google-Smtp-Source: AH8x227eq9kQfhaxntP7tcbhfG7egMMgzTprWbfWY3E/shQ0bsga38HZUHODCu/PZ7kWwzGIJEY0 X-Received: by 2002:a17:902:1746:: with SMTP id i64-v6mr3359474pli.53.1517945011594; Tue, 06 Feb 2018 11:23:31 -0800 (PST) ARC-Seal: i=1; a=rsa-sha256; t=1517945011; cv=none; d=google.com; s=arc-20160816; b=v80qcgvVvxD8Lo3ZzaZBpkQHEYjz24U/6kvqdk/TcRD276T0nq44NMITD8c41vgpiS ey3kdmNxeVpfpI+9RNR+GeMHmF2BWSYhYVSc/4nvsji19TvnOtnQ9+HLC6R+ZvEnwFeR cojmPPu1QUFietbgtEsy93G/AjNLE2E92R9eWOqumWnu0kKMlqzG4XP/ZUif3CC7Cy2H SXN5cgpAdsPZ5ffOagq9XCDszUqspKlcGHlhoFLwD4JYxy5JqlfyQCOABaZRo/nFElma fVcrXyb8JtLN79prxZEIhFoLPEWGU1JHMEDKDcyXkMVEB/PhiONvaOm0zU+LIGuBUwUV 1WKA== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=list-id:precedence:sender:references:in-reply-to:message-id:date :subject:cc:to:from:dkim-signature:arc-authentication-results; bh=tT7Qk2gRgh/hDQ3lYMdUYy4kTKgrgAd2GsvFw/+KYZg=; b=cTCxJYbaVUmUSm54CcsexinnreCgTNkfUc+gB8cIzzOfwQAXGuaehKmCGhgDG6c3+A giGxU1fd1GL8NlCFOsaRacyPYfmCNAILWPpw8a82t1anXsD/mhepQf4OoLIkMRVH4ySd pFdVRgh4BktLeYyP9931yPn3TVnRA0bA6cIBq4NIQ1y80tGWIHZcgktLXvuz/acGDZ2D wPCOD65mIoj3F4Y/6QO632rCK6HCJA7gPWTo9blZq/AGvzKEI6eVSwV7MGspQ6Fu6HwW IjuMOk5titHJEPf+G3F2XoM4Q3VyZAfOc3g6UGDFogQMyMxgbhKba7pxLCm3bOm74lU7 KwFA== ARC-Authentication-Results: i=1; mx.google.com; dkim=pass header.i=@linaro.org header.s=google header.b=A34b3Pok; spf=pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=pass (p=NONE sp=NONE dis=NONE) header.from=linaro.org Return-Path: Received: from vger.kernel.org (vger.kernel.org. [209.132.180.67]) by mx.google.com with ESMTP id k12si900171pgs.499.2018.02.06.11.23.31; Tue, 06 Feb 2018 11:23:31 -0800 (PST) Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67; Authentication-Results: mx.google.com; dkim=pass header.i=@linaro.org header.s=google header.b=A34b3Pok; spf=pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=pass (p=NONE sp=NONE dis=NONE) header.from=linaro.org Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753277AbeBFTX3 (ORCPT + 18 others); Tue, 6 Feb 2018 14:23:29 -0500 Received: from mail-wm0-f66.google.com ([74.125.82.66]:56049 "EHLO mail-wm0-f66.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752922AbeBFTXY (ORCPT ); Tue, 6 Feb 2018 14:23:24 -0500 Received: by mail-wm0-f66.google.com with SMTP id 143so5831104wma.5 for ; Tue, 06 Feb 2018 11:23:23 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linaro.org; s=google; h=from:to:cc:subject:date:message-id:in-reply-to:references; bh=tT7Qk2gRgh/hDQ3lYMdUYy4kTKgrgAd2GsvFw/+KYZg=; b=A34b3PokbLoY4fUwmypNsBV1uimEXVoqvQ6LFW5ejabMNhjXDb3IKZERwyTzPpSvm6 3Gnhc3D/B/1t+fdgCI+VNXQiqMoptmVz0MfBLS6PSltEQSYqIMRZw6UiGF2H9Z8eLQbx b8jLzlkHZLPcWo5Gv0bOjNwBvwPA8MAVdNCbs= X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references; bh=tT7Qk2gRgh/hDQ3lYMdUYy4kTKgrgAd2GsvFw/+KYZg=; b=kMS+MadHXxT/l7300T7Mwf3vqJwlPH/xmSxCBME0unNiTIxvg92aVdx0NK3lys/FHb VALxtbL8AyOn2fazn5VGekB8pzkYE09v85LldKZPHR93NbGTg2txxN36C9ATjzdH5OMb 2tzc9/rCD/3qV6STwi8AiYJQG01KZSR0zUOSmphTbDwuydS7uv8T5j3qnDeL61m4fqxJ 9iC8Txy0QpyvrFbbZgFpnw+HP7xm+ZkeGQGlQ4djG1YqDRz6AMblpcBLhAHUOwaQkgHa pWmPB5YZF3T39W1LuNHaSetoaQ/vx4ra1XfyIhh4HO8MW7CFdLkdJZQGWZ7gpSG0R2xa zncQ== X-Gm-Message-State: APf1xPDvyPta7xnrcRyvjUfgGzMBEi6AridFkO3OlxsF1I8rczcnGEnF r51Y8/0NWXWw19VWe3D2Z4HTmA== X-Received: by 10.80.224.195 with SMTP id j3mr5066887edl.50.1517945002655; Tue, 06 Feb 2018 11:23:22 -0800 (PST) Received: from localhost.localdomain ([2a01:e0a:f:6020:29bb:43dd:7d11:e095]) by smtp.gmail.com with ESMTPSA id s5sm78155eda.60.2018.02.06.11.23.21 (version=TLS1_2 cipher=ECDHE-RSA-AES128-SHA bits=128/128); Tue, 06 Feb 2018 11:23:21 -0800 (PST) From: Vincent Guittot To: peterz@infradead.org, mingo@kernel.org, linux-kernel@vger.kernel.org, valentin.schneider@arm.com Cc: morten.rasmussen@foss.arm.com, brendan.jackman@arm.com, dietmar.eggemann@arm.com, Vincent Guittot Subject: [PATCH v2 3/3] sched: update blocked load when newly idle Date: Tue, 6 Feb 2018 20:23:07 +0100 Message-Id: <1517944987-343-4-git-send-email-vincent.guittot@linaro.org> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1517944987-343-1-git-send-email-vincent.guittot@linaro.org> References: <1517944987-343-1-git-send-email-vincent.guittot@linaro.org> Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org When NEWLY_IDLE load balance is not triggered, we might need to update the blocked load anyway. We can kick an ilb so an idle CPU will take care of updating blocked load or we can try to update them locally before entering idle. In the latter case, we reuse part of the nohz_idle_balance. Signed-off-by: Vincent Guittot --- kernel/sched/fair.c | 102 ++++++++++++++++++++++++++++++++++++++++++---------- 1 file changed, 84 insertions(+), 18 deletions(-) -- 2.7.4 diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index e203297..dcac728 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -8829,6 +8829,9 @@ update_next_balance(struct sched_domain *sd, unsigned long *next_balance) *next_balance = next; } +static bool _nohz_idle_balance(struct rq *this_rq, unsigned int flags, enum cpu_idle_type idle); +static void kick_ilb(unsigned int flags); + /* * idle_balance is called by schedule() if this_cpu is about to become * idle. Attempts to pull tasks from other CPUs. @@ -8863,12 +8866,26 @@ static int idle_balance(struct rq *this_rq, struct rq_flags *rf) if (this_rq->avg_idle < sysctl_sched_migration_cost || !this_rq->rd->overload) { + unsigned long has_blocked = READ_ONCE(nohz.has_blocked); + unsigned long next_blocked = READ_ONCE(nohz.next_blocked); + rcu_read_lock(); sd = rcu_dereference_check_sched_domain(this_rq->sd); if (sd) update_next_balance(sd, &next_balance); rcu_read_unlock(); + /* + * Update blocked idle load if it has not been done for a + * while. Try to do it locally before entering idle but kick a + * ilb if it takes too much time and/or might delay next local + * wake up + */ + if (has_blocked && time_after_eq(jiffies, next_blocked) && + (this_rq->avg_idle < sysctl_sched_migration_cost || + !_nohz_idle_balance(this_rq, NOHZ_STATS_KICK, CPU_NEWLY_IDLE))) + kick_ilb(NOHZ_STATS_KICK); + goto out; } @@ -9393,30 +9410,24 @@ static void rebalance_domains(struct rq *rq, enum cpu_idle_type idle) #ifdef CONFIG_NO_HZ_COMMON /* - * In CONFIG_NO_HZ_COMMON case, the idle balance kickee will do the - * rebalancing for all the cpus for whom scheduler ticks are stopped. + * Internal function that runs load balance for all idle cpus. The load balance + * can be a simple update of blocked load or a complete load balance with + * tasks movement depending of flags. + * For newly idle mode, we abort the loop if it takes too much time and return + * false to notify that the loop has not be completed and a ilb should be kick. */ -static bool nohz_idle_balance(struct rq *this_rq, enum cpu_idle_type idle) +static bool _nohz_idle_balance(struct rq *this_rq, unsigned int flags, enum cpu_idle_type idle) { /* Earliest time when we have to do rebalance again */ unsigned long now = jiffies; unsigned long next_balance = now + 60*HZ; - unsigned long next_stats = now + msecs_to_jiffies(LOAD_AVG_PERIOD); + bool has_blocked_load = false; int update_next_balance = 0; int this_cpu = this_rq->cpu; - unsigned int flags; int balance_cpu; + int ret = false; struct rq *rq; - - if (!(atomic_read(nohz_flags(this_cpu)) & NOHZ_KICK_MASK)) - return false; - - if (idle != CPU_IDLE) { - atomic_andnot(NOHZ_KICK_MASK, nohz_flags(this_cpu)); - return false; - } - - flags = atomic_fetch_andnot(NOHZ_KICK_MASK, nohz_flags(this_cpu)); + u64 curr_cost = 0; SCHED_WARN_ON((flags & NOHZ_KICK_MASK) == NOHZ_BALANCE_KICK); @@ -9431,6 +9442,10 @@ static bool nohz_idle_balance(struct rq *this_rq, enum cpu_idle_type idle) WRITE_ONCE(nohz.has_blocked, 0); for_each_cpu(balance_cpu, nohz.idle_cpus_mask) { + u64 t0, domain_cost; + + t0 = sched_clock_cpu(this_cpu); + if (balance_cpu == this_cpu || !idle_cpu(balance_cpu)) continue; @@ -9444,6 +9459,16 @@ static bool nohz_idle_balance(struct rq *this_rq, enum cpu_idle_type idle) goto abort; } + /* + * If the update is done while CPU becomes idle, we abort + * the update when its cost is higher than the average idle + * time in orde to not delay a possible wake up. + */ + if (idle == CPU_NEWLY_IDLE && this_rq->avg_idle < curr_cost) { + has_blocked_load = true; + goto abort; + } + rq = cpu_rq(balance_cpu); update_blocked_averages(rq->cpu); @@ -9456,10 +9481,10 @@ static bool nohz_idle_balance(struct rq *this_rq, enum cpu_idle_type idle) if (time_after_eq(jiffies, rq->next_balance)) { struct rq_flags rf; - rq_lock_irq(rq, &rf); + rq_lock_irqsave(rq, &rf); update_rq_clock(rq); cpu_load_update_idle(rq); - rq_unlock_irq(rq, &rf); + rq_unlock_irqrestore(rq, &rf); if (flags & NOHZ_BALANCE_KICK) rebalance_domains(rq, CPU_IDLE); @@ -9469,15 +9494,27 @@ static bool nohz_idle_balance(struct rq *this_rq, enum cpu_idle_type idle) next_balance = rq->next_balance; update_next_balance = 1; } + + domain_cost = sched_clock_cpu(this_cpu) - t0; + curr_cost += domain_cost; + + } + + /* Newly idle CPU doesn't need an update */ + if (idle != CPU_NEWLY_IDLE) { + update_blocked_averages(this_cpu); + has_blocked_load |= this_rq->has_blocked_load; } - update_blocked_averages(this_cpu); if (flags & NOHZ_BALANCE_KICK) rebalance_domains(this_rq, CPU_IDLE); WRITE_ONCE(nohz.next_blocked, now + msecs_to_jiffies(LOAD_AVG_PERIOD)); + /* The full idle balance loop has been done */ + ret = true; + abort: /* There is still blocked load, enable periodic update */ if (has_blocked_load) @@ -9491,6 +9528,35 @@ static bool nohz_idle_balance(struct rq *this_rq, enum cpu_idle_type idle) if (likely(update_next_balance)) nohz.next_balance = next_balance; + return ret; +} + +/* + * In CONFIG_NO_HZ_COMMON case, the idle balance kickee will do the + * rebalancing for all the cpus for whom scheduler ticks are stopped. + */ +static bool nohz_idle_balance(struct rq *this_rq, enum cpu_idle_type idle) +{ + int this_cpu = this_rq->cpu; + unsigned int flags; + + if (!(atomic_read(nohz_flags(this_cpu)) & NOHZ_KICK_MASK)) + return false; + + if (idle != CPU_IDLE) { + atomic_andnot(NOHZ_KICK_MASK, nohz_flags(this_cpu)); + return false; + } + + /* + * barrier, pairs with nohz_balance_enter_idle(), ensures ... + */ + flags = atomic_fetch_andnot(NOHZ_KICK_MASK, nohz_flags(this_cpu)); + if (!(flags & NOHZ_KICK_MASK)) + return false; + + _nohz_idle_balance(this_rq, flags, idle); + return true; } #else