From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-15.2 required=3.0 tests=BAYES_00, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_CR_TRAILER,INCLUDES_PATCH, MAILING_LIST_MULTI,NICE_REPLY_A,SPF_HELO_NONE,SPF_PASS,URIBL_BLOCKED, USER_AGENT_SANE_1 autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 338A9C6369E for ; Wed, 18 Nov 2020 06:27:16 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id C64D82468F for ; Wed, 18 Nov 2020 06:27:15 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1726360AbgKRG1P (ORCPT ); Wed, 18 Nov 2020 01:27:15 -0500 Received: from szxga05-in.huawei.com ([45.249.212.191]:8106 "EHLO szxga05-in.huawei.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1725355AbgKRG1O (ORCPT ); Wed, 18 Nov 2020 01:27:14 -0500 Received: from DGGEMS412-HUB.china.huawei.com (unknown [172.30.72.58]) by szxga05-in.huawei.com (SkyGuard) with ESMTP id 4CbXs80gmZzLq99; Wed, 18 Nov 2020 14:26:52 +0800 (CST) Received: from [10.174.176.199] (10.174.176.199) by DGGEMS412-HUB.china.huawei.com (10.3.19.212) with Microsoft SMTP Server id 14.3.487.0; Wed, 18 Nov 2020 14:27:01 +0800 Subject: Re: workqueue: Only kick a worker after thawed or for an unbound workqueue To: Lai Jiangshan CC: Tejun Heo , LKML , Shiyuan Hu , Hewenliang References: From: Yunfeng Ye Message-ID: <6e174c9f-5436-7d1c-0443-3ca21ff8dad7@huawei.com> Date: Wed, 18 Nov 2020 14:26:49 +0800 User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:60.0) Gecko/20100101 Thunderbird/60.6.1 MIME-Version: 1.0 In-Reply-To: Content-Type: text/plain; charset="utf-8" Content-Language: en-US Content-Transfer-Encoding: 7bit X-Originating-IP: [10.174.176.199] X-CFilter-Loop: Reflected Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 2020/11/18 12:06, Lai Jiangshan wrote: > On Tue, Nov 17, 2020 at 3:33 PM Yunfeng Ye wrote: >> >> In realtime scenario, We do not want to have interference on the >> isolated cpu cores. but when invoking alloc_workqueue() for percpu wq >> on the housekeeping cpu, it kick a kworker on the isolated cpu. >> >> alloc_workqueue >> pwq_adjust_max_active >> wake_up_worker >> >> The comment in pwq_adjust_max_active() said: >> "Need to kick a worker after thawed or an unbound wq's >> max_active is bumped" >> >> So it is unnecessary to kick a kworker for percpu wq's when >> alloc_workqueue. this patch only kick a worker after thawed or for an >> unbound workqueue. >> >> Signed-off-by: Yunfeng Ye >> --- >> kernel/workqueue.c | 18 +++++++++++++----- >> 1 file changed, 13 insertions(+), 5 deletions(-) >> >> diff --git a/kernel/workqueue.c b/kernel/workqueue.c >> index c41c3c17b86a..80f7bbd4889f 100644 >> --- a/kernel/workqueue.c >> +++ b/kernel/workqueue.c >> @@ -3696,14 +3696,16 @@ static void pwq_unbound_release_workfn(struct work_struct *work) >> } >> >> /** >> - * pwq_adjust_max_active - update a pwq's max_active to the current setting >> + * pwq_adjust_max_active_and_kick - update a pwq's max_active to the current setting >> * @pwq: target pool_workqueue >> + * @force_kick: force to kick a worker >> * >> * If @pwq isn't freezing, set @pwq->max_active to the associated >> * workqueue's saved_max_active and activate delayed work items >> * accordingly. If @pwq is freezing, clear @pwq->max_active to zero. >> */ >> -static void pwq_adjust_max_active(struct pool_workqueue *pwq) >> +static void pwq_adjust_max_active_and_kick(struct pool_workqueue *pwq, >> + bool force_kick) >> { >> struct workqueue_struct *wq = pwq->wq; >> bool freezable = wq->flags & WQ_FREEZABLE; >> @@ -3733,9 +3735,10 @@ static void pwq_adjust_max_active(struct pool_workqueue *pwq) >> >> /* >> * Need to kick a worker after thawed or an unbound wq's >> - * max_active is bumped. It's a slow path. Do it always. >> + * max_active is bumped. > > > Hello > > Thanks for reporting the problem. > > But I don't like to add an argument. The waking up is called > always just because it was considered no harm and it is slow > path. But it can still be possible to detect if the waking up > is really needed based on the actual activation of delayed works. > > The previous lines are: > > while (!list_empty(&pwq->delayed_works) && > pwq->nr_active < pwq->max_active) > pwq_activate_first_delayed(pwq); > > And you can record the old pwq->nr_active before these lines: > > int old_nr_active = pwq->nr_active; > > while (!list_empty(&pwq->delayed_works) && > pwq->nr_active < pwq->max_active) > pwq_activate_first_delayed(pwq); > > /* please add more comments here, see 951a078a5 */ > if (old_nr_active < pwq->nr_active) { > if (!old_nr_active || (wq->flags & WQ_UNBOUND)) > wake_up_worker(pwq->pool); > } > Ok, I will send a patch v2. Thanks. > > Thanks for your work. > Lai. > >> */ >> - wake_up_worker(pwq->pool); >> + if (force_kick || (wq->flags & WQ_UNBOUND)) >> + wake_up_worker(pwq->pool); >> } else { >> pwq->max_active = 0; >> } >> @@ -3743,6 +3746,11 @@ static void pwq_adjust_max_active(struct pool_workqueue *pwq) >> raw_spin_unlock_irqrestore(&pwq->pool->lock, flags); >> } >> >> +static void pwq_adjust_max_active(struct pool_workqueue *pwq) >> +{ >> + pwq_adjust_max_active_and_kick(pwq, false); >> +} >> + >> /* initialize newly alloced @pwq which is associated with @wq and @pool */ >> static void init_pwq(struct pool_workqueue *pwq, struct workqueue_struct *wq, >> struct worker_pool *pool) >> @@ -5252,7 +5260,7 @@ void thaw_workqueues(void) >> list_for_each_entry(wq, &workqueues, list) { >> mutex_lock(&wq->mutex); >> for_each_pwq(pwq, wq) >> - pwq_adjust_max_active(pwq); >> + pwq_adjust_max_active_and_kick(pwq, true); >> mutex_unlock(&wq->mutex); >> } >> >> -- >> 2.18.4 > . >