From nobody Mon Apr 13 08:22:59 2026 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id B0DC4C74A5B for ; Wed, 22 Mar 2023 14:00:21 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S231193AbjCVOAU (ORCPT ); Wed, 22 Mar 2023 10:00:20 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:42184 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S231137AbjCVOAQ (ORCPT ); Wed, 22 Mar 2023 10:00:16 -0400 Received: from foss.arm.com (foss.arm.com [217.140.110.172]) by lindbergh.monkeyblade.net (Postfix) with ESMTP id BAF885FE6; Wed, 22 Mar 2023 07:00:14 -0700 (PDT) Received: from usa-sjc-imap-foss1.foss.arm.com (unknown [10.121.207.14]) by usa-sjc-mx-foss1.foss.arm.com (Postfix) with ESMTP id 8590315A1; Wed, 22 Mar 2023 07:00:58 -0700 (PDT) Received: from e125579.fritz.box (unknown [172.31.20.19]) by usa-sjc-imap-foss1.foss.arm.com (Postfix) with ESMTPA id EF77B3F71E; Wed, 22 Mar 2023 07:00:11 -0700 (PDT) From: Dietmar Eggemann To: Ingo Molnar , Peter Zijlstra , Juri Lelli , Waiman Long , Tejun Heo , Zefan Li , Johannes Weiner , Hao Luo Cc: Steven Rostedt , Vincent Guittot , Daniel Bristot de Oliveira , Luca Abeni , Tommaso Cucinotta , Qais Yousef , Wei Wang , linux-kernel@vger.kernel.org, cgroups@vger.kernel.org Subject: [RFC PATCH 2/2] cgroup/cpuset: Free DL BW in case can_attach() fails Date: Wed, 22 Mar 2023 14:59:59 +0100 Message-Id: <20230322135959.1998790-3-dietmar.eggemann@arm.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20230322135959.1998790-1-dietmar.eggemann@arm.com> References: <20230322135959.1998790-1-dietmar.eggemann@arm.com> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Type: text/plain; charset="utf-8" cpuset_can_attach() can fail. Postpone DL BW allocation until all task have been checked. DL BW is not allocated per-task but as a sum over all DL tasks migrating. If multiple controllers are attached to the cgroup next to cuset a non-cpuset can_attach() can fail. In this case free DL BW in cpuset_cancel_attach(). Finally, update cpuset DL task count (nr_deadline_tasks) only in cpuset_attach(). Suggested-by: Waiman Long Signed-off-by: Dietmar Eggemann --- include/linux/sched.h | 2 +- kernel/cgroup/cpuset.c | 55 ++++++++++++++++++++++++++++++++++++++---- kernel/sched/core.c | 17 ++----------- 3 files changed, 53 insertions(+), 21 deletions(-) diff --git a/include/linux/sched.h b/include/linux/sched.h index 658e997ba057..675ec74469d7 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1846,7 +1846,7 @@ current_restore_flags(unsigned long orig_flags, unsig= ned long flags) } =20 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const stru= ct cpumask *trial); -extern int task_can_attach(struct task_struct *p, const struct cpumask *cs= _effective_cpus); +extern int task_can_attach(struct task_struct *p); extern int dl_bw_alloc(int cpu, u64 dl_bw); extern void dl_bw_free(int cpu, u64 dl_bw); #ifdef CONFIG_SMP diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index f46192d2e97e..fdc476eefbed 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -198,6 +198,8 @@ struct cpuset { * know when to rebuild associated root domain bandwidth information. */ int nr_deadline_tasks; + int nr_migrate_dl_tasks; + u64 sum_migrate_dl_bw; =20 /* Invalid partition error code, not lock protected */ enum prs_errcode prs_err; @@ -2462,16 +2464,23 @@ static int fmeter_getrate(struct fmeter *fmp) =20 static struct cpuset *cpuset_attach_old_cs; =20 +static void reset_migrate_dl_data(struct cpuset *cs) +{ + cs->nr_migrate_dl_tasks =3D 0; + cs->sum_migrate_dl_bw =3D 0; +} + /* Called by cgroups to determine if a cpuset is usable; cpuset_mutex held= */ static int cpuset_can_attach(struct cgroup_taskset *tset) { struct cgroup_subsys_state *css; - struct cpuset *cs; + struct cpuset *cs, *oldcs; struct task_struct *task; int ret; =20 /* used later by cpuset_attach() */ cpuset_attach_old_cs =3D task_cs(cgroup_taskset_first(tset, &css)); + oldcs =3D cpuset_attach_old_cs; cs =3D css_cs(css); =20 mutex_lock(&cpuset_mutex); @@ -2489,7 +2498,7 @@ static int cpuset_can_attach(struct cgroup_taskset *t= set) goto out_unlock; =20 cgroup_taskset_for_each(task, css, tset) { - ret =3D task_can_attach(task, cs->effective_cpus); + ret =3D task_can_attach(task); if (ret) goto out_unlock; ret =3D security_task_setscheduler(task); @@ -2497,11 +2506,31 @@ static int cpuset_can_attach(struct cgroup_taskset = *tset) goto out_unlock; =20 if (dl_task(task)) { - cs->nr_deadline_tasks++; - cpuset_attach_old_cs->nr_deadline_tasks--; + cs->nr_migrate_dl_tasks++; + cs->sum_migrate_dl_bw +=3D task->dl.dl_bw; + } + } + + if (!cs->nr_migrate_dl_tasks) + goto out_succes; + + if (!cpumask_intersects(oldcs->effective_cpus, cs->effective_cpus)) { + int cpu =3D cpumask_any_and(cpu_active_mask, cs->effective_cpus); + + if (unlikely(cpu >=3D nr_cpu_ids)) { + reset_migrate_dl_data(cs); + ret =3D -EINVAL; + goto out_unlock; + } + + ret =3D dl_bw_alloc(cpu, cs->sum_migrate_dl_bw); + if (ret) { + reset_migrate_dl_data(cs); + goto out_unlock; } } =20 +out_succes: /* * Mark attach is in progress. This makes validate_change() fail * changes which zero cpus/mems_allowed. @@ -2516,11 +2545,21 @@ static int cpuset_can_attach(struct cgroup_taskset = *tset) static void cpuset_cancel_attach(struct cgroup_taskset *tset) { struct cgroup_subsys_state *css; + struct cpuset *cs; =20 cgroup_taskset_first(tset, &css); + cs =3D css_cs(css); =20 mutex_lock(&cpuset_mutex); - css_cs(css)->attach_in_progress--; + cs->attach_in_progress--; + + if (cs->nr_migrate_dl_tasks) { + int cpu =3D cpumask_any(cs->effective_cpus); + + dl_bw_free(cpu, cs->sum_migrate_dl_bw); + reset_migrate_dl_data(cs); + } + mutex_unlock(&cpuset_mutex); } =20 @@ -2615,6 +2654,12 @@ static void cpuset_attach(struct cgroup_taskset *tse= t) out: cs->old_mems_allowed =3D cpuset_attach_nodemask_to; =20 + if (cs->nr_migrate_dl_tasks) { + cs->nr_deadline_tasks +=3D cs->nr_migrate_dl_tasks; + oldcs->nr_deadline_tasks -=3D cs->nr_migrate_dl_tasks; + reset_migrate_dl_data(cs); + } + cs->attach_in_progress--; if (!cs->attach_in_progress) wake_up(&cpuset_attach_wq); diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 2f07aecb7434..4fb058b72886 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -9201,8 +9201,7 @@ int cpuset_cpumask_can_shrink(const struct cpumask *c= ur, return ret; } =20 -int task_can_attach(struct task_struct *p, - const struct cpumask *cs_effective_cpus) +int task_can_attach(struct task_struct *p) { int ret =3D 0; =20 @@ -9215,21 +9214,9 @@ int task_can_attach(struct task_struct *p, * success of set_cpus_allowed_ptr() on all attached tasks * before cpus_mask may be changed. */ - if (p->flags & PF_NO_SETAFFINITY) { + if (p->flags & PF_NO_SETAFFINITY) ret =3D -EINVAL; - goto out; - } - - if (dl_task(p) && !cpumask_intersects(task_rq(p)->rd->span, - cs_effective_cpus)) { - int cpu =3D cpumask_any_and(cpu_active_mask, cs_effective_cpus); =20 - if (unlikely(cpu >=3D nr_cpu_ids)) - return -EINVAL; - ret =3D dl_bw_alloc(cpu, p->dl.dl_bw); - } - -out: return ret; } =20 --=20 2.25.1