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 Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 826C2C77B75 for ; Tue, 18 Apr 2023 12:47:34 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232033AbjDRMrd (ORCPT ); Tue, 18 Apr 2023 08:47:33 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:38388 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232002AbjDRMrc (ORCPT ); Tue, 18 Apr 2023 08:47:32 -0400 Received: from dfw.source.kernel.org (dfw.source.kernel.org [139.178.84.217]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 5E4B86A4E for ; Tue, 18 Apr 2023 05:47:19 -0700 (PDT) Received: from smtp.kernel.org (relay.kernel.org [52.25.139.140]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by dfw.source.kernel.org (Postfix) with ESMTPS id 9F511633C5 for ; Tue, 18 Apr 2023 12:47:18 +0000 (UTC) Received: by smtp.kernel.org (Postfix) with ESMTPSA id B20BEC433EF; Tue, 18 Apr 2023 12:47:17 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=linuxfoundation.org; s=korg; t=1681822038; bh=BsCfp74AGdXgZ/qhpWgg5qXAIsy0A35s6k5zs29Wwn8=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=ltVNKQqocK9/qr2dHZo4+6s+4ci6J2gYgIEWuuNPTM2arzfXTpdHxX9eYT/1Gr6uD TPrlcpz7nRz7/VVjg+HO2OSO1c43Mu/jN2KrMwgom8jIfcUsXflYYrSJI1h1WWPeZ4 zFCfa55YVZCgEAs/aPAGLYpKy7Yd4vtiozIqMpC8= From: Greg Kroah-Hartman To: stable@vger.kernel.org Cc: Greg Kroah-Hartman , patches@lists.linux.dev, Giuseppe Scrivano , Waiman Long , Tejun Heo , Sasha Levin Subject: [PATCH 6.1 133/134] cgroup/cpuset: Make cpuset_fork() handle CLONE_INTO_CGROUP properly Date: Tue, 18 Apr 2023 14:23:09 +0200 Message-Id: <20230418120317.802543750@linuxfoundation.org> X-Mailer: git-send-email 2.40.0 In-Reply-To: <20230418120313.001025904@linuxfoundation.org> References: <20230418120313.001025904@linuxfoundation.org> User-Agent: quilt/0.67 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Precedence: bulk List-ID: X-Mailing-List: stable@vger.kernel.org From: Waiman Long [ Upstream commit 42a11bf5c5436e91b040aeb04063be1710bb9f9c ] By default, the clone(2) syscall spawn a child process into the same cgroup as its parent. With the use of the CLONE_INTO_CGROUP flag introduced by commit ef2c41cf38a7 ("clone3: allow spawning processes into cgroups"), the child will be spawned into a different cgroup which is somewhat similar to writing the child's tid into "cgroup.threads". The current cpuset_fork() method does not properly handle the CLONE_INTO_CGROUP case where the cpuset of the child may be different from that of its parent. Update the cpuset_fork() method to treat the CLONE_INTO_CGROUP case similar to cpuset_attach(). Since the newly cloned task has not been running yet, its actual memory usage isn't known. So it is not necessary to make change to mm in cpuset_fork(). Fixes: ef2c41cf38a7 ("clone3: allow spawning processes into cgroups") Reported-by: Giuseppe Scrivano Signed-off-by: Waiman Long Cc: stable@vger.kernel.org # v5.7+ Signed-off-by: Tejun Heo Signed-off-by: Sasha Levin --- kernel/cgroup/cpuset.c | 62 ++++++++++++++++++++++++++++-------------- 1 file changed, 42 insertions(+), 20 deletions(-) diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index 1c549452c4abb..85f071fb1a414 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -2515,16 +2515,33 @@ static void cpuset_cancel_attach(struct cgroup_taskset *tset) } /* - * Protected by cpuset_rwsem. cpus_attach is used only by cpuset_attach() + * Protected by cpuset_rwsem. cpus_attach is used only by cpuset_attach_task() * but we can't allocate it dynamically there. Define it global and * allocate from cpuset_init(). */ static cpumask_var_t cpus_attach; +static nodemask_t cpuset_attach_nodemask_to; + +static void cpuset_attach_task(struct cpuset *cs, struct task_struct *task) +{ + percpu_rwsem_assert_held(&cpuset_rwsem); + + if (cs != &top_cpuset) + guarantee_online_cpus(task, cpus_attach); + else + cpumask_copy(cpus_attach, task_cpu_possible_mask(task)); + /* + * can_attach beforehand should guarantee that this doesn't + * fail. TODO: have a better way to handle failure here + */ + WARN_ON_ONCE(set_cpus_allowed_ptr(task, cpus_attach)); + + cpuset_change_task_nodemask(task, &cpuset_attach_nodemask_to); + cpuset_update_task_spread_flags(cs, task); +} static void cpuset_attach(struct cgroup_taskset *tset) { - /* static buf protected by cpuset_rwsem */ - static nodemask_t cpuset_attach_nodemask_to; struct task_struct *task; struct task_struct *leader; struct cgroup_subsys_state *css; @@ -2539,20 +2556,8 @@ static void cpuset_attach(struct cgroup_taskset *tset) guarantee_online_mems(cs, &cpuset_attach_nodemask_to); - cgroup_taskset_for_each(task, css, tset) { - if (cs != &top_cpuset) - guarantee_online_cpus(task, cpus_attach); - else - cpumask_copy(cpus_attach, task_cpu_possible_mask(task)); - /* - * can_attach beforehand should guarantee that this doesn't - * fail. TODO: have a better way to handle failure here - */ - WARN_ON_ONCE(set_cpus_allowed_ptr(task, cpus_attach)); - - cpuset_change_task_nodemask(task, &cpuset_attach_nodemask_to); - cpuset_update_task_spread_flags(cs, task); - } + cgroup_taskset_for_each(task, css, tset) + cpuset_attach_task(cs, task); /* * Change mm for all threadgroup leaders. This is expensive and may @@ -3240,11 +3245,28 @@ static void cpuset_bind(struct cgroup_subsys_state *root_css) */ static void cpuset_fork(struct task_struct *task) { - if (task_css_is_root(task, cpuset_cgrp_id)) + struct cpuset *cs; + bool same_cs; + + rcu_read_lock(); + cs = task_cs(task); + same_cs = (cs == task_cs(current)); + rcu_read_unlock(); + + if (same_cs) { + if (cs == &top_cpuset) + return; + + set_cpus_allowed_ptr(task, current->cpus_ptr); + task->mems_allowed = current->mems_allowed; return; + } - set_cpus_allowed_ptr(task, current->cpus_ptr); - task->mems_allowed = current->mems_allowed; + /* CLONE_INTO_CGROUP */ + percpu_down_write(&cpuset_rwsem); + guarantee_online_mems(cs, &cpuset_attach_nodemask_to); + cpuset_attach_task(cs, task); + percpu_up_write(&cpuset_rwsem); } struct cgroup_subsys cpuset_cgrp_subsys = { -- 2.39.2