From d24a3b0b6ec677a1b44d5db18864b5451fd5458c Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Tue, 7 Feb 2023 17:08:02 +0800 Subject: [PATCH] cgroup: Elide write-locking threadgroup_rwsem when updating csses on an empty subtree MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit stable inclusion from stable-v5.10.143 commit bfbacc2ef7b5b9261e795e361ee233e7d36efae6 category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I60G94 CVE: NA Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=bfbacc2ef7b5b9261e795e361ee233e7d36efae6 ------------------------------- [ Upstream commit 671c11f0619e5ccb380bcf0f062f69ba95fc974a ] cgroup_update_dfl_csses() write-lock the threadgroup_rwsem as updating the csses can trigger process migrations. However, if the subtree doesn't contain any tasks, there aren't gonna be any cgroup migrations. This condition can be trivially detected by testing whether mgctx.preloaded_src_csets is empty. Elide write-locking threadgroup_rwsem if the subtree is empty. After this optimization, the usage pattern of creating a cgroup, enabling the necessary controllers, and then seeding it with CLONE_INTO_CGROUP and then removing the cgroup after it becomes empty doesn't need to write-lock threadgroup_rwsem at all. Signed-off-by: Tejun Heo Cc: Christian Brauner Cc: Michal Koutný Signed-off-by: Sasha Levin Signed-off-by: Cai Xinchen Reviewed-by: Xiu Jianfeng Reviewed-by: Wang Weiyang Signed-off-by: Jialin Zhang (cherry picked from commit 00e6868baa998234155afe9781646201bfd87cbf) --- kernel/cgroup/cgroup.c | 16 +++++++++++++--- 1 file changed, 13 insertions(+), 3 deletions(-) diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c index a5dd82dd8b2c..d00c391ca19e 100644 --- a/kernel/cgroup/cgroup.c +++ b/kernel/cgroup/cgroup.c @@ -2954,12 +2954,11 @@ static int cgroup_update_dfl_csses(struct cgroup *cgrp) struct cgroup_subsys_state *d_css; struct cgroup *dsct; struct css_set *src_cset; + bool has_tasks; int ret; lockdep_assert_held(&cgroup_mutex); - percpu_down_write(&cgroup_threadgroup_rwsem); - /* look up all csses currently attached to @cgrp's subtree */ spin_lock_irq(&css_set_lock); cgroup_for_each_live_descendant_pre(dsct, d_css, cgrp) { @@ -2970,6 +2969,16 @@ static int cgroup_update_dfl_csses(struct cgroup *cgrp) } spin_unlock_irq(&css_set_lock); + /* + * We need to write-lock threadgroup_rwsem while migrating tasks. + * However, if there are no source csets for @cgrp, changing its + * controllers isn't gonna produce any task migrations and the + * write-locking can be skipped safely. + */ + has_tasks = !list_empty(&mgctx.preloaded_src_csets); + if (has_tasks) + percpu_down_write(&cgroup_threadgroup_rwsem); + /* NULL dst indicates self on default hierarchy */ ret = cgroup_migrate_prepare_dst(&mgctx); if (ret) @@ -2988,7 +2997,8 @@ static int cgroup_update_dfl_csses(struct cgroup *cgrp) ret = cgroup_migrate_execute(&mgctx); out_finish: cgroup_migrate_finish(&mgctx); - percpu_up_write(&cgroup_threadgroup_rwsem); + if (has_tasks) + percpu_up_write(&cgroup_threadgroup_rwsem); return ret; } -- GitLab