From 2ff899e3516437354204423ef0a94994717b8e6a Mon Sep 17 00:00:00 2001 From: Juri Lelli Date: Thu, 13 Mar 2025 18:10:21 +0100 Subject: sched/deadline: Rebuild root domain accounting after every update Rebuilding of root domains accounting information (total_bw) is currently broken on some cases, e.g. suspend/resume on aarch64. Problem is that the way we keep track of domain changes and try to add bandwidth back is convoluted and fragile. Fix it by simplify things by making sure bandwidth accounting is cleared and completely restored after root domains changes (after root domains are again stable). To be sure we always call dl_rebuild_rd_accounting while holding cpuset_mutex we also add cpuset_reset_sched_domains() wrapper. Fixes: 53916d5fd3c0 ("sched/deadline: Check bandwidth overflow earlier for hotplug") Reported-by: Jon Hunter Co-developed-by: Waiman Long Signed-off-by: Waiman Long Signed-off-by: Juri Lelli Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Dietmar Eggemann Tested-by: Dietmar Eggemann Link: https://lore.kernel.org/r/Z9MRfeJKJUOyUSto@jlelli-thinkpadt14gen4.remote.csb --- kernel/cgroup/cpuset.c | 23 ++++++++++++++++------- 1 file changed, 16 insertions(+), 7 deletions(-) (limited to 'kernel/cgroup') diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index f87526edb2a4..1892dc8cd211 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -954,10 +954,12 @@ static void dl_update_tasks_root_domain(struct cpuset *cs) css_task_iter_end(&it); } -static void dl_rebuild_rd_accounting(void) +void dl_rebuild_rd_accounting(void) { struct cpuset *cs = NULL; struct cgroup_subsys_state *pos_css; + int cpu; + u64 cookie = ++dl_cookie; lockdep_assert_held(&cpuset_mutex); lockdep_assert_cpus_held(); @@ -965,11 +967,12 @@ static void dl_rebuild_rd_accounting(void) rcu_read_lock(); - /* - * Clear default root domain DL accounting, it will be computed again - * if a task belongs to it. - */ - dl_clear_root_domain(&def_root_domain); + for_each_possible_cpu(cpu) { + if (dl_bw_visited(cpu, cookie)) + continue; + + dl_clear_root_domain_cpu(cpu); + } cpuset_for_each_descendant_pre(cs, pos_css, &top_cpuset) { @@ -996,7 +999,6 @@ partition_and_rebuild_sched_domains(int ndoms_new, cpumask_var_t doms_new[], { sched_domains_mutex_lock(); partition_sched_domains_locked(ndoms_new, doms_new, dattr_new); - dl_rebuild_rd_accounting(); sched_domains_mutex_unlock(); } @@ -1083,6 +1085,13 @@ void rebuild_sched_domains(void) cpus_read_unlock(); } +void cpuset_reset_sched_domains(void) +{ + mutex_lock(&cpuset_mutex); + partition_sched_domains(1, NULL, NULL); + mutex_unlock(&cpuset_mutex); +} + /** * cpuset_update_tasks_cpumask - Update the cpumasks of tasks in the cpuset. * @cs: the cpuset in which each task's cpus_allowed mask needs to be changed -- cgit v1.2.3