From aa24163b2ee5c92120e32e99b5a93143a0f4258e Mon Sep 17 00:00:00 2001 From: Prateek Sood Date: Wed, 15 Nov 2017 19:50:14 +0530 Subject: [PATCH 1/7] cgroup/cpuset: remove circular dependency deadlock Remove circular dependency deadlock in a scenario where hotplug of CPU is being done while there is updation in cgroup and cpuset triggered from userspace. Process A => kthreadd => Process B => Process C => Process A Process A cpu_subsys_offline(); cpu_down(); _cpu_down(); percpu_down_write(&cpu_hotplug_lock); //held cpuhp_invoke_callback(); workqueue_offline_cpu(); queue_work_on(); // unbind_work on system_highpri_wq __queue_work(); insert_work(); wake_up_worker(); flush_work(); wait_for_completion(); worker_thread(); manage_workers(); create_worker(); kthread_create_on_node(); wake_up_process(kthreadd_task); kthreadd kthreadd(); kernel_thread(); do_fork(); copy_process(); percpu_down_read(&cgroup_threadgroup_rwsem); __rwsem_down_read_failed_common(); //waiting Process B kernfs_fop_write(); cgroup_file_write(); cgroup_procs_write(); percpu_down_write(&cgroup_threadgroup_rwsem); //held cgroup_attach_task(); cgroup_migrate(); cgroup_migrate_execute(); cpuset_can_attach(); mutex_lock(&cpuset_mutex); //waiting Process C kernfs_fop_write(); cgroup_file_write(); cpuset_write_resmask(); mutex_lock(&cpuset_mutex); //held update_cpumask(); update_cpumasks_hier(); rebuild_sched_domains_locked(); get_online_cpus(); percpu_down_read(&cpu_hotplug_lock); //waiting Eliminating deadlock by reversing the locking order for cpuset_mutex and cpu_hotplug_lock. Signed-off-by: Prateek Sood Signed-off-by: Tejun Heo --- kernel/cgroup/cpuset.c | 53 ++++++++++++++++++++++++------------------ 1 file changed, 30 insertions(+), 23 deletions(-) diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index f7efa7b4d825..cab5fd1ee767 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -812,6 +812,18 @@ done: return ndoms; } +static void cpuset_sched_change_begin(void) +{ + cpus_read_lock(); + mutex_lock(&cpuset_mutex); +} + +static void cpuset_sched_change_end(void) +{ + mutex_unlock(&cpuset_mutex); + cpus_read_unlock(); +} + /* * Rebuild scheduler domains. * @@ -821,16 +833,14 @@ done: * 'cpus' is removed, then call this routine to rebuild the * scheduler's dynamic sched domains. * - * Call with cpuset_mutex held. Takes get_online_cpus(). */ -static void rebuild_sched_domains_locked(void) +static void rebuild_sched_domains_cpuslocked(void) { struct sched_domain_attr *attr; cpumask_var_t *doms; int ndoms; lockdep_assert_held(&cpuset_mutex); - get_online_cpus(); /* * We have raced with CPU hotplug. Don't do anything to avoid @@ -838,27 +848,25 @@ static void rebuild_sched_domains_locked(void) * Anyways, hotplug work item will rebuild sched domains. */ if (!cpumask_equal(top_cpuset.effective_cpus, cpu_active_mask)) - goto out; + return; /* Generate domain masks and attrs */ ndoms = generate_sched_domains(&doms, &attr); /* Have scheduler rebuild the domains */ partition_sched_domains(ndoms, doms, attr); -out: - put_online_cpus(); } #else /* !CONFIG_SMP */ -static void rebuild_sched_domains_locked(void) +static void rebuild_sched_domains_cpuslocked(void) { } #endif /* CONFIG_SMP */ void rebuild_sched_domains(void) { - mutex_lock(&cpuset_mutex); - rebuild_sched_domains_locked(); - mutex_unlock(&cpuset_mutex); + cpuset_sched_change_begin(); + rebuild_sched_domains_cpuslocked(); + cpuset_sched_change_end(); } /** @@ -944,7 +952,7 @@ static void update_cpumasks_hier(struct cpuset *cs, struct cpumask *new_cpus) rcu_read_unlock(); if (need_rebuild_sched_domains) - rebuild_sched_domains_locked(); + rebuild_sched_domains_cpuslocked(); } /** @@ -1276,7 +1284,7 @@ static int update_relax_domain_level(struct cpuset *cs, s64 val) cs->relax_domain_level = val; if (!cpumask_empty(cs->cpus_allowed) && is_sched_load_balance(cs)) - rebuild_sched_domains_locked(); + rebuild_sched_domains_cpuslocked(); } return 0; @@ -1309,7 +1317,6 @@ static void update_tasks_flags(struct cpuset *cs) * * Call with cpuset_mutex held. */ - static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs, int turning_on) { @@ -1342,7 +1349,7 @@ static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs, spin_unlock_irq(&callback_lock); if (!cpumask_empty(trialcs->cpus_allowed) && balance_flag_changed) - rebuild_sched_domains_locked(); + rebuild_sched_domains_cpuslocked(); if (spread_flag_changed) update_tasks_flags(cs); @@ -1610,7 +1617,7 @@ static int cpuset_write_u64(struct cgroup_subsys_state *css, struct cftype *cft, cpuset_filetype_t type = cft->private; int retval = 0; - mutex_lock(&cpuset_mutex); + cpuset_sched_change_begin(); if (!is_cpuset_online(cs)) { retval = -ENODEV; goto out_unlock; @@ -1646,7 +1653,7 @@ static int cpuset_write_u64(struct cgroup_subsys_state *css, struct cftype *cft, break; } out_unlock: - mutex_unlock(&cpuset_mutex); + cpuset_sched_change_end(); return retval; } @@ -1657,7 +1664,7 @@ static int cpuset_write_s64(struct cgroup_subsys_state *css, struct cftype *cft, cpuset_filetype_t type = cft->private; int retval = -ENODEV; - mutex_lock(&cpuset_mutex); + cpuset_sched_change_begin(); if (!is_cpuset_online(cs)) goto out_unlock; @@ -1670,7 +1677,7 @@ static int cpuset_write_s64(struct cgroup_subsys_state *css, struct cftype *cft, break; } out_unlock: - mutex_unlock(&cpuset_mutex); + cpuset_sched_change_end(); return retval; } @@ -1709,7 +1716,7 @@ static ssize_t cpuset_write_resmask(struct kernfs_open_file *of, kernfs_break_active_protection(of->kn); flush_work(&cpuset_hotplug_work); - mutex_lock(&cpuset_mutex); + cpuset_sched_change_begin(); if (!is_cpuset_online(cs)) goto out_unlock; @@ -1733,7 +1740,7 @@ static ssize_t cpuset_write_resmask(struct kernfs_open_file *of, free_trial_cpuset(trialcs); out_unlock: - mutex_unlock(&cpuset_mutex); + cpuset_sched_change_end(); kernfs_unbreak_active_protection(of->kn); css_put(&cs->css); flush_workqueue(cpuset_migrate_mm_wq); @@ -2034,14 +2041,14 @@ out_unlock: /* * If the cpuset being removed has its flag 'sched_load_balance' * enabled, then simulate turning sched_load_balance off, which - * will call rebuild_sched_domains_locked(). + * will call rebuild_sched_domains_cpuslocked(). */ static void cpuset_css_offline(struct cgroup_subsys_state *css) { struct cpuset *cs = css_cs(css); - mutex_lock(&cpuset_mutex); + cpuset_sched_change_begin(); if (is_sched_load_balance(cs)) update_flag(CS_SCHED_LOAD_BALANCE, cs, 0); @@ -2049,7 +2056,7 @@ static void cpuset_css_offline(struct cgroup_subsys_state *css) cpuset_dec(); clear_bit(CS_ONLINE, &cs->flags); - mutex_unlock(&cpuset_mutex); + cpuset_sched_change_end(); } static void cpuset_css_free(struct cgroup_subsys_state *css) From 1599a185f0e6113be185b9fb809c621c73865829 Mon Sep 17 00:00:00 2001 From: Prateek Sood Date: Wed, 15 Nov 2017 19:50:15 +0530 Subject: [PATCH 2/7] cpuset: Make cpuset hotplug synchronous Convert cpuset_hotplug_workfn() into synchronous call for cpu hotplug path. For memory hotplug path it still gets queued as a work item. Since cpuset_hotplug_workfn() can be made synchronous for cpu hotplug path, it is not required to wait for cpuset hotplug while thawing processes. Signed-off-by: Prateek Sood Signed-off-by: Tejun Heo --- include/linux/cpuset.h | 6 ------ kernel/cgroup/cpuset.c | 45 +++++++++++++++++++++--------------------- kernel/power/process.c | 2 -- kernel/sched/core.c | 1 - 4 files changed, 22 insertions(+), 32 deletions(-) diff --git a/include/linux/cpuset.h b/include/linux/cpuset.h index 1b8e41597ef5..2ab910f85154 100644 --- a/include/linux/cpuset.h +++ b/include/linux/cpuset.h @@ -52,9 +52,7 @@ static inline void cpuset_dec(void) extern int cpuset_init(void); extern void cpuset_init_smp(void); -extern void cpuset_force_rebuild(void); extern void cpuset_update_active_cpus(void); -extern void cpuset_wait_for_hotplug(void); extern void cpuset_cpus_allowed(struct task_struct *p, struct cpumask *mask); extern void cpuset_cpus_allowed_fallback(struct task_struct *p); extern nodemask_t cpuset_mems_allowed(struct task_struct *p); @@ -167,15 +165,11 @@ static inline bool cpusets_enabled(void) { return false; } static inline int cpuset_init(void) { return 0; } static inline void cpuset_init_smp(void) {} -static inline void cpuset_force_rebuild(void) { } - static inline void cpuset_update_active_cpus(void) { partition_sched_domains(1, NULL, NULL); } -static inline void cpuset_wait_for_hotplug(void) { } - static inline void cpuset_cpus_allowed(struct task_struct *p, struct cpumask *mask) { diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index cab5fd1ee767..227bc25d951d 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -2277,15 +2277,8 @@ retry: mutex_unlock(&cpuset_mutex); } -static bool force_rebuild; - -void cpuset_force_rebuild(void) -{ - force_rebuild = true; -} - /** - * cpuset_hotplug_workfn - handle CPU/memory hotunplug for a cpuset + * cpuset_hotplug - handle CPU/memory hotunplug for a cpuset * * This function is called after either CPU or memory configuration has * changed and updates cpuset accordingly. The top_cpuset is always @@ -2300,7 +2293,7 @@ void cpuset_force_rebuild(void) * Note that CPU offlining during suspend is ignored. We don't modify * cpusets across suspend/resume cycles at all. */ -static void cpuset_hotplug_workfn(struct work_struct *work) +static void cpuset_hotplug(bool use_cpu_hp_lock) { static cpumask_t new_cpus; static nodemask_t new_mems; @@ -2358,25 +2351,31 @@ static void cpuset_hotplug_workfn(struct work_struct *work) } /* rebuild sched domains if cpus_allowed has changed */ - if (cpus_updated || force_rebuild) { - force_rebuild = false; - rebuild_sched_domains(); + if (cpus_updated) { + if (use_cpu_hp_lock) + rebuild_sched_domains(); + else { + /* Acquiring cpu_hotplug_lock is not required. + * When cpuset_hotplug() is called in hotplug path, + * cpu_hotplug_lock is held by the hotplug context + * which is waiting for cpuhp_thread_fun to indicate + * completion of callback. + */ + mutex_lock(&cpuset_mutex); + rebuild_sched_domains_cpuslocked(); + mutex_unlock(&cpuset_mutex); + } } } +static void cpuset_hotplug_workfn(struct work_struct *work) +{ + cpuset_hotplug(true); +} + void cpuset_update_active_cpus(void) { - /* - * We're inside cpu hotplug critical region which usually nests - * inside cgroup synchronization. Bounce actual hotplug processing - * to a work item to avoid reverse locking order. - */ - schedule_work(&cpuset_hotplug_work); -} - -void cpuset_wait_for_hotplug(void) -{ - flush_work(&cpuset_hotplug_work); + cpuset_hotplug(false); } /* diff --git a/kernel/power/process.c b/kernel/power/process.c index 7381d49a44db..c326d7235c5f 100644 --- a/kernel/power/process.c +++ b/kernel/power/process.c @@ -204,8 +204,6 @@ void thaw_processes(void) __usermodehelper_set_disable_depth(UMH_FREEZING); thaw_workqueues(); - cpuset_wait_for_hotplug(); - read_lock(&tasklist_lock); for_each_process_thread(g, p) { /* No other threads should have PF_SUSPEND_TASK set */ diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 75554f366fd3..88b3450b29ab 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -5624,7 +5624,6 @@ static void cpuset_cpu_active(void) * restore the original sched domains by considering the * cpuset configurations. */ - cpuset_force_rebuild(); } cpuset_update_active_cpus(); } From ddf7005f32212f28669032651e09bd8d2245c35d Mon Sep 17 00:00:00 2001 From: Wang Long Date: Sun, 19 Nov 2017 16:08:37 -0500 Subject: [PATCH 3/7] debug cgroup: use task_css_set instead of rcu_dereference This macro `task_css_set` verifies that the caller is inside proper critical section if the kernel set CONFIG_PROVE_RCU=y. Signed-off-by: Wang Long Signed-off-by: Tejun Heo --- kernel/cgroup/debug.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/kernel/cgroup/debug.c b/kernel/cgroup/debug.c index 5f780d8f6a9d..9caeda610249 100644 --- a/kernel/cgroup/debug.c +++ b/kernel/cgroup/debug.c @@ -50,7 +50,7 @@ static int current_css_set_read(struct seq_file *seq, void *v) spin_lock_irq(&css_set_lock); rcu_read_lock(); - cset = rcu_dereference(current->cgroups); + cset = task_css_set(current); refcnt = refcount_read(&cset->refcount); seq_printf(seq, "css_set %pK %d", cset, refcnt); if (refcnt > cset->nr_tasks) @@ -96,7 +96,7 @@ static int current_css_set_cg_links_read(struct seq_file *seq, void *v) spin_lock_irq(&css_set_lock); rcu_read_lock(); - cset = rcu_dereference(current->cgroups); + cset = task_css_set(current); list_for_each_entry(link, &cset->cgrp_links, cgrp_link) { struct cgroup *c = link->cgrp; From 52cf373c37a684f8fc279d541307fad39d206376 Mon Sep 17 00:00:00 2001 From: Lucas Stach Date: Tue, 28 Nov 2017 13:59:25 +0100 Subject: [PATCH 4/7] cgroup: properly init u64_stats Lockdep complains that the stats update is trying to register a non-static key. This is because u64_stats are using a seqlock on 32bit arches, which needs to be initialized before usage. Fixes: 041cd640b2f3 (cgroup: Implement cgroup2 basic CPU usage accounting) Signed-off-by: Lucas Stach Signed-off-by: Tejun Heo --- kernel/cgroup/stat.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/kernel/cgroup/stat.c b/kernel/cgroup/stat.c index 133b465691d6..1e111dd455c4 100644 --- a/kernel/cgroup/stat.c +++ b/kernel/cgroup/stat.c @@ -296,8 +296,12 @@ int cgroup_stat_init(struct cgroup *cgrp) } /* ->updated_children list is self terminated */ - for_each_possible_cpu(cpu) - cgroup_cpu_stat(cgrp, cpu)->updated_children = cgrp; + for_each_possible_cpu(cpu) { + struct cgroup_cpu_stat *cstat = cgroup_cpu_stat(cgrp, cpu); + + cstat->updated_children = cgrp; + u64_stats_init(&cstat->sync); + } prev_cputime_init(&cgrp->stat.prev_cputime); From 11db855c3d06e82f432cb1bafd73296586d5ceec Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Mon, 4 Dec 2017 14:41:11 -0800 Subject: [PATCH 5/7] Revert "cpuset: Make cpuset hotplug synchronous" This reverts commit 1599a185f0e6113be185b9fb809c621c73865829. This and the previous commit led to another circular locking scenario and the scenario which is fixed by this commit no longer exists after e8b3f8db7aad ("workqueue/hotplug: simplify workqueue_offline_cpu()") which removes work item flushing from hotplug path. Revert it for now. Signed-off-by: Tejun Heo --- include/linux/cpuset.h | 6 ++++++ kernel/cgroup/cpuset.c | 45 +++++++++++++++++++++--------------------- kernel/power/process.c | 2 ++ kernel/sched/core.c | 1 + 4 files changed, 32 insertions(+), 22 deletions(-) diff --git a/include/linux/cpuset.h b/include/linux/cpuset.h index 2ab910f85154..1b8e41597ef5 100644 --- a/include/linux/cpuset.h +++ b/include/linux/cpuset.h @@ -52,7 +52,9 @@ static inline void cpuset_dec(void) extern int cpuset_init(void); extern void cpuset_init_smp(void); +extern void cpuset_force_rebuild(void); extern void cpuset_update_active_cpus(void); +extern void cpuset_wait_for_hotplug(void); extern void cpuset_cpus_allowed(struct task_struct *p, struct cpumask *mask); extern void cpuset_cpus_allowed_fallback(struct task_struct *p); extern nodemask_t cpuset_mems_allowed(struct task_struct *p); @@ -165,11 +167,15 @@ static inline bool cpusets_enabled(void) { return false; } static inline int cpuset_init(void) { return 0; } static inline void cpuset_init_smp(void) {} +static inline void cpuset_force_rebuild(void) { } + static inline void cpuset_update_active_cpus(void) { partition_sched_domains(1, NULL, NULL); } +static inline void cpuset_wait_for_hotplug(void) { } + static inline void cpuset_cpus_allowed(struct task_struct *p, struct cpumask *mask) { diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index 227bc25d951d..cab5fd1ee767 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -2277,8 +2277,15 @@ retry: mutex_unlock(&cpuset_mutex); } +static bool force_rebuild; + +void cpuset_force_rebuild(void) +{ + force_rebuild = true; +} + /** - * cpuset_hotplug - handle CPU/memory hotunplug for a cpuset + * cpuset_hotplug_workfn - handle CPU/memory hotunplug for a cpuset * * This function is called after either CPU or memory configuration has * changed and updates cpuset accordingly. The top_cpuset is always @@ -2293,7 +2300,7 @@ retry: * Note that CPU offlining during suspend is ignored. We don't modify * cpusets across suspend/resume cycles at all. */ -static void cpuset_hotplug(bool use_cpu_hp_lock) +static void cpuset_hotplug_workfn(struct work_struct *work) { static cpumask_t new_cpus; static nodemask_t new_mems; @@ -2351,31 +2358,25 @@ static void cpuset_hotplug(bool use_cpu_hp_lock) } /* rebuild sched domains if cpus_allowed has changed */ - if (cpus_updated) { - if (use_cpu_hp_lock) - rebuild_sched_domains(); - else { - /* Acquiring cpu_hotplug_lock is not required. - * When cpuset_hotplug() is called in hotplug path, - * cpu_hotplug_lock is held by the hotplug context - * which is waiting for cpuhp_thread_fun to indicate - * completion of callback. - */ - mutex_lock(&cpuset_mutex); - rebuild_sched_domains_cpuslocked(); - mutex_unlock(&cpuset_mutex); - } + if (cpus_updated || force_rebuild) { + force_rebuild = false; + rebuild_sched_domains(); } } -static void cpuset_hotplug_workfn(struct work_struct *work) -{ - cpuset_hotplug(true); -} - void cpuset_update_active_cpus(void) { - cpuset_hotplug(false); + /* + * We're inside cpu hotplug critical region which usually nests + * inside cgroup synchronization. Bounce actual hotplug processing + * to a work item to avoid reverse locking order. + */ + schedule_work(&cpuset_hotplug_work); +} + +void cpuset_wait_for_hotplug(void) +{ + flush_work(&cpuset_hotplug_work); } /* diff --git a/kernel/power/process.c b/kernel/power/process.c index c326d7235c5f..7381d49a44db 100644 --- a/kernel/power/process.c +++ b/kernel/power/process.c @@ -204,6 +204,8 @@ void thaw_processes(void) __usermodehelper_set_disable_depth(UMH_FREEZING); thaw_workqueues(); + cpuset_wait_for_hotplug(); + read_lock(&tasklist_lock); for_each_process_thread(g, p) { /* No other threads should have PF_SUSPEND_TASK set */ diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 88b3450b29ab..75554f366fd3 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -5624,6 +5624,7 @@ static void cpuset_cpu_active(void) * restore the original sched domains by considering the * cpuset configurations. */ + cpuset_force_rebuild(); } cpuset_update_active_cpus(); } From bdfbbda90aeb75ce0951413fd7f495d4d377bd5e Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Mon, 4 Dec 2017 14:55:59 -0800 Subject: [PATCH 6/7] Revert "cgroup/cpuset: remove circular dependency deadlock" This reverts commit aa24163b2ee5c92120e32e99b5a93143a0f4258e. This and the following commit led to another circular locking scenario and the scenario which is fixed by this commit no longer exists after e8b3f8db7aad ("workqueue/hotplug: simplify workqueue_offline_cpu()") which removes work item flushing from hotplug path. Revert it for now. Signed-off-by: Tejun Heo --- kernel/cgroup/cpuset.c | 53 ++++++++++++++++++------------------------ 1 file changed, 23 insertions(+), 30 deletions(-) diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index cab5fd1ee767..f7efa7b4d825 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -812,18 +812,6 @@ done: return ndoms; } -static void cpuset_sched_change_begin(void) -{ - cpus_read_lock(); - mutex_lock(&cpuset_mutex); -} - -static void cpuset_sched_change_end(void) -{ - mutex_unlock(&cpuset_mutex); - cpus_read_unlock(); -} - /* * Rebuild scheduler domains. * @@ -833,14 +821,16 @@ static void cpuset_sched_change_end(void) * 'cpus' is removed, then call this routine to rebuild the * scheduler's dynamic sched domains. * + * Call with cpuset_mutex held. Takes get_online_cpus(). */ -static void rebuild_sched_domains_cpuslocked(void) +static void rebuild_sched_domains_locked(void) { struct sched_domain_attr *attr; cpumask_var_t *doms; int ndoms; lockdep_assert_held(&cpuset_mutex); + get_online_cpus(); /* * We have raced with CPU hotplug. Don't do anything to avoid @@ -848,25 +838,27 @@ static void rebuild_sched_domains_cpuslocked(void) * Anyways, hotplug work item will rebuild sched domains. */ if (!cpumask_equal(top_cpuset.effective_cpus, cpu_active_mask)) - return; + goto out; /* Generate domain masks and attrs */ ndoms = generate_sched_domains(&doms, &attr); /* Have scheduler rebuild the domains */ partition_sched_domains(ndoms, doms, attr); +out: + put_online_cpus(); } #else /* !CONFIG_SMP */ -static void rebuild_sched_domains_cpuslocked(void) +static void rebuild_sched_domains_locked(void) { } #endif /* CONFIG_SMP */ void rebuild_sched_domains(void) { - cpuset_sched_change_begin(); - rebuild_sched_domains_cpuslocked(); - cpuset_sched_change_end(); + mutex_lock(&cpuset_mutex); + rebuild_sched_domains_locked(); + mutex_unlock(&cpuset_mutex); } /** @@ -952,7 +944,7 @@ static void update_cpumasks_hier(struct cpuset *cs, struct cpumask *new_cpus) rcu_read_unlock(); if (need_rebuild_sched_domains) - rebuild_sched_domains_cpuslocked(); + rebuild_sched_domains_locked(); } /** @@ -1284,7 +1276,7 @@ static int update_relax_domain_level(struct cpuset *cs, s64 val) cs->relax_domain_level = val; if (!cpumask_empty(cs->cpus_allowed) && is_sched_load_balance(cs)) - rebuild_sched_domains_cpuslocked(); + rebuild_sched_domains_locked(); } return 0; @@ -1317,6 +1309,7 @@ static void update_tasks_flags(struct cpuset *cs) * * Call with cpuset_mutex held. */ + static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs, int turning_on) { @@ -1349,7 +1342,7 @@ static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs, spin_unlock_irq(&callback_lock); if (!cpumask_empty(trialcs->cpus_allowed) && balance_flag_changed) - rebuild_sched_domains_cpuslocked(); + rebuild_sched_domains_locked(); if (spread_flag_changed) update_tasks_flags(cs); @@ -1617,7 +1610,7 @@ static int cpuset_write_u64(struct cgroup_subsys_state *css, struct cftype *cft, cpuset_filetype_t type = cft->private; int retval = 0; - cpuset_sched_change_begin(); + mutex_lock(&cpuset_mutex); if (!is_cpuset_online(cs)) { retval = -ENODEV; goto out_unlock; @@ -1653,7 +1646,7 @@ static int cpuset_write_u64(struct cgroup_subsys_state *css, struct cftype *cft, break; } out_unlock: - cpuset_sched_change_end(); + mutex_unlock(&cpuset_mutex); return retval; } @@ -1664,7 +1657,7 @@ static int cpuset_write_s64(struct cgroup_subsys_state *css, struct cftype *cft, cpuset_filetype_t type = cft->private; int retval = -ENODEV; - cpuset_sched_change_begin(); + mutex_lock(&cpuset_mutex); if (!is_cpuset_online(cs)) goto out_unlock; @@ -1677,7 +1670,7 @@ static int cpuset_write_s64(struct cgroup_subsys_state *css, struct cftype *cft, break; } out_unlock: - cpuset_sched_change_end(); + mutex_unlock(&cpuset_mutex); return retval; } @@ -1716,7 +1709,7 @@ static ssize_t cpuset_write_resmask(struct kernfs_open_file *of, kernfs_break_active_protection(of->kn); flush_work(&cpuset_hotplug_work); - cpuset_sched_change_begin(); + mutex_lock(&cpuset_mutex); if (!is_cpuset_online(cs)) goto out_unlock; @@ -1740,7 +1733,7 @@ static ssize_t cpuset_write_resmask(struct kernfs_open_file *of, free_trial_cpuset(trialcs); out_unlock: - cpuset_sched_change_end(); + mutex_unlock(&cpuset_mutex); kernfs_unbreak_active_protection(of->kn); css_put(&cs->css); flush_workqueue(cpuset_migrate_mm_wq); @@ -2041,14 +2034,14 @@ out_unlock: /* * If the cpuset being removed has its flag 'sched_load_balance' * enabled, then simulate turning sched_load_balance off, which - * will call rebuild_sched_domains_cpuslocked(). + * will call rebuild_sched_domains_locked(). */ static void cpuset_css_offline(struct cgroup_subsys_state *css) { struct cpuset *cs = css_cs(css); - cpuset_sched_change_begin(); + mutex_lock(&cpuset_mutex); if (is_sched_load_balance(cs)) update_flag(CS_SCHED_LOAD_BALANCE, cs, 0); @@ -2056,7 +2049,7 @@ static void cpuset_css_offline(struct cgroup_subsys_state *css) cpuset_dec(); clear_bit(CS_ONLINE, &cs->flags); - cpuset_sched_change_end(); + mutex_unlock(&cpuset_mutex); } static void cpuset_css_free(struct cgroup_subsys_state *css) From c2f31b79d510ec1a27138bdcf2d0ece1080be85e Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Tue, 5 Dec 2017 09:10:17 -0800 Subject: [PATCH 7/7] cgroup: add warning about RT not being supported on cgroup2 We haven't yet figured out what to do with RT threads on cgroup2. Document the limitation. v2: Included the warning about system management software behavior as suggested by Michael. Signed-off-by: Tejun Heo Reported-by: "Michael Kerrisk (man-pages)" --- Documentation/cgroup-v2.txt | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/Documentation/cgroup-v2.txt b/Documentation/cgroup-v2.txt index 779211fbb69f..2cddab7efb20 100644 --- a/Documentation/cgroup-v2.txt +++ b/Documentation/cgroup-v2.txt @@ -898,6 +898,13 @@ controller implements weight and absolute bandwidth limit models for normal scheduling policy and absolute bandwidth allocation model for realtime scheduling policy. +WARNING: cgroup2 doesn't yet support control of realtime processes and +the cpu controller can only be enabled when all RT processes are in +the root cgroup. Be aware that system management software may already +have placed RT processes into nonroot cgroups during the system boot +process, and these processes may need to be moved to the root cgroup +before the cpu controller can be enabled. + CPU Interface Files ~~~~~~~~~~~~~~~~~~~