diff options
Diffstat (limited to 'kernel/cgroup/cgroup.c')
| -rw-r--r-- | kernel/cgroup/cgroup.c | 1196 |
1 files changed, 791 insertions, 405 deletions
diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c index c099cf3fa02d..e717208cfb18 100644 --- a/kernel/cgroup/cgroup.c +++ b/kernel/cgroup/cgroup.c @@ -57,7 +57,10 @@ #include <linux/file.h> #include <linux/fs_parser.h> #include <linux/sched/cputime.h> +#include <linux/sched/deadline.h> #include <linux/psi.h> +#include <linux/nstree.h> +#include <linux/irq_work.h> #include <net/sock.h> #define CREATE_TRACE_POINTS @@ -89,11 +92,14 @@ DEFINE_MUTEX(cgroup_mutex); DEFINE_SPINLOCK(css_set_lock); -#ifdef CONFIG_PROVE_RCU +#if (defined CONFIG_PROVE_RCU || defined CONFIG_LOCKDEP) EXPORT_SYMBOL_GPL(cgroup_mutex); EXPORT_SYMBOL_GPL(css_set_lock); #endif +struct blocking_notifier_head cgroup_lifetime_notifier = + BLOCKING_NOTIFIER_INIT(cgroup_lifetime_notifier); + DEFINE_SPINLOCK(trace_cgroup_path_lock); char trace_cgroup_path[TRACE_CGROUP_PATH_LEN]; static bool cgroup_debug __read_mostly; @@ -120,10 +126,33 @@ DEFINE_PERCPU_RWSEM(cgroup_threadgroup_rwsem); /* * cgroup destruction makes heavy use of work items and there can be a lot * of concurrent destructions. Use a separate workqueue so that cgroup - * destruction work items don't end up filling up max_active of system_wq + * destruction work items don't end up filling up max_active of system_percpu_wq * which may lead to deadlock. + * + * A cgroup destruction should enqueue work sequentially to: + * cgroup_offline_wq: use for css offline work + * cgroup_release_wq: use for css release work + * cgroup_free_wq: use for free work + * + * Rationale for using separate workqueues: + * The cgroup root free work may depend on completion of other css offline + * operations. If all tasks were enqueued to a single workqueue, this could + * create a deadlock scenario where: + * - Free work waits for other css offline work to complete. + * - But other css offline work is queued after free work in the same queue. + * + * Example deadlock scenario with single workqueue (cgroup_destroy_wq): + * 1. umount net_prio + * 2. net_prio root destruction enqueues work to cgroup_destroy_wq (CPUx) + * 3. perf_event CSS A offline enqueues work to same cgroup_destroy_wq (CPUx) + * 4. net_prio cgroup_destroy_root->cgroup_lock_and_drain_offline. + * 5. net_prio root destruction blocks waiting for perf_event CSS A offline, + * which can never complete as it's behind in the same queue and + * workqueue's max_active is 1. */ -static struct workqueue_struct *cgroup_destroy_wq; +static struct workqueue_struct *cgroup_offline_wq; +static struct workqueue_struct *cgroup_release_wq; +static struct workqueue_struct *cgroup_free_wq; /* generate an array of cgroup subsystem pointers */ #define SUBSYS(_x) [_x ## _cgrp_id] = &_x ## _cgrp_subsys, @@ -160,17 +189,21 @@ static struct static_key_true *cgroup_subsys_on_dfl_key[] = { }; #undef SUBSYS -static DEFINE_PER_CPU(struct cgroup_rstat_cpu, cgrp_dfl_root_rstat_cpu); +static DEFINE_PER_CPU(struct css_rstat_cpu, root_rstat_cpu); +static DEFINE_PER_CPU(struct cgroup_rstat_base_cpu, root_rstat_base_cpu); /* the default hierarchy */ -struct cgroup_root cgrp_dfl_root = { .cgrp.rstat_cpu = &cgrp_dfl_root_rstat_cpu }; +struct cgroup_root cgrp_dfl_root = { + .cgrp.self.rstat_cpu = &root_rstat_cpu, + .cgrp.rstat_base_cpu = &root_rstat_base_cpu, +}; EXPORT_SYMBOL_GPL(cgrp_dfl_root); /* * The default hierarchy always exists but is hidden until mounted for the * first time. This is for backward compatibility. */ -static bool cgrp_dfl_visible; +bool cgrp_dfl_visible; /* some controllers are not supported in the default hierarchy */ static u16 cgrp_dfl_inhibit_ss_mask; @@ -206,12 +239,20 @@ static u16 have_exit_callback __read_mostly; static u16 have_release_callback __read_mostly; static u16 have_canfork_callback __read_mostly; +static bool have_favordynmods __ro_after_init = IS_ENABLED(CONFIG_CGROUP_FAVOR_DYNMODS); + +/* + * Write protected by cgroup_mutex and write-lock of cgroup_threadgroup_rwsem, + * read protected by either. + * + * Can only be turned on, but not turned off. + */ +bool cgroup_enable_per_threadgroup_rwsem __read_mostly; + /* cgroup namespace for init task */ struct cgroup_namespace init_cgroup_ns = { - .ns.count = REFCOUNT_INIT(2), + .ns = NS_COMMON_INIT(init_cgroup_ns), .user_ns = &init_user_ns, - .ns.ops = &cgroupns_operations, - .ns.inum = PROC_CGROUP_INIT_INO, .root_cset = &init_css_set, }; @@ -247,6 +288,7 @@ static void kill_css(struct cgroup_subsys_state *css); static int cgroup_addrm_files(struct cgroup_subsys_state *css, struct cgroup *cgrp, struct cftype cfts[], bool is_add); +static void cgroup_rt_init(void); #ifdef CONFIG_DEBUG_CGROUP_REF #define CGROUP_REF_FN_ATTRS noinline @@ -312,8 +354,6 @@ bool cgroup_ssid_enabled(int ssid) * masks of ancestors. * * - blkcg: blk-throttle becomes properly hierarchical. - * - * - debug: disallowed on the default hierarchy. */ bool cgroup_on_dfl(const struct cgroup *cgrp) { @@ -356,7 +396,7 @@ static bool cgroup_has_tasks(struct cgroup *cgrp) return cgrp->nr_populated_csets; } -bool cgroup_is_threaded(struct cgroup *cgrp) +static bool cgroup_is_threaded(struct cgroup *cgrp) { return cgrp->dom_cgrp != cgrp; } @@ -395,7 +435,7 @@ static bool cgroup_can_be_thread_root(struct cgroup *cgrp) } /* is @cgrp root of a threaded subtree? */ -bool cgroup_is_thread_root(struct cgroup *cgrp) +static bool cgroup_is_thread_root(struct cgroup *cgrp) { /* thread root should be a domain */ if (cgroup_is_threaded(cgrp)) @@ -494,28 +534,6 @@ static struct cgroup_subsys_state *cgroup_css(struct cgroup *cgrp, } /** - * cgroup_tryget_css - try to get a cgroup's css for the specified subsystem - * @cgrp: the cgroup of interest - * @ss: the subsystem of interest - * - * Find and get @cgrp's css associated with @ss. If the css doesn't exist - * or is offline, %NULL is returned. - */ -static struct cgroup_subsys_state *cgroup_tryget_css(struct cgroup *cgrp, - struct cgroup_subsys *ss) -{ - struct cgroup_subsys_state *css; - - rcu_read_lock(); - css = cgroup_css(cgrp, ss); - if (css && !css_tryget_online(css)) - css = NULL; - rcu_read_unlock(); - - return css; -} - -/** * cgroup_e_css_by_mask - obtain a cgroup's effective css for the specified ss * @cgrp: the cgroup of interest * @ss: the subsystem of interest (%NULL returns @cgrp->self) @@ -618,7 +636,7 @@ EXPORT_SYMBOL_GPL(cgroup_get_e_css); static void cgroup_get_live(struct cgroup *cgrp) { WARN_ON_ONCE(cgroup_is_dead(cgrp)); - css_get(&cgrp->self); + cgroup_get(cgrp); } /** @@ -654,9 +672,22 @@ int cgroup_task_count(const struct cgroup *cgrp) return count; } +static struct cgroup *kn_priv(struct kernfs_node *kn) +{ + struct kernfs_node *parent; + /* + * The parent can not be replaced due to KERNFS_ROOT_INVARIANT_PARENT. + * Therefore it is always safe to dereference this pointer outside of a + * RCU section. + */ + parent = rcu_dereference_check(kn->__parent, + kernfs_root_flags(kn) & KERNFS_ROOT_INVARIANT_PARENT); + return parent->priv; +} + struct cgroup_subsys_state *of_css(struct kernfs_open_file *of) { - struct cgroup *cgrp = of->kn->parent->priv; + struct cgroup *cgrp = kn_priv(of->kn); struct cftype *cft = of_cft(of); /* @@ -680,7 +711,7 @@ EXPORT_SYMBOL_GPL(of_css); * @ssid: the index of the subsystem, CGROUP_SUBSYS_COUNT after reaching the end * @cgrp: the target cgroup to iterate css's of * - * Should be called under cgroup_[tree_]mutex. + * Should be called under cgroup_mutex. */ #define for_each_css(css, ssid, cgrp) \ for ((ssid) = 0; (ssid) < CGROUP_SUBSYS_COUNT; (ssid)++) \ @@ -690,21 +721,6 @@ EXPORT_SYMBOL_GPL(of_css); else /** - * for_each_e_css - iterate all effective css's of a cgroup - * @css: the iteration cursor - * @ssid: the index of the subsystem, CGROUP_SUBSYS_COUNT after reaching the end - * @cgrp: the target cgroup to iterate css's of - * - * Should be called under cgroup_[tree_]mutex. - */ -#define for_each_e_css(css, ssid, cgrp) \ - for ((ssid) = 0; (ssid) < CGROUP_SUBSYS_COUNT; (ssid)++) \ - if (!((css) = cgroup_e_css_by_mask(cgrp, \ - cgroup_subsys[(ssid)]))) \ - ; \ - else - -/** * do_each_subsys_mask - filter for_each_subsys with a bitmask * @ss: the iteration cursor * @ssid: the index of @ss, CGROUP_SUBSYS_COUNT after reaching the end @@ -927,7 +943,8 @@ static void css_set_move_task(struct task_struct *task, /* * We are synchronized through cgroup_threadgroup_rwsem * against PF_EXITING setting such that we can't race - * against cgroup_exit()/cgroup_free() dropping the css_set. + * against cgroup_task_dead()/cgroup_task_free() dropping + * the css_set. */ WARN_ON_ONCE(task->flags & PF_EXITING); @@ -945,7 +962,7 @@ static void css_set_move_task(struct task_struct *task, #define CSS_SET_HASH_BITS 7 static DEFINE_HASHTABLE(css_set_table, CSS_SET_HASH_BITS); -static unsigned long css_set_hash(struct cgroup_subsys_state *css[]) +static unsigned long css_set_hash(struct cgroup_subsys_state **css) { unsigned long key = 0UL; struct cgroup_subsys *ss; @@ -1086,7 +1103,7 @@ static bool compare_css_sets(struct css_set *cset, */ static struct css_set *find_existing_css_set(struct css_set *old_cset, struct cgroup *cgrp, - struct cgroup_subsys_state *template[]) + struct cgroup_subsys_state **template) { struct cgroup_root *root = cgrp->root; struct cgroup_subsys *ss; @@ -1318,14 +1335,30 @@ void cgroup_favor_dynmods(struct cgroup_root *root, bool favor) { bool favoring = root->flags & CGRP_ROOT_FAVOR_DYNMODS; - /* see the comment above CGRP_ROOT_FAVOR_DYNMODS definition */ + /* + * see the comment above CGRP_ROOT_FAVOR_DYNMODS definition. + * favordynmods can flip while task is between + * cgroup_threadgroup_change_begin() and end(), so down_write global + * cgroup_threadgroup_rwsem to synchronize them. + * + * Once cgroup_enable_per_threadgroup_rwsem is enabled, holding + * cgroup_threadgroup_rwsem doesn't exlude tasks between + * cgroup_thread_group_change_begin() and end() and thus it's unsafe to + * turn off. As the scenario is unlikely, simply disallow disabling once + * enabled and print out a warning. + */ + percpu_down_write(&cgroup_threadgroup_rwsem); if (favor && !favoring) { + cgroup_enable_per_threadgroup_rwsem = true; rcu_sync_enter(&cgroup_threadgroup_rwsem.rss); root->flags |= CGRP_ROOT_FAVOR_DYNMODS; } else if (!favor && favoring) { + if (cgroup_enable_per_threadgroup_rwsem) + pr_warn_once("cgroup favordynmods: per threadgroup rwsem mechanism can't be disabled\n"); rcu_sync_exit(&cgroup_threadgroup_rwsem.rss); root->flags &= ~CGRP_ROOT_FAVOR_DYNMODS; } + percpu_up_write(&cgroup_threadgroup_rwsem); } static int cgroup_init_root_id(struct cgroup_root *root) @@ -1351,13 +1384,14 @@ static void cgroup_exit_root_id(struct cgroup_root *root) void cgroup_free_root(struct cgroup_root *root) { - kfree(root); + kfree_rcu(root, rcu); } static void cgroup_destroy_root(struct cgroup_root *root) { struct cgroup *cgrp = &root->cgrp; struct cgrp_cset_link *link, *tmp_link; + int ret; trace_cgroup_destroy_root(root); @@ -1366,6 +1400,10 @@ static void cgroup_destroy_root(struct cgroup_root *root) BUG_ON(atomic_read(&root->nr_cgrps)); BUG_ON(!list_empty(&cgrp->self.children)); + ret = blocking_notifier_call_chain(&cgroup_lifetime_notifier, + CGROUP_LIFETIME_OFFLINE, cgrp); + WARN_ON_ONCE(notifier_to_errno(ret)); + /* Rebind all subsystems back to the default hierarchy */ WARN_ON(rebind_subsystems(&cgrp_dfl_root, root->subsys_mask)); @@ -1383,17 +1421,17 @@ static void cgroup_destroy_root(struct cgroup_root *root) spin_unlock_irq(&css_set_lock); - if (!list_empty(&root->root_list)) { - list_del(&root->root_list); - cgroup_root_count--; - } + WARN_ON_ONCE(list_empty(&root->root_list)); + list_del_rcu(&root->root_list); + cgroup_root_count--; + + if (!have_favordynmods) + cgroup_favor_dynmods(root, false); - cgroup_favor_dynmods(root, false); cgroup_exit_root_id(root); - mutex_unlock(&cgroup_mutex); + cgroup_unlock(); - cgroup_rstat_exit(cgrp); kernfs_destroy_root(root->kf_root); cgroup_free_root(root); } @@ -1424,7 +1462,15 @@ static inline struct cgroup *__cset_cgroup_from_root(struct css_set *cset, } } - BUG_ON(!res_cgroup); + /* + * If cgroup_mutex is not held, the cgrp_cset_link will be freed + * before we remove the cgroup root from the root_list. Consequently, + * when accessing a cgroup root, the cset_link may have already been + * freed, resulting in a NULL res_cgroup. However, by holding the + * cgroup_mutex, we ensure that res_cgroup can't be NULL. + * If we don't hold cgroup_mutex in the caller, we must do the NULL + * check. + */ return res_cgroup; } @@ -1447,6 +1493,11 @@ current_cgns_cgroup_from_root(struct cgroup_root *root) rcu_read_unlock(); + /* + * The namespace_sem is held by current, so the root cgroup can't + * be umounted. Therefore, we can ensure that the res is non-NULL. + */ + WARN_ON_ONCE(!res); return res; } @@ -1465,15 +1516,24 @@ static struct cgroup *current_cgns_cgroup_dfl(void) { struct css_set *cset; - cset = current->nsproxy->cgroup_ns->root_cset; - return __cset_cgroup_from_root(cset, &cgrp_dfl_root); + if (current->nsproxy) { + cset = current->nsproxy->cgroup_ns->root_cset; + return __cset_cgroup_from_root(cset, &cgrp_dfl_root); + } else { + /* + * NOTE: This function may be called from bpf_cgroup_from_id() + * on a task which has already passed exit_nsproxy_namespaces() + * and nsproxy == NULL. Fall back to cgrp_dfl_root which will + * make all cgroups visible for lookups. + */ + return &cgrp_dfl_root.cgrp; + } } /* look up cgroup associated with given css_set on the specified hierarchy */ static struct cgroup *cset_cgroup_from_root(struct css_set *cset, struct cgroup_root *root) { - lockdep_assert_held(&cgroup_mutex); lockdep_assert_held(&css_set_lock); return __cset_cgroup_from_root(cset, root); @@ -1481,7 +1541,9 @@ static struct cgroup *cset_cgroup_from_root(struct css_set *cset, /* * Return the cgroup for "task" from the given hierarchy. Must be - * called with cgroup_mutex and css_set_lock held. + * called with css_set_lock held to prevent task's groups from being modified. + * Must be called with either cgroup_mutex or rcu read lock to prevent the + * cgroup root from being destroyed. */ struct cgroup *task_cgroup_from_root(struct task_struct *task, struct cgroup_root *root) @@ -1623,9 +1685,9 @@ void cgroup_kn_unlock(struct kernfs_node *kn) if (kernfs_type(kn) == KERNFS_DIR) cgrp = kn->priv; else - cgrp = kn->parent->priv; + cgrp = kn_priv(kn); - mutex_unlock(&cgroup_mutex); + cgroup_unlock(); kernfs_unbreak_active_protection(kn); cgroup_put(cgrp); @@ -1655,7 +1717,7 @@ struct cgroup *cgroup_kn_lock_live(struct kernfs_node *kn, bool drain_offline) if (kernfs_type(kn) == KERNFS_DIR) cgrp = kn->priv; else - cgrp = kn->parent->priv; + cgrp = kn_priv(kn); /* * We're gonna grab cgroup_mutex which nests outside kernfs @@ -1670,7 +1732,7 @@ struct cgroup *cgroup_kn_lock_live(struct kernfs_node *kn, bool drain_offline) if (drain_offline) cgroup_lock_and_drain_offline(cgrp); else - mutex_lock(&cgroup_mutex); + cgroup_lock(); if (!cgroup_is_dead(cgrp)) return cgrp; @@ -1693,7 +1755,7 @@ static void cgroup_rm_file(struct cgroup *cgrp, const struct cftype *cft) cfile->kn = NULL; spin_unlock_irq(&cgroup_file_kn_lock); - del_timer_sync(&cfile->notify_timer); + timer_delete_sync(&cfile->notify_timer); } kernfs_remove_by_name(cgrp->kn, cgroup_file_name(cgrp, cft, name)); @@ -1713,7 +1775,7 @@ static void css_clear_dir(struct cgroup_subsys_state *css) css->flags &= ~CSS_VISIBLE; - if (!css->ss) { + if (css_is_self(css)) { if (cgroup_on_dfl(cgrp)) { cgroup_addrm_files(css, cgrp, cgroup_base_files, false); @@ -1742,25 +1804,30 @@ static int css_populate_dir(struct cgroup_subsys_state *css) struct cftype *cfts, *failed_cfts; int ret; - if ((css->flags & CSS_VISIBLE) || !cgrp->kn) + if (css->flags & CSS_VISIBLE) return 0; - if (!css->ss) { + if (css_is_self(css)) { if (cgroup_on_dfl(cgrp)) { - ret = cgroup_addrm_files(&cgrp->self, cgrp, + ret = cgroup_addrm_files(css, cgrp, cgroup_base_files, true); if (ret < 0) return ret; if (cgroup_psi_enabled()) { - ret = cgroup_addrm_files(&cgrp->self, cgrp, + ret = cgroup_addrm_files(css, cgrp, cgroup_psi_files, true); - if (ret < 0) + if (ret < 0) { + cgroup_addrm_files(css, cgrp, + cgroup_base_files, false); return ret; + } } } else { - cgroup_addrm_files(css, cgrp, - cgroup1_base_files, true); + ret = cgroup_addrm_files(css, cgrp, + cgroup1_base_files, true); + if (ret < 0) + return ret; } } else { list_for_each_entry(cfts, &css->ss->cfts, node) { @@ -1788,7 +1855,7 @@ int rebind_subsystems(struct cgroup_root *dst_root, u16 ss_mask) { struct cgroup *dcgrp = &dst_root->cgrp; struct cgroup_subsys *ss; - int ssid, i, ret; + int ssid, ret; u16 dfl_disable_ss_mask = 0; lockdep_assert_held(&cgroup_mutex); @@ -1832,7 +1899,8 @@ int rebind_subsystems(struct cgroup_root *dst_root, u16 ss_mask) struct cgroup_root *src_root = ss->root; struct cgroup *scgrp = &src_root->cgrp; struct cgroup_subsys_state *css = cgroup_css(scgrp, ss); - struct css_set *cset; + struct css_set *cset, *cset_pos; + struct css_task_iter *it; WARN_ON(!css || cgroup_css(dcgrp, ss)); @@ -1847,20 +1915,26 @@ int rebind_subsystems(struct cgroup_root *dst_root, u16 ss_mask) RCU_INIT_POINTER(scgrp->subsys[ssid], NULL); rcu_assign_pointer(dcgrp->subsys[ssid], css); ss->root = dst_root; - css->cgroup = dcgrp; spin_lock_irq(&css_set_lock); - hash_for_each(css_set_table, i, cset, hlist) + css->cgroup = dcgrp; + WARN_ON(!list_empty(&dcgrp->e_csets[ss->id])); + list_for_each_entry_safe(cset, cset_pos, &scgrp->e_csets[ss->id], + e_cset_node[ss->id]) { list_move_tail(&cset->e_cset_node[ss->id], &dcgrp->e_csets[ss->id]); - spin_unlock_irq(&css_set_lock); - - if (ss->css_rstat_flush) { - list_del_rcu(&css->rstat_css_node); - synchronize_rcu(); - list_add_rcu(&css->rstat_css_node, - &dcgrp->rstat_css_list); + /* + * all css_sets of scgrp together in same order to dcgrp, + * patch in-flight iterators to preserve correct iteration. + * since the iterator is always advanced right away and + * finished when it->cset_pos meets it->cset_head, so only + * update it->cset_head is enough here. + */ + list_for_each_entry(it, &cset->task_iters, iters_node) + if (it->cset_head == &scgrp->e_csets[ss->id]) + it->cset_head = &dcgrp->e_csets[ss->id]; } + spin_unlock_irq(&css_set_lock); /* default hierarchy doesn't enable controllers by default */ dst_root->subsys_mask |= 1 << ssid; @@ -1901,7 +1975,7 @@ int cgroup_show_path(struct seq_file *sf, struct kernfs_node *kf_node, len = kernfs_path_from_node(kf_node, ns_cgroup->kn, buf, PATH_MAX); spin_unlock_irq(&css_set_lock); - if (len >= PATH_MAX) + if (len == -E2BIG) len = -ERANGE; else if (len > 0) { seq_escape(sf, buf, " \t\n\\"); @@ -1916,6 +1990,8 @@ enum cgroup2_param { Opt_favordynmods, Opt_memory_localevents, Opt_memory_recursiveprot, + Opt_memory_hugetlb_accounting, + Opt_pids_localevents, nr__cgroup2_params }; @@ -1924,6 +2000,8 @@ static const struct fs_parameter_spec cgroup2_fs_parameters[] = { fsparam_flag("favordynmods", Opt_favordynmods), fsparam_flag("memory_localevents", Opt_memory_localevents), fsparam_flag("memory_recursiveprot", Opt_memory_recursiveprot), + fsparam_flag("memory_hugetlb_accounting", Opt_memory_hugetlb_accounting), + fsparam_flag("pids_localevents", Opt_pids_localevents), {} }; @@ -1950,10 +2028,23 @@ static int cgroup2_parse_param(struct fs_context *fc, struct fs_parameter *param case Opt_memory_recursiveprot: ctx->flags |= CGRP_ROOT_MEMORY_RECURSIVE_PROT; return 0; + case Opt_memory_hugetlb_accounting: + ctx->flags |= CGRP_ROOT_MEMORY_HUGETLB_ACCOUNTING; + return 0; + case Opt_pids_localevents: + ctx->flags |= CGRP_ROOT_PIDS_LOCAL_EVENTS; + return 0; } return -EINVAL; } +struct cgroup_of_peak *of_peak(struct kernfs_open_file *of) +{ + struct cgroup_file_ctx *ctx = of->priv; + + return &ctx->peak; +} + static void apply_cgroup_root_flags(unsigned int root_flags) { if (current->nsproxy->cgroup_ns == &init_cgroup_ns) { @@ -1974,6 +2065,16 @@ static void apply_cgroup_root_flags(unsigned int root_flags) cgrp_dfl_root.flags |= CGRP_ROOT_MEMORY_RECURSIVE_PROT; else cgrp_dfl_root.flags &= ~CGRP_ROOT_MEMORY_RECURSIVE_PROT; + + if (root_flags & CGRP_ROOT_MEMORY_HUGETLB_ACCOUNTING) + cgrp_dfl_root.flags |= CGRP_ROOT_MEMORY_HUGETLB_ACCOUNTING; + else + cgrp_dfl_root.flags &= ~CGRP_ROOT_MEMORY_HUGETLB_ACCOUNTING; + + if (root_flags & CGRP_ROOT_PIDS_LOCAL_EVENTS) + cgrp_dfl_root.flags |= CGRP_ROOT_PIDS_LOCAL_EVENTS; + else + cgrp_dfl_root.flags &= ~CGRP_ROOT_PIDS_LOCAL_EVENTS; } } @@ -1987,6 +2088,10 @@ static int cgroup_show_options(struct seq_file *seq, struct kernfs_root *kf_root seq_puts(seq, ",memory_localevents"); if (cgrp_dfl_root.flags & CGRP_ROOT_MEMORY_RECURSIVE_PROT) seq_puts(seq, ",memory_recursiveprot"); + if (cgrp_dfl_root.flags & CGRP_ROOT_MEMORY_HUGETLB_ACCOUNTING) + seq_puts(seq, ",memory_hugetlb_accounting"); + if (cgrp_dfl_root.flags & CGRP_ROOT_PIDS_LOCAL_EVENTS) + seq_puts(seq, ",pids_localevents"); return 0; } @@ -2013,12 +2118,16 @@ static void init_cgroup_housekeeping(struct cgroup *cgrp) cgrp->dom_cgrp = cgrp; cgrp->max_descendants = INT_MAX; cgrp->max_depth = INT_MAX; - INIT_LIST_HEAD(&cgrp->rstat_css_list); prev_cputime_init(&cgrp->prev_cputime); for_each_subsys(ss, ssid) INIT_LIST_HEAD(&cgrp->e_csets[ssid]); +#ifdef CONFIG_CGROUP_BPF + for (int i = 0; i < ARRAY_SIZE(cgrp->bpf.revisions); i++) + cgrp->bpf.revisions[i] = 1; +#endif + init_waitqueue_head(&cgrp->offline_waitq); INIT_WORK(&cgrp->release_agent_work, cgroup1_release_agent); } @@ -2028,7 +2137,7 @@ void init_cgroup_root(struct cgroup_fs_context *ctx) struct cgroup_root *root = ctx->root; struct cgroup *cgrp = &root->cgrp; - INIT_LIST_HEAD(&root->root_list); + INIT_LIST_HEAD_RCU(&root->root_list); atomic_set(&root->nr_cgrps, 1); cgrp->root = root; init_cgroup_housekeeping(cgrp); @@ -2079,7 +2188,8 @@ int cgroup_setup_root(struct cgroup_root *root, u16 ss_mask) root->kf_root = kernfs_create_root(kf_sops, KERNFS_ROOT_CREATE_DEACTIVATED | KERNFS_ROOT_SUPPORT_EXPORTOP | - KERNFS_ROOT_SUPPORT_USER_XATTR, + KERNFS_ROOT_SUPPORT_USER_XATTR | + KERNFS_ROOT_INVARIANT_PARENT, root_cgrp); if (IS_ERR(root->kf_root)) { ret = PTR_ERR(root->kf_root); @@ -2093,7 +2203,7 @@ int cgroup_setup_root(struct cgroup_root *root, u16 ss_mask) if (ret) goto destroy_root; - ret = cgroup_rstat_init(root_cgrp); + ret = css_rstat_init(&root_cgrp->self); if (ret) goto destroy_root; @@ -2101,8 +2211,9 @@ int cgroup_setup_root(struct cgroup_root *root, u16 ss_mask) if (ret) goto exit_stats; - ret = cgroup_bpf_inherit(root_cgrp); - WARN_ON_ONCE(ret); + ret = blocking_notifier_call_chain(&cgroup_lifetime_notifier, + CGROUP_LIFETIME_ONLINE, root_cgrp); + WARN_ON_ONCE(notifier_to_errno(ret)); trace_cgroup_setup_root(root); @@ -2111,7 +2222,7 @@ int cgroup_setup_root(struct cgroup_root *root, u16 ss_mask) * care of subsystems' refcounts, which are explicitly dropped in * the failure exit path. */ - list_add(&root->root_list, &cgroup_roots); + list_add_rcu(&root->root_list, &cgroup_roots); cgroup_root_count++; /* @@ -2133,7 +2244,7 @@ int cgroup_setup_root(struct cgroup_root *root, u16 ss_mask) goto out; exit_stats: - cgroup_rstat_exit(root_cgrp); + css_rstat_exit(&root_cgrp->self); destroy_root: kernfs_destroy_root(root->kf_root); root->kf_root = NULL; @@ -2167,13 +2278,13 @@ int cgroup_do_get_tree(struct fs_context *fc) struct super_block *sb = fc->root->d_sb; struct cgroup *cgrp; - mutex_lock(&cgroup_mutex); + cgroup_lock(); spin_lock_irq(&css_set_lock); cgrp = cset_cgroup_from_root(ctx->ns->root_cset, ctx->root); spin_unlock_irq(&css_set_lock); - mutex_unlock(&cgroup_mutex); + cgroup_unlock(); nsdentry = kernfs_node_dentry(cgrp->kn, sb); dput(fc->root); @@ -2257,9 +2368,9 @@ static int cgroup_init_fs_context(struct fs_context *fc) fc->user_ns = get_user_ns(ctx->ns->user_ns); fc->global = true; -#ifdef CONFIG_CGROUP_FAVOR_DYNMODS - ctx->flags |= CGRP_ROOT_FAVOR_DYNMODS; -#endif + if (have_favordynmods) + ctx->flags |= CGRP_ROOT_FAVOR_DYNMODS; + return 0; } @@ -2275,10 +2386,8 @@ static void cgroup_kill_sb(struct super_block *sb) * And don't kill the default root. */ if (list_empty(&root->cgrp.self.children) && root != &cgrp_dfl_root && - !percpu_ref_is_dying(&root->cgrp.self.refcnt)) { - cgroup_bpf_offline(&root->cgrp); + !percpu_ref_is_dying(&root->cgrp.self.refcnt)) percpu_ref_kill(&root->cgrp.self.refcnt); - } cgroup_put(&root->cgrp); kernfs_kill_sb(sb); } @@ -2299,10 +2408,38 @@ static struct file_system_type cgroup2_fs_type = { .fs_flags = FS_USERNS_MOUNT, }; -#ifdef CONFIG_CPUSETS +#ifdef CONFIG_CPUSETS_V1 +enum cpuset_param { + Opt_cpuset_v2_mode, +}; + +static const struct fs_parameter_spec cpuset_fs_parameters[] = { + fsparam_flag ("cpuset_v2_mode", Opt_cpuset_v2_mode), + {} +}; + +static int cpuset_parse_param(struct fs_context *fc, struct fs_parameter *param) +{ + struct cgroup_fs_context *ctx = cgroup_fc2context(fc); + struct fs_parse_result result; + int opt; + + opt = fs_parse(fc, cpuset_fs_parameters, param, &result); + if (opt < 0) + return opt; + + switch (opt) { + case Opt_cpuset_v2_mode: + ctx->flags |= CGRP_ROOT_CPUSET_V2_MODE; + return 0; + } + return -EINVAL; +} + static const struct fs_context_operations cpuset_fs_context_ops = { .get_tree = cgroup1_get_tree, .free = cgroup_fs_context_free, + .parse_param = cpuset_parse_param, }; /* @@ -2339,6 +2476,7 @@ static int cpuset_init_fs_context(struct fs_context *fc) static struct file_system_type cpuset_fs_type = { .name = "cpuset", .init_fs_context = cpuset_init_fs_context, + .parameters = cpuset_fs_parameters, .fs_flags = FS_USERNS_MOUNT, }; #endif @@ -2356,60 +2494,22 @@ int cgroup_path_ns(struct cgroup *cgrp, char *buf, size_t buflen, { int ret; - mutex_lock(&cgroup_mutex); + cgroup_lock(); spin_lock_irq(&css_set_lock); ret = cgroup_path_ns_locked(cgrp, buf, buflen, ns); spin_unlock_irq(&css_set_lock); - mutex_unlock(&cgroup_mutex); + cgroup_unlock(); return ret; } EXPORT_SYMBOL_GPL(cgroup_path_ns); /** - * task_cgroup_path - cgroup path of a task in the first cgroup hierarchy - * @task: target task - * @buf: the buffer to write the path into - * @buflen: the length of the buffer - * - * Determine @task's cgroup on the first (the one with the lowest non-zero - * hierarchy_id) cgroup hierarchy and copy its path into @buf. This - * function grabs cgroup_mutex and shouldn't be used inside locks used by - * cgroup controller callbacks. - * - * Return value is the same as kernfs_path(). - */ -int task_cgroup_path(struct task_struct *task, char *buf, size_t buflen) -{ - struct cgroup_root *root; - struct cgroup *cgrp; - int hierarchy_id = 1; - int ret; - - mutex_lock(&cgroup_mutex); - spin_lock_irq(&css_set_lock); - - root = idr_get_next(&cgroup_hierarchy_idr, &hierarchy_id); - - if (root) { - cgrp = task_cgroup_from_root(task, root); - ret = cgroup_path_ns_locked(cgrp, buf, buflen, &init_cgroup_ns); - } else { - /* if no hierarchy exists, everyone is in "/" */ - ret = strscpy(buf, "/", buflen); - } - - spin_unlock_irq(&css_set_lock); - mutex_unlock(&cgroup_mutex); - return ret; -} -EXPORT_SYMBOL_GPL(task_cgroup_path); - -/** * cgroup_attach_lock - Lock for ->attach() - * @lock_threadgroup: whether to down_write cgroup_threadgroup_rwsem + * @lock_mode: whether acquire and acquire which rwsem + * @tsk: thread group to lock * * cgroup migration sometimes needs to stabilize threadgroups against forks and * exits by write-locking cgroup_threadgroup_rwsem. However, some ->attach() @@ -2429,22 +2529,55 @@ EXPORT_SYMBOL_GPL(task_cgroup_path); * Resolve the situation by always acquiring cpus_read_lock() before optionally * write-locking cgroup_threadgroup_rwsem. This allows ->attach() to assume that * CPU hotplug is disabled on entry. + * + * When favordynmods is enabled, take per threadgroup rwsem to reduce overhead + * on dynamic cgroup modifications. see the comment above + * CGRP_ROOT_FAVOR_DYNMODS definition. + * + * tsk is not NULL only when writing to cgroup.procs. */ -void cgroup_attach_lock(bool lock_threadgroup) +void cgroup_attach_lock(enum cgroup_attach_lock_mode lock_mode, + struct task_struct *tsk) { cpus_read_lock(); - if (lock_threadgroup) + + switch (lock_mode) { + case CGRP_ATTACH_LOCK_NONE: + break; + case CGRP_ATTACH_LOCK_GLOBAL: percpu_down_write(&cgroup_threadgroup_rwsem); + break; + case CGRP_ATTACH_LOCK_PER_THREADGROUP: + down_write(&tsk->signal->cgroup_threadgroup_rwsem); + break; + default: + pr_warn("cgroup: Unexpected attach lock mode."); + break; + } } /** * cgroup_attach_unlock - Undo cgroup_attach_lock() - * @lock_threadgroup: whether to up_write cgroup_threadgroup_rwsem + * @lock_mode: whether release and release which rwsem + * @tsk: thread group to lock */ -void cgroup_attach_unlock(bool lock_threadgroup) +void cgroup_attach_unlock(enum cgroup_attach_lock_mode lock_mode, + struct task_struct *tsk) { - if (lock_threadgroup) + switch (lock_mode) { + case CGRP_ATTACH_LOCK_NONE: + break; + case CGRP_ATTACH_LOCK_GLOBAL: percpu_up_write(&cgroup_threadgroup_rwsem); + break; + case CGRP_ATTACH_LOCK_PER_THREADGROUP: + up_write(&tsk->signal->cgroup_threadgroup_rwsem); + break; + default: + pr_warn("cgroup: Unexpected attach lock mode."); + break; + } + cpus_read_unlock(); } @@ -2530,7 +2663,7 @@ struct task_struct *cgroup_taskset_next(struct cgroup_taskset *tset, /* * This function may be called both before and - * after cgroup_taskset_migrate(). The two cases + * after cgroup_migrate_execute(). The two cases * can be distinguished by looking at whether @cset * has its ->mg_dst_cset set. */ @@ -2861,9 +2994,9 @@ int cgroup_migrate(struct task_struct *leader, bool threadgroup, struct task_struct *task; /* - * Prevent freeing of tasks while we take a snapshot. Tasks that are - * already PF_EXITING could be freed from underneath us unless we - * take an rcu_read_lock. + * The following thread iteration should be inside an RCU critical + * section to prevent tasks from being freed while taking the snapshot. + * spin_lock_irq() implies RCU critical section here. */ spin_lock_irq(&css_set_lock); task = leader; @@ -2894,14 +3027,12 @@ int cgroup_attach_task(struct cgroup *dst_cgrp, struct task_struct *leader, /* look up all src csets */ spin_lock_irq(&css_set_lock); - rcu_read_lock(); task = leader; do { cgroup_migrate_add_src(task_css_set(task), dst_cgrp, &mgctx); if (!threadgroup) break; } while_each_thread(leader, task); - rcu_read_unlock(); spin_unlock_irq(&css_set_lock); /* prepare dst csets and commit */ @@ -2918,7 +3049,7 @@ int cgroup_attach_task(struct cgroup *dst_cgrp, struct task_struct *leader, } struct task_struct *cgroup_procs_write_start(char *buf, bool threadgroup, - bool *threadgroup_locked) + enum cgroup_attach_lock_mode *lock_mode) { struct task_struct *tsk; pid_t pid; @@ -2926,24 +3057,13 @@ struct task_struct *cgroup_procs_write_start(char *buf, bool threadgroup, if (kstrtoint(strstrip(buf), 0, &pid) || pid < 0) return ERR_PTR(-EINVAL); - /* - * If we migrate a single thread, we don't care about threadgroup - * stability. If the thread is `current`, it won't exit(2) under our - * hands or change PID through exec(2). We exclude - * cgroup_update_dfl_csses and other cgroup_{proc,thread}s_write - * callers by cgroup_mutex. - * Therefore, we can skip the global lock. - */ - lockdep_assert_held(&cgroup_mutex); - *threadgroup_locked = pid || threadgroup; - cgroup_attach_lock(*threadgroup_locked); - +retry_find_task: rcu_read_lock(); if (pid) { tsk = find_task_by_vpid(pid); if (!tsk) { tsk = ERR_PTR(-ESRCH); - goto out_unlock_threadgroup; + goto out_unlock_rcu; } } else { tsk = current; @@ -2960,33 +3080,58 @@ struct task_struct *cgroup_procs_write_start(char *buf, bool threadgroup, */ if (tsk->no_cgroup_migration || (tsk->flags & PF_NO_SETAFFINITY)) { tsk = ERR_PTR(-EINVAL); - goto out_unlock_threadgroup; + goto out_unlock_rcu; } - get_task_struct(tsk); - goto out_unlock_rcu; + rcu_read_unlock(); + + /* + * If we migrate a single thread, we don't care about threadgroup + * stability. If the thread is `current`, it won't exit(2) under our + * hands or change PID through exec(2). We exclude + * cgroup_update_dfl_csses and other cgroup_{proc,thread}s_write callers + * by cgroup_mutex. Therefore, we can skip the global lock. + */ + lockdep_assert_held(&cgroup_mutex); + + if (pid || threadgroup) { + if (cgroup_enable_per_threadgroup_rwsem) + *lock_mode = CGRP_ATTACH_LOCK_PER_THREADGROUP; + else + *lock_mode = CGRP_ATTACH_LOCK_GLOBAL; + } else { + *lock_mode = CGRP_ATTACH_LOCK_NONE; + } + + cgroup_attach_lock(*lock_mode, tsk); + + if (threadgroup) { + if (!thread_group_leader(tsk)) { + /* + * A race with de_thread from another thread's exec() + * may strip us of our leadership. If this happens, + * throw this task away and try again. + */ + cgroup_attach_unlock(*lock_mode, tsk); + put_task_struct(tsk); + goto retry_find_task; + } + } + + return tsk; -out_unlock_threadgroup: - cgroup_attach_unlock(*threadgroup_locked); - *threadgroup_locked = false; out_unlock_rcu: rcu_read_unlock(); return tsk; } -void cgroup_procs_write_finish(struct task_struct *task, bool threadgroup_locked) +void cgroup_procs_write_finish(struct task_struct *task, + enum cgroup_attach_lock_mode lock_mode) { - struct cgroup_subsys *ss; - int ssid; + cgroup_attach_unlock(lock_mode, task); /* release reference from cgroup_procs_write_start() */ put_task_struct(task); - - cgroup_attach_unlock(threadgroup_locked); - - for_each_subsys(ss, ssid) - if (ss->post_attach) - ss->post_attach(); } static void cgroup_print_ss_mask(struct seq_file *seq, u16 ss_mask) @@ -3038,6 +3183,7 @@ static int cgroup_update_dfl_csses(struct cgroup *cgrp) struct cgroup_subsys_state *d_css; struct cgroup *dsct; struct css_set *src_cset; + enum cgroup_attach_lock_mode lock_mode; bool has_tasks; int ret; @@ -3069,7 +3215,13 @@ static int cgroup_update_dfl_csses(struct cgroup *cgrp) * write-locking can be skipped safely. */ has_tasks = !list_empty(&mgctx.preloaded_src_csets); - cgroup_attach_lock(has_tasks); + + if (has_tasks) + lock_mode = CGRP_ATTACH_LOCK_GLOBAL; + else + lock_mode = CGRP_ATTACH_LOCK_NONE; + + cgroup_attach_lock(lock_mode, NULL); /* NULL dst indicates self on default hierarchy */ ret = cgroup_migrate_prepare_dst(&mgctx); @@ -3090,7 +3242,7 @@ static int cgroup_update_dfl_csses(struct cgroup *cgrp) ret = cgroup_migrate_execute(&mgctx); out_finish: cgroup_migrate_finish(&mgctx); - cgroup_attach_unlock(has_tasks); + cgroup_attach_unlock(lock_mode, NULL); return ret; } @@ -3111,7 +3263,7 @@ void cgroup_lock_and_drain_offline(struct cgroup *cgrp) int ssid; restart: - mutex_lock(&cgroup_mutex); + cgroup_lock(); cgroup_for_each_live_descendant_post(dsct, d_css, cgrp) { for_each_subsys(ss, ssid) { @@ -3125,7 +3277,7 @@ restart: prepare_to_wait(&dsct->offline_waitq, &wait, TASK_UNINTERRUPTIBLE); - mutex_unlock(&cgroup_mutex); + cgroup_unlock(); schedule(); finish_wait(&dsct->offline_waitq, &wait); @@ -3676,18 +3828,90 @@ static int cgroup_events_show(struct seq_file *seq, void *v) static int cgroup_stat_show(struct seq_file *seq, void *v) { struct cgroup *cgroup = seq_css(seq)->cgroup; + struct cgroup_subsys_state *css; + int dying_cnt[CGROUP_SUBSYS_COUNT]; + int ssid; seq_printf(seq, "nr_descendants %d\n", cgroup->nr_descendants); + + /* + * Show the number of live and dying csses associated with each of + * non-inhibited cgroup subsystems that is bound to cgroup v2. + * + * Without proper lock protection, racing is possible. So the + * numbers may not be consistent when that happens. + */ + rcu_read_lock(); + for (ssid = 0; ssid < CGROUP_SUBSYS_COUNT; ssid++) { + dying_cnt[ssid] = -1; + if ((BIT(ssid) & cgrp_dfl_inhibit_ss_mask) || + (cgroup_subsys[ssid]->root != &cgrp_dfl_root)) + continue; + css = rcu_dereference_raw(cgroup->subsys[ssid]); + dying_cnt[ssid] = cgroup->nr_dying_subsys[ssid]; + seq_printf(seq, "nr_subsys_%s %d\n", cgroup_subsys[ssid]->name, + css ? (css->nr_descendants + 1) : 0); + } + seq_printf(seq, "nr_dying_descendants %d\n", cgroup->nr_dying_descendants); + for (ssid = 0; ssid < CGROUP_SUBSYS_COUNT; ssid++) { + if (dying_cnt[ssid] >= 0) + seq_printf(seq, "nr_dying_subsys_%s %d\n", + cgroup_subsys[ssid]->name, dying_cnt[ssid]); + } + rcu_read_unlock(); + return 0; +} + +static int cgroup_core_local_stat_show(struct seq_file *seq, void *v) +{ + struct cgroup *cgrp = seq_css(seq)->cgroup; + unsigned int sequence; + u64 freeze_time; + + do { + sequence = read_seqcount_begin(&cgrp->freezer.freeze_seq); + freeze_time = cgrp->freezer.frozen_nsec; + /* Add in current freezer interval if the cgroup is freezing. */ + if (test_bit(CGRP_FREEZE, &cgrp->flags)) + freeze_time += (ktime_get_ns() - + cgrp->freezer.freeze_start_nsec); + } while (read_seqcount_retry(&cgrp->freezer.freeze_seq, sequence)); + + do_div(freeze_time, NSEC_PER_USEC); + seq_printf(seq, "frozen_usec %llu\n", freeze_time); return 0; } -static int __maybe_unused cgroup_extra_stat_show(struct seq_file *seq, - struct cgroup *cgrp, int ssid) +#ifdef CONFIG_CGROUP_SCHED +/** + * cgroup_tryget_css - try to get a cgroup's css for the specified subsystem + * @cgrp: the cgroup of interest + * @ss: the subsystem of interest + * + * Find and get @cgrp's css associated with @ss. If the css doesn't exist + * or is offline, %NULL is returned. + */ +static struct cgroup_subsys_state *cgroup_tryget_css(struct cgroup *cgrp, + struct cgroup_subsys *ss) { + struct cgroup_subsys_state *css; + + rcu_read_lock(); + css = cgroup_css(cgrp, ss); + if (css && !css_tryget_online(css)) + css = NULL; + rcu_read_unlock(); + + return css; +} + +static int cgroup_extra_stat_show(struct seq_file *seq, int ssid) +{ + struct cgroup *cgrp = seq_css(seq)->cgroup; struct cgroup_subsys *ss = cgroup_subsys[ssid]; struct cgroup_subsys_state *css; int ret; @@ -3704,14 +3928,44 @@ static int __maybe_unused cgroup_extra_stat_show(struct seq_file *seq, return ret; } +static int cgroup_local_stat_show(struct seq_file *seq, + struct cgroup *cgrp, int ssid) +{ + struct cgroup_subsys *ss = cgroup_subsys[ssid]; + struct cgroup_subsys_state *css; + int ret; + + if (!ss->css_local_stat_show) + return 0; + + css = cgroup_tryget_css(cgrp, ss); + if (!css) + return 0; + + ret = ss->css_local_stat_show(seq, css); + css_put(css); + return ret; +} +#endif + static int cpu_stat_show(struct seq_file *seq, void *v) { - struct cgroup __maybe_unused *cgrp = seq_css(seq)->cgroup; int ret = 0; cgroup_base_stat_cputime_show(seq); #ifdef CONFIG_CGROUP_SCHED - ret = cgroup_extra_stat_show(seq, cgrp, cpu_cgrp_id); + ret = cgroup_extra_stat_show(seq, cpu_cgrp_id); +#endif + return ret; +} + +static int cpu_local_stat_show(struct seq_file *seq, void *v) +{ + struct cgroup __maybe_unused *cgrp = seq_css(seq)->cgroup; + int ret = 0; + +#ifdef CONFIG_CGROUP_SCHED + ret = cgroup_local_stat_show(seq, cgrp, cpu_cgrp_id); #endif return ret; } @@ -3761,7 +4015,7 @@ static ssize_t pressure_write(struct kernfs_open_file *of, char *buf, } psi = cgroup_psi(cgrp); - new = psi_trigger_create(psi, buf, res); + new = psi_trigger_create(psi, buf, res, of->file, of); if (IS_ERR(new)) { cgroup_put(cgrp); return PTR_ERR(new); @@ -3932,7 +4186,7 @@ static void __cgroup_kill(struct cgroup *cgrp) lockdep_assert_held(&cgroup_mutex); spin_lock_irq(&css_set_lock); - set_bit(CGRP_KILL, &cgrp->flags); + cgrp->kill_seq++; spin_unlock_irq(&css_set_lock); css_task_iter_start(&cgrp->self, CSS_TASK_ITER_PROCS | CSS_TASK_ITER_THREADED, &it); @@ -3948,10 +4202,6 @@ static void __cgroup_kill(struct cgroup *cgrp) send_sig(SIGKILL, task, 0); } css_task_iter_end(&it); - - spin_lock_irq(&css_set_lock); - clear_bit(CGRP_KILL, &cgrp->flags); - spin_unlock_irq(&css_set_lock); } static void cgroup_kill(struct cgroup *cgrp) @@ -4032,13 +4282,14 @@ static void cgroup_file_release(struct kernfs_open_file *of) cft->release(of); put_cgroup_ns(ctx->ns); kfree(ctx); + of->priv = NULL; } static ssize_t cgroup_file_write(struct kernfs_open_file *of, char *buf, size_t nbytes, loff_t off) { struct cgroup_file_ctx *ctx = of->priv; - struct cgroup *cgrp = of->kn->parent->priv; + struct cgroup *cgrp = kn_priv(of->kn); struct cftype *cft = of_cft(of); struct cgroup_subsys_state *css; int ret; @@ -4050,7 +4301,7 @@ static ssize_t cgroup_file_write(struct kernfs_open_file *of, char *buf, * If namespaces are delegation boundaries, disallow writes to * files in an non-init namespace root from inside the namespace * except for the files explicitly marked delegatable - - * cgroup.procs and cgroup.subtree_control. + * eg. cgroup.procs, cgroup.threads and cgroup.subtree_control. */ if ((cgrp->root->flags & CGRP_ROOT_NS_DELEGATE) && !(cft->flags & CFTYPE_NS_DELEGATABLE) && @@ -4151,20 +4402,6 @@ static struct kernfs_ops cgroup_kf_ops = { .seq_show = cgroup_seqfile_show, }; -/* set uid and gid of cgroup dirs and files to that of the creator */ -static int cgroup_kn_set_ugid(struct kernfs_node *kn) -{ - struct iattr iattr = { .ia_valid = ATTR_UID | ATTR_GID, - .ia_uid = current_fsuid(), - .ia_gid = current_fsgid(), }; - - if (uid_eq(iattr.ia_uid, GLOBAL_ROOT_UID) && - gid_eq(iattr.ia_gid, GLOBAL_ROOT_GID)) - return 0; - - return kernfs_setattr(kn, &iattr); -} - static void cgroup_file_notify_timer(struct timer_list *timer) { cgroup_file_notify(container_of(timer, struct cgroup_file, @@ -4177,25 +4414,18 @@ static int cgroup_add_file(struct cgroup_subsys_state *css, struct cgroup *cgrp, char name[CGROUP_FILE_NAME_MAX]; struct kernfs_node *kn; struct lock_class_key *key = NULL; - int ret; #ifdef CONFIG_DEBUG_LOCK_ALLOC key = &cft->lockdep_key; #endif kn = __kernfs_create_file(cgrp->kn, cgroup_file_name(cgrp, cft, name), cgroup_file_mode(cft), - GLOBAL_ROOT_UID, GLOBAL_ROOT_GID, + current_fsuid(), current_fsgid(), 0, cft->kf_ops, cft, NULL, key); if (IS_ERR(kn)) return PTR_ERR(kn); - ret = cgroup_kn_set_ugid(kn); - if (ret) { - kernfs_remove(kn); - return ret; - } - if (cft->file_offset) { struct cgroup_file *cfile = (void *)css + cft->file_offset; @@ -4343,14 +4573,13 @@ static int cgroup_init_cftypes(struct cgroup_subsys *ss, struct cftype *cfts) return ret; } -static int cgroup_rm_cftypes_locked(struct cftype *cfts) +static void cgroup_rm_cftypes_locked(struct cftype *cfts) { lockdep_assert_held(&cgroup_mutex); list_del(&cfts->node); cgroup_apply_cftypes(cfts, false); cgroup_exit_cftypes(cfts); - return 0; } /** @@ -4366,18 +4595,16 @@ static int cgroup_rm_cftypes_locked(struct cftype *cfts) */ int cgroup_rm_cftypes(struct cftype *cfts) { - int ret; - if (!cfts || cfts[0].name[0] == '\0') return 0; if (!(cfts[0].flags & __CFTYPE_ADDED)) return -ENOENT; - mutex_lock(&cgroup_mutex); - ret = cgroup_rm_cftypes_locked(cfts); - mutex_unlock(&cgroup_mutex); - return ret; + cgroup_lock(); + cgroup_rm_cftypes_locked(cfts); + cgroup_unlock(); + return 0; } /** @@ -4394,7 +4621,7 @@ int cgroup_rm_cftypes(struct cftype *cfts) * function currently returns 0 as long as @cfts registration is successful * even if some file creation attempts on existing cgroups fail. */ -static int cgroup_add_cftypes(struct cgroup_subsys *ss, struct cftype *cfts) +int cgroup_add_cftypes(struct cgroup_subsys *ss, struct cftype *cfts) { int ret; @@ -4408,14 +4635,14 @@ static int cgroup_add_cftypes(struct cgroup_subsys *ss, struct cftype *cfts) if (ret) return ret; - mutex_lock(&cgroup_mutex); + cgroup_lock(); list_add_tail(&cfts->node, &ss->cfts); ret = cgroup_apply_cftypes(cfts, true); if (ret) cgroup_rm_cftypes_locked(cfts); - mutex_unlock(&cgroup_mutex); + cgroup_unlock(); return ret; } @@ -4477,6 +4704,7 @@ void cgroup_file_notify(struct cgroup_file *cfile) } spin_unlock_irqrestore(&cgroup_file_kn_lock, flags); } +EXPORT_SYMBOL_GPL(cgroup_file_notify); /** * cgroup_file_show - show or hide a hidden cgroup file @@ -4573,8 +4801,9 @@ struct cgroup_subsys_state *css_next_child(struct cgroup_subsys_state *pos, * * While this function requires cgroup_mutex or RCU read locking, it * doesn't require the whole traversal to be contained in a single critical - * section. This function will return the correct next descendant as long - * as both @pos and @root are accessible and @pos is a descendant of @root. + * section. Additionally, it isn't necessary to hold onto a reference to @pos. + * This function will return the correct next descendant as long as both @pos + * and @root are accessible and @pos is a descendant of @root. * * If a subsystem synchronizes ->css_online() and the start of iteration, a * css which finished ->css_online() is guaranteed to be visible in the @@ -4622,8 +4851,9 @@ EXPORT_SYMBOL_GPL(css_next_descendant_pre); * * While this function requires cgroup_mutex or RCU read locking, it * doesn't require the whole traversal to be contained in a single critical - * section. This function will return the correct rightmost descendant as - * long as @pos is accessible. + * section. Additionally, it isn't necessary to hold onto a reference to @pos. + * This function will return the correct rightmost descendant as long as @pos + * is accessible. */ struct cgroup_subsys_state * css_rightmost_descendant(struct cgroup_subsys_state *pos) @@ -4667,9 +4897,9 @@ css_leftmost_descendant(struct cgroup_subsys_state *pos) * * While this function requires cgroup_mutex or RCU read locking, it * doesn't require the whole traversal to be contained in a single critical - * section. This function will return the correct next descendant as long - * as both @pos and @cgroup are accessible and @pos is a descendant of - * @cgroup. + * section. Additionally, it isn't necessary to hold onto a reference to @pos. + * This function will return the correct next descendant as long as both @pos + * and @cgroup are accessible and @pos is a descendant of @cgroup. * * If a subsystem synchronizes ->css_online() and the start of iteration, a * css which finished ->css_online() is guaranteed to be visible in the @@ -4912,9 +5142,11 @@ repeat: void css_task_iter_start(struct cgroup_subsys_state *css, unsigned int flags, struct css_task_iter *it) { + unsigned long irqflags; + memset(it, 0, sizeof(*it)); - spin_lock_irq(&css_set_lock); + spin_lock_irqsave(&css_set_lock, irqflags); it->ss = css->ss; it->flags = flags; @@ -4928,7 +5160,7 @@ void css_task_iter_start(struct cgroup_subsys_state *css, unsigned int flags, css_task_iter_advance(it); - spin_unlock_irq(&css_set_lock); + spin_unlock_irqrestore(&css_set_lock, irqflags); } /** @@ -4941,12 +5173,14 @@ void css_task_iter_start(struct cgroup_subsys_state *css, unsigned int flags, */ struct task_struct *css_task_iter_next(struct css_task_iter *it) { + unsigned long irqflags; + if (it->cur_task) { put_task_struct(it->cur_task); it->cur_task = NULL; } - spin_lock_irq(&css_set_lock); + spin_lock_irqsave(&css_set_lock, irqflags); /* @it may be half-advanced by skips, finish advancing */ if (it->flags & CSS_TASK_ITER_SKIPPED) @@ -4959,7 +5193,7 @@ struct task_struct *css_task_iter_next(struct css_task_iter *it) css_task_iter_advance(it); } - spin_unlock_irq(&css_set_lock); + spin_unlock_irqrestore(&css_set_lock, irqflags); return it->cur_task; } @@ -4972,11 +5206,13 @@ struct task_struct *css_task_iter_next(struct css_task_iter *it) */ void css_task_iter_end(struct css_task_iter *it) { + unsigned long irqflags; + if (it->cur_cset) { - spin_lock_irq(&css_set_lock); + spin_lock_irqsave(&css_set_lock, irqflags); list_del(&it->iters_node); put_css_set_locked(it->cur_cset); - spin_unlock_irq(&css_set_lock); + spin_unlock_irqrestore(&css_set_lock, irqflags); } if (it->cur_dcset) @@ -5065,7 +5301,7 @@ static int cgroup_may_write(const struct cgroup *cgrp, struct super_block *sb) if (!inode) return -ENOMEM; - ret = inode_permission(&init_user_ns, inode, MAY_WRITE); + ret = inode_permission(&nop_mnt_idmap, inode, MAY_WRITE); iput(inode); return ret; } @@ -5128,15 +5364,14 @@ static ssize_t __cgroup_procs_write(struct kernfs_open_file *of, char *buf, struct cgroup_file_ctx *ctx = of->priv; struct cgroup *src_cgrp, *dst_cgrp; struct task_struct *task; - const struct cred *saved_cred; ssize_t ret; - bool threadgroup_locked; + enum cgroup_attach_lock_mode lock_mode; dst_cgrp = cgroup_kn_lock_live(of->kn, false); if (!dst_cgrp) return -ENODEV; - task = cgroup_procs_write_start(buf, threadgroup, &threadgroup_locked); + task = cgroup_procs_write_start(buf, threadgroup, &lock_mode); ret = PTR_ERR_OR_ZERO(task); if (ret) goto out_unlock; @@ -5151,18 +5386,17 @@ static ssize_t __cgroup_procs_write(struct kernfs_open_file *of, char *buf, * permissions using the credentials from file open to protect against * inherited fd attacks. */ - saved_cred = override_creds(of->file->f_cred); - ret = cgroup_attach_permissions(src_cgrp, dst_cgrp, - of->file->f_path.dentry->d_sb, - threadgroup, ctx->ns); - revert_creds(saved_cred); + scoped_with_creds(of->file->f_cred) + ret = cgroup_attach_permissions(src_cgrp, dst_cgrp, + of->file->f_path.dentry->d_sb, + threadgroup, ctx->ns); if (ret) goto out_finish; ret = cgroup_attach_task(dst_cgrp, task, threadgroup); out_finish: - cgroup_procs_write_finish(task, threadgroup_locked); + cgroup_procs_write_finish(task, lock_mode); out_unlock: cgroup_kn_unlock(of->kn); @@ -5244,6 +5478,11 @@ static struct cftype cgroup_base_files[] = { .seq_show = cgroup_stat_show, }, { + .name = "cgroup.stat.local", + .flags = CFTYPE_NOT_ON_ROOT, + .seq_show = cgroup_core_local_stat_show, + }, + { .name = "cgroup.freeze", .flags = CFTYPE_NOT_ON_ROOT, .seq_show = cgroup_freeze_show, @@ -5258,6 +5497,10 @@ static struct cftype cgroup_base_files[] = { .name = "cpu.stat", .seq_show = cpu_stat_show, }, + { + .name = "cpu.stat.local", + .seq_show = cpu_local_stat_show, + }, { } /* terminate */ }; @@ -5322,7 +5565,7 @@ static struct cftype cgroup_psi_files[] = { * RCU callback. * * 4. After the grace period, the css can be freed. Implemented in - * css_free_work_fn(). + * css_free_rwork_fn(). * * It is actually hairier because both step 2 and 4 require process context * and thus involve punting to css->destroy_work adding two additional @@ -5336,8 +5579,9 @@ static void css_free_rwork_fn(struct work_struct *work) struct cgroup *cgrp = css->cgroup; percpu_ref_exit(&css->refcnt); + css_rstat_exit(css); - if (ss) { + if (!css_is_self(css)) { /* css free path */ struct cgroup_subsys_state *parent = css->parent; int id = css->id; @@ -5351,7 +5595,8 @@ static void css_free_rwork_fn(struct work_struct *work) } else { /* cgroup free path */ atomic_dec(&cgrp->root->nr_cgrps); - cgroup1_pidlist_destroy_all(cgrp); + if (!cgroup_on_dfl(cgrp)) + cgroup1_pidlist_destroy_all(cgrp); cancel_work_sync(&cgrp->release_agent_work); bpf_cgrp_storage_free(cgrp); @@ -5365,7 +5610,6 @@ static void css_free_rwork_fn(struct work_struct *work) cgroup_put(cgroup_parent(cgrp)); kernfs_put(cgrp->kn); psi_cgroup_free(cgrp); - cgroup_rstat_exit(cgrp); kfree(cgrp); } else { /* @@ -5385,28 +5629,41 @@ static void css_release_work_fn(struct work_struct *work) struct cgroup_subsys *ss = css->ss; struct cgroup *cgrp = css->cgroup; - mutex_lock(&cgroup_mutex); + cgroup_lock(); css->flags |= CSS_RELEASED; list_del_rcu(&css->sibling); - if (ss) { - /* css release path */ - if (!list_empty(&css->rstat_css_node)) { - cgroup_rstat_flush(cgrp); - list_del_rcu(&css->rstat_css_node); - } + if (!css_is_self(css)) { + struct cgroup *parent_cgrp; + + css_rstat_flush(css); cgroup_idr_replace(&ss->css_idr, NULL, css->id); if (ss->css_released) ss->css_released(css); + + cgrp->nr_dying_subsys[ss->id]--; + /* + * When a css is released and ready to be freed, its + * nr_descendants must be zero. However, the corresponding + * cgrp->nr_dying_subsys[ss->id] may not be 0 if a subsystem + * is activated and deactivated multiple times with one or + * more of its previous activation leaving behind dying csses. + */ + WARN_ON_ONCE(css->nr_descendants); + parent_cgrp = cgroup_parent(cgrp); + while (parent_cgrp) { + parent_cgrp->nr_dying_subsys[ss->id]--; + parent_cgrp = cgroup_parent(parent_cgrp); + } } else { struct cgroup *tcgrp; /* cgroup release path */ TRACE_CGROUP_PATH(release, cgrp); - cgroup_rstat_flush(cgrp); + css_rstat_flush(&cgrp->self); spin_lock_irq(&css_set_lock); for (tcgrp = cgroup_parent(cgrp); tcgrp; @@ -5426,10 +5683,10 @@ static void css_release_work_fn(struct work_struct *work) NULL); } - mutex_unlock(&cgroup_mutex); + cgroup_unlock(); INIT_RCU_WORK(&css->destroy_rwork, css_free_rwork_fn); - queue_rcu_work(cgroup_destroy_wq, &css->destroy_rwork); + queue_rcu_work(cgroup_free_wq, &css->destroy_rwork); } static void css_release(struct percpu_ref *ref) @@ -5438,7 +5695,7 @@ static void css_release(struct percpu_ref *ref) container_of(ref, struct cgroup_subsys_state, refcnt); INIT_WORK(&css->destroy_work, css_release_work_fn); - queue_work(cgroup_destroy_wq, &css->destroy_work); + queue_work(cgroup_release_wq, &css->destroy_work); } static void init_and_link_css(struct cgroup_subsys_state *css, @@ -5454,7 +5711,6 @@ static void init_and_link_css(struct cgroup_subsys_state *css, css->id = -1; INIT_LIST_HEAD(&css->sibling); INIT_LIST_HEAD(&css->children); - INIT_LIST_HEAD(&css->rstat_css_node); css->serial_nr = css_serial_nr_next++; atomic_set(&css->online_cnt, 0); @@ -5463,9 +5719,6 @@ static void init_and_link_css(struct cgroup_subsys_state *css, css_get(css->parent); } - if (ss->css_rstat_flush) - list_add_rcu(&css->rstat_css_node, &cgrp->rstat_css_list); - BUG_ON(cgroup_css(cgrp, ss)); } @@ -5484,8 +5737,11 @@ static int online_css(struct cgroup_subsys_state *css) rcu_assign_pointer(css->cgroup->subsys[ss->id], css); atomic_inc(&css->online_cnt); - if (css->parent) + if (css->parent) { atomic_inc(&css->parent->online_cnt); + while ((css = css->parent)) + css->nr_descendants++; + } } return ret; } @@ -5507,6 +5763,16 @@ static void offline_css(struct cgroup_subsys_state *css) RCU_INIT_POINTER(css->cgroup->subsys[ss->id], NULL); wake_up_all(&css->cgroup->offline_waitq); + + css->cgroup->nr_dying_subsys[ss->id]++; + /* + * Parent css and cgroup cannot be freed until after the freeing + * of child css, see css_free_rwork_fn(). + */ + while ((css = css->parent)) { + css->nr_descendants--; + css->cgroup->nr_dying_subsys[ss->id]++; + } } /** @@ -5545,6 +5811,10 @@ static struct cgroup_subsys_state *css_create(struct cgroup *cgrp, goto err_free_css; css->id = err; + err = css_rstat_init(css); + if (err) + goto err_free_css; + /* @css is ready to be brought online now, make it visible */ list_add_tail_rcu(&css->sibling, &parent_css->children); cgroup_idr_replace(&ss->css_idr, css, css->id); @@ -5558,16 +5828,14 @@ static struct cgroup_subsys_state *css_create(struct cgroup *cgrp, err_list_del: list_del_rcu(&css->sibling); err_free_css: - list_del_rcu(&css->rstat_css_node); INIT_RCU_WORK(&css->destroy_rwork, css_free_rwork_fn); - queue_rcu_work(cgroup_destroy_wq, &css->destroy_rwork); + queue_rcu_work(cgroup_free_wq, &css->destroy_rwork); return ERR_PTR(err); } /* * The returned cgroup is fully initialized including its control mask, but - * it isn't associated with its kernfs_node and doesn't have the control - * mask applied. + * it doesn't have the control mask applied. */ static struct cgroup *cgroup_create(struct cgroup *parent, const char *name, umode_t mode) @@ -5575,7 +5843,7 @@ static struct cgroup *cgroup_create(struct cgroup *parent, const char *name, struct cgroup_root *root = parent->root; struct cgroup *cgrp, *tcgrp; struct kernfs_node *kn; - int level = parent->level + 1; + int i, level = parent->level + 1; int ret; /* allocate the cgroup and its ID, 0 is reserved for the root */ @@ -5587,15 +5855,13 @@ static struct cgroup *cgroup_create(struct cgroup *parent, const char *name, if (ret) goto out_free_cgrp; - ret = cgroup_rstat_init(cgrp); - if (ret) - goto out_cancel_ref; - /* create the directory */ - kn = kernfs_create_dir(parent->kn, name, mode, cgrp); + kn = kernfs_create_dir_ns(parent->kn, name, mode, + current_fsuid(), current_fsgid(), + cgrp, NULL); if (IS_ERR(kn)) { ret = PTR_ERR(kn); - goto out_stat_exit; + goto out_cancel_ref; } cgrp->kn = kn; @@ -5605,19 +5871,27 @@ static struct cgroup *cgroup_create(struct cgroup *parent, const char *name, cgrp->root = root; cgrp->level = level; - ret = psi_cgroup_alloc(cgrp); + /* + * Now that init_cgroup_housekeeping() has been called and cgrp->self + * is setup, it is safe to perform rstat initialization on it. + */ + ret = css_rstat_init(&cgrp->self); if (ret) goto out_kernfs_remove; - ret = cgroup_bpf_inherit(cgrp); + ret = psi_cgroup_alloc(cgrp); if (ret) - goto out_psi_free; + goto out_stat_exit; + + for (tcgrp = cgrp; tcgrp; tcgrp = cgroup_parent(tcgrp)) + cgrp->ancestors[tcgrp->level] = tcgrp; /* * New cgroup inherits effective freeze counter, and * if the parent has to be frozen, the child has too. */ cgrp->freezer.e_freeze = parent->freezer.e_freeze; + seqcount_spinlock_init(&cgrp->freezer.freeze_seq, &css_set_lock); if (cgrp->freezer.e_freeze) { /* * Set the CGRP_FREEZE flag, so when a process will be @@ -5626,27 +5900,10 @@ static struct cgroup *cgroup_create(struct cgroup *parent, const char *name, * consider it frozen immediately. */ set_bit(CGRP_FREEZE, &cgrp->flags); + cgrp->freezer.freeze_start_nsec = ktime_get_ns(); set_bit(CGRP_FROZEN, &cgrp->flags); } - spin_lock_irq(&css_set_lock); - for (tcgrp = cgrp; tcgrp; tcgrp = cgroup_parent(tcgrp)) { - cgrp->ancestors[tcgrp->level] = tcgrp; - - if (tcgrp != cgrp) { - tcgrp->nr_descendants++; - - /* - * If the new cgroup is frozen, all ancestor cgroups - * get a new frozen descendant, but their state can't - * change because of this. - */ - if (cgrp->freezer.e_freeze) - tcgrp->freezer.nr_frozen_descendants++; - } - } - spin_unlock_irq(&css_set_lock); - if (notify_on_release(parent)) set_bit(CGRP_NOTIFY_ON_RELEASE, &cgrp->flags); @@ -5655,7 +5912,29 @@ static struct cgroup *cgroup_create(struct cgroup *parent, const char *name, cgrp->self.serial_nr = css_serial_nr_next++; + ret = blocking_notifier_call_chain_robust(&cgroup_lifetime_notifier, + CGROUP_LIFETIME_ONLINE, + CGROUP_LIFETIME_OFFLINE, cgrp); + ret = notifier_to_errno(ret); + if (ret) + goto out_psi_free; + /* allocation complete, commit to creation */ + spin_lock_irq(&css_set_lock); + for (i = 0; i < level; i++) { + tcgrp = cgrp->ancestors[i]; + tcgrp->nr_descendants++; + + /* + * If the new cgroup is frozen, all ancestor cgroups get a new + * frozen descendant, but their state can't change because of + * this. + */ + if (cgrp->freezer.e_freeze) + tcgrp->freezer.nr_frozen_descendants++; + } + spin_unlock_irq(&css_set_lock); + list_add_tail_rcu(&cgrp->self.sibling, &cgroup_parent(cgrp)->self.children); atomic_inc(&root->nr_cgrps); cgroup_get_live(parent); @@ -5673,10 +5952,10 @@ static struct cgroup *cgroup_create(struct cgroup *parent, const char *name, out_psi_free: psi_cgroup_free(cgrp); +out_stat_exit: + css_rstat_exit(&cgrp->self); out_kernfs_remove: kernfs_remove(cgrp->kn); -out_stat_exit: - cgroup_rstat_exit(cgrp); out_cancel_ref: percpu_ref_exit(&cgrp->self.refcnt); out_free_cgrp: @@ -5688,7 +5967,7 @@ static bool cgroup_check_hierarchy_limits(struct cgroup *parent) { struct cgroup *cgroup; int ret = false; - int level = 1; + int level = 0; lockdep_assert_held(&cgroup_mutex); @@ -5696,7 +5975,7 @@ static bool cgroup_check_hierarchy_limits(struct cgroup *parent) if (cgroup->nr_descendants >= cgroup->max_descendants) goto fail; - if (level > cgroup->max_depth) + if (level >= cgroup->max_depth) goto fail; level++; @@ -5732,15 +6011,11 @@ int cgroup_mkdir(struct kernfs_node *parent_kn, const char *name, umode_t mode) } /* - * This extra ref will be put in cgroup_free_fn() and guarantees + * This extra ref will be put in css_free_rwork_fn() and guarantees * that @cgrp->kn is always accessible. */ kernfs_get(cgrp->kn); - ret = cgroup_kn_set_ugid(cgrp->kn); - if (ret) - goto out_destroy; - ret = css_populate_dir(&cgrp->self); if (ret) goto out_destroy; @@ -5774,7 +6049,7 @@ static void css_killed_work_fn(struct work_struct *work) struct cgroup_subsys_state *css = container_of(work, struct cgroup_subsys_state, destroy_work); - mutex_lock(&cgroup_mutex); + cgroup_lock(); do { offline_css(css); @@ -5783,7 +6058,7 @@ static void css_killed_work_fn(struct work_struct *work) css = css->parent; } while (css && atomic_dec_and_test(&css->online_cnt)); - mutex_unlock(&cgroup_mutex); + cgroup_unlock(); } /* css kill confirmation processing requires process context, bounce */ @@ -5794,7 +6069,7 @@ static void css_killed_ref_fn(struct percpu_ref *ref) if (atomic_dec_and_test(&css->online_cnt)) { INIT_WORK(&css->destroy_work, css_killed_work_fn); - queue_work(cgroup_destroy_wq, &css->destroy_work); + queue_work(cgroup_offline_wq, &css->destroy_work); } } @@ -5814,6 +6089,12 @@ static void kill_css(struct cgroup_subsys_state *css) if (css->flags & CSS_DYING) return; + /* + * Call css_killed(), if defined, before setting the CSS_DYING flag + */ + if (css->ss->css_killed) + css->ss->css_killed(css); + css->flags |= CSS_DYING; /* @@ -5871,7 +6152,7 @@ static int cgroup_destroy_locked(struct cgroup *cgrp) struct cgroup *tcgrp, *parent = cgroup_parent(cgrp); struct cgroup_subsys_state *css; struct cgrp_cset_link *link; - int ssid; + int ssid, ret; lockdep_assert_held(&cgroup_mutex); @@ -5893,7 +6174,7 @@ static int cgroup_destroy_locked(struct cgroup *cgrp) /* * Mark @cgrp and the associated csets dead. The former prevents * further task migration and child creation by disabling - * cgroup_lock_live_group(). The latter makes the csets ignored by + * cgroup_kn_lock_live(). The latter makes the csets ignored by * the migration path. */ cgrp->self.flags &= ~CSS_ONLINE; @@ -5915,7 +6196,7 @@ static int cgroup_destroy_locked(struct cgroup *cgrp) parent->nr_threaded_children--; spin_lock_irq(&css_set_lock); - for (tcgrp = cgroup_parent(cgrp); tcgrp; tcgrp = cgroup_parent(tcgrp)) { + for (tcgrp = parent; tcgrp; tcgrp = cgroup_parent(tcgrp)) { tcgrp->nr_descendants--; tcgrp->nr_dying_descendants++; /* @@ -5929,7 +6210,9 @@ static int cgroup_destroy_locked(struct cgroup *cgrp) cgroup1_check_for_release(parent); - cgroup_bpf_offline(cgrp); + ret = blocking_notifier_call_chain(&cgroup_lifetime_notifier, + CGROUP_LIFETIME_OFFLINE, cgrp); + WARN_ON_ONCE(notifier_to_errno(ret)); /* put the base reference */ percpu_ref_kill(&cgrp->self.refcnt); @@ -5967,7 +6250,7 @@ static void __init cgroup_init_subsys(struct cgroup_subsys *ss, bool early) pr_debug("Initializing cgroup subsys %s\n", ss->name); - mutex_lock(&cgroup_mutex); + cgroup_lock(); idr_init(&ss->css_idr); INIT_LIST_HEAD(&ss->cfts); @@ -5991,6 +6274,9 @@ static void __init cgroup_init_subsys(struct cgroup_subsys *ss, bool early) } else { css->id = cgroup_idr_alloc(&ss->css_idr, css, 1, 2, GFP_KERNEL); BUG_ON(css->id < 0); + + BUG_ON(ss_rstat_init(ss)); + BUG_ON(css_rstat_init(css)); } /* Update the init_css_set to contain a subsys @@ -6011,7 +6297,7 @@ static void __init cgroup_init_subsys(struct cgroup_subsys *ss, bool early) BUG_ON(online_css(css)); - mutex_unlock(&cgroup_mutex); + cgroup_unlock(); } /** @@ -6039,6 +6325,8 @@ int __init cgroup_init_early(void) ss->id, ss->name); WARN(strlen(cgroup_subsys_name[i]) > MAX_CGROUP_TYPE_NAMELEN, "cgroup_subsys_name %s too long\n", cgroup_subsys_name[i]); + WARN(ss->early_init && ss->css_rstat_flush, + "cgroup rstat cannot be used with early init subsystem\n"); ss->id = i; ss->name = cgroup_subsys_name[i]; @@ -6067,11 +6355,12 @@ int __init cgroup_init(void) BUG_ON(cgroup_init_cftypes(NULL, cgroup_psi_files)); BUG_ON(cgroup_init_cftypes(NULL, cgroup1_base_files)); - cgroup_rstat_boot(); + BUG_ON(ss_rstat_init(NULL)); get_user_ns(init_cgroup_ns.user_ns); + cgroup_rt_init(); - mutex_lock(&cgroup_mutex); + cgroup_lock(); /* * Add init_css_set to the hash table so that dfl_root can link to @@ -6080,9 +6369,11 @@ int __init cgroup_init(void) hash_add(css_set_table, &init_css_set.hlist, css_set_hash(init_css_set.subsys)); + cgroup_bpf_lifetime_notifier_init(); + BUG_ON(cgroup_setup_root(&cgrp_dfl_root, 0)); - mutex_unlock(&cgroup_mutex); + cgroup_unlock(); for_each_subsys(ss, ssid) { if (ss->early_init) { @@ -6108,8 +6399,8 @@ int __init cgroup_init(void) continue; if (cgroup1_ssid_disabled(ssid)) - printk(KERN_INFO "Disabling %s control group subsystem in v1 mounts\n", - ss->name); + pr_info("Disabling %s control group subsystem in v1 mounts\n", + ss->legacy_name); cgrp_dfl_root.subsys_mask |= 1 << ss->id; @@ -6134,9 +6425,9 @@ int __init cgroup_init(void) if (ss->bind) ss->bind(init_css_set.subsys[ssid]); - mutex_lock(&cgroup_mutex); + cgroup_lock(); css_populate_dir(init_css_set.subsys[ssid]); - mutex_unlock(&cgroup_mutex); + cgroup_unlock(); } /* init_css_set.subsys[] has been updated, re-hash */ @@ -6148,10 +6439,11 @@ int __init cgroup_init(void) WARN_ON(register_filesystem(&cgroup_fs_type)); WARN_ON(register_filesystem(&cgroup2_fs_type)); WARN_ON(!proc_create_single("cgroups", 0, NULL, proc_cgroupstats_show)); -#ifdef CONFIG_CPUSETS +#ifdef CONFIG_CPUSETS_V1 WARN_ON(register_filesystem(&cpuset_fs_type)); #endif + ns_tree_add(&init_cgroup_ns); return 0; } @@ -6165,8 +6457,14 @@ static int __init cgroup_wq_init(void) * We would prefer to do this in cgroup_init() above, but that * is called before init_workqueues(): so leave this until after. */ - cgroup_destroy_wq = alloc_workqueue("cgroup_destroy", 0, 1); - BUG_ON(!cgroup_destroy_wq); + cgroup_offline_wq = alloc_workqueue("cgroup_offline", WQ_PERCPU, 1); + BUG_ON(!cgroup_offline_wq); + + cgroup_release_wq = alloc_workqueue("cgroup_release", WQ_PERCPU, 1); + BUG_ON(!cgroup_release_wq); + + cgroup_free_wq = alloc_workqueue("cgroup_free", WQ_PERCPU, 1); + BUG_ON(!cgroup_free_wq); return 0; } core_initcall(cgroup_wq_init); @@ -6183,15 +6481,15 @@ void cgroup_path_from_kernfs_id(u64 id, char *buf, size_t buflen) } /* - * cgroup_get_from_id : get the cgroup associated with cgroup id + * __cgroup_get_from_id : get the cgroup associated with cgroup id * @id: cgroup id * On success return the cgrp or ERR_PTR on failure - * Only cgroups within current task's cgroup NS are valid. + * There are no cgroup NS restrictions. */ -struct cgroup *cgroup_get_from_id(u64 id) +struct cgroup *__cgroup_get_from_id(u64 id) { struct kernfs_node *kn; - struct cgroup *cgrp, *root_cgrp; + struct cgroup *cgrp; kn = kernfs_find_and_get_node_by_id(cgrp_dfl_root.kf_root, id); if (!kn) @@ -6213,6 +6511,22 @@ struct cgroup *cgroup_get_from_id(u64 id) if (!cgrp) return ERR_PTR(-ENOENT); + return cgrp; +} + +/* + * cgroup_get_from_id : get the cgroup associated with cgroup id + * @id: cgroup id + * On success return the cgrp or ERR_PTR on failure + * Only cgroups within current task's cgroup NS are valid. + */ +struct cgroup *cgroup_get_from_id(u64 id) +{ + struct cgroup *cgrp, *root_cgrp; + + cgrp = __cgroup_get_from_id(id); + if (IS_ERR(cgrp)) + return cgrp; root_cgrp = current_cgns_cgroup_dfl(); if (!cgroup_is_descendant(cgrp, root_cgrp)) { @@ -6241,7 +6555,7 @@ int proc_cgroup_show(struct seq_file *m, struct pid_namespace *ns, if (!buf) goto out; - mutex_lock(&cgroup_mutex); + rcu_read_lock(); spin_lock_irq(&css_set_lock); for_each_root(root) { @@ -6252,6 +6566,11 @@ int proc_cgroup_show(struct seq_file *m, struct pid_namespace *ns, if (root == &cgrp_dfl_root && !READ_ONCE(cgrp_dfl_visible)) continue; + cgrp = task_cgroup_from_root(tsk, root); + /* The root has already been unmounted. */ + if (!cgrp) + continue; + seq_printf(m, "%d:", root->hierarchy_id); if (root != &cgrp_dfl_root) for_each_subsys(ss, ssid) @@ -6262,9 +6581,6 @@ int proc_cgroup_show(struct seq_file *m, struct pid_namespace *ns, seq_printf(m, "%sname=%s", count ? "," : "", root->name); seq_putc(m, ':'); - - cgrp = task_cgroup_from_root(tsk, root); - /* * On traditional hierarchies, all zombie tasks show up as * belonging to the root cgroup. On the default hierarchy, @@ -6277,7 +6593,7 @@ int proc_cgroup_show(struct seq_file *m, struct pid_namespace *ns, if (cgroup_on_dfl(cgrp) || !(tsk->flags & PF_EXITING)) { retval = cgroup_path_ns_locked(cgrp, buf, PATH_MAX, current->nsproxy->cgroup_ns); - if (retval >= PATH_MAX) + if (retval == -E2BIG) retval = -ENAMETOOLONG; if (retval < 0) goto out_unlock; @@ -6296,7 +6612,7 @@ int proc_cgroup_show(struct seq_file *m, struct pid_namespace *ns, retval = 0; out_unlock: spin_unlock_irq(&css_set_lock); - mutex_unlock(&cgroup_mutex); + rcu_read_unlock(); kfree(buf); out: return retval; @@ -6377,16 +6693,19 @@ static int cgroup_css_set_fork(struct kernel_clone_args *kargs) struct cgroup *dst_cgrp = NULL; struct css_set *cset; struct super_block *sb; - struct file *f; if (kargs->flags & CLONE_INTO_CGROUP) - mutex_lock(&cgroup_mutex); + cgroup_lock(); cgroup_threadgroup_change_begin(current); spin_lock_irq(&css_set_lock); cset = task_css_set(current); get_css_set(cset); + if (kargs->cgrp) + kargs->kill_seq = kargs->cgrp->kill_seq; + else + kargs->kill_seq = cset->dfl_cgrp->kill_seq; spin_unlock_irq(&css_set_lock); if (!(kargs->flags & CLONE_INTO_CGROUP)) { @@ -6394,14 +6713,14 @@ static int cgroup_css_set_fork(struct kernel_clone_args *kargs) return 0; } - f = fget_raw(kargs->cgroup); - if (!f) { + CLASS(fd_raw, f)(kargs->cgroup); + if (fd_empty(f)) { ret = -EBADF; goto err; } - sb = f->f_path.dentry->d_sb; + sb = fd_file(f)->f_path.dentry->d_sb; - dst_cgrp = cgroup_get_from_file(f); + dst_cgrp = cgroup_get_from_file(fd_file(f)); if (IS_ERR(dst_cgrp)) { ret = PTR_ERR(dst_cgrp); dst_cgrp = NULL; @@ -6449,15 +6768,12 @@ static int cgroup_css_set_fork(struct kernel_clone_args *kargs) } put_css_set(cset); - fput(f); kargs->cgrp = dst_cgrp; return ret; err: cgroup_threadgroup_change_end(current); - mutex_unlock(&cgroup_mutex); - if (f) - fput(f); + cgroup_unlock(); if (dst_cgrp) cgroup_put(dst_cgrp); put_css_set(cset); @@ -6476,19 +6792,18 @@ err: static void cgroup_css_set_put_fork(struct kernel_clone_args *kargs) __releases(&cgroup_threadgroup_rwsem) __releases(&cgroup_mutex) { - cgroup_threadgroup_change_end(current); - - if (kargs->flags & CLONE_INTO_CGROUP) { - struct cgroup *cgrp = kargs->cgrp; - struct css_set *cset = kargs->cset; + struct cgroup *cgrp = kargs->cgrp; + struct css_set *cset = kargs->cset; - mutex_unlock(&cgroup_mutex); + cgroup_threadgroup_change_end(current); - if (cset) { - put_css_set(cset); - kargs->cset = NULL; - } + if (cset) { + put_css_set(cset); + kargs->cset = NULL; + } + if (kargs->flags & CLONE_INTO_CGROUP) { + cgroup_unlock(); if (cgrp) { cgroup_put(cgrp); kargs->cgrp = NULL; @@ -6571,6 +6886,7 @@ void cgroup_post_fork(struct task_struct *child, struct kernel_clone_args *kargs) __releases(&cgroup_threadgroup_rwsem) __releases(&cgroup_mutex) { + unsigned int cgrp_kill_seq = 0; unsigned long cgrp_flags = 0; bool kill = false; struct cgroup_subsys *ss; @@ -6584,10 +6900,13 @@ void cgroup_post_fork(struct task_struct *child, /* init tasks are special, only link regular threads */ if (likely(child->pid)) { - if (kargs->cgrp) + if (kargs->cgrp) { cgrp_flags = kargs->cgrp->flags; - else + cgrp_kill_seq = kargs->cgrp->kill_seq; + } else { cgrp_flags = cset->dfl_cgrp->flags; + cgrp_kill_seq = cset->dfl_cgrp->kill_seq; + } WARN_ON_ONCE(!list_empty(&child->cg_list)); cset->nr_tasks++; @@ -6622,7 +6941,7 @@ void cgroup_post_fork(struct task_struct *child, * child down right after we finished preparing it for * userspace. */ - kill = test_bit(CGRP_KILL, &cgrp_flags); + kill = kargs->kill_seq != cgrp_kill_seq; } spin_unlock_irq(&css_set_lock); @@ -6653,40 +6972,101 @@ void cgroup_post_fork(struct task_struct *child, } /** - * cgroup_exit - detach cgroup from exiting task + * cgroup_task_exit - detach cgroup from exiting task * @tsk: pointer to task_struct of exiting process * * Description: Detach cgroup from @tsk. * */ -void cgroup_exit(struct task_struct *tsk) +void cgroup_task_exit(struct task_struct *tsk) { struct cgroup_subsys *ss; - struct css_set *cset; int i; - spin_lock_irq(&css_set_lock); + /* see cgroup_post_fork() for details */ + do_each_subsys_mask(ss, i, have_exit_callback) { + ss->exit(tsk); + } while_each_subsys_mask(); +} + +static void do_cgroup_task_dead(struct task_struct *tsk) +{ + struct css_set *cset; + unsigned long flags; + + spin_lock_irqsave(&css_set_lock, flags); WARN_ON_ONCE(list_empty(&tsk->cg_list)); cset = task_css_set(tsk); css_set_move_task(tsk, cset, NULL, false); - list_add_tail(&tsk->cg_list, &cset->dying_tasks); cset->nr_tasks--; + /* matches the signal->live check in css_task_iter_advance() */ + if (thread_group_leader(tsk) && atomic_read(&tsk->signal->live)) + list_add_tail(&tsk->cg_list, &cset->dying_tasks); + + if (dl_task(tsk)) + dec_dl_tasks_cs(tsk); WARN_ON_ONCE(cgroup_task_frozen(tsk)); if (unlikely(!(tsk->flags & PF_KTHREAD) && test_bit(CGRP_FREEZE, &task_dfl_cgroup(tsk)->flags))) cgroup_update_frozen(task_dfl_cgroup(tsk)); - spin_unlock_irq(&css_set_lock); + spin_unlock_irqrestore(&css_set_lock, flags); +} - /* see cgroup_post_fork() for details */ - do_each_subsys_mask(ss, i, have_exit_callback) { - ss->exit(tsk); - } while_each_subsys_mask(); +#ifdef CONFIG_PREEMPT_RT +/* + * cgroup_task_dead() is called from finish_task_switch() which doesn't allow + * scheduling even in RT. As the task_dead path requires grabbing css_set_lock, + * this lead to sleeping in the invalid context warning bug. css_set_lock is too + * big to become a raw_spinlock. The task_dead path doesn't need to run + * synchronously but can't be delayed indefinitely either as the dead task pins + * the cgroup and task_struct can be pinned indefinitely. Bounce through lazy + * irq_work to allow batching while ensuring timely completion. + */ +static DEFINE_PER_CPU(struct llist_head, cgrp_dead_tasks); +static DEFINE_PER_CPU(struct irq_work, cgrp_dead_tasks_iwork); + +static void cgrp_dead_tasks_iwork_fn(struct irq_work *iwork) +{ + struct llist_node *lnode; + struct task_struct *task, *next; + + lnode = llist_del_all(this_cpu_ptr(&cgrp_dead_tasks)); + llist_for_each_entry_safe(task, next, lnode, cg_dead_lnode) { + do_cgroup_task_dead(task); + put_task_struct(task); + } +} + +static void __init cgroup_rt_init(void) +{ + int cpu; + + for_each_possible_cpu(cpu) { + init_llist_head(per_cpu_ptr(&cgrp_dead_tasks, cpu)); + per_cpu(cgrp_dead_tasks_iwork, cpu) = + IRQ_WORK_INIT_LAZY(cgrp_dead_tasks_iwork_fn); + } +} + +void cgroup_task_dead(struct task_struct *task) +{ + get_task_struct(task); + llist_add(&task->cg_dead_lnode, this_cpu_ptr(&cgrp_dead_tasks)); + irq_work_queue(this_cpu_ptr(&cgrp_dead_tasks_iwork)); } +#else /* CONFIG_PREEMPT_RT */ +static void __init cgroup_rt_init(void) {} -void cgroup_release(struct task_struct *task) +void cgroup_task_dead(struct task_struct *task) +{ + do_cgroup_task_dead(task); +} +#endif /* CONFIG_PREEMPT_RT */ + +void cgroup_task_release(struct task_struct *task) { struct cgroup_subsys *ss; int ssid; @@ -6694,16 +7074,19 @@ void cgroup_release(struct task_struct *task) do_each_subsys_mask(ss, ssid, have_release_callback) { ss->release(task); } while_each_subsys_mask(); - - spin_lock_irq(&css_set_lock); - css_set_skip_task_iters(task_css_set(task), task); - list_del_init(&task->cg_list); - spin_unlock_irq(&css_set_lock); } -void cgroup_free(struct task_struct *task) +void cgroup_task_free(struct task_struct *task) { struct css_set *cset = task_css_set(task); + + if (!list_empty(&task->cg_list)) { + spin_lock_irq(&css_set_lock); + css_set_skip_task_iters(task_css_set(task), task); + list_del_init(&task->cg_list); + spin_unlock_irq(&css_set_lock); + } + put_css_set(cset); } @@ -6750,6 +7133,12 @@ static int __init enable_cgroup_debug(char *str) } __setup("cgroup_debug", enable_cgroup_debug); +static int __init cgroup_favordynmods_setup(char *str) +{ + return (kstrtobool(str, &have_favordynmods) == 0); +} +__setup("cgroup_favordynmods=", cgroup_favordynmods_setup); + /** * css_tryget_online_from_dir - get corresponding css from a cgroup dentry * @dentry: directory dentry of interest @@ -6855,16 +7244,11 @@ EXPORT_SYMBOL_GPL(cgroup_get_from_path); */ struct cgroup *cgroup_v1v2_get_from_fd(int fd) { - struct cgroup *cgrp; - struct file *f; - - f = fget_raw(fd); - if (!f) + CLASS(fd_raw, f)(fd); + if (fd_empty(f)) return ERR_PTR(-EBADF); - cgrp = cgroup_v1v2_get_from_file(f); - fput(f); - return cgrp; + return cgroup_v1v2_get_from_file(fd_file(f)); } /** @@ -7038,7 +7422,9 @@ static ssize_t features_show(struct kobject *kobj, struct kobj_attribute *attr, "nsdelegate\n" "favordynmods\n" "memory_localevents\n" - "memory_recursiveprot\n"); + "memory_recursiveprot\n" + "memory_hugetlb_accounting\n" + "pids_localevents\n"); } static struct kobj_attribute cgroup_features_attr = __ATTR_RO(features); |