diff options
Diffstat (limited to 'kernel/workqueue.c')
| -rw-r--r-- | kernel/workqueue.c | 848 |
1 files changed, 627 insertions, 221 deletions
diff --git a/kernel/workqueue.c b/kernel/workqueue.c index bfe030b443e2..33b721a9af02 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -41,6 +41,7 @@ #include <linux/mempolicy.h> #include <linux/freezer.h> #include <linux/debug_locks.h> +#include <linux/device/devres.h> #include <linux/lockdep.h> #include <linux/idr.h> #include <linux/jhash.h> @@ -117,6 +118,8 @@ enum wq_internal_consts { MAYDAY_INTERVAL = HZ / 10, /* and then every 100ms */ CREATE_COOLDOWN = HZ, /* time to breath after fail */ + RESCUER_BATCH = 16, /* process items per turn */ + /* * Rescue workers are used only on emergencies and shared by * all cpus. Give MIN_NICE. @@ -128,6 +131,14 @@ enum wq_internal_consts { WORKER_ID_LEN = 10 + WQ_NAME_LEN, /* "kworker/R-" + WQ_NAME_LEN */ }; +/* Layout of shards within one LLC pod */ +struct llc_shard_layout { + int nr_large_shards; /* number of large shards (cores_per_shard + 1) */ + int cores_per_shard; /* base number of cores per default shard */ + int nr_shards; /* total number of shards */ + /* nr_default shards = (nr_shards - nr_large_shards) */ +}; + /* * We don't want to trap softirq for too long. See MAX_SOFTIRQ_TIME and * MAX_SOFTIRQ_RESTART in kernel/softirq.c. These are macros because @@ -188,7 +199,7 @@ struct worker_pool { int id; /* I: pool ID */ unsigned int flags; /* L: flags */ - unsigned long watchdog_ts; /* L: watchdog timestamp */ + unsigned long last_progress_ts; /* L: last forward progress timestamp */ bool cpu_stall; /* WD: stalled cpu bound pool */ /* @@ -222,7 +233,9 @@ struct worker_pool { struct workqueue_attrs *attrs; /* I: worker attributes */ struct hlist_node hash_node; /* PL: unbound_pool_hash node */ int refcnt; /* PL: refcnt for unbound pools */ - +#ifdef CONFIG_PREEMPT_RT + spinlock_t cb_lock; /* BH worker cancel lock */ +#endif /* * Destruction of pool is RCU protected to allow dereferences * from get_work_pool(). @@ -284,6 +297,7 @@ struct pool_workqueue { struct list_head pending_node; /* LN: node on wq_node_nr_active->pending_pwqs */ struct list_head pwqs_node; /* WR: node on wq->pwqs */ struct list_head mayday_node; /* MD: node on wq->maydays */ + struct work_struct mayday_cursor; /* L: cursor on pool->worklist */ u64 stats[PWQ_NR_STATS]; @@ -399,11 +413,12 @@ struct work_offq_data { u32 flags; }; -static const char *wq_affn_names[WQ_AFFN_NR_TYPES] = { +static const char * const wq_affn_names[WQ_AFFN_NR_TYPES] = { [WQ_AFFN_DFL] = "default", [WQ_AFFN_CPU] = "cpu", [WQ_AFFN_SMT] = "smt", [WQ_AFFN_CACHE] = "cache", + [WQ_AFFN_CACHE_SHARD] = "cache_shard", [WQ_AFFN_NUMA] = "numa", [WQ_AFFN_SYSTEM] = "system", }; @@ -426,13 +441,16 @@ module_param_named(cpu_intensive_warning_thresh, wq_cpu_intensive_warning_thresh static bool wq_power_efficient = IS_ENABLED(CONFIG_WQ_POWER_EFFICIENT_DEFAULT); module_param_named(power_efficient, wq_power_efficient, bool, 0444); +static unsigned int wq_cache_shard_size = 8; +module_param_named(cache_shard_size, wq_cache_shard_size, uint, 0444); + static bool wq_online; /* can kworkers be created yet? */ static bool wq_topo_initialized __read_mostly = false; static struct kmem_cache *pwq_cache; static struct wq_pod_type wq_pod_types[WQ_AFFN_NR_TYPES]; -static enum wq_affn_scope wq_affn_dfl = WQ_AFFN_CACHE; +static enum wq_affn_scope wq_affn_dfl = WQ_AFFN_CACHE_SHARD; /* buf for wq_update_unbound_pod_attrs(), protected by CPU hotplug exclusion */ static struct workqueue_attrs *unbound_wq_update_pwq_attrs_buf; @@ -505,12 +523,16 @@ static struct kthread_worker *pwq_release_worker __ro_after_init; struct workqueue_struct *system_wq __ro_after_init; EXPORT_SYMBOL(system_wq); +struct workqueue_struct *system_percpu_wq __ro_after_init; +EXPORT_SYMBOL(system_percpu_wq); struct workqueue_struct *system_highpri_wq __ro_after_init; EXPORT_SYMBOL_GPL(system_highpri_wq); struct workqueue_struct *system_long_wq __ro_after_init; EXPORT_SYMBOL_GPL(system_long_wq); struct workqueue_struct *system_unbound_wq __ro_after_init; EXPORT_SYMBOL_GPL(system_unbound_wq); +struct workqueue_struct *system_dfl_wq __ro_after_init; +EXPORT_SYMBOL_GPL(system_dfl_wq); struct workqueue_struct *system_freezable_wq __ro_after_init; EXPORT_SYMBOL_GPL(system_freezable_wq); struct workqueue_struct *system_power_efficient_wq __ro_after_init; @@ -521,6 +543,8 @@ struct workqueue_struct *system_bh_wq; EXPORT_SYMBOL_GPL(system_bh_wq); struct workqueue_struct *system_bh_highpri_wq; EXPORT_SYMBOL_GPL(system_bh_highpri_wq); +struct workqueue_struct *system_dfl_long_wq __ro_after_init; +EXPORT_SYMBOL_GPL(system_dfl_long_wq); static int worker_thread(void *__worker); static void workqueue_sysfs_unregister(struct workqueue_struct *wq); @@ -535,12 +559,6 @@ static void show_one_worker_pool(struct worker_pool *pool); !lockdep_is_held(&wq_pool_mutex), \ "RCU or wq_pool_mutex should be held") -#define assert_rcu_or_wq_mutex_or_pool_mutex(wq) \ - RCU_LOCKDEP_WARN(!rcu_read_lock_any_held() && \ - !lockdep_is_held(&wq->mutex) && \ - !lockdep_is_held(&wq_pool_mutex), \ - "RCU, wq->mutex or wq_pool_mutex should be held") - #define for_each_bh_worker_pool(pool, cpu) \ for ((pool) = &per_cpu(bh_worker_pools, cpu)[0]; \ (pool) < &per_cpu(bh_worker_pools, cpu)[NR_STD_WORKER_POOLS]; \ @@ -686,7 +704,7 @@ EXPORT_SYMBOL_GPL(destroy_work_on_stack); void destroy_delayed_work_on_stack(struct delayed_work *work) { - destroy_timer_on_stack(&work->timer); + timer_destroy_on_stack(&work->timer); debug_object_free(&work->work, &work_debug_descr); } EXPORT_SYMBOL_GPL(destroy_delayed_work_on_stack); @@ -1120,6 +1138,12 @@ static struct worker *find_worker_executing_work(struct worker_pool *pool, return NULL; } +static void mayday_cursor_func(struct work_struct *work) +{ + /* should not be processed, only for marking position */ + BUG(); +} + /** * move_linked_works - move linked works to a list * @work: start of series of works to be scheduled @@ -1182,6 +1206,16 @@ static bool assign_work(struct work_struct *work, struct worker *worker, lockdep_assert_held(&pool->lock); + /* The cursor work should not be processed */ + if (unlikely(work->func == mayday_cursor_func)) { + /* only worker_thread() can possibly take this branch */ + WARN_ON_ONCE(worker->rescue_wq); + if (nextp) + *nextp = list_next_entry(work, entry); + list_del_init(&work->entry); + return false; + } + /* * A single work shouldn't be executed concurrently by multiple workers. * __queue_work() ensures that @work doesn't jump to a different pool @@ -1678,7 +1712,7 @@ static void __pwq_activate_work(struct pool_workqueue *pwq, WARN_ON_ONCE(!(*wdb & WORK_STRUCT_INACTIVE)); trace_workqueue_activate_work(work); if (list_empty(&pwq->pool->worklist)) - pwq->pool->watchdog_ts = jiffies; + pwq->pool->last_progress_ts = jiffies; move_linked_works(work, &pwq->pool->worklist, NULL); __clear_bit(WORK_STRUCT_INACTIVE_BIT, wdb); } @@ -1686,17 +1720,14 @@ static void __pwq_activate_work(struct pool_workqueue *pwq, static bool tryinc_node_nr_active(struct wq_node_nr_active *nna) { int max = READ_ONCE(nna->max); + int old = atomic_read(&nna->nr); - while (true) { - int old, tmp; - - old = atomic_read(&nna->nr); + do { if (old >= max) return false; - tmp = atomic_cmpxchg_relaxed(&nna->nr, old, old + 1); - if (tmp == old) - return true; - } + } while (!atomic_try_cmpxchg_relaxed(&nna->nr, &old, old + 1)); + + return true; } /** @@ -1833,8 +1864,20 @@ static void unplug_oldest_pwq(struct workqueue_struct *wq) raw_spin_lock_irq(&pwq->pool->lock); if (pwq->plugged) { pwq->plugged = false; - if (pwq_activate_first_inactive(pwq, true)) + if (pwq_activate_first_inactive(pwq, true)) { + /* + * While plugged, queueing skips activation which + * includes bumping the nr_active count and adding the + * pwq to nna->pending_pwqs if the count can't be + * obtained. We need to restore both for the pwq being + * unplugged. The first call activates the first + * inactive work item and the second, if there are more + * inactive, puts the pwq on pending_pwqs. + */ + pwq_activate_first_inactive(pwq, false); + kick_pool(pwq->pool); + } } raw_spin_unlock_irq(&pwq->pool->lock); } @@ -2057,11 +2100,11 @@ static int try_to_grab_pending(struct work_struct *work, u32 cflags, struct delayed_work *dwork = to_delayed_work(work); /* - * dwork->timer is irqsafe. If del_timer() fails, it's + * dwork->timer is irqsafe. If timer_delete() fails, it's * guaranteed that the timer is not queued anywhere and not * running on the local CPU. */ - if (likely(del_timer(&dwork->timer))) + if (likely(timer_delete(&dwork->timer))) return 1; } @@ -2221,12 +2264,9 @@ static int wq_select_unbound_cpu(int cpu) } new_cpu = __this_cpu_read(wq_rr_cpu_last); - new_cpu = cpumask_next_and(new_cpu, wq_unbound_cpumask, cpu_online_mask); - if (unlikely(new_cpu >= nr_cpu_ids)) { - new_cpu = cpumask_first_and(wq_unbound_cpumask, cpu_online_mask); - if (unlikely(new_cpu >= nr_cpu_ids)) - return cpu; - } + new_cpu = cpumask_next_and_wrap(new_cpu, wq_unbound_cpumask, cpu_online_mask); + if (unlikely(new_cpu >= nr_cpu_ids)) + return cpu; __this_cpu_write(wq_rr_cpu_last, new_cpu); return new_cpu; @@ -2256,6 +2296,18 @@ static void __queue_work(int cpu, struct workqueue_struct *wq, if (unlikely(wq->flags & (__WQ_DESTROYING | __WQ_DRAINING) && WARN_ONCE(!is_chained_work(wq), "workqueue: cannot queue %ps on wq %s\n", work->func, wq->name))) { + struct work_offq_data offqd; + + /* + * State on entry: PENDING is set, work is off-queue (no + * insert_work() has run). + * + * Returning without clearing PENDING would leave the work + * in a weird state (PENDING=1, PWQ=0, entry empty) + */ + work_offqd_unpack(&offqd, *work_data_bits(work)); + set_work_pool_and_clear_pending(work, offqd.pool_id, + work_offqd_pack_flags(&offqd)); return; } rcu_read_lock(); @@ -2335,7 +2387,7 @@ retry: */ if (list_empty(&pwq->inactive_works) && pwq_tryinc_nr_active(pwq, false)) { if (list_empty(&pool->worklist)) - pool->watchdog_ts = jiffies; + pool->last_progress_ts = jiffies; trace_workqueue_activate_work(work); insert_work(pwq, work, &pool->worklist, work_flags); @@ -2481,7 +2533,7 @@ EXPORT_SYMBOL_GPL(queue_work_node); void delayed_work_timer_fn(struct timer_list *t) { - struct delayed_work *dwork = from_timer(dwork, t, timer); + struct delayed_work *dwork = timer_container_of(dwork, t, timer); /* should have been called from irqsafe timer with irq already off */ __queue_work(dwork->cpu, dwork->wq, &dwork->work); @@ -2494,7 +2546,6 @@ static void __queue_delayed_work(int cpu, struct workqueue_struct *wq, struct timer_list *timer = &dwork->timer; struct work_struct *work = &dwork->work; - WARN_ON_ONCE(!wq); WARN_ON_ONCE(timer->function != delayed_work_timer_fn); WARN_ON_ONCE(timer_pending(timer)); WARN_ON_ONCE(!list_empty(&work->entry)); @@ -2909,7 +2960,7 @@ static void set_worker_dying(struct worker *worker, struct list_head *list) */ static void idle_worker_timeout(struct timer_list *t) { - struct worker_pool *pool = from_timer(pool, t, idle_timer); + struct worker_pool *pool = timer_container_of(pool, t, idle_timer); bool do_cull = false; if (work_pending(&pool->idle_cull_work)) @@ -2932,7 +2983,7 @@ static void idle_worker_timeout(struct timer_list *t) raw_spin_unlock_irq(&pool->lock); if (do_cull) - queue_work(system_unbound_wq, &pool->idle_cull_work); + queue_work(system_dfl_wq, &pool->idle_cull_work); } /** @@ -2982,9 +3033,8 @@ static void idle_cull_fn(struct work_struct *work) reap_dying_workers(&cull_list); } -static void send_mayday(struct work_struct *work) +static void send_mayday(struct pool_workqueue *pwq) { - struct pool_workqueue *pwq = get_work_pwq(work); struct workqueue_struct *wq = pwq->wq; lockdep_assert_held(&wq_mayday_lock); @@ -3008,7 +3058,7 @@ static void send_mayday(struct work_struct *work) static void pool_mayday_timeout(struct timer_list *t) { - struct worker_pool *pool = from_timer(pool, t, mayday_timer); + struct worker_pool *pool = timer_container_of(pool, t, mayday_timer); struct work_struct *work; raw_spin_lock_irq(&pool->lock); @@ -3022,7 +3072,7 @@ static void pool_mayday_timeout(struct timer_list *t) * rescuers. */ list_for_each_entry(work, &pool->worklist, entry) - send_mayday(work); + send_mayday(get_work_pwq(work)); } raw_spin_unlock(&wq_mayday_lock); @@ -3069,7 +3119,7 @@ restart: break; } - del_timer_sync(&pool->mayday_timer); + timer_delete_sync(&pool->mayday_timer); raw_spin_lock_irq(&pool->lock); /* * This is necessary even after a new worker was just successfully @@ -3080,6 +3130,31 @@ restart: goto restart; } +#ifdef CONFIG_PREEMPT_RT +static void worker_lock_callback(struct worker_pool *pool) +{ + spin_lock(&pool->cb_lock); +} + +static void worker_unlock_callback(struct worker_pool *pool) +{ + spin_unlock(&pool->cb_lock); +} + +static void workqueue_callback_cancel_wait_running(struct worker_pool *pool) +{ + spin_lock(&pool->cb_lock); + spin_unlock(&pool->cb_lock); +} + +#else + +static void worker_lock_callback(struct worker_pool *pool) { } +static void worker_unlock_callback(struct worker_pool *pool) { } +static void workqueue_callback_cancel_wait_running(struct worker_pool *pool) { } + +#endif + /** * manage_workers - manage worker pool * @worker: self @@ -3167,6 +3242,7 @@ __acquires(&pool->lock) worker->current_pwq = pwq; if (worker->task) worker->current_at = worker->task->se.sum_exec_runtime; + worker->current_start = jiffies; work_data = *work_data_bits(work); worker->current_color = get_work_color(work_data); @@ -3241,7 +3317,7 @@ __acquires(&pool->lock) * point will only record its address. */ trace_workqueue_execute_end(work, worker->current_func); - pwq->stats[PWQ_STAT_COMPLETED]++; + lock_map_release(&lockdep_map); if (!bh_draining) lock_map_release(pwq->wq->lockdep_map); @@ -3272,6 +3348,8 @@ __acquires(&pool->lock) raw_spin_lock_irq(&pool->lock); + pwq->stats[PWQ_STAT_COMPLETED]++; + /* * In addition to %WQ_CPU_INTENSIVE, @worker may also have been marked * CPU intensive by wq_worker_tick() if @work hogged CPU longer than @@ -3313,7 +3391,7 @@ static void process_scheduled_works(struct worker *worker) while ((work = list_first_entry_or_null(&worker->scheduled, struct work_struct, entry))) { if (first) { - worker->pool->watchdog_ts = jiffies; + worker->pool->last_progress_ts = jiffies; first = false; } process_one_work(worker, work); @@ -3416,6 +3494,62 @@ sleep: goto woke_up; } +static bool assign_rescuer_work(struct pool_workqueue *pwq, struct worker *rescuer) +{ + struct worker_pool *pool = pwq->pool; + struct work_struct *cursor = &pwq->mayday_cursor; + struct work_struct *work, *n; + + /* have work items to rescue? */ + if (!pwq->nr_active) + return false; + + /* need rescue? */ + if (!need_to_create_worker(pool)) { + /* + * The pool has idle workers and doesn't need the rescuer, so it + * could simply return false here. + * + * However, the memory pressure might not be fully relieved. + * In PERCPU pool with concurrency enabled, having idle workers + * does not necessarily mean memory pressure is gone; it may + * simply mean regular workers have woken up, completed their + * work, and gone idle again due to concurrency limits. + * + * In this case, those working workers may later sleep again, + * the pool may run out of idle workers, and it will have to + * allocate new ones and wait for the timer to send mayday, + * causing unnecessary delay - especially if memory pressure + * was never resolved throughout. + * + * Do more work if memory pressure is still on to reduce + * relapse, using (pool->flags & POOL_MANAGER_ACTIVE), though + * not precisely, unless there are other PWQs needing help. + */ + if (!(pool->flags & POOL_MANAGER_ACTIVE) || + !list_empty(&pwq->wq->maydays)) + return false; + } + + /* search from the start or cursor if available */ + if (list_empty(&cursor->entry)) + work = list_first_entry(&pool->worklist, struct work_struct, entry); + else + work = list_next_entry(cursor, entry); + + /* find the next work item to rescue */ + list_for_each_entry_safe_from(work, n, &pool->worklist, entry) { + if (get_work_pwq(work) == pwq && assign_work(work, rescuer, &n)) { + pwq->stats[PWQ_STAT_RESCUED]++; + /* put the cursor for next search */ + list_move_tail(&cursor->entry, &n->entry); + return true; + } + } + + return false; +} + /** * rescuer_thread - the rescuer thread function * @__rescuer: self @@ -3470,7 +3604,7 @@ repeat: struct pool_workqueue *pwq = list_first_entry(&wq->maydays, struct pool_workqueue, mayday_node); struct worker_pool *pool = pwq->pool; - struct work_struct *work, *n; + unsigned int count = 0; __set_current_state(TASK_RUNNING); list_del_init(&pwq->mayday_node); @@ -3481,43 +3615,29 @@ repeat: raw_spin_lock_irq(&pool->lock); - /* - * Slurp in all works issued via this workqueue and - * process'em. - */ WARN_ON_ONCE(!list_empty(&rescuer->scheduled)); - list_for_each_entry_safe(work, n, &pool->worklist, entry) { - if (get_work_pwq(work) == pwq && - assign_work(work, rescuer, &n)) - pwq->stats[PWQ_STAT_RESCUED]++; - } - if (!list_empty(&rescuer->scheduled)) { + while (assign_rescuer_work(pwq, rescuer)) { process_scheduled_works(rescuer); /* - * The above execution of rescued work items could - * have created more to rescue through - * pwq_activate_first_inactive() or chained - * queueing. Let's put @pwq back on mayday list so - * that such back-to-back work items, which may be - * being used to relieve memory pressure, don't - * incur MAYDAY_INTERVAL delay inbetween. + * If the per-turn work item limit is reached and other + * PWQs are in mayday, requeue mayday for this PWQ and + * let the rescuer handle the other PWQs first. */ - if (pwq->nr_active && need_to_create_worker(pool)) { + if (++count > RESCUER_BATCH && !list_empty(&pwq->wq->maydays) && + pwq->nr_active && need_to_create_worker(pool)) { raw_spin_lock(&wq_mayday_lock); - /* - * Queue iff we aren't racing destruction - * and somebody else hasn't queued it already. - */ - if (wq->rescuer && list_empty(&pwq->mayday_node)) { - get_pwq(pwq); - list_add_tail(&pwq->mayday_node, &wq->maydays); - } + send_mayday(pwq); raw_spin_unlock(&wq_mayday_lock); + break; } } + /* The cursor can not be left behind without the rescuer watching it. */ + if (!list_empty(&pwq->mayday_cursor.entry) && list_empty(&pwq->mayday_node)) + list_del_init(&pwq->mayday_cursor.entry); + /* * Leave this pool. Notify regular workers; otherwise, we end up * with 0 concurrency and stalling the execution. @@ -3557,6 +3677,7 @@ static void bh_worker(struct worker *worker) int nr_restarts = BH_WORKER_RESTARTS; unsigned long end = jiffies + BH_WORKER_JIFFIES; + worker_lock_callback(pool); raw_spin_lock_irq(&pool->lock); worker_leave_idle(worker); @@ -3585,6 +3706,7 @@ done: worker_enter_idle(worker); kick_pool(pool); raw_spin_unlock_irq(&pool->lock); + worker_unlock_callback(pool); } /* @@ -4222,17 +4344,17 @@ static bool __flush_work(struct work_struct *work, bool from_cancel) (data & WORK_OFFQ_BH)) { /* * On RT, prevent a live lock when %current preempted - * soft interrupt processing or prevents ksoftirqd from - * running by keeping flipping BH. If the BH work item - * runs on a different CPU then this has no effect other - * than doing the BH disable/enable dance for nothing. - * This is copied from - * kernel/softirq.c::tasklet_unlock_spin_wait(). + * soft interrupt processing by blocking on lock which + * is owned by the thread invoking the callback. */ while (!try_wait_for_completion(&barr.done)) { if (IS_ENABLED(CONFIG_PREEMPT_RT)) { - local_bh_disable(); - local_bh_enable(); + struct worker_pool *pool; + + guard(rcu)(); + pool = get_work_pool(work); + if (pool) + workqueue_callback_cancel_wait_running(pool); } else { cpu_relax(); } @@ -4281,7 +4403,7 @@ EXPORT_SYMBOL_GPL(flush_work); bool flush_delayed_work(struct delayed_work *dwork) { local_irq_disable(); - if (del_timer_sync(&dwork->timer)) + if (timer_delete_sync(&dwork->timer)) __queue_work(dwork->cpu, dwork->wq, &dwork->work); local_irq_enable(); return flush_work(&dwork->work); @@ -4627,11 +4749,11 @@ void free_workqueue_attrs(struct workqueue_attrs *attrs) * * Return: The allocated new workqueue_attr on success. %NULL on failure. */ -struct workqueue_attrs *alloc_workqueue_attrs(void) +struct workqueue_attrs *alloc_workqueue_attrs_noprof(void) { struct workqueue_attrs *attrs; - attrs = kzalloc(sizeof(*attrs), GFP_KERNEL); + attrs = kzalloc_obj(*attrs); if (!attrs) goto fail; if (!alloc_cpumask_var(&attrs->cpumask, GFP_KERNEL)) @@ -4767,7 +4889,7 @@ static int init_worker_pool(struct worker_pool *pool) pool->cpu = -1; pool->node = NUMA_NO_NODE; pool->flags |= POOL_DISASSOCIATED; - pool->watchdog_ts = jiffies; + pool->last_progress_ts = jiffies; INIT_LIST_HEAD(&pool->worklist); INIT_LIST_HEAD(&pool->idle_list); hash_init(pool->busy_hash); @@ -4782,6 +4904,9 @@ static int init_worker_pool(struct worker_pool *pool) ida_init(&pool->worker_ida); INIT_HLIST_NODE(&pool->hash_node); pool->refcnt = 1; +#ifdef CONFIG_PREEMPT_RT + spin_lock_init(&pool->cb_lock); +#endif /* shouldn't fail above this point */ pool->attrs = alloc_workqueue_attrs(); @@ -4984,9 +5109,9 @@ static void put_unbound_pool(struct worker_pool *pool) reap_dying_workers(&cull_list); /* shut down the timers */ - del_timer_sync(&pool->idle_timer); + timer_delete_sync(&pool->idle_timer); cancel_work_sync(&pool->idle_cull_work); - del_timer_sync(&pool->mayday_timer); + timer_delete_sync(&pool->mayday_timer); /* RCU protected to allow dereferences from get_work_pool() */ call_rcu(&pool->rcu, rcu_free_pool); @@ -5131,6 +5256,19 @@ static void init_pwq(struct pool_workqueue *pwq, struct workqueue_struct *wq, INIT_LIST_HEAD(&pwq->pwqs_node); INIT_LIST_HEAD(&pwq->mayday_node); kthread_init_work(&pwq->release_work, pwq_release_workfn); + + /* + * Set the dummy cursor work with valid function and get_work_pwq(). + * + * The cursor work should only be in the pwq->pool->worklist, and + * should not be treated as a processable work item. + * + * WORK_STRUCT_PENDING and WORK_STRUCT_INACTIVE just make it less + * surprise for kernel debugging tools and reviewers. + */ + INIT_WORK(&pwq->mayday_cursor, mayday_cursor_func); + atomic_long_set(&pwq->mayday_cursor.data, (unsigned long)pwq | + WORK_STRUCT_PENDING | WORK_STRUCT_PWQ | WORK_STRUCT_INACTIVE); } /* sync @pwq with the current state of its associated wq and link it */ @@ -5271,7 +5409,7 @@ apply_wqattrs_prepare(struct workqueue_struct *wq, attrs->affn_scope >= WQ_AFFN_NR_TYPES)) return ERR_PTR(-EINVAL); - ctx = kzalloc(struct_size(ctx, pwq_tbl, nr_cpu_ids), GFP_KERNEL); + ctx = kzalloc_flex(*ctx, pwq_tbl, nr_cpu_ids); new_attrs = alloc_workqueue_attrs(); if (!ctx || !new_attrs) @@ -5344,11 +5482,6 @@ static void apply_wqattrs_commit(struct apply_wqattrs_ctx *ctx) /* update node_nr_active->max */ wq_update_node_max_active(ctx->wq, -1); - /* rescuer needs to respect wq cpumask changes */ - if (ctx->wq->rescuer) - set_cpus_allowed_ptr(ctx->wq->rescuer->task, - unbound_effective_cpumask(ctx->wq)); - mutex_unlock(&ctx->wq->mutex); } @@ -5521,15 +5654,25 @@ static int alloc_and_link_pwqs(struct workqueue_struct *wq) ret = apply_workqueue_attrs_locked(wq, unbound_std_wq_attrs[highpri]); } - return ret; + if (ret) + goto enomem; + return 0; enomem: if (wq->cpu_pwq) { for_each_possible_cpu(cpu) { struct pool_workqueue *pwq = *per_cpu_ptr(wq->cpu_pwq, cpu); - if (pwq) + if (pwq) { + /* + * Unlink pwq from wq->pwqs since link_pwq() + * may have already added it. wq->mutex is not + * needed as the wq has not been published yet. + */ + if (!list_empty(&pwq->pwqs_node)) + list_del_rcu(&pwq->pwqs_node); kmem_cache_free(pwq_cache, pwq); + } } free_percpu(wq->cpu_pwq); wq->cpu_pwq = NULL; @@ -5582,10 +5725,13 @@ static int init_rescuer(struct workqueue_struct *wq) } wq->rescuer = rescuer; - if (wq->flags & WQ_UNBOUND) - kthread_bind_mask(rescuer->task, unbound_effective_cpumask(wq)); + + /* initial cpumask is consistent with the detached rescuer and unbind_worker() */ + if (cpumask_intersects(wq_unbound_cpumask, cpu_active_mask)) + kthread_bind_mask(rescuer->task, wq_unbound_cpumask); else kthread_bind_mask(rescuer->task, cpu_possible_mask); + wake_up_process(rescuer->task); return 0; @@ -5680,12 +5826,12 @@ static struct workqueue_struct *__alloc_workqueue(const char *fmt, else wq_size = sizeof(*wq); - wq = kzalloc(wq_size, GFP_KERNEL); + wq = kzalloc_noprof(wq_size, GFP_KERNEL); if (!wq) return NULL; if (flags & WQ_UNBOUND) { - wq->unbound_attrs = alloc_workqueue_attrs(); + wq->unbound_attrs = alloc_workqueue_attrs_noprof(); if (!wq->unbound_attrs) goto err_free_wq; } @@ -5774,25 +5920,63 @@ err_destroy: return NULL; } +__printf(1, 0) +static struct workqueue_struct *alloc_workqueue_va(const char *fmt, + unsigned int flags, + int max_active, + va_list args) +{ + struct workqueue_struct *wq; + + wq = __alloc_workqueue(fmt, flags, max_active, args); + if (wq) + wq_init_lockdep(wq); + + return wq; +} + __printf(1, 4) -struct workqueue_struct *alloc_workqueue(const char *fmt, - unsigned int flags, - int max_active, ...) +struct workqueue_struct *alloc_workqueue_noprof(const char *fmt, + unsigned int flags, + int max_active, ...) { struct workqueue_struct *wq; va_list args; va_start(args, max_active); - wq = __alloc_workqueue(fmt, flags, max_active, args); + wq = alloc_workqueue_va(fmt, flags, max_active, args); + va_end(args); + + return wq; +} +EXPORT_SYMBOL_GPL(alloc_workqueue_noprof); + +static void devm_workqueue_release(void *res) +{ + destroy_workqueue(res); +} + +__printf(2, 5) struct workqueue_struct * +devm_alloc_workqueue_noprof(struct device *dev, const char *fmt, + unsigned int flags, int max_active, ...) +{ + struct workqueue_struct *wq; + va_list args; + int ret; + + va_start(args, max_active); + wq = alloc_workqueue_va(fmt, flags, max_active, args); va_end(args); if (!wq) return NULL; - wq_init_lockdep(wq); + ret = devm_add_action_or_reset(dev, devm_workqueue_release, wq); + if (ret) + return NULL; return wq; } -EXPORT_SYMBOL_GPL(alloc_workqueue); +EXPORT_SYMBOL_GPL(devm_alloc_workqueue_noprof); #ifdef CONFIG_LOCKDEP __printf(1, 5) @@ -5837,6 +6021,17 @@ static bool pwq_busy(struct pool_workqueue *pwq) * @wq: target workqueue * * Safely destroy a workqueue. All work currently pending will be done first. + * + * This function does NOT guarantee that non-pending work that has been + * submitted with queue_delayed_work() and similar functions will be done + * before destroying the workqueue. The fundamental problem is that, currently, + * the workqueue has no way of accessing non-pending delayed_work. delayed_work + * is only linked on the timer-side. All delayed_work must, therefore, be + * canceled before calling this function. + * + * TODO: It would be better if the problem described above wouldn't exist and + * destroy_workqueue() would cleanly cancel all pending and non-pending + * delayed_work. */ void destroy_workqueue(struct workqueue_struct *wq) { @@ -5859,16 +6054,10 @@ void destroy_workqueue(struct workqueue_struct *wq) /* kill rescuer, if sanity checks fail, leave it w/o rescuer */ if (wq->rescuer) { - struct worker *rescuer = wq->rescuer; - - /* this prevents new queueing */ - raw_spin_lock_irq(&wq_mayday_lock); - wq->rescuer = NULL; - raw_spin_unlock_irq(&wq_mayday_lock); - /* rescuer will empty maydays list before exiting */ - kthread_stop(rescuer->task); - kfree(rescuer); + kthread_stop(wq->rescuer->task); + kfree(wq->rescuer); + wq->rescuer = NULL; } /* @@ -6035,7 +6224,6 @@ bool workqueue_congested(int cpu, struct workqueue_struct *wq) struct pool_workqueue *pwq; bool ret; - rcu_read_lock(); preempt_disable(); if (cpu == WORK_CPU_UNBOUND) @@ -6045,7 +6233,6 @@ bool workqueue_congested(int cpu, struct workqueue_struct *wq) ret = !list_empty(&pwq->inactive_works); preempt_enable(); - rcu_read_unlock(); return ret; } @@ -6174,7 +6361,7 @@ static void pr_cont_worker_id(struct worker *worker) { struct worker_pool *pool = worker->pool; - if (pool->flags & WQ_BH) + if (pool->flags & POOL_BH) pr_cont("bh%s", pool->attrs->nice == HIGHPRI_NICE_LEVEL ? "-hi" : ""); else @@ -6259,6 +6446,8 @@ static void show_pwq(struct pool_workqueue *pwq) pr_cont(" %s", comma ? "," : ""); pr_cont_worker_id(worker); pr_cont(":%ps", worker->current_func); + pr_cont(" for %us", + jiffies_to_msecs(jiffies - worker->current_start) / 1000); list_for_each_entry(work, &worker->scheduled, entry) pr_cont_work(false, work, &pcws); pr_cont_work_flush(comma, (work_func_t)-1L, &pcws); @@ -6362,7 +6551,7 @@ static void show_one_worker_pool(struct worker_pool *pool) /* How long the first pending work is waiting for a worker. */ if (!list_empty(&pool->worklist)) - hung = jiffies_to_msecs(jiffies - pool->watchdog_ts) / 1000; + hung = jiffies_to_msecs(jiffies - pool->last_progress_ts) / 1000; /* * Defer printing to avoid deadlocks in console drivers that @@ -6757,31 +6946,6 @@ long work_on_cpu_key(int cpu, long (*fn)(void *), return wfc.ret; } EXPORT_SYMBOL_GPL(work_on_cpu_key); - -/** - * work_on_cpu_safe_key - run a function in thread context on a particular cpu - * @cpu: the cpu to run on - * @fn: the function to run - * @arg: the function argument - * @key: The lock class key for lock debugging purposes - * - * Disables CPU hotplug and calls work_on_cpu(). The caller must not hold - * any locks which would prevent @fn from completing. - * - * Return: The value @fn returns. - */ -long work_on_cpu_safe_key(int cpu, long (*fn)(void *), - void *arg, struct lock_class_key *key) -{ - long ret = -ENODEV; - - cpus_read_lock(); - if (cpu_online(cpu)) - ret = work_on_cpu_key(cpu, fn, arg, key); - cpus_read_unlock(); - return ret; -} -EXPORT_SYMBOL_GPL(work_on_cpu_safe_key); #endif /* CONFIG_SMP */ #ifdef CONFIG_FREEZER @@ -6921,21 +7085,42 @@ static int workqueue_apply_unbound_cpumask(const cpumask_var_t unbound_cpumask) } if (!ret) { + int cpu; + struct worker_pool *pool; + struct worker *worker; + mutex_lock(&wq_pool_attach_mutex); cpumask_copy(wq_unbound_cpumask, unbound_cpumask); + /* rescuer needs to respect cpumask changes when it is not attached */ + list_for_each_entry(wq, &workqueues, list) { + if (wq->rescuer && !wq->rescuer->pool) + unbind_worker(wq->rescuer); + } + /* DISASSOCIATED worker needs to respect wq_unbound_cpumask */ + for_each_possible_cpu(cpu) { + for_each_cpu_worker_pool(pool, cpu) { + if (!(pool->flags & POOL_DISASSOCIATED)) + continue; + for_each_pool_worker(worker, pool) + unbind_worker(worker); + } + } mutex_unlock(&wq_pool_attach_mutex); } return ret; } /** - * workqueue_unbound_exclude_cpumask - Exclude given CPUs from unbound cpumask - * @exclude_cpumask: the cpumask to be excluded from wq_unbound_cpumask + * workqueue_unbound_housekeeping_update - Propagate housekeeping cpumask update + * @hk: the new housekeeping cpumask * - * This function can be called from cpuset code to provide a set of isolated - * CPUs that should be excluded from wq_unbound_cpumask. + * Update the unbound workqueue cpumask on top of the new housekeeping cpumask such + * that the effective unbound affinity is the intersection of the new housekeeping + * with the requested affinity set via nohz_full=/isolcpus= or sysfs. + * + * Return: 0 on success and -errno on failure. */ -int workqueue_unbound_exclude_cpumask(cpumask_var_t exclude_cpumask) +int workqueue_unbound_housekeeping_update(const struct cpumask *hk) { cpumask_var_t cpumask; int ret = 0; @@ -6948,17 +7133,17 @@ int workqueue_unbound_exclude_cpumask(cpumask_var_t exclude_cpumask) /* * If the operation fails, it will fall back to * wq_requested_unbound_cpumask which is initially set to - * (HK_TYPE_WQ ∩ HK_TYPE_DOMAIN) house keeping mask and rewritten + * HK_TYPE_DOMAIN house keeping mask and rewritten * by any subsequent write to workqueue/cpumask sysfs file. */ - if (!cpumask_andnot(cpumask, wq_requested_unbound_cpumask, exclude_cpumask)) + if (!cpumask_and(cpumask, wq_requested_unbound_cpumask, hk)) cpumask_copy(cpumask, wq_requested_unbound_cpumask); if (!cpumask_equal(cpumask, wq_unbound_cpumask)) ret = workqueue_apply_unbound_cpumask(cpumask); /* Save the current isolated cpumask & export it via sysfs */ if (!ret) - cpumask_copy(wq_isolated_cpumask, exclude_cpumask); + cpumask_andnot(wq_isolated_cpumask, cpu_possible_mask, hk); mutex_unlock(&wq_pool_mutex); free_cpumask_var(cpumask); @@ -6967,13 +7152,7 @@ int workqueue_unbound_exclude_cpumask(cpumask_var_t exclude_cpumask) static int parse_affn_scope(const char *val) { - int i; - - for (i = 0; i < ARRAY_SIZE(wq_affn_names); i++) { - if (!strncasecmp(val, wq_affn_names[i], strlen(wq_affn_names[i]))) - return i; - } - return -EINVAL; + return sysfs_match_string(wq_affn_names, val); } static int wq_affn_dfl_set(const char *val, const struct kernel_param *kp) @@ -7080,7 +7259,26 @@ static struct attribute *wq_sysfs_attrs[] = { &dev_attr_max_active.attr, NULL, }; -ATTRIBUTE_GROUPS(wq_sysfs); + +static umode_t wq_sysfs_is_visible(struct kobject *kobj, struct attribute *a, int n) +{ + struct device *dev = kobj_to_dev(kobj); + struct workqueue_struct *wq = dev_to_wq(dev); + + /* + * Adjusting max_active breaks ordering guarantee. Changing it has no + * effect on BH worker. Limit max_active to RO in such case. + */ + if (wq->flags & (WQ_BH | __WQ_ORDERED)) + return 0444; + return a->mode; +} + +static const struct attribute_group wq_sysfs_group = { + .is_visible = wq_sysfs_is_visible, + .attrs = wq_sysfs_attrs, +}; +__ATTRIBUTE_GROUPS(wq_sysfs); static ssize_t wq_nice_show(struct device *dev, struct device_attribute *attr, char *buf) @@ -7383,14 +7581,7 @@ int workqueue_sysfs_register(struct workqueue_struct *wq) struct wq_device *wq_dev; int ret; - /* - * Adjusting max_active breaks ordering guarantee. Disallow exposing - * ordered workqueues. - */ - if (WARN_ON(wq->flags & __WQ_ORDERED)) - return -EINVAL; - - wq->wq_dev = wq_dev = kzalloc(sizeof(*wq_dev), GFP_KERNEL); + wq->wq_dev = wq_dev = kzalloc_obj(*wq_dev); if (!wq_dev) return -ENOMEM; @@ -7475,16 +7666,20 @@ static struct timer_list wq_watchdog_timer; static unsigned long wq_watchdog_touched = INITIAL_JIFFIES; static DEFINE_PER_CPU(unsigned long, wq_watchdog_touched_cpu) = INITIAL_JIFFIES; -static unsigned int wq_panic_on_stall; +static unsigned int wq_panic_on_stall = CONFIG_BOOTPARAM_WQ_STALL_PANIC; module_param_named(panic_on_stall, wq_panic_on_stall, uint, 0644); +static unsigned int wq_panic_on_stall_time; +module_param_named(panic_on_stall_time, wq_panic_on_stall_time, uint, 0644); +MODULE_PARM_DESC(panic_on_stall_time, "Panic if stall exceeds this many seconds (0=disabled)"); + /* * Show workers that might prevent the processing of pending work items. - * The only candidates are CPU-bound workers in the running state. - * Pending work items should be handled by another idle worker - * in all other situations. + * A busy worker that is not running on the CPU (e.g. sleeping in + * wait_event_idle() with PF_WQ_WORKER cleared) can stall the pool just as + * effectively as a CPU-bound one, so dump every in-flight worker. */ -static void show_cpu_pool_hog(struct worker_pool *pool) +static void show_cpu_pool_busy_workers(struct worker_pool *pool) { struct worker *worker; unsigned long irq_flags; @@ -7493,50 +7688,59 @@ static void show_cpu_pool_hog(struct worker_pool *pool) raw_spin_lock_irqsave(&pool->lock, irq_flags); hash_for_each(pool->busy_hash, bkt, worker, hentry) { - if (task_is_running(worker->task)) { - /* - * Defer printing to avoid deadlocks in console - * drivers that queue work while holding locks - * also taken in their write paths. - */ - printk_deferred_enter(); + /* + * Defer printing to avoid deadlocks in console + * drivers that queue work while holding locks + * also taken in their write paths. + */ + printk_deferred_enter(); - pr_info("pool %d:\n", pool->id); - sched_show_task(worker->task); + pr_info("pool %d:\n", pool->id); + sched_show_task(worker->task); - printk_deferred_exit(); - } + printk_deferred_exit(); } raw_spin_unlock_irqrestore(&pool->lock, irq_flags); } -static void show_cpu_pools_hogs(void) +static void show_cpu_pools_busy_workers(void) { struct worker_pool *pool; int pi; - pr_info("Showing backtraces of running workers in stalled CPU-bound worker pools:\n"); + pr_info("Showing backtraces of busy workers in stalled worker pools:\n"); rcu_read_lock(); for_each_pool(pool, pi) { if (pool->cpu_stall) - show_cpu_pool_hog(pool); + show_cpu_pool_busy_workers(pool); } rcu_read_unlock(); } -static void panic_on_wq_watchdog(void) +/* + * It triggers a panic in two scenarios: when the total number of stalls + * exceeds a threshold, and when a stall lasts longer than + * wq_panic_on_stall_time + */ +static void panic_on_wq_watchdog(unsigned int stall_time_sec) { static unsigned int wq_stall; if (wq_panic_on_stall) { wq_stall++; - BUG_ON(wq_stall >= wq_panic_on_stall); + if (wq_stall >= wq_panic_on_stall) + panic("workqueue: %u stall(s) exceeded threshold %u\n", + wq_stall, wq_panic_on_stall); } + + if (wq_panic_on_stall_time && stall_time_sec >= wq_panic_on_stall_time) + panic("workqueue: stall lasted %us, exceeding threshold %us\n", + stall_time_sec, wq_panic_on_stall_time); } static void wq_watchdog_reset_touched(void) @@ -7551,17 +7755,17 @@ static void wq_watchdog_reset_touched(void) static void wq_watchdog_timer_fn(struct timer_list *unused) { unsigned long thresh = READ_ONCE(wq_watchdog_thresh) * HZ; + unsigned int max_stall_time = 0; bool lockup_detected = false; bool cpu_pool_stall = false; unsigned long now = jiffies; struct worker_pool *pool; + unsigned int stall_time; int pi; if (!thresh) return; - rcu_read_lock(); - for_each_pool(pool, pi) { unsigned long pool_ts, touched, ts; @@ -7580,39 +7784,57 @@ static void wq_watchdog_timer_fn(struct timer_list *unused) touched = READ_ONCE(per_cpu(wq_watchdog_touched_cpu, pool->cpu)); else touched = READ_ONCE(wq_watchdog_touched); - pool_ts = READ_ONCE(pool->watchdog_ts); + pool_ts = READ_ONCE(pool->last_progress_ts); if (time_after(pool_ts, touched)) ts = pool_ts; else ts = touched; - /* did we stall? */ + /* + * Did we stall? + * + * Do a lockless check first to do not disturb the system. + * + * Prevent false positives by double checking the timestamp + * under pool->lock. The lock makes sure that the check reads + * an updated pool->last_progress_ts when this CPU saw + * an already updated pool->worklist above. It seems better + * than adding another barrier into __queue_work() which + * is a hotter path. + */ if (time_after(now, ts + thresh)) { + scoped_guard(raw_spinlock_irqsave, &pool->lock) { + pool_ts = pool->last_progress_ts; + if (time_after(pool_ts, touched)) + ts = pool_ts; + else + ts = touched; + } + if (!time_after(now, ts + thresh)) + continue; + lockup_detected = true; + stall_time = jiffies_to_msecs(now - pool_ts) / 1000; + max_stall_time = max(max_stall_time, stall_time); if (pool->cpu >= 0 && !(pool->flags & POOL_BH)) { pool->cpu_stall = true; cpu_pool_stall = true; } pr_emerg("BUG: workqueue lockup - pool"); pr_cont_pool_info(pool); - pr_cont(" stuck for %us!\n", - jiffies_to_msecs(now - pool_ts) / 1000); + pr_cont(" stuck for %us!\n", stall_time); } - - } - rcu_read_unlock(); - if (lockup_detected) show_all_workqueues(); if (cpu_pool_stall) - show_cpu_pools_hogs(); + show_cpu_pools_busy_workers(); if (lockup_detected) - panic_on_wq_watchdog(); + panic_on_wq_watchdog(max_stall_time); wq_watchdog_reset_touched(); mod_timer(&wq_watchdog_timer, jiffies + thresh); @@ -7637,7 +7859,7 @@ notrace void wq_watchdog_touch(int cpu) static void wq_watchdog_set_thresh(unsigned long thresh) { wq_watchdog_thresh = 0; - del_timer_sync(&wq_watchdog_timer); + timer_delete_sync(&wq_watchdog_timer); if (thresh) { wq_watchdog_thresh = thresh; @@ -7656,7 +7878,7 @@ static int wq_watchdog_param_set_thresh(const char *val, if (ret) return ret; - if (system_wq) + if (system_percpu_wq) wq_watchdog_set_thresh(thresh); else wq_watchdog_thresh = thresh; @@ -7735,8 +7957,8 @@ void __init workqueue_init_early(void) { struct wq_pod_type *pt = &wq_pod_types[WQ_AFFN_SYSTEM]; int std_nice[NR_STD_WORKER_POOLS] = { 0, HIGHPRI_NICE_LEVEL }; - void (*irq_work_fns[2])(struct irq_work *) = { bh_pool_kick_normal, - bh_pool_kick_highpri }; + void (*irq_work_fns[NR_STD_WORKER_POOLS])(struct irq_work *) = + { bh_pool_kick_normal, bh_pool_kick_highpri }; int i, cpu; BUILD_BUG_ON(__alignof__(struct pool_workqueue) < __alignof__(long long)); @@ -7748,13 +7970,13 @@ void __init workqueue_init_early(void) cpumask_copy(wq_online_cpumask, cpu_online_mask); cpumask_copy(wq_unbound_cpumask, cpu_possible_mask); - restrict_unbound_cpumask("HK_TYPE_WQ", housekeeping_cpumask(HK_TYPE_WQ)); restrict_unbound_cpumask("HK_TYPE_DOMAIN", housekeeping_cpumask(HK_TYPE_DOMAIN)); if (!cpumask_empty(&wq_cmdline_cpumask)) restrict_unbound_cpumask("workqueue.unbound_cpus", &wq_cmdline_cpumask); cpumask_copy(wq_requested_unbound_cpumask, wq_unbound_cpumask); - + cpumask_andnot(wq_isolated_cpumask, cpu_possible_mask, + housekeeping_cpumask(HK_TYPE_DOMAIN)); pwq_cache = KMEM_CACHE(pool_workqueue, SLAB_PANIC); unbound_wq_update_pwq_attrs_buf = alloc_workqueue_attrs(); @@ -7768,9 +7990,9 @@ void __init workqueue_init_early(void) wq_power_efficient = true; /* initialize WQ_AFFN_SYSTEM pods */ - pt->pod_cpus = kcalloc(1, sizeof(pt->pod_cpus[0]), GFP_KERNEL); - pt->pod_node = kcalloc(1, sizeof(pt->pod_node[0]), GFP_KERNEL); - pt->cpu_pod = kcalloc(nr_cpu_ids, sizeof(pt->cpu_pod[0]), GFP_KERNEL); + pt->pod_cpus = kzalloc_objs(pt->pod_cpus[0], 1); + pt->pod_node = kzalloc_objs(pt->pod_node[0], 1); + pt->cpu_pod = kzalloc_objs(pt->cpu_pod[0], nr_cpu_ids); BUG_ON(!pt->pod_cpus || !pt->pod_node || !pt->cpu_pod); BUG_ON(!zalloc_cpumask_var_node(&pt->pod_cpus[0], GFP_KERNEL, NUMA_NO_NODE)); @@ -7815,26 +8037,28 @@ void __init workqueue_init_early(void) ordered_wq_attrs[i] = attrs; } - system_wq = alloc_workqueue("events", 0, 0); - system_highpri_wq = alloc_workqueue("events_highpri", WQ_HIGHPRI, 0); - system_long_wq = alloc_workqueue("events_long", 0, 0); - system_unbound_wq = alloc_workqueue("events_unbound", WQ_UNBOUND, - WQ_MAX_ACTIVE); + system_wq = alloc_workqueue("events", WQ_PERCPU, 0); + system_percpu_wq = alloc_workqueue("events", WQ_PERCPU, 0); + system_highpri_wq = alloc_workqueue("events_highpri", + WQ_HIGHPRI | WQ_PERCPU, 0); + system_long_wq = alloc_workqueue("events_long", WQ_PERCPU, 0); + system_unbound_wq = alloc_workqueue("events_unbound", WQ_UNBOUND, WQ_MAX_ACTIVE); + system_dfl_wq = alloc_workqueue("events_unbound", WQ_UNBOUND, WQ_MAX_ACTIVE); system_freezable_wq = alloc_workqueue("events_freezable", - WQ_FREEZABLE, 0); + WQ_FREEZABLE | WQ_PERCPU, 0); system_power_efficient_wq = alloc_workqueue("events_power_efficient", - WQ_POWER_EFFICIENT, 0); + WQ_POWER_EFFICIENT | WQ_PERCPU, 0); system_freezable_power_efficient_wq = alloc_workqueue("events_freezable_pwr_efficient", - WQ_FREEZABLE | WQ_POWER_EFFICIENT, - 0); - system_bh_wq = alloc_workqueue("events_bh", WQ_BH, 0); + WQ_FREEZABLE | WQ_POWER_EFFICIENT | WQ_PERCPU, 0); + system_bh_wq = alloc_workqueue("events_bh", WQ_BH | WQ_PERCPU, 0); system_bh_highpri_wq = alloc_workqueue("events_bh_highpri", - WQ_BH | WQ_HIGHPRI, 0); - BUG_ON(!system_wq || !system_highpri_wq || !system_long_wq || - !system_unbound_wq || !system_freezable_wq || + WQ_BH | WQ_HIGHPRI | WQ_PERCPU, 0); + system_dfl_long_wq = alloc_workqueue("events_dfl_long", WQ_UNBOUND, WQ_MAX_ACTIVE); + BUG_ON(!system_wq || !system_percpu_wq|| !system_highpri_wq || !system_long_wq || + !system_unbound_wq || !system_freezable_wq || !system_dfl_wq || !system_power_efficient_wq || !system_freezable_power_efficient_wq || - !system_bh_wq || !system_bh_highpri_wq); + !system_bh_wq || !system_bh_highpri_wq || !system_dfl_long_wq); } static void __init wq_cpu_intensive_thresh_init(void) @@ -7951,7 +8175,7 @@ static void __init init_pod_type(struct wq_pod_type *pt, pt->nr_pods = 0; /* init @pt->cpu_pod[] according to @cpus_share_pod() */ - pt->cpu_pod = kcalloc(nr_cpu_ids, sizeof(pt->cpu_pod[0]), GFP_KERNEL); + pt->cpu_pod = kzalloc_objs(pt->cpu_pod[0], nr_cpu_ids); BUG_ON(!pt->cpu_pod); for_each_possible_cpu(cur) { @@ -7968,8 +8192,8 @@ static void __init init_pod_type(struct wq_pod_type *pt, } /* init the rest to match @pt->cpu_pod[] */ - pt->pod_cpus = kcalloc(pt->nr_pods, sizeof(pt->pod_cpus[0]), GFP_KERNEL); - pt->pod_node = kcalloc(pt->nr_pods, sizeof(pt->pod_node[0]), GFP_KERNEL); + pt->pod_cpus = kzalloc_objs(pt->pod_cpus[0], pt->nr_pods); + pt->pod_node = kzalloc_objs(pt->pod_node[0], pt->nr_pods); BUG_ON(!pt->pod_cpus || !pt->pod_node); for (pod = 0; pod < pt->nr_pods; pod++) @@ -8000,6 +8224,186 @@ static bool __init cpus_share_numa(int cpu0, int cpu1) return cpu_to_node(cpu0) == cpu_to_node(cpu1); } +/* Maps each CPU to its shard index within the LLC pod it belongs to */ +static int cpu_shard_id[NR_CPUS] __initdata; + +/** + * llc_count_cores - count distinct cores (SMT groups) within an LLC pod + * @pod_cpus: the cpumask of CPUs in the LLC pod + * @smt_pods: the SMT pod type, used to identify sibling groups + * + * A core is represented by the lowest-numbered CPU in its SMT group. Returns + * the number of distinct cores found in @pod_cpus. + */ +static int __init llc_count_cores(const struct cpumask *pod_cpus, + struct wq_pod_type *smt_pods) +{ + const struct cpumask *sibling_cpus; + int nr_cores = 0, c; + + /* + * Count distinct cores by only counting the first CPU in each + * SMT sibling group. + */ + for_each_cpu(c, pod_cpus) { + sibling_cpus = smt_pods->pod_cpus[smt_pods->cpu_pod[c]]; + if (cpumask_first(sibling_cpus) == c) + nr_cores++; + } + + return nr_cores; +} + +/* + * llc_shard_size - number of cores in a given shard + * + * Cores are spread as evenly as possible. The first @nr_large_shards shards are + * "large shards" with (cores_per_shard + 1) cores; the rest are "default + * shards" with cores_per_shard cores. + */ +static int __init llc_shard_size(int shard_id, int cores_per_shard, int nr_large_shards) +{ + /* The first @nr_large_shards shards are large shards */ + if (shard_id < nr_large_shards) + return cores_per_shard + 1; + + /* The remaining shards are default shards */ + return cores_per_shard; +} + +/* + * llc_calc_shard_layout - compute the shard layout for an LLC pod + * @nr_cores: number of distinct cores in the LLC pod + * + * Chooses the number of shards that keeps average shard size closest to + * wq_cache_shard_size. Returns a struct describing the total number of shards, + * the base size of each, and how many are large shards. + */ +static struct llc_shard_layout __init llc_calc_shard_layout(int nr_cores) +{ + struct llc_shard_layout layout; + + /* Ensure at least one shard; pick the count closest to the target size */ + layout.nr_shards = max(1, DIV_ROUND_CLOSEST(nr_cores, wq_cache_shard_size)); + layout.cores_per_shard = nr_cores / layout.nr_shards; + layout.nr_large_shards = nr_cores % layout.nr_shards; + + return layout; +} + +/* + * llc_shard_is_full - check whether a shard has reached its core capacity + * @cores_in_shard: number of cores already assigned to this shard + * @shard_id: index of the shard being checked + * @layout: the shard layout computed by llc_calc_shard_layout() + * + * Returns true if @cores_in_shard equals the expected size for @shard_id. + */ +static bool __init llc_shard_is_full(int cores_in_shard, int shard_id, + const struct llc_shard_layout *layout) +{ + return cores_in_shard == llc_shard_size(shard_id, layout->cores_per_shard, + layout->nr_large_shards); +} + +/** + * llc_populate_cpu_shard_id - populate cpu_shard_id[] for each CPU in an LLC pod + * @pod_cpus: the cpumask of CPUs in the LLC pod + * @smt_pods: the SMT pod type, used to identify sibling groups + * @nr_cores: number of distinct cores in @pod_cpus (from llc_count_cores()) + * + * Walks @pod_cpus in order. At each SMT group leader, advances to the next + * shard once the current shard is full. Results are written to cpu_shard_id[]. + */ +static void __init llc_populate_cpu_shard_id(const struct cpumask *pod_cpus, + struct wq_pod_type *smt_pods, + int nr_cores) +{ + struct llc_shard_layout layout = llc_calc_shard_layout(nr_cores); + const struct cpumask *sibling_cpus; + /* Count the number of cores in the current shard_id */ + int cores_in_shard = 0; + unsigned int leader; + /* This is a cursor for the shards. Go from zero to nr_shards - 1*/ + int shard_id = 0; + int c; + + /* Iterate at every CPU for a given LLC pod, and assign it a shard */ + for_each_cpu(c, pod_cpus) { + sibling_cpus = smt_pods->pod_cpus[smt_pods->cpu_pod[c]]; + if (cpumask_first(sibling_cpus) == c) { + /* This is the CPU leader for the siblings */ + if (llc_shard_is_full(cores_in_shard, shard_id, &layout)) { + shard_id++; + cores_in_shard = 0; + } + cores_in_shard++; + cpu_shard_id[c] = shard_id; + } else { + /* + * The siblings' shard MUST be the same as the leader. + * never split threads in the same core. + */ + leader = cpumask_first(sibling_cpus); + + /* + * This check silences a Warray-bounds warning on UP + * configs where NR_CPUS=1 makes cpu_shard_id[] + * a single-element array, and the compiler can't + * prove the index is always 0. + */ + if (WARN_ON_ONCE(leader >= nr_cpu_ids)) + continue; + cpu_shard_id[c] = cpu_shard_id[leader]; + } + } + + WARN_ON_ONCE(shard_id != (layout.nr_shards - 1)); +} + +/** + * precompute_cache_shard_ids - assign each CPU its shard index within its LLC + * + * Iterates over all LLC pods. For each pod, counts distinct cores then assigns + * shard indices to all CPUs in the pod. Must be called after WQ_AFFN_CACHE and + * WQ_AFFN_SMT have been initialized. + */ +static void __init precompute_cache_shard_ids(void) +{ + struct wq_pod_type *llc_pods = &wq_pod_types[WQ_AFFN_CACHE]; + struct wq_pod_type *smt_pods = &wq_pod_types[WQ_AFFN_SMT]; + const struct cpumask *cpus_sharing_llc; + int nr_cores; + int pod; + + if (!wq_cache_shard_size) { + pr_warn("workqueue: cache_shard_size must be > 0, setting to 1\n"); + wq_cache_shard_size = 1; + } + + for (pod = 0; pod < llc_pods->nr_pods; pod++) { + cpus_sharing_llc = llc_pods->pod_cpus[pod]; + + /* Number of cores in this given LLC */ + nr_cores = llc_count_cores(cpus_sharing_llc, smt_pods); + llc_populate_cpu_shard_id(cpus_sharing_llc, smt_pods, nr_cores); + } +} + +/* + * cpus_share_cache_shard - test whether two CPUs belong to the same cache shard + * + * Two CPUs share a cache shard if they are in the same LLC and have the same + * shard index. Used as the pod affinity callback for WQ_AFFN_CACHE_SHARD. + */ +static bool __init cpus_share_cache_shard(int cpu0, int cpu1) +{ + if (!cpus_share_cache(cpu0, cpu1)) + return false; + + return cpu_shard_id[cpu0] == cpu_shard_id[cpu1]; +} + /** * workqueue_init_topology - initialize CPU pods for unbound workqueues * @@ -8015,6 +8419,8 @@ void __init workqueue_init_topology(void) init_pod_type(&wq_pod_types[WQ_AFFN_CPU], cpus_dont_share); init_pod_type(&wq_pod_types[WQ_AFFN_SMT], cpus_share_smt); init_pod_type(&wq_pod_types[WQ_AFFN_CACHE], cpus_share_cache); + precompute_cache_shard_ids(); + init_pod_type(&wq_pod_types[WQ_AFFN_CACHE_SHARD], cpus_share_cache_shard); init_pod_type(&wq_pod_types[WQ_AFFN_NUMA], cpus_share_numa); wq_topo_initialized = true; |