From d25302e46592c97d29f70ccb1be558df31a9a360 Mon Sep 17 00:00:00 2001 From: Menglong Dong Date: Sun, 17 Oct 2021 20:04:02 +0800 Subject: workqueue: make sysfs of unbound kworker cpumask more clever Some unfriendly component, such as dpdk, write the same mask to unbound kworker cpumask again and again. Every time it write to this interface some work is queue to cpu, even though the mask is same with the original mask. So, fix it by return success and do nothing if the cpumask is equal with the old one. Signed-off-by: Mengen Sun Signed-off-by: Menglong Dong Signed-off-by: Tejun Heo --- kernel/workqueue.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) (limited to 'kernel/workqueue.c') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 1b3eb1e9531f..76988f39ed5a 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -5384,9 +5384,6 @@ int workqueue_set_unbound_cpumask(cpumask_var_t cpumask) int ret = -EINVAL; cpumask_var_t saved_cpumask; - if (!zalloc_cpumask_var(&saved_cpumask, GFP_KERNEL)) - return -ENOMEM; - /* * Not excluding isolated cpus on purpose. * If the user wishes to include them, we allow that. @@ -5394,6 +5391,15 @@ int workqueue_set_unbound_cpumask(cpumask_var_t cpumask) cpumask_and(cpumask, cpumask, cpu_possible_mask); if (!cpumask_empty(cpumask)) { apply_wqattrs_lock(); + if (cpumask_equal(cpumask, wq_unbound_cpumask)) { + ret = 0; + goto out_unlock; + } + + if (!zalloc_cpumask_var(&saved_cpumask, GFP_KERNEL)) { + ret = -ENOMEM; + goto out_unlock; + } /* save the old wq_unbound_cpumask. */ cpumask_copy(saved_cpumask, wq_unbound_cpumask); @@ -5406,10 +5412,11 @@ int workqueue_set_unbound_cpumask(cpumask_var_t cpumask) if (ret < 0) cpumask_copy(wq_unbound_cpumask, saved_cpumask); + free_cpumask_var(saved_cpumask); +out_unlock: apply_wqattrs_unlock(); } - free_cpumask_var(saved_cpumask); return ret; } -- cgit v1.2.3 From 55df0933be74bd2e52aba0b67eb743ae0feabe7e Mon Sep 17 00:00:00 2001 From: Imran Khan Date: Wed, 20 Oct 2021 14:09:00 +1100 Subject: workqueue: Introduce show_one_worker_pool and show_one_workqueue. Currently show_workqueue_state shows the state of all workqueues and of all worker pools. In certain cases we may need to dump state of only a specific workqueue or worker pool. For example in destroy_workqueue we only need to show state of the workqueue which is getting destroyed. So rename show_workqueue_state to show_all_workqueues(to signify it dumps state of all busy workqueues) and divide it into more granular functions (show_one_workqueue and show_one_worker_pool), that would show states of individual workqueues and worker pools and can be used in cases such as the one mentioned above. Also, as mentioned earlier, make destroy_workqueue dump data pertaining to only the workqueue that is being destroyed and make user(s) of earlier interface(show_workqueue_state), use new interface (show_all_workqueues). Signed-off-by: Imran Khan Signed-off-by: Tejun Heo --- kernel/workqueue.c | 172 ++++++++++++++++++++++++++++++----------------------- 1 file changed, 96 insertions(+), 76 deletions(-) (limited to 'kernel/workqueue.c') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 76988f39ed5a..1a7df882f55e 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -375,6 +375,7 @@ EXPORT_SYMBOL_GPL(system_freezable_power_efficient_wq); static int worker_thread(void *__worker); static void workqueue_sysfs_unregister(struct workqueue_struct *wq); static void show_pwq(struct pool_workqueue *pwq); +static void show_one_worker_pool(struct worker_pool *pool); #define CREATE_TRACE_POINTS #include @@ -4447,7 +4448,7 @@ void destroy_workqueue(struct workqueue_struct *wq) raw_spin_unlock_irq(&pwq->pool->lock); mutex_unlock(&wq->mutex); mutex_unlock(&wq_pool_mutex); - show_workqueue_state(); + show_one_workqueue(wq); return; } raw_spin_unlock_irq(&pwq->pool->lock); @@ -4797,97 +4798,116 @@ static void show_pwq(struct pool_workqueue *pwq) } /** - * show_workqueue_state - dump workqueue state - * - * Called from a sysrq handler or try_to_freeze_tasks() and prints out - * all busy workqueues and pools. + * show_one_workqueue - dump state of specified workqueue + * @wq: workqueue whose state will be printed */ -void show_workqueue_state(void) +void show_one_workqueue(struct workqueue_struct *wq) { - struct workqueue_struct *wq; - struct worker_pool *pool; + struct pool_workqueue *pwq; + bool idle = true; unsigned long flags; - int pi; - - rcu_read_lock(); - pr_info("Showing busy workqueues and worker pools:\n"); - - list_for_each_entry_rcu(wq, &workqueues, list) { - struct pool_workqueue *pwq; - bool idle = true; - - for_each_pwq(pwq, wq) { - if (pwq->nr_active || !list_empty(&pwq->inactive_works)) { - idle = false; - break; - } + for_each_pwq(pwq, wq) { + if (pwq->nr_active || !list_empty(&pwq->inactive_works)) { + idle = false; + break; } - if (idle) - continue; + } + if (idle) /* Nothing to print for idle workqueue */ + return; - pr_info("workqueue %s: flags=0x%x\n", wq->name, wq->flags); + pr_info("workqueue %s: flags=0x%x\n", wq->name, wq->flags); - for_each_pwq(pwq, wq) { - raw_spin_lock_irqsave(&pwq->pool->lock, flags); - if (pwq->nr_active || !list_empty(&pwq->inactive_works)) { - /* - * Defer printing to avoid deadlocks in console - * drivers that queue work while holding locks - * also taken in their write paths. - */ - printk_deferred_enter(); - show_pwq(pwq); - printk_deferred_exit(); - } - raw_spin_unlock_irqrestore(&pwq->pool->lock, flags); + for_each_pwq(pwq, wq) { + raw_spin_lock_irqsave(&pwq->pool->lock, flags); + if (pwq->nr_active || !list_empty(&pwq->inactive_works)) { /* - * We could be printing a lot from atomic context, e.g. - * sysrq-t -> show_workqueue_state(). Avoid triggering - * hard lockup. + * Defer printing to avoid deadlocks in console + * drivers that queue work while holding locks + * also taken in their write paths. */ - touch_nmi_watchdog(); - } - } - - for_each_pool(pool, pi) { - struct worker *worker; - bool first = true; - - raw_spin_lock_irqsave(&pool->lock, flags); - if (pool->nr_workers == pool->nr_idle) - goto next_pool; - /* - * 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:", pool->id); - pr_cont_pool_info(pool); - pr_cont(" hung=%us workers=%d", - jiffies_to_msecs(jiffies - pool->watchdog_ts) / 1000, - pool->nr_workers); - if (pool->manager) - pr_cont(" manager: %d", - task_pid_nr(pool->manager->task)); - list_for_each_entry(worker, &pool->idle_list, entry) { - pr_cont(" %s%d", first ? "idle: " : "", - task_pid_nr(worker->task)); - first = false; + printk_deferred_enter(); + show_pwq(pwq); + printk_deferred_exit(); } - pr_cont("\n"); - printk_deferred_exit(); - next_pool: - raw_spin_unlock_irqrestore(&pool->lock, flags); + raw_spin_unlock_irqrestore(&pwq->pool->lock, flags); /* * We could be printing a lot from atomic context, e.g. - * sysrq-t -> show_workqueue_state(). Avoid triggering + * sysrq-t -> show_all_workqueues(). Avoid triggering * hard lockup. */ touch_nmi_watchdog(); } +} + +/** + * show_one_worker_pool - dump state of specified worker pool + * @pool: worker pool whose state will be printed + */ +static void show_one_worker_pool(struct worker_pool *pool) +{ + struct worker *worker; + bool first = true; + unsigned long flags; + + raw_spin_lock_irqsave(&pool->lock, flags); + if (pool->nr_workers == pool->nr_idle) + goto next_pool; + /* + * 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:", pool->id); + pr_cont_pool_info(pool); + pr_cont(" hung=%us workers=%d", + jiffies_to_msecs(jiffies - pool->watchdog_ts) / 1000, + pool->nr_workers); + if (pool->manager) + pr_cont(" manager: %d", + task_pid_nr(pool->manager->task)); + list_for_each_entry(worker, &pool->idle_list, entry) { + pr_cont(" %s%d", first ? "idle: " : "", + task_pid_nr(worker->task)); + first = false; + } + pr_cont("\n"); + printk_deferred_exit(); +next_pool: + raw_spin_unlock_irqrestore(&pool->lock, flags); + /* + * We could be printing a lot from atomic context, e.g. + * sysrq-t -> show_all_workqueues(). Avoid triggering + * hard lockup. + */ + touch_nmi_watchdog(); + +} + +/** + * show_all_workqueues - dump workqueue state + * + * Called from a sysrq handler or try_to_freeze_tasks() and prints out + * all busy workqueues and pools. + */ +void show_all_workqueues(void) +{ + struct workqueue_struct *wq; + struct worker_pool *pool; + int pi; + + rcu_read_lock(); + + pr_info("Showing busy workqueues and worker pools:\n"); + + list_for_each_entry_rcu(wq, &workqueues, list) + show_one_workqueue(wq); + + for_each_pool(pool, pi) + show_one_worker_pool(pool); + rcu_read_unlock(); } @@ -5876,7 +5896,7 @@ static void wq_watchdog_timer_fn(struct timer_list *unused) rcu_read_unlock(); if (lockup_detected) - show_workqueue_state(); + show_all_workqueues(); wq_watchdog_reset_touched(); mod_timer(&wq_watchdog_timer, jiffies + thresh); -- cgit v1.2.3