From 52bad64d95bd89e08c49ec5a071fa6dcbe5a1a9c Mon Sep 17 00:00:00 2001 From: David Howells Date: Wed, 22 Nov 2006 14:54:01 +0000 Subject: WorkStruct: Separate delayable and non-delayable events. Separate delayable work items from non-delayable work items be splitting them into a separate structure (delayed_work), which incorporates a work_struct and the timer_list removed from work_struct. The work_struct struct is huge, and this limits it's usefulness. On a 64-bit architecture it's nearly 100 bytes in size. This reduces that by half for the non-delayable type of event. Signed-Off-By: David Howells --- kernel/workqueue.c | 51 ++++++++++++++++++++++++++++----------------------- 1 file changed, 28 insertions(+), 23 deletions(-) (limited to 'kernel/workqueue.c') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 17c2f03d2c2..44fc54b7dec 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -122,29 +122,33 @@ EXPORT_SYMBOL_GPL(queue_work); static void delayed_work_timer_fn(unsigned long __data) { - struct work_struct *work = (struct work_struct *)__data; - struct workqueue_struct *wq = work->wq_data; + struct delayed_work *dwork = (struct delayed_work *)__data; + struct workqueue_struct *wq = dwork->work.wq_data; int cpu = smp_processor_id(); if (unlikely(is_single_threaded(wq))) cpu = singlethread_cpu; - __queue_work(per_cpu_ptr(wq->cpu_wq, cpu), work); + __queue_work(per_cpu_ptr(wq->cpu_wq, cpu), &dwork->work); } /** * queue_delayed_work - queue work on a workqueue after delay * @wq: workqueue to use - * @work: work to queue + * @work: delayable work to queue * @delay: number of jiffies to wait before queueing * * Returns 0 if @work was already on a queue, non-zero otherwise. */ int fastcall queue_delayed_work(struct workqueue_struct *wq, - struct work_struct *work, unsigned long delay) + struct delayed_work *dwork, unsigned long delay) { int ret = 0; - struct timer_list *timer = &work->timer; + struct timer_list *timer = &dwork->timer; + struct work_struct *work = &dwork->work; + + if (delay == 0) + return queue_work(wq, work); if (!test_and_set_bit(0, &work->pending)) { BUG_ON(timer_pending(timer)); @@ -153,7 +157,7 @@ int fastcall queue_delayed_work(struct workqueue_struct *wq, /* This stores wq for the moment, for the timer_fn */ work->wq_data = wq; timer->expires = jiffies + delay; - timer->data = (unsigned long)work; + timer->data = (unsigned long)dwork; timer->function = delayed_work_timer_fn; add_timer(timer); ret = 1; @@ -172,10 +176,11 @@ EXPORT_SYMBOL_GPL(queue_delayed_work); * Returns 0 if @work was already on a queue, non-zero otherwise. */ int queue_delayed_work_on(int cpu, struct workqueue_struct *wq, - struct work_struct *work, unsigned long delay) + struct delayed_work *dwork, unsigned long delay) { int ret = 0; - struct timer_list *timer = &work->timer; + struct timer_list *timer = &dwork->timer; + struct work_struct *work = &dwork->work; if (!test_and_set_bit(0, &work->pending)) { BUG_ON(timer_pending(timer)); @@ -184,7 +189,7 @@ int queue_delayed_work_on(int cpu, struct workqueue_struct *wq, /* This stores wq for the moment, for the timer_fn */ work->wq_data = wq; timer->expires = jiffies + delay; - timer->data = (unsigned long)work; + timer->data = (unsigned long)dwork; timer->function = delayed_work_timer_fn; add_timer_on(timer, cpu); ret = 1; @@ -468,31 +473,31 @@ EXPORT_SYMBOL(schedule_work); /** * schedule_delayed_work - put work task in global workqueue after delay - * @work: job to be done - * @delay: number of jiffies to wait + * @dwork: job to be done + * @delay: number of jiffies to wait or 0 for immediate execution * * After waiting for a given time this puts a job in the kernel-global * workqueue. */ -int fastcall schedule_delayed_work(struct work_struct *work, unsigned long delay) +int fastcall schedule_delayed_work(struct delayed_work *dwork, unsigned long delay) { - return queue_delayed_work(keventd_wq, work, delay); + return queue_delayed_work(keventd_wq, dwork, delay); } EXPORT_SYMBOL(schedule_delayed_work); /** * schedule_delayed_work_on - queue work in global workqueue on CPU after delay * @cpu: cpu to use - * @work: job to be done + * @dwork: job to be done * @delay: number of jiffies to wait * * After waiting for a given time this puts a job in the kernel-global * workqueue on the specified CPU. */ int schedule_delayed_work_on(int cpu, - struct work_struct *work, unsigned long delay) + struct delayed_work *dwork, unsigned long delay) { - return queue_delayed_work_on(cpu, keventd_wq, work, delay); + return queue_delayed_work_on(cpu, keventd_wq, dwork, delay); } EXPORT_SYMBOL(schedule_delayed_work_on); @@ -539,12 +544,12 @@ EXPORT_SYMBOL(flush_scheduled_work); * cancel_rearming_delayed_workqueue - reliably kill off a delayed * work whose handler rearms the delayed work. * @wq: the controlling workqueue structure - * @work: the delayed work struct + * @dwork: the delayed work struct */ void cancel_rearming_delayed_workqueue(struct workqueue_struct *wq, - struct work_struct *work) + struct delayed_work *dwork) { - while (!cancel_delayed_work(work)) + while (!cancel_delayed_work(dwork)) flush_workqueue(wq); } EXPORT_SYMBOL(cancel_rearming_delayed_workqueue); @@ -552,11 +557,11 @@ EXPORT_SYMBOL(cancel_rearming_delayed_workqueue); /** * cancel_rearming_delayed_work - reliably kill off a delayed keventd * work whose handler rearms the delayed work. - * @work: the delayed work struct + * @dwork: the delayed work struct */ -void cancel_rearming_delayed_work(struct work_struct *work) +void cancel_rearming_delayed_work(struct delayed_work *dwork) { - cancel_rearming_delayed_workqueue(keventd_wq, work); + cancel_rearming_delayed_workqueue(keventd_wq, dwork); } EXPORT_SYMBOL(cancel_rearming_delayed_work); -- cgit v1.2.3-18-g5258 From 6bb49e5965c1fc399b4d3cd2b5cf2da535b330c0 Mon Sep 17 00:00:00 2001 From: David Howells Date: Wed, 22 Nov 2006 14:54:45 +0000 Subject: WorkStruct: Typedef the work function prototype Define a type for the work function prototype. It's not only kept in the work_struct struct, it's also passed as an argument to several functions. This makes it easier to change it. Signed-Off-By: David Howells --- kernel/workqueue.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel/workqueue.c') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 44fc54b7dec..1e9d61ecf76 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -217,7 +217,7 @@ static void run_workqueue(struct cpu_workqueue_struct *cwq) while (!list_empty(&cwq->worklist)) { struct work_struct *work = list_entry(cwq->worklist.next, struct work_struct, entry); - void (*f) (void *) = work->func; + work_func_t f = work->func; void *data = work->data; list_del_init(cwq->worklist.next); @@ -513,7 +513,7 @@ EXPORT_SYMBOL(schedule_delayed_work_on); * * schedule_on_each_cpu() is very slow. */ -int schedule_on_each_cpu(void (*func)(void *info), void *info) +int schedule_on_each_cpu(work_func_t func, void *info) { int cpu; struct work_struct *works; @@ -578,7 +578,7 @@ EXPORT_SYMBOL(cancel_rearming_delayed_work); * Returns: 0 - function was executed * 1 - function was scheduled for execution */ -int execute_in_process_context(void (*fn)(void *data), void *data, +int execute_in_process_context(work_func_t fn, void *data, struct execute_work *ew) { if (!in_interrupt()) { -- cgit v1.2.3-18-g5258 From 365970a1ea76d81cb1ad2f652acb605f06dae256 Mon Sep 17 00:00:00 2001 From: David Howells Date: Wed, 22 Nov 2006 14:54:49 +0000 Subject: WorkStruct: Merge the pending bit into the wq_data pointer Reclaim a word from the size of the work_struct by folding the pending bit and the wq_data pointer together. This shouldn't cause misalignment problems as all pointers should be at least 4-byte aligned. Signed-Off-By: David Howells --- kernel/workqueue.c | 41 ++++++++++++++++++++++++++++++++--------- 1 file changed, 32 insertions(+), 9 deletions(-) (limited to 'kernel/workqueue.c') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 1e9d61ecf76..96747975651 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -80,6 +80,29 @@ static inline int is_single_threaded(struct workqueue_struct *wq) return list_empty(&wq->list); } +static inline void set_wq_data(struct work_struct *work, void *wq) +{ + unsigned long new, old, res; + + /* assume the pending flag is already set and that the task has already + * been queued on this workqueue */ + new = (unsigned long) wq | (1UL << WORK_STRUCT_PENDING); + res = work->management; + if (res != new) { + do { + old = res; + new = (unsigned long) wq; + new |= (old & WORK_STRUCT_FLAG_MASK); + res = cmpxchg(&work->management, old, new); + } while (res != old); + } +} + +static inline void *get_wq_data(struct work_struct *work) +{ + return (void *) (work->management & WORK_STRUCT_WQ_DATA_MASK); +} + /* Preempt must be disabled. */ static void __queue_work(struct cpu_workqueue_struct *cwq, struct work_struct *work) @@ -87,7 +110,7 @@ static void __queue_work(struct cpu_workqueue_struct *cwq, unsigned long flags; spin_lock_irqsave(&cwq->lock, flags); - work->wq_data = cwq; + set_wq_data(work, cwq); list_add_tail(&work->entry, &cwq->worklist); cwq->insert_sequence++; wake_up(&cwq->more_work); @@ -108,7 +131,7 @@ int fastcall queue_work(struct workqueue_struct *wq, struct work_struct *work) { int ret = 0, cpu = get_cpu(); - if (!test_and_set_bit(0, &work->pending)) { + if (!test_and_set_bit(WORK_STRUCT_PENDING, &work->management)) { if (unlikely(is_single_threaded(wq))) cpu = singlethread_cpu; BUG_ON(!list_empty(&work->entry)); @@ -123,7 +146,7 @@ EXPORT_SYMBOL_GPL(queue_work); static void delayed_work_timer_fn(unsigned long __data) { struct delayed_work *dwork = (struct delayed_work *)__data; - struct workqueue_struct *wq = dwork->work.wq_data; + struct workqueue_struct *wq = get_wq_data(&dwork->work); int cpu = smp_processor_id(); if (unlikely(is_single_threaded(wq))) @@ -150,12 +173,12 @@ int fastcall queue_delayed_work(struct workqueue_struct *wq, if (delay == 0) return queue_work(wq, work); - if (!test_and_set_bit(0, &work->pending)) { + if (!test_and_set_bit(WORK_STRUCT_PENDING, &work->management)) { BUG_ON(timer_pending(timer)); BUG_ON(!list_empty(&work->entry)); /* This stores wq for the moment, for the timer_fn */ - work->wq_data = wq; + set_wq_data(work, wq); timer->expires = jiffies + delay; timer->data = (unsigned long)dwork; timer->function = delayed_work_timer_fn; @@ -182,12 +205,12 @@ int queue_delayed_work_on(int cpu, struct workqueue_struct *wq, struct timer_list *timer = &dwork->timer; struct work_struct *work = &dwork->work; - if (!test_and_set_bit(0, &work->pending)) { + if (!test_and_set_bit(WORK_STRUCT_PENDING, &work->management)) { BUG_ON(timer_pending(timer)); BUG_ON(!list_empty(&work->entry)); /* This stores wq for the moment, for the timer_fn */ - work->wq_data = wq; + set_wq_data(work, wq); timer->expires = jiffies + delay; timer->data = (unsigned long)dwork; timer->function = delayed_work_timer_fn; @@ -223,8 +246,8 @@ static void run_workqueue(struct cpu_workqueue_struct *cwq) list_del_init(cwq->worklist.next); spin_unlock_irqrestore(&cwq->lock, flags); - BUG_ON(work->wq_data != cwq); - clear_bit(0, &work->pending); + BUG_ON(get_wq_data(work) != cwq); + clear_bit(WORK_STRUCT_PENDING, &work->management); f(data); spin_lock_irqsave(&cwq->lock, flags); -- cgit v1.2.3-18-g5258 From 65f27f38446e1976cc98fd3004b110fedcddd189 Mon Sep 17 00:00:00 2001 From: David Howells Date: Wed, 22 Nov 2006 14:55:48 +0000 Subject: WorkStruct: Pass the work_struct pointer instead of context data Pass the work_struct pointer to the work function rather than context data. The work function can use container_of() to work out the data. For the cases where the container of the work_struct may go away the moment the pending bit is cleared, it is made possible to defer the release of the structure by deferring the clearing of the pending bit. To make this work, an extra flag is introduced into the management side of the work_struct. This governs auto-release of the structure upon execution. Ordinarily, the work queue executor would release the work_struct for further scheduling or deallocation by clearing the pending bit prior to jumping to the work function. This means that, unless the driver makes some guarantee itself that the work_struct won't go away, the work function may not access anything else in the work_struct or its container lest they be deallocated.. This is a problem if the auxiliary data is taken away (as done by the last patch). However, if the pending bit is *not* cleared before jumping to the work function, then the work function *may* access the work_struct and its container with no problems. But then the work function must itself release the work_struct by calling work_release(). In most cases, automatic release is fine, so this is the default. Special initiators exist for the non-auto-release case (ending in _NAR). Signed-Off-By: David Howells --- kernel/workqueue.c | 19 ++++++++----------- 1 file changed, 8 insertions(+), 11 deletions(-) (limited to 'kernel/workqueue.c') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 96747975651..8d1e7cb8a51 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -241,14 +241,14 @@ static void run_workqueue(struct cpu_workqueue_struct *cwq) struct work_struct *work = list_entry(cwq->worklist.next, struct work_struct, entry); work_func_t f = work->func; - void *data = work->data; list_del_init(cwq->worklist.next); spin_unlock_irqrestore(&cwq->lock, flags); BUG_ON(get_wq_data(work) != cwq); - clear_bit(WORK_STRUCT_PENDING, &work->management); - f(data); + if (!test_bit(WORK_STRUCT_NOAUTOREL, &work->management)) + work_release(work); + f(work); spin_lock_irqsave(&cwq->lock, flags); cwq->remove_sequence++; @@ -527,7 +527,6 @@ EXPORT_SYMBOL(schedule_delayed_work_on); /** * schedule_on_each_cpu - call a function on each online CPU from keventd * @func: the function to call - * @info: a pointer to pass to func() * * Returns zero on success. * Returns -ve errno on failure. @@ -536,7 +535,7 @@ EXPORT_SYMBOL(schedule_delayed_work_on); * * schedule_on_each_cpu() is very slow. */ -int schedule_on_each_cpu(work_func_t func, void *info) +int schedule_on_each_cpu(work_func_t func) { int cpu; struct work_struct *works; @@ -547,7 +546,7 @@ int schedule_on_each_cpu(work_func_t func, void *info) mutex_lock(&workqueue_mutex); for_each_online_cpu(cpu) { - INIT_WORK(per_cpu_ptr(works, cpu), func, info); + INIT_WORK(per_cpu_ptr(works, cpu), func); __queue_work(per_cpu_ptr(keventd_wq->cpu_wq, cpu), per_cpu_ptr(works, cpu)); } @@ -591,7 +590,6 @@ EXPORT_SYMBOL(cancel_rearming_delayed_work); /** * execute_in_process_context - reliably execute the routine with user context * @fn: the function to execute - * @data: data to pass to the function * @ew: guaranteed storage for the execute work structure (must * be available when the work executes) * @@ -601,15 +599,14 @@ EXPORT_SYMBOL(cancel_rearming_delayed_work); * Returns: 0 - function was executed * 1 - function was scheduled for execution */ -int execute_in_process_context(work_func_t fn, void *data, - struct execute_work *ew) +int execute_in_process_context(work_func_t fn, struct execute_work *ew) { if (!in_interrupt()) { - fn(data); + fn(&ew->work); return 0; } - INIT_WORK(&ew->work, fn, data); + INIT_WORK(&ew->work, fn); schedule_work(&ew->work); return 1; -- cgit v1.2.3-18-g5258