Linux内核中工作队列(work_queue)的操作

本文档的Copyleft归yfydz所有，使用GPL发布，可以自由拷贝，转载，转载时请保持文档的完整性，严禁用于任何商业用途。
msn: [e-mail]
来源：http://yfydz.cublog.cn

1. 前言

工作队列(workqueue)的Linux内核中的定义的用来处理不是很紧急事件的回调方式处理方法.

以下代码的linux内核版本为2.6.19.2, 源代码文件主要为kernel/workqueue.c.

2. 数据结构
/* include/linux/workqueue.h */
// 工作节点结构
struct work_struct {
// 等待时间
 unsigned long pending;
// 链表节点
 struct list_head entry;
// workqueue回调函数
 void (*func)(void *);
// 回调函数func的数据
 void *data;
// 指向CPU相关数据, 一般指向struct cpu_workqueue_struct结构
 void *wq_data;
// 定时器
 struct timer_list timer;
};

struct execute_work {
 struct work_struct work;
};

/* kernel/workqueue.c */
/*
 * The per-CPU workqueue (if single thread, we always use the first
 * possible cpu).
 *
 * The sequence counters are for flush_scheduled_work().  It wants to wait
 * until all currently-scheduled works are completed, but it doesn't
 * want to be livelocked by new, incoming ones.  So it waits until
 * remove_sequence is >= the insert_sequence which pertained when
 * flush_scheduled_work() was called.
 */
// 这个结构是针对每个CPU的
struct cpu_workqueue_struct {
// 结构锁
 spinlock_t lock;
// 下一个要执行的节点序号
 long remove_sequence; /* Least-recently added (next to run) */
// 下一个要插入节点的序号
 long insert_sequence; /* Next to add */
// 工作机构链表节点
 struct list_head worklist;
// 要进行处理的等待队列
 wait_queue_head_t more_work;
// 处理完的等待队列
 wait_queue_head_t work_done;
// 工作队列节点
 struct workqueue_struct *wq;
// 进程指针
 struct task_struct *thread;
 int run_depth;  /* Detect run_workqueue() recursion depth */
} ____cacheline_aligned;
/*
 * The externally visible workqueue abstraction is an array of
 * per-CPU workqueues:
 */
// 工作队列结构
struct workqueue_struct {
 struct cpu_workqueue_struct *cpu_wq;
 const char *name;
 struct list_head list;  /* Empty if single thread */
};
 
kernel/workqueue.c中定义了一个工作队列链表, 所有工作队列可以挂接到这个链表中:
static LIST_HEAD(workqueues);

3. 一些宏定义
/* include/linux/workqueue.h */
// 初始化工作队列
#define __WORK_INITIALIZER(n, f, d) {    \
// 初始化list
        .entry = { &(n).entry, &(n).entry },   \
// 回调函数
 .func = (f),      \
// 回调函数参数
 .data = (d),      \
// 初始化定时器
 .timer = TIMER_INITIALIZER(NULL, 0, 0),   \
 }

// 声明工作队列并初始化
#define DECLARE_WORK(n, f, d)     \
 struct work_struct n = __WORK_INITIALIZER(n, f, d)
/*
 * initialize a work-struct's func and data pointers:
 */
// 重新定义工作结构参数
#define PREPARE_WORK(_work, _func, _data)   \
 do {       \
  (_work)->func = _func;    \
  (_work)->data = _data;    \
 } while (0)
/*
 * initialize all of a work-struct:
 */
// 初始化工作结构, 和__WORK_INITIALIZER功能相同,不过__WORK_INITIALIZER用在
// 参数初始化定义, 而该宏用在程序之中对工作结构赋值
#define INIT_WORK(_work, _func, _data)    \
 do {       \
  INIT_LIST_HEAD(&(_work)->entry);  \
  (_work)->pending = 0;    \
  PREPARE_WORK((_work), (_func), (_data)); \
  init_timer(&(_work)->timer);   \
 } while (0)
 
4. 操作函数

4.1 创建工作队列

一般的创建函数是create_workqueue, 但这其实只是一个宏:
/* include/linux/workqueue.h */
#define create_workqueue(name) __create_workqueue((name), 0)
在workqueue的初始化函数中, 定义了一个针对内核中所有线程可用的事件工作队列, 其他内核线程建立的事件工作结构就都挂接到该队列:
void init_workqueues(void)
{
...
 keventd_wq = create_workqueue("events");
...
}

核心创建函数是__create_workqueue:

struct workqueue_struct *__create_workqueue(const char *name,
         int singlethread)
{
 int cpu, destroy = 0;
 struct workqueue_struct *wq;
 struct task_struct *p;
// 分配工作队列结构空间
 wq = kzalloc(sizeof(*wq), GFP_KERNEL);
 if (!wq)
  return NULL;
// 为每个CPU分配单独的工作队列空间
 wq->cpu_wq = alloc_percpu(struct cpu_workqueue_struct);
 if (!wq->cpu_wq) {
  kfree(wq);
  return NULL;
 }
 wq->name = name;
 mutex_lock(&workqueue_mutex);
 if (singlethread) {
// 使用create_workqueue宏时该参数始终为0
// 如果是单一线程模式, 在单线程中调用各个工作队列
// 建立一个的工作队列内核线程
  INIT_LIST_HEAD(&wq->list);
// 建立工作队列的线程
  p = create_workqueue_thread(wq, singlethread_cpu);
  if (!p)
   destroy = 1;
  else
// 唤醒该线程
   wake_up_process(p);
 } else {
// 链表模式, 将工作队列添加到工作队列链表
  list_add(&wq->list, &workqueues);
// 为每个CPU建立一个工作队列线程
  for_each_online_cpu(cpu) {
   p = create_workqueue_thread(wq, cpu);
   if (p) {
// 绑定CPU
    kthread_bind(p, cpu);
// 唤醒线程
    wake_up_process(p);
   } else
    destroy = 1;
  }
 }
 mutex_unlock(&workqueue_mutex);
 /*
  * Was there any error during startup? If yes then clean up:
  */
 if (destroy) {
// 建立线程失败, 释放工作队列
  destroy_workqueue(wq);
  wq = NULL;
 }
 return wq;
}
EXPORT_SYMBOL_GPL(__create_workqueue);

// 创建工作队列线程
static struct task_struct *create_workqueue_thread(struct workqueue_struct *wq,
         int cpu)
{
// 每个CPU的工作队列
 struct cpu_workqueue_struct *cwq = per_cpu_ptr(wq->cpu_wq, cpu);
 struct task_struct *p;
 spin_lock_init(&cwq->lock);
// 初始化
 cwq->wq = wq;
 cwq->thread = NULL;
 cwq->insert_sequence = 0;
 cwq->remove_sequence = 0;
 INIT_LIST_HEAD(&cwq->worklist);
// 初始化等待队列more_work, 该队列处理要执行的工作结构
 init_waitqueue_head(&cwq->more_work);
// 初始化等待队列work_done, 该队列处理执行完的工作结构
 init_waitqueue_head(&cwq->work_done);
// 建立内核线程work_thread
 if (is_single_threaded(wq))
  p = kthread_create(worker_thread, cwq, "%s", wq->name);
 else
  p = kthread_create(worker_thread, cwq, "%s/%d", wq->name, cpu);
 if (IS_ERR(p))
  return NULL;
// 保存线程指针
 cwq->thread = p;
 return p;
}
static int worker_thread(void *__cwq)
{
 struct cpu_workqueue_struct *cwq = __cwq;
// 声明一个等待队列
 DECLARE_WAITQUEUE(wait, current);
// 信号
 struct k_sigaction sa;
 sigset_t blocked;
 current->flags |= PF_NOFREEZE;
// 降低进程优先级, 工作进程不是个很紧急的进程,不和其他进程抢占CPU,通常在系统空闲时运行
 set_user_nice(current, -5);
 /* Block and flush all signals */
// 阻塞所有信号
 sigfillset(&blocked);
 sigprocmask(SIG_BLOCK, &blocked, NULL);
 flush_signals(current);
 /*
  * We inherited MPOL_INTERLEAVE from the booting kernel.
  * Set MPOL_DEFAULT to insure node local allocations.
  */
 numa_default_policy();
 /* SIG_IGN makes children autoreap: see do_notify_parent(). */
// 信号处理都是忽略
 sa.sa.sa_handler = SIG_IGN;
 sa.sa.sa_flags = 0;
 siginitset(&sa.sa.sa_mask, sigmask(SIGCHLD));
 do_sigaction(SIGCHLD, &sa, (struct k_sigaction *)0);
// 进程可中断
 set_current_state(TASK_INTERRUPTIBLE);
// 进入循环, 没明确停止该进程就一直运行
 while (!kthread_should_stop()) {
// 设置more_work等待队列, 当有新work结构链入队列中时会激发此等待队列
  add_wait_queue(&cwq->more_work, &wait);
  if (list_empty(&cwq->worklist))
// 工作队列为空, 睡眠
   schedule();
  else
// 进行运行状态
   __set_current_state(TASK_RUNNING);
// 删除等待队列
  remove_wait_queue(&cwq->more_work, &wait);
// 按链表遍历执行工作任务
  if (!list_empty(&cwq->worklist))
   run_workqueue(cwq);
// 执行完工作, 设置进程是可中断的, 重新循环等待工作
  set_current_state(TASK_INTERRUPTIBLE);
 }
 __set_current_state(TASK_RUNNING);
 return 0;
}

// 运行工作结构
static void run_workqueue(struct cpu_workqueue_struct *cwq)
{
 unsigned long flags;
 /*
  * Keep taking off work from the queue until
  * done.
  */
// 加锁
 spin_lock_irqsave(&cwq->lock, flags);
// 统计已经递归调用了多少次了
 cwq->run_depth++;
 if (cwq->run_depth > 3) {
// 递归调用此时太多
  /* morton gets to eat his hat */
  printk("%s: recursion depth exceeded: %d\n",
   __FUNCTION__, cwq->run_depth);
  dump_stack();
 }
// 遍历工作链表
 while (!list_empty(&cwq->worklist)) {
// 获取的是next节点的
  struct work_struct *work = list_entry(cwq->worklist.next,
      struct work_struct, entry);
  void (*f) (void *) = work->func;
  void *data = work->data;
// 删除节点, 同时节点中的list参数清空
  list_del_init(cwq->worklist.next);
// 解锁
// 现在在执行以下代码时可以中断,run_workqueue本身可能会重新被调用, 所以要判断递归深度
  spin_unlock_irqrestore(&cwq->lock, flags);
  BUG_ON(work->wq_data != cwq);
// 工作结构已经不在链表中
  clear_bit(0, &work->pending);
// 执行工作函数
  f(data);
// 重新加锁
  spin_lock_irqsave(&cwq->lock, flags);
// 执行完的工作序列号递增
  cwq->remove_sequence++;
// 唤醒工作完成等待队列, 供释放工作队列
  wake_up(&cwq->work_done);
 }
// 减少递归深度
 cwq->run_depth--;
// 解锁
 spin_unlock_irqrestore(&cwq->lock, flags);
}

4.2 释放工作队列
/**
 * destroy_workqueue - safely terminate a workqueue
 * @wq: target workqueue
 *
 * Safely destroy a workqueue. All work currently pending will be done first.
 */
void destroy_workqueue(struct workqueue_struct *wq)
{
 int cpu;
// 清除当前工作队列中的所有工作
 flush_workqueue(wq);
 /* We don't need the distraction of CPUs appearing and vanishing. */
 mutex_lock(&workqueue_mutex);
// 结束该工作队列的线程
 if (is_single_threaded(wq))
  cleanup_workqueue_thread(wq, singlethread_cpu);
 else {
  for_each_online_cpu(cpu)
   cleanup_workqueue_thread(wq, cpu);
  list_del(&wq->list);
 }
 mutex_unlock(&workqueue_mutex);
// 释放工作队列中对应每个CPU的工作队列数据
 free_percpu(wq->cpu_wq);
 kfree(wq);
}
EXPORT_SYMBOL_GPL(destroy_workqueue);

/**
 * flush_workqueue - ensure that any scheduled work has run to completion.
 * @wq: workqueue to flush
 *
 * Forces execution of the workqueue and blocks until its completion.
 * This is typically used in driver shutdown handlers.
 *
 * This function will sample each workqueue's current insert_sequence number and
 * will sleep until the head sequence is greater than or equal to that.  This
 * means that we sleep until all works which were queued on entry have been
 * handled, but we are not livelocked by new incoming ones.
 *
 * This function used to run the workqueues itself.  Now we just wait for the
 * helper threads to do it.
 */
void fastcall flush_workqueue(struct workqueue_struct *wq)
{
// 该进程可以睡眠
 might_sleep();
// 清空每个CPU上的工作队列
 if (is_single_threaded(wq)) {
  /* Always use first cpu's area. */
  flush_cpu_workqueue(per_cpu_ptr(wq->cpu_wq, singlethread_cpu));
 } else {
  int cpu;
  mutex_lock(&workqueue_mutex);
  for_each_online_cpu(cpu)
   flush_cpu_workqueue(per_cpu_ptr(wq->cpu_wq, cpu));
  mutex_unlock(&workqueue_mutex);
 }
}
EXPORT_SYMBOL_GPL(flush_workqueue);
 
flush_workqueue的核心处理函数为flush_cpu_workqueue:
static void flush_cpu_workqueue(struct cpu_workqueue_struct *cwq)
{
 if (cwq->thread == current) {
// 如果是工作队列进程正在被调度
  /*
   * Probably keventd trying to flush its own queue. So simply run
   * it by hand rather than deadlocking.
   */
// 执行完该工作队列
  run_workqueue(cwq);
 } else {
// 定义等待
  DEFINE_WAIT(wait);
  long sequence_needed;
// 加锁
  spin_lock_irq(&cwq->lock);
// 最新工作结构序号
  sequence_needed = cwq->insert_sequence;
// 该条件是判断队列中是否还有没有执行的工作结构
  while (sequence_needed - cwq->remove_sequence > 0) {
// 有为执行的工作结构
// 通过work_done等待队列等待
   prepare_to_wait(&cwq->work_done, &wait,
     TASK_UNINTERRUPTIBLE);
// 解锁
   spin_unlock_irq(&cwq->lock);
// 睡眠, 由wake_up(&cwq->work_done)来唤醒
   schedule();
// 重新加锁
   spin_lock_irq(&cwq->lock);
  }
// 等待清除
  finish_wait(&cwq->work_done, &wait);
  spin_unlock_irq(&cwq->lock);
 }
}
 
4.3 调度工作

在大多数情况下, 并不需要自己建立工作队列,而是只定义工作, 将工作结构挂接到内核预定义的事件工作队列中调度, 在kernel/workqueue.c中定义了一个静态全局量的工作队列keventd_wq:
static struct workqueue_struct *keventd_wq;
 
4.3.1 立即调度
// 在其他函数中使用以下函数来调度工作结构, 是把工作结构挂接到工作队列中进行调度
/**
 * schedule_work - put work task in global workqueue
 * @work: job to be done
 *
 * This puts a job in the kernel-global workqueue.
 */
// 调度工作结构, 将工作结构添加到事件工作队列keventd_wq
int fastcall schedule_work(struct work_struct *work)
{
 return queue_work(keventd_wq, work);
}
EXPORT_SYMBOL(schedule_work);
 
/**
 * queue_work - queue work on a workqueue
 * @wq: workqueue to use
 * @work: work to queue
 *
 * Returns 0 if @work was already on a queue, non-zero otherwise.
 *
 * We queue the work to the CPU it was submitted, but there is no
 * guarantee that it will be processed by that CPU.
 */
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)) {
// 工作结构还没在队列, 设置pending标志表示把工作结构挂接到队列中
  if (unlikely(is_single_threaded(wq)))
   cpu = singlethread_cpu;
  BUG_ON(!list_empty(&work->entry));
// 进行具体的排队
  __queue_work(per_cpu_ptr(wq->cpu_wq, cpu), work);
  ret = 1;
 }
 put_cpu();
 return ret;
}
EXPORT_SYMBOL_GPL(queue_work);
/* Preempt must be disabled. */
// 不能被抢占
static void __queue_work(struct cpu_workqueue_struct *cwq,
    struct work_struct *work)
{
 unsigned long flags;
// 加锁
 spin_lock_irqsave(&cwq->lock, flags);
// 指向CPU工作队列
 work->wq_data = cwq;
// 挂接到工作链表
 list_add_tail(&work->entry, &cwq->worklist);
// 递增插入的序列号
 cwq->insert_sequence++;
// 唤醒等待队列准备处理工作结构
 wake_up(&cwq->more_work);
 spin_unlock_irqrestore(&cwq->lock, flags);
}

4.3.2 延迟调度

4.3.2.1 schedule_delayed_work
/**
 * schedule_delayed_work - put work task in global workqueue after delay
 * @work: 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.
 */
// 延迟调度工作, 延迟一定时间后再将工作结构挂接到工作队列
int fastcall schedule_delayed_work(struct work_struct *work, unsigned long delay)
{
 return queue_delayed_work(keventd_wq, work, delay);
}
EXPORT_SYMBOL(schedule_delayed_work);

/**
 * queue_delayed_work - queue work on a workqueue after delay
 * @wq: workqueue to use
 * @work: 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)
{
 int ret = 0;
// 定时器, 此时的定时器应该是不起效的, 延迟将通过该定时器来实现
 struct timer_list *timer = &work->timer;
 if (!test_and_set_bit(0, &work->pending)) {
// 工作结构还没在队列, 设置pending标志表示把工作结构挂接到队列中
// 如果现在定时器已经起效, 出错
  BUG_ON(timer_pending(timer));
// 工作结构已经挂接到链表, 出错
  BUG_ON(!list_empty(&work->entry));
  /* This stores wq for the moment, for the timer_fn */
// 保存工作队列的指针
  work->wq_data = wq;
// 定时器初始化
  timer->expires = jiffies + delay;
  timer->data = (unsigned long)work;
// 定时函数
  timer->function = delayed_work_timer_fn;
// 定时器生效, 定时到期后再添加到工作队列
  add_timer(timer);
  ret = 1;
 }
 return ret;
}
EXPORT_SYMBOL_GPL(queue_delayed_work);
 

// 定时中断函数
static void delayed_work_timer_fn(unsigned long __data)
{
 struct work_struct *work = (struct work_struct *)__data;
 struct workqueue_struct *wq = work->wq_data;
// 获取CPU
 int cpu = smp_processor_id();
 if (unlikely(is_single_threaded(wq)))
  cpu = singlethread_cpu;
// 将工作结构添加到工作队列,注意这是在时间中断调用
 __queue_work(per_cpu_ptr(wq->cpu_wq, cpu), work);
}

4.3.2.2 schedule_delayed_work_on

指定CPU的延迟调度工作结构, 和schedule_delayed_work相比增加了一个CPU参数, 其他都相同
/**
 * schedule_delayed_work_on - queue work in global workqueue on CPU after delay
 * @cpu: cpu to use
 * @work: 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)
{
 return queue_delayed_work_on(cpu, keventd_wq, work, delay);
}

/**
 * queue_delayed_work_on - queue work on specific CPU after delay
 * @cpu: CPU number to execute work on
 * @wq: workqueue to use
 * @work: work to queue
 * @delay: number of jiffies to wait before queueing
 *
 * 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)
{
 int ret = 0;
 struct timer_list *timer = &work->timer;
 if (!test_and_set_bit(0, &work->pending)) {
  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;
  timer->expires = jiffies + delay;
  timer->data = (unsigned long)work;
  timer->function = delayed_work_timer_fn;
  add_timer_on(timer, cpu);
  ret = 1;
 }
 return ret;
}
EXPORT_SYMBOL_GPL(queue_delayed_work_on);
 
5. 结论

工作队列和定时器函数处理有点类似, 都是执行一定的回调函数, 但和定时器处理函数不同的是定时器回调函数只执行一次, 而且执行定时器回调函数的时候是在时钟中断中, 限制比较多, 因此回调程序不能太复杂; 而工作队列是通过内核线程实现,  一直有效, 可重复执行, 由于执行时降低了线程的优先级, 执行时可能休眠, 因此工作队列处理的应该是那些不是很紧急的任务, 如垃圾回收处理等, 通常在系统空闲时执行，在xfrm库中就广泛使用了workqueue，使用时，只需要定义work结构，然后调用 schedule_(delayed_)work即可。