Fix stop_machine_run problem with naughty real time process
[sfrench/cifs-2.6.git] / kernel / workqueue.c
index b6fa5e63085d65b42f46fe97735751b6f402cab4..3bebf73be976592cab56a218e93b7449a5de7df4 100644 (file)
 /*
  * 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.
  */
 struct cpu_workqueue_struct {
 
        spinlock_t lock;
 
-       long remove_sequence;   /* Least-recently added (next to run) */
-       long insert_sequence;   /* Next to add */
-
        struct list_head worklist;
        wait_queue_head_t more_work;
-       wait_queue_head_t work_done;
+       struct work_struct *current_work;
 
        struct workqueue_struct *wq;
        struct task_struct *thread;
 
        int run_depth;          /* Detect run_workqueue() recursion depth */
-
-       int freezeable;         /* Freeze the thread during suspend */
 } ____cacheline_aligned;
 
 /*
@@ -68,8 +57,10 @@ struct cpu_workqueue_struct {
  */
 struct workqueue_struct {
        struct cpu_workqueue_struct *cpu_wq;
+       struct list_head list;
        const char *name;
-       struct list_head list;  /* Empty if single thread */
+       int singlethread;
+       int freezeable;         /* Freeze threads during suspend */
 };
 
 /* All the per-cpu workqueues on the system, for hotplug cpu to add/remove
@@ -77,106 +68,74 @@ struct workqueue_struct {
 static DEFINE_MUTEX(workqueue_mutex);
 static LIST_HEAD(workqueues);
 
-static int singlethread_cpu;
+static int singlethread_cpu __read_mostly;
+static cpumask_t cpu_singlethread_map __read_mostly;
+/*
+ * _cpu_down() first removes CPU from cpu_online_map, then CPU_DEAD
+ * flushes cwq->worklist. This means that flush_workqueue/wait_on_work
+ * which comes in between can't use for_each_online_cpu(). We could
+ * use cpu_possible_map, the cpumask below is more a documentation
+ * than optimization.
+ */
+static cpumask_t cpu_populated_map __read_mostly;
 
 /* If it's single threaded, it isn't in the list of workqueues. */
 static inline int is_single_threaded(struct workqueue_struct *wq)
 {
-       return list_empty(&wq->list);
+       return wq->singlethread;
+}
+
+static const cpumask_t *wq_cpu_map(struct workqueue_struct *wq)
+{
+       return is_single_threaded(wq)
+               ? &cpu_singlethread_map : &cpu_populated_map;
+}
+
+static
+struct cpu_workqueue_struct *wq_per_cpu(struct workqueue_struct *wq, int cpu)
+{
+       if (unlikely(is_single_threaded(wq)))
+               cpu = singlethread_cpu;
+       return per_cpu_ptr(wq->cpu_wq, cpu);
 }
 
 /*
  * Set the workqueue on which a work item is to be run
  * - Must *only* be called if the pending flag is set
  */
-static inline void set_wq_data(struct work_struct *work, void *wq)
+static inline void set_wq_data(struct work_struct *work,
+                               struct cpu_workqueue_struct *cwq)
 {
        unsigned long new;
 
        BUG_ON(!work_pending(work));
 
-       new = (unsigned long) wq | (1UL << WORK_STRUCT_PENDING);
+       new = (unsigned long) cwq | (1UL << WORK_STRUCT_PENDING);
        new |= WORK_STRUCT_FLAG_MASK & *work_data_bits(work);
        atomic_long_set(&work->data, new);
 }
 
-static inline void *get_wq_data(struct work_struct *work)
+static inline
+struct cpu_workqueue_struct *get_wq_data(struct work_struct *work)
 {
        return (void *) (atomic_long_read(&work->data) & WORK_STRUCT_WQ_DATA_MASK);
 }
 
-static int __run_work(struct cpu_workqueue_struct *cwq, struct work_struct *work)
+static void insert_work(struct cpu_workqueue_struct *cwq,
+                               struct work_struct *work, int tail)
 {
-       int ret = 0;
-       unsigned long flags;
-
-       spin_lock_irqsave(&cwq->lock, flags);
+       set_wq_data(work, cwq);
        /*
-        * We need to re-validate the work info after we've gotten
-        * the cpu_workqueue lock. We can run the work now iff:
-        *
-        *  - the wq_data still matches the cpu_workqueue_struct
-        *  - AND the work is still marked pending
-        *  - AND the work is still on a list (which will be this
-        *    workqueue_struct list)
-        *
-        * All these conditions are important, because we
-        * need to protect against the work being run right
-        * now on another CPU (all but the last one might be
-        * true if it's currently running and has not been
-        * released yet, for example).
+        * Ensure that we get the right work->data if we see the
+        * result of list_add() below, see try_to_grab_pending().
         */
-       if (get_wq_data(work) == cwq
-           && work_pending(work)
-           && !list_empty(&work->entry)) {
-               work_func_t f = work->func;
-               list_del_init(&work->entry);
-               spin_unlock_irqrestore(&cwq->lock, flags);
-
-               if (!test_bit(WORK_STRUCT_NOAUTOREL, work_data_bits(work)))
-                       work_release(work);
-               f(work);
-
-               spin_lock_irqsave(&cwq->lock, flags);
-               cwq->remove_sequence++;
-               wake_up(&cwq->work_done);
-               ret = 1;
-       }
-       spin_unlock_irqrestore(&cwq->lock, flags);
-       return ret;
-}
-
-/**
- * run_scheduled_work - run scheduled work synchronously
- * @work: work to run
- *
- * This checks if the work was pending, and runs it
- * synchronously if so. It returns a boolean to indicate
- * whether it had any scheduled work to run or not.
- *
- * NOTE! This _only_ works for normal work_structs. You
- * CANNOT use this for delayed work, because the wq data
- * for delayed work will not point properly to the per-
- * CPU workqueue struct, but will change!
- */
-int fastcall run_scheduled_work(struct work_struct *work)
-{
-       for (;;) {
-               struct cpu_workqueue_struct *cwq;
-
-               if (!work_pending(work))
-                       return 0;
-               if (list_empty(&work->entry))
-                       return 0;
-               /* NOTE! This depends intimately on __queue_work! */
-               cwq = get_wq_data(work);
-               if (!cwq)
-                       return 0;
-               if (__run_work(cwq, work))
-                       return 1;
-       }
+       smp_wmb();
+       if (tail)
+               list_add_tail(&work->entry, &cwq->worklist);
+       else
+               list_add(&work->entry, &cwq->worklist);
+       wake_up(&cwq->more_work);
 }
-EXPORT_SYMBOL(run_scheduled_work);
 
 /* Preempt must be disabled. */
 static void __queue_work(struct cpu_workqueue_struct *cwq,
@@ -185,10 +144,7 @@ static void __queue_work(struct cpu_workqueue_struct *cwq,
        unsigned long flags;
 
        spin_lock_irqsave(&cwq->lock, flags);
-       set_wq_data(work, cwq);
-       list_add_tail(&work->entry, &cwq->worklist);
-       cwq->insert_sequence++;
-       wake_up(&cwq->more_work);
+       insert_work(cwq, work, 1);
        spin_unlock_irqrestore(&cwq->lock, flags);
 }
 
@@ -204,16 +160,14 @@ static void __queue_work(struct cpu_workqueue_struct *cwq,
  */
 int fastcall queue_work(struct workqueue_struct *wq, struct work_struct *work)
 {
-       int ret = 0, cpu = get_cpu();
+       int ret = 0;
 
        if (!test_and_set_bit(WORK_STRUCT_PENDING, work_data_bits(work))) {
-               if (unlikely(is_single_threaded(wq)))
-                       cpu = singlethread_cpu;
                BUG_ON(!list_empty(&work->entry));
-               __queue_work(per_cpu_ptr(wq->cpu_wq, cpu), work);
+               __queue_work(wq_per_cpu(wq, get_cpu()), work);
+               put_cpu();
                ret = 1;
        }
-       put_cpu();
        return ret;
 }
 EXPORT_SYMBOL_GPL(queue_work);
@@ -221,13 +175,10 @@ EXPORT_SYMBOL_GPL(queue_work);
 void delayed_work_timer_fn(unsigned long __data)
 {
        struct delayed_work *dwork = (struct delayed_work *)__data;
-       struct workqueue_struct *wq = get_wq_data(&dwork->work);
-       int cpu = smp_processor_id();
-
-       if (unlikely(is_single_threaded(wq)))
-               cpu = singlethread_cpu;
+       struct cpu_workqueue_struct *cwq = get_wq_data(&dwork->work);
+       struct workqueue_struct *wq = cwq->wq;
 
-       __queue_work(per_cpu_ptr(wq->cpu_wq, cpu), &dwork->work);
+       __queue_work(wq_per_cpu(wq, smp_processor_id()), &dwork->work);
 }
 
 /**
@@ -241,27 +192,11 @@ void delayed_work_timer_fn(unsigned long __data)
 int fastcall queue_delayed_work(struct workqueue_struct *wq,
                        struct delayed_work *dwork, unsigned long delay)
 {
-       int ret = 0;
-       struct timer_list *timer = &dwork->timer;
-       struct work_struct *work = &dwork->work;
-
-       timer_stats_timer_set_start_info(timer);
+       timer_stats_timer_set_start_info(&dwork->timer);
        if (delay == 0)
-               return queue_work(wq, work);
+               return queue_work(wq, &dwork->work);
 
-       if (!test_and_set_bit(WORK_STRUCT_PENDING, work_data_bits(work))) {
-               BUG_ON(timer_pending(timer));
-               BUG_ON(!list_empty(&work->entry));
-
-               /* This stores wq for the moment, for the timer_fn */
-               set_wq_data(work, wq);
-               timer->expires = jiffies + delay;
-               timer->data = (unsigned long)dwork;
-               timer->function = delayed_work_timer_fn;
-               add_timer(timer);
-               ret = 1;
-       }
-       return ret;
+       return queue_delayed_work_on(-1, wq, dwork, delay);
 }
 EXPORT_SYMBOL_GPL(queue_delayed_work);
 
@@ -285,12 +220,16 @@ int queue_delayed_work_on(int cpu, struct workqueue_struct *wq,
                BUG_ON(timer_pending(timer));
                BUG_ON(!list_empty(&work->entry));
 
-               /* This stores wq for the moment, for the timer_fn */
-               set_wq_data(work, wq);
+               /* This stores cwq for the moment, for the timer_fn */
+               set_wq_data(work, wq_per_cpu(wq, raw_smp_processor_id()));
                timer->expires = jiffies + delay;
                timer->data = (unsigned long)dwork;
                timer->function = delayed_work_timer_fn;
-               add_timer_on(timer, cpu);
+
+               if (unlikely(cpu >= 0))
+                       add_timer_on(timer, cpu);
+               else
+                       add_timer(timer);
                ret = 1;
        }
        return ret;
@@ -299,13 +238,7 @@ EXPORT_SYMBOL_GPL(queue_delayed_work_on);
 
 static void run_workqueue(struct cpu_workqueue_struct *cwq)
 {
-       unsigned long flags;
-
-       /*
-        * Keep taking off work from the queue until
-        * done.
-        */
-       spin_lock_irqsave(&cwq->lock, flags);
+       spin_lock_irq(&cwq->lock);
        cwq->run_depth++;
        if (cwq->run_depth > 3) {
                /* morton gets to eat his hat */
@@ -318,12 +251,12 @@ static void run_workqueue(struct cpu_workqueue_struct *cwq)
                                                struct work_struct, entry);
                work_func_t f = work->func;
 
+               cwq->current_work = work;
                list_del_init(cwq->worklist.next);
-               spin_unlock_irqrestore(&cwq->lock, flags);
+               spin_unlock_irq(&cwq->lock);
 
                BUG_ON(get_wq_data(work) != cwq);
-               if (!test_bit(WORK_STRUCT_NOAUTOREL, work_data_bits(work)))
-                       work_release(work);
+               work_clear_pending(work);
                f(work);
 
                if (unlikely(in_atomic() || lockdep_depth(current) > 0)) {
@@ -337,88 +270,91 @@ static void run_workqueue(struct cpu_workqueue_struct *cwq)
                        dump_stack();
                }
 
-               spin_lock_irqsave(&cwq->lock, flags);
-               cwq->remove_sequence++;
-               wake_up(&cwq->work_done);
+               spin_lock_irq(&cwq->lock);
+               cwq->current_work = NULL;
        }
        cwq->run_depth--;
-       spin_unlock_irqrestore(&cwq->lock, flags);
+       spin_unlock_irq(&cwq->lock);
 }
 
 static int worker_thread(void *__cwq)
 {
        struct cpu_workqueue_struct *cwq = __cwq;
-       DECLARE_WAITQUEUE(wait, current);
-       struct k_sigaction sa;
-       sigset_t blocked;
+       DEFINE_WAIT(wait);
 
-       if (!cwq->freezeable)
+       if (!cwq->wq->freezeable)
                current->flags |= PF_NOFREEZE;
 
        set_user_nice(current, -5);
 
-       /* Block and flush all signals */
-       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();
+       for (;;) {
+               prepare_to_wait(&cwq->more_work, &wait, TASK_INTERRUPTIBLE);
+               if (!freezing(current) &&
+                   !kthread_should_stop() &&
+                   list_empty(&cwq->worklist))
+                       schedule();
+               finish_wait(&cwq->more_work, &wait);
 
-       /* SIG_IGN makes children autoreap: see do_notify_parent(). */
-       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);
+               try_to_freeze();
 
-       set_current_state(TASK_INTERRUPTIBLE);
-       while (!kthread_should_stop()) {
-               if (cwq->freezeable)
-                       try_to_freeze();
+               if (kthread_should_stop())
+                       break;
 
-               add_wait_queue(&cwq->more_work, &wait);
-               if (list_empty(&cwq->worklist))
-                       schedule();
-               else
-                       __set_current_state(TASK_RUNNING);
-               remove_wait_queue(&cwq->more_work, &wait);
-
-               if (!list_empty(&cwq->worklist))
-                       run_workqueue(cwq);
-               set_current_state(TASK_INTERRUPTIBLE);
+               run_workqueue(cwq);
        }
-       __set_current_state(TASK_RUNNING);
+
        return 0;
 }
 
-static void flush_cpu_workqueue(struct cpu_workqueue_struct *cwq)
+struct wq_barrier {
+       struct work_struct      work;
+       struct completion       done;
+};
+
+static void wq_barrier_func(struct work_struct *work)
+{
+       struct wq_barrier *barr = container_of(work, struct wq_barrier, work);
+       complete(&barr->done);
+}
+
+static void insert_wq_barrier(struct cpu_workqueue_struct *cwq,
+                                       struct wq_barrier *barr, int tail)
+{
+       INIT_WORK(&barr->work, wq_barrier_func);
+       __set_bit(WORK_STRUCT_PENDING, work_data_bits(&barr->work));
+
+       init_completion(&barr->done);
+
+       insert_work(cwq, &barr->work, tail);
+}
+
+static int flush_cpu_workqueue(struct cpu_workqueue_struct *cwq)
 {
+       int active;
+
        if (cwq->thread == current) {
                /*
                 * Probably keventd trying to flush its own queue. So simply run
                 * it by hand rather than deadlocking.
                 */
                run_workqueue(cwq);
+               active = 1;
        } else {
-               DEFINE_WAIT(wait);
-               long sequence_needed;
+               struct wq_barrier barr;
 
+               active = 0;
                spin_lock_irq(&cwq->lock);
-               sequence_needed = cwq->insert_sequence;
-
-               while (sequence_needed - cwq->remove_sequence > 0) {
-                       prepare_to_wait(&cwq->work_done, &wait,
-                                       TASK_UNINTERRUPTIBLE);
-                       spin_unlock_irq(&cwq->lock);
-                       schedule();
-                       spin_lock_irq(&cwq->lock);
+               if (!list_empty(&cwq->worklist) || cwq->current_work != NULL) {
+                       insert_wq_barrier(cwq, &barr, 1);
+                       active = 1;
                }
-               finish_wait(&cwq->work_done, &wait);
                spin_unlock_irq(&cwq->lock);
+
+               if (active)
+                       wait_for_completion(&barr.done);
        }
+
+       return active;
 }
 
 /**
@@ -428,151 +364,145 @@ static void flush_cpu_workqueue(struct cpu_workqueue_struct *cwq)
  * 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.
+ * 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)
 {
+       const cpumask_t *cpu_map = wq_cpu_map(wq);
+       int cpu;
+
        might_sleep();
+       for_each_cpu_mask(cpu, *cpu_map)
+               flush_cpu_workqueue(per_cpu_ptr(wq->cpu_wq, cpu));
+}
+EXPORT_SYMBOL_GPL(flush_workqueue);
 
-       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;
+/*
+ * Upon a successful return, the caller "owns" WORK_STRUCT_PENDING bit,
+ * so this work can't be re-armed in any way.
+ */
+static int try_to_grab_pending(struct work_struct *work)
+{
+       struct cpu_workqueue_struct *cwq;
+       int ret = 0;
 
-               mutex_lock(&workqueue_mutex);
-               for_each_online_cpu(cpu)
-                       flush_cpu_workqueue(per_cpu_ptr(wq->cpu_wq, cpu));
-               mutex_unlock(&workqueue_mutex);
+       if (!test_and_set_bit(WORK_STRUCT_PENDING, work_data_bits(work)))
+               return 1;
+
+       /*
+        * The queueing is in progress, or it is already queued. Try to
+        * steal it from ->worklist without clearing WORK_STRUCT_PENDING.
+        */
+
+       cwq = get_wq_data(work);
+       if (!cwq)
+               return ret;
+
+       spin_lock_irq(&cwq->lock);
+       if (!list_empty(&work->entry)) {
+               /*
+                * This work is queued, but perhaps we locked the wrong cwq.
+                * In that case we must see the new value after rmb(), see
+                * insert_work()->wmb().
+                */
+               smp_rmb();
+               if (cwq == get_wq_data(work)) {
+                       list_del_init(&work->entry);
+                       ret = 1;
+               }
        }
+       spin_unlock_irq(&cwq->lock);
+
+       return ret;
 }
-EXPORT_SYMBOL_GPL(flush_workqueue);
 
-static struct task_struct *create_workqueue_thread(struct workqueue_struct *wq,
-                                                  int cpu, int freezeable)
+static void wait_on_cpu_work(struct cpu_workqueue_struct *cwq,
+                               struct work_struct *work)
 {
-       struct cpu_workqueue_struct *cwq = per_cpu_ptr(wq->cpu_wq, cpu);
-       struct task_struct *p;
+       struct wq_barrier barr;
+       int running = 0;
 
-       spin_lock_init(&cwq->lock);
-       cwq->wq = wq;
-       cwq->thread = NULL;
-       cwq->insert_sequence = 0;
-       cwq->remove_sequence = 0;
-       cwq->freezeable = freezeable;
-       INIT_LIST_HEAD(&cwq->worklist);
-       init_waitqueue_head(&cwq->more_work);
-       init_waitqueue_head(&cwq->work_done);
+       spin_lock_irq(&cwq->lock);
+       if (unlikely(cwq->current_work == work)) {
+               insert_wq_barrier(cwq, &barr, 0);
+               running = 1;
+       }
+       spin_unlock_irq(&cwq->lock);
 
-       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;
-       cwq->thread = p;
-       return p;
+       if (unlikely(running))
+               wait_for_completion(&barr.done);
 }
 
-struct workqueue_struct *__create_workqueue(const char *name,
-                                           int singlethread, int freezeable)
+static void wait_on_work(struct work_struct *work)
 {
-       int cpu, destroy = 0;
+       struct cpu_workqueue_struct *cwq;
        struct workqueue_struct *wq;
-       struct task_struct *p;
+       const cpumask_t *cpu_map;
+       int cpu;
 
-       wq = kzalloc(sizeof(*wq), GFP_KERNEL);
-       if (!wq)
-               return NULL;
+       might_sleep();
 
-       wq->cpu_wq = alloc_percpu(struct cpu_workqueue_struct);
-       if (!wq->cpu_wq) {
-               kfree(wq);
-               return NULL;
-       }
+       cwq = get_wq_data(work);
+       if (!cwq)
+               return;
 
-       wq->name = name;
-       mutex_lock(&workqueue_mutex);
-       if (singlethread) {
-               INIT_LIST_HEAD(&wq->list);
-               p = create_workqueue_thread(wq, singlethread_cpu, freezeable);
-               if (!p)
-                       destroy = 1;
-               else
-                       wake_up_process(p);
-       } else {
-               list_add(&wq->list, &workqueues);
-               for_each_online_cpu(cpu) {
-                       p = create_workqueue_thread(wq, cpu, freezeable);
-                       if (p) {
-                               kthread_bind(p, cpu);
-                               wake_up_process(p);
-                       } else
-                               destroy = 1;
-               }
-       }
-       mutex_unlock(&workqueue_mutex);
+       wq = cwq->wq;
+       cpu_map = wq_cpu_map(wq);
 
-       /*
-        * Was there any error during startup? If yes then clean up:
-        */
-       if (destroy) {
-               destroy_workqueue(wq);
-               wq = NULL;
-       }
-       return wq;
+       for_each_cpu_mask(cpu, *cpu_map)
+               wait_on_cpu_work(per_cpu_ptr(wq->cpu_wq, cpu), work);
 }
-EXPORT_SYMBOL_GPL(__create_workqueue);
 
-static void cleanup_workqueue_thread(struct workqueue_struct *wq, int cpu)
+/**
+ * cancel_work_sync - block until a work_struct's callback has terminated
+ * @work: the work which is to be flushed
+ *
+ * cancel_work_sync() will cancel the work if it is queued. If the work's
+ * callback appears to be running, cancel_work_sync() will block until it
+ * has completed.
+ *
+ * It is possible to use this function if the work re-queues itself. It can
+ * cancel the work even if it migrates to another workqueue, however in that
+ * case it only guarantees that work->func() has completed on the last queued
+ * workqueue.
+ *
+ * cancel_work_sync(&delayed_work->work) should be used only if ->timer is not
+ * pending, otherwise it goes into a busy-wait loop until the timer expires.
+ *
+ * The caller must ensure that workqueue_struct on which this work was last
+ * queued can't be destroyed before this function returns.
+ */
+void cancel_work_sync(struct work_struct *work)
 {
-       struct cpu_workqueue_struct *cwq;
-       unsigned long flags;
-       struct task_struct *p;
-
-       cwq = per_cpu_ptr(wq->cpu_wq, cpu);
-       spin_lock_irqsave(&cwq->lock, flags);
-       p = cwq->thread;
-       cwq->thread = NULL;
-       spin_unlock_irqrestore(&cwq->lock, flags);
-       if (p)
-               kthread_stop(p);
+       while (!try_to_grab_pending(work))
+               cpu_relax();
+       wait_on_work(work);
+       work_clear_pending(work);
 }
+EXPORT_SYMBOL_GPL(cancel_work_sync);
 
 /**
- * destroy_workqueue - safely terminate a workqueue
- * @wq: target workqueue
+ * cancel_rearming_delayed_work - reliably kill off a delayed work.
+ * @dwork: the delayed work struct
  *
- * Safely destroy a workqueue. All work currently pending will be done first.
+ * It is possible to use this function if @dwork rearms itself via queue_work()
+ * or queue_delayed_work(). See also the comment for cancel_work_sync().
  */
-void destroy_workqueue(struct workqueue_struct *wq)
+void cancel_rearming_delayed_work(struct delayed_work *dwork)
 {
-       int cpu;
-
-       flush_workqueue(wq);
-
-       /* We don't need the distraction of CPUs appearing and vanishing. */
-       mutex_lock(&workqueue_mutex);
-       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);
-       free_percpu(wq->cpu_wq);
-       kfree(wq);
+       while (!del_timer(&dwork->timer) &&
+              !try_to_grab_pending(&dwork->work))
+               cpu_relax();
+       wait_on_work(&dwork->work);
+       work_clear_pending(&dwork->work);
 }
-EXPORT_SYMBOL_GPL(destroy_workqueue);
+EXPORT_SYMBOL(cancel_rearming_delayed_work);
 
-static struct workqueue_struct *keventd_wq;
+static struct workqueue_struct *keventd_wq __read_mostly;
 
 /**
  * schedule_work - put work task in global workqueue
@@ -638,7 +568,7 @@ int schedule_on_each_cpu(work_func_t func)
        if (!works)
                return -ENOMEM;
 
-       mutex_lock(&workqueue_mutex);
+       preempt_disable();              /* CPU hotplug */
        for_each_online_cpu(cpu) {
                struct work_struct *work = per_cpu_ptr(works, cpu);
 
@@ -646,7 +576,7 @@ int schedule_on_each_cpu(work_func_t func)
                set_bit(WORK_STRUCT_PENDING, work_data_bits(work));
                __queue_work(per_cpu_ptr(keventd_wq->cpu_wq, cpu), work);
        }
-       mutex_unlock(&workqueue_mutex);
+       preempt_enable();
        flush_workqueue(keventd_wq);
        free_percpu(works);
        return 0;
@@ -658,29 +588,6 @@ void flush_scheduled_work(void)
 }
 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
- * @dwork: the delayed work struct
- */
-void cancel_rearming_delayed_workqueue(struct workqueue_struct *wq,
-                                      struct delayed_work *dwork)
-{
-       while (!cancel_delayed_work(dwork))
-               flush_workqueue(wq);
-}
-EXPORT_SYMBOL(cancel_rearming_delayed_workqueue);
-
-/**
- * cancel_rearming_delayed_work - reliably kill off a delayed keventd work whose handler rearms the delayed work.
- * @dwork: the delayed work struct
- */
-void cancel_rearming_delayed_work(struct delayed_work *dwork)
-{
-       cancel_rearming_delayed_workqueue(keventd_wq, dwork);
-}
-EXPORT_SYMBOL(cancel_rearming_delayed_work);
-
 /**
  * execute_in_process_context - reliably execute the routine with user context
  * @fn:                the function to execute
@@ -728,94 +635,206 @@ int current_is_keventd(void)
 
 }
 
-/* Take the work from this (downed) CPU. */
-static void take_over_work(struct workqueue_struct *wq, unsigned int cpu)
+static struct cpu_workqueue_struct *
+init_cpu_workqueue(struct workqueue_struct *wq, int cpu)
 {
        struct cpu_workqueue_struct *cwq = per_cpu_ptr(wq->cpu_wq, cpu);
-       struct list_head list;
-       struct work_struct *work;
 
-       spin_lock_irq(&cwq->lock);
-       list_replace_init(&cwq->worklist, &list);
+       cwq->wq = wq;
+       spin_lock_init(&cwq->lock);
+       INIT_LIST_HEAD(&cwq->worklist);
+       init_waitqueue_head(&cwq->more_work);
+
+       return cwq;
+}
+
+static int create_workqueue_thread(struct cpu_workqueue_struct *cwq, int cpu)
+{
+       struct workqueue_struct *wq = cwq->wq;
+       const char *fmt = is_single_threaded(wq) ? "%s" : "%s/%d";
+       struct task_struct *p;
+
+       p = kthread_create(worker_thread, cwq, fmt, wq->name, cpu);
+       /*
+        * Nobody can add the work_struct to this cwq,
+        *      if (caller is __create_workqueue)
+        *              nobody should see this wq
+        *      else // caller is CPU_UP_PREPARE
+        *              cpu is not on cpu_online_map
+        * so we can abort safely.
+        */
+       if (IS_ERR(p))
+               return PTR_ERR(p);
+
+       cwq->thread = p;
+
+       return 0;
+}
+
+static void start_workqueue_thread(struct cpu_workqueue_struct *cwq, int cpu)
+{
+       struct task_struct *p = cwq->thread;
 
-       while (!list_empty(&list)) {
-               printk("Taking work for %s\n", wq->name);
-               work = list_entry(list.next,struct work_struct,entry);
-               list_del(&work->entry);
-               __queue_work(per_cpu_ptr(wq->cpu_wq, smp_processor_id()), work);
+       if (p != NULL) {
+               if (cpu >= 0)
+                       kthread_bind(p, cpu);
+               wake_up_process(p);
        }
-       spin_unlock_irq(&cwq->lock);
 }
 
-/* We're holding the cpucontrol mutex here */
-static int __devinit workqueue_cpu_callback(struct notifier_block *nfb,
-                                 unsigned long action,
-                                 void *hcpu)
+struct workqueue_struct *__create_workqueue(const char *name,
+                                           int singlethread, int freezeable)
 {
-       unsigned int hotcpu = (unsigned long)hcpu;
        struct workqueue_struct *wq;
+       struct cpu_workqueue_struct *cwq;
+       int err = 0, cpu;
 
-       switch (action) {
-       case CPU_UP_PREPARE:
-               mutex_lock(&workqueue_mutex);
-               /* Create a new workqueue thread for it. */
-               list_for_each_entry(wq, &workqueues, list) {
-                       if (!create_workqueue_thread(wq, hotcpu, 0)) {
-                               printk("workqueue for %i failed\n", hotcpu);
-                               return NOTIFY_BAD;
-                       }
-               }
-               break;
+       wq = kzalloc(sizeof(*wq), GFP_KERNEL);
+       if (!wq)
+               return NULL;
 
-       case CPU_ONLINE:
-               /* Kick off worker threads. */
-               list_for_each_entry(wq, &workqueues, list) {
-                       struct cpu_workqueue_struct *cwq;
+       wq->cpu_wq = alloc_percpu(struct cpu_workqueue_struct);
+       if (!wq->cpu_wq) {
+               kfree(wq);
+               return NULL;
+       }
 
-                       cwq = per_cpu_ptr(wq->cpu_wq, hotcpu);
-                       kthread_bind(cwq->thread, hotcpu);
-                       wake_up_process(cwq->thread);
-               }
-               mutex_unlock(&workqueue_mutex);
-               break;
+       wq->name = name;
+       wq->singlethread = singlethread;
+       wq->freezeable = freezeable;
+       INIT_LIST_HEAD(&wq->list);
 
-       case CPU_UP_CANCELED:
-               list_for_each_entry(wq, &workqueues, list) {
-                       if (!per_cpu_ptr(wq->cpu_wq, hotcpu)->thread)
+       if (singlethread) {
+               cwq = init_cpu_workqueue(wq, singlethread_cpu);
+               err = create_workqueue_thread(cwq, singlethread_cpu);
+               start_workqueue_thread(cwq, -1);
+       } else {
+               mutex_lock(&workqueue_mutex);
+               list_add(&wq->list, &workqueues);
+
+               for_each_possible_cpu(cpu) {
+                       cwq = init_cpu_workqueue(wq, cpu);
+                       if (err || !cpu_online(cpu))
                                continue;
-                       /* Unbind so it can run. */
-                       kthread_bind(per_cpu_ptr(wq->cpu_wq, hotcpu)->thread,
-                                    any_online_cpu(cpu_online_map));
-                       cleanup_workqueue_thread(wq, hotcpu);
+                       err = create_workqueue_thread(cwq, cpu);
+                       start_workqueue_thread(cwq, cpu);
                }
                mutex_unlock(&workqueue_mutex);
-               break;
+       }
 
-       case CPU_DOWN_PREPARE:
+       if (err) {
+               destroy_workqueue(wq);
+               wq = NULL;
+       }
+       return wq;
+}
+EXPORT_SYMBOL_GPL(__create_workqueue);
+
+static void cleanup_workqueue_thread(struct cpu_workqueue_struct *cwq, int cpu)
+{
+       /*
+        * Our caller is either destroy_workqueue() or CPU_DEAD,
+        * workqueue_mutex protects cwq->thread
+        */
+       if (cwq->thread == NULL)
+               return;
+
+       /*
+        * If the caller is CPU_DEAD the single flush_cpu_workqueue()
+        * is not enough, a concurrent flush_workqueue() can insert a
+        * barrier after us.
+        * When ->worklist becomes empty it is safe to exit because no
+        * more work_structs can be queued on this cwq: flush_workqueue
+        * checks list_empty(), and a "normal" queue_work() can't use
+        * a dead CPU.
+        */
+       while (flush_cpu_workqueue(cwq))
+               ;
+
+       kthread_stop(cwq->thread);
+       cwq->thread = NULL;
+}
+
+/**
+ * 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)
+{
+       const cpumask_t *cpu_map = wq_cpu_map(wq);
+       struct cpu_workqueue_struct *cwq;
+       int cpu;
+
+       mutex_lock(&workqueue_mutex);
+       list_del(&wq->list);
+       mutex_unlock(&workqueue_mutex);
+
+       for_each_cpu_mask(cpu, *cpu_map) {
+               cwq = per_cpu_ptr(wq->cpu_wq, cpu);
+               cleanup_workqueue_thread(cwq, cpu);
+       }
+
+       free_percpu(wq->cpu_wq);
+       kfree(wq);
+}
+EXPORT_SYMBOL_GPL(destroy_workqueue);
+
+static int __devinit workqueue_cpu_callback(struct notifier_block *nfb,
+                                               unsigned long action,
+                                               void *hcpu)
+{
+       unsigned int cpu = (unsigned long)hcpu;
+       struct cpu_workqueue_struct *cwq;
+       struct workqueue_struct *wq;
+
+       action &= ~CPU_TASKS_FROZEN;
+
+       switch (action) {
+       case CPU_LOCK_ACQUIRE:
                mutex_lock(&workqueue_mutex);
-               break;
+               return NOTIFY_OK;
 
-       case CPU_DOWN_FAILED:
+       case CPU_LOCK_RELEASE:
                mutex_unlock(&workqueue_mutex);
-               break;
+               return NOTIFY_OK;
 
-       case CPU_DEAD:
-               list_for_each_entry(wq, &workqueues, list)
-                       cleanup_workqueue_thread(wq, hotcpu);
-               list_for_each_entry(wq, &workqueues, list)
-                       take_over_work(wq, hotcpu);
-               mutex_unlock(&workqueue_mutex);
-               break;
+       case CPU_UP_PREPARE:
+               cpu_set(cpu, cpu_populated_map);
+       }
+
+       list_for_each_entry(wq, &workqueues, list) {
+               cwq = per_cpu_ptr(wq->cpu_wq, cpu);
+
+               switch (action) {
+               case CPU_UP_PREPARE:
+                       if (!create_workqueue_thread(cwq, cpu))
+                               break;
+                       printk(KERN_ERR "workqueue for %i failed\n", cpu);
+                       return NOTIFY_BAD;
+
+               case CPU_ONLINE:
+                       start_workqueue_thread(cwq, cpu);
+                       break;
+
+               case CPU_UP_CANCELED:
+                       start_workqueue_thread(cwq, -1);
+               case CPU_DEAD:
+                       cleanup_workqueue_thread(cwq, cpu);
+                       break;
+               }
        }
 
        return NOTIFY_OK;
 }
 
-void init_workqueues(void)
+void __init init_workqueues(void)
 {
+       cpu_populated_map = cpu_online_map;
        singlethread_cpu = first_cpu(cpu_possible_map);
+       cpu_singlethread_map = cpumask_of_cpu(singlethread_cpu);
        hotcpu_notifier(workqueue_cpu_callback, 0);
        keventd_wq = create_workqueue("events");
        BUG_ON(!keventd_wq);
 }
-