/*
       *	linux/kernel/softirq.c
       *
       *	Copyright (C) 1992 Linus Torvalds
       *
       * Fixed a disable_bh()/enable_bh() race (was causing a console lockup)
       * due bh_mask_count not atomic handling. Copyright (C) 1998  Andrea Arcangeli
       *
       * Rewritten. Old one was good in 2.2, but in 2.3 it was immoral. --ANK (990903)
       */
      
      #include <linux/config.h>
      #include <linux/mm.h>
      #include <linux/kernel_stat.h>
      #include <linux/interrupt.h>
      #include <linux/smp_lock.h>
      #include <linux/init.h>
      #include <linux/tqueue.h>
      
      /*
         - No shared variables, all the data are CPU local.
         - If a softirq needs serialization, let it serialize itself
           by its own spinlocks.
         - Even if softirq is serialized, only local cpu is marked for
           execution. Hence, we get something sort of weak cpu binding.
           Though it is still not clear, will it result in better locality
           or will not.
         - These softirqs are not masked by global cli() and start_bh_atomic()
           (by clear reasons). Hence, old parts of code still using global locks
           MUST NOT use softirqs, but insert interfacing routines acquiring
           global locks. F.e. look at BHs implementation.
      
         Examples:
         - NET RX softirq. It is multithreaded and does not require
           any global serialization.
         - NET TX softirq. It kicks software netdevice queues, hence
           it is logically serialized per device, but this serialization
           is invisible to common code.
         - Tasklets: serialized wrt itself.
         - Bottom halves: globally serialized, grr...
       */
      
      /* No separate irq_stat for s390, it is part of PSA */
      #if !defined(CONFIG_ARCH_S390)
      irq_cpustat_t irq_stat[NR_CPUS];
      #endif	/* CONFIG_ARCH_S390 */
      
      static struct softirq_action softirq_vec[32] __cacheline_aligned;
      
  50  asmlinkage void do_softirq()
      {
      	int cpu = smp_processor_id();
      	__u32 active, mask;
      
  55  	if (in_interrupt())
  56  		return;
      
  58  	local_bh_disable();
      
      	local_irq_disable();
      	mask = softirq_mask(cpu);
      	active = softirq_active(cpu) & mask;
      
  64  	if (active) {
      		struct softirq_action *h;
      
      restart:
      		/* Reset active bitmask before enabling irqs */
      		softirq_active(cpu) &= ~active;
      
      		local_irq_enable();
      
      		h = softirq_vec;
      		mask &= ~active;
      
  76  		do {
  77  			if (active & 1)
      				h->action(h);
      			h++;
      			active >>= 1;
  81  		} while (active);
      
      		local_irq_disable();
      
      		active = softirq_active(cpu);
  86  		if ((active &= mask) != 0)
  87  			goto retry;
      	}
      
  90  	local_bh_enable();
      
      	/* Leave with locally disabled hard irqs. It is critical to close
      	 * window for infinite recursion, while we help local bh count,
      	 * it protected us. Now we are defenceless.
      	 */
  96  	return;
      
      retry:
  99  	goto restart;
      }
      
      
      static spinlock_t softirq_mask_lock = SPIN_LOCK_UNLOCKED;
      
 105  void open_softirq(int nr, void (*action)(struct softirq_action*), void *data)
      {
      	unsigned long flags;
      	int i;
      
 110  	spin_lock_irqsave(&softirq_mask_lock, flags);
      	softirq_vec[nr].data = data;
      	softirq_vec[nr].action = action;
      
 114  	for (i=0; i<NR_CPUS; i++)
      		softirq_mask(i) |= (1<<nr);
 116  	spin_unlock_irqrestore(&softirq_mask_lock, flags);
      }
      
      
      /* Tasklets */
      
      struct tasklet_head tasklet_vec[NR_CPUS] __cacheline_aligned;
      
 124  static void tasklet_action(struct softirq_action *a)
      {
      	int cpu = smp_processor_id();
      	struct tasklet_struct *list;
      
      	local_irq_disable();
      	list = tasklet_vec[cpu].list;
      	tasklet_vec[cpu].list = NULL;
      	local_irq_enable();
      
 134  	while (list != NULL) {
      		struct tasklet_struct *t = list;
      
      		list = list->next;
      
 139  		if (tasklet_trylock(t)) {
 140  			if (atomic_read(&t->count) == 0) {
      				clear_bit(TASKLET_STATE_SCHED, &t->state);
      
      				t->func(t->data);
      				/*
      				 * talklet_trylock() uses test_and_set_bit that imply
      				 * an mb when it returns zero, thus we need the explicit
      				 * mb only here: while closing the critical section.
      				 */
      #ifdef CONFIG_SMP
      				smp_mb__before_clear_bit();
      #endif
 152  				tasklet_unlock(t);
 153  				continue;
      			}
 155  			tasklet_unlock(t);
      		}
      		local_irq_disable();
      		t->next = tasklet_vec[cpu].list;
      		tasklet_vec[cpu].list = t;
      		__cpu_raise_softirq(cpu, TASKLET_SOFTIRQ);
      		local_irq_enable();
      	}
      }
      
      
      
      struct tasklet_head tasklet_hi_vec[NR_CPUS] __cacheline_aligned;
      
 169  static void tasklet_hi_action(struct softirq_action *a)
      {
      	int cpu = smp_processor_id();
      	struct tasklet_struct *list;
      
      	local_irq_disable();
      	list = tasklet_hi_vec[cpu].list;
      	tasklet_hi_vec[cpu].list = NULL;
      	local_irq_enable();
      
 179  	while (list != NULL) {
      		struct tasklet_struct *t = list;
      
      		list = list->next;
      
 184  		if (tasklet_trylock(t)) {
 185  			if (atomic_read(&t->count) == 0) {
      				clear_bit(TASKLET_STATE_SCHED, &t->state);
      
      				t->func(t->data);
 189  				tasklet_unlock(t);
 190  				continue;
      			}
 192  			tasklet_unlock(t);
      		}
      		local_irq_disable();
      		t->next = tasklet_hi_vec[cpu].list;
      		tasklet_hi_vec[cpu].list = t;
      		__cpu_raise_softirq(cpu, HI_SOFTIRQ);
      		local_irq_enable();
      	}
      }
      
      
 203  void tasklet_init(struct tasklet_struct *t,
      		  void (*func)(unsigned long), unsigned long data)
      {
      	t->func = func;
      	t->data = data;
      	t->state = 0;
      	atomic_set(&t->count, 0);
      }
      
 212  void tasklet_kill(struct tasklet_struct *t)
      {
 214  	if (in_interrupt())
      		printk("Attempt to kill tasklet from interrupt\n");
      
 217  	while (test_and_set_bit(TASKLET_STATE_SCHED, &t->state)) {
      		current->state = TASK_RUNNING;
 219  		do {
      			current->policy |= SCHED_YIELD;
      			schedule();
 222  		} while (test_bit(TASKLET_STATE_SCHED, &t->state));
      	}
 224  	tasklet_unlock_wait(t);
      	clear_bit(TASKLET_STATE_SCHED, &t->state);
      }
      
      
      
      /* Old style BHs */
      
      static void (*bh_base[32])(void);
      struct tasklet_struct bh_task_vec[32];
      
      /* BHs are serialized by spinlock global_bh_lock.
      
         It is still possible to make synchronize_bh() as
         spin_unlock_wait(&global_bh_lock). This operation is not used
         by kernel now, so that this lock is not made private only
         due to wait_on_irq().
      
         It can be removed only after auditing all the BHs.
       */
      spinlock_t global_bh_lock = SPIN_LOCK_UNLOCKED;
      
 246  static void bh_action(unsigned long nr)
      {
      	int cpu = smp_processor_id();
      
 250  	if (!spin_trylock(&global_bh_lock))
 251  		goto resched;
      
 253  	if (!hardirq_trylock(cpu))
 254  		goto resched_unlock;
      
 256  	if (bh_base[nr])
      		bh_base[nr]();
      
 259  	hardirq_endlock(cpu);
 260  	spin_unlock(&global_bh_lock);
 261  	return;
      
      resched_unlock:
 264  	spin_unlock(&global_bh_lock);
      resched:
      	mark_bh(nr);
      }
      
 269  void init_bh(int nr, void (*routine)(void))
      {
      	bh_base[nr] = routine;
      	mb();
      }
      
 275  void remove_bh(int nr)
      {
      	tasklet_kill(bh_task_vec+nr);
      	bh_base[nr] = NULL;
      }
      
 281  void __init softirq_init()
      {
      	int i;
      
 285  	for (i=0; i<32; i++)
      		tasklet_init(bh_task_vec+i, bh_action, i);
      
      	open_softirq(TASKLET_SOFTIRQ, tasklet_action, NULL);
      	open_softirq(HI_SOFTIRQ, tasklet_hi_action, NULL);
      }
      
 292  void __run_task_queue(task_queue *list)
      {
      	struct list_head head, *next;
      	unsigned long flags;
      
 297  	spin_lock_irqsave(&tqueue_lock, flags);
      	list_add(&head, list);
      	list_del_init(list);
 300  	spin_unlock_irqrestore(&tqueue_lock, flags);
      
      	next = head.next;
 303  	while (next != &head) {
      		void (*f) (void *);
      		struct tq_struct *p;
      		void *data;
      
      		p = list_entry(next, struct tq_struct, list);
      		next = next->next;
      		f = p->routine;
      		data = p->data;
      		wmb();
      		p->sync = 0;
 314  		if (f)
      			f(data);
      	}
      }