[linux-2.6-block.git] / kernel / softirq.c

/*
 *	linux/kernel/softirq.c
 *
 *	Copyright (C) 1992 Linus Torvalds
 *
 *	Distribute under GPLv2.
 *
 *	Rewritten. Old one was good in 2.2, but in 2.3 it was immoral. --ANK (990903)
 */

#include <linux/module.h>
#include <linux/kernel_stat.h>
#include <linux/interrupt.h>
#include <linux/init.h>
#include <linux/mm.h>
#include <linux/notifier.h>
#include <linux/percpu.h>
#include <linux/cpu.h>
#include <linux/freezer.h>
#include <linux/kthread.h>
#include <linux/rcupdate.h>
#include <linux/smp.h>
#include <linux/tick.h>

#include <asm/irq.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.

   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.
 */

#ifndef __ARCH_IRQ_STAT
irq_cpustat_t irq_stat[NR_CPUS] ____cacheline_aligned;
EXPORT_SYMBOL(irq_stat);
#endif

static struct softirq_action softirq_vec[32] __cacheline_aligned_in_smp;

static DEFINE_PER_CPU(struct task_struct *, ksoftirqd);

/*
 * we cannot loop indefinitely here to avoid userspace starvation,
 * but we also don't want to introduce a worst case 1/HZ latency
 * to the pending events, so lets the scheduler to balance
 * the softirq load for us.
 */
static inline void wakeup_softirqd(void)
{
	/* Interrupts are disabled: no need to stop preemption */
	struct task_struct *tsk = __get_cpu_var(ksoftirqd);

	if (tsk && tsk->state != TASK_RUNNING)
		wake_up_process(tsk);
}

/*
 * This one is for softirq.c-internal use,
 * where hardirqs are disabled legitimately:
 */
#ifdef CONFIG_TRACE_IRQFLAGS
static void __local_bh_disable(unsigned long ip)
{
	unsigned long flags;

	WARN_ON_ONCE(in_irq());

	raw_local_irq_save(flags);
	add_preempt_count(SOFTIRQ_OFFSET);
	/*
	 * Were softirqs turned off above:
	 */
	if (softirq_count() == SOFTIRQ_OFFSET)
		trace_softirqs_off(ip);
	raw_local_irq_restore(flags);
}
#else /* !CONFIG_TRACE_IRQFLAGS */
static inline void __local_bh_disable(unsigned long ip)
{
	add_preempt_count(SOFTIRQ_OFFSET);
	barrier();
}
#endif /* CONFIG_TRACE_IRQFLAGS */

void local_bh_disable(void)
{
	__local_bh_disable((unsigned long)__builtin_return_address(0));
}

EXPORT_SYMBOL(local_bh_disable);

void __local_bh_enable(void)
{
	WARN_ON_ONCE(in_irq());

	/*
	 * softirqs should never be enabled by __local_bh_enable(),
	 * it always nests inside local_bh_enable() sections:
	 */
	WARN_ON_ONCE(softirq_count() == SOFTIRQ_OFFSET);

	sub_preempt_count(SOFTIRQ_OFFSET);
}
EXPORT_SYMBOL_GPL(__local_bh_enable);

/*
 * Special-case - softirqs can safely be enabled in
 * cond_resched_softirq(), or by __do_softirq(),
 * without processing still-pending softirqs:
 */
void _local_bh_enable(void)
{
	WARN_ON_ONCE(in_irq());
	WARN_ON_ONCE(!irqs_disabled());

	if (softirq_count() == SOFTIRQ_OFFSET)
		trace_softirqs_on((unsigned long)__builtin_return_address(0));
	sub_preempt_count(SOFTIRQ_OFFSET);
}

EXPORT_SYMBOL(_local_bh_enable);

void local_bh_enable(void)
{
#ifdef CONFIG_TRACE_IRQFLAGS
	unsigned long flags;

	WARN_ON_ONCE(in_irq());
#endif
	WARN_ON_ONCE(irqs_disabled());

#ifdef CONFIG_TRACE_IRQFLAGS
	local_irq_save(flags);
#endif
	/*
	 * Are softirqs going to be turned on now:
	 */
	if (softirq_count() == SOFTIRQ_OFFSET)
		trace_softirqs_on((unsigned long)__builtin_return_address(0));
	/*
	 * Keep preemption disabled until we are done with
	 * softirq processing:
 	 */
 	sub_preempt_count(SOFTIRQ_OFFSET - 1);

	if (unlikely(!in_interrupt() && local_softirq_pending()))
		do_softirq();

	dec_preempt_count();
#ifdef CONFIG_TRACE_IRQFLAGS
	local_irq_restore(flags);
#endif
	preempt_check_resched();
}
EXPORT_SYMBOL(local_bh_enable);

void local_bh_enable_ip(unsigned long ip)
{
#ifdef CONFIG_TRACE_IRQFLAGS
	unsigned long flags;

	WARN_ON_ONCE(in_irq());

	local_irq_save(flags);
#endif
	/*
	 * Are softirqs going to be turned on now:
	 */
	if (softirq_count() == SOFTIRQ_OFFSET)
		trace_softirqs_on(ip);
	/*
	 * Keep preemption disabled until we are done with
	 * softirq processing:
 	 */
 	sub_preempt_count(SOFTIRQ_OFFSET - 1);

	if (unlikely(!in_interrupt() && local_softirq_pending()))
		do_softirq();

	dec_preempt_count();
#ifdef CONFIG_TRACE_IRQFLAGS
	local_irq_restore(flags);
#endif
	preempt_check_resched();
}
EXPORT_SYMBOL(local_bh_enable_ip);

/*
 * We restart softirq processing MAX_SOFTIRQ_RESTART times,
 * and we fall back to softirqd after that.
 *
 * This number has been established via experimentation.
 * The two things to balance is latency against fairness -
 * we want to handle softirqs as soon as possible, but they
 * should not be able to lock up the box.
 */
#define MAX_SOFTIRQ_RESTART 10

asmlinkage void __do_softirq(void)
{
	struct softirq_action *h;
	__u32 pending;
	int max_restart = MAX_SOFTIRQ_RESTART;
	int cpu;

	pending = local_softirq_pending();
	account_system_vtime(current);

	__local_bh_disable((unsigned long)__builtin_return_address(0));
	trace_softirq_enter();

	cpu = smp_processor_id();
restart:
	/* Reset the pending bitmask before enabling irqs */
	set_softirq_pending(0);

	local_irq_enable();

	h = softirq_vec;

	do {
		if (pending & 1) {
			h->action(h);
			rcu_bh_qsctr_inc(cpu);
		}
		h++;
		pending >>= 1;
	} while (pending);

	local_irq_disable();

	pending = local_softirq_pending();
	if (pending && --max_restart)
		goto restart;

	if (pending)
		wakeup_softirqd();

	trace_softirq_exit();

	account_system_vtime(current);
	_local_bh_enable();
}

#ifndef __ARCH_HAS_DO_SOFTIRQ

asmlinkage void do_softirq(void)
{
	__u32 pending;
	unsigned long flags;

	if (in_interrupt())
		return;

	local_irq_save(flags);

	pending = local_softirq_pending();

	if (pending)
		__do_softirq();

	local_irq_restore(flags);
}

#endif

/*
 * Enter an interrupt context.
 */
void irq_enter(void)
{
	__irq_enter();
#ifdef CONFIG_NO_HZ
	if (idle_cpu(smp_processor_id()))
		tick_nohz_update_jiffies();
#endif
}

#ifdef __ARCH_IRQ_EXIT_IRQS_DISABLED
# define invoke_softirq()	__do_softirq()
#else
# define invoke_softirq()	do_softirq()
#endif

/*
 * Exit an interrupt context. Process softirqs if needed and possible:
 */
void irq_exit(void)
{
	account_system_vtime(current);
	trace_hardirq_exit();
	sub_preempt_count(IRQ_EXIT_OFFSET);
	if (!in_interrupt() && local_softirq_pending())
		invoke_softirq();

#ifdef CONFIG_NO_HZ
	/* Make sure that timer wheel updates are propagated */
	if (!in_interrupt() && idle_cpu(smp_processor_id()) && !need_resched())
		tick_nohz_stop_sched_tick();
#endif
	preempt_enable_no_resched();
}

/*
 * This function must run with irqs disabled!
 */
inline fastcall void raise_softirq_irqoff(unsigned int nr)
{
	__raise_softirq_irqoff(nr);

	/*
	 * If we're in an interrupt or softirq, we're done
	 * (this also catches softirq-disabled code). We will
	 * actually run the softirq once we return from
	 * the irq or softirq.
	 *
	 * Otherwise we wake up ksoftirqd to make sure we
	 * schedule the softirq soon.
	 */
	if (!in_interrupt())
		wakeup_softirqd();
}

void fastcall raise_softirq(unsigned int nr)
{
	unsigned long flags;

	local_irq_save(flags);
	raise_softirq_irqoff(nr);
	local_irq_restore(flags);
}

void open_softirq(int nr, void (*action)(struct softirq_action*), void *data)
{
	softirq_vec[nr].data = data;
	softirq_vec[nr].action = action;
}

/* Tasklets */
struct tasklet_head
{
	struct tasklet_struct *list;
};

/* Some compilers disobey section attribute on statics when not
   initialized -- RR */
static DEFINE_PER_CPU(struct tasklet_head, tasklet_vec) = { NULL };
static DEFINE_PER_CPU(struct tasklet_head, tasklet_hi_vec) = { NULL };

void fastcall __tasklet_schedule(struct tasklet_struct *t)
{
	unsigned long flags;

	local_irq_save(flags);
	t->next = __get_cpu_var(tasklet_vec).list;
	__get_cpu_var(tasklet_vec).list = t;
	raise_softirq_irqoff(TASKLET_SOFTIRQ);
	local_irq_restore(flags);
}

EXPORT_SYMBOL(__tasklet_schedule);

void fastcall __tasklet_hi_schedule(struct tasklet_struct *t)
{
	unsigned long flags;

	local_irq_save(flags);
	t->next = __get_cpu_var(tasklet_hi_vec).list;
	__get_cpu_var(tasklet_hi_vec).list = t;
	raise_softirq_irqoff(HI_SOFTIRQ);
	local_irq_restore(flags);
}

EXPORT_SYMBOL(__tasklet_hi_schedule);

static void tasklet_action(struct softirq_action *a)
{
	struct tasklet_struct *list;

	local_irq_disable();
	list = __get_cpu_var(tasklet_vec).list;
	__get_cpu_var(tasklet_vec).list = NULL;
	local_irq_enable();

	while (list) {
		struct tasklet_struct *t = list;

		list = list->next;

		if (tasklet_trylock(t)) {
			if (!atomic_read(&t->count)) {
				if (!test_and_clear_bit(TASKLET_STATE_SCHED, &t->state))
					BUG();
				t->func(t->data);
				tasklet_unlock(t);
				continue;
			}
			tasklet_unlock(t);
		}

		local_irq_disable();
		t->next = __get_cpu_var(tasklet_vec).list;
		__get_cpu_var(tasklet_vec).list = t;
		__raise_softirq_irqoff(TASKLET_SOFTIRQ);
		local_irq_enable();
	}
}

static void tasklet_hi_action(struct softirq_action *a)
{
	struct tasklet_struct *list;

	local_irq_disable();
	list = __get_cpu_var(tasklet_hi_vec).list;
	__get_cpu_var(tasklet_hi_vec).list = NULL;
	local_irq_enable();

	while (list) {
		struct tasklet_struct *t = list;

		list = list->next;

		if (tasklet_trylock(t)) {
			if (!atomic_read(&t->count)) {
				if (!test_and_clear_bit(TASKLET_STATE_SCHED, &t->state))
					BUG();
				t->func(t->data);
				tasklet_unlock(t);
				continue;
			}
			tasklet_unlock(t);
		}

		local_irq_disable();
		t->next = __get_cpu_var(tasklet_hi_vec).list;
		__get_cpu_var(tasklet_hi_vec).list = t;
		__raise_softirq_irqoff(HI_SOFTIRQ);
		local_irq_enable();
	}
}


void tasklet_init(struct tasklet_struct *t,
		  void (*func)(unsigned long), unsigned long data)
{
	t->next = NULL;
	t->state = 0;
	atomic_set(&t->count, 0);
	t->func = func;
	t->data = data;
}

EXPORT_SYMBOL(tasklet_init);

void tasklet_kill(struct tasklet_struct *t)
{
	if (in_interrupt())
		printk("Attempt to kill tasklet from interrupt\n");

	while (test_and_set_bit(TASKLET_STATE_SCHED, &t->state)) {
		do
			yield();
		while (test_bit(TASKLET_STATE_SCHED, &t->state));
	}
	tasklet_unlock_wait(t);
	clear_bit(TASKLET_STATE_SCHED, &t->state);
}

EXPORT_SYMBOL(tasklet_kill);

void __init softirq_init(void)
{
	open_softirq(TASKLET_SOFTIRQ, tasklet_action, NULL);
	open_softirq(HI_SOFTIRQ, tasklet_hi_action, NULL);
}

static int ksoftirqd(void * __bind_cpu)
{
	set_current_state(TASK_INTERRUPTIBLE);

	while (!kthread_should_stop()) {
		preempt_disable();
		if (!local_softirq_pending()) {
			preempt_enable_no_resched();
			schedule();
			preempt_disable();
		}

		__set_current_state(TASK_RUNNING);

		while (local_softirq_pending()) {
			/* Preempt disable stops cpu going offline.
			   If already offline, we'll be on wrong CPU:
			   don't process */
			if (cpu_is_offline((long)__bind_cpu))
				goto wait_to_die;
			do_softirq();
			preempt_enable_no_resched();
			cond_resched();
			preempt_disable();
		}
		preempt_enable();
		set_current_state(TASK_INTERRUPTIBLE);
	}
	__set_current_state(TASK_RUNNING);
	return 0;

wait_to_die:
	preempt_enable();
	/* Wait for kthread_stop */
	set_current_state(TASK_INTERRUPTIBLE);
	while (!kthread_should_stop()) {
		schedule();
		set_current_state(TASK_INTERRUPTIBLE);
	}
	__set_current_state(TASK_RUNNING);
	return 0;
}

#ifdef CONFIG_HOTPLUG_CPU
/*
 * tasklet_kill_immediate is called to remove a tasklet which can already be
 * scheduled for execution on @cpu.
 *
 * Unlike tasklet_kill, this function removes the tasklet
 * _immediately_, even if the tasklet is in TASKLET_STATE_SCHED state.
 *
 * When this function is called, @cpu must be in the CPU_DEAD state.
 */
void tasklet_kill_immediate(struct tasklet_struct *t, unsigned int cpu)
{
	struct tasklet_struct **i;

	BUG_ON(cpu_online(cpu));
	BUG_ON(test_bit(TASKLET_STATE_RUN, &t->state));

	if (!test_bit(TASKLET_STATE_SCHED, &t->state))
		return;

	/* CPU is dead, so no lock needed. */
	for (i = &per_cpu(tasklet_vec, cpu).list; *i; i = &(*i)->next) {
		if (*i == t) {
			*i = t->next;
			return;
		}
	}
	BUG();
}

static void takeover_tasklets(unsigned int cpu)
{
	struct tasklet_struct **i;

	/* CPU is dead, so no lock needed. */
	local_irq_disable();

	/* Find end, append list for that CPU. */
	for (i = &__get_cpu_var(tasklet_vec).list; *i; i = &(*i)->next);
	*i = per_cpu(tasklet_vec, cpu).list;
	per_cpu(tasklet_vec, cpu).list = NULL;
	raise_softirq_irqoff(TASKLET_SOFTIRQ);

	for (i = &__get_cpu_var(tasklet_hi_vec).list; *i; i = &(*i)->next);
	*i = per_cpu(tasklet_hi_vec, cpu).list;
	per_cpu(tasklet_hi_vec, cpu).list = NULL;
	raise_softirq_irqoff(HI_SOFTIRQ);

	local_irq_enable();
}
#endif /* CONFIG_HOTPLUG_CPU */

static int __cpuinit cpu_callback(struct notifier_block *nfb,
				  unsigned long action,
				  void *hcpu)
{
	int hotcpu = (unsigned long)hcpu;
	struct task_struct *p;

	switch (action) {
	case CPU_UP_PREPARE:
	case CPU_UP_PREPARE_FROZEN:
		p = kthread_create(ksoftirqd, hcpu, "ksoftirqd/%d", hotcpu);
		if (IS_ERR(p)) {
			printk("ksoftirqd for %i failed\n", hotcpu);
			return NOTIFY_BAD;
		}
		kthread_bind(p, hotcpu);
  		per_cpu(ksoftirqd, hotcpu) = p;
 		break;
	case CPU_ONLINE:
	case CPU_ONLINE_FROZEN:
		wake_up_process(per_cpu(ksoftirqd, hotcpu));
		break;
#ifdef CONFIG_HOTPLUG_CPU
	case CPU_UP_CANCELED:
	case CPU_UP_CANCELED_FROZEN:
		if (!per_cpu(ksoftirqd, hotcpu))
			break;
		/* Unbind so it can run.  Fall thru. */
		kthread_bind(per_cpu(ksoftirqd, hotcpu),
			     any_online_cpu(cpu_online_map));
	case CPU_DEAD:
	case CPU_DEAD_FROZEN: {
		struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 };

		p = per_cpu(ksoftirqd, hotcpu);
		per_cpu(ksoftirqd, hotcpu) = NULL;
		sched_setscheduler(p, SCHED_FIFO, &param);
		kthread_stop(p);
		takeover_tasklets(hotcpu);
		break;
	}
#endif /* CONFIG_HOTPLUG_CPU */
 	}
	return NOTIFY_OK;
}

static struct notifier_block __cpuinitdata cpu_nfb = {
	.notifier_call = cpu_callback
};

__init int spawn_ksoftirqd(void)
{
	void *cpu = (void *)(long)smp_processor_id();
	int err = cpu_callback(&cpu_nfb, CPU_UP_PREPARE, cpu);

	BUG_ON(err == NOTIFY_BAD);
	cpu_callback(&cpu_nfb, CPU_ONLINE, cpu);
	register_cpu_notifier(&cpu_nfb);
	return 0;
}

#ifdef CONFIG_SMP
/*
 * Call a function on all processors
 */
int on_each_cpu(void (*func) (void *info), void *info, int retry, int wait)
{
	int ret = 0;

	preempt_disable();
	ret = smp_call_function(func, info, retry, wait);
	local_irq_disable();
	func(info);
	local_irq_enable();
	preempt_enable();
	return ret;
}
EXPORT_SYMBOL(on_each_cpu);
#endif
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/kernel/softirq.c
	3	*
	4	* Copyright (C) 1992 Linus Torvalds
	5	*
b10db7f0 PM	6	* Distribute under GPLv2.
	7	*
	8	* Rewritten. Old one was good in 2.2, but in 2.3 it was immoral. --ANK (990903)
1da177e4 LT	9	*/
	10
	11	#include <linux/module.h>
	12	#include <linux/kernel_stat.h>
	13	#include <linux/interrupt.h>
	14	#include <linux/init.h>
	15	#include <linux/mm.h>
	16	#include <linux/notifier.h>
	17	#include <linux/percpu.h>
	18	#include <linux/cpu.h>
83144186	19	#include <linux/freezer.h>
1da177e4 LT	20	#include <linux/kthread.h>
1da177e4 LT	21	#include <linux/rcupdate.h>
78eef01b	22	#include <linux/smp.h>
79bf2bb3	23	#include <linux/tick.h>
1da177e4 LT	24
	25	#include <asm/irq.h>
	26	/*
	27	- No shared variables, all the data are CPU local.
	28	- If a softirq needs serialization, let it serialize itself
	29	by its own spinlocks.
	30	- Even if softirq is serialized, only local cpu is marked for
	31	execution. Hence, we get something sort of weak cpu binding.
	32	Though it is still not clear, will it result in better locality
	33	or will not.
	34
	35	Examples:
	36	- NET RX softirq. It is multithreaded and does not require
	37	any global serialization.
	38	- NET TX softirq. It kicks software netdevice queues, hence
	39	it is logically serialized per device, but this serialization
	40	is invisible to common code.
	41	- Tasklets: serialized wrt itself.
	42	*/
	43
	44	#ifndef __ARCH_IRQ_STAT
	45	irq_cpustat_t irq_stat[NR_CPUS] ____cacheline_aligned;
	46	EXPORT_SYMBOL(irq_stat);
	47	#endif
	48
	49	static struct softirq_action softirq_vec[32] __cacheline_aligned_in_smp;
	50
	51	static DEFINE_PER_CPU(struct task_struct *, ksoftirqd);
	52
	53	/*
	54	* we cannot loop indefinitely here to avoid userspace starvation,
	55	* but we also don't want to introduce a worst case 1/HZ latency
	56	* to the pending events, so lets the scheduler to balance
	57	* the softirq load for us.
	58	*/
	59	static inline void wakeup_softirqd(void)
	60	{
	61	/* Interrupts are disabled: no need to stop preemption */
	62	struct task_struct *tsk = __get_cpu_var(ksoftirqd);
	63
	64	if (tsk && tsk->state != TASK_RUNNING)
	65	wake_up_process(tsk);
	66	}
	67
de30a2b3 IM	68	/*
	69	* This one is for softirq.c-internal use,
	70	* where hardirqs are disabled legitimately:
	71	*/
3c829c36	72	#ifdef CONFIG_TRACE_IRQFLAGS
de30a2b3 IM	73	static void __local_bh_disable(unsigned long ip)
	74	{
	75	unsigned long flags;
	76
	77	WARN_ON_ONCE(in_irq());
	78
	79	raw_local_irq_save(flags);
	80	add_preempt_count(SOFTIRQ_OFFSET);
	81	/*
	82	* Were softirqs turned off above:
	83	*/
	84	if (softirq_count() == SOFTIRQ_OFFSET)
	85	trace_softirqs_off(ip);
	86	raw_local_irq_restore(flags);
	87	}
3c829c36 TC	88	#else /* !CONFIG_TRACE_IRQFLAGS */
	89	static inline void __local_bh_disable(unsigned long ip)
	90	{
	91	add_preempt_count(SOFTIRQ_OFFSET);
	92	barrier();
	93	}
	94	#endif /* CONFIG_TRACE_IRQFLAGS */
de30a2b3 IM	95
	96	void local_bh_disable(void)
	97	{
	98	__local_bh_disable((unsigned long)__builtin_return_address(0));
	99	}
	100
	101	EXPORT_SYMBOL(local_bh_disable);
	102
	103	void __local_bh_enable(void)
	104	{
	105	WARN_ON_ONCE(in_irq());
	106
	107	/*
	108	* softirqs should never be enabled by __local_bh_enable(),
	109	* it always nests inside local_bh_enable() sections:
	110	*/
	111	WARN_ON_ONCE(softirq_count() == SOFTIRQ_OFFSET);
	112
	113	sub_preempt_count(SOFTIRQ_OFFSET);
	114	}
	115	EXPORT_SYMBOL_GPL(__local_bh_enable);
	116
	117	/*
	118	* Special-case - softirqs can safely be enabled in
	119	* cond_resched_softirq(), or by __do_softirq(),
	120	* without processing still-pending softirqs:
	121	*/
	122	void _local_bh_enable(void)
	123	{
	124	WARN_ON_ONCE(in_irq());
	125	WARN_ON_ONCE(!irqs_disabled());
	126
	127	if (softirq_count() == SOFTIRQ_OFFSET)
	128	trace_softirqs_on((unsigned long)__builtin_return_address(0));
	129	sub_preempt_count(SOFTIRQ_OFFSET);
	130	}
	131
	132	EXPORT_SYMBOL(_local_bh_enable);
	133
	134	void local_bh_enable(void)
	135	{
3c829c36	136	#ifdef CONFIG_TRACE_IRQFLAGS
de30a2b3 IM	137	unsigned long flags;
	138
	139	WARN_ON_ONCE(in_irq());
3c829c36	140	#endif
de30a2b3 IM	141	WARN_ON_ONCE(irqs_disabled());
de30a2b3 IM	142
3c829c36	143	#ifdef CONFIG_TRACE_IRQFLAGS
de30a2b3	144	local_irq_save(flags);
3c829c36	145	#endif
de30a2b3 IM	146	/*
	147	* Are softirqs going to be turned on now:
	148	*/
	149	if (softirq_count() == SOFTIRQ_OFFSET)
	150	trace_softirqs_on((unsigned long)__builtin_return_address(0));
	151	/*
	152	* Keep preemption disabled until we are done with
	153	* softirq processing:
	154	*/
	155	sub_preempt_count(SOFTIRQ_OFFSET - 1);
	156
	157	if (unlikely(!in_interrupt() && local_softirq_pending()))
	158	do_softirq();
	159
	160	dec_preempt_count();
3c829c36	161	#ifdef CONFIG_TRACE_IRQFLAGS
de30a2b3	162	local_irq_restore(flags);
3c829c36	163	#endif
de30a2b3 IM	164	preempt_check_resched();
	165	}
	166	EXPORT_SYMBOL(local_bh_enable);
	167
	168	void local_bh_enable_ip(unsigned long ip)
	169	{
3c829c36	170	#ifdef CONFIG_TRACE_IRQFLAGS
de30a2b3 IM	171	unsigned long flags;
	172
	173	WARN_ON_ONCE(in_irq());
	174
	175	local_irq_save(flags);
3c829c36	176	#endif
de30a2b3 IM	177	/*
	178	* Are softirqs going to be turned on now:
	179	*/
	180	if (softirq_count() == SOFTIRQ_OFFSET)
	181	trace_softirqs_on(ip);
	182	/*
	183	* Keep preemption disabled until we are done with
	184	* softirq processing:
	185	*/
	186	sub_preempt_count(SOFTIRQ_OFFSET - 1);
	187
	188	if (unlikely(!in_interrupt() && local_softirq_pending()))
	189	do_softirq();
	190
	191	dec_preempt_count();
3c829c36	192	#ifdef CONFIG_TRACE_IRQFLAGS
de30a2b3	193	local_irq_restore(flags);
3c829c36	194	#endif
de30a2b3 IM	195	preempt_check_resched();
	196	}
	197	EXPORT_SYMBOL(local_bh_enable_ip);
	198
1da177e4 LT	199	/*
	200	* We restart softirq processing MAX_SOFTIRQ_RESTART times,
	201	* and we fall back to softirqd after that.
	202	*
	203	* This number has been established via experimentation.
	204	* The two things to balance is latency against fairness -
	205	* we want to handle softirqs as soon as possible, but they
	206	* should not be able to lock up the box.
	207	*/
	208	#define MAX_SOFTIRQ_RESTART 10
	209
	210	asmlinkage void __do_softirq(void)
	211	{
	212	struct softirq_action *h;
	213	__u32 pending;
	214	int max_restart = MAX_SOFTIRQ_RESTART;
	215	int cpu;
	216
	217	pending = local_softirq_pending();
829035fd PM	218	account_system_vtime(current);
829035fd PM	219
de30a2b3 IM	220	__local_bh_disable((unsigned long)__builtin_return_address(0));
de30a2b3 IM	221	trace_softirq_enter();
1da177e4	222
1da177e4 LT	223	cpu = smp_processor_id();
	224	restart:
	225	/* Reset the pending bitmask before enabling irqs */
3f74478b	226	set_softirq_pending(0);
1da177e4	227
c70f5d66	228	local_irq_enable();
1da177e4 LT	229
	230	h = softirq_vec;
	231
	232	do {
	233	if (pending & 1) {
	234	h->action(h);
	235	rcu_bh_qsctr_inc(cpu);
	236	}
	237	h++;
	238	pending >>= 1;
	239	} while (pending);
	240
c70f5d66	241	local_irq_disable();
1da177e4 LT	242
	243	pending = local_softirq_pending();
	244	if (pending && --max_restart)
	245	goto restart;
	246
	247	if (pending)
	248	wakeup_softirqd();
	249
de30a2b3	250	trace_softirq_exit();
829035fd PM	251
829035fd PM	252	account_system_vtime(current);
de30a2b3	253	_local_bh_enable();
1da177e4 LT	254	}
	255
	256	#ifndef __ARCH_HAS_DO_SOFTIRQ
	257
	258	asmlinkage void do_softirq(void)
	259	{
	260	__u32 pending;
	261	unsigned long flags;
	262
	263	if (in_interrupt())
	264	return;
	265
	266	local_irq_save(flags);
	267
	268	pending = local_softirq_pending();
	269
	270	if (pending)
	271	__do_softirq();
	272
	273	local_irq_restore(flags);
	274	}
	275
1da177e4 LT	276	#endif
1da177e4 LT	277
dde4b2b5 IM	278	/*
	279	* Enter an interrupt context.
	280	*/
	281	void irq_enter(void)
	282	{
79bf2bb3 TG	283	__irq_enter();
	284	#ifdef CONFIG_NO_HZ
	285	if (idle_cpu(smp_processor_id()))
	286	tick_nohz_update_jiffies();
	287	#endif
dde4b2b5 IM	288	}
dde4b2b5 IM	289
1da177e4 LT	290	#ifdef __ARCH_IRQ_EXIT_IRQS_DISABLED
	291	# define invoke_softirq() __do_softirq()
	292	#else
	293	# define invoke_softirq() do_softirq()
	294	#endif
	295
	296	/*
	297	* Exit an interrupt context. Process softirqs if needed and possible:
	298	*/
	299	void irq_exit(void)
	300	{
	301	account_system_vtime(current);
de30a2b3	302	trace_hardirq_exit();
1da177e4 LT	303	sub_preempt_count(IRQ_EXIT_OFFSET);
	304	if (!in_interrupt() && local_softirq_pending())
	305	invoke_softirq();
79bf2bb3 TG	306
	307	#ifdef CONFIG_NO_HZ
	308	/* Make sure that timer wheel updates are propagated */
	309	if (!in_interrupt() && idle_cpu(smp_processor_id()) && !need_resched())
	310	tick_nohz_stop_sched_tick();
	311	#endif
1da177e4 LT	312	preempt_enable_no_resched();
	313	}
	314
	315	/*
	316	* This function must run with irqs disabled!
	317	*/
	318	inline fastcall void raise_softirq_irqoff(unsigned int nr)
	319	{
	320	__raise_softirq_irqoff(nr);
	321
	322	/*
	323	* If we're in an interrupt or softirq, we're done
	324	* (this also catches softirq-disabled code). We will
	325	* actually run the softirq once we return from
	326	* the irq or softirq.
	327	*
	328	* Otherwise we wake up ksoftirqd to make sure we
	329	* schedule the softirq soon.
	330	*/
	331	if (!in_interrupt())
	332	wakeup_softirqd();
	333	}
	334
1da177e4 LT	335	void fastcall raise_softirq(unsigned int nr)
	336	{
	337	unsigned long flags;
	338
	339	local_irq_save(flags);
	340	raise_softirq_irqoff(nr);
	341	local_irq_restore(flags);
	342	}
	343
	344	void open_softirq(int nr, void (action)(struct softirq_action), void *data)
	345	{
	346	softirq_vec[nr].data = data;
	347	softirq_vec[nr].action = action;
	348	}
	349
1da177e4 LT	350	/* Tasklets */
	351	struct tasklet_head
	352	{
	353	struct tasklet_struct *list;
	354	};
	355
	356	/* Some compilers disobey section attribute on statics when not
	357	initialized -- RR */
	358	static DEFINE_PER_CPU(struct tasklet_head, tasklet_vec) = { NULL };
	359	static DEFINE_PER_CPU(struct tasklet_head, tasklet_hi_vec) = { NULL };
	360
	361	void fastcall __tasklet_schedule(struct tasklet_struct *t)
	362	{
	363	unsigned long flags;
	364
	365	local_irq_save(flags);
	366	t->next = __get_cpu_var(tasklet_vec).list;
	367	__get_cpu_var(tasklet_vec).list = t;
	368	raise_softirq_irqoff(TASKLET_SOFTIRQ);
	369	local_irq_restore(flags);
	370	}
	371
	372	EXPORT_SYMBOL(__tasklet_schedule);
	373
	374	void fastcall __tasklet_hi_schedule(struct tasklet_struct *t)
	375	{
	376	unsigned long flags;
	377
	378	local_irq_save(flags);
	379	t->next = __get_cpu_var(tasklet_hi_vec).list;
	380	__get_cpu_var(tasklet_hi_vec).list = t;
	381	raise_softirq_irqoff(HI_SOFTIRQ);
	382	local_irq_restore(flags);
	383	}
	384
	385	EXPORT_SYMBOL(__tasklet_hi_schedule);
	386
	387	static void tasklet_action(struct softirq_action *a)
	388	{
	389	struct tasklet_struct *list;
	390
	391	local_irq_disable();
	392	list = __get_cpu_var(tasklet_vec).list;
	393	__get_cpu_var(tasklet_vec).list = NULL;
	394	local_irq_enable();
	395
	396	while (list) {
	397	struct tasklet_struct *t = list;
	398
	399	list = list->next;
	400
	401	if (tasklet_trylock(t)) {
	402	if (!atomic_read(&t->count)) {
	403	if (!test_and_clear_bit(TASKLET_STATE_SCHED, &t->state))
	404	BUG();
	405	t->func(t->data);
	406	tasklet_unlock(t);
	407	continue;
	408	}
	409	tasklet_unlock(t);
	410	}
	411
	412	local_irq_disable();
	413	t->next = __get_cpu_var(tasklet_vec).list;
414	__get_cpu_var(tasklet_vec).list = t;
415	__raise_softirq_irqoff(TASKLET_SOFTIRQ);
416	local_irq_enable();
417	}
418	}
419
420	static void tasklet_hi_action(struct softirq_action *a)
421	{
422	struct tasklet_struct *list;
423
424	local_irq_disable();
425	list = __get_cpu_var(tasklet_hi_vec).list;
426	__get_cpu_var(tasklet_hi_vec).list = NULL;
427	local_irq_enable();
428
429	while (list) {
430	struct tasklet_struct *t = list;
431
432	list = list->next;
433
434	if (tasklet_trylock(t)) {
435	if (!atomic_read(&t->count)) {
436	if (!test_and_clear_bit(TASKLET_STATE_SCHED, &t->state))
437	BUG();
438	t->func(t->data);
439	tasklet_unlock(t);
440	continue;
441	}
442	tasklet_unlock(t);
443	}
444
445	local_irq_disable();
446	t->next = __get_cpu_var(tasklet_hi_vec).list;
447	__get_cpu_var(tasklet_hi_vec).list = t;
448	__raise_softirq_irqoff(HI_SOFTIRQ);
449	local_irq_enable();
450	}
451	}
452
453
454	void tasklet_init(struct tasklet_struct *t,
455	void (*func)(unsigned long), unsigned long data)
456	{
457	t->next = NULL;
458	t->state = 0;
459	atomic_set(&t->count, 0);
460	t->func = func;
461	t->data = data;
462	}
463
464	EXPORT_SYMBOL(tasklet_init);
465
466	void tasklet_kill(struct tasklet_struct *t)
467	{
468	if (in_interrupt())
469	printk("Attempt to kill tasklet from interrupt\n");
470
471	while (test_and_set_bit(TASKLET_STATE_SCHED, &t->state)) {
472	do
473	yield();
474	while (test_bit(TASKLET_STATE_SCHED, &t->state));
475	}
476	tasklet_unlock_wait(t);
477	clear_bit(TASKLET_STATE_SCHED, &t->state);
478	}
479
480	EXPORT_SYMBOL(tasklet_kill);
481
482	void __init softirq_init(void)
483	{
484	open_softirq(TASKLET_SOFTIRQ, tasklet_action, NULL);
485	open_softirq(HI_SOFTIRQ, tasklet_hi_action, NULL);
486	}
487
488	static int ksoftirqd(void * __bind_cpu)
489	{
1da177e4 LT	490	set_current_state(TASK_INTERRUPTIBLE);
	491
	492	while (!kthread_should_stop()) {
	493	preempt_disable();
	494	if (!local_softirq_pending()) {
	495	preempt_enable_no_resched();
	496	schedule();
	497	preempt_disable();
	498	}
	499
	500	__set_current_state(TASK_RUNNING);
	501
	502	while (local_softirq_pending()) {
	503	/* Preempt disable stops cpu going offline.
	504	If already offline, we'll be on wrong CPU:
	505	don't process */
	506	if (cpu_is_offline((long)__bind_cpu))
	507	goto wait_to_die;
	508	do_softirq();
	509	preempt_enable_no_resched();
	510	cond_resched();
	511	preempt_disable();
	512	}
	513	preempt_enable();
	514	set_current_state(TASK_INTERRUPTIBLE);
	515	}
	516	__set_current_state(TASK_RUNNING);
	517	return 0;
	518
	519	wait_to_die:
	520	preempt_enable();
	521	/* Wait for kthread_stop */
	522	set_current_state(TASK_INTERRUPTIBLE);
	523	while (!kthread_should_stop()) {
	524	schedule();
	525	set_current_state(TASK_INTERRUPTIBLE);
	526	}
	527	__set_current_state(TASK_RUNNING);
	528	return 0;
	529	}
	530
	531	#ifdef CONFIG_HOTPLUG_CPU
	532	/*
	533	* tasklet_kill_immediate is called to remove a tasklet which can already be
	534	* scheduled for execution on @cpu.
	535	*
	536	* Unlike tasklet_kill, this function removes the tasklet
	537	* _immediately_, even if the tasklet is in TASKLET_STATE_SCHED state.
	538	*
	539	* When this function is called, @cpu must be in the CPU_DEAD state.
	540	*/
	541	void tasklet_kill_immediate(struct tasklet_struct *t, unsigned int cpu)
	542	{
	543	struct tasklet_struct **i;
	544
	545	BUG_ON(cpu_online(cpu));
	546	BUG_ON(test_bit(TASKLET_STATE_RUN, &t->state));
	547
	548	if (!test_bit(TASKLET_STATE_SCHED, &t->state))
	549	return;
	550
	551	/* CPU is dead, so no lock needed. */
	552	for (i = &per_cpu(tasklet_vec, cpu).list; i; i = &(i)->next) {
	553	if (*i == t) {
554	*i = t->next;
555	return;
556	}
557	}
558	BUG();
559	}
560
561	static void takeover_tasklets(unsigned int cpu)
562	{
563	struct tasklet_struct **i;
564
565	/* CPU is dead, so no lock needed. */
566	local_irq_disable();
567
568	/* Find end, append list for that CPU. */
569	for (i = &__get_cpu_var(tasklet_vec).list; i; i = &(i)->next);
570	*i = per_cpu(tasklet_vec, cpu).list;
571	per_cpu(tasklet_vec, cpu).list = NULL;
572	raise_softirq_irqoff(TASKLET_SOFTIRQ);
573
574	for (i = &__get_cpu_var(tasklet_hi_vec).list; i; i = &(i)->next);
575	*i = per_cpu(tasklet_hi_vec, cpu).list;
576	per_cpu(tasklet_hi_vec, cpu).list = NULL;
577	raise_softirq_irqoff(HI_SOFTIRQ);
578
579	local_irq_enable();
580	}
581	#endif /* CONFIG_HOTPLUG_CPU */
582
8c78f307	583	static int __cpuinit cpu_callback(struct notifier_block *nfb,
1da177e4 LT	584	unsigned long action,
	585	void *hcpu)
	586	{
	587	int hotcpu = (unsigned long)hcpu;
	588	struct task_struct *p;
	589
	590	switch (action) {
	591	case CPU_UP_PREPARE:
8bb78442	592	case CPU_UP_PREPARE_FROZEN:
1da177e4 LT	593	p = kthread_create(ksoftirqd, hcpu, "ksoftirqd/%d", hotcpu);
	594	if (IS_ERR(p)) {
	595	printk("ksoftirqd for %i failed\n", hotcpu);
	596	return NOTIFY_BAD;
	597	}
	598	kthread_bind(p, hotcpu);
	599	per_cpu(ksoftirqd, hotcpu) = p;
	600	break;
	601	case CPU_ONLINE:
8bb78442	602	case CPU_ONLINE_FROZEN:
1da177e4 LT	603	wake_up_process(per_cpu(ksoftirqd, hotcpu));
	604	break;
	605	#ifdef CONFIG_HOTPLUG_CPU
	606	case CPU_UP_CANCELED:
8bb78442	607	case CPU_UP_CANCELED_FROZEN:
fc75cdfa HC	608	if (!per_cpu(ksoftirqd, hotcpu))
fc75cdfa HC	609	break;
1da177e4	610	/* Unbind so it can run. Fall thru. */
a4c4af7c HC	611	kthread_bind(per_cpu(ksoftirqd, hotcpu),
a4c4af7c HC	612	any_online_cpu(cpu_online_map));
1da177e4	613	case CPU_DEAD:
1c6b4aa9 ST	614	case CPU_DEAD_FROZEN: {
	615	struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 };
	616
1da177e4 LT	617	p = per_cpu(ksoftirqd, hotcpu);
1da177e4 LT	618	per_cpu(ksoftirqd, hotcpu) = NULL;
1c6b4aa9	619	sched_setscheduler(p, SCHED_FIFO, &param);
1da177e4 LT	620	kthread_stop(p);
	621	takeover_tasklets(hotcpu);
	622	break;
1c6b4aa9	623	}
1da177e4 LT	624	#endif /* CONFIG_HOTPLUG_CPU */
	625	}
	626	return NOTIFY_OK;
	627	}
	628
8c78f307	629	static struct notifier_block __cpuinitdata cpu_nfb = {
1da177e4 LT	630	.notifier_call = cpu_callback
	631	};
	632
	633	__init int spawn_ksoftirqd(void)
	634	{
	635	void cpu = (void )(long)smp_processor_id();
07dccf33 AM	636	int err = cpu_callback(&cpu_nfb, CPU_UP_PREPARE, cpu);
	637
	638	BUG_ON(err == NOTIFY_BAD);
1da177e4 LT	639	cpu_callback(&cpu_nfb, CPU_ONLINE, cpu);
	640	register_cpu_notifier(&cpu_nfb);
	641	return 0;
	642	}
78eef01b AM	643
	644	#ifdef CONFIG_SMP
	645	/*
	646	* Call a function on all processors
	647	*/
	648	int on_each_cpu(void (func) (void info), void *info, int retry, int wait)
	649	{
	650	int ret = 0;
	651
	652	preempt_disable();
	653	ret = smp_call_function(func, info, retry, wait);
	654	local_irq_disable();
	655	func(info);
	656	local_irq_enable();
	657	preempt_enable();
	658	return ret;
	659	}
	660	EXPORT_SYMBOL(on_each_cpu);
	661	#endif