manage.c 32.4 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3
/*
 * linux/kernel/irq/manage.c
 *
I
Ingo Molnar 已提交
4 5
 * Copyright (C) 1992, 1998-2006 Linus Torvalds, Ingo Molnar
 * Copyright (C) 2005-2006 Thomas Gleixner
L
Linus Torvalds 已提交
6 7 8 9 10
 *
 * This file contains driver APIs to the irq subsystem.
 */

#include <linux/irq.h>
11
#include <linux/kthread.h>
L
Linus Torvalds 已提交
12 13 14
#include <linux/module.h>
#include <linux/random.h>
#include <linux/interrupt.h>
15
#include <linux/slab.h>
16
#include <linux/sched.h>
L
Linus Torvalds 已提交
17 18 19 20 21

#include "internals.h"

/**
 *	synchronize_irq - wait for pending IRQ handlers (on other CPUs)
22
 *	@irq: interrupt number to wait for
L
Linus Torvalds 已提交
23 24 25 26 27 28 29 30 31
 *
 *	This function waits for any pending IRQ handlers for this interrupt
 *	to complete before returning. If you use this function while
 *	holding a resource the IRQ handler may need you will deadlock.
 *
 *	This function may be called - with care - from IRQ context.
 */
void synchronize_irq(unsigned int irq)
{
32
	struct irq_desc *desc = irq_to_desc(irq);
33
	unsigned int state;
L
Linus Torvalds 已提交
34

35
	if (!desc)
36 37
		return;

38 39 40 41 42 43 44
	do {
		unsigned long flags;

		/*
		 * Wait until we're out of the critical section.  This might
		 * give the wrong answer due to the lack of memory barriers.
		 */
45
		while (desc->istate & IRQS_INPROGRESS)
46 47 48
			cpu_relax();

		/* Ok, that indicated we're done: double-check carefully. */
49
		raw_spin_lock_irqsave(&desc->lock, flags);
50
		state = desc->istate;
51
		raw_spin_unlock_irqrestore(&desc->lock, flags);
52 53

		/* Oops, that failed? */
54
	} while (state & IRQS_INPROGRESS);
55 56 57 58 59 60

	/*
	 * We made sure that no hardirq handler is running. Now verify
	 * that no threaded handlers are active.
	 */
	wait_event(desc->wait_for_threads, !atomic_read(&desc->threads_active));
L
Linus Torvalds 已提交
61 62 63
}
EXPORT_SYMBOL(synchronize_irq);

64 65 66
#ifdef CONFIG_SMP
cpumask_var_t irq_default_affinity;

67 68 69 70 71 72 73
/**
 *	irq_can_set_affinity - Check if the affinity of a given irq can be set
 *	@irq:		Interrupt to check
 *
 */
int irq_can_set_affinity(unsigned int irq)
{
74
	struct irq_desc *desc = irq_to_desc(irq);
75

76 77
	if (!desc || !irqd_can_balance(&desc->irq_data) ||
	    !desc->irq_data.chip || !desc->irq_data.chip->irq_set_affinity)
78 79 80 81 82
		return 0;

	return 1;
}

83 84 85 86 87 88 89 90 91 92
/**
 *	irq_set_thread_affinity - Notify irq threads to adjust affinity
 *	@desc:		irq descriptor which has affitnity changed
 *
 *	We just set IRQTF_AFFINITY and delegate the affinity setting
 *	to the interrupt thread itself. We can not call
 *	set_cpus_allowed_ptr() here as we hold desc->lock and this
 *	code can be called from hard interrupt context.
 */
void irq_set_thread_affinity(struct irq_desc *desc)
93 94 95 96 97
{
	struct irqaction *action = desc->action;

	while (action) {
		if (action->thread)
98
			set_bit(IRQTF_AFFINITY, &action->thread_flags);
99 100 101 102
		action = action->next;
	}
}

103 104 105
#ifdef CONFIG_GENERIC_PENDING_IRQ
static inline bool irq_can_move_pcntxt(struct irq_desc *desc)
{
106
	return irq_settings_can_move_pcntxt(desc);
107 108 109
}
static inline bool irq_move_pending(struct irq_desc *desc)
{
110
	return irqd_is_setaffinity_pending(&desc->irq_data);
111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
}
static inline void
irq_copy_pending(struct irq_desc *desc, const struct cpumask *mask)
{
	cpumask_copy(desc->pending_mask, mask);
}
static inline void
irq_get_pending(struct cpumask *mask, struct irq_desc *desc)
{
	cpumask_copy(mask, desc->pending_mask);
}
#else
static inline bool irq_can_move_pcntxt(struct irq_desc *desc) { return true; }
static inline bool irq_move_pending(struct irq_desc *desc) { return false; }
static inline void
irq_copy_pending(struct irq_desc *desc, const struct cpumask *mask) { }
static inline void
irq_get_pending(struct cpumask *mask, struct irq_desc *desc) { }
#endif

131 132 133 134 135 136
/**
 *	irq_set_affinity - Set the irq affinity of a given irq
 *	@irq:		Interrupt to set affinity
 *	@cpumask:	cpumask
 *
 */
137
int irq_set_affinity(unsigned int irq, const struct cpumask *mask)
138
{
139
	struct irq_desc *desc = irq_to_desc(irq);
140
	struct irq_chip *chip = desc->irq_data.chip;
141
	unsigned long flags;
142
	int ret = 0;
143

144
	if (!chip->irq_set_affinity)
145 146
		return -EINVAL;

147
	raw_spin_lock_irqsave(&desc->lock, flags);
148

149 150
	if (irq_can_move_pcntxt(desc)) {
		ret = chip->irq_set_affinity(&desc->irq_data, mask, false);
151 152
		switch (ret) {
		case IRQ_SET_MASK_OK:
153
			cpumask_copy(desc->irq_data.affinity, mask);
154
		case IRQ_SET_MASK_OK_NOCOPY:
155
			irq_set_thread_affinity(desc);
156
			ret = 0;
157
		}
158
	} else {
159
		irqd_set_move_pending(&desc->irq_data);
160
		irq_copy_pending(desc, mask);
161
	}
162

163 164 165 166
	if (desc->affinity_notify) {
		kref_get(&desc->affinity_notify->kref);
		schedule_work(&desc->affinity_notify->work);
	}
167 168
	irq_compat_set_affinity(desc);
	irqd_set(&desc->irq_data, IRQD_AFFINITY_SET);
169
	raw_spin_unlock_irqrestore(&desc->lock, flags);
170
	return ret;
171 172
}

173 174 175
int irq_set_affinity_hint(unsigned int irq, const struct cpumask *m)
{
	unsigned long flags;
176
	struct irq_desc *desc = irq_get_desc_lock(irq, &flags);
177 178 179 180

	if (!desc)
		return -EINVAL;
	desc->affinity_hint = m;
181
	irq_put_desc_unlock(desc, flags);
182 183 184 185
	return 0;
}
EXPORT_SYMBOL_GPL(irq_set_affinity_hint);

186 187 188 189 190 191 192 193
static void irq_affinity_notify(struct work_struct *work)
{
	struct irq_affinity_notify *notify =
		container_of(work, struct irq_affinity_notify, work);
	struct irq_desc *desc = irq_to_desc(notify->irq);
	cpumask_var_t cpumask;
	unsigned long flags;

194
	if (!desc || !alloc_cpumask_var(&cpumask, GFP_KERNEL))
195 196 197
		goto out;

	raw_spin_lock_irqsave(&desc->lock, flags);
198 199
	if (irq_move_pending(desc))
		irq_get_pending(cpumask, desc);
200
	else
201
		cpumask_copy(cpumask, desc->irq_data.affinity);
202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253
	raw_spin_unlock_irqrestore(&desc->lock, flags);

	notify->notify(notify, cpumask);

	free_cpumask_var(cpumask);
out:
	kref_put(&notify->kref, notify->release);
}

/**
 *	irq_set_affinity_notifier - control notification of IRQ affinity changes
 *	@irq:		Interrupt for which to enable/disable notification
 *	@notify:	Context for notification, or %NULL to disable
 *			notification.  Function pointers must be initialised;
 *			the other fields will be initialised by this function.
 *
 *	Must be called in process context.  Notification may only be enabled
 *	after the IRQ is allocated and must be disabled before the IRQ is
 *	freed using free_irq().
 */
int
irq_set_affinity_notifier(unsigned int irq, struct irq_affinity_notify *notify)
{
	struct irq_desc *desc = irq_to_desc(irq);
	struct irq_affinity_notify *old_notify;
	unsigned long flags;

	/* The release function is promised process context */
	might_sleep();

	if (!desc)
		return -EINVAL;

	/* Complete initialisation of *notify */
	if (notify) {
		notify->irq = irq;
		kref_init(&notify->kref);
		INIT_WORK(&notify->work, irq_affinity_notify);
	}

	raw_spin_lock_irqsave(&desc->lock, flags);
	old_notify = desc->affinity_notify;
	desc->affinity_notify = notify;
	raw_spin_unlock_irqrestore(&desc->lock, flags);

	if (old_notify)
		kref_put(&old_notify->kref, old_notify->release);

	return 0;
}
EXPORT_SYMBOL_GPL(irq_set_affinity_notifier);

254 255 256 257
#ifndef CONFIG_AUTO_IRQ_AFFINITY
/*
 * Generic version of the affinity autoselector.
 */
258 259
static int
setup_affinity(unsigned int irq, struct irq_desc *desc, struct cpumask *mask)
260
{
261
	struct irq_chip *chip = irq_desc_get_chip(desc);
262
	struct cpumask *set = irq_default_affinity;
263
	int ret;
264

265
	/* Excludes PER_CPU and NO_BALANCE interrupts */
266 267 268
	if (!irq_can_set_affinity(irq))
		return 0;

269 270 271 272
	/*
	 * Preserve an userspace affinity setup, but make sure that
	 * one of the targets is online.
	 */
273
	if (irqd_has_set(&desc->irq_data, IRQD_AFFINITY_SET)) {
274 275 276
		if (cpumask_intersects(desc->irq_data.affinity,
				       cpu_online_mask))
			set = desc->irq_data.affinity;
277 278 279 280
		else {
			irq_compat_clr_affinity(desc);
			irqd_clear(&desc->irq_data, IRQD_AFFINITY_SET);
		}
281
	}
282

283 284 285 286 287 288 289 290
	cpumask_and(mask, cpu_online_mask, set);
	ret = chip->irq_set_affinity(&desc->irq_data, mask, false);
	switch (ret) {
	case IRQ_SET_MASK_OK:
		cpumask_copy(desc->irq_data.affinity, mask);
	case IRQ_SET_MASK_OK_NOCOPY:
		irq_set_thread_affinity(desc);
	}
291 292
	return 0;
}
293
#else
294 295
static inline int
setup_affinity(unsigned int irq, struct irq_desc *d, struct cpumask *mask)
296 297 298
{
	return irq_select_affinity(irq);
}
299 300
#endif

301 302 303
/*
 * Called when affinity is set via /proc/irq
 */
304
int irq_select_affinity_usr(unsigned int irq, struct cpumask *mask)
305 306 307 308 309
{
	struct irq_desc *desc = irq_to_desc(irq);
	unsigned long flags;
	int ret;

310
	raw_spin_lock_irqsave(&desc->lock, flags);
311
	ret = setup_affinity(irq, desc, mask);
312
	raw_spin_unlock_irqrestore(&desc->lock, flags);
313 314 315 316
	return ret;
}

#else
317 318
static inline int
setup_affinity(unsigned int irq, struct irq_desc *desc, struct cpumask *mask)
319 320 321
{
	return 0;
}
L
Linus Torvalds 已提交
322 323
#endif

324 325 326
void __disable_irq(struct irq_desc *desc, unsigned int irq, bool suspend)
{
	if (suspend) {
327
		if (!desc->action || (desc->action->flags & IRQF_NO_SUSPEND))
328
			return;
329
		desc->istate |= IRQS_SUSPENDED;
330 331
	}

332
	if (!desc->depth++)
333
		irq_disable(desc);
334 335
}

336 337 338 339 340 341 342 343 344 345 346 347
static int __disable_irq_nosync(unsigned int irq)
{
	unsigned long flags;
	struct irq_desc *desc = irq_get_desc_buslock(irq, &flags);

	if (!desc)
		return -EINVAL;
	__disable_irq(desc, irq, false);
	irq_put_desc_busunlock(desc, flags);
	return 0;
}

L
Linus Torvalds 已提交
348 349 350 351 352 353 354 355 356 357 358 359 360
/**
 *	disable_irq_nosync - disable an irq without waiting
 *	@irq: Interrupt to disable
 *
 *	Disable the selected interrupt line.  Disables and Enables are
 *	nested.
 *	Unlike disable_irq(), this function does not ensure existing
 *	instances of the IRQ handler have completed before returning.
 *
 *	This function may be called from IRQ context.
 */
void disable_irq_nosync(unsigned int irq)
{
361
	__disable_irq_nosync(irq);
L
Linus Torvalds 已提交
362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378
}
EXPORT_SYMBOL(disable_irq_nosync);

/**
 *	disable_irq - disable an irq and wait for completion
 *	@irq: Interrupt to disable
 *
 *	Disable the selected interrupt line.  Enables and Disables are
 *	nested.
 *	This function waits for any pending IRQ handlers for this interrupt
 *	to complete before returning. If you use this function while
 *	holding a resource the IRQ handler may need you will deadlock.
 *
 *	This function may be called - with care - from IRQ context.
 */
void disable_irq(unsigned int irq)
{
379
	if (!__disable_irq_nosync(irq))
L
Linus Torvalds 已提交
380 381 382 383
		synchronize_irq(irq);
}
EXPORT_SYMBOL(disable_irq);

384
void __enable_irq(struct irq_desc *desc, unsigned int irq, bool resume)
385
{
T
Thomas Gleixner 已提交
386
	if (resume) {
387
		if (!(desc->istate & IRQS_SUSPENDED)) {
T
Thomas Gleixner 已提交
388 389 390 391 392 393 394
			if (!desc->action)
				return;
			if (!(desc->action->flags & IRQF_FORCE_RESUME))
				return;
			/* Pretend that it got disabled ! */
			desc->depth++;
		}
395
		desc->istate &= ~IRQS_SUSPENDED;
T
Thomas Gleixner 已提交
396
	}
397

398 399
	switch (desc->depth) {
	case 0:
400
 err_out:
401
		WARN(1, KERN_WARNING "Unbalanced enable for IRQ %d\n", irq);
402 403
		break;
	case 1: {
404
		if (desc->istate & IRQS_SUSPENDED)
405
			goto err_out;
406
		/* Prevent probing on this irq: */
407
		irq_settings_set_noprobe(desc);
408
		irq_enable(desc);
409 410 411 412 413 414 415 416
		check_irq_resend(desc, irq);
		/* fall-through */
	}
	default:
		desc->depth--;
	}
}

L
Linus Torvalds 已提交
417 418 419 420 421 422 423 424
/**
 *	enable_irq - enable handling of an irq
 *	@irq: Interrupt to enable
 *
 *	Undoes the effect of one call to disable_irq().  If this
 *	matches the last disable, processing of interrupts on this
 *	IRQ line is re-enabled.
 *
T
Thomas Gleixner 已提交
425
 *	This function may be called from IRQ context only when
426
 *	desc->irq_data.chip->bus_lock and desc->chip->bus_sync_unlock are NULL !
L
Linus Torvalds 已提交
427 428 429 430
 */
void enable_irq(unsigned int irq)
{
	unsigned long flags;
431
	struct irq_desc *desc = irq_get_desc_buslock(irq, &flags);
L
Linus Torvalds 已提交
432

433
	if (!desc)
434
		return;
T
Thomas Gleixner 已提交
435 436
	if (WARN(!desc->irq_data.chip,
		 KERN_ERR "enable_irq before setup/request_irq: irq %u\n", irq))
437
		goto out;
438

439
	__enable_irq(desc, irq, false);
440 441
out:
	irq_put_desc_busunlock(desc, flags);
L
Linus Torvalds 已提交
442 443 444
}
EXPORT_SYMBOL(enable_irq);

D
David Brownell 已提交
445
static int set_irq_wake_real(unsigned int irq, unsigned int on)
446
{
447
	struct irq_desc *desc = irq_to_desc(irq);
448 449
	int ret = -ENXIO;

450 451
	if (desc->irq_data.chip->irq_set_wake)
		ret = desc->irq_data.chip->irq_set_wake(&desc->irq_data, on);
452 453 454 455

	return ret;
}

456
/**
T
Thomas Gleixner 已提交
457
 *	irq_set_irq_wake - control irq power management wakeup
458 459 460
 *	@irq:	interrupt to control
 *	@on:	enable/disable power management wakeup
 *
461 462 463 464 465 466
 *	Enable/disable power management wakeup mode, which is
 *	disabled by default.  Enables and disables must match,
 *	just as they match for non-wakeup mode support.
 *
 *	Wakeup mode lets this IRQ wake the system from sleep
 *	states like "suspend to RAM".
467
 */
T
Thomas Gleixner 已提交
468
int irq_set_irq_wake(unsigned int irq, unsigned int on)
469 470
{
	unsigned long flags;
471
	struct irq_desc *desc = irq_get_desc_buslock(irq, &flags);
472
	int ret = 0;
473

474 475 476 477
	/* wakeup-capable irqs can be shared between drivers that
	 * don't need to have the same sleep mode behaviors.
	 */
	if (on) {
478 479 480 481 482
		if (desc->wake_depth++ == 0) {
			ret = set_irq_wake_real(irq, on);
			if (ret)
				desc->wake_depth = 0;
			else
483
				irqd_set(&desc->irq_data, IRQD_WAKEUP_STATE);
484
		}
485 486
	} else {
		if (desc->wake_depth == 0) {
487
			WARN(1, "Unbalanced IRQ %d wake disable\n", irq);
488 489 490 491 492
		} else if (--desc->wake_depth == 0) {
			ret = set_irq_wake_real(irq, on);
			if (ret)
				desc->wake_depth = 1;
			else
493
				irqd_clear(&desc->irq_data, IRQD_WAKEUP_STATE);
494
		}
495
	}
496
	irq_put_desc_busunlock(desc, flags);
497 498
	return ret;
}
T
Thomas Gleixner 已提交
499
EXPORT_SYMBOL(irq_set_irq_wake);
500

L
Linus Torvalds 已提交
501 502 503 504 505 506 507
/*
 * Internal function that tells the architecture code whether a
 * particular irq has been exclusively allocated or is available
 * for driver use.
 */
int can_request_irq(unsigned int irq, unsigned long irqflags)
{
508
	unsigned long flags;
509 510
	struct irq_desc *desc = irq_get_desc_lock(irq, &flags);
	int canrequest = 0;
L
Linus Torvalds 已提交
511

512 513 514
	if (!desc)
		return 0;

515 516 517 518 519 520 521
	if (irq_settings_can_request(desc)) {
		if (desc->action)
			if (irqflags & desc->action->flags & IRQF_SHARED)
				canrequest =1;
	}
	irq_put_desc_unlock(desc, flags);
	return canrequest;
L
Linus Torvalds 已提交
522 523
}

D
David Brownell 已提交
524
int __irq_set_trigger(struct irq_desc *desc, unsigned int irq,
525
		      unsigned long flags)
526
{
527
	struct irq_chip *chip = desc->irq_data.chip;
528
	int ret, unmask = 0;
529

530
	if (!chip || !chip->irq_set_type) {
531 532 533 534
		/*
		 * IRQF_TRIGGER_* but the PIC does not support multiple
		 * flow-types?
		 */
535
		pr_debug("No set_type function for IRQ %d (%s)\n", irq,
536 537 538 539
				chip ? (chip->name ? : "unknown") : "unknown");
		return 0;
	}

540
	flags &= IRQ_TYPE_SENSE_MASK;
541 542 543 544 545 546 547 548

	if (chip->flags & IRQCHIP_SET_TYPE_MASKED) {
		if (!(desc->istate & IRQS_MASKED))
			mask_irq(desc);
		if (!(desc->istate & IRQS_DISABLED))
			unmask = 1;
	}

549
	/* caller masked out all except trigger mode flags */
550
	ret = chip->irq_set_type(&desc->irq_data, flags);
551

552 553 554 555 556 557 558 559 560 561 562 563 564 565
	switch (ret) {
	case IRQ_SET_MASK_OK:
		irqd_clear(&desc->irq_data, IRQD_TRIGGER_MASK);
		irqd_set(&desc->irq_data, flags);

	case IRQ_SET_MASK_OK_NOCOPY:
		flags = irqd_get_trigger_type(&desc->irq_data);
		irq_settings_set_trigger_mask(desc, flags);
		irqd_clear(&desc->irq_data, IRQD_LEVEL);
		irq_settings_clr_level(desc);
		if (flags & IRQ_TYPE_LEVEL_MASK) {
			irq_settings_set_level(desc);
			irqd_set(&desc->irq_data, IRQD_LEVEL);
		}
566

567 568
		if (chip != desc->irq_data.chip)
			irq_chip_set_defaults(desc->irq_data.chip);
569
		ret = 0;
570 571 572
	default:
		pr_err("setting trigger mode %lu for irq %u failed (%pF)\n",
		       flags, irq, chip->irq_set_type);
D
David Brownell 已提交
573
	}
574 575
	if (unmask)
		unmask_irq(desc);
576 577 578
	return ret;
}

T
Thomas Gleixner 已提交
579 580 581 582 583 584 585 586 587 588
/*
 * Default primary interrupt handler for threaded interrupts. Is
 * assigned as primary handler when request_threaded_irq is called
 * with handler == NULL. Useful for oneshot interrupts.
 */
static irqreturn_t irq_default_primary_handler(int irq, void *dev_id)
{
	return IRQ_WAKE_THREAD;
}

589 590 591 592 593 594 595 596 597 598
/*
 * Primary handler for nested threaded interrupts. Should never be
 * called.
 */
static irqreturn_t irq_nested_primary_handler(int irq, void *dev_id)
{
	WARN(1, "Primary handler called for nested irq %d\n", irq);
	return IRQ_NONE;
}

599 600 601 602
static int irq_wait_for_interrupt(struct irqaction *action)
{
	while (!kthread_should_stop()) {
		set_current_state(TASK_INTERRUPTIBLE);
603 604 605

		if (test_and_clear_bit(IRQTF_RUNTHREAD,
				       &action->thread_flags)) {
606 607
			__set_current_state(TASK_RUNNING);
			return 0;
608 609
		}
		schedule();
610 611 612 613
	}
	return -1;
}

T
Thomas Gleixner 已提交
614 615 616 617 618 619 620
/*
 * Oneshot interrupts keep the irq line masked until the threaded
 * handler finished. unmask if the interrupt has not been disabled and
 * is marked MASKED.
 */
static void irq_finalize_oneshot(unsigned int irq, struct irq_desc *desc)
{
621
again:
622
	chip_bus_lock(desc);
623
	raw_spin_lock_irq(&desc->lock);
624 625 626 627 628 629 630 631

	/*
	 * Implausible though it may be we need to protect us against
	 * the following scenario:
	 *
	 * The thread is faster done than the hard interrupt handler
	 * on the other CPU. If we unmask the irq line then the
	 * interrupt can come in again and masks the line, leaves due
632
	 * to IRQS_INPROGRESS and the irq line is masked forever.
633
	 */
634
	if (unlikely(desc->istate & IRQS_INPROGRESS)) {
635
		raw_spin_unlock_irq(&desc->lock);
636
		chip_bus_sync_unlock(desc);
637 638 639 640
		cpu_relax();
		goto again;
	}

641 642 643
	if (!(desc->istate & IRQS_DISABLED) && (desc->istate & IRQS_MASKED)) {
		irq_compat_clr_masked(desc);
		desc->istate &= ~IRQS_MASKED;
644
		desc->irq_data.chip->irq_unmask(&desc->irq_data);
T
Thomas Gleixner 已提交
645
	}
646
	raw_spin_unlock_irq(&desc->lock);
647
	chip_bus_sync_unlock(desc);
T
Thomas Gleixner 已提交
648 649
}

650
#ifdef CONFIG_SMP
651
/*
652
 * Check whether we need to chasnge the affinity of the interrupt thread.
653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670
 */
static void
irq_thread_check_affinity(struct irq_desc *desc, struct irqaction *action)
{
	cpumask_var_t mask;

	if (!test_and_clear_bit(IRQTF_AFFINITY, &action->thread_flags))
		return;

	/*
	 * In case we are out of memory we set IRQTF_AFFINITY again and
	 * try again next time
	 */
	if (!alloc_cpumask_var(&mask, GFP_KERNEL)) {
		set_bit(IRQTF_AFFINITY, &action->thread_flags);
		return;
	}

671
	raw_spin_lock_irq(&desc->lock);
672
	cpumask_copy(mask, desc->irq_data.affinity);
673
	raw_spin_unlock_irq(&desc->lock);
674 675 676 677

	set_cpus_allowed_ptr(current, mask);
	free_cpumask_var(mask);
}
678 679 680 681
#else
static inline void
irq_thread_check_affinity(struct irq_desc *desc, struct irqaction *action) { }
#endif
682

683 684 685 686 687
/*
 * Interrupt handler thread
 */
static int irq_thread(void *data)
{
688
	static const struct sched_param param = {
689 690
		.sched_priority = MAX_USER_RT_PRIO/2,
	};
691 692
	struct irqaction *action = data;
	struct irq_desc *desc = irq_to_desc(action->irq);
693
	int wake, oneshot = desc->istate & IRQS_ONESHOT;
694 695 696 697 698 699

	sched_setscheduler(current, SCHED_FIFO, &param);
	current->irqaction = action;

	while (!irq_wait_for_interrupt(action)) {

700 701
		irq_thread_check_affinity(desc, action);

702 703
		atomic_inc(&desc->threads_active);

704
		raw_spin_lock_irq(&desc->lock);
705
		if (unlikely(desc->istate & IRQS_DISABLED)) {
706 707 708 709
			/*
			 * CHECKME: We might need a dedicated
			 * IRQ_THREAD_PENDING flag here, which
			 * retriggers the thread in check_irq_resend()
710
			 * but AFAICT IRQS_PENDING should be fine as it
711 712
			 * retriggers the interrupt itself --- tglx
			 */
713 714
			irq_compat_set_pending(desc);
			desc->istate |= IRQS_PENDING;
715
			raw_spin_unlock_irq(&desc->lock);
716
		} else {
717
			raw_spin_unlock_irq(&desc->lock);
718 719

			action->thread_fn(action->irq, action->dev_id);
T
Thomas Gleixner 已提交
720 721 722

			if (oneshot)
				irq_finalize_oneshot(action->irq, desc);
723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759
		}

		wake = atomic_dec_and_test(&desc->threads_active);

		if (wake && waitqueue_active(&desc->wait_for_threads))
			wake_up(&desc->wait_for_threads);
	}

	/*
	 * Clear irqaction. Otherwise exit_irq_thread() would make
	 * fuzz about an active irq thread going into nirvana.
	 */
	current->irqaction = NULL;
	return 0;
}

/*
 * Called from do_exit()
 */
void exit_irq_thread(void)
{
	struct task_struct *tsk = current;

	if (!tsk->irqaction)
		return;

	printk(KERN_ERR
	       "exiting task \"%s\" (%d) is an active IRQ thread (irq %d)\n",
	       tsk->comm ? tsk->comm : "", tsk->pid, tsk->irqaction->irq);

	/*
	 * Set the THREAD DIED flag to prevent further wakeups of the
	 * soon to be gone threaded handler.
	 */
	set_bit(IRQTF_DIED, &tsk->irqaction->flags);
}

L
Linus Torvalds 已提交
760 761 762 763
/*
 * Internal function to register an irqaction - typically used to
 * allocate special interrupts that are part of the architecture.
 */
764
static int
I
Ingo Molnar 已提交
765
__setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new)
L
Linus Torvalds 已提交
766
{
767
	struct irqaction *old, **old_ptr;
768
	const char *old_name = NULL;
L
Linus Torvalds 已提交
769
	unsigned long flags;
770 771
	int ret, nested, shared = 0;
	cpumask_var_t mask;
L
Linus Torvalds 已提交
772

773
	if (!desc)
774 775
		return -EINVAL;

776
	if (desc->irq_data.chip == &no_irq_chip)
L
Linus Torvalds 已提交
777 778 779 780 781 782
		return -ENOSYS;
	/*
	 * Some drivers like serial.c use request_irq() heavily,
	 * so we have to be careful not to interfere with a
	 * running system.
	 */
783
	if (new->flags & IRQF_SAMPLE_RANDOM) {
L
Linus Torvalds 已提交
784 785 786 787 788 789 790 791 792 793 794
		/*
		 * This function might sleep, we want to call it first,
		 * outside of the atomic block.
		 * Yes, this might clear the entropy pool if the wrong
		 * driver is attempted to be loaded, without actually
		 * installing a new handler, but is this really a problem,
		 * only the sysadmin is able to do this.
		 */
		rand_initialize_irq(irq);
	}

T
Thomas Gleixner 已提交
795 796 797 798
	/* Oneshot interrupts are not allowed with shared */
	if ((new->flags & IRQF_ONESHOT) && (new->flags & IRQF_SHARED))
		return -EINVAL;

799
	/*
800 801 802
	 * Check whether the interrupt nests into another interrupt
	 * thread.
	 */
803
	nested = irq_settings_is_nested_thread(desc);
804 805 806 807 808 809 810 811 812 813 814
	if (nested) {
		if (!new->thread_fn)
			return -EINVAL;
		/*
		 * Replace the primary handler which was provided from
		 * the driver for non nested interrupt handling by the
		 * dummy function which warns when called.
		 */
		new->handler = irq_nested_primary_handler;
	}

815
	/*
816 817 818
	 * Create a handler thread when a thread function is supplied
	 * and the interrupt does not nest into another interrupt
	 * thread.
819
	 */
820
	if (new->thread_fn && !nested) {
821 822 823 824 825 826 827 828 829 830 831 832 833 834 835
		struct task_struct *t;

		t = kthread_create(irq_thread, new, "irq/%d-%s", irq,
				   new->name);
		if (IS_ERR(t))
			return PTR_ERR(t);
		/*
		 * We keep the reference to the task struct even if
		 * the thread dies to avoid that the interrupt code
		 * references an already freed task_struct.
		 */
		get_task_struct(t);
		new->thread = t;
	}

836 837 838 839 840
	if (!alloc_cpumask_var(&mask, GFP_KERNEL)) {
		ret = -ENOMEM;
		goto out_thread;
	}

L
Linus Torvalds 已提交
841 842 843
	/*
	 * The following block of code has to be executed atomically
	 */
844
	raw_spin_lock_irqsave(&desc->lock, flags);
845 846
	old_ptr = &desc->action;
	old = *old_ptr;
847
	if (old) {
848 849 850
		/*
		 * Can't share interrupts unless both agree to and are
		 * the same type (level, edge, polarity). So both flag
851
		 * fields must have IRQF_SHARED set and the bits which
852 853
		 * set the trigger type must match.
		 */
854
		if (!((old->flags & new->flags) & IRQF_SHARED) ||
855 856
		    ((old->flags ^ new->flags) & IRQF_TRIGGER_MASK)) {
			old_name = old->name;
857
			goto mismatch;
858
		}
859 860

		/* All handlers must agree on per-cpuness */
861 862
		if ((old->flags & IRQF_PERCPU) !=
		    (new->flags & IRQF_PERCPU))
863
			goto mismatch;
L
Linus Torvalds 已提交
864 865 866

		/* add new interrupt at end of irq queue */
		do {
867 868
			old_ptr = &old->next;
			old = *old_ptr;
L
Linus Torvalds 已提交
869 870 871 872 873
		} while (old);
		shared = 1;
	}

	if (!shared) {
874
		irq_chip_set_defaults(desc->irq_data.chip);
875

876 877
		init_waitqueue_head(&desc->wait_for_threads);

878
		/* Setup the type (level, edge polarity) if configured: */
879
		if (new->flags & IRQF_TRIGGER_MASK) {
880 881
			ret = __irq_set_trigger(desc, irq,
					new->flags & IRQF_TRIGGER_MASK);
882

883
			if (ret)
884
				goto out_mask;
885
		}
T
Thomas Gleixner 已提交
886

887
		desc->istate &= ~(IRQS_AUTODETECT | IRQS_SPURIOUS_DISABLED | \
888 889
				  IRQS_INPROGRESS | IRQS_ONESHOT | \
				  IRQS_WAITING);
890

891 892 893 894
		if (new->flags & IRQF_PERCPU) {
			irqd_set(&desc->irq_data, IRQD_PER_CPU);
			irq_settings_set_per_cpu(desc);
		}
895

T
Thomas Gleixner 已提交
896
		if (new->flags & IRQF_ONESHOT)
897
			desc->istate |= IRQS_ONESHOT;
T
Thomas Gleixner 已提交
898

899
		if (irq_settings_can_autoenable(desc))
900 901
			irq_startup(desc);
		else
902 903
			/* Undo nested disables: */
			desc->depth = 1;
904

905
		/* Exclude IRQ from balancing if requested */
906 907 908 909
		if (new->flags & IRQF_NOBALANCING) {
			irq_settings_set_no_balancing(desc);
			irqd_set(&desc->irq_data, IRQD_NO_BALANCING);
		}
910

911
		/* Set default affinity mask once everything is setup */
912
		setup_affinity(irq, desc, mask);
D
David Brownell 已提交
913

914 915 916 917 918 919 920 921
	} else if (new->flags & IRQF_TRIGGER_MASK) {
		unsigned int nmsk = new->flags & IRQF_TRIGGER_MASK;
		unsigned int omsk = irq_settings_get_trigger_mask(desc);

		if (nmsk != omsk)
			/* hope the handler works with current  trigger mode */
			pr_warning("IRQ %d uses trigger mode %u; requested %u\n",
				   irq, nmsk, omsk);
L
Linus Torvalds 已提交
922
	}
923

924
	new->irq = irq;
925
	*old_ptr = new;
926

927 928 929
	/* Reset broken irq detection when installing new handler */
	desc->irq_count = 0;
	desc->irqs_unhandled = 0;
930 931 932 933 934

	/*
	 * Check whether we disabled the irq via the spurious handler
	 * before. Reenable it and give it another chance.
	 */
935 936
	if (shared && (desc->istate & IRQS_SPURIOUS_DISABLED)) {
		desc->istate &= ~IRQS_SPURIOUS_DISABLED;
937
		__enable_irq(desc, irq, false);
938 939
	}

940
	raw_spin_unlock_irqrestore(&desc->lock, flags);
L
Linus Torvalds 已提交
941

942 943 944 945 946 947 948
	/*
	 * Strictly no need to wake it up, but hung_task complains
	 * when no hard interrupt wakes the thread up.
	 */
	if (new->thread)
		wake_up_process(new->thread);

949
	register_irq_proc(irq, desc);
L
Linus Torvalds 已提交
950 951 952 953
	new->dir = NULL;
	register_handler_proc(irq, new);

	return 0;
954 955

mismatch:
956
#ifdef CONFIG_DEBUG_SHIRQ
957
	if (!(new->flags & IRQF_PROBE_SHARED)) {
958
		printk(KERN_ERR "IRQ handler type mismatch for IRQ %d\n", irq);
959 960
		if (old_name)
			printk(KERN_ERR "current handler: %s\n", old_name);
961 962
		dump_stack();
	}
963
#endif
964 965
	ret = -EBUSY;

966 967 968
out_mask:
	free_cpumask_var(mask);

969
out_thread:
970
	raw_spin_unlock_irqrestore(&desc->lock, flags);
971 972 973 974 975 976 977 978 979
	if (new->thread) {
		struct task_struct *t = new->thread;

		new->thread = NULL;
		if (likely(!test_bit(IRQTF_DIED, &new->thread_flags)))
			kthread_stop(t);
		put_task_struct(t);
	}
	return ret;
L
Linus Torvalds 已提交
980 981
}

982 983 984 985 986 987 988 989 990
/**
 *	setup_irq - setup an interrupt
 *	@irq: Interrupt line to setup
 *	@act: irqaction for the interrupt
 *
 * Used to statically setup interrupts in the early boot process.
 */
int setup_irq(unsigned int irq, struct irqaction *act)
{
991
	int retval;
992 993
	struct irq_desc *desc = irq_to_desc(irq);

994 995 996 997 998
	chip_bus_lock(desc);
	retval = __setup_irq(irq, desc, act);
	chip_bus_sync_unlock(desc);

	return retval;
999
}
1000
EXPORT_SYMBOL_GPL(setup_irq);
1001

1002 1003 1004
 /*
 * Internal function to unregister an irqaction - used to free
 * regular and special interrupts that are part of the architecture.
L
Linus Torvalds 已提交
1005
 */
1006
static struct irqaction *__free_irq(unsigned int irq, void *dev_id)
L
Linus Torvalds 已提交
1007
{
1008
	struct irq_desc *desc = irq_to_desc(irq);
1009
	struct irqaction *action, **action_ptr;
L
Linus Torvalds 已提交
1010 1011
	unsigned long flags;

1012
	WARN(in_interrupt(), "Trying to free IRQ %d from IRQ context!\n", irq);
1013 1014

	if (!desc)
1015
		return NULL;
L
Linus Torvalds 已提交
1016

1017
	raw_spin_lock_irqsave(&desc->lock, flags);
1018 1019 1020 1021 1022

	/*
	 * There can be multiple actions per IRQ descriptor, find the right
	 * one based on the dev_id:
	 */
1023
	action_ptr = &desc->action;
L
Linus Torvalds 已提交
1024
	for (;;) {
1025
		action = *action_ptr;
L
Linus Torvalds 已提交
1026

1027 1028
		if (!action) {
			WARN(1, "Trying to free already-free IRQ %d\n", irq);
1029
			raw_spin_unlock_irqrestore(&desc->lock, flags);
L
Linus Torvalds 已提交
1030

1031
			return NULL;
1032
		}
L
Linus Torvalds 已提交
1033

1034 1035
		if (action->dev_id == dev_id)
			break;
1036
		action_ptr = &action->next;
1037
	}
1038

1039
	/* Found it - now remove it from the list of entries: */
1040
	*action_ptr = action->next;
1041 1042

	/* Currently used only by UML, might disappear one day: */
1043
#ifdef CONFIG_IRQ_RELEASE_METHOD
1044 1045
	if (desc->irq_data.chip->release)
		desc->irq_data.chip->release(irq, dev_id);
1046
#endif
1047

1048
	/* If this was the last handler, shut down the IRQ line: */
1049 1050
	if (!desc->action)
		irq_shutdown(desc);
1051

1052 1053 1054 1055 1056 1057
#ifdef CONFIG_SMP
	/* make sure affinity_hint is cleaned up */
	if (WARN_ON_ONCE(desc->affinity_hint))
		desc->affinity_hint = NULL;
#endif

1058
	raw_spin_unlock_irqrestore(&desc->lock, flags);
1059 1060 1061 1062 1063

	unregister_handler_proc(irq, action);

	/* Make sure it's not being used on another CPU: */
	synchronize_irq(irq);
L
Linus Torvalds 已提交
1064

1065
#ifdef CONFIG_DEBUG_SHIRQ
1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077
	/*
	 * It's a shared IRQ -- the driver ought to be prepared for an IRQ
	 * event to happen even now it's being freed, so let's make sure that
	 * is so by doing an extra call to the handler ....
	 *
	 * ( We do this after actually deregistering it, to make sure that a
	 *   'real' IRQ doesn't run in * parallel with our fake. )
	 */
	if (action->flags & IRQF_SHARED) {
		local_irq_save(flags);
		action->handler(irq, dev_id);
		local_irq_restore(flags);
L
Linus Torvalds 已提交
1078
	}
1079
#endif
1080 1081 1082 1083 1084 1085 1086

	if (action->thread) {
		if (!test_bit(IRQTF_DIED, &action->thread_flags))
			kthread_stop(action->thread);
		put_task_struct(action->thread);
	}

1087 1088 1089
	return action;
}

1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100
/**
 *	remove_irq - free an interrupt
 *	@irq: Interrupt line to free
 *	@act: irqaction for the interrupt
 *
 * Used to remove interrupts statically setup by the early boot process.
 */
void remove_irq(unsigned int irq, struct irqaction *act)
{
	__free_irq(irq, act->dev_id);
}
1101
EXPORT_SYMBOL_GPL(remove_irq);
1102

1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118
/**
 *	free_irq - free an interrupt allocated with request_irq
 *	@irq: Interrupt line to free
 *	@dev_id: Device identity to free
 *
 *	Remove an interrupt handler. The handler is removed and if the
 *	interrupt line is no longer in use by any driver it is disabled.
 *	On a shared IRQ the caller must ensure the interrupt is disabled
 *	on the card it drives before calling this function. The function
 *	does not return until any executing interrupts for this IRQ
 *	have completed.
 *
 *	This function must not be called from interrupt context.
 */
void free_irq(unsigned int irq, void *dev_id)
{
T
Thomas Gleixner 已提交
1119 1120 1121 1122 1123
	struct irq_desc *desc = irq_to_desc(irq);

	if (!desc)
		return;

1124 1125 1126 1127 1128
#ifdef CONFIG_SMP
	if (WARN_ON(desc->affinity_notify))
		desc->affinity_notify = NULL;
#endif

1129
	chip_bus_lock(desc);
1130
	kfree(__free_irq(irq, dev_id));
1131
	chip_bus_sync_unlock(desc);
L
Linus Torvalds 已提交
1132 1133 1134 1135
}
EXPORT_SYMBOL(free_irq);

/**
1136
 *	request_threaded_irq - allocate an interrupt line
L
Linus Torvalds 已提交
1137
 *	@irq: Interrupt line to allocate
1138 1139
 *	@handler: Function to be called when the IRQ occurs.
 *		  Primary handler for threaded interrupts
T
Thomas Gleixner 已提交
1140 1141
 *		  If NULL and thread_fn != NULL the default
 *		  primary handler is installed
1142 1143
 *	@thread_fn: Function called from the irq handler thread
 *		    If NULL, no irq thread is created
L
Linus Torvalds 已提交
1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154
 *	@irqflags: Interrupt type flags
 *	@devname: An ascii name for the claiming device
 *	@dev_id: A cookie passed back to the handler function
 *
 *	This call allocates interrupt resources and enables the
 *	interrupt line and IRQ handling. From the point this
 *	call is made your handler function may be invoked. Since
 *	your handler function must clear any interrupt the board
 *	raises, you must take care both to initialise your hardware
 *	and to set up the interrupt handler in the right order.
 *
1155 1156 1157 1158 1159
 *	If you want to set up a threaded irq handler for your device
 *	then you need to supply @handler and @thread_fn. @handler ist
 *	still called in hard interrupt context and has to check
 *	whether the interrupt originates from the device. If yes it
 *	needs to disable the interrupt on the device and return
1160
 *	IRQ_WAKE_THREAD which will wake up the handler thread and run
1161 1162 1163
 *	@thread_fn. This split handler design is necessary to support
 *	shared interrupts.
 *
L
Linus Torvalds 已提交
1164 1165 1166 1167 1168 1169 1170 1171 1172
 *	Dev_id must be globally unique. Normally the address of the
 *	device data structure is used as the cookie. Since the handler
 *	receives this value it makes sense to use it.
 *
 *	If your interrupt is shared you must pass a non NULL dev_id
 *	as this is required when freeing the interrupt.
 *
 *	Flags:
 *
1173 1174
 *	IRQF_SHARED		Interrupt is shared
 *	IRQF_SAMPLE_RANDOM	The interrupt can be used for entropy
D
David Brownell 已提交
1175
 *	IRQF_TRIGGER_*		Specify active edge(s) or level
L
Linus Torvalds 已提交
1176 1177
 *
 */
1178 1179 1180
int request_threaded_irq(unsigned int irq, irq_handler_t handler,
			 irq_handler_t thread_fn, unsigned long irqflags,
			 const char *devname, void *dev_id)
L
Linus Torvalds 已提交
1181
{
1182
	struct irqaction *action;
1183
	struct irq_desc *desc;
1184
	int retval;
L
Linus Torvalds 已提交
1185 1186 1187 1188 1189 1190 1191

	/*
	 * Sanity-check: shared interrupts must pass in a real dev-ID,
	 * otherwise we'll have trouble later trying to figure out
	 * which interrupt is which (messes up the interrupt freeing
	 * logic etc).
	 */
1192
	if ((irqflags & IRQF_SHARED) && !dev_id)
L
Linus Torvalds 已提交
1193
		return -EINVAL;
1194

1195
	desc = irq_to_desc(irq);
1196
	if (!desc)
L
Linus Torvalds 已提交
1197
		return -EINVAL;
1198

1199
	if (!irq_settings_can_request(desc))
1200
		return -EINVAL;
T
Thomas Gleixner 已提交
1201 1202 1203 1204 1205 1206

	if (!handler) {
		if (!thread_fn)
			return -EINVAL;
		handler = irq_default_primary_handler;
	}
L
Linus Torvalds 已提交
1207

1208
	action = kzalloc(sizeof(struct irqaction), GFP_KERNEL);
L
Linus Torvalds 已提交
1209 1210 1211 1212
	if (!action)
		return -ENOMEM;

	action->handler = handler;
1213
	action->thread_fn = thread_fn;
L
Linus Torvalds 已提交
1214 1215 1216 1217
	action->flags = irqflags;
	action->name = devname;
	action->dev_id = dev_id;

1218
	chip_bus_lock(desc);
1219
	retval = __setup_irq(irq, desc, action);
1220
	chip_bus_sync_unlock(desc);
T
Thomas Gleixner 已提交
1221

1222 1223 1224
	if (retval)
		kfree(action);

1225
#ifdef CONFIG_DEBUG_SHIRQ_FIXME
1226
	if (!retval && (irqflags & IRQF_SHARED)) {
D
David Woodhouse 已提交
1227 1228 1229
		/*
		 * It's a shared IRQ -- the driver ought to be prepared for it
		 * to happen immediately, so let's make sure....
1230 1231
		 * We disable the irq to make sure that a 'real' IRQ doesn't
		 * run in parallel with our fake.
D
David Woodhouse 已提交
1232
		 */
1233
		unsigned long flags;
D
David Woodhouse 已提交
1234

1235
		disable_irq(irq);
1236
		local_irq_save(flags);
1237

1238
		handler(irq, dev_id);
1239

1240
		local_irq_restore(flags);
1241
		enable_irq(irq);
D
David Woodhouse 已提交
1242 1243
	}
#endif
L
Linus Torvalds 已提交
1244 1245
	return retval;
}
1246
EXPORT_SYMBOL(request_threaded_irq);
1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273

/**
 *	request_any_context_irq - allocate an interrupt line
 *	@irq: Interrupt line to allocate
 *	@handler: Function to be called when the IRQ occurs.
 *		  Threaded handler for threaded interrupts.
 *	@flags: Interrupt type flags
 *	@name: An ascii name for the claiming device
 *	@dev_id: A cookie passed back to the handler function
 *
 *	This call allocates interrupt resources and enables the
 *	interrupt line and IRQ handling. It selects either a
 *	hardirq or threaded handling method depending on the
 *	context.
 *
 *	On failure, it returns a negative value. On success,
 *	it returns either IRQC_IS_HARDIRQ or IRQC_IS_NESTED.
 */
int request_any_context_irq(unsigned int irq, irq_handler_t handler,
			    unsigned long flags, const char *name, void *dev_id)
{
	struct irq_desc *desc = irq_to_desc(irq);
	int ret;

	if (!desc)
		return -EINVAL;

1274
	if (irq_settings_is_nested_thread(desc)) {
1275 1276 1277 1278 1279 1280 1281 1282 1283
		ret = request_threaded_irq(irq, NULL, handler,
					   flags, name, dev_id);
		return !ret ? IRQC_IS_NESTED : ret;
	}

	ret = request_irq(irq, handler, flags, name, dev_id);
	return !ret ? IRQC_IS_HARDIRQ : ret;
}
EXPORT_SYMBOL_GPL(request_any_context_irq);