events.c 37.0 KB
Newer Older
J
Jeremy Fitzhardinge 已提交
1 2 3 4 5 6 7
/*
 * Xen event channels
 *
 * Xen models interrupts with abstract event channels.  Because each
 * domain gets 1024 event channels, but NR_IRQ is not that large, we
 * must dynamically map irqs<->event channels.  The event channels
 * interface with the rest of the kernel by defining a xen interrupt
L
Lucas De Marchi 已提交
8
 * chip.  When an event is received, it is mapped to an irq and sent
J
Jeremy Fitzhardinge 已提交
9 10 11 12 13 14 15 16 17 18
 * through the normal interrupt processing path.
 *
 * There are four kinds of events which can be mapped to an event
 * channel:
 *
 * 1. Inter-domain notifications.  This includes all the virtual
 *    device events, since they're driven by front-ends in another domain
 *    (typically dom0).
 * 2. VIRQs, typically used for timers.  These are per-cpu events.
 * 3. IPIs.
19
 * 4. PIRQs - Hardware interrupts.
J
Jeremy Fitzhardinge 已提交
20 21 22 23 24 25 26 27 28
 *
 * Jeremy Fitzhardinge <jeremy@xensource.com>, XenSource Inc, 2007
 */

#include <linux/linkage.h>
#include <linux/interrupt.h>
#include <linux/irq.h>
#include <linux/module.h>
#include <linux/string.h>
29
#include <linux/bootmem.h>
30
#include <linux/slab.h>
31
#include <linux/irqnr.h>
32
#include <linux/pci.h>
J
Jeremy Fitzhardinge 已提交
33

34
#include <asm/desc.h>
J
Jeremy Fitzhardinge 已提交
35 36
#include <asm/ptrace.h>
#include <asm/irq.h>
37
#include <asm/idle.h>
38
#include <asm/io_apic.h>
J
Jeremy Fitzhardinge 已提交
39
#include <asm/sync_bitops.h>
40
#include <asm/xen/pci.h>
J
Jeremy Fitzhardinge 已提交
41
#include <asm/xen/hypercall.h>
42
#include <asm/xen/hypervisor.h>
J
Jeremy Fitzhardinge 已提交
43

44 45
#include <xen/xen.h>
#include <xen/hvm.h>
46
#include <xen/xen-ops.h>
J
Jeremy Fitzhardinge 已提交
47 48 49
#include <xen/events.h>
#include <xen/interface/xen.h>
#include <xen/interface/event_channel.h>
50 51
#include <xen/interface/hvm/hvm_op.h>
#include <xen/interface/hvm/params.h>
J
Jeremy Fitzhardinge 已提交
52 53 54 55 56 57 58

/*
 * This lock protects updates to the following mapping and reference-count
 * arrays. The lock does not need to be acquired to read the mapping tables.
 */
static DEFINE_SPINLOCK(irq_mapping_update_lock);

59 60
static LIST_HEAD(xen_irq_list_head);

J
Jeremy Fitzhardinge 已提交
61
/* IRQ <-> VIRQ mapping. */
62
static DEFINE_PER_CPU(int [NR_VIRQS], virq_to_irq) = {[0 ... NR_VIRQS-1] = -1};
J
Jeremy Fitzhardinge 已提交
63

J
Jeremy Fitzhardinge 已提交
64
/* IRQ <-> IPI mapping */
65
static DEFINE_PER_CPU(int [XEN_NR_IPIS], ipi_to_irq) = {[0 ... XEN_NR_IPIS-1] = -1};
J
Jeremy Fitzhardinge 已提交
66

67 68
/* Interrupt types. */
enum xen_irq_type {
J
Jeremy Fitzhardinge 已提交
69
	IRQT_UNBOUND = 0,
J
Jeremy Fitzhardinge 已提交
70 71 72 73 74
	IRQT_PIRQ,
	IRQT_VIRQ,
	IRQT_IPI,
	IRQT_EVTCHN
};
J
Jeremy Fitzhardinge 已提交
75

76 77 78 79 80 81
/*
 * Packed IRQ information:
 * type - enum xen_irq_type
 * event channel - irq->event channel mapping
 * cpu - cpu this event channel is bound to
 * index - type-specific information:
82 83
 *    PIRQ - vector, with MSB being "needs EIO", or physical IRQ of the HVM
 *           guest, or GSI (real passthrough IRQ) of the device.
84 85 86 87 88 89
 *    VIRQ - virq number
 *    IPI - IPI vector
 *    EVTCHN -
 */
struct irq_info
{
90
	struct list_head list;
91
	enum xen_irq_type type;	/* type */
92
	unsigned irq;
93 94 95 96 97 98 99
	unsigned short evtchn;	/* event channel */
	unsigned short cpu;	/* cpu bound */

	union {
		unsigned short virq;
		enum ipi_vector ipi;
		struct {
S
Stefano Stabellini 已提交
100
			unsigned short pirq;
101
			unsigned short gsi;
102 103
			unsigned char vector;
			unsigned char flags;
104 105 106
		} pirq;
	} u;
};
107
#define PIRQ_NEEDS_EOI	(1 << 0)
108
#define PIRQ_SHAREABLE	(1 << 1)
109

110
static int *evtchn_to_irq;
111

112 113
static DEFINE_PER_CPU(unsigned long [NR_EVENT_CHANNELS/BITS_PER_LONG],
		      cpu_evtchn_mask);
J
Jeremy Fitzhardinge 已提交
114 115 116 117 118

/* Xen will never allocate port zero for any purpose. */
#define VALID_EVTCHN(chn)	((chn) != 0)

static struct irq_chip xen_dynamic_chip;
119
static struct irq_chip xen_percpu_chip;
120
static struct irq_chip xen_pirq_chip;
J
Jeremy Fitzhardinge 已提交
121

122 123
/* Get info for IRQ */
static struct irq_info *info_for_irq(unsigned irq)
124
{
T
Thomas Gleixner 已提交
125
	return irq_get_handler_data(irq);
126 127
}

128 129
/* Constructors for packed IRQ information. */
static void xen_irq_info_common_init(struct irq_info *info,
130
				     unsigned irq,
131 132 133
				     enum xen_irq_type type,
				     unsigned short evtchn,
				     unsigned short cpu)
134
{
135 136 137 138

	BUG_ON(info->type != IRQT_UNBOUND && info->type != type);

	info->type = type;
139
	info->irq = irq;
140 141
	info->evtchn = evtchn;
	info->cpu = cpu;
142 143

	evtchn_to_irq[evtchn] = irq;
144 145
}

146 147
static void xen_irq_info_evtchn_init(unsigned irq,
				     unsigned short evtchn)
148
{
149 150
	struct irq_info *info = info_for_irq(irq);

151
	xen_irq_info_common_init(info, irq, IRQT_EVTCHN, evtchn, 0);
152 153
}

154 155
static void xen_irq_info_ipi_init(unsigned cpu,
				  unsigned irq,
156 157
				  unsigned short evtchn,
				  enum ipi_vector ipi)
J
Jeremy Fitzhardinge 已提交
158
{
159 160
	struct irq_info *info = info_for_irq(irq);

161
	xen_irq_info_common_init(info, irq, IRQT_IPI, evtchn, 0);
162 163

	info->u.ipi = ipi;
164 165

	per_cpu(ipi_to_irq, cpu)[ipi] = irq;
166 167
}

168 169
static void xen_irq_info_virq_init(unsigned cpu,
				   unsigned irq,
170 171
				   unsigned short evtchn,
				   unsigned short virq)
172
{
173 174
	struct irq_info *info = info_for_irq(irq);

175
	xen_irq_info_common_init(info, irq, IRQT_VIRQ, evtchn, 0);
176 177

	info->u.virq = virq;
178 179

	per_cpu(virq_to_irq, cpu)[virq] = irq;
180 181
}

182 183 184 185 186 187
static void xen_irq_info_pirq_init(unsigned irq,
				   unsigned short evtchn,
				   unsigned short pirq,
				   unsigned short gsi,
				   unsigned short vector,
				   unsigned char flags)
188
{
189 190
	struct irq_info *info = info_for_irq(irq);

191
	xen_irq_info_common_init(info, irq, IRQT_PIRQ, evtchn, 0);
192 193 194 195 196

	info->u.pirq.pirq = pirq;
	info->u.pirq.gsi = gsi;
	info->u.pirq.vector = vector;
	info->u.pirq.flags = flags;
J
Jeremy Fitzhardinge 已提交
197 198 199 200 201
}

/*
 * Accessors for packed IRQ information.
 */
202
static unsigned int evtchn_from_irq(unsigned irq)
J
Jeremy Fitzhardinge 已提交
203
{
204 205 206
	if (unlikely(WARN(irq < 0 || irq >= nr_irqs, "Invalid irq %d!\n", irq)))
		return 0;

207
	return info_for_irq(irq)->evtchn;
J
Jeremy Fitzhardinge 已提交
208 209
}

I
Ian Campbell 已提交
210 211 212 213 214 215
unsigned irq_from_evtchn(unsigned int evtchn)
{
	return evtchn_to_irq[evtchn];
}
EXPORT_SYMBOL_GPL(irq_from_evtchn);

216
static enum ipi_vector ipi_from_irq(unsigned irq)
J
Jeremy Fitzhardinge 已提交
217
{
218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235
	struct irq_info *info = info_for_irq(irq);

	BUG_ON(info == NULL);
	BUG_ON(info->type != IRQT_IPI);

	return info->u.ipi;
}

static unsigned virq_from_irq(unsigned irq)
{
	struct irq_info *info = info_for_irq(irq);

	BUG_ON(info == NULL);
	BUG_ON(info->type != IRQT_VIRQ);

	return info->u.virq;
}

S
Stefano Stabellini 已提交
236 237 238 239 240 241 242 243 244 245
static unsigned pirq_from_irq(unsigned irq)
{
	struct irq_info *info = info_for_irq(irq);

	BUG_ON(info == NULL);
	BUG_ON(info->type != IRQT_PIRQ);

	return info->u.pirq.pirq;
}

246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264
static enum xen_irq_type type_from_irq(unsigned irq)
{
	return info_for_irq(irq)->type;
}

static unsigned cpu_from_irq(unsigned irq)
{
	return info_for_irq(irq)->cpu;
}

static unsigned int cpu_from_evtchn(unsigned int evtchn)
{
	int irq = evtchn_to_irq[evtchn];
	unsigned ret = 0;

	if (irq != -1)
		ret = cpu_from_irq(irq);

	return ret;
J
Jeremy Fitzhardinge 已提交
265 266
}

267 268 269 270 271 272 273 274 275
static bool pirq_needs_eoi(unsigned irq)
{
	struct irq_info *info = info_for_irq(irq);

	BUG_ON(info->type != IRQT_PIRQ);

	return info->u.pirq.flags & PIRQ_NEEDS_EOI;
}

J
Jeremy Fitzhardinge 已提交
276 277 278 279 280
static inline unsigned long active_evtchns(unsigned int cpu,
					   struct shared_info *sh,
					   unsigned int idx)
{
	return (sh->evtchn_pending[idx] &
281
		per_cpu(cpu_evtchn_mask, cpu)[idx] &
J
Jeremy Fitzhardinge 已提交
282 283 284 285 286 287 288 289 290
		~sh->evtchn_mask[idx]);
}

static void bind_evtchn_to_cpu(unsigned int chn, unsigned int cpu)
{
	int irq = evtchn_to_irq[chn];

	BUG_ON(irq == -1);
#ifdef CONFIG_SMP
291
	cpumask_copy(irq_to_desc(irq)->irq_data.affinity, cpumask_of(cpu));
J
Jeremy Fitzhardinge 已提交
292 293
#endif

294 295
	clear_bit(chn, per_cpu(cpu_evtchn_mask, cpu_from_irq(irq)));
	set_bit(chn, per_cpu(cpu_evtchn_mask, cpu));
J
Jeremy Fitzhardinge 已提交
296

297
	info_for_irq(irq)->cpu = cpu;
J
Jeremy Fitzhardinge 已提交
298 299 300 301
}

static void init_evtchn_cpu_bindings(void)
{
302
	int i;
J
Jeremy Fitzhardinge 已提交
303
#ifdef CONFIG_SMP
304
	struct irq_info *info;
305

J
Jeremy Fitzhardinge 已提交
306
	/* By default all event channels notify CPU#0. */
307 308
	list_for_each_entry(info, &xen_irq_list_head, list) {
		struct irq_desc *desc = irq_to_desc(info->irq);
309
		cpumask_copy(desc->irq_data.affinity, cpumask_of(0));
310
	}
J
Jeremy Fitzhardinge 已提交
311 312
#endif

313
	for_each_possible_cpu(i)
314 315
		memset(per_cpu(cpu_evtchn_mask, i),
		       (i == 0) ? ~0 : 0, sizeof(*per_cpu(cpu_evtchn_mask, i)));
J
Jeremy Fitzhardinge 已提交
316 317 318 319 320 321 322 323 324 325 326 327 328 329
}

static inline void clear_evtchn(int port)
{
	struct shared_info *s = HYPERVISOR_shared_info;
	sync_clear_bit(port, &s->evtchn_pending[0]);
}

static inline void set_evtchn(int port)
{
	struct shared_info *s = HYPERVISOR_shared_info;
	sync_set_bit(port, &s->evtchn_pending[0]);
}

330 331 332 333 334 335
static inline int test_evtchn(int port)
{
	struct shared_info *s = HYPERVISOR_shared_info;
	return sync_test_bit(port, &s->evtchn_pending[0]);
}

J
Jeremy Fitzhardinge 已提交
336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371

/**
 * notify_remote_via_irq - send event to remote end of event channel via irq
 * @irq: irq of event channel to send event to
 *
 * Unlike notify_remote_via_evtchn(), this is safe to use across
 * save/restore. Notifications on a broken connection are silently
 * dropped.
 */
void notify_remote_via_irq(int irq)
{
	int evtchn = evtchn_from_irq(irq);

	if (VALID_EVTCHN(evtchn))
		notify_remote_via_evtchn(evtchn);
}
EXPORT_SYMBOL_GPL(notify_remote_via_irq);

static void mask_evtchn(int port)
{
	struct shared_info *s = HYPERVISOR_shared_info;
	sync_set_bit(port, &s->evtchn_mask[0]);
}

static void unmask_evtchn(int port)
{
	struct shared_info *s = HYPERVISOR_shared_info;
	unsigned int cpu = get_cpu();

	BUG_ON(!irqs_disabled());

	/* Slow path (hypercall) if this is a non-local port. */
	if (unlikely(cpu != cpu_from_evtchn(port))) {
		struct evtchn_unmask unmask = { .port = port };
		(void)HYPERVISOR_event_channel_op(EVTCHNOP_unmask, &unmask);
	} else {
C
Christoph Lameter 已提交
372
		struct vcpu_info *vcpu_info = __this_cpu_read(xen_vcpu);
J
Jeremy Fitzhardinge 已提交
373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389

		sync_clear_bit(port, &s->evtchn_mask[0]);

		/*
		 * The following is basically the equivalent of
		 * 'hw_resend_irq'. Just like a real IO-APIC we 'lose
		 * the interrupt edge' if the channel is masked.
		 */
		if (sync_test_bit(port, &s->evtchn_pending[0]) &&
		    !sync_test_and_set_bit(port / BITS_PER_LONG,
					   &vcpu_info->evtchn_pending_sel))
			vcpu_info->evtchn_upcall_pending = 1;
	}

	put_cpu();
}

390 391 392 393 394
static void xen_irq_init(unsigned irq)
{
	struct irq_info *info;
	struct irq_desc *desc = irq_to_desc(irq);

395
#ifdef CONFIG_SMP
396 397
	/* By default all event channels notify CPU#0. */
	cpumask_copy(desc->irq_data.affinity, cpumask_of(0));
398
#endif
399

400 401 402
	info = kzalloc(sizeof(*info), GFP_KERNEL);
	if (info == NULL)
		panic("Unable to allocate metadata for IRQ%d\n", irq);
403 404 405

	info->type = IRQT_UNBOUND;

T
Thomas Gleixner 已提交
406
	irq_set_handler_data(irq, info);
407

408 409 410
	list_add_tail(&info->list, &xen_irq_list_head);
}

411
static int __must_check xen_allocate_irq_dynamic(void)
412
{
413 414
	int first = 0;
	int irq;
415 416

#ifdef CONFIG_X86_IO_APIC
417 418
	/*
	 * For an HVM guest or domain 0 which see "real" (emulated or
L
Lucas De Marchi 已提交
419
	 * actual respectively) GSIs we allocate dynamic IRQs
420 421 422 423 424 425
	 * e.g. those corresponding to event channels or MSIs
	 * etc. from the range above those "real" GSIs to avoid
	 * collisions.
	 */
	if (xen_initial_domain() || xen_hvm_domain())
		first = get_nr_irqs_gsi();
426 427
#endif

428
	irq = irq_alloc_desc_from(first, -1);
429

430
	xen_irq_init(irq);
431

J
Jeremy Fitzhardinge 已提交
432
	return irq;
433 434
}

435
static int __must_check xen_allocate_irq_gsi(unsigned gsi)
436 437 438
{
	int irq;

439 440 441 442 443 444 445
	/*
	 * A PV guest has no concept of a GSI (since it has no ACPI
	 * nor access to/knowledge of the physical APICs). Therefore
	 * all IRQs are dynamically allocated from the entire IRQ
	 * space.
	 */
	if (xen_pv_domain() && !xen_initial_domain())
446 447 448 449
		return xen_allocate_irq_dynamic();

	/* Legacy IRQ descriptors are already allocated by the arch. */
	if (gsi < NR_IRQS_LEGACY)
450 451 452
		irq = gsi;
	else
		irq = irq_alloc_desc_at(gsi, -1);
453

454
	xen_irq_init(irq);
455 456 457 458 459 460

	return irq;
}

static void xen_free_irq(unsigned irq)
{
T
Thomas Gleixner 已提交
461
	struct irq_info *info = irq_get_handler_data(irq);
462 463

	list_del(&info->list);
464

T
Thomas Gleixner 已提交
465
	irq_set_handler_data(irq, NULL);
466 467 468

	kfree(info);

469 470 471 472
	/* Legacy IRQ descriptors are managed by the arch. */
	if (irq < NR_IRQS_LEGACY)
		return;

473 474 475
	irq_free_desc(irq);
}

476 477
static void pirq_unmask_notify(int irq)
{
S
Stefano Stabellini 已提交
478
	struct physdev_eoi eoi = { .irq = pirq_from_irq(irq) };
479 480 481 482 483 484 485 486 487 488 489 490 491 492

	if (unlikely(pirq_needs_eoi(irq))) {
		int rc = HYPERVISOR_physdev_op(PHYSDEVOP_eoi, &eoi);
		WARN_ON(rc);
	}
}

static void pirq_query_unmask(int irq)
{
	struct physdev_irq_status_query irq_status;
	struct irq_info *info = info_for_irq(irq);

	BUG_ON(info->type != IRQT_PIRQ);

S
Stefano Stabellini 已提交
493
	irq_status.irq = pirq_from_irq(irq);
494 495 496 497 498 499 500 501 502 503 504 505 506 507 508
	if (HYPERVISOR_physdev_op(PHYSDEVOP_irq_status_query, &irq_status))
		irq_status.flags = 0;

	info->u.pirq.flags &= ~PIRQ_NEEDS_EOI;
	if (irq_status.flags & XENIRQSTAT_needs_eoi)
		info->u.pirq.flags |= PIRQ_NEEDS_EOI;
}

static bool probing_irq(int irq)
{
	struct irq_desc *desc = irq_to_desc(irq);

	return desc && desc->action == NULL;
}

509
static unsigned int __startup_pirq(unsigned int irq)
510 511 512 513
{
	struct evtchn_bind_pirq bind_pirq;
	struct irq_info *info = info_for_irq(irq);
	int evtchn = evtchn_from_irq(irq);
514
	int rc;
515 516 517 518 519 520

	BUG_ON(info->type != IRQT_PIRQ);

	if (VALID_EVTCHN(evtchn))
		goto out;

S
Stefano Stabellini 已提交
521
	bind_pirq.pirq = pirq_from_irq(irq);
522
	/* NB. We are happy to share unless we are probing. */
523 524 525 526
	bind_pirq.flags = info->u.pirq.flags & PIRQ_SHAREABLE ?
					BIND_PIRQ__WILL_SHARE : 0;
	rc = HYPERVISOR_event_channel_op(EVTCHNOP_bind_pirq, &bind_pirq);
	if (rc != 0) {
527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546
		if (!probing_irq(irq))
			printk(KERN_INFO "Failed to obtain physical IRQ %d\n",
			       irq);
		return 0;
	}
	evtchn = bind_pirq.port;

	pirq_query_unmask(irq);

	evtchn_to_irq[evtchn] = irq;
	bind_evtchn_to_cpu(evtchn, 0);
	info->evtchn = evtchn;

out:
	unmask_evtchn(evtchn);
	pirq_unmask_notify(irq);

	return 0;
}

547 548 549 550 551 552
static unsigned int startup_pirq(struct irq_data *data)
{
	return __startup_pirq(data->irq);
}

static void shutdown_pirq(struct irq_data *data)
553 554
{
	struct evtchn_close close;
555
	unsigned int irq = data->irq;
556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574
	struct irq_info *info = info_for_irq(irq);
	int evtchn = evtchn_from_irq(irq);

	BUG_ON(info->type != IRQT_PIRQ);

	if (!VALID_EVTCHN(evtchn))
		return;

	mask_evtchn(evtchn);

	close.port = evtchn;
	if (HYPERVISOR_event_channel_op(EVTCHNOP_close, &close) != 0)
		BUG();

	bind_evtchn_to_cpu(evtchn, 0);
	evtchn_to_irq[evtchn] = -1;
	info->evtchn = 0;
}

575
static void enable_pirq(struct irq_data *data)
576
{
577
	startup_pirq(data);
578 579
}

580
static void disable_pirq(struct irq_data *data)
581 582 583
{
}

584
static void ack_pirq(struct irq_data *data)
585
{
586
	int evtchn = evtchn_from_irq(data->irq);
587

588
	irq_move_irq(data);
589 590 591 592 593 594 595 596 597

	if (VALID_EVTCHN(evtchn)) {
		mask_evtchn(evtchn);
		clear_evtchn(evtchn);
	}
}

static int find_irq_by_gsi(unsigned gsi)
{
598
	struct irq_info *info;
599

600 601
	list_for_each_entry(info, &xen_irq_list_head, list) {
		if (info->type != IRQT_PIRQ)
602 603
			continue;

604 605
		if (info->u.pirq.gsi == gsi)
			return info->irq;
606 607 608 609 610
	}

	return -1;
}

611
int xen_allocate_pirq_gsi(unsigned gsi)
S
Stefano Stabellini 已提交
612
{
613
	return gsi;
S
Stefano Stabellini 已提交
614 615
}

I
Ian Campbell 已提交
616 617 618
/*
 * Do not make any assumptions regarding the relationship between the
 * IRQ number returned here and the Xen pirq argument.
S
Stefano Stabellini 已提交
619 620 621
 *
 * Note: We don't assign an event channel until the irq actually started
 * up.  Return an existing irq if we've already got one for the gsi.
622
 */
623 624
int xen_bind_pirq_gsi_to_irq(unsigned gsi,
			     unsigned pirq, int shareable, char *name)
625
{
626
	int irq = -1;
627 628 629 630 631 632
	struct physdev_irq irq_op;

	spin_lock(&irq_mapping_update_lock);

	irq = find_irq_by_gsi(gsi);
	if (irq != -1) {
S
Stefano Stabellini 已提交
633
		printk(KERN_INFO "xen_map_pirq_gsi: returning irq %d for gsi %u\n",
634 635 636 637
		       irq, gsi);
		goto out;	/* XXX need refcount? */
	}

638
	irq = xen_allocate_irq_gsi(gsi);
639 640
	if (irq < 0)
		goto out;
641

T
Thomas Gleixner 已提交
642 643
	irq_set_chip_and_handler_name(irq, &xen_pirq_chip, handle_level_irq,
				      name);
644 645

	irq_op.irq = irq;
646 647 648 649 650 651 652
	irq_op.vector = 0;

	/* Only the privileged domain can do this. For non-priv, the pcifront
	 * driver provides a PCI bus that does the call to do exactly
	 * this in the priv domain. */
	if (xen_initial_domain() &&
	    HYPERVISOR_physdev_op(PHYSDEVOP_alloc_irq_vector, &irq_op)) {
653
		xen_free_irq(irq);
654 655 656 657
		irq = -ENOSPC;
		goto out;
	}

658 659
	xen_irq_info_pirq_init(irq, 0, pirq, gsi, irq_op.vector,
			       shareable ? PIRQ_SHAREABLE : 0);
660 661 662 663 664 665 666

out:
	spin_unlock(&irq_mapping_update_lock);

	return irq;
}

667
#ifdef CONFIG_PCI_MSI
668
int xen_allocate_pirq_msi(struct pci_dev *dev, struct msi_desc *msidesc)
669
{
670
	int rc;
671 672
	struct physdev_get_free_pirq op_get_free_pirq;

673
	op_get_free_pirq.type = MAP_PIRQ_TYPE_MSI;
674 675
	rc = HYPERVISOR_physdev_op(PHYSDEVOP_get_free_pirq, &op_get_free_pirq);

676 677 678 679
	WARN_ONCE(rc == -ENOSYS,
		  "hypervisor does not support the PHYSDEVOP_get_free_pirq interface\n");

	return rc ? -1 : op_get_free_pirq.pirq;
680 681
}

682
int xen_bind_pirq_msi_to_irq(struct pci_dev *dev, struct msi_desc *msidesc,
683
			     int pirq, int vector, const char *name)
S
Stefano Stabellini 已提交
684
{
685
	int irq, ret;
686

S
Stefano Stabellini 已提交
687 688
	spin_lock(&irq_mapping_update_lock);

689 690
	irq = xen_allocate_irq_dynamic();
	if (irq == -1)
691
		goto out;
S
Stefano Stabellini 已提交
692

T
Thomas Gleixner 已提交
693 694
	irq_set_chip_and_handler_name(irq, &xen_pirq_chip, handle_level_irq,
				      name);
S
Stefano Stabellini 已提交
695

696
	xen_irq_info_pirq_init(irq, 0, pirq, 0, vector, 0);
697
	ret = irq_set_msi_desc(irq, msidesc);
698 699
	if (ret < 0)
		goto error_irq;
S
Stefano Stabellini 已提交
700 701
out:
	spin_unlock(&irq_mapping_update_lock);
702
	return irq;
703 704 705 706
error_irq:
	spin_unlock(&irq_mapping_update_lock);
	xen_free_irq(irq);
	return -1;
S
Stefano Stabellini 已提交
707
}
708 709
#endif

710 711 712
int xen_destroy_irq(int irq)
{
	struct irq_desc *desc;
713 714
	struct physdev_unmap_pirq unmap_irq;
	struct irq_info *info = info_for_irq(irq);
715 716 717 718 719 720 721 722
	int rc = -ENOENT;

	spin_lock(&irq_mapping_update_lock);

	desc = irq_to_desc(irq);
	if (!desc)
		goto out;

723
	if (xen_initial_domain()) {
724
		unmap_irq.pirq = info->u.pirq.pirq;
725 726 727 728 729 730 731
		unmap_irq.domid = DOMID_SELF;
		rc = HYPERVISOR_physdev_op(PHYSDEVOP_unmap_pirq, &unmap_irq);
		if (rc) {
			printk(KERN_WARNING "unmap irq failed %d\n", rc);
			goto out;
		}
	}
732

733
	xen_free_irq(irq);
734 735 736 737 738 739

out:
	spin_unlock(&irq_mapping_update_lock);
	return rc;
}

740
int xen_irq_from_pirq(unsigned pirq)
741
{
742
	int irq;
743

744
	struct irq_info *info;
J
Jeremy Fitzhardinge 已提交
745

746 747 748 749 750 751 752 753 754 755 756
	spin_lock(&irq_mapping_update_lock);

	list_for_each_entry(info, &xen_irq_list_head, list) {
		if (info == NULL || info->type != IRQT_PIRQ)
			continue;
		irq = info->irq;
		if (info->u.pirq.pirq == pirq)
			goto out;
	}
	irq = -1;
out:
757
	spin_unlock(&irq_mapping_update_lock);
758 759

	return irq;
760 761
}

762
int bind_evtchn_to_irq(unsigned int evtchn)
J
Jeremy Fitzhardinge 已提交
763 764 765 766 767 768 769 770
{
	int irq;

	spin_lock(&irq_mapping_update_lock);

	irq = evtchn_to_irq[evtchn];

	if (irq == -1) {
771
		irq = xen_allocate_irq_dynamic();
772 773
		if (irq == -1)
			goto out;
J
Jeremy Fitzhardinge 已提交
774

T
Thomas Gleixner 已提交
775
		irq_set_chip_and_handler_name(irq, &xen_dynamic_chip,
776
					      handle_fasteoi_irq, "event");
J
Jeremy Fitzhardinge 已提交
777

778
		xen_irq_info_evtchn_init(irq, evtchn);
J
Jeremy Fitzhardinge 已提交
779 780
	}

781
out:
J
Jeremy Fitzhardinge 已提交
782 783 784 785
	spin_unlock(&irq_mapping_update_lock);

	return irq;
}
786
EXPORT_SYMBOL_GPL(bind_evtchn_to_irq);
J
Jeremy Fitzhardinge 已提交
787

J
Jeremy Fitzhardinge 已提交
788 789 790 791 792 793 794 795
static int bind_ipi_to_irq(unsigned int ipi, unsigned int cpu)
{
	struct evtchn_bind_ipi bind_ipi;
	int evtchn, irq;

	spin_lock(&irq_mapping_update_lock);

	irq = per_cpu(ipi_to_irq, cpu)[ipi];
796

J
Jeremy Fitzhardinge 已提交
797
	if (irq == -1) {
798
		irq = xen_allocate_irq_dynamic();
J
Jeremy Fitzhardinge 已提交
799 800 801
		if (irq < 0)
			goto out;

T
Thomas Gleixner 已提交
802
		irq_set_chip_and_handler_name(irq, &xen_percpu_chip,
803
					      handle_percpu_irq, "ipi");
J
Jeremy Fitzhardinge 已提交
804 805 806 807 808 809 810

		bind_ipi.vcpu = cpu;
		if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_ipi,
						&bind_ipi) != 0)
			BUG();
		evtchn = bind_ipi.port;

811
		xen_irq_info_ipi_init(cpu, irq, evtchn, ipi);
J
Jeremy Fitzhardinge 已提交
812 813 814 815 816 817 818 819 820

		bind_evtchn_to_cpu(evtchn, cpu);
	}

 out:
	spin_unlock(&irq_mapping_update_lock);
	return irq;
}

821 822 823 824 825 826 827 828 829 830 831 832 833 834 835
static int bind_interdomain_evtchn_to_irq(unsigned int remote_domain,
					  unsigned int remote_port)
{
	struct evtchn_bind_interdomain bind_interdomain;
	int err;

	bind_interdomain.remote_dom  = remote_domain;
	bind_interdomain.remote_port = remote_port;

	err = HYPERVISOR_event_channel_op(EVTCHNOP_bind_interdomain,
					  &bind_interdomain);

	return err ? : bind_evtchn_to_irq(bind_interdomain.local_port);
}

J
Jeremy Fitzhardinge 已提交
836

837
int bind_virq_to_irq(unsigned int virq, unsigned int cpu)
J
Jeremy Fitzhardinge 已提交
838 839 840 841 842 843 844 845 846
{
	struct evtchn_bind_virq bind_virq;
	int evtchn, irq;

	spin_lock(&irq_mapping_update_lock);

	irq = per_cpu(virq_to_irq, cpu)[virq];

	if (irq == -1) {
847
		irq = xen_allocate_irq_dynamic();
848 849
		if (irq == -1)
			goto out;
850

T
Thomas Gleixner 已提交
851
		irq_set_chip_and_handler_name(irq, &xen_percpu_chip,
852 853
					      handle_percpu_irq, "virq");

J
Jeremy Fitzhardinge 已提交
854 855 856 857 858 859 860
		bind_virq.virq = virq;
		bind_virq.vcpu = cpu;
		if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_virq,
						&bind_virq) != 0)
			BUG();
		evtchn = bind_virq.port;

861
		xen_irq_info_virq_init(cpu, irq, evtchn, virq);
J
Jeremy Fitzhardinge 已提交
862 863 864 865

		bind_evtchn_to_cpu(evtchn, cpu);
	}

866
out:
J
Jeremy Fitzhardinge 已提交
867 868 869 870 871 872 873 874 875 876 877 878
	spin_unlock(&irq_mapping_update_lock);

	return irq;
}

static void unbind_from_irq(unsigned int irq)
{
	struct evtchn_close close;
	int evtchn = evtchn_from_irq(irq);

	spin_lock(&irq_mapping_update_lock);

J
Jeremy Fitzhardinge 已提交
879
	if (VALID_EVTCHN(evtchn)) {
J
Jeremy Fitzhardinge 已提交
880 881 882 883 884 885 886
		close.port = evtchn;
		if (HYPERVISOR_event_channel_op(EVTCHNOP_close, &close) != 0)
			BUG();

		switch (type_from_irq(irq)) {
		case IRQT_VIRQ:
			per_cpu(virq_to_irq, cpu_from_evtchn(evtchn))
887
				[virq_from_irq(irq)] = -1;
J
Jeremy Fitzhardinge 已提交
888
			break;
A
Alex Nixon 已提交
889 890
		case IRQT_IPI:
			per_cpu(ipi_to_irq, cpu_from_evtchn(evtchn))
891
				[ipi_from_irq(irq)] = -1;
A
Alex Nixon 已提交
892
			break;
J
Jeremy Fitzhardinge 已提交
893 894 895 896 897 898 899 900
		default:
			break;
		}

		/* Closed ports are implicitly re-bound to VCPU0. */
		bind_evtchn_to_cpu(evtchn, 0);

		evtchn_to_irq[evtchn] = -1;
901 902
	}

903
	BUG_ON(info_for_irq(irq)->type == IRQT_UNBOUND);
J
Jeremy Fitzhardinge 已提交
904

905
	xen_free_irq(irq);
J
Jeremy Fitzhardinge 已提交
906 907 908 909 910

	spin_unlock(&irq_mapping_update_lock);
}

int bind_evtchn_to_irqhandler(unsigned int evtchn,
911
			      irq_handler_t handler,
J
Jeremy Fitzhardinge 已提交
912 913 914 915 916 917 918
			      unsigned long irqflags,
			      const char *devname, void *dev_id)
{
	unsigned int irq;
	int retval;

	irq = bind_evtchn_to_irq(evtchn);
919 920
	if (irq < 0)
		return irq;
J
Jeremy Fitzhardinge 已提交
921 922 923 924 925 926 927 928 929 930
	retval = request_irq(irq, handler, irqflags, devname, dev_id);
	if (retval != 0) {
		unbind_from_irq(irq);
		return retval;
	}

	return irq;
}
EXPORT_SYMBOL_GPL(bind_evtchn_to_irqhandler);

931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953
int bind_interdomain_evtchn_to_irqhandler(unsigned int remote_domain,
					  unsigned int remote_port,
					  irq_handler_t handler,
					  unsigned long irqflags,
					  const char *devname,
					  void *dev_id)
{
	int irq, retval;

	irq = bind_interdomain_evtchn_to_irq(remote_domain, remote_port);
	if (irq < 0)
		return irq;

	retval = request_irq(irq, handler, irqflags, devname, dev_id);
	if (retval != 0) {
		unbind_from_irq(irq);
		return retval;
	}

	return irq;
}
EXPORT_SYMBOL_GPL(bind_interdomain_evtchn_to_irqhandler);

J
Jeremy Fitzhardinge 已提交
954
int bind_virq_to_irqhandler(unsigned int virq, unsigned int cpu,
955
			    irq_handler_t handler,
J
Jeremy Fitzhardinge 已提交
956 957 958 959 960 961
			    unsigned long irqflags, const char *devname, void *dev_id)
{
	unsigned int irq;
	int retval;

	irq = bind_virq_to_irq(virq, cpu);
962 963
	if (irq < 0)
		return irq;
J
Jeremy Fitzhardinge 已提交
964 965 966 967 968 969 970 971 972 973
	retval = request_irq(irq, handler, irqflags, devname, dev_id);
	if (retval != 0) {
		unbind_from_irq(irq);
		return retval;
	}

	return irq;
}
EXPORT_SYMBOL_GPL(bind_virq_to_irqhandler);

J
Jeremy Fitzhardinge 已提交
974 975 976 977 978 979 980 981 982 983 984 985 986
int bind_ipi_to_irqhandler(enum ipi_vector ipi,
			   unsigned int cpu,
			   irq_handler_t handler,
			   unsigned long irqflags,
			   const char *devname,
			   void *dev_id)
{
	int irq, retval;

	irq = bind_ipi_to_irq(ipi, cpu);
	if (irq < 0)
		return irq;

T
Thomas Gleixner 已提交
987
	irqflags |= IRQF_NO_SUSPEND | IRQF_FORCE_RESUME;
J
Jeremy Fitzhardinge 已提交
988 989 990 991 992 993 994 995 996
	retval = request_irq(irq, handler, irqflags, devname, dev_id);
	if (retval != 0) {
		unbind_from_irq(irq);
		return retval;
	}

	return irq;
}

J
Jeremy Fitzhardinge 已提交
997 998 999 1000 1001 1002 1003
void unbind_from_irqhandler(unsigned int irq, void *dev_id)
{
	free_irq(irq, dev_id);
	unbind_from_irq(irq);
}
EXPORT_SYMBOL_GPL(unbind_from_irqhandler);

J
Jeremy Fitzhardinge 已提交
1004 1005 1006 1007 1008 1009 1010
void xen_send_IPI_one(unsigned int cpu, enum ipi_vector vector)
{
	int irq = per_cpu(ipi_to_irq, cpu)[vector];
	BUG_ON(irq < 0);
	notify_remote_via_irq(irq);
}

1011 1012 1013 1014
irqreturn_t xen_debug_interrupt(int irq, void *dev_id)
{
	struct shared_info *sh = HYPERVISOR_shared_info;
	int cpu = smp_processor_id();
1015
	unsigned long *cpu_evtchn = per_cpu(cpu_evtchn_mask, cpu);
1016 1017 1018
	int i;
	unsigned long flags;
	static DEFINE_SPINLOCK(debug_lock);
1019
	struct vcpu_info *v;
1020 1021 1022

	spin_lock_irqsave(&debug_lock, flags);

1023
	printk("\nvcpu %d\n  ", cpu);
1024 1025

	for_each_online_cpu(i) {
1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068
		int pending;
		v = per_cpu(xen_vcpu, i);
		pending = (get_irq_regs() && i == cpu)
			? xen_irqs_disabled(get_irq_regs())
			: v->evtchn_upcall_mask;
		printk("%d: masked=%d pending=%d event_sel %0*lx\n  ", i,
		       pending, v->evtchn_upcall_pending,
		       (int)(sizeof(v->evtchn_pending_sel)*2),
		       v->evtchn_pending_sel);
	}
	v = per_cpu(xen_vcpu, cpu);

	printk("\npending:\n   ");
	for (i = ARRAY_SIZE(sh->evtchn_pending)-1; i >= 0; i--)
		printk("%0*lx%s", (int)sizeof(sh->evtchn_pending[0])*2,
		       sh->evtchn_pending[i],
		       i % 8 == 0 ? "\n   " : " ");
	printk("\nglobal mask:\n   ");
	for (i = ARRAY_SIZE(sh->evtchn_mask)-1; i >= 0; i--)
		printk("%0*lx%s",
		       (int)(sizeof(sh->evtchn_mask[0])*2),
		       sh->evtchn_mask[i],
		       i % 8 == 0 ? "\n   " : " ");

	printk("\nglobally unmasked:\n   ");
	for (i = ARRAY_SIZE(sh->evtchn_mask)-1; i >= 0; i--)
		printk("%0*lx%s", (int)(sizeof(sh->evtchn_mask[0])*2),
		       sh->evtchn_pending[i] & ~sh->evtchn_mask[i],
		       i % 8 == 0 ? "\n   " : " ");

	printk("\nlocal cpu%d mask:\n   ", cpu);
	for (i = (NR_EVENT_CHANNELS/BITS_PER_LONG)-1; i >= 0; i--)
		printk("%0*lx%s", (int)(sizeof(cpu_evtchn[0])*2),
		       cpu_evtchn[i],
		       i % 8 == 0 ? "\n   " : " ");

	printk("\nlocally unmasked:\n   ");
	for (i = ARRAY_SIZE(sh->evtchn_mask)-1; i >= 0; i--) {
		unsigned long pending = sh->evtchn_pending[i]
			& ~sh->evtchn_mask[i]
			& cpu_evtchn[i];
		printk("%0*lx%s", (int)(sizeof(sh->evtchn_mask[0])*2),
		       pending, i % 8 == 0 ? "\n   " : " ");
1069 1070 1071
	}

	printk("\npending list:\n");
1072
	for (i = 0; i < NR_EVENT_CHANNELS; i++) {
1073
		if (sync_test_bit(i, sh->evtchn_pending)) {
1074 1075
			int word_idx = i / BITS_PER_LONG;
			printk("  %d: event %d -> irq %d%s%s%s\n",
1076
			       cpu_from_evtchn(i), i,
1077 1078 1079 1080 1081 1082 1083
			       evtchn_to_irq[i],
			       sync_test_bit(word_idx, &v->evtchn_pending_sel)
					     ? "" : " l2-clear",
			       !sync_test_bit(i, sh->evtchn_mask)
					     ? "" : " globally-masked",
			       sync_test_bit(i, cpu_evtchn)
					     ? "" : " locally-masked");
1084 1085 1086 1087 1088 1089 1090 1091
		}
	}

	spin_unlock_irqrestore(&debug_lock, flags);

	return IRQ_HANDLED;
}

1092
static DEFINE_PER_CPU(unsigned, xed_nesting_count);
1093 1094
static DEFINE_PER_CPU(unsigned int, current_word_idx);
static DEFINE_PER_CPU(unsigned int, current_bit_idx);
1095

1096 1097 1098 1099
/*
 * Mask out the i least significant bits of w
 */
#define MASK_LSBS(w, i) (w & ((~0UL) << i))
1100

J
Jeremy Fitzhardinge 已提交
1101 1102 1103 1104 1105 1106 1107 1108 1109
/*
 * Search the CPUs pending events bitmasks.  For each one found, map
 * the event number to an irq, and feed it into do_IRQ() for
 * handling.
 *
 * Xen uses a two-level bitmap to speed searching.  The first level is
 * a bitset of words which contain pending event bits.  The second
 * level is a bitset of pending events themselves.
 */
1110
static void __xen_evtchn_do_upcall(void)
J
Jeremy Fitzhardinge 已提交
1111
{
1112
	int start_word_idx, start_bit_idx;
1113
	int word_idx, bit_idx;
1114
	int i;
J
Jeremy Fitzhardinge 已提交
1115 1116
	int cpu = get_cpu();
	struct shared_info *s = HYPERVISOR_shared_info;
C
Christoph Lameter 已提交
1117
	struct vcpu_info *vcpu_info = __this_cpu_read(xen_vcpu);
1118
 	unsigned count;
J
Jeremy Fitzhardinge 已提交
1119

1120 1121
	do {
		unsigned long pending_words;
J
Jeremy Fitzhardinge 已提交
1122

1123
		vcpu_info->evtchn_upcall_pending = 0;
J
Jeremy Fitzhardinge 已提交
1124

C
Christoph Lameter 已提交
1125
		if (__this_cpu_inc_return(xed_nesting_count) - 1)
1126
			goto out;
J
Jeremy Fitzhardinge 已提交
1127

1128 1129
#ifndef CONFIG_X86 /* No need for a barrier -- XCHG is a barrier on x86. */
		/* Clear master flag /before/ clearing selector flag. */
1130
		wmb();
1131
#endif
1132
		pending_words = xchg(&vcpu_info->evtchn_pending_sel, 0);
1133

1134 1135 1136 1137
		start_word_idx = __this_cpu_read(current_word_idx);
		start_bit_idx = __this_cpu_read(current_bit_idx);

		word_idx = start_word_idx;
1138

1139
		for (i = 0; pending_words != 0; i++) {
1140
			unsigned long pending_bits;
1141
			unsigned long words;
1142

1143 1144 1145
			words = MASK_LSBS(pending_words, word_idx);

			/*
1146
			 * If we masked out all events, wrap to beginning.
1147 1148
			 */
			if (words == 0) {
1149 1150
				word_idx = 0;
				bit_idx = 0;
1151 1152 1153
				continue;
			}
			word_idx = __ffs(words);
1154

1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166
			pending_bits = active_evtchns(cpu, s, word_idx);
			bit_idx = 0; /* usually scan entire word from start */
			if (word_idx == start_word_idx) {
				/* We scan the starting word in two parts */
				if (i == 0)
					/* 1st time: start in the middle */
					bit_idx = start_bit_idx;
				else
					/* 2nd time: mask bits done already */
					bit_idx &= (1UL << start_bit_idx) - 1;
			}

1167 1168 1169
			do {
				unsigned long bits;
				int port, irq;
1170
				struct irq_desc *desc;
1171

1172 1173 1174
				bits = MASK_LSBS(pending_bits, bit_idx);

				/* If we masked out all events, move on. */
1175
				if (bits == 0)
1176 1177 1178 1179 1180 1181 1182 1183
					break;

				bit_idx = __ffs(bits);

				/* Process port. */
				port = (word_idx * BITS_PER_LONG) + bit_idx;
				irq = evtchn_to_irq[port];

1184 1185 1186
				mask_evtchn(port);
				clear_evtchn(port);

1187 1188 1189 1190 1191
				if (irq != -1) {
					desc = irq_to_desc(irq);
					if (desc)
						generic_handle_irq_desc(irq, desc);
				}
1192

1193 1194 1195 1196 1197 1198 1199 1200
				bit_idx = (bit_idx + 1) % BITS_PER_LONG;

				/* Next caller starts at last processed + 1 */
				__this_cpu_write(current_word_idx,
						 bit_idx ? word_idx :
						 (word_idx+1) % BITS_PER_LONG);
				__this_cpu_write(current_bit_idx, bit_idx);
			} while (bit_idx != 0);
1201

1202 1203
			/* Scan start_l1i twice; all others once. */
			if ((word_idx != start_word_idx) || (i != 0))
1204
				pending_words &= ~(1UL << word_idx);
1205 1206

			word_idx = (word_idx + 1) % BITS_PER_LONG;
J
Jeremy Fitzhardinge 已提交
1207 1208
		}

1209 1210
		BUG_ON(!irqs_disabled());

C
Christoph Lameter 已提交
1211 1212
		count = __this_cpu_read(xed_nesting_count);
		__this_cpu_write(xed_nesting_count, 0);
1213
	} while (count != 1 || vcpu_info->evtchn_upcall_pending);
1214 1215

out:
1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228

	put_cpu();
}

void xen_evtchn_do_upcall(struct pt_regs *regs)
{
	struct pt_regs *old_regs = set_irq_regs(regs);

	exit_idle();
	irq_enter();

	__xen_evtchn_do_upcall();

1229 1230
	irq_exit();
	set_irq_regs(old_regs);
1231
}
1232

1233 1234 1235
void xen_hvm_evtchn_do_upcall(void)
{
	__xen_evtchn_do_upcall();
J
Jeremy Fitzhardinge 已提交
1236
}
1237
EXPORT_SYMBOL_GPL(xen_hvm_evtchn_do_upcall);
J
Jeremy Fitzhardinge 已提交
1238

J
Jeremy Fitzhardinge 已提交
1239 1240 1241
/* Rebind a new event channel to an existing irq. */
void rebind_evtchn_irq(int evtchn, int irq)
{
J
Jeremy Fitzhardinge 已提交
1242 1243
	struct irq_info *info = info_for_irq(irq);

J
Jeremy Fitzhardinge 已提交
1244 1245 1246 1247 1248 1249 1250 1251 1252
	/* Make sure the irq is masked, since the new event channel
	   will also be masked. */
	disable_irq(irq);

	spin_lock(&irq_mapping_update_lock);

	/* After resume the irq<->evtchn mappings are all cleared out */
	BUG_ON(evtchn_to_irq[evtchn] != -1);
	/* Expect irq to have been bound before,
J
Jeremy Fitzhardinge 已提交
1253 1254
	   so there should be a proper type */
	BUG_ON(info->type == IRQT_UNBOUND);
J
Jeremy Fitzhardinge 已提交
1255

1256
	xen_irq_info_evtchn_init(irq, evtchn);
J
Jeremy Fitzhardinge 已提交
1257 1258 1259 1260

	spin_unlock(&irq_mapping_update_lock);

	/* new event channels are always bound to cpu 0 */
1261
	irq_set_affinity(irq, cpumask_of(0));
J
Jeremy Fitzhardinge 已提交
1262 1263 1264 1265 1266

	/* Unmask the event channel. */
	enable_irq(irq);
}

J
Jeremy Fitzhardinge 已提交
1267
/* Rebind an evtchn so that it gets delivered to a specific cpu */
1268
static int rebind_irq_to_cpu(unsigned irq, unsigned tcpu)
J
Jeremy Fitzhardinge 已提交
1269 1270 1271 1272
{
	struct evtchn_bind_vcpu bind_vcpu;
	int evtchn = evtchn_from_irq(irq);

1273 1274 1275 1276 1277 1278 1279 1280
	if (!VALID_EVTCHN(evtchn))
		return -1;

	/*
	 * Events delivered via platform PCI interrupts are always
	 * routed to vcpu 0 and hence cannot be rebound.
	 */
	if (xen_hvm_domain() && !xen_have_vector_callback)
1281
		return -1;
J
Jeremy Fitzhardinge 已提交
1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294

	/* Send future instances of this interrupt to other vcpu. */
	bind_vcpu.port = evtchn;
	bind_vcpu.vcpu = tcpu;

	/*
	 * If this fails, it usually just indicates that we're dealing with a
	 * virq or IPI channel, which don't actually need to be rebound. Ignore
	 * it, but don't do the xenlinux-level rebind in that case.
	 */
	if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_vcpu, &bind_vcpu) >= 0)
		bind_evtchn_to_cpu(evtchn, tcpu);

1295 1296
	return 0;
}
J
Jeremy Fitzhardinge 已提交
1297

1298 1299
static int set_affinity_irq(struct irq_data *data, const struct cpumask *dest,
			    bool force)
J
Jeremy Fitzhardinge 已提交
1300
{
1301
	unsigned tcpu = cpumask_first(dest);
1302

1303
	return rebind_irq_to_cpu(data->irq, tcpu);
J
Jeremy Fitzhardinge 已提交
1304 1305
}

1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321
int resend_irq_on_evtchn(unsigned int irq)
{
	int masked, evtchn = evtchn_from_irq(irq);
	struct shared_info *s = HYPERVISOR_shared_info;

	if (!VALID_EVTCHN(evtchn))
		return 1;

	masked = sync_test_and_set_bit(evtchn, s->evtchn_mask);
	sync_set_bit(evtchn, s->evtchn_pending);
	if (!masked)
		unmask_evtchn(evtchn);

	return 1;
}

1322
static void enable_dynirq(struct irq_data *data)
J
Jeremy Fitzhardinge 已提交
1323
{
1324
	int evtchn = evtchn_from_irq(data->irq);
J
Jeremy Fitzhardinge 已提交
1325 1326 1327 1328 1329

	if (VALID_EVTCHN(evtchn))
		unmask_evtchn(evtchn);
}

1330
static void disable_dynirq(struct irq_data *data)
J
Jeremy Fitzhardinge 已提交
1331
{
1332
	int evtchn = evtchn_from_irq(data->irq);
J
Jeremy Fitzhardinge 已提交
1333 1334 1335 1336 1337

	if (VALID_EVTCHN(evtchn))
		mask_evtchn(evtchn);
}

1338
static void ack_dynirq(struct irq_data *data)
J
Jeremy Fitzhardinge 已提交
1339
{
1340
	int evtchn = evtchn_from_irq(data->irq);
J
Jeremy Fitzhardinge 已提交
1341

1342
	irq_move_masked_irq(data);
J
Jeremy Fitzhardinge 已提交
1343 1344

	if (VALID_EVTCHN(evtchn))
1345
		unmask_evtchn(evtchn);
J
Jeremy Fitzhardinge 已提交
1346 1347
}

1348
static int retrigger_dynirq(struct irq_data *data)
J
Jeremy Fitzhardinge 已提交
1349
{
1350
	int evtchn = evtchn_from_irq(data->irq);
1351
	struct shared_info *sh = HYPERVISOR_shared_info;
J
Jeremy Fitzhardinge 已提交
1352 1353 1354
	int ret = 0;

	if (VALID_EVTCHN(evtchn)) {
1355 1356 1357 1358 1359 1360
		int masked;

		masked = sync_test_and_set_bit(evtchn, sh->evtchn_mask);
		sync_set_bit(evtchn, sh->evtchn_pending);
		if (!masked)
			unmask_evtchn(evtchn);
J
Jeremy Fitzhardinge 已提交
1361 1362 1363 1364 1365 1366
		ret = 1;
	}

	return ret;
}

1367
static void restore_pirqs(void)
1368 1369 1370
{
	int pirq, rc, irq, gsi;
	struct physdev_map_pirq map_irq;
1371
	struct irq_info *info;
1372

1373 1374
	list_for_each_entry(info, &xen_irq_list_head, list) {
		if (info->type != IRQT_PIRQ)
1375 1376
			continue;

1377 1378 1379 1380
		pirq = info->u.pirq.pirq;
		gsi = info->u.pirq.gsi;
		irq = info->irq;

1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394
		/* save/restore of PT devices doesn't work, so at this point the
		 * only devices present are GSI based emulated devices */
		if (!gsi)
			continue;

		map_irq.domid = DOMID_SELF;
		map_irq.type = MAP_PIRQ_TYPE_GSI;
		map_irq.index = gsi;
		map_irq.pirq = pirq;

		rc = HYPERVISOR_physdev_op(PHYSDEVOP_map_pirq, &map_irq);
		if (rc) {
			printk(KERN_WARNING "xen map irq failed gsi=%d irq=%d pirq=%d rc=%d\n",
					gsi, irq, pirq, rc);
1395
			xen_free_irq(irq);
1396 1397 1398 1399 1400
			continue;
		}

		printk(KERN_DEBUG "xen: --> irq=%d, pirq=%d\n", irq, map_irq.pirq);

1401
		__startup_pirq(irq);
1402 1403 1404
	}
}

1405 1406 1407 1408 1409 1410 1411 1412 1413
static void restore_cpu_virqs(unsigned int cpu)
{
	struct evtchn_bind_virq bind_virq;
	int virq, irq, evtchn;

	for (virq = 0; virq < NR_VIRQS; virq++) {
		if ((irq = per_cpu(virq_to_irq, cpu)[virq]) == -1)
			continue;

1414
		BUG_ON(virq_from_irq(irq) != virq);
1415 1416 1417 1418 1419 1420 1421 1422 1423 1424

		/* Get a new binding from Xen. */
		bind_virq.virq = virq;
		bind_virq.vcpu = cpu;
		if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_virq,
						&bind_virq) != 0)
			BUG();
		evtchn = bind_virq.port;

		/* Record the new mapping. */
1425
		xen_irq_info_virq_init(cpu, irq, evtchn, virq);
1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438
		bind_evtchn_to_cpu(evtchn, cpu);
	}
}

static void restore_cpu_ipis(unsigned int cpu)
{
	struct evtchn_bind_ipi bind_ipi;
	int ipi, irq, evtchn;

	for (ipi = 0; ipi < XEN_NR_IPIS; ipi++) {
		if ((irq = per_cpu(ipi_to_irq, cpu)[ipi]) == -1)
			continue;

1439
		BUG_ON(ipi_from_irq(irq) != ipi);
1440 1441 1442 1443 1444 1445 1446 1447 1448

		/* Get a new binding from Xen. */
		bind_ipi.vcpu = cpu;
		if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_ipi,
						&bind_ipi) != 0)
			BUG();
		evtchn = bind_ipi.port;

		/* Record the new mapping. */
1449
		xen_irq_info_ipi_init(cpu, irq, evtchn, ipi);
1450 1451 1452 1453
		bind_evtchn_to_cpu(evtchn, cpu);
	}
}

1454 1455 1456 1457 1458 1459 1460 1461
/* Clear an irq's pending state, in preparation for polling on it */
void xen_clear_irq_pending(int irq)
{
	int evtchn = evtchn_from_irq(irq);

	if (VALID_EVTCHN(evtchn))
		clear_evtchn(evtchn);
}
1462
EXPORT_SYMBOL(xen_clear_irq_pending);
1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481
void xen_set_irq_pending(int irq)
{
	int evtchn = evtchn_from_irq(irq);

	if (VALID_EVTCHN(evtchn))
		set_evtchn(evtchn);
}

bool xen_test_irq_pending(int irq)
{
	int evtchn = evtchn_from_irq(irq);
	bool ret = false;

	if (VALID_EVTCHN(evtchn))
		ret = test_evtchn(evtchn);

	return ret;
}

1482 1483 1484
/* Poll waiting for an irq to become pending with timeout.  In the usual case,
 * the irq will be disabled so it won't deliver an interrupt. */
void xen_poll_irq_timeout(int irq, u64 timeout)
1485 1486 1487 1488 1489 1490 1491
{
	evtchn_port_t evtchn = evtchn_from_irq(irq);

	if (VALID_EVTCHN(evtchn)) {
		struct sched_poll poll;

		poll.nr_ports = 1;
1492
		poll.timeout = timeout;
1493
		set_xen_guest_handle(poll.ports, &evtchn);
1494 1495 1496 1497 1498

		if (HYPERVISOR_sched_op(SCHEDOP_poll, &poll) != 0)
			BUG();
	}
}
1499 1500 1501 1502 1503 1504 1505
EXPORT_SYMBOL(xen_poll_irq_timeout);
/* Poll waiting for an irq to become pending.  In the usual case, the
 * irq will be disabled so it won't deliver an interrupt. */
void xen_poll_irq(int irq)
{
	xen_poll_irq_timeout(irq, 0 /* no timeout */);
}
1506

1507 1508
void xen_irq_resume(void)
{
1509 1510
	unsigned int cpu, evtchn;
	struct irq_info *info;
1511 1512 1513 1514 1515 1516 1517 1518

	init_evtchn_cpu_bindings();

	/* New event-channel space is not 'live' yet. */
	for (evtchn = 0; evtchn < NR_EVENT_CHANNELS; evtchn++)
		mask_evtchn(evtchn);

	/* No IRQ <-> event-channel mappings. */
1519 1520
	list_for_each_entry(info, &xen_irq_list_head, list)
		info->evtchn = 0; /* zap event-channel binding */
1521 1522 1523 1524 1525 1526 1527 1528

	for (evtchn = 0; evtchn < NR_EVENT_CHANNELS; evtchn++)
		evtchn_to_irq[evtchn] = -1;

	for_each_possible_cpu(cpu) {
		restore_cpu_virqs(cpu);
		restore_cpu_ipis(cpu);
	}
1529

1530
	restore_pirqs();
1531 1532
}

J
Jeremy Fitzhardinge 已提交
1533
static struct irq_chip xen_dynamic_chip __read_mostly = {
1534
	.name			= "xen-dyn",
J
Jeremy Fitzhardinge 已提交
1535

1536 1537 1538
	.irq_disable		= disable_dynirq,
	.irq_mask		= disable_dynirq,
	.irq_unmask		= enable_dynirq,
J
Jeremy Fitzhardinge 已提交
1539

1540 1541 1542
	.irq_eoi		= ack_dynirq,
	.irq_set_affinity	= set_affinity_irq,
	.irq_retrigger		= retrigger_dynirq,
J
Jeremy Fitzhardinge 已提交
1543 1544
};

1545
static struct irq_chip xen_pirq_chip __read_mostly = {
1546
	.name			= "xen-pirq",
1547

1548 1549
	.irq_startup		= startup_pirq,
	.irq_shutdown		= shutdown_pirq,
1550

1551 1552
	.irq_enable		= enable_pirq,
	.irq_unmask		= enable_pirq,
1553

1554 1555
	.irq_disable		= disable_pirq,
	.irq_mask		= disable_pirq,
1556

1557
	.irq_ack		= ack_pirq,
1558

1559
	.irq_set_affinity	= set_affinity_irq,
1560

1561
	.irq_retrigger		= retrigger_dynirq,
1562 1563
};

1564
static struct irq_chip xen_percpu_chip __read_mostly = {
1565
	.name			= "xen-percpu",
1566

1567 1568 1569
	.irq_disable		= disable_dynirq,
	.irq_mask		= disable_dynirq,
	.irq_unmask		= enable_dynirq,
1570

1571
	.irq_ack		= ack_dynirq,
1572 1573
};

1574 1575 1576 1577 1578 1579 1580 1581 1582 1583
int xen_set_callback_via(uint64_t via)
{
	struct xen_hvm_param a;
	a.domid = DOMID_SELF;
	a.index = HVM_PARAM_CALLBACK_IRQ;
	a.value = via;
	return HYPERVISOR_hvm_op(HVMOP_set_param, &a);
}
EXPORT_SYMBOL_GPL(xen_set_callback_via);

1584
#ifdef CONFIG_XEN_PVHVM
1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607
/* Vector callbacks are better than PCI interrupts to receive event
 * channel notifications because we can receive vector callbacks on any
 * vcpu and we don't need PCI support or APIC interactions. */
void xen_callback_vector(void)
{
	int rc;
	uint64_t callback_via;
	if (xen_have_vector_callback) {
		callback_via = HVM_CALLBACK_VECTOR(XEN_HVM_EVTCHN_CALLBACK);
		rc = xen_set_callback_via(callback_via);
		if (rc) {
			printk(KERN_ERR "Request for Xen HVM callback vector"
					" failed.\n");
			xen_have_vector_callback = 0;
			return;
		}
		printk(KERN_INFO "Xen HVM callback vector for event delivery is "
				"enabled\n");
		/* in the restore case the vector has already been allocated */
		if (!test_bit(XEN_HVM_EVTCHN_CALLBACK, used_vectors))
			alloc_intr_gate(XEN_HVM_EVTCHN_CALLBACK, xen_hvm_callback_vector);
	}
}
1608 1609 1610
#else
void xen_callback_vector(void) {}
#endif
1611

J
Jeremy Fitzhardinge 已提交
1612 1613
void __init xen_init_IRQ(void)
{
1614
	int i;
1615

1616 1617 1618 1619
	evtchn_to_irq = kcalloc(NR_EVENT_CHANNELS, sizeof(*evtchn_to_irq),
				    GFP_KERNEL);
	for (i = 0; i < NR_EVENT_CHANNELS; i++)
		evtchn_to_irq[i] = -1;
J
Jeremy Fitzhardinge 已提交
1620 1621 1622 1623 1624 1625 1626

	init_evtchn_cpu_bindings();

	/* No event channels are 'live' right now. */
	for (i = 0; i < NR_EVENT_CHANNELS; i++)
		mask_evtchn(i);

1627 1628 1629
	if (xen_hvm_domain()) {
		xen_callback_vector();
		native_init_IRQ();
1630 1631 1632
		/* pci_xen_hvm_init must be called after native_init_IRQ so that
		 * __acpi_register_gsi can point at the right function */
		pci_xen_hvm_init();
1633 1634
	} else {
		irq_ctx_init(smp_processor_id());
1635 1636
		if (xen_initial_domain())
			xen_setup_pirqs();
1637
	}
J
Jeremy Fitzhardinge 已提交
1638
}