ioapic.c 19.6 KB
Newer Older
E
Eddie Dong 已提交
1 2
/*
 *  Copyright (C) 2001  MandrakeSoft S.A.
A
Avi Kivity 已提交
3
 *  Copyright 2010 Red Hat, Inc. and/or its affiliates.
E
Eddie Dong 已提交
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
 *
 *    MandrakeSoft S.A.
 *    43, rue d'Aboukir
 *    75002 Paris - France
 *    http://www.linux-mandrake.com/
 *    http://www.mandrakesoft.com/
 *
 *  This library is free software; you can redistribute it and/or
 *  modify it under the terms of the GNU Lesser General Public
 *  License as published by the Free Software Foundation; either
 *  version 2 of the License, or (at your option) any later version.
 *
 *  This library is distributed in the hope that it will be useful,
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 *  Lesser General Public License for more details.
 *
 *  You should have received a copy of the GNU Lesser General Public
 *  License along with this library; if not, write to the Free Software
 *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
 *
 *  Yunhong Jiang <yunhong.jiang@intel.com>
 *  Yaozu (Eddie) Dong <eddie.dong@intel.com>
 *  Based on Xen 3.1 code.
 */

30
#include <linux/kvm_host.h>
E
Eddie Dong 已提交
31 32 33 34 35 36
#include <linux/kvm.h>
#include <linux/mm.h>
#include <linux/highmem.h>
#include <linux/smp.h>
#include <linux/hrtimer.h>
#include <linux/io.h>
37
#include <linux/slab.h>
38
#include <linux/export.h>
39
#include <linux/nospec.h>
E
Eddie Dong 已提交
40 41 42
#include <asm/processor.h>
#include <asm/page.h>
#include <asm/current.h>
43
#include <trace/events/kvm.h>
44 45 46

#include "ioapic.h"
#include "lapic.h"
M
Marcelo Tosatti 已提交
47
#include "irq.h"
48

49
static int ioapic_service(struct kvm_ioapic *vioapic, int irq,
50
		bool line_status);
E
Eddie Dong 已提交
51

52 53 54 55 56
static void kvm_ioapic_update_eoi_one(struct kvm_vcpu *vcpu,
				      struct kvm_ioapic *ioapic,
				      int trigger_mode,
				      int pin);

E
Eddie Dong 已提交
57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76
static unsigned long ioapic_read_indirect(struct kvm_ioapic *ioapic,
					  unsigned long addr,
					  unsigned long length)
{
	unsigned long result = 0;

	switch (ioapic->ioregsel) {
	case IOAPIC_REG_VERSION:
		result = ((((IOAPIC_NUM_PINS - 1) & 0xff) << 16)
			  | (IOAPIC_VERSION_ID & 0xff));
		break;

	case IOAPIC_REG_APIC_ID:
	case IOAPIC_REG_ARB_ID:
		result = ((ioapic->id & 0xf) << 24);
		break;

	default:
		{
			u32 redir_index = (ioapic->ioregsel - 0x10) >> 1;
77
			u64 redir_content = ~0ULL;
E
Eddie Dong 已提交
78

79 80 81 82 83 84
			if (redir_index < IOAPIC_NUM_PINS) {
				u32 index = array_index_nospec(
					redir_index, IOAPIC_NUM_PINS);

				redir_content = ioapic->redirtbl[index].bits;
			}
E
Eddie Dong 已提交
85 86 87 88 89 90 91 92 93 94 95

			result = (ioapic->ioregsel & 0x1) ?
			    (redir_content >> 32) & 0xffffffff :
			    redir_content & 0xffffffff;
			break;
		}
	}

	return result;
}

96 97 98
static void rtc_irq_eoi_tracking_reset(struct kvm_ioapic *ioapic)
{
	ioapic->rtc_status.pending_eoi = 0;
99
	bitmap_zero(ioapic->rtc_status.dest_map.map, KVM_MAX_VCPU_ID + 1);
100 101
}

102 103 104 105 106 107 108 109
static void kvm_rtc_eoi_tracking_restore_all(struct kvm_ioapic *ioapic);

static void rtc_status_pending_eoi_check_valid(struct kvm_ioapic *ioapic)
{
	if (WARN_ON(ioapic->rtc_status.pending_eoi < 0))
		kvm_rtc_eoi_tracking_restore_all(ioapic);
}

110 111 112 113
static void __rtc_irq_eoi_tracking_restore_one(struct kvm_vcpu *vcpu)
{
	bool new_val, old_val;
	struct kvm_ioapic *ioapic = vcpu->kvm->arch.vioapic;
114
	struct dest_map *dest_map = &ioapic->rtc_status.dest_map;
115 116 117
	union kvm_ioapic_redirect_entry *e;

	e = &ioapic->redirtbl[RTC_GSI];
118 119 120
	if (!kvm_apic_match_dest(vcpu, NULL, APIC_DEST_NOSHORT,
				 e->fields.dest_id,
				 kvm_lapic_irq_dest_mode(!!e->fields.dest_mode)))
121 122 123
		return;

	new_val = kvm_apic_pending_eoi(vcpu, e->fields.vector);
124
	old_val = test_bit(vcpu->vcpu_id, dest_map->map);
125 126 127 128 129

	if (new_val == old_val)
		return;

	if (new_val) {
130 131
		__set_bit(vcpu->vcpu_id, dest_map->map);
		dest_map->vectors[vcpu->vcpu_id] = e->fields.vector;
132 133
		ioapic->rtc_status.pending_eoi++;
	} else {
134
		__clear_bit(vcpu->vcpu_id, dest_map->map);
135
		ioapic->rtc_status.pending_eoi--;
136
		rtc_status_pending_eoi_check_valid(ioapic);
137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161
	}
}

void kvm_rtc_eoi_tracking_restore_one(struct kvm_vcpu *vcpu)
{
	struct kvm_ioapic *ioapic = vcpu->kvm->arch.vioapic;

	spin_lock(&ioapic->lock);
	__rtc_irq_eoi_tracking_restore_one(vcpu);
	spin_unlock(&ioapic->lock);
}

static void kvm_rtc_eoi_tracking_restore_all(struct kvm_ioapic *ioapic)
{
	struct kvm_vcpu *vcpu;
	int i;

	if (RTC_GSI >= IOAPIC_NUM_PINS)
		return;

	rtc_irq_eoi_tracking_reset(ioapic);
	kvm_for_each_vcpu(i, vcpu, ioapic->kvm)
	    __rtc_irq_eoi_tracking_restore_one(vcpu);
}

162 163
static void rtc_irq_eoi(struct kvm_ioapic *ioapic, struct kvm_vcpu *vcpu,
			int vector)
164
{
165 166 167 168 169 170 171
	struct dest_map *dest_map = &ioapic->rtc_status.dest_map;

	/* RTC special handling */
	if (test_bit(vcpu->vcpu_id, dest_map->map) &&
	    (vector == dest_map->vectors[vcpu->vcpu_id]) &&
	    (test_and_clear_bit(vcpu->vcpu_id,
				ioapic->rtc_status.dest_map.map))) {
172
		--ioapic->rtc_status.pending_eoi;
173 174
		rtc_status_pending_eoi_check_valid(ioapic);
	}
175 176 177 178 179 180 181 182 183 184
}

static bool rtc_irq_check_coalesced(struct kvm_ioapic *ioapic)
{
	if (ioapic->rtc_status.pending_eoi > 0)
		return true; /* coalesced */

	return false;
}

185 186 187 188 189 190 191 192 193 194 195 196 197 198 199
static void ioapic_lazy_update_eoi(struct kvm_ioapic *ioapic, int irq)
{
	int i;
	struct kvm_vcpu *vcpu;
	union kvm_ioapic_redirect_entry *entry = &ioapic->redirtbl[irq];

	kvm_for_each_vcpu(i, vcpu, ioapic->kvm) {
		if (!kvm_apic_match_dest(vcpu, NULL, APIC_DEST_NOSHORT,
					 entry->fields.dest_id,
					 entry->fields.dest_mode) ||
		    kvm_apic_pending_eoi(vcpu, entry->fields.vector))
			continue;

		/*
		 * If no longer has pending EOI in LAPICs, update
200
		 * EOI for this vector.
201 202 203 204 205 206
		 */
		rtc_irq_eoi(ioapic, vcpu, entry->fields.vector);
		break;
	}
}

207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223
static int ioapic_set_irq(struct kvm_ioapic *ioapic, unsigned int irq,
		int irq_level, bool line_status)
{
	union kvm_ioapic_redirect_entry entry;
	u32 mask = 1 << irq;
	u32 old_irr;
	int edge, ret;

	entry = ioapic->redirtbl[irq];
	edge = (entry.fields.trig_mode == IOAPIC_EDGE_TRIG);

	if (!irq_level) {
		ioapic->irr &= ~mask;
		ret = 1;
		goto out;
	}

224
	/*
225 226 227 228
	 * AMD SVM AVIC accelerate EOI write iff the interrupt is edge
	 * triggered, in which case the in-kernel IOAPIC will not be able
	 * to receive the EOI.  In this case, we do a lazy update of the
	 * pending EOI when trying to set IOAPIC irq.
229
	 */
230
	if (edge && kvm_apicv_activated(ioapic->kvm))
231 232
		ioapic_lazy_update_eoi(ioapic, irq);

233 234 235
	/*
	 * Return 0 for coalesced interrupts; for edge-triggered interrupts,
	 * this only happens if a previous edge has not been delivered due
M
Miaohe Lin 已提交
236
	 * to masking.  For level interrupts, the remote_irr field tells
237 238 239 240 241 242 243 244 245 246 247 248 249 250 251
	 * us if the interrupt is waiting for an EOI.
	 *
	 * RTC is special: it is edge-triggered, but userspace likes to know
	 * if it has been already ack-ed via EOI because coalesced RTC
	 * interrupts lead to time drift in Windows guests.  So we track
	 * EOI manually for the RTC interrupt.
	 */
	if (irq == RTC_GSI && line_status &&
		rtc_irq_check_coalesced(ioapic)) {
		ret = 0;
		goto out;
	}

	old_irr = ioapic->irr;
	ioapic->irr |= mask;
252
	if (edge) {
253
		ioapic->irr_delivered &= ~mask;
254 255 256 257
		if (old_irr == ioapic->irr) {
			ret = 0;
			goto out;
		}
258 259 260 261 262 263 264 265 266
	}

	ret = ioapic_service(ioapic, irq, line_status);

out:
	trace_kvm_ioapic_set_irq(entry.bits, irq, ret == 0);
	return ret;
}

267 268 269 270 271 272 273 274 275 276 277 278
static void kvm_ioapic_inject_all(struct kvm_ioapic *ioapic, unsigned long irr)
{
	u32 idx;

	rtc_irq_eoi_tracking_reset(ioapic);
	for_each_set_bit(idx, &irr, IOAPIC_NUM_PINS)
		ioapic_set_irq(ioapic, idx, 1, true);

	kvm_rtc_eoi_tracking_restore_all(ioapic);
}


279
void kvm_ioapic_scan_entry(struct kvm_vcpu *vcpu, ulong *ioapic_handled_vectors)
280 281
{
	struct kvm_ioapic *ioapic = vcpu->kvm->arch.vioapic;
282
	struct dest_map *dest_map = &ioapic->rtc_status.dest_map;
283 284 285 286
	union kvm_ioapic_redirect_entry *e;
	int index;

	spin_lock(&ioapic->lock);
287 288 289 290 291 292

	/* Make sure we see any missing RTC EOI */
	if (test_bit(vcpu->vcpu_id, dest_map->map))
		__set_bit(dest_map->vectors[vcpu->vcpu_id],
			  ioapic_handled_vectors);

293 294
	for (index = 0; index < IOAPIC_NUM_PINS; index++) {
		e = &ioapic->redirtbl[index];
295 296 297
		if (e->fields.trig_mode == IOAPIC_LEVEL_TRIG ||
		    kvm_irq_has_notifier(ioapic->kvm, KVM_IRQCHIP_IOAPIC, index) ||
		    index == RTC_GSI) {
298 299 300 301
			u16 dm = kvm_lapic_irq_dest_mode(!!e->fields.dest_mode);

			if (kvm_apic_match_dest(vcpu, NULL, APIC_DEST_NOSHORT,
						e->fields.dest_id, dm) ||
302
			    kvm_apic_pending_eoi(vcpu, e->fields.vector))
303
				__set_bit(e->fields.vector,
304
					  ioapic_handled_vectors);
305 306 307 308 309
		}
	}
	spin_unlock(&ioapic->lock);
}

310
void kvm_arch_post_irq_ack_notifier_list_update(struct kvm *kvm)
311
{
312
	if (!ioapic_in_kernel(kvm))
313
		return;
314
	kvm_make_scan_ioapic_request(kvm);
315 316
}

E
Eddie Dong 已提交
317 318 319
static void ioapic_write_indirect(struct kvm_ioapic *ioapic, u32 val)
{
	unsigned index;
320
	bool mask_before, mask_after;
321
	union kvm_ioapic_redirect_entry *e;
322 323
	unsigned long vcpu_bitmap;
	int old_remote_irr, old_delivery_status, old_dest_id, old_dest_mode;
E
Eddie Dong 已提交
324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341

	switch (ioapic->ioregsel) {
	case IOAPIC_REG_VERSION:
		/* Writes are ignored. */
		break;

	case IOAPIC_REG_APIC_ID:
		ioapic->id = (val >> 24) & 0xf;
		break;

	case IOAPIC_REG_ARB_ID:
		break;

	default:
		index = (ioapic->ioregsel - 0x10) >> 1;

		if (index >= IOAPIC_NUM_PINS)
			return;
342
		index = array_index_nospec(index, IOAPIC_NUM_PINS);
343 344
		e = &ioapic->redirtbl[index];
		mask_before = e->fields.mask;
345 346 347
		/* Preserve read-only fields */
		old_remote_irr = e->fields.remote_irr;
		old_delivery_status = e->fields.delivery_status;
348 349
		old_dest_id = e->fields.dest_id;
		old_dest_mode = e->fields.dest_mode;
E
Eddie Dong 已提交
350
		if (ioapic->ioregsel & 1) {
351 352
			e->bits &= 0xffffffff;
			e->bits |= (u64) val << 32;
E
Eddie Dong 已提交
353
		} else {
354 355
			e->bits &= ~0xffffffffULL;
			e->bits |= (u32) val;
E
Eddie Dong 已提交
356
		}
357 358
		e->fields.remote_irr = old_remote_irr;
		e->fields.delivery_status = old_delivery_status;
359 360 361 362 363 364 365 366 367 368

		/*
		 * Some OSes (Linux, Xen) assume that Remote IRR bit will
		 * be cleared by IOAPIC hardware when the entry is configured
		 * as edge-triggered. This behavior is used to simulate an
		 * explicit EOI on IOAPICs that don't have the EOI register.
		 */
		if (e->fields.trig_mode == IOAPIC_EDGE_TRIG)
			e->fields.remote_irr = 0;

369
		mask_after = e->fields.mask;
370
		if (mask_before != mask_after)
371
			kvm_fire_mask_notifiers(ioapic->kvm, KVM_IRQCHIP_IOAPIC, index, mask_after);
372
		if (e->fields.trig_mode == IOAPIC_LEVEL_TRIG
373
		    && ioapic->irr & (1 << index))
374
			ioapic_service(ioapic, index, false);
375 376 377 378 379
		if (e->fields.delivery_mode == APIC_DM_FIXED) {
			struct kvm_lapic_irq irq;

			irq.vector = e->fields.vector;
			irq.delivery_mode = e->fields.delivery_mode << 8;
380 381
			irq.dest_mode =
			    kvm_lapic_irq_dest_mode(!!e->fields.dest_mode);
382 383 384 385 386
			irq.level = false;
			irq.trig_mode = e->fields.trig_mode;
			irq.shorthand = APIC_DEST_NOSHORT;
			irq.dest_id = e->fields.dest_id;
			irq.msi_redir_hint = false;
387
			bitmap_zero(&vcpu_bitmap, 16);
388 389 390 391 392 393 394 395 396 397
			kvm_bitmap_or_dest_vcpus(ioapic->kvm, &irq,
						 &vcpu_bitmap);
			if (old_dest_mode != e->fields.dest_mode ||
			    old_dest_id != e->fields.dest_id) {
				/*
				 * Update vcpu_bitmap with vcpus specified in
				 * the previous request as well. This is done to
				 * keep ioapic_handled_vectors synchronized.
				 */
				irq.dest_id = old_dest_id;
398 399 400
				irq.dest_mode =
				    kvm_lapic_irq_dest_mode(
					!!e->fields.dest_mode);
401 402 403 404 405 406 407 408
				kvm_bitmap_or_dest_vcpus(ioapic->kvm, &irq,
							 &vcpu_bitmap);
			}
			kvm_make_scan_ioapic_request_mask(ioapic->kvm,
							  &vcpu_bitmap);
		} else {
			kvm_make_scan_ioapic_request(ioapic->kvm);
		}
E
Eddie Dong 已提交
409 410 411 412
		break;
	}
}

413
static int ioapic_service(struct kvm_ioapic *ioapic, int irq, bool line_status)
414
{
415 416
	union kvm_ioapic_redirect_entry *entry = &ioapic->redirtbl[irq];
	struct kvm_lapic_irq irqe;
417
	int ret;
418

419 420 421
	if (entry->fields.mask ||
	    (entry->fields.trig_mode == IOAPIC_LEVEL_TRIG &&
	    entry->fields.remote_irr))
422 423
		return -1;

424 425
	irqe.dest_id = entry->fields.dest_id;
	irqe.vector = entry->fields.vector;
426
	irqe.dest_mode = kvm_lapic_irq_dest_mode(!!entry->fields.dest_mode);
427 428 429
	irqe.trig_mode = entry->fields.trig_mode;
	irqe.delivery_mode = entry->fields.delivery_mode << 8;
	irqe.level = 1;
430
	irqe.shorthand = APIC_DEST_NOSHORT;
431
	irqe.msi_redir_hint = false;
432

433
	if (irqe.trig_mode == IOAPIC_EDGE_TRIG)
434
		ioapic->irr_delivered |= 1 << irq;
435

436
	if (irq == RTC_GSI && line_status) {
437 438 439 440 441 442
		/*
		 * pending_eoi cannot ever become negative (see
		 * rtc_status_pending_eoi_check_valid) and the caller
		 * ensures that it is only called if it is >= zero, namely
		 * if rtc_irq_check_coalesced returns false).
		 */
443 444
		BUG_ON(ioapic->rtc_status.pending_eoi != 0);
		ret = kvm_irq_delivery_to_apic(ioapic->kvm, NULL, &irqe,
445
					       &ioapic->rtc_status.dest_map);
446
		ioapic->rtc_status.pending_eoi = (ret < 0 ? 0 : ret);
447 448 449
	} else
		ret = kvm_irq_delivery_to_apic(ioapic->kvm, NULL, &irqe, NULL);

450 451 452
	if (ret && irqe.trig_mode == IOAPIC_LEVEL_TRIG)
		entry->fields.remote_irr = 1;

453
	return ret;
454 455
}

456
int kvm_ioapic_set_irq(struct kvm_ioapic *ioapic, int irq, int irq_source_id,
457
		       int level, bool line_status)
E
Eddie Dong 已提交
458
{
459 460 461
	int ret, irq_level;

	BUG_ON(irq < 0 || irq >= IOAPIC_NUM_PINS);
E
Eddie Dong 已提交
462

463
	spin_lock(&ioapic->lock);
464 465
	irq_level = __kvm_irq_line_state(&ioapic->irq_states[irq],
					 irq_source_id, level);
466
	ret = ioapic_set_irq(ioapic, irq, irq_level, line_status);
467

468
	spin_unlock(&ioapic->lock);
G
Gleb Natapov 已提交
469

470
	return ret;
E
Eddie Dong 已提交
471 472
}

473 474 475 476 477 478 479 480 481 482
void kvm_ioapic_clear_all(struct kvm_ioapic *ioapic, int irq_source_id)
{
	int i;

	spin_lock(&ioapic->lock);
	for (i = 0; i < KVM_IOAPIC_NUM_PINS; i++)
		__clear_bit(irq_source_id, &ioapic->irq_states[i]);
	spin_unlock(&ioapic->lock);
}

483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501
static void kvm_ioapic_eoi_inject_work(struct work_struct *work)
{
	int i;
	struct kvm_ioapic *ioapic = container_of(work, struct kvm_ioapic,
						 eoi_inject.work);
	spin_lock(&ioapic->lock);
	for (i = 0; i < IOAPIC_NUM_PINS; i++) {
		union kvm_ioapic_redirect_entry *ent = &ioapic->redirtbl[i];

		if (ent->fields.trig_mode != IOAPIC_LEVEL_TRIG)
			continue;

		if (ioapic->irr & (1 << i) && !ent->fields.remote_irr)
			ioapic_service(ioapic, i, false);
	}
	spin_unlock(&ioapic->lock);
}

#define IOAPIC_SUCCESSIVE_IRQ_MAX_COUNT 10000
502 503 504 505
static void kvm_ioapic_update_eoi_one(struct kvm_vcpu *vcpu,
				      struct kvm_ioapic *ioapic,
				      int trigger_mode,
				      int pin)
E
Eddie Dong 已提交
506
{
507
	struct kvm_lapic *apic = vcpu->arch.apic;
508
	union kvm_ioapic_redirect_entry *ent = &ioapic->redirtbl[pin];
G
Gleb Natapov 已提交
509

510 511 512 513 514 515 516 517 518 519 520
	/*
	 * We are dropping lock while calling ack notifiers because ack
	 * notifier callbacks for assigned devices call into IOAPIC
	 * recursively. Since remote_irr is cleared only after call
	 * to notifiers if the same vector will be delivered while lock
	 * is dropped it will be put into irr and will be delivered
	 * after ack notifier returns.
	 */
	spin_unlock(&ioapic->lock);
	kvm_notify_acked_irq(ioapic->kvm, KVM_IRQCHIP_IOAPIC, pin);
	spin_lock(&ioapic->lock);
G
Gleb Natapov 已提交
521

522 523 524
	if (trigger_mode != IOAPIC_LEVEL_TRIG ||
	    kvm_lapic_get_reg(apic, APIC_SPIV) & APIC_SPIV_DIRECTED_EOI)
		return;
M
Marcelo Tosatti 已提交
525

526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541
	ASSERT(ent->fields.trig_mode == IOAPIC_LEVEL_TRIG);
	ent->fields.remote_irr = 0;
	if (!ent->fields.mask && (ioapic->irr & (1 << pin))) {
		++ioapic->irq_eoi[pin];
		if (ioapic->irq_eoi[pin] == IOAPIC_SUCCESSIVE_IRQ_MAX_COUNT) {
			/*
			 * Real hardware does not deliver the interrupt
			 * immediately during eoi broadcast, and this
			 * lets a buggy guest make slow progress
			 * even if it does not correctly handle a
			 * level-triggered interrupt.  Emulate this
			 * behavior if we detect an interrupt storm.
			 */
			schedule_delayed_work(&ioapic->eoi_inject, HZ / 100);
			ioapic->irq_eoi[pin] = 0;
			trace_kvm_ioapic_delayed_eoi_inj(ent->bits);
542
		} else {
543
			ioapic_service(ioapic, pin, false);
544
		}
545 546
	} else {
		ioapic->irq_eoi[pin] = 0;
M
Marcelo Tosatti 已提交
547
	}
E
Eddie Dong 已提交
548 549
}

550
void kvm_ioapic_update_eoi(struct kvm_vcpu *vcpu, int vector, int trigger_mode)
551
{
552
	int i;
553
	struct kvm_ioapic *ioapic = vcpu->kvm->arch.vioapic;
554

555
	spin_lock(&ioapic->lock);
556 557 558 559 560 561 562 563
	rtc_irq_eoi(ioapic, vcpu, vector);
	for (i = 0; i < IOAPIC_NUM_PINS; i++) {
		union kvm_ioapic_redirect_entry *ent = &ioapic->redirtbl[i];

		if (ent->fields.vector != vector)
			continue;
		kvm_ioapic_update_eoi_one(vcpu, ioapic, trigger_mode, i);
	}
564
	spin_unlock(&ioapic->lock);
565 566
}

G
Gregory Haskins 已提交
567 568 569 570 571
static inline struct kvm_ioapic *to_ioapic(struct kvm_io_device *dev)
{
	return container_of(dev, struct kvm_ioapic, dev);
}

572
static inline int ioapic_in_range(struct kvm_ioapic *ioapic, gpa_t addr)
E
Eddie Dong 已提交
573 574 575 576 577
{
	return ((addr >= ioapic->base_address &&
		 (addr < ioapic->base_address + IOAPIC_MEM_LENGTH)));
}

578 579
static int ioapic_mmio_read(struct kvm_vcpu *vcpu, struct kvm_io_device *this,
				gpa_t addr, int len, void *val)
E
Eddie Dong 已提交
580
{
G
Gregory Haskins 已提交
581
	struct kvm_ioapic *ioapic = to_ioapic(this);
E
Eddie Dong 已提交
582
	u32 result;
583 584
	if (!ioapic_in_range(ioapic, addr))
		return -EOPNOTSUPP;
E
Eddie Dong 已提交
585 586 587 588

	ASSERT(!(addr & 0xf));	/* check alignment */

	addr &= 0xff;
589
	spin_lock(&ioapic->lock);
E
Eddie Dong 已提交
590 591 592 593 594 595 596 597 598 599 600 601 602
	switch (addr) {
	case IOAPIC_REG_SELECT:
		result = ioapic->ioregsel;
		break;

	case IOAPIC_REG_WINDOW:
		result = ioapic_read_indirect(ioapic, addr, len);
		break;

	default:
		result = 0;
		break;
	}
603
	spin_unlock(&ioapic->lock);
G
Gleb Natapov 已提交
604

E
Eddie Dong 已提交
605 606 607 608 609 610 611 612 613 614 615 616
	switch (len) {
	case 8:
		*(u64 *) val = result;
		break;
	case 1:
	case 2:
	case 4:
		memcpy(val, (char *)&result, len);
		break;
	default:
		printk(KERN_WARNING "ioapic: wrong length %d\n", len);
	}
617
	return 0;
E
Eddie Dong 已提交
618 619
}

620 621
static int ioapic_mmio_write(struct kvm_vcpu *vcpu, struct kvm_io_device *this,
				 gpa_t addr, int len, const void *val)
E
Eddie Dong 已提交
622
{
G
Gregory Haskins 已提交
623
	struct kvm_ioapic *ioapic = to_ioapic(this);
E
Eddie Dong 已提交
624
	u32 data;
625 626
	if (!ioapic_in_range(ioapic, addr))
		return -EOPNOTSUPP;
E
Eddie Dong 已提交
627 628

	ASSERT(!(addr & 0xf));	/* check alignment */
629

630 631 632
	switch (len) {
	case 8:
	case 4:
E
Eddie Dong 已提交
633
		data = *(u32 *) val;
634 635 636 637 638 639 640 641
		break;
	case 2:
		data = *(u16 *) val;
		break;
	case 1:
		data = *(u8  *) val;
		break;
	default:
E
Eddie Dong 已提交
642
		printk(KERN_WARNING "ioapic: Unsupported size %d\n", len);
G
Gleb Natapov 已提交
643
		return 0;
E
Eddie Dong 已提交
644 645 646
	}

	addr &= 0xff;
647
	spin_lock(&ioapic->lock);
E
Eddie Dong 已提交
648 649
	switch (addr) {
	case IOAPIC_REG_SELECT:
650
		ioapic->ioregsel = data & 0xFF; /* 8-bit register */
E
Eddie Dong 已提交
651 652 653 654 655 656 657 658 659
		break;

	case IOAPIC_REG_WINDOW:
		ioapic_write_indirect(ioapic, data);
		break;

	default:
		break;
	}
660
	spin_unlock(&ioapic->lock);
661
	return 0;
E
Eddie Dong 已提交
662 663
}

664
static void kvm_ioapic_reset(struct kvm_ioapic *ioapic)
665 666 667
{
	int i;

668
	cancel_delayed_work_sync(&ioapic->eoi_inject);
669 670 671 672 673
	for (i = 0; i < IOAPIC_NUM_PINS; i++)
		ioapic->redirtbl[i].fields.mask = 1;
	ioapic->base_address = IOAPIC_DEFAULT_BASE_ADDRESS;
	ioapic->ioregsel = 0;
	ioapic->irr = 0;
674
	ioapic->irr_delivered = 0;
675
	ioapic->id = 0;
J
Jiri Slaby 已提交
676
	memset(ioapic->irq_eoi, 0x00, sizeof(ioapic->irq_eoi));
677
	rtc_irq_eoi_tracking_reset(ioapic);
678 679
}

G
Gregory Haskins 已提交
680 681 682 683 684
static const struct kvm_io_device_ops ioapic_mmio_ops = {
	.read     = ioapic_mmio_read,
	.write    = ioapic_mmio_write,
};

E
Eddie Dong 已提交
685 686 687
int kvm_ioapic_init(struct kvm *kvm)
{
	struct kvm_ioapic *ioapic;
688
	int ret;
E
Eddie Dong 已提交
689

690
	ioapic = kzalloc(sizeof(struct kvm_ioapic), GFP_KERNEL_ACCOUNT);
E
Eddie Dong 已提交
691 692
	if (!ioapic)
		return -ENOMEM;
693
	spin_lock_init(&ioapic->lock);
694
	INIT_DELAYED_WORK(&ioapic->eoi_inject, kvm_ioapic_eoi_inject_work);
695
	kvm->arch.vioapic = ioapic;
696
	kvm_ioapic_reset(ioapic);
G
Gregory Haskins 已提交
697
	kvm_iodevice_init(&ioapic->dev, &ioapic_mmio_ops);
E
Eddie Dong 已提交
698
	ioapic->kvm = kvm;
699
	mutex_lock(&kvm->slots_lock);
700 701
	ret = kvm_io_bus_register_dev(kvm, KVM_MMIO_BUS, ioapic->base_address,
				      IOAPIC_MEM_LENGTH, &ioapic->dev);
702
	mutex_unlock(&kvm->slots_lock);
703 704
	if (ret < 0) {
		kvm->arch.vioapic = NULL;
705
		kfree(ioapic);
706
	}
707 708

	return ret;
E
Eddie Dong 已提交
709
}
710

711 712 713 714
void kvm_ioapic_destroy(struct kvm *kvm)
{
	struct kvm_ioapic *ioapic = kvm->arch.vioapic;

715 716 717
	if (!ioapic)
		return;

718
	cancel_delayed_work_sync(&ioapic->eoi_inject);
719
	mutex_lock(&kvm->slots_lock);
J
Julia Lawall 已提交
720
	kvm_io_bus_unregister_dev(kvm, KVM_MMIO_BUS, &ioapic->dev);
721
	mutex_unlock(&kvm->slots_lock);
J
Julia Lawall 已提交
722 723
	kvm->arch.vioapic = NULL;
	kfree(ioapic);
724 725
}

726
void kvm_get_ioapic(struct kvm *kvm, struct kvm_ioapic_state *state)
G
Gleb Natapov 已提交
727
{
728
	struct kvm_ioapic *ioapic = kvm->arch.vioapic;
G
Gleb Natapov 已提交
729

730
	spin_lock(&ioapic->lock);
G
Gleb Natapov 已提交
731
	memcpy(state, ioapic, sizeof(struct kvm_ioapic_state));
732
	state->irr &= ~ioapic->irr_delivered;
733
	spin_unlock(&ioapic->lock);
G
Gleb Natapov 已提交
734 735
}

736
void kvm_set_ioapic(struct kvm *kvm, struct kvm_ioapic_state *state)
G
Gleb Natapov 已提交
737
{
738
	struct kvm_ioapic *ioapic = kvm->arch.vioapic;
G
Gleb Natapov 已提交
739

740
	spin_lock(&ioapic->lock);
G
Gleb Natapov 已提交
741
	memcpy(ioapic, state, sizeof(struct kvm_ioapic_state));
742
	ioapic->irr = 0;
743
	ioapic->irr_delivered = 0;
744
	kvm_make_scan_ioapic_request(kvm);
745
	kvm_ioapic_inject_all(ioapic, state->irr);
746
	spin_unlock(&ioapic->lock);
G
Gleb Natapov 已提交
747
}