kvm_main.c 84.6 KB
Newer Older
1 2 3 4 5 6 7
/*
 * Kernel-based Virtual Machine driver for Linux
 *
 * This module enables machines with Intel VT-x extensions to run virtual
 * machines without emulation or binary translation.
 *
 * Copyright (C) 2006 Qumranet, Inc.
8
 * Copyright 2010 Red Hat, Inc. and/or its affiliates.
9 10 11 12 13 14 15 16 17 18
 *
 * Authors:
 *   Avi Kivity   <avi@qumranet.com>
 *   Yaniv Kamay  <yaniv@qumranet.com>
 *
 * This work is licensed under the terms of the GNU GPL, version 2.  See
 * the COPYING file in the top-level directory.
 *
 */

19
#include <kvm/iodev.h>
20

21
#include <linux/kvm_host.h>
22 23 24 25 26 27 28 29 30 31 32
#include <linux/kvm.h>
#include <linux/module.h>
#include <linux/errno.h>
#include <linux/percpu.h>
#include <linux/mm.h>
#include <linux/miscdevice.h>
#include <linux/vmalloc.h>
#include <linux/reboot.h>
#include <linux/debugfs.h>
#include <linux/highmem.h>
#include <linux/file.h>
33
#include <linux/syscore_ops.h>
34
#include <linux/cpu.h>
Alexey Dobriyan's avatar
Alexey Dobriyan committed
35
#include <linux/sched.h>
36 37
#include <linux/cpumask.h>
#include <linux/smp.h>
38
#include <linux/anon_inodes.h>
39
#include <linux/profile.h>
40
#include <linux/kvm_para.h>
41
#include <linux/pagemap.h>
42
#include <linux/mman.h>
43
#include <linux/swap.h>
44
#include <linux/bitops.h>
45
#include <linux/spinlock.h>
46
#include <linux/compat.h>
47
#include <linux/srcu.h>
48
#include <linux/hugetlb.h>
49
#include <linux/slab.h>
50 51
#include <linux/sort.h>
#include <linux/bsearch.h>
52

Avi Kivity's avatar
Avi Kivity committed
53 54
#include <asm/processor.h>
#include <asm/io.h>
55
#include <asm/ioctl.h>
Avi Kivity's avatar
Avi Kivity committed
56
#include <asm/uaccess.h>
57
#include <asm/pgtable.h>
58

59
#include "coalesced_mmio.h"
60
#include "async_pf.h"
61
#include "vfio.h"
62

63 64 65
#define CREATE_TRACE_POINTS
#include <trace/events/kvm.h>

66 67 68
MODULE_AUTHOR("Qumranet");
MODULE_LICENSE("GPL");

69 70
/* Architectures should define their poll value according to the halt latency */
static unsigned int halt_poll_ns = KVM_HALT_POLL_NS_DEFAULT;
71 72
module_param(halt_poll_ns, uint, S_IRUGO | S_IWUSR);

Wanpeng Li's avatar
Wanpeng Li committed
73 74 75 76 77 78 79 80
/* Default doubles per-vcpu halt_poll_ns. */
static unsigned int halt_poll_ns_grow = 2;
module_param(halt_poll_ns_grow, int, S_IRUGO);

/* Default resets per-vcpu halt_poll_ns . */
static unsigned int halt_poll_ns_shrink;
module_param(halt_poll_ns_shrink, int, S_IRUGO);

81 82 83
/*
 * Ordering of locks:
 *
84
 *	kvm->lock --> kvm->slots_lock --> kvm->irq_lock
85 86
 */

87
DEFINE_MUTEX(kvm_lock);
88
static DEFINE_RAW_SPINLOCK(kvm_count_lock);
89
LIST_HEAD(vm_list);
90

91
static cpumask_var_t cpus_hardware_enabled;
92
static int kvm_usage_count;
93
static atomic_t hardware_enable_failed;
94

95 96
struct kmem_cache *kvm_vcpu_cache;
EXPORT_SYMBOL_GPL(kvm_vcpu_cache);
Avi Kivity's avatar
Avi Kivity committed
97

98 99
static __read_mostly struct preempt_ops kvm_preempt_ops;

100
struct dentry *kvm_debugfs_dir;
101
EXPORT_SYMBOL_GPL(kvm_debugfs_dir);
102

Avi Kivity's avatar
Avi Kivity committed
103 104
static long kvm_vcpu_ioctl(struct file *file, unsigned int ioctl,
			   unsigned long arg);
105
#ifdef CONFIG_KVM_COMPAT
106 107 108
static long kvm_vcpu_compat_ioctl(struct file *file, unsigned int ioctl,
				  unsigned long arg);
#endif
109 110
static int hardware_enable_all(void);
static void hardware_disable_all(void);
Avi Kivity's avatar
Avi Kivity committed
111

112
static void kvm_io_bus_destroy(struct kvm_io_bus *bus);
113 114

static void kvm_release_pfn_dirty(pfn_t pfn);
115
static void mark_page_dirty_in_slot(struct kvm_memory_slot *memslot, gfn_t gfn);
116

117
__visible bool kvm_rebooting;
118
EXPORT_SYMBOL_GPL(kvm_rebooting);
119

120 121
static bool largepages_enabled = true;

122
bool kvm_is_reserved_pfn(pfn_t pfn)
Ben-Ami Yassour's avatar
Ben-Ami Yassour committed
123
{
124
	if (pfn_valid(pfn))
125
		return PageReserved(pfn_to_page(pfn));
Ben-Ami Yassour's avatar
Ben-Ami Yassour committed
126 127 128 129

	return true;
}

Avi Kivity's avatar
Avi Kivity committed
130 131 132
/*
 * Switches to specified vcpu, until a matching vcpu_put()
 */
133
int vcpu_load(struct kvm_vcpu *vcpu)
134
{
135 136
	int cpu;

137 138
	if (mutex_lock_killable(&vcpu->mutex))
		return -EINTR;
139 140
	cpu = get_cpu();
	preempt_notifier_register(&vcpu->preempt_notifier);
141
	kvm_arch_vcpu_load(vcpu, cpu);
142
	put_cpu();
143
	return 0;
144
}
145
EXPORT_SYMBOL_GPL(vcpu_load);
146

147
void vcpu_put(struct kvm_vcpu *vcpu)
148
{
149
	preempt_disable();
150
	kvm_arch_vcpu_put(vcpu);
151 152
	preempt_notifier_unregister(&vcpu->preempt_notifier);
	preempt_enable();
153 154
	mutex_unlock(&vcpu->mutex);
}
155
EXPORT_SYMBOL_GPL(vcpu_put);
156

157 158 159 160
static void ack_flush(void *_completed)
{
}

161
bool kvm_make_all_cpus_request(struct kvm *kvm, unsigned int req)
162
{
163
	int i, cpu, me;
164 165
	cpumask_var_t cpus;
	bool called = true;
166 167
	struct kvm_vcpu *vcpu;

168
	zalloc_cpumask_var(&cpus, GFP_ATOMIC);
169

170
	me = get_cpu();
171
	kvm_for_each_vcpu(i, vcpu, kvm) {
172
		kvm_make_request(req, vcpu);
173
		cpu = vcpu->cpu;
174 175 176 177 178 179

		/* Set ->requests bit before we read ->mode */
		smp_mb();

		if (cpus != NULL && cpu != -1 && cpu != me &&
		      kvm_vcpu_exiting_guest_mode(vcpu) != OUTSIDE_GUEST_MODE)
180
			cpumask_set_cpu(cpu, cpus);
181
	}
182 183 184 185 186 187
	if (unlikely(cpus == NULL))
		smp_call_function_many(cpu_online_mask, ack_flush, NULL, 1);
	else if (!cpumask_empty(cpus))
		smp_call_function_many(cpus, ack_flush, NULL, 1);
	else
		called = false;
188
	put_cpu();
189
	free_cpumask_var(cpus);
190
	return called;
191 192
}

193
#ifndef CONFIG_HAVE_KVM_ARCH_TLB_FLUSH_ALL
194
void kvm_flush_remote_tlbs(struct kvm *kvm)
195
{
196 197 198
	long dirty_count = kvm->tlbs_dirty;

	smp_mb();
199
	if (kvm_make_all_cpus_request(kvm, KVM_REQ_TLB_FLUSH))
200
		++kvm->stat.remote_tlb_flush;
201
	cmpxchg(&kvm->tlbs_dirty, dirty_count, 0);
202
}
203
EXPORT_SYMBOL_GPL(kvm_flush_remote_tlbs);
204
#endif
205

206 207
void kvm_reload_remote_mmus(struct kvm *kvm)
{
208
	kvm_make_all_cpus_request(kvm, KVM_REQ_MMU_RELOAD);
209
}
210

211 212
void kvm_make_mclock_inprogress_request(struct kvm *kvm)
{
213
	kvm_make_all_cpus_request(kvm, KVM_REQ_MCLOCK_INPROGRESS);
214 215
}

216
void kvm_make_scan_ioapic_request(struct kvm *kvm)
217
{
218
	kvm_make_all_cpus_request(kvm, KVM_REQ_SCAN_IOAPIC);
219 220
}

221 222 223 224 225 226 227 228 229
int kvm_vcpu_init(struct kvm_vcpu *vcpu, struct kvm *kvm, unsigned id)
{
	struct page *page;
	int r;

	mutex_init(&vcpu->mutex);
	vcpu->cpu = -1;
	vcpu->kvm = kvm;
	vcpu->vcpu_id = id;
230
	vcpu->pid = NULL;
231
	vcpu->halt_poll_ns = 0;
Eddie Dong's avatar
Eddie Dong committed
232
	init_waitqueue_head(&vcpu->wq);
233
	kvm_async_pf_vcpu_init(vcpu);
234

235 236 237
	vcpu->pre_pcpu = -1;
	INIT_LIST_HEAD(&vcpu->blocked_vcpu_list);

238 239 240 241 242 243 244
	page = alloc_page(GFP_KERNEL | __GFP_ZERO);
	if (!page) {
		r = -ENOMEM;
		goto fail;
	}
	vcpu->run = page_address(page);

245 246
	kvm_vcpu_set_in_spin_loop(vcpu, false);
	kvm_vcpu_set_dy_eligible(vcpu, false);
247
	vcpu->preempted = false;
248

249
	r = kvm_arch_vcpu_init(vcpu);
250
	if (r < 0)
251
		goto fail_free_run;
252 253 254 255 256
	return 0;

fail_free_run:
	free_page((unsigned long)vcpu->run);
fail:
257
	return r;
258 259 260 261 262
}
EXPORT_SYMBOL_GPL(kvm_vcpu_init);

void kvm_vcpu_uninit(struct kvm_vcpu *vcpu)
{
263
	put_pid(vcpu->pid);
264
	kvm_arch_vcpu_uninit(vcpu);
265 266 267 268
	free_page((unsigned long)vcpu->run);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_uninit);

269 270 271 272 273 274 275 276 277 278 279
#if defined(CONFIG_MMU_NOTIFIER) && defined(KVM_ARCH_WANT_MMU_NOTIFIER)
static inline struct kvm *mmu_notifier_to_kvm(struct mmu_notifier *mn)
{
	return container_of(mn, struct kvm, mmu_notifier);
}

static void kvm_mmu_notifier_invalidate_page(struct mmu_notifier *mn,
					     struct mm_struct *mm,
					     unsigned long address)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
280
	int need_tlb_flush, idx;
281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299

	/*
	 * When ->invalidate_page runs, the linux pte has been zapped
	 * already but the page is still allocated until
	 * ->invalidate_page returns. So if we increase the sequence
	 * here the kvm page fault will notice if the spte can't be
	 * established because the page is going to be freed. If
	 * instead the kvm page fault establishes the spte before
	 * ->invalidate_page runs, kvm_unmap_hva will release it
	 * before returning.
	 *
	 * The sequence increase only need to be seen at spin_unlock
	 * time, and not at spin_lock time.
	 *
	 * Increasing the sequence after the spin_unlock would be
	 * unsafe because the kvm page fault could then establish the
	 * pte after kvm_unmap_hva returned, without noticing the page
	 * is going to be freed.
	 */
300
	idx = srcu_read_lock(&kvm->srcu);
301
	spin_lock(&kvm->mmu_lock);
302

303
	kvm->mmu_notifier_seq++;
304
	need_tlb_flush = kvm_unmap_hva(kvm, address) | kvm->tlbs_dirty;
305 306 307 308
	/* we've to flush the tlb before the pages can be freed */
	if (need_tlb_flush)
		kvm_flush_remote_tlbs(kvm);

309
	spin_unlock(&kvm->mmu_lock);
310 311 312

	kvm_arch_mmu_notifier_invalidate_page(kvm, address);

313
	srcu_read_unlock(&kvm->srcu, idx);
314 315
}

316 317 318 319 320 321
static void kvm_mmu_notifier_change_pte(struct mmu_notifier *mn,
					struct mm_struct *mm,
					unsigned long address,
					pte_t pte)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
322
	int idx;
323

324
	idx = srcu_read_lock(&kvm->srcu);
325 326 327 328
	spin_lock(&kvm->mmu_lock);
	kvm->mmu_notifier_seq++;
	kvm_set_spte_hva(kvm, address, pte);
	spin_unlock(&kvm->mmu_lock);
329
	srcu_read_unlock(&kvm->srcu, idx);
330 331
}

332 333 334 335 336 337
static void kvm_mmu_notifier_invalidate_range_start(struct mmu_notifier *mn,
						    struct mm_struct *mm,
						    unsigned long start,
						    unsigned long end)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
338
	int need_tlb_flush = 0, idx;
339

340
	idx = srcu_read_lock(&kvm->srcu);
341 342 343 344 345 346 347
	spin_lock(&kvm->mmu_lock);
	/*
	 * The count increase must become visible at unlock time as no
	 * spte can be established without taking the mmu_lock and
	 * count is also read inside the mmu_lock critical section.
	 */
	kvm->mmu_notifier_count++;
348
	need_tlb_flush = kvm_unmap_hva_range(kvm, start, end);
349
	need_tlb_flush |= kvm->tlbs_dirty;
350 351 352
	/* we've to flush the tlb before the pages can be freed */
	if (need_tlb_flush)
		kvm_flush_remote_tlbs(kvm);
353 354 355

	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);
356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371
}

static void kvm_mmu_notifier_invalidate_range_end(struct mmu_notifier *mn,
						  struct mm_struct *mm,
						  unsigned long start,
						  unsigned long end)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);

	spin_lock(&kvm->mmu_lock);
	/*
	 * This sequence increase will notify the kvm page fault that
	 * the page that is going to be mapped in the spte could have
	 * been freed.
	 */
	kvm->mmu_notifier_seq++;
372
	smp_wmb();
373 374
	/*
	 * The above sequence increase must be visible before the
375 376
	 * below count decrease, which is ensured by the smp_wmb above
	 * in conjunction with the smp_rmb in mmu_notifier_retry().
377 378 379 380 381 382 383 384 385
	 */
	kvm->mmu_notifier_count--;
	spin_unlock(&kvm->mmu_lock);

	BUG_ON(kvm->mmu_notifier_count < 0);
}

static int kvm_mmu_notifier_clear_flush_young(struct mmu_notifier *mn,
					      struct mm_struct *mm,
386 387
					      unsigned long start,
					      unsigned long end)
388 389
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
390
	int young, idx;
391

392
	idx = srcu_read_lock(&kvm->srcu);
393 394
	spin_lock(&kvm->mmu_lock);

395
	young = kvm_age_hva(kvm, start, end);
396 397 398
	if (young)
		kvm_flush_remote_tlbs(kvm);

399 400 401
	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);

402 403 404
	return young;
}

405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434
static int kvm_mmu_notifier_clear_young(struct mmu_notifier *mn,
					struct mm_struct *mm,
					unsigned long start,
					unsigned long end)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
	int young, idx;

	idx = srcu_read_lock(&kvm->srcu);
	spin_lock(&kvm->mmu_lock);
	/*
	 * Even though we do not flush TLB, this will still adversely
	 * affect performance on pre-Haswell Intel EPT, where there is
	 * no EPT Access Bit to clear so that we have to tear down EPT
	 * tables instead. If we find this unacceptable, we can always
	 * add a parameter to kvm_age_hva so that it effectively doesn't
	 * do anything on clear_young.
	 *
	 * Also note that currently we never issue secondary TLB flushes
	 * from clear_young, leaving this job up to the regular system
	 * cadence. If we find this inaccurate, we might come up with a
	 * more sophisticated heuristic later.
	 */
	young = kvm_age_hva(kvm, start, end);
	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);

	return young;
}

435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450
static int kvm_mmu_notifier_test_young(struct mmu_notifier *mn,
				       struct mm_struct *mm,
				       unsigned long address)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
	int young, idx;

	idx = srcu_read_lock(&kvm->srcu);
	spin_lock(&kvm->mmu_lock);
	young = kvm_test_age_hva(kvm, address);
	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);

	return young;
}

451 452 453 454
static void kvm_mmu_notifier_release(struct mmu_notifier *mn,
				     struct mm_struct *mm)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
455 456 457
	int idx;

	idx = srcu_read_lock(&kvm->srcu);
458
	kvm_arch_flush_shadow_all(kvm);
459
	srcu_read_unlock(&kvm->srcu, idx);
460 461
}

462 463 464 465 466
static const struct mmu_notifier_ops kvm_mmu_notifier_ops = {
	.invalidate_page	= kvm_mmu_notifier_invalidate_page,
	.invalidate_range_start	= kvm_mmu_notifier_invalidate_range_start,
	.invalidate_range_end	= kvm_mmu_notifier_invalidate_range_end,
	.clear_flush_young	= kvm_mmu_notifier_clear_flush_young,
467
	.clear_young		= kvm_mmu_notifier_clear_young,
468
	.test_young		= kvm_mmu_notifier_test_young,
469
	.change_pte		= kvm_mmu_notifier_change_pte,
470
	.release		= kvm_mmu_notifier_release,
471
};
472 473 474 475 476 477 478 479 480 481 482 483 484 485

static int kvm_init_mmu_notifier(struct kvm *kvm)
{
	kvm->mmu_notifier.ops = &kvm_mmu_notifier_ops;
	return mmu_notifier_register(&kvm->mmu_notifier, current->mm);
}

#else  /* !(CONFIG_MMU_NOTIFIER && KVM_ARCH_WANT_MMU_NOTIFIER) */

static int kvm_init_mmu_notifier(struct kvm *kvm)
{
	return 0;
}

486 487
#endif /* CONFIG_MMU_NOTIFIER && KVM_ARCH_WANT_MMU_NOTIFIER */

488
static struct kvm_memslots *kvm_alloc_memslots(void)
489 490
{
	int i;
491
	struct kvm_memslots *slots;
492

493 494 495 496 497 498 499 500 501
	slots = kvm_kvzalloc(sizeof(struct kvm_memslots));
	if (!slots)
		return NULL;

	/*
	 * Init kvm generation close to the maximum to easily test the
	 * code of handling generation number wrap-around.
	 */
	slots->generation = -150;
502
	for (i = 0; i < KVM_MEM_SLOTS_NUM; i++)
503
		slots->id_to_index[i] = slots->memslots[i].id = i;
504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541

	return slots;
}

static void kvm_destroy_dirty_bitmap(struct kvm_memory_slot *memslot)
{
	if (!memslot->dirty_bitmap)
		return;

	kvfree(memslot->dirty_bitmap);
	memslot->dirty_bitmap = NULL;
}

/*
 * Free any memory in @free but not in @dont.
 */
static void kvm_free_memslot(struct kvm *kvm, struct kvm_memory_slot *free,
			      struct kvm_memory_slot *dont)
{
	if (!dont || free->dirty_bitmap != dont->dirty_bitmap)
		kvm_destroy_dirty_bitmap(free);

	kvm_arch_free_memslot(kvm, free, dont);

	free->npages = 0;
}

static void kvm_free_memslots(struct kvm *kvm, struct kvm_memslots *slots)
{
	struct kvm_memory_slot *memslot;

	if (!slots)
		return;

	kvm_for_each_memslot(memslot, slots)
		kvm_free_memslot(kvm, memslot, NULL);

	kvfree(slots);
542 543
}

544
static struct kvm *kvm_create_vm(unsigned long type)
545
{
546 547
	int r, i;
	struct kvm *kvm = kvm_arch_alloc_vm();
548

549 550 551
	if (!kvm)
		return ERR_PTR(-ENOMEM);

552 553 554 555 556 557 558 559 560 561
	spin_lock_init(&kvm->mmu_lock);
	atomic_inc(&current->mm->mm_count);
	kvm->mm = current->mm;
	kvm_eventfd_init(kvm);
	mutex_init(&kvm->lock);
	mutex_init(&kvm->irq_lock);
	mutex_init(&kvm->slots_lock);
	atomic_set(&kvm->users_count, 1);
	INIT_LIST_HEAD(&kvm->devices);

562
	r = kvm_arch_init_vm(kvm, type);
563
	if (r)
564
		goto out_err_no_disable;
565 566 567

	r = hardware_enable_all();
	if (r)
568
		goto out_err_no_disable;
569

570
#ifdef CONFIG_HAVE_KVM_IRQFD
571
	INIT_HLIST_HEAD(&kvm->irq_ack_notifier_list);
572
#endif
573

574 575
	BUILD_BUG_ON(KVM_MEM_SLOTS_NUM > SHRT_MAX);

576
	r = -ENOMEM;
577 578 579 580 581
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++) {
		kvm->memslots[i] = kvm_alloc_memslots();
		if (!kvm->memslots[i])
			goto out_err_no_srcu;
	}
582

583
	if (init_srcu_struct(&kvm->srcu))
584 585 586
		goto out_err_no_srcu;
	if (init_srcu_struct(&kvm->irq_srcu))
		goto out_err_no_irq_srcu;
587 588 589
	for (i = 0; i < KVM_NR_BUSES; i++) {
		kvm->buses[i] = kzalloc(sizeof(struct kvm_io_bus),
					GFP_KERNEL);
590
		if (!kvm->buses[i])
591 592
			goto out_err;
	}
593

594 595 596 597
	r = kvm_init_mmu_notifier(kvm);
	if (r)
		goto out_err;

598
	mutex_lock(&kvm_lock);
599
	list_add(&kvm->vm_list, &vm_list);
600
	mutex_unlock(&kvm_lock);
601

602 603
	preempt_notifier_inc();

604
	return kvm;
605 606

out_err:
607 608
	cleanup_srcu_struct(&kvm->irq_srcu);
out_err_no_irq_srcu:
609
	cleanup_srcu_struct(&kvm->srcu);
610
out_err_no_srcu:
611
	hardware_disable_all();
612
out_err_no_disable:
613 614
	for (i = 0; i < KVM_NR_BUSES; i++)
		kfree(kvm->buses[i]);
615 616
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++)
		kvm_free_memslots(kvm, kvm->memslots[i]);
617
	kvm_arch_free_vm(kvm);
618
	mmdrop(current->mm);
619
	return ERR_PTR(r);
620 621
}

622 623 624 625
/*
 * Avoid using vmalloc for a small buffer.
 * Should not be used when the size is statically known.
 */
626
void *kvm_kvzalloc(unsigned long size)
627 628 629 630 631 632 633
{
	if (size > PAGE_SIZE)
		return vzalloc(size);
	else
		return kzalloc(size, GFP_KERNEL);
}

634 635 636 637 638 639 640 641 642 643 644 645 646
static void kvm_destroy_devices(struct kvm *kvm)
{
	struct list_head *node, *tmp;

	list_for_each_safe(node, tmp, &kvm->devices) {
		struct kvm_device *dev =
			list_entry(node, struct kvm_device, vm_node);

		list_del(node);
		dev->ops->destroy(dev);
	}
}

647 648
static void kvm_destroy_vm(struct kvm *kvm)
{
649
	int i;
650 651
	struct mm_struct *mm = kvm->mm;

652
	kvm_arch_sync_events(kvm);
653
	mutex_lock(&kvm_lock);
654
	list_del(&kvm->vm_list);
655
	mutex_unlock(&kvm_lock);
656
	kvm_free_irq_routing(kvm);
657
	for (i = 0; i < KVM_NR_BUSES; i++) {
658 659
		if (kvm->buses[i])
			kvm_io_bus_destroy(kvm->buses[i]);
660 661
		kvm->buses[i] = NULL;
	}
662
	kvm_coalesced_mmio_free(kvm);
663 664
#if defined(CONFIG_MMU_NOTIFIER) && defined(KVM_ARCH_WANT_MMU_NOTIFIER)
	mmu_notifier_unregister(&kvm->mmu_notifier, kvm->mm);
665
#else
666
	kvm_arch_flush_shadow_all(kvm);
667
#endif
668
	kvm_arch_destroy_vm(kvm);
669
	kvm_destroy_devices(kvm);
670 671
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++)
		kvm_free_memslots(kvm, kvm->memslots[i]);
672
	cleanup_srcu_struct(&kvm->irq_srcu);
673 674
	cleanup_srcu_struct(&kvm->srcu);
	kvm_arch_free_vm(kvm);
675
	preempt_notifier_dec();
676
	hardware_disable_all();
677
	mmdrop(mm);
678 679
}

Izik Eidus's avatar
Izik Eidus committed
680 681 682 683 684 685 686 687 688 689 690 691 692 693
void kvm_get_kvm(struct kvm *kvm)
{
	atomic_inc(&kvm->users_count);
}
EXPORT_SYMBOL_GPL(kvm_get_kvm);

void kvm_put_kvm(struct kvm *kvm)
{
	if (atomic_dec_and_test(&kvm->users_count))
		kvm_destroy_vm(kvm);
}
EXPORT_SYMBOL_GPL(kvm_put_kvm);


694 695 696 697
static int kvm_vm_release(struct inode *inode, struct file *filp)
{
	struct kvm *kvm = filp->private_data;

Gregory Haskins's avatar
Gregory Haskins committed
698 699
	kvm_irqfd_release(kvm);

Izik Eidus's avatar
Izik Eidus committed
700
	kvm_put_kvm(kvm);
701 702 703
	return 0;
}

704 705
/*
 * Allocation size is twice as large as the actual dirty bitmap size.
706
 * See x86's kvm_vm_ioctl_get_dirty_log() why this is needed.
707
 */
708 709
static int kvm_create_dirty_bitmap(struct kvm_memory_slot *memslot)
{
710
	unsigned long dirty_bytes = 2 * kvm_dirty_bitmap_bytes(memslot);
711

712
	memslot->dirty_bitmap = kvm_kvzalloc(dirty_bytes);
713 714 715 716 717 718
	if (!memslot->dirty_bitmap)
		return -ENOMEM;

	return 0;
}

719
/*
720 721 722 723
 * Insert memslot and re-sort memslots based on their GFN,
 * so binary search could be used to lookup GFN.
 * Sorting algorithm takes advantage of having initially
 * sorted array and known changed memslot position.
724
 */
725 726
static void update_memslots(struct kvm_memslots *slots,
			    struct kvm_memory_slot *new)
727
{
728 729
	int id = new->id;
	int i = slots->id_to_index[id];
730
	struct kvm_memory_slot *mslots = slots->memslots;
731

732
	WARN_ON(mslots[i].id != id);
733
	if (!new->npages) {
734
		WARN_ON(!mslots[i].npages);
735 736 737 738 739 740
		if (mslots[i].npages)
			slots->used_slots--;
	} else {
		if (!mslots[i].npages)
			slots->used_slots++;
	}
741

742
	while (i < KVM_MEM_SLOTS_NUM - 1 &&
743 744 745
	       new->base_gfn <= mslots[i + 1].base_gfn) {
		if (!mslots[i + 1].npages)
			break;
746 747 748 749
		mslots[i] = mslots[i + 1];
		slots->id_to_index[mslots[i].id] = i;
		i++;
	}
750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766

	/*
	 * The ">=" is needed when creating a slot with base_gfn == 0,
	 * so that it moves before all those with base_gfn == npages == 0.
	 *
	 * On the other hand, if new->npages is zero, the above loop has
	 * already left i pointing to the beginning of the empty part of
	 * mslots, and the ">=" would move the hole backwards in this
	 * case---which is wrong.  So skip the loop when deleting a slot.
	 */
	if (new->npages) {
		while (i > 0 &&
		       new->base_gfn >= mslots[i - 1].base_gfn) {
			mslots[i] = mslots[i - 1];
			slots->id_to_index[mslots[i].id] = i;
			i--;
		}
767 768
	} else
		WARN_ON_ONCE(i != slots->used_slots);
769

770 771
	mslots[i] = *new;
	slots->id_to_index[mslots[i].id] = i;
772 773
}

774
static int check_memory_region_flags(const struct kvm_userspace_memory_region *mem)
775
{
776 777
	u32 valid_flags = KVM_MEM_LOG_DIRTY_PAGES;

778
#ifdef __KVM_HAVE_READONLY_MEM
779 780 781 782
	valid_flags |= KVM_MEM_READONLY;
#endif

	if (mem->flags & ~valid_flags)
783 784 785 786 787
		return -EINVAL;

	return 0;
}

788
static struct kvm_memslots *install_new_memslots(struct kvm *kvm,
789
		int as_id, struct kvm_memslots *slots)
790
{
791
	struct kvm_memslots *old_memslots = __kvm_memslots(kvm, as_id);
792

793 794 795 796 797 798 799
	/*
	 * Set the low bit in the generation, which disables SPTE caching
	 * until the end of synchronize_srcu_expedited.
	 */
	WARN_ON(old_memslots->generation & 1);
	slots->generation = old_memslots->generation + 1;

800
	rcu_assign_pointer(kvm->memslots[as_id], slots);
801
	synchronize_srcu_expedited(&kvm->srcu);
802

803 804 805 806 807 808 809
	/*
	 * Increment the new memslot generation a second time. This prevents
	 * vm exits that race with memslot updates from caching a memslot
	 * generation that will (potentially) be valid forever.
	 */
	slots->generation++;

810
	kvm_arch_memslots_updated(kvm, slots);
811 812

	return old_memslots;
813 814
}

815 816 817 818 819
/*
 * Allocate some memory and give it an address in the guest physical address
 * space.
 *
 * Discontiguous memory is allowed, mostly for framebuffers.
820
 *
821
 * Must be called holding kvm->slots_lock for write.
822
 */
823
int __kvm_set_memory_region(struct kvm *kvm,
824
			    const struct kvm_userspace_memory_region *mem)
825
{
826
	int r;
827
	gfn_t base_gfn;
828
	unsigned long npages;
829
	struct kvm_memory_slot *slot;
830
	struct kvm_memory_slot old, new;
831
	struct kvm_memslots *slots = NULL, *old_memslots;
832
	int as_id, id;
833
	enum kvm_mr_change change;
834

835 836 837 838
	r = check_memory_region_flags(mem);
	if (r)
		goto out;

839
	r = -EINVAL;
840 841 842
	as_id = mem->slot >> 16;
	id = (u16)mem->slot;

843 844 845 846 847
	/* General sanity checks */
	if (mem->memory_size & (PAGE_SIZE - 1))
		goto out;
	if (mem->guest_phys_addr & (PAGE_SIZE - 1))
		goto out;
848
	/* We can read the guest memory with __xxx_user() later on. */
849
	if ((id < KVM_USER_MEM_SLOTS) &&
850
	    ((mem->userspace_addr & (PAGE_SIZE - 1)) ||
851 852 853
	     !access_ok(VERIFY_WRITE,
			(void __user *)(unsigned long)mem->userspace_addr,
			mem->memory_size)))
854
		goto out;
855
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_MEM_SLOTS_NUM)
856 857 858 859
		goto out;
	if (mem->guest_phys_addr + mem->memory_size < mem->guest_phys_addr)
		goto out;

860
	slot = id_to_memslot(__kvm_memslots(kvm, as_id), id);
861 862 863
	base_gfn = mem->guest_phys_addr >> PAGE_SHIFT;
	npages = mem->memory_size >> PAGE_SHIFT;

864 865 866
	if (npages > KVM_MEM_MAX_NR_PAGES)
		goto out;

867
	new = old = *slot;
868

869
	new.id = id;
870 871 872 873
	new.base_gfn = base_gfn;
	new.npages = npages;
	new.flags = mem->flags;

874 875 876 877 878
	if (npages) {
		if (!old.npages)
			change = KVM_MR_CREATE;
		else { /* Modify an existing slot. */
			if ((mem->userspace_addr != old.userspace_addr) ||
879 880
			    (npages != old.npages) ||
			    ((new.flags ^ old.flags) & KVM_MEM_READONLY))
881 882 883 884 885 886 887 888 889 890 891
				goto out;

			if (base_gfn != old.base_gfn)
				change = KVM_MR_MOVE;
			else if (new.flags != old.flags)
				change = KVM_MR_FLAGS_ONLY;
			else { /* Nothing to change. */
				r = 0;
				goto out;
			}
		}
892 893 894 895
	} else {
		if (!old.npages)
			goto out;

896
		change = KVM_MR_DELETE;
897 898 899
		new.base_gfn = 0;
		new.flags = 0;
	}
900

901
	if ((change == KVM_MR_CREATE) || (change == KVM_MR_MOVE)) {
902 903
		/* Check for overlaps */
		r = -EEXIST;
904
		kvm_for_each_memslot(slot, __kvm_memslots(kvm, as_id)) {
905
			if (slot->id == id)
906 907 908 909 910
				continue;
			if (!((base_gfn + npages <= slot->base_gfn) ||
			      (base_gfn >= slot->base_gfn + slot->npages)))
				goto out;
		}
911 912 913 914
	}

	/* Free page dirty bitmap if unneeded */
	if (!(new.flags & KVM_MEM_LOG_DIRTY_PAGES))
Al Viro's avatar
Al Viro committed
915
		new.dirty_bitmap = NULL;
916 917

	r = -ENOMEM;
918
	if (change == KVM_MR_CREATE) {
919
		new.userspace_addr = mem->userspace_addr;
920

921
		if (kvm_arch_create_memslot(kvm, &new, npages))
922
			goto out_free;
923
	}
924

925 926
	/* Allocate page dirty bitmap if needed */
	if ((new.flags & KVM_MEM_LOG_DIRTY_PAGES) && !new.dirty_bitmap) {
927
		if (kvm_create_dirty_bitmap(&new) < 0)
928
			goto out_free;
929 930
	}

931
	slots = kvm_kvzalloc(sizeof(struct kvm_memslots));
932 933
	if (!slots)
		goto out_free;
934
	memcpy(slots, __kvm_memslots(kvm, as_id), sizeof(struct kvm_memslots));
935

936
	if ((change == KVM_MR_DELETE) || (change == KVM_MR_MOVE)) {
937
		slot = id_to_memslot(slots, id);
938 939
		slot->flags |= KVM_MEMSLOT_INVALID;

940
		old_memslots = install_new_memslots(kvm, as_id, slots);
941

942 943
		/* slot was deleted or moved, clear iommu mapping */
		kvm_iommu_unmap_pages(kvm, &old);
944 945
		/* From this point no new shadow pages pointing to a deleted,
		 * or moved, memslot will be created.
946 947
		 *
		 * validation of sp->gfn happens in:
948 949
		 *	- gfn_to_hva (kvm_read_guest, gfn_to_pfn)
		 *	- kvm_is_visible_gfn (mmu_check_roots)
950
		 */
951
		kvm_arch_flush_shadow_memslot(kvm, slot);
952 953 954 955 956 957

		/*
		 * We can re-use the old_memslots from above, the only difference
		 * from the currently installed memslots is the invalid flag.  This
		 * will get overwritten by update_memslots anyway.
		 */
958
		slots = old_memslots;
959
	}
960

961
	r = kvm_arch_prepare_memory_region(kvm, &new, mem, change);
962
	if (r)
963
		goto out_slots;
964

965
	/* actual memory is freed via old in kvm_free_memslot below */
966
	if (change == KVM_MR_DELETE) {
967
		new.dirty_bitmap = NULL;
968
		memset(&new.arch, 0, sizeof(new.arch));
969 970
	}

971
	update_memslots(slots, &new);
972
	old_memslots = install_new_memslots(kvm, as_id, slots);
973

974
	kvm_arch_commit_memory_region(kvm, mem, &old, &new, change);
975

976
	kvm_free_memslot(kvm, &old, &new);
977
	kvfree(old_memslots);
978

979 980
	/*
	 * IOMMU mapping:  New slots need to be mapped.  Old slots need to be
981 982 983 984 985 986
	 * un-mapped and re-mapped if their base changes.  Since base change
	 * unmapping is handled above with slot deletion, mapping alone is
	 * needed here.  Anything else the iommu might care about for existing
	 * slots (size changes, userspace addr changes and read-only flag
	 * changes) is disallowed above, so any other attribute changes getting
	 * here can be skipped.
987
	 */
988
	if (as_id == 0 && (change == KVM_MR_CREATE || change == KVM_MR_MOVE)) {
989
		r = kvm_iommu_map_pages(kvm, &new);
990
		return r;
991 992
	}

993 994
	return 0;

995
out_slots:
996
	kvfree(slots);
997
out_free:
998
	kvm_free_memslot(kvm, &new, &old);
999 1000
out:
	return r;
1001
}
1002 1003 1004
EXPORT_SYMBOL_GPL(__kvm_set_memory_region);

int kvm_set_memory_region(struct kvm *kvm,
1005
			  const struct kvm_userspace_memory_region *mem)
1006 1007 1008
{
	int r;

1009
	mutex_lock(&kvm->slots_lock);
1010
	r = __kvm_set_memory_region(kvm, mem);
1011
	mutex_unlock(&kvm->slots_lock);
1012 1013
	return r;
}
1014 1015
EXPORT_SYMBOL_GPL(kvm_set_memory_region);

1016 1017
static int kvm_vm_ioctl_set_memory_region(struct kvm *kvm,
					  struct kvm_userspace_memory_region *mem)
1018
{
1019
	if ((u16)mem->slot >= KVM_USER_MEM_SLOTS)
1020
		return -EINVAL;
1021

1022
	return kvm_set_memory_region(kvm, mem);
1023 1024
}

1025 1026
int kvm_get_dirty_log(struct kvm *kvm,
			struct kvm_dirty_log *log, int *is_dirty)
1027
{
1028
	struct kvm_memslots *slots;
1029
	struct kvm_memory_slot *memslot;
1030
	int r, i, as_id, id;
1031
	unsigned long n;
1032 1033 1034
	unsigned long any = 0;

	r = -EINVAL;
1035 1036 1037
	as_id = log->slot >> 16;
	id = (u16)log->slot;
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_USER_MEM_SLOTS)
1038 1039
		goto out;

1040 1041
	slots = __kvm_memslots(kvm, as_id);
	memslot = id_to_memslot(slots, id);
1042 1043 1044 1045
	r = -ENOENT;
	if (!memslot->dirty_bitmap)
		goto out;

1046
	n = kvm_dirty_bitmap_bytes(memslot);
1047

1048
	for (i = 0; !any && i < n/sizeof(long); ++i)
1049 1050 1051 1052 1053 1054
		any = memslot->dirty_bitmap[i];

	r = -EFAULT;
	if (copy_to_user(log->dirty_bitmap, memslot->dirty_bitmap, n))
		goto out;

1055 1056
	if (any)
		*is_dirty = 1;
1057 1058 1059 1060 1061

	r = 0;
out:
	return r;
}
1062
EXPORT_SYMBOL_GPL(kvm_get_dirty_log);
1063

1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089
#ifdef CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT
/**
 * kvm_get_dirty_log_protect - get a snapshot of dirty pages, and if any pages
 *	are dirty write protect them for next write.
 * @kvm:	pointer to kvm instance
 * @log:	slot id and address to which we copy the log
 * @is_dirty:	flag set if any page is dirty
 *
 * We need to keep it in mind that VCPU threads can write to the bitmap
 * concurrently. So, to avoid losing track of dirty pages we keep the
 * following order:
 *
 *    1. Take a snapshot of the bit and clear it if needed.
 *    2. Write protect the corresponding page.
 *    3. Copy the snapshot to the userspace.
 *    4. Upon return caller flushes TLB's if needed.
 *
 * Between 2 and 4, the guest may write to the page using the remaining TLB
 * entry.  This is not a problem because the page is reported dirty using
 * the snapshot taken before and step 4 ensures that writes done after
 * exiting to userspace will be logged for the next call.
 *
 */
int kvm_get_dirty_log_protect(struct kvm *kvm,
			struct kvm_dirty_log *log, bool *is_dirty)
{
1090
	struct kvm_memslots *slots;
1091
	struct kvm_memory_slot *memslot;
1092
	int r, i, as_id, id;
1093 1094 1095 1096 1097
	unsigned long n;
	unsigned long *dirty_bitmap;
	unsigned long *dirty_bitmap_buffer;

	r = -EINVAL;
1098 1099 1100
	as_id = log->slot >> 16;
	id = (u16)log->slot;
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_USER_MEM_SLOTS)
1101 1102
		goto out;

1103 1104
	slots = __kvm_memslots(kvm, as_id);
	memslot = id_to_memslot(slots, id);
1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129

	dirty_bitmap = memslot->dirty_bitmap;
	r = -ENOENT;
	if (!dirty_bitmap)
		goto out;

	n = kvm_dirty_bitmap_bytes(memslot);

	dirty_bitmap_buffer = dirty_bitmap + n / sizeof(long);
	memset(dirty_bitmap_buffer, 0, n);

	spin_lock(&kvm->mmu_lock);
	*is_dirty = false;
	for (i = 0; i < n / sizeof(long); i++) {
		unsigned long mask;
		gfn_t offset;

		if (!dirty_bitmap[i])
			continue;

		*is_dirty = true;

		mask = xchg(&dirty_bitmap[i], 0);
		dirty_bitmap_buffer[i] = mask;

1130 1131 1132 1133 1134
		if (mask) {
			offset = i * BITS_PER_LONG;
			kvm_arch_mmu_enable_log_dirty_pt_masked(kvm, memslot,
								offset, mask);
		}
1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149
	}

	spin_unlock(&kvm->mmu_lock);

	r = -EFAULT;
	if (copy_to_user(log->dirty_bitmap, dirty_bitmap_buffer, n))
		goto out;

	r = 0;
out:
	return r;
}
EXPORT_SYMBOL_GPL(kvm_get_dirty_log_protect);
#endif

1150 1151 1152 1153 1154
bool kvm_largepages_enabled(void)
{
	return largepages_enabled;
}

1155 1156 1157 1158 1159 1160
void kvm_disable_largepages(void)
{
	largepages_enabled = false;
}
EXPORT_SYMBOL_GPL(kvm_disable_largepages);

1161 1162 1163 1164
struct kvm_memory_slot *gfn_to_memslot(struct kvm *kvm, gfn_t gfn)
{
	return __gfn_to_memslot(kvm_memslots(kvm), gfn);
}
1165
EXPORT_SYMBOL_GPL(gfn_to_memslot);
1166

1167 1168 1169 1170 1171
struct kvm_memory_slot *kvm_vcpu_gfn_to_memslot(struct kvm_vcpu *vcpu, gfn_t gfn)
{
	return __gfn_to_memslot(kvm_vcpu_memslots(vcpu), gfn);
}

1172 1173
int kvm_is_visible_gfn(struct kvm *kvm, gfn_t gfn)
{
1174
	struct kvm_memory_slot *memslot = gfn_to_memslot(kvm, gfn);
1175

1176
	if (!memslot || memslot->id >= KVM_USER_MEM_SLOTS ||
1177 1178
	      memslot->flags & KVM_MEMSLOT_INVALID)
		return 0;
1179

1180
	return 1;
1181 1182 1183
}
EXPORT_SYMBOL_GPL(kvm_is_visible_gfn);

1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207
unsigned long kvm_host_page_size(struct kvm *kvm, gfn_t gfn)
{
	struct vm_area_struct *vma;
	unsigned long addr, size;

	size = PAGE_SIZE;

	addr = gfn_to_hva(kvm, gfn);
	if (kvm_is_error_hva(addr))
		return PAGE_SIZE;

	down_read(&current->mm->mmap_sem);
	vma = find_vma(current->mm, addr);
	if (!vma)
		goto out;

	size = vma_kernel_pagesize(vma);

out:
	up_read(&current->mm->mmap_sem);

	return size;
}

1208 1209 1210 1211 1212 1213 1214
static bool memslot_is_readonly(struct kvm_memory_slot *slot)
{
	return slot->flags & KVM_MEM_READONLY;
}

static unsigned long __gfn_to_hva_many(struct kvm_memory_slot *slot, gfn_t gfn,
				       gfn_t *nr_pages, bool write)
Izik Eidus's avatar
Izik Eidus committed
1215
{
1216
	if (!slot || slot->flags & KVM_MEMSLOT_INVALID)
1217
		return KVM_HVA_ERR_BAD;
1218

1219 1220
	if (memslot_is_readonly(slot) && write)
		return KVM_HVA_ERR_RO_BAD;
1221 1222 1223 1224

	if (nr_pages)
		*nr_pages = slot->npages - (gfn - slot->base_gfn);

1225
	return __gfn_to_hva_memslot(slot, gfn);
Izik Eidus's avatar
Izik Eidus committed
1226
}
1227

1228 1229 1230 1231
static unsigned long gfn_to_hva_many(struct kvm_memory_slot *slot, gfn_t gfn,
				     gfn_t *nr_pages)
{
	return __gfn_to_hva_many(slot, gfn, nr_pages, true);
Izik Eidus's avatar
Izik Eidus committed
1232
}
1233

1234
unsigned long gfn_to_hva_memslot(struct kvm_memory_slot *slot,
1235
					gfn_t gfn)
1236 1237 1238 1239 1240
{
	return gfn_to_hva_many(slot, gfn, NULL);
}
EXPORT_SYMBOL_GPL(gfn_to_hva_memslot);

1241 1242
unsigned long gfn_to_hva(struct kvm *kvm, gfn_t gfn)
{
1243
	return gfn_to_hva_many(gfn_to_memslot(kvm, gfn), gfn, NULL);
1244
}
1245
EXPORT_SYMBOL_GPL(gfn_to_hva);
Izik Eidus's avatar
Izik Eidus committed
1246

1247 1248 1249 1250 1251 1252
unsigned long kvm_vcpu_gfn_to_hva(struct kvm_vcpu *vcpu, gfn_t gfn)
{
	return gfn_to_hva_many(kvm_vcpu_gfn_to_memslot(vcpu, gfn), gfn, NULL);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_gfn_to_hva);

1253
/*
1254 1255
 * If writable is set to false, the hva returned by this function is only
 * allowed to be read.
1256
 */
1257 1258
unsigned long gfn_to_hva_memslot_prot(struct kvm_memory_slot *slot,
				      gfn_t gfn, bool *writable)
1259
{
1260 1261 1262
	unsigned long hva = __gfn_to_hva_many(slot, gfn, NULL, false);

	if (!kvm_is_error_hva(hva) && writable)
1263 1264
		*writable = !memslot_is_readonly(slot);

1265
	return hva;
1266 1267
}

1268 1269 1270 1271 1272 1273 1274
unsigned long gfn_to_hva_prot(struct kvm *kvm, gfn_t gfn, bool *writable)
{
	struct kvm_memory_slot *slot = gfn_to_memslot(kvm, gfn);

	return gfn_to_hva_memslot_prot(slot, gfn, writable);
}

1275 1276 1277 1278 1279 1280 1281
unsigned long kvm_vcpu_gfn_to_hva_prot(struct kvm_vcpu *vcpu, gfn_t gfn, bool *writable)
{
	struct kvm_memory_slot *slot = kvm_vcpu_gfn_to_memslot(vcpu, gfn);

	return gfn_to_hva_memslot_prot(slot, gfn, writable);
}

1282
static int get_user_page_nowait(struct task_struct *tsk, struct mm_struct *mm,
1283 1284 1285 1286 1287 1288 1289 1290 1291 1292
	unsigned long start, int write, struct page **page)
{
	int flags = FOLL_TOUCH | FOLL_NOWAIT | FOLL_HWPOISON | FOLL_GET;

	if (write)
		flags |= FOLL_WRITE;

	return __get_user_pages(tsk, mm, start, 1, flags, page, NULL, NULL);
}

1293 1294 1295 1296 1297 1298 1299 1300 1301
static inline int check_user_page_hwpoison(unsigned long addr)
{
	int rc, flags = FOLL_TOUCH | FOLL_HWPOISON | FOLL_WRITE;

	rc = __get_user_pages(current, current->mm, addr, 1,
			      flags, NULL, NULL, NULL);
	return rc == -EHWPOISON;
}

Xiao Guangrong's avatar
Xiao Guangrong committed
1302 1303 1304 1305 1306 1307
/*
 * The atomic path to get the writable pfn which will be stored in @pfn,
 * true indicates success, otherwise false is returned.
 */
static bool hva_to_pfn_fast(unsigned long addr, bool atomic, bool *async,
			    bool write_fault, bool *writable, pfn_t *pfn)
Avi Kivity's avatar
Avi Kivity committed
1308
{
1309
	struct page *page[1];
Xiao Guangrong's avatar
Xiao Guangrong committed
1310
	int npages;
Avi Kivity's avatar
Avi Kivity committed
1311

Xiao Guangrong's avatar
Xiao Guangrong committed
1312 1313
	if (!(async || atomic))
		return false;
1314

1315 1316 1317 1318 1319 1320 1321
	/*
	 * Fast pin a writable pfn only if it is a write fault request
	 * or the caller allows to map a writable pfn for a read fault
	 * request.
	 */
	if (!(write_fault || writable))
		return false;
1322

Xiao Guangrong's avatar
Xiao Guangrong committed
1323 1324 1325
	npages = __get_user_pages_fast(addr, 1, 1, page);
	if (npages == 1) {
		*pfn = page_to_pfn(page[0]);
1326

Xiao Guangrong's avatar
Xiao Guangrong committed
1327 1328 1329 1330
		if (writable)
			*writable = true;
		return true;
	}
1331

Xiao Guangrong's avatar
Xiao Guangrong committed
1332 1333
	return false;
}
1334

Xiao Guangrong's avatar
Xiao Guangrong committed
1335 1336 1337 1338 1339 1340 1341 1342 1343
/*
 * The slow path to get the pfn of the specified host virtual address,
 * 1 indicates success, -errno is returned if error is detected.
 */
static int hva_to_pfn_slow(unsigned long addr, bool *async, bool write_fault,
			   bool *writable, pfn_t *pfn)
{
	struct page *page[1];
	int npages = 0;
1344

Xiao Guangrong's avatar
Xiao Guangrong committed
1345 1346 1347 1348 1349 1350 1351 1352 1353 1354
	might_sleep();

	if (writable)
		*writable = write_fault;

	if (async) {
		down_read(&current->mm->mmap_sem);
		npages = get_user_page_nowait(current, current->mm,
					      addr, write_fault, page);
		up_read(&current->mm->mmap_sem);
1355 1356 1357 1358 1359 1360
	} else {
		unsigned int flags = FOLL_TOUCH | FOLL_HWPOISON;

		if (write_fault)
			flags |= FOLL_WRITE;

1361
		npages = __get_user_pages_unlocked(current, current->mm, addr, 1,
1362 1363
						   page, flags);
	}
Xiao Guangrong's avatar
Xiao Guangrong committed
1364 1365 1366 1367
	if (npages != 1)
		return npages;

	/* map read fault as writable if possible */
1368
	if (unlikely(!write_fault) && writable) {
Xiao Guangrong's avatar
Xiao Guangrong committed
1369 1370 1371 1372 1373 1374 1375
		struct page *wpage[1];

		npages = __get_user_pages_fast(addr, 1, 1, wpage);
		if (npages == 1) {
			*writable = true;
			put_page(page[0]);
			page[0] = wpage[0];
1376
		}
Xiao Guangrong's avatar
Xiao Guangrong committed
1377 1378

		npages = 1;
1379
	}
Xiao Guangrong's avatar
Xiao Guangrong committed
1380 1381 1382
	*pfn = page_to_pfn(page[0]);
	return npages;
}
Izik Eidus's avatar
Izik Eidus committed
1383

1384 1385 1386 1387
static bool vma_is_valid(struct vm_area_struct *vma, bool write_fault)
{
	if (unlikely(!(vma->vm_flags & VM_READ)))
		return false;
1388

1389 1390
	if (write_fault && (unlikely(!(vma->vm_flags & VM_WRITE))))
		return false;
1391

1392 1393
	return true;
}
1394

1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408
/*
 * Pin guest page in memory and return its pfn.
 * @addr: host virtual address which maps memory to the guest
 * @atomic: whether this function can sleep
 * @async: whether this function need to wait IO complete if the
 *         host page is not in the memory
 * @write_fault: whether we should get a writable host page
 * @writable: whether it allows to map a writable host page for !@write_fault
 *
 * The function will map a writable host page for these two cases:
 * 1): @write_fault = true
 * 2): @write_fault = false && @writable, @writable will tell the caller
 *     whether the mapping is writable.
 */
Xiao Guangrong's avatar
Xiao Guangrong committed
1409 1410 1411 1412 1413 1414
static pfn_t hva_to_pfn(unsigned long addr, bool atomic, bool *async,
			bool write_fault, bool *writable)
{
	struct vm_area_struct *vma;
	pfn_t pfn = 0;
	int npages;
1415

Xiao Guangrong's avatar
Xiao Guangrong committed
1416 1417
	/* we can do it either atomically or asynchronously, not both */
	BUG_ON(atomic && async);
1418

Xiao Guangrong's avatar
Xiao Guangrong committed
1419 1420 1421 1422 1423 1424 1425 1426 1427
	if (hva_to_pfn_fast(addr, atomic, async, write_fault, writable, &pfn))
		return pfn;

	if (atomic)
		return KVM_PFN_ERR_FAULT;

	npages = hva_to_pfn_slow(addr, async, write_fault, writable, &pfn);
	if (npages == 1)
		return pfn;
1428

Xiao Guangrong's avatar
Xiao Guangrong committed
1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442
	down_read(&current->mm->mmap_sem);
	if (npages == -EHWPOISON ||
	      (!async && check_user_page_hwpoison(addr))) {
		pfn = KVM_PFN_ERR_HWPOISON;
		goto exit;
	}

	vma = find_vma_intersection(current->mm, addr, addr + 1);

	if (vma == NULL)
		pfn = KVM_PFN_ERR_FAULT;
	else if ((vma->vm_flags & VM_PFNMAP)) {
		pfn = ((addr - vma->vm_start) >> PAGE_SHIFT) +
			vma->vm_pgoff;
1443
		BUG_ON(!kvm_is_reserved_pfn(pfn));
Xiao Guangrong's avatar
Xiao Guangrong committed
1444
	} else {
1445
		if (async && vma_is_valid(vma, write_fault))
Xiao Guangrong's avatar
Xiao Guangrong committed
1446 1447 1448 1449 1450
			*async = true;
		pfn = KVM_PFN_ERR_FAULT;
	}
exit:
	up_read(&current->mm->mmap_sem);
1451
	return pfn;
1452 1453
}

1454 1455
pfn_t __gfn_to_pfn_memslot(struct kvm_memory_slot *slot, gfn_t gfn, bool atomic,
			   bool *async, bool write_fault, bool *writable)
1456
{
1457 1458 1459 1460 1461 1462
	unsigned long addr = __gfn_to_hva_many(slot, gfn, NULL, write_fault);

	if (addr == KVM_HVA_ERR_RO_BAD)
		return KVM_PFN_ERR_RO_FAULT;

	if (kvm_is_error_hva(addr))
1463
		return KVM_PFN_NOSLOT;
1464 1465 1466 1467 1468 1469 1470 1471 1472

	/* Do not map writable pfn in the readonly memslot. */
	if (writable && memslot_is_readonly(slot)) {
		*writable = false;
		writable = NULL;
	}

	return hva_to_pfn(addr, atomic, async, write_fault,
			  writable);
1473
}
1474
EXPORT_SYMBOL_GPL(__gfn_to_pfn_memslot);
1475

1476 1477 1478
pfn_t gfn_to_pfn_prot(struct kvm *kvm, gfn_t gfn, bool write_fault,
		      bool *writable)
{
Paolo Bonzini's avatar
Paolo Bonzini committed
1479 1480
	return __gfn_to_pfn_memslot(gfn_to_memslot(kvm, gfn), gfn, false, NULL,
				    write_fault, writable);
1481 1482 1483
}
EXPORT_SYMBOL_GPL(gfn_to_pfn_prot);

1484
pfn_t gfn_to_pfn_memslot(struct kvm_memory_slot *slot, gfn_t gfn)
1485
{
1486
	return __gfn_to_pfn_memslot(slot, gfn, false, NULL, true, NULL);
1487
}
Paolo Bonzini's avatar
Paolo Bonzini committed
1488
EXPORT_SYMBOL_GPL(gfn_to_pfn_memslot);
1489

1490
pfn_t gfn_to_pfn_memslot_atomic(struct kvm_memory_slot *slot, gfn_t gfn)
1491
{
1492
	return __gfn_to_pfn_memslot(slot, gfn, true, NULL, true, NULL);
1493
}
1494
EXPORT_SYMBOL_GPL(gfn_to_pfn_memslot_atomic);
1495

Paolo Bonzini's avatar
Paolo Bonzini committed
1496 1497 1498 1499 1500 1501
pfn_t gfn_to_pfn_atomic(struct kvm *kvm, gfn_t gfn)
{
	return gfn_to_pfn_memslot_atomic(gfn_to_memslot(kvm, gfn), gfn);
}
EXPORT_SYMBOL_GPL(gfn_to_pfn_atomic);

1502 1503 1504 1505 1506 1507
pfn_t kvm_vcpu_gfn_to_pfn_atomic(struct kvm_vcpu *vcpu, gfn_t gfn)
{
	return gfn_to_pfn_memslot_atomic(kvm_vcpu_gfn_to_memslot(vcpu, gfn), gfn);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_gfn_to_pfn_atomic);

Paolo Bonzini's avatar
Paolo Bonzini committed
1508 1509 1510 1511 1512 1513
pfn_t gfn_to_pfn(struct kvm *kvm, gfn_t gfn)
{
	return gfn_to_pfn_memslot(gfn_to_memslot(kvm, gfn), gfn);
}
EXPORT_SYMBOL_GPL(gfn_to_pfn);

1514 1515 1516 1517 1518 1519
pfn_t kvm_vcpu_gfn_to_pfn(struct kvm_vcpu *vcpu, gfn_t gfn)
{
	return gfn_to_pfn_memslot(kvm_vcpu_gfn_to_memslot(vcpu, gfn), gfn);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_gfn_to_pfn);

1520 1521
int gfn_to_page_many_atomic(struct kvm_memory_slot *slot, gfn_t gfn,
			    struct page **pages, int nr_pages)
1522 1523 1524 1525
{
	unsigned long addr;
	gfn_t entry;

1526
	addr = gfn_to_hva_many(slot, gfn, &entry);
1527 1528 1529 1530 1531 1532 1533 1534 1535 1536
	if (kvm_is_error_hva(addr))
		return -1;

	if (entry < nr_pages)
		return 0;

	return __get_user_pages_fast(addr, nr_pages, 1, pages);
}
EXPORT_SYMBOL_GPL(gfn_to_page_many_atomic);

Xiao Guangrong's avatar
Xiao Guangrong committed
1537 1538
static struct page *kvm_pfn_to_page(pfn_t pfn)
{
1539
	if (is_error_noslot_pfn(pfn))
1540
		return KVM_ERR_PTR_BAD_PAGE;
Xiao Guangrong's avatar
Xiao Guangrong committed
1541

1542
	if (kvm_is_reserved_pfn(pfn)) {
1543
		WARN_ON(1);
1544
		return KVM_ERR_PTR_BAD_PAGE;
1545
	}
Xiao Guangrong's avatar
Xiao Guangrong committed
1546 1547 1548 1549

	return pfn_to_page(pfn);
}

1550 1551
struct page *gfn_to_page(struct kvm *kvm, gfn_t gfn)
{
1552 1553 1554 1555
	pfn_t pfn;

	pfn = gfn_to_pfn(kvm, gfn);

Xiao Guangrong's avatar
Xiao Guangrong committed
1556
	return kvm_pfn_to_page(pfn);
Avi Kivity's avatar
Avi Kivity committed
1557 1558 1559
}
EXPORT_SYMBOL_GPL(gfn_to_page);

1560 1561 1562 1563 1564 1565 1566 1567 1568 1569
struct page *kvm_vcpu_gfn_to_page(struct kvm_vcpu *vcpu, gfn_t gfn)
{
	pfn_t pfn;

	pfn = kvm_vcpu_gfn_to_pfn(vcpu, gfn);

	return kvm_pfn_to_page(pfn);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_gfn_to_page);

1570 1571
void kvm_release_page_clean(struct page *page)
{
1572 1573
	WARN_ON(is_error_page(page));

1574
	kvm_release_pfn_clean(page_to_pfn(page));
1575 1576 1577
}
EXPORT_SYMBOL_GPL(kvm_release_page_clean);

1578 1579
void kvm_release_pfn_clean(pfn_t pfn)
{
1580
	if (!is_error_noslot_pfn(pfn) && !kvm_is_reserved_pfn(pfn))
1581
		put_page(pfn_to_page(pfn));
1582 1583 1584
}
EXPORT_SYMBOL_GPL(kvm_release_pfn_clean);

1585
void kvm_release_page_dirty(struct page *page)
1586
{
Xiao Guangrong's avatar
Xiao Guangrong committed
1587 1588
	WARN_ON(is_error_page(page));

1589 1590 1591 1592
	kvm_release_pfn_dirty(page_to_pfn(page));
}
EXPORT_SYMBOL_GPL(kvm_release_page_dirty);

1593
static void kvm_release_pfn_dirty(pfn_t pfn)
1594 1595 1596 1597 1598 1599 1600
{
	kvm_set_pfn_dirty(pfn);
	kvm_release_pfn_clean(pfn);
}

void kvm_set_pfn_dirty(pfn_t pfn)
{
1601
	if (!kvm_is_reserved_pfn(pfn)) {
1602
		struct page *page = pfn_to_page(pfn);
1603

1604 1605 1606
		if (!PageReserved(page))
			SetPageDirty(page);
	}
1607
}
1608 1609 1610 1611
EXPORT_SYMBOL_GPL(kvm_set_pfn_dirty);

void kvm_set_pfn_accessed(pfn_t pfn)
{
1612
	if (!kvm_is_reserved_pfn(pfn))
1613
		mark_page_accessed(pfn_to_page(pfn));
1614 1615 1616 1617 1618
}
EXPORT_SYMBOL_GPL(kvm_set_pfn_accessed);

void kvm_get_pfn(pfn_t pfn)
{
1619
	if (!kvm_is_reserved_pfn(pfn))
1620
		get_page(pfn_to_page(pfn));
1621 1622
}
EXPORT_SYMBOL_GPL(kvm_get_pfn);
1623

1624 1625 1626 1627 1628 1629 1630 1631
static int next_segment(unsigned long len, int offset)
{
	if (len > PAGE_SIZE - offset)
		return PAGE_SIZE - offset;
	else
		return len;
}

1632 1633
static int __kvm_read_guest_page(struct kvm_memory_slot *slot, gfn_t gfn,
				 void *data, int offset, int len)
1634
{
1635 1636
	int r;
	unsigned long addr;
1637

1638
	addr = gfn_to_hva_memslot_prot(slot, gfn, NULL);
1639 1640
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1641
	r = __copy_from_user(data, (void __user *)addr + offset, len);
1642
	if (r)
1643 1644 1645
		return -EFAULT;
	return 0;
}
1646 1647 1648 1649 1650 1651 1652 1653

int kvm_read_guest_page(struct kvm *kvm, gfn_t gfn, void *data, int offset,
			int len)
{
	struct kvm_memory_slot *slot = gfn_to_memslot(kvm, gfn);

	return __kvm_read_guest_page(slot, gfn, data, offset, len);
}
1654 1655
EXPORT_SYMBOL_GPL(kvm_read_guest_page);

1656 1657 1658 1659 1660 1661 1662 1663 1664
int kvm_vcpu_read_guest_page(struct kvm_vcpu *vcpu, gfn_t gfn, void *data,
			     int offset, int len)
{
	struct kvm_memory_slot *slot = kvm_vcpu_gfn_to_memslot(vcpu, gfn);

	return __kvm_read_guest_page(slot, gfn, data, offset, len);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_read_guest_page);

1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684
int kvm_read_guest(struct kvm *kvm, gpa_t gpa, void *data, unsigned long len)
{
	gfn_t gfn = gpa >> PAGE_SHIFT;
	int seg;
	int offset = offset_in_page(gpa);
	int ret;

	while ((seg = next_segment(len, offset)) != 0) {
		ret = kvm_read_guest_page(kvm, gfn, data, offset, seg);
		if (ret < 0)
			return ret;
		offset = 0;
		len -= seg;
		data += seg;
		++gfn;
	}
	return 0;
}
EXPORT_SYMBOL_GPL(kvm_read_guest);

1685
int kvm_vcpu_read_guest(struct kvm_vcpu *vcpu, gpa_t gpa, void *data, unsigned long len)
1686 1687
{
	gfn_t gfn = gpa >> PAGE_SHIFT;
1688
	int seg;
1689
	int offset = offset_in_page(gpa);
1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703
	int ret;

	while ((seg = next_segment(len, offset)) != 0) {
		ret = kvm_vcpu_read_guest_page(vcpu, gfn, data, offset, seg);
		if (ret < 0)
			return ret;
		offset = 0;
		len -= seg;
		data += seg;
		++gfn;
	}
	return 0;
}
EXPORT_SYMBOL_GPL(kvm_vcpu_read_guest);
1704

1705 1706 1707 1708 1709 1710 1711
static int __kvm_read_guest_atomic(struct kvm_memory_slot *slot, gfn_t gfn,
			           void *data, int offset, unsigned long len)
{
	int r;
	unsigned long addr;

	addr = gfn_to_hva_memslot_prot(slot, gfn, NULL);
1712 1713
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1714
	pagefault_disable();
1715
	r = __copy_from_user_inatomic(data, (void __user *)addr + offset, len);
1716
	pagefault_enable();
1717 1718 1719 1720 1721
	if (r)
		return -EFAULT;
	return 0;
}

1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745
int kvm_read_guest_atomic(struct kvm *kvm, gpa_t gpa, void *data,
			  unsigned long len)
{
	gfn_t gfn = gpa >> PAGE_SHIFT;
	struct kvm_memory_slot *slot = gfn_to_memslot(kvm, gfn);
	int offset = offset_in_page(gpa);

	return __kvm_read_guest_atomic(slot, gfn, data, offset, len);
}
EXPORT_SYMBOL_GPL(kvm_read_guest_atomic);

int kvm_vcpu_read_guest_atomic(struct kvm_vcpu *vcpu, gpa_t gpa,
			       void *data, unsigned long len)
{
	gfn_t gfn = gpa >> PAGE_SHIFT;
	struct kvm_memory_slot *slot = kvm_vcpu_gfn_to_memslot(vcpu, gfn);
	int offset = offset_in_page(gpa);

	return __kvm_read_guest_atomic(slot, gfn, data, offset, len);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_read_guest_atomic);

static int __kvm_write_guest_page(struct kvm_memory_slot *memslot, gfn_t gfn,
			          const void *data, int offset, int len)
1746
{
1747 1748
	int r;
	unsigned long addr;
1749

1750
	addr = gfn_to_hva_memslot(memslot, gfn);
1751 1752
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1753
	r = __copy_to_user((void __user *)addr + offset, data, len);
1754
	if (r)
1755
		return -EFAULT;
1756
	mark_page_dirty_in_slot(memslot, gfn);
1757 1758
	return 0;
}
1759 1760 1761 1762 1763 1764 1765 1766

int kvm_write_guest_page(struct kvm *kvm, gfn_t gfn,
			 const void *data, int offset, int len)
{
	struct kvm_memory_slot *slot = gfn_to_memslot(kvm, gfn);

	return __kvm_write_guest_page(slot, gfn, data, offset, len);
}
1767 1768
EXPORT_SYMBOL_GPL(kvm_write_guest_page);

1769 1770 1771 1772 1773 1774 1775 1776 1777
int kvm_vcpu_write_guest_page(struct kvm_vcpu *vcpu, gfn_t gfn,
			      const void *data, int offset, int len)
{
	struct kvm_memory_slot *slot = kvm_vcpu_gfn_to_memslot(vcpu, gfn);

	return __kvm_write_guest_page(slot, gfn, data, offset, len);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_write_guest_page);

1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796
int kvm_write_guest(struct kvm *kvm, gpa_t gpa, const void *data,
		    unsigned long len)
{
	gfn_t gfn = gpa >> PAGE_SHIFT;
	int seg;
	int offset = offset_in_page(gpa);
	int ret;

	while ((seg = next_segment(len, offset)) != 0) {
		ret = kvm_write_guest_page(kvm, gfn, data, offset, seg);
		if (ret < 0)
			return ret;
		offset = 0;
		len -= seg;
		data += seg;
		++gfn;
	}
	return 0;
}
1797
EXPORT_SYMBOL_GPL(kvm_write_guest);
1798

1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819
int kvm_vcpu_write_guest(struct kvm_vcpu *vcpu, gpa_t gpa, const void *data,
		         unsigned long len)
{
	gfn_t gfn = gpa >> PAGE_SHIFT;
	int seg;
	int offset = offset_in_page(gpa);
	int ret;

	while ((seg = next_segment(len, offset)) != 0) {
		ret = kvm_vcpu_write_guest_page(vcpu, gfn, data, offset, seg);
		if (ret < 0)
			return ret;
		offset = 0;
		len -= seg;
		data += seg;
		++gfn;
	}
	return 0;
}
EXPORT_SYMBOL_GPL(kvm_vcpu_write_guest);

1820
int kvm_gfn_to_hva_cache_init(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
1821
			      gpa_t gpa, unsigned long len)
1822 1823 1824
{
	struct kvm_memslots *slots = kvm_memslots(kvm);
	int offset = offset_in_page(gpa);
1825 1826 1827 1828
	gfn_t start_gfn = gpa >> PAGE_SHIFT;
	gfn_t end_gfn = (gpa + len - 1) >> PAGE_SHIFT;
	gfn_t nr_pages_needed = end_gfn - start_gfn + 1;
	gfn_t nr_pages_avail;
1829 1830 1831

	ghc->gpa = gpa;
	ghc->generation = slots->generation;
1832 1833
	ghc->len = len;
	ghc->memslot = gfn_to_memslot(kvm, start_gfn);
1834 1835
	ghc->hva = gfn_to_hva_many(ghc->memslot, start_gfn, NULL);
	if (!kvm_is_error_hva(ghc->hva) && nr_pages_needed <= 1) {
1836
		ghc->hva += offset;
1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852
	} else {
		/*
		 * If the requested region crosses two memslots, we still
		 * verify that the entire region is valid here.
		 */
		while (start_gfn <= end_gfn) {
			ghc->memslot = gfn_to_memslot(kvm, start_gfn);
			ghc->hva = gfn_to_hva_many(ghc->memslot, start_gfn,
						   &nr_pages_avail);
			if (kvm_is_error_hva(ghc->hva))
				return -EFAULT;
			start_gfn += nr_pages_avail;
		}
		/* Use the slow path for cross page reads and writes. */
		ghc->memslot = NULL;
	}
1853 1854 1855 1856 1857 1858 1859 1860 1861 1862
	return 0;
}
EXPORT_SYMBOL_GPL(kvm_gfn_to_hva_cache_init);

int kvm_write_guest_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
			   void *data, unsigned long len)
{
	struct kvm_memslots *slots = kvm_memslots(kvm);
	int r;

1863 1864
	BUG_ON(len > ghc->len);

1865
	if (slots->generation != ghc->generation)
1866 1867 1868 1869
		kvm_gfn_to_hva_cache_init(kvm, ghc, ghc->gpa, ghc->len);

	if (unlikely(!ghc->memslot))
		return kvm_write_guest(kvm, ghc->gpa, data, len);
1870 1871 1872 1873

	if (kvm_is_error_hva(ghc->hva))
		return -EFAULT;

1874
	r = __copy_to_user((void __user *)ghc->hva, data, len);
1875 1876
	if (r)
		return -EFAULT;
1877
	mark_page_dirty_in_slot(ghc->memslot, ghc->gpa >> PAGE_SHIFT);
1878 1879 1880 1881 1882

	return 0;
}
EXPORT_SYMBOL_GPL(kvm_write_guest_cached);

1883 1884 1885 1886 1887 1888
int kvm_read_guest_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
			   void *data, unsigned long len)
{
	struct kvm_memslots *slots = kvm_memslots(kvm);
	int r;

1889 1890
	BUG_ON(len > ghc->len);

1891
	if (slots->generation != ghc->generation)
1892 1893 1894 1895
		kvm_gfn_to_hva_cache_init(kvm, ghc, ghc->gpa, ghc->len);

	if (unlikely(!ghc->memslot))
		return kvm_read_guest(kvm, ghc->gpa, data, len);
1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907

	if (kvm_is_error_hva(ghc->hva))
		return -EFAULT;

	r = __copy_from_user(data, (void __user *)ghc->hva, len);
	if (r)
		return -EFAULT;

	return 0;
}
EXPORT_SYMBOL_GPL(kvm_read_guest_cached);

1908 1909
int kvm_clear_guest_page(struct kvm *kvm, gfn_t gfn, int offset, int len)
{
1910 1911 1912
	const void *zero_page = (const void *) __va(page_to_phys(ZERO_PAGE(0)));

	return kvm_write_guest_page(kvm, gfn, zero_page, offset, len);
1913 1914 1915 1916 1917 1918 1919 1920 1921 1922
}
EXPORT_SYMBOL_GPL(kvm_clear_guest_page);

int kvm_clear_guest(struct kvm *kvm, gpa_t gpa, unsigned long len)
{
	gfn_t gfn = gpa >> PAGE_SHIFT;
	int seg;
	int offset = offset_in_page(gpa);
	int ret;

1923
	while ((seg = next_segment(len, offset)) != 0) {
1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934
		ret = kvm_clear_guest_page(kvm, gfn, offset, seg);
		if (ret < 0)
			return ret;
		offset = 0;
		len -= seg;
		++gfn;
	}
	return 0;
}
EXPORT_SYMBOL_GPL(kvm_clear_guest);

1935
static void mark_page_dirty_in_slot(struct kvm_memory_slot *memslot,
1936
				    gfn_t gfn)
1937
{
Rusty Russell's avatar
Rusty Russell committed
1938 1939
	if (memslot && memslot->dirty_bitmap) {
		unsigned long rel_gfn = gfn - memslot->base_gfn;
1940

1941
		set_bit_le(rel_gfn, memslot->dirty_bitmap);
1942 1943 1944
	}
}

1945 1946 1947 1948 1949
void mark_page_dirty(struct kvm *kvm, gfn_t gfn)
{
	struct kvm_memory_slot *memslot;

	memslot = gfn_to_memslot(kvm, gfn);
1950
	mark_page_dirty_in_slot(memslot, gfn);
1951
}
1952
EXPORT_SYMBOL_GPL(mark_page_dirty);
1953

1954 1955 1956 1957 1958 1959 1960 1961 1962
void kvm_vcpu_mark_page_dirty(struct kvm_vcpu *vcpu, gfn_t gfn)
{
	struct kvm_memory_slot *memslot;

	memslot = kvm_vcpu_gfn_to_memslot(vcpu, gfn);
	mark_page_dirty_in_slot(memslot, gfn);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_mark_page_dirty);

Wanpeng Li's avatar
Wanpeng Li committed
1963 1964
static void grow_halt_poll_ns(struct kvm_vcpu *vcpu)
{
1965
	int old, val;
Wanpeng Li's avatar
Wanpeng Li committed
1966

1967
	old = val = vcpu->halt_poll_ns;
Wanpeng Li's avatar
Wanpeng Li committed
1968 1969 1970 1971 1972 1973
	/* 10us base */
	if (val == 0 && halt_poll_ns_grow)
		val = 10000;
	else
		val *= halt_poll_ns_grow;

1974 1975 1976
	if (val > halt_poll_ns)
		val = halt_poll_ns;

Wanpeng Li's avatar
Wanpeng Li committed
1977
	vcpu->halt_poll_ns = val;
1978
	trace_kvm_halt_poll_ns_grow(vcpu->vcpu_id, val, old);
Wanpeng Li's avatar
Wanpeng Li committed
1979 1980 1981 1982
}

static void shrink_halt_poll_ns(struct kvm_vcpu *vcpu)
{
1983
	int old, val;
Wanpeng Li's avatar
Wanpeng Li committed
1984

1985
	old = val = vcpu->halt_poll_ns;
Wanpeng Li's avatar
Wanpeng Li committed
1986 1987 1988 1989 1990 1991
	if (halt_poll_ns_shrink == 0)
		val = 0;
	else
		val /= halt_poll_ns_shrink;

	vcpu->halt_poll_ns = val;
1992
	trace_kvm_halt_poll_ns_shrink(vcpu->vcpu_id, val, old);
Wanpeng Li's avatar
Wanpeng Li committed
1993 1994
}

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
static int kvm_vcpu_check_block(struct kvm_vcpu *vcpu)
{
	if (kvm_arch_vcpu_runnable(vcpu)) {
		kvm_make_request(KVM_REQ_UNHALT, vcpu);
		return -EINTR;
	}
	if (kvm_cpu_has_pending_timer(vcpu))
		return -EINTR;
	if (signal_pending(current))
		return -EINTR;

	return 0;
}

Eddie Dong's avatar
Eddie Dong committed
2009 2010 2011
/*
 * The vCPU has executed a HLT instruction with in-kernel mode enabled.
 */
2012
void kvm_vcpu_block(struct kvm_vcpu *vcpu)
2013
{
2014
	ktime_t start, cur;
2015
	DEFINE_WAIT(wait);
2016
	bool waited = false;
Wanpeng Li's avatar
Wanpeng Li committed
2017
	u64 block_ns;
2018 2019

	start = cur = ktime_get();
2020 2021
	if (vcpu->halt_poll_ns) {
		ktime_t stop = ktime_add_ns(ktime_get(), vcpu->halt_poll_ns);
2022

2023
		++vcpu->stat.halt_attempted_poll;
2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035
		do {
			/*
			 * This sets KVM_REQ_UNHALT if an interrupt
			 * arrives.
			 */
			if (kvm_vcpu_check_block(vcpu) < 0) {
				++vcpu->stat.halt_successful_poll;
				goto out;
			}
			cur = ktime_get();
		} while (single_task_running() && ktime_before(cur, stop));
	}
2036

2037 2038
	kvm_arch_vcpu_blocking(vcpu);

2039 2040 2041
	for (;;) {
		prepare_to_wait(&vcpu->wq, &wait, TASK_INTERRUPTIBLE);

2042
		if (kvm_vcpu_check_block(vcpu) < 0)
2043 2044
			break;

2045
		waited = true;
Eddie Dong's avatar
Eddie Dong committed
2046 2047
		schedule();
	}
2048

2049
	finish_wait(&vcpu->wq, &wait);
2050 2051
	cur = ktime_get();

2052
	kvm_arch_vcpu_unblocking(vcpu);
2053
out:
Wanpeng Li's avatar
Wanpeng Li committed
2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065
	block_ns = ktime_to_ns(cur) - ktime_to_ns(start);

	if (halt_poll_ns) {
		if (block_ns <= vcpu->halt_poll_ns)
			;
		/* we had a long block, shrink polling */
		else if (vcpu->halt_poll_ns && block_ns > halt_poll_ns)
			shrink_halt_poll_ns(vcpu);
		/* we had a short halt and our poll time is too small */
		else if (vcpu->halt_poll_ns < halt_poll_ns &&
			block_ns < halt_poll_ns)
			grow_halt_poll_ns(vcpu);
2066 2067
	} else
		vcpu->halt_poll_ns = 0;
Wanpeng Li's avatar
Wanpeng Li committed
2068 2069

	trace_kvm_vcpu_wakeup(block_ns, waited);
Eddie Dong's avatar
Eddie Dong committed
2070
}
2071
EXPORT_SYMBOL_GPL(kvm_vcpu_block);
Eddie Dong's avatar
Eddie Dong committed
2072

2073
#ifndef CONFIG_S390
2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094
/*
 * Kick a sleeping VCPU, or a guest VCPU in guest mode, into host kernel mode.
 */
void kvm_vcpu_kick(struct kvm_vcpu *vcpu)
{
	int me;
	int cpu = vcpu->cpu;
	wait_queue_head_t *wqp;

	wqp = kvm_arch_vcpu_wq(vcpu);
	if (waitqueue_active(wqp)) {
		wake_up_interruptible(wqp);
		++vcpu->stat.halt_wakeup;
	}

	me = get_cpu();
	if (cpu != me && (unsigned)cpu < nr_cpu_ids && cpu_online(cpu))
		if (kvm_arch_vcpu_should_kick(vcpu))
			smp_send_reschedule(cpu);
	put_cpu();
}
2095
EXPORT_SYMBOL_GPL(kvm_vcpu_kick);
2096
#endif /* !CONFIG_S390 */
2097

2098
int kvm_vcpu_yield_to(struct kvm_vcpu *target)
2099 2100 2101
{
	struct pid *pid;
	struct task_struct *task = NULL;
2102
	int ret = 0;
2103 2104 2105 2106

	rcu_read_lock();
	pid = rcu_dereference(target->pid);
	if (pid)
2107
		task = get_pid_task(pid, PIDTYPE_PID);
2108 2109
	rcu_read_unlock();
	if (!task)
2110 2111
		return ret;
	ret = yield_to(task, 1);
2112
	put_task_struct(task);
2113 2114

	return ret;
2115 2116 2117
}
EXPORT_SYMBOL_GPL(kvm_vcpu_yield_to);

2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139
/*
 * Helper that checks whether a VCPU is eligible for directed yield.
 * Most eligible candidate to yield is decided by following heuristics:
 *
 *  (a) VCPU which has not done pl-exit or cpu relax intercepted recently
 *  (preempted lock holder), indicated by @in_spin_loop.
 *  Set at the beiginning and cleared at the end of interception/PLE handler.
 *
 *  (b) VCPU which has done pl-exit/ cpu relax intercepted but did not get
 *  chance last time (mostly it has become eligible now since we have probably
 *  yielded to lockholder in last iteration. This is done by toggling
 *  @dy_eligible each time a VCPU checked for eligibility.)
 *
 *  Yielding to a recently pl-exited/cpu relax intercepted VCPU before yielding
 *  to preempted lock-holder could result in wrong VCPU selection and CPU
 *  burning. Giving priority for a potential lock-holder increases lock
 *  progress.
 *
 *  Since algorithm is based on heuristics, accessing another VCPU data without
 *  locking does not harm. It may result in trying to yield to  same VCPU, fail
 *  and continue with next VCPU and so on.
 */
2140
static bool kvm_vcpu_eligible_for_directed_yield(struct kvm_vcpu *vcpu)
2141
{
2142
#ifdef CONFIG_HAVE_KVM_CPU_RELAX_INTERCEPT
2143 2144 2145
	bool eligible;

	eligible = !vcpu->spin_loop.in_spin_loop ||
2146
		    vcpu->spin_loop.dy_eligible;
2147 2148 2149 2150 2151

	if (vcpu->spin_loop.in_spin_loop)
		kvm_vcpu_set_dy_eligible(vcpu, !vcpu->spin_loop.dy_eligible);

	return eligible;
2152 2153
#else
	return true;
2154
#endif
2155
}
2156

2157
void kvm_vcpu_on_spin(struct kvm_vcpu *me)
2158
{
2159 2160 2161 2162
	struct kvm *kvm = me->kvm;
	struct kvm_vcpu *vcpu;
	int last_boosted_vcpu = me->kvm->last_boosted_vcpu;
	int yielded = 0;
2163
	int try = 3;
2164 2165
	int pass;
	int i;
2166

2167
	kvm_vcpu_set_in_spin_loop(me, true);
2168 2169 2170 2171 2172 2173 2174
	/*
	 * We boost the priority of a VCPU that is runnable but not
	 * currently running, because it got preempted by something
	 * else and called schedule in __vcpu_run.  Hopefully that
	 * VCPU is holding the lock that we need and will release it.
	 * We approximate round-robin by starting at the last boosted VCPU.
	 */
2175
	for (pass = 0; pass < 2 && !yielded && try; pass++) {
2176
		kvm_for_each_vcpu(i, vcpu, kvm) {
2177
			if (!pass && i <= last_boosted_vcpu) {
2178 2179 2180 2181
				i = last_boosted_vcpu;
				continue;
			} else if (pass && i > last_boosted_vcpu)
				break;
2182 2183
			if (!ACCESS_ONCE(vcpu->preempted))
				continue;
2184 2185
			if (vcpu == me)
				continue;
2186
			if (waitqueue_active(&vcpu->wq) && !kvm_arch_vcpu_runnable(vcpu))
2187
				continue;
2188 2189
			if (!kvm_vcpu_eligible_for_directed_yield(vcpu))
				continue;
2190 2191 2192

			yielded = kvm_vcpu_yield_to(vcpu);
			if (yielded > 0) {
2193 2194
				kvm->last_boosted_vcpu = i;
				break;
2195 2196 2197 2198
			} else if (yielded < 0) {
				try--;
				if (!try)
					break;
2199 2200 2201
			}
		}
	}
2202
	kvm_vcpu_set_in_spin_loop(me, false);
2203 2204 2205

	/* Ensure vcpu is not eligible during next spinloop */
	kvm_vcpu_set_dy_eligible(me, false);
2206 2207 2208
}
EXPORT_SYMBOL_GPL(kvm_vcpu_on_spin);

2209
static int kvm_vcpu_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
2210 2211 2212 2213
{
	struct kvm_vcpu *vcpu = vma->vm_file->private_data;
	struct page *page;

2214
	if (vmf->pgoff == 0)
2215
		page = virt_to_page(vcpu->run);
Avi Kivity's avatar
Avi Kivity committed
2216
#ifdef CONFIG_X86
2217
	else if (vmf->pgoff == KVM_PIO_PAGE_OFFSET)
2218
		page = virt_to_page(vcpu->arch.pio_data);
2219 2220 2221 2222
#endif
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
	else if (vmf->pgoff == KVM_COALESCED_MMIO_PAGE_OFFSET)
		page = virt_to_page(vcpu->kvm->coalesced_mmio_ring);
Avi Kivity's avatar
Avi Kivity committed
2223
#endif
2224
	else
2225
		return kvm_arch_vcpu_fault(vcpu, vmf);
2226
	get_page(page);
2227 2228
	vmf->page = page;
	return 0;
2229 2230
}

2231
static const struct vm_operations_struct kvm_vcpu_vm_ops = {
2232
	.fault = kvm_vcpu_fault,
2233 2234 2235 2236 2237 2238 2239 2240
};

static int kvm_vcpu_mmap(struct file *file, struct vm_area_struct *vma)
{
	vma->vm_ops = &kvm_vcpu_vm_ops;
	return 0;
}

Avi Kivity's avatar
Avi Kivity committed
2241 2242 2243 2244
static int kvm_vcpu_release(struct inode *inode, struct file *filp)
{
	struct kvm_vcpu *vcpu = filp->private_data;

2245
	kvm_put_kvm(vcpu->kvm);
Avi Kivity's avatar
Avi Kivity committed
2246 2247 2248
	return 0;
}

2249
static struct file_operations kvm_vcpu_fops = {
Avi Kivity's avatar
Avi Kivity committed
2250 2251
	.release        = kvm_vcpu_release,
	.unlocked_ioctl = kvm_vcpu_ioctl,
2252
#ifdef CONFIG_KVM_COMPAT
2253 2254
	.compat_ioctl   = kvm_vcpu_compat_ioctl,
#endif
2255
	.mmap           = kvm_vcpu_mmap,
2256
	.llseek		= noop_llseek,
Avi Kivity's avatar
Avi Kivity committed
2257 2258 2259 2260 2261 2262 2263
};

/*
 * Allocates an inode for the vcpu.
 */
static int create_vcpu_fd(struct kvm_vcpu *vcpu)
{
2264
	return anon_inode_getfd("kvm-vcpu", &kvm_vcpu_fops, vcpu, O_RDWR | O_CLOEXEC);
Avi Kivity's avatar
Avi Kivity committed
2265 2266
}

2267 2268 2269
/*
 * Creates some virtual cpus.  Good luck creating more than one.
 */
2270
static int kvm_vm_ioctl_create_vcpu(struct kvm *kvm, u32 id)
2271 2272
{
	int r;
2273
	struct kvm_vcpu *vcpu, *v;
2274

2275 2276 2277
	if (id >= KVM_MAX_VCPUS)
		return -EINVAL;

2278
	vcpu = kvm_arch_vcpu_create(kvm, id);
2279 2280
	if (IS_ERR(vcpu))
		return PTR_ERR(vcpu);
2281

2282 2283
	preempt_notifier_init(&vcpu->preempt_notifier, &kvm_preempt_ops);

2284 2285
	r = kvm_arch_vcpu_setup(vcpu);
	if (r)
2286
		goto vcpu_destroy;
2287

Shaohua Li's avatar
Shaohua Li committed
2288
	mutex_lock(&kvm->lock);
2289 2290 2291 2292
	if (!kvm_vcpu_compatible(vcpu)) {
		r = -EINVAL;
		goto unlock_vcpu_destroy;
	}
2293 2294
	if (atomic_read(&kvm->online_vcpus) == KVM_MAX_VCPUS) {
		r = -EINVAL;
2295
		goto unlock_vcpu_destroy;
2296
	}
2297

2298 2299
	kvm_for_each_vcpu(r, v, kvm)
		if (v->vcpu_id == id) {
2300
			r = -EEXIST;
2301
			goto unlock_vcpu_destroy;
2302 2303 2304
		}

	BUG_ON(kvm->vcpus[atomic_read(&kvm->online_vcpus)]);
2305

2306
	/* Now it's all set up, let userspace reach it */
2307
	kvm_get_kvm(kvm);
Avi Kivity's avatar
Avi Kivity committed
2308
	r = create_vcpu_fd(vcpu);
2309 2310
	if (r < 0) {
		kvm_put_kvm(kvm);
2311
		goto unlock_vcpu_destroy;
2312 2313 2314
	}

	kvm->vcpus[atomic_read(&kvm->online_vcpus)] = vcpu;
2315 2316 2317 2318 2319

	/*
	 * Pairs with smp_rmb() in kvm_get_vcpu.  Write kvm->vcpus
	 * before kvm->online_vcpu's incremented value.
	 */
2320 2321 2322 2323
	smp_wmb();
	atomic_inc(&kvm->online_vcpus);

	mutex_unlock(&kvm->lock);
2324
	kvm_arch_vcpu_postcreate(vcpu);
2325
	return r;
2326

2327
unlock_vcpu_destroy:
2328
	mutex_unlock(&kvm->lock);
2329
vcpu_destroy:
2330
	kvm_arch_vcpu_destroy(vcpu);
2331 2332 2333
	return r;
}

Avi Kivity's avatar
Avi Kivity committed
2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344
static int kvm_vcpu_ioctl_set_sigmask(struct kvm_vcpu *vcpu, sigset_t *sigset)
{
	if (sigset) {
		sigdelsetmask(sigset, sigmask(SIGKILL)|sigmask(SIGSTOP));
		vcpu->sigset_active = 1;
		vcpu->sigset = *sigset;
	} else
		vcpu->sigset_active = 0;
	return 0;
}

Avi Kivity's avatar
Avi Kivity committed
2345 2346
static long kvm_vcpu_ioctl(struct file *filp,
			   unsigned int ioctl, unsigned long arg)
2347
{
Avi Kivity's avatar
Avi Kivity committed
2348
	struct kvm_vcpu *vcpu = filp->private_data;
Al Viro's avatar
Al Viro committed
2349
	void __user *argp = (void __user *)arg;
2350
	int r;
2351 2352
	struct kvm_fpu *fpu = NULL;
	struct kvm_sregs *kvm_sregs = NULL;
2353

2354 2355
	if (vcpu->kvm->mm != current->mm)
		return -EIO;
2356

2357 2358 2359
	if (unlikely(_IOC_TYPE(ioctl) != KVMIO))
		return -EINVAL;

2360
#if defined(CONFIG_S390) || defined(CONFIG_PPC) || defined(CONFIG_MIPS)
2361 2362 2363 2364
	/*
	 * Special cases: vcpu ioctls that are asynchronous to vcpu execution,
	 * so vcpu_load() would break it.
	 */
2365
	if (ioctl == KVM_S390_INTERRUPT || ioctl == KVM_S390_IRQ || ioctl == KVM_INTERRUPT)
2366 2367 2368 2369
		return kvm_arch_vcpu_ioctl(filp, ioctl, arg);
#endif


2370 2371 2372
	r = vcpu_load(vcpu);
	if (r)
		return r;
2373
	switch (ioctl) {
2374
	case KVM_RUN:
2375 2376 2377
		r = -EINVAL;
		if (arg)
			goto out;
2378 2379 2380 2381
		if (unlikely(vcpu->pid != current->pids[PIDTYPE_PID].pid)) {
			/* The thread running this VCPU changed. */
			struct pid *oldpid = vcpu->pid;
			struct pid *newpid = get_task_pid(current, PIDTYPE_PID);
2382

2383 2384 2385 2386 2387
			rcu_assign_pointer(vcpu->pid, newpid);
			if (oldpid)
				synchronize_rcu();
			put_pid(oldpid);
		}
2388
		r = kvm_arch_vcpu_ioctl_run(vcpu, vcpu->run);
2389
		trace_kvm_userspace_exit(vcpu->run->exit_reason, r);
2390 2391
		break;
	case KVM_GET_REGS: {
2392
		struct kvm_regs *kvm_regs;
2393

2394 2395 2396
		r = -ENOMEM;
		kvm_regs = kzalloc(sizeof(struct kvm_regs), GFP_KERNEL);
		if (!kvm_regs)
2397
			goto out;
2398 2399 2400
		r = kvm_arch_vcpu_ioctl_get_regs(vcpu, kvm_regs);
		if (r)
			goto out_free1;
2401
		r = -EFAULT;
2402 2403
		if (copy_to_user(argp, kvm_regs, sizeof(struct kvm_regs)))
			goto out_free1;
2404
		r = 0;
2405 2406
out_free1:
		kfree(kvm_regs);
2407 2408 2409
		break;
	}
	case KVM_SET_REGS: {
2410
		struct kvm_regs *kvm_regs;
2411

2412
		r = -ENOMEM;
2413 2414 2415
		kvm_regs = memdup_user(argp, sizeof(*kvm_regs));
		if (IS_ERR(kvm_regs)) {
			r = PTR_ERR(kvm_regs);
2416
			goto out;
2417
		}
2418 2419
		r = kvm_arch_vcpu_ioctl_set_regs(vcpu, kvm_regs);
		kfree(kvm_regs);
2420 2421 2422
		break;
	}
	case KVM_GET_SREGS: {
2423 2424 2425 2426 2427
		kvm_sregs = kzalloc(sizeof(struct kvm_sregs), GFP_KERNEL);
		r = -ENOMEM;
		if (!kvm_sregs)
			goto out;
		r = kvm_arch_vcpu_ioctl_get_sregs(vcpu, kvm_sregs);
2428 2429 2430
		if (r)
			goto out;
		r = -EFAULT;
2431
		if (copy_to_user(argp, kvm_sregs, sizeof(struct kvm_sregs)))
2432 2433 2434 2435 2436
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_SREGS: {
2437 2438 2439
		kvm_sregs = memdup_user(argp, sizeof(*kvm_sregs));
		if (IS_ERR(kvm_sregs)) {
			r = PTR_ERR(kvm_sregs);
Guo Chao's avatar
Guo Chao committed
2440
			kvm_sregs = NULL;
2441
			goto out;
2442
		}
2443
		r = kvm_arch_vcpu_ioctl_set_sregs(vcpu, kvm_sregs);
2444 2445
		break;
	}
2446 2447 2448 2449 2450 2451 2452
	case KVM_GET_MP_STATE: {
		struct kvm_mp_state mp_state;

		r = kvm_arch_vcpu_ioctl_get_mpstate(vcpu, &mp_state);
		if (r)
			goto out;
		r = -EFAULT;
2453
		if (copy_to_user(argp, &mp_state, sizeof(mp_state)))
2454 2455 2456 2457 2458 2459 2460 2461
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_MP_STATE: {
		struct kvm_mp_state mp_state;

		r = -EFAULT;
2462
		if (copy_from_user(&mp_state, argp, sizeof(mp_state)))
2463 2464 2465 2466
			goto out;
		r = kvm_arch_vcpu_ioctl_set_mpstate(vcpu, &mp_state);
		break;
	}
2467 2468 2469 2470
	case KVM_TRANSLATE: {
		struct kvm_translation tr;

		r = -EFAULT;
2471
		if (copy_from_user(&tr, argp, sizeof(tr)))
2472
			goto out;
2473
		r = kvm_arch_vcpu_ioctl_translate(vcpu, &tr);
2474 2475 2476
		if (r)
			goto out;
		r = -EFAULT;
2477
		if (copy_to_user(argp, &tr, sizeof(tr)))
2478 2479 2480 2481
			goto out;
		r = 0;
		break;
	}
Jan Kiszka's avatar
Jan Kiszka committed
2482 2483
	case KVM_SET_GUEST_DEBUG: {
		struct kvm_guest_debug dbg;
2484 2485

		r = -EFAULT;
2486
		if (copy_from_user(&dbg, argp, sizeof(dbg)))
2487
			goto out;
Jan Kiszka's avatar
Jan Kiszka committed
2488
		r = kvm_arch_vcpu_ioctl_set_guest_debug(vcpu, &dbg);
2489 2490
		break;
	}
Avi Kivity's avatar
Avi Kivity committed
2491 2492 2493 2494 2495 2496 2497 2498 2499
	case KVM_SET_SIGNAL_MASK: {
		struct kvm_signal_mask __user *sigmask_arg = argp;
		struct kvm_signal_mask kvm_sigmask;
		sigset_t sigset, *p;

		p = NULL;
		if (argp) {
			r = -EFAULT;
			if (copy_from_user(&kvm_sigmask, argp,
2500
					   sizeof(kvm_sigmask)))
Avi Kivity's avatar
Avi Kivity committed
2501 2502
				goto out;
			r = -EINVAL;
2503
			if (kvm_sigmask.len != sizeof(sigset))
Avi Kivity's avatar
Avi Kivity committed
2504 2505 2506
				goto out;
			r = -EFAULT;
			if (copy_from_user(&sigset, sigmask_arg->sigset,
2507
					   sizeof(sigset)))
Avi Kivity's avatar
Avi Kivity committed
2508 2509 2510
				goto out;
			p = &sigset;
		}
2511
		r = kvm_vcpu_ioctl_set_sigmask(vcpu, p);
Avi Kivity's avatar
Avi Kivity committed
2512 2513
		break;
	}
Avi Kivity's avatar
Avi Kivity committed
2514
	case KVM_GET_FPU: {
2515 2516 2517 2518 2519
		fpu = kzalloc(sizeof(struct kvm_fpu), GFP_KERNEL);
		r = -ENOMEM;
		if (!fpu)
			goto out;
		r = kvm_arch_vcpu_ioctl_get_fpu(vcpu, fpu);
Avi Kivity's avatar
Avi Kivity committed
2520 2521 2522
		if (r)
			goto out;
		r = -EFAULT;
2523
		if (copy_to_user(argp, fpu, sizeof(struct kvm_fpu)))
Avi Kivity's avatar
Avi Kivity committed
2524 2525 2526 2527 2528
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_FPU: {
2529 2530 2531
		fpu = memdup_user(argp, sizeof(*fpu));
		if (IS_ERR(fpu)) {
			r = PTR_ERR(fpu);
Guo Chao's avatar
Guo Chao committed
2532
			fpu = NULL;
Avi Kivity's avatar
Avi Kivity committed
2533
			goto out;
2534
		}
2535
		r = kvm_arch_vcpu_ioctl_set_fpu(vcpu, fpu);
Avi Kivity's avatar
Avi Kivity committed
2536 2537
		break;
	}
Avi Kivity's avatar
Avi Kivity committed
2538
	default:
2539
		r = kvm_arch_vcpu_ioctl(filp, ioctl, arg);
Avi Kivity's avatar
Avi Kivity committed
2540 2541
	}
out:
2542
	vcpu_put(vcpu);
2543 2544
	kfree(fpu);
	kfree(kvm_sregs);
Avi Kivity's avatar
Avi Kivity committed
2545 2546 2547
	return r;
}

2548
#ifdef CONFIG_KVM_COMPAT
2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568
static long kvm_vcpu_compat_ioctl(struct file *filp,
				  unsigned int ioctl, unsigned long arg)
{
	struct kvm_vcpu *vcpu = filp->private_data;
	void __user *argp = compat_ptr(arg);
	int r;

	if (vcpu->kvm->mm != current->mm)
		return -EIO;

	switch (ioctl) {
	case KVM_SET_SIGNAL_MASK: {
		struct kvm_signal_mask __user *sigmask_arg = argp;
		struct kvm_signal_mask kvm_sigmask;
		compat_sigset_t csigset;
		sigset_t sigset;

		if (argp) {
			r = -EFAULT;
			if (copy_from_user(&kvm_sigmask, argp,
2569
					   sizeof(kvm_sigmask)))
2570 2571
				goto out;
			r = -EINVAL;
2572
			if (kvm_sigmask.len != sizeof(csigset))
2573 2574 2575
				goto out;
			r = -EFAULT;
			if (copy_from_user(&csigset, sigmask_arg->sigset,
2576
					   sizeof(csigset)))
2577
				goto out;
2578 2579 2580 2581
			sigset_from_compat(&sigset, &csigset);
			r = kvm_vcpu_ioctl_set_sigmask(vcpu, &sigset);
		} else
			r = kvm_vcpu_ioctl_set_sigmask(vcpu, NULL);
2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592
		break;
	}
	default:
		r = kvm_vcpu_ioctl(filp, ioctl, arg);
	}

out:
	return r;
}
#endif

Scott Wood's avatar
Scott Wood committed
2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613
static int kvm_device_ioctl_attr(struct kvm_device *dev,
				 int (*accessor)(struct kvm_device *dev,
						 struct kvm_device_attr *attr),
				 unsigned long arg)
{
	struct kvm_device_attr attr;

	if (!accessor)
		return -EPERM;

	if (copy_from_user(&attr, (void __user *)arg, sizeof(attr)))
		return -EFAULT;

	return accessor(dev, &attr);
}

static long kvm_device_ioctl(struct file *filp, unsigned int ioctl,
			     unsigned long arg)
{
	struct kvm_device *dev = filp->private_data;

2614 2615 2616
	if (dev->kvm->mm != current->mm)
		return -EIO;

Scott Wood's avatar
Scott Wood committed
2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642
	switch (ioctl) {
	case KVM_SET_DEVICE_ATTR:
		return kvm_device_ioctl_attr(dev, dev->ops->set_attr, arg);
	case KVM_GET_DEVICE_ATTR:
		return kvm_device_ioctl_attr(dev, dev->ops->get_attr, arg);
	case KVM_HAS_DEVICE_ATTR:
		return kvm_device_ioctl_attr(dev, dev->ops->has_attr, arg);
	default:
		if (dev->ops->ioctl)
			return dev->ops->ioctl(dev, ioctl, arg);

		return -ENOTTY;
	}
}

static int kvm_device_release(struct inode *inode, struct file *filp)
{
	struct kvm_device *dev = filp->private_data;
	struct kvm *kvm = dev->kvm;

	kvm_put_kvm(kvm);
	return 0;
}

static const struct file_operations kvm_device_fops = {
	.unlocked_ioctl = kvm_device_ioctl,
2643
#ifdef CONFIG_KVM_COMPAT
2644 2645
	.compat_ioctl = kvm_device_ioctl,
#endif
Scott Wood's avatar
Scott Wood committed
2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656
	.release = kvm_device_release,
};

struct kvm_device *kvm_device_from_filp(struct file *filp)
{
	if (filp->f_op != &kvm_device_fops)
		return NULL;

	return filp->private_data;
}

2657
static struct kvm_device_ops *kvm_device_ops_table[KVM_DEV_TYPE_MAX] = {
2658
#ifdef CONFIG_KVM_MPIC
2659 2660
	[KVM_DEV_TYPE_FSL_MPIC_20]	= &kvm_mpic_ops,
	[KVM_DEV_TYPE_FSL_MPIC_42]	= &kvm_mpic_ops,
2661
#endif
2662

2663
#ifdef CONFIG_KVM_XICS
2664
	[KVM_DEV_TYPE_XICS]		= &kvm_xics_ops,
Alex Williamson's avatar
Alex Williamson committed
2665
#endif
2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679
};

int kvm_register_device_ops(struct kvm_device_ops *ops, u32 type)
{
	if (type >= ARRAY_SIZE(kvm_device_ops_table))
		return -ENOSPC;

	if (kvm_device_ops_table[type] != NULL)
		return -EEXIST;

	kvm_device_ops_table[type] = ops;
	return 0;
}

2680 2681 2682 2683 2684 2685
void kvm_unregister_device_ops(u32 type)
{
	if (kvm_device_ops_table[type] != NULL)
		kvm_device_ops_table[type] = NULL;
}

Scott Wood's avatar
Scott Wood committed
2686 2687 2688 2689 2690 2691 2692 2693
static int kvm_ioctl_create_device(struct kvm *kvm,
				   struct kvm_create_device *cd)
{
	struct kvm_device_ops *ops = NULL;
	struct kvm_device *dev;
	bool test = cd->flags & KVM_CREATE_DEVICE_TEST;
	int ret;

2694 2695 2696 2697 2698
	if (cd->type >= ARRAY_SIZE(kvm_device_ops_table))
		return -ENODEV;

	ops = kvm_device_ops_table[cd->type];
	if (ops == NULL)
Scott Wood's avatar
Scott Wood committed
2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716
		return -ENODEV;

	if (test)
		return 0;

	dev = kzalloc(sizeof(*dev), GFP_KERNEL);
	if (!dev)
		return -ENOMEM;

	dev->ops = ops;
	dev->kvm = kvm;

	ret = ops->create(dev, cd->type);
	if (ret < 0) {
		kfree(dev);
		return ret;
	}

2717
	kvm_get_kvm(kvm);
2718
	ret = anon_inode_getfd(ops->name, &kvm_device_fops, dev, O_RDWR | O_CLOEXEC);
Scott Wood's avatar
Scott Wood committed
2719
	if (ret < 0) {
2720
		kvm_put_kvm(kvm);
Scott Wood's avatar
Scott Wood committed
2721 2722 2723 2724
		ops->destroy(dev);
		return ret;
	}

2725
	list_add(&dev->vm_node, &kvm->devices);
Scott Wood's avatar
Scott Wood committed
2726 2727 2728 2729
	cd->fd = ret;
	return 0;
}

2730 2731 2732 2733 2734 2735 2736 2737 2738 2739
static long kvm_vm_ioctl_check_extension_generic(struct kvm *kvm, long arg)
{
	switch (arg) {
	case KVM_CAP_USER_MEMORY:
	case KVM_CAP_DESTROY_MEMORY_REGION_WORKS:
	case KVM_CAP_JOIN_MEMORY_REGIONS_WORKS:
	case KVM_CAP_INTERNAL_ERROR_DATA:
#ifdef CONFIG_HAVE_KVM_MSI
	case KVM_CAP_SIGNAL_MSI:
#endif
2740
#ifdef CONFIG_HAVE_KVM_IRQFD
2741
	case KVM_CAP_IRQFD:
2742 2743
	case KVM_CAP_IRQFD_RESAMPLE:
#endif
2744
	case KVM_CAP_IOEVENTFD_ANY_LENGTH:
2745 2746 2747 2748 2749
	case KVM_CAP_CHECK_EXTENSION_VM:
		return 1;
#ifdef CONFIG_HAVE_KVM_IRQ_ROUTING
	case KVM_CAP_IRQ_ROUTING:
		return KVM_MAX_IRQ_ROUTES;
2750 2751 2752 2753
#endif
#if KVM_ADDRESS_SPACE_NUM > 1
	case KVM_CAP_MULTI_ADDRESS_SPACE:
		return KVM_ADDRESS_SPACE_NUM;
2754 2755 2756 2757 2758 2759 2760
#endif
	default:
		break;
	}
	return kvm_vm_ioctl_check_extension(kvm, arg);
}

Avi Kivity's avatar
Avi Kivity committed
2761 2762 2763 2764 2765
static long kvm_vm_ioctl(struct file *filp,
			   unsigned int ioctl, unsigned long arg)
{
	struct kvm *kvm = filp->private_data;
	void __user *argp = (void __user *)arg;
2766
	int r;
Avi Kivity's avatar
Avi Kivity committed
2767

2768 2769
	if (kvm->mm != current->mm)
		return -EIO;
Avi Kivity's avatar
Avi Kivity committed
2770 2771 2772 2773
	switch (ioctl) {
	case KVM_CREATE_VCPU:
		r = kvm_vm_ioctl_create_vcpu(kvm, arg);
		break;
2774 2775 2776 2777 2778
	case KVM_SET_USER_MEMORY_REGION: {
		struct kvm_userspace_memory_region kvm_userspace_mem;

		r = -EFAULT;
		if (copy_from_user(&kvm_userspace_mem, argp,
2779
						sizeof(kvm_userspace_mem)))
2780 2781
			goto out;

2782
		r = kvm_vm_ioctl_set_memory_region(kvm, &kvm_userspace_mem);
2783 2784 2785 2786 2787 2788
		break;
	}
	case KVM_GET_DIRTY_LOG: {
		struct kvm_dirty_log log;

		r = -EFAULT;
2789
		if (copy_from_user(&log, argp, sizeof(log)))
2790
			goto out;
2791
		r = kvm_vm_ioctl_get_dirty_log(kvm, &log);
2792 2793
		break;
	}
2794 2795 2796
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
	case KVM_REGISTER_COALESCED_MMIO: {
		struct kvm_coalesced_mmio_zone zone;
2797

2798
		r = -EFAULT;
2799
		if (copy_from_user(&zone, argp, sizeof(zone)))
2800 2801 2802 2803 2804 2805
			goto out;
		r = kvm_vm_ioctl_register_coalesced_mmio(kvm, &zone);
		break;
	}
	case KVM_UNREGISTER_COALESCED_MMIO: {
		struct kvm_coalesced_mmio_zone zone;
2806

2807
		r = -EFAULT;
2808
		if (copy_from_user(&zone, argp, sizeof(zone)))
2809 2810 2811 2812 2813
			goto out;
		r = kvm_vm_ioctl_unregister_coalesced_mmio(kvm, &zone);
		break;
	}
#endif
Gregory Haskins's avatar
Gregory Haskins committed
2814 2815 2816 2817
	case KVM_IRQFD: {
		struct kvm_irqfd data;

		r = -EFAULT;
2818
		if (copy_from_user(&data, argp, sizeof(data)))
Gregory Haskins's avatar
Gregory Haskins committed
2819
			goto out;
2820
		r = kvm_irqfd(kvm, &data);
Gregory Haskins's avatar
Gregory Haskins committed
2821 2822
		break;
	}
Gregory Haskins's avatar
Gregory Haskins committed
2823 2824 2825 2826
	case KVM_IOEVENTFD: {
		struct kvm_ioeventfd data;

		r = -EFAULT;
2827
		if (copy_from_user(&data, argp, sizeof(data)))
Gregory Haskins's avatar
Gregory Haskins committed
2828 2829 2830 2831
			goto out;
		r = kvm_ioeventfd(kvm, &data);
		break;
	}
2832 2833 2834 2835 2836
#ifdef CONFIG_HAVE_KVM_MSI
	case KVM_SIGNAL_MSI: {
		struct kvm_msi msi;

		r = -EFAULT;
2837
		if (copy_from_user(&msi, argp, sizeof(msi)))
2838 2839 2840 2841
			goto out;
		r = kvm_send_userspace_msi(kvm, &msi);
		break;
	}
2842 2843 2844 2845 2846 2847 2848
#endif
#ifdef __KVM_HAVE_IRQ_LINE
	case KVM_IRQ_LINE_STATUS:
	case KVM_IRQ_LINE: {
		struct kvm_irq_level irq_event;

		r = -EFAULT;
2849
		if (copy_from_user(&irq_event, argp, sizeof(irq_event)))
2850 2851
			goto out;

2852 2853
		r = kvm_vm_ioctl_irq_line(kvm, &irq_event,
					ioctl == KVM_IRQ_LINE_STATUS);
2854 2855 2856 2857 2858
		if (r)
			goto out;

		r = -EFAULT;
		if (ioctl == KVM_IRQ_LINE_STATUS) {
2859
			if (copy_to_user(argp, &irq_event, sizeof(irq_event)))
2860 2861 2862 2863 2864 2865
				goto out;
		}

		r = 0;
		break;
	}
2866
#endif
2867 2868 2869 2870 2871 2872 2873 2874 2875 2876
#ifdef CONFIG_HAVE_KVM_IRQ_ROUTING
	case KVM_SET_GSI_ROUTING: {
		struct kvm_irq_routing routing;
		struct kvm_irq_routing __user *urouting;
		struct kvm_irq_routing_entry *entries;

		r = -EFAULT;
		if (copy_from_user(&routing, argp, sizeof(routing)))
			goto out;
		r = -EINVAL;
2877
		if (routing.nr > KVM_MAX_IRQ_ROUTES)
2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891
			goto out;
		if (routing.flags)
			goto out;
		r = -ENOMEM;
		entries = vmalloc(routing.nr * sizeof(*entries));
		if (!entries)
			goto out;
		r = -EFAULT;
		urouting = argp;
		if (copy_from_user(entries, urouting->entries,
				   routing.nr * sizeof(*entries)))
			goto out_free_irq_routing;
		r = kvm_set_irq_routing(kvm, entries, routing.nr,
					routing.flags);
2892
out_free_irq_routing:
2893 2894 2895 2896
		vfree(entries);
		break;
	}
#endif /* CONFIG_HAVE_KVM_IRQ_ROUTING */
Scott Wood's avatar
Scott Wood committed
2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914
	case KVM_CREATE_DEVICE: {
		struct kvm_create_device cd;

		r = -EFAULT;
		if (copy_from_user(&cd, argp, sizeof(cd)))
			goto out;

		r = kvm_ioctl_create_device(kvm, &cd);
		if (r)
			goto out;

		r = -EFAULT;
		if (copy_to_user(argp, &cd, sizeof(cd)))
			goto out;

		r = 0;
		break;
	}
2915 2916 2917
	case KVM_CHECK_EXTENSION:
		r = kvm_vm_ioctl_check_extension_generic(kvm, arg);
		break;
2918
	default:
2919
		r = kvm_arch_vm_ioctl(filp, ioctl, arg);
2920 2921 2922 2923 2924
	}
out:
	return r;
}

2925
#ifdef CONFIG_KVM_COMPAT
2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968
struct compat_kvm_dirty_log {
	__u32 slot;
	__u32 padding1;
	union {
		compat_uptr_t dirty_bitmap; /* one bit per page */
		__u64 padding2;
	};
};

static long kvm_vm_compat_ioctl(struct file *filp,
			   unsigned int ioctl, unsigned long arg)
{
	struct kvm *kvm = filp->private_data;
	int r;

	if (kvm->mm != current->mm)
		return -EIO;
	switch (ioctl) {
	case KVM_GET_DIRTY_LOG: {
		struct compat_kvm_dirty_log compat_log;
		struct kvm_dirty_log log;

		r = -EFAULT;
		if (copy_from_user(&compat_log, (void __user *)arg,
				   sizeof(compat_log)))
			goto out;
		log.slot	 = compat_log.slot;
		log.padding1	 = compat_log.padding1;
		log.padding2	 = compat_log.padding2;
		log.dirty_bitmap = compat_ptr(compat_log.dirty_bitmap);

		r = kvm_vm_ioctl_get_dirty_log(kvm, &log);
		break;
	}
	default:
		r = kvm_vm_ioctl(filp, ioctl, arg);
	}

out:
	return r;
}
#endif

2969
static struct file_operations kvm_vm_fops = {
2970 2971
	.release        = kvm_vm_release,
	.unlocked_ioctl = kvm_vm_ioctl,
2972
#ifdef CONFIG_KVM_COMPAT
2973 2974
	.compat_ioctl   = kvm_vm_compat_ioctl,
#endif
2975
	.llseek		= noop_llseek,
2976 2977
};

2978
static int kvm_dev_ioctl_create_vm(unsigned long type)
2979
{
2980
	int r;
2981 2982
	struct kvm *kvm;

2983
	kvm = kvm_create_vm(type);
2984 2985
	if (IS_ERR(kvm))
		return PTR_ERR(kvm);
2986 2987 2988 2989 2990 2991 2992
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
	r = kvm_coalesced_mmio_init(kvm);
	if (r < 0) {
		kvm_put_kvm(kvm);
		return r;
	}
#endif
2993
	r = anon_inode_getfd("kvm-vm", &kvm_vm_fops, kvm, O_RDWR | O_CLOEXEC);
2994
	if (r < 0)
2995
		kvm_put_kvm(kvm);
2996

2997
	return r;
2998 2999 3000 3001 3002
}

static long kvm_dev_ioctl(struct file *filp,
			  unsigned int ioctl, unsigned long arg)
{
3003
	long r = -EINVAL;
3004 3005 3006

	switch (ioctl) {
	case KVM_GET_API_VERSION:
3007 3008
		if (arg)
			goto out;
3009 3010 3011
		r = KVM_API_VERSION;
		break;
	case KVM_CREATE_VM:
3012
		r = kvm_dev_ioctl_create_vm(arg);
3013
		break;
3014
	case KVM_CHECK_EXTENSION:
3015
		r = kvm_vm_ioctl_check_extension_generic(NULL, arg);
3016
		break;
3017 3018 3019
	case KVM_GET_VCPU_MMAP_SIZE:
		if (arg)
			goto out;
3020 3021 3022
		r = PAGE_SIZE;     /* struct kvm_run */
#ifdef CONFIG_X86
		r += PAGE_SIZE;    /* pio data page */
3023 3024 3025
#endif
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
		r += PAGE_SIZE;    /* coalesced mmio ring page */
3026
#endif
3027
		break;
3028 3029 3030
	case KVM_TRACE_ENABLE:
	case KVM_TRACE_PAUSE:
	case KVM_TRACE_DISABLE:
3031
		r = -EOPNOTSUPP;
3032
		break;
3033
	default:
3034
		return kvm_arch_dev_ioctl(filp, ioctl, arg);
3035 3036 3037 3038 3039 3040 3041 3042
	}
out:
	return r;
}

static struct file_operations kvm_chardev_ops = {
	.unlocked_ioctl = kvm_dev_ioctl,
	.compat_ioctl   = kvm_dev_ioctl,
3043
	.llseek		= noop_llseek,
3044 3045 3046
};

static struct miscdevice kvm_dev = {
Avi Kivity's avatar
Avi Kivity committed
3047
	KVM_MINOR,
3048 3049 3050 3051
	"kvm",
	&kvm_chardev_ops,
};

3052
static void hardware_enable_nolock(void *junk)
3053 3054
{
	int cpu = raw_smp_processor_id();
3055
	int r;
3056

3057
	if (cpumask_test_cpu(cpu, cpus_hardware_enabled))
3058
		return;
3059

3060
	cpumask_set_cpu(cpu, cpus_hardware_enabled);
3061

3062
	r = kvm_arch_hardware_enable();
3063 3064 3065 3066

	if (r) {
		cpumask_clear_cpu(cpu, cpus_hardware_enabled);
		atomic_inc(&hardware_enable_failed);
3067
		pr_info("kvm: enabling virtualization on CPU%d failed\n", cpu);
3068
	}
3069 3070
}

3071
static void hardware_enable(void)
3072
{
3073
	raw_spin_lock(&kvm_count_lock);
3074 3075
	if (kvm_usage_count)
		hardware_enable_nolock(NULL);
3076
	raw_spin_unlock(&kvm_count_lock);
3077 3078 3079
}

static void hardware_disable_nolock(void *junk)
3080 3081 3082
{
	int cpu = raw_smp_processor_id();

3083
	if (!cpumask_test_cpu(cpu, cpus_hardware_enabled))
3084
		return;
3085
	cpumask_clear_cpu(cpu, cpus_hardware_enabled);
3086
	kvm_arch_hardware_disable();
3087 3088
}

3089
static void hardware_disable(void)
3090
{
3091
	raw_spin_lock(&kvm_count_lock);
3092 3093
	if (kvm_usage_count)
		hardware_disable_nolock(NULL);
3094
	raw_spin_unlock(&kvm_count_lock);
3095 3096
}

3097 3098 3099 3100 3101 3102
static void hardware_disable_all_nolock(void)
{
	BUG_ON(!kvm_usage_count);

	kvm_usage_count--;
	if (!kvm_usage_count)
3103
		on_each_cpu(hardware_disable_nolock, NULL, 1);
3104 3105 3106 3107
}

static void hardware_disable_all(void)
{
3108
	raw_spin_lock(&kvm_count_lock);
3109
	hardware_disable_all_nolock();
3110
	raw_spin_unlock(&kvm_count_lock);
3111 3112 3113 3114 3115 3116
}

static int hardware_enable_all(void)
{
	int r = 0;

3117
	raw_spin_lock(&kvm_count_lock);
3118 3119 3120 3121

	kvm_usage_count++;
	if (kvm_usage_count == 1) {
		atomic_set(&hardware_enable_failed, 0);
3122
		on_each_cpu(hardware_enable_nolock, NULL, 1);
3123 3124 3125 3126 3127 3128 3129

		if (atomic_read(&hardware_enable_failed)) {
			hardware_disable_all_nolock();
			r = -EBUSY;
		}
	}

3130
	raw_spin_unlock(&kvm_count_lock);
3131 3132 3133 3134

	return r;
}

3135 3136 3137
static int kvm_cpu_hotplug(struct notifier_block *notifier, unsigned long val,
			   void *v)
{
3138
	val &= ~CPU_TASKS_FROZEN;
3139
	switch (val) {
3140
	case CPU_DYING:
3141
		hardware_disable();
3142
		break;
3143
	case CPU_STARTING:
3144
		hardware_enable();
3145 3146 3147 3148 3149
		break;
	}
	return NOTIFY_OK;
}

3150
static int kvm_reboot(struct notifier_block *notifier, unsigned long val,
Mike Day's avatar
Mike Day committed
3151
		      void *v)
3152
{
3153 3154 3155 3156 3157 3158
	/*
	 * Some (well, at least mine) BIOSes hang on reboot if
	 * in vmx root mode.
	 *
	 * And Intel TXT required VMX off for all cpu when system shutdown.
	 */
3159
	pr_info("kvm: exiting hardware virtualization\n");
3160
	kvm_rebooting = true;
3161
	on_each_cpu(hardware_disable_nolock, NULL, 1);
3162 3163 3164 3165 3166 3167 3168 3169
	return NOTIFY_OK;
}

static struct notifier_block kvm_reboot_notifier = {
	.notifier_call = kvm_reboot,
	.priority = 0,
};

3170
static void kvm_io_bus_destroy(struct kvm_io_bus *bus)
3171 3172 3173 3174
{
	int i;

	for (i = 0; i < bus->dev_count; i++) {
3175
		struct kvm_io_device *pos = bus->range[i].dev;
3176 3177 3178

		kvm_iodevice_destructor(pos);
	}
3179
	kfree(bus);
3180 3181
}

3182
static inline int kvm_io_bus_cmp(const struct kvm_io_range *r1,
3183
				 const struct kvm_io_range *r2)
3184
{
3185 3186 3187 3188
	gpa_t addr1 = r1->addr;
	gpa_t addr2 = r2->addr;

	if (addr1 < addr2)
3189
		return -1;
3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201

	/* If r2->len == 0, match the exact address.  If r2->len != 0,
	 * accept any overlapping write.  Any order is acceptable for
	 * overlapping ranges, because kvm_io_bus_get_first_dev ensures
	 * we process all of them.
	 */
	if (r2->len) {
		addr1 += r1->len;
		addr2 += r2->len;
	}

	if (addr1 > addr2)
3202
		return 1;
3203

3204 3205 3206
	return 0;
}

3207 3208
static int kvm_io_bus_sort_cmp(const void *p1, const void *p2)
{
3209
	return kvm_io_bus_cmp(p1, p2);
3210 3211
}

3212
static int kvm_io_bus_insert_dev(struct kvm_io_bus *bus, struct kvm_io_device *dev,
3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226
			  gpa_t addr, int len)
{
	bus->range[bus->dev_count++] = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
		.dev = dev,
	};

	sort(bus->range, bus->dev_count, sizeof(struct kvm_io_range),
		kvm_io_bus_sort_cmp, NULL);

	return 0;
}

3227
static int kvm_io_bus_get_first_dev(struct kvm_io_bus *bus,
3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244
			     gpa_t addr, int len)
{
	struct kvm_io_range *range, key;
	int off;

	key = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};

	range = bsearch(&key, bus->range, bus->dev_count,
			sizeof(struct kvm_io_range), kvm_io_bus_sort_cmp);
	if (range == NULL)
		return -ENOENT;

	off = range - bus->range;

3245
	while (off > 0 && kvm_io_bus_cmp(&key, &bus->range[off-1]) == 0)
3246 3247 3248 3249 3250
		off--;

	return off;
}

3251
static int __kvm_io_bus_write(struct kvm_vcpu *vcpu, struct kvm_io_bus *bus,
Cornelia Huck's avatar
Cornelia Huck committed
3252 3253 3254 3255 3256 3257 3258 3259 3260
			      struct kvm_io_range *range, const void *val)
{
	int idx;

	idx = kvm_io_bus_get_first_dev(bus, range->addr, range->len);
	if (idx < 0)
		return -EOPNOTSUPP;

	while (idx < bus->dev_count &&
3261
		kvm_io_bus_cmp(range, &bus->range[idx]) == 0) {
3262
		if (!kvm_iodevice_write(vcpu, bus->range[idx].dev, range->addr,
Cornelia Huck's avatar
Cornelia Huck committed
3263 3264 3265 3266 3267 3268 3269 3270
					range->len, val))
			return idx;
		idx++;
	}

	return -EOPNOTSUPP;
}

3271
/* kvm_io_bus_write - called under kvm->slots_lock */
3272
int kvm_io_bus_write(struct kvm_vcpu *vcpu, enum kvm_bus bus_idx, gpa_t addr,
3273
		     int len, const void *val)
3274
{
3275
	struct kvm_io_bus *bus;
3276
	struct kvm_io_range range;
Cornelia Huck's avatar
Cornelia Huck committed
3277
	int r;
3278 3279 3280 3281 3282

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};
3283

3284
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
3285 3286
	if (!bus)
		return -ENOMEM;
3287
	r = __kvm_io_bus_write(vcpu, bus, &range, val);
Cornelia Huck's avatar
Cornelia Huck committed
3288 3289 3290 3291
	return r < 0 ? r : 0;
}

/* kvm_io_bus_write_cookie - called under kvm->slots_lock */
3292 3293
int kvm_io_bus_write_cookie(struct kvm_vcpu *vcpu, enum kvm_bus bus_idx,
			    gpa_t addr, int len, const void *val, long cookie)
Cornelia Huck's avatar
Cornelia Huck committed
3294 3295 3296 3297 3298 3299 3300 3301 3302
{
	struct kvm_io_bus *bus;
	struct kvm_io_range range;

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};

3303
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
3304 3305
	if (!bus)
		return -ENOMEM;
Cornelia Huck's avatar
Cornelia Huck committed
3306 3307 3308

	/* First try the device referenced by cookie. */
	if ((cookie >= 0) && (cookie < bus->dev_count) &&
3309
	    (kvm_io_bus_cmp(&range, &bus->range[cookie]) == 0))
3310
		if (!kvm_iodevice_write(vcpu, bus->range[cookie].dev, addr, len,
Cornelia Huck's avatar
Cornelia Huck committed
3311 3312 3313 3314 3315 3316 3317
					val))
			return cookie;

	/*
	 * cookie contained garbage; fall back to search and return the
	 * correct cookie value.
	 */
3318
	return __kvm_io_bus_write(vcpu, bus, &range, val);
Cornelia Huck's avatar
Cornelia Huck committed
3319 3320
}

3321 3322
static int __kvm_io_bus_read(struct kvm_vcpu *vcpu, struct kvm_io_bus *bus,
			     struct kvm_io_range *range, void *val)
Cornelia Huck's avatar
Cornelia Huck committed
3323 3324 3325 3326
{
	int idx;

	idx = kvm_io_bus_get_first_dev(bus, range->addr, range->len);
3327 3328 3329 3330
	if (idx < 0)
		return -EOPNOTSUPP;

	while (idx < bus->dev_count &&
3331
		kvm_io_bus_cmp(range, &bus->range[idx]) == 0) {
3332
		if (!kvm_iodevice_read(vcpu, bus->range[idx].dev, range->addr,
Cornelia Huck's avatar
Cornelia Huck committed
3333 3334
				       range->len, val))
			return idx;
3335 3336 3337
		idx++;
	}

3338 3339
	return -EOPNOTSUPP;
}
3340
EXPORT_SYMBOL_GPL(kvm_io_bus_write);
3341

3342
/* kvm_io_bus_read - called under kvm->slots_lock */
3343
int kvm_io_bus_read(struct kvm_vcpu *vcpu, enum kvm_bus bus_idx, gpa_t addr,
3344
		    int len, void *val)
3345
{
3346
	struct kvm_io_bus *bus;
3347
	struct kvm_io_range range;
Cornelia Huck's avatar
Cornelia Huck committed
3348
	int r;
3349 3350 3351 3352 3353

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};
3354

3355
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
3356 3357
	if (!bus)
		return -ENOMEM;
3358
	r = __kvm_io_bus_read(vcpu, bus, &range, val);
Cornelia Huck's avatar
Cornelia Huck committed
3359 3360
	return r < 0 ? r : 0;
}
3361

3362

3363
/* Caller must hold slots_lock. */
3364 3365
int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
			    int len, struct kvm_io_device *dev)
3366
{
3367
	struct kvm_io_bus *new_bus, *bus;
3368

3369
	bus = kvm->buses[bus_idx];
3370 3371 3372
	if (!bus)
		return -ENOMEM;

3373 3374
	/* exclude ioeventfd which is limited by maximum fd */
	if (bus->dev_count - bus->ioeventfd_count > NR_IOBUS_DEVS - 1)
3375
		return -ENOSPC;
3376

3377
	new_bus = kmalloc(sizeof(*bus) + ((bus->dev_count + 1) *
3378
			  sizeof(struct kvm_io_range)), GFP_KERNEL);
3379 3380
	if (!new_bus)
		return -ENOMEM;
3381 3382
	memcpy(new_bus, bus, sizeof(*bus) + (bus->dev_count *
	       sizeof(struct kvm_io_range)));
3383
	kvm_io_bus_insert_dev(new_bus, dev, addr, len);
3384 3385 3386
	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
	synchronize_srcu_expedited(&kvm->srcu);
	kfree(bus);
3387 3388 3389 3390

	return 0;
}

3391
/* Caller must hold slots_lock. */
3392 3393
void kvm_io_bus_unregister_dev(struct kvm *kvm, enum kvm_bus bus_idx,
			       struct kvm_io_device *dev)
3394
{
3395
	int i;
3396
	struct kvm_io_bus *new_bus, *bus;
3397

3398
	bus = kvm->buses[bus_idx];
3399
	if (!bus)
3400
		return;
3401

3402 3403
	for (i = 0; i < bus->dev_count; i++)
		if (bus->range[i].dev == dev) {
3404 3405
			break;
		}
3406

3407 3408
	if (i == bus->dev_count)
		return;
3409

3410
	new_bus = kmalloc(sizeof(*bus) + ((bus->dev_count - 1) *
3411
			  sizeof(struct kvm_io_range)), GFP_KERNEL);
3412 3413 3414 3415
	if (!new_bus)  {
		pr_err("kvm: failed to shrink bus, removing it completely\n");
		goto broken;
	}
3416 3417 3418 3419 3420

	memcpy(new_bus, bus, sizeof(*bus) + i * sizeof(struct kvm_io_range));
	new_bus->dev_count--;
	memcpy(new_bus->range + i, bus->range + i + 1,
	       (new_bus->dev_count - i) * sizeof(struct kvm_io_range));
3421

3422
broken:
3423 3424 3425
	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
	synchronize_srcu_expedited(&kvm->srcu);
	kfree(bus);
3426
	return;
3427 3428
}

3429 3430 3431 3432
static struct notifier_block kvm_cpu_notifier = {
	.notifier_call = kvm_cpu_hotplug,
};

3433
static int vm_stat_get(void *_offset, u64 *val)
3434 3435 3436 3437
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;

3438
	*val = 0;
3439
	mutex_lock(&kvm_lock);
3440
	list_for_each_entry(kvm, &vm_list, vm_list)
3441
		*val += *(u32 *)((void *)kvm + offset);
3442
	mutex_unlock(&kvm_lock);
3443
	return 0;
3444 3445 3446 3447
}

DEFINE_SIMPLE_ATTRIBUTE(vm_stat_fops, vm_stat_get, NULL, "%llu\n");

3448
static int vcpu_stat_get(void *_offset, u64 *val)
Avi Kivity's avatar
Avi Kivity committed
3449 3450 3451 3452 3453 3454
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;
	struct kvm_vcpu *vcpu;
	int i;

3455
	*val = 0;
3456
	mutex_lock(&kvm_lock);
Avi Kivity's avatar
Avi Kivity committed
3457
	list_for_each_entry(kvm, &vm_list, vm_list)
3458 3459 3460
		kvm_for_each_vcpu(i, vcpu, kvm)
			*val += *(u32 *)((void *)vcpu + offset);

3461
	mutex_unlock(&kvm_lock);
3462
	return 0;
Avi Kivity's avatar
Avi Kivity committed
3463 3464
}

3465 3466
DEFINE_SIMPLE_ATTRIBUTE(vcpu_stat_fops, vcpu_stat_get, NULL, "%llu\n");

3467
static const struct file_operations *stat_fops[] = {
3468 3469 3470
	[KVM_STAT_VCPU] = &vcpu_stat_fops,
	[KVM_STAT_VM]   = &vm_stat_fops,
};
Avi Kivity's avatar
Avi Kivity committed
3471

3472
static int kvm_init_debug(void)
3473
{
3474
	int r = -EEXIST;
3475 3476
	struct kvm_stats_debugfs_item *p;

3477
	kvm_debugfs_dir = debugfs_create_dir("kvm", NULL);
3478 3479 3480 3481
	if (kvm_debugfs_dir == NULL)
		goto out;

	for (p = debugfs_entries; p->name; ++p) {
3482
		p->dentry = debugfs_create_file(p->name, 0444, kvm_debugfs_dir,
Avi Kivity's avatar
Avi Kivity committed
3483
						(void *)(long)p->offset,
3484
						stat_fops[p->kind]);
3485 3486 3487 3488 3489 3490 3491 3492 3493 3494
		if (p->dentry == NULL)
			goto out_dir;
	}

	return 0;

out_dir:
	debugfs_remove_recursive(kvm_debugfs_dir);
out:
	return r;
3495 3496 3497 3498 3499 3500 3501 3502
}

static void kvm_exit_debug(void)
{
	struct kvm_stats_debugfs_item *p;

	for (p = debugfs_entries; p->name; ++p)
		debugfs_remove(p->dentry);
3503
	debugfs_remove(kvm_debugfs_dir);
3504 3505
}

3506
static int kvm_suspend(void)
3507
{
3508
	if (kvm_usage_count)
3509
		hardware_disable_nolock(NULL);
3510 3511 3512
	return 0;
}

3513
static void kvm_resume(void)
3514
{
3515
	if (kvm_usage_count) {
3516
		WARN_ON(raw_spin_is_locked(&kvm_count_lock));
3517
		hardware_enable_nolock(NULL);
3518
	}
3519 3520
}

3521
static struct syscore_ops kvm_syscore_ops = {
3522 3523 3524 3525
	.suspend = kvm_suspend,
	.resume = kvm_resume,
};

3526 3527 3528 3529 3530 3531 3532 3533 3534
static inline
struct kvm_vcpu *preempt_notifier_to_vcpu(struct preempt_notifier *pn)
{
	return container_of(pn, struct kvm_vcpu, preempt_notifier);
}

static void kvm_sched_in(struct preempt_notifier *pn, int cpu)
{
	struct kvm_vcpu *vcpu = preempt_notifier_to_vcpu(pn);
3535

3536 3537
	if (vcpu->preempted)
		vcpu->preempted = false;
3538

Radim Krčmář's avatar
Radim Krčmář committed
3539 3540
	kvm_arch_sched_in(vcpu, cpu);

3541
	kvm_arch_vcpu_load(vcpu, cpu);
3542 3543 3544 3545 3546 3547 3548
}

static void kvm_sched_out(struct preempt_notifier *pn,
			  struct task_struct *next)
{
	struct kvm_vcpu *vcpu = preempt_notifier_to_vcpu(pn);

3549 3550
	if (current->state == TASK_RUNNING)
		vcpu->preempted = true;
3551
	kvm_arch_vcpu_put(vcpu);
3552 3553
}

3554
int kvm_init(void *opaque, unsigned vcpu_size, unsigned vcpu_align,
3555
		  struct module *module)
3556 3557
{
	int r;
3558
	int cpu;
3559

3560 3561
	r = kvm_arch_init(opaque);
	if (r)
3562
		goto out_fail;
3563

3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574
	/*
	 * kvm_arch_init makes sure there's at most one caller
	 * for architectures that support multiple implementations,
	 * like intel and amd on x86.
	 * kvm_arch_init must be called before kvm_irqfd_init to avoid creating
	 * conflicts in case kvm is already setup for another implementation.
	 */
	r = kvm_irqfd_init();
	if (r)
		goto out_irqfd;

3575
	if (!zalloc_cpumask_var(&cpus_hardware_enabled, GFP_KERNEL)) {
3576 3577 3578 3579
		r = -ENOMEM;
		goto out_free_0;
	}

3580
	r = kvm_arch_hardware_setup();
3581
	if (r < 0)
3582
		goto out_free_0a;
3583

3584 3585
	for_each_online_cpu(cpu) {
		smp_call_function_single(cpu,
3586
				kvm_arch_check_processor_compat,
3587
				&r, 1);
3588
		if (r < 0)
3589
			goto out_free_1;
3590 3591
	}

3592 3593
	r = register_cpu_notifier(&kvm_cpu_notifier);
	if (r)
3594
		goto out_free_2;
3595 3596
	register_reboot_notifier(&kvm_reboot_notifier);

3597
	/* A kmem cache lets us meet the alignment requirements of fx_save. */
3598 3599 3600
	if (!vcpu_align)
		vcpu_align = __alignof__(struct kvm_vcpu);
	kvm_vcpu_cache = kmem_cache_create("kvm_vcpu", vcpu_size, vcpu_align,
3601
					   0, NULL);
3602 3603
	if (!kvm_vcpu_cache) {
		r = -ENOMEM;
3604
		goto out_free_3;
3605 3606
	}

3607 3608 3609 3610
	r = kvm_async_pf_init();
	if (r)
		goto out_free;

3611
	kvm_chardev_ops.owner = module;
3612 3613
	kvm_vm_fops.owner = module;
	kvm_vcpu_fops.owner = module;
3614 3615 3616

	r = misc_register(&kvm_dev);
	if (r) {
3617
		pr_err("kvm: misc device register failed\n");
3618
		goto out_unreg;
3619 3620
	}

3621 3622
	register_syscore_ops(&kvm_syscore_ops);

3623 3624 3625
	kvm_preempt_ops.sched_in = kvm_sched_in;
	kvm_preempt_ops.sched_out = kvm_sched_out;

3626 3627
	r = kvm_init_debug();
	if (r) {
3628
		pr_err("kvm: create debugfs files failed\n");
3629 3630
		goto out_undebugfs;
	}
3631

3632 3633 3634
	r = kvm_vfio_ops_init();
	WARN_ON(r);

3635
	return 0;
3636

3637 3638
out_undebugfs:
	unregister_syscore_ops(&kvm_syscore_ops);
3639
	misc_deregister(&kvm_dev);
3640 3641
out_unreg:
	kvm_async_pf_deinit();
3642
out_free:
3643
	kmem_cache_destroy(kvm_vcpu_cache);
3644
out_free_3:
3645
	unregister_reboot_notifier(&kvm_reboot_notifier);
3646
	unregister_cpu_notifier(&kvm_cpu_notifier);
3647 3648
out_free_2:
out_free_1:
3649
	kvm_arch_hardware_unsetup();
3650 3651
out_free_0a:
	free_cpumask_var(cpus_hardware_enabled);
3652
out_free_0:
3653 3654
	kvm_irqfd_exit();
out_irqfd:
3655 3656
	kvm_arch_exit();
out_fail:
3657 3658
	return r;
}
3659
EXPORT_SYMBOL_GPL(kvm_init);
3660

3661
void kvm_exit(void)
3662
{
3663
	kvm_exit_debug();
3664
	misc_deregister(&kvm_dev);
3665
	kmem_cache_destroy(kvm_vcpu_cache);
3666
	kvm_async_pf_deinit();
3667
	unregister_syscore_ops(&kvm_syscore_ops);
3668
	unregister_reboot_notifier(&kvm_reboot_notifier);
3669
	unregister_cpu_notifier(&kvm_cpu_notifier);
3670
	on_each_cpu(hardware_disable_nolock, NULL, 1);
3671
	kvm_arch_hardware_unsetup();
3672
	kvm_arch_exit();
3673
	kvm_irqfd_exit();
3674
	free_cpumask_var(cpus_hardware_enabled);
3675
	kvm_vfio_ops_exit();
3676
}
3677
EXPORT_SYMBOL_GPL(kvm_exit);