kvm_main.c 94.2 KB
Newer Older
A
Avi Kivity 已提交
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.
N
Nicolas Kaiser 已提交
8
 * Copyright 2010 Red Hat, Inc. and/or its affiliates.
A
Avi Kivity 已提交
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>
A
Avi Kivity 已提交
20

21
#include <linux/kvm_host.h>
A
Avi Kivity 已提交
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>
A
Avi Kivity 已提交
34
#include <linux/cpu.h>
A
Alexey Dobriyan 已提交
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>
J
Joerg Roedel 已提交
48
#include <linux/hugetlb.h>
49
#include <linux/slab.h>
50 51
#include <linux/sort.h>
#include <linux/bsearch.h>
A
Avi Kivity 已提交
52

A
Avi Kivity 已提交
53 54
#include <asm/processor.h>
#include <asm/io.h>
55
#include <asm/ioctl.h>
56
#include <linux/uaccess.h>
57
#include <asm/pgtable.h>
A
Avi Kivity 已提交
58

59
#include "coalesced_mmio.h"
60
#include "async_pf.h"
P
Paolo Bonzini 已提交
61
#include "vfio.h"
62

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

66 67 68
/* Worst case buffer size needed for holding an integer. */
#define ITOA_MAX_LEN 12

A
Avi Kivity 已提交
69 70 71
MODULE_AUTHOR("Qumranet");
MODULE_LICENSE("GPL");

72
/* Architectures should define their poll value according to the halt latency */
73
unsigned int halt_poll_ns = KVM_HALT_POLL_NS_DEFAULT;
74
module_param(halt_poll_ns, uint, S_IRUGO | S_IWUSR);
75
EXPORT_SYMBOL_GPL(halt_poll_ns);
76

W
Wanpeng Li 已提交
77
/* Default doubles per-vcpu halt_poll_ns. */
78
unsigned int halt_poll_ns_grow = 2;
79
module_param(halt_poll_ns_grow, uint, S_IRUGO | S_IWUSR);
80
EXPORT_SYMBOL_GPL(halt_poll_ns_grow);
W
Wanpeng Li 已提交
81 82

/* Default resets per-vcpu halt_poll_ns . */
83
unsigned int halt_poll_ns_shrink;
84
module_param(halt_poll_ns_shrink, uint, S_IRUGO | S_IWUSR);
85
EXPORT_SYMBOL_GPL(halt_poll_ns_shrink);
W
Wanpeng Li 已提交
86

87 88 89
/*
 * Ordering of locks:
 *
90
 *	kvm->lock --> kvm->slots_lock --> kvm->irq_lock
91 92
 */

93
DEFINE_SPINLOCK(kvm_lock);
94
static DEFINE_RAW_SPINLOCK(kvm_count_lock);
95
LIST_HEAD(vm_list);
96

97
static cpumask_var_t cpus_hardware_enabled;
98
static int kvm_usage_count;
99
static atomic_t hardware_enable_failed;
100

101 102
struct kmem_cache *kvm_vcpu_cache;
EXPORT_SYMBOL_GPL(kvm_vcpu_cache);
A
Avi Kivity 已提交
103

104 105
static __read_mostly struct preempt_ops kvm_preempt_ops;

106
struct dentry *kvm_debugfs_dir;
107
EXPORT_SYMBOL_GPL(kvm_debugfs_dir);
A
Avi Kivity 已提交
108

109 110 111
static int kvm_debugfs_num_entries;
static const struct file_operations *stat_fops_per_vm[];

A
Avi Kivity 已提交
112 113
static long kvm_vcpu_ioctl(struct file *file, unsigned int ioctl,
			   unsigned long arg);
114
#ifdef CONFIG_KVM_COMPAT
115 116 117
static long kvm_vcpu_compat_ioctl(struct file *file, unsigned int ioctl,
				  unsigned long arg);
#endif
118 119
static int hardware_enable_all(void);
static void hardware_disable_all(void);
A
Avi Kivity 已提交
120

M
Marcelo Tosatti 已提交
121
static void kvm_io_bus_destroy(struct kvm_io_bus *bus);
122

D
Dan Williams 已提交
123
static void kvm_release_pfn_dirty(kvm_pfn_t pfn);
124
static void mark_page_dirty_in_slot(struct kvm_memory_slot *memslot, gfn_t gfn);
M
Marcelo Tosatti 已提交
125

126
__visible bool kvm_rebooting;
127
EXPORT_SYMBOL_GPL(kvm_rebooting);
128

129 130
static bool largepages_enabled = true;

D
Dan Williams 已提交
131
bool kvm_is_reserved_pfn(kvm_pfn_t pfn)
B
Ben-Ami Yassour 已提交
132
{
133
	if (pfn_valid(pfn))
134
		return PageReserved(pfn_to_page(pfn));
B
Ben-Ami Yassour 已提交
135 136 137 138

	return true;
}

A
Avi Kivity 已提交
139 140 141
/*
 * Switches to specified vcpu, until a matching vcpu_put()
 */
142
int vcpu_load(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
143
{
144 145
	int cpu;

146 147
	if (mutex_lock_killable(&vcpu->mutex))
		return -EINTR;
148 149
	cpu = get_cpu();
	preempt_notifier_register(&vcpu->preempt_notifier);
150
	kvm_arch_vcpu_load(vcpu, cpu);
151
	put_cpu();
152
	return 0;
A
Avi Kivity 已提交
153
}
154
EXPORT_SYMBOL_GPL(vcpu_load);
A
Avi Kivity 已提交
155

156
void vcpu_put(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
157
{
158
	preempt_disable();
159
	kvm_arch_vcpu_put(vcpu);
160 161
	preempt_notifier_unregister(&vcpu->preempt_notifier);
	preempt_enable();
A
Avi Kivity 已提交
162 163
	mutex_unlock(&vcpu->mutex);
}
164
EXPORT_SYMBOL_GPL(vcpu_put);
A
Avi Kivity 已提交
165

166 167 168 169
static void ack_flush(void *_completed)
{
}

170
bool kvm_make_all_cpus_request(struct kvm *kvm, unsigned int req)
171
{
172
	int i, cpu, me;
173 174
	cpumask_var_t cpus;
	bool called = true;
175 176
	struct kvm_vcpu *vcpu;

177
	zalloc_cpumask_var(&cpus, GFP_ATOMIC);
178

179
	me = get_cpu();
180
	kvm_for_each_vcpu(i, vcpu, kvm) {
181
		kvm_make_request(req, vcpu);
182
		cpu = vcpu->cpu;
183

184 185
		/* Set ->requests bit before we read ->mode. */
		smp_mb__after_atomic();
186 187 188

		if (cpus != NULL && cpu != -1 && cpu != me &&
		      kvm_vcpu_exiting_guest_mode(vcpu) != OUTSIDE_GUEST_MODE)
189
			cpumask_set_cpu(cpu, cpus);
190
	}
191 192 193 194 195 196
	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;
197
	put_cpu();
198
	free_cpumask_var(cpus);
199
	return called;
200 201
}

202
#ifndef CONFIG_HAVE_KVM_ARCH_TLB_FLUSH_ALL
203
void kvm_flush_remote_tlbs(struct kvm *kvm)
204
{
205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221
	/*
	 * Read tlbs_dirty before setting KVM_REQ_TLB_FLUSH in
	 * kvm_make_all_cpus_request.
	 */
	long dirty_count = smp_load_acquire(&kvm->tlbs_dirty);

	/*
	 * We want to publish modifications to the page tables before reading
	 * mode. Pairs with a memory barrier in arch-specific code.
	 * - x86: smp_mb__after_srcu_read_unlock in vcpu_enter_guest
	 * and smp_mb in walk_shadow_page_lockless_begin/end.
	 * - powerpc: smp_mb in kvmppc_prepare_to_enter.
	 *
	 * There is already an smp_mb__after_atomic() before
	 * kvm_make_all_cpus_request() reads vcpu->mode. We reuse that
	 * barrier here.
	 */
222
	if (kvm_make_all_cpus_request(kvm, KVM_REQ_TLB_FLUSH))
223
		++kvm->stat.remote_tlb_flush;
224
	cmpxchg(&kvm->tlbs_dirty, dirty_count, 0);
225
}
226
EXPORT_SYMBOL_GPL(kvm_flush_remote_tlbs);
227
#endif
228

229 230
void kvm_reload_remote_mmus(struct kvm *kvm)
{
231
	kvm_make_all_cpus_request(kvm, KVM_REQ_MMU_RELOAD);
232
}
233

R
Rusty Russell 已提交
234 235 236 237 238 239 240 241 242
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;
243
	vcpu->pid = NULL;
244
	init_swait_queue_head(&vcpu->wq);
245
	kvm_async_pf_vcpu_init(vcpu);
R
Rusty Russell 已提交
246

247 248 249
	vcpu->pre_pcpu = -1;
	INIT_LIST_HEAD(&vcpu->blocked_vcpu_list);

R
Rusty Russell 已提交
250 251 252 253 254 255 256
	page = alloc_page(GFP_KERNEL | __GFP_ZERO);
	if (!page) {
		r = -ENOMEM;
		goto fail;
	}
	vcpu->run = page_address(page);

257 258
	kvm_vcpu_set_in_spin_loop(vcpu, false);
	kvm_vcpu_set_dy_eligible(vcpu, false);
259
	vcpu->preempted = false;
260

261
	r = kvm_arch_vcpu_init(vcpu);
R
Rusty Russell 已提交
262
	if (r < 0)
263
		goto fail_free_run;
R
Rusty Russell 已提交
264 265 266 267 268
	return 0;

fail_free_run:
	free_page((unsigned long)vcpu->run);
fail:
269
	return r;
R
Rusty Russell 已提交
270 271 272 273 274
}
EXPORT_SYMBOL_GPL(kvm_vcpu_init);

void kvm_vcpu_uninit(struct kvm_vcpu *vcpu)
{
275
	put_pid(vcpu->pid);
276
	kvm_arch_vcpu_uninit(vcpu);
R
Rusty Russell 已提交
277 278 279 280
	free_page((unsigned long)vcpu->run);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_uninit);

281 282 283 284 285 286 287 288 289 290 291
#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);
292
	int need_tlb_flush, idx;
293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311

	/*
	 * 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.
	 */
312
	idx = srcu_read_lock(&kvm->srcu);
313
	spin_lock(&kvm->mmu_lock);
314

315
	kvm->mmu_notifier_seq++;
316
	need_tlb_flush = kvm_unmap_hva(kvm, address) | kvm->tlbs_dirty;
317 318 319 320
	/* we've to flush the tlb before the pages can be freed */
	if (need_tlb_flush)
		kvm_flush_remote_tlbs(kvm);

321
	spin_unlock(&kvm->mmu_lock);
322 323 324

	kvm_arch_mmu_notifier_invalidate_page(kvm, address);

325
	srcu_read_unlock(&kvm->srcu, idx);
326 327
}

328 329 330 331 332 333
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);
334
	int idx;
335

336
	idx = srcu_read_lock(&kvm->srcu);
337 338 339 340
	spin_lock(&kvm->mmu_lock);
	kvm->mmu_notifier_seq++;
	kvm_set_spte_hva(kvm, address, pte);
	spin_unlock(&kvm->mmu_lock);
341
	srcu_read_unlock(&kvm->srcu, idx);
342 343
}

344 345 346 347 348 349
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);
350
	int need_tlb_flush = 0, idx;
351

352
	idx = srcu_read_lock(&kvm->srcu);
353 354 355 356 357 358 359
	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++;
360
	need_tlb_flush = kvm_unmap_hva_range(kvm, start, end);
361
	need_tlb_flush |= kvm->tlbs_dirty;
362 363 364
	/* we've to flush the tlb before the pages can be freed */
	if (need_tlb_flush)
		kvm_flush_remote_tlbs(kvm);
365 366 367

	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);
368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383
}

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++;
384
	smp_wmb();
385 386
	/*
	 * The above sequence increase must be visible before the
387 388
	 * below count decrease, which is ensured by the smp_wmb above
	 * in conjunction with the smp_rmb in mmu_notifier_retry().
389 390 391 392 393 394 395 396 397
	 */
	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,
A
Andres Lagar-Cavilla 已提交
398 399
					      unsigned long start,
					      unsigned long end)
400 401
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
402
	int young, idx;
403

404
	idx = srcu_read_lock(&kvm->srcu);
405 406
	spin_lock(&kvm->mmu_lock);

A
Andres Lagar-Cavilla 已提交
407
	young = kvm_age_hva(kvm, start, end);
408 409 410
	if (young)
		kvm_flush_remote_tlbs(kvm);

411 412 413
	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);

414 415 416
	return young;
}

417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446
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;
}

A
Andrea Arcangeli 已提交
447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462
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;
}

463 464 465 466
static void kvm_mmu_notifier_release(struct mmu_notifier *mn,
				     struct mm_struct *mm)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
467 468 469
	int idx;

	idx = srcu_read_lock(&kvm->srcu);
470
	kvm_arch_flush_shadow_all(kvm);
471
	srcu_read_unlock(&kvm->srcu, idx);
472 473
}

474 475 476 477 478
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,
479
	.clear_young		= kvm_mmu_notifier_clear_young,
A
Andrea Arcangeli 已提交
480
	.test_young		= kvm_mmu_notifier_test_young,
481
	.change_pte		= kvm_mmu_notifier_change_pte,
482
	.release		= kvm_mmu_notifier_release,
483
};
484 485 486 487 488 489 490 491 492 493 494 495 496 497

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;
}

498 499
#endif /* CONFIG_MMU_NOTIFIER && KVM_ARCH_WANT_MMU_NOTIFIER */

500
static struct kvm_memslots *kvm_alloc_memslots(void)
501 502
{
	int i;
503
	struct kvm_memslots *slots;
504

505 506 507 508
	slots = kvm_kvzalloc(sizeof(struct kvm_memslots));
	if (!slots)
		return NULL;

509
	for (i = 0; i < KVM_MEM_SLOTS_NUM; i++)
510
		slots->id_to_index[i] = slots->memslots[i].id = i;
511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548

	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);
549 550
}

551 552 553 554 555 556 557 558 559
static void kvm_destroy_vm_debugfs(struct kvm *kvm)
{
	int i;

	if (!kvm->debugfs_dentry)
		return;

	debugfs_remove_recursive(kvm->debugfs_dentry);

560 561 562 563 564
	if (kvm->debugfs_stat_data) {
		for (i = 0; i < kvm_debugfs_num_entries; i++)
			kfree(kvm->debugfs_stat_data[i]);
		kfree(kvm->debugfs_stat_data);
	}
565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595
}

static int kvm_create_vm_debugfs(struct kvm *kvm, int fd)
{
	char dir_name[ITOA_MAX_LEN * 2];
	struct kvm_stat_data *stat_data;
	struct kvm_stats_debugfs_item *p;

	if (!debugfs_initialized())
		return 0;

	snprintf(dir_name, sizeof(dir_name), "%d-%d", task_pid_nr(current), fd);
	kvm->debugfs_dentry = debugfs_create_dir(dir_name,
						 kvm_debugfs_dir);
	if (!kvm->debugfs_dentry)
		return -ENOMEM;

	kvm->debugfs_stat_data = kcalloc(kvm_debugfs_num_entries,
					 sizeof(*kvm->debugfs_stat_data),
					 GFP_KERNEL);
	if (!kvm->debugfs_stat_data)
		return -ENOMEM;

	for (p = debugfs_entries; p->name; p++) {
		stat_data = kzalloc(sizeof(*stat_data), GFP_KERNEL);
		if (!stat_data)
			return -ENOMEM;

		stat_data->kvm = kvm;
		stat_data->offset = p->offset;
		kvm->debugfs_stat_data[p - debugfs_entries] = stat_data;
596
		if (!debugfs_create_file(p->name, 0644,
597 598 599 600 601 602 603 604
					 kvm->debugfs_dentry,
					 stat_data,
					 stat_fops_per_vm[p->kind]))
			return -ENOMEM;
	}
	return 0;
}

605
static struct kvm *kvm_create_vm(unsigned long type)
A
Avi Kivity 已提交
606
{
607 608
	int r, i;
	struct kvm *kvm = kvm_arch_alloc_vm();
A
Avi Kivity 已提交
609

610 611 612
	if (!kvm)
		return ERR_PTR(-ENOMEM);

613 614 615 616 617 618 619 620 621 622
	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);

623
	r = kvm_arch_init_vm(kvm, type);
624
	if (r)
625
		goto out_err_no_disable;
626 627 628

	r = hardware_enable_all();
	if (r)
629
		goto out_err_no_disable;
630

631
#ifdef CONFIG_HAVE_KVM_IRQFD
632
	INIT_HLIST_HEAD(&kvm->irq_ack_notifier_list);
633
#endif
A
Avi Kivity 已提交
634

635 636
	BUILD_BUG_ON(KVM_MEM_SLOTS_NUM > SHRT_MAX);

637
	r = -ENOMEM;
638
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++) {
639 640
		struct kvm_memslots *slots = kvm_alloc_memslots();
		if (!slots)
641
			goto out_err_no_srcu;
642 643 644 645 646 647 648
		/*
		 * Generations must be different for each address space.
		 * Init kvm generation close to the maximum to easily test the
		 * code of handling generation number wrap-around.
		 */
		slots->generation = i * 2 - 150;
		rcu_assign_pointer(kvm->memslots[i], slots);
649
	}
650

651
	if (init_srcu_struct(&kvm->srcu))
652 653 654
		goto out_err_no_srcu;
	if (init_srcu_struct(&kvm->irq_srcu))
		goto out_err_no_irq_srcu;
M
Marcelo Tosatti 已提交
655 656 657
	for (i = 0; i < KVM_NR_BUSES; i++) {
		kvm->buses[i] = kzalloc(sizeof(struct kvm_io_bus),
					GFP_KERNEL);
658
		if (!kvm->buses[i])
M
Marcelo Tosatti 已提交
659 660
			goto out_err;
	}
661

662 663 664 665
	r = kvm_init_mmu_notifier(kvm);
	if (r)
		goto out_err;

666
	spin_lock(&kvm_lock);
667
	list_add(&kvm->vm_list, &vm_list);
668
	spin_unlock(&kvm_lock);
669

670 671
	preempt_notifier_inc();

672
	return kvm;
673 674

out_err:
675 676
	cleanup_srcu_struct(&kvm->irq_srcu);
out_err_no_irq_srcu:
677
	cleanup_srcu_struct(&kvm->srcu);
678
out_err_no_srcu:
679
	hardware_disable_all();
680
out_err_no_disable:
M
Marcelo Tosatti 已提交
681 682
	for (i = 0; i < KVM_NR_BUSES; i++)
		kfree(kvm->buses[i]);
683 684
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++)
		kvm_free_memslots(kvm, kvm->memslots[i]);
685
	kvm_arch_free_vm(kvm);
686
	mmdrop(current->mm);
687
	return ERR_PTR(r);
688 689
}

690 691 692 693
/*
 * Avoid using vmalloc for a small buffer.
 * Should not be used when the size is statically known.
 */
694
void *kvm_kvzalloc(unsigned long size)
695 696 697 698 699 700 701
{
	if (size > PAGE_SIZE)
		return vzalloc(size);
	else
		return kzalloc(size, GFP_KERNEL);
}

702 703
static void kvm_destroy_devices(struct kvm *kvm)
{
G
Geliang Tang 已提交
704
	struct kvm_device *dev, *tmp;
705

706 707 708 709 710
	/*
	 * We do not need to take the kvm->lock here, because nobody else
	 * has a reference to the struct kvm at this point and therefore
	 * cannot access the devices list anyhow.
	 */
G
Geliang Tang 已提交
711 712
	list_for_each_entry_safe(dev, tmp, &kvm->devices, vm_node) {
		list_del(&dev->vm_node);
713 714 715 716
		dev->ops->destroy(dev);
	}
}

717 718
static void kvm_destroy_vm(struct kvm *kvm)
{
M
Marcelo Tosatti 已提交
719
	int i;
720 721
	struct mm_struct *mm = kvm->mm;

722
	kvm_destroy_vm_debugfs(kvm);
723
	kvm_arch_sync_events(kvm);
724
	spin_lock(&kvm_lock);
725
	list_del(&kvm->vm_list);
726
	spin_unlock(&kvm_lock);
727
	kvm_free_irq_routing(kvm);
M
Marcelo Tosatti 已提交
728 729
	for (i = 0; i < KVM_NR_BUSES; i++)
		kvm_io_bus_destroy(kvm->buses[i]);
730
	kvm_coalesced_mmio_free(kvm);
731 732
#if defined(CONFIG_MMU_NOTIFIER) && defined(KVM_ARCH_WANT_MMU_NOTIFIER)
	mmu_notifier_unregister(&kvm->mmu_notifier, kvm->mm);
733
#else
734
	kvm_arch_flush_shadow_all(kvm);
735
#endif
736
	kvm_arch_destroy_vm(kvm);
737
	kvm_destroy_devices(kvm);
738 739
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++)
		kvm_free_memslots(kvm, kvm->memslots[i]);
740
	cleanup_srcu_struct(&kvm->irq_srcu);
741 742
	cleanup_srcu_struct(&kvm->srcu);
	kvm_arch_free_vm(kvm);
743
	preempt_notifier_dec();
744
	hardware_disable_all();
745
	mmdrop(mm);
746 747
}

I
Izik Eidus 已提交
748 749 750 751 752 753 754 755 756 757 758 759 760 761
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);


762 763 764 765
static int kvm_vm_release(struct inode *inode, struct file *filp)
{
	struct kvm *kvm = filp->private_data;

G
Gregory Haskins 已提交
766 767
	kvm_irqfd_release(kvm);

I
Izik Eidus 已提交
768
	kvm_put_kvm(kvm);
A
Avi Kivity 已提交
769 770 771
	return 0;
}

772 773
/*
 * Allocation size is twice as large as the actual dirty bitmap size.
774
 * See x86's kvm_vm_ioctl_get_dirty_log() why this is needed.
775
 */
776 777
static int kvm_create_dirty_bitmap(struct kvm_memory_slot *memslot)
{
778
	unsigned long dirty_bytes = 2 * kvm_dirty_bitmap_bytes(memslot);
779

780
	memslot->dirty_bitmap = kvm_kvzalloc(dirty_bytes);
781 782 783 784 785 786
	if (!memslot->dirty_bitmap)
		return -ENOMEM;

	return 0;
}

787
/*
788 789 790 791
 * 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.
792
 */
793 794
static void update_memslots(struct kvm_memslots *slots,
			    struct kvm_memory_slot *new)
795
{
796 797
	int id = new->id;
	int i = slots->id_to_index[id];
798
	struct kvm_memory_slot *mslots = slots->memslots;
799

800
	WARN_ON(mslots[i].id != id);
801
	if (!new->npages) {
802
		WARN_ON(!mslots[i].npages);
803 804 805 806 807 808
		if (mslots[i].npages)
			slots->used_slots--;
	} else {
		if (!mslots[i].npages)
			slots->used_slots++;
	}
809

810
	while (i < KVM_MEM_SLOTS_NUM - 1 &&
811 812 813
	       new->base_gfn <= mslots[i + 1].base_gfn) {
		if (!mslots[i + 1].npages)
			break;
814 815 816 817
		mslots[i] = mslots[i + 1];
		slots->id_to_index[mslots[i].id] = i;
		i++;
	}
818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834

	/*
	 * 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--;
		}
835 836
	} else
		WARN_ON_ONCE(i != slots->used_slots);
837

838 839
	mslots[i] = *new;
	slots->id_to_index[mslots[i].id] = i;
840 841
}

842
static int check_memory_region_flags(const struct kvm_userspace_memory_region *mem)
843
{
X
Xiao Guangrong 已提交
844 845
	u32 valid_flags = KVM_MEM_LOG_DIRTY_PAGES;

846
#ifdef __KVM_HAVE_READONLY_MEM
X
Xiao Guangrong 已提交
847 848 849 850
	valid_flags |= KVM_MEM_READONLY;
#endif

	if (mem->flags & ~valid_flags)
851 852 853 854 855
		return -EINVAL;

	return 0;
}

856
static struct kvm_memslots *install_new_memslots(struct kvm *kvm,
857
		int as_id, struct kvm_memslots *slots)
858
{
859
	struct kvm_memslots *old_memslots = __kvm_memslots(kvm, as_id);
860

861 862 863 864 865 866 867
	/*
	 * 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;

868
	rcu_assign_pointer(kvm->memslots[as_id], slots);
869
	synchronize_srcu_expedited(&kvm->srcu);
870

871 872 873 874
	/*
	 * 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.
875 876 877 878 879 880
	 *
	 * Generations must be unique even across address spaces.  We do not need
	 * a global counter for that, instead the generation space is evenly split
	 * across address spaces.  For example, with two address spaces, address
	 * space 0 will use generations 0, 4, 8, ... while * address space 1 will
	 * use generations 2, 6, 10, 14, ...
881
	 */
882
	slots->generation += KVM_ADDRESS_SPACE_NUM * 2 - 1;
883

884
	kvm_arch_memslots_updated(kvm, slots);
885 886

	return old_memslots;
887 888
}

A
Avi Kivity 已提交
889 890 891 892 893
/*
 * Allocate some memory and give it an address in the guest physical address
 * space.
 *
 * Discontiguous memory is allowed, mostly for framebuffers.
894
 *
895
 * Must be called holding kvm->slots_lock for write.
A
Avi Kivity 已提交
896
 */
897
int __kvm_set_memory_region(struct kvm *kvm,
898
			    const struct kvm_userspace_memory_region *mem)
A
Avi Kivity 已提交
899
{
900
	int r;
A
Avi Kivity 已提交
901
	gfn_t base_gfn;
902
	unsigned long npages;
903
	struct kvm_memory_slot *slot;
A
Avi Kivity 已提交
904
	struct kvm_memory_slot old, new;
905
	struct kvm_memslots *slots = NULL, *old_memslots;
906
	int as_id, id;
907
	enum kvm_mr_change change;
A
Avi Kivity 已提交
908

909 910 911 912
	r = check_memory_region_flags(mem);
	if (r)
		goto out;

A
Avi Kivity 已提交
913
	r = -EINVAL;
914 915 916
	as_id = mem->slot >> 16;
	id = (u16)mem->slot;

A
Avi Kivity 已提交
917 918 919 920 921
	/* General sanity checks */
	if (mem->memory_size & (PAGE_SIZE - 1))
		goto out;
	if (mem->guest_phys_addr & (PAGE_SIZE - 1))
		goto out;
922
	/* We can read the guest memory with __xxx_user() later on. */
923
	if ((id < KVM_USER_MEM_SLOTS) &&
924
	    ((mem->userspace_addr & (PAGE_SIZE - 1)) ||
925 926 927
	     !access_ok(VERIFY_WRITE,
			(void __user *)(unsigned long)mem->userspace_addr,
			mem->memory_size)))
928
		goto out;
929
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_MEM_SLOTS_NUM)
A
Avi Kivity 已提交
930 931 932 933
		goto out;
	if (mem->guest_phys_addr + mem->memory_size < mem->guest_phys_addr)
		goto out;

934
	slot = id_to_memslot(__kvm_memslots(kvm, as_id), id);
A
Avi Kivity 已提交
935 936 937
	base_gfn = mem->guest_phys_addr >> PAGE_SHIFT;
	npages = mem->memory_size >> PAGE_SHIFT;

938 939 940
	if (npages > KVM_MEM_MAX_NR_PAGES)
		goto out;

941
	new = old = *slot;
A
Avi Kivity 已提交
942

943
	new.id = id;
A
Avi Kivity 已提交
944 945 946 947
	new.base_gfn = base_gfn;
	new.npages = npages;
	new.flags = mem->flags;

948 949 950 951 952
	if (npages) {
		if (!old.npages)
			change = KVM_MR_CREATE;
		else { /* Modify an existing slot. */
			if ((mem->userspace_addr != old.userspace_addr) ||
953 954
			    (npages != old.npages) ||
			    ((new.flags ^ old.flags) & KVM_MEM_READONLY))
955 956 957 958 959 960 961 962 963 964 965
				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;
			}
		}
966 967 968 969
	} else {
		if (!old.npages)
			goto out;

970
		change = KVM_MR_DELETE;
971 972 973
		new.base_gfn = 0;
		new.flags = 0;
	}
A
Avi Kivity 已提交
974

975
	if ((change == KVM_MR_CREATE) || (change == KVM_MR_MOVE)) {
976 977
		/* Check for overlaps */
		r = -EEXIST;
978
		kvm_for_each_memslot(slot, __kvm_memslots(kvm, as_id)) {
979
			if ((slot->id >= KVM_USER_MEM_SLOTS) ||
980
			    (slot->id == id))
981 982 983 984 985
				continue;
			if (!((base_gfn + npages <= slot->base_gfn) ||
			      (base_gfn >= slot->base_gfn + slot->npages)))
				goto out;
		}
A
Avi Kivity 已提交
986 987 988 989
	}

	/* Free page dirty bitmap if unneeded */
	if (!(new.flags & KVM_MEM_LOG_DIRTY_PAGES))
A
Al Viro 已提交
990
		new.dirty_bitmap = NULL;
A
Avi Kivity 已提交
991 992

	r = -ENOMEM;
993
	if (change == KVM_MR_CREATE) {
994
		new.userspace_addr = mem->userspace_addr;
995

996
		if (kvm_arch_create_memslot(kvm, &new, npages))
997
			goto out_free;
A
Avi Kivity 已提交
998
	}
999

A
Avi Kivity 已提交
1000 1001
	/* Allocate page dirty bitmap if needed */
	if ((new.flags & KVM_MEM_LOG_DIRTY_PAGES) && !new.dirty_bitmap) {
1002
		if (kvm_create_dirty_bitmap(&new) < 0)
1003
			goto out_free;
A
Avi Kivity 已提交
1004 1005
	}

1006
	slots = kvm_kvzalloc(sizeof(struct kvm_memslots));
1007 1008
	if (!slots)
		goto out_free;
1009
	memcpy(slots, __kvm_memslots(kvm, as_id), sizeof(struct kvm_memslots));
1010

1011
	if ((change == KVM_MR_DELETE) || (change == KVM_MR_MOVE)) {
1012
		slot = id_to_memslot(slots, id);
1013 1014
		slot->flags |= KVM_MEMSLOT_INVALID;

1015
		old_memslots = install_new_memslots(kvm, as_id, slots);
1016

1017 1018
		/* slot was deleted or moved, clear iommu mapping */
		kvm_iommu_unmap_pages(kvm, &old);
1019 1020
		/* From this point no new shadow pages pointing to a deleted,
		 * or moved, memslot will be created.
1021 1022
		 *
		 * validation of sp->gfn happens in:
1023 1024
		 *	- gfn_to_hva (kvm_read_guest, gfn_to_pfn)
		 *	- kvm_is_visible_gfn (mmu_check_roots)
1025
		 */
1026
		kvm_arch_flush_shadow_memslot(kvm, slot);
1027 1028 1029 1030 1031 1032

		/*
		 * 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.
		 */
1033
		slots = old_memslots;
1034
	}
1035

1036
	r = kvm_arch_prepare_memory_region(kvm, &new, mem, change);
1037
	if (r)
1038
		goto out_slots;
1039

1040
	/* actual memory is freed via old in kvm_free_memslot below */
1041
	if (change == KVM_MR_DELETE) {
1042
		new.dirty_bitmap = NULL;
1043
		memset(&new.arch, 0, sizeof(new.arch));
1044 1045
	}

1046
	update_memslots(slots, &new);
1047
	old_memslots = install_new_memslots(kvm, as_id, slots);
1048

1049
	kvm_arch_commit_memory_region(kvm, mem, &old, &new, change);
1050

1051
	kvm_free_memslot(kvm, &old, &new);
1052
	kvfree(old_memslots);
1053

1054 1055
	/*
	 * IOMMU mapping:  New slots need to be mapped.  Old slots need to be
1056 1057 1058 1059 1060 1061
	 * 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.
1062
	 */
1063 1064
	if ((change == KVM_MR_CREATE) || (change == KVM_MR_MOVE)) {
		r = kvm_iommu_map_pages(kvm, &new);
1065
		return r;
1066 1067
	}

A
Avi Kivity 已提交
1068 1069
	return 0;

1070
out_slots:
1071
	kvfree(slots);
1072
out_free:
1073
	kvm_free_memslot(kvm, &new, &old);
A
Avi Kivity 已提交
1074 1075
out:
	return r;
1076
}
1077 1078 1079
EXPORT_SYMBOL_GPL(__kvm_set_memory_region);

int kvm_set_memory_region(struct kvm *kvm,
1080
			  const struct kvm_userspace_memory_region *mem)
1081 1082 1083
{
	int r;

1084
	mutex_lock(&kvm->slots_lock);
1085
	r = __kvm_set_memory_region(kvm, mem);
1086
	mutex_unlock(&kvm->slots_lock);
1087 1088
	return r;
}
1089 1090
EXPORT_SYMBOL_GPL(kvm_set_memory_region);

1091 1092
static int kvm_vm_ioctl_set_memory_region(struct kvm *kvm,
					  struct kvm_userspace_memory_region *mem)
1093
{
1094
	if ((u16)mem->slot >= KVM_USER_MEM_SLOTS)
1095
		return -EINVAL;
1096

1097
	return kvm_set_memory_region(kvm, mem);
A
Avi Kivity 已提交
1098 1099
}

1100 1101
int kvm_get_dirty_log(struct kvm *kvm,
			struct kvm_dirty_log *log, int *is_dirty)
A
Avi Kivity 已提交
1102
{
1103
	struct kvm_memslots *slots;
A
Avi Kivity 已提交
1104
	struct kvm_memory_slot *memslot;
1105
	int r, i, as_id, id;
1106
	unsigned long n;
A
Avi Kivity 已提交
1107 1108 1109
	unsigned long any = 0;

	r = -EINVAL;
1110 1111 1112
	as_id = log->slot >> 16;
	id = (u16)log->slot;
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_USER_MEM_SLOTS)
A
Avi Kivity 已提交
1113 1114
		goto out;

1115 1116
	slots = __kvm_memslots(kvm, as_id);
	memslot = id_to_memslot(slots, id);
A
Avi Kivity 已提交
1117 1118 1119 1120
	r = -ENOENT;
	if (!memslot->dirty_bitmap)
		goto out;

1121
	n = kvm_dirty_bitmap_bytes(memslot);
A
Avi Kivity 已提交
1122

1123
	for (i = 0; !any && i < n/sizeof(long); ++i)
A
Avi Kivity 已提交
1124 1125 1126 1127 1128 1129
		any = memslot->dirty_bitmap[i];

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

1130 1131
	if (any)
		*is_dirty = 1;
A
Avi Kivity 已提交
1132 1133 1134 1135 1136

	r = 0;
out:
	return r;
}
1137
EXPORT_SYMBOL_GPL(kvm_get_dirty_log);
A
Avi Kivity 已提交
1138

1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164
#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)
{
1165
	struct kvm_memslots *slots;
1166
	struct kvm_memory_slot *memslot;
1167
	int r, i, as_id, id;
1168 1169 1170 1171 1172
	unsigned long n;
	unsigned long *dirty_bitmap;
	unsigned long *dirty_bitmap_buffer;

	r = -EINVAL;
1173 1174 1175
	as_id = log->slot >> 16;
	id = (u16)log->slot;
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_USER_MEM_SLOTS)
1176 1177
		goto out;

1178 1179
	slots = __kvm_memslots(kvm, as_id);
	memslot = id_to_memslot(slots, id);
1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204

	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;

1205 1206 1207 1208 1209
		if (mask) {
			offset = i * BITS_PER_LONG;
			kvm_arch_mmu_enable_log_dirty_pt_masked(kvm, memslot,
								offset, mask);
		}
1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224
	}

	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

1225 1226 1227 1228 1229
bool kvm_largepages_enabled(void)
{
	return largepages_enabled;
}

1230 1231 1232 1233 1234 1235
void kvm_disable_largepages(void)
{
	largepages_enabled = false;
}
EXPORT_SYMBOL_GPL(kvm_disable_largepages);

1236 1237 1238 1239
struct kvm_memory_slot *gfn_to_memslot(struct kvm *kvm, gfn_t gfn)
{
	return __gfn_to_memslot(kvm_memslots(kvm), gfn);
}
A
Avi Kivity 已提交
1240
EXPORT_SYMBOL_GPL(gfn_to_memslot);
A
Avi Kivity 已提交
1241

1242 1243 1244 1245 1246
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);
}

1247
bool kvm_is_visible_gfn(struct kvm *kvm, gfn_t gfn)
1248
{
1249
	struct kvm_memory_slot *memslot = gfn_to_memslot(kvm, gfn);
1250

1251
	if (!memslot || memslot->id >= KVM_USER_MEM_SLOTS ||
1252
	      memslot->flags & KVM_MEMSLOT_INVALID)
1253
		return false;
1254

1255
	return true;
1256 1257 1258
}
EXPORT_SYMBOL_GPL(kvm_is_visible_gfn);

J
Joerg Roedel 已提交
1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282
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;
}

X
Xiao Guangrong 已提交
1283 1284 1285 1286 1287 1288 1289
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)
I
Izik Eidus 已提交
1290
{
1291
	if (!slot || slot->flags & KVM_MEMSLOT_INVALID)
X
Xiao Guangrong 已提交
1292
		return KVM_HVA_ERR_BAD;
1293

X
Xiao Guangrong 已提交
1294 1295
	if (memslot_is_readonly(slot) && write)
		return KVM_HVA_ERR_RO_BAD;
1296 1297 1298 1299

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

X
Xiao Guangrong 已提交
1300
	return __gfn_to_hva_memslot(slot, gfn);
I
Izik Eidus 已提交
1301
}
1302

X
Xiao Guangrong 已提交
1303 1304 1305 1306
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);
I
Izik Eidus 已提交
1307
}
1308

X
Xiao Guangrong 已提交
1309
unsigned long gfn_to_hva_memslot(struct kvm_memory_slot *slot,
1310
					gfn_t gfn)
X
Xiao Guangrong 已提交
1311 1312 1313 1314 1315
{
	return gfn_to_hva_many(slot, gfn, NULL);
}
EXPORT_SYMBOL_GPL(gfn_to_hva_memslot);

1316 1317
unsigned long gfn_to_hva(struct kvm *kvm, gfn_t gfn)
{
1318
	return gfn_to_hva_many(gfn_to_memslot(kvm, gfn), gfn, NULL);
1319
}
1320
EXPORT_SYMBOL_GPL(gfn_to_hva);
I
Izik Eidus 已提交
1321

1322 1323 1324 1325 1326 1327
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);

1328
/*
1329 1330
 * If writable is set to false, the hva returned by this function is only
 * allowed to be read.
1331
 */
1332 1333
unsigned long gfn_to_hva_memslot_prot(struct kvm_memory_slot *slot,
				      gfn_t gfn, bool *writable)
1334
{
1335 1336 1337
	unsigned long hva = __gfn_to_hva_many(slot, gfn, NULL, false);

	if (!kvm_is_error_hva(hva) && writable)
1338 1339
		*writable = !memslot_is_readonly(slot);

1340
	return hva;
1341 1342
}

1343 1344 1345 1346 1347 1348 1349
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);
}

1350 1351 1352 1353 1354 1355 1356
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);
}

1357 1358
static int get_user_page_nowait(unsigned long start, int write,
		struct page **page)
1359
{
L
Lorenzo Stoakes 已提交
1360
	int flags = FOLL_NOWAIT | FOLL_HWPOISON;
1361 1362 1363 1364

	if (write)
		flags |= FOLL_WRITE;

L
Lorenzo Stoakes 已提交
1365
	return get_user_pages(start, 1, flags, page, NULL);
1366 1367
}

1368 1369
static inline int check_user_page_hwpoison(unsigned long addr)
{
L
Lorenzo Stoakes 已提交
1370
	int rc, flags = FOLL_HWPOISON | FOLL_WRITE;
1371

L
Lorenzo Stoakes 已提交
1372
	rc = get_user_pages(addr, 1, flags, NULL, NULL);
1373 1374 1375
	return rc == -EHWPOISON;
}

X
Xiao Guangrong 已提交
1376 1377 1378 1379 1380
/*
 * 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,
D
Dan Williams 已提交
1381
			    bool write_fault, bool *writable, kvm_pfn_t *pfn)
A
Avi Kivity 已提交
1382
{
1383
	struct page *page[1];
X
Xiao Guangrong 已提交
1384
	int npages;
A
Avi Kivity 已提交
1385

X
Xiao Guangrong 已提交
1386 1387
	if (!(async || atomic))
		return false;
1388

1389 1390 1391 1392 1393 1394 1395
	/*
	 * 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;
1396

X
Xiao Guangrong 已提交
1397 1398 1399
	npages = __get_user_pages_fast(addr, 1, 1, page);
	if (npages == 1) {
		*pfn = page_to_pfn(page[0]);
1400

X
Xiao Guangrong 已提交
1401 1402 1403 1404
		if (writable)
			*writable = true;
		return true;
	}
1405

X
Xiao Guangrong 已提交
1406 1407
	return false;
}
1408

X
Xiao Guangrong 已提交
1409 1410 1411 1412 1413
/*
 * 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,
D
Dan Williams 已提交
1414
			   bool *writable, kvm_pfn_t *pfn)
X
Xiao Guangrong 已提交
1415 1416 1417
{
	struct page *page[1];
	int npages = 0;
1418

X
Xiao Guangrong 已提交
1419 1420 1421 1422 1423 1424 1425
	might_sleep();

	if (writable)
		*writable = write_fault;

	if (async) {
		down_read(&current->mm->mmap_sem);
1426
		npages = get_user_page_nowait(addr, write_fault, page);
X
Xiao Guangrong 已提交
1427
		up_read(&current->mm->mmap_sem);
1428
	} else {
1429
		unsigned int flags = FOLL_HWPOISON;
1430 1431 1432 1433

		if (write_fault)
			flags |= FOLL_WRITE;

1434
		npages = get_user_pages_unlocked(addr, 1, page, flags);
1435
	}
X
Xiao Guangrong 已提交
1436 1437 1438 1439
	if (npages != 1)
		return npages;

	/* map read fault as writable if possible */
1440
	if (unlikely(!write_fault) && writable) {
X
Xiao Guangrong 已提交
1441 1442 1443 1444 1445 1446 1447
		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];
1448
		}
X
Xiao Guangrong 已提交
1449 1450

		npages = 1;
1451
	}
X
Xiao Guangrong 已提交
1452 1453 1454
	*pfn = page_to_pfn(page[0]);
	return npages;
}
I
Izik Eidus 已提交
1455

X
Xiao Guangrong 已提交
1456 1457 1458 1459
static bool vma_is_valid(struct vm_area_struct *vma, bool write_fault)
{
	if (unlikely(!(vma->vm_flags & VM_READ)))
		return false;
1460

X
Xiao Guangrong 已提交
1461 1462
	if (write_fault && (unlikely(!(vma->vm_flags & VM_WRITE))))
		return false;
1463

X
Xiao Guangrong 已提交
1464 1465
	return true;
}
1466

1467 1468 1469 1470
static int hva_to_pfn_remapped(struct vm_area_struct *vma,
			       unsigned long addr, bool *async,
			       bool write_fault, kvm_pfn_t *p_pfn)
{
1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509
	unsigned long pfn;
	int r;

	r = follow_pfn(vma, addr, &pfn);
	if (r) {
		/*
		 * get_user_pages fails for VM_IO and VM_PFNMAP vmas and does
		 * not call the fault handler, so do it here.
		 */
		bool unlocked = false;
		r = fixup_user_fault(current, current->mm, addr,
				     (write_fault ? FAULT_FLAG_WRITE : 0),
				     &unlocked);
		if (unlocked)
			return -EAGAIN;
		if (r)
			return r;

		r = follow_pfn(vma, addr, &pfn);
		if (r)
			return r;

	}


	/*
	 * Get a reference here because callers of *hva_to_pfn* and
	 * *gfn_to_pfn* ultimately call kvm_release_pfn_clean on the
	 * returned pfn.  This is only needed if the VMA has VM_MIXEDMAP
	 * set, but the kvm_get_pfn/kvm_release_pfn_clean pair will
	 * simply do nothing for reserved pfns.
	 *
	 * Whoever called remap_pfn_range is also going to call e.g.
	 * unmap_mapping_range before the underlying pages are freed,
	 * causing a call to our MMU notifier.
	 */ 
	kvm_get_pfn(pfn);

	*p_pfn = pfn;
1510 1511 1512
	return 0;
}

1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526
/*
 * 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.
 */
D
Dan Williams 已提交
1527
static kvm_pfn_t hva_to_pfn(unsigned long addr, bool atomic, bool *async,
X
Xiao Guangrong 已提交
1528 1529 1530
			bool write_fault, bool *writable)
{
	struct vm_area_struct *vma;
D
Dan Williams 已提交
1531
	kvm_pfn_t pfn = 0;
1532
	int npages, r;
1533

X
Xiao Guangrong 已提交
1534 1535
	/* we can do it either atomically or asynchronously, not both */
	BUG_ON(atomic && async);
1536

X
Xiao Guangrong 已提交
1537 1538 1539 1540 1541 1542 1543 1544 1545
	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;
1546

X
Xiao Guangrong 已提交
1547 1548 1549 1550 1551 1552 1553
	down_read(&current->mm->mmap_sem);
	if (npages == -EHWPOISON ||
	      (!async && check_user_page_hwpoison(addr))) {
		pfn = KVM_PFN_ERR_HWPOISON;
		goto exit;
	}

1554
retry:
X
Xiao Guangrong 已提交
1555 1556 1557 1558
	vma = find_vma_intersection(current->mm, addr, addr + 1);

	if (vma == NULL)
		pfn = KVM_PFN_ERR_FAULT;
1559 1560
	else if (vma->vm_flags & (VM_IO | VM_PFNMAP)) {
		r = hva_to_pfn_remapped(vma, addr, async, write_fault, &pfn);
1561 1562
		if (r == -EAGAIN)
			goto retry;
1563 1564
		if (r < 0)
			pfn = KVM_PFN_ERR_FAULT;
X
Xiao Guangrong 已提交
1565
	} else {
X
Xiao Guangrong 已提交
1566
		if (async && vma_is_valid(vma, write_fault))
X
Xiao Guangrong 已提交
1567 1568 1569 1570 1571
			*async = true;
		pfn = KVM_PFN_ERR_FAULT;
	}
exit:
	up_read(&current->mm->mmap_sem);
1572
	return pfn;
1573 1574
}

D
Dan Williams 已提交
1575 1576 1577
kvm_pfn_t __gfn_to_pfn_memslot(struct kvm_memory_slot *slot, gfn_t gfn,
			       bool atomic, bool *async, bool write_fault,
			       bool *writable)
1578
{
X
Xiao Guangrong 已提交
1579 1580
	unsigned long addr = __gfn_to_hva_many(slot, gfn, NULL, write_fault);

1581 1582 1583
	if (addr == KVM_HVA_ERR_RO_BAD) {
		if (writable)
			*writable = false;
X
Xiao Guangrong 已提交
1584
		return KVM_PFN_ERR_RO_FAULT;
1585
	}
X
Xiao Guangrong 已提交
1586

1587 1588 1589
	if (kvm_is_error_hva(addr)) {
		if (writable)
			*writable = false;
1590
		return KVM_PFN_NOSLOT;
1591
	}
X
Xiao Guangrong 已提交
1592 1593 1594 1595 1596 1597 1598 1599 1600

	/* 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);
1601
}
1602
EXPORT_SYMBOL_GPL(__gfn_to_pfn_memslot);
1603

D
Dan Williams 已提交
1604
kvm_pfn_t gfn_to_pfn_prot(struct kvm *kvm, gfn_t gfn, bool write_fault,
1605 1606
		      bool *writable)
{
P
Paolo Bonzini 已提交
1607 1608
	return __gfn_to_pfn_memslot(gfn_to_memslot(kvm, gfn), gfn, false, NULL,
				    write_fault, writable);
1609 1610 1611
}
EXPORT_SYMBOL_GPL(gfn_to_pfn_prot);

D
Dan Williams 已提交
1612
kvm_pfn_t gfn_to_pfn_memslot(struct kvm_memory_slot *slot, gfn_t gfn)
1613
{
X
Xiao Guangrong 已提交
1614
	return __gfn_to_pfn_memslot(slot, gfn, false, NULL, true, NULL);
1615
}
P
Paolo Bonzini 已提交
1616
EXPORT_SYMBOL_GPL(gfn_to_pfn_memslot);
1617

D
Dan Williams 已提交
1618
kvm_pfn_t gfn_to_pfn_memslot_atomic(struct kvm_memory_slot *slot, gfn_t gfn)
1619
{
X
Xiao Guangrong 已提交
1620
	return __gfn_to_pfn_memslot(slot, gfn, true, NULL, true, NULL);
1621
}
1622
EXPORT_SYMBOL_GPL(gfn_to_pfn_memslot_atomic);
1623

D
Dan Williams 已提交
1624
kvm_pfn_t gfn_to_pfn_atomic(struct kvm *kvm, gfn_t gfn)
P
Paolo Bonzini 已提交
1625 1626 1627 1628 1629
{
	return gfn_to_pfn_memslot_atomic(gfn_to_memslot(kvm, gfn), gfn);
}
EXPORT_SYMBOL_GPL(gfn_to_pfn_atomic);

D
Dan Williams 已提交
1630
kvm_pfn_t kvm_vcpu_gfn_to_pfn_atomic(struct kvm_vcpu *vcpu, gfn_t gfn)
1631 1632 1633 1634 1635
{
	return gfn_to_pfn_memslot_atomic(kvm_vcpu_gfn_to_memslot(vcpu, gfn), gfn);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_gfn_to_pfn_atomic);

D
Dan Williams 已提交
1636
kvm_pfn_t gfn_to_pfn(struct kvm *kvm, gfn_t gfn)
P
Paolo Bonzini 已提交
1637 1638 1639 1640 1641
{
	return gfn_to_pfn_memslot(gfn_to_memslot(kvm, gfn), gfn);
}
EXPORT_SYMBOL_GPL(gfn_to_pfn);

D
Dan Williams 已提交
1642
kvm_pfn_t kvm_vcpu_gfn_to_pfn(struct kvm_vcpu *vcpu, gfn_t gfn)
1643 1644 1645 1646 1647
{
	return gfn_to_pfn_memslot(kvm_vcpu_gfn_to_memslot(vcpu, gfn), gfn);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_gfn_to_pfn);

1648 1649
int gfn_to_page_many_atomic(struct kvm_memory_slot *slot, gfn_t gfn,
			    struct page **pages, int nr_pages)
1650 1651 1652 1653
{
	unsigned long addr;
	gfn_t entry;

1654
	addr = gfn_to_hva_many(slot, gfn, &entry);
1655 1656 1657 1658 1659 1660 1661 1662 1663 1664
	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);

D
Dan Williams 已提交
1665
static struct page *kvm_pfn_to_page(kvm_pfn_t pfn)
X
Xiao Guangrong 已提交
1666
{
1667
	if (is_error_noslot_pfn(pfn))
1668
		return KVM_ERR_PTR_BAD_PAGE;
X
Xiao Guangrong 已提交
1669

1670
	if (kvm_is_reserved_pfn(pfn)) {
1671
		WARN_ON(1);
1672
		return KVM_ERR_PTR_BAD_PAGE;
1673
	}
X
Xiao Guangrong 已提交
1674 1675 1676 1677

	return pfn_to_page(pfn);
}

1678 1679
struct page *gfn_to_page(struct kvm *kvm, gfn_t gfn)
{
D
Dan Williams 已提交
1680
	kvm_pfn_t pfn;
1681 1682 1683

	pfn = gfn_to_pfn(kvm, gfn);

X
Xiao Guangrong 已提交
1684
	return kvm_pfn_to_page(pfn);
A
Avi Kivity 已提交
1685 1686 1687
}
EXPORT_SYMBOL_GPL(gfn_to_page);

1688 1689
struct page *kvm_vcpu_gfn_to_page(struct kvm_vcpu *vcpu, gfn_t gfn)
{
D
Dan Williams 已提交
1690
	kvm_pfn_t pfn;
1691 1692 1693 1694 1695 1696 1697

	pfn = kvm_vcpu_gfn_to_pfn(vcpu, gfn);

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

1698 1699
void kvm_release_page_clean(struct page *page)
{
1700 1701
	WARN_ON(is_error_page(page));

1702
	kvm_release_pfn_clean(page_to_pfn(page));
1703 1704 1705
}
EXPORT_SYMBOL_GPL(kvm_release_page_clean);

D
Dan Williams 已提交
1706
void kvm_release_pfn_clean(kvm_pfn_t pfn)
1707
{
1708
	if (!is_error_noslot_pfn(pfn) && !kvm_is_reserved_pfn(pfn))
1709
		put_page(pfn_to_page(pfn));
1710 1711 1712
}
EXPORT_SYMBOL_GPL(kvm_release_pfn_clean);

1713
void kvm_release_page_dirty(struct page *page)
1714
{
X
Xiao Guangrong 已提交
1715 1716
	WARN_ON(is_error_page(page));

1717 1718 1719 1720
	kvm_release_pfn_dirty(page_to_pfn(page));
}
EXPORT_SYMBOL_GPL(kvm_release_page_dirty);

D
Dan Williams 已提交
1721
static void kvm_release_pfn_dirty(kvm_pfn_t pfn)
1722 1723 1724 1725 1726
{
	kvm_set_pfn_dirty(pfn);
	kvm_release_pfn_clean(pfn);
}

D
Dan Williams 已提交
1727
void kvm_set_pfn_dirty(kvm_pfn_t pfn)
1728
{
1729
	if (!kvm_is_reserved_pfn(pfn)) {
1730
		struct page *page = pfn_to_page(pfn);
1731

1732 1733 1734
		if (!PageReserved(page))
			SetPageDirty(page);
	}
1735
}
1736 1737
EXPORT_SYMBOL_GPL(kvm_set_pfn_dirty);

D
Dan Williams 已提交
1738
void kvm_set_pfn_accessed(kvm_pfn_t pfn)
1739
{
1740
	if (!kvm_is_reserved_pfn(pfn))
1741
		mark_page_accessed(pfn_to_page(pfn));
1742 1743 1744
}
EXPORT_SYMBOL_GPL(kvm_set_pfn_accessed);

D
Dan Williams 已提交
1745
void kvm_get_pfn(kvm_pfn_t pfn)
1746
{
1747
	if (!kvm_is_reserved_pfn(pfn))
1748
		get_page(pfn_to_page(pfn));
1749 1750
}
EXPORT_SYMBOL_GPL(kvm_get_pfn);
1751

1752 1753 1754 1755 1756 1757 1758 1759
static int next_segment(unsigned long len, int offset)
{
	if (len > PAGE_SIZE - offset)
		return PAGE_SIZE - offset;
	else
		return len;
}

1760 1761
static int __kvm_read_guest_page(struct kvm_memory_slot *slot, gfn_t gfn,
				 void *data, int offset, int len)
1762
{
1763 1764
	int r;
	unsigned long addr;
1765

1766
	addr = gfn_to_hva_memslot_prot(slot, gfn, NULL);
1767 1768
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1769
	r = __copy_from_user(data, (void __user *)addr + offset, len);
1770
	if (r)
1771 1772 1773
		return -EFAULT;
	return 0;
}
1774 1775 1776 1777 1778 1779 1780 1781

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);
}
1782 1783
EXPORT_SYMBOL_GPL(kvm_read_guest_page);

1784 1785 1786 1787 1788 1789 1790 1791 1792
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);

1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812
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);

1813
int kvm_vcpu_read_guest(struct kvm_vcpu *vcpu, gpa_t gpa, void *data, unsigned long len)
1814 1815
{
	gfn_t gfn = gpa >> PAGE_SHIFT;
1816
	int seg;
1817
	int offset = offset_in_page(gpa);
1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831
	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);
1832

1833 1834 1835 1836 1837 1838 1839
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);
1840 1841
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1842
	pagefault_disable();
1843
	r = __copy_from_user_inatomic(data, (void __user *)addr + offset, len);
1844
	pagefault_enable();
1845 1846 1847 1848 1849
	if (r)
		return -EFAULT;
	return 0;
}

1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873
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)
1874
{
1875 1876
	int r;
	unsigned long addr;
1877

1878
	addr = gfn_to_hva_memslot(memslot, gfn);
1879 1880
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1881
	r = __copy_to_user((void __user *)addr + offset, data, len);
1882
	if (r)
1883
		return -EFAULT;
1884
	mark_page_dirty_in_slot(memslot, gfn);
1885 1886
	return 0;
}
1887 1888 1889 1890 1891 1892 1893 1894

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);
}
1895 1896
EXPORT_SYMBOL_GPL(kvm_write_guest_page);

1897 1898 1899 1900 1901 1902 1903 1904 1905
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);

1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924
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;
}
1925
EXPORT_SYMBOL_GPL(kvm_write_guest);
1926

1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947
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);

1948 1949 1950
static int __kvm_gfn_to_hva_cache_init(struct kvm_memslots *slots,
				       struct gfn_to_hva_cache *ghc,
				       gpa_t gpa, unsigned long len)
1951 1952
{
	int offset = offset_in_page(gpa);
1953 1954 1955 1956
	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;
1957 1958 1959

	ghc->gpa = gpa;
	ghc->generation = slots->generation;
1960
	ghc->len = len;
1961
	ghc->memslot = __gfn_to_memslot(slots, start_gfn);
1962 1963
	ghc->hva = gfn_to_hva_many(ghc->memslot, start_gfn, NULL);
	if (!kvm_is_error_hva(ghc->hva) && nr_pages_needed <= 1) {
1964
		ghc->hva += offset;
1965 1966 1967 1968 1969 1970
	} else {
		/*
		 * If the requested region crosses two memslots, we still
		 * verify that the entire region is valid here.
		 */
		while (start_gfn <= end_gfn) {
1971
			ghc->memslot = __gfn_to_memslot(slots, start_gfn);
1972 1973 1974 1975 1976 1977 1978 1979 1980
			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;
	}
1981 1982
	return 0;
}
1983 1984 1985 1986 1987 1988 1989

int kvm_gfn_to_hva_cache_init(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
			      gpa_t gpa, unsigned long len)
{
	struct kvm_memslots *slots = kvm_memslots(kvm);
	return __kvm_gfn_to_hva_cache_init(slots, ghc, gpa, len);
}
1990 1991
EXPORT_SYMBOL_GPL(kvm_gfn_to_hva_cache_init);

1992 1993
int kvm_write_guest_offset_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
			   void *data, int offset, unsigned long len)
1994 1995 1996
{
	struct kvm_memslots *slots = kvm_memslots(kvm);
	int r;
1997
	gpa_t gpa = ghc->gpa + offset;
1998

1999
	BUG_ON(len + offset > ghc->len);
2000

2001
	if (slots->generation != ghc->generation)
2002
		__kvm_gfn_to_hva_cache_init(slots, ghc, ghc->gpa, ghc->len);
2003 2004

	if (unlikely(!ghc->memslot))
2005
		return kvm_write_guest(kvm, gpa, data, len);
2006 2007 2008 2009

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

2010
	r = __copy_to_user((void __user *)ghc->hva + offset, data, len);
2011 2012
	if (r)
		return -EFAULT;
2013
	mark_page_dirty_in_slot(ghc->memslot, gpa >> PAGE_SHIFT);
2014 2015 2016

	return 0;
}
2017 2018 2019 2020 2021 2022 2023
EXPORT_SYMBOL_GPL(kvm_write_guest_offset_cached);

int kvm_write_guest_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
			   void *data, unsigned long len)
{
	return kvm_write_guest_offset_cached(kvm, ghc, data, 0, len);
}
2024 2025
EXPORT_SYMBOL_GPL(kvm_write_guest_cached);

2026 2027 2028 2029 2030 2031
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;

2032 2033
	BUG_ON(len > ghc->len);

2034
	if (slots->generation != ghc->generation)
2035
		__kvm_gfn_to_hva_cache_init(slots, ghc, ghc->gpa, ghc->len);
2036 2037 2038

	if (unlikely(!ghc->memslot))
		return kvm_read_guest(kvm, ghc->gpa, data, len);
2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050

	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);

2051 2052
int kvm_clear_guest_page(struct kvm *kvm, gfn_t gfn, int offset, int len)
{
2053 2054 2055
	const void *zero_page = (const void *) __va(page_to_phys(ZERO_PAGE(0)));

	return kvm_write_guest_page(kvm, gfn, zero_page, offset, len);
2056 2057 2058 2059 2060 2061 2062 2063 2064 2065
}
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;

2066
	while ((seg = next_segment(len, offset)) != 0) {
2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077
		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);

2078
static void mark_page_dirty_in_slot(struct kvm_memory_slot *memslot,
2079
				    gfn_t gfn)
A
Avi Kivity 已提交
2080
{
R
Rusty Russell 已提交
2081 2082
	if (memslot && memslot->dirty_bitmap) {
		unsigned long rel_gfn = gfn - memslot->base_gfn;
A
Avi Kivity 已提交
2083

2084
		set_bit_le(rel_gfn, memslot->dirty_bitmap);
A
Avi Kivity 已提交
2085 2086 2087
	}
}

2088 2089 2090 2091 2092
void mark_page_dirty(struct kvm *kvm, gfn_t gfn)
{
	struct kvm_memory_slot *memslot;

	memslot = gfn_to_memslot(kvm, gfn);
2093
	mark_page_dirty_in_slot(memslot, gfn);
2094
}
2095
EXPORT_SYMBOL_GPL(mark_page_dirty);
2096

2097 2098 2099 2100 2101 2102 2103 2104 2105
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);

W
Wanpeng Li 已提交
2106 2107
static void grow_halt_poll_ns(struct kvm_vcpu *vcpu)
{
2108
	unsigned int old, val, grow;
W
Wanpeng Li 已提交
2109

2110
	old = val = vcpu->halt_poll_ns;
2111
	grow = READ_ONCE(halt_poll_ns_grow);
W
Wanpeng Li 已提交
2112
	/* 10us base */
2113
	if (val == 0 && grow)
W
Wanpeng Li 已提交
2114 2115
		val = 10000;
	else
2116
		val *= grow;
W
Wanpeng Li 已提交
2117

2118 2119 2120
	if (val > halt_poll_ns)
		val = halt_poll_ns;

W
Wanpeng Li 已提交
2121
	vcpu->halt_poll_ns = val;
2122
	trace_kvm_halt_poll_ns_grow(vcpu->vcpu_id, val, old);
W
Wanpeng Li 已提交
2123 2124 2125 2126
}

static void shrink_halt_poll_ns(struct kvm_vcpu *vcpu)
{
2127
	unsigned int old, val, shrink;
W
Wanpeng Li 已提交
2128

2129
	old = val = vcpu->halt_poll_ns;
2130 2131
	shrink = READ_ONCE(halt_poll_ns_shrink);
	if (shrink == 0)
W
Wanpeng Li 已提交
2132 2133
		val = 0;
	else
2134
		val /= shrink;
W
Wanpeng Li 已提交
2135 2136

	vcpu->halt_poll_ns = val;
2137
	trace_kvm_halt_poll_ns_shrink(vcpu->vcpu_id, val, old);
W
Wanpeng Li 已提交
2138 2139
}

2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153
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;
}

E
Eddie Dong 已提交
2154 2155 2156
/*
 * The vCPU has executed a HLT instruction with in-kernel mode enabled.
 */
2157
void kvm_vcpu_block(struct kvm_vcpu *vcpu)
2158
{
2159
	ktime_t start, cur;
2160
	DECLARE_SWAITQUEUE(wait);
2161
	bool waited = false;
W
Wanpeng Li 已提交
2162
	u64 block_ns;
2163 2164

	start = cur = ktime_get();
W
Wanpeng Li 已提交
2165 2166
	if (vcpu->halt_poll_ns) {
		ktime_t stop = ktime_add_ns(ktime_get(), vcpu->halt_poll_ns);
2167

2168
		++vcpu->stat.halt_attempted_poll;
2169 2170 2171 2172 2173 2174 2175
		do {
			/*
			 * This sets KVM_REQ_UNHALT if an interrupt
			 * arrives.
			 */
			if (kvm_vcpu_check_block(vcpu) < 0) {
				++vcpu->stat.halt_successful_poll;
2176 2177
				if (!vcpu_valid_wakeup(vcpu))
					++vcpu->stat.halt_poll_invalid;
2178 2179 2180 2181 2182
				goto out;
			}
			cur = ktime_get();
		} while (single_task_running() && ktime_before(cur, stop));
	}
2183

2184 2185
	kvm_arch_vcpu_blocking(vcpu);

2186
	for (;;) {
2187
		prepare_to_swait(&vcpu->wq, &wait, TASK_INTERRUPTIBLE);
2188

2189
		if (kvm_vcpu_check_block(vcpu) < 0)
2190 2191
			break;

2192
		waited = true;
E
Eddie Dong 已提交
2193 2194
		schedule();
	}
2195

2196
	finish_swait(&vcpu->wq, &wait);
2197 2198
	cur = ktime_get();

2199
	kvm_arch_vcpu_unblocking(vcpu);
2200
out:
W
Wanpeng Li 已提交
2201 2202
	block_ns = ktime_to_ns(cur) - ktime_to_ns(start);

2203 2204 2205
	if (!vcpu_valid_wakeup(vcpu))
		shrink_halt_poll_ns(vcpu);
	else if (halt_poll_ns) {
W
Wanpeng Li 已提交
2206 2207 2208
		if (block_ns <= vcpu->halt_poll_ns)
			;
		/* we had a long block, shrink polling */
2209
		else if (vcpu->halt_poll_ns && block_ns > halt_poll_ns)
W
Wanpeng Li 已提交
2210 2211 2212 2213 2214
			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);
2215 2216
	} else
		vcpu->halt_poll_ns = 0;
W
Wanpeng Li 已提交
2217

2218 2219
	trace_kvm_vcpu_wakeup(block_ns, waited, vcpu_valid_wakeup(vcpu));
	kvm_arch_vcpu_block_finish(vcpu);
E
Eddie Dong 已提交
2220
}
2221
EXPORT_SYMBOL_GPL(kvm_vcpu_block);
E
Eddie Dong 已提交
2222

2223
#ifndef CONFIG_S390
2224
void kvm_vcpu_wake_up(struct kvm_vcpu *vcpu)
2225
{
2226
	struct swait_queue_head *wqp;
2227 2228

	wqp = kvm_arch_vcpu_wq(vcpu);
2229 2230
	if (swait_active(wqp)) {
		swake_up(wqp);
2231 2232 2233
		++vcpu->stat.halt_wakeup;
	}

2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245
}
EXPORT_SYMBOL_GPL(kvm_vcpu_wake_up);

/*
 * 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;

	kvm_vcpu_wake_up(vcpu);
2246 2247 2248 2249 2250 2251
	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();
}
2252
EXPORT_SYMBOL_GPL(kvm_vcpu_kick);
2253
#endif /* !CONFIG_S390 */
2254

2255
int kvm_vcpu_yield_to(struct kvm_vcpu *target)
2256 2257 2258
{
	struct pid *pid;
	struct task_struct *task = NULL;
2259
	int ret = 0;
2260 2261 2262 2263

	rcu_read_lock();
	pid = rcu_dereference(target->pid);
	if (pid)
2264
		task = get_pid_task(pid, PIDTYPE_PID);
2265 2266
	rcu_read_unlock();
	if (!task)
2267 2268
		return ret;
	ret = yield_to(task, 1);
2269
	put_task_struct(task);
2270 2271

	return ret;
2272 2273 2274
}
EXPORT_SYMBOL_GPL(kvm_vcpu_yield_to);

2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296
/*
 * 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.
 */
2297
static bool kvm_vcpu_eligible_for_directed_yield(struct kvm_vcpu *vcpu)
2298
{
2299
#ifdef CONFIG_HAVE_KVM_CPU_RELAX_INTERCEPT
2300 2301 2302
	bool eligible;

	eligible = !vcpu->spin_loop.in_spin_loop ||
2303
		    vcpu->spin_loop.dy_eligible;
2304 2305 2306 2307 2308

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

	return eligible;
2309 2310
#else
	return true;
2311
#endif
2312
}
2313

2314
void kvm_vcpu_on_spin(struct kvm_vcpu *me)
Z
Zhai, Edwin 已提交
2315
{
2316 2317 2318 2319
	struct kvm *kvm = me->kvm;
	struct kvm_vcpu *vcpu;
	int last_boosted_vcpu = me->kvm->last_boosted_vcpu;
	int yielded = 0;
2320
	int try = 3;
2321 2322
	int pass;
	int i;
Z
Zhai, Edwin 已提交
2323

2324
	kvm_vcpu_set_in_spin_loop(me, true);
2325 2326 2327 2328 2329 2330 2331
	/*
	 * 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.
	 */
2332
	for (pass = 0; pass < 2 && !yielded && try; pass++) {
2333
		kvm_for_each_vcpu(i, vcpu, kvm) {
2334
			if (!pass && i <= last_boosted_vcpu) {
2335 2336 2337 2338
				i = last_boosted_vcpu;
				continue;
			} else if (pass && i > last_boosted_vcpu)
				break;
2339 2340
			if (!ACCESS_ONCE(vcpu->preempted))
				continue;
2341 2342
			if (vcpu == me)
				continue;
2343
			if (swait_active(&vcpu->wq) && !kvm_arch_vcpu_runnable(vcpu))
2344
				continue;
2345 2346
			if (!kvm_vcpu_eligible_for_directed_yield(vcpu))
				continue;
2347 2348 2349

			yielded = kvm_vcpu_yield_to(vcpu);
			if (yielded > 0) {
2350 2351
				kvm->last_boosted_vcpu = i;
				break;
2352 2353 2354 2355
			} else if (yielded < 0) {
				try--;
				if (!try)
					break;
2356 2357 2358
			}
		}
	}
2359
	kvm_vcpu_set_in_spin_loop(me, false);
2360 2361 2362

	/* Ensure vcpu is not eligible during next spinloop */
	kvm_vcpu_set_dy_eligible(me, false);
Z
Zhai, Edwin 已提交
2363 2364 2365
}
EXPORT_SYMBOL_GPL(kvm_vcpu_on_spin);

2366
static int kvm_vcpu_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
2367 2368 2369 2370
{
	struct kvm_vcpu *vcpu = vma->vm_file->private_data;
	struct page *page;

2371
	if (vmf->pgoff == 0)
2372
		page = virt_to_page(vcpu->run);
A
Avi Kivity 已提交
2373
#ifdef CONFIG_X86
2374
	else if (vmf->pgoff == KVM_PIO_PAGE_OFFSET)
2375
		page = virt_to_page(vcpu->arch.pio_data);
2376 2377 2378 2379
#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);
A
Avi Kivity 已提交
2380
#endif
2381
	else
2382
		return kvm_arch_vcpu_fault(vcpu, vmf);
2383
	get_page(page);
2384 2385
	vmf->page = page;
	return 0;
2386 2387
}

2388
static const struct vm_operations_struct kvm_vcpu_vm_ops = {
2389
	.fault = kvm_vcpu_fault,
2390 2391 2392 2393 2394 2395 2396 2397
};

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

A
Avi Kivity 已提交
2398 2399 2400 2401
static int kvm_vcpu_release(struct inode *inode, struct file *filp)
{
	struct kvm_vcpu *vcpu = filp->private_data;

2402
	debugfs_remove_recursive(vcpu->debugfs_dentry);
A
Al Viro 已提交
2403
	kvm_put_kvm(vcpu->kvm);
A
Avi Kivity 已提交
2404 2405 2406
	return 0;
}

2407
static struct file_operations kvm_vcpu_fops = {
A
Avi Kivity 已提交
2408 2409
	.release        = kvm_vcpu_release,
	.unlocked_ioctl = kvm_vcpu_ioctl,
2410
#ifdef CONFIG_KVM_COMPAT
2411 2412
	.compat_ioctl   = kvm_vcpu_compat_ioctl,
#endif
2413
	.mmap           = kvm_vcpu_mmap,
2414
	.llseek		= noop_llseek,
A
Avi Kivity 已提交
2415 2416 2417 2418 2419 2420 2421
};

/*
 * Allocates an inode for the vcpu.
 */
static int create_vcpu_fd(struct kvm_vcpu *vcpu)
{
2422
	return anon_inode_getfd("kvm-vcpu", &kvm_vcpu_fops, vcpu, O_RDWR | O_CLOEXEC);
A
Avi Kivity 已提交
2423 2424
}

2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450
static int kvm_create_vcpu_debugfs(struct kvm_vcpu *vcpu)
{
	char dir_name[ITOA_MAX_LEN * 2];
	int ret;

	if (!kvm_arch_has_vcpu_debugfs())
		return 0;

	if (!debugfs_initialized())
		return 0;

	snprintf(dir_name, sizeof(dir_name), "vcpu%d", vcpu->vcpu_id);
	vcpu->debugfs_dentry = debugfs_create_dir(dir_name,
								vcpu->kvm->debugfs_dentry);
	if (!vcpu->debugfs_dentry)
		return -ENOMEM;

	ret = kvm_arch_create_vcpu_debugfs(vcpu);
	if (ret < 0) {
		debugfs_remove_recursive(vcpu->debugfs_dentry);
		return ret;
	}

	return 0;
}

2451 2452 2453
/*
 * Creates some virtual cpus.  Good luck creating more than one.
 */
2454
static int kvm_vm_ioctl_create_vcpu(struct kvm *kvm, u32 id)
2455 2456
{
	int r;
2457
	struct kvm_vcpu *vcpu;
2458

G
Greg Kurz 已提交
2459
	if (id >= KVM_MAX_VCPU_ID)
2460 2461
		return -EINVAL;

2462 2463 2464 2465 2466 2467 2468 2469 2470
	mutex_lock(&kvm->lock);
	if (kvm->created_vcpus == KVM_MAX_VCPUS) {
		mutex_unlock(&kvm->lock);
		return -EINVAL;
	}

	kvm->created_vcpus++;
	mutex_unlock(&kvm->lock);

2471
	vcpu = kvm_arch_vcpu_create(kvm, id);
2472 2473 2474 2475
	if (IS_ERR(vcpu)) {
		r = PTR_ERR(vcpu);
		goto vcpu_decrement;
	}
2476

2477 2478
	preempt_notifier_init(&vcpu->preempt_notifier, &kvm_preempt_ops);

2479 2480
	r = kvm_arch_vcpu_setup(vcpu);
	if (r)
2481
		goto vcpu_destroy;
2482

2483 2484 2485 2486
	r = kvm_create_vcpu_debugfs(vcpu);
	if (r)
		goto vcpu_destroy;

S
Shaohua Li 已提交
2487
	mutex_lock(&kvm->lock);
2488 2489 2490 2491
	if (kvm_get_vcpu_by_id(kvm, id)) {
		r = -EEXIST;
		goto unlock_vcpu_destroy;
	}
2492 2493

	BUG_ON(kvm->vcpus[atomic_read(&kvm->online_vcpus)]);
2494

R
Rusty Russell 已提交
2495
	/* Now it's all set up, let userspace reach it */
A
Al Viro 已提交
2496
	kvm_get_kvm(kvm);
A
Avi Kivity 已提交
2497
	r = create_vcpu_fd(vcpu);
2498 2499
	if (r < 0) {
		kvm_put_kvm(kvm);
2500
		goto unlock_vcpu_destroy;
2501 2502 2503
	}

	kvm->vcpus[atomic_read(&kvm->online_vcpus)] = vcpu;
2504 2505 2506 2507 2508

	/*
	 * Pairs with smp_rmb() in kvm_get_vcpu.  Write kvm->vcpus
	 * before kvm->online_vcpu's incremented value.
	 */
2509 2510 2511 2512
	smp_wmb();
	atomic_inc(&kvm->online_vcpus);

	mutex_unlock(&kvm->lock);
2513
	kvm_arch_vcpu_postcreate(vcpu);
R
Rusty Russell 已提交
2514
	return r;
2515

2516
unlock_vcpu_destroy:
2517
	mutex_unlock(&kvm->lock);
2518
	debugfs_remove_recursive(vcpu->debugfs_dentry);
2519
vcpu_destroy:
2520
	kvm_arch_vcpu_destroy(vcpu);
2521 2522 2523 2524
vcpu_decrement:
	mutex_lock(&kvm->lock);
	kvm->created_vcpus--;
	mutex_unlock(&kvm->lock);
2525 2526 2527
	return r;
}

A
Avi Kivity 已提交
2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538
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;
}

A
Avi Kivity 已提交
2539 2540
static long kvm_vcpu_ioctl(struct file *filp,
			   unsigned int ioctl, unsigned long arg)
A
Avi Kivity 已提交
2541
{
A
Avi Kivity 已提交
2542
	struct kvm_vcpu *vcpu = filp->private_data;
A
Al Viro 已提交
2543
	void __user *argp = (void __user *)arg;
2544
	int r;
2545 2546
	struct kvm_fpu *fpu = NULL;
	struct kvm_sregs *kvm_sregs = NULL;
A
Avi Kivity 已提交
2547

2548 2549
	if (vcpu->kvm->mm != current->mm)
		return -EIO;
2550

2551 2552 2553
	if (unlikely(_IOC_TYPE(ioctl) != KVMIO))
		return -EINVAL;

2554
#if defined(CONFIG_S390) || defined(CONFIG_PPC) || defined(CONFIG_MIPS)
2555 2556 2557 2558
	/*
	 * Special cases: vcpu ioctls that are asynchronous to vcpu execution,
	 * so vcpu_load() would break it.
	 */
2559
	if (ioctl == KVM_S390_INTERRUPT || ioctl == KVM_S390_IRQ || ioctl == KVM_INTERRUPT)
2560 2561 2562 2563
		return kvm_arch_vcpu_ioctl(filp, ioctl, arg);
#endif


2564 2565 2566
	r = vcpu_load(vcpu);
	if (r)
		return r;
A
Avi Kivity 已提交
2567
	switch (ioctl) {
2568
	case KVM_RUN:
2569 2570 2571
		r = -EINVAL;
		if (arg)
			goto out;
2572 2573 2574 2575
		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);
2576

2577 2578 2579 2580 2581
			rcu_assign_pointer(vcpu->pid, newpid);
			if (oldpid)
				synchronize_rcu();
			put_pid(oldpid);
		}
2582
		r = kvm_arch_vcpu_ioctl_run(vcpu, vcpu->run);
2583
		trace_kvm_userspace_exit(vcpu->run->exit_reason, r);
A
Avi Kivity 已提交
2584 2585
		break;
	case KVM_GET_REGS: {
2586
		struct kvm_regs *kvm_regs;
A
Avi Kivity 已提交
2587

2588 2589 2590
		r = -ENOMEM;
		kvm_regs = kzalloc(sizeof(struct kvm_regs), GFP_KERNEL);
		if (!kvm_regs)
A
Avi Kivity 已提交
2591
			goto out;
2592 2593 2594
		r = kvm_arch_vcpu_ioctl_get_regs(vcpu, kvm_regs);
		if (r)
			goto out_free1;
A
Avi Kivity 已提交
2595
		r = -EFAULT;
2596 2597
		if (copy_to_user(argp, kvm_regs, sizeof(struct kvm_regs)))
			goto out_free1;
A
Avi Kivity 已提交
2598
		r = 0;
2599 2600
out_free1:
		kfree(kvm_regs);
A
Avi Kivity 已提交
2601 2602 2603
		break;
	}
	case KVM_SET_REGS: {
2604
		struct kvm_regs *kvm_regs;
A
Avi Kivity 已提交
2605

2606
		r = -ENOMEM;
2607 2608 2609
		kvm_regs = memdup_user(argp, sizeof(*kvm_regs));
		if (IS_ERR(kvm_regs)) {
			r = PTR_ERR(kvm_regs);
A
Avi Kivity 已提交
2610
			goto out;
2611
		}
2612 2613
		r = kvm_arch_vcpu_ioctl_set_regs(vcpu, kvm_regs);
		kfree(kvm_regs);
A
Avi Kivity 已提交
2614 2615 2616
		break;
	}
	case KVM_GET_SREGS: {
2617 2618 2619 2620 2621
		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);
A
Avi Kivity 已提交
2622 2623 2624
		if (r)
			goto out;
		r = -EFAULT;
2625
		if (copy_to_user(argp, kvm_sregs, sizeof(struct kvm_sregs)))
A
Avi Kivity 已提交
2626 2627 2628 2629 2630
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_SREGS: {
2631 2632 2633
		kvm_sregs = memdup_user(argp, sizeof(*kvm_sregs));
		if (IS_ERR(kvm_sregs)) {
			r = PTR_ERR(kvm_sregs);
G
Guo Chao 已提交
2634
			kvm_sregs = NULL;
A
Avi Kivity 已提交
2635
			goto out;
2636
		}
2637
		r = kvm_arch_vcpu_ioctl_set_sregs(vcpu, kvm_sregs);
A
Avi Kivity 已提交
2638 2639
		break;
	}
2640 2641 2642 2643 2644 2645 2646
	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;
2647
		if (copy_to_user(argp, &mp_state, sizeof(mp_state)))
2648 2649 2650 2651 2652 2653 2654 2655
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_MP_STATE: {
		struct kvm_mp_state mp_state;

		r = -EFAULT;
2656
		if (copy_from_user(&mp_state, argp, sizeof(mp_state)))
2657 2658 2659 2660
			goto out;
		r = kvm_arch_vcpu_ioctl_set_mpstate(vcpu, &mp_state);
		break;
	}
A
Avi Kivity 已提交
2661 2662 2663 2664
	case KVM_TRANSLATE: {
		struct kvm_translation tr;

		r = -EFAULT;
2665
		if (copy_from_user(&tr, argp, sizeof(tr)))
A
Avi Kivity 已提交
2666
			goto out;
2667
		r = kvm_arch_vcpu_ioctl_translate(vcpu, &tr);
A
Avi Kivity 已提交
2668 2669 2670
		if (r)
			goto out;
		r = -EFAULT;
2671
		if (copy_to_user(argp, &tr, sizeof(tr)))
A
Avi Kivity 已提交
2672 2673 2674 2675
			goto out;
		r = 0;
		break;
	}
J
Jan Kiszka 已提交
2676 2677
	case KVM_SET_GUEST_DEBUG: {
		struct kvm_guest_debug dbg;
A
Avi Kivity 已提交
2678 2679

		r = -EFAULT;
2680
		if (copy_from_user(&dbg, argp, sizeof(dbg)))
A
Avi Kivity 已提交
2681
			goto out;
J
Jan Kiszka 已提交
2682
		r = kvm_arch_vcpu_ioctl_set_guest_debug(vcpu, &dbg);
A
Avi Kivity 已提交
2683 2684
		break;
	}
A
Avi Kivity 已提交
2685 2686 2687 2688 2689 2690 2691 2692 2693
	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,
2694
					   sizeof(kvm_sigmask)))
A
Avi Kivity 已提交
2695 2696
				goto out;
			r = -EINVAL;
2697
			if (kvm_sigmask.len != sizeof(sigset))
A
Avi Kivity 已提交
2698 2699 2700
				goto out;
			r = -EFAULT;
			if (copy_from_user(&sigset, sigmask_arg->sigset,
2701
					   sizeof(sigset)))
A
Avi Kivity 已提交
2702 2703 2704
				goto out;
			p = &sigset;
		}
2705
		r = kvm_vcpu_ioctl_set_sigmask(vcpu, p);
A
Avi Kivity 已提交
2706 2707
		break;
	}
A
Avi Kivity 已提交
2708
	case KVM_GET_FPU: {
2709 2710 2711 2712 2713
		fpu = kzalloc(sizeof(struct kvm_fpu), GFP_KERNEL);
		r = -ENOMEM;
		if (!fpu)
			goto out;
		r = kvm_arch_vcpu_ioctl_get_fpu(vcpu, fpu);
A
Avi Kivity 已提交
2714 2715 2716
		if (r)
			goto out;
		r = -EFAULT;
2717
		if (copy_to_user(argp, fpu, sizeof(struct kvm_fpu)))
A
Avi Kivity 已提交
2718 2719 2720 2721 2722
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_FPU: {
2723 2724 2725
		fpu = memdup_user(argp, sizeof(*fpu));
		if (IS_ERR(fpu)) {
			r = PTR_ERR(fpu);
G
Guo Chao 已提交
2726
			fpu = NULL;
A
Avi Kivity 已提交
2727
			goto out;
2728
		}
2729
		r = kvm_arch_vcpu_ioctl_set_fpu(vcpu, fpu);
A
Avi Kivity 已提交
2730 2731
		break;
	}
A
Avi Kivity 已提交
2732
	default:
2733
		r = kvm_arch_vcpu_ioctl(filp, ioctl, arg);
A
Avi Kivity 已提交
2734 2735
	}
out:
2736
	vcpu_put(vcpu);
2737 2738
	kfree(fpu);
	kfree(kvm_sregs);
A
Avi Kivity 已提交
2739 2740 2741
	return r;
}

2742
#ifdef CONFIG_KVM_COMPAT
2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762
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,
2763
					   sizeof(kvm_sigmask)))
2764 2765
				goto out;
			r = -EINVAL;
2766
			if (kvm_sigmask.len != sizeof(csigset))
2767 2768 2769
				goto out;
			r = -EFAULT;
			if (copy_from_user(&csigset, sigmask_arg->sigset,
2770
					   sizeof(csigset)))
2771
				goto out;
2772 2773 2774 2775
			sigset_from_compat(&sigset, &csigset);
			r = kvm_vcpu_ioctl_set_sigmask(vcpu, &sigset);
		} else
			r = kvm_vcpu_ioctl_set_sigmask(vcpu, NULL);
2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786
		break;
	}
	default:
		r = kvm_vcpu_ioctl(filp, ioctl, arg);
	}

out:
	return r;
}
#endif

S
Scott Wood 已提交
2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833
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;

	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,
2834
#ifdef CONFIG_KVM_COMPAT
2835 2836
	.compat_ioctl = kvm_device_ioctl,
#endif
S
Scott Wood 已提交
2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847
	.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;
}

2848
static struct kvm_device_ops *kvm_device_ops_table[KVM_DEV_TYPE_MAX] = {
2849
#ifdef CONFIG_KVM_MPIC
2850 2851
	[KVM_DEV_TYPE_FSL_MPIC_20]	= &kvm_mpic_ops,
	[KVM_DEV_TYPE_FSL_MPIC_42]	= &kvm_mpic_ops,
2852
#endif
2853

2854
#ifdef CONFIG_KVM_XICS
2855
	[KVM_DEV_TYPE_XICS]		= &kvm_xics_ops,
A
Alex Williamson 已提交
2856
#endif
2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870
};

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;
}

2871 2872 2873 2874 2875 2876
void kvm_unregister_device_ops(u32 type)
{
	if (kvm_device_ops_table[type] != NULL)
		kvm_device_ops_table[type] = NULL;
}

S
Scott Wood 已提交
2877 2878 2879 2880 2881 2882 2883 2884
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;

2885 2886 2887 2888 2889
	if (cd->type >= ARRAY_SIZE(kvm_device_ops_table))
		return -ENODEV;

	ops = kvm_device_ops_table[cd->type];
	if (ops == NULL)
S
Scott Wood 已提交
2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901
		return -ENODEV;

	if (test)
		return 0;

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

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

2902
	mutex_lock(&kvm->lock);
S
Scott Wood 已提交
2903 2904
	ret = ops->create(dev, cd->type);
	if (ret < 0) {
2905
		mutex_unlock(&kvm->lock);
S
Scott Wood 已提交
2906 2907 2908
		kfree(dev);
		return ret;
	}
2909 2910
	list_add(&dev->vm_node, &kvm->devices);
	mutex_unlock(&kvm->lock);
S
Scott Wood 已提交
2911

2912 2913 2914
	if (ops->init)
		ops->init(dev);

2915
	ret = anon_inode_getfd(ops->name, &kvm_device_fops, dev, O_RDWR | O_CLOEXEC);
S
Scott Wood 已提交
2916
	if (ret < 0) {
2917 2918 2919
		mutex_lock(&kvm->lock);
		list_del(&dev->vm_node);
		mutex_unlock(&kvm->lock);
2920
		ops->destroy(dev);
S
Scott Wood 已提交
2921 2922 2923 2924 2925 2926 2927 2928
		return ret;
	}

	kvm_get_kvm(kvm);
	cd->fd = ret;
	return 0;
}

2929 2930 2931 2932 2933 2934 2935 2936 2937 2938
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
2939
#ifdef CONFIG_HAVE_KVM_IRQFD
2940
	case KVM_CAP_IRQFD:
2941 2942
	case KVM_CAP_IRQFD_RESAMPLE:
#endif
2943
	case KVM_CAP_IOEVENTFD_ANY_LENGTH:
2944 2945 2946 2947 2948
	case KVM_CAP_CHECK_EXTENSION_VM:
		return 1;
#ifdef CONFIG_HAVE_KVM_IRQ_ROUTING
	case KVM_CAP_IRQ_ROUTING:
		return KVM_MAX_IRQ_ROUTES;
2949 2950 2951 2952
#endif
#if KVM_ADDRESS_SPACE_NUM > 1
	case KVM_CAP_MULTI_ADDRESS_SPACE:
		return KVM_ADDRESS_SPACE_NUM;
2953
#endif
G
Greg Kurz 已提交
2954 2955
	case KVM_CAP_MAX_VCPU_ID:
		return KVM_MAX_VCPU_ID;
2956 2957 2958 2959 2960 2961
	default:
		break;
	}
	return kvm_vm_ioctl_check_extension(kvm, arg);
}

A
Avi Kivity 已提交
2962 2963 2964 2965 2966
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;
2967
	int r;
A
Avi Kivity 已提交
2968

2969 2970
	if (kvm->mm != current->mm)
		return -EIO;
A
Avi Kivity 已提交
2971 2972 2973 2974
	switch (ioctl) {
	case KVM_CREATE_VCPU:
		r = kvm_vm_ioctl_create_vcpu(kvm, arg);
		break;
2975 2976 2977 2978 2979
	case KVM_SET_USER_MEMORY_REGION: {
		struct kvm_userspace_memory_region kvm_userspace_mem;

		r = -EFAULT;
		if (copy_from_user(&kvm_userspace_mem, argp,
2980
						sizeof(kvm_userspace_mem)))
2981 2982
			goto out;

2983
		r = kvm_vm_ioctl_set_memory_region(kvm, &kvm_userspace_mem);
A
Avi Kivity 已提交
2984 2985 2986 2987 2988 2989
		break;
	}
	case KVM_GET_DIRTY_LOG: {
		struct kvm_dirty_log log;

		r = -EFAULT;
2990
		if (copy_from_user(&log, argp, sizeof(log)))
A
Avi Kivity 已提交
2991
			goto out;
2992
		r = kvm_vm_ioctl_get_dirty_log(kvm, &log);
A
Avi Kivity 已提交
2993 2994
		break;
	}
2995 2996 2997
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
	case KVM_REGISTER_COALESCED_MMIO: {
		struct kvm_coalesced_mmio_zone zone;
2998

2999
		r = -EFAULT;
3000
		if (copy_from_user(&zone, argp, sizeof(zone)))
3001 3002 3003 3004 3005 3006
			goto out;
		r = kvm_vm_ioctl_register_coalesced_mmio(kvm, &zone);
		break;
	}
	case KVM_UNREGISTER_COALESCED_MMIO: {
		struct kvm_coalesced_mmio_zone zone;
3007

3008
		r = -EFAULT;
3009
		if (copy_from_user(&zone, argp, sizeof(zone)))
3010 3011 3012 3013 3014
			goto out;
		r = kvm_vm_ioctl_unregister_coalesced_mmio(kvm, &zone);
		break;
	}
#endif
G
Gregory Haskins 已提交
3015 3016 3017 3018
	case KVM_IRQFD: {
		struct kvm_irqfd data;

		r = -EFAULT;
3019
		if (copy_from_user(&data, argp, sizeof(data)))
G
Gregory Haskins 已提交
3020
			goto out;
3021
		r = kvm_irqfd(kvm, &data);
G
Gregory Haskins 已提交
3022 3023
		break;
	}
G
Gregory Haskins 已提交
3024 3025 3026 3027
	case KVM_IOEVENTFD: {
		struct kvm_ioeventfd data;

		r = -EFAULT;
3028
		if (copy_from_user(&data, argp, sizeof(data)))
G
Gregory Haskins 已提交
3029 3030 3031 3032
			goto out;
		r = kvm_ioeventfd(kvm, &data);
		break;
	}
3033 3034 3035 3036 3037
#ifdef CONFIG_HAVE_KVM_MSI
	case KVM_SIGNAL_MSI: {
		struct kvm_msi msi;

		r = -EFAULT;
3038
		if (copy_from_user(&msi, argp, sizeof(msi)))
3039 3040 3041 3042
			goto out;
		r = kvm_send_userspace_msi(kvm, &msi);
		break;
	}
3043 3044 3045 3046 3047 3048 3049
#endif
#ifdef __KVM_HAVE_IRQ_LINE
	case KVM_IRQ_LINE_STATUS:
	case KVM_IRQ_LINE: {
		struct kvm_irq_level irq_event;

		r = -EFAULT;
3050
		if (copy_from_user(&irq_event, argp, sizeof(irq_event)))
3051 3052
			goto out;

3053 3054
		r = kvm_vm_ioctl_irq_line(kvm, &irq_event,
					ioctl == KVM_IRQ_LINE_STATUS);
3055 3056 3057 3058 3059
		if (r)
			goto out;

		r = -EFAULT;
		if (ioctl == KVM_IRQ_LINE_STATUS) {
3060
			if (copy_to_user(argp, &irq_event, sizeof(irq_event)))
3061 3062 3063 3064 3065 3066
				goto out;
		}

		r = 0;
		break;
	}
3067
#endif
3068 3069 3070 3071
#ifdef CONFIG_HAVE_KVM_IRQ_ROUTING
	case KVM_SET_GSI_ROUTING: {
		struct kvm_irq_routing routing;
		struct kvm_irq_routing __user *urouting;
3072
		struct kvm_irq_routing_entry *entries = NULL;
3073 3074 3075 3076 3077

		r = -EFAULT;
		if (copy_from_user(&routing, argp, sizeof(routing)))
			goto out;
		r = -EINVAL;
3078
		if (routing.nr > KVM_MAX_IRQ_ROUTES)
3079 3080 3081
			goto out;
		if (routing.flags)
			goto out;
3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092
		if (routing.nr) {
			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;
		}
3093 3094
		r = kvm_set_irq_routing(kvm, entries, routing.nr,
					routing.flags);
3095
out_free_irq_routing:
3096 3097 3098 3099
		vfree(entries);
		break;
	}
#endif /* CONFIG_HAVE_KVM_IRQ_ROUTING */
S
Scott Wood 已提交
3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117
	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;
	}
3118 3119 3120
	case KVM_CHECK_EXTENSION:
		r = kvm_vm_ioctl_check_extension_generic(kvm, arg);
		break;
3121
	default:
3122
		r = kvm_arch_vm_ioctl(filp, ioctl, arg);
3123 3124 3125 3126 3127
	}
out:
	return r;
}

3128
#ifdef CONFIG_KVM_COMPAT
3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171
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

3172
static struct file_operations kvm_vm_fops = {
3173 3174
	.release        = kvm_vm_release,
	.unlocked_ioctl = kvm_vm_ioctl,
3175
#ifdef CONFIG_KVM_COMPAT
3176 3177
	.compat_ioctl   = kvm_vm_compat_ioctl,
#endif
3178
	.llseek		= noop_llseek,
3179 3180
};

3181
static int kvm_dev_ioctl_create_vm(unsigned long type)
3182
{
3183
	int r;
3184
	struct kvm *kvm;
3185
	struct file *file;
3186

3187
	kvm = kvm_create_vm(type);
3188 3189
	if (IS_ERR(kvm))
		return PTR_ERR(kvm);
3190 3191 3192 3193 3194 3195 3196
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
	r = kvm_coalesced_mmio_init(kvm);
	if (r < 0) {
		kvm_put_kvm(kvm);
		return r;
	}
#endif
3197
	r = get_unused_fd_flags(O_CLOEXEC);
3198
	if (r < 0) {
A
Al Viro 已提交
3199
		kvm_put_kvm(kvm);
3200 3201
		return r;
	}
3202 3203 3204 3205 3206 3207
	file = anon_inode_getfile("kvm-vm", &kvm_vm_fops, kvm, O_RDWR);
	if (IS_ERR(file)) {
		put_unused_fd(r);
		kvm_put_kvm(kvm);
		return PTR_ERR(file);
	}
3208 3209

	if (kvm_create_vm_debugfs(kvm, r) < 0) {
3210 3211
		put_unused_fd(r);
		fput(file);
3212 3213
		return -ENOMEM;
	}
3214

3215
	fd_install(r, file);
3216
	return r;
3217 3218 3219 3220 3221
}

static long kvm_dev_ioctl(struct file *filp,
			  unsigned int ioctl, unsigned long arg)
{
3222
	long r = -EINVAL;
3223 3224 3225

	switch (ioctl) {
	case KVM_GET_API_VERSION:
3226 3227
		if (arg)
			goto out;
3228 3229 3230
		r = KVM_API_VERSION;
		break;
	case KVM_CREATE_VM:
3231
		r = kvm_dev_ioctl_create_vm(arg);
3232
		break;
3233
	case KVM_CHECK_EXTENSION:
3234
		r = kvm_vm_ioctl_check_extension_generic(NULL, arg);
3235
		break;
3236 3237 3238
	case KVM_GET_VCPU_MMAP_SIZE:
		if (arg)
			goto out;
3239 3240 3241
		r = PAGE_SIZE;     /* struct kvm_run */
#ifdef CONFIG_X86
		r += PAGE_SIZE;    /* pio data page */
3242 3243 3244
#endif
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
		r += PAGE_SIZE;    /* coalesced mmio ring page */
3245
#endif
3246
		break;
3247 3248 3249
	case KVM_TRACE_ENABLE:
	case KVM_TRACE_PAUSE:
	case KVM_TRACE_DISABLE:
3250
		r = -EOPNOTSUPP;
3251
		break;
A
Avi Kivity 已提交
3252
	default:
3253
		return kvm_arch_dev_ioctl(filp, ioctl, arg);
A
Avi Kivity 已提交
3254 3255 3256 3257 3258 3259 3260 3261
	}
out:
	return r;
}

static struct file_operations kvm_chardev_ops = {
	.unlocked_ioctl = kvm_dev_ioctl,
	.compat_ioctl   = kvm_dev_ioctl,
3262
	.llseek		= noop_llseek,
A
Avi Kivity 已提交
3263 3264 3265
};

static struct miscdevice kvm_dev = {
A
Avi Kivity 已提交
3266
	KVM_MINOR,
A
Avi Kivity 已提交
3267 3268 3269 3270
	"kvm",
	&kvm_chardev_ops,
};

3271
static void hardware_enable_nolock(void *junk)
3272 3273
{
	int cpu = raw_smp_processor_id();
3274
	int r;
3275

3276
	if (cpumask_test_cpu(cpu, cpus_hardware_enabled))
3277
		return;
3278

3279
	cpumask_set_cpu(cpu, cpus_hardware_enabled);
3280

3281
	r = kvm_arch_hardware_enable();
3282 3283 3284 3285

	if (r) {
		cpumask_clear_cpu(cpu, cpus_hardware_enabled);
		atomic_inc(&hardware_enable_failed);
X
Xiubo Li 已提交
3286
		pr_info("kvm: enabling virtualization on CPU%d failed\n", cpu);
3287
	}
3288 3289
}

3290
static int kvm_starting_cpu(unsigned int cpu)
3291
{
3292
	raw_spin_lock(&kvm_count_lock);
3293 3294
	if (kvm_usage_count)
		hardware_enable_nolock(NULL);
3295
	raw_spin_unlock(&kvm_count_lock);
3296
	return 0;
3297 3298 3299
}

static void hardware_disable_nolock(void *junk)
3300 3301 3302
{
	int cpu = raw_smp_processor_id();

3303
	if (!cpumask_test_cpu(cpu, cpus_hardware_enabled))
3304
		return;
3305
	cpumask_clear_cpu(cpu, cpus_hardware_enabled);
3306
	kvm_arch_hardware_disable();
3307 3308
}

3309
static int kvm_dying_cpu(unsigned int cpu)
3310
{
3311
	raw_spin_lock(&kvm_count_lock);
3312 3313
	if (kvm_usage_count)
		hardware_disable_nolock(NULL);
3314
	raw_spin_unlock(&kvm_count_lock);
3315
	return 0;
3316 3317
}

3318 3319 3320 3321 3322 3323
static void hardware_disable_all_nolock(void)
{
	BUG_ON(!kvm_usage_count);

	kvm_usage_count--;
	if (!kvm_usage_count)
3324
		on_each_cpu(hardware_disable_nolock, NULL, 1);
3325 3326 3327 3328
}

static void hardware_disable_all(void)
{
3329
	raw_spin_lock(&kvm_count_lock);
3330
	hardware_disable_all_nolock();
3331
	raw_spin_unlock(&kvm_count_lock);
3332 3333 3334 3335 3336 3337
}

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

3338
	raw_spin_lock(&kvm_count_lock);
3339 3340 3341 3342

	kvm_usage_count++;
	if (kvm_usage_count == 1) {
		atomic_set(&hardware_enable_failed, 0);
3343
		on_each_cpu(hardware_enable_nolock, NULL, 1);
3344 3345 3346 3347 3348 3349 3350

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

3351
	raw_spin_unlock(&kvm_count_lock);
3352 3353 3354 3355

	return r;
}

3356
static int kvm_reboot(struct notifier_block *notifier, unsigned long val,
M
Mike Day 已提交
3357
		      void *v)
3358
{
3359 3360 3361 3362 3363 3364
	/*
	 * 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.
	 */
X
Xiubo Li 已提交
3365
	pr_info("kvm: exiting hardware virtualization\n");
3366
	kvm_rebooting = true;
3367
	on_each_cpu(hardware_disable_nolock, NULL, 1);
3368 3369 3370 3371 3372 3373 3374 3375
	return NOTIFY_OK;
}

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

M
Marcelo Tosatti 已提交
3376
static void kvm_io_bus_destroy(struct kvm_io_bus *bus)
3377 3378 3379 3380
{
	int i;

	for (i = 0; i < bus->dev_count; i++) {
3381
		struct kvm_io_device *pos = bus->range[i].dev;
3382 3383 3384

		kvm_iodevice_destructor(pos);
	}
M
Marcelo Tosatti 已提交
3385
	kfree(bus);
3386 3387
}

3388
static inline int kvm_io_bus_cmp(const struct kvm_io_range *r1,
X
Xiubo Li 已提交
3389
				 const struct kvm_io_range *r2)
3390
{
J
Jason Wang 已提交
3391 3392 3393 3394
	gpa_t addr1 = r1->addr;
	gpa_t addr2 = r2->addr;

	if (addr1 < addr2)
3395
		return -1;
J
Jason Wang 已提交
3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407

	/* 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)
3408
		return 1;
J
Jason Wang 已提交
3409

3410 3411 3412
	return 0;
}

3413 3414
static int kvm_io_bus_sort_cmp(const void *p1, const void *p2)
{
3415
	return kvm_io_bus_cmp(p1, p2);
3416 3417
}

G
Geoff Levand 已提交
3418
static int kvm_io_bus_insert_dev(struct kvm_io_bus *bus, struct kvm_io_device *dev,
3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432
			  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;
}

G
Geoff Levand 已提交
3433
static int kvm_io_bus_get_first_dev(struct kvm_io_bus *bus,
3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450
			     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;

3451
	while (off > 0 && kvm_io_bus_cmp(&key, &bus->range[off-1]) == 0)
3452 3453 3454 3455 3456
		off--;

	return off;
}

3457
static int __kvm_io_bus_write(struct kvm_vcpu *vcpu, struct kvm_io_bus *bus,
C
Cornelia Huck 已提交
3458 3459 3460 3461 3462 3463 3464 3465 3466
			      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 &&
3467
		kvm_io_bus_cmp(range, &bus->range[idx]) == 0) {
3468
		if (!kvm_iodevice_write(vcpu, bus->range[idx].dev, range->addr,
C
Cornelia Huck 已提交
3469 3470 3471 3472 3473 3474 3475 3476
					range->len, val))
			return idx;
		idx++;
	}

	return -EOPNOTSUPP;
}

3477
/* kvm_io_bus_write - called under kvm->slots_lock */
3478
int kvm_io_bus_write(struct kvm_vcpu *vcpu, enum kvm_bus bus_idx, gpa_t addr,
3479
		     int len, const void *val)
3480
{
3481
	struct kvm_io_bus *bus;
3482
	struct kvm_io_range range;
C
Cornelia Huck 已提交
3483
	int r;
3484 3485 3486 3487 3488

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};
3489

3490 3491
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
	r = __kvm_io_bus_write(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
3492 3493 3494 3495
	return r < 0 ? r : 0;
}

/* kvm_io_bus_write_cookie - called under kvm->slots_lock */
3496 3497
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)
C
Cornelia Huck 已提交
3498 3499 3500 3501 3502 3503 3504 3505 3506
{
	struct kvm_io_bus *bus;
	struct kvm_io_range range;

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

3507
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
C
Cornelia Huck 已提交
3508 3509 3510

	/* First try the device referenced by cookie. */
	if ((cookie >= 0) && (cookie < bus->dev_count) &&
3511
	    (kvm_io_bus_cmp(&range, &bus->range[cookie]) == 0))
3512
		if (!kvm_iodevice_write(vcpu, bus->range[cookie].dev, addr, len,
C
Cornelia Huck 已提交
3513 3514 3515 3516 3517 3518 3519
					val))
			return cookie;

	/*
	 * cookie contained garbage; fall back to search and return the
	 * correct cookie value.
	 */
3520
	return __kvm_io_bus_write(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
3521 3522
}

3523 3524
static int __kvm_io_bus_read(struct kvm_vcpu *vcpu, struct kvm_io_bus *bus,
			     struct kvm_io_range *range, void *val)
C
Cornelia Huck 已提交
3525 3526 3527 3528
{
	int idx;

	idx = kvm_io_bus_get_first_dev(bus, range->addr, range->len);
3529 3530 3531 3532
	if (idx < 0)
		return -EOPNOTSUPP;

	while (idx < bus->dev_count &&
3533
		kvm_io_bus_cmp(range, &bus->range[idx]) == 0) {
3534
		if (!kvm_iodevice_read(vcpu, bus->range[idx].dev, range->addr,
C
Cornelia Huck 已提交
3535 3536
				       range->len, val))
			return idx;
3537 3538 3539
		idx++;
	}

3540 3541
	return -EOPNOTSUPP;
}
3542
EXPORT_SYMBOL_GPL(kvm_io_bus_write);
3543

3544
/* kvm_io_bus_read - called under kvm->slots_lock */
3545
int kvm_io_bus_read(struct kvm_vcpu *vcpu, enum kvm_bus bus_idx, gpa_t addr,
M
Marcelo Tosatti 已提交
3546
		    int len, void *val)
3547
{
3548
	struct kvm_io_bus *bus;
3549
	struct kvm_io_range range;
C
Cornelia Huck 已提交
3550
	int r;
3551 3552 3553 3554 3555

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};
M
Marcelo Tosatti 已提交
3556

3557 3558
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
	r = __kvm_io_bus_read(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
3559 3560
	return r < 0 ? r : 0;
}
3561

3562

3563
/* Caller must hold slots_lock. */
3564 3565
int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
			    int len, struct kvm_io_device *dev)
3566
{
M
Marcelo Tosatti 已提交
3567
	struct kvm_io_bus *new_bus, *bus;
3568

M
Marcelo Tosatti 已提交
3569
	bus = kvm->buses[bus_idx];
3570 3571
	/* exclude ioeventfd which is limited by maximum fd */
	if (bus->dev_count - bus->ioeventfd_count > NR_IOBUS_DEVS - 1)
3572
		return -ENOSPC;
3573

3574
	new_bus = kmalloc(sizeof(*bus) + ((bus->dev_count + 1) *
3575
			  sizeof(struct kvm_io_range)), GFP_KERNEL);
M
Marcelo Tosatti 已提交
3576 3577
	if (!new_bus)
		return -ENOMEM;
3578 3579
	memcpy(new_bus, bus, sizeof(*bus) + (bus->dev_count *
	       sizeof(struct kvm_io_range)));
3580
	kvm_io_bus_insert_dev(new_bus, dev, addr, len);
M
Marcelo Tosatti 已提交
3581 3582 3583
	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
	synchronize_srcu_expedited(&kvm->srcu);
	kfree(bus);
3584 3585 3586 3587

	return 0;
}

3588
/* Caller must hold slots_lock. */
M
Marcelo Tosatti 已提交
3589 3590
int kvm_io_bus_unregister_dev(struct kvm *kvm, enum kvm_bus bus_idx,
			      struct kvm_io_device *dev)
3591
{
M
Marcelo Tosatti 已提交
3592 3593
	int i, r;
	struct kvm_io_bus *new_bus, *bus;
3594

3595
	bus = kvm->buses[bus_idx];
M
Marcelo Tosatti 已提交
3596
	r = -ENOENT;
3597 3598
	for (i = 0; i < bus->dev_count; i++)
		if (bus->range[i].dev == dev) {
M
Marcelo Tosatti 已提交
3599
			r = 0;
3600 3601
			break;
		}
M
Marcelo Tosatti 已提交
3602

3603
	if (r)
M
Marcelo Tosatti 已提交
3604
		return r;
3605

3606
	new_bus = kmalloc(sizeof(*bus) + ((bus->dev_count - 1) *
3607 3608 3609 3610 3611 3612 3613 3614
			  sizeof(struct kvm_io_range)), GFP_KERNEL);
	if (!new_bus)
		return -ENOMEM;

	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));
M
Marcelo Tosatti 已提交
3615 3616 3617 3618 3619

	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
	synchronize_srcu_expedited(&kvm->srcu);
	kfree(bus);
	return r;
3620 3621
}

3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645
struct kvm_io_device *kvm_io_bus_get_dev(struct kvm *kvm, enum kvm_bus bus_idx,
					 gpa_t addr)
{
	struct kvm_io_bus *bus;
	int dev_idx, srcu_idx;
	struct kvm_io_device *iodev = NULL;

	srcu_idx = srcu_read_lock(&kvm->srcu);

	bus = srcu_dereference(kvm->buses[bus_idx], &kvm->srcu);

	dev_idx = kvm_io_bus_get_first_dev(bus, addr, 1);
	if (dev_idx < 0)
		goto out_unlock;

	iodev = bus->range[dev_idx].dev;

out_unlock:
	srcu_read_unlock(&kvm->srcu, srcu_idx);

	return iodev;
}
EXPORT_SYMBOL_GPL(kvm_io_bus_get_dev);

3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683
static int kvm_debugfs_open(struct inode *inode, struct file *file,
			   int (*get)(void *, u64 *), int (*set)(void *, u64),
			   const char *fmt)
{
	struct kvm_stat_data *stat_data = (struct kvm_stat_data *)
					  inode->i_private;

	/* The debugfs files are a reference to the kvm struct which
	 * is still valid when kvm_destroy_vm is called.
	 * To avoid the race between open and the removal of the debugfs
	 * directory we test against the users count.
	 */
	if (!atomic_add_unless(&stat_data->kvm->users_count, 1, 0))
		return -ENOENT;

	if (simple_attr_open(inode, file, get, set, fmt)) {
		kvm_put_kvm(stat_data->kvm);
		return -ENOMEM;
	}

	return 0;
}

static int kvm_debugfs_release(struct inode *inode, struct file *file)
{
	struct kvm_stat_data *stat_data = (struct kvm_stat_data *)
					  inode->i_private;

	simple_attr_release(inode, file);
	kvm_put_kvm(stat_data->kvm);

	return 0;
}

static int vm_stat_get_per_vm(void *data, u64 *val)
{
	struct kvm_stat_data *stat_data = (struct kvm_stat_data *)data;

3684
	*val = *(ulong *)((void *)stat_data->kvm + stat_data->offset);
3685 3686 3687 3688

	return 0;
}

3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700
static int vm_stat_clear_per_vm(void *data, u64 val)
{
	struct kvm_stat_data *stat_data = (struct kvm_stat_data *)data;

	if (val)
		return -EINVAL;

	*(ulong *)((void *)stat_data->kvm + stat_data->offset) = 0;

	return 0;
}

3701 3702 3703 3704
static int vm_stat_get_per_vm_open(struct inode *inode, struct file *file)
{
	__simple_attr_check_format("%llu\n", 0ull);
	return kvm_debugfs_open(inode, file, vm_stat_get_per_vm,
3705
				vm_stat_clear_per_vm, "%llu\n");
3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725
}

static const struct file_operations vm_stat_get_per_vm_fops = {
	.owner   = THIS_MODULE,
	.open    = vm_stat_get_per_vm_open,
	.release = kvm_debugfs_release,
	.read    = simple_attr_read,
	.write   = simple_attr_write,
	.llseek  = generic_file_llseek,
};

static int vcpu_stat_get_per_vm(void *data, u64 *val)
{
	int i;
	struct kvm_stat_data *stat_data = (struct kvm_stat_data *)data;
	struct kvm_vcpu *vcpu;

	*val = 0;

	kvm_for_each_vcpu(i, vcpu, stat_data->kvm)
3726
		*val += *(u64 *)((void *)vcpu + stat_data->offset);
3727 3728 3729 3730

	return 0;
}

3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745
static int vcpu_stat_clear_per_vm(void *data, u64 val)
{
	int i;
	struct kvm_stat_data *stat_data = (struct kvm_stat_data *)data;
	struct kvm_vcpu *vcpu;

	if (val)
		return -EINVAL;

	kvm_for_each_vcpu(i, vcpu, stat_data->kvm)
		*(u64 *)((void *)vcpu + stat_data->offset) = 0;

	return 0;
}

3746 3747 3748 3749
static int vcpu_stat_get_per_vm_open(struct inode *inode, struct file *file)
{
	__simple_attr_check_format("%llu\n", 0ull);
	return kvm_debugfs_open(inode, file, vcpu_stat_get_per_vm,
3750
				 vcpu_stat_clear_per_vm, "%llu\n");
3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766
}

static const struct file_operations vcpu_stat_get_per_vm_fops = {
	.owner   = THIS_MODULE,
	.open    = vcpu_stat_get_per_vm_open,
	.release = kvm_debugfs_release,
	.read    = simple_attr_read,
	.write   = simple_attr_write,
	.llseek  = generic_file_llseek,
};

static const struct file_operations *stat_fops_per_vm[] = {
	[KVM_STAT_VCPU] = &vcpu_stat_get_per_vm_fops,
	[KVM_STAT_VM]   = &vm_stat_get_per_vm_fops,
};

3767
static int vm_stat_get(void *_offset, u64 *val)
3768 3769 3770
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;
3771 3772
	struct kvm_stat_data stat_tmp = {.offset = offset};
	u64 tmp_val;
3773

3774
	*val = 0;
3775
	spin_lock(&kvm_lock);
3776 3777 3778 3779 3780
	list_for_each_entry(kvm, &vm_list, vm_list) {
		stat_tmp.kvm = kvm;
		vm_stat_get_per_vm((void *)&stat_tmp, &tmp_val);
		*val += tmp_val;
	}
3781
	spin_unlock(&kvm_lock);
3782
	return 0;
3783 3784
}

3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804
static int vm_stat_clear(void *_offset, u64 val)
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;
	struct kvm_stat_data stat_tmp = {.offset = offset};

	if (val)
		return -EINVAL;

	spin_lock(&kvm_lock);
	list_for_each_entry(kvm, &vm_list, vm_list) {
		stat_tmp.kvm = kvm;
		vm_stat_clear_per_vm((void *)&stat_tmp, 0);
	}
	spin_unlock(&kvm_lock);

	return 0;
}

DEFINE_SIMPLE_ATTRIBUTE(vm_stat_fops, vm_stat_get, vm_stat_clear, "%llu\n");
3805

3806
static int vcpu_stat_get(void *_offset, u64 *val)
A
Avi Kivity 已提交
3807 3808 3809
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;
3810 3811
	struct kvm_stat_data stat_tmp = {.offset = offset};
	u64 tmp_val;
A
Avi Kivity 已提交
3812

3813
	*val = 0;
3814
	spin_lock(&kvm_lock);
3815 3816 3817 3818 3819
	list_for_each_entry(kvm, &vm_list, vm_list) {
		stat_tmp.kvm = kvm;
		vcpu_stat_get_per_vm((void *)&stat_tmp, &tmp_val);
		*val += tmp_val;
	}
3820
	spin_unlock(&kvm_lock);
3821
	return 0;
A
Avi Kivity 已提交
3822 3823
}

3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844
static int vcpu_stat_clear(void *_offset, u64 val)
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;
	struct kvm_stat_data stat_tmp = {.offset = offset};

	if (val)
		return -EINVAL;

	spin_lock(&kvm_lock);
	list_for_each_entry(kvm, &vm_list, vm_list) {
		stat_tmp.kvm = kvm;
		vcpu_stat_clear_per_vm((void *)&stat_tmp, 0);
	}
	spin_unlock(&kvm_lock);

	return 0;
}

DEFINE_SIMPLE_ATTRIBUTE(vcpu_stat_fops, vcpu_stat_get, vcpu_stat_clear,
			"%llu\n");
3845

3846
static const struct file_operations *stat_fops[] = {
3847 3848 3849
	[KVM_STAT_VCPU] = &vcpu_stat_fops,
	[KVM_STAT_VM]   = &vm_stat_fops,
};
A
Avi Kivity 已提交
3850

3851
static int kvm_init_debug(void)
A
Avi Kivity 已提交
3852
{
3853
	int r = -EEXIST;
A
Avi Kivity 已提交
3854 3855
	struct kvm_stats_debugfs_item *p;

3856
	kvm_debugfs_dir = debugfs_create_dir("kvm", NULL);
3857 3858 3859
	if (kvm_debugfs_dir == NULL)
		goto out;

3860 3861
	kvm_debugfs_num_entries = 0;
	for (p = debugfs_entries; p->name; ++p, kvm_debugfs_num_entries++) {
3862
		if (!debugfs_create_file(p->name, 0644, kvm_debugfs_dir,
3863 3864
					 (void *)(long)p->offset,
					 stat_fops[p->kind]))
3865 3866 3867 3868 3869 3870 3871 3872 3873
			goto out_dir;
	}

	return 0;

out_dir:
	debugfs_remove_recursive(kvm_debugfs_dir);
out:
	return r;
A
Avi Kivity 已提交
3874 3875
}

3876
static int kvm_suspend(void)
3877
{
3878
	if (kvm_usage_count)
3879
		hardware_disable_nolock(NULL);
3880 3881 3882
	return 0;
}

3883
static void kvm_resume(void)
3884
{
3885
	if (kvm_usage_count) {
3886
		WARN_ON(raw_spin_is_locked(&kvm_count_lock));
3887
		hardware_enable_nolock(NULL);
3888
	}
3889 3890
}

3891
static struct syscore_ops kvm_syscore_ops = {
3892 3893 3894 3895
	.suspend = kvm_suspend,
	.resume = kvm_resume,
};

3896 3897 3898 3899 3900 3901 3902 3903 3904
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);
3905

3906 3907
	if (vcpu->preempted)
		vcpu->preempted = false;
3908

R
Radim Krčmář 已提交
3909 3910
	kvm_arch_sched_in(vcpu, cpu);

3911
	kvm_arch_vcpu_load(vcpu, cpu);
3912 3913 3914 3915 3916 3917 3918
}

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

3919 3920
	if (current->state == TASK_RUNNING)
		vcpu->preempted = true;
3921
	kvm_arch_vcpu_put(vcpu);
3922 3923
}

3924
int kvm_init(void *opaque, unsigned vcpu_size, unsigned vcpu_align,
3925
		  struct module *module)
A
Avi Kivity 已提交
3926 3927
{
	int r;
Y
Yang, Sheng 已提交
3928
	int cpu;
A
Avi Kivity 已提交
3929

3930 3931
	r = kvm_arch_init(opaque);
	if (r)
3932
		goto out_fail;
3933

3934 3935 3936 3937
	/*
	 * kvm_arch_init makes sure there's at most one caller
	 * for architectures that support multiple implementations,
	 * like intel and amd on x86.
P
Paolo Bonzini 已提交
3938 3939
	 * kvm_arch_init must be called before kvm_irqfd_init to avoid creating
	 * conflicts in case kvm is already setup for another implementation.
3940
	 */
P
Paolo Bonzini 已提交
3941 3942 3943
	r = kvm_irqfd_init();
	if (r)
		goto out_irqfd;
3944

3945
	if (!zalloc_cpumask_var(&cpus_hardware_enabled, GFP_KERNEL)) {
3946 3947 3948 3949
		r = -ENOMEM;
		goto out_free_0;
	}

3950
	r = kvm_arch_hardware_setup();
A
Avi Kivity 已提交
3951
	if (r < 0)
3952
		goto out_free_0a;
A
Avi Kivity 已提交
3953

Y
Yang, Sheng 已提交
3954 3955
	for_each_online_cpu(cpu) {
		smp_call_function_single(cpu,
3956
				kvm_arch_check_processor_compat,
3957
				&r, 1);
Y
Yang, Sheng 已提交
3958
		if (r < 0)
3959
			goto out_free_1;
Y
Yang, Sheng 已提交
3960 3961
	}

T
Thomas Gleixner 已提交
3962
	r = cpuhp_setup_state_nocalls(CPUHP_AP_KVM_STARTING, "kvm/cpu:starting",
3963
				      kvm_starting_cpu, kvm_dying_cpu);
A
Avi Kivity 已提交
3964
	if (r)
3965
		goto out_free_2;
A
Avi Kivity 已提交
3966 3967
	register_reboot_notifier(&kvm_reboot_notifier);

3968
	/* A kmem cache lets us meet the alignment requirements of fx_save. */
3969 3970 3971
	if (!vcpu_align)
		vcpu_align = __alignof__(struct kvm_vcpu);
	kvm_vcpu_cache = kmem_cache_create("kvm_vcpu", vcpu_size, vcpu_align,
J
Joe Perches 已提交
3972
					   0, NULL);
3973 3974
	if (!kvm_vcpu_cache) {
		r = -ENOMEM;
3975
		goto out_free_3;
3976 3977
	}

3978 3979 3980 3981
	r = kvm_async_pf_init();
	if (r)
		goto out_free;

A
Avi Kivity 已提交
3982
	kvm_chardev_ops.owner = module;
3983 3984
	kvm_vm_fops.owner = module;
	kvm_vcpu_fops.owner = module;
A
Avi Kivity 已提交
3985 3986 3987

	r = misc_register(&kvm_dev);
	if (r) {
X
Xiubo Li 已提交
3988
		pr_err("kvm: misc device register failed\n");
3989
		goto out_unreg;
A
Avi Kivity 已提交
3990 3991
	}

3992 3993
	register_syscore_ops(&kvm_syscore_ops);

3994 3995 3996
	kvm_preempt_ops.sched_in = kvm_sched_in;
	kvm_preempt_ops.sched_out = kvm_sched_out;

3997 3998
	r = kvm_init_debug();
	if (r) {
X
Xiubo Li 已提交
3999
		pr_err("kvm: create debugfs files failed\n");
4000 4001
		goto out_undebugfs;
	}
4002

P
Paolo Bonzini 已提交
4003 4004 4005
	r = kvm_vfio_ops_init();
	WARN_ON(r);

4006
	return 0;
A
Avi Kivity 已提交
4007

4008 4009
out_undebugfs:
	unregister_syscore_ops(&kvm_syscore_ops);
4010
	misc_deregister(&kvm_dev);
4011 4012
out_unreg:
	kvm_async_pf_deinit();
A
Avi Kivity 已提交
4013
out_free:
4014
	kmem_cache_destroy(kvm_vcpu_cache);
4015
out_free_3:
A
Avi Kivity 已提交
4016
	unregister_reboot_notifier(&kvm_reboot_notifier);
4017
	cpuhp_remove_state_nocalls(CPUHP_AP_KVM_STARTING);
4018 4019
out_free_2:
out_free_1:
4020
	kvm_arch_hardware_unsetup();
4021 4022
out_free_0a:
	free_cpumask_var(cpus_hardware_enabled);
4023
out_free_0:
4024
	kvm_irqfd_exit();
P
Paolo Bonzini 已提交
4025
out_irqfd:
4026 4027
	kvm_arch_exit();
out_fail:
A
Avi Kivity 已提交
4028 4029
	return r;
}
4030
EXPORT_SYMBOL_GPL(kvm_init);
A
Avi Kivity 已提交
4031

4032
void kvm_exit(void)
A
Avi Kivity 已提交
4033
{
4034
	debugfs_remove_recursive(kvm_debugfs_dir);
A
Avi Kivity 已提交
4035
	misc_deregister(&kvm_dev);
4036
	kmem_cache_destroy(kvm_vcpu_cache);
4037
	kvm_async_pf_deinit();
4038
	unregister_syscore_ops(&kvm_syscore_ops);
A
Avi Kivity 已提交
4039
	unregister_reboot_notifier(&kvm_reboot_notifier);
4040
	cpuhp_remove_state_nocalls(CPUHP_AP_KVM_STARTING);
4041
	on_each_cpu(hardware_disable_nolock, NULL, 1);
4042
	kvm_arch_hardware_unsetup();
4043
	kvm_arch_exit();
4044
	kvm_irqfd_exit();
4045
	free_cpumask_var(cpus_hardware_enabled);
4046
	kvm_vfio_ops_exit();
A
Avi Kivity 已提交
4047
}
4048
EXPORT_SYMBOL_GPL(kvm_exit);