kvm_main.c 101.4 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>
35
#include <linux/sched/signal.h>
36
#include <linux/sched/mm.h>
37
#include <linux/sched/stat.h>
38 39
#include <linux/cpumask.h>
#include <linux/smp.h>
40
#include <linux/anon_inodes.h>
41
#include <linux/profile.h>
42
#include <linux/kvm_para.h>
43
#include <linux/pagemap.h>
44
#include <linux/mman.h>
45
#include <linux/swap.h>
46
#include <linux/bitops.h>
47
#include <linux/spinlock.h>
48
#include <linux/compat.h>
49
#include <linux/srcu.h>
J
Joerg Roedel 已提交
50
#include <linux/hugetlb.h>
51
#include <linux/slab.h>
52 53
#include <linux/sort.h>
#include <linux/bsearch.h>
P
Paolo Bonzini 已提交
54
#include <linux/io.h>
55
#include <linux/lockdep.h>
A
Avi Kivity 已提交
56

A
Avi Kivity 已提交
57
#include <asm/processor.h>
58
#include <asm/ioctl.h>
59
#include <linux/uaccess.h>
60
#include <asm/pgtable.h>
A
Avi Kivity 已提交
61

62
#include "coalesced_mmio.h"
63
#include "async_pf.h"
P
Paolo Bonzini 已提交
64
#include "vfio.h"
65

66 67 68
#define CREATE_TRACE_POINTS
#include <trace/events/kvm.h>

69 70 71
/* Worst case buffer size needed for holding an integer. */
#define ITOA_MAX_LEN 12

A
Avi Kivity 已提交
72 73 74
MODULE_AUTHOR("Qumranet");
MODULE_LICENSE("GPL");

75
/* Architectures should define their poll value according to the halt latency */
76
unsigned int halt_poll_ns = KVM_HALT_POLL_NS_DEFAULT;
77
module_param(halt_poll_ns, uint, 0644);
78
EXPORT_SYMBOL_GPL(halt_poll_ns);
79

W
Wanpeng Li 已提交
80
/* Default doubles per-vcpu halt_poll_ns. */
81
unsigned int halt_poll_ns_grow = 2;
82
module_param(halt_poll_ns_grow, uint, 0644);
83
EXPORT_SYMBOL_GPL(halt_poll_ns_grow);
W
Wanpeng Li 已提交
84

85 86 87 88 89
/* The start value to grow halt_poll_ns from */
unsigned int halt_poll_ns_grow_start = 10000; /* 10us */
module_param(halt_poll_ns_grow_start, uint, 0644);
EXPORT_SYMBOL_GPL(halt_poll_ns_grow_start);

W
Wanpeng Li 已提交
90
/* Default resets per-vcpu halt_poll_ns . */
91
unsigned int halt_poll_ns_shrink;
92
module_param(halt_poll_ns_shrink, uint, 0644);
93
EXPORT_SYMBOL_GPL(halt_poll_ns_shrink);
W
Wanpeng Li 已提交
94

95 96 97
/*
 * Ordering of locks:
 *
98
 *	kvm->lock --> kvm->slots_lock --> kvm->irq_lock
99 100
 */

101
DEFINE_SPINLOCK(kvm_lock);
102
static DEFINE_RAW_SPINLOCK(kvm_count_lock);
103
LIST_HEAD(vm_list);
104

105
static cpumask_var_t cpus_hardware_enabled;
106
static int kvm_usage_count;
107
static atomic_t hardware_enable_failed;
108

109 110
struct kmem_cache *kvm_vcpu_cache;
EXPORT_SYMBOL_GPL(kvm_vcpu_cache);
A
Avi Kivity 已提交
111

112 113
static __read_mostly struct preempt_ops kvm_preempt_ops;

114
struct dentry *kvm_debugfs_dir;
115
EXPORT_SYMBOL_GPL(kvm_debugfs_dir);
A
Avi Kivity 已提交
116

117 118 119
static int kvm_debugfs_num_entries;
static const struct file_operations *stat_fops_per_vm[];

A
Avi Kivity 已提交
120 121
static long kvm_vcpu_ioctl(struct file *file, unsigned int ioctl,
			   unsigned long arg);
122
#ifdef CONFIG_KVM_COMPAT
123 124
static long kvm_vcpu_compat_ioctl(struct file *file, unsigned int ioctl,
				  unsigned long arg);
125 126 127 128 129
#define KVM_COMPAT(c)	.compat_ioctl	= (c)
#else
static long kvm_no_compat_ioctl(struct file *file, unsigned int ioctl,
				unsigned long arg) { return -EINVAL; }
#define KVM_COMPAT(c)	.compat_ioctl	= kvm_no_compat_ioctl
130
#endif
131 132
static int hardware_enable_all(void);
static void hardware_disable_all(void);
A
Avi Kivity 已提交
133

M
Marcelo Tosatti 已提交
134
static void kvm_io_bus_destroy(struct kvm_io_bus *bus);
135

136
static void mark_page_dirty_in_slot(struct kvm_memory_slot *memslot, gfn_t gfn);
M
Marcelo Tosatti 已提交
137

138
__visible bool kvm_rebooting;
139
EXPORT_SYMBOL_GPL(kvm_rebooting);
140

141 142
static bool largepages_enabled = true;

143 144 145 146 147 148
#define KVM_EVENT_CREATE_VM 0
#define KVM_EVENT_DESTROY_VM 1
static void kvm_uevent_notify_change(unsigned int type, struct kvm *kvm);
static unsigned long long kvm_createvm_count;
static unsigned long long kvm_active_vms;

149 150
__weak int kvm_arch_mmu_notifier_invalidate_range(struct kvm *kvm,
		unsigned long start, unsigned long end, bool blockable)
151
{
152
	return 0;
153 154
}

D
Dan Williams 已提交
155
bool kvm_is_reserved_pfn(kvm_pfn_t pfn)
B
Ben-Ami Yassour 已提交
156
{
157
	if (pfn_valid(pfn))
158
		return PageReserved(pfn_to_page(pfn));
B
Ben-Ami Yassour 已提交
159 160 161 162

	return true;
}

A
Avi Kivity 已提交
163 164 165
/*
 * Switches to specified vcpu, until a matching vcpu_put()
 */
166
void vcpu_load(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
167
{
168
	int cpu = get_cpu();
169
	preempt_notifier_register(&vcpu->preempt_notifier);
170
	kvm_arch_vcpu_load(vcpu, cpu);
171
	put_cpu();
A
Avi Kivity 已提交
172
}
173
EXPORT_SYMBOL_GPL(vcpu_load);
A
Avi Kivity 已提交
174

175
void vcpu_put(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
176
{
177
	preempt_disable();
178
	kvm_arch_vcpu_put(vcpu);
179 180
	preempt_notifier_unregister(&vcpu->preempt_notifier);
	preempt_enable();
A
Avi Kivity 已提交
181
}
182
EXPORT_SYMBOL_GPL(vcpu_put);
A
Avi Kivity 已提交
183

184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201
/* TODO: merge with kvm_arch_vcpu_should_kick */
static bool kvm_request_needs_ipi(struct kvm_vcpu *vcpu, unsigned req)
{
	int mode = kvm_vcpu_exiting_guest_mode(vcpu);

	/*
	 * We need to wait for the VCPU to reenable interrupts and get out of
	 * READING_SHADOW_PAGE_TABLES mode.
	 */
	if (req & KVM_REQUEST_WAIT)
		return mode != OUTSIDE_GUEST_MODE;

	/*
	 * Need to kick a running VCPU, but otherwise there is nothing to do.
	 */
	return mode == IN_GUEST_MODE;
}

202 203 204 205
static void ack_flush(void *_completed)
{
}

206 207 208 209 210 211 212 213 214 215 216 217
static inline bool kvm_kick_many_cpus(const struct cpumask *cpus, bool wait)
{
	if (unlikely(!cpus))
		cpus = cpu_online_mask;

	if (cpumask_empty(cpus))
		return false;

	smp_call_function_many(cpus, ack_flush, NULL, wait);
	return true;
}

218 219
bool kvm_make_vcpus_request_mask(struct kvm *kvm, unsigned int req,
				 unsigned long *vcpu_bitmap, cpumask_var_t tmp)
220
{
221
	int i, cpu, me;
222
	struct kvm_vcpu *vcpu;
223
	bool called;
224

225
	me = get_cpu();
226

227
	kvm_for_each_vcpu(i, vcpu, kvm) {
228
		if (vcpu_bitmap && !test_bit(i, vcpu_bitmap))
229 230
			continue;

231
		kvm_make_request(req, vcpu);
232
		cpu = vcpu->cpu;
233

234 235
		if (!(req & KVM_REQUEST_NO_WAKEUP) && kvm_vcpu_wake_up(vcpu))
			continue;
236

237
		if (tmp != NULL && cpu != -1 && cpu != me &&
238
		    kvm_request_needs_ipi(vcpu, req))
239
			__cpumask_set_cpu(cpu, tmp);
240
	}
241 242

	called = kvm_kick_many_cpus(tmp, !!(req & KVM_REQUEST_WAIT));
243
	put_cpu();
244 245 246 247 248 249 250 251 252 253 254

	return called;
}

bool kvm_make_all_cpus_request(struct kvm *kvm, unsigned int req)
{
	cpumask_var_t cpus;
	bool called;

	zalloc_cpumask_var(&cpus, GFP_ATOMIC);

255
	called = kvm_make_vcpus_request_mask(kvm, req, NULL, cpus);
256

257
	free_cpumask_var(cpus);
258
	return called;
259 260
}

261
#ifndef CONFIG_HAVE_KVM_ARCH_TLB_FLUSH_ALL
262
void kvm_flush_remote_tlbs(struct kvm *kvm)
263
{
264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280
	/*
	 * 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.
	 */
281 282
	if (!kvm_arch_flush_remote_tlb(kvm)
	    || kvm_make_all_cpus_request(kvm, KVM_REQ_TLB_FLUSH))
283
		++kvm->stat.remote_tlb_flush;
284
	cmpxchg(&kvm->tlbs_dirty, dirty_count, 0);
285
}
286
EXPORT_SYMBOL_GPL(kvm_flush_remote_tlbs);
287
#endif
288

289 290
void kvm_reload_remote_mmus(struct kvm *kvm)
{
291
	kvm_make_all_cpus_request(kvm, KVM_REQ_MMU_RELOAD);
292
}
293

R
Rusty Russell 已提交
294 295 296 297 298 299 300 301 302
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;
303
	vcpu->pid = NULL;
304
	init_swait_queue_head(&vcpu->wq);
305
	kvm_async_pf_vcpu_init(vcpu);
R
Rusty Russell 已提交
306

307 308 309
	vcpu->pre_pcpu = -1;
	INIT_LIST_HEAD(&vcpu->blocked_vcpu_list);

R
Rusty Russell 已提交
310 311 312 313 314 315 316
	page = alloc_page(GFP_KERNEL | __GFP_ZERO);
	if (!page) {
		r = -ENOMEM;
		goto fail;
	}
	vcpu->run = page_address(page);

317 318
	kvm_vcpu_set_in_spin_loop(vcpu, false);
	kvm_vcpu_set_dy_eligible(vcpu, false);
319
	vcpu->preempted = false;
320

321
	r = kvm_arch_vcpu_init(vcpu);
R
Rusty Russell 已提交
322
	if (r < 0)
323
		goto fail_free_run;
R
Rusty Russell 已提交
324 325 326 327 328
	return 0;

fail_free_run:
	free_page((unsigned long)vcpu->run);
fail:
329
	return r;
R
Rusty Russell 已提交
330 331 332 333 334
}
EXPORT_SYMBOL_GPL(kvm_vcpu_init);

void kvm_vcpu_uninit(struct kvm_vcpu *vcpu)
{
335 336 337 338 339 340
	/*
	 * no need for rcu_read_lock as VCPU_RUN is the only place that
	 * will change the vcpu->pid pointer and on uninit all file
	 * descriptors are already gone.
	 */
	put_pid(rcu_dereference_protected(vcpu->pid, 1));
341
	kvm_arch_vcpu_uninit(vcpu);
R
Rusty Russell 已提交
342 343 344 345
	free_page((unsigned long)vcpu->run);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_uninit);

346 347 348 349 350 351
#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);
}

352 353 354 355 356 357
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);
358
	int idx;
359

360
	idx = srcu_read_lock(&kvm->srcu);
361 362
	spin_lock(&kvm->mmu_lock);
	kvm->mmu_notifier_seq++;
363 364 365 366

	if (kvm_set_spte_hva(kvm, address, pte))
		kvm_flush_remote_tlbs(kvm);

367
	spin_unlock(&kvm->mmu_lock);
368
	srcu_read_unlock(&kvm->srcu, idx);
369 370
}

371
static int kvm_mmu_notifier_invalidate_range_start(struct mmu_notifier *mn,
372
					const struct mmu_notifier_range *range)
373 374
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
375
	int need_tlb_flush = 0, idx;
376
	int ret;
377

378
	idx = srcu_read_lock(&kvm->srcu);
379 380 381 382 383 384 385
	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++;
386
	need_tlb_flush = kvm_unmap_hva_range(kvm, range->start, range->end);
387
	need_tlb_flush |= kvm->tlbs_dirty;
388 389 390
	/* we've to flush the tlb before the pages can be freed */
	if (need_tlb_flush)
		kvm_flush_remote_tlbs(kvm);
391 392

	spin_unlock(&kvm->mmu_lock);
393

394
	ret = kvm_arch_mmu_notifier_invalidate_range(kvm, range->start,
395 396
					range->end,
					mmu_notifier_range_blockable(range));
397

398
	srcu_read_unlock(&kvm->srcu, idx);
399 400

	return ret;
401 402 403
}

static void kvm_mmu_notifier_invalidate_range_end(struct mmu_notifier *mn,
404
					const struct mmu_notifier_range *range)
405 406 407 408 409 410 411 412 413 414
{
	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++;
415
	smp_wmb();
416 417
	/*
	 * The above sequence increase must be visible before the
418 419
	 * below count decrease, which is ensured by the smp_wmb above
	 * in conjunction with the smp_rmb in mmu_notifier_retry().
420 421 422 423 424 425 426 427 428
	 */
	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
429 430
					      unsigned long start,
					      unsigned long end)
431 432
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
433
	int young, idx;
434

435
	idx = srcu_read_lock(&kvm->srcu);
436 437
	spin_lock(&kvm->mmu_lock);

A
Andres Lagar-Cavilla 已提交
438
	young = kvm_age_hva(kvm, start, end);
439 440 441
	if (young)
		kvm_flush_remote_tlbs(kvm);

442 443 444
	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);

445 446 447
	return young;
}

448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477
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
478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493
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;
}

494 495 496 497
static void kvm_mmu_notifier_release(struct mmu_notifier *mn,
				     struct mm_struct *mm)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
498 499 500
	int idx;

	idx = srcu_read_lock(&kvm->srcu);
501
	kvm_arch_flush_shadow_all(kvm);
502
	srcu_read_unlock(&kvm->srcu, idx);
503 504
}

505 506 507 508
static const struct mmu_notifier_ops kvm_mmu_notifier_ops = {
	.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,
509
	.clear_young		= kvm_mmu_notifier_clear_young,
A
Andrea Arcangeli 已提交
510
	.test_young		= kvm_mmu_notifier_test_young,
511
	.change_pte		= kvm_mmu_notifier_change_pte,
512
	.release		= kvm_mmu_notifier_release,
513
};
514 515 516 517 518 519 520 521 522 523 524 525 526 527

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;
}

528 529
#endif /* CONFIG_MMU_NOTIFIER && KVM_ARCH_WANT_MMU_NOTIFIER */

530
static struct kvm_memslots *kvm_alloc_memslots(void)
531 532
{
	int i;
533
	struct kvm_memslots *slots;
534

535
	slots = kvzalloc(sizeof(struct kvm_memslots), GFP_KERNEL_ACCOUNT);
536 537 538
	if (!slots)
		return NULL;

539
	for (i = 0; i < KVM_MEM_SLOTS_NUM; i++)
540
		slots->id_to_index[i] = slots->memslots[i].id = i;
541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578

	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);
579 580
}

581 582 583 584 585 586 587 588 589
static void kvm_destroy_vm_debugfs(struct kvm *kvm)
{
	int i;

	if (!kvm->debugfs_dentry)
		return;

	debugfs_remove_recursive(kvm->debugfs_dentry);

590 591 592 593 594
	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);
	}
595 596 597 598 599 600 601 602 603 604 605 606
}

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);
607
	kvm->debugfs_dentry = debugfs_create_dir(dir_name, kvm_debugfs_dir);
608 609 610

	kvm->debugfs_stat_data = kcalloc(kvm_debugfs_num_entries,
					 sizeof(*kvm->debugfs_stat_data),
611
					 GFP_KERNEL_ACCOUNT);
612 613 614 615
	if (!kvm->debugfs_stat_data)
		return -ENOMEM;

	for (p = debugfs_entries; p->name; p++) {
616
		stat_data = kzalloc(sizeof(*stat_data), GFP_KERNEL_ACCOUNT);
617 618 619 620 621 622
		if (!stat_data)
			return -ENOMEM;

		stat_data->kvm = kvm;
		stat_data->offset = p->offset;
		kvm->debugfs_stat_data[p - debugfs_entries] = stat_data;
623 624
		debugfs_create_file(p->name, 0644, kvm->debugfs_dentry,
				    stat_data, stat_fops_per_vm[p->kind]);
625 626 627 628
	}
	return 0;
}

629
static struct kvm *kvm_create_vm(unsigned long type)
A
Avi Kivity 已提交
630
{
631 632
	int r, i;
	struct kvm *kvm = kvm_arch_alloc_vm();
A
Avi Kivity 已提交
633

634 635 636
	if (!kvm)
		return ERR_PTR(-ENOMEM);

637
	spin_lock_init(&kvm->mmu_lock);
V
Vegard Nossum 已提交
638
	mmgrab(current->mm);
639 640 641 642 643
	kvm->mm = current->mm;
	kvm_eventfd_init(kvm);
	mutex_init(&kvm->lock);
	mutex_init(&kvm->irq_lock);
	mutex_init(&kvm->slots_lock);
644
	refcount_set(&kvm->users_count, 1);
645 646
	INIT_LIST_HEAD(&kvm->devices);

647
	r = kvm_arch_init_vm(kvm, type);
648
	if (r)
649
		goto out_err_no_disable;
650 651 652

	r = hardware_enable_all();
	if (r)
653
		goto out_err_no_disable;
654

655
#ifdef CONFIG_HAVE_KVM_IRQFD
656
	INIT_HLIST_HEAD(&kvm->irq_ack_notifier_list);
657
#endif
A
Avi Kivity 已提交
658

659 660
	BUILD_BUG_ON(KVM_MEM_SLOTS_NUM > SHRT_MAX);

661
	r = -ENOMEM;
662
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++) {
663 664
		struct kvm_memslots *slots = kvm_alloc_memslots();
		if (!slots)
665
			goto out_err_no_srcu;
666
		/* Generations must be different for each address space. */
667
		slots->generation = i;
668
		rcu_assign_pointer(kvm->memslots[i], slots);
669
	}
670

671
	if (init_srcu_struct(&kvm->srcu))
672 673 674
		goto out_err_no_srcu;
	if (init_srcu_struct(&kvm->irq_srcu))
		goto out_err_no_irq_srcu;
M
Marcelo Tosatti 已提交
675
	for (i = 0; i < KVM_NR_BUSES; i++) {
676
		rcu_assign_pointer(kvm->buses[i],
677
			kzalloc(sizeof(struct kvm_io_bus), GFP_KERNEL_ACCOUNT));
678
		if (!kvm->buses[i])
M
Marcelo Tosatti 已提交
679 680
			goto out_err;
	}
681

682 683 684 685
	r = kvm_init_mmu_notifier(kvm);
	if (r)
		goto out_err;

686
	spin_lock(&kvm_lock);
687
	list_add(&kvm->vm_list, &vm_list);
688
	spin_unlock(&kvm_lock);
689

690 691
	preempt_notifier_inc();

692
	return kvm;
693 694

out_err:
695 696
	cleanup_srcu_struct(&kvm->irq_srcu);
out_err_no_irq_srcu:
697
	cleanup_srcu_struct(&kvm->srcu);
698
out_err_no_srcu:
699
	hardware_disable_all();
700
out_err_no_disable:
701
	refcount_set(&kvm->users_count, 0);
M
Marcelo Tosatti 已提交
702
	for (i = 0; i < KVM_NR_BUSES; i++)
703
		kfree(kvm_get_bus(kvm, i));
704
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++)
705
		kvm_free_memslots(kvm, __kvm_memslots(kvm, i));
706
	kvm_arch_free_vm(kvm);
707
	mmdrop(current->mm);
708
	return ERR_PTR(r);
709 710
}

711 712
static void kvm_destroy_devices(struct kvm *kvm)
{
G
Geliang Tang 已提交
713
	struct kvm_device *dev, *tmp;
714

715 716 717 718 719
	/*
	 * 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
720 721
	list_for_each_entry_safe(dev, tmp, &kvm->devices, vm_node) {
		list_del(&dev->vm_node);
722 723 724 725
		dev->ops->destroy(dev);
	}
}

726 727
static void kvm_destroy_vm(struct kvm *kvm)
{
M
Marcelo Tosatti 已提交
728
	int i;
729 730
	struct mm_struct *mm = kvm->mm;

731
	kvm_uevent_notify_change(KVM_EVENT_DESTROY_VM, kvm);
732
	kvm_destroy_vm_debugfs(kvm);
733
	kvm_arch_sync_events(kvm);
734
	spin_lock(&kvm_lock);
735
	list_del(&kvm->vm_list);
736
	spin_unlock(&kvm_lock);
737
	kvm_free_irq_routing(kvm);
738
	for (i = 0; i < KVM_NR_BUSES; i++) {
739
		struct kvm_io_bus *bus = kvm_get_bus(kvm, i);
740 741 742

		if (bus)
			kvm_io_bus_destroy(bus);
743 744
		kvm->buses[i] = NULL;
	}
745
	kvm_coalesced_mmio_free(kvm);
746 747
#if defined(CONFIG_MMU_NOTIFIER) && defined(KVM_ARCH_WANT_MMU_NOTIFIER)
	mmu_notifier_unregister(&kvm->mmu_notifier, kvm->mm);
748
#else
749
	kvm_arch_flush_shadow_all(kvm);
750
#endif
751
	kvm_arch_destroy_vm(kvm);
752
	kvm_destroy_devices(kvm);
753
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++)
754
		kvm_free_memslots(kvm, __kvm_memslots(kvm, i));
755
	cleanup_srcu_struct(&kvm->irq_srcu);
756 757
	cleanup_srcu_struct(&kvm->srcu);
	kvm_arch_free_vm(kvm);
758
	preempt_notifier_dec();
759
	hardware_disable_all();
760
	mmdrop(mm);
761 762
}

I
Izik Eidus 已提交
763 764
void kvm_get_kvm(struct kvm *kvm)
{
765
	refcount_inc(&kvm->users_count);
I
Izik Eidus 已提交
766 767 768 769 770
}
EXPORT_SYMBOL_GPL(kvm_get_kvm);

void kvm_put_kvm(struct kvm *kvm)
{
771
	if (refcount_dec_and_test(&kvm->users_count))
I
Izik Eidus 已提交
772 773 774 775 776
		kvm_destroy_vm(kvm);
}
EXPORT_SYMBOL_GPL(kvm_put_kvm);


777 778 779 780
static int kvm_vm_release(struct inode *inode, struct file *filp)
{
	struct kvm *kvm = filp->private_data;

G
Gregory Haskins 已提交
781 782
	kvm_irqfd_release(kvm);

I
Izik Eidus 已提交
783
	kvm_put_kvm(kvm);
A
Avi Kivity 已提交
784 785 786
	return 0;
}

787 788
/*
 * Allocation size is twice as large as the actual dirty bitmap size.
789
 * See x86's kvm_vm_ioctl_get_dirty_log() why this is needed.
790
 */
791 792
static int kvm_create_dirty_bitmap(struct kvm_memory_slot *memslot)
{
793
	unsigned long dirty_bytes = 2 * kvm_dirty_bitmap_bytes(memslot);
794

795
	memslot->dirty_bitmap = kvzalloc(dirty_bytes, GFP_KERNEL_ACCOUNT);
796 797 798 799 800 801
	if (!memslot->dirty_bitmap)
		return -ENOMEM;

	return 0;
}

802
/*
803 804 805 806
 * 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.
807
 */
808
static void update_memslots(struct kvm_memslots *slots,
809 810
			    struct kvm_memory_slot *new,
			    enum kvm_mr_change change)
811
{
812 813
	int id = new->id;
	int i = slots->id_to_index[id];
814
	struct kvm_memory_slot *mslots = slots->memslots;
815

816
	WARN_ON(mslots[i].id != id);
817 818 819 820 821 822 823 824 825 826 827
	switch (change) {
	case KVM_MR_CREATE:
		slots->used_slots++;
		WARN_ON(mslots[i].npages || !new->npages);
		break;
	case KVM_MR_DELETE:
		slots->used_slots--;
		WARN_ON(new->npages || !mslots[i].npages);
		break;
	default:
		break;
828
	}
829

830
	while (i < KVM_MEM_SLOTS_NUM - 1 &&
831 832 833
	       new->base_gfn <= mslots[i + 1].base_gfn) {
		if (!mslots[i + 1].npages)
			break;
834 835 836 837
		mslots[i] = mslots[i + 1];
		slots->id_to_index[mslots[i].id] = i;
		i++;
	}
838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854

	/*
	 * 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--;
		}
855 856
	} else
		WARN_ON_ONCE(i != slots->used_slots);
857

858 859
	mslots[i] = *new;
	slots->id_to_index[mslots[i].id] = i;
860 861
}

862
static int check_memory_region_flags(const struct kvm_userspace_memory_region *mem)
863
{
X
Xiao Guangrong 已提交
864 865
	u32 valid_flags = KVM_MEM_LOG_DIRTY_PAGES;

866
#ifdef __KVM_HAVE_READONLY_MEM
X
Xiao Guangrong 已提交
867 868 869 870
	valid_flags |= KVM_MEM_READONLY;
#endif

	if (mem->flags & ~valid_flags)
871 872 873 874 875
		return -EINVAL;

	return 0;
}

876
static struct kvm_memslots *install_new_memslots(struct kvm *kvm,
877
		int as_id, struct kvm_memslots *slots)
878
{
879
	struct kvm_memslots *old_memslots = __kvm_memslots(kvm, as_id);
880
	u64 gen = old_memslots->generation;
881

882 883
	WARN_ON(gen & KVM_MEMSLOT_GEN_UPDATE_IN_PROGRESS);
	slots->generation = gen | KVM_MEMSLOT_GEN_UPDATE_IN_PROGRESS;
884

885
	rcu_assign_pointer(kvm->memslots[as_id], slots);
886
	synchronize_srcu_expedited(&kvm->srcu);
887

888
	/*
889 890 891 892 893 894 895 896
	 * Increment the new memslot generation a second time, dropping the
	 * update in-progress flag and incrementing then generation based on
	 * the number of address spaces.  This provides a unique and easily
	 * identifiable generation number while the memslots are in flux.
	 */
	gen = slots->generation & ~KVM_MEMSLOT_GEN_UPDATE_IN_PROGRESS;

	/*
897 898 899
	 * 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
900 901
	 * space 0 will use generations 0, 2, 4, ... while address space 1 will
	 * use generations 1, 3, 5, ...
902
	 */
903
	gen += KVM_ADDRESS_SPACE_NUM;
904

905
	kvm_arch_memslots_updated(kvm, gen);
906

907
	slots->generation = gen;
908 909

	return old_memslots;
910 911
}

A
Avi Kivity 已提交
912 913 914 915 916
/*
 * Allocate some memory and give it an address in the guest physical address
 * space.
 *
 * Discontiguous memory is allowed, mostly for framebuffers.
917
 *
918
 * Must be called holding kvm->slots_lock for write.
A
Avi Kivity 已提交
919
 */
920
int __kvm_set_memory_region(struct kvm *kvm,
921
			    const struct kvm_userspace_memory_region *mem)
A
Avi Kivity 已提交
922
{
923
	int r;
A
Avi Kivity 已提交
924
	gfn_t base_gfn;
925
	unsigned long npages;
926
	struct kvm_memory_slot *slot;
A
Avi Kivity 已提交
927
	struct kvm_memory_slot old, new;
928
	struct kvm_memslots *slots = NULL, *old_memslots;
929
	int as_id, id;
930
	enum kvm_mr_change change;
A
Avi Kivity 已提交
931

932 933 934 935
	r = check_memory_region_flags(mem);
	if (r)
		goto out;

A
Avi Kivity 已提交
936
	r = -EINVAL;
937 938 939
	as_id = mem->slot >> 16;
	id = (u16)mem->slot;

A
Avi Kivity 已提交
940 941 942 943 944
	/* General sanity checks */
	if (mem->memory_size & (PAGE_SIZE - 1))
		goto out;
	if (mem->guest_phys_addr & (PAGE_SIZE - 1))
		goto out;
945
	/* We can read the guest memory with __xxx_user() later on. */
946
	if ((id < KVM_USER_MEM_SLOTS) &&
947
	    ((mem->userspace_addr & (PAGE_SIZE - 1)) ||
948
	     !access_ok((void __user *)(unsigned long)mem->userspace_addr,
949
			mem->memory_size)))
950
		goto out;
951
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_MEM_SLOTS_NUM)
A
Avi Kivity 已提交
952 953 954 955
		goto out;
	if (mem->guest_phys_addr + mem->memory_size < mem->guest_phys_addr)
		goto out;

956
	slot = id_to_memslot(__kvm_memslots(kvm, as_id), id);
A
Avi Kivity 已提交
957 958 959
	base_gfn = mem->guest_phys_addr >> PAGE_SHIFT;
	npages = mem->memory_size >> PAGE_SHIFT;

960 961 962
	if (npages > KVM_MEM_MAX_NR_PAGES)
		goto out;

963
	new = old = *slot;
A
Avi Kivity 已提交
964

965
	new.id = id;
A
Avi Kivity 已提交
966 967 968 969
	new.base_gfn = base_gfn;
	new.npages = npages;
	new.flags = mem->flags;

970 971 972 973 974
	if (npages) {
		if (!old.npages)
			change = KVM_MR_CREATE;
		else { /* Modify an existing slot. */
			if ((mem->userspace_addr != old.userspace_addr) ||
975 976
			    (npages != old.npages) ||
			    ((new.flags ^ old.flags) & KVM_MEM_READONLY))
977 978 979 980 981 982 983 984 985 986 987
				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;
			}
		}
988 989 990 991
	} else {
		if (!old.npages)
			goto out;

992
		change = KVM_MR_DELETE;
993 994 995
		new.base_gfn = 0;
		new.flags = 0;
	}
A
Avi Kivity 已提交
996

997
	if ((change == KVM_MR_CREATE) || (change == KVM_MR_MOVE)) {
998 999
		/* Check for overlaps */
		r = -EEXIST;
1000
		kvm_for_each_memslot(slot, __kvm_memslots(kvm, as_id)) {
1001
			if (slot->id == id)
1002 1003 1004 1005 1006
				continue;
			if (!((base_gfn + npages <= slot->base_gfn) ||
			      (base_gfn >= slot->base_gfn + slot->npages)))
				goto out;
		}
A
Avi Kivity 已提交
1007 1008 1009 1010
	}

	/* Free page dirty bitmap if unneeded */
	if (!(new.flags & KVM_MEM_LOG_DIRTY_PAGES))
A
Al Viro 已提交
1011
		new.dirty_bitmap = NULL;
A
Avi Kivity 已提交
1012 1013

	r = -ENOMEM;
1014
	if (change == KVM_MR_CREATE) {
1015
		new.userspace_addr = mem->userspace_addr;
1016

1017
		if (kvm_arch_create_memslot(kvm, &new, npages))
1018
			goto out_free;
A
Avi Kivity 已提交
1019
	}
1020

A
Avi Kivity 已提交
1021 1022
	/* Allocate page dirty bitmap if needed */
	if ((new.flags & KVM_MEM_LOG_DIRTY_PAGES) && !new.dirty_bitmap) {
1023
		if (kvm_create_dirty_bitmap(&new) < 0)
1024
			goto out_free;
A
Avi Kivity 已提交
1025 1026
	}

1027
	slots = kvzalloc(sizeof(struct kvm_memslots), GFP_KERNEL_ACCOUNT);
1028 1029
	if (!slots)
		goto out_free;
1030
	memcpy(slots, __kvm_memslots(kvm, as_id), sizeof(struct kvm_memslots));
1031

1032
	if ((change == KVM_MR_DELETE) || (change == KVM_MR_MOVE)) {
1033
		slot = id_to_memslot(slots, id);
1034 1035
		slot->flags |= KVM_MEMSLOT_INVALID;

1036
		old_memslots = install_new_memslots(kvm, as_id, slots);
1037

1038 1039
		/* From this point no new shadow pages pointing to a deleted,
		 * or moved, memslot will be created.
1040 1041
		 *
		 * validation of sp->gfn happens in:
1042 1043
		 *	- gfn_to_hva (kvm_read_guest, gfn_to_pfn)
		 *	- kvm_is_visible_gfn (mmu_check_roots)
1044
		 */
1045
		kvm_arch_flush_shadow_memslot(kvm, slot);
1046 1047 1048 1049 1050 1051

		/*
		 * 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.
		 */
1052
		slots = old_memslots;
1053
	}
1054

1055
	r = kvm_arch_prepare_memory_region(kvm, &new, mem, change);
1056
	if (r)
1057
		goto out_slots;
1058

1059
	/* actual memory is freed via old in kvm_free_memslot below */
1060
	if (change == KVM_MR_DELETE) {
1061
		new.dirty_bitmap = NULL;
1062
		memset(&new.arch, 0, sizeof(new.arch));
1063 1064
	}

1065
	update_memslots(slots, &new, change);
1066
	old_memslots = install_new_memslots(kvm, as_id, slots);
1067

1068
	kvm_arch_commit_memory_region(kvm, mem, &old, &new, change);
1069

1070
	kvm_free_memslot(kvm, &old, &new);
1071
	kvfree(old_memslots);
A
Avi Kivity 已提交
1072 1073
	return 0;

1074
out_slots:
1075
	kvfree(slots);
1076
out_free:
1077
	kvm_free_memslot(kvm, &new, &old);
A
Avi Kivity 已提交
1078 1079
out:
	return r;
1080
}
1081 1082 1083
EXPORT_SYMBOL_GPL(__kvm_set_memory_region);

int kvm_set_memory_region(struct kvm *kvm,
1084
			  const struct kvm_userspace_memory_region *mem)
1085 1086 1087
{
	int r;

1088
	mutex_lock(&kvm->slots_lock);
1089
	r = __kvm_set_memory_region(kvm, mem);
1090
	mutex_unlock(&kvm->slots_lock);
1091 1092
	return r;
}
1093 1094
EXPORT_SYMBOL_GPL(kvm_set_memory_region);

1095 1096
static int kvm_vm_ioctl_set_memory_region(struct kvm *kvm,
					  struct kvm_userspace_memory_region *mem)
1097
{
1098
	if ((u16)mem->slot >= KVM_USER_MEM_SLOTS)
1099
		return -EINVAL;
1100

1101
	return kvm_set_memory_region(kvm, mem);
A
Avi Kivity 已提交
1102 1103
}

1104 1105
int kvm_get_dirty_log(struct kvm *kvm,
			struct kvm_dirty_log *log, int *is_dirty)
A
Avi Kivity 已提交
1106
{
1107
	struct kvm_memslots *slots;
A
Avi Kivity 已提交
1108
	struct kvm_memory_slot *memslot;
1109
	int i, as_id, id;
1110
	unsigned long n;
A
Avi Kivity 已提交
1111 1112
	unsigned long any = 0;

1113 1114 1115
	as_id = log->slot >> 16;
	id = (u16)log->slot;
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_USER_MEM_SLOTS)
1116
		return -EINVAL;
A
Avi Kivity 已提交
1117

1118 1119
	slots = __kvm_memslots(kvm, as_id);
	memslot = id_to_memslot(slots, id);
A
Avi Kivity 已提交
1120
	if (!memslot->dirty_bitmap)
1121
		return -ENOENT;
A
Avi Kivity 已提交
1122

1123
	n = kvm_dirty_bitmap_bytes(memslot);
A
Avi Kivity 已提交
1124

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

	if (copy_to_user(log->dirty_bitmap, memslot->dirty_bitmap, n))
1129
		return -EFAULT;
A
Avi Kivity 已提交
1130

1131 1132
	if (any)
		*is_dirty = 1;
1133
	return 0;
A
Avi Kivity 已提交
1134
}
1135
EXPORT_SYMBOL_GPL(kvm_get_dirty_log);
A
Avi Kivity 已提交
1136

1137 1138
#ifdef CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT
/**
J
Jiang Biao 已提交
1139
 * kvm_get_dirty_log_protect - get a snapshot of dirty pages
1140
 *	and reenable dirty page tracking for the corresponding pages.
1141 1142
 * @kvm:	pointer to kvm instance
 * @log:	slot id and address to which we copy the log
J
Jiang Biao 已提交
1143
 * @flush:	true if TLB flush is needed by caller
1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160
 *
 * 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,
1161
			struct kvm_dirty_log *log, bool *flush)
1162
{
1163
	struct kvm_memslots *slots;
1164
	struct kvm_memory_slot *memslot;
1165
	int i, as_id, id;
1166 1167 1168 1169
	unsigned long n;
	unsigned long *dirty_bitmap;
	unsigned long *dirty_bitmap_buffer;

1170 1171 1172
	as_id = log->slot >> 16;
	id = (u16)log->slot;
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_USER_MEM_SLOTS)
1173
		return -EINVAL;
1174

1175 1176
	slots = __kvm_memslots(kvm, as_id);
	memslot = id_to_memslot(slots, id);
1177 1178 1179

	dirty_bitmap = memslot->dirty_bitmap;
	if (!dirty_bitmap)
1180
		return -ENOENT;
1181 1182

	n = kvm_dirty_bitmap_bytes(memslot);
1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196
	*flush = false;
	if (kvm->manual_dirty_log_protect) {
		/*
		 * Unlike kvm_get_dirty_log, we always return false in *flush,
		 * because no flush is needed until KVM_CLEAR_DIRTY_LOG.  There
		 * is some code duplication between this function and
		 * kvm_get_dirty_log, but hopefully all architecture
		 * transition to kvm_get_dirty_log_protect and kvm_get_dirty_log
		 * can be eliminated.
		 */
		dirty_bitmap_buffer = dirty_bitmap;
	} else {
		dirty_bitmap_buffer = kvm_second_dirty_bitmap(memslot);
		memset(dirty_bitmap_buffer, 0, n);
1197

1198 1199 1200 1201
		spin_lock(&kvm->mmu_lock);
		for (i = 0; i < n / sizeof(long); i++) {
			unsigned long mask;
			gfn_t offset;
1202

1203 1204 1205 1206 1207 1208 1209
			if (!dirty_bitmap[i])
				continue;

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

1210 1211 1212
			offset = i * BITS_PER_LONG;
			kvm_arch_mmu_enable_log_dirty_pt_masked(kvm, memslot,
								offset, mask);
1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227
		}
		spin_unlock(&kvm->mmu_lock);
	}

	if (copy_to_user(log->dirty_bitmap, dirty_bitmap_buffer, n))
		return -EFAULT;
	return 0;
}
EXPORT_SYMBOL_GPL(kvm_get_dirty_log_protect);

/**
 * kvm_clear_dirty_log_protect - clear dirty bits in the bitmap
 *	and reenable dirty page tracking for the corresponding pages.
 * @kvm:	pointer to kvm instance
 * @log:	slot id and address from which to fetch the bitmap of dirty pages
J
Jiang Biao 已提交
1228
 * @flush:	true if TLB flush is needed by caller
1229 1230 1231 1232 1233 1234
 */
int kvm_clear_dirty_log_protect(struct kvm *kvm,
				struct kvm_clear_dirty_log *log, bool *flush)
{
	struct kvm_memslots *slots;
	struct kvm_memory_slot *memslot;
1235
	int as_id, id;
1236
	gfn_t offset;
1237
	unsigned long i, n;
1238 1239 1240 1241 1242 1243 1244 1245
	unsigned long *dirty_bitmap;
	unsigned long *dirty_bitmap_buffer;

	as_id = log->slot >> 16;
	id = (u16)log->slot;
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_USER_MEM_SLOTS)
		return -EINVAL;

1246
	if (log->first_page & 63)
1247 1248 1249 1250 1251 1252 1253 1254 1255
		return -EINVAL;

	slots = __kvm_memslots(kvm, as_id);
	memslot = id_to_memslot(slots, id);

	dirty_bitmap = memslot->dirty_bitmap;
	if (!dirty_bitmap)
		return -ENOENT;

1256
	n = ALIGN(log->num_pages, BITS_PER_LONG) / 8;
1257 1258

	if (log->first_page > memslot->npages ||
1259 1260 1261
	    log->num_pages > memslot->npages - log->first_page ||
	    (log->num_pages < memslot->npages - log->first_page && (log->num_pages & 63)))
	    return -EINVAL;
1262

1263
	*flush = false;
1264 1265 1266
	dirty_bitmap_buffer = kvm_second_dirty_bitmap(memslot);
	if (copy_from_user(dirty_bitmap_buffer, log->dirty_bitmap, n))
		return -EFAULT;
1267

1268
	spin_lock(&kvm->mmu_lock);
1269 1270
	for (offset = log->first_page, i = offset / BITS_PER_LONG,
		 n = DIV_ROUND_UP(log->num_pages, BITS_PER_LONG); n--;
1271 1272 1273 1274
	     i++, offset += BITS_PER_LONG) {
		unsigned long mask = *dirty_bitmap_buffer++;
		atomic_long_t *p = (atomic_long_t *) &dirty_bitmap[i];
		if (!mask)
1275 1276
			continue;

1277
		mask &= atomic_long_fetch_andnot(mask, p);
1278

1279 1280 1281 1282 1283 1284
		/*
		 * mask contains the bits that really have been cleared.  This
		 * never includes any bits beyond the length of the memslot (if
		 * the length is not aligned to 64 pages), therefore it is not
		 * a problem if userspace sets them in log->dirty_bitmap.
		*/
1285
		if (mask) {
1286
			*flush = true;
1287 1288 1289
			kvm_arch_mmu_enable_log_dirty_pt_masked(kvm, memslot,
								offset, mask);
		}
1290 1291
	}
	spin_unlock(&kvm->mmu_lock);
1292

1293
	return 0;
1294
}
1295
EXPORT_SYMBOL_GPL(kvm_clear_dirty_log_protect);
1296 1297
#endif

1298 1299 1300 1301 1302
bool kvm_largepages_enabled(void)
{
	return largepages_enabled;
}

1303 1304 1305 1306 1307 1308
void kvm_disable_largepages(void)
{
	largepages_enabled = false;
}
EXPORT_SYMBOL_GPL(kvm_disable_largepages);

1309 1310 1311 1312
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
1313
EXPORT_SYMBOL_GPL(gfn_to_memslot);
A
Avi Kivity 已提交
1314

1315 1316 1317 1318 1319
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);
}

1320
bool kvm_is_visible_gfn(struct kvm *kvm, gfn_t gfn)
1321
{
1322
	struct kvm_memory_slot *memslot = gfn_to_memslot(kvm, gfn);
1323

1324
	if (!memslot || memslot->id >= KVM_USER_MEM_SLOTS ||
1325
	      memslot->flags & KVM_MEMSLOT_INVALID)
1326
		return false;
1327

1328
	return true;
1329 1330 1331
}
EXPORT_SYMBOL_GPL(kvm_is_visible_gfn);

J
Joerg Roedel 已提交
1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355
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
1356 1357 1358 1359 1360 1361 1362
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
1363
{
1364
	if (!slot || slot->flags & KVM_MEMSLOT_INVALID)
X
Xiao Guangrong 已提交
1365
		return KVM_HVA_ERR_BAD;
1366

X
Xiao Guangrong 已提交
1367 1368
	if (memslot_is_readonly(slot) && write)
		return KVM_HVA_ERR_RO_BAD;
1369 1370 1371 1372

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

X
Xiao Guangrong 已提交
1373
	return __gfn_to_hva_memslot(slot, gfn);
I
Izik Eidus 已提交
1374
}
1375

X
Xiao Guangrong 已提交
1376 1377 1378 1379
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
1380
}
1381

X
Xiao Guangrong 已提交
1382
unsigned long gfn_to_hva_memslot(struct kvm_memory_slot *slot,
1383
					gfn_t gfn)
X
Xiao Guangrong 已提交
1384 1385 1386 1387 1388
{
	return gfn_to_hva_many(slot, gfn, NULL);
}
EXPORT_SYMBOL_GPL(gfn_to_hva_memslot);

1389 1390
unsigned long gfn_to_hva(struct kvm *kvm, gfn_t gfn)
{
1391
	return gfn_to_hva_many(gfn_to_memslot(kvm, gfn), gfn, NULL);
1392
}
1393
EXPORT_SYMBOL_GPL(gfn_to_hva);
I
Izik Eidus 已提交
1394

1395 1396 1397 1398 1399 1400
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);

1401
/*
1402 1403 1404 1405 1406 1407
 * Return the hva of a @gfn and the R/W attribute if possible.
 *
 * @slot: the kvm_memory_slot which contains @gfn
 * @gfn: the gfn to be translated
 * @writable: used to return the read/write attribute of the @slot if the hva
 * is valid and @writable is not NULL
1408
 */
1409 1410
unsigned long gfn_to_hva_memslot_prot(struct kvm_memory_slot *slot,
				      gfn_t gfn, bool *writable)
1411
{
1412 1413 1414
	unsigned long hva = __gfn_to_hva_many(slot, gfn, NULL, false);

	if (!kvm_is_error_hva(hva) && writable)
1415 1416
		*writable = !memslot_is_readonly(slot);

1417
	return hva;
1418 1419
}

1420 1421 1422 1423 1424 1425 1426
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);
}

1427 1428 1429 1430 1431 1432 1433
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);
}

1434 1435
static inline int check_user_page_hwpoison(unsigned long addr)
{
L
Lorenzo Stoakes 已提交
1436
	int rc, flags = FOLL_HWPOISON | FOLL_WRITE;
1437

L
Lorenzo Stoakes 已提交
1438
	rc = get_user_pages(addr, 1, flags, NULL, NULL);
1439 1440 1441
	return rc == -EHWPOISON;
}

X
Xiao Guangrong 已提交
1442
/*
1443 1444 1445
 * The fast path to get the writable pfn which will be stored in @pfn,
 * true indicates success, otherwise false is returned.  It's also the
 * only part that runs if we can are in atomic context.
X
Xiao Guangrong 已提交
1446
 */
1447 1448
static bool hva_to_pfn_fast(unsigned long addr, bool write_fault,
			    bool *writable, kvm_pfn_t *pfn)
A
Avi Kivity 已提交
1449
{
1450
	struct page *page[1];
X
Xiao Guangrong 已提交
1451
	int npages;
A
Avi Kivity 已提交
1452

1453 1454 1455 1456 1457 1458 1459
	/*
	 * 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;
1460

X
Xiao Guangrong 已提交
1461 1462 1463
	npages = __get_user_pages_fast(addr, 1, 1, page);
	if (npages == 1) {
		*pfn = page_to_pfn(page[0]);
1464

X
Xiao Guangrong 已提交
1465 1466 1467 1468
		if (writable)
			*writable = true;
		return true;
	}
1469

X
Xiao Guangrong 已提交
1470 1471
	return false;
}
1472

X
Xiao Guangrong 已提交
1473 1474 1475 1476 1477
/*
 * 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
1478
			   bool *writable, kvm_pfn_t *pfn)
X
Xiao Guangrong 已提交
1479
{
1480 1481
	unsigned int flags = FOLL_HWPOISON;
	struct page *page;
X
Xiao Guangrong 已提交
1482
	int npages = 0;
1483

X
Xiao Guangrong 已提交
1484 1485 1486 1487 1488
	might_sleep();

	if (writable)
		*writable = write_fault;

1489 1490 1491 1492
	if (write_fault)
		flags |= FOLL_WRITE;
	if (async)
		flags |= FOLL_NOWAIT;
1493

1494
	npages = get_user_pages_unlocked(addr, 1, &page, flags);
X
Xiao Guangrong 已提交
1495 1496 1497 1498
	if (npages != 1)
		return npages;

	/* map read fault as writable if possible */
1499
	if (unlikely(!write_fault) && writable) {
1500
		struct page *wpage;
X
Xiao Guangrong 已提交
1501

1502
		if (__get_user_pages_fast(addr, 1, 1, &wpage) == 1) {
X
Xiao Guangrong 已提交
1503
			*writable = true;
1504 1505
			put_page(page);
			page = wpage;
1506
		}
1507
	}
1508
	*pfn = page_to_pfn(page);
X
Xiao Guangrong 已提交
1509 1510
	return npages;
}
I
Izik Eidus 已提交
1511

X
Xiao Guangrong 已提交
1512 1513 1514 1515
static bool vma_is_valid(struct vm_area_struct *vma, bool write_fault)
{
	if (unlikely(!(vma->vm_flags & VM_READ)))
		return false;
1516

X
Xiao Guangrong 已提交
1517 1518
	if (write_fault && (unlikely(!(vma->vm_flags & VM_WRITE))))
		return false;
1519

X
Xiao Guangrong 已提交
1520 1521
	return true;
}
1522

1523 1524
static int hva_to_pfn_remapped(struct vm_area_struct *vma,
			       unsigned long addr, bool *async,
1525 1526
			       bool write_fault, bool *writable,
			       kvm_pfn_t *p_pfn)
1527
{
1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551
	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;

	}

1552 1553
	if (writable)
		*writable = true;
1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568

	/*
	 * 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;
1569 1570 1571
	return 0;
}

1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585
/*
 * 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
1586
static kvm_pfn_t hva_to_pfn(unsigned long addr, bool atomic, bool *async,
X
Xiao Guangrong 已提交
1587 1588 1589
			bool write_fault, bool *writable)
{
	struct vm_area_struct *vma;
D
Dan Williams 已提交
1590
	kvm_pfn_t pfn = 0;
1591
	int npages, r;
1592

X
Xiao Guangrong 已提交
1593 1594
	/* we can do it either atomically or asynchronously, not both */
	BUG_ON(atomic && async);
1595

1596
	if (hva_to_pfn_fast(addr, write_fault, writable, &pfn))
X
Xiao Guangrong 已提交
1597 1598 1599 1600 1601 1602 1603 1604
		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;
1605

X
Xiao Guangrong 已提交
1606 1607 1608 1609 1610 1611 1612
	down_read(&current->mm->mmap_sem);
	if (npages == -EHWPOISON ||
	      (!async && check_user_page_hwpoison(addr))) {
		pfn = KVM_PFN_ERR_HWPOISON;
		goto exit;
	}

1613
retry:
X
Xiao Guangrong 已提交
1614 1615 1616 1617
	vma = find_vma_intersection(current->mm, addr, addr + 1);

	if (vma == NULL)
		pfn = KVM_PFN_ERR_FAULT;
1618
	else if (vma->vm_flags & (VM_IO | VM_PFNMAP)) {
1619
		r = hva_to_pfn_remapped(vma, addr, async, write_fault, writable, &pfn);
1620 1621
		if (r == -EAGAIN)
			goto retry;
1622 1623
		if (r < 0)
			pfn = KVM_PFN_ERR_FAULT;
X
Xiao Guangrong 已提交
1624
	} else {
X
Xiao Guangrong 已提交
1625
		if (async && vma_is_valid(vma, write_fault))
X
Xiao Guangrong 已提交
1626 1627 1628 1629 1630
			*async = true;
		pfn = KVM_PFN_ERR_FAULT;
	}
exit:
	up_read(&current->mm->mmap_sem);
1631
	return pfn;
1632 1633
}

D
Dan Williams 已提交
1634 1635 1636
kvm_pfn_t __gfn_to_pfn_memslot(struct kvm_memory_slot *slot, gfn_t gfn,
			       bool atomic, bool *async, bool write_fault,
			       bool *writable)
1637
{
X
Xiao Guangrong 已提交
1638 1639
	unsigned long addr = __gfn_to_hva_many(slot, gfn, NULL, write_fault);

1640 1641 1642
	if (addr == KVM_HVA_ERR_RO_BAD) {
		if (writable)
			*writable = false;
X
Xiao Guangrong 已提交
1643
		return KVM_PFN_ERR_RO_FAULT;
1644
	}
X
Xiao Guangrong 已提交
1645

1646 1647 1648
	if (kvm_is_error_hva(addr)) {
		if (writable)
			*writable = false;
1649
		return KVM_PFN_NOSLOT;
1650
	}
X
Xiao Guangrong 已提交
1651 1652 1653 1654 1655 1656 1657 1658 1659

	/* 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);
1660
}
1661
EXPORT_SYMBOL_GPL(__gfn_to_pfn_memslot);
1662

D
Dan Williams 已提交
1663
kvm_pfn_t gfn_to_pfn_prot(struct kvm *kvm, gfn_t gfn, bool write_fault,
1664 1665
		      bool *writable)
{
P
Paolo Bonzini 已提交
1666 1667
	return __gfn_to_pfn_memslot(gfn_to_memslot(kvm, gfn), gfn, false, NULL,
				    write_fault, writable);
1668 1669 1670
}
EXPORT_SYMBOL_GPL(gfn_to_pfn_prot);

D
Dan Williams 已提交
1671
kvm_pfn_t gfn_to_pfn_memslot(struct kvm_memory_slot *slot, gfn_t gfn)
1672
{
X
Xiao Guangrong 已提交
1673
	return __gfn_to_pfn_memslot(slot, gfn, false, NULL, true, NULL);
1674
}
P
Paolo Bonzini 已提交
1675
EXPORT_SYMBOL_GPL(gfn_to_pfn_memslot);
1676

D
Dan Williams 已提交
1677
kvm_pfn_t gfn_to_pfn_memslot_atomic(struct kvm_memory_slot *slot, gfn_t gfn)
1678
{
X
Xiao Guangrong 已提交
1679
	return __gfn_to_pfn_memslot(slot, gfn, true, NULL, true, NULL);
1680
}
1681
EXPORT_SYMBOL_GPL(gfn_to_pfn_memslot_atomic);
1682

D
Dan Williams 已提交
1683
kvm_pfn_t gfn_to_pfn_atomic(struct kvm *kvm, gfn_t gfn)
P
Paolo Bonzini 已提交
1684 1685 1686 1687 1688
{
	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
1689
kvm_pfn_t kvm_vcpu_gfn_to_pfn_atomic(struct kvm_vcpu *vcpu, gfn_t gfn)
1690 1691 1692 1693 1694
{
	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
1695
kvm_pfn_t gfn_to_pfn(struct kvm *kvm, gfn_t gfn)
P
Paolo Bonzini 已提交
1696 1697 1698 1699 1700
{
	return gfn_to_pfn_memslot(gfn_to_memslot(kvm, gfn), gfn);
}
EXPORT_SYMBOL_GPL(gfn_to_pfn);

D
Dan Williams 已提交
1701
kvm_pfn_t kvm_vcpu_gfn_to_pfn(struct kvm_vcpu *vcpu, gfn_t gfn)
1702 1703 1704 1705 1706
{
	return gfn_to_pfn_memslot(kvm_vcpu_gfn_to_memslot(vcpu, gfn), gfn);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_gfn_to_pfn);

1707 1708
int gfn_to_page_many_atomic(struct kvm_memory_slot *slot, gfn_t gfn,
			    struct page **pages, int nr_pages)
1709 1710
{
	unsigned long addr;
1711
	gfn_t entry = 0;
1712

1713
	addr = gfn_to_hva_many(slot, gfn, &entry);
1714 1715 1716 1717 1718 1719 1720 1721 1722 1723
	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
1724
static struct page *kvm_pfn_to_page(kvm_pfn_t pfn)
X
Xiao Guangrong 已提交
1725
{
1726
	if (is_error_noslot_pfn(pfn))
1727
		return KVM_ERR_PTR_BAD_PAGE;
X
Xiao Guangrong 已提交
1728

1729
	if (kvm_is_reserved_pfn(pfn)) {
1730
		WARN_ON(1);
1731
		return KVM_ERR_PTR_BAD_PAGE;
1732
	}
X
Xiao Guangrong 已提交
1733 1734 1735 1736

	return pfn_to_page(pfn);
}

1737 1738
struct page *gfn_to_page(struct kvm *kvm, gfn_t gfn)
{
D
Dan Williams 已提交
1739
	kvm_pfn_t pfn;
1740 1741 1742

	pfn = gfn_to_pfn(kvm, gfn);

X
Xiao Guangrong 已提交
1743
	return kvm_pfn_to_page(pfn);
A
Avi Kivity 已提交
1744 1745 1746
}
EXPORT_SYMBOL_GPL(gfn_to_page);

1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810
static int __kvm_map_gfn(struct kvm_memory_slot *slot, gfn_t gfn,
			 struct kvm_host_map *map)
{
	kvm_pfn_t pfn;
	void *hva = NULL;
	struct page *page = KVM_UNMAPPED_PAGE;

	if (!map)
		return -EINVAL;

	pfn = gfn_to_pfn_memslot(slot, gfn);
	if (is_error_noslot_pfn(pfn))
		return -EINVAL;

	if (pfn_valid(pfn)) {
		page = pfn_to_page(pfn);
		hva = kmap(page);
	} else {
		hva = memremap(pfn_to_hpa(pfn), PAGE_SIZE, MEMREMAP_WB);
	}

	if (!hva)
		return -EFAULT;

	map->page = page;
	map->hva = hva;
	map->pfn = pfn;
	map->gfn = gfn;

	return 0;
}

int kvm_vcpu_map(struct kvm_vcpu *vcpu, gfn_t gfn, struct kvm_host_map *map)
{
	return __kvm_map_gfn(kvm_vcpu_gfn_to_memslot(vcpu, gfn), gfn, map);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_map);

void kvm_vcpu_unmap(struct kvm_vcpu *vcpu, struct kvm_host_map *map,
		    bool dirty)
{
	if (!map)
		return;

	if (!map->hva)
		return;

	if (map->page)
		kunmap(map->page);
	else
		memunmap(map->hva);

	if (dirty) {
		kvm_vcpu_mark_page_dirty(vcpu, map->gfn);
		kvm_release_pfn_dirty(map->pfn);
	} else {
		kvm_release_pfn_clean(map->pfn);
	}

	map->hva = NULL;
	map->page = NULL;
}
EXPORT_SYMBOL_GPL(kvm_vcpu_unmap);

1811 1812
struct page *kvm_vcpu_gfn_to_page(struct kvm_vcpu *vcpu, gfn_t gfn)
{
D
Dan Williams 已提交
1813
	kvm_pfn_t pfn;
1814 1815 1816 1817 1818 1819 1820

	pfn = kvm_vcpu_gfn_to_pfn(vcpu, gfn);

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

1821 1822
void kvm_release_page_clean(struct page *page)
{
1823 1824
	WARN_ON(is_error_page(page));

1825
	kvm_release_pfn_clean(page_to_pfn(page));
1826 1827 1828
}
EXPORT_SYMBOL_GPL(kvm_release_page_clean);

D
Dan Williams 已提交
1829
void kvm_release_pfn_clean(kvm_pfn_t pfn)
1830
{
1831
	if (!is_error_noslot_pfn(pfn) && !kvm_is_reserved_pfn(pfn))
1832
		put_page(pfn_to_page(pfn));
1833 1834 1835
}
EXPORT_SYMBOL_GPL(kvm_release_pfn_clean);

1836
void kvm_release_page_dirty(struct page *page)
1837
{
X
Xiao Guangrong 已提交
1838 1839
	WARN_ON(is_error_page(page));

1840 1841 1842 1843
	kvm_release_pfn_dirty(page_to_pfn(page));
}
EXPORT_SYMBOL_GPL(kvm_release_page_dirty);

1844
void kvm_release_pfn_dirty(kvm_pfn_t pfn)
1845 1846 1847 1848
{
	kvm_set_pfn_dirty(pfn);
	kvm_release_pfn_clean(pfn);
}
1849
EXPORT_SYMBOL_GPL(kvm_release_pfn_dirty);
1850

D
Dan Williams 已提交
1851
void kvm_set_pfn_dirty(kvm_pfn_t pfn)
1852
{
1853
	if (!kvm_is_reserved_pfn(pfn)) {
1854
		struct page *page = pfn_to_page(pfn);
1855

1856 1857 1858
		if (!PageReserved(page))
			SetPageDirty(page);
	}
1859
}
1860 1861
EXPORT_SYMBOL_GPL(kvm_set_pfn_dirty);

D
Dan Williams 已提交
1862
void kvm_set_pfn_accessed(kvm_pfn_t pfn)
1863
{
1864
	if (!kvm_is_reserved_pfn(pfn))
1865
		mark_page_accessed(pfn_to_page(pfn));
1866 1867 1868
}
EXPORT_SYMBOL_GPL(kvm_set_pfn_accessed);

D
Dan Williams 已提交
1869
void kvm_get_pfn(kvm_pfn_t pfn)
1870
{
1871
	if (!kvm_is_reserved_pfn(pfn))
1872
		get_page(pfn_to_page(pfn));
1873 1874
}
EXPORT_SYMBOL_GPL(kvm_get_pfn);
1875

1876 1877 1878 1879 1880 1881 1882 1883
static int next_segment(unsigned long len, int offset)
{
	if (len > PAGE_SIZE - offset)
		return PAGE_SIZE - offset;
	else
		return len;
}

1884 1885
static int __kvm_read_guest_page(struct kvm_memory_slot *slot, gfn_t gfn,
				 void *data, int offset, int len)
1886
{
1887 1888
	int r;
	unsigned long addr;
1889

1890
	addr = gfn_to_hva_memslot_prot(slot, gfn, NULL);
1891 1892
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1893
	r = __copy_from_user(data, (void __user *)addr + offset, len);
1894
	if (r)
1895 1896 1897
		return -EFAULT;
	return 0;
}
1898 1899 1900 1901 1902 1903 1904 1905

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);
}
1906 1907
EXPORT_SYMBOL_GPL(kvm_read_guest_page);

1908 1909 1910 1911 1912 1913 1914 1915 1916
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);

1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936
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);

1937
int kvm_vcpu_read_guest(struct kvm_vcpu *vcpu, gpa_t gpa, void *data, unsigned long len)
1938 1939
{
	gfn_t gfn = gpa >> PAGE_SHIFT;
1940
	int seg;
1941
	int offset = offset_in_page(gpa);
1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955
	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);
1956

1957 1958 1959 1960 1961 1962 1963
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);
1964 1965
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1966
	pagefault_disable();
1967
	r = __copy_from_user_inatomic(data, (void __user *)addr + offset, len);
1968
	pagefault_enable();
1969 1970 1971 1972 1973
	if (r)
		return -EFAULT;
	return 0;
}

1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997
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)
1998
{
1999 2000
	int r;
	unsigned long addr;
2001

2002
	addr = gfn_to_hva_memslot(memslot, gfn);
2003 2004
	if (kvm_is_error_hva(addr))
		return -EFAULT;
2005
	r = __copy_to_user((void __user *)addr + offset, data, len);
2006
	if (r)
2007
		return -EFAULT;
2008
	mark_page_dirty_in_slot(memslot, gfn);
2009 2010
	return 0;
}
2011 2012 2013 2014 2015 2016 2017 2018

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);
}
2019 2020
EXPORT_SYMBOL_GPL(kvm_write_guest_page);

2021 2022 2023 2024 2025 2026 2027 2028 2029
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);

2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048
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;
}
2049
EXPORT_SYMBOL_GPL(kvm_write_guest);
2050

2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071
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);

2072 2073 2074
static int __kvm_gfn_to_hva_cache_init(struct kvm_memslots *slots,
				       struct gfn_to_hva_cache *ghc,
				       gpa_t gpa, unsigned long len)
2075 2076
{
	int offset = offset_in_page(gpa);
2077 2078 2079 2080
	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;
2081
	int r = start_gfn <= end_gfn ? 0 : -EINVAL;
2082 2083 2084

	ghc->gpa = gpa;
	ghc->generation = slots->generation;
2085
	ghc->len = len;
2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102
	ghc->hva = KVM_HVA_ERR_BAD;

	/*
	 * If the requested region crosses two memslots, we still
	 * verify that the entire region is valid here.
	 */
	while (!r && start_gfn <= end_gfn) {
		ghc->memslot = __gfn_to_memslot(slots, start_gfn);
		ghc->hva = gfn_to_hva_many(ghc->memslot, start_gfn,
					   &nr_pages_avail);
		if (kvm_is_error_hva(ghc->hva))
			r = -EFAULT;
		start_gfn += nr_pages_avail;
	}

	/* Use the slow path for cross page reads and writes. */
	if (!r && nr_pages_needed == 1)
2103
		ghc->hva += offset;
2104
	else
2105
		ghc->memslot = NULL;
2106 2107

	return r;
2108
}
2109

2110
int kvm_gfn_to_hva_cache_init(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
2111 2112
			      gpa_t gpa, unsigned long len)
{
2113
	struct kvm_memslots *slots = kvm_memslots(kvm);
2114 2115
	return __kvm_gfn_to_hva_cache_init(slots, ghc, gpa, len);
}
2116
EXPORT_SYMBOL_GPL(kvm_gfn_to_hva_cache_init);
2117

2118
int kvm_write_guest_offset_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
2119 2120
				  void *data, unsigned int offset,
				  unsigned long len)
2121
{
2122
	struct kvm_memslots *slots = kvm_memslots(kvm);
2123
	int r;
2124
	gpa_t gpa = ghc->gpa + offset;
2125

2126
	BUG_ON(len + offset > ghc->len);
2127

2128
	if (slots->generation != ghc->generation)
2129
		__kvm_gfn_to_hva_cache_init(slots, ghc, ghc->gpa, ghc->len);
2130 2131

	if (unlikely(!ghc->memslot))
2132
		return kvm_write_guest(kvm, gpa, data, len);
2133 2134 2135 2136

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

2137
	r = __copy_to_user((void __user *)ghc->hva + offset, data, len);
2138 2139
	if (r)
		return -EFAULT;
2140
	mark_page_dirty_in_slot(ghc->memslot, gpa >> PAGE_SHIFT);
2141 2142 2143

	return 0;
}
2144
EXPORT_SYMBOL_GPL(kvm_write_guest_offset_cached);
2145

2146 2147
int kvm_write_guest_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
			   void *data, unsigned long len)
2148
{
2149
	return kvm_write_guest_offset_cached(kvm, ghc, data, 0, len);
2150
}
2151
EXPORT_SYMBOL_GPL(kvm_write_guest_cached);
2152

2153 2154
int kvm_read_guest_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
			   void *data, unsigned long len)
2155
{
2156
	struct kvm_memslots *slots = kvm_memslots(kvm);
2157 2158
	int r;

2159 2160
	BUG_ON(len > ghc->len);

2161
	if (slots->generation != ghc->generation)
2162
		__kvm_gfn_to_hva_cache_init(slots, ghc, ghc->gpa, ghc->len);
2163 2164

	if (unlikely(!ghc->memslot))
2165
		return kvm_read_guest(kvm, ghc->gpa, data, len);
2166 2167 2168 2169 2170 2171 2172 2173 2174 2175

	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;
}
2176
EXPORT_SYMBOL_GPL(kvm_read_guest_cached);
2177

2178 2179
int kvm_clear_guest_page(struct kvm *kvm, gfn_t gfn, int offset, int len)
{
2180 2181 2182
	const void *zero_page = (const void *) __va(page_to_phys(ZERO_PAGE(0)));

	return kvm_write_guest_page(kvm, gfn, zero_page, offset, len);
2183 2184 2185 2186 2187 2188 2189 2190 2191 2192
}
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;

2193
	while ((seg = next_segment(len, offset)) != 0) {
2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204
		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);

2205
static void mark_page_dirty_in_slot(struct kvm_memory_slot *memslot,
2206
				    gfn_t gfn)
A
Avi Kivity 已提交
2207
{
R
Rusty Russell 已提交
2208 2209
	if (memslot && memslot->dirty_bitmap) {
		unsigned long rel_gfn = gfn - memslot->base_gfn;
A
Avi Kivity 已提交
2210

2211
		set_bit_le(rel_gfn, memslot->dirty_bitmap);
A
Avi Kivity 已提交
2212 2213 2214
	}
}

2215 2216 2217 2218 2219
void mark_page_dirty(struct kvm *kvm, gfn_t gfn)
{
	struct kvm_memory_slot *memslot;

	memslot = gfn_to_memslot(kvm, gfn);
2220
	mark_page_dirty_in_slot(memslot, gfn);
2221
}
2222
EXPORT_SYMBOL_GPL(mark_page_dirty);
2223

2224 2225 2226 2227 2228 2229 2230 2231 2232
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);

2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255
void kvm_sigset_activate(struct kvm_vcpu *vcpu)
{
	if (!vcpu->sigset_active)
		return;

	/*
	 * This does a lockless modification of ->real_blocked, which is fine
	 * because, only current can change ->real_blocked and all readers of
	 * ->real_blocked don't care as long ->real_blocked is always a subset
	 * of ->blocked.
	 */
	sigprocmask(SIG_SETMASK, &vcpu->sigset, &current->real_blocked);
}

void kvm_sigset_deactivate(struct kvm_vcpu *vcpu)
{
	if (!vcpu->sigset_active)
		return;

	sigprocmask(SIG_SETMASK, &current->real_blocked, NULL);
	sigemptyset(&current->real_blocked);
}

W
Wanpeng Li 已提交
2256 2257
static void grow_halt_poll_ns(struct kvm_vcpu *vcpu)
{
2258
	unsigned int old, val, grow, grow_start;
W
Wanpeng Li 已提交
2259

2260
	old = val = vcpu->halt_poll_ns;
2261
	grow_start = READ_ONCE(halt_poll_ns_grow_start);
2262
	grow = READ_ONCE(halt_poll_ns_grow);
2263 2264 2265
	if (!grow)
		goto out;

2266 2267 2268
	val *= grow;
	if (val < grow_start)
		val = grow_start;
W
Wanpeng Li 已提交
2269

2270 2271 2272
	if (val > halt_poll_ns)
		val = halt_poll_ns;

W
Wanpeng Li 已提交
2273
	vcpu->halt_poll_ns = val;
2274
out:
2275
	trace_kvm_halt_poll_ns_grow(vcpu->vcpu_id, val, old);
W
Wanpeng Li 已提交
2276 2277 2278 2279
}

static void shrink_halt_poll_ns(struct kvm_vcpu *vcpu)
{
2280
	unsigned int old, val, shrink;
W
Wanpeng Li 已提交
2281

2282
	old = val = vcpu->halt_poll_ns;
2283 2284
	shrink = READ_ONCE(halt_poll_ns_shrink);
	if (shrink == 0)
W
Wanpeng Li 已提交
2285 2286
		val = 0;
	else
2287
		val /= shrink;
W
Wanpeng Li 已提交
2288 2289

	vcpu->halt_poll_ns = val;
2290
	trace_kvm_halt_poll_ns_shrink(vcpu->vcpu_id, val, old);
W
Wanpeng Li 已提交
2291 2292
}

2293 2294
static int kvm_vcpu_check_block(struct kvm_vcpu *vcpu)
{
2295 2296 2297
	int ret = -EINTR;
	int idx = srcu_read_lock(&vcpu->kvm->srcu);

2298 2299
	if (kvm_arch_vcpu_runnable(vcpu)) {
		kvm_make_request(KVM_REQ_UNHALT, vcpu);
2300
		goto out;
2301 2302
	}
	if (kvm_cpu_has_pending_timer(vcpu))
2303
		goto out;
2304
	if (signal_pending(current))
2305
		goto out;
2306

2307 2308 2309 2310
	ret = 0;
out:
	srcu_read_unlock(&vcpu->kvm->srcu, idx);
	return ret;
2311 2312
}

E
Eddie Dong 已提交
2313 2314 2315
/*
 * The vCPU has executed a HLT instruction with in-kernel mode enabled.
 */
2316
void kvm_vcpu_block(struct kvm_vcpu *vcpu)
2317
{
2318
	ktime_t start, cur;
2319
	DECLARE_SWAITQUEUE(wait);
2320
	bool waited = false;
W
Wanpeng Li 已提交
2321
	u64 block_ns;
2322 2323

	start = cur = ktime_get();
2324
	if (vcpu->halt_poll_ns && !kvm_arch_no_poll(vcpu)) {
W
Wanpeng Li 已提交
2325
		ktime_t stop = ktime_add_ns(ktime_get(), vcpu->halt_poll_ns);
2326

2327
		++vcpu->stat.halt_attempted_poll;
2328 2329 2330 2331 2332 2333 2334
		do {
			/*
			 * This sets KVM_REQ_UNHALT if an interrupt
			 * arrives.
			 */
			if (kvm_vcpu_check_block(vcpu) < 0) {
				++vcpu->stat.halt_successful_poll;
2335 2336
				if (!vcpu_valid_wakeup(vcpu))
					++vcpu->stat.halt_poll_invalid;
2337 2338 2339 2340 2341
				goto out;
			}
			cur = ktime_get();
		} while (single_task_running() && ktime_before(cur, stop));
	}
2342

2343 2344
	kvm_arch_vcpu_blocking(vcpu);

2345
	for (;;) {
2346
		prepare_to_swait_exclusive(&vcpu->wq, &wait, TASK_INTERRUPTIBLE);
2347

2348
		if (kvm_vcpu_check_block(vcpu) < 0)
2349 2350
			break;

2351
		waited = true;
E
Eddie Dong 已提交
2352 2353
		schedule();
	}
2354

2355
	finish_swait(&vcpu->wq, &wait);
2356 2357
	cur = ktime_get();

2358
	kvm_arch_vcpu_unblocking(vcpu);
2359
out:
W
Wanpeng Li 已提交
2360 2361
	block_ns = ktime_to_ns(cur) - ktime_to_ns(start);

2362 2363 2364
	if (!vcpu_valid_wakeup(vcpu))
		shrink_halt_poll_ns(vcpu);
	else if (halt_poll_ns) {
W
Wanpeng Li 已提交
2365 2366 2367
		if (block_ns <= vcpu->halt_poll_ns)
			;
		/* we had a long block, shrink polling */
2368
		else if (vcpu->halt_poll_ns && block_ns > halt_poll_ns)
W
Wanpeng Li 已提交
2369 2370 2371 2372 2373
			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);
2374 2375
	} else
		vcpu->halt_poll_ns = 0;
W
Wanpeng Li 已提交
2376

2377 2378
	trace_kvm_vcpu_wakeup(block_ns, waited, vcpu_valid_wakeup(vcpu));
	kvm_arch_vcpu_block_finish(vcpu);
E
Eddie Dong 已提交
2379
}
2380
EXPORT_SYMBOL_GPL(kvm_vcpu_block);
E
Eddie Dong 已提交
2381

2382
bool kvm_vcpu_wake_up(struct kvm_vcpu *vcpu)
2383
{
2384
	struct swait_queue_head *wqp;
2385 2386

	wqp = kvm_arch_vcpu_wq(vcpu);
2387
	if (swq_has_sleeper(wqp)) {
2388
		swake_up_one(wqp);
2389
		++vcpu->stat.halt_wakeup;
2390
		return true;
2391 2392
	}

2393
	return false;
2394 2395 2396
}
EXPORT_SYMBOL_GPL(kvm_vcpu_wake_up);

2397
#ifndef CONFIG_S390
2398 2399 2400 2401 2402 2403 2404 2405
/*
 * 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;

2406 2407 2408
	if (kvm_vcpu_wake_up(vcpu))
		return;

2409 2410 2411 2412 2413 2414
	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();
}
2415
EXPORT_SYMBOL_GPL(kvm_vcpu_kick);
2416
#endif /* !CONFIG_S390 */
2417

2418
int kvm_vcpu_yield_to(struct kvm_vcpu *target)
2419 2420 2421
{
	struct pid *pid;
	struct task_struct *task = NULL;
2422
	int ret = 0;
2423 2424 2425 2426

	rcu_read_lock();
	pid = rcu_dereference(target->pid);
	if (pid)
2427
		task = get_pid_task(pid, PIDTYPE_PID);
2428 2429
	rcu_read_unlock();
	if (!task)
2430 2431
		return ret;
	ret = yield_to(task, 1);
2432
	put_task_struct(task);
2433 2434

	return ret;
2435 2436 2437
}
EXPORT_SYMBOL_GPL(kvm_vcpu_yield_to);

2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459
/*
 * 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.
 */
2460
static bool kvm_vcpu_eligible_for_directed_yield(struct kvm_vcpu *vcpu)
2461
{
2462
#ifdef CONFIG_HAVE_KVM_CPU_RELAX_INTERCEPT
2463 2464 2465
	bool eligible;

	eligible = !vcpu->spin_loop.in_spin_loop ||
2466
		    vcpu->spin_loop.dy_eligible;
2467 2468 2469 2470 2471

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

	return eligible;
2472 2473
#else
	return true;
2474
#endif
2475
}
2476

2477
void kvm_vcpu_on_spin(struct kvm_vcpu *me, bool yield_to_kernel_mode)
Z
Zhai, Edwin 已提交
2478
{
2479 2480 2481 2482
	struct kvm *kvm = me->kvm;
	struct kvm_vcpu *vcpu;
	int last_boosted_vcpu = me->kvm->last_boosted_vcpu;
	int yielded = 0;
2483
	int try = 3;
2484 2485
	int pass;
	int i;
Z
Zhai, Edwin 已提交
2486

2487
	kvm_vcpu_set_in_spin_loop(me, true);
2488 2489 2490 2491 2492 2493 2494
	/*
	 * 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.
	 */
2495
	for (pass = 0; pass < 2 && !yielded && try; pass++) {
2496
		kvm_for_each_vcpu(i, vcpu, kvm) {
2497
			if (!pass && i <= last_boosted_vcpu) {
2498 2499 2500 2501
				i = last_boosted_vcpu;
				continue;
			} else if (pass && i > last_boosted_vcpu)
				break;
2502
			if (!READ_ONCE(vcpu->preempted))
2503
				continue;
2504 2505
			if (vcpu == me)
				continue;
2506
			if (swait_active(&vcpu->wq) && !kvm_arch_vcpu_runnable(vcpu))
2507
				continue;
2508 2509
			if (yield_to_kernel_mode && !kvm_arch_vcpu_in_kernel(vcpu))
				continue;
2510 2511
			if (!kvm_vcpu_eligible_for_directed_yield(vcpu))
				continue;
2512 2513 2514

			yielded = kvm_vcpu_yield_to(vcpu);
			if (yielded > 0) {
2515 2516
				kvm->last_boosted_vcpu = i;
				break;
2517 2518 2519 2520
			} else if (yielded < 0) {
				try--;
				if (!try)
					break;
2521 2522 2523
			}
		}
	}
2524
	kvm_vcpu_set_in_spin_loop(me, false);
2525 2526 2527

	/* Ensure vcpu is not eligible during next spinloop */
	kvm_vcpu_set_dy_eligible(me, false);
Z
Zhai, Edwin 已提交
2528 2529 2530
}
EXPORT_SYMBOL_GPL(kvm_vcpu_on_spin);

2531
static vm_fault_t kvm_vcpu_fault(struct vm_fault *vmf)
2532
{
2533
	struct kvm_vcpu *vcpu = vmf->vma->vm_file->private_data;
2534 2535
	struct page *page;

2536
	if (vmf->pgoff == 0)
2537
		page = virt_to_page(vcpu->run);
A
Avi Kivity 已提交
2538
#ifdef CONFIG_X86
2539
	else if (vmf->pgoff == KVM_PIO_PAGE_OFFSET)
2540
		page = virt_to_page(vcpu->arch.pio_data);
2541
#endif
2542
#ifdef CONFIG_KVM_MMIO
2543 2544
	else if (vmf->pgoff == KVM_COALESCED_MMIO_PAGE_OFFSET)
		page = virt_to_page(vcpu->kvm->coalesced_mmio_ring);
A
Avi Kivity 已提交
2545
#endif
2546
	else
2547
		return kvm_arch_vcpu_fault(vcpu, vmf);
2548
	get_page(page);
2549 2550
	vmf->page = page;
	return 0;
2551 2552
}

2553
static const struct vm_operations_struct kvm_vcpu_vm_ops = {
2554
	.fault = kvm_vcpu_fault,
2555 2556 2557 2558 2559 2560 2561 2562
};

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
2563 2564 2565 2566
static int kvm_vcpu_release(struct inode *inode, struct file *filp)
{
	struct kvm_vcpu *vcpu = filp->private_data;

2567
	debugfs_remove_recursive(vcpu->debugfs_dentry);
A
Al Viro 已提交
2568
	kvm_put_kvm(vcpu->kvm);
A
Avi Kivity 已提交
2569 2570 2571
	return 0;
}

2572
static struct file_operations kvm_vcpu_fops = {
A
Avi Kivity 已提交
2573 2574
	.release        = kvm_vcpu_release,
	.unlocked_ioctl = kvm_vcpu_ioctl,
2575
	.mmap           = kvm_vcpu_mmap,
2576
	.llseek		= noop_llseek,
2577
	KVM_COMPAT(kvm_vcpu_compat_ioctl),
A
Avi Kivity 已提交
2578 2579 2580 2581 2582 2583 2584
};

/*
 * Allocates an inode for the vcpu.
 */
static int create_vcpu_fd(struct kvm_vcpu *vcpu)
{
2585 2586 2587 2588
	char name[8 + 1 + ITOA_MAX_LEN + 1];

	snprintf(name, sizeof(name), "kvm-vcpu:%d", vcpu->vcpu_id);
	return anon_inode_getfd(name, &kvm_vcpu_fops, vcpu, O_RDWR | O_CLOEXEC);
A
Avi Kivity 已提交
2589 2590
}

2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616
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;
}

2617 2618 2619
/*
 * Creates some virtual cpus.  Good luck creating more than one.
 */
2620
static int kvm_vm_ioctl_create_vcpu(struct kvm *kvm, u32 id)
2621 2622
{
	int r;
2623
	struct kvm_vcpu *vcpu;
2624

G
Greg Kurz 已提交
2625
	if (id >= KVM_MAX_VCPU_ID)
2626 2627
		return -EINVAL;

2628 2629 2630 2631 2632 2633 2634 2635 2636
	mutex_lock(&kvm->lock);
	if (kvm->created_vcpus == KVM_MAX_VCPUS) {
		mutex_unlock(&kvm->lock);
		return -EINVAL;
	}

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

2637
	vcpu = kvm_arch_vcpu_create(kvm, id);
2638 2639 2640 2641
	if (IS_ERR(vcpu)) {
		r = PTR_ERR(vcpu);
		goto vcpu_decrement;
	}
2642

2643 2644
	preempt_notifier_init(&vcpu->preempt_notifier, &kvm_preempt_ops);

2645 2646
	r = kvm_arch_vcpu_setup(vcpu);
	if (r)
2647
		goto vcpu_destroy;
2648

2649 2650 2651 2652
	r = kvm_create_vcpu_debugfs(vcpu);
	if (r)
		goto vcpu_destroy;

S
Shaohua Li 已提交
2653
	mutex_lock(&kvm->lock);
2654 2655 2656 2657
	if (kvm_get_vcpu_by_id(kvm, id)) {
		r = -EEXIST;
		goto unlock_vcpu_destroy;
	}
2658 2659

	BUG_ON(kvm->vcpus[atomic_read(&kvm->online_vcpus)]);
2660

R
Rusty Russell 已提交
2661
	/* Now it's all set up, let userspace reach it */
A
Al Viro 已提交
2662
	kvm_get_kvm(kvm);
A
Avi Kivity 已提交
2663
	r = create_vcpu_fd(vcpu);
2664 2665
	if (r < 0) {
		kvm_put_kvm(kvm);
2666
		goto unlock_vcpu_destroy;
2667 2668 2669
	}

	kvm->vcpus[atomic_read(&kvm->online_vcpus)] = vcpu;
2670 2671 2672 2673 2674

	/*
	 * Pairs with smp_rmb() in kvm_get_vcpu.  Write kvm->vcpus
	 * before kvm->online_vcpu's incremented value.
	 */
2675 2676 2677 2678
	smp_wmb();
	atomic_inc(&kvm->online_vcpus);

	mutex_unlock(&kvm->lock);
2679
	kvm_arch_vcpu_postcreate(vcpu);
R
Rusty Russell 已提交
2680
	return r;
2681

2682
unlock_vcpu_destroy:
2683
	mutex_unlock(&kvm->lock);
2684
	debugfs_remove_recursive(vcpu->debugfs_dentry);
2685
vcpu_destroy:
2686
	kvm_arch_vcpu_destroy(vcpu);
2687 2688 2689 2690
vcpu_decrement:
	mutex_lock(&kvm->lock);
	kvm->created_vcpus--;
	mutex_unlock(&kvm->lock);
2691 2692 2693
	return r;
}

A
Avi Kivity 已提交
2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704
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
2705 2706
static long kvm_vcpu_ioctl(struct file *filp,
			   unsigned int ioctl, unsigned long arg)
A
Avi Kivity 已提交
2707
{
A
Avi Kivity 已提交
2708
	struct kvm_vcpu *vcpu = filp->private_data;
A
Al Viro 已提交
2709
	void __user *argp = (void __user *)arg;
2710
	int r;
2711 2712
	struct kvm_fpu *fpu = NULL;
	struct kvm_sregs *kvm_sregs = NULL;
A
Avi Kivity 已提交
2713

2714 2715
	if (vcpu->kvm->mm != current->mm)
		return -EIO;
2716

2717 2718 2719
	if (unlikely(_IOC_TYPE(ioctl) != KVMIO))
		return -EINVAL;

2720
	/*
2721 2722
	 * Some architectures have vcpu ioctls that are asynchronous to vcpu
	 * execution; mutex_lock() would break them.
2723
	 */
2724 2725
	r = kvm_arch_vcpu_async_ioctl(filp, ioctl, arg);
	if (r != -ENOIOCTLCMD)
2726
		return r;
2727

2728 2729
	if (mutex_lock_killable(&vcpu->mutex))
		return -EINTR;
A
Avi Kivity 已提交
2730
	switch (ioctl) {
2731 2732
	case KVM_RUN: {
		struct pid *oldpid;
2733 2734 2735
		r = -EINVAL;
		if (arg)
			goto out;
2736
		oldpid = rcu_access_pointer(vcpu->pid);
2737
		if (unlikely(oldpid != task_pid(current))) {
2738
			/* The thread running this VCPU changed. */
2739
			struct pid *newpid;
2740

2741 2742 2743 2744 2745
			r = kvm_arch_vcpu_run_pid_change(vcpu);
			if (r)
				break;

			newpid = get_task_pid(current, PIDTYPE_PID);
2746 2747 2748 2749 2750
			rcu_assign_pointer(vcpu->pid, newpid);
			if (oldpid)
				synchronize_rcu();
			put_pid(oldpid);
		}
2751
		r = kvm_arch_vcpu_ioctl_run(vcpu, vcpu->run);
2752
		trace_kvm_userspace_exit(vcpu->run->exit_reason, r);
A
Avi Kivity 已提交
2753
		break;
2754
	}
A
Avi Kivity 已提交
2755
	case KVM_GET_REGS: {
2756
		struct kvm_regs *kvm_regs;
A
Avi Kivity 已提交
2757

2758
		r = -ENOMEM;
2759
		kvm_regs = kzalloc(sizeof(struct kvm_regs), GFP_KERNEL_ACCOUNT);
2760
		if (!kvm_regs)
A
Avi Kivity 已提交
2761
			goto out;
2762 2763 2764
		r = kvm_arch_vcpu_ioctl_get_regs(vcpu, kvm_regs);
		if (r)
			goto out_free1;
A
Avi Kivity 已提交
2765
		r = -EFAULT;
2766 2767
		if (copy_to_user(argp, kvm_regs, sizeof(struct kvm_regs)))
			goto out_free1;
A
Avi Kivity 已提交
2768
		r = 0;
2769 2770
out_free1:
		kfree(kvm_regs);
A
Avi Kivity 已提交
2771 2772 2773
		break;
	}
	case KVM_SET_REGS: {
2774
		struct kvm_regs *kvm_regs;
A
Avi Kivity 已提交
2775

2776
		r = -ENOMEM;
2777 2778 2779
		kvm_regs = memdup_user(argp, sizeof(*kvm_regs));
		if (IS_ERR(kvm_regs)) {
			r = PTR_ERR(kvm_regs);
A
Avi Kivity 已提交
2780
			goto out;
2781
		}
2782 2783
		r = kvm_arch_vcpu_ioctl_set_regs(vcpu, kvm_regs);
		kfree(kvm_regs);
A
Avi Kivity 已提交
2784 2785 2786
		break;
	}
	case KVM_GET_SREGS: {
2787 2788
		kvm_sregs = kzalloc(sizeof(struct kvm_sregs),
				    GFP_KERNEL_ACCOUNT);
2789 2790 2791 2792
		r = -ENOMEM;
		if (!kvm_sregs)
			goto out;
		r = kvm_arch_vcpu_ioctl_get_sregs(vcpu, kvm_sregs);
A
Avi Kivity 已提交
2793 2794 2795
		if (r)
			goto out;
		r = -EFAULT;
2796
		if (copy_to_user(argp, kvm_sregs, sizeof(struct kvm_sregs)))
A
Avi Kivity 已提交
2797 2798 2799 2800 2801
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_SREGS: {
2802 2803 2804
		kvm_sregs = memdup_user(argp, sizeof(*kvm_sregs));
		if (IS_ERR(kvm_sregs)) {
			r = PTR_ERR(kvm_sregs);
G
Guo Chao 已提交
2805
			kvm_sregs = NULL;
A
Avi Kivity 已提交
2806
			goto out;
2807
		}
2808
		r = kvm_arch_vcpu_ioctl_set_sregs(vcpu, kvm_sregs);
A
Avi Kivity 已提交
2809 2810
		break;
	}
2811 2812 2813 2814 2815 2816 2817
	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;
2818
		if (copy_to_user(argp, &mp_state, sizeof(mp_state)))
2819 2820 2821 2822 2823 2824 2825 2826
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_MP_STATE: {
		struct kvm_mp_state mp_state;

		r = -EFAULT;
2827
		if (copy_from_user(&mp_state, argp, sizeof(mp_state)))
2828 2829 2830 2831
			goto out;
		r = kvm_arch_vcpu_ioctl_set_mpstate(vcpu, &mp_state);
		break;
	}
A
Avi Kivity 已提交
2832 2833 2834 2835
	case KVM_TRANSLATE: {
		struct kvm_translation tr;

		r = -EFAULT;
2836
		if (copy_from_user(&tr, argp, sizeof(tr)))
A
Avi Kivity 已提交
2837
			goto out;
2838
		r = kvm_arch_vcpu_ioctl_translate(vcpu, &tr);
A
Avi Kivity 已提交
2839 2840 2841
		if (r)
			goto out;
		r = -EFAULT;
2842
		if (copy_to_user(argp, &tr, sizeof(tr)))
A
Avi Kivity 已提交
2843 2844 2845 2846
			goto out;
		r = 0;
		break;
	}
J
Jan Kiszka 已提交
2847 2848
	case KVM_SET_GUEST_DEBUG: {
		struct kvm_guest_debug dbg;
A
Avi Kivity 已提交
2849 2850

		r = -EFAULT;
2851
		if (copy_from_user(&dbg, argp, sizeof(dbg)))
A
Avi Kivity 已提交
2852
			goto out;
J
Jan Kiszka 已提交
2853
		r = kvm_arch_vcpu_ioctl_set_guest_debug(vcpu, &dbg);
A
Avi Kivity 已提交
2854 2855
		break;
	}
A
Avi Kivity 已提交
2856 2857 2858 2859 2860 2861 2862 2863 2864
	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,
2865
					   sizeof(kvm_sigmask)))
A
Avi Kivity 已提交
2866 2867
				goto out;
			r = -EINVAL;
2868
			if (kvm_sigmask.len != sizeof(sigset))
A
Avi Kivity 已提交
2869 2870 2871
				goto out;
			r = -EFAULT;
			if (copy_from_user(&sigset, sigmask_arg->sigset,
2872
					   sizeof(sigset)))
A
Avi Kivity 已提交
2873 2874 2875
				goto out;
			p = &sigset;
		}
2876
		r = kvm_vcpu_ioctl_set_sigmask(vcpu, p);
A
Avi Kivity 已提交
2877 2878
		break;
	}
A
Avi Kivity 已提交
2879
	case KVM_GET_FPU: {
2880
		fpu = kzalloc(sizeof(struct kvm_fpu), GFP_KERNEL_ACCOUNT);
2881 2882 2883 2884
		r = -ENOMEM;
		if (!fpu)
			goto out;
		r = kvm_arch_vcpu_ioctl_get_fpu(vcpu, fpu);
A
Avi Kivity 已提交
2885 2886 2887
		if (r)
			goto out;
		r = -EFAULT;
2888
		if (copy_to_user(argp, fpu, sizeof(struct kvm_fpu)))
A
Avi Kivity 已提交
2889 2890 2891 2892 2893
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_FPU: {
2894 2895 2896
		fpu = memdup_user(argp, sizeof(*fpu));
		if (IS_ERR(fpu)) {
			r = PTR_ERR(fpu);
G
Guo Chao 已提交
2897
			fpu = NULL;
A
Avi Kivity 已提交
2898
			goto out;
2899
		}
2900
		r = kvm_arch_vcpu_ioctl_set_fpu(vcpu, fpu);
A
Avi Kivity 已提交
2901 2902
		break;
	}
A
Avi Kivity 已提交
2903
	default:
2904
		r = kvm_arch_vcpu_ioctl(filp, ioctl, arg);
A
Avi Kivity 已提交
2905 2906
	}
out:
2907
	mutex_unlock(&vcpu->mutex);
2908 2909
	kfree(fpu);
	kfree(kvm_sregs);
A
Avi Kivity 已提交
2910 2911 2912
	return r;
}

2913
#ifdef CONFIG_KVM_COMPAT
2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932
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;
		sigset_t sigset;

		if (argp) {
			r = -EFAULT;
			if (copy_from_user(&kvm_sigmask, argp,
2933
					   sizeof(kvm_sigmask)))
2934 2935
				goto out;
			r = -EINVAL;
A
Al Viro 已提交
2936
			if (kvm_sigmask.len != sizeof(compat_sigset_t))
2937 2938
				goto out;
			r = -EFAULT;
A
Al Viro 已提交
2939
			if (get_compat_sigset(&sigset, (void *)sigmask_arg->sigset))
2940
				goto out;
2941 2942 2943
			r = kvm_vcpu_ioctl_set_sigmask(vcpu, &sigset);
		} else
			r = kvm_vcpu_ioctl_set_sigmask(vcpu, NULL);
2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954
		break;
	}
	default:
		r = kvm_vcpu_ioctl(filp, ioctl, arg);
	}

out:
	return r;
}
#endif

2955 2956 2957 2958 2959 2960 2961 2962 2963 2964
static int kvm_device_mmap(struct file *filp, struct vm_area_struct *vma)
{
	struct kvm_device *dev = filp->private_data;

	if (dev->ops->mmap)
		return dev->ops->mmap(dev, vma);

	return -ENODEV;
}

S
Scott Wood 已提交
2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985
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;

2986 2987 2988
	if (dev->kvm->mm != current->mm)
		return -EIO;

S
Scott Wood 已提交
2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008
	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;

3009 3010 3011 3012 3013 3014 3015
	if (dev->ops->release) {
		mutex_lock(&kvm->lock);
		list_del(&dev->vm_node);
		dev->ops->release(dev);
		mutex_unlock(&kvm->lock);
	}

S
Scott Wood 已提交
3016 3017 3018 3019 3020 3021 3022
	kvm_put_kvm(kvm);
	return 0;
}

static const struct file_operations kvm_device_fops = {
	.unlocked_ioctl = kvm_device_ioctl,
	.release = kvm_device_release,
3023
	KVM_COMPAT(kvm_device_ioctl),
3024
	.mmap = kvm_device_mmap,
S
Scott Wood 已提交
3025 3026 3027 3028 3029 3030 3031 3032 3033 3034
};

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

	return filp->private_data;
}

3035
static struct kvm_device_ops *kvm_device_ops_table[KVM_DEV_TYPE_MAX] = {
3036
#ifdef CONFIG_KVM_MPIC
3037 3038
	[KVM_DEV_TYPE_FSL_MPIC_20]	= &kvm_mpic_ops,
	[KVM_DEV_TYPE_FSL_MPIC_42]	= &kvm_mpic_ops,
3039
#endif
3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053
};

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;
}

3054 3055 3056 3057 3058 3059
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
3060 3061 3062 3063 3064 3065
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;
P
Paolo Bonzini 已提交
3066
	int type;
S
Scott Wood 已提交
3067 3068
	int ret;

3069 3070 3071
	if (cd->type >= ARRAY_SIZE(kvm_device_ops_table))
		return -ENODEV;

P
Paolo Bonzini 已提交
3072 3073
	type = array_index_nospec(cd->type, ARRAY_SIZE(kvm_device_ops_table));
	ops = kvm_device_ops_table[type];
3074
	if (ops == NULL)
S
Scott Wood 已提交
3075 3076 3077 3078 3079
		return -ENODEV;

	if (test)
		return 0;

3080
	dev = kzalloc(sizeof(*dev), GFP_KERNEL_ACCOUNT);
S
Scott Wood 已提交
3081 3082 3083 3084 3085 3086
	if (!dev)
		return -ENOMEM;

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

3087
	mutex_lock(&kvm->lock);
P
Paolo Bonzini 已提交
3088
	ret = ops->create(dev, type);
S
Scott Wood 已提交
3089
	if (ret < 0) {
3090
		mutex_unlock(&kvm->lock);
S
Scott Wood 已提交
3091 3092 3093
		kfree(dev);
		return ret;
	}
3094 3095
	list_add(&dev->vm_node, &kvm->devices);
	mutex_unlock(&kvm->lock);
S
Scott Wood 已提交
3096

3097 3098 3099
	if (ops->init)
		ops->init(dev);

3100
	kvm_get_kvm(kvm);
3101
	ret = anon_inode_getfd(ops->name, &kvm_device_fops, dev, O_RDWR | O_CLOEXEC);
S
Scott Wood 已提交
3102
	if (ret < 0) {
3103
		kvm_put_kvm(kvm);
3104 3105 3106
		mutex_lock(&kvm->lock);
		list_del(&dev->vm_node);
		mutex_unlock(&kvm->lock);
3107
		ops->destroy(dev);
S
Scott Wood 已提交
3108 3109 3110 3111 3112 3113 3114
		return ret;
	}

	cd->fd = ret;
	return 0;
}

3115 3116 3117 3118 3119 3120 3121 3122 3123 3124
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
3125
#ifdef CONFIG_HAVE_KVM_IRQFD
3126
	case KVM_CAP_IRQFD:
3127 3128
	case KVM_CAP_IRQFD_RESAMPLE:
#endif
3129
	case KVM_CAP_IOEVENTFD_ANY_LENGTH:
3130
	case KVM_CAP_CHECK_EXTENSION_VM:
3131
	case KVM_CAP_ENABLE_CAP_VM:
3132
#ifdef CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT
3133
	case KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2:
3134
#endif
3135
		return 1;
3136
#ifdef CONFIG_KVM_MMIO
3137 3138
	case KVM_CAP_COALESCED_MMIO:
		return KVM_COALESCED_MMIO_PAGE_OFFSET;
P
Peng Hao 已提交
3139 3140
	case KVM_CAP_COALESCED_PIO:
		return 1;
3141
#endif
3142 3143 3144
#ifdef CONFIG_HAVE_KVM_IRQ_ROUTING
	case KVM_CAP_IRQ_ROUTING:
		return KVM_MAX_IRQ_ROUTES;
3145 3146 3147 3148
#endif
#if KVM_ADDRESS_SPACE_NUM > 1
	case KVM_CAP_MULTI_ADDRESS_SPACE:
		return KVM_ADDRESS_SPACE_NUM;
3149
#endif
G
Greg Kurz 已提交
3150 3151
	case KVM_CAP_MAX_VCPU_ID:
		return KVM_MAX_VCPU_ID;
3152 3153
	case KVM_CAP_NR_MEMSLOTS:
		return KVM_USER_MEM_SLOTS;
3154 3155 3156 3157 3158 3159
	default:
		break;
	}
	return kvm_vm_ioctl_check_extension(kvm, arg);
}

3160 3161 3162 3163 3164 3165 3166 3167 3168 3169
int __attribute__((weak)) kvm_vm_ioctl_enable_cap(struct kvm *kvm,
						  struct kvm_enable_cap *cap)
{
	return -EINVAL;
}

static int kvm_vm_ioctl_enable_cap_generic(struct kvm *kvm,
					   struct kvm_enable_cap *cap)
{
	switch (cap->cap) {
3170
#ifdef CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT
3171
	case KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2:
3172 3173 3174 3175 3176
		if (cap->flags || (cap->args[0] & ~1))
			return -EINVAL;
		kvm->manual_dirty_log_protect = cap->args[0];
		return 0;
#endif
3177 3178 3179 3180 3181
	default:
		return kvm_vm_ioctl_enable_cap(kvm, cap);
	}
}

A
Avi Kivity 已提交
3182 3183 3184 3185 3186
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;
3187
	int r;
A
Avi Kivity 已提交
3188

3189 3190
	if (kvm->mm != current->mm)
		return -EIO;
A
Avi Kivity 已提交
3191 3192 3193 3194
	switch (ioctl) {
	case KVM_CREATE_VCPU:
		r = kvm_vm_ioctl_create_vcpu(kvm, arg);
		break;
3195 3196 3197 3198 3199 3200 3201 3202 3203
	case KVM_ENABLE_CAP: {
		struct kvm_enable_cap cap;

		r = -EFAULT;
		if (copy_from_user(&cap, argp, sizeof(cap)))
			goto out;
		r = kvm_vm_ioctl_enable_cap_generic(kvm, &cap);
		break;
	}
3204 3205 3206 3207 3208
	case KVM_SET_USER_MEMORY_REGION: {
		struct kvm_userspace_memory_region kvm_userspace_mem;

		r = -EFAULT;
		if (copy_from_user(&kvm_userspace_mem, argp,
3209
						sizeof(kvm_userspace_mem)))
3210 3211
			goto out;

3212
		r = kvm_vm_ioctl_set_memory_region(kvm, &kvm_userspace_mem);
A
Avi Kivity 已提交
3213 3214 3215 3216 3217 3218
		break;
	}
	case KVM_GET_DIRTY_LOG: {
		struct kvm_dirty_log log;

		r = -EFAULT;
3219
		if (copy_from_user(&log, argp, sizeof(log)))
A
Avi Kivity 已提交
3220
			goto out;
3221
		r = kvm_vm_ioctl_get_dirty_log(kvm, &log);
A
Avi Kivity 已提交
3222 3223
		break;
	}
3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234
#ifdef CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT
	case KVM_CLEAR_DIRTY_LOG: {
		struct kvm_clear_dirty_log log;

		r = -EFAULT;
		if (copy_from_user(&log, argp, sizeof(log)))
			goto out;
		r = kvm_vm_ioctl_clear_dirty_log(kvm, &log);
		break;
	}
#endif
3235
#ifdef CONFIG_KVM_MMIO
3236 3237
	case KVM_REGISTER_COALESCED_MMIO: {
		struct kvm_coalesced_mmio_zone zone;
3238

3239
		r = -EFAULT;
3240
		if (copy_from_user(&zone, argp, sizeof(zone)))
3241 3242 3243 3244 3245 3246
			goto out;
		r = kvm_vm_ioctl_register_coalesced_mmio(kvm, &zone);
		break;
	}
	case KVM_UNREGISTER_COALESCED_MMIO: {
		struct kvm_coalesced_mmio_zone zone;
3247

3248
		r = -EFAULT;
3249
		if (copy_from_user(&zone, argp, sizeof(zone)))
3250 3251 3252 3253 3254
			goto out;
		r = kvm_vm_ioctl_unregister_coalesced_mmio(kvm, &zone);
		break;
	}
#endif
G
Gregory Haskins 已提交
3255 3256 3257 3258
	case KVM_IRQFD: {
		struct kvm_irqfd data;

		r = -EFAULT;
3259
		if (copy_from_user(&data, argp, sizeof(data)))
G
Gregory Haskins 已提交
3260
			goto out;
3261
		r = kvm_irqfd(kvm, &data);
G
Gregory Haskins 已提交
3262 3263
		break;
	}
G
Gregory Haskins 已提交
3264 3265 3266 3267
	case KVM_IOEVENTFD: {
		struct kvm_ioeventfd data;

		r = -EFAULT;
3268
		if (copy_from_user(&data, argp, sizeof(data)))
G
Gregory Haskins 已提交
3269 3270 3271 3272
			goto out;
		r = kvm_ioeventfd(kvm, &data);
		break;
	}
3273 3274 3275 3276 3277
#ifdef CONFIG_HAVE_KVM_MSI
	case KVM_SIGNAL_MSI: {
		struct kvm_msi msi;

		r = -EFAULT;
3278
		if (copy_from_user(&msi, argp, sizeof(msi)))
3279 3280 3281 3282
			goto out;
		r = kvm_send_userspace_msi(kvm, &msi);
		break;
	}
3283 3284 3285 3286 3287 3288 3289
#endif
#ifdef __KVM_HAVE_IRQ_LINE
	case KVM_IRQ_LINE_STATUS:
	case KVM_IRQ_LINE: {
		struct kvm_irq_level irq_event;

		r = -EFAULT;
3290
		if (copy_from_user(&irq_event, argp, sizeof(irq_event)))
3291 3292
			goto out;

3293 3294
		r = kvm_vm_ioctl_irq_line(kvm, &irq_event,
					ioctl == KVM_IRQ_LINE_STATUS);
3295 3296 3297 3298 3299
		if (r)
			goto out;

		r = -EFAULT;
		if (ioctl == KVM_IRQ_LINE_STATUS) {
3300
			if (copy_to_user(argp, &irq_event, sizeof(irq_event)))
3301 3302 3303 3304 3305 3306
				goto out;
		}

		r = 0;
		break;
	}
3307
#endif
3308 3309 3310 3311
#ifdef CONFIG_HAVE_KVM_IRQ_ROUTING
	case KVM_SET_GSI_ROUTING: {
		struct kvm_irq_routing routing;
		struct kvm_irq_routing __user *urouting;
3312
		struct kvm_irq_routing_entry *entries = NULL;
3313 3314 3315 3316 3317

		r = -EFAULT;
		if (copy_from_user(&routing, argp, sizeof(routing)))
			goto out;
		r = -EINVAL;
3318 3319
		if (!kvm_arch_can_set_irq_routing(kvm))
			goto out;
3320
		if (routing.nr > KVM_MAX_IRQ_ROUTES)
3321 3322 3323
			goto out;
		if (routing.flags)
			goto out;
3324 3325
		if (routing.nr) {
			r = -ENOMEM;
3326 3327
			entries = vmalloc(array_size(sizeof(*entries),
						     routing.nr));
3328 3329 3330 3331 3332 3333 3334 3335
			if (!entries)
				goto out;
			r = -EFAULT;
			urouting = argp;
			if (copy_from_user(entries, urouting->entries,
					   routing.nr * sizeof(*entries)))
				goto out_free_irq_routing;
		}
3336 3337
		r = kvm_set_irq_routing(kvm, entries, routing.nr,
					routing.flags);
3338
out_free_irq_routing:
3339 3340 3341 3342
		vfree(entries);
		break;
	}
#endif /* CONFIG_HAVE_KVM_IRQ_ROUTING */
S
Scott Wood 已提交
3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360
	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;
	}
3361 3362 3363
	case KVM_CHECK_EXTENSION:
		r = kvm_vm_ioctl_check_extension_generic(kvm, arg);
		break;
3364
	default:
3365
		r = kvm_arch_vm_ioctl(filp, ioctl, arg);
3366 3367 3368 3369 3370
	}
out:
	return r;
}

3371
#ifdef CONFIG_KVM_COMPAT
3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395
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;

		if (copy_from_user(&compat_log, (void __user *)arg,
				   sizeof(compat_log)))
3396
			return -EFAULT;
3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411
		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);
	}
	return r;
}
#endif

3412
static struct file_operations kvm_vm_fops = {
3413 3414
	.release        = kvm_vm_release,
	.unlocked_ioctl = kvm_vm_ioctl,
3415
	.llseek		= noop_llseek,
3416
	KVM_COMPAT(kvm_vm_compat_ioctl),
3417 3418
};

3419
static int kvm_dev_ioctl_create_vm(unsigned long type)
3420
{
3421
	int r;
3422
	struct kvm *kvm;
3423
	struct file *file;
3424

3425
	kvm = kvm_create_vm(type);
3426 3427
	if (IS_ERR(kvm))
		return PTR_ERR(kvm);
3428
#ifdef CONFIG_KVM_MMIO
3429
	r = kvm_coalesced_mmio_init(kvm);
3430 3431
	if (r < 0)
		goto put_kvm;
3432
#endif
3433
	r = get_unused_fd_flags(O_CLOEXEC);
3434 3435 3436
	if (r < 0)
		goto put_kvm;

3437 3438 3439
	file = anon_inode_getfile("kvm-vm", &kvm_vm_fops, kvm, O_RDWR);
	if (IS_ERR(file)) {
		put_unused_fd(r);
3440 3441
		r = PTR_ERR(file);
		goto put_kvm;
3442
	}
3443

3444 3445 3446 3447 3448 3449
	/*
	 * Don't call kvm_put_kvm anymore at this point; file->f_op is
	 * already set, with ->release() being kvm_vm_release().  In error
	 * cases it will be called by the final fput(file) and will take
	 * care of doing kvm_put_kvm(kvm).
	 */
3450
	if (kvm_create_vm_debugfs(kvm, r) < 0) {
3451 3452
		put_unused_fd(r);
		fput(file);
3453 3454
		return -ENOMEM;
	}
3455
	kvm_uevent_notify_change(KVM_EVENT_CREATE_VM, kvm);
3456

3457
	fd_install(r, file);
3458
	return r;
3459 3460 3461 3462

put_kvm:
	kvm_put_kvm(kvm);
	return r;
3463 3464 3465 3466 3467
}

static long kvm_dev_ioctl(struct file *filp,
			  unsigned int ioctl, unsigned long arg)
{
3468
	long r = -EINVAL;
3469 3470 3471

	switch (ioctl) {
	case KVM_GET_API_VERSION:
3472 3473
		if (arg)
			goto out;
3474 3475 3476
		r = KVM_API_VERSION;
		break;
	case KVM_CREATE_VM:
3477
		r = kvm_dev_ioctl_create_vm(arg);
3478
		break;
3479
	case KVM_CHECK_EXTENSION:
3480
		r = kvm_vm_ioctl_check_extension_generic(NULL, arg);
3481
		break;
3482 3483 3484
	case KVM_GET_VCPU_MMAP_SIZE:
		if (arg)
			goto out;
3485 3486 3487
		r = PAGE_SIZE;     /* struct kvm_run */
#ifdef CONFIG_X86
		r += PAGE_SIZE;    /* pio data page */
3488
#endif
3489
#ifdef CONFIG_KVM_MMIO
3490
		r += PAGE_SIZE;    /* coalesced mmio ring page */
3491
#endif
3492
		break;
3493 3494 3495
	case KVM_TRACE_ENABLE:
	case KVM_TRACE_PAUSE:
	case KVM_TRACE_DISABLE:
3496
		r = -EOPNOTSUPP;
3497
		break;
A
Avi Kivity 已提交
3498
	default:
3499
		return kvm_arch_dev_ioctl(filp, ioctl, arg);
A
Avi Kivity 已提交
3500 3501 3502 3503 3504 3505 3506
	}
out:
	return r;
}

static struct file_operations kvm_chardev_ops = {
	.unlocked_ioctl = kvm_dev_ioctl,
3507
	.llseek		= noop_llseek,
3508
	KVM_COMPAT(kvm_dev_ioctl),
A
Avi Kivity 已提交
3509 3510 3511
};

static struct miscdevice kvm_dev = {
A
Avi Kivity 已提交
3512
	KVM_MINOR,
A
Avi Kivity 已提交
3513 3514 3515 3516
	"kvm",
	&kvm_chardev_ops,
};

3517
static void hardware_enable_nolock(void *junk)
3518 3519
{
	int cpu = raw_smp_processor_id();
3520
	int r;
3521

3522
	if (cpumask_test_cpu(cpu, cpus_hardware_enabled))
3523
		return;
3524

3525
	cpumask_set_cpu(cpu, cpus_hardware_enabled);
3526

3527
	r = kvm_arch_hardware_enable();
3528 3529 3530 3531

	if (r) {
		cpumask_clear_cpu(cpu, cpus_hardware_enabled);
		atomic_inc(&hardware_enable_failed);
X
Xiubo Li 已提交
3532
		pr_info("kvm: enabling virtualization on CPU%d failed\n", cpu);
3533
	}
3534 3535
}

3536
static int kvm_starting_cpu(unsigned int cpu)
3537
{
3538
	raw_spin_lock(&kvm_count_lock);
3539 3540
	if (kvm_usage_count)
		hardware_enable_nolock(NULL);
3541
	raw_spin_unlock(&kvm_count_lock);
3542
	return 0;
3543 3544 3545
}

static void hardware_disable_nolock(void *junk)
3546 3547 3548
{
	int cpu = raw_smp_processor_id();

3549
	if (!cpumask_test_cpu(cpu, cpus_hardware_enabled))
3550
		return;
3551
	cpumask_clear_cpu(cpu, cpus_hardware_enabled);
3552
	kvm_arch_hardware_disable();
3553 3554
}

3555
static int kvm_dying_cpu(unsigned int cpu)
3556
{
3557
	raw_spin_lock(&kvm_count_lock);
3558 3559
	if (kvm_usage_count)
		hardware_disable_nolock(NULL);
3560
	raw_spin_unlock(&kvm_count_lock);
3561
	return 0;
3562 3563
}

3564 3565 3566 3567 3568 3569
static void hardware_disable_all_nolock(void)
{
	BUG_ON(!kvm_usage_count);

	kvm_usage_count--;
	if (!kvm_usage_count)
3570
		on_each_cpu(hardware_disable_nolock, NULL, 1);
3571 3572 3573 3574
}

static void hardware_disable_all(void)
{
3575
	raw_spin_lock(&kvm_count_lock);
3576
	hardware_disable_all_nolock();
3577
	raw_spin_unlock(&kvm_count_lock);
3578 3579 3580 3581 3582 3583
}

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

3584
	raw_spin_lock(&kvm_count_lock);
3585 3586 3587 3588

	kvm_usage_count++;
	if (kvm_usage_count == 1) {
		atomic_set(&hardware_enable_failed, 0);
3589
		on_each_cpu(hardware_enable_nolock, NULL, 1);
3590 3591 3592 3593 3594 3595 3596

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

3597
	raw_spin_unlock(&kvm_count_lock);
3598 3599 3600 3601

	return r;
}

3602
static int kvm_reboot(struct notifier_block *notifier, unsigned long val,
M
Mike Day 已提交
3603
		      void *v)
3604
{
3605 3606 3607 3608 3609 3610
	/*
	 * 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
3611
	pr_info("kvm: exiting hardware virtualization\n");
3612
	kvm_rebooting = true;
3613
	on_each_cpu(hardware_disable_nolock, NULL, 1);
3614 3615 3616 3617 3618 3619 3620 3621
	return NOTIFY_OK;
}

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

M
Marcelo Tosatti 已提交
3622
static void kvm_io_bus_destroy(struct kvm_io_bus *bus)
3623 3624 3625 3626
{
	int i;

	for (i = 0; i < bus->dev_count; i++) {
3627
		struct kvm_io_device *pos = bus->range[i].dev;
3628 3629 3630

		kvm_iodevice_destructor(pos);
	}
M
Marcelo Tosatti 已提交
3631
	kfree(bus);
3632 3633
}

3634
static inline int kvm_io_bus_cmp(const struct kvm_io_range *r1,
X
Xiubo Li 已提交
3635
				 const struct kvm_io_range *r2)
3636
{
J
Jason Wang 已提交
3637 3638 3639 3640
	gpa_t addr1 = r1->addr;
	gpa_t addr2 = r2->addr;

	if (addr1 < addr2)
3641
		return -1;
J
Jason Wang 已提交
3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653

	/* 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)
3654
		return 1;
J
Jason Wang 已提交
3655

3656 3657 3658
	return 0;
}

3659 3660
static int kvm_io_bus_sort_cmp(const void *p1, const void *p2)
{
3661
	return kvm_io_bus_cmp(p1, p2);
3662 3663
}

G
Geoff Levand 已提交
3664
static int kvm_io_bus_get_first_dev(struct kvm_io_bus *bus,
3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681
			     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;

3682
	while (off > 0 && kvm_io_bus_cmp(&key, &bus->range[off-1]) == 0)
3683 3684 3685 3686 3687
		off--;

	return off;
}

3688
static int __kvm_io_bus_write(struct kvm_vcpu *vcpu, struct kvm_io_bus *bus,
C
Cornelia Huck 已提交
3689 3690 3691 3692 3693 3694 3695 3696 3697
			      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 &&
3698
		kvm_io_bus_cmp(range, &bus->range[idx]) == 0) {
3699
		if (!kvm_iodevice_write(vcpu, bus->range[idx].dev, range->addr,
C
Cornelia Huck 已提交
3700 3701 3702 3703 3704 3705 3706 3707
					range->len, val))
			return idx;
		idx++;
	}

	return -EOPNOTSUPP;
}

3708
/* kvm_io_bus_write - called under kvm->slots_lock */
3709
int kvm_io_bus_write(struct kvm_vcpu *vcpu, enum kvm_bus bus_idx, gpa_t addr,
3710
		     int len, const void *val)
3711
{
3712
	struct kvm_io_bus *bus;
3713
	struct kvm_io_range range;
C
Cornelia Huck 已提交
3714
	int r;
3715 3716 3717 3718 3719

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};
3720

3721
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
3722 3723
	if (!bus)
		return -ENOMEM;
3724
	r = __kvm_io_bus_write(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
3725 3726
	return r < 0 ? r : 0;
}
L
Leo Yan 已提交
3727
EXPORT_SYMBOL_GPL(kvm_io_bus_write);
C
Cornelia Huck 已提交
3728 3729

/* kvm_io_bus_write_cookie - called under kvm->slots_lock */
3730 3731
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
3732 3733 3734 3735 3736 3737 3738 3739 3740
{
	struct kvm_io_bus *bus;
	struct kvm_io_range range;

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

3741
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
3742 3743
	if (!bus)
		return -ENOMEM;
C
Cornelia Huck 已提交
3744 3745 3746

	/* First try the device referenced by cookie. */
	if ((cookie >= 0) && (cookie < bus->dev_count) &&
3747
	    (kvm_io_bus_cmp(&range, &bus->range[cookie]) == 0))
3748
		if (!kvm_iodevice_write(vcpu, bus->range[cookie].dev, addr, len,
C
Cornelia Huck 已提交
3749 3750 3751 3752 3753 3754 3755
					val))
			return cookie;

	/*
	 * cookie contained garbage; fall back to search and return the
	 * correct cookie value.
	 */
3756
	return __kvm_io_bus_write(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
3757 3758
}

3759 3760
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
3761 3762 3763 3764
{
	int idx;

	idx = kvm_io_bus_get_first_dev(bus, range->addr, range->len);
3765 3766 3767 3768
	if (idx < 0)
		return -EOPNOTSUPP;

	while (idx < bus->dev_count &&
3769
		kvm_io_bus_cmp(range, &bus->range[idx]) == 0) {
3770
		if (!kvm_iodevice_read(vcpu, bus->range[idx].dev, range->addr,
C
Cornelia Huck 已提交
3771 3772
				       range->len, val))
			return idx;
3773 3774 3775
		idx++;
	}

3776 3777
	return -EOPNOTSUPP;
}
3778

3779
/* kvm_io_bus_read - called under kvm->slots_lock */
3780
int kvm_io_bus_read(struct kvm_vcpu *vcpu, enum kvm_bus bus_idx, gpa_t addr,
M
Marcelo Tosatti 已提交
3781
		    int len, void *val)
3782
{
3783
	struct kvm_io_bus *bus;
3784
	struct kvm_io_range range;
C
Cornelia Huck 已提交
3785
	int r;
3786 3787 3788 3789 3790

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};
M
Marcelo Tosatti 已提交
3791

3792
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
3793 3794
	if (!bus)
		return -ENOMEM;
3795
	r = __kvm_io_bus_read(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
3796 3797
	return r < 0 ? r : 0;
}
3798

3799
/* Caller must hold slots_lock. */
3800 3801
int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
			    int len, struct kvm_io_device *dev)
3802
{
3803
	int i;
M
Marcelo Tosatti 已提交
3804
	struct kvm_io_bus *new_bus, *bus;
3805
	struct kvm_io_range range;
3806

3807
	bus = kvm_get_bus(kvm, bus_idx);
3808 3809 3810
	if (!bus)
		return -ENOMEM;

3811 3812
	/* exclude ioeventfd which is limited by maximum fd */
	if (bus->dev_count - bus->ioeventfd_count > NR_IOBUS_DEVS - 1)
3813
		return -ENOSPC;
3814

3815
	new_bus = kmalloc(struct_size(bus, range, bus->dev_count + 1),
3816
			  GFP_KERNEL_ACCOUNT);
M
Marcelo Tosatti 已提交
3817 3818
	if (!new_bus)
		return -ENOMEM;
3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834

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

	for (i = 0; i < bus->dev_count; i++)
		if (kvm_io_bus_cmp(&bus->range[i], &range) > 0)
			break;

	memcpy(new_bus, bus, sizeof(*bus) + i * sizeof(struct kvm_io_range));
	new_bus->dev_count++;
	new_bus->range[i] = range;
	memcpy(new_bus->range + i + 1, bus->range + i,
		(bus->dev_count - i) * sizeof(struct kvm_io_range));
M
Marcelo Tosatti 已提交
3835 3836 3837
	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
	synchronize_srcu_expedited(&kvm->srcu);
	kfree(bus);
3838 3839 3840 3841

	return 0;
}

3842
/* Caller must hold slots_lock. */
3843 3844
void kvm_io_bus_unregister_dev(struct kvm *kvm, enum kvm_bus bus_idx,
			       struct kvm_io_device *dev)
3845
{
3846
	int i;
M
Marcelo Tosatti 已提交
3847
	struct kvm_io_bus *new_bus, *bus;
3848

3849
	bus = kvm_get_bus(kvm, bus_idx);
3850
	if (!bus)
3851
		return;
3852

3853 3854
	for (i = 0; i < bus->dev_count; i++)
		if (bus->range[i].dev == dev) {
3855 3856
			break;
		}
M
Marcelo Tosatti 已提交
3857

3858 3859
	if (i == bus->dev_count)
		return;
3860

3861
	new_bus = kmalloc(struct_size(bus, range, bus->dev_count - 1),
3862
			  GFP_KERNEL_ACCOUNT);
3863 3864 3865 3866
	if (!new_bus)  {
		pr_err("kvm: failed to shrink bus, removing it completely\n");
		goto broken;
	}
3867 3868 3869 3870 3871

	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
3872

3873
broken:
M
Marcelo Tosatti 已提交
3874 3875 3876
	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
	synchronize_srcu_expedited(&kvm->srcu);
	kfree(bus);
3877
	return;
3878 3879
}

3880 3881 3882 3883 3884 3885 3886 3887 3888 3889
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);
3890 3891
	if (!bus)
		goto out_unlock;
3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905

	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);

3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917
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.
	 */
3918
	if (!refcount_inc_not_zero(&stat_data->kvm->users_count))
3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943
		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;

3944
	*val = *(ulong *)((void *)stat_data->kvm + stat_data->offset);
3945 3946 3947 3948

	return 0;
}

3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960
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;
}

3961 3962 3963 3964
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,
3965
				vm_stat_clear_per_vm, "%llu\n");
3966 3967 3968 3969 3970 3971 3972 3973
}

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,
3974
	.llseek  = no_llseek,
3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985
};

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)
3986
		*val += *(u64 *)((void *)vcpu + stat_data->offset);
3987 3988 3989 3990

	return 0;
}

3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005
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;
}

4006 4007 4008 4009
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,
4010
				 vcpu_stat_clear_per_vm, "%llu\n");
4011 4012 4013 4014 4015 4016 4017 4018
}

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,
4019
	.llseek  = no_llseek,
4020 4021 4022 4023 4024 4025 4026
};

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,
};

4027
static int vm_stat_get(void *_offset, u64 *val)
4028 4029 4030
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;
4031 4032
	struct kvm_stat_data stat_tmp = {.offset = offset};
	u64 tmp_val;
4033

4034
	*val = 0;
4035
	spin_lock(&kvm_lock);
4036 4037 4038 4039 4040
	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;
	}
4041
	spin_unlock(&kvm_lock);
4042
	return 0;
4043 4044
}

4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064
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");
4065

4066
static int vcpu_stat_get(void *_offset, u64 *val)
A
Avi Kivity 已提交
4067 4068 4069
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;
4070 4071
	struct kvm_stat_data stat_tmp = {.offset = offset};
	u64 tmp_val;
A
Avi Kivity 已提交
4072

4073
	*val = 0;
4074
	spin_lock(&kvm_lock);
4075 4076 4077 4078 4079
	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;
	}
4080
	spin_unlock(&kvm_lock);
4081
	return 0;
A
Avi Kivity 已提交
4082 4083
}

4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104
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");
4105

4106
static const struct file_operations *stat_fops[] = {
4107 4108 4109
	[KVM_STAT_VCPU] = &vcpu_stat_fops,
	[KVM_STAT_VM]   = &vm_stat_fops,
};
A
Avi Kivity 已提交
4110

4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129
static void kvm_uevent_notify_change(unsigned int type, struct kvm *kvm)
{
	struct kobj_uevent_env *env;
	unsigned long long created, active;

	if (!kvm_dev.this_device || !kvm)
		return;

	spin_lock(&kvm_lock);
	if (type == KVM_EVENT_CREATE_VM) {
		kvm_createvm_count++;
		kvm_active_vms++;
	} else if (type == KVM_EVENT_DESTROY_VM) {
		kvm_active_vms--;
	}
	created = kvm_createvm_count;
	active = kvm_active_vms;
	spin_unlock(&kvm_lock);

4130
	env = kzalloc(sizeof(*env), GFP_KERNEL_ACCOUNT);
4131 4132 4133 4134 4135 4136
	if (!env)
		return;

	add_uevent_var(env, "CREATED=%llu", created);
	add_uevent_var(env, "COUNT=%llu", active);

4137
	if (type == KVM_EVENT_CREATE_VM) {
4138
		add_uevent_var(env, "EVENT=create");
4139 4140
		kvm->userspace_pid = task_pid_nr(current);
	} else if (type == KVM_EVENT_DESTROY_VM) {
4141
		add_uevent_var(env, "EVENT=destroy");
4142 4143
	}
	add_uevent_var(env, "PID=%d", kvm->userspace_pid);
4144

4145
	if (!IS_ERR_OR_NULL(kvm->debugfs_dentry)) {
4146
		char *tmp, *p = kmalloc(PATH_MAX, GFP_KERNEL_ACCOUNT);
4147 4148 4149 4150 4151 4152

		if (p) {
			tmp = dentry_path_raw(kvm->debugfs_dentry, p, PATH_MAX);
			if (!IS_ERR(tmp))
				add_uevent_var(env, "STATS_PATH=%s", tmp);
			kfree(p);
4153 4154 4155 4156 4157 4158 4159 4160
		}
	}
	/* no need for checks, since we are adding at most only 5 keys */
	env->envp[env->envp_idx++] = NULL;
	kobject_uevent_env(&kvm_dev.this_device->kobj, KOBJ_CHANGE, env->envp);
	kfree(env);
}

4161
static void kvm_init_debug(void)
A
Avi Kivity 已提交
4162 4163 4164
{
	struct kvm_stats_debugfs_item *p;

4165
	kvm_debugfs_dir = debugfs_create_dir("kvm", NULL);
4166

4167 4168
	kvm_debugfs_num_entries = 0;
	for (p = debugfs_entries; p->name; ++p, kvm_debugfs_num_entries++) {
4169 4170 4171
		debugfs_create_file(p->name, 0644, kvm_debugfs_dir,
				    (void *)(long)p->offset,
				    stat_fops[p->kind]);
4172
	}
A
Avi Kivity 已提交
4173 4174
}

4175
static int kvm_suspend(void)
4176
{
4177
	if (kvm_usage_count)
4178
		hardware_disable_nolock(NULL);
4179 4180 4181
	return 0;
}

4182
static void kvm_resume(void)
4183
{
4184
	if (kvm_usage_count) {
4185 4186 4187
#ifdef CONFIG_LOCKDEP
		WARN_ON(lockdep_is_held(&kvm_count_lock));
#endif
4188
		hardware_enable_nolock(NULL);
4189
	}
4190 4191
}

4192
static struct syscore_ops kvm_syscore_ops = {
4193 4194 4195 4196
	.suspend = kvm_suspend,
	.resume = kvm_resume,
};

4197 4198 4199 4200 4201 4202 4203 4204 4205
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);
4206

4207 4208
	if (vcpu->preempted)
		vcpu->preempted = false;
4209

R
Radim Krčmář 已提交
4210 4211
	kvm_arch_sched_in(vcpu, cpu);

4212
	kvm_arch_vcpu_load(vcpu, cpu);
4213 4214 4215 4216 4217 4218 4219
}

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

4220 4221
	if (current->state == TASK_RUNNING)
		vcpu->preempted = true;
4222
	kvm_arch_vcpu_put(vcpu);
4223 4224
}

4225
int kvm_init(void *opaque, unsigned vcpu_size, unsigned vcpu_align,
4226
		  struct module *module)
A
Avi Kivity 已提交
4227 4228
{
	int r;
Y
Yang, Sheng 已提交
4229
	int cpu;
A
Avi Kivity 已提交
4230

4231 4232
	r = kvm_arch_init(opaque);
	if (r)
4233
		goto out_fail;
4234

4235 4236 4237 4238
	/*
	 * 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
4239 4240
	 * kvm_arch_init must be called before kvm_irqfd_init to avoid creating
	 * conflicts in case kvm is already setup for another implementation.
4241
	 */
P
Paolo Bonzini 已提交
4242 4243 4244
	r = kvm_irqfd_init();
	if (r)
		goto out_irqfd;
4245

4246
	if (!zalloc_cpumask_var(&cpus_hardware_enabled, GFP_KERNEL)) {
4247 4248 4249 4250
		r = -ENOMEM;
		goto out_free_0;
	}

4251
	r = kvm_arch_hardware_setup();
A
Avi Kivity 已提交
4252
	if (r < 0)
4253
		goto out_free_0a;
A
Avi Kivity 已提交
4254

Y
Yang, Sheng 已提交
4255 4256
	for_each_online_cpu(cpu) {
		smp_call_function_single(cpu,
4257
				kvm_arch_check_processor_compat,
4258
				&r, 1);
Y
Yang, Sheng 已提交
4259
		if (r < 0)
4260
			goto out_free_1;
Y
Yang, Sheng 已提交
4261 4262
	}

T
Thomas Gleixner 已提交
4263
	r = cpuhp_setup_state_nocalls(CPUHP_AP_KVM_STARTING, "kvm/cpu:starting",
4264
				      kvm_starting_cpu, kvm_dying_cpu);
A
Avi Kivity 已提交
4265
	if (r)
4266
		goto out_free_2;
A
Avi Kivity 已提交
4267 4268
	register_reboot_notifier(&kvm_reboot_notifier);

4269
	/* A kmem cache lets us meet the alignment requirements of fx_save. */
4270 4271
	if (!vcpu_align)
		vcpu_align = __alignof__(struct kvm_vcpu);
4272 4273 4274 4275 4276 4277
	kvm_vcpu_cache =
		kmem_cache_create_usercopy("kvm_vcpu", vcpu_size, vcpu_align,
					   SLAB_ACCOUNT,
					   offsetof(struct kvm_vcpu, arch),
					   sizeof_field(struct kvm_vcpu, arch),
					   NULL);
4278 4279
	if (!kvm_vcpu_cache) {
		r = -ENOMEM;
4280
		goto out_free_3;
4281 4282
	}

4283 4284 4285 4286
	r = kvm_async_pf_init();
	if (r)
		goto out_free;

A
Avi Kivity 已提交
4287
	kvm_chardev_ops.owner = module;
4288 4289
	kvm_vm_fops.owner = module;
	kvm_vcpu_fops.owner = module;
A
Avi Kivity 已提交
4290 4291 4292

	r = misc_register(&kvm_dev);
	if (r) {
X
Xiubo Li 已提交
4293
		pr_err("kvm: misc device register failed\n");
4294
		goto out_unreg;
A
Avi Kivity 已提交
4295 4296
	}

4297 4298
	register_syscore_ops(&kvm_syscore_ops);

4299 4300 4301
	kvm_preempt_ops.sched_in = kvm_sched_in;
	kvm_preempt_ops.sched_out = kvm_sched_out;

4302
	kvm_init_debug();
4303

P
Paolo Bonzini 已提交
4304 4305 4306
	r = kvm_vfio_ops_init();
	WARN_ON(r);

4307
	return 0;
A
Avi Kivity 已提交
4308

4309 4310
out_unreg:
	kvm_async_pf_deinit();
A
Avi Kivity 已提交
4311
out_free:
4312
	kmem_cache_destroy(kvm_vcpu_cache);
4313
out_free_3:
A
Avi Kivity 已提交
4314
	unregister_reboot_notifier(&kvm_reboot_notifier);
4315
	cpuhp_remove_state_nocalls(CPUHP_AP_KVM_STARTING);
4316 4317
out_free_2:
out_free_1:
4318
	kvm_arch_hardware_unsetup();
4319 4320
out_free_0a:
	free_cpumask_var(cpus_hardware_enabled);
4321
out_free_0:
4322
	kvm_irqfd_exit();
P
Paolo Bonzini 已提交
4323
out_irqfd:
4324 4325
	kvm_arch_exit();
out_fail:
A
Avi Kivity 已提交
4326 4327
	return r;
}
4328
EXPORT_SYMBOL_GPL(kvm_init);
A
Avi Kivity 已提交
4329

4330
void kvm_exit(void)
A
Avi Kivity 已提交
4331
{
4332
	debugfs_remove_recursive(kvm_debugfs_dir);
A
Avi Kivity 已提交
4333
	misc_deregister(&kvm_dev);
4334
	kmem_cache_destroy(kvm_vcpu_cache);
4335
	kvm_async_pf_deinit();
4336
	unregister_syscore_ops(&kvm_syscore_ops);
A
Avi Kivity 已提交
4337
	unregister_reboot_notifier(&kvm_reboot_notifier);
4338
	cpuhp_remove_state_nocalls(CPUHP_AP_KVM_STARTING);
4339
	on_each_cpu(hardware_disable_nolock, NULL, 1);
4340
	kvm_arch_hardware_unsetup();
4341
	kvm_arch_exit();
4342
	kvm_irqfd_exit();
4343
	free_cpumask_var(cpus_hardware_enabled);
4344
	kvm_vfio_ops_exit();
A
Avi Kivity 已提交
4345
}
4346
EXPORT_SYMBOL_GPL(kvm_exit);