kvm_main.c 93.8 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>
A
Avi Kivity 已提交
54

A
Avi Kivity 已提交
55 56
#include <asm/processor.h>
#include <asm/io.h>
57
#include <asm/ioctl.h>
58
#include <linux/uaccess.h>
59
#include <asm/pgtable.h>
A
Avi Kivity 已提交
60

61
#include "coalesced_mmio.h"
62
#include "async_pf.h"
P
Paolo Bonzini 已提交
63
#include "vfio.h"
64

65 66 67
#define CREATE_TRACE_POINTS
#include <trace/events/kvm.h>

68 69 70
/* Worst case buffer size needed for holding an integer. */
#define ITOA_MAX_LEN 12

A
Avi Kivity 已提交
71 72 73
MODULE_AUTHOR("Qumranet");
MODULE_LICENSE("GPL");

74
/* Architectures should define their poll value according to the halt latency */
75
unsigned int halt_poll_ns = KVM_HALT_POLL_NS_DEFAULT;
76
module_param(halt_poll_ns, uint, S_IRUGO | S_IWUSR);
77
EXPORT_SYMBOL_GPL(halt_poll_ns);
78

W
Wanpeng Li 已提交
79
/* Default doubles per-vcpu halt_poll_ns. */
80
unsigned int halt_poll_ns_grow = 2;
81
module_param(halt_poll_ns_grow, uint, S_IRUGO | S_IWUSR);
82
EXPORT_SYMBOL_GPL(halt_poll_ns_grow);
W
Wanpeng Li 已提交
83 84

/* Default resets per-vcpu halt_poll_ns . */
85
unsigned int halt_poll_ns_shrink;
86
module_param(halt_poll_ns_shrink, uint, S_IRUGO | S_IWUSR);
87
EXPORT_SYMBOL_GPL(halt_poll_ns_shrink);
W
Wanpeng Li 已提交
88

89 90 91
/*
 * Ordering of locks:
 *
92
 *	kvm->lock --> kvm->slots_lock --> kvm->irq_lock
93 94
 */

95
DEFINE_SPINLOCK(kvm_lock);
96
static DEFINE_RAW_SPINLOCK(kvm_count_lock);
97
LIST_HEAD(vm_list);
98

99
static cpumask_var_t cpus_hardware_enabled;
100
static int kvm_usage_count;
101
static atomic_t hardware_enable_failed;
102

103 104
struct kmem_cache *kvm_vcpu_cache;
EXPORT_SYMBOL_GPL(kvm_vcpu_cache);
A
Avi Kivity 已提交
105

106 107
static __read_mostly struct preempt_ops kvm_preempt_ops;

108
struct dentry *kvm_debugfs_dir;
109
EXPORT_SYMBOL_GPL(kvm_debugfs_dir);
A
Avi Kivity 已提交
110

111 112 113
static int kvm_debugfs_num_entries;
static const struct file_operations *stat_fops_per_vm[];

A
Avi Kivity 已提交
114 115
static long kvm_vcpu_ioctl(struct file *file, unsigned int ioctl,
			   unsigned long arg);
116
#ifdef CONFIG_KVM_COMPAT
117 118 119
static long kvm_vcpu_compat_ioctl(struct file *file, unsigned int ioctl,
				  unsigned long arg);
#endif
120 121
static int hardware_enable_all(void);
static void hardware_disable_all(void);
A
Avi Kivity 已提交
122

M
Marcelo Tosatti 已提交
123
static void kvm_io_bus_destroy(struct kvm_io_bus *bus);
124

D
Dan Williams 已提交
125
static void kvm_release_pfn_dirty(kvm_pfn_t pfn);
126
static void mark_page_dirty_in_slot(struct kvm_memory_slot *memslot, gfn_t gfn);
M
Marcelo Tosatti 已提交
127

128
__visible bool kvm_rebooting;
129
EXPORT_SYMBOL_GPL(kvm_rebooting);
130

131 132
static bool largepages_enabled = true;

D
Dan Williams 已提交
133
bool kvm_is_reserved_pfn(kvm_pfn_t pfn)
B
Ben-Ami Yassour 已提交
134
{
135
	if (pfn_valid(pfn))
136
		return PageReserved(pfn_to_page(pfn));
B
Ben-Ami Yassour 已提交
137 138 139 140

	return true;
}

A
Avi Kivity 已提交
141 142 143
/*
 * Switches to specified vcpu, until a matching vcpu_put()
 */
144
int vcpu_load(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
145
{
146 147
	int cpu;

148 149
	if (mutex_lock_killable(&vcpu->mutex))
		return -EINTR;
150 151
	cpu = get_cpu();
	preempt_notifier_register(&vcpu->preempt_notifier);
152
	kvm_arch_vcpu_load(vcpu, cpu);
153
	put_cpu();
154
	return 0;
A
Avi Kivity 已提交
155
}
156
EXPORT_SYMBOL_GPL(vcpu_load);
A
Avi Kivity 已提交
157

158
void vcpu_put(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
159
{
160
	preempt_disable();
161
	kvm_arch_vcpu_put(vcpu);
162 163
	preempt_notifier_unregister(&vcpu->preempt_notifier);
	preempt_enable();
A
Avi Kivity 已提交
164 165
	mutex_unlock(&vcpu->mutex);
}
166
EXPORT_SYMBOL_GPL(vcpu_put);
A
Avi Kivity 已提交
167

168 169 170 171
static void ack_flush(void *_completed)
{
}

172
bool kvm_make_all_cpus_request(struct kvm *kvm, unsigned int req)
173
{
174
	int i, cpu, me;
175 176
	cpumask_var_t cpus;
	bool called = true;
177 178
	struct kvm_vcpu *vcpu;

179
	zalloc_cpumask_var(&cpus, GFP_ATOMIC);
180

181
	me = get_cpu();
182
	kvm_for_each_vcpu(i, vcpu, kvm) {
183
		kvm_make_request(req, vcpu);
184
		cpu = vcpu->cpu;
185

186 187
		/* Set ->requests bit before we read ->mode. */
		smp_mb__after_atomic();
188 189 190

		if (cpus != NULL && cpu != -1 && cpu != me &&
		      kvm_vcpu_exiting_guest_mode(vcpu) != OUTSIDE_GUEST_MODE)
191
			cpumask_set_cpu(cpu, cpus);
192
	}
193 194 195 196 197 198
	if (unlikely(cpus == NULL))
		smp_call_function_many(cpu_online_mask, ack_flush, NULL, 1);
	else if (!cpumask_empty(cpus))
		smp_call_function_many(cpus, ack_flush, NULL, 1);
	else
		called = false;
199
	put_cpu();
200
	free_cpumask_var(cpus);
201
	return called;
202 203
}

204
#ifndef CONFIG_HAVE_KVM_ARCH_TLB_FLUSH_ALL
205
void kvm_flush_remote_tlbs(struct kvm *kvm)
206
{
207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223
	/*
	 * 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.
	 */
224
	if (kvm_make_all_cpus_request(kvm, KVM_REQ_TLB_FLUSH))
225
		++kvm->stat.remote_tlb_flush;
226
	cmpxchg(&kvm->tlbs_dirty, dirty_count, 0);
227
}
228
EXPORT_SYMBOL_GPL(kvm_flush_remote_tlbs);
229
#endif
230

231 232
void kvm_reload_remote_mmus(struct kvm *kvm)
{
233
	kvm_make_all_cpus_request(kvm, KVM_REQ_MMU_RELOAD);
234
}
235

R
Rusty Russell 已提交
236 237 238 239 240 241 242 243 244
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;
245
	vcpu->pid = NULL;
246
	init_swait_queue_head(&vcpu->wq);
247
	kvm_async_pf_vcpu_init(vcpu);
R
Rusty Russell 已提交
248

249 250 251
	vcpu->pre_pcpu = -1;
	INIT_LIST_HEAD(&vcpu->blocked_vcpu_list);

R
Rusty Russell 已提交
252 253 254 255 256 257 258
	page = alloc_page(GFP_KERNEL | __GFP_ZERO);
	if (!page) {
		r = -ENOMEM;
		goto fail;
	}
	vcpu->run = page_address(page);

259 260
	kvm_vcpu_set_in_spin_loop(vcpu, false);
	kvm_vcpu_set_dy_eligible(vcpu, false);
261
	vcpu->preempted = false;
262

263
	r = kvm_arch_vcpu_init(vcpu);
R
Rusty Russell 已提交
264
	if (r < 0)
265
		goto fail_free_run;
R
Rusty Russell 已提交
266 267 268 269 270
	return 0;

fail_free_run:
	free_page((unsigned long)vcpu->run);
fail:
271
	return r;
R
Rusty Russell 已提交
272 273 274 275 276
}
EXPORT_SYMBOL_GPL(kvm_vcpu_init);

void kvm_vcpu_uninit(struct kvm_vcpu *vcpu)
{
277
	put_pid(vcpu->pid);
278
	kvm_arch_vcpu_uninit(vcpu);
R
Rusty Russell 已提交
279 280 281 282
	free_page((unsigned long)vcpu->run);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_uninit);

283 284 285 286 287 288 289 290 291 292 293
#if defined(CONFIG_MMU_NOTIFIER) && defined(KVM_ARCH_WANT_MMU_NOTIFIER)
static inline struct kvm *mmu_notifier_to_kvm(struct mmu_notifier *mn)
{
	return container_of(mn, struct kvm, mmu_notifier);
}

static void kvm_mmu_notifier_invalidate_page(struct mmu_notifier *mn,
					     struct mm_struct *mm,
					     unsigned long address)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
294
	int need_tlb_flush, idx;
295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313

	/*
	 * When ->invalidate_page runs, the linux pte has been zapped
	 * already but the page is still allocated until
	 * ->invalidate_page returns. So if we increase the sequence
	 * here the kvm page fault will notice if the spte can't be
	 * established because the page is going to be freed. If
	 * instead the kvm page fault establishes the spte before
	 * ->invalidate_page runs, kvm_unmap_hva will release it
	 * before returning.
	 *
	 * The sequence increase only need to be seen at spin_unlock
	 * time, and not at spin_lock time.
	 *
	 * Increasing the sequence after the spin_unlock would be
	 * unsafe because the kvm page fault could then establish the
	 * pte after kvm_unmap_hva returned, without noticing the page
	 * is going to be freed.
	 */
314
	idx = srcu_read_lock(&kvm->srcu);
315
	spin_lock(&kvm->mmu_lock);
316

317
	kvm->mmu_notifier_seq++;
318
	need_tlb_flush = kvm_unmap_hva(kvm, address) | kvm->tlbs_dirty;
319 320 321 322
	/* we've to flush the tlb before the pages can be freed */
	if (need_tlb_flush)
		kvm_flush_remote_tlbs(kvm);

323
	spin_unlock(&kvm->mmu_lock);
324 325 326

	kvm_arch_mmu_notifier_invalidate_page(kvm, address);

327
	srcu_read_unlock(&kvm->srcu, idx);
328 329
}

330 331 332 333 334 335
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);
336
	int idx;
337

338
	idx = srcu_read_lock(&kvm->srcu);
339 340 341 342
	spin_lock(&kvm->mmu_lock);
	kvm->mmu_notifier_seq++;
	kvm_set_spte_hva(kvm, address, pte);
	spin_unlock(&kvm->mmu_lock);
343
	srcu_read_unlock(&kvm->srcu, idx);
344 345
}

346 347 348 349 350 351
static void kvm_mmu_notifier_invalidate_range_start(struct mmu_notifier *mn,
						    struct mm_struct *mm,
						    unsigned long start,
						    unsigned long end)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
352
	int need_tlb_flush = 0, idx;
353

354
	idx = srcu_read_lock(&kvm->srcu);
355 356 357 358 359 360 361
	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++;
362
	need_tlb_flush = kvm_unmap_hva_range(kvm, start, end);
363
	need_tlb_flush |= kvm->tlbs_dirty;
364 365 366
	/* we've to flush the tlb before the pages can be freed */
	if (need_tlb_flush)
		kvm_flush_remote_tlbs(kvm);
367 368 369

	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);
370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385
}

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

	spin_lock(&kvm->mmu_lock);
	/*
	 * This sequence increase will notify the kvm page fault that
	 * the page that is going to be mapped in the spte could have
	 * been freed.
	 */
	kvm->mmu_notifier_seq++;
386
	smp_wmb();
387 388
	/*
	 * The above sequence increase must be visible before the
389 390
	 * below count decrease, which is ensured by the smp_wmb above
	 * in conjunction with the smp_rmb in mmu_notifier_retry().
391 392 393 394 395 396 397 398 399
	 */
	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
400 401
					      unsigned long start,
					      unsigned long end)
402 403
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
404
	int young, idx;
405

406
	idx = srcu_read_lock(&kvm->srcu);
407 408
	spin_lock(&kvm->mmu_lock);

A
Andres Lagar-Cavilla 已提交
409
	young = kvm_age_hva(kvm, start, end);
410 411 412
	if (young)
		kvm_flush_remote_tlbs(kvm);

413 414 415
	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);

416 417 418
	return young;
}

419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448
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
449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464
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;
}

465 466 467 468
static void kvm_mmu_notifier_release(struct mmu_notifier *mn,
				     struct mm_struct *mm)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
469 470 471
	int idx;

	idx = srcu_read_lock(&kvm->srcu);
472
	kvm_arch_flush_shadow_all(kvm);
473
	srcu_read_unlock(&kvm->srcu, idx);
474 475
}

476 477 478 479 480
static const struct mmu_notifier_ops kvm_mmu_notifier_ops = {
	.invalidate_page	= kvm_mmu_notifier_invalidate_page,
	.invalidate_range_start	= kvm_mmu_notifier_invalidate_range_start,
	.invalidate_range_end	= kvm_mmu_notifier_invalidate_range_end,
	.clear_flush_young	= kvm_mmu_notifier_clear_flush_young,
481
	.clear_young		= kvm_mmu_notifier_clear_young,
A
Andrea Arcangeli 已提交
482
	.test_young		= kvm_mmu_notifier_test_young,
483
	.change_pte		= kvm_mmu_notifier_change_pte,
484
	.release		= kvm_mmu_notifier_release,
485
};
486 487 488 489 490 491 492 493 494 495 496 497 498 499

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;
}

500 501
#endif /* CONFIG_MMU_NOTIFIER && KVM_ARCH_WANT_MMU_NOTIFIER */

502
static struct kvm_memslots *kvm_alloc_memslots(void)
503 504
{
	int i;
505
	struct kvm_memslots *slots;
506

507 508 509 510
	slots = kvm_kvzalloc(sizeof(struct kvm_memslots));
	if (!slots)
		return NULL;

511
	for (i = 0; i < KVM_MEM_SLOTS_NUM; i++)
512
		slots->id_to_index[i] = slots->memslots[i].id = i;
513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550

	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);
551 552
}

553 554 555 556 557 558 559 560 561
static void kvm_destroy_vm_debugfs(struct kvm *kvm)
{
	int i;

	if (!kvm->debugfs_dentry)
		return;

	debugfs_remove_recursive(kvm->debugfs_dentry);

562 563 564 565 566
	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);
	}
567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597
}

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

	if (!debugfs_initialized())
		return 0;

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

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

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

		stat_data->kvm = kvm;
		stat_data->offset = p->offset;
		kvm->debugfs_stat_data[p - debugfs_entries] = stat_data;
598
		if (!debugfs_create_file(p->name, 0644,
599 600 601 602 603 604 605 606
					 kvm->debugfs_dentry,
					 stat_data,
					 stat_fops_per_vm[p->kind]))
			return -ENOMEM;
	}
	return 0;
}

607
static struct kvm *kvm_create_vm(unsigned long type)
A
Avi Kivity 已提交
608
{
609 610
	int r, i;
	struct kvm *kvm = kvm_arch_alloc_vm();
A
Avi Kivity 已提交
611

612 613 614
	if (!kvm)
		return ERR_PTR(-ENOMEM);

615
	spin_lock_init(&kvm->mmu_lock);
V
Vegard Nossum 已提交
616
	mmgrab(current->mm);
617 618 619 620 621
	kvm->mm = current->mm;
	kvm_eventfd_init(kvm);
	mutex_init(&kvm->lock);
	mutex_init(&kvm->irq_lock);
	mutex_init(&kvm->slots_lock);
622
	refcount_set(&kvm->users_count, 1);
623 624
	INIT_LIST_HEAD(&kvm->devices);

625
	r = kvm_arch_init_vm(kvm, type);
626
	if (r)
627
		goto out_err_no_disable;
628 629 630

	r = hardware_enable_all();
	if (r)
631
		goto out_err_no_disable;
632

633
#ifdef CONFIG_HAVE_KVM_IRQFD
634
	INIT_HLIST_HEAD(&kvm->irq_ack_notifier_list);
635
#endif
A
Avi Kivity 已提交
636

637 638
	BUILD_BUG_ON(KVM_MEM_SLOTS_NUM > SHRT_MAX);

639
	r = -ENOMEM;
640
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++) {
641 642
		struct kvm_memslots *slots = kvm_alloc_memslots();
		if (!slots)
643
			goto out_err_no_srcu;
644 645 646 647 648 649 650
		/*
		 * Generations must be different for each address space.
		 * Init kvm generation close to the maximum to easily test the
		 * code of handling generation number wrap-around.
		 */
		slots->generation = i * 2 - 150;
		rcu_assign_pointer(kvm->memslots[i], slots);
651
	}
652

653
	if (init_srcu_struct(&kvm->srcu))
654 655 656
		goto out_err_no_srcu;
	if (init_srcu_struct(&kvm->irq_srcu))
		goto out_err_no_irq_srcu;
M
Marcelo Tosatti 已提交
657 658 659
	for (i = 0; i < KVM_NR_BUSES; i++) {
		kvm->buses[i] = kzalloc(sizeof(struct kvm_io_bus),
					GFP_KERNEL);
660
		if (!kvm->buses[i])
M
Marcelo Tosatti 已提交
661 662
			goto out_err;
	}
663

664 665 666 667
	r = kvm_init_mmu_notifier(kvm);
	if (r)
		goto out_err;

668
	spin_lock(&kvm_lock);
669
	list_add(&kvm->vm_list, &vm_list);
670
	spin_unlock(&kvm_lock);
671

672 673
	preempt_notifier_inc();

674
	return kvm;
675 676

out_err:
677 678
	cleanup_srcu_struct(&kvm->irq_srcu);
out_err_no_irq_srcu:
679
	cleanup_srcu_struct(&kvm->srcu);
680
out_err_no_srcu:
681
	hardware_disable_all();
682
out_err_no_disable:
M
Marcelo Tosatti 已提交
683 684
	for (i = 0; i < KVM_NR_BUSES; i++)
		kfree(kvm->buses[i]);
685 686
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++)
		kvm_free_memslots(kvm, kvm->memslots[i]);
687
	kvm_arch_free_vm(kvm);
688
	mmdrop(current->mm);
689
	return ERR_PTR(r);
690 691
}

692 693 694 695
/*
 * Avoid using vmalloc for a small buffer.
 * Should not be used when the size is statically known.
 */
696
void *kvm_kvzalloc(unsigned long size)
697 698 699 700 701 702 703
{
	if (size > PAGE_SIZE)
		return vzalloc(size);
	else
		return kzalloc(size, GFP_KERNEL);
}

704 705
static void kvm_destroy_devices(struct kvm *kvm)
{
G
Geliang Tang 已提交
706
	struct kvm_device *dev, *tmp;
707

708 709 710 711 712
	/*
	 * 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
713 714
	list_for_each_entry_safe(dev, tmp, &kvm->devices, vm_node) {
		list_del(&dev->vm_node);
715 716 717 718
		dev->ops->destroy(dev);
	}
}

719 720
static void kvm_destroy_vm(struct kvm *kvm)
{
M
Marcelo Tosatti 已提交
721
	int i;
722 723
	struct mm_struct *mm = kvm->mm;

724
	kvm_destroy_vm_debugfs(kvm);
725
	kvm_arch_sync_events(kvm);
726
	spin_lock(&kvm_lock);
727
	list_del(&kvm->vm_list);
728
	spin_unlock(&kvm_lock);
729
	kvm_free_irq_routing(kvm);
730
	for (i = 0; i < KVM_NR_BUSES; i++) {
731 732
		if (kvm->buses[i])
			kvm_io_bus_destroy(kvm->buses[i]);
733 734
		kvm->buses[i] = NULL;
	}
735
	kvm_coalesced_mmio_free(kvm);
736 737
#if defined(CONFIG_MMU_NOTIFIER) && defined(KVM_ARCH_WANT_MMU_NOTIFIER)
	mmu_notifier_unregister(&kvm->mmu_notifier, kvm->mm);
738
#else
739
	kvm_arch_flush_shadow_all(kvm);
740
#endif
741
	kvm_arch_destroy_vm(kvm);
742
	kvm_destroy_devices(kvm);
743 744
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++)
		kvm_free_memslots(kvm, kvm->memslots[i]);
745
	cleanup_srcu_struct(&kvm->irq_srcu);
746 747
	cleanup_srcu_struct(&kvm->srcu);
	kvm_arch_free_vm(kvm);
748
	preempt_notifier_dec();
749
	hardware_disable_all();
750
	mmdrop(mm);
751 752
}

I
Izik Eidus 已提交
753 754
void kvm_get_kvm(struct kvm *kvm)
{
755
	refcount_inc(&kvm->users_count);
I
Izik Eidus 已提交
756 757 758 759 760
}
EXPORT_SYMBOL_GPL(kvm_get_kvm);

void kvm_put_kvm(struct kvm *kvm)
{
761
	if (refcount_dec_and_test(&kvm->users_count))
I
Izik Eidus 已提交
762 763 764 765 766
		kvm_destroy_vm(kvm);
}
EXPORT_SYMBOL_GPL(kvm_put_kvm);


767 768 769 770
static int kvm_vm_release(struct inode *inode, struct file *filp)
{
	struct kvm *kvm = filp->private_data;

G
Gregory Haskins 已提交
771 772
	kvm_irqfd_release(kvm);

I
Izik Eidus 已提交
773
	kvm_put_kvm(kvm);
A
Avi Kivity 已提交
774 775 776
	return 0;
}

777 778
/*
 * Allocation size is twice as large as the actual dirty bitmap size.
779
 * See x86's kvm_vm_ioctl_get_dirty_log() why this is needed.
780
 */
781 782
static int kvm_create_dirty_bitmap(struct kvm_memory_slot *memslot)
{
783
	unsigned long dirty_bytes = 2 * kvm_dirty_bitmap_bytes(memslot);
784

785
	memslot->dirty_bitmap = kvm_kvzalloc(dirty_bytes);
786 787 788 789 790 791
	if (!memslot->dirty_bitmap)
		return -ENOMEM;

	return 0;
}

792
/*
793 794 795 796
 * 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.
797
 */
798 799
static void update_memslots(struct kvm_memslots *slots,
			    struct kvm_memory_slot *new)
800
{
801 802
	int id = new->id;
	int i = slots->id_to_index[id];
803
	struct kvm_memory_slot *mslots = slots->memslots;
804

805
	WARN_ON(mslots[i].id != id);
806
	if (!new->npages) {
807
		WARN_ON(!mslots[i].npages);
808 809 810 811 812 813
		if (mslots[i].npages)
			slots->used_slots--;
	} else {
		if (!mslots[i].npages)
			slots->used_slots++;
	}
814

815
	while (i < KVM_MEM_SLOTS_NUM - 1 &&
816 817 818
	       new->base_gfn <= mslots[i + 1].base_gfn) {
		if (!mslots[i + 1].npages)
			break;
819 820 821 822
		mslots[i] = mslots[i + 1];
		slots->id_to_index[mslots[i].id] = i;
		i++;
	}
823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839

	/*
	 * 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--;
		}
840 841
	} else
		WARN_ON_ONCE(i != slots->used_slots);
842

843 844
	mslots[i] = *new;
	slots->id_to_index[mslots[i].id] = i;
845 846
}

847
static int check_memory_region_flags(const struct kvm_userspace_memory_region *mem)
848
{
X
Xiao Guangrong 已提交
849 850
	u32 valid_flags = KVM_MEM_LOG_DIRTY_PAGES;

851
#ifdef __KVM_HAVE_READONLY_MEM
X
Xiao Guangrong 已提交
852 853 854 855
	valid_flags |= KVM_MEM_READONLY;
#endif

	if (mem->flags & ~valid_flags)
856 857 858 859 860
		return -EINVAL;

	return 0;
}

861
static struct kvm_memslots *install_new_memslots(struct kvm *kvm,
862
		int as_id, struct kvm_memslots *slots)
863
{
864
	struct kvm_memslots *old_memslots = __kvm_memslots(kvm, as_id);
865

866 867 868 869 870 871 872
	/*
	 * Set the low bit in the generation, which disables SPTE caching
	 * until the end of synchronize_srcu_expedited.
	 */
	WARN_ON(old_memslots->generation & 1);
	slots->generation = old_memslots->generation + 1;

873
	rcu_assign_pointer(kvm->memslots[as_id], slots);
874
	synchronize_srcu_expedited(&kvm->srcu);
875

876 877 878 879
	/*
	 * Increment the new memslot generation a second time. This prevents
	 * vm exits that race with memslot updates from caching a memslot
	 * generation that will (potentially) be valid forever.
880 881 882 883 884 885
	 *
	 * Generations must be unique even across address spaces.  We do not need
	 * a global counter for that, instead the generation space is evenly split
	 * across address spaces.  For example, with two address spaces, address
	 * space 0 will use generations 0, 4, 8, ... while * address space 1 will
	 * use generations 2, 6, 10, 14, ...
886
	 */
887
	slots->generation += KVM_ADDRESS_SPACE_NUM * 2 - 1;
888

889
	kvm_arch_memslots_updated(kvm, slots);
890 891

	return old_memslots;
892 893
}

A
Avi Kivity 已提交
894 895 896 897 898
/*
 * Allocate some memory and give it an address in the guest physical address
 * space.
 *
 * Discontiguous memory is allowed, mostly for framebuffers.
899
 *
900
 * Must be called holding kvm->slots_lock for write.
A
Avi Kivity 已提交
901
 */
902
int __kvm_set_memory_region(struct kvm *kvm,
903
			    const struct kvm_userspace_memory_region *mem)
A
Avi Kivity 已提交
904
{
905
	int r;
A
Avi Kivity 已提交
906
	gfn_t base_gfn;
907
	unsigned long npages;
908
	struct kvm_memory_slot *slot;
A
Avi Kivity 已提交
909
	struct kvm_memory_slot old, new;
910
	struct kvm_memslots *slots = NULL, *old_memslots;
911
	int as_id, id;
912
	enum kvm_mr_change change;
A
Avi Kivity 已提交
913

914 915 916 917
	r = check_memory_region_flags(mem);
	if (r)
		goto out;

A
Avi Kivity 已提交
918
	r = -EINVAL;
919 920 921
	as_id = mem->slot >> 16;
	id = (u16)mem->slot;

A
Avi Kivity 已提交
922 923 924 925 926
	/* General sanity checks */
	if (mem->memory_size & (PAGE_SIZE - 1))
		goto out;
	if (mem->guest_phys_addr & (PAGE_SIZE - 1))
		goto out;
927
	/* We can read the guest memory with __xxx_user() later on. */
928
	if ((id < KVM_USER_MEM_SLOTS) &&
929
	    ((mem->userspace_addr & (PAGE_SIZE - 1)) ||
930 931 932
	     !access_ok(VERIFY_WRITE,
			(void __user *)(unsigned long)mem->userspace_addr,
			mem->memory_size)))
933
		goto out;
934
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_MEM_SLOTS_NUM)
A
Avi Kivity 已提交
935 936 937 938
		goto out;
	if (mem->guest_phys_addr + mem->memory_size < mem->guest_phys_addr)
		goto out;

939
	slot = id_to_memslot(__kvm_memslots(kvm, as_id), id);
A
Avi Kivity 已提交
940 941 942
	base_gfn = mem->guest_phys_addr >> PAGE_SHIFT;
	npages = mem->memory_size >> PAGE_SHIFT;

943 944 945
	if (npages > KVM_MEM_MAX_NR_PAGES)
		goto out;

946
	new = old = *slot;
A
Avi Kivity 已提交
947

948
	new.id = id;
A
Avi Kivity 已提交
949 950 951 952
	new.base_gfn = base_gfn;
	new.npages = npages;
	new.flags = mem->flags;

953 954 955 956 957
	if (npages) {
		if (!old.npages)
			change = KVM_MR_CREATE;
		else { /* Modify an existing slot. */
			if ((mem->userspace_addr != old.userspace_addr) ||
958 959
			    (npages != old.npages) ||
			    ((new.flags ^ old.flags) & KVM_MEM_READONLY))
960 961 962 963 964 965 966 967 968 969 970
				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;
			}
		}
971 972 973 974
	} else {
		if (!old.npages)
			goto out;

975
		change = KVM_MR_DELETE;
976 977 978
		new.base_gfn = 0;
		new.flags = 0;
	}
A
Avi Kivity 已提交
979

980
	if ((change == KVM_MR_CREATE) || (change == KVM_MR_MOVE)) {
981 982
		/* Check for overlaps */
		r = -EEXIST;
983
		kvm_for_each_memslot(slot, __kvm_memslots(kvm, as_id)) {
984
			if ((slot->id >= KVM_USER_MEM_SLOTS) ||
985
			    (slot->id == id))
986 987 988 989 990
				continue;
			if (!((base_gfn + npages <= slot->base_gfn) ||
			      (base_gfn >= slot->base_gfn + slot->npages)))
				goto out;
		}
A
Avi Kivity 已提交
991 992 993 994
	}

	/* Free page dirty bitmap if unneeded */
	if (!(new.flags & KVM_MEM_LOG_DIRTY_PAGES))
A
Al Viro 已提交
995
		new.dirty_bitmap = NULL;
A
Avi Kivity 已提交
996 997

	r = -ENOMEM;
998
	if (change == KVM_MR_CREATE) {
999
		new.userspace_addr = mem->userspace_addr;
1000

1001
		if (kvm_arch_create_memslot(kvm, &new, npages))
1002
			goto out_free;
A
Avi Kivity 已提交
1003
	}
1004

A
Avi Kivity 已提交
1005 1006
	/* Allocate page dirty bitmap if needed */
	if ((new.flags & KVM_MEM_LOG_DIRTY_PAGES) && !new.dirty_bitmap) {
1007
		if (kvm_create_dirty_bitmap(&new) < 0)
1008
			goto out_free;
A
Avi Kivity 已提交
1009 1010
	}

1011
	slots = kvm_kvzalloc(sizeof(struct kvm_memslots));
1012 1013
	if (!slots)
		goto out_free;
1014
	memcpy(slots, __kvm_memslots(kvm, as_id), sizeof(struct kvm_memslots));
1015

1016
	if ((change == KVM_MR_DELETE) || (change == KVM_MR_MOVE)) {
1017
		slot = id_to_memslot(slots, id);
1018 1019
		slot->flags |= KVM_MEMSLOT_INVALID;

1020
		old_memslots = install_new_memslots(kvm, as_id, slots);
1021

1022 1023
		/* From this point no new shadow pages pointing to a deleted,
		 * or moved, memslot will be created.
1024 1025
		 *
		 * validation of sp->gfn happens in:
1026 1027
		 *	- gfn_to_hva (kvm_read_guest, gfn_to_pfn)
		 *	- kvm_is_visible_gfn (mmu_check_roots)
1028
		 */
1029
		kvm_arch_flush_shadow_memslot(kvm, slot);
1030 1031 1032 1033 1034 1035

		/*
		 * 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.
		 */
1036
		slots = old_memslots;
1037
	}
1038

1039
	r = kvm_arch_prepare_memory_region(kvm, &new, mem, change);
1040
	if (r)
1041
		goto out_slots;
1042

1043
	/* actual memory is freed via old in kvm_free_memslot below */
1044
	if (change == KVM_MR_DELETE) {
1045
		new.dirty_bitmap = NULL;
1046
		memset(&new.arch, 0, sizeof(new.arch));
1047 1048
	}

1049
	update_memslots(slots, &new);
1050
	old_memslots = install_new_memslots(kvm, as_id, slots);
1051

1052
	kvm_arch_commit_memory_region(kvm, mem, &old, &new, change);
1053

1054
	kvm_free_memslot(kvm, &old, &new);
1055
	kvfree(old_memslots);
A
Avi Kivity 已提交
1056 1057
	return 0;

1058
out_slots:
1059
	kvfree(slots);
1060
out_free:
1061
	kvm_free_memslot(kvm, &new, &old);
A
Avi Kivity 已提交
1062 1063
out:
	return r;
1064
}
1065 1066 1067
EXPORT_SYMBOL_GPL(__kvm_set_memory_region);

int kvm_set_memory_region(struct kvm *kvm,
1068
			  const struct kvm_userspace_memory_region *mem)
1069 1070 1071
{
	int r;

1072
	mutex_lock(&kvm->slots_lock);
1073
	r = __kvm_set_memory_region(kvm, mem);
1074
	mutex_unlock(&kvm->slots_lock);
1075 1076
	return r;
}
1077 1078
EXPORT_SYMBOL_GPL(kvm_set_memory_region);

1079 1080
static int kvm_vm_ioctl_set_memory_region(struct kvm *kvm,
					  struct kvm_userspace_memory_region *mem)
1081
{
1082
	if ((u16)mem->slot >= KVM_USER_MEM_SLOTS)
1083
		return -EINVAL;
1084

1085
	return kvm_set_memory_region(kvm, mem);
A
Avi Kivity 已提交
1086 1087
}

1088 1089
int kvm_get_dirty_log(struct kvm *kvm,
			struct kvm_dirty_log *log, int *is_dirty)
A
Avi Kivity 已提交
1090
{
1091
	struct kvm_memslots *slots;
A
Avi Kivity 已提交
1092
	struct kvm_memory_slot *memslot;
1093
	int i, as_id, id;
1094
	unsigned long n;
A
Avi Kivity 已提交
1095 1096
	unsigned long any = 0;

1097 1098 1099
	as_id = log->slot >> 16;
	id = (u16)log->slot;
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_USER_MEM_SLOTS)
1100
		return -EINVAL;
A
Avi Kivity 已提交
1101

1102 1103
	slots = __kvm_memslots(kvm, as_id);
	memslot = id_to_memslot(slots, id);
A
Avi Kivity 已提交
1104
	if (!memslot->dirty_bitmap)
1105
		return -ENOENT;
A
Avi Kivity 已提交
1106

1107
	n = kvm_dirty_bitmap_bytes(memslot);
A
Avi Kivity 已提交
1108

1109
	for (i = 0; !any && i < n/sizeof(long); ++i)
A
Avi Kivity 已提交
1110 1111 1112
		any = memslot->dirty_bitmap[i];

	if (copy_to_user(log->dirty_bitmap, memslot->dirty_bitmap, n))
1113
		return -EFAULT;
A
Avi Kivity 已提交
1114

1115 1116
	if (any)
		*is_dirty = 1;
1117
	return 0;
A
Avi Kivity 已提交
1118
}
1119
EXPORT_SYMBOL_GPL(kvm_get_dirty_log);
A
Avi Kivity 已提交
1120

1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146
#ifdef CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT
/**
 * kvm_get_dirty_log_protect - get a snapshot of dirty pages, and if any pages
 *	are dirty write protect them for next write.
 * @kvm:	pointer to kvm instance
 * @log:	slot id and address to which we copy the log
 * @is_dirty:	flag set if any page is dirty
 *
 * We need to keep it in mind that VCPU threads can write to the bitmap
 * concurrently. So, to avoid losing track of dirty pages we keep the
 * following order:
 *
 *    1. Take a snapshot of the bit and clear it if needed.
 *    2. Write protect the corresponding page.
 *    3. Copy the snapshot to the userspace.
 *    4. Upon return caller flushes TLB's if needed.
 *
 * Between 2 and 4, the guest may write to the page using the remaining TLB
 * entry.  This is not a problem because the page is reported dirty using
 * the snapshot taken before and step 4 ensures that writes done after
 * exiting to userspace will be logged for the next call.
 *
 */
int kvm_get_dirty_log_protect(struct kvm *kvm,
			struct kvm_dirty_log *log, bool *is_dirty)
{
1147
	struct kvm_memslots *slots;
1148
	struct kvm_memory_slot *memslot;
1149
	int i, as_id, id;
1150 1151 1152 1153
	unsigned long n;
	unsigned long *dirty_bitmap;
	unsigned long *dirty_bitmap_buffer;

1154 1155 1156
	as_id = log->slot >> 16;
	id = (u16)log->slot;
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_USER_MEM_SLOTS)
1157
		return -EINVAL;
1158

1159 1160
	slots = __kvm_memslots(kvm, as_id);
	memslot = id_to_memslot(slots, id);
1161 1162 1163

	dirty_bitmap = memslot->dirty_bitmap;
	if (!dirty_bitmap)
1164
		return -ENOENT;
1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184

	n = kvm_dirty_bitmap_bytes(memslot);

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

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

		if (!dirty_bitmap[i])
			continue;

		*is_dirty = true;

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

1185 1186 1187 1188 1189
		if (mask) {
			offset = i * BITS_PER_LONG;
			kvm_arch_mmu_enable_log_dirty_pt_masked(kvm, memslot,
								offset, mask);
		}
1190 1191 1192 1193
	}

	spin_unlock(&kvm->mmu_lock);
	if (copy_to_user(log->dirty_bitmap, dirty_bitmap_buffer, n))
1194 1195
		return -EFAULT;
	return 0;
1196 1197 1198 1199
}
EXPORT_SYMBOL_GPL(kvm_get_dirty_log_protect);
#endif

1200 1201 1202 1203 1204
bool kvm_largepages_enabled(void)
{
	return largepages_enabled;
}

1205 1206 1207 1208 1209 1210
void kvm_disable_largepages(void)
{
	largepages_enabled = false;
}
EXPORT_SYMBOL_GPL(kvm_disable_largepages);

1211 1212 1213 1214
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
1215
EXPORT_SYMBOL_GPL(gfn_to_memslot);
A
Avi Kivity 已提交
1216

1217 1218 1219 1220 1221
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);
}

1222
bool kvm_is_visible_gfn(struct kvm *kvm, gfn_t gfn)
1223
{
1224
	struct kvm_memory_slot *memslot = gfn_to_memslot(kvm, gfn);
1225

1226
	if (!memslot || memslot->id >= KVM_USER_MEM_SLOTS ||
1227
	      memslot->flags & KVM_MEMSLOT_INVALID)
1228
		return false;
1229

1230
	return true;
1231 1232 1233
}
EXPORT_SYMBOL_GPL(kvm_is_visible_gfn);

J
Joerg Roedel 已提交
1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257
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
1258 1259 1260 1261 1262 1263 1264
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
1265
{
1266
	if (!slot || slot->flags & KVM_MEMSLOT_INVALID)
X
Xiao Guangrong 已提交
1267
		return KVM_HVA_ERR_BAD;
1268

X
Xiao Guangrong 已提交
1269 1270
	if (memslot_is_readonly(slot) && write)
		return KVM_HVA_ERR_RO_BAD;
1271 1272 1273 1274

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

X
Xiao Guangrong 已提交
1275
	return __gfn_to_hva_memslot(slot, gfn);
I
Izik Eidus 已提交
1276
}
1277

X
Xiao Guangrong 已提交
1278 1279 1280 1281
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
1282
}
1283

X
Xiao Guangrong 已提交
1284
unsigned long gfn_to_hva_memslot(struct kvm_memory_slot *slot,
1285
					gfn_t gfn)
X
Xiao Guangrong 已提交
1286 1287 1288 1289 1290
{
	return gfn_to_hva_many(slot, gfn, NULL);
}
EXPORT_SYMBOL_GPL(gfn_to_hva_memslot);

1291 1292
unsigned long gfn_to_hva(struct kvm *kvm, gfn_t gfn)
{
1293
	return gfn_to_hva_many(gfn_to_memslot(kvm, gfn), gfn, NULL);
1294
}
1295
EXPORT_SYMBOL_GPL(gfn_to_hva);
I
Izik Eidus 已提交
1296

1297 1298 1299 1300 1301 1302
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);

1303
/*
1304 1305
 * If writable is set to false, the hva returned by this function is only
 * allowed to be read.
1306
 */
1307 1308
unsigned long gfn_to_hva_memslot_prot(struct kvm_memory_slot *slot,
				      gfn_t gfn, bool *writable)
1309
{
1310 1311 1312
	unsigned long hva = __gfn_to_hva_many(slot, gfn, NULL, false);

	if (!kvm_is_error_hva(hva) && writable)
1313 1314
		*writable = !memslot_is_readonly(slot);

1315
	return hva;
1316 1317
}

1318 1319 1320 1321 1322 1323 1324
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);
}

1325 1326 1327 1328 1329 1330 1331
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);
}

1332 1333
static int get_user_page_nowait(unsigned long start, int write,
		struct page **page)
1334
{
L
Lorenzo Stoakes 已提交
1335
	int flags = FOLL_NOWAIT | FOLL_HWPOISON;
1336 1337 1338 1339

	if (write)
		flags |= FOLL_WRITE;

L
Lorenzo Stoakes 已提交
1340
	return get_user_pages(start, 1, flags, page, NULL);
1341 1342
}

1343 1344
static inline int check_user_page_hwpoison(unsigned long addr)
{
L
Lorenzo Stoakes 已提交
1345
	int rc, flags = FOLL_HWPOISON | FOLL_WRITE;
1346

L
Lorenzo Stoakes 已提交
1347
	rc = get_user_pages(addr, 1, flags, NULL, NULL);
1348 1349 1350
	return rc == -EHWPOISON;
}

X
Xiao Guangrong 已提交
1351 1352 1353 1354 1355
/*
 * The atomic path to get the writable pfn which will be stored in @pfn,
 * true indicates success, otherwise false is returned.
 */
static bool hva_to_pfn_fast(unsigned long addr, bool atomic, bool *async,
D
Dan Williams 已提交
1356
			    bool write_fault, bool *writable, kvm_pfn_t *pfn)
A
Avi Kivity 已提交
1357
{
1358
	struct page *page[1];
X
Xiao Guangrong 已提交
1359
	int npages;
A
Avi Kivity 已提交
1360

X
Xiao Guangrong 已提交
1361 1362
	if (!(async || atomic))
		return false;
1363

1364 1365 1366 1367 1368 1369 1370
	/*
	 * 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;
1371

X
Xiao Guangrong 已提交
1372 1373 1374
	npages = __get_user_pages_fast(addr, 1, 1, page);
	if (npages == 1) {
		*pfn = page_to_pfn(page[0]);
1375

X
Xiao Guangrong 已提交
1376 1377 1378 1379
		if (writable)
			*writable = true;
		return true;
	}
1380

X
Xiao Guangrong 已提交
1381 1382
	return false;
}
1383

X
Xiao Guangrong 已提交
1384 1385 1386 1387 1388
/*
 * 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
1389
			   bool *writable, kvm_pfn_t *pfn)
X
Xiao Guangrong 已提交
1390 1391 1392
{
	struct page *page[1];
	int npages = 0;
1393

X
Xiao Guangrong 已提交
1394 1395 1396 1397 1398 1399 1400
	might_sleep();

	if (writable)
		*writable = write_fault;

	if (async) {
		down_read(&current->mm->mmap_sem);
1401
		npages = get_user_page_nowait(addr, write_fault, page);
X
Xiao Guangrong 已提交
1402
		up_read(&current->mm->mmap_sem);
1403
	} else {
1404
		unsigned int flags = FOLL_HWPOISON;
1405 1406 1407 1408

		if (write_fault)
			flags |= FOLL_WRITE;

1409
		npages = get_user_pages_unlocked(addr, 1, page, flags);
1410
	}
X
Xiao Guangrong 已提交
1411 1412 1413 1414
	if (npages != 1)
		return npages;

	/* map read fault as writable if possible */
1415
	if (unlikely(!write_fault) && writable) {
X
Xiao Guangrong 已提交
1416 1417 1418 1419 1420 1421 1422
		struct page *wpage[1];

		npages = __get_user_pages_fast(addr, 1, 1, wpage);
		if (npages == 1) {
			*writable = true;
			put_page(page[0]);
			page[0] = wpage[0];
1423
		}
X
Xiao Guangrong 已提交
1424 1425

		npages = 1;
1426
	}
X
Xiao Guangrong 已提交
1427 1428 1429
	*pfn = page_to_pfn(page[0]);
	return npages;
}
I
Izik Eidus 已提交
1430

X
Xiao Guangrong 已提交
1431 1432 1433 1434
static bool vma_is_valid(struct vm_area_struct *vma, bool write_fault)
{
	if (unlikely(!(vma->vm_flags & VM_READ)))
		return false;
1435

X
Xiao Guangrong 已提交
1436 1437
	if (write_fault && (unlikely(!(vma->vm_flags & VM_WRITE))))
		return false;
1438

X
Xiao Guangrong 已提交
1439 1440
	return true;
}
1441

1442 1443 1444 1445
static int hva_to_pfn_remapped(struct vm_area_struct *vma,
			       unsigned long addr, bool *async,
			       bool write_fault, kvm_pfn_t *p_pfn)
{
1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484
	unsigned long pfn;
	int r;

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

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

	}


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

	*p_pfn = pfn;
1485 1486 1487
	return 0;
}

1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501
/*
 * 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
1502
static kvm_pfn_t hva_to_pfn(unsigned long addr, bool atomic, bool *async,
X
Xiao Guangrong 已提交
1503 1504 1505
			bool write_fault, bool *writable)
{
	struct vm_area_struct *vma;
D
Dan Williams 已提交
1506
	kvm_pfn_t pfn = 0;
1507
	int npages, r;
1508

X
Xiao Guangrong 已提交
1509 1510
	/* we can do it either atomically or asynchronously, not both */
	BUG_ON(atomic && async);
1511

X
Xiao Guangrong 已提交
1512 1513 1514 1515 1516 1517 1518 1519 1520
	if (hva_to_pfn_fast(addr, atomic, async, write_fault, writable, &pfn))
		return pfn;

	if (atomic)
		return KVM_PFN_ERR_FAULT;

	npages = hva_to_pfn_slow(addr, async, write_fault, writable, &pfn);
	if (npages == 1)
		return pfn;
1521

X
Xiao Guangrong 已提交
1522 1523 1524 1525 1526 1527 1528
	down_read(&current->mm->mmap_sem);
	if (npages == -EHWPOISON ||
	      (!async && check_user_page_hwpoison(addr))) {
		pfn = KVM_PFN_ERR_HWPOISON;
		goto exit;
	}

1529
retry:
X
Xiao Guangrong 已提交
1530 1531 1532 1533
	vma = find_vma_intersection(current->mm, addr, addr + 1);

	if (vma == NULL)
		pfn = KVM_PFN_ERR_FAULT;
1534 1535
	else if (vma->vm_flags & (VM_IO | VM_PFNMAP)) {
		r = hva_to_pfn_remapped(vma, addr, async, write_fault, &pfn);
1536 1537
		if (r == -EAGAIN)
			goto retry;
1538 1539
		if (r < 0)
			pfn = KVM_PFN_ERR_FAULT;
X
Xiao Guangrong 已提交
1540
	} else {
X
Xiao Guangrong 已提交
1541
		if (async && vma_is_valid(vma, write_fault))
X
Xiao Guangrong 已提交
1542 1543 1544 1545 1546
			*async = true;
		pfn = KVM_PFN_ERR_FAULT;
	}
exit:
	up_read(&current->mm->mmap_sem);
1547
	return pfn;
1548 1549
}

D
Dan Williams 已提交
1550 1551 1552
kvm_pfn_t __gfn_to_pfn_memslot(struct kvm_memory_slot *slot, gfn_t gfn,
			       bool atomic, bool *async, bool write_fault,
			       bool *writable)
1553
{
X
Xiao Guangrong 已提交
1554 1555
	unsigned long addr = __gfn_to_hva_many(slot, gfn, NULL, write_fault);

1556 1557 1558
	if (addr == KVM_HVA_ERR_RO_BAD) {
		if (writable)
			*writable = false;
X
Xiao Guangrong 已提交
1559
		return KVM_PFN_ERR_RO_FAULT;
1560
	}
X
Xiao Guangrong 已提交
1561

1562 1563 1564
	if (kvm_is_error_hva(addr)) {
		if (writable)
			*writable = false;
1565
		return KVM_PFN_NOSLOT;
1566
	}
X
Xiao Guangrong 已提交
1567 1568 1569 1570 1571 1572 1573 1574 1575

	/* 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);
1576
}
1577
EXPORT_SYMBOL_GPL(__gfn_to_pfn_memslot);
1578

D
Dan Williams 已提交
1579
kvm_pfn_t gfn_to_pfn_prot(struct kvm *kvm, gfn_t gfn, bool write_fault,
1580 1581
		      bool *writable)
{
P
Paolo Bonzini 已提交
1582 1583
	return __gfn_to_pfn_memslot(gfn_to_memslot(kvm, gfn), gfn, false, NULL,
				    write_fault, writable);
1584 1585 1586
}
EXPORT_SYMBOL_GPL(gfn_to_pfn_prot);

D
Dan Williams 已提交
1587
kvm_pfn_t gfn_to_pfn_memslot(struct kvm_memory_slot *slot, gfn_t gfn)
1588
{
X
Xiao Guangrong 已提交
1589
	return __gfn_to_pfn_memslot(slot, gfn, false, NULL, true, NULL);
1590
}
P
Paolo Bonzini 已提交
1591
EXPORT_SYMBOL_GPL(gfn_to_pfn_memslot);
1592

D
Dan Williams 已提交
1593
kvm_pfn_t gfn_to_pfn_memslot_atomic(struct kvm_memory_slot *slot, gfn_t gfn)
1594
{
X
Xiao Guangrong 已提交
1595
	return __gfn_to_pfn_memslot(slot, gfn, true, NULL, true, NULL);
1596
}
1597
EXPORT_SYMBOL_GPL(gfn_to_pfn_memslot_atomic);
1598

D
Dan Williams 已提交
1599
kvm_pfn_t gfn_to_pfn_atomic(struct kvm *kvm, gfn_t gfn)
P
Paolo Bonzini 已提交
1600 1601 1602 1603 1604
{
	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
1605
kvm_pfn_t kvm_vcpu_gfn_to_pfn_atomic(struct kvm_vcpu *vcpu, gfn_t gfn)
1606 1607 1608 1609 1610
{
	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
1611
kvm_pfn_t gfn_to_pfn(struct kvm *kvm, gfn_t gfn)
P
Paolo Bonzini 已提交
1612 1613 1614 1615 1616
{
	return gfn_to_pfn_memslot(gfn_to_memslot(kvm, gfn), gfn);
}
EXPORT_SYMBOL_GPL(gfn_to_pfn);

D
Dan Williams 已提交
1617
kvm_pfn_t kvm_vcpu_gfn_to_pfn(struct kvm_vcpu *vcpu, gfn_t gfn)
1618 1619 1620 1621 1622
{
	return gfn_to_pfn_memslot(kvm_vcpu_gfn_to_memslot(vcpu, gfn), gfn);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_gfn_to_pfn);

1623 1624
int gfn_to_page_many_atomic(struct kvm_memory_slot *slot, gfn_t gfn,
			    struct page **pages, int nr_pages)
1625 1626 1627 1628
{
	unsigned long addr;
	gfn_t entry;

1629
	addr = gfn_to_hva_many(slot, gfn, &entry);
1630 1631 1632 1633 1634 1635 1636 1637 1638 1639
	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
1640
static struct page *kvm_pfn_to_page(kvm_pfn_t pfn)
X
Xiao Guangrong 已提交
1641
{
1642
	if (is_error_noslot_pfn(pfn))
1643
		return KVM_ERR_PTR_BAD_PAGE;
X
Xiao Guangrong 已提交
1644

1645
	if (kvm_is_reserved_pfn(pfn)) {
1646
		WARN_ON(1);
1647
		return KVM_ERR_PTR_BAD_PAGE;
1648
	}
X
Xiao Guangrong 已提交
1649 1650 1651 1652

	return pfn_to_page(pfn);
}

1653 1654
struct page *gfn_to_page(struct kvm *kvm, gfn_t gfn)
{
D
Dan Williams 已提交
1655
	kvm_pfn_t pfn;
1656 1657 1658

	pfn = gfn_to_pfn(kvm, gfn);

X
Xiao Guangrong 已提交
1659
	return kvm_pfn_to_page(pfn);
A
Avi Kivity 已提交
1660 1661 1662
}
EXPORT_SYMBOL_GPL(gfn_to_page);

1663 1664
struct page *kvm_vcpu_gfn_to_page(struct kvm_vcpu *vcpu, gfn_t gfn)
{
D
Dan Williams 已提交
1665
	kvm_pfn_t pfn;
1666 1667 1668 1669 1670 1671 1672

	pfn = kvm_vcpu_gfn_to_pfn(vcpu, gfn);

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

1673 1674
void kvm_release_page_clean(struct page *page)
{
1675 1676
	WARN_ON(is_error_page(page));

1677
	kvm_release_pfn_clean(page_to_pfn(page));
1678 1679 1680
}
EXPORT_SYMBOL_GPL(kvm_release_page_clean);

D
Dan Williams 已提交
1681
void kvm_release_pfn_clean(kvm_pfn_t pfn)
1682
{
1683
	if (!is_error_noslot_pfn(pfn) && !kvm_is_reserved_pfn(pfn))
1684
		put_page(pfn_to_page(pfn));
1685 1686 1687
}
EXPORT_SYMBOL_GPL(kvm_release_pfn_clean);

1688
void kvm_release_page_dirty(struct page *page)
1689
{
X
Xiao Guangrong 已提交
1690 1691
	WARN_ON(is_error_page(page));

1692 1693 1694 1695
	kvm_release_pfn_dirty(page_to_pfn(page));
}
EXPORT_SYMBOL_GPL(kvm_release_page_dirty);

D
Dan Williams 已提交
1696
static void kvm_release_pfn_dirty(kvm_pfn_t pfn)
1697 1698 1699 1700 1701
{
	kvm_set_pfn_dirty(pfn);
	kvm_release_pfn_clean(pfn);
}

D
Dan Williams 已提交
1702
void kvm_set_pfn_dirty(kvm_pfn_t pfn)
1703
{
1704
	if (!kvm_is_reserved_pfn(pfn)) {
1705
		struct page *page = pfn_to_page(pfn);
1706

1707 1708 1709
		if (!PageReserved(page))
			SetPageDirty(page);
	}
1710
}
1711 1712
EXPORT_SYMBOL_GPL(kvm_set_pfn_dirty);

D
Dan Williams 已提交
1713
void kvm_set_pfn_accessed(kvm_pfn_t pfn)
1714
{
1715
	if (!kvm_is_reserved_pfn(pfn))
1716
		mark_page_accessed(pfn_to_page(pfn));
1717 1718 1719
}
EXPORT_SYMBOL_GPL(kvm_set_pfn_accessed);

D
Dan Williams 已提交
1720
void kvm_get_pfn(kvm_pfn_t pfn)
1721
{
1722
	if (!kvm_is_reserved_pfn(pfn))
1723
		get_page(pfn_to_page(pfn));
1724 1725
}
EXPORT_SYMBOL_GPL(kvm_get_pfn);
1726

1727 1728 1729 1730 1731 1732 1733 1734
static int next_segment(unsigned long len, int offset)
{
	if (len > PAGE_SIZE - offset)
		return PAGE_SIZE - offset;
	else
		return len;
}

1735 1736
static int __kvm_read_guest_page(struct kvm_memory_slot *slot, gfn_t gfn,
				 void *data, int offset, int len)
1737
{
1738 1739
	int r;
	unsigned long addr;
1740

1741
	addr = gfn_to_hva_memslot_prot(slot, gfn, NULL);
1742 1743
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1744
	r = __copy_from_user(data, (void __user *)addr + offset, len);
1745
	if (r)
1746 1747 1748
		return -EFAULT;
	return 0;
}
1749 1750 1751 1752 1753 1754 1755 1756

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);
}
1757 1758
EXPORT_SYMBOL_GPL(kvm_read_guest_page);

1759 1760 1761 1762 1763 1764 1765 1766 1767
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);

1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787
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);

1788
int kvm_vcpu_read_guest(struct kvm_vcpu *vcpu, gpa_t gpa, void *data, unsigned long len)
1789 1790
{
	gfn_t gfn = gpa >> PAGE_SHIFT;
1791
	int seg;
1792
	int offset = offset_in_page(gpa);
1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806
	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);
1807

1808 1809 1810 1811 1812 1813 1814
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);
1815 1816
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1817
	pagefault_disable();
1818
	r = __copy_from_user_inatomic(data, (void __user *)addr + offset, len);
1819
	pagefault_enable();
1820 1821 1822 1823 1824
	if (r)
		return -EFAULT;
	return 0;
}

1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848
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)
1849
{
1850 1851
	int r;
	unsigned long addr;
1852

1853
	addr = gfn_to_hva_memslot(memslot, gfn);
1854 1855
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1856
	r = __copy_to_user((void __user *)addr + offset, data, len);
1857
	if (r)
1858
		return -EFAULT;
1859
	mark_page_dirty_in_slot(memslot, gfn);
1860 1861
	return 0;
}
1862 1863 1864 1865 1866 1867 1868 1869

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);
}
1870 1871
EXPORT_SYMBOL_GPL(kvm_write_guest_page);

1872 1873 1874 1875 1876 1877 1878 1879 1880
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);

1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899
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;
}
1900
EXPORT_SYMBOL_GPL(kvm_write_guest);
1901

1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922
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);

1923 1924 1925
static int __kvm_gfn_to_hva_cache_init(struct kvm_memslots *slots,
				       struct gfn_to_hva_cache *ghc,
				       gpa_t gpa, unsigned long len)
1926 1927
{
	int offset = offset_in_page(gpa);
1928 1929 1930 1931
	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;
1932 1933 1934

	ghc->gpa = gpa;
	ghc->generation = slots->generation;
1935
	ghc->len = len;
1936
	ghc->memslot = __gfn_to_memslot(slots, start_gfn);
1937 1938
	ghc->hva = gfn_to_hva_many(ghc->memslot, start_gfn, NULL);
	if (!kvm_is_error_hva(ghc->hva) && nr_pages_needed <= 1) {
1939
		ghc->hva += offset;
1940 1941 1942 1943 1944 1945
	} else {
		/*
		 * If the requested region crosses two memslots, we still
		 * verify that the entire region is valid here.
		 */
		while (start_gfn <= end_gfn) {
1946
			ghc->memslot = __gfn_to_memslot(slots, start_gfn);
1947 1948 1949 1950 1951 1952 1953 1954 1955
			ghc->hva = gfn_to_hva_many(ghc->memslot, start_gfn,
						   &nr_pages_avail);
			if (kvm_is_error_hva(ghc->hva))
				return -EFAULT;
			start_gfn += nr_pages_avail;
		}
		/* Use the slow path for cross page reads and writes. */
		ghc->memslot = NULL;
	}
1956 1957
	return 0;
}
1958

1959
int kvm_vcpu_gfn_to_hva_cache_init(struct kvm_vcpu *vcpu, struct gfn_to_hva_cache *ghc,
1960 1961
			      gpa_t gpa, unsigned long len)
{
1962
	struct kvm_memslots *slots = kvm_vcpu_memslots(vcpu);
1963 1964
	return __kvm_gfn_to_hva_cache_init(slots, ghc, gpa, len);
}
1965
EXPORT_SYMBOL_GPL(kvm_vcpu_gfn_to_hva_cache_init);
1966

1967 1968
int kvm_vcpu_write_guest_offset_cached(struct kvm_vcpu *vcpu, struct gfn_to_hva_cache *ghc,
				       void *data, int offset, unsigned long len)
1969
{
1970
	struct kvm_memslots *slots = kvm_vcpu_memslots(vcpu);
1971
	int r;
1972
	gpa_t gpa = ghc->gpa + offset;
1973

1974
	BUG_ON(len + offset > ghc->len);
1975

1976
	if (slots->generation != ghc->generation)
1977
		__kvm_gfn_to_hva_cache_init(slots, ghc, ghc->gpa, ghc->len);
1978 1979

	if (unlikely(!ghc->memslot))
1980
		return kvm_vcpu_write_guest(vcpu, gpa, data, len);
1981 1982 1983 1984

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

1985
	r = __copy_to_user((void __user *)ghc->hva + offset, data, len);
1986 1987
	if (r)
		return -EFAULT;
1988
	mark_page_dirty_in_slot(ghc->memslot, gpa >> PAGE_SHIFT);
1989 1990 1991

	return 0;
}
1992
EXPORT_SYMBOL_GPL(kvm_vcpu_write_guest_offset_cached);
1993

1994 1995
int kvm_vcpu_write_guest_cached(struct kvm_vcpu *vcpu, struct gfn_to_hva_cache *ghc,
			       void *data, unsigned long len)
1996
{
1997
	return kvm_vcpu_write_guest_offset_cached(vcpu, ghc, data, 0, len);
1998
}
1999
EXPORT_SYMBOL_GPL(kvm_vcpu_write_guest_cached);
2000

2001 2002
int kvm_vcpu_read_guest_cached(struct kvm_vcpu *vcpu, struct gfn_to_hva_cache *ghc,
			       void *data, unsigned long len)
2003
{
2004
	struct kvm_memslots *slots = kvm_vcpu_memslots(vcpu);
2005 2006
	int r;

2007 2008
	BUG_ON(len > ghc->len);

2009
	if (slots->generation != ghc->generation)
2010
		__kvm_gfn_to_hva_cache_init(slots, ghc, ghc->gpa, ghc->len);
2011 2012

	if (unlikely(!ghc->memslot))
2013
		return kvm_vcpu_read_guest(vcpu, ghc->gpa, data, len);
2014 2015 2016 2017 2018 2019 2020 2021 2022 2023

	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;
}
2024
EXPORT_SYMBOL_GPL(kvm_vcpu_read_guest_cached);
2025

2026 2027
int kvm_clear_guest_page(struct kvm *kvm, gfn_t gfn, int offset, int len)
{
2028 2029 2030
	const void *zero_page = (const void *) __va(page_to_phys(ZERO_PAGE(0)));

	return kvm_write_guest_page(kvm, gfn, zero_page, offset, len);
2031 2032 2033 2034 2035 2036 2037 2038 2039 2040
}
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;

2041
	while ((seg = next_segment(len, offset)) != 0) {
2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052
		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);

2053
static void mark_page_dirty_in_slot(struct kvm_memory_slot *memslot,
2054
				    gfn_t gfn)
A
Avi Kivity 已提交
2055
{
R
Rusty Russell 已提交
2056 2057
	if (memslot && memslot->dirty_bitmap) {
		unsigned long rel_gfn = gfn - memslot->base_gfn;
A
Avi Kivity 已提交
2058

2059
		set_bit_le(rel_gfn, memslot->dirty_bitmap);
A
Avi Kivity 已提交
2060 2061 2062
	}
}

2063 2064 2065 2066 2067
void mark_page_dirty(struct kvm *kvm, gfn_t gfn)
{
	struct kvm_memory_slot *memslot;

	memslot = gfn_to_memslot(kvm, gfn);
2068
	mark_page_dirty_in_slot(memslot, gfn);
2069
}
2070
EXPORT_SYMBOL_GPL(mark_page_dirty);
2071

2072 2073 2074 2075 2076 2077 2078 2079 2080
void kvm_vcpu_mark_page_dirty(struct kvm_vcpu *vcpu, gfn_t gfn)
{
	struct kvm_memory_slot *memslot;

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

W
Wanpeng Li 已提交
2081 2082
static void grow_halt_poll_ns(struct kvm_vcpu *vcpu)
{
2083
	unsigned int old, val, grow;
W
Wanpeng Li 已提交
2084

2085
	old = val = vcpu->halt_poll_ns;
2086
	grow = READ_ONCE(halt_poll_ns_grow);
W
Wanpeng Li 已提交
2087
	/* 10us base */
2088
	if (val == 0 && grow)
W
Wanpeng Li 已提交
2089 2090
		val = 10000;
	else
2091
		val *= grow;
W
Wanpeng Li 已提交
2092

2093 2094 2095
	if (val > halt_poll_ns)
		val = halt_poll_ns;

W
Wanpeng Li 已提交
2096
	vcpu->halt_poll_ns = val;
2097
	trace_kvm_halt_poll_ns_grow(vcpu->vcpu_id, val, old);
W
Wanpeng Li 已提交
2098 2099 2100 2101
}

static void shrink_halt_poll_ns(struct kvm_vcpu *vcpu)
{
2102
	unsigned int old, val, shrink;
W
Wanpeng Li 已提交
2103

2104
	old = val = vcpu->halt_poll_ns;
2105 2106
	shrink = READ_ONCE(halt_poll_ns_shrink);
	if (shrink == 0)
W
Wanpeng Li 已提交
2107 2108
		val = 0;
	else
2109
		val /= shrink;
W
Wanpeng Li 已提交
2110 2111

	vcpu->halt_poll_ns = val;
2112
	trace_kvm_halt_poll_ns_shrink(vcpu->vcpu_id, val, old);
W
Wanpeng Li 已提交
2113 2114
}

2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128
static int kvm_vcpu_check_block(struct kvm_vcpu *vcpu)
{
	if (kvm_arch_vcpu_runnable(vcpu)) {
		kvm_make_request(KVM_REQ_UNHALT, vcpu);
		return -EINTR;
	}
	if (kvm_cpu_has_pending_timer(vcpu))
		return -EINTR;
	if (signal_pending(current))
		return -EINTR;

	return 0;
}

E
Eddie Dong 已提交
2129 2130 2131
/*
 * The vCPU has executed a HLT instruction with in-kernel mode enabled.
 */
2132
void kvm_vcpu_block(struct kvm_vcpu *vcpu)
2133
{
2134
	ktime_t start, cur;
2135
	DECLARE_SWAITQUEUE(wait);
2136
	bool waited = false;
W
Wanpeng Li 已提交
2137
	u64 block_ns;
2138 2139

	start = cur = ktime_get();
W
Wanpeng Li 已提交
2140 2141
	if (vcpu->halt_poll_ns) {
		ktime_t stop = ktime_add_ns(ktime_get(), vcpu->halt_poll_ns);
2142

2143
		++vcpu->stat.halt_attempted_poll;
2144 2145 2146 2147 2148 2149 2150
		do {
			/*
			 * This sets KVM_REQ_UNHALT if an interrupt
			 * arrives.
			 */
			if (kvm_vcpu_check_block(vcpu) < 0) {
				++vcpu->stat.halt_successful_poll;
2151 2152
				if (!vcpu_valid_wakeup(vcpu))
					++vcpu->stat.halt_poll_invalid;
2153 2154 2155 2156 2157
				goto out;
			}
			cur = ktime_get();
		} while (single_task_running() && ktime_before(cur, stop));
	}
2158

2159 2160
	kvm_arch_vcpu_blocking(vcpu);

2161
	for (;;) {
2162
		prepare_to_swait(&vcpu->wq, &wait, TASK_INTERRUPTIBLE);
2163

2164
		if (kvm_vcpu_check_block(vcpu) < 0)
2165 2166
			break;

2167
		waited = true;
E
Eddie Dong 已提交
2168 2169
		schedule();
	}
2170

2171
	finish_swait(&vcpu->wq, &wait);
2172 2173
	cur = ktime_get();

2174
	kvm_arch_vcpu_unblocking(vcpu);
2175
out:
W
Wanpeng Li 已提交
2176 2177
	block_ns = ktime_to_ns(cur) - ktime_to_ns(start);

2178 2179 2180
	if (!vcpu_valid_wakeup(vcpu))
		shrink_halt_poll_ns(vcpu);
	else if (halt_poll_ns) {
W
Wanpeng Li 已提交
2181 2182 2183
		if (block_ns <= vcpu->halt_poll_ns)
			;
		/* we had a long block, shrink polling */
2184
		else if (vcpu->halt_poll_ns && block_ns > halt_poll_ns)
W
Wanpeng Li 已提交
2185 2186 2187 2188 2189
			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);
2190 2191
	} else
		vcpu->halt_poll_ns = 0;
W
Wanpeng Li 已提交
2192

2193 2194
	trace_kvm_vcpu_wakeup(block_ns, waited, vcpu_valid_wakeup(vcpu));
	kvm_arch_vcpu_block_finish(vcpu);
E
Eddie Dong 已提交
2195
}
2196
EXPORT_SYMBOL_GPL(kvm_vcpu_block);
E
Eddie Dong 已提交
2197

2198
#ifndef CONFIG_S390
2199
void kvm_vcpu_wake_up(struct kvm_vcpu *vcpu)
2200
{
2201
	struct swait_queue_head *wqp;
2202 2203

	wqp = kvm_arch_vcpu_wq(vcpu);
2204 2205
	if (swait_active(wqp)) {
		swake_up(wqp);
2206 2207 2208
		++vcpu->stat.halt_wakeup;
	}

2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220
}
EXPORT_SYMBOL_GPL(kvm_vcpu_wake_up);

/*
 * Kick a sleeping VCPU, or a guest VCPU in guest mode, into host kernel mode.
 */
void kvm_vcpu_kick(struct kvm_vcpu *vcpu)
{
	int me;
	int cpu = vcpu->cpu;

	kvm_vcpu_wake_up(vcpu);
2221 2222 2223 2224 2225 2226
	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();
}
2227
EXPORT_SYMBOL_GPL(kvm_vcpu_kick);
2228
#endif /* !CONFIG_S390 */
2229

2230
int kvm_vcpu_yield_to(struct kvm_vcpu *target)
2231 2232 2233
{
	struct pid *pid;
	struct task_struct *task = NULL;
2234
	int ret = 0;
2235 2236 2237 2238

	rcu_read_lock();
	pid = rcu_dereference(target->pid);
	if (pid)
2239
		task = get_pid_task(pid, PIDTYPE_PID);
2240 2241
	rcu_read_unlock();
	if (!task)
2242 2243
		return ret;
	ret = yield_to(task, 1);
2244
	put_task_struct(task);
2245 2246

	return ret;
2247 2248 2249
}
EXPORT_SYMBOL_GPL(kvm_vcpu_yield_to);

2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271
/*
 * 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.
 */
2272
static bool kvm_vcpu_eligible_for_directed_yield(struct kvm_vcpu *vcpu)
2273
{
2274
#ifdef CONFIG_HAVE_KVM_CPU_RELAX_INTERCEPT
2275 2276 2277
	bool eligible;

	eligible = !vcpu->spin_loop.in_spin_loop ||
2278
		    vcpu->spin_loop.dy_eligible;
2279 2280 2281 2282 2283

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

	return eligible;
2284 2285
#else
	return true;
2286
#endif
2287
}
2288

2289
void kvm_vcpu_on_spin(struct kvm_vcpu *me)
Z
Zhai, Edwin 已提交
2290
{
2291 2292 2293 2294
	struct kvm *kvm = me->kvm;
	struct kvm_vcpu *vcpu;
	int last_boosted_vcpu = me->kvm->last_boosted_vcpu;
	int yielded = 0;
2295
	int try = 3;
2296 2297
	int pass;
	int i;
Z
Zhai, Edwin 已提交
2298

2299
	kvm_vcpu_set_in_spin_loop(me, true);
2300 2301 2302 2303 2304 2305 2306
	/*
	 * 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.
	 */
2307
	for (pass = 0; pass < 2 && !yielded && try; pass++) {
2308
		kvm_for_each_vcpu(i, vcpu, kvm) {
2309
			if (!pass && i <= last_boosted_vcpu) {
2310 2311 2312 2313
				i = last_boosted_vcpu;
				continue;
			} else if (pass && i > last_boosted_vcpu)
				break;
2314 2315
			if (!ACCESS_ONCE(vcpu->preempted))
				continue;
2316 2317
			if (vcpu == me)
				continue;
2318
			if (swait_active(&vcpu->wq) && !kvm_arch_vcpu_runnable(vcpu))
2319
				continue;
2320 2321
			if (!kvm_vcpu_eligible_for_directed_yield(vcpu))
				continue;
2322 2323 2324

			yielded = kvm_vcpu_yield_to(vcpu);
			if (yielded > 0) {
2325 2326
				kvm->last_boosted_vcpu = i;
				break;
2327 2328 2329 2330
			} else if (yielded < 0) {
				try--;
				if (!try)
					break;
2331 2332 2333
			}
		}
	}
2334
	kvm_vcpu_set_in_spin_loop(me, false);
2335 2336 2337

	/* Ensure vcpu is not eligible during next spinloop */
	kvm_vcpu_set_dy_eligible(me, false);
Z
Zhai, Edwin 已提交
2338 2339 2340
}
EXPORT_SYMBOL_GPL(kvm_vcpu_on_spin);

2341
static int kvm_vcpu_fault(struct vm_fault *vmf)
2342
{
2343
	struct kvm_vcpu *vcpu = vmf->vma->vm_file->private_data;
2344 2345
	struct page *page;

2346
	if (vmf->pgoff == 0)
2347
		page = virt_to_page(vcpu->run);
A
Avi Kivity 已提交
2348
#ifdef CONFIG_X86
2349
	else if (vmf->pgoff == KVM_PIO_PAGE_OFFSET)
2350
		page = virt_to_page(vcpu->arch.pio_data);
2351 2352 2353 2354
#endif
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
	else if (vmf->pgoff == KVM_COALESCED_MMIO_PAGE_OFFSET)
		page = virt_to_page(vcpu->kvm->coalesced_mmio_ring);
A
Avi Kivity 已提交
2355
#endif
2356
	else
2357
		return kvm_arch_vcpu_fault(vcpu, vmf);
2358
	get_page(page);
2359 2360
	vmf->page = page;
	return 0;
2361 2362
}

2363
static const struct vm_operations_struct kvm_vcpu_vm_ops = {
2364
	.fault = kvm_vcpu_fault,
2365 2366 2367 2368 2369 2370 2371 2372
};

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
2373 2374 2375 2376
static int kvm_vcpu_release(struct inode *inode, struct file *filp)
{
	struct kvm_vcpu *vcpu = filp->private_data;

2377
	debugfs_remove_recursive(vcpu->debugfs_dentry);
A
Al Viro 已提交
2378
	kvm_put_kvm(vcpu->kvm);
A
Avi Kivity 已提交
2379 2380 2381
	return 0;
}

2382
static struct file_operations kvm_vcpu_fops = {
A
Avi Kivity 已提交
2383 2384
	.release        = kvm_vcpu_release,
	.unlocked_ioctl = kvm_vcpu_ioctl,
2385
#ifdef CONFIG_KVM_COMPAT
2386 2387
	.compat_ioctl   = kvm_vcpu_compat_ioctl,
#endif
2388
	.mmap           = kvm_vcpu_mmap,
2389
	.llseek		= noop_llseek,
A
Avi Kivity 已提交
2390 2391 2392 2393 2394 2395 2396
};

/*
 * Allocates an inode for the vcpu.
 */
static int create_vcpu_fd(struct kvm_vcpu *vcpu)
{
2397
	return anon_inode_getfd("kvm-vcpu", &kvm_vcpu_fops, vcpu, O_RDWR | O_CLOEXEC);
A
Avi Kivity 已提交
2398 2399
}

2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425
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;
}

2426 2427 2428
/*
 * Creates some virtual cpus.  Good luck creating more than one.
 */
2429
static int kvm_vm_ioctl_create_vcpu(struct kvm *kvm, u32 id)
2430 2431
{
	int r;
2432
	struct kvm_vcpu *vcpu;
2433

G
Greg Kurz 已提交
2434
	if (id >= KVM_MAX_VCPU_ID)
2435 2436
		return -EINVAL;

2437 2438 2439 2440 2441 2442 2443 2444 2445
	mutex_lock(&kvm->lock);
	if (kvm->created_vcpus == KVM_MAX_VCPUS) {
		mutex_unlock(&kvm->lock);
		return -EINVAL;
	}

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

2446
	vcpu = kvm_arch_vcpu_create(kvm, id);
2447 2448 2449 2450
	if (IS_ERR(vcpu)) {
		r = PTR_ERR(vcpu);
		goto vcpu_decrement;
	}
2451

2452 2453
	preempt_notifier_init(&vcpu->preempt_notifier, &kvm_preempt_ops);

2454 2455
	r = kvm_arch_vcpu_setup(vcpu);
	if (r)
2456
		goto vcpu_destroy;
2457

2458 2459 2460 2461
	r = kvm_create_vcpu_debugfs(vcpu);
	if (r)
		goto vcpu_destroy;

S
Shaohua Li 已提交
2462
	mutex_lock(&kvm->lock);
2463 2464 2465 2466
	if (kvm_get_vcpu_by_id(kvm, id)) {
		r = -EEXIST;
		goto unlock_vcpu_destroy;
	}
2467 2468

	BUG_ON(kvm->vcpus[atomic_read(&kvm->online_vcpus)]);
2469

R
Rusty Russell 已提交
2470
	/* Now it's all set up, let userspace reach it */
A
Al Viro 已提交
2471
	kvm_get_kvm(kvm);
A
Avi Kivity 已提交
2472
	r = create_vcpu_fd(vcpu);
2473 2474
	if (r < 0) {
		kvm_put_kvm(kvm);
2475
		goto unlock_vcpu_destroy;
2476 2477 2478
	}

	kvm->vcpus[atomic_read(&kvm->online_vcpus)] = vcpu;
2479 2480 2481 2482 2483

	/*
	 * Pairs with smp_rmb() in kvm_get_vcpu.  Write kvm->vcpus
	 * before kvm->online_vcpu's incremented value.
	 */
2484 2485 2486 2487
	smp_wmb();
	atomic_inc(&kvm->online_vcpus);

	mutex_unlock(&kvm->lock);
2488
	kvm_arch_vcpu_postcreate(vcpu);
R
Rusty Russell 已提交
2489
	return r;
2490

2491
unlock_vcpu_destroy:
2492
	mutex_unlock(&kvm->lock);
2493
	debugfs_remove_recursive(vcpu->debugfs_dentry);
2494
vcpu_destroy:
2495
	kvm_arch_vcpu_destroy(vcpu);
2496 2497 2498 2499
vcpu_decrement:
	mutex_lock(&kvm->lock);
	kvm->created_vcpus--;
	mutex_unlock(&kvm->lock);
2500 2501 2502
	return r;
}

A
Avi Kivity 已提交
2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513
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
2514 2515
static long kvm_vcpu_ioctl(struct file *filp,
			   unsigned int ioctl, unsigned long arg)
A
Avi Kivity 已提交
2516
{
A
Avi Kivity 已提交
2517
	struct kvm_vcpu *vcpu = filp->private_data;
A
Al Viro 已提交
2518
	void __user *argp = (void __user *)arg;
2519
	int r;
2520 2521
	struct kvm_fpu *fpu = NULL;
	struct kvm_sregs *kvm_sregs = NULL;
A
Avi Kivity 已提交
2522

2523 2524
	if (vcpu->kvm->mm != current->mm)
		return -EIO;
2525

2526 2527 2528
	if (unlikely(_IOC_TYPE(ioctl) != KVMIO))
		return -EINVAL;

2529
#if defined(CONFIG_S390) || defined(CONFIG_PPC) || defined(CONFIG_MIPS)
2530 2531 2532 2533
	/*
	 * Special cases: vcpu ioctls that are asynchronous to vcpu execution,
	 * so vcpu_load() would break it.
	 */
2534
	if (ioctl == KVM_S390_INTERRUPT || ioctl == KVM_S390_IRQ || ioctl == KVM_INTERRUPT)
2535 2536 2537 2538
		return kvm_arch_vcpu_ioctl(filp, ioctl, arg);
#endif


2539 2540 2541
	r = vcpu_load(vcpu);
	if (r)
		return r;
A
Avi Kivity 已提交
2542
	switch (ioctl) {
2543
	case KVM_RUN:
2544 2545 2546
		r = -EINVAL;
		if (arg)
			goto out;
2547 2548 2549 2550
		if (unlikely(vcpu->pid != current->pids[PIDTYPE_PID].pid)) {
			/* The thread running this VCPU changed. */
			struct pid *oldpid = vcpu->pid;
			struct pid *newpid = get_task_pid(current, PIDTYPE_PID);
2551

2552 2553 2554 2555 2556
			rcu_assign_pointer(vcpu->pid, newpid);
			if (oldpid)
				synchronize_rcu();
			put_pid(oldpid);
		}
2557
		r = kvm_arch_vcpu_ioctl_run(vcpu, vcpu->run);
2558
		trace_kvm_userspace_exit(vcpu->run->exit_reason, r);
A
Avi Kivity 已提交
2559 2560
		break;
	case KVM_GET_REGS: {
2561
		struct kvm_regs *kvm_regs;
A
Avi Kivity 已提交
2562

2563 2564 2565
		r = -ENOMEM;
		kvm_regs = kzalloc(sizeof(struct kvm_regs), GFP_KERNEL);
		if (!kvm_regs)
A
Avi Kivity 已提交
2566
			goto out;
2567 2568 2569
		r = kvm_arch_vcpu_ioctl_get_regs(vcpu, kvm_regs);
		if (r)
			goto out_free1;
A
Avi Kivity 已提交
2570
		r = -EFAULT;
2571 2572
		if (copy_to_user(argp, kvm_regs, sizeof(struct kvm_regs)))
			goto out_free1;
A
Avi Kivity 已提交
2573
		r = 0;
2574 2575
out_free1:
		kfree(kvm_regs);
A
Avi Kivity 已提交
2576 2577 2578
		break;
	}
	case KVM_SET_REGS: {
2579
		struct kvm_regs *kvm_regs;
A
Avi Kivity 已提交
2580

2581
		r = -ENOMEM;
2582 2583 2584
		kvm_regs = memdup_user(argp, sizeof(*kvm_regs));
		if (IS_ERR(kvm_regs)) {
			r = PTR_ERR(kvm_regs);
A
Avi Kivity 已提交
2585
			goto out;
2586
		}
2587 2588
		r = kvm_arch_vcpu_ioctl_set_regs(vcpu, kvm_regs);
		kfree(kvm_regs);
A
Avi Kivity 已提交
2589 2590 2591
		break;
	}
	case KVM_GET_SREGS: {
2592 2593 2594 2595 2596
		kvm_sregs = kzalloc(sizeof(struct kvm_sregs), GFP_KERNEL);
		r = -ENOMEM;
		if (!kvm_sregs)
			goto out;
		r = kvm_arch_vcpu_ioctl_get_sregs(vcpu, kvm_sregs);
A
Avi Kivity 已提交
2597 2598 2599
		if (r)
			goto out;
		r = -EFAULT;
2600
		if (copy_to_user(argp, kvm_sregs, sizeof(struct kvm_sregs)))
A
Avi Kivity 已提交
2601 2602 2603 2604 2605
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_SREGS: {
2606 2607 2608
		kvm_sregs = memdup_user(argp, sizeof(*kvm_sregs));
		if (IS_ERR(kvm_sregs)) {
			r = PTR_ERR(kvm_sregs);
G
Guo Chao 已提交
2609
			kvm_sregs = NULL;
A
Avi Kivity 已提交
2610
			goto out;
2611
		}
2612
		r = kvm_arch_vcpu_ioctl_set_sregs(vcpu, kvm_sregs);
A
Avi Kivity 已提交
2613 2614
		break;
	}
2615 2616 2617 2618 2619 2620 2621
	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;
2622
		if (copy_to_user(argp, &mp_state, sizeof(mp_state)))
2623 2624 2625 2626 2627 2628 2629 2630
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_MP_STATE: {
		struct kvm_mp_state mp_state;

		r = -EFAULT;
2631
		if (copy_from_user(&mp_state, argp, sizeof(mp_state)))
2632 2633 2634 2635
			goto out;
		r = kvm_arch_vcpu_ioctl_set_mpstate(vcpu, &mp_state);
		break;
	}
A
Avi Kivity 已提交
2636 2637 2638 2639
	case KVM_TRANSLATE: {
		struct kvm_translation tr;

		r = -EFAULT;
2640
		if (copy_from_user(&tr, argp, sizeof(tr)))
A
Avi Kivity 已提交
2641
			goto out;
2642
		r = kvm_arch_vcpu_ioctl_translate(vcpu, &tr);
A
Avi Kivity 已提交
2643 2644 2645
		if (r)
			goto out;
		r = -EFAULT;
2646
		if (copy_to_user(argp, &tr, sizeof(tr)))
A
Avi Kivity 已提交
2647 2648 2649 2650
			goto out;
		r = 0;
		break;
	}
J
Jan Kiszka 已提交
2651 2652
	case KVM_SET_GUEST_DEBUG: {
		struct kvm_guest_debug dbg;
A
Avi Kivity 已提交
2653 2654

		r = -EFAULT;
2655
		if (copy_from_user(&dbg, argp, sizeof(dbg)))
A
Avi Kivity 已提交
2656
			goto out;
J
Jan Kiszka 已提交
2657
		r = kvm_arch_vcpu_ioctl_set_guest_debug(vcpu, &dbg);
A
Avi Kivity 已提交
2658 2659
		break;
	}
A
Avi Kivity 已提交
2660 2661 2662 2663 2664 2665 2666 2667 2668
	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,
2669
					   sizeof(kvm_sigmask)))
A
Avi Kivity 已提交
2670 2671
				goto out;
			r = -EINVAL;
2672
			if (kvm_sigmask.len != sizeof(sigset))
A
Avi Kivity 已提交
2673 2674 2675
				goto out;
			r = -EFAULT;
			if (copy_from_user(&sigset, sigmask_arg->sigset,
2676
					   sizeof(sigset)))
A
Avi Kivity 已提交
2677 2678 2679
				goto out;
			p = &sigset;
		}
2680
		r = kvm_vcpu_ioctl_set_sigmask(vcpu, p);
A
Avi Kivity 已提交
2681 2682
		break;
	}
A
Avi Kivity 已提交
2683
	case KVM_GET_FPU: {
2684 2685 2686 2687 2688
		fpu = kzalloc(sizeof(struct kvm_fpu), GFP_KERNEL);
		r = -ENOMEM;
		if (!fpu)
			goto out;
		r = kvm_arch_vcpu_ioctl_get_fpu(vcpu, fpu);
A
Avi Kivity 已提交
2689 2690 2691
		if (r)
			goto out;
		r = -EFAULT;
2692
		if (copy_to_user(argp, fpu, sizeof(struct kvm_fpu)))
A
Avi Kivity 已提交
2693 2694 2695 2696 2697
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_FPU: {
2698 2699 2700
		fpu = memdup_user(argp, sizeof(*fpu));
		if (IS_ERR(fpu)) {
			r = PTR_ERR(fpu);
G
Guo Chao 已提交
2701
			fpu = NULL;
A
Avi Kivity 已提交
2702
			goto out;
2703
		}
2704
		r = kvm_arch_vcpu_ioctl_set_fpu(vcpu, fpu);
A
Avi Kivity 已提交
2705 2706
		break;
	}
A
Avi Kivity 已提交
2707
	default:
2708
		r = kvm_arch_vcpu_ioctl(filp, ioctl, arg);
A
Avi Kivity 已提交
2709 2710
	}
out:
2711
	vcpu_put(vcpu);
2712 2713
	kfree(fpu);
	kfree(kvm_sregs);
A
Avi Kivity 已提交
2714 2715 2716
	return r;
}

2717
#ifdef CONFIG_KVM_COMPAT
2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737
static long kvm_vcpu_compat_ioctl(struct file *filp,
				  unsigned int ioctl, unsigned long arg)
{
	struct kvm_vcpu *vcpu = filp->private_data;
	void __user *argp = compat_ptr(arg);
	int r;

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

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

		if (argp) {
			r = -EFAULT;
			if (copy_from_user(&kvm_sigmask, argp,
2738
					   sizeof(kvm_sigmask)))
2739 2740
				goto out;
			r = -EINVAL;
2741
			if (kvm_sigmask.len != sizeof(csigset))
2742 2743 2744
				goto out;
			r = -EFAULT;
			if (copy_from_user(&csigset, sigmask_arg->sigset,
2745
					   sizeof(csigset)))
2746
				goto out;
2747 2748 2749 2750
			sigset_from_compat(&sigset, &csigset);
			r = kvm_vcpu_ioctl_set_sigmask(vcpu, &sigset);
		} else
			r = kvm_vcpu_ioctl_set_sigmask(vcpu, NULL);
2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761
		break;
	}
	default:
		r = kvm_vcpu_ioctl(filp, ioctl, arg);
	}

out:
	return r;
}
#endif

S
Scott Wood 已提交
2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808
static int kvm_device_ioctl_attr(struct kvm_device *dev,
				 int (*accessor)(struct kvm_device *dev,
						 struct kvm_device_attr *attr),
				 unsigned long arg)
{
	struct kvm_device_attr attr;

	if (!accessor)
		return -EPERM;

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

	return accessor(dev, &attr);
}

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

	switch (ioctl) {
	case KVM_SET_DEVICE_ATTR:
		return kvm_device_ioctl_attr(dev, dev->ops->set_attr, arg);
	case KVM_GET_DEVICE_ATTR:
		return kvm_device_ioctl_attr(dev, dev->ops->get_attr, arg);
	case KVM_HAS_DEVICE_ATTR:
		return kvm_device_ioctl_attr(dev, dev->ops->has_attr, arg);
	default:
		if (dev->ops->ioctl)
			return dev->ops->ioctl(dev, ioctl, arg);

		return -ENOTTY;
	}
}

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

	kvm_put_kvm(kvm);
	return 0;
}

static const struct file_operations kvm_device_fops = {
	.unlocked_ioctl = kvm_device_ioctl,
2809
#ifdef CONFIG_KVM_COMPAT
2810 2811
	.compat_ioctl = kvm_device_ioctl,
#endif
S
Scott Wood 已提交
2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822
	.release = kvm_device_release,
};

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

	return filp->private_data;
}

2823
static struct kvm_device_ops *kvm_device_ops_table[KVM_DEV_TYPE_MAX] = {
2824
#ifdef CONFIG_KVM_MPIC
2825 2826
	[KVM_DEV_TYPE_FSL_MPIC_20]	= &kvm_mpic_ops,
	[KVM_DEV_TYPE_FSL_MPIC_42]	= &kvm_mpic_ops,
2827
#endif
2828

2829
#ifdef CONFIG_KVM_XICS
2830
	[KVM_DEV_TYPE_XICS]		= &kvm_xics_ops,
A
Alex Williamson 已提交
2831
#endif
2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845
};

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;
}

2846 2847 2848 2849 2850 2851
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
2852 2853 2854 2855 2856 2857 2858 2859
static int kvm_ioctl_create_device(struct kvm *kvm,
				   struct kvm_create_device *cd)
{
	struct kvm_device_ops *ops = NULL;
	struct kvm_device *dev;
	bool test = cd->flags & KVM_CREATE_DEVICE_TEST;
	int ret;

2860 2861 2862 2863 2864
	if (cd->type >= ARRAY_SIZE(kvm_device_ops_table))
		return -ENODEV;

	ops = kvm_device_ops_table[cd->type];
	if (ops == NULL)
S
Scott Wood 已提交
2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876
		return -ENODEV;

	if (test)
		return 0;

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

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

2877
	mutex_lock(&kvm->lock);
S
Scott Wood 已提交
2878 2879
	ret = ops->create(dev, cd->type);
	if (ret < 0) {
2880
		mutex_unlock(&kvm->lock);
S
Scott Wood 已提交
2881 2882 2883
		kfree(dev);
		return ret;
	}
2884 2885
	list_add(&dev->vm_node, &kvm->devices);
	mutex_unlock(&kvm->lock);
S
Scott Wood 已提交
2886

2887 2888 2889
	if (ops->init)
		ops->init(dev);

2890
	ret = anon_inode_getfd(ops->name, &kvm_device_fops, dev, O_RDWR | O_CLOEXEC);
S
Scott Wood 已提交
2891
	if (ret < 0) {
2892 2893 2894
		mutex_lock(&kvm->lock);
		list_del(&dev->vm_node);
		mutex_unlock(&kvm->lock);
2895
		ops->destroy(dev);
S
Scott Wood 已提交
2896 2897 2898 2899 2900 2901 2902 2903
		return ret;
	}

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

2904 2905 2906 2907 2908 2909 2910 2911 2912 2913
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
2914
#ifdef CONFIG_HAVE_KVM_IRQFD
2915
	case KVM_CAP_IRQFD:
2916 2917
	case KVM_CAP_IRQFD_RESAMPLE:
#endif
2918
	case KVM_CAP_IOEVENTFD_ANY_LENGTH:
2919 2920 2921 2922 2923
	case KVM_CAP_CHECK_EXTENSION_VM:
		return 1;
#ifdef CONFIG_HAVE_KVM_IRQ_ROUTING
	case KVM_CAP_IRQ_ROUTING:
		return KVM_MAX_IRQ_ROUTES;
2924 2925 2926 2927
#endif
#if KVM_ADDRESS_SPACE_NUM > 1
	case KVM_CAP_MULTI_ADDRESS_SPACE:
		return KVM_ADDRESS_SPACE_NUM;
2928
#endif
G
Greg Kurz 已提交
2929 2930
	case KVM_CAP_MAX_VCPU_ID:
		return KVM_MAX_VCPU_ID;
2931 2932 2933 2934 2935 2936
	default:
		break;
	}
	return kvm_vm_ioctl_check_extension(kvm, arg);
}

A
Avi Kivity 已提交
2937 2938 2939 2940 2941
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;
2942
	int r;
A
Avi Kivity 已提交
2943

2944 2945
	if (kvm->mm != current->mm)
		return -EIO;
A
Avi Kivity 已提交
2946 2947 2948 2949
	switch (ioctl) {
	case KVM_CREATE_VCPU:
		r = kvm_vm_ioctl_create_vcpu(kvm, arg);
		break;
2950 2951 2952 2953 2954
	case KVM_SET_USER_MEMORY_REGION: {
		struct kvm_userspace_memory_region kvm_userspace_mem;

		r = -EFAULT;
		if (copy_from_user(&kvm_userspace_mem, argp,
2955
						sizeof(kvm_userspace_mem)))
2956 2957
			goto out;

2958
		r = kvm_vm_ioctl_set_memory_region(kvm, &kvm_userspace_mem);
A
Avi Kivity 已提交
2959 2960 2961 2962 2963 2964
		break;
	}
	case KVM_GET_DIRTY_LOG: {
		struct kvm_dirty_log log;

		r = -EFAULT;
2965
		if (copy_from_user(&log, argp, sizeof(log)))
A
Avi Kivity 已提交
2966
			goto out;
2967
		r = kvm_vm_ioctl_get_dirty_log(kvm, &log);
A
Avi Kivity 已提交
2968 2969
		break;
	}
2970 2971 2972
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
	case KVM_REGISTER_COALESCED_MMIO: {
		struct kvm_coalesced_mmio_zone zone;
2973

2974
		r = -EFAULT;
2975
		if (copy_from_user(&zone, argp, sizeof(zone)))
2976 2977 2978 2979 2980 2981
			goto out;
		r = kvm_vm_ioctl_register_coalesced_mmio(kvm, &zone);
		break;
	}
	case KVM_UNREGISTER_COALESCED_MMIO: {
		struct kvm_coalesced_mmio_zone zone;
2982

2983
		r = -EFAULT;
2984
		if (copy_from_user(&zone, argp, sizeof(zone)))
2985 2986 2987 2988 2989
			goto out;
		r = kvm_vm_ioctl_unregister_coalesced_mmio(kvm, &zone);
		break;
	}
#endif
G
Gregory Haskins 已提交
2990 2991 2992 2993
	case KVM_IRQFD: {
		struct kvm_irqfd data;

		r = -EFAULT;
2994
		if (copy_from_user(&data, argp, sizeof(data)))
G
Gregory Haskins 已提交
2995
			goto out;
2996
		r = kvm_irqfd(kvm, &data);
G
Gregory Haskins 已提交
2997 2998
		break;
	}
G
Gregory Haskins 已提交
2999 3000 3001 3002
	case KVM_IOEVENTFD: {
		struct kvm_ioeventfd data;

		r = -EFAULT;
3003
		if (copy_from_user(&data, argp, sizeof(data)))
G
Gregory Haskins 已提交
3004 3005 3006 3007
			goto out;
		r = kvm_ioeventfd(kvm, &data);
		break;
	}
3008 3009 3010 3011 3012
#ifdef CONFIG_HAVE_KVM_MSI
	case KVM_SIGNAL_MSI: {
		struct kvm_msi msi;

		r = -EFAULT;
3013
		if (copy_from_user(&msi, argp, sizeof(msi)))
3014 3015 3016 3017
			goto out;
		r = kvm_send_userspace_msi(kvm, &msi);
		break;
	}
3018 3019 3020 3021 3022 3023 3024
#endif
#ifdef __KVM_HAVE_IRQ_LINE
	case KVM_IRQ_LINE_STATUS:
	case KVM_IRQ_LINE: {
		struct kvm_irq_level irq_event;

		r = -EFAULT;
3025
		if (copy_from_user(&irq_event, argp, sizeof(irq_event)))
3026 3027
			goto out;

3028 3029
		r = kvm_vm_ioctl_irq_line(kvm, &irq_event,
					ioctl == KVM_IRQ_LINE_STATUS);
3030 3031 3032 3033 3034
		if (r)
			goto out;

		r = -EFAULT;
		if (ioctl == KVM_IRQ_LINE_STATUS) {
3035
			if (copy_to_user(argp, &irq_event, sizeof(irq_event)))
3036 3037 3038 3039 3040 3041
				goto out;
		}

		r = 0;
		break;
	}
3042
#endif
3043 3044 3045 3046
#ifdef CONFIG_HAVE_KVM_IRQ_ROUTING
	case KVM_SET_GSI_ROUTING: {
		struct kvm_irq_routing routing;
		struct kvm_irq_routing __user *urouting;
3047
		struct kvm_irq_routing_entry *entries = NULL;
3048 3049 3050 3051 3052

		r = -EFAULT;
		if (copy_from_user(&routing, argp, sizeof(routing)))
			goto out;
		r = -EINVAL;
3053
		if (routing.nr > KVM_MAX_IRQ_ROUTES)
3054 3055 3056
			goto out;
		if (routing.flags)
			goto out;
3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067
		if (routing.nr) {
			r = -ENOMEM;
			entries = vmalloc(routing.nr * sizeof(*entries));
			if (!entries)
				goto out;
			r = -EFAULT;
			urouting = argp;
			if (copy_from_user(entries, urouting->entries,
					   routing.nr * sizeof(*entries)))
				goto out_free_irq_routing;
		}
3068 3069
		r = kvm_set_irq_routing(kvm, entries, routing.nr,
					routing.flags);
3070
out_free_irq_routing:
3071 3072 3073 3074
		vfree(entries);
		break;
	}
#endif /* CONFIG_HAVE_KVM_IRQ_ROUTING */
S
Scott Wood 已提交
3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092
	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;
	}
3093 3094 3095
	case KVM_CHECK_EXTENSION:
		r = kvm_vm_ioctl_check_extension_generic(kvm, arg);
		break;
3096
	default:
3097
		r = kvm_arch_vm_ioctl(filp, ioctl, arg);
3098 3099 3100 3101 3102
	}
out:
	return r;
}

3103
#ifdef CONFIG_KVM_COMPAT
3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127
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)))
3128
			return -EFAULT;
3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143
		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

3144
static struct file_operations kvm_vm_fops = {
3145 3146
	.release        = kvm_vm_release,
	.unlocked_ioctl = kvm_vm_ioctl,
3147
#ifdef CONFIG_KVM_COMPAT
3148 3149
	.compat_ioctl   = kvm_vm_compat_ioctl,
#endif
3150
	.llseek		= noop_llseek,
3151 3152
};

3153
static int kvm_dev_ioctl_create_vm(unsigned long type)
3154
{
3155
	int r;
3156
	struct kvm *kvm;
3157
	struct file *file;
3158

3159
	kvm = kvm_create_vm(type);
3160 3161
	if (IS_ERR(kvm))
		return PTR_ERR(kvm);
3162 3163 3164 3165 3166 3167 3168
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
	r = kvm_coalesced_mmio_init(kvm);
	if (r < 0) {
		kvm_put_kvm(kvm);
		return r;
	}
#endif
3169
	r = get_unused_fd_flags(O_CLOEXEC);
3170
	if (r < 0) {
A
Al Viro 已提交
3171
		kvm_put_kvm(kvm);
3172 3173
		return r;
	}
3174 3175 3176 3177 3178 3179
	file = anon_inode_getfile("kvm-vm", &kvm_vm_fops, kvm, O_RDWR);
	if (IS_ERR(file)) {
		put_unused_fd(r);
		kvm_put_kvm(kvm);
		return PTR_ERR(file);
	}
3180 3181

	if (kvm_create_vm_debugfs(kvm, r) < 0) {
3182 3183
		put_unused_fd(r);
		fput(file);
3184 3185
		return -ENOMEM;
	}
3186

3187
	fd_install(r, file);
3188
	return r;
3189 3190 3191 3192 3193
}

static long kvm_dev_ioctl(struct file *filp,
			  unsigned int ioctl, unsigned long arg)
{
3194
	long r = -EINVAL;
3195 3196 3197

	switch (ioctl) {
	case KVM_GET_API_VERSION:
3198 3199
		if (arg)
			goto out;
3200 3201 3202
		r = KVM_API_VERSION;
		break;
	case KVM_CREATE_VM:
3203
		r = kvm_dev_ioctl_create_vm(arg);
3204
		break;
3205
	case KVM_CHECK_EXTENSION:
3206
		r = kvm_vm_ioctl_check_extension_generic(NULL, arg);
3207
		break;
3208 3209 3210
	case KVM_GET_VCPU_MMAP_SIZE:
		if (arg)
			goto out;
3211 3212 3213
		r = PAGE_SIZE;     /* struct kvm_run */
#ifdef CONFIG_X86
		r += PAGE_SIZE;    /* pio data page */
3214 3215 3216
#endif
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
		r += PAGE_SIZE;    /* coalesced mmio ring page */
3217
#endif
3218
		break;
3219 3220 3221
	case KVM_TRACE_ENABLE:
	case KVM_TRACE_PAUSE:
	case KVM_TRACE_DISABLE:
3222
		r = -EOPNOTSUPP;
3223
		break;
A
Avi Kivity 已提交
3224
	default:
3225
		return kvm_arch_dev_ioctl(filp, ioctl, arg);
A
Avi Kivity 已提交
3226 3227 3228 3229 3230 3231 3232 3233
	}
out:
	return r;
}

static struct file_operations kvm_chardev_ops = {
	.unlocked_ioctl = kvm_dev_ioctl,
	.compat_ioctl   = kvm_dev_ioctl,
3234
	.llseek		= noop_llseek,
A
Avi Kivity 已提交
3235 3236 3237
};

static struct miscdevice kvm_dev = {
A
Avi Kivity 已提交
3238
	KVM_MINOR,
A
Avi Kivity 已提交
3239 3240 3241 3242
	"kvm",
	&kvm_chardev_ops,
};

3243
static void hardware_enable_nolock(void *junk)
3244 3245
{
	int cpu = raw_smp_processor_id();
3246
	int r;
3247

3248
	if (cpumask_test_cpu(cpu, cpus_hardware_enabled))
3249
		return;
3250

3251
	cpumask_set_cpu(cpu, cpus_hardware_enabled);
3252

3253
	r = kvm_arch_hardware_enable();
3254 3255 3256 3257

	if (r) {
		cpumask_clear_cpu(cpu, cpus_hardware_enabled);
		atomic_inc(&hardware_enable_failed);
X
Xiubo Li 已提交
3258
		pr_info("kvm: enabling virtualization on CPU%d failed\n", cpu);
3259
	}
3260 3261
}

3262
static int kvm_starting_cpu(unsigned int cpu)
3263
{
3264
	raw_spin_lock(&kvm_count_lock);
3265 3266
	if (kvm_usage_count)
		hardware_enable_nolock(NULL);
3267
	raw_spin_unlock(&kvm_count_lock);
3268
	return 0;
3269 3270 3271
}

static void hardware_disable_nolock(void *junk)
3272 3273 3274
{
	int cpu = raw_smp_processor_id();

3275
	if (!cpumask_test_cpu(cpu, cpus_hardware_enabled))
3276
		return;
3277
	cpumask_clear_cpu(cpu, cpus_hardware_enabled);
3278
	kvm_arch_hardware_disable();
3279 3280
}

3281
static int kvm_dying_cpu(unsigned int cpu)
3282
{
3283
	raw_spin_lock(&kvm_count_lock);
3284 3285
	if (kvm_usage_count)
		hardware_disable_nolock(NULL);
3286
	raw_spin_unlock(&kvm_count_lock);
3287
	return 0;
3288 3289
}

3290 3291 3292 3293 3294 3295
static void hardware_disable_all_nolock(void)
{
	BUG_ON(!kvm_usage_count);

	kvm_usage_count--;
	if (!kvm_usage_count)
3296
		on_each_cpu(hardware_disable_nolock, NULL, 1);
3297 3298 3299 3300
}

static void hardware_disable_all(void)
{
3301
	raw_spin_lock(&kvm_count_lock);
3302
	hardware_disable_all_nolock();
3303
	raw_spin_unlock(&kvm_count_lock);
3304 3305 3306 3307 3308 3309
}

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

3310
	raw_spin_lock(&kvm_count_lock);
3311 3312 3313 3314

	kvm_usage_count++;
	if (kvm_usage_count == 1) {
		atomic_set(&hardware_enable_failed, 0);
3315
		on_each_cpu(hardware_enable_nolock, NULL, 1);
3316 3317 3318 3319 3320 3321 3322

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

3323
	raw_spin_unlock(&kvm_count_lock);
3324 3325 3326 3327

	return r;
}

3328
static int kvm_reboot(struct notifier_block *notifier, unsigned long val,
M
Mike Day 已提交
3329
		      void *v)
3330
{
3331 3332 3333 3334 3335 3336
	/*
	 * 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
3337
	pr_info("kvm: exiting hardware virtualization\n");
3338
	kvm_rebooting = true;
3339
	on_each_cpu(hardware_disable_nolock, NULL, 1);
3340 3341 3342 3343 3344 3345 3346 3347
	return NOTIFY_OK;
}

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

M
Marcelo Tosatti 已提交
3348
static void kvm_io_bus_destroy(struct kvm_io_bus *bus)
3349 3350 3351 3352
{
	int i;

	for (i = 0; i < bus->dev_count; i++) {
3353
		struct kvm_io_device *pos = bus->range[i].dev;
3354 3355 3356

		kvm_iodevice_destructor(pos);
	}
M
Marcelo Tosatti 已提交
3357
	kfree(bus);
3358 3359
}

3360
static inline int kvm_io_bus_cmp(const struct kvm_io_range *r1,
X
Xiubo Li 已提交
3361
				 const struct kvm_io_range *r2)
3362
{
J
Jason Wang 已提交
3363 3364 3365 3366
	gpa_t addr1 = r1->addr;
	gpa_t addr2 = r2->addr;

	if (addr1 < addr2)
3367
		return -1;
J
Jason Wang 已提交
3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379

	/* 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)
3380
		return 1;
J
Jason Wang 已提交
3381

3382 3383 3384
	return 0;
}

3385 3386
static int kvm_io_bus_sort_cmp(const void *p1, const void *p2)
{
3387
	return kvm_io_bus_cmp(p1, p2);
3388 3389
}

G
Geoff Levand 已提交
3390
static int kvm_io_bus_insert_dev(struct kvm_io_bus *bus, struct kvm_io_device *dev,
3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404
			  gpa_t addr, int len)
{
	bus->range[bus->dev_count++] = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
		.dev = dev,
	};

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

	return 0;
}

G
Geoff Levand 已提交
3405
static int kvm_io_bus_get_first_dev(struct kvm_io_bus *bus,
3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422
			     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;

3423
	while (off > 0 && kvm_io_bus_cmp(&key, &bus->range[off-1]) == 0)
3424 3425 3426 3427 3428
		off--;

	return off;
}

3429
static int __kvm_io_bus_write(struct kvm_vcpu *vcpu, struct kvm_io_bus *bus,
C
Cornelia Huck 已提交
3430 3431 3432 3433 3434 3435 3436 3437 3438
			      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 &&
3439
		kvm_io_bus_cmp(range, &bus->range[idx]) == 0) {
3440
		if (!kvm_iodevice_write(vcpu, bus->range[idx].dev, range->addr,
C
Cornelia Huck 已提交
3441 3442 3443 3444 3445 3446 3447 3448
					range->len, val))
			return idx;
		idx++;
	}

	return -EOPNOTSUPP;
}

3449
/* kvm_io_bus_write - called under kvm->slots_lock */
3450
int kvm_io_bus_write(struct kvm_vcpu *vcpu, enum kvm_bus bus_idx, gpa_t addr,
3451
		     int len, const void *val)
3452
{
3453
	struct kvm_io_bus *bus;
3454
	struct kvm_io_range range;
C
Cornelia Huck 已提交
3455
	int r;
3456 3457 3458 3459 3460

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};
3461

3462
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
3463 3464
	if (!bus)
		return -ENOMEM;
3465
	r = __kvm_io_bus_write(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
3466 3467 3468 3469
	return r < 0 ? r : 0;
}

/* kvm_io_bus_write_cookie - called under kvm->slots_lock */
3470 3471
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
3472 3473 3474 3475 3476 3477 3478 3479 3480
{
	struct kvm_io_bus *bus;
	struct kvm_io_range range;

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

3481
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
3482 3483
	if (!bus)
		return -ENOMEM;
C
Cornelia Huck 已提交
3484 3485 3486

	/* First try the device referenced by cookie. */
	if ((cookie >= 0) && (cookie < bus->dev_count) &&
3487
	    (kvm_io_bus_cmp(&range, &bus->range[cookie]) == 0))
3488
		if (!kvm_iodevice_write(vcpu, bus->range[cookie].dev, addr, len,
C
Cornelia Huck 已提交
3489 3490 3491 3492 3493 3494 3495
					val))
			return cookie;

	/*
	 * cookie contained garbage; fall back to search and return the
	 * correct cookie value.
	 */
3496
	return __kvm_io_bus_write(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
3497 3498
}

3499 3500
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
3501 3502 3503 3504
{
	int idx;

	idx = kvm_io_bus_get_first_dev(bus, range->addr, range->len);
3505 3506 3507 3508
	if (idx < 0)
		return -EOPNOTSUPP;

	while (idx < bus->dev_count &&
3509
		kvm_io_bus_cmp(range, &bus->range[idx]) == 0) {
3510
		if (!kvm_iodevice_read(vcpu, bus->range[idx].dev, range->addr,
C
Cornelia Huck 已提交
3511 3512
				       range->len, val))
			return idx;
3513 3514 3515
		idx++;
	}

3516 3517
	return -EOPNOTSUPP;
}
3518
EXPORT_SYMBOL_GPL(kvm_io_bus_write);
3519

3520
/* kvm_io_bus_read - called under kvm->slots_lock */
3521
int kvm_io_bus_read(struct kvm_vcpu *vcpu, enum kvm_bus bus_idx, gpa_t addr,
M
Marcelo Tosatti 已提交
3522
		    int len, void *val)
3523
{
3524
	struct kvm_io_bus *bus;
3525
	struct kvm_io_range range;
C
Cornelia Huck 已提交
3526
	int r;
3527 3528 3529 3530 3531

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};
M
Marcelo Tosatti 已提交
3532

3533
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
3534 3535
	if (!bus)
		return -ENOMEM;
3536
	r = __kvm_io_bus_read(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
3537 3538
	return r < 0 ? r : 0;
}
3539

3540

3541
/* Caller must hold slots_lock. */
3542 3543
int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
			    int len, struct kvm_io_device *dev)
3544
{
M
Marcelo Tosatti 已提交
3545
	struct kvm_io_bus *new_bus, *bus;
3546

M
Marcelo Tosatti 已提交
3547
	bus = kvm->buses[bus_idx];
3548 3549 3550
	if (!bus)
		return -ENOMEM;

3551 3552
	/* exclude ioeventfd which is limited by maximum fd */
	if (bus->dev_count - bus->ioeventfd_count > NR_IOBUS_DEVS - 1)
3553
		return -ENOSPC;
3554

3555
	new_bus = kmalloc(sizeof(*bus) + ((bus->dev_count + 1) *
3556
			  sizeof(struct kvm_io_range)), GFP_KERNEL);
M
Marcelo Tosatti 已提交
3557 3558
	if (!new_bus)
		return -ENOMEM;
3559 3560
	memcpy(new_bus, bus, sizeof(*bus) + (bus->dev_count *
	       sizeof(struct kvm_io_range)));
3561
	kvm_io_bus_insert_dev(new_bus, dev, addr, len);
M
Marcelo Tosatti 已提交
3562 3563 3564
	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
	synchronize_srcu_expedited(&kvm->srcu);
	kfree(bus);
3565 3566 3567 3568

	return 0;
}

3569
/* Caller must hold slots_lock. */
3570 3571
void kvm_io_bus_unregister_dev(struct kvm *kvm, enum kvm_bus bus_idx,
			       struct kvm_io_device *dev)
3572
{
3573
	int i;
M
Marcelo Tosatti 已提交
3574
	struct kvm_io_bus *new_bus, *bus;
3575

3576
	bus = kvm->buses[bus_idx];
3577
	if (!bus)
3578
		return;
3579

3580 3581
	for (i = 0; i < bus->dev_count; i++)
		if (bus->range[i].dev == dev) {
3582 3583
			break;
		}
M
Marcelo Tosatti 已提交
3584

3585 3586
	if (i == bus->dev_count)
		return;
3587

3588
	new_bus = kmalloc(sizeof(*bus) + ((bus->dev_count - 1) *
3589
			  sizeof(struct kvm_io_range)), GFP_KERNEL);
3590 3591 3592 3593
	if (!new_bus)  {
		pr_err("kvm: failed to shrink bus, removing it completely\n");
		goto broken;
	}
3594 3595 3596 3597 3598

	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
3599

3600
broken:
M
Marcelo Tosatti 已提交
3601 3602 3603
	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
	synchronize_srcu_expedited(&kvm->srcu);
	kfree(bus);
3604
	return;
3605 3606
}

3607 3608 3609 3610 3611 3612 3613 3614 3615 3616
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);
3617 3618
	if (!bus)
		goto out_unlock;
3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632

	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);

3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644
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.
	 */
3645
	if (!refcount_inc_not_zero(&stat_data->kvm->users_count))
3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670
		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;

3671
	*val = *(ulong *)((void *)stat_data->kvm + stat_data->offset);
3672 3673 3674 3675

	return 0;
}

3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687
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;
}

3688 3689 3690 3691
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,
3692
				vm_stat_clear_per_vm, "%llu\n");
3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712
}

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

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

	*val = 0;

	kvm_for_each_vcpu(i, vcpu, stat_data->kvm)
3713
		*val += *(u64 *)((void *)vcpu + stat_data->offset);
3714 3715 3716 3717

	return 0;
}

3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732
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;
}

3733 3734 3735 3736
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,
3737
				 vcpu_stat_clear_per_vm, "%llu\n");
3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753
}

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

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

3754
static int vm_stat_get(void *_offset, u64 *val)
3755 3756 3757
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;
3758 3759
	struct kvm_stat_data stat_tmp = {.offset = offset};
	u64 tmp_val;
3760

3761
	*val = 0;
3762
	spin_lock(&kvm_lock);
3763 3764 3765 3766 3767
	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;
	}
3768
	spin_unlock(&kvm_lock);
3769
	return 0;
3770 3771
}

3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791
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");
3792

3793
static int vcpu_stat_get(void *_offset, u64 *val)
A
Avi Kivity 已提交
3794 3795 3796
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;
3797 3798
	struct kvm_stat_data stat_tmp = {.offset = offset};
	u64 tmp_val;
A
Avi Kivity 已提交
3799

3800
	*val = 0;
3801
	spin_lock(&kvm_lock);
3802 3803 3804 3805 3806
	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;
	}
3807
	spin_unlock(&kvm_lock);
3808
	return 0;
A
Avi Kivity 已提交
3809 3810
}

3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831
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");
3832

3833
static const struct file_operations *stat_fops[] = {
3834 3835 3836
	[KVM_STAT_VCPU] = &vcpu_stat_fops,
	[KVM_STAT_VM]   = &vm_stat_fops,
};
A
Avi Kivity 已提交
3837

3838
static int kvm_init_debug(void)
A
Avi Kivity 已提交
3839
{
3840
	int r = -EEXIST;
A
Avi Kivity 已提交
3841 3842
	struct kvm_stats_debugfs_item *p;

3843
	kvm_debugfs_dir = debugfs_create_dir("kvm", NULL);
3844 3845 3846
	if (kvm_debugfs_dir == NULL)
		goto out;

3847 3848
	kvm_debugfs_num_entries = 0;
	for (p = debugfs_entries; p->name; ++p, kvm_debugfs_num_entries++) {
3849
		if (!debugfs_create_file(p->name, 0644, kvm_debugfs_dir,
3850 3851
					 (void *)(long)p->offset,
					 stat_fops[p->kind]))
3852 3853 3854 3855 3856 3857 3858 3859 3860
			goto out_dir;
	}

	return 0;

out_dir:
	debugfs_remove_recursive(kvm_debugfs_dir);
out:
	return r;
A
Avi Kivity 已提交
3861 3862
}

3863
static int kvm_suspend(void)
3864
{
3865
	if (kvm_usage_count)
3866
		hardware_disable_nolock(NULL);
3867 3868 3869
	return 0;
}

3870
static void kvm_resume(void)
3871
{
3872
	if (kvm_usage_count) {
3873
		WARN_ON(raw_spin_is_locked(&kvm_count_lock));
3874
		hardware_enable_nolock(NULL);
3875
	}
3876 3877
}

3878
static struct syscore_ops kvm_syscore_ops = {
3879 3880 3881 3882
	.suspend = kvm_suspend,
	.resume = kvm_resume,
};

3883 3884 3885 3886 3887 3888 3889 3890 3891
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);
3892

3893 3894
	if (vcpu->preempted)
		vcpu->preempted = false;
3895

R
Radim Krčmář 已提交
3896 3897
	kvm_arch_sched_in(vcpu, cpu);

3898
	kvm_arch_vcpu_load(vcpu, cpu);
3899 3900 3901 3902 3903 3904 3905
}

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

3906 3907
	if (current->state == TASK_RUNNING)
		vcpu->preempted = true;
3908
	kvm_arch_vcpu_put(vcpu);
3909 3910
}

3911
int kvm_init(void *opaque, unsigned vcpu_size, unsigned vcpu_align,
3912
		  struct module *module)
A
Avi Kivity 已提交
3913 3914
{
	int r;
Y
Yang, Sheng 已提交
3915
	int cpu;
A
Avi Kivity 已提交
3916

3917 3918
	r = kvm_arch_init(opaque);
	if (r)
3919
		goto out_fail;
3920

3921 3922 3923 3924
	/*
	 * 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
3925 3926
	 * kvm_arch_init must be called before kvm_irqfd_init to avoid creating
	 * conflicts in case kvm is already setup for another implementation.
3927
	 */
P
Paolo Bonzini 已提交
3928 3929 3930
	r = kvm_irqfd_init();
	if (r)
		goto out_irqfd;
3931

3932
	if (!zalloc_cpumask_var(&cpus_hardware_enabled, GFP_KERNEL)) {
3933 3934 3935 3936
		r = -ENOMEM;
		goto out_free_0;
	}

3937
	r = kvm_arch_hardware_setup();
A
Avi Kivity 已提交
3938
	if (r < 0)
3939
		goto out_free_0a;
A
Avi Kivity 已提交
3940

Y
Yang, Sheng 已提交
3941 3942
	for_each_online_cpu(cpu) {
		smp_call_function_single(cpu,
3943
				kvm_arch_check_processor_compat,
3944
				&r, 1);
Y
Yang, Sheng 已提交
3945
		if (r < 0)
3946
			goto out_free_1;
Y
Yang, Sheng 已提交
3947 3948
	}

T
Thomas Gleixner 已提交
3949
	r = cpuhp_setup_state_nocalls(CPUHP_AP_KVM_STARTING, "kvm/cpu:starting",
3950
				      kvm_starting_cpu, kvm_dying_cpu);
A
Avi Kivity 已提交
3951
	if (r)
3952
		goto out_free_2;
A
Avi Kivity 已提交
3953 3954
	register_reboot_notifier(&kvm_reboot_notifier);

3955
	/* A kmem cache lets us meet the alignment requirements of fx_save. */
3956 3957 3958
	if (!vcpu_align)
		vcpu_align = __alignof__(struct kvm_vcpu);
	kvm_vcpu_cache = kmem_cache_create("kvm_vcpu", vcpu_size, vcpu_align,
J
Joe Perches 已提交
3959
					   0, NULL);
3960 3961
	if (!kvm_vcpu_cache) {
		r = -ENOMEM;
3962
		goto out_free_3;
3963 3964
	}

3965 3966 3967 3968
	r = kvm_async_pf_init();
	if (r)
		goto out_free;

A
Avi Kivity 已提交
3969
	kvm_chardev_ops.owner = module;
3970 3971
	kvm_vm_fops.owner = module;
	kvm_vcpu_fops.owner = module;
A
Avi Kivity 已提交
3972 3973 3974

	r = misc_register(&kvm_dev);
	if (r) {
X
Xiubo Li 已提交
3975
		pr_err("kvm: misc device register failed\n");
3976
		goto out_unreg;
A
Avi Kivity 已提交
3977 3978
	}

3979 3980
	register_syscore_ops(&kvm_syscore_ops);

3981 3982 3983
	kvm_preempt_ops.sched_in = kvm_sched_in;
	kvm_preempt_ops.sched_out = kvm_sched_out;

3984 3985
	r = kvm_init_debug();
	if (r) {
X
Xiubo Li 已提交
3986
		pr_err("kvm: create debugfs files failed\n");
3987 3988
		goto out_undebugfs;
	}
3989

P
Paolo Bonzini 已提交
3990 3991 3992
	r = kvm_vfio_ops_init();
	WARN_ON(r);

3993
	return 0;
A
Avi Kivity 已提交
3994

3995 3996
out_undebugfs:
	unregister_syscore_ops(&kvm_syscore_ops);
3997
	misc_deregister(&kvm_dev);
3998 3999
out_unreg:
	kvm_async_pf_deinit();
A
Avi Kivity 已提交
4000
out_free:
4001
	kmem_cache_destroy(kvm_vcpu_cache);
4002
out_free_3:
A
Avi Kivity 已提交
4003
	unregister_reboot_notifier(&kvm_reboot_notifier);
4004
	cpuhp_remove_state_nocalls(CPUHP_AP_KVM_STARTING);
4005 4006
out_free_2:
out_free_1:
4007
	kvm_arch_hardware_unsetup();
4008 4009
out_free_0a:
	free_cpumask_var(cpus_hardware_enabled);
4010
out_free_0:
4011
	kvm_irqfd_exit();
P
Paolo Bonzini 已提交
4012
out_irqfd:
4013 4014
	kvm_arch_exit();
out_fail:
A
Avi Kivity 已提交
4015 4016
	return r;
}
4017
EXPORT_SYMBOL_GPL(kvm_init);
A
Avi Kivity 已提交
4018

4019
void kvm_exit(void)
A
Avi Kivity 已提交
4020
{
4021
	debugfs_remove_recursive(kvm_debugfs_dir);
A
Avi Kivity 已提交
4022
	misc_deregister(&kvm_dev);
4023
	kmem_cache_destroy(kvm_vcpu_cache);
4024
	kvm_async_pf_deinit();
4025
	unregister_syscore_ops(&kvm_syscore_ops);
A
Avi Kivity 已提交
4026
	unregister_reboot_notifier(&kvm_reboot_notifier);
4027
	cpuhp_remove_state_nocalls(CPUHP_AP_KVM_STARTING);
4028
	on_each_cpu(hardware_disable_nolock, NULL, 1);
4029
	kvm_arch_hardware_unsetup();
4030
	kvm_arch_exit();
4031
	kvm_irqfd_exit();
4032
	free_cpumask_var(cpus_hardware_enabled);
4033
	kvm_vfio_ops_exit();
A
Avi Kivity 已提交
4034
}
4035
EXPORT_SYMBOL_GPL(kvm_exit);