kvm_main.c 101.4 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0-only
A
Avi Kivity 已提交
2 3 4 5 6 7 8
/*
 * 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
9
 * Copyright 2010 Red Hat, Inc. and/or its affiliates.
A
Avi Kivity 已提交
10 11 12 13 14 15
 *
 * Authors:
 *   Avi Kivity   <avi@qumranet.com>
 *   Yaniv Kamay  <yaniv@qumranet.com>
 */

16
#include <kvm/iodev.h>
A
Avi Kivity 已提交
17

18
#include <linux/kvm_host.h>
A
Avi Kivity 已提交
19 20 21 22 23 24 25 26 27 28 29
#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>
30
#include <linux/syscore_ops.h>
A
Avi Kivity 已提交
31
#include <linux/cpu.h>
32
#include <linux/sched/signal.h>
33
#include <linux/sched/mm.h>
34
#include <linux/sched/stat.h>
35 36
#include <linux/cpumask.h>
#include <linux/smp.h>
37
#include <linux/anon_inodes.h>
38
#include <linux/profile.h>
39
#include <linux/kvm_para.h>
40
#include <linux/pagemap.h>
41
#include <linux/mman.h>
42
#include <linux/swap.h>
43
#include <linux/bitops.h>
44
#include <linux/spinlock.h>
45
#include <linux/compat.h>
46
#include <linux/srcu.h>
J
Joerg Roedel 已提交
47
#include <linux/hugetlb.h>
48
#include <linux/slab.h>
49 50
#include <linux/sort.h>
#include <linux/bsearch.h>
P
Paolo Bonzini 已提交
51
#include <linux/io.h>
52
#include <linux/lockdep.h>
A
Avi Kivity 已提交
53

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

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

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

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

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

72
/* Architectures should define their poll value according to the halt latency */
73
unsigned int halt_poll_ns = KVM_HALT_POLL_NS_DEFAULT;
74
module_param(halt_poll_ns, uint, 0644);
75
EXPORT_SYMBOL_GPL(halt_poll_ns);
76

W
Wanpeng Li 已提交
77
/* Default doubles per-vcpu halt_poll_ns. */
78
unsigned int halt_poll_ns_grow = 2;
79
module_param(halt_poll_ns_grow, uint, 0644);
80
EXPORT_SYMBOL_GPL(halt_poll_ns_grow);
W
Wanpeng Li 已提交
81

82 83 84 85 86
/* The start value to grow halt_poll_ns from */
unsigned int halt_poll_ns_grow_start = 10000; /* 10us */
module_param(halt_poll_ns_grow_start, uint, 0644);
EXPORT_SYMBOL_GPL(halt_poll_ns_grow_start);

W
Wanpeng Li 已提交
87
/* Default resets per-vcpu halt_poll_ns . */
88
unsigned int halt_poll_ns_shrink;
89
module_param(halt_poll_ns_shrink, uint, 0644);
90
EXPORT_SYMBOL_GPL(halt_poll_ns_shrink);
W
Wanpeng Li 已提交
91

92 93 94
/*
 * Ordering of locks:
 *
95
 *	kvm->lock --> kvm->slots_lock --> kvm->irq_lock
96 97
 */

J
Junaid Shahid 已提交
98
DEFINE_MUTEX(kvm_lock);
99
static DEFINE_RAW_SPINLOCK(kvm_count_lock);
100
LIST_HEAD(vm_list);
101

102
static cpumask_var_t cpus_hardware_enabled;
103
static int kvm_usage_count;
104
static atomic_t hardware_enable_failed;
105

106 107
struct kmem_cache *kvm_vcpu_cache;
EXPORT_SYMBOL_GPL(kvm_vcpu_cache);
A
Avi Kivity 已提交
108

109 110
static __read_mostly struct preempt_ops kvm_preempt_ops;

111
struct dentry *kvm_debugfs_dir;
112
EXPORT_SYMBOL_GPL(kvm_debugfs_dir);
A
Avi Kivity 已提交
113

114 115 116
static int kvm_debugfs_num_entries;
static const struct file_operations *stat_fops_per_vm[];

A
Avi Kivity 已提交
117 118
static long kvm_vcpu_ioctl(struct file *file, unsigned int ioctl,
			   unsigned long arg);
119
#ifdef CONFIG_KVM_COMPAT
120 121
static long kvm_vcpu_compat_ioctl(struct file *file, unsigned int ioctl,
				  unsigned long arg);
122 123 124 125 126
#define KVM_COMPAT(c)	.compat_ioctl	= (c)
#else
static long kvm_no_compat_ioctl(struct file *file, unsigned int ioctl,
				unsigned long arg) { return -EINVAL; }
#define KVM_COMPAT(c)	.compat_ioctl	= kvm_no_compat_ioctl
127
#endif
128 129
static int hardware_enable_all(void);
static void hardware_disable_all(void);
A
Avi Kivity 已提交
130

M
Marcelo Tosatti 已提交
131
static void kvm_io_bus_destroy(struct kvm_io_bus *bus);
132

133
static void mark_page_dirty_in_slot(struct kvm_memory_slot *memslot, gfn_t gfn);
M
Marcelo Tosatti 已提交
134

135
__visible bool kvm_rebooting;
136
EXPORT_SYMBOL_GPL(kvm_rebooting);
137

138 139
static bool largepages_enabled = true;

140 141 142 143 144 145
#define KVM_EVENT_CREATE_VM 0
#define KVM_EVENT_DESTROY_VM 1
static void kvm_uevent_notify_change(unsigned int type, struct kvm *kvm);
static unsigned long long kvm_createvm_count;
static unsigned long long kvm_active_vms;

146 147
__weak int kvm_arch_mmu_notifier_invalidate_range(struct kvm *kvm,
		unsigned long start, unsigned long end, bool blockable)
148
{
149
	return 0;
150 151
}

D
Dan Williams 已提交
152
bool kvm_is_reserved_pfn(kvm_pfn_t pfn)
B
Ben-Ami Yassour 已提交
153
{
154
	if (pfn_valid(pfn))
155
		return PageReserved(pfn_to_page(pfn));
B
Ben-Ami Yassour 已提交
156 157 158 159

	return true;
}

A
Avi Kivity 已提交
160 161 162
/*
 * Switches to specified vcpu, until a matching vcpu_put()
 */
163
void vcpu_load(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
164
{
165
	int cpu = get_cpu();
166
	preempt_notifier_register(&vcpu->preempt_notifier);
167
	kvm_arch_vcpu_load(vcpu, cpu);
168
	put_cpu();
A
Avi Kivity 已提交
169
}
170
EXPORT_SYMBOL_GPL(vcpu_load);
A
Avi Kivity 已提交
171

172
void vcpu_put(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
173
{
174
	preempt_disable();
175
	kvm_arch_vcpu_put(vcpu);
176 177
	preempt_notifier_unregister(&vcpu->preempt_notifier);
	preempt_enable();
A
Avi Kivity 已提交
178
}
179
EXPORT_SYMBOL_GPL(vcpu_put);
A
Avi Kivity 已提交
180

181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198
/* TODO: merge with kvm_arch_vcpu_should_kick */
static bool kvm_request_needs_ipi(struct kvm_vcpu *vcpu, unsigned req)
{
	int mode = kvm_vcpu_exiting_guest_mode(vcpu);

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

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

199 200 201 202
static void ack_flush(void *_completed)
{
}

203 204 205 206 207 208 209 210 211 212 213 214
static inline bool kvm_kick_many_cpus(const struct cpumask *cpus, bool wait)
{
	if (unlikely(!cpus))
		cpus = cpu_online_mask;

	if (cpumask_empty(cpus))
		return false;

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

215 216
bool kvm_make_vcpus_request_mask(struct kvm *kvm, unsigned int req,
				 unsigned long *vcpu_bitmap, cpumask_var_t tmp)
217
{
218
	int i, cpu, me;
219
	struct kvm_vcpu *vcpu;
220
	bool called;
221

222
	me = get_cpu();
223

224
	kvm_for_each_vcpu(i, vcpu, kvm) {
225
		if (vcpu_bitmap && !test_bit(i, vcpu_bitmap))
226 227
			continue;

228
		kvm_make_request(req, vcpu);
229
		cpu = vcpu->cpu;
230

231 232
		if (!(req & KVM_REQUEST_NO_WAKEUP) && kvm_vcpu_wake_up(vcpu))
			continue;
233

234
		if (tmp != NULL && cpu != -1 && cpu != me &&
235
		    kvm_request_needs_ipi(vcpu, req))
236
			__cpumask_set_cpu(cpu, tmp);
237
	}
238 239

	called = kvm_kick_many_cpus(tmp, !!(req & KVM_REQUEST_WAIT));
240
	put_cpu();
241 242 243 244 245 246 247 248 249 250 251

	return called;
}

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

	zalloc_cpumask_var(&cpus, GFP_ATOMIC);

252
	called = kvm_make_vcpus_request_mask(kvm, req, NULL, cpus);
253

254
	free_cpumask_var(cpus);
255
	return called;
256 257
}

258
#ifndef CONFIG_HAVE_KVM_ARCH_TLB_FLUSH_ALL
259
void kvm_flush_remote_tlbs(struct kvm *kvm)
260
{
261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277
	/*
	 * 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.
	 */
278 279
	if (!kvm_arch_flush_remote_tlb(kvm)
	    || kvm_make_all_cpus_request(kvm, KVM_REQ_TLB_FLUSH))
280
		++kvm->stat.remote_tlb_flush;
281
	cmpxchg(&kvm->tlbs_dirty, dirty_count, 0);
282
}
283
EXPORT_SYMBOL_GPL(kvm_flush_remote_tlbs);
284
#endif
285

286 287
void kvm_reload_remote_mmus(struct kvm *kvm)
{
288
	kvm_make_all_cpus_request(kvm, KVM_REQ_MMU_RELOAD);
289
}
290

R
Rusty Russell 已提交
291 292 293 294 295 296 297 298 299
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;
300
	vcpu->pid = NULL;
301
	init_swait_queue_head(&vcpu->wq);
302
	kvm_async_pf_vcpu_init(vcpu);
R
Rusty Russell 已提交
303

304 305 306
	vcpu->pre_pcpu = -1;
	INIT_LIST_HEAD(&vcpu->blocked_vcpu_list);

R
Rusty Russell 已提交
307 308 309 310 311 312 313
	page = alloc_page(GFP_KERNEL | __GFP_ZERO);
	if (!page) {
		r = -ENOMEM;
		goto fail;
	}
	vcpu->run = page_address(page);

314 315
	kvm_vcpu_set_in_spin_loop(vcpu, false);
	kvm_vcpu_set_dy_eligible(vcpu, false);
316
	vcpu->preempted = false;
317

318
	r = kvm_arch_vcpu_init(vcpu);
R
Rusty Russell 已提交
319
	if (r < 0)
320
		goto fail_free_run;
R
Rusty Russell 已提交
321 322 323 324 325
	return 0;

fail_free_run:
	free_page((unsigned long)vcpu->run);
fail:
326
	return r;
R
Rusty Russell 已提交
327 328 329 330 331
}
EXPORT_SYMBOL_GPL(kvm_vcpu_init);

void kvm_vcpu_uninit(struct kvm_vcpu *vcpu)
{
332 333 334 335 336 337
	/*
	 * no need for rcu_read_lock as VCPU_RUN is the only place that
	 * will change the vcpu->pid pointer and on uninit all file
	 * descriptors are already gone.
	 */
	put_pid(rcu_dereference_protected(vcpu->pid, 1));
338
	kvm_arch_vcpu_uninit(vcpu);
R
Rusty Russell 已提交
339 340 341 342
	free_page((unsigned long)vcpu->run);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_uninit);

343 344 345 346 347 348
#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);
}

349 350 351 352 353 354
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);
355
	int idx;
356

357
	idx = srcu_read_lock(&kvm->srcu);
358 359
	spin_lock(&kvm->mmu_lock);
	kvm->mmu_notifier_seq++;
360 361 362 363

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

364
	spin_unlock(&kvm->mmu_lock);
365
	srcu_read_unlock(&kvm->srcu, idx);
366 367
}

368
static int kvm_mmu_notifier_invalidate_range_start(struct mmu_notifier *mn,
369
					const struct mmu_notifier_range *range)
370 371
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
372
	int need_tlb_flush = 0, idx;
373
	int ret;
374

375
	idx = srcu_read_lock(&kvm->srcu);
376 377 378 379 380 381 382
	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++;
383
	need_tlb_flush = kvm_unmap_hva_range(kvm, range->start, range->end);
384
	need_tlb_flush |= kvm->tlbs_dirty;
385 386 387
	/* we've to flush the tlb before the pages can be freed */
	if (need_tlb_flush)
		kvm_flush_remote_tlbs(kvm);
388 389

	spin_unlock(&kvm->mmu_lock);
390

391
	ret = kvm_arch_mmu_notifier_invalidate_range(kvm, range->start,
392 393
					range->end,
					mmu_notifier_range_blockable(range));
394

395
	srcu_read_unlock(&kvm->srcu, idx);
396 397

	return ret;
398 399 400
}

static void kvm_mmu_notifier_invalidate_range_end(struct mmu_notifier *mn,
401
					const struct mmu_notifier_range *range)
402 403 404 405 406 407 408 409 410 411
{
	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++;
412
	smp_wmb();
413 414
	/*
	 * The above sequence increase must be visible before the
415 416
	 * below count decrease, which is ensured by the smp_wmb above
	 * in conjunction with the smp_rmb in mmu_notifier_retry().
417 418 419 420 421 422 423 424 425
	 */
	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
426 427
					      unsigned long start,
					      unsigned long end)
428 429
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
430
	int young, idx;
431

432
	idx = srcu_read_lock(&kvm->srcu);
433 434
	spin_lock(&kvm->mmu_lock);

A
Andres Lagar-Cavilla 已提交
435
	young = kvm_age_hva(kvm, start, end);
436 437 438
	if (young)
		kvm_flush_remote_tlbs(kvm);

439 440 441
	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);

442 443 444
	return young;
}

445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474
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
475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490
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;
}

491 492 493 494
static void kvm_mmu_notifier_release(struct mmu_notifier *mn,
				     struct mm_struct *mm)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
495 496 497
	int idx;

	idx = srcu_read_lock(&kvm->srcu);
498
	kvm_arch_flush_shadow_all(kvm);
499
	srcu_read_unlock(&kvm->srcu, idx);
500 501
}

502 503 504 505
static const struct mmu_notifier_ops kvm_mmu_notifier_ops = {
	.invalidate_range_start	= kvm_mmu_notifier_invalidate_range_start,
	.invalidate_range_end	= kvm_mmu_notifier_invalidate_range_end,
	.clear_flush_young	= kvm_mmu_notifier_clear_flush_young,
506
	.clear_young		= kvm_mmu_notifier_clear_young,
A
Andrea Arcangeli 已提交
507
	.test_young		= kvm_mmu_notifier_test_young,
508
	.change_pte		= kvm_mmu_notifier_change_pte,
509
	.release		= kvm_mmu_notifier_release,
510
};
511 512 513 514 515 516 517 518 519 520 521 522 523 524

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;
}

525 526
#endif /* CONFIG_MMU_NOTIFIER && KVM_ARCH_WANT_MMU_NOTIFIER */

527
static struct kvm_memslots *kvm_alloc_memslots(void)
528 529
{
	int i;
530
	struct kvm_memslots *slots;
531

532
	slots = kvzalloc(sizeof(struct kvm_memslots), GFP_KERNEL_ACCOUNT);
533 534 535
	if (!slots)
		return NULL;

536
	for (i = 0; i < KVM_MEM_SLOTS_NUM; i++)
537
		slots->id_to_index[i] = slots->memslots[i].id = i;
538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575

	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);
576 577
}

578 579 580 581 582 583 584 585 586
static void kvm_destroy_vm_debugfs(struct kvm *kvm)
{
	int i;

	if (!kvm->debugfs_dentry)
		return;

	debugfs_remove_recursive(kvm->debugfs_dentry);

587 588 589 590 591
	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);
	}
592 593 594 595 596 597 598 599 600 601 602 603
}

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);
604
	kvm->debugfs_dentry = debugfs_create_dir(dir_name, kvm_debugfs_dir);
605 606 607

	kvm->debugfs_stat_data = kcalloc(kvm_debugfs_num_entries,
					 sizeof(*kvm->debugfs_stat_data),
608
					 GFP_KERNEL_ACCOUNT);
609 610 611 612
	if (!kvm->debugfs_stat_data)
		return -ENOMEM;

	for (p = debugfs_entries; p->name; p++) {
613
		stat_data = kzalloc(sizeof(*stat_data), GFP_KERNEL_ACCOUNT);
614 615 616 617 618 619
		if (!stat_data)
			return -ENOMEM;

		stat_data->kvm = kvm;
		stat_data->offset = p->offset;
		kvm->debugfs_stat_data[p - debugfs_entries] = stat_data;
620 621
		debugfs_create_file(p->name, 0644, kvm->debugfs_dentry,
				    stat_data, stat_fops_per_vm[p->kind]);
622 623 624 625
	}
	return 0;
}

626
static struct kvm *kvm_create_vm(unsigned long type)
A
Avi Kivity 已提交
627
{
628 629
	int r, i;
	struct kvm *kvm = kvm_arch_alloc_vm();
A
Avi Kivity 已提交
630

631 632 633
	if (!kvm)
		return ERR_PTR(-ENOMEM);

634
	spin_lock_init(&kvm->mmu_lock);
V
Vegard Nossum 已提交
635
	mmgrab(current->mm);
636 637 638 639 640
	kvm->mm = current->mm;
	kvm_eventfd_init(kvm);
	mutex_init(&kvm->lock);
	mutex_init(&kvm->irq_lock);
	mutex_init(&kvm->slots_lock);
641
	refcount_set(&kvm->users_count, 1);
642 643
	INIT_LIST_HEAD(&kvm->devices);

644
	r = kvm_arch_init_vm(kvm, type);
645
	if (r)
646
		goto out_err_no_disable;
647 648 649

	r = hardware_enable_all();
	if (r)
650
		goto out_err_no_disable;
651

652
#ifdef CONFIG_HAVE_KVM_IRQFD
653
	INIT_HLIST_HEAD(&kvm->irq_ack_notifier_list);
654
#endif
A
Avi Kivity 已提交
655

656 657
	BUILD_BUG_ON(KVM_MEM_SLOTS_NUM > SHRT_MAX);

658
	r = -ENOMEM;
659
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++) {
660 661
		struct kvm_memslots *slots = kvm_alloc_memslots();
		if (!slots)
662
			goto out_err_no_srcu;
663
		/* Generations must be different for each address space. */
664
		slots->generation = i;
665
		rcu_assign_pointer(kvm->memslots[i], slots);
666
	}
667

668
	if (init_srcu_struct(&kvm->srcu))
669 670 671
		goto out_err_no_srcu;
	if (init_srcu_struct(&kvm->irq_srcu))
		goto out_err_no_irq_srcu;
M
Marcelo Tosatti 已提交
672
	for (i = 0; i < KVM_NR_BUSES; i++) {
673
		rcu_assign_pointer(kvm->buses[i],
674
			kzalloc(sizeof(struct kvm_io_bus), GFP_KERNEL_ACCOUNT));
675
		if (!kvm->buses[i])
M
Marcelo Tosatti 已提交
676 677
			goto out_err;
	}
678

679 680 681 682
	r = kvm_init_mmu_notifier(kvm);
	if (r)
		goto out_err;

J
Junaid Shahid 已提交
683
	mutex_lock(&kvm_lock);
684
	list_add(&kvm->vm_list, &vm_list);
J
Junaid Shahid 已提交
685
	mutex_unlock(&kvm_lock);
686

687 688
	preempt_notifier_inc();

689
	return kvm;
690 691

out_err:
692 693
	cleanup_srcu_struct(&kvm->irq_srcu);
out_err_no_irq_srcu:
694
	cleanup_srcu_struct(&kvm->srcu);
695
out_err_no_srcu:
696
	hardware_disable_all();
697
out_err_no_disable:
698
	refcount_set(&kvm->users_count, 0);
M
Marcelo Tosatti 已提交
699
	for (i = 0; i < KVM_NR_BUSES; i++)
700
		kfree(kvm_get_bus(kvm, i));
701
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++)
702
		kvm_free_memslots(kvm, __kvm_memslots(kvm, i));
703
	kvm_arch_free_vm(kvm);
704
	mmdrop(current->mm);
705
	return ERR_PTR(r);
706 707
}

708 709
static void kvm_destroy_devices(struct kvm *kvm)
{
G
Geliang Tang 已提交
710
	struct kvm_device *dev, *tmp;
711

712 713 714 715 716
	/*
	 * 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
717 718
	list_for_each_entry_safe(dev, tmp, &kvm->devices, vm_node) {
		list_del(&dev->vm_node);
719 720 721 722
		dev->ops->destroy(dev);
	}
}

723 724
static void kvm_destroy_vm(struct kvm *kvm)
{
M
Marcelo Tosatti 已提交
725
	int i;
726 727
	struct mm_struct *mm = kvm->mm;

728
	kvm_uevent_notify_change(KVM_EVENT_DESTROY_VM, kvm);
729
	kvm_destroy_vm_debugfs(kvm);
730
	kvm_arch_sync_events(kvm);
J
Junaid Shahid 已提交
731
	mutex_lock(&kvm_lock);
732
	list_del(&kvm->vm_list);
J
Junaid Shahid 已提交
733
	mutex_unlock(&kvm_lock);
734
	kvm_free_irq_routing(kvm);
735
	for (i = 0; i < KVM_NR_BUSES; i++) {
736
		struct kvm_io_bus *bus = kvm_get_bus(kvm, i);
737 738 739

		if (bus)
			kvm_io_bus_destroy(bus);
740 741
		kvm->buses[i] = NULL;
	}
742
	kvm_coalesced_mmio_free(kvm);
743 744
#if defined(CONFIG_MMU_NOTIFIER) && defined(KVM_ARCH_WANT_MMU_NOTIFIER)
	mmu_notifier_unregister(&kvm->mmu_notifier, kvm->mm);
745
#else
746
	kvm_arch_flush_shadow_all(kvm);
747
#endif
748
	kvm_arch_destroy_vm(kvm);
749
	kvm_destroy_devices(kvm);
750
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++)
751
		kvm_free_memslots(kvm, __kvm_memslots(kvm, i));
752
	cleanup_srcu_struct(&kvm->irq_srcu);
753 754
	cleanup_srcu_struct(&kvm->srcu);
	kvm_arch_free_vm(kvm);
755
	preempt_notifier_dec();
756
	hardware_disable_all();
757
	mmdrop(mm);
758 759
}

I
Izik Eidus 已提交
760 761
void kvm_get_kvm(struct kvm *kvm)
{
762
	refcount_inc(&kvm->users_count);
I
Izik Eidus 已提交
763 764 765 766 767
}
EXPORT_SYMBOL_GPL(kvm_get_kvm);

void kvm_put_kvm(struct kvm *kvm)
{
768
	if (refcount_dec_and_test(&kvm->users_count))
I
Izik Eidus 已提交
769 770 771 772 773
		kvm_destroy_vm(kvm);
}
EXPORT_SYMBOL_GPL(kvm_put_kvm);


774 775 776 777
static int kvm_vm_release(struct inode *inode, struct file *filp)
{
	struct kvm *kvm = filp->private_data;

G
Gregory Haskins 已提交
778 779
	kvm_irqfd_release(kvm);

I
Izik Eidus 已提交
780
	kvm_put_kvm(kvm);
A
Avi Kivity 已提交
781 782 783
	return 0;
}

784 785
/*
 * Allocation size is twice as large as the actual dirty bitmap size.
786
 * See x86's kvm_vm_ioctl_get_dirty_log() why this is needed.
787
 */
788 789
static int kvm_create_dirty_bitmap(struct kvm_memory_slot *memslot)
{
790
	unsigned long dirty_bytes = 2 * kvm_dirty_bitmap_bytes(memslot);
791

792
	memslot->dirty_bitmap = kvzalloc(dirty_bytes, GFP_KERNEL_ACCOUNT);
793 794 795 796 797 798
	if (!memslot->dirty_bitmap)
		return -ENOMEM;

	return 0;
}

799
/*
800 801 802 803
 * 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.
804
 */
805
static void update_memslots(struct kvm_memslots *slots,
806 807
			    struct kvm_memory_slot *new,
			    enum kvm_mr_change change)
808
{
809 810
	int id = new->id;
	int i = slots->id_to_index[id];
811
	struct kvm_memory_slot *mslots = slots->memslots;
812

813
	WARN_ON(mslots[i].id != id);
814 815 816 817 818 819 820 821 822 823 824
	switch (change) {
	case KVM_MR_CREATE:
		slots->used_slots++;
		WARN_ON(mslots[i].npages || !new->npages);
		break;
	case KVM_MR_DELETE:
		slots->used_slots--;
		WARN_ON(new->npages || !mslots[i].npages);
		break;
	default:
		break;
825
	}
826

827
	while (i < KVM_MEM_SLOTS_NUM - 1 &&
828 829 830
	       new->base_gfn <= mslots[i + 1].base_gfn) {
		if (!mslots[i + 1].npages)
			break;
831 832 833 834
		mslots[i] = mslots[i + 1];
		slots->id_to_index[mslots[i].id] = i;
		i++;
	}
835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851

	/*
	 * 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--;
		}
852 853
	} else
		WARN_ON_ONCE(i != slots->used_slots);
854

855 856
	mslots[i] = *new;
	slots->id_to_index[mslots[i].id] = i;
857 858
}

859
static int check_memory_region_flags(const struct kvm_userspace_memory_region *mem)
860
{
X
Xiao Guangrong 已提交
861 862
	u32 valid_flags = KVM_MEM_LOG_DIRTY_PAGES;

863
#ifdef __KVM_HAVE_READONLY_MEM
X
Xiao Guangrong 已提交
864 865 866 867
	valid_flags |= KVM_MEM_READONLY;
#endif

	if (mem->flags & ~valid_flags)
868 869 870 871 872
		return -EINVAL;

	return 0;
}

873
static struct kvm_memslots *install_new_memslots(struct kvm *kvm,
874
		int as_id, struct kvm_memslots *slots)
875
{
876
	struct kvm_memslots *old_memslots = __kvm_memslots(kvm, as_id);
877
	u64 gen = old_memslots->generation;
878

879 880
	WARN_ON(gen & KVM_MEMSLOT_GEN_UPDATE_IN_PROGRESS);
	slots->generation = gen | KVM_MEMSLOT_GEN_UPDATE_IN_PROGRESS;
881

882
	rcu_assign_pointer(kvm->memslots[as_id], slots);
883
	synchronize_srcu_expedited(&kvm->srcu);
884

885
	/*
886 887 888 889 890 891 892 893
	 * Increment the new memslot generation a second time, dropping the
	 * update in-progress flag and incrementing then generation based on
	 * the number of address spaces.  This provides a unique and easily
	 * identifiable generation number while the memslots are in flux.
	 */
	gen = slots->generation & ~KVM_MEMSLOT_GEN_UPDATE_IN_PROGRESS;

	/*
894 895 896
	 * 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
897 898
	 * space 0 will use generations 0, 2, 4, ... while address space 1 will
	 * use generations 1, 3, 5, ...
899
	 */
900
	gen += KVM_ADDRESS_SPACE_NUM;
901

902
	kvm_arch_memslots_updated(kvm, gen);
903

904
	slots->generation = gen;
905 906

	return old_memslots;
907 908
}

A
Avi Kivity 已提交
909 910 911 912 913
/*
 * Allocate some memory and give it an address in the guest physical address
 * space.
 *
 * Discontiguous memory is allowed, mostly for framebuffers.
914
 *
915
 * Must be called holding kvm->slots_lock for write.
A
Avi Kivity 已提交
916
 */
917
int __kvm_set_memory_region(struct kvm *kvm,
918
			    const struct kvm_userspace_memory_region *mem)
A
Avi Kivity 已提交
919
{
920
	int r;
A
Avi Kivity 已提交
921
	gfn_t base_gfn;
922
	unsigned long npages;
923
	struct kvm_memory_slot *slot;
A
Avi Kivity 已提交
924
	struct kvm_memory_slot old, new;
925
	struct kvm_memslots *slots = NULL, *old_memslots;
926
	int as_id, id;
927
	enum kvm_mr_change change;
A
Avi Kivity 已提交
928

929 930 931 932
	r = check_memory_region_flags(mem);
	if (r)
		goto out;

A
Avi Kivity 已提交
933
	r = -EINVAL;
934 935 936
	as_id = mem->slot >> 16;
	id = (u16)mem->slot;

A
Avi Kivity 已提交
937 938 939 940 941
	/* General sanity checks */
	if (mem->memory_size & (PAGE_SIZE - 1))
		goto out;
	if (mem->guest_phys_addr & (PAGE_SIZE - 1))
		goto out;
942
	/* We can read the guest memory with __xxx_user() later on. */
943
	if ((id < KVM_USER_MEM_SLOTS) &&
944
	    ((mem->userspace_addr & (PAGE_SIZE - 1)) ||
945
	     !access_ok((void __user *)(unsigned long)mem->userspace_addr,
946
			mem->memory_size)))
947
		goto out;
948
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_MEM_SLOTS_NUM)
A
Avi Kivity 已提交
949 950 951 952
		goto out;
	if (mem->guest_phys_addr + mem->memory_size < mem->guest_phys_addr)
		goto out;

953
	slot = id_to_memslot(__kvm_memslots(kvm, as_id), id);
A
Avi Kivity 已提交
954 955 956
	base_gfn = mem->guest_phys_addr >> PAGE_SHIFT;
	npages = mem->memory_size >> PAGE_SHIFT;

957 958 959
	if (npages > KVM_MEM_MAX_NR_PAGES)
		goto out;

960
	new = old = *slot;
A
Avi Kivity 已提交
961

962
	new.id = id;
A
Avi Kivity 已提交
963 964 965 966
	new.base_gfn = base_gfn;
	new.npages = npages;
	new.flags = mem->flags;

967 968 969 970 971
	if (npages) {
		if (!old.npages)
			change = KVM_MR_CREATE;
		else { /* Modify an existing slot. */
			if ((mem->userspace_addr != old.userspace_addr) ||
972 973
			    (npages != old.npages) ||
			    ((new.flags ^ old.flags) & KVM_MEM_READONLY))
974 975 976 977 978 979 980 981 982 983 984
				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;
			}
		}
985 986 987 988
	} else {
		if (!old.npages)
			goto out;

989
		change = KVM_MR_DELETE;
990 991 992
		new.base_gfn = 0;
		new.flags = 0;
	}
A
Avi Kivity 已提交
993

994
	if ((change == KVM_MR_CREATE) || (change == KVM_MR_MOVE)) {
995 996
		/* Check for overlaps */
		r = -EEXIST;
997
		kvm_for_each_memslot(slot, __kvm_memslots(kvm, as_id)) {
998
			if (slot->id == id)
999 1000 1001 1002 1003
				continue;
			if (!((base_gfn + npages <= slot->base_gfn) ||
			      (base_gfn >= slot->base_gfn + slot->npages)))
				goto out;
		}
A
Avi Kivity 已提交
1004 1005 1006 1007
	}

	/* Free page dirty bitmap if unneeded */
	if (!(new.flags & KVM_MEM_LOG_DIRTY_PAGES))
A
Al Viro 已提交
1008
		new.dirty_bitmap = NULL;
A
Avi Kivity 已提交
1009 1010

	r = -ENOMEM;
1011
	if (change == KVM_MR_CREATE) {
1012
		new.userspace_addr = mem->userspace_addr;
1013

1014
		if (kvm_arch_create_memslot(kvm, &new, npages))
1015
			goto out_free;
A
Avi Kivity 已提交
1016
	}
1017

A
Avi Kivity 已提交
1018 1019
	/* Allocate page dirty bitmap if needed */
	if ((new.flags & KVM_MEM_LOG_DIRTY_PAGES) && !new.dirty_bitmap) {
1020
		if (kvm_create_dirty_bitmap(&new) < 0)
1021
			goto out_free;
A
Avi Kivity 已提交
1022 1023
	}

1024
	slots = kvzalloc(sizeof(struct kvm_memslots), GFP_KERNEL_ACCOUNT);
1025 1026
	if (!slots)
		goto out_free;
1027
	memcpy(slots, __kvm_memslots(kvm, as_id), sizeof(struct kvm_memslots));
1028

1029
	if ((change == KVM_MR_DELETE) || (change == KVM_MR_MOVE)) {
1030
		slot = id_to_memslot(slots, id);
1031 1032
		slot->flags |= KVM_MEMSLOT_INVALID;

1033
		old_memslots = install_new_memslots(kvm, as_id, slots);
1034

1035 1036
		/* From this point no new shadow pages pointing to a deleted,
		 * or moved, memslot will be created.
1037 1038
		 *
		 * validation of sp->gfn happens in:
1039 1040
		 *	- gfn_to_hva (kvm_read_guest, gfn_to_pfn)
		 *	- kvm_is_visible_gfn (mmu_check_roots)
1041
		 */
1042
		kvm_arch_flush_shadow_memslot(kvm, slot);
1043 1044 1045 1046 1047 1048

		/*
		 * 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.
		 */
1049
		slots = old_memslots;
1050
	}
1051

1052
	r = kvm_arch_prepare_memory_region(kvm, &new, mem, change);
1053
	if (r)
1054
		goto out_slots;
1055

1056
	/* actual memory is freed via old in kvm_free_memslot below */
1057
	if (change == KVM_MR_DELETE) {
1058
		new.dirty_bitmap = NULL;
1059
		memset(&new.arch, 0, sizeof(new.arch));
1060 1061
	}

1062
	update_memslots(slots, &new, change);
1063
	old_memslots = install_new_memslots(kvm, as_id, slots);
1064

1065
	kvm_arch_commit_memory_region(kvm, mem, &old, &new, change);
1066

1067
	kvm_free_memslot(kvm, &old, &new);
1068
	kvfree(old_memslots);
A
Avi Kivity 已提交
1069 1070
	return 0;

1071
out_slots:
1072
	kvfree(slots);
1073
out_free:
1074
	kvm_free_memslot(kvm, &new, &old);
A
Avi Kivity 已提交
1075 1076
out:
	return r;
1077
}
1078 1079 1080
EXPORT_SYMBOL_GPL(__kvm_set_memory_region);

int kvm_set_memory_region(struct kvm *kvm,
1081
			  const struct kvm_userspace_memory_region *mem)
1082 1083 1084
{
	int r;

1085
	mutex_lock(&kvm->slots_lock);
1086
	r = __kvm_set_memory_region(kvm, mem);
1087
	mutex_unlock(&kvm->slots_lock);
1088 1089
	return r;
}
1090 1091
EXPORT_SYMBOL_GPL(kvm_set_memory_region);

1092 1093
static int kvm_vm_ioctl_set_memory_region(struct kvm *kvm,
					  struct kvm_userspace_memory_region *mem)
1094
{
1095
	if ((u16)mem->slot >= KVM_USER_MEM_SLOTS)
1096
		return -EINVAL;
1097

1098
	return kvm_set_memory_region(kvm, mem);
A
Avi Kivity 已提交
1099 1100
}

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

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

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

1120
	n = kvm_dirty_bitmap_bytes(memslot);
A
Avi Kivity 已提交
1121

1122
	for (i = 0; !any && i < n/sizeof(long); ++i)
A
Avi Kivity 已提交
1123 1124 1125
		any = memslot->dirty_bitmap[i];

	if (copy_to_user(log->dirty_bitmap, memslot->dirty_bitmap, n))
1126
		return -EFAULT;
A
Avi Kivity 已提交
1127

1128 1129
	if (any)
		*is_dirty = 1;
1130
	return 0;
A
Avi Kivity 已提交
1131
}
1132
EXPORT_SYMBOL_GPL(kvm_get_dirty_log);
A
Avi Kivity 已提交
1133

1134 1135
#ifdef CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT
/**
J
Jiang Biao 已提交
1136
 * kvm_get_dirty_log_protect - get a snapshot of dirty pages
1137
 *	and reenable dirty page tracking for the corresponding pages.
1138 1139
 * @kvm:	pointer to kvm instance
 * @log:	slot id and address to which we copy the log
J
Jiang Biao 已提交
1140
 * @flush:	true if TLB flush is needed by caller
1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157
 *
 * 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,
1158
			struct kvm_dirty_log *log, bool *flush)
1159
{
1160
	struct kvm_memslots *slots;
1161
	struct kvm_memory_slot *memslot;
1162
	int i, as_id, id;
1163 1164 1165 1166
	unsigned long n;
	unsigned long *dirty_bitmap;
	unsigned long *dirty_bitmap_buffer;

1167 1168 1169
	as_id = log->slot >> 16;
	id = (u16)log->slot;
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_USER_MEM_SLOTS)
1170
		return -EINVAL;
1171

1172 1173
	slots = __kvm_memslots(kvm, as_id);
	memslot = id_to_memslot(slots, id);
1174 1175 1176

	dirty_bitmap = memslot->dirty_bitmap;
	if (!dirty_bitmap)
1177
		return -ENOENT;
1178 1179

	n = kvm_dirty_bitmap_bytes(memslot);
1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193
	*flush = false;
	if (kvm->manual_dirty_log_protect) {
		/*
		 * Unlike kvm_get_dirty_log, we always return false in *flush,
		 * because no flush is needed until KVM_CLEAR_DIRTY_LOG.  There
		 * is some code duplication between this function and
		 * kvm_get_dirty_log, but hopefully all architecture
		 * transition to kvm_get_dirty_log_protect and kvm_get_dirty_log
		 * can be eliminated.
		 */
		dirty_bitmap_buffer = dirty_bitmap;
	} else {
		dirty_bitmap_buffer = kvm_second_dirty_bitmap(memslot);
		memset(dirty_bitmap_buffer, 0, n);
1194

1195 1196 1197 1198
		spin_lock(&kvm->mmu_lock);
		for (i = 0; i < n / sizeof(long); i++) {
			unsigned long mask;
			gfn_t offset;
1199

1200 1201 1202 1203 1204 1205 1206
			if (!dirty_bitmap[i])
				continue;

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

1207 1208 1209
			offset = i * BITS_PER_LONG;
			kvm_arch_mmu_enable_log_dirty_pt_masked(kvm, memslot,
								offset, mask);
1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224
		}
		spin_unlock(&kvm->mmu_lock);
	}

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

/**
 * kvm_clear_dirty_log_protect - clear dirty bits in the bitmap
 *	and reenable dirty page tracking for the corresponding pages.
 * @kvm:	pointer to kvm instance
 * @log:	slot id and address from which to fetch the bitmap of dirty pages
J
Jiang Biao 已提交
1225
 * @flush:	true if TLB flush is needed by caller
1226 1227 1228 1229 1230 1231
 */
int kvm_clear_dirty_log_protect(struct kvm *kvm,
				struct kvm_clear_dirty_log *log, bool *flush)
{
	struct kvm_memslots *slots;
	struct kvm_memory_slot *memslot;
1232
	int as_id, id;
1233
	gfn_t offset;
1234
	unsigned long i, n;
1235 1236 1237 1238 1239 1240 1241 1242
	unsigned long *dirty_bitmap;
	unsigned long *dirty_bitmap_buffer;

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

1243
	if (log->first_page & 63)
1244 1245 1246 1247 1248 1249 1250 1251 1252
		return -EINVAL;

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

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

1253
	n = ALIGN(log->num_pages, BITS_PER_LONG) / 8;
1254 1255

	if (log->first_page > memslot->npages ||
1256 1257 1258
	    log->num_pages > memslot->npages - log->first_page ||
	    (log->num_pages < memslot->npages - log->first_page && (log->num_pages & 63)))
	    return -EINVAL;
1259

1260
	*flush = false;
1261 1262 1263
	dirty_bitmap_buffer = kvm_second_dirty_bitmap(memslot);
	if (copy_from_user(dirty_bitmap_buffer, log->dirty_bitmap, n))
		return -EFAULT;
1264

1265
	spin_lock(&kvm->mmu_lock);
1266 1267
	for (offset = log->first_page, i = offset / BITS_PER_LONG,
		 n = DIV_ROUND_UP(log->num_pages, BITS_PER_LONG); n--;
1268 1269 1270 1271
	     i++, offset += BITS_PER_LONG) {
		unsigned long mask = *dirty_bitmap_buffer++;
		atomic_long_t *p = (atomic_long_t *) &dirty_bitmap[i];
		if (!mask)
1272 1273
			continue;

1274
		mask &= atomic_long_fetch_andnot(mask, p);
1275

1276 1277 1278 1279 1280 1281
		/*
		 * mask contains the bits that really have been cleared.  This
		 * never includes any bits beyond the length of the memslot (if
		 * the length is not aligned to 64 pages), therefore it is not
		 * a problem if userspace sets them in log->dirty_bitmap.
		*/
1282
		if (mask) {
1283
			*flush = true;
1284 1285 1286
			kvm_arch_mmu_enable_log_dirty_pt_masked(kvm, memslot,
								offset, mask);
		}
1287 1288
	}
	spin_unlock(&kvm->mmu_lock);
1289

1290
	return 0;
1291
}
1292
EXPORT_SYMBOL_GPL(kvm_clear_dirty_log_protect);
1293 1294
#endif

1295 1296 1297 1298 1299
bool kvm_largepages_enabled(void)
{
	return largepages_enabled;
}

1300 1301 1302 1303 1304 1305
void kvm_disable_largepages(void)
{
	largepages_enabled = false;
}
EXPORT_SYMBOL_GPL(kvm_disable_largepages);

1306 1307 1308 1309
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
1310
EXPORT_SYMBOL_GPL(gfn_to_memslot);
A
Avi Kivity 已提交
1311

1312 1313 1314 1315 1316
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);
}

1317
bool kvm_is_visible_gfn(struct kvm *kvm, gfn_t gfn)
1318
{
1319
	struct kvm_memory_slot *memslot = gfn_to_memslot(kvm, gfn);
1320

1321
	if (!memslot || memslot->id >= KVM_USER_MEM_SLOTS ||
1322
	      memslot->flags & KVM_MEMSLOT_INVALID)
1323
		return false;
1324

1325
	return true;
1326 1327 1328
}
EXPORT_SYMBOL_GPL(kvm_is_visible_gfn);

J
Joerg Roedel 已提交
1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352
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
1353 1354 1355 1356 1357 1358 1359
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
1360
{
1361
	if (!slot || slot->flags & KVM_MEMSLOT_INVALID)
X
Xiao Guangrong 已提交
1362
		return KVM_HVA_ERR_BAD;
1363

X
Xiao Guangrong 已提交
1364 1365
	if (memslot_is_readonly(slot) && write)
		return KVM_HVA_ERR_RO_BAD;
1366 1367 1368 1369

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

X
Xiao Guangrong 已提交
1370
	return __gfn_to_hva_memslot(slot, gfn);
I
Izik Eidus 已提交
1371
}
1372

X
Xiao Guangrong 已提交
1373 1374 1375 1376
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
1377
}
1378

X
Xiao Guangrong 已提交
1379
unsigned long gfn_to_hva_memslot(struct kvm_memory_slot *slot,
1380
					gfn_t gfn)
X
Xiao Guangrong 已提交
1381 1382 1383 1384 1385
{
	return gfn_to_hva_many(slot, gfn, NULL);
}
EXPORT_SYMBOL_GPL(gfn_to_hva_memslot);

1386 1387
unsigned long gfn_to_hva(struct kvm *kvm, gfn_t gfn)
{
1388
	return gfn_to_hva_many(gfn_to_memslot(kvm, gfn), gfn, NULL);
1389
}
1390
EXPORT_SYMBOL_GPL(gfn_to_hva);
I
Izik Eidus 已提交
1391

1392 1393 1394 1395 1396 1397
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);

1398
/*
1399 1400 1401 1402 1403 1404
 * Return the hva of a @gfn and the R/W attribute if possible.
 *
 * @slot: the kvm_memory_slot which contains @gfn
 * @gfn: the gfn to be translated
 * @writable: used to return the read/write attribute of the @slot if the hva
 * is valid and @writable is not NULL
1405
 */
1406 1407
unsigned long gfn_to_hva_memslot_prot(struct kvm_memory_slot *slot,
				      gfn_t gfn, bool *writable)
1408
{
1409 1410 1411
	unsigned long hva = __gfn_to_hva_many(slot, gfn, NULL, false);

	if (!kvm_is_error_hva(hva) && writable)
1412 1413
		*writable = !memslot_is_readonly(slot);

1414
	return hva;
1415 1416
}

1417 1418 1419 1420 1421 1422 1423
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);
}

1424 1425 1426 1427 1428 1429 1430
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);
}

1431 1432
static inline int check_user_page_hwpoison(unsigned long addr)
{
L
Lorenzo Stoakes 已提交
1433
	int rc, flags = FOLL_HWPOISON | FOLL_WRITE;
1434

L
Lorenzo Stoakes 已提交
1435
	rc = get_user_pages(addr, 1, flags, NULL, NULL);
1436 1437 1438
	return rc == -EHWPOISON;
}

X
Xiao Guangrong 已提交
1439
/*
1440 1441 1442
 * The fast path to get the writable pfn which will be stored in @pfn,
 * true indicates success, otherwise false is returned.  It's also the
 * only part that runs if we can are in atomic context.
X
Xiao Guangrong 已提交
1443
 */
1444 1445
static bool hva_to_pfn_fast(unsigned long addr, bool write_fault,
			    bool *writable, kvm_pfn_t *pfn)
A
Avi Kivity 已提交
1446
{
1447
	struct page *page[1];
X
Xiao Guangrong 已提交
1448
	int npages;
A
Avi Kivity 已提交
1449

1450 1451 1452 1453 1454 1455 1456
	/*
	 * 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;
1457

X
Xiao Guangrong 已提交
1458 1459 1460
	npages = __get_user_pages_fast(addr, 1, 1, page);
	if (npages == 1) {
		*pfn = page_to_pfn(page[0]);
1461

X
Xiao Guangrong 已提交
1462 1463 1464 1465
		if (writable)
			*writable = true;
		return true;
	}
1466

X
Xiao Guangrong 已提交
1467 1468
	return false;
}
1469

X
Xiao Guangrong 已提交
1470 1471 1472 1473 1474
/*
 * 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
1475
			   bool *writable, kvm_pfn_t *pfn)
X
Xiao Guangrong 已提交
1476
{
1477 1478
	unsigned int flags = FOLL_HWPOISON;
	struct page *page;
X
Xiao Guangrong 已提交
1479
	int npages = 0;
1480

X
Xiao Guangrong 已提交
1481 1482 1483 1484 1485
	might_sleep();

	if (writable)
		*writable = write_fault;

1486 1487 1488 1489
	if (write_fault)
		flags |= FOLL_WRITE;
	if (async)
		flags |= FOLL_NOWAIT;
1490

1491
	npages = get_user_pages_unlocked(addr, 1, &page, flags);
X
Xiao Guangrong 已提交
1492 1493 1494 1495
	if (npages != 1)
		return npages;

	/* map read fault as writable if possible */
1496
	if (unlikely(!write_fault) && writable) {
1497
		struct page *wpage;
X
Xiao Guangrong 已提交
1498

1499
		if (__get_user_pages_fast(addr, 1, 1, &wpage) == 1) {
X
Xiao Guangrong 已提交
1500
			*writable = true;
1501 1502
			put_page(page);
			page = wpage;
1503
		}
1504
	}
1505
	*pfn = page_to_pfn(page);
X
Xiao Guangrong 已提交
1506 1507
	return npages;
}
I
Izik Eidus 已提交
1508

X
Xiao Guangrong 已提交
1509 1510 1511 1512
static bool vma_is_valid(struct vm_area_struct *vma, bool write_fault)
{
	if (unlikely(!(vma->vm_flags & VM_READ)))
		return false;
1513

X
Xiao Guangrong 已提交
1514 1515
	if (write_fault && (unlikely(!(vma->vm_flags & VM_WRITE))))
		return false;
1516

X
Xiao Guangrong 已提交
1517 1518
	return true;
}
1519

1520 1521
static int hva_to_pfn_remapped(struct vm_area_struct *vma,
			       unsigned long addr, bool *async,
1522 1523
			       bool write_fault, bool *writable,
			       kvm_pfn_t *p_pfn)
1524
{
1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548
	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;

	}

1549 1550
	if (writable)
		*writable = true;
1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565

	/*
	 * 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;
1566 1567 1568
	return 0;
}

1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582
/*
 * 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
1583
static kvm_pfn_t hva_to_pfn(unsigned long addr, bool atomic, bool *async,
X
Xiao Guangrong 已提交
1584 1585 1586
			bool write_fault, bool *writable)
{
	struct vm_area_struct *vma;
D
Dan Williams 已提交
1587
	kvm_pfn_t pfn = 0;
1588
	int npages, r;
1589

X
Xiao Guangrong 已提交
1590 1591
	/* we can do it either atomically or asynchronously, not both */
	BUG_ON(atomic && async);
1592

1593
	if (hva_to_pfn_fast(addr, write_fault, writable, &pfn))
X
Xiao Guangrong 已提交
1594 1595 1596 1597 1598 1599 1600 1601
		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;
1602

X
Xiao Guangrong 已提交
1603 1604 1605 1606 1607 1608 1609
	down_read(&current->mm->mmap_sem);
	if (npages == -EHWPOISON ||
	      (!async && check_user_page_hwpoison(addr))) {
		pfn = KVM_PFN_ERR_HWPOISON;
		goto exit;
	}

1610
retry:
X
Xiao Guangrong 已提交
1611 1612 1613 1614
	vma = find_vma_intersection(current->mm, addr, addr + 1);

	if (vma == NULL)
		pfn = KVM_PFN_ERR_FAULT;
1615
	else if (vma->vm_flags & (VM_IO | VM_PFNMAP)) {
1616
		r = hva_to_pfn_remapped(vma, addr, async, write_fault, writable, &pfn);
1617 1618
		if (r == -EAGAIN)
			goto retry;
1619 1620
		if (r < 0)
			pfn = KVM_PFN_ERR_FAULT;
X
Xiao Guangrong 已提交
1621
	} else {
X
Xiao Guangrong 已提交
1622
		if (async && vma_is_valid(vma, write_fault))
X
Xiao Guangrong 已提交
1623 1624 1625 1626 1627
			*async = true;
		pfn = KVM_PFN_ERR_FAULT;
	}
exit:
	up_read(&current->mm->mmap_sem);
1628
	return pfn;
1629 1630
}

D
Dan Williams 已提交
1631 1632 1633
kvm_pfn_t __gfn_to_pfn_memslot(struct kvm_memory_slot *slot, gfn_t gfn,
			       bool atomic, bool *async, bool write_fault,
			       bool *writable)
1634
{
X
Xiao Guangrong 已提交
1635 1636
	unsigned long addr = __gfn_to_hva_many(slot, gfn, NULL, write_fault);

1637 1638 1639
	if (addr == KVM_HVA_ERR_RO_BAD) {
		if (writable)
			*writable = false;
X
Xiao Guangrong 已提交
1640
		return KVM_PFN_ERR_RO_FAULT;
1641
	}
X
Xiao Guangrong 已提交
1642

1643 1644 1645
	if (kvm_is_error_hva(addr)) {
		if (writable)
			*writable = false;
1646
		return KVM_PFN_NOSLOT;
1647
	}
X
Xiao Guangrong 已提交
1648 1649 1650 1651 1652 1653 1654 1655 1656

	/* 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);
1657
}
1658
EXPORT_SYMBOL_GPL(__gfn_to_pfn_memslot);
1659

D
Dan Williams 已提交
1660
kvm_pfn_t gfn_to_pfn_prot(struct kvm *kvm, gfn_t gfn, bool write_fault,
1661 1662
		      bool *writable)
{
P
Paolo Bonzini 已提交
1663 1664
	return __gfn_to_pfn_memslot(gfn_to_memslot(kvm, gfn), gfn, false, NULL,
				    write_fault, writable);
1665 1666 1667
}
EXPORT_SYMBOL_GPL(gfn_to_pfn_prot);

D
Dan Williams 已提交
1668
kvm_pfn_t gfn_to_pfn_memslot(struct kvm_memory_slot *slot, gfn_t gfn)
1669
{
X
Xiao Guangrong 已提交
1670
	return __gfn_to_pfn_memslot(slot, gfn, false, NULL, true, NULL);
1671
}
P
Paolo Bonzini 已提交
1672
EXPORT_SYMBOL_GPL(gfn_to_pfn_memslot);
1673

D
Dan Williams 已提交
1674
kvm_pfn_t gfn_to_pfn_memslot_atomic(struct kvm_memory_slot *slot, gfn_t gfn)
1675
{
X
Xiao Guangrong 已提交
1676
	return __gfn_to_pfn_memslot(slot, gfn, true, NULL, true, NULL);
1677
}
1678
EXPORT_SYMBOL_GPL(gfn_to_pfn_memslot_atomic);
1679

D
Dan Williams 已提交
1680
kvm_pfn_t gfn_to_pfn_atomic(struct kvm *kvm, gfn_t gfn)
P
Paolo Bonzini 已提交
1681 1682 1683 1684 1685
{
	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
1686
kvm_pfn_t kvm_vcpu_gfn_to_pfn_atomic(struct kvm_vcpu *vcpu, gfn_t gfn)
1687 1688 1689 1690 1691
{
	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
1692
kvm_pfn_t gfn_to_pfn(struct kvm *kvm, gfn_t gfn)
P
Paolo Bonzini 已提交
1693 1694 1695 1696 1697
{
	return gfn_to_pfn_memslot(gfn_to_memslot(kvm, gfn), gfn);
}
EXPORT_SYMBOL_GPL(gfn_to_pfn);

D
Dan Williams 已提交
1698
kvm_pfn_t kvm_vcpu_gfn_to_pfn(struct kvm_vcpu *vcpu, gfn_t gfn)
1699 1700 1701 1702 1703
{
	return gfn_to_pfn_memslot(kvm_vcpu_gfn_to_memslot(vcpu, gfn), gfn);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_gfn_to_pfn);

1704 1705
int gfn_to_page_many_atomic(struct kvm_memory_slot *slot, gfn_t gfn,
			    struct page **pages, int nr_pages)
1706 1707
{
	unsigned long addr;
1708
	gfn_t entry = 0;
1709

1710
	addr = gfn_to_hva_many(slot, gfn, &entry);
1711 1712 1713 1714 1715 1716 1717 1718 1719 1720
	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
1721
static struct page *kvm_pfn_to_page(kvm_pfn_t pfn)
X
Xiao Guangrong 已提交
1722
{
1723
	if (is_error_noslot_pfn(pfn))
1724
		return KVM_ERR_PTR_BAD_PAGE;
X
Xiao Guangrong 已提交
1725

1726
	if (kvm_is_reserved_pfn(pfn)) {
1727
		WARN_ON(1);
1728
		return KVM_ERR_PTR_BAD_PAGE;
1729
	}
X
Xiao Guangrong 已提交
1730 1731 1732 1733

	return pfn_to_page(pfn);
}

1734 1735
struct page *gfn_to_page(struct kvm *kvm, gfn_t gfn)
{
D
Dan Williams 已提交
1736
	kvm_pfn_t pfn;
1737 1738 1739

	pfn = gfn_to_pfn(kvm, gfn);

X
Xiao Guangrong 已提交
1740
	return kvm_pfn_to_page(pfn);
A
Avi Kivity 已提交
1741 1742 1743
}
EXPORT_SYMBOL_GPL(gfn_to_page);

1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760
static int __kvm_map_gfn(struct kvm_memory_slot *slot, gfn_t gfn,
			 struct kvm_host_map *map)
{
	kvm_pfn_t pfn;
	void *hva = NULL;
	struct page *page = KVM_UNMAPPED_PAGE;

	if (!map)
		return -EINVAL;

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

	if (pfn_valid(pfn)) {
		page = pfn_to_page(pfn);
		hva = kmap(page);
P
Paolo Bonzini 已提交
1761
#ifdef CONFIG_HAS_IOMEM
1762 1763
	} else {
		hva = memremap(pfn_to_hpa(pfn), PAGE_SIZE, MEMREMAP_WB);
P
Paolo Bonzini 已提交
1764
#endif
1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792
	}

	if (!hva)
		return -EFAULT;

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

	return 0;
}

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

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

	if (!map->hva)
		return;

1793
	if (map->page != KVM_UNMAPPED_PAGE)
1794
		kunmap(map->page);
1795
#ifdef CONFIG_HAS_IOMEM
1796 1797
	else
		memunmap(map->hva);
1798
#endif
1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811

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

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

1812 1813
struct page *kvm_vcpu_gfn_to_page(struct kvm_vcpu *vcpu, gfn_t gfn)
{
D
Dan Williams 已提交
1814
	kvm_pfn_t pfn;
1815 1816 1817 1818 1819 1820 1821

	pfn = kvm_vcpu_gfn_to_pfn(vcpu, gfn);

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

1822 1823
void kvm_release_page_clean(struct page *page)
{
1824 1825
	WARN_ON(is_error_page(page));

1826
	kvm_release_pfn_clean(page_to_pfn(page));
1827 1828 1829
}
EXPORT_SYMBOL_GPL(kvm_release_page_clean);

D
Dan Williams 已提交
1830
void kvm_release_pfn_clean(kvm_pfn_t pfn)
1831
{
1832
	if (!is_error_noslot_pfn(pfn) && !kvm_is_reserved_pfn(pfn))
1833
		put_page(pfn_to_page(pfn));
1834 1835 1836
}
EXPORT_SYMBOL_GPL(kvm_release_pfn_clean);

1837
void kvm_release_page_dirty(struct page *page)
1838
{
X
Xiao Guangrong 已提交
1839 1840
	WARN_ON(is_error_page(page));

1841 1842 1843 1844
	kvm_release_pfn_dirty(page_to_pfn(page));
}
EXPORT_SYMBOL_GPL(kvm_release_page_dirty);

1845
void kvm_release_pfn_dirty(kvm_pfn_t pfn)
1846 1847 1848 1849
{
	kvm_set_pfn_dirty(pfn);
	kvm_release_pfn_clean(pfn);
}
1850
EXPORT_SYMBOL_GPL(kvm_release_pfn_dirty);
1851

D
Dan Williams 已提交
1852
void kvm_set_pfn_dirty(kvm_pfn_t pfn)
1853
{
1854
	if (!kvm_is_reserved_pfn(pfn)) {
1855
		struct page *page = pfn_to_page(pfn);
1856

1857 1858 1859
		if (!PageReserved(page))
			SetPageDirty(page);
	}
1860
}
1861 1862
EXPORT_SYMBOL_GPL(kvm_set_pfn_dirty);

D
Dan Williams 已提交
1863
void kvm_set_pfn_accessed(kvm_pfn_t pfn)
1864
{
1865
	if (!kvm_is_reserved_pfn(pfn))
1866
		mark_page_accessed(pfn_to_page(pfn));
1867 1868 1869
}
EXPORT_SYMBOL_GPL(kvm_set_pfn_accessed);

D
Dan Williams 已提交
1870
void kvm_get_pfn(kvm_pfn_t pfn)
1871
{
1872
	if (!kvm_is_reserved_pfn(pfn))
1873
		get_page(pfn_to_page(pfn));
1874 1875
}
EXPORT_SYMBOL_GPL(kvm_get_pfn);
1876

1877 1878 1879 1880 1881 1882 1883 1884
static int next_segment(unsigned long len, int offset)
{
	if (len > PAGE_SIZE - offset)
		return PAGE_SIZE - offset;
	else
		return len;
}

1885 1886
static int __kvm_read_guest_page(struct kvm_memory_slot *slot, gfn_t gfn,
				 void *data, int offset, int len)
1887
{
1888 1889
	int r;
	unsigned long addr;
1890

1891
	addr = gfn_to_hva_memslot_prot(slot, gfn, NULL);
1892 1893
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1894
	r = __copy_from_user(data, (void __user *)addr + offset, len);
1895
	if (r)
1896 1897 1898
		return -EFAULT;
	return 0;
}
1899 1900 1901 1902 1903 1904 1905 1906

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);
}
1907 1908
EXPORT_SYMBOL_GPL(kvm_read_guest_page);

1909 1910 1911 1912 1913 1914 1915 1916 1917
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);

1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937
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);

1938
int kvm_vcpu_read_guest(struct kvm_vcpu *vcpu, gpa_t gpa, void *data, unsigned long len)
1939 1940
{
	gfn_t gfn = gpa >> PAGE_SHIFT;
1941
	int seg;
1942
	int offset = offset_in_page(gpa);
1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956
	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);
1957

1958 1959 1960 1961 1962 1963 1964
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);
1965 1966
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1967
	pagefault_disable();
1968
	r = __copy_from_user_inatomic(data, (void __user *)addr + offset, len);
1969
	pagefault_enable();
1970 1971 1972 1973 1974
	if (r)
		return -EFAULT;
	return 0;
}

1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998
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)
1999
{
2000 2001
	int r;
	unsigned long addr;
2002

2003
	addr = gfn_to_hva_memslot(memslot, gfn);
2004 2005
	if (kvm_is_error_hva(addr))
		return -EFAULT;
2006
	r = __copy_to_user((void __user *)addr + offset, data, len);
2007
	if (r)
2008
		return -EFAULT;
2009
	mark_page_dirty_in_slot(memslot, gfn);
2010 2011
	return 0;
}
2012 2013 2014 2015 2016 2017 2018 2019

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);
}
2020 2021
EXPORT_SYMBOL_GPL(kvm_write_guest_page);

2022 2023 2024 2025 2026 2027 2028 2029 2030
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);

2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049
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;
}
2050
EXPORT_SYMBOL_GPL(kvm_write_guest);
2051

2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072
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);

2073 2074 2075
static int __kvm_gfn_to_hva_cache_init(struct kvm_memslots *slots,
				       struct gfn_to_hva_cache *ghc,
				       gpa_t gpa, unsigned long len)
2076 2077
{
	int offset = offset_in_page(gpa);
2078 2079 2080 2081
	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;
2082
	int r = start_gfn <= end_gfn ? 0 : -EINVAL;
2083 2084 2085

	ghc->gpa = gpa;
	ghc->generation = slots->generation;
2086
	ghc->len = len;
2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103
	ghc->hva = KVM_HVA_ERR_BAD;

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

	/* Use the slow path for cross page reads and writes. */
	if (!r && nr_pages_needed == 1)
2104
		ghc->hva += offset;
2105
	else
2106
		ghc->memslot = NULL;
2107 2108

	return r;
2109
}
2110

2111
int kvm_gfn_to_hva_cache_init(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
2112 2113
			      gpa_t gpa, unsigned long len)
{
2114
	struct kvm_memslots *slots = kvm_memslots(kvm);
2115 2116
	return __kvm_gfn_to_hva_cache_init(slots, ghc, gpa, len);
}
2117
EXPORT_SYMBOL_GPL(kvm_gfn_to_hva_cache_init);
2118

2119
int kvm_write_guest_offset_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
2120 2121
				  void *data, unsigned int offset,
				  unsigned long len)
2122
{
2123
	struct kvm_memslots *slots = kvm_memslots(kvm);
2124
	int r;
2125
	gpa_t gpa = ghc->gpa + offset;
2126

2127
	BUG_ON(len + offset > ghc->len);
2128

2129
	if (slots->generation != ghc->generation)
2130
		__kvm_gfn_to_hva_cache_init(slots, ghc, ghc->gpa, ghc->len);
2131 2132

	if (unlikely(!ghc->memslot))
2133
		return kvm_write_guest(kvm, gpa, data, len);
2134 2135 2136 2137

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

2138
	r = __copy_to_user((void __user *)ghc->hva + offset, data, len);
2139 2140
	if (r)
		return -EFAULT;
2141
	mark_page_dirty_in_slot(ghc->memslot, gpa >> PAGE_SHIFT);
2142 2143 2144

	return 0;
}
2145
EXPORT_SYMBOL_GPL(kvm_write_guest_offset_cached);
2146

2147 2148
int kvm_write_guest_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
			   void *data, unsigned long len)
2149
{
2150
	return kvm_write_guest_offset_cached(kvm, ghc, data, 0, len);
2151
}
2152
EXPORT_SYMBOL_GPL(kvm_write_guest_cached);
2153

2154 2155
int kvm_read_guest_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
			   void *data, unsigned long len)
2156
{
2157
	struct kvm_memslots *slots = kvm_memslots(kvm);
2158 2159
	int r;

2160 2161
	BUG_ON(len > ghc->len);

2162
	if (slots->generation != ghc->generation)
2163
		__kvm_gfn_to_hva_cache_init(slots, ghc, ghc->gpa, ghc->len);
2164 2165

	if (unlikely(!ghc->memslot))
2166
		return kvm_read_guest(kvm, ghc->gpa, data, len);
2167 2168 2169 2170 2171 2172 2173 2174 2175 2176

	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;
}
2177
EXPORT_SYMBOL_GPL(kvm_read_guest_cached);
2178

2179 2180
int kvm_clear_guest_page(struct kvm *kvm, gfn_t gfn, int offset, int len)
{
2181 2182 2183
	const void *zero_page = (const void *) __va(page_to_phys(ZERO_PAGE(0)));

	return kvm_write_guest_page(kvm, gfn, zero_page, offset, len);
2184 2185 2186 2187 2188 2189 2190 2191 2192 2193
}
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;

2194
	while ((seg = next_segment(len, offset)) != 0) {
2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205
		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);

2206
static void mark_page_dirty_in_slot(struct kvm_memory_slot *memslot,
2207
				    gfn_t gfn)
A
Avi Kivity 已提交
2208
{
R
Rusty Russell 已提交
2209 2210
	if (memslot && memslot->dirty_bitmap) {
		unsigned long rel_gfn = gfn - memslot->base_gfn;
A
Avi Kivity 已提交
2211

2212
		set_bit_le(rel_gfn, memslot->dirty_bitmap);
A
Avi Kivity 已提交
2213 2214 2215
	}
}

2216 2217 2218 2219 2220
void mark_page_dirty(struct kvm *kvm, gfn_t gfn)
{
	struct kvm_memory_slot *memslot;

	memslot = gfn_to_memslot(kvm, gfn);
2221
	mark_page_dirty_in_slot(memslot, gfn);
2222
}
2223
EXPORT_SYMBOL_GPL(mark_page_dirty);
2224

2225 2226 2227 2228 2229 2230 2231 2232 2233
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);

2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256
void kvm_sigset_activate(struct kvm_vcpu *vcpu)
{
	if (!vcpu->sigset_active)
		return;

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

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

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

W
Wanpeng Li 已提交
2257 2258
static void grow_halt_poll_ns(struct kvm_vcpu *vcpu)
{
2259
	unsigned int old, val, grow, grow_start;
W
Wanpeng Li 已提交
2260

2261
	old = val = vcpu->halt_poll_ns;
2262
	grow_start = READ_ONCE(halt_poll_ns_grow_start);
2263
	grow = READ_ONCE(halt_poll_ns_grow);
2264 2265 2266
	if (!grow)
		goto out;

2267 2268 2269
	val *= grow;
	if (val < grow_start)
		val = grow_start;
W
Wanpeng Li 已提交
2270

2271 2272 2273
	if (val > halt_poll_ns)
		val = halt_poll_ns;

W
Wanpeng Li 已提交
2274
	vcpu->halt_poll_ns = val;
2275
out:
2276
	trace_kvm_halt_poll_ns_grow(vcpu->vcpu_id, val, old);
W
Wanpeng Li 已提交
2277 2278 2279 2280
}

static void shrink_halt_poll_ns(struct kvm_vcpu *vcpu)
{
2281
	unsigned int old, val, shrink;
W
Wanpeng Li 已提交
2282

2283
	old = val = vcpu->halt_poll_ns;
2284 2285
	shrink = READ_ONCE(halt_poll_ns_shrink);
	if (shrink == 0)
W
Wanpeng Li 已提交
2286 2287
		val = 0;
	else
2288
		val /= shrink;
W
Wanpeng Li 已提交
2289 2290

	vcpu->halt_poll_ns = val;
2291
	trace_kvm_halt_poll_ns_shrink(vcpu->vcpu_id, val, old);
W
Wanpeng Li 已提交
2292 2293
}

2294 2295
static int kvm_vcpu_check_block(struct kvm_vcpu *vcpu)
{
2296 2297 2298
	int ret = -EINTR;
	int idx = srcu_read_lock(&vcpu->kvm->srcu);

2299 2300
	if (kvm_arch_vcpu_runnable(vcpu)) {
		kvm_make_request(KVM_REQ_UNHALT, vcpu);
2301
		goto out;
2302 2303
	}
	if (kvm_cpu_has_pending_timer(vcpu))
2304
		goto out;
2305
	if (signal_pending(current))
2306
		goto out;
2307

2308 2309 2310 2311
	ret = 0;
out:
	srcu_read_unlock(&vcpu->kvm->srcu, idx);
	return ret;
2312 2313
}

E
Eddie Dong 已提交
2314 2315 2316
/*
 * The vCPU has executed a HLT instruction with in-kernel mode enabled.
 */
2317
void kvm_vcpu_block(struct kvm_vcpu *vcpu)
2318
{
2319
	ktime_t start, cur;
2320
	DECLARE_SWAITQUEUE(wait);
2321
	bool waited = false;
W
Wanpeng Li 已提交
2322
	u64 block_ns;
2323 2324

	start = cur = ktime_get();
2325
	if (vcpu->halt_poll_ns && !kvm_arch_no_poll(vcpu)) {
W
Wanpeng Li 已提交
2326
		ktime_t stop = ktime_add_ns(ktime_get(), vcpu->halt_poll_ns);
2327

2328
		++vcpu->stat.halt_attempted_poll;
2329 2330 2331 2332 2333 2334 2335
		do {
			/*
			 * This sets KVM_REQ_UNHALT if an interrupt
			 * arrives.
			 */
			if (kvm_vcpu_check_block(vcpu) < 0) {
				++vcpu->stat.halt_successful_poll;
2336 2337
				if (!vcpu_valid_wakeup(vcpu))
					++vcpu->stat.halt_poll_invalid;
2338 2339 2340 2341 2342
				goto out;
			}
			cur = ktime_get();
		} while (single_task_running() && ktime_before(cur, stop));
	}
2343

2344 2345
	kvm_arch_vcpu_blocking(vcpu);

2346
	for (;;) {
2347
		prepare_to_swait_exclusive(&vcpu->wq, &wait, TASK_INTERRUPTIBLE);
2348

2349
		if (kvm_vcpu_check_block(vcpu) < 0)
2350 2351
			break;

2352
		waited = true;
E
Eddie Dong 已提交
2353 2354
		schedule();
	}
2355

2356
	finish_swait(&vcpu->wq, &wait);
2357 2358
	cur = ktime_get();

2359
	kvm_arch_vcpu_unblocking(vcpu);
2360
out:
W
Wanpeng Li 已提交
2361 2362
	block_ns = ktime_to_ns(cur) - ktime_to_ns(start);

2363 2364 2365
	if (!vcpu_valid_wakeup(vcpu))
		shrink_halt_poll_ns(vcpu);
	else if (halt_poll_ns) {
W
Wanpeng Li 已提交
2366 2367 2368
		if (block_ns <= vcpu->halt_poll_ns)
			;
		/* we had a long block, shrink polling */
2369
		else if (vcpu->halt_poll_ns && block_ns > halt_poll_ns)
W
Wanpeng Li 已提交
2370 2371 2372 2373 2374
			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);
2375 2376
	} else
		vcpu->halt_poll_ns = 0;
W
Wanpeng Li 已提交
2377

2378 2379
	trace_kvm_vcpu_wakeup(block_ns, waited, vcpu_valid_wakeup(vcpu));
	kvm_arch_vcpu_block_finish(vcpu);
E
Eddie Dong 已提交
2380
}
2381
EXPORT_SYMBOL_GPL(kvm_vcpu_block);
E
Eddie Dong 已提交
2382

2383
bool kvm_vcpu_wake_up(struct kvm_vcpu *vcpu)
2384
{
2385
	struct swait_queue_head *wqp;
2386 2387

	wqp = kvm_arch_vcpu_wq(vcpu);
2388
	if (swq_has_sleeper(wqp)) {
2389
		swake_up_one(wqp);
2390
		++vcpu->stat.halt_wakeup;
2391
		return true;
2392 2393
	}

2394
	return false;
2395 2396 2397
}
EXPORT_SYMBOL_GPL(kvm_vcpu_wake_up);

2398
#ifndef CONFIG_S390
2399 2400 2401 2402 2403 2404 2405 2406
/*
 * 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;

2407 2408 2409
	if (kvm_vcpu_wake_up(vcpu))
		return;

2410 2411 2412 2413 2414 2415
	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();
}
2416
EXPORT_SYMBOL_GPL(kvm_vcpu_kick);
2417
#endif /* !CONFIG_S390 */
2418

2419
int kvm_vcpu_yield_to(struct kvm_vcpu *target)
2420 2421 2422
{
	struct pid *pid;
	struct task_struct *task = NULL;
2423
	int ret = 0;
2424 2425 2426 2427

	rcu_read_lock();
	pid = rcu_dereference(target->pid);
	if (pid)
2428
		task = get_pid_task(pid, PIDTYPE_PID);
2429 2430
	rcu_read_unlock();
	if (!task)
2431 2432
		return ret;
	ret = yield_to(task, 1);
2433
	put_task_struct(task);
2434 2435

	return ret;
2436 2437 2438
}
EXPORT_SYMBOL_GPL(kvm_vcpu_yield_to);

2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460
/*
 * 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.
 */
2461
static bool kvm_vcpu_eligible_for_directed_yield(struct kvm_vcpu *vcpu)
2462
{
2463
#ifdef CONFIG_HAVE_KVM_CPU_RELAX_INTERCEPT
2464 2465 2466
	bool eligible;

	eligible = !vcpu->spin_loop.in_spin_loop ||
2467
		    vcpu->spin_loop.dy_eligible;
2468 2469 2470 2471 2472

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

	return eligible;
2473 2474
#else
	return true;
2475
#endif
2476
}
2477

2478
void kvm_vcpu_on_spin(struct kvm_vcpu *me, bool yield_to_kernel_mode)
Z
Zhai, Edwin 已提交
2479
{
2480 2481 2482 2483
	struct kvm *kvm = me->kvm;
	struct kvm_vcpu *vcpu;
	int last_boosted_vcpu = me->kvm->last_boosted_vcpu;
	int yielded = 0;
2484
	int try = 3;
2485 2486
	int pass;
	int i;
Z
Zhai, Edwin 已提交
2487

2488
	kvm_vcpu_set_in_spin_loop(me, true);
2489 2490 2491 2492 2493 2494 2495
	/*
	 * 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.
	 */
2496
	for (pass = 0; pass < 2 && !yielded && try; pass++) {
2497
		kvm_for_each_vcpu(i, vcpu, kvm) {
2498
			if (!pass && i <= last_boosted_vcpu) {
2499 2500 2501 2502
				i = last_boosted_vcpu;
				continue;
			} else if (pass && i > last_boosted_vcpu)
				break;
2503
			if (!READ_ONCE(vcpu->preempted))
2504
				continue;
2505 2506
			if (vcpu == me)
				continue;
2507
			if (swait_active(&vcpu->wq) && !kvm_arch_vcpu_runnable(vcpu))
2508
				continue;
2509 2510
			if (yield_to_kernel_mode && !kvm_arch_vcpu_in_kernel(vcpu))
				continue;
2511 2512
			if (!kvm_vcpu_eligible_for_directed_yield(vcpu))
				continue;
2513 2514 2515

			yielded = kvm_vcpu_yield_to(vcpu);
			if (yielded > 0) {
2516 2517
				kvm->last_boosted_vcpu = i;
				break;
2518 2519 2520 2521
			} else if (yielded < 0) {
				try--;
				if (!try)
					break;
2522 2523 2524
			}
		}
	}
2525
	kvm_vcpu_set_in_spin_loop(me, false);
2526 2527 2528

	/* Ensure vcpu is not eligible during next spinloop */
	kvm_vcpu_set_dy_eligible(me, false);
Z
Zhai, Edwin 已提交
2529 2530 2531
}
EXPORT_SYMBOL_GPL(kvm_vcpu_on_spin);

2532
static vm_fault_t kvm_vcpu_fault(struct vm_fault *vmf)
2533
{
2534
	struct kvm_vcpu *vcpu = vmf->vma->vm_file->private_data;
2535 2536
	struct page *page;

2537
	if (vmf->pgoff == 0)
2538
		page = virt_to_page(vcpu->run);
A
Avi Kivity 已提交
2539
#ifdef CONFIG_X86
2540
	else if (vmf->pgoff == KVM_PIO_PAGE_OFFSET)
2541
		page = virt_to_page(vcpu->arch.pio_data);
2542
#endif
2543
#ifdef CONFIG_KVM_MMIO
2544 2545
	else if (vmf->pgoff == KVM_COALESCED_MMIO_PAGE_OFFSET)
		page = virt_to_page(vcpu->kvm->coalesced_mmio_ring);
A
Avi Kivity 已提交
2546
#endif
2547
	else
2548
		return kvm_arch_vcpu_fault(vcpu, vmf);
2549
	get_page(page);
2550 2551
	vmf->page = page;
	return 0;
2552 2553
}

2554
static const struct vm_operations_struct kvm_vcpu_vm_ops = {
2555
	.fault = kvm_vcpu_fault,
2556 2557 2558 2559 2560 2561 2562 2563
};

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
2564 2565 2566 2567
static int kvm_vcpu_release(struct inode *inode, struct file *filp)
{
	struct kvm_vcpu *vcpu = filp->private_data;

2568
	debugfs_remove_recursive(vcpu->debugfs_dentry);
A
Al Viro 已提交
2569
	kvm_put_kvm(vcpu->kvm);
A
Avi Kivity 已提交
2570 2571 2572
	return 0;
}

2573
static struct file_operations kvm_vcpu_fops = {
A
Avi Kivity 已提交
2574 2575
	.release        = kvm_vcpu_release,
	.unlocked_ioctl = kvm_vcpu_ioctl,
2576
	.mmap           = kvm_vcpu_mmap,
2577
	.llseek		= noop_llseek,
2578
	KVM_COMPAT(kvm_vcpu_compat_ioctl),
A
Avi Kivity 已提交
2579 2580 2581 2582 2583 2584 2585
};

/*
 * Allocates an inode for the vcpu.
 */
static int create_vcpu_fd(struct kvm_vcpu *vcpu)
{
2586 2587 2588 2589
	char name[8 + 1 + ITOA_MAX_LEN + 1];

	snprintf(name, sizeof(name), "kvm-vcpu:%d", vcpu->vcpu_id);
	return anon_inode_getfd(name, &kvm_vcpu_fops, vcpu, O_RDWR | O_CLOEXEC);
A
Avi Kivity 已提交
2590 2591
}

2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617
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;
}

2618 2619 2620
/*
 * Creates some virtual cpus.  Good luck creating more than one.
 */
2621
static int kvm_vm_ioctl_create_vcpu(struct kvm *kvm, u32 id)
2622 2623
{
	int r;
2624
	struct kvm_vcpu *vcpu;
2625

G
Greg Kurz 已提交
2626
	if (id >= KVM_MAX_VCPU_ID)
2627 2628
		return -EINVAL;

2629 2630 2631 2632 2633 2634 2635 2636 2637
	mutex_lock(&kvm->lock);
	if (kvm->created_vcpus == KVM_MAX_VCPUS) {
		mutex_unlock(&kvm->lock);
		return -EINVAL;
	}

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

2638
	vcpu = kvm_arch_vcpu_create(kvm, id);
2639 2640 2641 2642
	if (IS_ERR(vcpu)) {
		r = PTR_ERR(vcpu);
		goto vcpu_decrement;
	}
2643

2644 2645
	preempt_notifier_init(&vcpu->preempt_notifier, &kvm_preempt_ops);

2646 2647
	r = kvm_arch_vcpu_setup(vcpu);
	if (r)
2648
		goto vcpu_destroy;
2649

2650 2651 2652 2653
	r = kvm_create_vcpu_debugfs(vcpu);
	if (r)
		goto vcpu_destroy;

S
Shaohua Li 已提交
2654
	mutex_lock(&kvm->lock);
2655 2656 2657 2658
	if (kvm_get_vcpu_by_id(kvm, id)) {
		r = -EEXIST;
		goto unlock_vcpu_destroy;
	}
2659 2660

	BUG_ON(kvm->vcpus[atomic_read(&kvm->online_vcpus)]);
2661

R
Rusty Russell 已提交
2662
	/* Now it's all set up, let userspace reach it */
A
Al Viro 已提交
2663
	kvm_get_kvm(kvm);
A
Avi Kivity 已提交
2664
	r = create_vcpu_fd(vcpu);
2665 2666
	if (r < 0) {
		kvm_put_kvm(kvm);
2667
		goto unlock_vcpu_destroy;
2668 2669 2670
	}

	kvm->vcpus[atomic_read(&kvm->online_vcpus)] = vcpu;
2671 2672 2673 2674 2675

	/*
	 * Pairs with smp_rmb() in kvm_get_vcpu.  Write kvm->vcpus
	 * before kvm->online_vcpu's incremented value.
	 */
2676 2677 2678 2679
	smp_wmb();
	atomic_inc(&kvm->online_vcpus);

	mutex_unlock(&kvm->lock);
2680
	kvm_arch_vcpu_postcreate(vcpu);
R
Rusty Russell 已提交
2681
	return r;
2682

2683
unlock_vcpu_destroy:
2684
	mutex_unlock(&kvm->lock);
2685
	debugfs_remove_recursive(vcpu->debugfs_dentry);
2686
vcpu_destroy:
2687
	kvm_arch_vcpu_destroy(vcpu);
2688 2689 2690 2691
vcpu_decrement:
	mutex_lock(&kvm->lock);
	kvm->created_vcpus--;
	mutex_unlock(&kvm->lock);
2692 2693 2694
	return r;
}

A
Avi Kivity 已提交
2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705
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
2706 2707
static long kvm_vcpu_ioctl(struct file *filp,
			   unsigned int ioctl, unsigned long arg)
A
Avi Kivity 已提交
2708
{
A
Avi Kivity 已提交
2709
	struct kvm_vcpu *vcpu = filp->private_data;
A
Al Viro 已提交
2710
	void __user *argp = (void __user *)arg;
2711
	int r;
2712 2713
	struct kvm_fpu *fpu = NULL;
	struct kvm_sregs *kvm_sregs = NULL;
A
Avi Kivity 已提交
2714

2715 2716
	if (vcpu->kvm->mm != current->mm)
		return -EIO;
2717

2718 2719 2720
	if (unlikely(_IOC_TYPE(ioctl) != KVMIO))
		return -EINVAL;

2721
	/*
2722 2723
	 * Some architectures have vcpu ioctls that are asynchronous to vcpu
	 * execution; mutex_lock() would break them.
2724
	 */
2725 2726
	r = kvm_arch_vcpu_async_ioctl(filp, ioctl, arg);
	if (r != -ENOIOCTLCMD)
2727
		return r;
2728

2729 2730
	if (mutex_lock_killable(&vcpu->mutex))
		return -EINTR;
A
Avi Kivity 已提交
2731
	switch (ioctl) {
2732 2733
	case KVM_RUN: {
		struct pid *oldpid;
2734 2735 2736
		r = -EINVAL;
		if (arg)
			goto out;
2737
		oldpid = rcu_access_pointer(vcpu->pid);
2738
		if (unlikely(oldpid != task_pid(current))) {
2739
			/* The thread running this VCPU changed. */
2740
			struct pid *newpid;
2741

2742 2743 2744 2745 2746
			r = kvm_arch_vcpu_run_pid_change(vcpu);
			if (r)
				break;

			newpid = get_task_pid(current, PIDTYPE_PID);
2747 2748 2749 2750 2751
			rcu_assign_pointer(vcpu->pid, newpid);
			if (oldpid)
				synchronize_rcu();
			put_pid(oldpid);
		}
2752
		r = kvm_arch_vcpu_ioctl_run(vcpu, vcpu->run);
2753
		trace_kvm_userspace_exit(vcpu->run->exit_reason, r);
A
Avi Kivity 已提交
2754
		break;
2755
	}
A
Avi Kivity 已提交
2756
	case KVM_GET_REGS: {
2757
		struct kvm_regs *kvm_regs;
A
Avi Kivity 已提交
2758

2759
		r = -ENOMEM;
2760
		kvm_regs = kzalloc(sizeof(struct kvm_regs), GFP_KERNEL_ACCOUNT);
2761
		if (!kvm_regs)
A
Avi Kivity 已提交
2762
			goto out;
2763 2764 2765
		r = kvm_arch_vcpu_ioctl_get_regs(vcpu, kvm_regs);
		if (r)
			goto out_free1;
A
Avi Kivity 已提交
2766
		r = -EFAULT;
2767 2768
		if (copy_to_user(argp, kvm_regs, sizeof(struct kvm_regs)))
			goto out_free1;
A
Avi Kivity 已提交
2769
		r = 0;
2770 2771
out_free1:
		kfree(kvm_regs);
A
Avi Kivity 已提交
2772 2773 2774
		break;
	}
	case KVM_SET_REGS: {
2775
		struct kvm_regs *kvm_regs;
A
Avi Kivity 已提交
2776

2777
		r = -ENOMEM;
2778 2779 2780
		kvm_regs = memdup_user(argp, sizeof(*kvm_regs));
		if (IS_ERR(kvm_regs)) {
			r = PTR_ERR(kvm_regs);
A
Avi Kivity 已提交
2781
			goto out;
2782
		}
2783 2784
		r = kvm_arch_vcpu_ioctl_set_regs(vcpu, kvm_regs);
		kfree(kvm_regs);
A
Avi Kivity 已提交
2785 2786 2787
		break;
	}
	case KVM_GET_SREGS: {
2788 2789
		kvm_sregs = kzalloc(sizeof(struct kvm_sregs),
				    GFP_KERNEL_ACCOUNT);
2790 2791 2792 2793
		r = -ENOMEM;
		if (!kvm_sregs)
			goto out;
		r = kvm_arch_vcpu_ioctl_get_sregs(vcpu, kvm_sregs);
A
Avi Kivity 已提交
2794 2795 2796
		if (r)
			goto out;
		r = -EFAULT;
2797
		if (copy_to_user(argp, kvm_sregs, sizeof(struct kvm_sregs)))
A
Avi Kivity 已提交
2798 2799 2800 2801 2802
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_SREGS: {
2803 2804 2805
		kvm_sregs = memdup_user(argp, sizeof(*kvm_sregs));
		if (IS_ERR(kvm_sregs)) {
			r = PTR_ERR(kvm_sregs);
G
Guo Chao 已提交
2806
			kvm_sregs = NULL;
A
Avi Kivity 已提交
2807
			goto out;
2808
		}
2809
		r = kvm_arch_vcpu_ioctl_set_sregs(vcpu, kvm_sregs);
A
Avi Kivity 已提交
2810 2811
		break;
	}
2812 2813 2814 2815 2816 2817 2818
	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;
2819
		if (copy_to_user(argp, &mp_state, sizeof(mp_state)))
2820 2821 2822 2823 2824 2825 2826 2827
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_MP_STATE: {
		struct kvm_mp_state mp_state;

		r = -EFAULT;
2828
		if (copy_from_user(&mp_state, argp, sizeof(mp_state)))
2829 2830 2831 2832
			goto out;
		r = kvm_arch_vcpu_ioctl_set_mpstate(vcpu, &mp_state);
		break;
	}
A
Avi Kivity 已提交
2833 2834 2835 2836
	case KVM_TRANSLATE: {
		struct kvm_translation tr;

		r = -EFAULT;
2837
		if (copy_from_user(&tr, argp, sizeof(tr)))
A
Avi Kivity 已提交
2838
			goto out;
2839
		r = kvm_arch_vcpu_ioctl_translate(vcpu, &tr);
A
Avi Kivity 已提交
2840 2841 2842
		if (r)
			goto out;
		r = -EFAULT;
2843
		if (copy_to_user(argp, &tr, sizeof(tr)))
A
Avi Kivity 已提交
2844 2845 2846 2847
			goto out;
		r = 0;
		break;
	}
J
Jan Kiszka 已提交
2848 2849
	case KVM_SET_GUEST_DEBUG: {
		struct kvm_guest_debug dbg;
A
Avi Kivity 已提交
2850 2851

		r = -EFAULT;
2852
		if (copy_from_user(&dbg, argp, sizeof(dbg)))
A
Avi Kivity 已提交
2853
			goto out;
J
Jan Kiszka 已提交
2854
		r = kvm_arch_vcpu_ioctl_set_guest_debug(vcpu, &dbg);
A
Avi Kivity 已提交
2855 2856
		break;
	}
A
Avi Kivity 已提交
2857 2858 2859 2860 2861 2862 2863 2864 2865
	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,
2866
					   sizeof(kvm_sigmask)))
A
Avi Kivity 已提交
2867 2868
				goto out;
			r = -EINVAL;
2869
			if (kvm_sigmask.len != sizeof(sigset))
A
Avi Kivity 已提交
2870 2871 2872
				goto out;
			r = -EFAULT;
			if (copy_from_user(&sigset, sigmask_arg->sigset,
2873
					   sizeof(sigset)))
A
Avi Kivity 已提交
2874 2875 2876
				goto out;
			p = &sigset;
		}
2877
		r = kvm_vcpu_ioctl_set_sigmask(vcpu, p);
A
Avi Kivity 已提交
2878 2879
		break;
	}
A
Avi Kivity 已提交
2880
	case KVM_GET_FPU: {
2881
		fpu = kzalloc(sizeof(struct kvm_fpu), GFP_KERNEL_ACCOUNT);
2882 2883 2884 2885
		r = -ENOMEM;
		if (!fpu)
			goto out;
		r = kvm_arch_vcpu_ioctl_get_fpu(vcpu, fpu);
A
Avi Kivity 已提交
2886 2887 2888
		if (r)
			goto out;
		r = -EFAULT;
2889
		if (copy_to_user(argp, fpu, sizeof(struct kvm_fpu)))
A
Avi Kivity 已提交
2890 2891 2892 2893 2894
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_FPU: {
2895 2896 2897
		fpu = memdup_user(argp, sizeof(*fpu));
		if (IS_ERR(fpu)) {
			r = PTR_ERR(fpu);
G
Guo Chao 已提交
2898
			fpu = NULL;
A
Avi Kivity 已提交
2899
			goto out;
2900
		}
2901
		r = kvm_arch_vcpu_ioctl_set_fpu(vcpu, fpu);
A
Avi Kivity 已提交
2902 2903
		break;
	}
A
Avi Kivity 已提交
2904
	default:
2905
		r = kvm_arch_vcpu_ioctl(filp, ioctl, arg);
A
Avi Kivity 已提交
2906 2907
	}
out:
2908
	mutex_unlock(&vcpu->mutex);
2909 2910
	kfree(fpu);
	kfree(kvm_sregs);
A
Avi Kivity 已提交
2911 2912 2913
	return r;
}

2914
#ifdef CONFIG_KVM_COMPAT
2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933
static long kvm_vcpu_compat_ioctl(struct file *filp,
				  unsigned int ioctl, unsigned long arg)
{
	struct kvm_vcpu *vcpu = filp->private_data;
	void __user *argp = compat_ptr(arg);
	int r;

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

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

		if (argp) {
			r = -EFAULT;
			if (copy_from_user(&kvm_sigmask, argp,
2934
					   sizeof(kvm_sigmask)))
2935 2936
				goto out;
			r = -EINVAL;
A
Al Viro 已提交
2937
			if (kvm_sigmask.len != sizeof(compat_sigset_t))
2938 2939
				goto out;
			r = -EFAULT;
A
Al Viro 已提交
2940
			if (get_compat_sigset(&sigset, (void *)sigmask_arg->sigset))
2941
				goto out;
2942 2943 2944
			r = kvm_vcpu_ioctl_set_sigmask(vcpu, &sigset);
		} else
			r = kvm_vcpu_ioctl_set_sigmask(vcpu, NULL);
2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955
		break;
	}
	default:
		r = kvm_vcpu_ioctl(filp, ioctl, arg);
	}

out:
	return r;
}
#endif

2956 2957 2958 2959 2960 2961 2962 2963 2964 2965
static int kvm_device_mmap(struct file *filp, struct vm_area_struct *vma)
{
	struct kvm_device *dev = filp->private_data;

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

	return -ENODEV;
}

S
Scott Wood 已提交
2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986
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;

2987 2988 2989
	if (dev->kvm->mm != current->mm)
		return -EIO;

S
Scott Wood 已提交
2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009
	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;

3010 3011 3012 3013 3014 3015 3016
	if (dev->ops->release) {
		mutex_lock(&kvm->lock);
		list_del(&dev->vm_node);
		dev->ops->release(dev);
		mutex_unlock(&kvm->lock);
	}

S
Scott Wood 已提交
3017 3018 3019 3020 3021 3022 3023
	kvm_put_kvm(kvm);
	return 0;
}

static const struct file_operations kvm_device_fops = {
	.unlocked_ioctl = kvm_device_ioctl,
	.release = kvm_device_release,
3024
	KVM_COMPAT(kvm_device_ioctl),
3025
	.mmap = kvm_device_mmap,
S
Scott Wood 已提交
3026 3027 3028 3029 3030 3031 3032 3033 3034 3035
};

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

	return filp->private_data;
}

3036
static struct kvm_device_ops *kvm_device_ops_table[KVM_DEV_TYPE_MAX] = {
3037
#ifdef CONFIG_KVM_MPIC
3038 3039
	[KVM_DEV_TYPE_FSL_MPIC_20]	= &kvm_mpic_ops,
	[KVM_DEV_TYPE_FSL_MPIC_42]	= &kvm_mpic_ops,
3040
#endif
3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054
};

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;
}

3055 3056 3057 3058 3059 3060
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
3061 3062 3063 3064 3065 3066
static int kvm_ioctl_create_device(struct kvm *kvm,
				   struct kvm_create_device *cd)
{
	struct kvm_device_ops *ops = NULL;
	struct kvm_device *dev;
	bool test = cd->flags & KVM_CREATE_DEVICE_TEST;
P
Paolo Bonzini 已提交
3067
	int type;
S
Scott Wood 已提交
3068 3069
	int ret;

3070 3071 3072
	if (cd->type >= ARRAY_SIZE(kvm_device_ops_table))
		return -ENODEV;

P
Paolo Bonzini 已提交
3073 3074
	type = array_index_nospec(cd->type, ARRAY_SIZE(kvm_device_ops_table));
	ops = kvm_device_ops_table[type];
3075
	if (ops == NULL)
S
Scott Wood 已提交
3076 3077 3078 3079 3080
		return -ENODEV;

	if (test)
		return 0;

3081
	dev = kzalloc(sizeof(*dev), GFP_KERNEL_ACCOUNT);
S
Scott Wood 已提交
3082 3083 3084 3085 3086 3087
	if (!dev)
		return -ENOMEM;

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

3088
	mutex_lock(&kvm->lock);
P
Paolo Bonzini 已提交
3089
	ret = ops->create(dev, type);
S
Scott Wood 已提交
3090
	if (ret < 0) {
3091
		mutex_unlock(&kvm->lock);
S
Scott Wood 已提交
3092 3093 3094
		kfree(dev);
		return ret;
	}
3095 3096
	list_add(&dev->vm_node, &kvm->devices);
	mutex_unlock(&kvm->lock);
S
Scott Wood 已提交
3097

3098 3099 3100
	if (ops->init)
		ops->init(dev);

3101
	kvm_get_kvm(kvm);
3102
	ret = anon_inode_getfd(ops->name, &kvm_device_fops, dev, O_RDWR | O_CLOEXEC);
S
Scott Wood 已提交
3103
	if (ret < 0) {
3104
		kvm_put_kvm(kvm);
3105 3106 3107
		mutex_lock(&kvm->lock);
		list_del(&dev->vm_node);
		mutex_unlock(&kvm->lock);
3108
		ops->destroy(dev);
S
Scott Wood 已提交
3109 3110 3111 3112 3113 3114 3115
		return ret;
	}

	cd->fd = ret;
	return 0;
}

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

3159 3160 3161 3162 3163 3164 3165 3166 3167 3168
int __attribute__((weak)) kvm_vm_ioctl_enable_cap(struct kvm *kvm,
						  struct kvm_enable_cap *cap)
{
	return -EINVAL;
}

static int kvm_vm_ioctl_enable_cap_generic(struct kvm *kvm,
					   struct kvm_enable_cap *cap)
{
	switch (cap->cap) {
3169
#ifdef CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT
3170
	case KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2:
3171 3172 3173 3174 3175
		if (cap->flags || (cap->args[0] & ~1))
			return -EINVAL;
		kvm->manual_dirty_log_protect = cap->args[0];
		return 0;
#endif
3176 3177 3178 3179 3180
	default:
		return kvm_vm_ioctl_enable_cap(kvm, cap);
	}
}

A
Avi Kivity 已提交
3181 3182 3183 3184 3185
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;
3186
	int r;
A
Avi Kivity 已提交
3187

3188 3189
	if (kvm->mm != current->mm)
		return -EIO;
A
Avi Kivity 已提交
3190 3191 3192 3193
	switch (ioctl) {
	case KVM_CREATE_VCPU:
		r = kvm_vm_ioctl_create_vcpu(kvm, arg);
		break;
3194 3195 3196 3197 3198 3199 3200 3201 3202
	case KVM_ENABLE_CAP: {
		struct kvm_enable_cap cap;

		r = -EFAULT;
		if (copy_from_user(&cap, argp, sizeof(cap)))
			goto out;
		r = kvm_vm_ioctl_enable_cap_generic(kvm, &cap);
		break;
	}
3203 3204 3205 3206 3207
	case KVM_SET_USER_MEMORY_REGION: {
		struct kvm_userspace_memory_region kvm_userspace_mem;

		r = -EFAULT;
		if (copy_from_user(&kvm_userspace_mem, argp,
3208
						sizeof(kvm_userspace_mem)))
3209 3210
			goto out;

3211
		r = kvm_vm_ioctl_set_memory_region(kvm, &kvm_userspace_mem);
A
Avi Kivity 已提交
3212 3213 3214 3215 3216 3217
		break;
	}
	case KVM_GET_DIRTY_LOG: {
		struct kvm_dirty_log log;

		r = -EFAULT;
3218
		if (copy_from_user(&log, argp, sizeof(log)))
A
Avi Kivity 已提交
3219
			goto out;
3220
		r = kvm_vm_ioctl_get_dirty_log(kvm, &log);
A
Avi Kivity 已提交
3221 3222
		break;
	}
3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233
#ifdef CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT
	case KVM_CLEAR_DIRTY_LOG: {
		struct kvm_clear_dirty_log log;

		r = -EFAULT;
		if (copy_from_user(&log, argp, sizeof(log)))
			goto out;
		r = kvm_vm_ioctl_clear_dirty_log(kvm, &log);
		break;
	}
#endif
3234
#ifdef CONFIG_KVM_MMIO
3235 3236
	case KVM_REGISTER_COALESCED_MMIO: {
		struct kvm_coalesced_mmio_zone zone;
3237

3238
		r = -EFAULT;
3239
		if (copy_from_user(&zone, argp, sizeof(zone)))
3240 3241 3242 3243 3244 3245
			goto out;
		r = kvm_vm_ioctl_register_coalesced_mmio(kvm, &zone);
		break;
	}
	case KVM_UNREGISTER_COALESCED_MMIO: {
		struct kvm_coalesced_mmio_zone zone;
3246

3247
		r = -EFAULT;
3248
		if (copy_from_user(&zone, argp, sizeof(zone)))
3249 3250 3251 3252 3253
			goto out;
		r = kvm_vm_ioctl_unregister_coalesced_mmio(kvm, &zone);
		break;
	}
#endif
G
Gregory Haskins 已提交
3254 3255 3256 3257
	case KVM_IRQFD: {
		struct kvm_irqfd data;

		r = -EFAULT;
3258
		if (copy_from_user(&data, argp, sizeof(data)))
G
Gregory Haskins 已提交
3259
			goto out;
3260
		r = kvm_irqfd(kvm, &data);
G
Gregory Haskins 已提交
3261 3262
		break;
	}
G
Gregory Haskins 已提交
3263 3264 3265 3266
	case KVM_IOEVENTFD: {
		struct kvm_ioeventfd data;

		r = -EFAULT;
3267
		if (copy_from_user(&data, argp, sizeof(data)))
G
Gregory Haskins 已提交
3268 3269 3270 3271
			goto out;
		r = kvm_ioeventfd(kvm, &data);
		break;
	}
3272 3273 3274 3275 3276
#ifdef CONFIG_HAVE_KVM_MSI
	case KVM_SIGNAL_MSI: {
		struct kvm_msi msi;

		r = -EFAULT;
3277
		if (copy_from_user(&msi, argp, sizeof(msi)))
3278 3279 3280 3281
			goto out;
		r = kvm_send_userspace_msi(kvm, &msi);
		break;
	}
3282 3283 3284 3285 3286 3287 3288
#endif
#ifdef __KVM_HAVE_IRQ_LINE
	case KVM_IRQ_LINE_STATUS:
	case KVM_IRQ_LINE: {
		struct kvm_irq_level irq_event;

		r = -EFAULT;
3289
		if (copy_from_user(&irq_event, argp, sizeof(irq_event)))
3290 3291
			goto out;

3292 3293
		r = kvm_vm_ioctl_irq_line(kvm, &irq_event,
					ioctl == KVM_IRQ_LINE_STATUS);
3294 3295 3296 3297 3298
		if (r)
			goto out;

		r = -EFAULT;
		if (ioctl == KVM_IRQ_LINE_STATUS) {
3299
			if (copy_to_user(argp, &irq_event, sizeof(irq_event)))
3300 3301 3302 3303 3304 3305
				goto out;
		}

		r = 0;
		break;
	}
3306
#endif
3307 3308 3309 3310
#ifdef CONFIG_HAVE_KVM_IRQ_ROUTING
	case KVM_SET_GSI_ROUTING: {
		struct kvm_irq_routing routing;
		struct kvm_irq_routing __user *urouting;
3311
		struct kvm_irq_routing_entry *entries = NULL;
3312 3313 3314 3315 3316

		r = -EFAULT;
		if (copy_from_user(&routing, argp, sizeof(routing)))
			goto out;
		r = -EINVAL;
3317 3318
		if (!kvm_arch_can_set_irq_routing(kvm))
			goto out;
3319
		if (routing.nr > KVM_MAX_IRQ_ROUTES)
3320 3321 3322
			goto out;
		if (routing.flags)
			goto out;
3323 3324
		if (routing.nr) {
			r = -ENOMEM;
3325 3326
			entries = vmalloc(array_size(sizeof(*entries),
						     routing.nr));
3327 3328 3329 3330 3331 3332 3333 3334
			if (!entries)
				goto out;
			r = -EFAULT;
			urouting = argp;
			if (copy_from_user(entries, urouting->entries,
					   routing.nr * sizeof(*entries)))
				goto out_free_irq_routing;
		}
3335 3336
		r = kvm_set_irq_routing(kvm, entries, routing.nr,
					routing.flags);
3337
out_free_irq_routing:
3338 3339 3340 3341
		vfree(entries);
		break;
	}
#endif /* CONFIG_HAVE_KVM_IRQ_ROUTING */
S
Scott Wood 已提交
3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359
	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;
	}
3360 3361 3362
	case KVM_CHECK_EXTENSION:
		r = kvm_vm_ioctl_check_extension_generic(kvm, arg);
		break;
3363
	default:
3364
		r = kvm_arch_vm_ioctl(filp, ioctl, arg);
3365 3366 3367 3368 3369
	}
out:
	return r;
}

3370
#ifdef CONFIG_KVM_COMPAT
3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394
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)))
3395
			return -EFAULT;
3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410
		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

3411
static struct file_operations kvm_vm_fops = {
3412 3413
	.release        = kvm_vm_release,
	.unlocked_ioctl = kvm_vm_ioctl,
3414
	.llseek		= noop_llseek,
3415
	KVM_COMPAT(kvm_vm_compat_ioctl),
3416 3417
};

3418
static int kvm_dev_ioctl_create_vm(unsigned long type)
3419
{
3420
	int r;
3421
	struct kvm *kvm;
3422
	struct file *file;
3423

3424
	kvm = kvm_create_vm(type);
3425 3426
	if (IS_ERR(kvm))
		return PTR_ERR(kvm);
3427
#ifdef CONFIG_KVM_MMIO
3428
	r = kvm_coalesced_mmio_init(kvm);
3429 3430
	if (r < 0)
		goto put_kvm;
3431
#endif
3432
	r = get_unused_fd_flags(O_CLOEXEC);
3433 3434 3435
	if (r < 0)
		goto put_kvm;

3436 3437 3438
	file = anon_inode_getfile("kvm-vm", &kvm_vm_fops, kvm, O_RDWR);
	if (IS_ERR(file)) {
		put_unused_fd(r);
3439 3440
		r = PTR_ERR(file);
		goto put_kvm;
3441
	}
3442

3443 3444 3445 3446 3447 3448
	/*
	 * Don't call kvm_put_kvm anymore at this point; file->f_op is
	 * already set, with ->release() being kvm_vm_release().  In error
	 * cases it will be called by the final fput(file) and will take
	 * care of doing kvm_put_kvm(kvm).
	 */
3449
	if (kvm_create_vm_debugfs(kvm, r) < 0) {
3450 3451
		put_unused_fd(r);
		fput(file);
3452 3453
		return -ENOMEM;
	}
3454
	kvm_uevent_notify_change(KVM_EVENT_CREATE_VM, kvm);
3455

3456
	fd_install(r, file);
3457
	return r;
3458 3459 3460 3461

put_kvm:
	kvm_put_kvm(kvm);
	return r;
3462 3463 3464 3465 3466
}

static long kvm_dev_ioctl(struct file *filp,
			  unsigned int ioctl, unsigned long arg)
{
3467
	long r = -EINVAL;
3468 3469 3470

	switch (ioctl) {
	case KVM_GET_API_VERSION:
3471 3472
		if (arg)
			goto out;
3473 3474 3475
		r = KVM_API_VERSION;
		break;
	case KVM_CREATE_VM:
3476
		r = kvm_dev_ioctl_create_vm(arg);
3477
		break;
3478
	case KVM_CHECK_EXTENSION:
3479
		r = kvm_vm_ioctl_check_extension_generic(NULL, arg);
3480
		break;
3481 3482 3483
	case KVM_GET_VCPU_MMAP_SIZE:
		if (arg)
			goto out;
3484 3485 3486
		r = PAGE_SIZE;     /* struct kvm_run */
#ifdef CONFIG_X86
		r += PAGE_SIZE;    /* pio data page */
3487
#endif
3488
#ifdef CONFIG_KVM_MMIO
3489
		r += PAGE_SIZE;    /* coalesced mmio ring page */
3490
#endif
3491
		break;
3492 3493 3494
	case KVM_TRACE_ENABLE:
	case KVM_TRACE_PAUSE:
	case KVM_TRACE_DISABLE:
3495
		r = -EOPNOTSUPP;
3496
		break;
A
Avi Kivity 已提交
3497
	default:
3498
		return kvm_arch_dev_ioctl(filp, ioctl, arg);
A
Avi Kivity 已提交
3499 3500 3501 3502 3503 3504 3505
	}
out:
	return r;
}

static struct file_operations kvm_chardev_ops = {
	.unlocked_ioctl = kvm_dev_ioctl,
3506
	.llseek		= noop_llseek,
3507
	KVM_COMPAT(kvm_dev_ioctl),
A
Avi Kivity 已提交
3508 3509 3510
};

static struct miscdevice kvm_dev = {
A
Avi Kivity 已提交
3511
	KVM_MINOR,
A
Avi Kivity 已提交
3512 3513 3514 3515
	"kvm",
	&kvm_chardev_ops,
};

3516
static void hardware_enable_nolock(void *junk)
3517 3518
{
	int cpu = raw_smp_processor_id();
3519
	int r;
3520

3521
	if (cpumask_test_cpu(cpu, cpus_hardware_enabled))
3522
		return;
3523

3524
	cpumask_set_cpu(cpu, cpus_hardware_enabled);
3525

3526
	r = kvm_arch_hardware_enable();
3527 3528 3529 3530

	if (r) {
		cpumask_clear_cpu(cpu, cpus_hardware_enabled);
		atomic_inc(&hardware_enable_failed);
X
Xiubo Li 已提交
3531
		pr_info("kvm: enabling virtualization on CPU%d failed\n", cpu);
3532
	}
3533 3534
}

3535
static int kvm_starting_cpu(unsigned int cpu)
3536
{
3537
	raw_spin_lock(&kvm_count_lock);
3538 3539
	if (kvm_usage_count)
		hardware_enable_nolock(NULL);
3540
	raw_spin_unlock(&kvm_count_lock);
3541
	return 0;
3542 3543 3544
}

static void hardware_disable_nolock(void *junk)
3545 3546 3547
{
	int cpu = raw_smp_processor_id();

3548
	if (!cpumask_test_cpu(cpu, cpus_hardware_enabled))
3549
		return;
3550
	cpumask_clear_cpu(cpu, cpus_hardware_enabled);
3551
	kvm_arch_hardware_disable();
3552 3553
}

3554
static int kvm_dying_cpu(unsigned int cpu)
3555
{
3556
	raw_spin_lock(&kvm_count_lock);
3557 3558
	if (kvm_usage_count)
		hardware_disable_nolock(NULL);
3559
	raw_spin_unlock(&kvm_count_lock);
3560
	return 0;
3561 3562
}

3563 3564 3565 3566 3567 3568
static void hardware_disable_all_nolock(void)
{
	BUG_ON(!kvm_usage_count);

	kvm_usage_count--;
	if (!kvm_usage_count)
3569
		on_each_cpu(hardware_disable_nolock, NULL, 1);
3570 3571 3572 3573
}

static void hardware_disable_all(void)
{
3574
	raw_spin_lock(&kvm_count_lock);
3575
	hardware_disable_all_nolock();
3576
	raw_spin_unlock(&kvm_count_lock);
3577 3578 3579 3580 3581 3582
}

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

3583
	raw_spin_lock(&kvm_count_lock);
3584 3585 3586 3587

	kvm_usage_count++;
	if (kvm_usage_count == 1) {
		atomic_set(&hardware_enable_failed, 0);
3588
		on_each_cpu(hardware_enable_nolock, NULL, 1);
3589 3590 3591 3592 3593 3594 3595

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

3596
	raw_spin_unlock(&kvm_count_lock);
3597 3598 3599 3600

	return r;
}

3601
static int kvm_reboot(struct notifier_block *notifier, unsigned long val,
M
Mike Day 已提交
3602
		      void *v)
3603
{
3604 3605 3606 3607 3608 3609
	/*
	 * 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
3610
	pr_info("kvm: exiting hardware virtualization\n");
3611
	kvm_rebooting = true;
3612
	on_each_cpu(hardware_disable_nolock, NULL, 1);
3613 3614 3615 3616 3617 3618 3619 3620
	return NOTIFY_OK;
}

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

M
Marcelo Tosatti 已提交
3621
static void kvm_io_bus_destroy(struct kvm_io_bus *bus)
3622 3623 3624 3625
{
	int i;

	for (i = 0; i < bus->dev_count; i++) {
3626
		struct kvm_io_device *pos = bus->range[i].dev;
3627 3628 3629

		kvm_iodevice_destructor(pos);
	}
M
Marcelo Tosatti 已提交
3630
	kfree(bus);
3631 3632
}

3633
static inline int kvm_io_bus_cmp(const struct kvm_io_range *r1,
X
Xiubo Li 已提交
3634
				 const struct kvm_io_range *r2)
3635
{
J
Jason Wang 已提交
3636 3637 3638 3639
	gpa_t addr1 = r1->addr;
	gpa_t addr2 = r2->addr;

	if (addr1 < addr2)
3640
		return -1;
J
Jason Wang 已提交
3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652

	/* 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)
3653
		return 1;
J
Jason Wang 已提交
3654

3655 3656 3657
	return 0;
}

3658 3659
static int kvm_io_bus_sort_cmp(const void *p1, const void *p2)
{
3660
	return kvm_io_bus_cmp(p1, p2);
3661 3662
}

G
Geoff Levand 已提交
3663
static int kvm_io_bus_get_first_dev(struct kvm_io_bus *bus,
3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680
			     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;

3681
	while (off > 0 && kvm_io_bus_cmp(&key, &bus->range[off-1]) == 0)
3682 3683 3684 3685 3686
		off--;

	return off;
}

3687
static int __kvm_io_bus_write(struct kvm_vcpu *vcpu, struct kvm_io_bus *bus,
C
Cornelia Huck 已提交
3688 3689 3690 3691 3692 3693 3694 3695 3696
			      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 &&
3697
		kvm_io_bus_cmp(range, &bus->range[idx]) == 0) {
3698
		if (!kvm_iodevice_write(vcpu, bus->range[idx].dev, range->addr,
C
Cornelia Huck 已提交
3699 3700 3701 3702 3703 3704 3705 3706
					range->len, val))
			return idx;
		idx++;
	}

	return -EOPNOTSUPP;
}

3707
/* kvm_io_bus_write - called under kvm->slots_lock */
3708
int kvm_io_bus_write(struct kvm_vcpu *vcpu, enum kvm_bus bus_idx, gpa_t addr,
3709
		     int len, const void *val)
3710
{
3711
	struct kvm_io_bus *bus;
3712
	struct kvm_io_range range;
C
Cornelia Huck 已提交
3713
	int r;
3714 3715 3716 3717 3718

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};
3719

3720
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
3721 3722
	if (!bus)
		return -ENOMEM;
3723
	r = __kvm_io_bus_write(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
3724 3725
	return r < 0 ? r : 0;
}
L
Leo Yan 已提交
3726
EXPORT_SYMBOL_GPL(kvm_io_bus_write);
C
Cornelia Huck 已提交
3727 3728

/* kvm_io_bus_write_cookie - called under kvm->slots_lock */
3729 3730
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
3731 3732 3733 3734 3735 3736 3737 3738 3739
{
	struct kvm_io_bus *bus;
	struct kvm_io_range range;

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

3740
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
3741 3742
	if (!bus)
		return -ENOMEM;
C
Cornelia Huck 已提交
3743 3744 3745

	/* First try the device referenced by cookie. */
	if ((cookie >= 0) && (cookie < bus->dev_count) &&
3746
	    (kvm_io_bus_cmp(&range, &bus->range[cookie]) == 0))
3747
		if (!kvm_iodevice_write(vcpu, bus->range[cookie].dev, addr, len,
C
Cornelia Huck 已提交
3748 3749 3750 3751 3752 3753 3754
					val))
			return cookie;

	/*
	 * cookie contained garbage; fall back to search and return the
	 * correct cookie value.
	 */
3755
	return __kvm_io_bus_write(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
3756 3757
}

3758 3759
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
3760 3761 3762 3763
{
	int idx;

	idx = kvm_io_bus_get_first_dev(bus, range->addr, range->len);
3764 3765 3766 3767
	if (idx < 0)
		return -EOPNOTSUPP;

	while (idx < bus->dev_count &&
3768
		kvm_io_bus_cmp(range, &bus->range[idx]) == 0) {
3769
		if (!kvm_iodevice_read(vcpu, bus->range[idx].dev, range->addr,
C
Cornelia Huck 已提交
3770 3771
				       range->len, val))
			return idx;
3772 3773 3774
		idx++;
	}

3775 3776
	return -EOPNOTSUPP;
}
3777

3778
/* kvm_io_bus_read - called under kvm->slots_lock */
3779
int kvm_io_bus_read(struct kvm_vcpu *vcpu, enum kvm_bus bus_idx, gpa_t addr,
M
Marcelo Tosatti 已提交
3780
		    int len, void *val)
3781
{
3782
	struct kvm_io_bus *bus;
3783
	struct kvm_io_range range;
C
Cornelia Huck 已提交
3784
	int r;
3785 3786 3787 3788 3789

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};
M
Marcelo Tosatti 已提交
3790

3791
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
3792 3793
	if (!bus)
		return -ENOMEM;
3794
	r = __kvm_io_bus_read(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
3795 3796
	return r < 0 ? r : 0;
}
3797

3798
/* Caller must hold slots_lock. */
3799 3800
int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
			    int len, struct kvm_io_device *dev)
3801
{
3802
	int i;
M
Marcelo Tosatti 已提交
3803
	struct kvm_io_bus *new_bus, *bus;
3804
	struct kvm_io_range range;
3805

3806
	bus = kvm_get_bus(kvm, bus_idx);
3807 3808 3809
	if (!bus)
		return -ENOMEM;

3810 3811
	/* exclude ioeventfd which is limited by maximum fd */
	if (bus->dev_count - bus->ioeventfd_count > NR_IOBUS_DEVS - 1)
3812
		return -ENOSPC;
3813

3814
	new_bus = kmalloc(struct_size(bus, range, bus->dev_count + 1),
3815
			  GFP_KERNEL_ACCOUNT);
M
Marcelo Tosatti 已提交
3816 3817
	if (!new_bus)
		return -ENOMEM;
3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833

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

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

	memcpy(new_bus, bus, sizeof(*bus) + i * sizeof(struct kvm_io_range));
	new_bus->dev_count++;
	new_bus->range[i] = range;
	memcpy(new_bus->range + i + 1, bus->range + i,
		(bus->dev_count - i) * sizeof(struct kvm_io_range));
M
Marcelo Tosatti 已提交
3834 3835 3836
	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
	synchronize_srcu_expedited(&kvm->srcu);
	kfree(bus);
3837 3838 3839 3840

	return 0;
}

3841
/* Caller must hold slots_lock. */
3842 3843
void kvm_io_bus_unregister_dev(struct kvm *kvm, enum kvm_bus bus_idx,
			       struct kvm_io_device *dev)
3844
{
3845
	int i;
M
Marcelo Tosatti 已提交
3846
	struct kvm_io_bus *new_bus, *bus;
3847

3848
	bus = kvm_get_bus(kvm, bus_idx);
3849
	if (!bus)
3850
		return;
3851

3852 3853
	for (i = 0; i < bus->dev_count; i++)
		if (bus->range[i].dev == dev) {
3854 3855
			break;
		}
M
Marcelo Tosatti 已提交
3856

3857 3858
	if (i == bus->dev_count)
		return;
3859

3860
	new_bus = kmalloc(struct_size(bus, range, bus->dev_count - 1),
3861
			  GFP_KERNEL_ACCOUNT);
3862 3863 3864 3865
	if (!new_bus)  {
		pr_err("kvm: failed to shrink bus, removing it completely\n");
		goto broken;
	}
3866 3867 3868 3869 3870

	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
3871

3872
broken:
M
Marcelo Tosatti 已提交
3873 3874 3875
	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
	synchronize_srcu_expedited(&kvm->srcu);
	kfree(bus);
3876
	return;
3877 3878
}

3879 3880 3881 3882 3883 3884 3885 3886 3887 3888
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);
3889 3890
	if (!bus)
		goto out_unlock;
3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904

	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);

3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916
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.
	 */
3917
	if (!refcount_inc_not_zero(&stat_data->kvm->users_count))
3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942
		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;

3943
	*val = *(ulong *)((void *)stat_data->kvm + stat_data->offset);
3944 3945 3946 3947

	return 0;
}

3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959
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;
}

3960 3961 3962 3963
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,
3964
				vm_stat_clear_per_vm, "%llu\n");
3965 3966 3967 3968 3969 3970 3971 3972
}

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,
3973
	.llseek  = no_llseek,
3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984
};

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)
3985
		*val += *(u64 *)((void *)vcpu + stat_data->offset);
3986 3987 3988 3989

	return 0;
}

3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004
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;
}

4005 4006 4007 4008
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,
4009
				 vcpu_stat_clear_per_vm, "%llu\n");
4010 4011 4012 4013 4014 4015 4016 4017
}

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,
4018
	.llseek  = no_llseek,
4019 4020 4021 4022 4023 4024 4025
};

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,
};

4026
static int vm_stat_get(void *_offset, u64 *val)
4027 4028 4029
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;
4030 4031
	struct kvm_stat_data stat_tmp = {.offset = offset};
	u64 tmp_val;
4032

4033
	*val = 0;
J
Junaid Shahid 已提交
4034
	mutex_lock(&kvm_lock);
4035 4036 4037 4038 4039
	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;
	}
J
Junaid Shahid 已提交
4040
	mutex_unlock(&kvm_lock);
4041
	return 0;
4042 4043
}

4044 4045 4046 4047 4048 4049 4050 4051 4052
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;

J
Junaid Shahid 已提交
4053
	mutex_lock(&kvm_lock);
4054 4055 4056 4057
	list_for_each_entry(kvm, &vm_list, vm_list) {
		stat_tmp.kvm = kvm;
		vm_stat_clear_per_vm((void *)&stat_tmp, 0);
	}
J
Junaid Shahid 已提交
4058
	mutex_unlock(&kvm_lock);
4059 4060 4061 4062 4063

	return 0;
}

DEFINE_SIMPLE_ATTRIBUTE(vm_stat_fops, vm_stat_get, vm_stat_clear, "%llu\n");
4064

4065
static int vcpu_stat_get(void *_offset, u64 *val)
A
Avi Kivity 已提交
4066 4067 4068
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;
4069 4070
	struct kvm_stat_data stat_tmp = {.offset = offset};
	u64 tmp_val;
A
Avi Kivity 已提交
4071

4072
	*val = 0;
J
Junaid Shahid 已提交
4073
	mutex_lock(&kvm_lock);
4074 4075 4076 4077 4078
	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;
	}
J
Junaid Shahid 已提交
4079
	mutex_unlock(&kvm_lock);
4080
	return 0;
A
Avi Kivity 已提交
4081 4082
}

4083 4084 4085 4086 4087 4088 4089 4090 4091
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;

J
Junaid Shahid 已提交
4092
	mutex_lock(&kvm_lock);
4093 4094 4095 4096
	list_for_each_entry(kvm, &vm_list, vm_list) {
		stat_tmp.kvm = kvm;
		vcpu_stat_clear_per_vm((void *)&stat_tmp, 0);
	}
J
Junaid Shahid 已提交
4097
	mutex_unlock(&kvm_lock);
4098 4099 4100 4101 4102 4103

	return 0;
}

DEFINE_SIMPLE_ATTRIBUTE(vcpu_stat_fops, vcpu_stat_get, vcpu_stat_clear,
			"%llu\n");
4104

4105
static const struct file_operations *stat_fops[] = {
4106 4107 4108
	[KVM_STAT_VCPU] = &vcpu_stat_fops,
	[KVM_STAT_VM]   = &vm_stat_fops,
};
A
Avi Kivity 已提交
4109

4110 4111 4112 4113 4114 4115 4116 4117
static void kvm_uevent_notify_change(unsigned int type, struct kvm *kvm)
{
	struct kobj_uevent_env *env;
	unsigned long long created, active;

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

J
Junaid Shahid 已提交
4118
	mutex_lock(&kvm_lock);
4119 4120 4121 4122 4123 4124 4125 4126
	if (type == KVM_EVENT_CREATE_VM) {
		kvm_createvm_count++;
		kvm_active_vms++;
	} else if (type == KVM_EVENT_DESTROY_VM) {
		kvm_active_vms--;
	}
	created = kvm_createvm_count;
	active = kvm_active_vms;
J
Junaid Shahid 已提交
4127
	mutex_unlock(&kvm_lock);
4128

4129
	env = kzalloc(sizeof(*env), GFP_KERNEL_ACCOUNT);
4130 4131 4132 4133 4134 4135
	if (!env)
		return;

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

4136
	if (type == KVM_EVENT_CREATE_VM) {
4137
		add_uevent_var(env, "EVENT=create");
4138 4139
		kvm->userspace_pid = task_pid_nr(current);
	} else if (type == KVM_EVENT_DESTROY_VM) {
4140
		add_uevent_var(env, "EVENT=destroy");
4141 4142
	}
	add_uevent_var(env, "PID=%d", kvm->userspace_pid);
4143

4144
	if (!IS_ERR_OR_NULL(kvm->debugfs_dentry)) {
4145
		char *tmp, *p = kmalloc(PATH_MAX, GFP_KERNEL_ACCOUNT);
4146 4147 4148 4149 4150 4151

		if (p) {
			tmp = dentry_path_raw(kvm->debugfs_dentry, p, PATH_MAX);
			if (!IS_ERR(tmp))
				add_uevent_var(env, "STATS_PATH=%s", tmp);
			kfree(p);
4152 4153 4154 4155 4156 4157 4158 4159
		}
	}
	/* no need for checks, since we are adding at most only 5 keys */
	env->envp[env->envp_idx++] = NULL;
	kobject_uevent_env(&kvm_dev.this_device->kobj, KOBJ_CHANGE, env->envp);
	kfree(env);
}

4160
static void kvm_init_debug(void)
A
Avi Kivity 已提交
4161 4162 4163
{
	struct kvm_stats_debugfs_item *p;

4164
	kvm_debugfs_dir = debugfs_create_dir("kvm", NULL);
4165

4166 4167
	kvm_debugfs_num_entries = 0;
	for (p = debugfs_entries; p->name; ++p, kvm_debugfs_num_entries++) {
4168 4169 4170
		debugfs_create_file(p->name, 0644, kvm_debugfs_dir,
				    (void *)(long)p->offset,
				    stat_fops[p->kind]);
4171
	}
A
Avi Kivity 已提交
4172 4173
}

4174
static int kvm_suspend(void)
4175
{
4176
	if (kvm_usage_count)
4177
		hardware_disable_nolock(NULL);
4178 4179 4180
	return 0;
}

4181
static void kvm_resume(void)
4182
{
4183
	if (kvm_usage_count) {
4184 4185 4186
#ifdef CONFIG_LOCKDEP
		WARN_ON(lockdep_is_held(&kvm_count_lock));
#endif
4187
		hardware_enable_nolock(NULL);
4188
	}
4189 4190
}

4191
static struct syscore_ops kvm_syscore_ops = {
4192 4193 4194 4195
	.suspend = kvm_suspend,
	.resume = kvm_resume,
};

4196 4197 4198 4199 4200 4201 4202 4203 4204
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);
4205

4206 4207
	if (vcpu->preempted)
		vcpu->preempted = false;
4208

R
Radim Krčmář 已提交
4209 4210
	kvm_arch_sched_in(vcpu, cpu);

4211
	kvm_arch_vcpu_load(vcpu, cpu);
4212 4213 4214 4215 4216 4217 4218
}

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

4219 4220
	if (current->state == TASK_RUNNING)
		vcpu->preempted = true;
4221
	kvm_arch_vcpu_put(vcpu);
4222 4223
}

4224 4225 4226 4227 4228
static void check_processor_compat(void *rtn)
{
	*(int *)rtn = kvm_arch_check_processor_compat();
}

4229
int kvm_init(void *opaque, unsigned vcpu_size, unsigned vcpu_align,
4230
		  struct module *module)
A
Avi Kivity 已提交
4231 4232
{
	int r;
Y
Yang, Sheng 已提交
4233
	int cpu;
A
Avi Kivity 已提交
4234

4235 4236
	r = kvm_arch_init(opaque);
	if (r)
4237
		goto out_fail;
4238

4239 4240 4241 4242
	/*
	 * 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
4243 4244
	 * kvm_arch_init must be called before kvm_irqfd_init to avoid creating
	 * conflicts in case kvm is already setup for another implementation.
4245
	 */
P
Paolo Bonzini 已提交
4246 4247 4248
	r = kvm_irqfd_init();
	if (r)
		goto out_irqfd;
4249

4250
	if (!zalloc_cpumask_var(&cpus_hardware_enabled, GFP_KERNEL)) {
4251 4252 4253 4254
		r = -ENOMEM;
		goto out_free_0;
	}

4255
	r = kvm_arch_hardware_setup();
A
Avi Kivity 已提交
4256
	if (r < 0)
4257
		goto out_free_0a;
A
Avi Kivity 已提交
4258

Y
Yang, Sheng 已提交
4259
	for_each_online_cpu(cpu) {
4260
		smp_call_function_single(cpu, check_processor_compat, &r, 1);
Y
Yang, Sheng 已提交
4261
		if (r < 0)
4262
			goto out_free_1;
Y
Yang, Sheng 已提交
4263 4264
	}

T
Thomas Gleixner 已提交
4265
	r = cpuhp_setup_state_nocalls(CPUHP_AP_KVM_STARTING, "kvm/cpu:starting",
4266
				      kvm_starting_cpu, kvm_dying_cpu);
A
Avi Kivity 已提交
4267
	if (r)
4268
		goto out_free_2;
A
Avi Kivity 已提交
4269 4270
	register_reboot_notifier(&kvm_reboot_notifier);

4271
	/* A kmem cache lets us meet the alignment requirements of fx_save. */
4272 4273
	if (!vcpu_align)
		vcpu_align = __alignof__(struct kvm_vcpu);
4274 4275 4276 4277 4278 4279
	kvm_vcpu_cache =
		kmem_cache_create_usercopy("kvm_vcpu", vcpu_size, vcpu_align,
					   SLAB_ACCOUNT,
					   offsetof(struct kvm_vcpu, arch),
					   sizeof_field(struct kvm_vcpu, arch),
					   NULL);
4280 4281
	if (!kvm_vcpu_cache) {
		r = -ENOMEM;
4282
		goto out_free_3;
4283 4284
	}

4285 4286 4287 4288
	r = kvm_async_pf_init();
	if (r)
		goto out_free;

A
Avi Kivity 已提交
4289
	kvm_chardev_ops.owner = module;
4290 4291
	kvm_vm_fops.owner = module;
	kvm_vcpu_fops.owner = module;
A
Avi Kivity 已提交
4292 4293 4294

	r = misc_register(&kvm_dev);
	if (r) {
X
Xiubo Li 已提交
4295
		pr_err("kvm: misc device register failed\n");
4296
		goto out_unreg;
A
Avi Kivity 已提交
4297 4298
	}

4299 4300
	register_syscore_ops(&kvm_syscore_ops);

4301 4302 4303
	kvm_preempt_ops.sched_in = kvm_sched_in;
	kvm_preempt_ops.sched_out = kvm_sched_out;

4304
	kvm_init_debug();
4305

P
Paolo Bonzini 已提交
4306 4307 4308
	r = kvm_vfio_ops_init();
	WARN_ON(r);

4309
	return 0;
A
Avi Kivity 已提交
4310

4311 4312
out_unreg:
	kvm_async_pf_deinit();
A
Avi Kivity 已提交
4313
out_free:
4314
	kmem_cache_destroy(kvm_vcpu_cache);
4315
out_free_3:
A
Avi Kivity 已提交
4316
	unregister_reboot_notifier(&kvm_reboot_notifier);
4317
	cpuhp_remove_state_nocalls(CPUHP_AP_KVM_STARTING);
4318 4319
out_free_2:
out_free_1:
4320
	kvm_arch_hardware_unsetup();
4321 4322
out_free_0a:
	free_cpumask_var(cpus_hardware_enabled);
4323
out_free_0:
4324
	kvm_irqfd_exit();
P
Paolo Bonzini 已提交
4325
out_irqfd:
4326 4327
	kvm_arch_exit();
out_fail:
A
Avi Kivity 已提交
4328 4329
	return r;
}
4330
EXPORT_SYMBOL_GPL(kvm_init);
A
Avi Kivity 已提交
4331

4332
void kvm_exit(void)
A
Avi Kivity 已提交
4333
{
4334
	debugfs_remove_recursive(kvm_debugfs_dir);
A
Avi Kivity 已提交
4335
	misc_deregister(&kvm_dev);
4336
	kmem_cache_destroy(kvm_vcpu_cache);
4337
	kvm_async_pf_deinit();
4338
	unregister_syscore_ops(&kvm_syscore_ops);
A
Avi Kivity 已提交
4339
	unregister_reboot_notifier(&kvm_reboot_notifier);
4340
	cpuhp_remove_state_nocalls(CPUHP_AP_KVM_STARTING);
4341
	on_each_cpu(hardware_disable_nolock, NULL, 1);
4342
	kvm_arch_hardware_unsetup();
4343
	kvm_arch_exit();
4344
	kvm_irqfd_exit();
4345
	free_cpumask_var(cpus_hardware_enabled);
4346
	kvm_vfio_ops_exit();
A
Avi Kivity 已提交
4347
}
4348
EXPORT_SYMBOL_GPL(kvm_exit);