kvm_main.c 122.1 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>
53
#include <linux/kthread.h>
A
Avi Kivity 已提交
54

A
Avi Kivity 已提交
55
#include <asm/processor.h>
56
#include <asm/ioctl.h>
57
#include <linux/uaccess.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
static struct kmem_cache *kvm_vcpu_cache;
A
Avi Kivity 已提交
107

108
static __read_mostly struct preempt_ops kvm_preempt_ops;
109
static DEFINE_PER_CPU(struct kvm_vcpu *, kvm_running_vcpu);
110

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

114
static int kvm_debugfs_num_entries;
115
static const struct file_operations stat_fops_per_vm;
116

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
#define KVM_COMPAT(c)	.compat_ioctl	= (c)
#else
124 125 126 127 128 129 130
/*
 * For architectures that don't implement a compat infrastructure,
 * adopt a double line of defense:
 * - Prevent a compat task from opening /dev/kvm
 * - If the open has been done by a 64bit task, and the KVM fd
 *   passed to a compat task, let the ioctls fail.
 */
131 132
static long kvm_no_compat_ioctl(struct file *file, unsigned int ioctl,
				unsigned long arg) { return -EINVAL; }
133 134 135 136 137 138 139

static int kvm_no_compat_open(struct inode *inode, struct file *file)
{
	return is_compat_task() ? -ENODEV : 0;
}
#define KVM_COMPAT(c)	.compat_ioctl	= kvm_no_compat_ioctl,	\
			.open		= kvm_no_compat_open
140
#endif
141 142
static int hardware_enable_all(void);
static void hardware_disable_all(void);
A
Avi Kivity 已提交
143

M
Marcelo Tosatti 已提交
144
static void kvm_io_bus_destroy(struct kvm_io_bus *bus);
145

146
__visible bool kvm_rebooting;
147
EXPORT_SYMBOL_GPL(kvm_rebooting);
148

149 150 151 152 153 154
#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;

155 156 157 158 159
/* debugfs entries of Detail For vcpu stat EXtension */
__weak struct dfx_kvm_stats_debugfs_item dfx_debugfs_entries[] = {
	{ NULL }
};

160 161
__weak void kvm_arch_mmu_notifier_invalidate_range(struct kvm *kvm,
						   unsigned long start, unsigned long end)
162 163 164
{
}

165 166 167 168 169 170 171 172 173 174 175 176 177 178
bool kvm_is_zone_device_pfn(kvm_pfn_t pfn)
{
	/*
	 * The metadata used by is_zone_device_page() to determine whether or
	 * not a page is ZONE_DEVICE is guaranteed to be valid if and only if
	 * the device has been pinned, e.g. by get_user_pages().  WARN if the
	 * page_count() is zero to help detect bad usage of this helper.
	 */
	if (!pfn_valid(pfn) || WARN_ON_ONCE(!page_count(pfn_to_page(pfn))))
		return false;

	return is_zone_device_page(pfn_to_page(pfn));
}

D
Dan Williams 已提交
179
bool kvm_is_reserved_pfn(kvm_pfn_t pfn)
B
Ben-Ami Yassour 已提交
180
{
181 182 183 184 185
	/*
	 * ZONE_DEVICE pages currently set PG_reserved, but from a refcounting
	 * perspective they are "normal" pages, albeit with slightly different
	 * usage rules.
	 */
186
	if (pfn_valid(pfn))
187
		return PageReserved(pfn_to_page(pfn)) &&
188
		       !is_zero_pfn(pfn) &&
189
		       !kvm_is_zone_device_pfn(pfn);
B
Ben-Ami Yassour 已提交
190 191 192 193

	return true;
}

194 195 196 197 198 199 200 201 202 203
bool kvm_is_transparent_hugepage(kvm_pfn_t pfn)
{
	struct page *page = pfn_to_page(pfn);

	if (!PageTransCompoundMap(page))
		return false;

	return is_transparent_hugepage(compound_head(page));
}

A
Avi Kivity 已提交
204 205 206
/*
 * Switches to specified vcpu, until a matching vcpu_put()
 */
207
void vcpu_load(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
208
{
209
	int cpu = get_cpu();
210 211

	__this_cpu_write(kvm_running_vcpu, vcpu);
212
	preempt_notifier_register(&vcpu->preempt_notifier);
213
	kvm_arch_vcpu_load(vcpu, cpu);
214
	put_cpu();
A
Avi Kivity 已提交
215
}
216
EXPORT_SYMBOL_GPL(vcpu_load);
A
Avi Kivity 已提交
217

218
void vcpu_put(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
219
{
220
	preempt_disable();
221
	kvm_arch_vcpu_put(vcpu);
222
	preempt_notifier_unregister(&vcpu->preempt_notifier);
223
	__this_cpu_write(kvm_running_vcpu, NULL);
224
	preempt_enable();
A
Avi Kivity 已提交
225
}
226
EXPORT_SYMBOL_GPL(vcpu_put);
A
Avi Kivity 已提交
227

228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245
/* 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;
}

246 247 248 249
static void ack_flush(void *_completed)
{
}

250 251 252 253 254 255 256 257 258 259 260 261
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;
}

262
bool kvm_make_vcpus_request_mask(struct kvm *kvm, unsigned int req,
263
				 struct kvm_vcpu *except,
264
				 unsigned long *vcpu_bitmap, cpumask_var_t tmp)
265
{
266
	int i, cpu, me;
267
	struct kvm_vcpu *vcpu;
268
	bool called;
269

270
	me = get_cpu();
271

272
	kvm_for_each_vcpu(i, vcpu, kvm) {
273 274
		if ((vcpu_bitmap && !test_bit(i, vcpu_bitmap)) ||
		    vcpu == except)
275 276
			continue;

277
		kvm_make_request(req, vcpu);
278
		cpu = vcpu->cpu;
279

280 281
		if (!(req & KVM_REQUEST_NO_WAKEUP) && kvm_vcpu_wake_up(vcpu))
			continue;
282

283
		if (tmp != NULL && cpu != -1 && cpu != me &&
284
		    kvm_request_needs_ipi(vcpu, req))
285
			__cpumask_set_cpu(cpu, tmp);
286
	}
287 288

	called = kvm_kick_many_cpus(tmp, !!(req & KVM_REQUEST_WAIT));
289
	put_cpu();
290 291 292 293

	return called;
}

294 295
bool kvm_make_all_cpus_request_except(struct kvm *kvm, unsigned int req,
				      struct kvm_vcpu *except)
296 297 298 299 300 301
{
	cpumask_var_t cpus;
	bool called;

	zalloc_cpumask_var(&cpus, GFP_ATOMIC);

302
	called = kvm_make_vcpus_request_mask(kvm, req, except, NULL, cpus);
303

304
	free_cpumask_var(cpus);
305
	return called;
306 307
}

308 309 310 311 312
bool kvm_make_all_cpus_request(struct kvm *kvm, unsigned int req)
{
	return kvm_make_all_cpus_request_except(kvm, req, NULL);
}

313
#ifndef CONFIG_HAVE_KVM_ARCH_TLB_FLUSH_ALL
314
void kvm_flush_remote_tlbs(struct kvm *kvm)
315
{
316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332
	/*
	 * 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.
	 */
333 334
	if (!kvm_arch_flush_remote_tlb(kvm)
	    || kvm_make_all_cpus_request(kvm, KVM_REQ_TLB_FLUSH))
335
		++kvm->stat.remote_tlb_flush;
336
	cmpxchg(&kvm->tlbs_dirty, dirty_count, 0);
337
}
338
EXPORT_SYMBOL_GPL(kvm_flush_remote_tlbs);
339
#endif
340

341 342
void kvm_reload_remote_mmus(struct kvm *kvm)
{
343
	kvm_make_all_cpus_request(kvm, KVM_REQ_MMU_RELOAD);
344
}
345

346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400
#ifdef KVM_ARCH_NR_OBJS_PER_MEMORY_CACHE
static inline void *mmu_memory_cache_alloc_obj(struct kvm_mmu_memory_cache *mc,
					       gfp_t gfp_flags)
{
	gfp_flags |= mc->gfp_zero;

	if (mc->kmem_cache)
		return kmem_cache_alloc(mc->kmem_cache, gfp_flags);
	else
		return (void *)__get_free_page(gfp_flags);
}

int kvm_mmu_topup_memory_cache(struct kvm_mmu_memory_cache *mc, int min)
{
	void *obj;

	if (mc->nobjs >= min)
		return 0;
	while (mc->nobjs < ARRAY_SIZE(mc->objects)) {
		obj = mmu_memory_cache_alloc_obj(mc, GFP_KERNEL_ACCOUNT);
		if (!obj)
			return mc->nobjs >= min ? 0 : -ENOMEM;
		mc->objects[mc->nobjs++] = obj;
	}
	return 0;
}

int kvm_mmu_memory_cache_nr_free_objects(struct kvm_mmu_memory_cache *mc)
{
	return mc->nobjs;
}

void kvm_mmu_free_memory_cache(struct kvm_mmu_memory_cache *mc)
{
	while (mc->nobjs) {
		if (mc->kmem_cache)
			kmem_cache_free(mc->kmem_cache, mc->objects[--mc->nobjs]);
		else
			free_page((unsigned long)mc->objects[--mc->nobjs]);
	}
}

void *kvm_mmu_memory_cache_alloc(struct kvm_mmu_memory_cache *mc)
{
	void *p;

	if (WARN_ON(!mc->nobjs))
		p = mmu_memory_cache_alloc_obj(mc, GFP_ATOMIC | __GFP_ACCOUNT);
	else
		p = mc->objects[--mc->nobjs];
	BUG_ON(!p);
	return p;
}
#endif

401
static void kvm_vcpu_init(struct kvm_vcpu *vcpu, struct kvm *kvm, unsigned id)
R
Rusty Russell 已提交
402 403 404 405 406
{
	mutex_init(&vcpu->mutex);
	vcpu->cpu = -1;
	vcpu->kvm = kvm;
	vcpu->vcpu_id = id;
407
	vcpu->pid = NULL;
408
	rcuwait_init(&vcpu->wait);
409
	kvm_async_pf_vcpu_init(vcpu);
R
Rusty Russell 已提交
410

411 412 413
	vcpu->pre_pcpu = -1;
	INIT_LIST_HEAD(&vcpu->blocked_vcpu_list);

414 415
	kvm_vcpu_set_in_spin_loop(vcpu, false);
	kvm_vcpu_set_dy_eligible(vcpu, false);
416
	vcpu->preempted = false;
417
	vcpu->ready = false;
418
	preempt_notifier_init(&vcpu->preempt_notifier, &kvm_preempt_ops);
R
Rusty Russell 已提交
419 420
}

421 422 423
void kvm_vcpu_destroy(struct kvm_vcpu *vcpu)
{
	kvm_arch_vcpu_destroy(vcpu);
424

425 426 427 428 429 430 431
	/*
	 * No need for rcu_read_lock as VCPU_RUN is the only place that changes
	 * the vcpu->pid pointer, and at destruction time all file descriptors
	 * are already gone.
	 */
	put_pid(rcu_dereference_protected(vcpu->pid, 1));

432
	free_page((unsigned long)vcpu->run);
433
	kmem_cache_free(kvm_vcpu_cache, vcpu);
434 435 436
}
EXPORT_SYMBOL_GPL(kvm_vcpu_destroy);

437 438 439 440 441 442
#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);
}

443 444 445 446 447 448 449 450 451 452 453 454
static void kvm_mmu_notifier_invalidate_range(struct mmu_notifier *mn,
					      struct mm_struct *mm,
					      unsigned long start, unsigned long end)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
	int idx;

	idx = srcu_read_lock(&kvm->srcu);
	kvm_arch_mmu_notifier_invalidate_range(kvm, start, end);
	srcu_read_unlock(&kvm->srcu, idx);
}

455 456 457 458 459 460
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);
461
	int idx;
462

463
	idx = srcu_read_lock(&kvm->srcu);
464 465
	spin_lock(&kvm->mmu_lock);
	kvm->mmu_notifier_seq++;
466 467 468 469

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

470
	spin_unlock(&kvm->mmu_lock);
471
	srcu_read_unlock(&kvm->srcu, idx);
472 473
}

474
static int kvm_mmu_notifier_invalidate_range_start(struct mmu_notifier *mn,
475
					const struct mmu_notifier_range *range)
476 477
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
478
	int need_tlb_flush = 0, idx;
479

480
	idx = srcu_read_lock(&kvm->srcu);
481 482 483 484 485 486 487
	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++;
488 489
	need_tlb_flush = kvm_unmap_hva_range(kvm, range->start, range->end,
					     range->flags);
490
	need_tlb_flush |= kvm->tlbs_dirty;
491 492 493
	/* we've to flush the tlb before the pages can be freed */
	if (need_tlb_flush)
		kvm_flush_remote_tlbs(kvm);
494 495 496

	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);
497

498
	return 0;
499 500 501
}

static void kvm_mmu_notifier_invalidate_range_end(struct mmu_notifier *mn,
502
					const struct mmu_notifier_range *range)
503 504 505 506 507 508 509 510 511 512
{
	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++;
513
	smp_wmb();
514 515
	/*
	 * The above sequence increase must be visible before the
516 517
	 * below count decrease, which is ensured by the smp_wmb above
	 * in conjunction with the smp_rmb in mmu_notifier_retry().
518 519 520 521 522 523 524 525 526
	 */
	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
527 528
					      unsigned long start,
					      unsigned long end)
529 530
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
531
	int young, idx;
532

533
	idx = srcu_read_lock(&kvm->srcu);
534 535
	spin_lock(&kvm->mmu_lock);

A
Andres Lagar-Cavilla 已提交
536
	young = kvm_age_hva(kvm, start, end);
537 538 539
	if (young)
		kvm_flush_remote_tlbs(kvm);

540 541 542
	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);

543 544 545
	return young;
}

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
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
576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591
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;
}

592 593 594 595
static void kvm_mmu_notifier_release(struct mmu_notifier *mn,
				     struct mm_struct *mm)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
596 597 598
	int idx;

	idx = srcu_read_lock(&kvm->srcu);
599
	kvm_arch_flush_shadow_all(kvm);
600
	srcu_read_unlock(&kvm->srcu, idx);
601 602
}

603
static const struct mmu_notifier_ops kvm_mmu_notifier_ops = {
604
	.invalidate_range	= kvm_mmu_notifier_invalidate_range,
605 606 607
	.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,
608
	.clear_young		= kvm_mmu_notifier_clear_young,
A
Andrea Arcangeli 已提交
609
	.test_young		= kvm_mmu_notifier_test_young,
610
	.change_pte		= kvm_mmu_notifier_change_pte,
611
	.release		= kvm_mmu_notifier_release,
612
};
613 614 615 616 617 618 619 620 621 622 623 624 625 626

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;
}

627 628
#endif /* CONFIG_MMU_NOTIFIER && KVM_ARCH_WANT_MMU_NOTIFIER */

629
static struct kvm_memslots *kvm_alloc_memslots(void)
630 631
{
	int i;
632
	struct kvm_memslots *slots;
633

634
	slots = kvzalloc(sizeof(struct kvm_memslots), GFP_KERNEL_ACCOUNT);
635 636 637
	if (!slots)
		return NULL;

638
	for (i = 0; i < KVM_MEM_SLOTS_NUM; i++)
639
		slots->id_to_index[i] = -1;
640 641 642 643 644 645 646 647 648 649 650 651 652

	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;
}

653
static void kvm_free_memslot(struct kvm *kvm, struct kvm_memory_slot *slot)
654
{
655
	kvm_destroy_dirty_bitmap(slot);
656

657
	kvm_arch_free_memslot(kvm, slot);
658

659 660
	slot->flags = 0;
	slot->npages = 0;
661 662 663 664 665 666 667 668 669 670
}

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)
671
		kvm_free_memslot(kvm, memslot);
672 673

	kvfree(slots);
674 675
}

676 677 678 679 680 681 682 683 684
static void kvm_destroy_vm_debugfs(struct kvm *kvm)
{
	int i;

	if (!kvm->debugfs_dentry)
		return;

	debugfs_remove_recursive(kvm->debugfs_dentry);

685 686 687 688 689
	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);
	}
690 691 692 693 694 695 696 697 698 699 700 701
}

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);
702
	kvm->debugfs_dentry = debugfs_create_dir(dir_name, kvm_debugfs_dir);
703 704 705

	kvm->debugfs_stat_data = kcalloc(kvm_debugfs_num_entries,
					 sizeof(*kvm->debugfs_stat_data),
706
					 GFP_KERNEL_ACCOUNT);
707 708 709 710
	if (!kvm->debugfs_stat_data)
		return -ENOMEM;

	for (p = debugfs_entries; p->name; p++) {
711
		stat_data = kzalloc(sizeof(*stat_data), GFP_KERNEL_ACCOUNT);
712 713 714 715
		if (!stat_data)
			return -ENOMEM;

		stat_data->kvm = kvm;
716
		stat_data->dbgfs_item = p;
717
		kvm->debugfs_stat_data[p - debugfs_entries] = stat_data;
718 719 720
		debugfs_create_file(p->name, KVM_DBGFS_GET_MODE(p),
				    kvm->debugfs_dentry, stat_data,
				    &stat_fops_per_vm);
721 722 723 724
	}
	return 0;
}

725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741
/*
 * Called after the VM is otherwise initialized, but just before adding it to
 * the vm_list.
 */
int __weak kvm_arch_post_init_vm(struct kvm *kvm)
{
	return 0;
}

/*
 * Called just after removing the VM from the vm_list, but before doing any
 * other destruction.
 */
void __weak kvm_arch_pre_destroy_vm(struct kvm *kvm)
{
}

742
static struct kvm *kvm_create_vm(unsigned long type)
A
Avi Kivity 已提交
743
{
744
	struct kvm *kvm = kvm_arch_alloc_vm();
745 746
	int r = -ENOMEM;
	int i;
A
Avi Kivity 已提交
747

748 749 750
	if (!kvm)
		return ERR_PTR(-ENOMEM);

751
	spin_lock_init(&kvm->mmu_lock);
V
Vegard Nossum 已提交
752
	mmgrab(current->mm);
753 754 755 756 757 758 759
	kvm->mm = current->mm;
	kvm_eventfd_init(kvm);
	mutex_init(&kvm->lock);
	mutex_init(&kvm->irq_lock);
	mutex_init(&kvm->slots_lock);
	INIT_LIST_HEAD(&kvm->devices);

760 761
	BUILD_BUG_ON(KVM_MEM_SLOTS_NUM > SHRT_MAX);

762 763 764 765 766
	if (init_srcu_struct(&kvm->srcu))
		goto out_err_no_srcu;
	if (init_srcu_struct(&kvm->irq_srcu))
		goto out_err_no_irq_srcu;

767
	refcount_set(&kvm->users_count, 1);
768
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++) {
769
		struct kvm_memslots *slots = kvm_alloc_memslots();
770

771
		if (!slots)
772
			goto out_err_no_arch_destroy_vm;
773
		/* Generations must be different for each address space. */
774
		slots->generation = i;
775
		rcu_assign_pointer(kvm->memslots[i], slots);
776
	}
777

M
Marcelo Tosatti 已提交
778
	for (i = 0; i < KVM_NR_BUSES; i++) {
779
		rcu_assign_pointer(kvm->buses[i],
780
			kzalloc(sizeof(struct kvm_io_bus), GFP_KERNEL_ACCOUNT));
781
		if (!kvm->buses[i])
782
			goto out_err_no_arch_destroy_vm;
M
Marcelo Tosatti 已提交
783
	}
784

785 786
	kvm->max_halt_poll_ns = halt_poll_ns;

787
	r = kvm_arch_init_vm(kvm, type);
788
	if (r)
789
		goto out_err_no_arch_destroy_vm;
790 791 792

	r = hardware_enable_all();
	if (r)
793
		goto out_err_no_disable;
794

795
#ifdef CONFIG_HAVE_KVM_IRQFD
796
	INIT_HLIST_HEAD(&kvm->irq_ack_notifier_list);
797
#endif
A
Avi Kivity 已提交
798

799
	r = kvm_init_mmu_notifier(kvm);
800 801 802 803
	if (r)
		goto out_err_no_mmu_notifier;

	r = kvm_arch_post_init_vm(kvm);
804 805 806
	if (r)
		goto out_err;

J
Junaid Shahid 已提交
807
	mutex_lock(&kvm_lock);
808
	list_add(&kvm->vm_list, &vm_list);
J
Junaid Shahid 已提交
809
	mutex_unlock(&kvm_lock);
810

811 812
	preempt_notifier_inc();

813
	return kvm;
814 815

out_err:
816 817 818 819 820
#if defined(CONFIG_MMU_NOTIFIER) && defined(KVM_ARCH_WANT_MMU_NOTIFIER)
	if (kvm->mmu_notifier.ops)
		mmu_notifier_unregister(&kvm->mmu_notifier, current->mm);
#endif
out_err_no_mmu_notifier:
821
	hardware_disable_all();
822
out_err_no_disable:
823 824
	kvm_arch_destroy_vm(kvm);
out_err_no_arch_destroy_vm:
825
	WARN_ON_ONCE(!refcount_dec_and_test(&kvm->users_count));
M
Marcelo Tosatti 已提交
826
	for (i = 0; i < KVM_NR_BUSES; i++)
827
		kfree(kvm_get_bus(kvm, i));
828
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++)
829
		kvm_free_memslots(kvm, __kvm_memslots(kvm, i));
830 831 832 833
	cleanup_srcu_struct(&kvm->irq_srcu);
out_err_no_irq_srcu:
	cleanup_srcu_struct(&kvm->srcu);
out_err_no_srcu:
834
	kvm_arch_free_vm(kvm);
835
	mmdrop(current->mm);
836
	return ERR_PTR(r);
837 838
}

839 840
static void kvm_destroy_devices(struct kvm *kvm)
{
G
Geliang Tang 已提交
841
	struct kvm_device *dev, *tmp;
842

843 844 845 846 847
	/*
	 * 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
848 849
	list_for_each_entry_safe(dev, tmp, &kvm->devices, vm_node) {
		list_del(&dev->vm_node);
850 851 852 853
		dev->ops->destroy(dev);
	}
}

854 855
static void kvm_destroy_vm(struct kvm *kvm)
{
M
Marcelo Tosatti 已提交
856
	int i;
857 858
	struct mm_struct *mm = kvm->mm;

859
	kvm_uevent_notify_change(KVM_EVENT_DESTROY_VM, kvm);
860
	kvm_destroy_vm_debugfs(kvm);
861
	kvm_arch_sync_events(kvm);
J
Junaid Shahid 已提交
862
	mutex_lock(&kvm_lock);
863
	list_del(&kvm->vm_list);
J
Junaid Shahid 已提交
864
	mutex_unlock(&kvm_lock);
865 866
	kvm_arch_pre_destroy_vm(kvm);

867
	kvm_free_irq_routing(kvm);
868
	for (i = 0; i < KVM_NR_BUSES; i++) {
869
		struct kvm_io_bus *bus = kvm_get_bus(kvm, i);
870 871 872

		if (bus)
			kvm_io_bus_destroy(bus);
873 874
		kvm->buses[i] = NULL;
	}
875
	kvm_coalesced_mmio_free(kvm);
876 877
#if defined(CONFIG_MMU_NOTIFIER) && defined(KVM_ARCH_WANT_MMU_NOTIFIER)
	mmu_notifier_unregister(&kvm->mmu_notifier, kvm->mm);
878
#else
879
	kvm_arch_flush_shadow_all(kvm);
880
#endif
881
	kvm_arch_destroy_vm(kvm);
882
	kvm_destroy_devices(kvm);
883
	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++)
884
		kvm_free_memslots(kvm, __kvm_memslots(kvm, i));
885
	cleanup_srcu_struct(&kvm->irq_srcu);
886 887
	cleanup_srcu_struct(&kvm->srcu);
	kvm_arch_free_vm(kvm);
888
	preempt_notifier_dec();
889
	hardware_disable_all();
890
	mmdrop(mm);
891 892
}

I
Izik Eidus 已提交
893 894
void kvm_get_kvm(struct kvm *kvm)
{
895
	refcount_inc(&kvm->users_count);
I
Izik Eidus 已提交
896 897 898 899 900
}
EXPORT_SYMBOL_GPL(kvm_get_kvm);

void kvm_put_kvm(struct kvm *kvm)
{
901
	if (refcount_dec_and_test(&kvm->users_count))
I
Izik Eidus 已提交
902 903 904 905
		kvm_destroy_vm(kvm);
}
EXPORT_SYMBOL_GPL(kvm_put_kvm);

906 907 908 909 910 911 912 913 914 915 916 917
/*
 * Used to put a reference that was taken on behalf of an object associated
 * with a user-visible file descriptor, e.g. a vcpu or device, if installation
 * of the new file descriptor fails and the reference cannot be transferred to
 * its final owner.  In such cases, the caller is still actively using @kvm and
 * will fail miserably if the refcount unexpectedly hits zero.
 */
void kvm_put_kvm_no_destroy(struct kvm *kvm)
{
	WARN_ON(refcount_dec_and_test(&kvm->users_count));
}
EXPORT_SYMBOL_GPL(kvm_put_kvm_no_destroy);
I
Izik Eidus 已提交
918

919 920 921 922
static int kvm_vm_release(struct inode *inode, struct file *filp)
{
	struct kvm *kvm = filp->private_data;

G
Gregory Haskins 已提交
923 924
	kvm_irqfd_release(kvm);

I
Izik Eidus 已提交
925
	kvm_put_kvm(kvm);
A
Avi Kivity 已提交
926 927 928
	return 0;
}

929 930
/*
 * Allocation size is twice as large as the actual dirty bitmap size.
931
 * See kvm_vm_ioctl_get_dirty_log() why this is needed.
932
 */
933
static int kvm_alloc_dirty_bitmap(struct kvm_memory_slot *memslot)
934
{
935
	unsigned long dirty_bytes = 2 * kvm_dirty_bitmap_bytes(memslot);
936

937
	memslot->dirty_bitmap = kvzalloc(dirty_bytes, GFP_KERNEL_ACCOUNT);
938 939 940 941 942 943
	if (!memslot->dirty_bitmap)
		return -ENOMEM;

	return 0;
}

944
/*
945 946
 * Delete a memslot by decrementing the number of used slots and shifting all
 * other entries in the array forward one spot.
947
 */
948 949
static inline void kvm_memslot_delete(struct kvm_memslots *slots,
				      struct kvm_memory_slot *memslot)
950
{
951
	struct kvm_memory_slot *mslots = slots->memslots;
952
	int i;
953

954 955
	if (WARN_ON(slots->id_to_index[memslot->id] == -1))
		return;
956

957 958
	slots->used_slots--;

959 960 961
	if (atomic_read(&slots->lru_slot) >= slots->used_slots)
		atomic_set(&slots->lru_slot, 0);

962
	for (i = slots->id_to_index[memslot->id]; i < slots->used_slots; i++) {
963 964 965
		mslots[i] = mslots[i + 1];
		slots->id_to_index[mslots[i].id] = i;
	}
966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994
	mslots[i] = *memslot;
	slots->id_to_index[memslot->id] = -1;
}

/*
 * "Insert" a new memslot by incrementing the number of used slots.  Returns
 * the new slot's initial index into the memslots array.
 */
static inline int kvm_memslot_insert_back(struct kvm_memslots *slots)
{
	return slots->used_slots++;
}

/*
 * Move a changed memslot backwards in the array by shifting existing slots
 * with a higher GFN toward the front of the array.  Note, the changed memslot
 * itself is not preserved in the array, i.e. not swapped at this time, only
 * its new index into the array is tracked.  Returns the changed memslot's
 * current index into the memslots array.
 */
static inline int kvm_memslot_move_backward(struct kvm_memslots *slots,
					    struct kvm_memory_slot *memslot)
{
	struct kvm_memory_slot *mslots = slots->memslots;
	int i;

	if (WARN_ON_ONCE(slots->id_to_index[memslot->id] == -1) ||
	    WARN_ON_ONCE(!slots->used_slots))
		return -1;
995 996

	/*
997 998 999
	 * Move the target memslot backward in the array by shifting existing
	 * memslots with a higher GFN (than the target memslot) towards the
	 * front of the array.
1000
	 */
1001 1002 1003 1004 1005
	for (i = slots->id_to_index[memslot->id]; i < slots->used_slots - 1; i++) {
		if (memslot->base_gfn > mslots[i + 1].base_gfn)
			break;

		WARN_ON_ONCE(memslot->base_gfn == mslots[i + 1].base_gfn);
1006

1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103
		/* Shift the next memslot forward one and update its index. */
		mslots[i] = mslots[i + 1];
		slots->id_to_index[mslots[i].id] = i;
	}
	return i;
}

/*
 * Move a changed memslot forwards in the array by shifting existing slots with
 * a lower GFN toward the back of the array.  Note, the changed memslot itself
 * is not preserved in the array, i.e. not swapped at this time, only its new
 * index into the array is tracked.  Returns the changed memslot's final index
 * into the memslots array.
 */
static inline int kvm_memslot_move_forward(struct kvm_memslots *slots,
					   struct kvm_memory_slot *memslot,
					   int start)
{
	struct kvm_memory_slot *mslots = slots->memslots;
	int i;

	for (i = start; i > 0; i--) {
		if (memslot->base_gfn < mslots[i - 1].base_gfn)
			break;

		WARN_ON_ONCE(memslot->base_gfn == mslots[i - 1].base_gfn);

		/* Shift the next memslot back one and update its index. */
		mslots[i] = mslots[i - 1];
		slots->id_to_index[mslots[i].id] = i;
	}
	return i;
}

/*
 * Re-sort memslots based on their GFN to account for an added, deleted, or
 * moved memslot.  Sorting memslots by GFN allows using a binary search during
 * memslot lookup.
 *
 * IMPORTANT: Slots are sorted from highest GFN to lowest GFN!  I.e. the entry
 * at memslots[0] has the highest GFN.
 *
 * The sorting algorithm takes advantage of having initially sorted memslots
 * and knowing the position of the changed memslot.  Sorting is also optimized
 * by not swapping the updated memslot and instead only shifting other memslots
 * and tracking the new index for the update memslot.  Only once its final
 * index is known is the updated memslot copied into its position in the array.
 *
 *  - When deleting a memslot, the deleted memslot simply needs to be moved to
 *    the end of the array.
 *
 *  - When creating a memslot, the algorithm "inserts" the new memslot at the
 *    end of the array and then it forward to its correct location.
 *
 *  - When moving a memslot, the algorithm first moves the updated memslot
 *    backward to handle the scenario where the memslot's GFN was changed to a
 *    lower value.  update_memslots() then falls through and runs the same flow
 *    as creating a memslot to move the memslot forward to handle the scenario
 *    where its GFN was changed to a higher value.
 *
 * Note, slots are sorted from highest->lowest instead of lowest->highest for
 * historical reasons.  Originally, invalid memslots where denoted by having
 * GFN=0, thus sorting from highest->lowest naturally sorted invalid memslots
 * to the end of the array.  The current algorithm uses dedicated logic to
 * delete a memslot and thus does not rely on invalid memslots having GFN=0.
 *
 * The other historical motiviation for highest->lowest was to improve the
 * performance of memslot lookup.  KVM originally used a linear search starting
 * at memslots[0].  On x86, the largest memslot usually has one of the highest,
 * if not *the* highest, GFN, as the bulk of the guest's RAM is located in a
 * single memslot above the 4gb boundary.  As the largest memslot is also the
 * most likely to be referenced, sorting it to the front of the array was
 * advantageous.  The current binary search starts from the middle of the array
 * and uses an LRU pointer to improve performance for all memslots and GFNs.
 */
static void update_memslots(struct kvm_memslots *slots,
			    struct kvm_memory_slot *memslot,
			    enum kvm_mr_change change)
{
	int i;

	if (change == KVM_MR_DELETE) {
		kvm_memslot_delete(slots, memslot);
	} else {
		if (change == KVM_MR_CREATE)
			i = kvm_memslot_insert_back(slots);
		else
			i = kvm_memslot_move_backward(slots, memslot);
		i = kvm_memslot_move_forward(slots, memslot, i);

		/*
		 * Copy the memslot to its new position in memslots and update
		 * its index accordingly.
		 */
		slots->memslots[i] = *memslot;
		slots->id_to_index[memslot->id] = i;
	}
1104 1105
}

1106
static int check_memory_region_flags(const struct kvm_userspace_memory_region *mem)
1107
{
X
Xiao Guangrong 已提交
1108 1109
	u32 valid_flags = KVM_MEM_LOG_DIRTY_PAGES;

1110
#ifdef __KVM_HAVE_READONLY_MEM
X
Xiao Guangrong 已提交
1111 1112 1113 1114
	valid_flags |= KVM_MEM_READONLY;
#endif

	if (mem->flags & ~valid_flags)
1115 1116 1117 1118 1119
		return -EINVAL;

	return 0;
}

1120
static struct kvm_memslots *install_new_memslots(struct kvm *kvm,
1121
		int as_id, struct kvm_memslots *slots)
1122
{
1123
	struct kvm_memslots *old_memslots = __kvm_memslots(kvm, as_id);
1124
	u64 gen = old_memslots->generation;
1125

1126 1127
	WARN_ON(gen & KVM_MEMSLOT_GEN_UPDATE_IN_PROGRESS);
	slots->generation = gen | KVM_MEMSLOT_GEN_UPDATE_IN_PROGRESS;
1128

1129
	rcu_assign_pointer(kvm->memslots[as_id], slots);
1130
	synchronize_srcu_expedited(&kvm->srcu);
1131

1132
	/*
1133
	 * Increment the new memslot generation a second time, dropping the
M
Miaohe Lin 已提交
1134
	 * update in-progress flag and incrementing the generation based on
1135 1136 1137 1138 1139 1140
	 * 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;

	/*
1141 1142 1143
	 * 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
1144 1145
	 * space 0 will use generations 0, 2, 4, ... while address space 1 will
	 * use generations 1, 3, 5, ...
1146
	 */
1147
	gen += KVM_ADDRESS_SPACE_NUM;
1148

1149
	kvm_arch_memslots_updated(kvm, gen);
1150

1151
	slots->generation = gen;
1152 1153

	return old_memslots;
1154 1155
}

1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181
/*
 * Note, at a minimum, the current number of used slots must be allocated, even
 * when deleting a memslot, as we need a complete duplicate of the memslots for
 * use when invalidating a memslot prior to deleting/moving the memslot.
 */
static struct kvm_memslots *kvm_dup_memslots(struct kvm_memslots *old,
					     enum kvm_mr_change change)
{
	struct kvm_memslots *slots;
	size_t old_size, new_size;

	old_size = sizeof(struct kvm_memslots) +
		   (sizeof(struct kvm_memory_slot) * old->used_slots);

	if (change == KVM_MR_CREATE)
		new_size = old_size + sizeof(struct kvm_memory_slot);
	else
		new_size = old_size;

	slots = kvzalloc(new_size, GFP_KERNEL_ACCOUNT);
	if (likely(slots))
		memcpy(slots, old, old_size);

	return slots;
}

1182 1183
static int kvm_set_memslot(struct kvm *kvm,
			   const struct kvm_userspace_memory_region *mem,
1184
			   struct kvm_memory_slot *old,
1185 1186 1187 1188 1189 1190 1191
			   struct kvm_memory_slot *new, int as_id,
			   enum kvm_mr_change change)
{
	struct kvm_memory_slot *slot;
	struct kvm_memslots *slots;
	int r;

1192
	slots = kvm_dup_memslots(__kvm_memslots(kvm, as_id), change);
1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240
	if (!slots)
		return -ENOMEM;

	if (change == KVM_MR_DELETE || change == KVM_MR_MOVE) {
		/*
		 * Note, the INVALID flag needs to be in the appropriate entry
		 * in the freshly allocated memslots, not in @old or @new.
		 */
		slot = id_to_memslot(slots, old->id);
		slot->flags |= KVM_MEMSLOT_INVALID;

		/*
		 * We can re-use the old memslots, the only difference from the
		 * newly installed memslots is the invalid flag, which will get
		 * dropped by update_memslots anyway.  We'll also revert to the
		 * old memslots if preparing the new memory region fails.
		 */
		slots = install_new_memslots(kvm, as_id, slots);

		/* From this point no new shadow pages pointing to a deleted,
		 * or moved, memslot will be created.
		 *
		 * validation of sp->gfn happens in:
		 *	- gfn_to_hva (kvm_read_guest, gfn_to_pfn)
		 *	- kvm_is_visible_gfn (mmu_check_root)
		 */
		kvm_arch_flush_shadow_memslot(kvm, slot);
	}

	r = kvm_arch_prepare_memory_region(kvm, new, mem, change);
	if (r)
		goto out_slots;

	update_memslots(slots, new, change);
	slots = install_new_memslots(kvm, as_id, slots);

	kvm_arch_commit_memory_region(kvm, mem, old, new, change);

	kvfree(slots);
	return 0;

out_slots:
	if (change == KVM_MR_DELETE || change == KVM_MR_MOVE)
		slots = install_new_memslots(kvm, as_id, slots);
	kvfree(slots);
	return r;
}

1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252
static int kvm_delete_memslot(struct kvm *kvm,
			      const struct kvm_userspace_memory_region *mem,
			      struct kvm_memory_slot *old, int as_id)
{
	struct kvm_memory_slot new;
	int r;

	if (!old->npages)
		return -EINVAL;

	memset(&new, 0, sizeof(new));
	new.id = old->id;
P
Peter Xu 已提交
1253 1254 1255 1256 1257
	/*
	 * This is only for debugging purpose; it should never be referenced
	 * for a removed memslot.
	 */
	new.as_id = as_id;
1258 1259 1260 1261 1262

	r = kvm_set_memslot(kvm, mem, old, &new, as_id, KVM_MR_DELETE);
	if (r)
		return r;

1263
	kvm_free_memslot(kvm, old);
1264 1265 1266
	return 0;
}

A
Avi Kivity 已提交
1267 1268 1269 1270 1271
/*
 * Allocate some memory and give it an address in the guest physical address
 * space.
 *
 * Discontiguous memory is allowed, mostly for framebuffers.
1272
 *
1273
 * Must be called holding kvm->slots_lock for write.
A
Avi Kivity 已提交
1274
 */
1275
int __kvm_set_memory_region(struct kvm *kvm,
1276
			    const struct kvm_userspace_memory_region *mem)
A
Avi Kivity 已提交
1277 1278
{
	struct kvm_memory_slot old, new;
1279
	struct kvm_memory_slot *tmp;
1280
	enum kvm_mr_change change;
1281 1282
	int as_id, id;
	int r;
A
Avi Kivity 已提交
1283

1284 1285
	r = check_memory_region_flags(mem);
	if (r)
1286
		return r;
1287

1288 1289 1290
	as_id = mem->slot >> 16;
	id = (u16)mem->slot;

A
Avi Kivity 已提交
1291 1292
	/* General sanity checks */
	if (mem->memory_size & (PAGE_SIZE - 1))
1293
		return -EINVAL;
A
Avi Kivity 已提交
1294
	if (mem->guest_phys_addr & (PAGE_SIZE - 1))
1295
		return -EINVAL;
1296
	/* We can read the guest memory with __xxx_user() later on. */
1297
	if ((mem->userspace_addr & (PAGE_SIZE - 1)) ||
1298
	     !access_ok((void __user *)(unsigned long)mem->userspace_addr,
1299
			mem->memory_size))
1300
		return -EINVAL;
1301
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_MEM_SLOTS_NUM)
1302
		return -EINVAL;
A
Avi Kivity 已提交
1303
	if (mem->guest_phys_addr + mem->memory_size < mem->guest_phys_addr)
1304
		return -EINVAL;
A
Avi Kivity 已提交
1305

1306 1307 1308 1309
	/*
	 * Make a full copy of the old memslot, the pointer will become stale
	 * when the memslots are re-sorted by update_memslots(), and the old
	 * memslot needs to be referenced after calling update_memslots(), e.g.
1310
	 * to free its resources and for arch specific behavior.
1311
	 */
1312 1313 1314 1315 1316 1317 1318 1319
	tmp = id_to_memslot(__kvm_memslots(kvm, as_id), id);
	if (tmp) {
		old = *tmp;
		tmp = NULL;
	} else {
		memset(&old, 0, sizeof(old));
		old.id = id;
	}
1320

1321 1322 1323
	if (!mem->memory_size)
		return kvm_delete_memslot(kvm, mem, &old, as_id);

P
Peter Xu 已提交
1324
	new.as_id = as_id;
1325
	new.id = id;
1326 1327
	new.base_gfn = mem->guest_phys_addr >> PAGE_SHIFT;
	new.npages = mem->memory_size >> PAGE_SHIFT;
A
Avi Kivity 已提交
1328
	new.flags = mem->flags;
1329
	new.userspace_addr = mem->userspace_addr;
A
Avi Kivity 已提交
1330

1331 1332 1333
	if (new.npages > KVM_MEM_MAX_NR_PAGES)
		return -EINVAL;

1334 1335
	if (!old.npages) {
		change = KVM_MR_CREATE;
1336 1337
		new.dirty_bitmap = NULL;
		memset(&new.arch, 0, sizeof(new.arch));
1338 1339
	} else { /* Modify an existing slot. */
		if ((new.userspace_addr != old.userspace_addr) ||
1340
		    (new.npages != old.npages) ||
1341
		    ((new.flags ^ old.flags) & KVM_MEM_READONLY))
1342
			return -EINVAL;
1343

1344
		if (new.base_gfn != old.base_gfn)
1345 1346 1347 1348 1349
			change = KVM_MR_MOVE;
		else if (new.flags != old.flags)
			change = KVM_MR_FLAGS_ONLY;
		else /* Nothing to change. */
			return 0;
1350 1351 1352 1353

		/* Copy dirty_bitmap and arch from the current memslot. */
		new.dirty_bitmap = old.dirty_bitmap;
		memcpy(&new.arch, &old.arch, sizeof(new.arch));
1354
	}
A
Avi Kivity 已提交
1355

1356
	if ((change == KVM_MR_CREATE) || (change == KVM_MR_MOVE)) {
1357
		/* Check for overlaps */
1358 1359
		kvm_for_each_memslot(tmp, __kvm_memslots(kvm, as_id)) {
			if (tmp->id == id)
1360
				continue;
1361 1362
			if (!((new.base_gfn + new.npages <= tmp->base_gfn) ||
			      (new.base_gfn >= tmp->base_gfn + tmp->npages)))
1363
				return -EEXIST;
1364
		}
A
Avi Kivity 已提交
1365 1366
	}

1367 1368 1369 1370
	/* Allocate/free page dirty bitmap as needed */
	if (!(new.flags & KVM_MEM_LOG_DIRTY_PAGES))
		new.dirty_bitmap = NULL;
	else if (!new.dirty_bitmap) {
1371
		r = kvm_alloc_dirty_bitmap(&new);
1372 1373
		if (r)
			return r;
1374 1375 1376

		if (kvm_dirty_log_manual_protect_and_init_set(kvm))
			bitmap_set(new.dirty_bitmap, 0, new.npages);
A
Avi Kivity 已提交
1377 1378
	}

1379 1380 1381
	r = kvm_set_memslot(kvm, mem, &old, &new, as_id, change);
	if (r)
		goto out_bitmap;
1382

1383 1384
	if (old.dirty_bitmap && !new.dirty_bitmap)
		kvm_destroy_dirty_bitmap(&old);
A
Avi Kivity 已提交
1385 1386
	return 0;

1387 1388 1389
out_bitmap:
	if (new.dirty_bitmap && !old.dirty_bitmap)
		kvm_destroy_dirty_bitmap(&new);
A
Avi Kivity 已提交
1390
	return r;
1391
}
1392 1393 1394
EXPORT_SYMBOL_GPL(__kvm_set_memory_region);

int kvm_set_memory_region(struct kvm *kvm,
1395
			  const struct kvm_userspace_memory_region *mem)
1396 1397 1398
{
	int r;

1399
	mutex_lock(&kvm->slots_lock);
1400
	r = __kvm_set_memory_region(kvm, mem);
1401
	mutex_unlock(&kvm->slots_lock);
1402 1403
	return r;
}
1404 1405
EXPORT_SYMBOL_GPL(kvm_set_memory_region);

1406 1407
static int kvm_vm_ioctl_set_memory_region(struct kvm *kvm,
					  struct kvm_userspace_memory_region *mem)
1408
{
1409
	if ((u16)mem->slot >= KVM_USER_MEM_SLOTS)
1410
		return -EINVAL;
1411

1412
	return kvm_set_memory_region(kvm, mem);
A
Avi Kivity 已提交
1413 1414
}

1415
#ifndef CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT
1416 1417 1418 1419 1420 1421 1422 1423 1424
/**
 * kvm_get_dirty_log - get a snapshot of dirty pages
 * @kvm:	pointer to kvm instance
 * @log:	slot id and address to which we copy the log
 * @is_dirty:	set to '1' if any dirty pages were found
 * @memslot:	set to the associated memslot, always valid on success
 */
int kvm_get_dirty_log(struct kvm *kvm, struct kvm_dirty_log *log,
		      int *is_dirty, struct kvm_memory_slot **memslot)
A
Avi Kivity 已提交
1425
{
1426
	struct kvm_memslots *slots;
1427
	int i, as_id, id;
1428
	unsigned long n;
A
Avi Kivity 已提交
1429 1430
	unsigned long any = 0;

1431 1432 1433
	*memslot = NULL;
	*is_dirty = 0;

1434 1435 1436
	as_id = log->slot >> 16;
	id = (u16)log->slot;
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_USER_MEM_SLOTS)
1437
		return -EINVAL;
A
Avi Kivity 已提交
1438

1439
	slots = __kvm_memslots(kvm, as_id);
1440
	*memslot = id_to_memslot(slots, id);
1441
	if (!(*memslot) || !(*memslot)->dirty_bitmap)
1442
		return -ENOENT;
A
Avi Kivity 已提交
1443

1444 1445 1446
	kvm_arch_sync_dirty_log(kvm, *memslot);

	n = kvm_dirty_bitmap_bytes(*memslot);
A
Avi Kivity 已提交
1447

1448
	for (i = 0; !any && i < n/sizeof(long); ++i)
1449
		any = (*memslot)->dirty_bitmap[i];
A
Avi Kivity 已提交
1450

1451
	if (copy_to_user(log->dirty_bitmap, (*memslot)->dirty_bitmap, n))
1452
		return -EFAULT;
A
Avi Kivity 已提交
1453

1454 1455
	if (any)
		*is_dirty = 1;
1456
	return 0;
A
Avi Kivity 已提交
1457
}
1458
EXPORT_SYMBOL_GPL(kvm_get_dirty_log);
A
Avi Kivity 已提交
1459

1460
#else /* CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT */
1461
/**
J
Jiang Biao 已提交
1462
 * kvm_get_dirty_log_protect - get a snapshot of dirty pages
1463
 *	and reenable dirty page tracking for the corresponding pages.
1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481
 * @kvm:	pointer to kvm instance
 * @log:	slot id and address to which we copy the log
 *
 * 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.
 *
 */
1482
static int kvm_get_dirty_log_protect(struct kvm *kvm, struct kvm_dirty_log *log)
1483
{
1484
	struct kvm_memslots *slots;
1485
	struct kvm_memory_slot *memslot;
1486
	int i, as_id, id;
1487 1488 1489
	unsigned long n;
	unsigned long *dirty_bitmap;
	unsigned long *dirty_bitmap_buffer;
1490
	bool flush;
1491

1492 1493 1494
	as_id = log->slot >> 16;
	id = (u16)log->slot;
	if (as_id >= KVM_ADDRESS_SPACE_NUM || id >= KVM_USER_MEM_SLOTS)
1495
		return -EINVAL;
1496

1497 1498
	slots = __kvm_memslots(kvm, as_id);
	memslot = id_to_memslot(slots, id);
1499 1500
	if (!memslot || !memslot->dirty_bitmap)
		return -ENOENT;
1501 1502 1503

	dirty_bitmap = memslot->dirty_bitmap;

1504 1505
	kvm_arch_sync_dirty_log(kvm, memslot);

1506
	n = kvm_dirty_bitmap_bytes(memslot);
1507
	flush = false;
1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520
	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);
1521

1522 1523 1524 1525
		spin_lock(&kvm->mmu_lock);
		for (i = 0; i < n / sizeof(long); i++) {
			unsigned long mask;
			gfn_t offset;
1526

1527 1528 1529
			if (!dirty_bitmap[i])
				continue;

1530
			flush = true;
1531 1532 1533
			mask = xchg(&dirty_bitmap[i], 0);
			dirty_bitmap_buffer[i] = mask;

1534 1535 1536
			offset = i * BITS_PER_LONG;
			kvm_arch_mmu_enable_log_dirty_pt_masked(kvm, memslot,
								offset, mask);
1537 1538 1539 1540
		}
		spin_unlock(&kvm->mmu_lock);
	}

1541 1542 1543
	if (flush)
		kvm_arch_flush_remote_tlbs_memslot(kvm, memslot);

1544 1545 1546 1547
	if (copy_to_user(log->dirty_bitmap, dirty_bitmap_buffer, n))
		return -EFAULT;
	return 0;
}
1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580


/**
 * kvm_vm_ioctl_get_dirty_log - get and clear the log of dirty pages in a slot
 * @kvm: kvm instance
 * @log: slot id and address to which we copy the log
 *
 * Steps 1-4 below provide general overview of dirty page logging. See
 * kvm_get_dirty_log_protect() function description for additional details.
 *
 * We call kvm_get_dirty_log_protect() to handle steps 1-3, upon return we
 * always flush the TLB (step 4) even if previous step failed  and the dirty
 * bitmap may be corrupt. Regardless of previous outcome the KVM logging API
 * does not preclude user space subsequent dirty log read. Flushing TLB ensures
 * writes will be marked dirty for next log read.
 *
 *   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. Flush TLB's if needed.
 */
static int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm,
				      struct kvm_dirty_log *log)
{
	int r;

	mutex_lock(&kvm->slots_lock);

	r = kvm_get_dirty_log_protect(kvm, log);

	mutex_unlock(&kvm->slots_lock);
	return r;
}
1581 1582 1583 1584 1585 1586 1587

/**
 * 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
 */
1588 1589
static int kvm_clear_dirty_log_protect(struct kvm *kvm,
				       struct kvm_clear_dirty_log *log)
1590 1591 1592
{
	struct kvm_memslots *slots;
	struct kvm_memory_slot *memslot;
1593
	int as_id, id;
1594
	gfn_t offset;
1595
	unsigned long i, n;
1596 1597
	unsigned long *dirty_bitmap;
	unsigned long *dirty_bitmap_buffer;
1598
	bool flush;
1599 1600 1601 1602 1603 1604

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

1605
	if (log->first_page & 63)
1606 1607 1608 1609
		return -EINVAL;

	slots = __kvm_memslots(kvm, as_id);
	memslot = id_to_memslot(slots, id);
1610 1611
	if (!memslot || !memslot->dirty_bitmap)
		return -ENOENT;
1612 1613 1614

	dirty_bitmap = memslot->dirty_bitmap;

1615
	n = ALIGN(log->num_pages, BITS_PER_LONG) / 8;
1616 1617

	if (log->first_page > memslot->npages ||
1618 1619 1620
	    log->num_pages > memslot->npages - log->first_page ||
	    (log->num_pages < memslot->npages - log->first_page && (log->num_pages & 63)))
	    return -EINVAL;
1621

1622 1623 1624
	kvm_arch_sync_dirty_log(kvm, memslot);

	flush = false;
1625 1626 1627
	dirty_bitmap_buffer = kvm_second_dirty_bitmap(memslot);
	if (copy_from_user(dirty_bitmap_buffer, log->dirty_bitmap, n))
		return -EFAULT;
1628

1629
	spin_lock(&kvm->mmu_lock);
1630 1631
	for (offset = log->first_page, i = offset / BITS_PER_LONG,
		 n = DIV_ROUND_UP(log->num_pages, BITS_PER_LONG); n--;
1632 1633 1634 1635
	     i++, offset += BITS_PER_LONG) {
		unsigned long mask = *dirty_bitmap_buffer++;
		atomic_long_t *p = (atomic_long_t *) &dirty_bitmap[i];
		if (!mask)
1636 1637
			continue;

1638
		mask &= atomic_long_fetch_andnot(mask, p);
1639

1640 1641 1642 1643 1644 1645
		/*
		 * 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.
		*/
1646
		if (mask) {
1647
			flush = true;
1648 1649 1650
			kvm_arch_mmu_enable_log_dirty_pt_masked(kvm, memslot,
								offset, mask);
		}
1651 1652
	}
	spin_unlock(&kvm->mmu_lock);
1653

1654 1655 1656
	if (flush)
		kvm_arch_flush_remote_tlbs_memslot(kvm, memslot);

1657
	return 0;
1658
}
1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672

static int kvm_vm_ioctl_clear_dirty_log(struct kvm *kvm,
					struct kvm_clear_dirty_log *log)
{
	int r;

	mutex_lock(&kvm->slots_lock);

	r = kvm_clear_dirty_log_protect(kvm, log);

	mutex_unlock(&kvm->slots_lock);
	return r;
}
#endif /* CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT */
1673

1674 1675 1676 1677
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
1678
EXPORT_SYMBOL_GPL(gfn_to_memslot);
A
Avi Kivity 已提交
1679

1680 1681 1682 1683
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);
}
1684
EXPORT_SYMBOL_GPL(kvm_vcpu_gfn_to_memslot);
1685

1686
bool kvm_is_visible_gfn(struct kvm *kvm, gfn_t gfn)
1687
{
1688
	struct kvm_memory_slot *memslot = gfn_to_memslot(kvm, gfn);
1689

1690
	return kvm_is_visible_memslot(memslot);
1691 1692 1693
}
EXPORT_SYMBOL_GPL(kvm_is_visible_gfn);

1694 1695 1696 1697 1698 1699 1700 1701
bool kvm_vcpu_is_visible_gfn(struct kvm_vcpu *vcpu, gfn_t gfn)
{
	struct kvm_memory_slot *memslot = kvm_vcpu_gfn_to_memslot(vcpu, gfn);

	return kvm_is_visible_memslot(memslot);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_is_visible_gfn);

1702
unsigned long kvm_host_page_size(struct kvm_vcpu *vcpu, gfn_t gfn)
J
Joerg Roedel 已提交
1703 1704 1705 1706 1707 1708
{
	struct vm_area_struct *vma;
	unsigned long addr, size;

	size = PAGE_SIZE;

1709
	addr = kvm_vcpu_gfn_to_hva_prot(vcpu, gfn, NULL);
J
Joerg Roedel 已提交
1710 1711 1712
	if (kvm_is_error_hva(addr))
		return PAGE_SIZE;

1713
	mmap_read_lock(current->mm);
J
Joerg Roedel 已提交
1714 1715 1716 1717 1718 1719 1720
	vma = find_vma(current->mm, addr);
	if (!vma)
		goto out;

	size = vma_kernel_pagesize(vma);

out:
1721
	mmap_read_unlock(current->mm);
J
Joerg Roedel 已提交
1722 1723 1724 1725

	return size;
}

X
Xiao Guangrong 已提交
1726 1727 1728 1729 1730 1731 1732
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
1733
{
1734
	if (!slot || slot->flags & KVM_MEMSLOT_INVALID)
X
Xiao Guangrong 已提交
1735
		return KVM_HVA_ERR_BAD;
1736

X
Xiao Guangrong 已提交
1737 1738
	if (memslot_is_readonly(slot) && write)
		return KVM_HVA_ERR_RO_BAD;
1739 1740 1741 1742

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

X
Xiao Guangrong 已提交
1743
	return __gfn_to_hva_memslot(slot, gfn);
I
Izik Eidus 已提交
1744
}
1745

X
Xiao Guangrong 已提交
1746 1747 1748 1749
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
1750
}
1751

X
Xiao Guangrong 已提交
1752
unsigned long gfn_to_hva_memslot(struct kvm_memory_slot *slot,
1753
					gfn_t gfn)
X
Xiao Guangrong 已提交
1754 1755 1756 1757 1758
{
	return gfn_to_hva_many(slot, gfn, NULL);
}
EXPORT_SYMBOL_GPL(gfn_to_hva_memslot);

1759 1760
unsigned long gfn_to_hva(struct kvm *kvm, gfn_t gfn)
{
1761
	return gfn_to_hva_many(gfn_to_memslot(kvm, gfn), gfn, NULL);
1762
}
1763
EXPORT_SYMBOL_GPL(gfn_to_hva);
I
Izik Eidus 已提交
1764

1765 1766 1767 1768 1769 1770
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);

1771
/*
1772 1773 1774 1775 1776 1777
 * 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
1778
 */
1779 1780
unsigned long gfn_to_hva_memslot_prot(struct kvm_memory_slot *slot,
				      gfn_t gfn, bool *writable)
1781
{
1782 1783 1784
	unsigned long hva = __gfn_to_hva_many(slot, gfn, NULL, false);

	if (!kvm_is_error_hva(hva) && writable)
1785 1786
		*writable = !memslot_is_readonly(slot);

1787
	return hva;
1788 1789
}

1790 1791 1792 1793 1794 1795 1796
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);
}

1797 1798 1799 1800 1801 1802 1803
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);
}

1804 1805
static inline int check_user_page_hwpoison(unsigned long addr)
{
L
Lorenzo Stoakes 已提交
1806
	int rc, flags = FOLL_HWPOISON | FOLL_WRITE;
1807

L
Lorenzo Stoakes 已提交
1808
	rc = get_user_pages(addr, 1, flags, NULL, NULL);
1809 1810 1811
	return rc == -EHWPOISON;
}

X
Xiao Guangrong 已提交
1812
/*
1813 1814
 * 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
M
Miaohe Lin 已提交
1815
 * only part that runs if we can in atomic context.
X
Xiao Guangrong 已提交
1816
 */
1817 1818
static bool hva_to_pfn_fast(unsigned long addr, bool write_fault,
			    bool *writable, kvm_pfn_t *pfn)
A
Avi Kivity 已提交
1819
{
1820
	struct page *page[1];
A
Avi Kivity 已提交
1821

1822 1823 1824 1825 1826 1827 1828
	/*
	 * 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;
1829

1830
	if (get_user_page_fast_only(addr, FOLL_WRITE, page)) {
X
Xiao Guangrong 已提交
1831
		*pfn = page_to_pfn(page[0]);
1832

X
Xiao Guangrong 已提交
1833 1834 1835 1836
		if (writable)
			*writable = true;
		return true;
	}
1837

X
Xiao Guangrong 已提交
1838 1839
	return false;
}
1840

X
Xiao Guangrong 已提交
1841 1842 1843 1844 1845
/*
 * 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
1846
			   bool *writable, kvm_pfn_t *pfn)
X
Xiao Guangrong 已提交
1847
{
1848 1849
	unsigned int flags = FOLL_HWPOISON;
	struct page *page;
X
Xiao Guangrong 已提交
1850
	int npages = 0;
1851

X
Xiao Guangrong 已提交
1852 1853 1854 1855 1856
	might_sleep();

	if (writable)
		*writable = write_fault;

1857 1858 1859 1860
	if (write_fault)
		flags |= FOLL_WRITE;
	if (async)
		flags |= FOLL_NOWAIT;
1861

1862
	npages = get_user_pages_unlocked(addr, 1, &page, flags);
X
Xiao Guangrong 已提交
1863 1864 1865 1866
	if (npages != 1)
		return npages;

	/* map read fault as writable if possible */
1867
	if (unlikely(!write_fault) && writable) {
1868
		struct page *wpage;
X
Xiao Guangrong 已提交
1869

1870
		if (get_user_page_fast_only(addr, FOLL_WRITE, &wpage)) {
X
Xiao Guangrong 已提交
1871
			*writable = true;
1872 1873
			put_page(page);
			page = wpage;
1874
		}
1875
	}
1876
	*pfn = page_to_pfn(page);
X
Xiao Guangrong 已提交
1877 1878
	return npages;
}
I
Izik Eidus 已提交
1879

X
Xiao Guangrong 已提交
1880 1881 1882 1883
static bool vma_is_valid(struct vm_area_struct *vma, bool write_fault)
{
	if (unlikely(!(vma->vm_flags & VM_READ)))
		return false;
1884

X
Xiao Guangrong 已提交
1885 1886
	if (write_fault && (unlikely(!(vma->vm_flags & VM_WRITE))))
		return false;
1887

X
Xiao Guangrong 已提交
1888 1889
	return true;
}
1890

1891 1892
static int hva_to_pfn_remapped(struct vm_area_struct *vma,
			       unsigned long addr, bool *async,
1893 1894
			       bool write_fault, bool *writable,
			       kvm_pfn_t *p_pfn)
1895
{
1896 1897 1898 1899 1900 1901 1902 1903 1904 1905
	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;
1906
		r = fixup_user_fault(current->mm, addr,
1907 1908
				     (write_fault ? FAULT_FLAG_WRITE : 0),
				     &unlocked);
1909 1910
		if (unlocked)
			return -EAGAIN;
1911 1912 1913 1914 1915 1916 1917 1918 1919
		if (r)
			return r;

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

	}

1920 1921
	if (writable)
		*writable = true;
1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936

	/*
	 * 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;
1937 1938 1939
	return 0;
}

1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953
/*
 * 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
1954
static kvm_pfn_t hva_to_pfn(unsigned long addr, bool atomic, bool *async,
X
Xiao Guangrong 已提交
1955 1956 1957
			bool write_fault, bool *writable)
{
	struct vm_area_struct *vma;
D
Dan Williams 已提交
1958
	kvm_pfn_t pfn = 0;
1959
	int npages, r;
1960

X
Xiao Guangrong 已提交
1961 1962
	/* we can do it either atomically or asynchronously, not both */
	BUG_ON(atomic && async);
1963

1964
	if (hva_to_pfn_fast(addr, write_fault, writable, &pfn))
X
Xiao Guangrong 已提交
1965 1966 1967 1968 1969 1970 1971 1972
		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;
1973

1974
	mmap_read_lock(current->mm);
X
Xiao Guangrong 已提交
1975 1976 1977 1978 1979 1980
	if (npages == -EHWPOISON ||
	      (!async && check_user_page_hwpoison(addr))) {
		pfn = KVM_PFN_ERR_HWPOISON;
		goto exit;
	}

1981
retry:
X
Xiao Guangrong 已提交
1982 1983 1984 1985
	vma = find_vma_intersection(current->mm, addr, addr + 1);

	if (vma == NULL)
		pfn = KVM_PFN_ERR_FAULT;
1986
	else if (vma->vm_flags & (VM_IO | VM_PFNMAP)) {
1987
		r = hva_to_pfn_remapped(vma, addr, async, write_fault, writable, &pfn);
1988 1989
		if (r == -EAGAIN)
			goto retry;
1990 1991
		if (r < 0)
			pfn = KVM_PFN_ERR_FAULT;
X
Xiao Guangrong 已提交
1992
	} else {
X
Xiao Guangrong 已提交
1993
		if (async && vma_is_valid(vma, write_fault))
X
Xiao Guangrong 已提交
1994 1995 1996 1997
			*async = true;
		pfn = KVM_PFN_ERR_FAULT;
	}
exit:
1998
	mmap_read_unlock(current->mm);
1999
	return pfn;
2000 2001
}

D
Dan Williams 已提交
2002 2003 2004
kvm_pfn_t __gfn_to_pfn_memslot(struct kvm_memory_slot *slot, gfn_t gfn,
			       bool atomic, bool *async, bool write_fault,
			       bool *writable)
2005
{
X
Xiao Guangrong 已提交
2006 2007
	unsigned long addr = __gfn_to_hva_many(slot, gfn, NULL, write_fault);

2008 2009 2010
	if (addr == KVM_HVA_ERR_RO_BAD) {
		if (writable)
			*writable = false;
X
Xiao Guangrong 已提交
2011
		return KVM_PFN_ERR_RO_FAULT;
2012
	}
X
Xiao Guangrong 已提交
2013

2014 2015 2016
	if (kvm_is_error_hva(addr)) {
		if (writable)
			*writable = false;
2017
		return KVM_PFN_NOSLOT;
2018
	}
X
Xiao Guangrong 已提交
2019 2020 2021 2022 2023 2024 2025 2026 2027

	/* 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);
2028
}
2029
EXPORT_SYMBOL_GPL(__gfn_to_pfn_memslot);
2030

D
Dan Williams 已提交
2031
kvm_pfn_t gfn_to_pfn_prot(struct kvm *kvm, gfn_t gfn, bool write_fault,
2032 2033
		      bool *writable)
{
P
Paolo Bonzini 已提交
2034 2035
	return __gfn_to_pfn_memslot(gfn_to_memslot(kvm, gfn), gfn, false, NULL,
				    write_fault, writable);
2036 2037 2038
}
EXPORT_SYMBOL_GPL(gfn_to_pfn_prot);

D
Dan Williams 已提交
2039
kvm_pfn_t gfn_to_pfn_memslot(struct kvm_memory_slot *slot, gfn_t gfn)
2040
{
X
Xiao Guangrong 已提交
2041
	return __gfn_to_pfn_memslot(slot, gfn, false, NULL, true, NULL);
2042
}
P
Paolo Bonzini 已提交
2043
EXPORT_SYMBOL_GPL(gfn_to_pfn_memslot);
2044

D
Dan Williams 已提交
2045
kvm_pfn_t gfn_to_pfn_memslot_atomic(struct kvm_memory_slot *slot, gfn_t gfn)
2046
{
X
Xiao Guangrong 已提交
2047
	return __gfn_to_pfn_memslot(slot, gfn, true, NULL, true, NULL);
2048
}
2049
EXPORT_SYMBOL_GPL(gfn_to_pfn_memslot_atomic);
2050

D
Dan Williams 已提交
2051
kvm_pfn_t kvm_vcpu_gfn_to_pfn_atomic(struct kvm_vcpu *vcpu, gfn_t gfn)
2052 2053 2054 2055 2056
{
	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
2057
kvm_pfn_t gfn_to_pfn(struct kvm *kvm, gfn_t gfn)
P
Paolo Bonzini 已提交
2058 2059 2060 2061 2062
{
	return gfn_to_pfn_memslot(gfn_to_memslot(kvm, gfn), gfn);
}
EXPORT_SYMBOL_GPL(gfn_to_pfn);

D
Dan Williams 已提交
2063
kvm_pfn_t kvm_vcpu_gfn_to_pfn(struct kvm_vcpu *vcpu, gfn_t gfn)
2064 2065 2066 2067 2068
{
	return gfn_to_pfn_memslot(kvm_vcpu_gfn_to_memslot(vcpu, gfn), gfn);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_gfn_to_pfn);

2069 2070
int gfn_to_page_many_atomic(struct kvm_memory_slot *slot, gfn_t gfn,
			    struct page **pages, int nr_pages)
2071 2072
{
	unsigned long addr;
2073
	gfn_t entry = 0;
2074

2075
	addr = gfn_to_hva_many(slot, gfn, &entry);
2076 2077 2078 2079 2080 2081
	if (kvm_is_error_hva(addr))
		return -1;

	if (entry < nr_pages)
		return 0;

2082
	return get_user_pages_fast_only(addr, nr_pages, FOLL_WRITE, pages);
2083 2084 2085
}
EXPORT_SYMBOL_GPL(gfn_to_page_many_atomic);

D
Dan Williams 已提交
2086
static struct page *kvm_pfn_to_page(kvm_pfn_t pfn)
X
Xiao Guangrong 已提交
2087
{
2088
	if (is_error_noslot_pfn(pfn))
2089
		return KVM_ERR_PTR_BAD_PAGE;
X
Xiao Guangrong 已提交
2090

2091
	if (kvm_is_reserved_pfn(pfn)) {
2092
		WARN_ON(1);
2093
		return KVM_ERR_PTR_BAD_PAGE;
2094
	}
X
Xiao Guangrong 已提交
2095 2096 2097 2098

	return pfn_to_page(pfn);
}

2099 2100
struct page *gfn_to_page(struct kvm *kvm, gfn_t gfn)
{
D
Dan Williams 已提交
2101
	kvm_pfn_t pfn;
2102 2103 2104

	pfn = gfn_to_pfn(kvm, gfn);

X
Xiao Guangrong 已提交
2105
	return kvm_pfn_to_page(pfn);
A
Avi Kivity 已提交
2106 2107 2108
}
EXPORT_SYMBOL_GPL(gfn_to_page);

2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133
void kvm_release_pfn(kvm_pfn_t pfn, bool dirty, struct gfn_to_pfn_cache *cache)
{
	if (pfn == 0)
		return;

	if (cache)
		cache->pfn = cache->gfn = 0;

	if (dirty)
		kvm_release_pfn_dirty(pfn);
	else
		kvm_release_pfn_clean(pfn);
}

static void kvm_cache_gfn_to_pfn(struct kvm_memory_slot *slot, gfn_t gfn,
				 struct gfn_to_pfn_cache *cache, u64 gen)
{
	kvm_release_pfn(cache->pfn, cache->dirty, cache);

	cache->pfn = gfn_to_pfn_memslot(slot, gfn);
	cache->gfn = gfn;
	cache->dirty = false;
	cache->generation = gen;
}

2134
static int __kvm_map_gfn(struct kvm_memslots *slots, gfn_t gfn,
2135 2136 2137
			 struct kvm_host_map *map,
			 struct gfn_to_pfn_cache *cache,
			 bool atomic)
2138 2139 2140 2141
{
	kvm_pfn_t pfn;
	void *hva = NULL;
	struct page *page = KVM_UNMAPPED_PAGE;
2142
	struct kvm_memory_slot *slot = __gfn_to_memslot(slots, gfn);
2143
	u64 gen = slots->generation;
2144 2145 2146 2147

	if (!map)
		return -EINVAL;

2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160
	if (cache) {
		if (!cache->pfn || cache->gfn != gfn ||
			cache->generation != gen) {
			if (atomic)
				return -EAGAIN;
			kvm_cache_gfn_to_pfn(slot, gfn, cache, gen);
		}
		pfn = cache->pfn;
	} else {
		if (atomic)
			return -EAGAIN;
		pfn = gfn_to_pfn_memslot(slot, gfn);
	}
2161 2162 2163 2164 2165
	if (is_error_noslot_pfn(pfn))
		return -EINVAL;

	if (pfn_valid(pfn)) {
		page = pfn_to_page(pfn);
2166 2167 2168 2169
		if (atomic)
			hva = kmap_atomic(page);
		else
			hva = kmap(page);
P
Paolo Bonzini 已提交
2170
#ifdef CONFIG_HAS_IOMEM
2171
	} else if (!atomic) {
2172
		hva = memremap(pfn_to_hpa(pfn), PAGE_SIZE, MEMREMAP_WB);
2173 2174
	} else {
		return -EINVAL;
P
Paolo Bonzini 已提交
2175
#endif
2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188
	}

	if (!hva)
		return -EFAULT;

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

	return 0;
}

2189 2190
int kvm_map_gfn(struct kvm_vcpu *vcpu, gfn_t gfn, struct kvm_host_map *map,
		struct gfn_to_pfn_cache *cache, bool atomic)
2191
{
2192 2193
	return __kvm_map_gfn(kvm_memslots(vcpu->kvm), gfn, map,
			cache, atomic);
2194 2195 2196
}
EXPORT_SYMBOL_GPL(kvm_map_gfn);

2197 2198
int kvm_vcpu_map(struct kvm_vcpu *vcpu, gfn_t gfn, struct kvm_host_map *map)
{
2199 2200
	return __kvm_map_gfn(kvm_vcpu_memslots(vcpu), gfn, map,
		NULL, false);
2201 2202 2203
}
EXPORT_SYMBOL_GPL(kvm_vcpu_map);

2204
static void __kvm_unmap_gfn(struct kvm_memory_slot *memslot,
2205 2206 2207
			struct kvm_host_map *map,
			struct gfn_to_pfn_cache *cache,
			bool dirty, bool atomic)
2208 2209 2210 2211 2212 2213 2214
{
	if (!map)
		return;

	if (!map->hva)
		return;

2215 2216 2217 2218 2219 2220
	if (map->page != KVM_UNMAPPED_PAGE) {
		if (atomic)
			kunmap_atomic(map->hva);
		else
			kunmap(map->page);
	}
2221
#ifdef CONFIG_HAS_IOMEM
2222
	else if (!atomic)
2223
		memunmap(map->hva);
2224 2225
	else
		WARN_ONCE(1, "Unexpected unmapping in atomic context");
2226
#endif
2227

2228
	if (dirty)
2229
		mark_page_dirty_in_slot(memslot, map->gfn);
2230 2231 2232 2233 2234

	if (cache)
		cache->dirty |= dirty;
	else
		kvm_release_pfn(map->pfn, dirty, NULL);
2235 2236 2237 2238

	map->hva = NULL;
	map->page = NULL;
}
2239

2240 2241
int kvm_unmap_gfn(struct kvm_vcpu *vcpu, struct kvm_host_map *map, 
		  struct gfn_to_pfn_cache *cache, bool dirty, bool atomic)
2242
{
2243 2244
	__kvm_unmap_gfn(gfn_to_memslot(vcpu->kvm, map->gfn), map,
			cache, dirty, atomic);
2245 2246 2247 2248 2249 2250
	return 0;
}
EXPORT_SYMBOL_GPL(kvm_unmap_gfn);

void kvm_vcpu_unmap(struct kvm_vcpu *vcpu, struct kvm_host_map *map, bool dirty)
{
2251 2252
	__kvm_unmap_gfn(kvm_vcpu_gfn_to_memslot(vcpu, map->gfn), map, NULL,
			dirty, false);
2253
}
2254 2255
EXPORT_SYMBOL_GPL(kvm_vcpu_unmap);

2256 2257
struct page *kvm_vcpu_gfn_to_page(struct kvm_vcpu *vcpu, gfn_t gfn)
{
D
Dan Williams 已提交
2258
	kvm_pfn_t pfn;
2259 2260 2261 2262 2263 2264 2265

	pfn = kvm_vcpu_gfn_to_pfn(vcpu, gfn);

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

2266 2267
void kvm_release_page_clean(struct page *page)
{
2268 2269
	WARN_ON(is_error_page(page));

2270
	kvm_release_pfn_clean(page_to_pfn(page));
2271 2272 2273
}
EXPORT_SYMBOL_GPL(kvm_release_page_clean);

D
Dan Williams 已提交
2274
void kvm_release_pfn_clean(kvm_pfn_t pfn)
2275
{
2276
	if (!is_error_noslot_pfn(pfn) && !kvm_is_reserved_pfn(pfn))
2277
		put_page(pfn_to_page(pfn));
2278 2279 2280
}
EXPORT_SYMBOL_GPL(kvm_release_pfn_clean);

2281
void kvm_release_page_dirty(struct page *page)
2282
{
X
Xiao Guangrong 已提交
2283 2284
	WARN_ON(is_error_page(page));

2285 2286 2287 2288
	kvm_release_pfn_dirty(page_to_pfn(page));
}
EXPORT_SYMBOL_GPL(kvm_release_page_dirty);

2289
void kvm_release_pfn_dirty(kvm_pfn_t pfn)
2290 2291 2292 2293
{
	kvm_set_pfn_dirty(pfn);
	kvm_release_pfn_clean(pfn);
}
2294
EXPORT_SYMBOL_GPL(kvm_release_pfn_dirty);
2295

D
Dan Williams 已提交
2296
void kvm_set_pfn_dirty(kvm_pfn_t pfn)
2297
{
2298 2299
	if (!kvm_is_reserved_pfn(pfn) && !kvm_is_zone_device_pfn(pfn))
		SetPageDirty(pfn_to_page(pfn));
2300
}
2301 2302
EXPORT_SYMBOL_GPL(kvm_set_pfn_dirty);

D
Dan Williams 已提交
2303
void kvm_set_pfn_accessed(kvm_pfn_t pfn)
2304
{
2305
	if (!kvm_is_reserved_pfn(pfn) && !kvm_is_zone_device_pfn(pfn))
2306
		mark_page_accessed(pfn_to_page(pfn));
2307 2308 2309
}
EXPORT_SYMBOL_GPL(kvm_set_pfn_accessed);

D
Dan Williams 已提交
2310
void kvm_get_pfn(kvm_pfn_t pfn)
2311
{
2312
	if (!kvm_is_reserved_pfn(pfn))
2313
		get_page(pfn_to_page(pfn));
2314 2315
}
EXPORT_SYMBOL_GPL(kvm_get_pfn);
2316

2317 2318 2319 2320 2321 2322 2323 2324
static int next_segment(unsigned long len, int offset)
{
	if (len > PAGE_SIZE - offset)
		return PAGE_SIZE - offset;
	else
		return len;
}

2325 2326
static int __kvm_read_guest_page(struct kvm_memory_slot *slot, gfn_t gfn,
				 void *data, int offset, int len)
2327
{
2328 2329
	int r;
	unsigned long addr;
2330

2331
	addr = gfn_to_hva_memslot_prot(slot, gfn, NULL);
2332 2333
	if (kvm_is_error_hva(addr))
		return -EFAULT;
2334
	r = __copy_from_user(data, (void __user *)addr + offset, len);
2335
	if (r)
2336 2337 2338
		return -EFAULT;
	return 0;
}
2339 2340 2341 2342 2343 2344 2345 2346

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);
}
2347 2348
EXPORT_SYMBOL_GPL(kvm_read_guest_page);

2349 2350 2351 2352 2353 2354 2355 2356 2357
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);

2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377
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);

2378
int kvm_vcpu_read_guest(struct kvm_vcpu *vcpu, gpa_t gpa, void *data, unsigned long len)
2379 2380
{
	gfn_t gfn = gpa >> PAGE_SHIFT;
2381
	int seg;
2382
	int offset = offset_in_page(gpa);
2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396
	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);
2397

2398 2399 2400 2401 2402 2403 2404
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);
2405 2406
	if (kvm_is_error_hva(addr))
		return -EFAULT;
2407
	pagefault_disable();
2408
	r = __copy_from_user_inatomic(data, (void __user *)addr + offset, len);
2409
	pagefault_enable();
2410 2411 2412 2413 2414
	if (r)
		return -EFAULT;
	return 0;
}

2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427
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)
2428
{
2429 2430
	int r;
	unsigned long addr;
2431

2432
	addr = gfn_to_hva_memslot(memslot, gfn);
2433 2434
	if (kvm_is_error_hva(addr))
		return -EFAULT;
2435
	r = __copy_to_user((void __user *)addr + offset, data, len);
2436
	if (r)
2437
		return -EFAULT;
2438
	mark_page_dirty_in_slot(memslot, gfn);
2439 2440
	return 0;
}
2441 2442 2443 2444 2445 2446 2447 2448

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);
}
2449 2450
EXPORT_SYMBOL_GPL(kvm_write_guest_page);

2451 2452 2453 2454 2455 2456 2457 2458 2459
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);

2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478
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;
}
2479
EXPORT_SYMBOL_GPL(kvm_write_guest);
2480

2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501
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);

2502 2503 2504
static int __kvm_gfn_to_hva_cache_init(struct kvm_memslots *slots,
				       struct gfn_to_hva_cache *ghc,
				       gpa_t gpa, unsigned long len)
2505 2506
{
	int offset = offset_in_page(gpa);
2507 2508 2509 2510
	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;
2511

2512
	/* Update ghc->generation before performing any error checks. */
2513
	ghc->generation = slots->generation;
2514 2515 2516 2517 2518

	if (start_gfn > end_gfn) {
		ghc->hva = KVM_HVA_ERR_BAD;
		return -EINVAL;
	}
2519 2520 2521 2522 2523

	/*
	 * If the requested region crosses two memslots, we still
	 * verify that the entire region is valid here.
	 */
2524
	for ( ; start_gfn <= end_gfn; start_gfn += nr_pages_avail) {
2525 2526 2527 2528
		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))
2529
			return -EFAULT;
2530 2531 2532
	}

	/* Use the slow path for cross page reads and writes. */
2533
	if (nr_pages_needed == 1)
2534
		ghc->hva += offset;
2535
	else
2536
		ghc->memslot = NULL;
2537

2538 2539 2540
	ghc->gpa = gpa;
	ghc->len = len;
	return 0;
2541
}
2542

2543
int kvm_gfn_to_hva_cache_init(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
2544 2545
			      gpa_t gpa, unsigned long len)
{
2546
	struct kvm_memslots *slots = kvm_memslots(kvm);
2547 2548
	return __kvm_gfn_to_hva_cache_init(slots, ghc, gpa, len);
}
2549
EXPORT_SYMBOL_GPL(kvm_gfn_to_hva_cache_init);
2550

2551
int kvm_write_guest_offset_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
2552 2553
				  void *data, unsigned int offset,
				  unsigned long len)
2554
{
2555
	struct kvm_memslots *slots = kvm_memslots(kvm);
2556
	int r;
2557
	gpa_t gpa = ghc->gpa + offset;
2558

2559
	BUG_ON(len + offset > ghc->len);
2560

2561 2562 2563 2564
	if (slots->generation != ghc->generation) {
		if (__kvm_gfn_to_hva_cache_init(slots, ghc, ghc->gpa, ghc->len))
			return -EFAULT;
	}
2565

2566 2567 2568
	if (kvm_is_error_hva(ghc->hva))
		return -EFAULT;

2569 2570 2571
	if (unlikely(!ghc->memslot))
		return kvm_write_guest(kvm, gpa, data, len);

2572
	r = __copy_to_user((void __user *)ghc->hva + offset, data, len);
2573 2574
	if (r)
		return -EFAULT;
2575
	mark_page_dirty_in_slot(ghc->memslot, gpa >> PAGE_SHIFT);
2576 2577 2578

	return 0;
}
2579
EXPORT_SYMBOL_GPL(kvm_write_guest_offset_cached);
2580

2581 2582
int kvm_write_guest_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
			   void *data, unsigned long len)
2583
{
2584
	return kvm_write_guest_offset_cached(kvm, ghc, data, 0, len);
2585
}
2586
EXPORT_SYMBOL_GPL(kvm_write_guest_cached);
2587

2588 2589 2590
int kvm_read_guest_offset_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
				 void *data, unsigned int offset,
				 unsigned long len)
2591
{
2592
	struct kvm_memslots *slots = kvm_memslots(kvm);
2593
	int r;
2594
	gpa_t gpa = ghc->gpa + offset;
2595

2596
	BUG_ON(len + offset > ghc->len);
2597

2598 2599 2600 2601
	if (slots->generation != ghc->generation) {
		if (__kvm_gfn_to_hva_cache_init(slots, ghc, ghc->gpa, ghc->len))
			return -EFAULT;
	}
2602

2603 2604 2605
	if (kvm_is_error_hva(ghc->hva))
		return -EFAULT;

2606
	if (unlikely(!ghc->memslot))
2607
		return kvm_read_guest(kvm, gpa, data, len);
2608

2609
	r = __copy_from_user(data, (void __user *)ghc->hva + offset, len);
2610 2611 2612 2613 2614
	if (r)
		return -EFAULT;

	return 0;
}
2615 2616 2617 2618 2619 2620 2621
EXPORT_SYMBOL_GPL(kvm_read_guest_offset_cached);

int kvm_read_guest_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
			  void *data, unsigned long len)
{
	return kvm_read_guest_offset_cached(kvm, ghc, data, 0, len);
}
2622
EXPORT_SYMBOL_GPL(kvm_read_guest_cached);
2623

2624 2625
int kvm_clear_guest_page(struct kvm *kvm, gfn_t gfn, int offset, int len)
{
2626 2627 2628
	const void *zero_page = (const void *) __va(page_to_phys(ZERO_PAGE(0)));

	return kvm_write_guest_page(kvm, gfn, zero_page, offset, len);
2629 2630 2631 2632 2633 2634 2635 2636 2637 2638
}
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;

2639
	while ((seg = next_segment(len, offset)) != 0) {
2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650
		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);

2651
void mark_page_dirty_in_slot(struct kvm_memory_slot *memslot, gfn_t gfn)
A
Avi Kivity 已提交
2652
{
R
Rusty Russell 已提交
2653 2654
	if (memslot && memslot->dirty_bitmap) {
		unsigned long rel_gfn = gfn - memslot->base_gfn;
A
Avi Kivity 已提交
2655

2656
		set_bit_le(rel_gfn, memslot->dirty_bitmap);
A
Avi Kivity 已提交
2657 2658
	}
}
2659
EXPORT_SYMBOL_GPL(mark_page_dirty_in_slot);
A
Avi Kivity 已提交
2660

2661 2662 2663 2664 2665
void mark_page_dirty(struct kvm *kvm, gfn_t gfn)
{
	struct kvm_memory_slot *memslot;

	memslot = gfn_to_memslot(kvm, gfn);
2666
	mark_page_dirty_in_slot(memslot, gfn);
2667
}
2668
EXPORT_SYMBOL_GPL(mark_page_dirty);
2669

2670 2671 2672 2673 2674 2675 2676 2677 2678
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);

2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701
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
2702 2703
static void grow_halt_poll_ns(struct kvm_vcpu *vcpu)
{
2704
	unsigned int old, val, grow, grow_start;
W
Wanpeng Li 已提交
2705

2706
	old = val = vcpu->halt_poll_ns;
2707
	grow_start = READ_ONCE(halt_poll_ns_grow_start);
2708
	grow = READ_ONCE(halt_poll_ns_grow);
2709 2710 2711
	if (!grow)
		goto out;

2712 2713 2714
	val *= grow;
	if (val < grow_start)
		val = grow_start;
W
Wanpeng Li 已提交
2715

2716 2717 2718
	if (val > halt_poll_ns)
		val = halt_poll_ns;

W
Wanpeng Li 已提交
2719
	vcpu->halt_poll_ns = val;
2720
out:
2721
	trace_kvm_halt_poll_ns_grow(vcpu->vcpu_id, val, old);
W
Wanpeng Li 已提交
2722 2723 2724 2725
}

static void shrink_halt_poll_ns(struct kvm_vcpu *vcpu)
{
2726
	unsigned int old, val, shrink;
W
Wanpeng Li 已提交
2727

2728
	old = val = vcpu->halt_poll_ns;
2729 2730
	shrink = READ_ONCE(halt_poll_ns_shrink);
	if (shrink == 0)
W
Wanpeng Li 已提交
2731 2732
		val = 0;
	else
2733
		val /= shrink;
W
Wanpeng Li 已提交
2734 2735

	vcpu->halt_poll_ns = val;
2736
	trace_kvm_halt_poll_ns_shrink(vcpu->vcpu_id, val, old);
W
Wanpeng Li 已提交
2737 2738
}

2739 2740
static int kvm_vcpu_check_block(struct kvm_vcpu *vcpu)
{
2741 2742 2743
	int ret = -EINTR;
	int idx = srcu_read_lock(&vcpu->kvm->srcu);

2744 2745
	if (kvm_arch_vcpu_runnable(vcpu)) {
		kvm_make_request(KVM_REQ_UNHALT, vcpu);
2746
		goto out;
2747 2748
	}
	if (kvm_cpu_has_pending_timer(vcpu))
2749
		goto out;
2750
	if (signal_pending(current))
2751
		goto out;
2752

2753 2754 2755 2756
	ret = 0;
out:
	srcu_read_unlock(&vcpu->kvm->srcu, idx);
	return ret;
2757 2758
}

2759 2760 2761 2762 2763 2764 2765 2766 2767
static inline void
update_halt_poll_stats(struct kvm_vcpu *vcpu, u64 poll_ns, bool waited)
{
	if (waited)
		vcpu->stat.halt_poll_fail_ns += poll_ns;
	else
		vcpu->stat.halt_poll_success_ns += poll_ns;
}

E
Eddie Dong 已提交
2768 2769 2770
/*
 * The vCPU has executed a HLT instruction with in-kernel mode enabled.
 */
2771
void kvm_vcpu_block(struct kvm_vcpu *vcpu)
2772
{
2773
	ktime_t start, cur, poll_end;
2774
	bool waited = false;
W
Wanpeng Li 已提交
2775
	u64 block_ns;
2776

2777 2778
	kvm_arch_vcpu_blocking(vcpu);

2779
	start = cur = poll_end = ktime_get();
2780
	if (vcpu->halt_poll_ns && !kvm_arch_no_poll(vcpu)) {
W
Wanpeng Li 已提交
2781
		ktime_t stop = ktime_add_ns(ktime_get(), vcpu->halt_poll_ns);
2782

2783
		++vcpu->stat.halt_attempted_poll;
2784 2785 2786 2787 2788 2789 2790
		do {
			/*
			 * This sets KVM_REQ_UNHALT if an interrupt
			 * arrives.
			 */
			if (kvm_vcpu_check_block(vcpu) < 0) {
				++vcpu->stat.halt_successful_poll;
2791 2792
				if (!vcpu_valid_wakeup(vcpu))
					++vcpu->stat.halt_poll_invalid;
2793 2794
				goto out;
			}
2795
			poll_end = cur = ktime_get();
2796 2797
		} while (single_task_running() && ktime_before(cur, stop));
	}
2798

2799
	prepare_to_rcuwait(&vcpu->wait);
2800
	for (;;) {
2801
		set_current_state(TASK_INTERRUPTIBLE);
2802

2803
		if (kvm_vcpu_check_block(vcpu) < 0)
2804 2805
			break;

2806
		waited = true;
E
Eddie Dong 已提交
2807 2808
		schedule();
	}
2809
	finish_rcuwait(&vcpu->wait);
2810 2811
	cur = ktime_get();
out:
2812
	kvm_arch_vcpu_unblocking(vcpu);
W
Wanpeng Li 已提交
2813 2814
	block_ns = ktime_to_ns(cur) - ktime_to_ns(start);

2815 2816 2817
	update_halt_poll_stats(
		vcpu, ktime_to_ns(ktime_sub(poll_end, start)), waited);

2818 2819
	if (!kvm_arch_no_poll(vcpu)) {
		if (!vcpu_valid_wakeup(vcpu)) {
W
Wanpeng Li 已提交
2820
			shrink_halt_poll_ns(vcpu);
2821
		} else if (vcpu->kvm->max_halt_poll_ns) {
2822 2823 2824
			if (block_ns <= vcpu->halt_poll_ns)
				;
			/* we had a long block, shrink polling */
2825 2826
			else if (vcpu->halt_poll_ns &&
					block_ns > vcpu->kvm->max_halt_poll_ns)
2827 2828
				shrink_halt_poll_ns(vcpu);
			/* we had a short halt and our poll time is too small */
2829 2830
			else if (vcpu->halt_poll_ns < vcpu->kvm->max_halt_poll_ns &&
					block_ns < vcpu->kvm->max_halt_poll_ns)
2831 2832 2833 2834 2835
				grow_halt_poll_ns(vcpu);
		} else {
			vcpu->halt_poll_ns = 0;
		}
	}
W
Wanpeng Li 已提交
2836

2837 2838
	trace_kvm_vcpu_wakeup(block_ns, waited, vcpu_valid_wakeup(vcpu));
	kvm_arch_vcpu_block_finish(vcpu);
E
Eddie Dong 已提交
2839
}
2840
EXPORT_SYMBOL_GPL(kvm_vcpu_block);
E
Eddie Dong 已提交
2841

2842
bool kvm_vcpu_wake_up(struct kvm_vcpu *vcpu)
2843
{
2844
	struct rcuwait *waitp;
2845

2846 2847
	waitp = kvm_arch_vcpu_get_wait(vcpu);
	if (rcuwait_wake_up(waitp)) {
2848
		WRITE_ONCE(vcpu->ready, true);
2849
		++vcpu->stat.halt_wakeup;
2850
		return true;
2851 2852
	}

2853
	return false;
2854 2855 2856
}
EXPORT_SYMBOL_GPL(kvm_vcpu_wake_up);

2857
#ifndef CONFIG_S390
2858 2859 2860 2861 2862 2863 2864 2865
/*
 * 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;

2866 2867 2868
	if (kvm_vcpu_wake_up(vcpu))
		return;

2869 2870 2871 2872 2873 2874
	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();
}
2875
EXPORT_SYMBOL_GPL(kvm_vcpu_kick);
2876
#endif /* !CONFIG_S390 */
2877

2878
int kvm_vcpu_yield_to(struct kvm_vcpu *target)
2879 2880 2881
{
	struct pid *pid;
	struct task_struct *task = NULL;
2882
	int ret = 0;
2883 2884 2885 2886

	rcu_read_lock();
	pid = rcu_dereference(target->pid);
	if (pid)
2887
		task = get_pid_task(pid, PIDTYPE_PID);
2888 2889
	rcu_read_unlock();
	if (!task)
2890 2891
		return ret;
	ret = yield_to(task, 1);
2892
	put_task_struct(task);
2893 2894

	return ret;
2895 2896 2897
}
EXPORT_SYMBOL_GPL(kvm_vcpu_yield_to);

2898 2899 2900 2901 2902 2903
/*
 * 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.
F
Fuad Tabba 已提交
2904
 *  Set at the beginning and cleared at the end of interception/PLE handler.
2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919
 *
 *  (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.
 */
2920
static bool kvm_vcpu_eligible_for_directed_yield(struct kvm_vcpu *vcpu)
2921
{
2922
#ifdef CONFIG_HAVE_KVM_CPU_RELAX_INTERCEPT
2923 2924 2925
	bool eligible;

	eligible = !vcpu->spin_loop.in_spin_loop ||
2926
		    vcpu->spin_loop.dy_eligible;
2927 2928 2929 2930 2931

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

	return eligible;
2932 2933
#else
	return true;
2934
#endif
2935
}
2936

2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959
/*
 * Unlike kvm_arch_vcpu_runnable, this function is called outside
 * a vcpu_load/vcpu_put pair.  However, for most architectures
 * kvm_arch_vcpu_runnable does not require vcpu_load.
 */
bool __weak kvm_arch_dy_runnable(struct kvm_vcpu *vcpu)
{
	return kvm_arch_vcpu_runnable(vcpu);
}

static bool vcpu_dy_runnable(struct kvm_vcpu *vcpu)
{
	if (kvm_arch_dy_runnable(vcpu))
		return true;

#ifdef CONFIG_KVM_ASYNC_PF
	if (!list_empty_careful(&vcpu->async_pf.done))
		return true;
#endif

	return false;
}

2960
void kvm_vcpu_on_spin(struct kvm_vcpu *me, bool yield_to_kernel_mode)
Z
Zhai, Edwin 已提交
2961
{
2962 2963 2964 2965
	struct kvm *kvm = me->kvm;
	struct kvm_vcpu *vcpu;
	int last_boosted_vcpu = me->kvm->last_boosted_vcpu;
	int yielded = 0;
2966
	int try = 3;
2967 2968
	int pass;
	int i;
Z
Zhai, Edwin 已提交
2969

2970
	kvm_vcpu_set_in_spin_loop(me, true);
2971 2972 2973 2974 2975 2976 2977
	/*
	 * 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.
	 */
2978
	for (pass = 0; pass < 2 && !yielded && try; pass++) {
2979
		kvm_for_each_vcpu(i, vcpu, kvm) {
2980
			if (!pass && i <= last_boosted_vcpu) {
2981 2982 2983 2984
				i = last_boosted_vcpu;
				continue;
			} else if (pass && i > last_boosted_vcpu)
				break;
2985
			if (!READ_ONCE(vcpu->ready))
2986
				continue;
2987 2988
			if (vcpu == me)
				continue;
2989 2990
			if (rcuwait_active(&vcpu->wait) &&
			    !vcpu_dy_runnable(vcpu))
2991
				continue;
2992 2993
			if (READ_ONCE(vcpu->preempted) && yield_to_kernel_mode &&
				!kvm_arch_vcpu_in_kernel(vcpu))
2994
				continue;
2995 2996
			if (!kvm_vcpu_eligible_for_directed_yield(vcpu))
				continue;
2997 2998 2999

			yielded = kvm_vcpu_yield_to(vcpu);
			if (yielded > 0) {
3000 3001
				kvm->last_boosted_vcpu = i;
				break;
3002 3003 3004 3005
			} else if (yielded < 0) {
				try--;
				if (!try)
					break;
3006 3007 3008
			}
		}
	}
3009
	kvm_vcpu_set_in_spin_loop(me, false);
3010 3011 3012

	/* Ensure vcpu is not eligible during next spinloop */
	kvm_vcpu_set_dy_eligible(me, false);
Z
Zhai, Edwin 已提交
3013 3014 3015
}
EXPORT_SYMBOL_GPL(kvm_vcpu_on_spin);

3016
static vm_fault_t kvm_vcpu_fault(struct vm_fault *vmf)
3017
{
3018
	struct kvm_vcpu *vcpu = vmf->vma->vm_file->private_data;
3019 3020
	struct page *page;

3021
	if (vmf->pgoff == 0)
3022
		page = virt_to_page(vcpu->run);
A
Avi Kivity 已提交
3023
#ifdef CONFIG_X86
3024
	else if (vmf->pgoff == KVM_PIO_PAGE_OFFSET)
3025
		page = virt_to_page(vcpu->arch.pio_data);
3026
#endif
3027
#ifdef CONFIG_KVM_MMIO
3028 3029
	else if (vmf->pgoff == KVM_COALESCED_MMIO_PAGE_OFFSET)
		page = virt_to_page(vcpu->kvm->coalesced_mmio_ring);
A
Avi Kivity 已提交
3030
#endif
3031
	else
3032
		return kvm_arch_vcpu_fault(vcpu, vmf);
3033
	get_page(page);
3034 3035
	vmf->page = page;
	return 0;
3036 3037
}

3038
static const struct vm_operations_struct kvm_vcpu_vm_ops = {
3039
	.fault = kvm_vcpu_fault,
3040 3041 3042 3043 3044 3045 3046 3047
};

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
3048 3049 3050 3051
static int kvm_vcpu_release(struct inode *inode, struct file *filp)
{
	struct kvm_vcpu *vcpu = filp->private_data;

A
Al Viro 已提交
3052
	kvm_put_kvm(vcpu->kvm);
A
Avi Kivity 已提交
3053 3054 3055
	return 0;
}

3056
static struct file_operations kvm_vcpu_fops = {
A
Avi Kivity 已提交
3057 3058
	.release        = kvm_vcpu_release,
	.unlocked_ioctl = kvm_vcpu_ioctl,
3059
	.mmap           = kvm_vcpu_mmap,
3060
	.llseek		= noop_llseek,
3061
	KVM_COMPAT(kvm_vcpu_compat_ioctl),
A
Avi Kivity 已提交
3062 3063 3064 3065 3066 3067 3068
};

/*
 * Allocates an inode for the vcpu.
 */
static int create_vcpu_fd(struct kvm_vcpu *vcpu)
{
3069 3070 3071 3072
	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
3073 3074
}

3075
static void kvm_create_vcpu_debugfs(struct kvm_vcpu *vcpu)
3076
{
3077
#ifdef __KVM_HAVE_ARCH_VCPU_DEBUGFS
3078
	struct dentry *debugfs_dentry;
3079 3080 3081
	char dir_name[ITOA_MAX_LEN * 2];

	if (!debugfs_initialized())
3082
		return;
3083 3084

	snprintf(dir_name, sizeof(dir_name), "vcpu%d", vcpu->vcpu_id);
3085 3086
	debugfs_dentry = debugfs_create_dir(dir_name,
					    vcpu->kvm->debugfs_dentry);
3087

3088
	kvm_arch_create_vcpu_debugfs(vcpu, debugfs_dentry);
3089
#endif
3090 3091
}

3092 3093 3094
/*
 * Creates some virtual cpus.  Good luck creating more than one.
 */
3095
static int kvm_vm_ioctl_create_vcpu(struct kvm *kvm, u32 id)
3096 3097
{
	int r;
3098
	struct kvm_vcpu *vcpu;
3099
	struct page *page;
3100

G
Greg Kurz 已提交
3101
	if (id >= KVM_MAX_VCPU_ID)
3102 3103
		return -EINVAL;

3104 3105 3106 3107 3108 3109 3110 3111 3112
	mutex_lock(&kvm->lock);
	if (kvm->created_vcpus == KVM_MAX_VCPUS) {
		mutex_unlock(&kvm->lock);
		return -EINVAL;
	}

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

3113 3114 3115 3116
	r = kvm_arch_vcpu_precreate(kvm, id);
	if (r)
		goto vcpu_decrement;

3117 3118 3119
	vcpu = kmem_cache_zalloc(kvm_vcpu_cache, GFP_KERNEL);
	if (!vcpu) {
		r = -ENOMEM;
3120 3121
		goto vcpu_decrement;
	}
3122

3123
	BUILD_BUG_ON(sizeof(struct kvm_run) > PAGE_SIZE);
3124 3125 3126
	page = alloc_page(GFP_KERNEL | __GFP_ZERO);
	if (!page) {
		r = -ENOMEM;
3127
		goto vcpu_free;
3128 3129 3130 3131
	}
	vcpu->run = page_address(page);

	kvm_vcpu_init(vcpu, kvm, id);
3132 3133 3134

	r = kvm_arch_vcpu_create(vcpu);
	if (r)
3135
		goto vcpu_free_run_page;
3136

S
Shaohua Li 已提交
3137
	mutex_lock(&kvm->lock);
3138 3139 3140 3141
	if (kvm_get_vcpu_by_id(kvm, id)) {
		r = -EEXIST;
		goto unlock_vcpu_destroy;
	}
3142

3143 3144
	vcpu->vcpu_idx = atomic_read(&kvm->online_vcpus);
	BUG_ON(kvm->vcpus[vcpu->vcpu_idx]);
3145

R
Rusty Russell 已提交
3146
	/* Now it's all set up, let userspace reach it */
A
Al Viro 已提交
3147
	kvm_get_kvm(kvm);
A
Avi Kivity 已提交
3148
	r = create_vcpu_fd(vcpu);
3149
	if (r < 0) {
3150
		kvm_put_kvm_no_destroy(kvm);
3151
		goto unlock_vcpu_destroy;
3152 3153
	}

3154
	kvm->vcpus[vcpu->vcpu_idx] = vcpu;
3155 3156 3157 3158 3159

	/*
	 * Pairs with smp_rmb() in kvm_get_vcpu.  Write kvm->vcpus
	 * before kvm->online_vcpu's incremented value.
	 */
3160 3161 3162 3163
	smp_wmb();
	atomic_inc(&kvm->online_vcpus);

	mutex_unlock(&kvm->lock);
3164
	kvm_arch_vcpu_postcreate(vcpu);
3165
	kvm_create_vcpu_debugfs(vcpu);
R
Rusty Russell 已提交
3166
	return r;
3167

3168
unlock_vcpu_destroy:
3169
	mutex_unlock(&kvm->lock);
3170
	kvm_arch_vcpu_destroy(vcpu);
3171 3172
vcpu_free_run_page:
	free_page((unsigned long)vcpu->run);
3173 3174
vcpu_free:
	kmem_cache_free(kvm_vcpu_cache, vcpu);
3175 3176 3177 3178
vcpu_decrement:
	mutex_lock(&kvm->lock);
	kvm->created_vcpus--;
	mutex_unlock(&kvm->lock);
3179 3180 3181
	return r;
}

A
Avi Kivity 已提交
3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192
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
3193 3194
static long kvm_vcpu_ioctl(struct file *filp,
			   unsigned int ioctl, unsigned long arg)
A
Avi Kivity 已提交
3195
{
A
Avi Kivity 已提交
3196
	struct kvm_vcpu *vcpu = filp->private_data;
A
Al Viro 已提交
3197
	void __user *argp = (void __user *)arg;
3198
	int r;
3199 3200
	struct kvm_fpu *fpu = NULL;
	struct kvm_sregs *kvm_sregs = NULL;
A
Avi Kivity 已提交
3201

3202 3203
	if (vcpu->kvm->mm != current->mm)
		return -EIO;
3204

3205 3206 3207
	if (unlikely(_IOC_TYPE(ioctl) != KVMIO))
		return -EINVAL;

3208
	/*
3209 3210
	 * Some architectures have vcpu ioctls that are asynchronous to vcpu
	 * execution; mutex_lock() would break them.
3211
	 */
3212 3213
	r = kvm_arch_vcpu_async_ioctl(filp, ioctl, arg);
	if (r != -ENOIOCTLCMD)
3214
		return r;
3215

3216 3217
	if (mutex_lock_killable(&vcpu->mutex))
		return -EINTR;
A
Avi Kivity 已提交
3218
	switch (ioctl) {
3219 3220
	case KVM_RUN: {
		struct pid *oldpid;
3221 3222 3223
		r = -EINVAL;
		if (arg)
			goto out;
3224
		oldpid = rcu_access_pointer(vcpu->pid);
3225
		if (unlikely(oldpid != task_pid(current))) {
3226
			/* The thread running this VCPU changed. */
3227
			struct pid *newpid;
3228

3229 3230 3231 3232 3233
			r = kvm_arch_vcpu_run_pid_change(vcpu);
			if (r)
				break;

			newpid = get_task_pid(current, PIDTYPE_PID);
3234 3235 3236 3237
			rcu_assign_pointer(vcpu->pid, newpid);
			if (oldpid)
				synchronize_rcu();
			put_pid(oldpid);
3238 3239 3240
#if defined(CONFIG_X86) || defined(CONFIG_ARM64)
			vcpu->stat.pid = current->pid;
#endif /* defined(CONFIG_X86) || defined (CONFIG_ARM64) */
3241
		}
3242
		r = kvm_arch_vcpu_ioctl_run(vcpu);
3243
		trace_kvm_userspace_exit(vcpu->run->exit_reason, r);
A
Avi Kivity 已提交
3244
		break;
3245
	}
A
Avi Kivity 已提交
3246
	case KVM_GET_REGS: {
3247
		struct kvm_regs *kvm_regs;
A
Avi Kivity 已提交
3248

3249
		r = -ENOMEM;
3250
		kvm_regs = kzalloc(sizeof(struct kvm_regs), GFP_KERNEL_ACCOUNT);
3251
		if (!kvm_regs)
A
Avi Kivity 已提交
3252
			goto out;
3253 3254 3255
		r = kvm_arch_vcpu_ioctl_get_regs(vcpu, kvm_regs);
		if (r)
			goto out_free1;
A
Avi Kivity 已提交
3256
		r = -EFAULT;
3257 3258
		if (copy_to_user(argp, kvm_regs, sizeof(struct kvm_regs)))
			goto out_free1;
A
Avi Kivity 已提交
3259
		r = 0;
3260 3261
out_free1:
		kfree(kvm_regs);
A
Avi Kivity 已提交
3262 3263 3264
		break;
	}
	case KVM_SET_REGS: {
3265
		struct kvm_regs *kvm_regs;
A
Avi Kivity 已提交
3266

3267 3268 3269
		kvm_regs = memdup_user(argp, sizeof(*kvm_regs));
		if (IS_ERR(kvm_regs)) {
			r = PTR_ERR(kvm_regs);
A
Avi Kivity 已提交
3270
			goto out;
3271
		}
3272 3273
		r = kvm_arch_vcpu_ioctl_set_regs(vcpu, kvm_regs);
		kfree(kvm_regs);
A
Avi Kivity 已提交
3274 3275 3276
		break;
	}
	case KVM_GET_SREGS: {
3277 3278
		kvm_sregs = kzalloc(sizeof(struct kvm_sregs),
				    GFP_KERNEL_ACCOUNT);
3279 3280 3281 3282
		r = -ENOMEM;
		if (!kvm_sregs)
			goto out;
		r = kvm_arch_vcpu_ioctl_get_sregs(vcpu, kvm_sregs);
A
Avi Kivity 已提交
3283 3284 3285
		if (r)
			goto out;
		r = -EFAULT;
3286
		if (copy_to_user(argp, kvm_sregs, sizeof(struct kvm_sregs)))
A
Avi Kivity 已提交
3287 3288 3289 3290 3291
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_SREGS: {
3292 3293 3294
		kvm_sregs = memdup_user(argp, sizeof(*kvm_sregs));
		if (IS_ERR(kvm_sregs)) {
			r = PTR_ERR(kvm_sregs);
G
Guo Chao 已提交
3295
			kvm_sregs = NULL;
A
Avi Kivity 已提交
3296
			goto out;
3297
		}
3298
		r = kvm_arch_vcpu_ioctl_set_sregs(vcpu, kvm_sregs);
A
Avi Kivity 已提交
3299 3300
		break;
	}
3301 3302 3303 3304 3305 3306 3307
	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;
3308
		if (copy_to_user(argp, &mp_state, sizeof(mp_state)))
3309 3310 3311 3312 3313 3314 3315 3316
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_MP_STATE: {
		struct kvm_mp_state mp_state;

		r = -EFAULT;
3317
		if (copy_from_user(&mp_state, argp, sizeof(mp_state)))
3318 3319 3320 3321
			goto out;
		r = kvm_arch_vcpu_ioctl_set_mpstate(vcpu, &mp_state);
		break;
	}
A
Avi Kivity 已提交
3322 3323 3324 3325
	case KVM_TRANSLATE: {
		struct kvm_translation tr;

		r = -EFAULT;
3326
		if (copy_from_user(&tr, argp, sizeof(tr)))
A
Avi Kivity 已提交
3327
			goto out;
3328
		r = kvm_arch_vcpu_ioctl_translate(vcpu, &tr);
A
Avi Kivity 已提交
3329 3330 3331
		if (r)
			goto out;
		r = -EFAULT;
3332
		if (copy_to_user(argp, &tr, sizeof(tr)))
A
Avi Kivity 已提交
3333 3334 3335 3336
			goto out;
		r = 0;
		break;
	}
J
Jan Kiszka 已提交
3337 3338
	case KVM_SET_GUEST_DEBUG: {
		struct kvm_guest_debug dbg;
A
Avi Kivity 已提交
3339 3340

		r = -EFAULT;
3341
		if (copy_from_user(&dbg, argp, sizeof(dbg)))
A
Avi Kivity 已提交
3342
			goto out;
J
Jan Kiszka 已提交
3343
		r = kvm_arch_vcpu_ioctl_set_guest_debug(vcpu, &dbg);
A
Avi Kivity 已提交
3344 3345
		break;
	}
A
Avi Kivity 已提交
3346 3347 3348 3349 3350 3351 3352 3353 3354
	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,
3355
					   sizeof(kvm_sigmask)))
A
Avi Kivity 已提交
3356 3357
				goto out;
			r = -EINVAL;
3358
			if (kvm_sigmask.len != sizeof(sigset))
A
Avi Kivity 已提交
3359 3360 3361
				goto out;
			r = -EFAULT;
			if (copy_from_user(&sigset, sigmask_arg->sigset,
3362
					   sizeof(sigset)))
A
Avi Kivity 已提交
3363 3364 3365
				goto out;
			p = &sigset;
		}
3366
		r = kvm_vcpu_ioctl_set_sigmask(vcpu, p);
A
Avi Kivity 已提交
3367 3368
		break;
	}
A
Avi Kivity 已提交
3369
	case KVM_GET_FPU: {
3370
		fpu = kzalloc(sizeof(struct kvm_fpu), GFP_KERNEL_ACCOUNT);
3371 3372 3373 3374
		r = -ENOMEM;
		if (!fpu)
			goto out;
		r = kvm_arch_vcpu_ioctl_get_fpu(vcpu, fpu);
A
Avi Kivity 已提交
3375 3376 3377
		if (r)
			goto out;
		r = -EFAULT;
3378
		if (copy_to_user(argp, fpu, sizeof(struct kvm_fpu)))
A
Avi Kivity 已提交
3379 3380 3381 3382 3383
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_FPU: {
3384 3385 3386
		fpu = memdup_user(argp, sizeof(*fpu));
		if (IS_ERR(fpu)) {
			r = PTR_ERR(fpu);
G
Guo Chao 已提交
3387
			fpu = NULL;
A
Avi Kivity 已提交
3388
			goto out;
3389
		}
3390
		r = kvm_arch_vcpu_ioctl_set_fpu(vcpu, fpu);
A
Avi Kivity 已提交
3391 3392
		break;
	}
A
Avi Kivity 已提交
3393
	default:
3394
		r = kvm_arch_vcpu_ioctl(filp, ioctl, arg);
A
Avi Kivity 已提交
3395 3396
	}
out:
3397
	mutex_unlock(&vcpu->mutex);
3398 3399
	kfree(fpu);
	kfree(kvm_sregs);
A
Avi Kivity 已提交
3400 3401 3402
	return r;
}

3403
#ifdef CONFIG_KVM_COMPAT
3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422
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,
3423
					   sizeof(kvm_sigmask)))
3424 3425
				goto out;
			r = -EINVAL;
A
Al Viro 已提交
3426
			if (kvm_sigmask.len != sizeof(compat_sigset_t))
3427 3428
				goto out;
			r = -EFAULT;
3429 3430
			if (get_compat_sigset(&sigset,
					      (compat_sigset_t __user *)sigmask_arg->sigset))
3431
				goto out;
3432 3433 3434
			r = kvm_vcpu_ioctl_set_sigmask(vcpu, &sigset);
		} else
			r = kvm_vcpu_ioctl_set_sigmask(vcpu, NULL);
3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445
		break;
	}
	default:
		r = kvm_vcpu_ioctl(filp, ioctl, arg);
	}

out:
	return r;
}
#endif

3446 3447 3448 3449 3450 3451 3452 3453 3454 3455
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
3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476
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;

3477 3478 3479
	if (dev->kvm->mm != current->mm)
		return -EIO;

S
Scott Wood 已提交
3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499
	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;

3500 3501 3502 3503 3504 3505 3506
	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
3507 3508 3509 3510 3511 3512 3513
	kvm_put_kvm(kvm);
	return 0;
}

static const struct file_operations kvm_device_fops = {
	.unlocked_ioctl = kvm_device_ioctl,
	.release = kvm_device_release,
3514
	KVM_COMPAT(kvm_device_ioctl),
3515
	.mmap = kvm_device_mmap,
S
Scott Wood 已提交
3516 3517 3518 3519 3520 3521 3522 3523 3524 3525
};

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

	return filp->private_data;
}

3526
static const struct kvm_device_ops *kvm_device_ops_table[KVM_DEV_TYPE_MAX] = {
3527
#ifdef CONFIG_KVM_MPIC
3528 3529
	[KVM_DEV_TYPE_FSL_MPIC_20]	= &kvm_mpic_ops,
	[KVM_DEV_TYPE_FSL_MPIC_42]	= &kvm_mpic_ops,
3530
#endif
3531 3532
};

3533
int kvm_register_device_ops(const struct kvm_device_ops *ops, u32 type)
3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544
{
	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;
}

3545 3546 3547 3548 3549 3550
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
3551 3552 3553
static int kvm_ioctl_create_device(struct kvm *kvm,
				   struct kvm_create_device *cd)
{
3554
	const struct kvm_device_ops *ops = NULL;
S
Scott Wood 已提交
3555 3556
	struct kvm_device *dev;
	bool test = cd->flags & KVM_CREATE_DEVICE_TEST;
P
Paolo Bonzini 已提交
3557
	int type;
S
Scott Wood 已提交
3558 3559
	int ret;

3560 3561 3562
	if (cd->type >= ARRAY_SIZE(kvm_device_ops_table))
		return -ENODEV;

P
Paolo Bonzini 已提交
3563 3564
	type = array_index_nospec(cd->type, ARRAY_SIZE(kvm_device_ops_table));
	ops = kvm_device_ops_table[type];
3565
	if (ops == NULL)
S
Scott Wood 已提交
3566 3567 3568 3569 3570
		return -ENODEV;

	if (test)
		return 0;

3571
	dev = kzalloc(sizeof(*dev), GFP_KERNEL_ACCOUNT);
S
Scott Wood 已提交
3572 3573 3574 3575 3576 3577
	if (!dev)
		return -ENOMEM;

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

3578
	mutex_lock(&kvm->lock);
P
Paolo Bonzini 已提交
3579
	ret = ops->create(dev, type);
S
Scott Wood 已提交
3580
	if (ret < 0) {
3581
		mutex_unlock(&kvm->lock);
S
Scott Wood 已提交
3582 3583 3584
		kfree(dev);
		return ret;
	}
3585 3586
	list_add(&dev->vm_node, &kvm->devices);
	mutex_unlock(&kvm->lock);
S
Scott Wood 已提交
3587

3588 3589 3590
	if (ops->init)
		ops->init(dev);

3591
	kvm_get_kvm(kvm);
3592
	ret = anon_inode_getfd(ops->name, &kvm_device_fops, dev, O_RDWR | O_CLOEXEC);
S
Scott Wood 已提交
3593
	if (ret < 0) {
3594
		kvm_put_kvm_no_destroy(kvm);
3595 3596 3597
		mutex_lock(&kvm->lock);
		list_del(&dev->vm_node);
		mutex_unlock(&kvm->lock);
3598
		ops->destroy(dev);
S
Scott Wood 已提交
3599 3600 3601 3602 3603 3604 3605
		return ret;
	}

	cd->fd = ret;
	return 0;
}

3606 3607 3608 3609 3610 3611 3612 3613 3614 3615
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
3616
#ifdef CONFIG_HAVE_KVM_IRQFD
3617
	case KVM_CAP_IRQFD:
3618 3619
	case KVM_CAP_IRQFD_RESAMPLE:
#endif
3620
	case KVM_CAP_IOEVENTFD_ANY_LENGTH:
3621
	case KVM_CAP_CHECK_EXTENSION_VM:
3622
	case KVM_CAP_ENABLE_CAP_VM:
3623
	case KVM_CAP_HALT_POLL:
3624
		return 1;
3625
#ifdef CONFIG_KVM_MMIO
3626 3627
	case KVM_CAP_COALESCED_MMIO:
		return KVM_COALESCED_MMIO_PAGE_OFFSET;
P
Peng Hao 已提交
3628 3629
	case KVM_CAP_COALESCED_PIO:
		return 1;
3630
#endif
3631 3632 3633 3634
#ifdef CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT
	case KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2:
		return KVM_DIRTY_LOG_MANUAL_CAPS;
#endif
3635 3636 3637
#ifdef CONFIG_HAVE_KVM_IRQ_ROUTING
	case KVM_CAP_IRQ_ROUTING:
		return KVM_MAX_IRQ_ROUTES;
3638 3639 3640 3641
#endif
#if KVM_ADDRESS_SPACE_NUM > 1
	case KVM_CAP_MULTI_ADDRESS_SPACE:
		return KVM_ADDRESS_SPACE_NUM;
3642
#endif
3643 3644
	case KVM_CAP_NR_MEMSLOTS:
		return KVM_USER_MEM_SLOTS;
3645 3646 3647 3648 3649 3650
	default:
		break;
	}
	return kvm_vm_ioctl_check_extension(kvm, arg);
}

3651 3652 3653 3654 3655 3656 3657 3658 3659 3660
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) {
3661
#ifdef CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT
3662 3663 3664 3665 3666 3667 3668
	case KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2: {
		u64 allowed_options = KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE;

		if (cap->args[0] & KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE)
			allowed_options = KVM_DIRTY_LOG_MANUAL_CAPS;

		if (cap->flags || (cap->args[0] & ~allowed_options))
3669 3670 3671
			return -EINVAL;
		kvm->manual_dirty_log_protect = cap->args[0];
		return 0;
3672
	}
3673
#endif
3674 3675 3676 3677 3678 3679 3680
	case KVM_CAP_HALT_POLL: {
		if (cap->flags || cap->args[0] != (unsigned int)cap->args[0])
			return -EINVAL;

		kvm->max_halt_poll_ns = cap->args[0];
		return 0;
	}
3681 3682 3683 3684 3685
	default:
		return kvm_vm_ioctl_enable_cap(kvm, cap);
	}
}

A
Avi Kivity 已提交
3686 3687 3688 3689 3690
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;
3691
	int r;
A
Avi Kivity 已提交
3692

3693 3694
	if (kvm->mm != current->mm)
		return -EIO;
A
Avi Kivity 已提交
3695 3696 3697 3698
	switch (ioctl) {
	case KVM_CREATE_VCPU:
		r = kvm_vm_ioctl_create_vcpu(kvm, arg);
		break;
3699 3700 3701 3702 3703 3704 3705 3706 3707
	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;
	}
3708 3709 3710 3711 3712
	case KVM_SET_USER_MEMORY_REGION: {
		struct kvm_userspace_memory_region kvm_userspace_mem;

		r = -EFAULT;
		if (copy_from_user(&kvm_userspace_mem, argp,
3713
						sizeof(kvm_userspace_mem)))
3714 3715
			goto out;

3716
		r = kvm_vm_ioctl_set_memory_region(kvm, &kvm_userspace_mem);
A
Avi Kivity 已提交
3717 3718 3719 3720 3721 3722
		break;
	}
	case KVM_GET_DIRTY_LOG: {
		struct kvm_dirty_log log;

		r = -EFAULT;
3723
		if (copy_from_user(&log, argp, sizeof(log)))
A
Avi Kivity 已提交
3724
			goto out;
3725
		r = kvm_vm_ioctl_get_dirty_log(kvm, &log);
A
Avi Kivity 已提交
3726 3727
		break;
	}
3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738
#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
3739
#ifdef CONFIG_KVM_MMIO
3740 3741
	case KVM_REGISTER_COALESCED_MMIO: {
		struct kvm_coalesced_mmio_zone zone;
3742

3743
		r = -EFAULT;
3744
		if (copy_from_user(&zone, argp, sizeof(zone)))
3745 3746 3747 3748 3749 3750
			goto out;
		r = kvm_vm_ioctl_register_coalesced_mmio(kvm, &zone);
		break;
	}
	case KVM_UNREGISTER_COALESCED_MMIO: {
		struct kvm_coalesced_mmio_zone zone;
3751

3752
		r = -EFAULT;
3753
		if (copy_from_user(&zone, argp, sizeof(zone)))
3754 3755 3756 3757 3758
			goto out;
		r = kvm_vm_ioctl_unregister_coalesced_mmio(kvm, &zone);
		break;
	}
#endif
G
Gregory Haskins 已提交
3759 3760 3761 3762
	case KVM_IRQFD: {
		struct kvm_irqfd data;

		r = -EFAULT;
3763
		if (copy_from_user(&data, argp, sizeof(data)))
G
Gregory Haskins 已提交
3764
			goto out;
3765
		r = kvm_irqfd(kvm, &data);
G
Gregory Haskins 已提交
3766 3767
		break;
	}
G
Gregory Haskins 已提交
3768 3769 3770 3771
	case KVM_IOEVENTFD: {
		struct kvm_ioeventfd data;

		r = -EFAULT;
3772
		if (copy_from_user(&data, argp, sizeof(data)))
G
Gregory Haskins 已提交
3773 3774 3775 3776
			goto out;
		r = kvm_ioeventfd(kvm, &data);
		break;
	}
3777 3778 3779 3780 3781
#ifdef CONFIG_HAVE_KVM_MSI
	case KVM_SIGNAL_MSI: {
		struct kvm_msi msi;

		r = -EFAULT;
3782
		if (copy_from_user(&msi, argp, sizeof(msi)))
3783 3784 3785 3786
			goto out;
		r = kvm_send_userspace_msi(kvm, &msi);
		break;
	}
3787 3788 3789 3790 3791 3792 3793
#endif
#ifdef __KVM_HAVE_IRQ_LINE
	case KVM_IRQ_LINE_STATUS:
	case KVM_IRQ_LINE: {
		struct kvm_irq_level irq_event;

		r = -EFAULT;
3794
		if (copy_from_user(&irq_event, argp, sizeof(irq_event)))
3795 3796
			goto out;

3797 3798
		r = kvm_vm_ioctl_irq_line(kvm, &irq_event,
					ioctl == KVM_IRQ_LINE_STATUS);
3799 3800 3801 3802 3803
		if (r)
			goto out;

		r = -EFAULT;
		if (ioctl == KVM_IRQ_LINE_STATUS) {
3804
			if (copy_to_user(argp, &irq_event, sizeof(irq_event)))
3805 3806 3807 3808 3809 3810
				goto out;
		}

		r = 0;
		break;
	}
3811
#endif
3812 3813 3814 3815
#ifdef CONFIG_HAVE_KVM_IRQ_ROUTING
	case KVM_SET_GSI_ROUTING: {
		struct kvm_irq_routing routing;
		struct kvm_irq_routing __user *urouting;
3816
		struct kvm_irq_routing_entry *entries = NULL;
3817 3818 3819 3820 3821

		r = -EFAULT;
		if (copy_from_user(&routing, argp, sizeof(routing)))
			goto out;
		r = -EINVAL;
3822 3823
		if (!kvm_arch_can_set_irq_routing(kvm))
			goto out;
3824
		if (routing.nr > KVM_MAX_IRQ_ROUTES)
3825 3826 3827
			goto out;
		if (routing.flags)
			goto out;
3828 3829
		if (routing.nr) {
			urouting = argp;
D
Denis Efremov 已提交
3830 3831 3832 3833 3834 3835 3836
			entries = vmemdup_user(urouting->entries,
					       array_size(sizeof(*entries),
							  routing.nr));
			if (IS_ERR(entries)) {
				r = PTR_ERR(entries);
				goto out;
			}
3837
		}
3838 3839
		r = kvm_set_irq_routing(kvm, entries, routing.nr,
					routing.flags);
D
Denis Efremov 已提交
3840
		kvfree(entries);
3841 3842 3843
		break;
	}
#endif /* CONFIG_HAVE_KVM_IRQ_ROUTING */
S
Scott Wood 已提交
3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861
	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;
	}
3862 3863 3864
	case KVM_CHECK_EXTENSION:
		r = kvm_vm_ioctl_check_extension_generic(kvm, arg);
		break;
3865
	default:
3866
		r = kvm_arch_vm_ioctl(filp, ioctl, arg);
3867 3868 3869 3870 3871
	}
out:
	return r;
}

3872
#ifdef CONFIG_KVM_COMPAT
3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896
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)))
3897
			return -EFAULT;
3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912
		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

3913
static struct file_operations kvm_vm_fops = {
3914 3915
	.release        = kvm_vm_release,
	.unlocked_ioctl = kvm_vm_ioctl,
3916
	.llseek		= noop_llseek,
3917
	KVM_COMPAT(kvm_vm_compat_ioctl),
3918 3919
};

3920
static int kvm_dev_ioctl_create_vm(unsigned long type)
3921
{
3922
	int r;
3923
	struct kvm *kvm;
3924
	struct file *file;
3925

3926
	kvm = kvm_create_vm(type);
3927 3928
	if (IS_ERR(kvm))
		return PTR_ERR(kvm);
3929
#ifdef CONFIG_KVM_MMIO
3930
	r = kvm_coalesced_mmio_init(kvm);
3931 3932
	if (r < 0)
		goto put_kvm;
3933
#endif
3934
	r = get_unused_fd_flags(O_CLOEXEC);
3935 3936 3937
	if (r < 0)
		goto put_kvm;

3938 3939 3940
	file = anon_inode_getfile("kvm-vm", &kvm_vm_fops, kvm, O_RDWR);
	if (IS_ERR(file)) {
		put_unused_fd(r);
3941 3942
		r = PTR_ERR(file);
		goto put_kvm;
3943
	}
3944

3945 3946 3947 3948 3949 3950
	/*
	 * 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).
	 */
3951
	if (kvm_create_vm_debugfs(kvm, r) < 0) {
3952 3953
		put_unused_fd(r);
		fput(file);
3954 3955
		return -ENOMEM;
	}
3956
	kvm_uevent_notify_change(KVM_EVENT_CREATE_VM, kvm);
3957

3958
	fd_install(r, file);
3959
	return r;
3960 3961 3962 3963

put_kvm:
	kvm_put_kvm(kvm);
	return r;
3964 3965 3966 3967 3968
}

static long kvm_dev_ioctl(struct file *filp,
			  unsigned int ioctl, unsigned long arg)
{
3969
	long r = -EINVAL;
3970 3971 3972

	switch (ioctl) {
	case KVM_GET_API_VERSION:
3973 3974
		if (arg)
			goto out;
3975 3976 3977
		r = KVM_API_VERSION;
		break;
	case KVM_CREATE_VM:
3978
		r = kvm_dev_ioctl_create_vm(arg);
3979
		break;
3980
	case KVM_CHECK_EXTENSION:
3981
		r = kvm_vm_ioctl_check_extension_generic(NULL, arg);
3982
		break;
3983 3984 3985
	case KVM_GET_VCPU_MMAP_SIZE:
		if (arg)
			goto out;
3986 3987 3988
		r = PAGE_SIZE;     /* struct kvm_run */
#ifdef CONFIG_X86
		r += PAGE_SIZE;    /* pio data page */
3989
#endif
3990
#ifdef CONFIG_KVM_MMIO
3991
		r += PAGE_SIZE;    /* coalesced mmio ring page */
3992
#endif
3993
		break;
3994 3995 3996
	case KVM_TRACE_ENABLE:
	case KVM_TRACE_PAUSE:
	case KVM_TRACE_DISABLE:
3997
		r = -EOPNOTSUPP;
3998
		break;
A
Avi Kivity 已提交
3999
	default:
4000
		return kvm_arch_dev_ioctl(filp, ioctl, arg);
A
Avi Kivity 已提交
4001 4002 4003 4004 4005 4006 4007
	}
out:
	return r;
}

static struct file_operations kvm_chardev_ops = {
	.unlocked_ioctl = kvm_dev_ioctl,
4008
	.llseek		= noop_llseek,
4009
	KVM_COMPAT(kvm_dev_ioctl),
A
Avi Kivity 已提交
4010 4011 4012
};

static struct miscdevice kvm_dev = {
A
Avi Kivity 已提交
4013
	KVM_MINOR,
A
Avi Kivity 已提交
4014 4015 4016 4017
	"kvm",
	&kvm_chardev_ops,
};

4018
static void hardware_enable_nolock(void *junk)
4019 4020
{
	int cpu = raw_smp_processor_id();
4021
	int r;
4022

4023
	if (cpumask_test_cpu(cpu, cpus_hardware_enabled))
4024
		return;
4025

4026
	cpumask_set_cpu(cpu, cpus_hardware_enabled);
4027

4028
	r = kvm_arch_hardware_enable();
4029 4030 4031 4032

	if (r) {
		cpumask_clear_cpu(cpu, cpus_hardware_enabled);
		atomic_inc(&hardware_enable_failed);
X
Xiubo Li 已提交
4033
		pr_info("kvm: enabling virtualization on CPU%d failed\n", cpu);
4034
	}
4035 4036
}

4037
static int kvm_starting_cpu(unsigned int cpu)
4038
{
4039
	raw_spin_lock(&kvm_count_lock);
4040 4041
	if (kvm_usage_count)
		hardware_enable_nolock(NULL);
4042
	raw_spin_unlock(&kvm_count_lock);
4043
	return 0;
4044 4045 4046
}

static void hardware_disable_nolock(void *junk)
4047 4048 4049
{
	int cpu = raw_smp_processor_id();

4050
	if (!cpumask_test_cpu(cpu, cpus_hardware_enabled))
4051
		return;
4052
	cpumask_clear_cpu(cpu, cpus_hardware_enabled);
4053
	kvm_arch_hardware_disable();
4054 4055
}

4056
static int kvm_dying_cpu(unsigned int cpu)
4057
{
4058
	raw_spin_lock(&kvm_count_lock);
4059 4060
	if (kvm_usage_count)
		hardware_disable_nolock(NULL);
4061
	raw_spin_unlock(&kvm_count_lock);
4062
	return 0;
4063 4064
}

4065 4066 4067 4068 4069 4070
static void hardware_disable_all_nolock(void)
{
	BUG_ON(!kvm_usage_count);

	kvm_usage_count--;
	if (!kvm_usage_count)
4071
		on_each_cpu(hardware_disable_nolock, NULL, 1);
4072 4073 4074 4075
}

static void hardware_disable_all(void)
{
4076
	raw_spin_lock(&kvm_count_lock);
4077
	hardware_disable_all_nolock();
4078
	raw_spin_unlock(&kvm_count_lock);
4079 4080 4081 4082 4083 4084
}

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

4085
	raw_spin_lock(&kvm_count_lock);
4086 4087 4088 4089

	kvm_usage_count++;
	if (kvm_usage_count == 1) {
		atomic_set(&hardware_enable_failed, 0);
4090
		on_each_cpu(hardware_enable_nolock, NULL, 1);
4091 4092 4093 4094 4095 4096 4097

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

4098
	raw_spin_unlock(&kvm_count_lock);
4099 4100 4101 4102

	return r;
}

4103
static int kvm_reboot(struct notifier_block *notifier, unsigned long val,
M
Mike Day 已提交
4104
		      void *v)
4105
{
4106 4107 4108 4109 4110 4111
	/*
	 * 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
4112
	pr_info("kvm: exiting hardware virtualization\n");
4113
	kvm_rebooting = true;
4114
	on_each_cpu(hardware_disable_nolock, NULL, 1);
4115 4116 4117 4118 4119 4120 4121 4122
	return NOTIFY_OK;
}

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

M
Marcelo Tosatti 已提交
4123
static void kvm_io_bus_destroy(struct kvm_io_bus *bus)
4124 4125 4126 4127
{
	int i;

	for (i = 0; i < bus->dev_count; i++) {
4128
		struct kvm_io_device *pos = bus->range[i].dev;
4129 4130 4131

		kvm_iodevice_destructor(pos);
	}
M
Marcelo Tosatti 已提交
4132
	kfree(bus);
4133 4134
}

4135
static inline int kvm_io_bus_cmp(const struct kvm_io_range *r1,
X
Xiubo Li 已提交
4136
				 const struct kvm_io_range *r2)
4137
{
J
Jason Wang 已提交
4138 4139 4140 4141
	gpa_t addr1 = r1->addr;
	gpa_t addr2 = r2->addr;

	if (addr1 < addr2)
4142
		return -1;
J
Jason Wang 已提交
4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154

	/* 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)
4155
		return 1;
J
Jason Wang 已提交
4156

4157 4158 4159
	return 0;
}

4160 4161
static int kvm_io_bus_sort_cmp(const void *p1, const void *p2)
{
4162
	return kvm_io_bus_cmp(p1, p2);
4163 4164
}

G
Geoff Levand 已提交
4165
static int kvm_io_bus_get_first_dev(struct kvm_io_bus *bus,
4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182
			     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;

4183
	while (off > 0 && kvm_io_bus_cmp(&key, &bus->range[off-1]) == 0)
4184 4185 4186 4187 4188
		off--;

	return off;
}

4189
static int __kvm_io_bus_write(struct kvm_vcpu *vcpu, struct kvm_io_bus *bus,
C
Cornelia Huck 已提交
4190 4191 4192 4193 4194 4195 4196 4197 4198
			      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 &&
4199
		kvm_io_bus_cmp(range, &bus->range[idx]) == 0) {
4200
		if (!kvm_iodevice_write(vcpu, bus->range[idx].dev, range->addr,
C
Cornelia Huck 已提交
4201 4202 4203 4204 4205 4206 4207 4208
					range->len, val))
			return idx;
		idx++;
	}

	return -EOPNOTSUPP;
}

4209
/* kvm_io_bus_write - called under kvm->slots_lock */
4210
int kvm_io_bus_write(struct kvm_vcpu *vcpu, enum kvm_bus bus_idx, gpa_t addr,
4211
		     int len, const void *val)
4212
{
4213
	struct kvm_io_bus *bus;
4214
	struct kvm_io_range range;
C
Cornelia Huck 已提交
4215
	int r;
4216 4217 4218 4219 4220

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};
4221

4222
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
4223 4224
	if (!bus)
		return -ENOMEM;
4225
	r = __kvm_io_bus_write(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
4226 4227
	return r < 0 ? r : 0;
}
L
Leo Yan 已提交
4228
EXPORT_SYMBOL_GPL(kvm_io_bus_write);
C
Cornelia Huck 已提交
4229 4230

/* kvm_io_bus_write_cookie - called under kvm->slots_lock */
4231 4232
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
4233 4234 4235 4236 4237 4238 4239 4240 4241
{
	struct kvm_io_bus *bus;
	struct kvm_io_range range;

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

4242
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
4243 4244
	if (!bus)
		return -ENOMEM;
C
Cornelia Huck 已提交
4245 4246 4247

	/* First try the device referenced by cookie. */
	if ((cookie >= 0) && (cookie < bus->dev_count) &&
4248
	    (kvm_io_bus_cmp(&range, &bus->range[cookie]) == 0))
4249
		if (!kvm_iodevice_write(vcpu, bus->range[cookie].dev, addr, len,
C
Cornelia Huck 已提交
4250 4251 4252 4253 4254 4255 4256
					val))
			return cookie;

	/*
	 * cookie contained garbage; fall back to search and return the
	 * correct cookie value.
	 */
4257
	return __kvm_io_bus_write(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
4258 4259
}

4260 4261
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
4262 4263 4264 4265
{
	int idx;

	idx = kvm_io_bus_get_first_dev(bus, range->addr, range->len);
4266 4267 4268 4269
	if (idx < 0)
		return -EOPNOTSUPP;

	while (idx < bus->dev_count &&
4270
		kvm_io_bus_cmp(range, &bus->range[idx]) == 0) {
4271
		if (!kvm_iodevice_read(vcpu, bus->range[idx].dev, range->addr,
C
Cornelia Huck 已提交
4272 4273
				       range->len, val))
			return idx;
4274 4275 4276
		idx++;
	}

4277 4278
	return -EOPNOTSUPP;
}
4279

4280
/* kvm_io_bus_read - called under kvm->slots_lock */
4281
int kvm_io_bus_read(struct kvm_vcpu *vcpu, enum kvm_bus bus_idx, gpa_t addr,
M
Marcelo Tosatti 已提交
4282
		    int len, void *val)
4283
{
4284
	struct kvm_io_bus *bus;
4285
	struct kvm_io_range range;
C
Cornelia Huck 已提交
4286
	int r;
4287 4288 4289 4290 4291

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};
M
Marcelo Tosatti 已提交
4292

4293
	bus = srcu_dereference(vcpu->kvm->buses[bus_idx], &vcpu->kvm->srcu);
4294 4295
	if (!bus)
		return -ENOMEM;
4296
	r = __kvm_io_bus_read(vcpu, bus, &range, val);
C
Cornelia Huck 已提交
4297 4298
	return r < 0 ? r : 0;
}
4299

4300
/* Caller must hold slots_lock. */
4301 4302
int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
			    int len, struct kvm_io_device *dev)
4303
{
4304
	int i;
M
Marcelo Tosatti 已提交
4305
	struct kvm_io_bus *new_bus, *bus;
4306
	struct kvm_io_range range;
4307

4308
	bus = kvm_get_bus(kvm, bus_idx);
4309 4310 4311
	if (!bus)
		return -ENOMEM;

4312 4313
	/* exclude ioeventfd which is limited by maximum fd */
	if (bus->dev_count - bus->ioeventfd_count > NR_IOBUS_DEVS - 1)
4314
		return -ENOSPC;
4315

4316
	new_bus = kmalloc(struct_size(bus, range, bus->dev_count + 1),
4317
			  GFP_KERNEL_ACCOUNT);
M
Marcelo Tosatti 已提交
4318 4319
	if (!new_bus)
		return -ENOMEM;
4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335

	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
4336 4337 4338
	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
	synchronize_srcu_expedited(&kvm->srcu);
	kfree(bus);
4339 4340 4341 4342

	return 0;
}

4343
/* Caller must hold slots_lock. */
4344 4345
void kvm_io_bus_unregister_dev(struct kvm *kvm, enum kvm_bus bus_idx,
			       struct kvm_io_device *dev)
4346
{
4347
	int i, j;
M
Marcelo Tosatti 已提交
4348
	struct kvm_io_bus *new_bus, *bus;
4349

4350
	bus = kvm_get_bus(kvm, bus_idx);
4351
	if (!bus)
4352
		return;
4353

4354 4355
	for (i = 0; i < bus->dev_count; i++)
		if (bus->range[i].dev == dev) {
4356 4357
			break;
		}
M
Marcelo Tosatti 已提交
4358

4359 4360
	if (i == bus->dev_count)
		return;
4361

4362
	new_bus = kmalloc(struct_size(bus, range, bus->dev_count - 1),
4363
			  GFP_KERNEL_ACCOUNT);
4364
	if (new_bus) {
4365
		memcpy(new_bus, bus, struct_size(bus, range, i));
4366 4367
		new_bus->dev_count--;
		memcpy(new_bus->range + i, bus->range + i + 1,
4368
				flex_array_size(new_bus, range, new_bus->dev_count - i));
4369
	} else {
4370
		pr_err("kvm: failed to shrink bus, removing it completely\n");
4371 4372 4373 4374 4375
		for (j = 0; j < bus->dev_count; j++) {
			if (j == i)
				continue;
			kvm_iodevice_destructor(bus->range[j].dev);
		}
4376
	}
4377

M
Marcelo Tosatti 已提交
4378 4379 4380
	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
	synchronize_srcu_expedited(&kvm->srcu);
	kfree(bus);
4381
	return;
4382 4383
}

4384 4385 4386 4387 4388 4389 4390 4391 4392 4393
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);
4394 4395
	if (!bus)
		goto out_unlock;
4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409

	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);

4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421
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.
	 */
4422
	if (!refcount_inc_not_zero(&stat_data->kvm->users_count))
4423 4424
		return -ENOENT;

4425
	if (simple_attr_open(inode, file, get,
4426 4427 4428
		    KVM_DBGFS_GET_MODE(stat_data->dbgfs_item) & 0222
		    ? set : NULL,
		    fmt)) {
4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446
		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;
}

4447
static int kvm_get_stat_per_vm(struct kvm *kvm, size_t offset, u64 *val)
4448
{
4449
	*val = *(ulong *)((void *)kvm + offset);
4450

4451 4452 4453 4454 4455 4456
	return 0;
}

static int kvm_clear_stat_per_vm(struct kvm *kvm, size_t offset)
{
	*(ulong *)((void *)kvm + offset) = 0;
4457 4458 4459 4460

	return 0;
}

4461
static int kvm_get_stat_per_vcpu(struct kvm *kvm, size_t offset, u64 *val)
4462
{
4463 4464
	int i;
	struct kvm_vcpu *vcpu;
4465

4466
	*val = 0;
4467

4468 4469
	kvm_for_each_vcpu(i, vcpu, kvm)
		*val += *(u64 *)((void *)vcpu + offset);
4470 4471 4472 4473

	return 0;
}

4474
static int kvm_clear_stat_per_vcpu(struct kvm *kvm, size_t offset)
4475
{
4476 4477
	int i;
	struct kvm_vcpu *vcpu;
4478

4479 4480 4481 4482 4483
	kvm_for_each_vcpu(i, vcpu, kvm)
		*(u64 *)((void *)vcpu + offset) = 0;

	return 0;
}
4484

4485
static int kvm_stat_data_get(void *data, u64 *val)
4486
{
4487
	int r = -EFAULT;
4488 4489
	struct kvm_stat_data *stat_data = (struct kvm_stat_data *)data;

4490 4491 4492 4493 4494 4495 4496 4497 4498
	switch (stat_data->dbgfs_item->kind) {
	case KVM_STAT_VM:
		r = kvm_get_stat_per_vm(stat_data->kvm,
					stat_data->dbgfs_item->offset, val);
		break;
	case KVM_STAT_VCPU:
		r = kvm_get_stat_per_vcpu(stat_data->kvm,
					  stat_data->dbgfs_item->offset, val);
		break;
4499 4500 4501
	case KVM_STAT_DFX:
		r = -ENOSYS;
		break;
4502
	}
4503

4504
	return r;
4505 4506
}

4507
static int kvm_stat_data_clear(void *data, u64 val)
4508
{
4509
	int r = -EFAULT;
4510 4511 4512 4513 4514
	struct kvm_stat_data *stat_data = (struct kvm_stat_data *)data;

	if (val)
		return -EINVAL;

4515 4516 4517 4518 4519 4520 4521 4522 4523
	switch (stat_data->dbgfs_item->kind) {
	case KVM_STAT_VM:
		r = kvm_clear_stat_per_vm(stat_data->kvm,
					  stat_data->dbgfs_item->offset);
		break;
	case KVM_STAT_VCPU:
		r = kvm_clear_stat_per_vcpu(stat_data->kvm,
					    stat_data->dbgfs_item->offset);
		break;
4524 4525 4526
	case KVM_STAT_DFX:
		r = -ENOSYS;
		break;
4527
	}
4528

4529
	return r;
4530 4531
}

4532
static int kvm_stat_data_open(struct inode *inode, struct file *file)
4533 4534
{
	__simple_attr_check_format("%llu\n", 0ull);
4535 4536
	return kvm_debugfs_open(inode, file, kvm_stat_data_get,
				kvm_stat_data_clear, "%llu\n");
4537 4538
}

4539 4540 4541
static const struct file_operations stat_fops_per_vm = {
	.owner = THIS_MODULE,
	.open = kvm_stat_data_open,
4542
	.release = kvm_debugfs_release,
4543 4544 4545
	.read = simple_attr_read,
	.write = simple_attr_write,
	.llseek = no_llseek,
4546 4547
};

4548
static int vm_stat_get(void *_offset, u64 *val)
4549 4550 4551
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;
4552
	u64 tmp_val;
4553

4554
	*val = 0;
J
Junaid Shahid 已提交
4555
	mutex_lock(&kvm_lock);
4556
	list_for_each_entry(kvm, &vm_list, vm_list) {
4557
		kvm_get_stat_per_vm(kvm, offset, &tmp_val);
4558 4559
		*val += tmp_val;
	}
J
Junaid Shahid 已提交
4560
	mutex_unlock(&kvm_lock);
4561
	return 0;
4562 4563
}

4564 4565 4566 4567 4568 4569 4570 4571
static int vm_stat_clear(void *_offset, u64 val)
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;

	if (val)
		return -EINVAL;

J
Junaid Shahid 已提交
4572
	mutex_lock(&kvm_lock);
4573
	list_for_each_entry(kvm, &vm_list, vm_list) {
4574
		kvm_clear_stat_per_vm(kvm, offset);
4575
	}
J
Junaid Shahid 已提交
4576
	mutex_unlock(&kvm_lock);
4577 4578 4579 4580 4581

	return 0;
}

DEFINE_SIMPLE_ATTRIBUTE(vm_stat_fops, vm_stat_get, vm_stat_clear, "%llu\n");
4582

4583
static int vcpu_stat_get(void *_offset, u64 *val)
A
Avi Kivity 已提交
4584 4585 4586
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;
4587
	u64 tmp_val;
A
Avi Kivity 已提交
4588

4589
	*val = 0;
J
Junaid Shahid 已提交
4590
	mutex_lock(&kvm_lock);
4591
	list_for_each_entry(kvm, &vm_list, vm_list) {
4592
		kvm_get_stat_per_vcpu(kvm, offset, &tmp_val);
4593 4594
		*val += tmp_val;
	}
J
Junaid Shahid 已提交
4595
	mutex_unlock(&kvm_lock);
4596
	return 0;
A
Avi Kivity 已提交
4597 4598
}

4599 4600 4601 4602 4603 4604 4605 4606
static int vcpu_stat_clear(void *_offset, u64 val)
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;

	if (val)
		return -EINVAL;

J
Junaid Shahid 已提交
4607
	mutex_lock(&kvm_lock);
4608
	list_for_each_entry(kvm, &vm_list, vm_list) {
4609
		kvm_clear_stat_per_vcpu(kvm, offset);
4610
	}
J
Junaid Shahid 已提交
4611
	mutex_unlock(&kvm_lock);
4612 4613 4614 4615 4616 4617

	return 0;
}

DEFINE_SIMPLE_ATTRIBUTE(vcpu_stat_fops, vcpu_stat_get, vcpu_stat_clear,
			"%llu\n");
4618

4619 4620 4621 4622
void __attribute__((weak)) kvm_arch_vcpu_stat_reset(struct kvm_vcpu_stat *vcpu_stat)
{
}

4623 4624 4625
#define DFX_MAX_VCPU		1024
#define DFX_MAX_VCPU_STAT_SIZE	1024

4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654
/*
 * copy of seq_buf_alloc of kernel, kernel not export it
 */
static void *dfx_seq_buf_alloc(unsigned long size)
{
	return kvmalloc(size, GFP_KERNEL_ACCOUNT);
}

static void dfx_seq_buf_free(const void *buf)
{
	kvfree(buf);
}

static int dfx_seq_buf_alloc_vcpu(struct seq_file *p, int vcpu_nr)
{
	char *buf;
	size_t size;

	size = (vcpu_nr + 1) * DFX_MAX_VCPU_STAT_SIZE;
	buf = dfx_seq_buf_alloc(size);
	if (!buf)
		return -ENOMEM;
	if (p->buf)
		dfx_seq_buf_free(p->buf);
	p->buf = buf;
	p->size = size;
	return 0;
}

4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665
static int __dfx_vcpu_stats_get(struct seq_file *p, void *v)
{
	struct kvm *kvm;
	struct kvm_vcpu *vcpu;
	struct kvm_vcpu_stat *vcpu_stats;
	struct dfx_kvm_stats_debugfs_item *dp;
	int vcpu_nr = 0;
	int i, index = 0;

	mutex_lock(&kvm_lock);
	list_for_each_entry(kvm, &vm_list, vm_list)
4666
		kvm_for_each_vcpu(i, vcpu, kvm) {
4667
			vcpu_nr++;
4668
		}
4669 4670
	mutex_unlock(&kvm_lock);
	vcpu_nr = min(vcpu_nr, DFX_MAX_VCPU);
4671 4672 4673 4674 4675 4676 4677 4678
	if (!vcpu_nr) {
		seq_putc(p, '\n');
		return 0;
	}

	if (dfx_seq_buf_alloc_vcpu(p, vcpu_nr))
		return -ENOMEM;

4679 4680 4681 4682 4683
	vcpu_stats = vmalloc(vcpu_nr * sizeof(struct kvm_vcpu_stat));
	if (!vcpu_stats)
		return -ENOMEM;

	mutex_lock(&kvm_lock);
4684
	list_for_each_entry(kvm, &vm_list, vm_list) {
4685 4686 4687
		kvm_for_each_vcpu(i, vcpu, kvm) {
			if (index >= vcpu_nr)
				break;
4688
			memcpy(vcpu_stats + index, &(vcpu->stat),
4689
			       sizeof(struct kvm_vcpu_stat));
4690
			kvm_arch_vcpu_stat_reset(&vcpu->stat);
4691 4692
			++index;
		}
4693
	}
4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718
	mutex_unlock(&kvm_lock);
	for (i = 0; i < vcpu_nr; i++) {
		for (dp = dfx_debugfs_entries; dp->name; ++dp) {
			switch (dp->dfx_kind) {
			case DFX_STAT_U64:
				seq_put_decimal_ull(p, " ",
						*(u64 *)((void *)&vcpu_stats[i] + dp->offset));
				break;
			case DFX_STAT_CPUTIME:
				pr_warn("DFX_STAT_CPUTIME not supported currently!");
				break;
			default:
				pr_warn("Bad dfx_kind in dfx_debugfs_entries!");
				break;
			}
		}
		seq_putc(p, '\n');
	}

	vfree(vcpu_stats);
	return 0;
}

static int dfx_vcpu_stats_open(struct inode *inode, struct file *file)
{
4719
	return single_open(file, __dfx_vcpu_stats_get, NULL);
4720 4721 4722 4723 4724 4725 4726 4727 4728
}

static const struct file_operations dfx_stat_fops = {
	.open		= dfx_vcpu_stats_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
	.release	= single_release,
};

4729
static const struct file_operations *stat_fops[] = {
4730 4731
	[KVM_STAT_VCPU] = &vcpu_stat_fops,
	[KVM_STAT_VM]   = &vm_stat_fops,
4732
	[KVM_STAT_DFX]	= &dfx_stat_fops,
4733
};
A
Avi Kivity 已提交
4734

4735 4736 4737 4738 4739 4740 4741 4742
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
4743
	mutex_lock(&kvm_lock);
4744 4745 4746 4747 4748 4749 4750 4751
	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
4752
	mutex_unlock(&kvm_lock);
4753

4754
	env = kzalloc(sizeof(*env), GFP_KERNEL_ACCOUNT);
4755 4756 4757 4758 4759 4760
	if (!env)
		return;

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

4761
	if (type == KVM_EVENT_CREATE_VM) {
4762
		add_uevent_var(env, "EVENT=create");
4763 4764
		kvm->userspace_pid = task_pid_nr(current);
	} else if (type == KVM_EVENT_DESTROY_VM) {
4765
		add_uevent_var(env, "EVENT=destroy");
4766 4767
	}
	add_uevent_var(env, "PID=%d", kvm->userspace_pid);
4768

4769
	if (!IS_ERR_OR_NULL(kvm->debugfs_dentry)) {
4770
		char *tmp, *p = kmalloc(PATH_MAX, GFP_KERNEL_ACCOUNT);
4771 4772 4773 4774 4775 4776

		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);
4777 4778 4779 4780 4781 4782 4783 4784
		}
	}
	/* 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);
}

4785
static void kvm_init_debug(void)
A
Avi Kivity 已提交
4786 4787 4788
{
	struct kvm_stats_debugfs_item *p;

4789
	kvm_debugfs_dir = debugfs_create_dir("kvm", NULL);
4790

4791 4792
	kvm_debugfs_num_entries = 0;
	for (p = debugfs_entries; p->name; ++p, kvm_debugfs_num_entries++) {
4793 4794
		debugfs_create_file(p->name, KVM_DBGFS_GET_MODE(p),
				    kvm_debugfs_dir, (void *)(long)p->offset,
4795
				    stat_fops[p->kind]);
4796
	}
A
Avi Kivity 已提交
4797 4798
}

4799
static int kvm_suspend(void)
4800
{
4801
	if (kvm_usage_count)
4802
		hardware_disable_nolock(NULL);
4803 4804 4805
	return 0;
}

4806
static void kvm_resume(void)
4807
{
4808
	if (kvm_usage_count) {
4809 4810 4811
#ifdef CONFIG_LOCKDEP
		WARN_ON(lockdep_is_held(&kvm_count_lock));
#endif
4812
		hardware_enable_nolock(NULL);
4813
	}
4814 4815
}

4816
static struct syscore_ops kvm_syscore_ops = {
4817 4818 4819 4820
	.suspend = kvm_suspend,
	.resume = kvm_resume,
};

4821 4822 4823 4824 4825 4826 4827 4828 4829
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);
4830

4831
	WRITE_ONCE(vcpu->preempted, false);
4832
	WRITE_ONCE(vcpu->ready, false);
4833

4834
	__this_cpu_write(kvm_running_vcpu, vcpu);
R
Radim Krčmář 已提交
4835
	kvm_arch_sched_in(vcpu, cpu);
4836
	kvm_arch_vcpu_load(vcpu, cpu);
4837 4838 4839 4840 4841 4842 4843
}

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

4844
	if (current->state == TASK_RUNNING) {
4845
		WRITE_ONCE(vcpu->preempted, true);
4846 4847
		WRITE_ONCE(vcpu->ready, true);
	}
4848
	kvm_arch_vcpu_put(vcpu);
4849 4850 4851 4852 4853
	__this_cpu_write(kvm_running_vcpu, NULL);
}

/**
 * kvm_get_running_vcpu - get the vcpu running on the current CPU.
4854 4855 4856 4857 4858 4859
 *
 * We can disable preemption locally around accessing the per-CPU variable,
 * and use the resolved vcpu pointer after enabling preemption again,
 * because even if the current thread is migrated to another CPU, reading
 * the per-CPU value later will give us the same value as we update the
 * per-CPU variable in the preempt notifier handlers.
4860 4861 4862
 */
struct kvm_vcpu *kvm_get_running_vcpu(void)
{
4863 4864 4865 4866 4867 4868 4869
	struct kvm_vcpu *vcpu;

	preempt_disable();
	vcpu = __this_cpu_read(kvm_running_vcpu);
	preempt_enable();

	return vcpu;
4870
}
4871
EXPORT_SYMBOL_GPL(kvm_get_running_vcpu);
4872 4873 4874 4875 4876 4877 4878

/**
 * kvm_get_running_vcpus - get the per-CPU array of currently running vcpus.
 */
struct kvm_vcpu * __percpu *kvm_get_running_vcpus(void)
{
        return &kvm_running_vcpu;
4879 4880
}

4881 4882 4883 4884 4885 4886
struct kvm_cpu_compat_check {
	void *opaque;
	int *ret;
};

static void check_processor_compat(void *data)
4887
{
4888 4889 4890
	struct kvm_cpu_compat_check *c = data;

	*c->ret = kvm_arch_check_processor_compat(c->opaque);
4891 4892
}

4893
int kvm_init(void *opaque, unsigned vcpu_size, unsigned vcpu_align,
4894
		  struct module *module)
A
Avi Kivity 已提交
4895
{
4896
	struct kvm_cpu_compat_check c;
A
Avi Kivity 已提交
4897
	int r;
Y
Yang, Sheng 已提交
4898
	int cpu;
A
Avi Kivity 已提交
4899

4900 4901
	r = kvm_arch_init(opaque);
	if (r)
4902
		goto out_fail;
4903

4904 4905 4906 4907
	/*
	 * 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
4908 4909
	 * kvm_arch_init must be called before kvm_irqfd_init to avoid creating
	 * conflicts in case kvm is already setup for another implementation.
4910
	 */
P
Paolo Bonzini 已提交
4911 4912 4913
	r = kvm_irqfd_init();
	if (r)
		goto out_irqfd;
4914

4915
	if (!zalloc_cpumask_var(&cpus_hardware_enabled, GFP_KERNEL)) {
4916 4917 4918 4919
		r = -ENOMEM;
		goto out_free_0;
	}

4920
	r = kvm_arch_hardware_setup(opaque);
A
Avi Kivity 已提交
4921
	if (r < 0)
4922
		goto out_free_1;
A
Avi Kivity 已提交
4923

4924 4925
	c.ret = &r;
	c.opaque = opaque;
Y
Yang, Sheng 已提交
4926
	for_each_online_cpu(cpu) {
4927
		smp_call_function_single(cpu, check_processor_compat, &c, 1);
Y
Yang, Sheng 已提交
4928
		if (r < 0)
4929
			goto out_free_2;
Y
Yang, Sheng 已提交
4930 4931
	}

T
Thomas Gleixner 已提交
4932
	r = cpuhp_setup_state_nocalls(CPUHP_AP_KVM_STARTING, "kvm/cpu:starting",
4933
				      kvm_starting_cpu, kvm_dying_cpu);
A
Avi Kivity 已提交
4934
	if (r)
4935
		goto out_free_2;
A
Avi Kivity 已提交
4936 4937
	register_reboot_notifier(&kvm_reboot_notifier);

4938
	/* A kmem cache lets us meet the alignment requirements of fx_save. */
4939 4940
	if (!vcpu_align)
		vcpu_align = __alignof__(struct kvm_vcpu);
4941 4942 4943 4944 4945 4946
	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);
4947 4948
	if (!kvm_vcpu_cache) {
		r = -ENOMEM;
4949
		goto out_free_3;
4950 4951
	}

4952 4953 4954 4955
	r = kvm_async_pf_init();
	if (r)
		goto out_free;

A
Avi Kivity 已提交
4956
	kvm_chardev_ops.owner = module;
4957 4958
	kvm_vm_fops.owner = module;
	kvm_vcpu_fops.owner = module;
A
Avi Kivity 已提交
4959 4960 4961

	r = misc_register(&kvm_dev);
	if (r) {
X
Xiubo Li 已提交
4962
		pr_err("kvm: misc device register failed\n");
4963
		goto out_unreg;
A
Avi Kivity 已提交
4964 4965
	}

4966 4967
	register_syscore_ops(&kvm_syscore_ops);

4968 4969 4970
	kvm_preempt_ops.sched_in = kvm_sched_in;
	kvm_preempt_ops.sched_out = kvm_sched_out;

4971
	kvm_init_debug();
4972

P
Paolo Bonzini 已提交
4973 4974 4975
	r = kvm_vfio_ops_init();
	WARN_ON(r);

4976
	return 0;
A
Avi Kivity 已提交
4977

4978 4979
out_unreg:
	kvm_async_pf_deinit();
A
Avi Kivity 已提交
4980
out_free:
4981
	kmem_cache_destroy(kvm_vcpu_cache);
4982
out_free_3:
A
Avi Kivity 已提交
4983
	unregister_reboot_notifier(&kvm_reboot_notifier);
4984
	cpuhp_remove_state_nocalls(CPUHP_AP_KVM_STARTING);
4985
out_free_2:
4986
	kvm_arch_hardware_unsetup();
4987
out_free_1:
4988
	free_cpumask_var(cpus_hardware_enabled);
4989
out_free_0:
4990
	kvm_irqfd_exit();
P
Paolo Bonzini 已提交
4991
out_irqfd:
4992 4993
	kvm_arch_exit();
out_fail:
A
Avi Kivity 已提交
4994 4995
	return r;
}
4996
EXPORT_SYMBOL_GPL(kvm_init);
A
Avi Kivity 已提交
4997

4998
void kvm_exit(void)
A
Avi Kivity 已提交
4999
{
5000
	debugfs_remove_recursive(kvm_debugfs_dir);
A
Avi Kivity 已提交
5001
	misc_deregister(&kvm_dev);
5002
	kmem_cache_destroy(kvm_vcpu_cache);
5003
	kvm_async_pf_deinit();
5004
	unregister_syscore_ops(&kvm_syscore_ops);
A
Avi Kivity 已提交
5005
	unregister_reboot_notifier(&kvm_reboot_notifier);
5006
	cpuhp_remove_state_nocalls(CPUHP_AP_KVM_STARTING);
5007
	on_each_cpu(hardware_disable_nolock, NULL, 1);
5008
	kvm_arch_hardware_unsetup();
5009
	kvm_arch_exit();
5010
	kvm_irqfd_exit();
5011
	free_cpumask_var(cpus_hardware_enabled);
5012
	kvm_vfio_ops_exit();
A
Avi Kivity 已提交
5013
}
5014
EXPORT_SYMBOL_GPL(kvm_exit);
5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097

struct kvm_vm_worker_thread_context {
	struct kvm *kvm;
	struct task_struct *parent;
	struct completion init_done;
	kvm_vm_thread_fn_t thread_fn;
	uintptr_t data;
	int err;
};

static int kvm_vm_worker_thread(void *context)
{
	/*
	 * The init_context is allocated on the stack of the parent thread, so
	 * we have to locally copy anything that is needed beyond initialization
	 */
	struct kvm_vm_worker_thread_context *init_context = context;
	struct kvm *kvm = init_context->kvm;
	kvm_vm_thread_fn_t thread_fn = init_context->thread_fn;
	uintptr_t data = init_context->data;
	int err;

	err = kthread_park(current);
	/* kthread_park(current) is never supposed to return an error */
	WARN_ON(err != 0);
	if (err)
		goto init_complete;

	err = cgroup_attach_task_all(init_context->parent, current);
	if (err) {
		kvm_err("%s: cgroup_attach_task_all failed with err %d\n",
			__func__, err);
		goto init_complete;
	}

	set_user_nice(current, task_nice(init_context->parent));

init_complete:
	init_context->err = err;
	complete(&init_context->init_done);
	init_context = NULL;

	if (err)
		return err;

	/* Wait to be woken up by the spawner before proceeding. */
	kthread_parkme();

	if (!kthread_should_stop())
		err = thread_fn(kvm, data);

	return err;
}

int kvm_vm_create_worker_thread(struct kvm *kvm, kvm_vm_thread_fn_t thread_fn,
				uintptr_t data, const char *name,
				struct task_struct **thread_ptr)
{
	struct kvm_vm_worker_thread_context init_context = {};
	struct task_struct *thread;

	*thread_ptr = NULL;
	init_context.kvm = kvm;
	init_context.parent = current;
	init_context.thread_fn = thread_fn;
	init_context.data = data;
	init_completion(&init_context.init_done);

	thread = kthread_run(kvm_vm_worker_thread, &init_context,
			     "%s-%d", name, task_pid_nr(current));
	if (IS_ERR(thread))
		return PTR_ERR(thread);

	/* kthread_run is never supposed to return NULL */
	WARN_ON(thread == NULL);

	wait_for_completion(&init_context.init_done);

	if (!init_context.err)
		*thread_ptr = thread;

	return init_context.err;
}