kvm_main.c 65.8 KB
Newer Older
A
Avi Kivity 已提交
1 2 3 4 5 6 7
/*
 * Kernel-based Virtual Machine driver for Linux
 *
 * This module enables machines with Intel VT-x extensions to run virtual
 * machines without emulation or binary translation.
 *
 * Copyright (C) 2006 Qumranet, Inc.
N
Nicolas Kaiser 已提交
8
 * Copyright 2010 Red Hat, Inc. and/or its affiliates.
A
Avi Kivity 已提交
9 10 11 12 13 14 15 16 17 18
 *
 * Authors:
 *   Avi Kivity   <avi@qumranet.com>
 *   Yaniv Kamay  <yaniv@qumranet.com>
 *
 * This work is licensed under the terms of the GNU GPL, version 2.  See
 * the COPYING file in the top-level directory.
 *
 */

19
#include "iodev.h"
A
Avi Kivity 已提交
20

21
#include <linux/kvm_host.h>
A
Avi Kivity 已提交
22 23 24 25 26 27 28 29 30 31 32
#include <linux/kvm.h>
#include <linux/module.h>
#include <linux/errno.h>
#include <linux/percpu.h>
#include <linux/mm.h>
#include <linux/miscdevice.h>
#include <linux/vmalloc.h>
#include <linux/reboot.h>
#include <linux/debugfs.h>
#include <linux/highmem.h>
#include <linux/file.h>
33
#include <linux/syscore_ops.h>
A
Avi Kivity 已提交
34
#include <linux/cpu.h>
A
Alexey Dobriyan 已提交
35
#include <linux/sched.h>
36 37
#include <linux/cpumask.h>
#include <linux/smp.h>
38
#include <linux/anon_inodes.h>
39
#include <linux/profile.h>
40
#include <linux/kvm_para.h>
41
#include <linux/pagemap.h>
42
#include <linux/mman.h>
43
#include <linux/swap.h>
44
#include <linux/bitops.h>
45
#include <linux/spinlock.h>
46
#include <linux/compat.h>
47
#include <linux/srcu.h>
J
Joerg Roedel 已提交
48
#include <linux/hugetlb.h>
49
#include <linux/slab.h>
50 51
#include <linux/sort.h>
#include <linux/bsearch.h>
A
Avi Kivity 已提交
52

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

58
#include "coalesced_mmio.h"
59
#include "async_pf.h"
60

61 62 63
#define CREATE_TRACE_POINTS
#include <trace/events/kvm.h>

A
Avi Kivity 已提交
64 65 66
MODULE_AUTHOR("Qumranet");
MODULE_LICENSE("GPL");

67 68 69
/*
 * Ordering of locks:
 *
70
 * 		kvm->lock --> kvm->slots_lock --> kvm->irq_lock
71 72
 */

73
DEFINE_RAW_SPINLOCK(kvm_lock);
74
LIST_HEAD(vm_list);
75

76
static cpumask_var_t cpus_hardware_enabled;
77 78
static int kvm_usage_count = 0;
static atomic_t hardware_enable_failed;
79

80 81
struct kmem_cache *kvm_vcpu_cache;
EXPORT_SYMBOL_GPL(kvm_vcpu_cache);
A
Avi Kivity 已提交
82

83 84
static __read_mostly struct preempt_ops kvm_preempt_ops;

85
struct dentry *kvm_debugfs_dir;
A
Avi Kivity 已提交
86

A
Avi Kivity 已提交
87 88
static long kvm_vcpu_ioctl(struct file *file, unsigned int ioctl,
			   unsigned long arg);
89 90 91 92
#ifdef CONFIG_COMPAT
static long kvm_vcpu_compat_ioctl(struct file *file, unsigned int ioctl,
				  unsigned long arg);
#endif
93 94
static int hardware_enable_all(void);
static void hardware_disable_all(void);
A
Avi Kivity 已提交
95

M
Marcelo Tosatti 已提交
96 97
static void kvm_io_bus_destroy(struct kvm_io_bus *bus);

98 99
bool kvm_rebooting;
EXPORT_SYMBOL_GPL(kvm_rebooting);
100

101 102
static bool largepages_enabled = true;

X
Xiao Guangrong 已提交
103
bool kvm_is_mmio_pfn(pfn_t pfn)
B
Ben-Ami Yassour 已提交
104
{
105
	if (pfn_valid(pfn)) {
106
		int reserved;
107
		struct page *tail = pfn_to_page(pfn);
108 109
		struct page *head = compound_trans_head(tail);
		reserved = PageReserved(head);
110 111
		if (head != tail) {
			/*
112 113 114 115 116 117 118 119
			 * "head" is not a dangling pointer
			 * (compound_trans_head takes care of that)
			 * but the hugepage may have been splitted
			 * from under us (and we may not hold a
			 * reference count on the head page so it can
			 * be reused before we run PageReferenced), so
			 * we've to check PageTail before returning
			 * what we just read.
120
			 */
121 122 123
			smp_rmb();
			if (PageTail(tail))
				return reserved;
124 125
		}
		return PageReserved(tail);
126
	}
B
Ben-Ami Yassour 已提交
127 128 129 130

	return true;
}

A
Avi Kivity 已提交
131 132 133
/*
 * Switches to specified vcpu, until a matching vcpu_put()
 */
134
void vcpu_load(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
135
{
136 137
	int cpu;

A
Avi Kivity 已提交
138
	mutex_lock(&vcpu->mutex);
139 140 141 142 143 144 145 146
	if (unlikely(vcpu->pid != current->pids[PIDTYPE_PID].pid)) {
		/* The thread running this VCPU changed. */
		struct pid *oldpid = vcpu->pid;
		struct pid *newpid = get_task_pid(current, PIDTYPE_PID);
		rcu_assign_pointer(vcpu->pid, newpid);
		synchronize_rcu();
		put_pid(oldpid);
	}
147 148
	cpu = get_cpu();
	preempt_notifier_register(&vcpu->preempt_notifier);
149
	kvm_arch_vcpu_load(vcpu, cpu);
150
	put_cpu();
A
Avi Kivity 已提交
151 152
}

153
void vcpu_put(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
154
{
155
	preempt_disable();
156
	kvm_arch_vcpu_put(vcpu);
157 158
	preempt_notifier_unregister(&vcpu->preempt_notifier);
	preempt_enable();
A
Avi Kivity 已提交
159 160 161
	mutex_unlock(&vcpu->mutex);
}

162 163 164 165
static void ack_flush(void *_completed)
{
}

166
static bool make_all_cpus_request(struct kvm *kvm, unsigned int req)
167
{
168
	int i, cpu, me;
169 170
	cpumask_var_t cpus;
	bool called = true;
171 172
	struct kvm_vcpu *vcpu;

173
	zalloc_cpumask_var(&cpus, GFP_ATOMIC);
174

175
	me = get_cpu();
176
	kvm_for_each_vcpu(i, vcpu, kvm) {
177
		kvm_make_request(req, vcpu);
178
		cpu = vcpu->cpu;
179 180 181 182 183 184

		/* Set ->requests bit before we read ->mode */
		smp_mb();

		if (cpus != NULL && cpu != -1 && cpu != me &&
		      kvm_vcpu_exiting_guest_mode(vcpu) != OUTSIDE_GUEST_MODE)
185
			cpumask_set_cpu(cpu, cpus);
186
	}
187 188 189 190 191 192
	if (unlikely(cpus == NULL))
		smp_call_function_many(cpu_online_mask, ack_flush, NULL, 1);
	else if (!cpumask_empty(cpus))
		smp_call_function_many(cpus, ack_flush, NULL, 1);
	else
		called = false;
193
	put_cpu();
194
	free_cpumask_var(cpus);
195
	return called;
196 197
}

198
void kvm_flush_remote_tlbs(struct kvm *kvm)
199
{
200
	long dirty_count = kvm->tlbs_dirty;
201 202

	smp_mb();
203 204
	if (make_all_cpus_request(kvm, KVM_REQ_TLB_FLUSH))
		++kvm->stat.remote_tlb_flush;
205
	cmpxchg(&kvm->tlbs_dirty, dirty_count, 0);
206 207
}

208 209 210 211
void kvm_reload_remote_mmus(struct kvm *kvm)
{
	make_all_cpus_request(kvm, KVM_REQ_MMU_RELOAD);
}
212

R
Rusty Russell 已提交
213 214 215 216 217 218 219 220 221
int kvm_vcpu_init(struct kvm_vcpu *vcpu, struct kvm *kvm, unsigned id)
{
	struct page *page;
	int r;

	mutex_init(&vcpu->mutex);
	vcpu->cpu = -1;
	vcpu->kvm = kvm;
	vcpu->vcpu_id = id;
222
	vcpu->pid = NULL;
E
Eddie Dong 已提交
223
	init_waitqueue_head(&vcpu->wq);
224
	kvm_async_pf_vcpu_init(vcpu);
R
Rusty Russell 已提交
225 226 227 228 229 230 231 232

	page = alloc_page(GFP_KERNEL | __GFP_ZERO);
	if (!page) {
		r = -ENOMEM;
		goto fail;
	}
	vcpu->run = page_address(page);

233 234 235
	kvm_vcpu_set_in_spin_loop(vcpu, false);
	kvm_vcpu_set_dy_eligible(vcpu, false);

236
	r = kvm_arch_vcpu_init(vcpu);
R
Rusty Russell 已提交
237
	if (r < 0)
238
		goto fail_free_run;
R
Rusty Russell 已提交
239 240 241 242 243
	return 0;

fail_free_run:
	free_page((unsigned long)vcpu->run);
fail:
244
	return r;
R
Rusty Russell 已提交
245 246 247 248 249
}
EXPORT_SYMBOL_GPL(kvm_vcpu_init);

void kvm_vcpu_uninit(struct kvm_vcpu *vcpu)
{
250
	put_pid(vcpu->pid);
251
	kvm_arch_vcpu_uninit(vcpu);
R
Rusty Russell 已提交
252 253 254 255
	free_page((unsigned long)vcpu->run);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_uninit);

256 257 258 259 260 261 262 263 264 265 266
#if defined(CONFIG_MMU_NOTIFIER) && defined(KVM_ARCH_WANT_MMU_NOTIFIER)
static inline struct kvm *mmu_notifier_to_kvm(struct mmu_notifier *mn)
{
	return container_of(mn, struct kvm, mmu_notifier);
}

static void kvm_mmu_notifier_invalidate_page(struct mmu_notifier *mn,
					     struct mm_struct *mm,
					     unsigned long address)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
267
	int need_tlb_flush, idx;
268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286

	/*
	 * When ->invalidate_page runs, the linux pte has been zapped
	 * already but the page is still allocated until
	 * ->invalidate_page returns. So if we increase the sequence
	 * here the kvm page fault will notice if the spte can't be
	 * established because the page is going to be freed. If
	 * instead the kvm page fault establishes the spte before
	 * ->invalidate_page runs, kvm_unmap_hva will release it
	 * before returning.
	 *
	 * The sequence increase only need to be seen at spin_unlock
	 * time, and not at spin_lock time.
	 *
	 * Increasing the sequence after the spin_unlock would be
	 * unsafe because the kvm page fault could then establish the
	 * pte after kvm_unmap_hva returned, without noticing the page
	 * is going to be freed.
	 */
287
	idx = srcu_read_lock(&kvm->srcu);
288
	spin_lock(&kvm->mmu_lock);
289

290
	kvm->mmu_notifier_seq++;
291
	need_tlb_flush = kvm_unmap_hva(kvm, address) | kvm->tlbs_dirty;
292 293 294 295
	/* we've to flush the tlb before the pages can be freed */
	if (need_tlb_flush)
		kvm_flush_remote_tlbs(kvm);

296 297
	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);
298 299
}

300 301 302 303 304 305
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);
306
	int idx;
307

308
	idx = srcu_read_lock(&kvm->srcu);
309 310 311 312
	spin_lock(&kvm->mmu_lock);
	kvm->mmu_notifier_seq++;
	kvm_set_spte_hva(kvm, address, pte);
	spin_unlock(&kvm->mmu_lock);
313
	srcu_read_unlock(&kvm->srcu, idx);
314 315
}

316 317 318 319 320 321
static void kvm_mmu_notifier_invalidate_range_start(struct mmu_notifier *mn,
						    struct mm_struct *mm,
						    unsigned long start,
						    unsigned long end)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
322
	int need_tlb_flush = 0, idx;
323

324
	idx = srcu_read_lock(&kvm->srcu);
325 326 327 328 329 330 331
	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++;
332
	need_tlb_flush = kvm_unmap_hva_range(kvm, start, end);
333
	need_tlb_flush |= kvm->tlbs_dirty;
334 335 336
	/* we've to flush the tlb before the pages can be freed */
	if (need_tlb_flush)
		kvm_flush_remote_tlbs(kvm);
337 338 339

	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);
340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355
}

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

	spin_lock(&kvm->mmu_lock);
	/*
	 * This sequence increase will notify the kvm page fault that
	 * the page that is going to be mapped in the spte could have
	 * been freed.
	 */
	kvm->mmu_notifier_seq++;
356
	smp_wmb();
357 358
	/*
	 * The above sequence increase must be visible before the
359 360
	 * below count decrease, which is ensured by the smp_wmb above
	 * in conjunction with the smp_rmb in mmu_notifier_retry().
361 362 363 364 365 366 367 368 369 370 371 372
	 */
	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,
					      unsigned long address)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
373
	int young, idx;
374

375
	idx = srcu_read_lock(&kvm->srcu);
376 377
	spin_lock(&kvm->mmu_lock);

378
	young = kvm_age_hva(kvm, address);
379 380 381
	if (young)
		kvm_flush_remote_tlbs(kvm);

382 383 384
	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);

385 386 387
	return young;
}

A
Andrea Arcangeli 已提交
388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403
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;
}

404 405 406 407
static void kvm_mmu_notifier_release(struct mmu_notifier *mn,
				     struct mm_struct *mm)
{
	struct kvm *kvm = mmu_notifier_to_kvm(mn);
408 409 410
	int idx;

	idx = srcu_read_lock(&kvm->srcu);
411
	kvm_arch_flush_shadow_all(kvm);
412
	srcu_read_unlock(&kvm->srcu, idx);
413 414
}

415 416 417 418 419
static const struct mmu_notifier_ops kvm_mmu_notifier_ops = {
	.invalidate_page	= kvm_mmu_notifier_invalidate_page,
	.invalidate_range_start	= kvm_mmu_notifier_invalidate_range_start,
	.invalidate_range_end	= kvm_mmu_notifier_invalidate_range_end,
	.clear_flush_young	= kvm_mmu_notifier_clear_flush_young,
A
Andrea Arcangeli 已提交
420
	.test_young		= kvm_mmu_notifier_test_young,
421
	.change_pte		= kvm_mmu_notifier_change_pte,
422
	.release		= kvm_mmu_notifier_release,
423
};
424 425 426 427 428 429 430 431 432 433 434 435 436 437

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;
}

438 439
#endif /* CONFIG_MMU_NOTIFIER && KVM_ARCH_WANT_MMU_NOTIFIER */

440 441 442 443 444 445
static void kvm_init_memslots_id(struct kvm *kvm)
{
	int i;
	struct kvm_memslots *slots = kvm->memslots;

	for (i = 0; i < KVM_MEM_SLOTS_NUM; i++)
446
		slots->id_to_index[i] = slots->memslots[i].id = i;
447 448
}

449
static struct kvm *kvm_create_vm(unsigned long type)
A
Avi Kivity 已提交
450
{
451 452
	int r, i;
	struct kvm *kvm = kvm_arch_alloc_vm();
A
Avi Kivity 已提交
453

454 455 456
	if (!kvm)
		return ERR_PTR(-ENOMEM);

457
	r = kvm_arch_init_vm(kvm, type);
458 459
	if (r)
		goto out_err_nodisable;
460 461 462 463 464

	r = hardware_enable_all();
	if (r)
		goto out_err_nodisable;

465 466
#ifdef CONFIG_HAVE_KVM_IRQCHIP
	INIT_HLIST_HEAD(&kvm->mask_notifier_list);
467
	INIT_HLIST_HEAD(&kvm->irq_ack_notifier_list);
468
#endif
A
Avi Kivity 已提交
469

470 471 472
	r = -ENOMEM;
	kvm->memslots = kzalloc(sizeof(struct kvm_memslots), GFP_KERNEL);
	if (!kvm->memslots)
473
		goto out_err_nosrcu;
474
	kvm_init_memslots_id(kvm);
475
	if (init_srcu_struct(&kvm->srcu))
476
		goto out_err_nosrcu;
M
Marcelo Tosatti 已提交
477 478 479
	for (i = 0; i < KVM_NR_BUSES; i++) {
		kvm->buses[i] = kzalloc(sizeof(struct kvm_io_bus),
					GFP_KERNEL);
480
		if (!kvm->buses[i])
M
Marcelo Tosatti 已提交
481 482
			goto out_err;
	}
483

484
	spin_lock_init(&kvm->mmu_lock);
485 486
	kvm->mm = current->mm;
	atomic_inc(&kvm->mm->mm_count);
G
Gregory Haskins 已提交
487
	kvm_eventfd_init(kvm);
S
Shaohua Li 已提交
488
	mutex_init(&kvm->lock);
489
	mutex_init(&kvm->irq_lock);
490
	mutex_init(&kvm->slots_lock);
I
Izik Eidus 已提交
491
	atomic_set(&kvm->users_count, 1);
492 493 494 495 496

	r = kvm_init_mmu_notifier(kvm);
	if (r)
		goto out_err;

497
	raw_spin_lock(&kvm_lock);
498
	list_add(&kvm->vm_list, &vm_list);
499
	raw_spin_unlock(&kvm_lock);
500

501
	return kvm;
502 503

out_err:
504 505
	cleanup_srcu_struct(&kvm->srcu);
out_err_nosrcu:
506 507
	hardware_disable_all();
out_err_nodisable:
M
Marcelo Tosatti 已提交
508 509
	for (i = 0; i < KVM_NR_BUSES; i++)
		kfree(kvm->buses[i]);
510
	kfree(kvm->memslots);
511
	kvm_arch_free_vm(kvm);
512
	return ERR_PTR(r);
513 514
}

515 516 517 518
/*
 * Avoid using vmalloc for a small buffer.
 * Should not be used when the size is statically known.
 */
519
void *kvm_kvzalloc(unsigned long size)
520 521 522 523 524 525 526
{
	if (size > PAGE_SIZE)
		return vzalloc(size);
	else
		return kzalloc(size, GFP_KERNEL);
}

527
void kvm_kvfree(const void *addr)
528 529 530 531 532 533 534
{
	if (is_vmalloc_addr(addr))
		vfree(addr);
	else
		kfree(addr);
}

535 536 537 538 539
static void kvm_destroy_dirty_bitmap(struct kvm_memory_slot *memslot)
{
	if (!memslot->dirty_bitmap)
		return;

540
	kvm_kvfree(memslot->dirty_bitmap);
541 542 543
	memslot->dirty_bitmap = NULL;
}

A
Avi Kivity 已提交
544 545 546 547 548 549 550
/*
 * Free any memory in @free but not in @dont.
 */
static void kvm_free_physmem_slot(struct kvm_memory_slot *free,
				  struct kvm_memory_slot *dont)
{
	if (!dont || free->dirty_bitmap != dont->dirty_bitmap)
551
		kvm_destroy_dirty_bitmap(free);
A
Avi Kivity 已提交
552

553
	kvm_arch_free_memslot(free, dont);
M
Marcelo Tosatti 已提交
554

A
Avi Kivity 已提交
555 556 557
	free->npages = 0;
}

558
void kvm_free_physmem(struct kvm *kvm)
A
Avi Kivity 已提交
559
{
560
	struct kvm_memslots *slots = kvm->memslots;
561
	struct kvm_memory_slot *memslot;
562

563 564
	kvm_for_each_memslot(memslot, slots)
		kvm_free_physmem_slot(memslot, NULL);
A
Avi Kivity 已提交
565

566
	kfree(kvm->memslots);
A
Avi Kivity 已提交
567 568
}

569 570
static void kvm_destroy_vm(struct kvm *kvm)
{
M
Marcelo Tosatti 已提交
571
	int i;
572 573
	struct mm_struct *mm = kvm->mm;

574
	kvm_arch_sync_events(kvm);
575
	raw_spin_lock(&kvm_lock);
576
	list_del(&kvm->vm_list);
577
	raw_spin_unlock(&kvm_lock);
578
	kvm_free_irq_routing(kvm);
M
Marcelo Tosatti 已提交
579 580
	for (i = 0; i < KVM_NR_BUSES; i++)
		kvm_io_bus_destroy(kvm->buses[i]);
581
	kvm_coalesced_mmio_free(kvm);
582 583
#if defined(CONFIG_MMU_NOTIFIER) && defined(KVM_ARCH_WANT_MMU_NOTIFIER)
	mmu_notifier_unregister(&kvm->mmu_notifier, kvm->mm);
584
#else
585
	kvm_arch_flush_shadow_all(kvm);
586
#endif
587
	kvm_arch_destroy_vm(kvm);
588 589 590
	kvm_free_physmem(kvm);
	cleanup_srcu_struct(&kvm->srcu);
	kvm_arch_free_vm(kvm);
591
	hardware_disable_all();
592
	mmdrop(mm);
593 594
}

I
Izik Eidus 已提交
595 596 597 598 599 600 601 602 603 604 605 606 607 608
void kvm_get_kvm(struct kvm *kvm)
{
	atomic_inc(&kvm->users_count);
}
EXPORT_SYMBOL_GPL(kvm_get_kvm);

void kvm_put_kvm(struct kvm *kvm)
{
	if (atomic_dec_and_test(&kvm->users_count))
		kvm_destroy_vm(kvm);
}
EXPORT_SYMBOL_GPL(kvm_put_kvm);


609 610 611 612
static int kvm_vm_release(struct inode *inode, struct file *filp)
{
	struct kvm *kvm = filp->private_data;

G
Gregory Haskins 已提交
613 614
	kvm_irqfd_release(kvm);

I
Izik Eidus 已提交
615
	kvm_put_kvm(kvm);
A
Avi Kivity 已提交
616 617 618
	return 0;
}

619 620
/*
 * Allocation size is twice as large as the actual dirty bitmap size.
621
 * See x86's kvm_vm_ioctl_get_dirty_log() why this is needed.
622
 */
623 624
static int kvm_create_dirty_bitmap(struct kvm_memory_slot *memslot)
{
625
#ifndef CONFIG_S390
626
	unsigned long dirty_bytes = 2 * kvm_dirty_bitmap_bytes(memslot);
627

628
	memslot->dirty_bitmap = kvm_kvzalloc(dirty_bytes);
629 630 631
	if (!memslot->dirty_bitmap)
		return -ENOMEM;

632
#endif /* !CONFIG_S390 */
633 634 635
	return 0;
}

636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656
static int cmp_memslot(const void *slot1, const void *slot2)
{
	struct kvm_memory_slot *s1, *s2;

	s1 = (struct kvm_memory_slot *)slot1;
	s2 = (struct kvm_memory_slot *)slot2;

	if (s1->npages < s2->npages)
		return 1;
	if (s1->npages > s2->npages)
		return -1;

	return 0;
}

/*
 * Sort the memslots base on its size, so the larger slots
 * will get better fit.
 */
static void sort_memslots(struct kvm_memslots *slots)
{
657 658
	int i;

659 660
	sort(slots->memslots, KVM_MEM_SLOTS_NUM,
	      sizeof(struct kvm_memory_slot), cmp_memslot, NULL);
661 662 663

	for (i = 0; i < KVM_MEM_SLOTS_NUM; i++)
		slots->id_to_index[slots->memslots[i].id] = i;
664 665
}

666 667 668 669
void update_memslots(struct kvm_memslots *slots, struct kvm_memory_slot *new)
{
	if (new) {
		int id = new->id;
670
		struct kvm_memory_slot *old = id_to_memslot(slots, id);
671
		unsigned long npages = old->npages;
672

673
		*old = *new;
674 675
		if (new->npages != npages)
			sort_memslots(slots);
676 677 678 679 680
	}

	slots->generation++;
}

681 682
static int check_memory_region_flags(struct kvm_userspace_memory_region *mem)
{
X
Xiao Guangrong 已提交
683 684 685 686 687 688 689
	u32 valid_flags = KVM_MEM_LOG_DIRTY_PAGES;

#ifdef KVM_CAP_READONLY_MEM
	valid_flags |= KVM_MEM_READONLY;
#endif

	if (mem->flags & ~valid_flags)
690 691 692 693 694
		return -EINVAL;

	return 0;
}

A
Avi Kivity 已提交
695 696 697 698 699
/*
 * Allocate some memory and give it an address in the guest physical address
 * space.
 *
 * Discontiguous memory is allowed, mostly for framebuffers.
700
 *
701
 * Must be called holding mmap_sem for write.
A
Avi Kivity 已提交
702
 */
703 704 705
int __kvm_set_memory_region(struct kvm *kvm,
			    struct kvm_userspace_memory_region *mem,
			    int user_alloc)
A
Avi Kivity 已提交
706
{
707
	int r;
A
Avi Kivity 已提交
708
	gfn_t base_gfn;
709 710
	unsigned long npages;
	unsigned long i;
A
Avi Kivity 已提交
711 712
	struct kvm_memory_slot *memslot;
	struct kvm_memory_slot old, new;
713
	struct kvm_memslots *slots, *old_memslots;
A
Avi Kivity 已提交
714

715 716 717 718
	r = check_memory_region_flags(mem);
	if (r)
		goto out;

A
Avi Kivity 已提交
719 720 721 722 723 724
	r = -EINVAL;
	/* General sanity checks */
	if (mem->memory_size & (PAGE_SIZE - 1))
		goto out;
	if (mem->guest_phys_addr & (PAGE_SIZE - 1))
		goto out;
725 726 727
	/* We can read the guest memory with __xxx_user() later on. */
	if (user_alloc &&
	    ((mem->userspace_addr & (PAGE_SIZE - 1)) ||
728 729 730
	     !access_ok(VERIFY_WRITE,
			(void __user *)(unsigned long)mem->userspace_addr,
			mem->memory_size)))
731
		goto out;
732
	if (mem->slot >= KVM_MEM_SLOTS_NUM)
A
Avi Kivity 已提交
733 734 735 736
		goto out;
	if (mem->guest_phys_addr + mem->memory_size < mem->guest_phys_addr)
		goto out;

737
	memslot = id_to_memslot(kvm->memslots, mem->slot);
A
Avi Kivity 已提交
738 739 740
	base_gfn = mem->guest_phys_addr >> PAGE_SHIFT;
	npages = mem->memory_size >> PAGE_SHIFT;

741 742 743 744
	r = -EINVAL;
	if (npages > KVM_MEM_MAX_NR_PAGES)
		goto out;

A
Avi Kivity 已提交
745 746 747 748 749
	if (!npages)
		mem->flags &= ~KVM_MEM_LOG_DIRTY_PAGES;

	new = old = *memslot;

750
	new.id = mem->slot;
A
Avi Kivity 已提交
751 752 753 754 755 756 757
	new.base_gfn = base_gfn;
	new.npages = npages;
	new.flags = mem->flags;

	/* Disallow changing a memory slot's size. */
	r = -EINVAL;
	if (npages && old.npages && npages != old.npages)
758
		goto out_free;
A
Avi Kivity 已提交
759 760 761 762

	/* Check for overlaps */
	r = -EEXIST;
	for (i = 0; i < KVM_MEMORY_SLOTS; ++i) {
763
		struct kvm_memory_slot *s = &kvm->memslots->memslots[i];
A
Avi Kivity 已提交
764

765
		if (s == memslot || !s->npages)
A
Avi Kivity 已提交
766 767 768
			continue;
		if (!((base_gfn + npages <= s->base_gfn) ||
		      (base_gfn >= s->base_gfn + s->npages)))
769
			goto out_free;
A
Avi Kivity 已提交
770 771 772 773
	}

	/* Free page dirty bitmap if unneeded */
	if (!(new.flags & KVM_MEM_LOG_DIRTY_PAGES))
A
Al Viro 已提交
774
		new.dirty_bitmap = NULL;
A
Avi Kivity 已提交
775 776 777 778

	r = -ENOMEM;

	/* Allocate if a slot is being created */
779 780 781
	if (npages && !old.npages) {
		new.user_alloc = user_alloc;
		new.userspace_addr = mem->userspace_addr;
782

783 784
		if (kvm_arch_create_memslot(&new, npages))
			goto out_free;
A
Avi Kivity 已提交
785
	}
786

A
Avi Kivity 已提交
787 788
	/* Allocate page dirty bitmap if needed */
	if ((new.flags & KVM_MEM_LOG_DIRTY_PAGES) && !new.dirty_bitmap) {
789
		if (kvm_create_dirty_bitmap(&new) < 0)
790
			goto out_free;
791
		/* destroy any largepage mappings for dirty tracking */
A
Avi Kivity 已提交
792 793
	}

794
	if (!npages) {
795 796
		struct kvm_memory_slot *slot;

797
		r = -ENOMEM;
798 799
		slots = kmemdup(kvm->memslots, sizeof(struct kvm_memslots),
				GFP_KERNEL);
800 801
		if (!slots)
			goto out_free;
802 803 804
		slot = id_to_memslot(slots, mem->slot);
		slot->flags |= KVM_MEMSLOT_INVALID;

805
		update_memslots(slots, NULL);
806 807 808 809 810 811 812 813 814 815 816

		old_memslots = kvm->memslots;
		rcu_assign_pointer(kvm->memslots, slots);
		synchronize_srcu_expedited(&kvm->srcu);
		/* From this point no new shadow pages pointing to a deleted
		 * 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_roots)
		 */
817
		kvm_arch_flush_shadow_memslot(kvm, slot);
818 819
		kfree(old_memslots);
	}
820

821 822 823 824
	r = kvm_arch_prepare_memory_region(kvm, &new, old, mem, user_alloc);
	if (r)
		goto out_free;

825
	/* map/unmap the pages in iommu page table */
826 827 828 829
	if (npages) {
		r = kvm_iommu_map_pages(kvm, &new);
		if (r)
			goto out_free;
830 831
	} else
		kvm_iommu_unmap_pages(kvm, &old);
832

833
	r = -ENOMEM;
834 835
	slots = kmemdup(kvm->memslots, sizeof(struct kvm_memslots),
			GFP_KERNEL);
836 837 838 839 840 841
	if (!slots)
		goto out_free;

	/* actual memory is freed via old in kvm_free_physmem_slot below */
	if (!npages) {
		new.dirty_bitmap = NULL;
842
		memset(&new.arch, 0, sizeof(new.arch));
843 844
	}

845
	update_memslots(slots, &new);
846 847 848
	old_memslots = kvm->memslots;
	rcu_assign_pointer(kvm->memslots, slots);
	synchronize_srcu_expedited(&kvm->srcu);
849

850
	kvm_arch_commit_memory_region(kvm, mem, old, user_alloc);
851

852 853 854 855 856
	/*
	 * If the new memory slot is created, we need to clear all
	 * mmio sptes.
	 */
	if (npages && old.base_gfn != mem->guest_phys_addr >> PAGE_SHIFT)
857
		kvm_arch_flush_shadow_all(kvm);
858

859 860 861
	kvm_free_physmem_slot(&old, &new);
	kfree(old_memslots);

A
Avi Kivity 已提交
862 863
	return 0;

864
out_free:
A
Avi Kivity 已提交
865 866 867
	kvm_free_physmem_slot(&new, &old);
out:
	return r;
868 869

}
870 871 872 873 874 875 876 877
EXPORT_SYMBOL_GPL(__kvm_set_memory_region);

int kvm_set_memory_region(struct kvm *kvm,
			  struct kvm_userspace_memory_region *mem,
			  int user_alloc)
{
	int r;

878
	mutex_lock(&kvm->slots_lock);
879
	r = __kvm_set_memory_region(kvm, mem, user_alloc);
880
	mutex_unlock(&kvm->slots_lock);
881 882
	return r;
}
883 884
EXPORT_SYMBOL_GPL(kvm_set_memory_region);

885 886 887 888
int kvm_vm_ioctl_set_memory_region(struct kvm *kvm,
				   struct
				   kvm_userspace_memory_region *mem,
				   int user_alloc)
889
{
890 891
	if (mem->slot >= KVM_MEMORY_SLOTS)
		return -EINVAL;
892
	return kvm_set_memory_region(kvm, mem, user_alloc);
A
Avi Kivity 已提交
893 894
}

895 896
int kvm_get_dirty_log(struct kvm *kvm,
			struct kvm_dirty_log *log, int *is_dirty)
A
Avi Kivity 已提交
897 898 899
{
	struct kvm_memory_slot *memslot;
	int r, i;
900
	unsigned long n;
A
Avi Kivity 已提交
901 902 903 904 905 906
	unsigned long any = 0;

	r = -EINVAL;
	if (log->slot >= KVM_MEMORY_SLOTS)
		goto out;

907
	memslot = id_to_memslot(kvm->memslots, log->slot);
A
Avi Kivity 已提交
908 909 910 911
	r = -ENOENT;
	if (!memslot->dirty_bitmap)
		goto out;

912
	n = kvm_dirty_bitmap_bytes(memslot);
A
Avi Kivity 已提交
913

914
	for (i = 0; !any && i < n/sizeof(long); ++i)
A
Avi Kivity 已提交
915 916 917 918 919 920
		any = memslot->dirty_bitmap[i];

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

921 922
	if (any)
		*is_dirty = 1;
A
Avi Kivity 已提交
923 924 925 926 927 928

	r = 0;
out:
	return r;
}

929 930 931 932 933
bool kvm_largepages_enabled(void)
{
	return largepages_enabled;
}

934 935 936 937 938 939
void kvm_disable_largepages(void)
{
	largepages_enabled = false;
}
EXPORT_SYMBOL_GPL(kvm_disable_largepages);

940 941 942 943
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
944
EXPORT_SYMBOL_GPL(gfn_to_memslot);
A
Avi Kivity 已提交
945

946 947
int kvm_is_visible_gfn(struct kvm *kvm, gfn_t gfn)
{
948
	struct kvm_memory_slot *memslot = gfn_to_memslot(kvm, gfn);
949

950 951 952
	if (!memslot || memslot->id >= KVM_MEMORY_SLOTS ||
	      memslot->flags & KVM_MEMSLOT_INVALID)
		return 0;
953

954
	return 1;
955 956 957
}
EXPORT_SYMBOL_GPL(kvm_is_visible_gfn);

J
Joerg Roedel 已提交
958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981
unsigned long kvm_host_page_size(struct kvm *kvm, gfn_t gfn)
{
	struct vm_area_struct *vma;
	unsigned long addr, size;

	size = PAGE_SIZE;

	addr = gfn_to_hva(kvm, gfn);
	if (kvm_is_error_hva(addr))
		return PAGE_SIZE;

	down_read(&current->mm->mmap_sem);
	vma = find_vma(current->mm, addr);
	if (!vma)
		goto out;

	size = vma_kernel_pagesize(vma);

out:
	up_read(&current->mm->mmap_sem);

	return size;
}

X
Xiao Guangrong 已提交
982 983 984 985 986 987 988
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
989
{
990
	if (!slot || slot->flags & KVM_MEMSLOT_INVALID)
X
Xiao Guangrong 已提交
991
		return KVM_HVA_ERR_BAD;
992

X
Xiao Guangrong 已提交
993 994 995
	if (memslot_is_readonly(slot) && write)
		return KVM_HVA_ERR_RO_BAD;

996 997 998
	if (nr_pages)
		*nr_pages = slot->npages - (gfn - slot->base_gfn);

X
Xiao Guangrong 已提交
999 1000 1001 1002 1003 1004 1005
	return __gfn_to_hva_memslot(slot, gfn);
}

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
1006
}
1007

X
Xiao Guangrong 已提交
1008 1009 1010 1011 1012 1013 1014
unsigned long gfn_to_hva_memslot(struct kvm_memory_slot *slot,
				 gfn_t gfn)
{
	return gfn_to_hva_many(slot, gfn, NULL);
}
EXPORT_SYMBOL_GPL(gfn_to_hva_memslot);

1015 1016
unsigned long gfn_to_hva(struct kvm *kvm, gfn_t gfn)
{
1017
	return gfn_to_hva_many(gfn_to_memslot(kvm, gfn), gfn, NULL);
1018
}
1019
EXPORT_SYMBOL_GPL(gfn_to_hva);
I
Izik Eidus 已提交
1020

1021 1022 1023 1024 1025 1026
/*
 * The hva returned by this function is only allowed to be read.
 * It should pair with kvm_read_hva() or kvm_read_hva_atomic().
 */
static unsigned long gfn_to_hva_read(struct kvm *kvm, gfn_t gfn)
{
X
Xiao Guangrong 已提交
1027
	return __gfn_to_hva_many(gfn_to_memslot(kvm, gfn), gfn, NULL, false);
1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039
}

static int kvm_read_hva(void *data, void __user *hva, int len)
{
	return __copy_from_user(data, hva, len);
}

static int kvm_read_hva_atomic(void *data, void __user *hva, int len)
{
	return __copy_from_user_inatomic(data, hva, len);
}

1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050
int get_user_page_nowait(struct task_struct *tsk, struct mm_struct *mm,
	unsigned long start, int write, struct page **page)
{
	int flags = FOLL_TOUCH | FOLL_NOWAIT | FOLL_HWPOISON | FOLL_GET;

	if (write)
		flags |= FOLL_WRITE;

	return __get_user_pages(tsk, mm, start, 1, flags, page, NULL, NULL);
}

1051 1052 1053 1054 1055 1056 1057 1058 1059
static inline int check_user_page_hwpoison(unsigned long addr)
{
	int rc, flags = FOLL_TOUCH | FOLL_HWPOISON | FOLL_WRITE;

	rc = __get_user_pages(current, current->mm, addr, 1,
			      flags, NULL, NULL, NULL);
	return rc == -EHWPOISON;
}

X
Xiao Guangrong 已提交
1060 1061 1062 1063 1064 1065
/*
 * The atomic path to get the writable pfn which will be stored in @pfn,
 * true indicates success, otherwise false is returned.
 */
static bool hva_to_pfn_fast(unsigned long addr, bool atomic, bool *async,
			    bool write_fault, bool *writable, pfn_t *pfn)
A
Avi Kivity 已提交
1066
{
1067
	struct page *page[1];
X
Xiao Guangrong 已提交
1068
	int npages;
A
Avi Kivity 已提交
1069

X
Xiao Guangrong 已提交
1070 1071
	if (!(async || atomic))
		return false;
1072

1073 1074 1075 1076 1077 1078 1079 1080
	/*
	 * 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;

X
Xiao Guangrong 已提交
1081 1082 1083
	npages = __get_user_pages_fast(addr, 1, 1, page);
	if (npages == 1) {
		*pfn = page_to_pfn(page[0]);
1084

X
Xiao Guangrong 已提交
1085 1086 1087 1088
		if (writable)
			*writable = true;
		return true;
	}
1089

X
Xiao Guangrong 已提交
1090 1091
	return false;
}
1092

X
Xiao Guangrong 已提交
1093 1094 1095 1096 1097 1098 1099 1100 1101
/*
 * 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,
			   bool *writable, pfn_t *pfn)
{
	struct page *page[1];
	int npages = 0;
1102

X
Xiao Guangrong 已提交
1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119
	might_sleep();

	if (writable)
		*writable = write_fault;

	if (async) {
		down_read(&current->mm->mmap_sem);
		npages = get_user_page_nowait(current, current->mm,
					      addr, write_fault, page);
		up_read(&current->mm->mmap_sem);
	} else
		npages = get_user_pages_fast(addr, 1, write_fault,
					     page);
	if (npages != 1)
		return npages;

	/* map read fault as writable if possible */
1120
	if (unlikely(!write_fault) && writable) {
X
Xiao Guangrong 已提交
1121 1122 1123 1124 1125 1126 1127
		struct page *wpage[1];

		npages = __get_user_pages_fast(addr, 1, 1, wpage);
		if (npages == 1) {
			*writable = true;
			put_page(page[0]);
			page[0] = wpage[0];
1128
		}
X
Xiao Guangrong 已提交
1129 1130

		npages = 1;
1131
	}
X
Xiao Guangrong 已提交
1132 1133 1134
	*pfn = page_to_pfn(page[0]);
	return npages;
}
I
Izik Eidus 已提交
1135

X
Xiao Guangrong 已提交
1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146
static bool vma_is_valid(struct vm_area_struct *vma, bool write_fault)
{
	if (unlikely(!(vma->vm_flags & VM_READ)))
		return false;

	if (write_fault && (unlikely(!(vma->vm_flags & VM_WRITE))))
		return false;

	return true;
}

1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160
/*
 * 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.
 */
X
Xiao Guangrong 已提交
1161 1162 1163 1164 1165 1166
static pfn_t hva_to_pfn(unsigned long addr, bool atomic, bool *async,
			bool write_fault, bool *writable)
{
	struct vm_area_struct *vma;
	pfn_t pfn = 0;
	int npages;
1167

X
Xiao Guangrong 已提交
1168 1169
	/* we can do it either atomically or asynchronously, not both */
	BUG_ON(atomic && async);
1170

X
Xiao Guangrong 已提交
1171 1172 1173 1174 1175 1176 1177 1178 1179
	if (hva_to_pfn_fast(addr, atomic, async, write_fault, writable, &pfn))
		return pfn;

	if (atomic)
		return KVM_PFN_ERR_FAULT;

	npages = hva_to_pfn_slow(addr, async, write_fault, writable, &pfn);
	if (npages == 1)
		return pfn;
1180

X
Xiao Guangrong 已提交
1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196
	down_read(&current->mm->mmap_sem);
	if (npages == -EHWPOISON ||
	      (!async && check_user_page_hwpoison(addr))) {
		pfn = KVM_PFN_ERR_HWPOISON;
		goto exit;
	}

	vma = find_vma_intersection(current->mm, addr, addr + 1);

	if (vma == NULL)
		pfn = KVM_PFN_ERR_FAULT;
	else if ((vma->vm_flags & VM_PFNMAP)) {
		pfn = ((addr - vma->vm_start) >> PAGE_SHIFT) +
			vma->vm_pgoff;
		BUG_ON(!kvm_is_mmio_pfn(pfn));
	} else {
X
Xiao Guangrong 已提交
1197
		if (async && vma_is_valid(vma, write_fault))
X
Xiao Guangrong 已提交
1198 1199 1200 1201 1202
			*async = true;
		pfn = KVM_PFN_ERR_FAULT;
	}
exit:
	up_read(&current->mm->mmap_sem);
1203
	return pfn;
1204 1205
}

X
Xiao Guangrong 已提交
1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227
static pfn_t
__gfn_to_pfn_memslot(struct kvm_memory_slot *slot, gfn_t gfn, bool atomic,
		     bool *async, bool write_fault, bool *writable)
{
	unsigned long addr = __gfn_to_hva_many(slot, gfn, NULL, write_fault);

	if (addr == KVM_HVA_ERR_RO_BAD)
		return KVM_PFN_ERR_RO_FAULT;

	if (kvm_is_error_hva(addr))
		return KVM_PFN_ERR_BAD;

	/* 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);
}

1228 1229
static pfn_t __gfn_to_pfn(struct kvm *kvm, gfn_t gfn, bool atomic, bool *async,
			  bool write_fault, bool *writable)
1230
{
X
Xiao Guangrong 已提交
1231
	struct kvm_memory_slot *slot;
1232

1233 1234 1235
	if (async)
		*async = false;

X
Xiao Guangrong 已提交
1236
	slot = gfn_to_memslot(kvm, gfn);
1237

X
Xiao Guangrong 已提交
1238 1239
	return __gfn_to_pfn_memslot(slot, gfn, atomic, async, write_fault,
				    writable);
1240 1241 1242 1243
}

pfn_t gfn_to_pfn_atomic(struct kvm *kvm, gfn_t gfn)
{
1244
	return __gfn_to_pfn(kvm, gfn, true, NULL, true, NULL);
1245 1246 1247
}
EXPORT_SYMBOL_GPL(gfn_to_pfn_atomic);

1248 1249
pfn_t gfn_to_pfn_async(struct kvm *kvm, gfn_t gfn, bool *async,
		       bool write_fault, bool *writable)
1250
{
1251
	return __gfn_to_pfn(kvm, gfn, false, async, write_fault, writable);
1252 1253 1254
}
EXPORT_SYMBOL_GPL(gfn_to_pfn_async);

1255 1256
pfn_t gfn_to_pfn(struct kvm *kvm, gfn_t gfn)
{
1257
	return __gfn_to_pfn(kvm, gfn, false, NULL, true, NULL);
1258
}
1259 1260
EXPORT_SYMBOL_GPL(gfn_to_pfn);

1261 1262 1263 1264 1265 1266 1267
pfn_t gfn_to_pfn_prot(struct kvm *kvm, gfn_t gfn, bool write_fault,
		      bool *writable)
{
	return __gfn_to_pfn(kvm, gfn, false, NULL, write_fault, writable);
}
EXPORT_SYMBOL_GPL(gfn_to_pfn_prot);

1268
pfn_t gfn_to_pfn_memslot(struct kvm_memory_slot *slot, gfn_t gfn)
1269
{
X
Xiao Guangrong 已提交
1270
	return __gfn_to_pfn_memslot(slot, gfn, false, NULL, true, NULL);
1271 1272
}

1273 1274
pfn_t gfn_to_pfn_memslot_atomic(struct kvm_memory_slot *slot, gfn_t gfn)
{
X
Xiao Guangrong 已提交
1275
	return __gfn_to_pfn_memslot(slot, gfn, true, NULL, true, NULL);
1276 1277 1278
}
EXPORT_SYMBOL_GPL(gfn_to_pfn_memslot_atomic);

1279 1280 1281 1282 1283 1284
int gfn_to_page_many_atomic(struct kvm *kvm, gfn_t gfn, struct page **pages,
								  int nr_pages)
{
	unsigned long addr;
	gfn_t entry;

1285
	addr = gfn_to_hva_many(gfn_to_memslot(kvm, gfn), gfn, &entry);
1286 1287 1288 1289 1290 1291 1292 1293 1294 1295
	if (kvm_is_error_hva(addr))
		return -1;

	if (entry < nr_pages)
		return 0;

	return __get_user_pages_fast(addr, nr_pages, 1, pages);
}
EXPORT_SYMBOL_GPL(gfn_to_page_many_atomic);

X
Xiao Guangrong 已提交
1296 1297
static struct page *kvm_pfn_to_page(pfn_t pfn)
{
1298 1299
	if (is_error_pfn(pfn))
		return KVM_ERR_PTR_BAD_PAGE;
X
Xiao Guangrong 已提交
1300

1301 1302
	if (kvm_is_mmio_pfn(pfn)) {
		WARN_ON(1);
1303
		return KVM_ERR_PTR_BAD_PAGE;
1304
	}
X
Xiao Guangrong 已提交
1305 1306 1307 1308

	return pfn_to_page(pfn);
}

1309 1310
struct page *gfn_to_page(struct kvm *kvm, gfn_t gfn)
{
1311 1312 1313 1314
	pfn_t pfn;

	pfn = gfn_to_pfn(kvm, gfn);

X
Xiao Guangrong 已提交
1315
	return kvm_pfn_to_page(pfn);
A
Avi Kivity 已提交
1316
}
1317

A
Avi Kivity 已提交
1318 1319
EXPORT_SYMBOL_GPL(gfn_to_page);

1320 1321
void kvm_release_page_clean(struct page *page)
{
1322 1323 1324
	WARN_ON(is_error_page(page));

	kvm_release_pfn_clean(page_to_pfn(page));
1325 1326 1327
}
EXPORT_SYMBOL_GPL(kvm_release_page_clean);

1328 1329
void kvm_release_pfn_clean(pfn_t pfn)
{
1330 1331 1332
	WARN_ON(is_error_pfn(pfn));

	if (!kvm_is_mmio_pfn(pfn))
1333
		put_page(pfn_to_page(pfn));
1334 1335 1336
}
EXPORT_SYMBOL_GPL(kvm_release_pfn_clean);

1337
void kvm_release_page_dirty(struct page *page)
1338
{
X
Xiao Guangrong 已提交
1339 1340
	WARN_ON(is_error_page(page));

1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359
	kvm_release_pfn_dirty(page_to_pfn(page));
}
EXPORT_SYMBOL_GPL(kvm_release_page_dirty);

void kvm_release_pfn_dirty(pfn_t pfn)
{
	kvm_set_pfn_dirty(pfn);
	kvm_release_pfn_clean(pfn);
}
EXPORT_SYMBOL_GPL(kvm_release_pfn_dirty);

void kvm_set_page_dirty(struct page *page)
{
	kvm_set_pfn_dirty(page_to_pfn(page));
}
EXPORT_SYMBOL_GPL(kvm_set_page_dirty);

void kvm_set_pfn_dirty(pfn_t pfn)
{
1360
	if (!kvm_is_mmio_pfn(pfn)) {
1361 1362 1363 1364
		struct page *page = pfn_to_page(pfn);
		if (!PageReserved(page))
			SetPageDirty(page);
	}
1365
}
1366 1367 1368 1369
EXPORT_SYMBOL_GPL(kvm_set_pfn_dirty);

void kvm_set_pfn_accessed(pfn_t pfn)
{
1370
	if (!kvm_is_mmio_pfn(pfn))
1371
		mark_page_accessed(pfn_to_page(pfn));
1372 1373 1374 1375 1376
}
EXPORT_SYMBOL_GPL(kvm_set_pfn_accessed);

void kvm_get_pfn(pfn_t pfn)
{
1377
	if (!kvm_is_mmio_pfn(pfn))
1378
		get_page(pfn_to_page(pfn));
1379 1380
}
EXPORT_SYMBOL_GPL(kvm_get_pfn);
1381

1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392
static int next_segment(unsigned long len, int offset)
{
	if (len > PAGE_SIZE - offset)
		return PAGE_SIZE - offset;
	else
		return len;
}

int kvm_read_guest_page(struct kvm *kvm, gfn_t gfn, void *data, int offset,
			int len)
{
1393 1394
	int r;
	unsigned long addr;
1395

1396
	addr = gfn_to_hva_read(kvm, gfn);
1397 1398
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1399
	r = kvm_read_hva(data, (void __user *)addr + offset, len);
1400
	if (r)
1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425
		return -EFAULT;
	return 0;
}
EXPORT_SYMBOL_GPL(kvm_read_guest_page);

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);

1426 1427 1428 1429 1430 1431 1432 1433
int kvm_read_guest_atomic(struct kvm *kvm, gpa_t gpa, void *data,
			  unsigned long len)
{
	int r;
	unsigned long addr;
	gfn_t gfn = gpa >> PAGE_SHIFT;
	int offset = offset_in_page(gpa);

1434
	addr = gfn_to_hva_read(kvm, gfn);
1435 1436
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1437
	pagefault_disable();
1438
	r = kvm_read_hva_atomic(data, (void __user *)addr + offset, len);
1439
	pagefault_enable();
1440 1441 1442 1443 1444 1445
	if (r)
		return -EFAULT;
	return 0;
}
EXPORT_SYMBOL(kvm_read_guest_atomic);

1446 1447 1448
int kvm_write_guest_page(struct kvm *kvm, gfn_t gfn, const void *data,
			 int offset, int len)
{
1449 1450
	int r;
	unsigned long addr;
1451

1452 1453 1454
	addr = gfn_to_hva(kvm, gfn);
	if (kvm_is_error_hva(addr))
		return -EFAULT;
1455
	r = __copy_to_user((void __user *)addr + offset, data, len);
1456
	if (r)
1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482
		return -EFAULT;
	mark_page_dirty(kvm, gfn);
	return 0;
}
EXPORT_SYMBOL_GPL(kvm_write_guest_page);

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;
}

1483 1484 1485 1486 1487 1488 1489 1490 1491
int kvm_gfn_to_hva_cache_init(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
			      gpa_t gpa)
{
	struct kvm_memslots *slots = kvm_memslots(kvm);
	int offset = offset_in_page(gpa);
	gfn_t gfn = gpa >> PAGE_SHIFT;

	ghc->gpa = gpa;
	ghc->generation = slots->generation;
1492
	ghc->memslot = gfn_to_memslot(kvm, gfn);
1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514
	ghc->hva = gfn_to_hva_many(ghc->memslot, gfn, NULL);
	if (!kvm_is_error_hva(ghc->hva))
		ghc->hva += offset;
	else
		return -EFAULT;

	return 0;
}
EXPORT_SYMBOL_GPL(kvm_gfn_to_hva_cache_init);

int kvm_write_guest_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
			   void *data, unsigned long len)
{
	struct kvm_memslots *slots = kvm_memslots(kvm);
	int r;

	if (slots->generation != ghc->generation)
		kvm_gfn_to_hva_cache_init(kvm, ghc, ghc->gpa);

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

1515
	r = __copy_to_user((void __user *)ghc->hva, data, len);
1516 1517 1518 1519 1520 1521 1522 1523
	if (r)
		return -EFAULT;
	mark_page_dirty_in_slot(kvm, ghc->memslot, ghc->gpa >> PAGE_SHIFT);

	return 0;
}
EXPORT_SYMBOL_GPL(kvm_write_guest_cached);

1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543
int kvm_read_guest_cached(struct kvm *kvm, struct gfn_to_hva_cache *ghc,
			   void *data, unsigned long len)
{
	struct kvm_memslots *slots = kvm_memslots(kvm);
	int r;

	if (slots->generation != ghc->generation)
		kvm_gfn_to_hva_cache_init(kvm, ghc, ghc->gpa);

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

	r = __copy_from_user(data, (void __user *)ghc->hva, len);
	if (r)
		return -EFAULT;

	return 0;
}
EXPORT_SYMBOL_GPL(kvm_read_guest_cached);

1544 1545
int kvm_clear_guest_page(struct kvm *kvm, gfn_t gfn, int offset, int len)
{
1546 1547
	return kvm_write_guest_page(kvm, gfn, (const void *) empty_zero_page,
				    offset, len);
1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569
}
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;

        while ((seg = next_segment(len, offset)) != 0) {
		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);

1570 1571
void mark_page_dirty_in_slot(struct kvm *kvm, struct kvm_memory_slot *memslot,
			     gfn_t gfn)
A
Avi Kivity 已提交
1572
{
R
Rusty Russell 已提交
1573 1574
	if (memslot && memslot->dirty_bitmap) {
		unsigned long rel_gfn = gfn - memslot->base_gfn;
A
Avi Kivity 已提交
1575

1576 1577
		/* TODO: introduce set_bit_le() and use it */
		test_and_set_bit_le(rel_gfn, memslot->dirty_bitmap);
A
Avi Kivity 已提交
1578 1579 1580
	}
}

1581 1582 1583 1584 1585 1586 1587 1588
void mark_page_dirty(struct kvm *kvm, gfn_t gfn)
{
	struct kvm_memory_slot *memslot;

	memslot = gfn_to_memslot(kvm, gfn);
	mark_page_dirty_in_slot(kvm, memslot, gfn);
}

E
Eddie Dong 已提交
1589 1590 1591
/*
 * The vCPU has executed a HLT instruction with in-kernel mode enabled.
 */
1592
void kvm_vcpu_block(struct kvm_vcpu *vcpu)
1593
{
1594 1595 1596 1597 1598
	DEFINE_WAIT(wait);

	for (;;) {
		prepare_to_wait(&vcpu->wq, &wait, TASK_INTERRUPTIBLE);

1599
		if (kvm_arch_vcpu_runnable(vcpu)) {
1600
			kvm_make_request(KVM_REQ_UNHALT, vcpu);
1601
			break;
1602
		}
1603 1604
		if (kvm_cpu_has_pending_timer(vcpu))
			break;
1605 1606 1607
		if (signal_pending(current))
			break;

E
Eddie Dong 已提交
1608 1609
		schedule();
	}
1610

1611
	finish_wait(&vcpu->wq, &wait);
E
Eddie Dong 已提交
1612 1613
}

1614
#ifndef CONFIG_S390
1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635
/*
 * 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;
	wait_queue_head_t *wqp;

	wqp = kvm_arch_vcpu_wq(vcpu);
	if (waitqueue_active(wqp)) {
		wake_up_interruptible(wqp);
		++vcpu->stat.halt_wakeup;
	}

	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();
}
1636
#endif /* !CONFIG_S390 */
1637

A
Avi Kivity 已提交
1638 1639
void kvm_resched(struct kvm_vcpu *vcpu)
{
1640 1641
	if (!need_resched())
		return;
A
Avi Kivity 已提交
1642 1643 1644 1645
	cond_resched();
}
EXPORT_SYMBOL_GPL(kvm_resched);

1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670
bool kvm_vcpu_yield_to(struct kvm_vcpu *target)
{
	struct pid *pid;
	struct task_struct *task = NULL;

	rcu_read_lock();
	pid = rcu_dereference(target->pid);
	if (pid)
		task = get_pid_task(target->pid, PIDTYPE_PID);
	rcu_read_unlock();
	if (!task)
		return false;
	if (task->flags & PF_VCPU) {
		put_task_struct(task);
		return false;
	}
	if (yield_to(task, 1)) {
		put_task_struct(task);
		return true;
	}
	put_task_struct(task);
	return false;
}
EXPORT_SYMBOL_GPL(kvm_vcpu_yield_to);

1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707
#ifdef CONFIG_HAVE_KVM_CPU_RELAX_INTERCEPT
/*
 * Helper that checks whether a VCPU is eligible for directed yield.
 * Most eligible candidate to yield is decided by following heuristics:
 *
 *  (a) VCPU which has not done pl-exit or cpu relax intercepted recently
 *  (preempted lock holder), indicated by @in_spin_loop.
 *  Set at the beiginning and cleared at the end of interception/PLE handler.
 *
 *  (b) VCPU which has done pl-exit/ cpu relax intercepted but did not get
 *  chance last time (mostly it has become eligible now since we have probably
 *  yielded to lockholder in last iteration. This is done by toggling
 *  @dy_eligible each time a VCPU checked for eligibility.)
 *
 *  Yielding to a recently pl-exited/cpu relax intercepted VCPU before yielding
 *  to preempted lock-holder could result in wrong VCPU selection and CPU
 *  burning. Giving priority for a potential lock-holder increases lock
 *  progress.
 *
 *  Since algorithm is based on heuristics, accessing another VCPU data without
 *  locking does not harm. It may result in trying to yield to  same VCPU, fail
 *  and continue with next VCPU and so on.
 */
bool kvm_vcpu_eligible_for_directed_yield(struct kvm_vcpu *vcpu)
{
	bool eligible;

	eligible = !vcpu->spin_loop.in_spin_loop ||
			(vcpu->spin_loop.in_spin_loop &&
			 vcpu->spin_loop.dy_eligible);

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

	return eligible;
}
#endif
1708
void kvm_vcpu_on_spin(struct kvm_vcpu *me)
Z
Zhai, Edwin 已提交
1709
{
1710 1711 1712 1713 1714 1715
	struct kvm *kvm = me->kvm;
	struct kvm_vcpu *vcpu;
	int last_boosted_vcpu = me->kvm->last_boosted_vcpu;
	int yielded = 0;
	int pass;
	int i;
Z
Zhai, Edwin 已提交
1716

1717
	kvm_vcpu_set_in_spin_loop(me, true);
1718 1719 1720 1721 1722 1723 1724 1725 1726
	/*
	 * 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.
	 */
	for (pass = 0; pass < 2 && !yielded; pass++) {
		kvm_for_each_vcpu(i, vcpu, kvm) {
1727
			if (!pass && i <= last_boosted_vcpu) {
1728 1729 1730 1731 1732 1733 1734 1735
				i = last_boosted_vcpu;
				continue;
			} else if (pass && i > last_boosted_vcpu)
				break;
			if (vcpu == me)
				continue;
			if (waitqueue_active(&vcpu->wq))
				continue;
1736 1737
			if (!kvm_vcpu_eligible_for_directed_yield(vcpu))
				continue;
1738
			if (kvm_vcpu_yield_to(vcpu)) {
1739 1740 1741 1742 1743 1744
				kvm->last_boosted_vcpu = i;
				yielded = 1;
				break;
			}
		}
	}
1745
	kvm_vcpu_set_in_spin_loop(me, false);
1746 1747 1748

	/* Ensure vcpu is not eligible during next spinloop */
	kvm_vcpu_set_dy_eligible(me, false);
Z
Zhai, Edwin 已提交
1749 1750 1751
}
EXPORT_SYMBOL_GPL(kvm_vcpu_on_spin);

1752
static int kvm_vcpu_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
1753 1754 1755 1756
{
	struct kvm_vcpu *vcpu = vma->vm_file->private_data;
	struct page *page;

1757
	if (vmf->pgoff == 0)
1758
		page = virt_to_page(vcpu->run);
A
Avi Kivity 已提交
1759
#ifdef CONFIG_X86
1760
	else if (vmf->pgoff == KVM_PIO_PAGE_OFFSET)
1761
		page = virt_to_page(vcpu->arch.pio_data);
1762 1763 1764 1765
#endif
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
	else if (vmf->pgoff == KVM_COALESCED_MMIO_PAGE_OFFSET)
		page = virt_to_page(vcpu->kvm->coalesced_mmio_ring);
A
Avi Kivity 已提交
1766
#endif
1767
	else
1768
		return kvm_arch_vcpu_fault(vcpu, vmf);
1769
	get_page(page);
1770 1771
	vmf->page = page;
	return 0;
1772 1773
}

1774
static const struct vm_operations_struct kvm_vcpu_vm_ops = {
1775
	.fault = kvm_vcpu_fault,
1776 1777 1778 1779 1780 1781 1782 1783
};

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
1784 1785 1786 1787
static int kvm_vcpu_release(struct inode *inode, struct file *filp)
{
	struct kvm_vcpu *vcpu = filp->private_data;

A
Al Viro 已提交
1788
	kvm_put_kvm(vcpu->kvm);
A
Avi Kivity 已提交
1789 1790 1791
	return 0;
}

1792
static struct file_operations kvm_vcpu_fops = {
A
Avi Kivity 已提交
1793 1794
	.release        = kvm_vcpu_release,
	.unlocked_ioctl = kvm_vcpu_ioctl,
1795 1796 1797
#ifdef CONFIG_COMPAT
	.compat_ioctl   = kvm_vcpu_compat_ioctl,
#endif
1798
	.mmap           = kvm_vcpu_mmap,
1799
	.llseek		= noop_llseek,
A
Avi Kivity 已提交
1800 1801 1802 1803 1804 1805 1806
};

/*
 * Allocates an inode for the vcpu.
 */
static int create_vcpu_fd(struct kvm_vcpu *vcpu)
{
1807
	return anon_inode_getfd("kvm-vcpu", &kvm_vcpu_fops, vcpu, O_RDWR);
A
Avi Kivity 已提交
1808 1809
}

1810 1811 1812
/*
 * Creates some virtual cpus.  Good luck creating more than one.
 */
1813
static int kvm_vm_ioctl_create_vcpu(struct kvm *kvm, u32 id)
1814 1815
{
	int r;
1816
	struct kvm_vcpu *vcpu, *v;
1817

1818
	vcpu = kvm_arch_vcpu_create(kvm, id);
R
Rusty Russell 已提交
1819 1820
	if (IS_ERR(vcpu))
		return PTR_ERR(vcpu);
1821

1822 1823
	preempt_notifier_init(&vcpu->preempt_notifier, &kvm_preempt_ops);

1824 1825
	r = kvm_arch_vcpu_setup(vcpu);
	if (r)
1826
		goto vcpu_destroy;
1827

S
Shaohua Li 已提交
1828
	mutex_lock(&kvm->lock);
1829 1830 1831 1832
	if (!kvm_vcpu_compatible(vcpu)) {
		r = -EINVAL;
		goto unlock_vcpu_destroy;
	}
1833 1834
	if (atomic_read(&kvm->online_vcpus) == KVM_MAX_VCPUS) {
		r = -EINVAL;
1835
		goto unlock_vcpu_destroy;
R
Rusty Russell 已提交
1836
	}
1837

1838 1839
	kvm_for_each_vcpu(r, v, kvm)
		if (v->vcpu_id == id) {
1840
			r = -EEXIST;
1841
			goto unlock_vcpu_destroy;
1842 1843 1844
		}

	BUG_ON(kvm->vcpus[atomic_read(&kvm->online_vcpus)]);
1845

R
Rusty Russell 已提交
1846
	/* Now it's all set up, let userspace reach it */
A
Al Viro 已提交
1847
	kvm_get_kvm(kvm);
A
Avi Kivity 已提交
1848
	r = create_vcpu_fd(vcpu);
1849 1850
	if (r < 0) {
		kvm_put_kvm(kvm);
1851
		goto unlock_vcpu_destroy;
1852 1853 1854 1855 1856 1857 1858
	}

	kvm->vcpus[atomic_read(&kvm->online_vcpus)] = vcpu;
	smp_wmb();
	atomic_inc(&kvm->online_vcpus);

	mutex_unlock(&kvm->lock);
R
Rusty Russell 已提交
1859
	return r;
1860

1861
unlock_vcpu_destroy:
1862
	mutex_unlock(&kvm->lock);
1863
vcpu_destroy:
1864
	kvm_arch_vcpu_destroy(vcpu);
1865 1866 1867
	return r;
}

A
Avi Kivity 已提交
1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878
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
1879 1880
static long kvm_vcpu_ioctl(struct file *filp,
			   unsigned int ioctl, unsigned long arg)
A
Avi Kivity 已提交
1881
{
A
Avi Kivity 已提交
1882
	struct kvm_vcpu *vcpu = filp->private_data;
A
Al Viro 已提交
1883
	void __user *argp = (void __user *)arg;
1884
	int r;
1885 1886
	struct kvm_fpu *fpu = NULL;
	struct kvm_sregs *kvm_sregs = NULL;
A
Avi Kivity 已提交
1887

1888 1889
	if (vcpu->kvm->mm != current->mm)
		return -EIO;
1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901

#if defined(CONFIG_S390) || defined(CONFIG_PPC)
	/*
	 * Special cases: vcpu ioctls that are asynchronous to vcpu execution,
	 * so vcpu_load() would break it.
	 */
	if (ioctl == KVM_S390_INTERRUPT || ioctl == KVM_INTERRUPT)
		return kvm_arch_vcpu_ioctl(filp, ioctl, arg);
#endif


	vcpu_load(vcpu);
A
Avi Kivity 已提交
1902
	switch (ioctl) {
1903
	case KVM_RUN:
1904 1905 1906
		r = -EINVAL;
		if (arg)
			goto out;
1907
		r = kvm_arch_vcpu_ioctl_run(vcpu, vcpu->run);
1908
		trace_kvm_userspace_exit(vcpu->run->exit_reason, r);
A
Avi Kivity 已提交
1909 1910
		break;
	case KVM_GET_REGS: {
1911
		struct kvm_regs *kvm_regs;
A
Avi Kivity 已提交
1912

1913 1914 1915
		r = -ENOMEM;
		kvm_regs = kzalloc(sizeof(struct kvm_regs), GFP_KERNEL);
		if (!kvm_regs)
A
Avi Kivity 已提交
1916
			goto out;
1917 1918 1919
		r = kvm_arch_vcpu_ioctl_get_regs(vcpu, kvm_regs);
		if (r)
			goto out_free1;
A
Avi Kivity 已提交
1920
		r = -EFAULT;
1921 1922
		if (copy_to_user(argp, kvm_regs, sizeof(struct kvm_regs)))
			goto out_free1;
A
Avi Kivity 已提交
1923
		r = 0;
1924 1925
out_free1:
		kfree(kvm_regs);
A
Avi Kivity 已提交
1926 1927 1928
		break;
	}
	case KVM_SET_REGS: {
1929
		struct kvm_regs *kvm_regs;
A
Avi Kivity 已提交
1930

1931
		r = -ENOMEM;
1932 1933 1934
		kvm_regs = memdup_user(argp, sizeof(*kvm_regs));
		if (IS_ERR(kvm_regs)) {
			r = PTR_ERR(kvm_regs);
A
Avi Kivity 已提交
1935
			goto out;
1936
		}
1937
		r = kvm_arch_vcpu_ioctl_set_regs(vcpu, kvm_regs);
A
Avi Kivity 已提交
1938
		if (r)
1939
			goto out_free2;
A
Avi Kivity 已提交
1940
		r = 0;
1941 1942
out_free2:
		kfree(kvm_regs);
A
Avi Kivity 已提交
1943 1944 1945
		break;
	}
	case KVM_GET_SREGS: {
1946 1947 1948 1949 1950
		kvm_sregs = kzalloc(sizeof(struct kvm_sregs), GFP_KERNEL);
		r = -ENOMEM;
		if (!kvm_sregs)
			goto out;
		r = kvm_arch_vcpu_ioctl_get_sregs(vcpu, kvm_sregs);
A
Avi Kivity 已提交
1951 1952 1953
		if (r)
			goto out;
		r = -EFAULT;
1954
		if (copy_to_user(argp, kvm_sregs, sizeof(struct kvm_sregs)))
A
Avi Kivity 已提交
1955 1956 1957 1958 1959
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_SREGS: {
1960 1961 1962
		kvm_sregs = memdup_user(argp, sizeof(*kvm_sregs));
		if (IS_ERR(kvm_sregs)) {
			r = PTR_ERR(kvm_sregs);
A
Avi Kivity 已提交
1963
			goto out;
1964
		}
1965
		r = kvm_arch_vcpu_ioctl_set_sregs(vcpu, kvm_sregs);
A
Avi Kivity 已提交
1966 1967 1968 1969 1970
		if (r)
			goto out;
		r = 0;
		break;
	}
1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
	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;
		if (copy_to_user(argp, &mp_state, sizeof mp_state))
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_MP_STATE: {
		struct kvm_mp_state mp_state;

		r = -EFAULT;
		if (copy_from_user(&mp_state, argp, sizeof mp_state))
			goto out;
		r = kvm_arch_vcpu_ioctl_set_mpstate(vcpu, &mp_state);
		if (r)
			goto out;
		r = 0;
		break;
	}
A
Avi Kivity 已提交
1995 1996 1997 1998
	case KVM_TRANSLATE: {
		struct kvm_translation tr;

		r = -EFAULT;
A
Al Viro 已提交
1999
		if (copy_from_user(&tr, argp, sizeof tr))
A
Avi Kivity 已提交
2000
			goto out;
2001
		r = kvm_arch_vcpu_ioctl_translate(vcpu, &tr);
A
Avi Kivity 已提交
2002 2003 2004
		if (r)
			goto out;
		r = -EFAULT;
A
Al Viro 已提交
2005
		if (copy_to_user(argp, &tr, sizeof tr))
A
Avi Kivity 已提交
2006 2007 2008 2009
			goto out;
		r = 0;
		break;
	}
J
Jan Kiszka 已提交
2010 2011
	case KVM_SET_GUEST_DEBUG: {
		struct kvm_guest_debug dbg;
A
Avi Kivity 已提交
2012 2013

		r = -EFAULT;
A
Al Viro 已提交
2014
		if (copy_from_user(&dbg, argp, sizeof dbg))
A
Avi Kivity 已提交
2015
			goto out;
J
Jan Kiszka 已提交
2016
		r = kvm_arch_vcpu_ioctl_set_guest_debug(vcpu, &dbg);
A
Avi Kivity 已提交
2017 2018 2019 2020 2021
		if (r)
			goto out;
		r = 0;
		break;
	}
A
Avi Kivity 已提交
2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041
	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,
					   sizeof kvm_sigmask))
				goto out;
			r = -EINVAL;
			if (kvm_sigmask.len != sizeof sigset)
				goto out;
			r = -EFAULT;
			if (copy_from_user(&sigset, sigmask_arg->sigset,
					   sizeof sigset))
				goto out;
			p = &sigset;
		}
2042
		r = kvm_vcpu_ioctl_set_sigmask(vcpu, p);
A
Avi Kivity 已提交
2043 2044
		break;
	}
A
Avi Kivity 已提交
2045
	case KVM_GET_FPU: {
2046 2047 2048 2049 2050
		fpu = kzalloc(sizeof(struct kvm_fpu), GFP_KERNEL);
		r = -ENOMEM;
		if (!fpu)
			goto out;
		r = kvm_arch_vcpu_ioctl_get_fpu(vcpu, fpu);
A
Avi Kivity 已提交
2051 2052 2053
		if (r)
			goto out;
		r = -EFAULT;
2054
		if (copy_to_user(argp, fpu, sizeof(struct kvm_fpu)))
A
Avi Kivity 已提交
2055 2056 2057 2058 2059
			goto out;
		r = 0;
		break;
	}
	case KVM_SET_FPU: {
2060 2061 2062
		fpu = memdup_user(argp, sizeof(*fpu));
		if (IS_ERR(fpu)) {
			r = PTR_ERR(fpu);
A
Avi Kivity 已提交
2063
			goto out;
2064
		}
2065
		r = kvm_arch_vcpu_ioctl_set_fpu(vcpu, fpu);
A
Avi Kivity 已提交
2066 2067 2068 2069 2070
		if (r)
			goto out;
		r = 0;
		break;
	}
A
Avi Kivity 已提交
2071
	default:
2072
		r = kvm_arch_vcpu_ioctl(filp, ioctl, arg);
A
Avi Kivity 已提交
2073 2074
	}
out:
2075
	vcpu_put(vcpu);
2076 2077
	kfree(fpu);
	kfree(kvm_sregs);
A
Avi Kivity 已提交
2078 2079 2080
	return r;
}

2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124
#ifdef CONFIG_COMPAT
static long kvm_vcpu_compat_ioctl(struct file *filp,
				  unsigned int ioctl, unsigned long arg)
{
	struct kvm_vcpu *vcpu = filp->private_data;
	void __user *argp = compat_ptr(arg);
	int r;

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

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

		if (argp) {
			r = -EFAULT;
			if (copy_from_user(&kvm_sigmask, argp,
					   sizeof kvm_sigmask))
				goto out;
			r = -EINVAL;
			if (kvm_sigmask.len != sizeof csigset)
				goto out;
			r = -EFAULT;
			if (copy_from_user(&csigset, sigmask_arg->sigset,
					   sizeof csigset))
				goto out;
		}
		sigset_from_compat(&sigset, &csigset);
		r = kvm_vcpu_ioctl_set_sigmask(vcpu, &sigset);
		break;
	}
	default:
		r = kvm_vcpu_ioctl(filp, ioctl, arg);
	}

out:
	return r;
}
#endif

A
Avi Kivity 已提交
2125 2126 2127 2128 2129
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;
2130
	int r;
A
Avi Kivity 已提交
2131

2132 2133
	if (kvm->mm != current->mm)
		return -EIO;
A
Avi Kivity 已提交
2134 2135 2136 2137 2138 2139
	switch (ioctl) {
	case KVM_CREATE_VCPU:
		r = kvm_vm_ioctl_create_vcpu(kvm, arg);
		if (r < 0)
			goto out;
		break;
2140 2141 2142 2143 2144 2145 2146 2147 2148
	case KVM_SET_USER_MEMORY_REGION: {
		struct kvm_userspace_memory_region kvm_userspace_mem;

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

		r = kvm_vm_ioctl_set_memory_region(kvm, &kvm_userspace_mem, 1);
A
Avi Kivity 已提交
2149 2150 2151 2152 2153 2154 2155 2156
		if (r)
			goto out;
		break;
	}
	case KVM_GET_DIRTY_LOG: {
		struct kvm_dirty_log log;

		r = -EFAULT;
A
Al Viro 已提交
2157
		if (copy_from_user(&log, argp, sizeof log))
A
Avi Kivity 已提交
2158
			goto out;
2159
		r = kvm_vm_ioctl_get_dirty_log(kvm, &log);
A
Avi Kivity 已提交
2160 2161 2162 2163
		if (r)
			goto out;
		break;
	}
2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
	case KVM_REGISTER_COALESCED_MMIO: {
		struct kvm_coalesced_mmio_zone zone;
		r = -EFAULT;
		if (copy_from_user(&zone, argp, sizeof zone))
			goto out;
		r = kvm_vm_ioctl_register_coalesced_mmio(kvm, &zone);
		if (r)
			goto out;
		r = 0;
		break;
	}
	case KVM_UNREGISTER_COALESCED_MMIO: {
		struct kvm_coalesced_mmio_zone zone;
		r = -EFAULT;
		if (copy_from_user(&zone, argp, sizeof zone))
			goto out;
		r = kvm_vm_ioctl_unregister_coalesced_mmio(kvm, &zone);
		if (r)
			goto out;
		r = 0;
		break;
	}
#endif
G
Gregory Haskins 已提交
2188 2189 2190 2191 2192 2193
	case KVM_IRQFD: {
		struct kvm_irqfd data;

		r = -EFAULT;
		if (copy_from_user(&data, argp, sizeof data))
			goto out;
2194
		r = kvm_irqfd(kvm, &data);
G
Gregory Haskins 已提交
2195 2196
		break;
	}
G
Gregory Haskins 已提交
2197 2198 2199 2200 2201 2202 2203 2204 2205
	case KVM_IOEVENTFD: {
		struct kvm_ioeventfd data;

		r = -EFAULT;
		if (copy_from_user(&data, argp, sizeof data))
			goto out;
		r = kvm_ioeventfd(kvm, &data);
		break;
	}
2206 2207 2208
#ifdef CONFIG_KVM_APIC_ARCHITECTURE
	case KVM_SET_BOOT_CPU_ID:
		r = 0;
2209
		mutex_lock(&kvm->lock);
2210 2211 2212 2213
		if (atomic_read(&kvm->online_vcpus) != 0)
			r = -EBUSY;
		else
			kvm->bsp_vcpu_id = arg;
2214
		mutex_unlock(&kvm->lock);
2215
		break;
2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226
#endif
#ifdef CONFIG_HAVE_KVM_MSI
	case KVM_SIGNAL_MSI: {
		struct kvm_msi msi;

		r = -EFAULT;
		if (copy_from_user(&msi, argp, sizeof msi))
			goto out;
		r = kvm_send_userspace_msi(kvm, &msi);
		break;
	}
2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249
#endif
#ifdef __KVM_HAVE_IRQ_LINE
	case KVM_IRQ_LINE_STATUS:
	case KVM_IRQ_LINE: {
		struct kvm_irq_level irq_event;

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

		r = kvm_vm_ioctl_irq_line(kvm, &irq_event);
		if (r)
			goto out;

		r = -EFAULT;
		if (ioctl == KVM_IRQ_LINE_STATUS) {
			if (copy_to_user(argp, &irq_event, sizeof irq_event))
				goto out;
		}

		r = 0;
		break;
	}
2250
#endif
2251
	default:
2252
		r = kvm_arch_vm_ioctl(filp, ioctl, arg);
2253 2254
		if (r == -ENOTTY)
			r = kvm_vm_ioctl_assigned_device(kvm, ioctl, arg);
2255 2256 2257 2258 2259
	}
out:
	return r;
}

2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305
#ifdef CONFIG_COMPAT
struct compat_kvm_dirty_log {
	__u32 slot;
	__u32 padding1;
	union {
		compat_uptr_t dirty_bitmap; /* one bit per page */
		__u64 padding2;
	};
};

static long kvm_vm_compat_ioctl(struct file *filp,
			   unsigned int ioctl, unsigned long arg)
{
	struct kvm *kvm = filp->private_data;
	int r;

	if (kvm->mm != current->mm)
		return -EIO;
	switch (ioctl) {
	case KVM_GET_DIRTY_LOG: {
		struct compat_kvm_dirty_log compat_log;
		struct kvm_dirty_log log;

		r = -EFAULT;
		if (copy_from_user(&compat_log, (void __user *)arg,
				   sizeof(compat_log)))
			goto out;
		log.slot	 = compat_log.slot;
		log.padding1	 = compat_log.padding1;
		log.padding2	 = compat_log.padding2;
		log.dirty_bitmap = compat_ptr(compat_log.dirty_bitmap);

		r = kvm_vm_ioctl_get_dirty_log(kvm, &log);
		if (r)
			goto out;
		break;
	}
	default:
		r = kvm_vm_ioctl(filp, ioctl, arg);
	}

out:
	return r;
}
#endif

2306
static int kvm_vm_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
2307
{
2308 2309 2310 2311
	struct page *page[1];
	unsigned long addr;
	int npages;
	gfn_t gfn = vmf->pgoff;
2312 2313
	struct kvm *kvm = vma->vm_file->private_data;

2314 2315
	addr = gfn_to_hva(kvm, gfn);
	if (kvm_is_error_hva(addr))
2316
		return VM_FAULT_SIGBUS;
2317 2318 2319 2320

	npages = get_user_pages(current, current->mm, addr, 1, 1, 0, page,
				NULL);
	if (unlikely(npages != 1))
2321
		return VM_FAULT_SIGBUS;
2322 2323

	vmf->page = page[0];
2324
	return 0;
2325 2326
}

2327
static const struct vm_operations_struct kvm_vm_vm_ops = {
2328
	.fault = kvm_vm_fault,
2329 2330 2331 2332 2333 2334 2335 2336
};

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

2337
static struct file_operations kvm_vm_fops = {
2338 2339
	.release        = kvm_vm_release,
	.unlocked_ioctl = kvm_vm_ioctl,
2340 2341 2342
#ifdef CONFIG_COMPAT
	.compat_ioctl   = kvm_vm_compat_ioctl,
#endif
2343
	.mmap           = kvm_vm_mmap,
2344
	.llseek		= noop_llseek,
2345 2346
};

2347
static int kvm_dev_ioctl_create_vm(unsigned long type)
2348
{
2349
	int r;
2350 2351
	struct kvm *kvm;

2352
	kvm = kvm_create_vm(type);
2353 2354
	if (IS_ERR(kvm))
		return PTR_ERR(kvm);
2355 2356 2357 2358 2359 2360 2361
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
	r = kvm_coalesced_mmio_init(kvm);
	if (r < 0) {
		kvm_put_kvm(kvm);
		return r;
	}
#endif
2362 2363
	r = anon_inode_getfd("kvm-vm", &kvm_vm_fops, kvm, O_RDWR);
	if (r < 0)
A
Al Viro 已提交
2364
		kvm_put_kvm(kvm);
2365

2366
	return r;
2367 2368
}

2369 2370 2371
static long kvm_dev_ioctl_check_extension_generic(long arg)
{
	switch (arg) {
2372
	case KVM_CAP_USER_MEMORY:
2373
	case KVM_CAP_DESTROY_MEMORY_REGION_WORKS:
2374
	case KVM_CAP_JOIN_MEMORY_REGIONS_WORKS:
2375 2376 2377
#ifdef CONFIG_KVM_APIC_ARCHITECTURE
	case KVM_CAP_SET_BOOT_CPU_ID:
#endif
2378
	case KVM_CAP_INTERNAL_ERROR_DATA:
2379 2380 2381
#ifdef CONFIG_HAVE_KVM_MSI
	case KVM_CAP_SIGNAL_MSI:
#endif
2382
		return 1;
2383
#ifdef KVM_CAP_IRQ_ROUTING
2384
	case KVM_CAP_IRQ_ROUTING:
2385
		return KVM_MAX_IRQ_ROUTES;
2386
#endif
2387 2388 2389 2390 2391 2392
	default:
		break;
	}
	return kvm_dev_ioctl_check_extension(arg);
}

2393 2394 2395
static long kvm_dev_ioctl(struct file *filp,
			  unsigned int ioctl, unsigned long arg)
{
2396
	long r = -EINVAL;
2397 2398 2399

	switch (ioctl) {
	case KVM_GET_API_VERSION:
2400 2401 2402
		r = -EINVAL;
		if (arg)
			goto out;
2403 2404 2405
		r = KVM_API_VERSION;
		break;
	case KVM_CREATE_VM:
2406
		r = kvm_dev_ioctl_create_vm(arg);
2407
		break;
2408
	case KVM_CHECK_EXTENSION:
2409
		r = kvm_dev_ioctl_check_extension_generic(arg);
2410
		break;
2411 2412 2413 2414
	case KVM_GET_VCPU_MMAP_SIZE:
		r = -EINVAL;
		if (arg)
			goto out;
2415 2416 2417
		r = PAGE_SIZE;     /* struct kvm_run */
#ifdef CONFIG_X86
		r += PAGE_SIZE;    /* pio data page */
2418 2419 2420
#endif
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
		r += PAGE_SIZE;    /* coalesced mmio ring page */
2421
#endif
2422
		break;
2423 2424 2425
	case KVM_TRACE_ENABLE:
	case KVM_TRACE_PAUSE:
	case KVM_TRACE_DISABLE:
2426
		r = -EOPNOTSUPP;
2427
		break;
A
Avi Kivity 已提交
2428
	default:
2429
		return kvm_arch_dev_ioctl(filp, ioctl, arg);
A
Avi Kivity 已提交
2430 2431 2432 2433 2434 2435 2436 2437
	}
out:
	return r;
}

static struct file_operations kvm_chardev_ops = {
	.unlocked_ioctl = kvm_dev_ioctl,
	.compat_ioctl   = kvm_dev_ioctl,
2438
	.llseek		= noop_llseek,
A
Avi Kivity 已提交
2439 2440 2441
};

static struct miscdevice kvm_dev = {
A
Avi Kivity 已提交
2442
	KVM_MINOR,
A
Avi Kivity 已提交
2443 2444 2445 2446
	"kvm",
	&kvm_chardev_ops,
};

2447
static void hardware_enable_nolock(void *junk)
2448 2449
{
	int cpu = raw_smp_processor_id();
2450
	int r;
2451

2452
	if (cpumask_test_cpu(cpu, cpus_hardware_enabled))
2453
		return;
2454

2455
	cpumask_set_cpu(cpu, cpus_hardware_enabled);
2456 2457 2458 2459 2460 2461 2462 2463 2464

	r = kvm_arch_hardware_enable(NULL);

	if (r) {
		cpumask_clear_cpu(cpu, cpus_hardware_enabled);
		atomic_inc(&hardware_enable_failed);
		printk(KERN_INFO "kvm: enabling virtualization on "
				 "CPU%d failed\n", cpu);
	}
2465 2466
}

2467 2468
static void hardware_enable(void *junk)
{
2469
	raw_spin_lock(&kvm_lock);
2470
	hardware_enable_nolock(junk);
2471
	raw_spin_unlock(&kvm_lock);
2472 2473 2474
}

static void hardware_disable_nolock(void *junk)
2475 2476 2477
{
	int cpu = raw_smp_processor_id();

2478
	if (!cpumask_test_cpu(cpu, cpus_hardware_enabled))
2479
		return;
2480
	cpumask_clear_cpu(cpu, cpus_hardware_enabled);
2481
	kvm_arch_hardware_disable(NULL);
2482 2483
}

2484 2485
static void hardware_disable(void *junk)
{
2486
	raw_spin_lock(&kvm_lock);
2487
	hardware_disable_nolock(junk);
2488
	raw_spin_unlock(&kvm_lock);
2489 2490
}

2491 2492 2493 2494 2495 2496
static void hardware_disable_all_nolock(void)
{
	BUG_ON(!kvm_usage_count);

	kvm_usage_count--;
	if (!kvm_usage_count)
2497
		on_each_cpu(hardware_disable_nolock, NULL, 1);
2498 2499 2500 2501
}

static void hardware_disable_all(void)
{
2502
	raw_spin_lock(&kvm_lock);
2503
	hardware_disable_all_nolock();
2504
	raw_spin_unlock(&kvm_lock);
2505 2506 2507 2508 2509 2510
}

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

2511
	raw_spin_lock(&kvm_lock);
2512 2513 2514 2515

	kvm_usage_count++;
	if (kvm_usage_count == 1) {
		atomic_set(&hardware_enable_failed, 0);
2516
		on_each_cpu(hardware_enable_nolock, NULL, 1);
2517 2518 2519 2520 2521 2522 2523

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

2524
	raw_spin_unlock(&kvm_lock);
2525 2526 2527 2528

	return r;
}

A
Avi Kivity 已提交
2529 2530 2531 2532 2533
static int kvm_cpu_hotplug(struct notifier_block *notifier, unsigned long val,
			   void *v)
{
	int cpu = (long)v;

2534 2535 2536
	if (!kvm_usage_count)
		return NOTIFY_OK;

2537
	val &= ~CPU_TASKS_FROZEN;
A
Avi Kivity 已提交
2538
	switch (val) {
2539
	case CPU_DYING:
2540 2541 2542 2543
		printk(KERN_INFO "kvm: disabling virtualization on CPU%d\n",
		       cpu);
		hardware_disable(NULL);
		break;
2544
	case CPU_STARTING:
2545 2546
		printk(KERN_INFO "kvm: enabling virtualization on CPU%d\n",
		       cpu);
2547
		hardware_enable(NULL);
A
Avi Kivity 已提交
2548 2549 2550 2551 2552
		break;
	}
	return NOTIFY_OK;
}

2553

2554
asmlinkage void kvm_spurious_fault(void)
2555 2556 2557 2558
{
	/* Fault while not rebooting.  We want the trace. */
	BUG();
}
2559
EXPORT_SYMBOL_GPL(kvm_spurious_fault);
2560

2561
static int kvm_reboot(struct notifier_block *notifier, unsigned long val,
M
Mike Day 已提交
2562
		      void *v)
2563
{
2564 2565 2566 2567 2568 2569 2570 2571
	/*
	 * 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.
	 */
	printk(KERN_INFO "kvm: exiting hardware virtualization\n");
	kvm_rebooting = true;
2572
	on_each_cpu(hardware_disable_nolock, NULL, 1);
2573 2574 2575 2576 2577 2578 2579 2580
	return NOTIFY_OK;
}

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

M
Marcelo Tosatti 已提交
2581
static void kvm_io_bus_destroy(struct kvm_io_bus *bus)
2582 2583 2584 2585
{
	int i;

	for (i = 0; i < bus->dev_count; i++) {
2586
		struct kvm_io_device *pos = bus->range[i].dev;
2587 2588 2589

		kvm_iodevice_destructor(pos);
	}
M
Marcelo Tosatti 已提交
2590
	kfree(bus);
2591 2592
}

2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643
int kvm_io_bus_sort_cmp(const void *p1, const void *p2)
{
	const struct kvm_io_range *r1 = p1;
	const struct kvm_io_range *r2 = p2;

	if (r1->addr < r2->addr)
		return -1;
	if (r1->addr + r1->len > r2->addr + r2->len)
		return 1;
	return 0;
}

int kvm_io_bus_insert_dev(struct kvm_io_bus *bus, struct kvm_io_device *dev,
			  gpa_t addr, int len)
{
	bus->range[bus->dev_count++] = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
		.dev = dev,
	};

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

	return 0;
}

int kvm_io_bus_get_first_dev(struct kvm_io_bus *bus,
			     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;

	while (off > 0 && kvm_io_bus_sort_cmp(&key, &bus->range[off-1]) == 0)
		off--;

	return off;
}

2644
/* kvm_io_bus_write - called under kvm->slots_lock */
M
Marcelo Tosatti 已提交
2645
int kvm_io_bus_write(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
2646
		     int len, const void *val)
2647
{
2648
	int idx;
2649
	struct kvm_io_bus *bus;
2650 2651 2652 2653 2654 2655
	struct kvm_io_range range;

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};
2656 2657

	bus = srcu_dereference(kvm->buses[bus_idx], &kvm->srcu);
2658 2659 2660 2661 2662 2663 2664
	idx = kvm_io_bus_get_first_dev(bus, addr, len);
	if (idx < 0)
		return -EOPNOTSUPP;

	while (idx < bus->dev_count &&
		kvm_io_bus_sort_cmp(&range, &bus->range[idx]) == 0) {
		if (!kvm_iodevice_write(bus->range[idx].dev, addr, len, val))
2665
			return 0;
2666 2667 2668
		idx++;
	}

2669 2670
	return -EOPNOTSUPP;
}
2671

2672
/* kvm_io_bus_read - called under kvm->slots_lock */
M
Marcelo Tosatti 已提交
2673 2674
int kvm_io_bus_read(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
		    int len, void *val)
2675
{
2676
	int idx;
2677
	struct kvm_io_bus *bus;
2678 2679 2680 2681 2682 2683
	struct kvm_io_range range;

	range = (struct kvm_io_range) {
		.addr = addr,
		.len = len,
	};
M
Marcelo Tosatti 已提交
2684

2685
	bus = srcu_dereference(kvm->buses[bus_idx], &kvm->srcu);
2686 2687 2688 2689 2690 2691 2692
	idx = kvm_io_bus_get_first_dev(bus, addr, len);
	if (idx < 0)
		return -EOPNOTSUPP;

	while (idx < bus->dev_count &&
		kvm_io_bus_sort_cmp(&range, &bus->range[idx]) == 0) {
		if (!kvm_iodevice_read(bus->range[idx].dev, addr, len, val))
2693
			return 0;
2694 2695 2696
		idx++;
	}

2697
	return -EOPNOTSUPP;
2698 2699
}

2700
/* Caller must hold slots_lock. */
2701 2702
int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
			    int len, struct kvm_io_device *dev)
2703
{
M
Marcelo Tosatti 已提交
2704
	struct kvm_io_bus *new_bus, *bus;
2705

M
Marcelo Tosatti 已提交
2706
	bus = kvm->buses[bus_idx];
2707
	if (bus->dev_count > NR_IOBUS_DEVS - 1)
2708
		return -ENOSPC;
2709

2710 2711
	new_bus = kzalloc(sizeof(*bus) + ((bus->dev_count + 1) *
			  sizeof(struct kvm_io_range)), GFP_KERNEL);
M
Marcelo Tosatti 已提交
2712 2713
	if (!new_bus)
		return -ENOMEM;
2714 2715
	memcpy(new_bus, bus, sizeof(*bus) + (bus->dev_count *
	       sizeof(struct kvm_io_range)));
2716
	kvm_io_bus_insert_dev(new_bus, dev, addr, len);
M
Marcelo Tosatti 已提交
2717 2718 2719
	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
	synchronize_srcu_expedited(&kvm->srcu);
	kfree(bus);
2720 2721 2722 2723

	return 0;
}

2724
/* Caller must hold slots_lock. */
M
Marcelo Tosatti 已提交
2725 2726
int kvm_io_bus_unregister_dev(struct kvm *kvm, enum kvm_bus bus_idx,
			      struct kvm_io_device *dev)
2727
{
M
Marcelo Tosatti 已提交
2728 2729
	int i, r;
	struct kvm_io_bus *new_bus, *bus;
2730

2731
	bus = kvm->buses[bus_idx];
M
Marcelo Tosatti 已提交
2732
	r = -ENOENT;
2733 2734
	for (i = 0; i < bus->dev_count; i++)
		if (bus->range[i].dev == dev) {
M
Marcelo Tosatti 已提交
2735
			r = 0;
2736 2737
			break;
		}
M
Marcelo Tosatti 已提交
2738

2739
	if (r)
M
Marcelo Tosatti 已提交
2740
		return r;
2741 2742 2743 2744 2745 2746 2747 2748 2749 2750

	new_bus = kzalloc(sizeof(*bus) + ((bus->dev_count - 1) *
			  sizeof(struct kvm_io_range)), GFP_KERNEL);
	if (!new_bus)
		return -ENOMEM;

	memcpy(new_bus, bus, sizeof(*bus) + i * sizeof(struct kvm_io_range));
	new_bus->dev_count--;
	memcpy(new_bus->range + i, bus->range + i + 1,
	       (new_bus->dev_count - i) * sizeof(struct kvm_io_range));
M
Marcelo Tosatti 已提交
2751 2752 2753 2754 2755

	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
	synchronize_srcu_expedited(&kvm->srcu);
	kfree(bus);
	return r;
2756 2757
}

A
Avi Kivity 已提交
2758 2759 2760 2761
static struct notifier_block kvm_cpu_notifier = {
	.notifier_call = kvm_cpu_hotplug,
};

2762
static int vm_stat_get(void *_offset, u64 *val)
2763 2764 2765 2766
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;

2767
	*val = 0;
2768
	raw_spin_lock(&kvm_lock);
2769
	list_for_each_entry(kvm, &vm_list, vm_list)
2770
		*val += *(u32 *)((void *)kvm + offset);
2771
	raw_spin_unlock(&kvm_lock);
2772
	return 0;
2773 2774 2775 2776
}

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

2777
static int vcpu_stat_get(void *_offset, u64 *val)
A
Avi Kivity 已提交
2778 2779 2780 2781 2782 2783
{
	unsigned offset = (long)_offset;
	struct kvm *kvm;
	struct kvm_vcpu *vcpu;
	int i;

2784
	*val = 0;
2785
	raw_spin_lock(&kvm_lock);
A
Avi Kivity 已提交
2786
	list_for_each_entry(kvm, &vm_list, vm_list)
2787 2788 2789
		kvm_for_each_vcpu(i, vcpu, kvm)
			*val += *(u32 *)((void *)vcpu + offset);

2790
	raw_spin_unlock(&kvm_lock);
2791
	return 0;
A
Avi Kivity 已提交
2792 2793
}

2794 2795
DEFINE_SIMPLE_ATTRIBUTE(vcpu_stat_fops, vcpu_stat_get, NULL, "%llu\n");

2796
static const struct file_operations *stat_fops[] = {
2797 2798 2799
	[KVM_STAT_VCPU] = &vcpu_stat_fops,
	[KVM_STAT_VM]   = &vm_stat_fops,
};
A
Avi Kivity 已提交
2800

2801
static int kvm_init_debug(void)
A
Avi Kivity 已提交
2802
{
2803
	int r = -EFAULT;
A
Avi Kivity 已提交
2804 2805
	struct kvm_stats_debugfs_item *p;

2806
	kvm_debugfs_dir = debugfs_create_dir("kvm", NULL);
2807 2808 2809 2810
	if (kvm_debugfs_dir == NULL)
		goto out;

	for (p = debugfs_entries; p->name; ++p) {
2811
		p->dentry = debugfs_create_file(p->name, 0444, kvm_debugfs_dir,
A
Avi Kivity 已提交
2812
						(void *)(long)p->offset,
2813
						stat_fops[p->kind]);
2814 2815 2816 2817 2818 2819 2820 2821 2822 2823
		if (p->dentry == NULL)
			goto out_dir;
	}

	return 0;

out_dir:
	debugfs_remove_recursive(kvm_debugfs_dir);
out:
	return r;
A
Avi Kivity 已提交
2824 2825 2826 2827 2828 2829 2830 2831
}

static void kvm_exit_debug(void)
{
	struct kvm_stats_debugfs_item *p;

	for (p = debugfs_entries; p->name; ++p)
		debugfs_remove(p->dentry);
2832
	debugfs_remove(kvm_debugfs_dir);
A
Avi Kivity 已提交
2833 2834
}

2835
static int kvm_suspend(void)
2836
{
2837
	if (kvm_usage_count)
2838
		hardware_disable_nolock(NULL);
2839 2840 2841
	return 0;
}

2842
static void kvm_resume(void)
2843
{
2844
	if (kvm_usage_count) {
2845
		WARN_ON(raw_spin_is_locked(&kvm_lock));
2846
		hardware_enable_nolock(NULL);
2847
	}
2848 2849
}

2850
static struct syscore_ops kvm_syscore_ops = {
2851 2852 2853 2854
	.suspend = kvm_suspend,
	.resume = kvm_resume,
};

2855 2856 2857 2858 2859 2860 2861 2862 2863 2864
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);

2865
	kvm_arch_vcpu_load(vcpu, cpu);
2866 2867 2868 2869 2870 2871 2872
}

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

2873
	kvm_arch_vcpu_put(vcpu);
2874 2875
}

2876
int kvm_init(void *opaque, unsigned vcpu_size, unsigned vcpu_align,
2877
		  struct module *module)
A
Avi Kivity 已提交
2878 2879
{
	int r;
Y
Yang, Sheng 已提交
2880
	int cpu;
A
Avi Kivity 已提交
2881

2882 2883
	r = kvm_arch_init(opaque);
	if (r)
2884
		goto out_fail;
2885

2886
	if (!zalloc_cpumask_var(&cpus_hardware_enabled, GFP_KERNEL)) {
2887 2888 2889 2890
		r = -ENOMEM;
		goto out_free_0;
	}

2891
	r = kvm_arch_hardware_setup();
A
Avi Kivity 已提交
2892
	if (r < 0)
2893
		goto out_free_0a;
A
Avi Kivity 已提交
2894

Y
Yang, Sheng 已提交
2895 2896
	for_each_online_cpu(cpu) {
		smp_call_function_single(cpu,
2897
				kvm_arch_check_processor_compat,
2898
				&r, 1);
Y
Yang, Sheng 已提交
2899
		if (r < 0)
2900
			goto out_free_1;
Y
Yang, Sheng 已提交
2901 2902
	}

A
Avi Kivity 已提交
2903 2904
	r = register_cpu_notifier(&kvm_cpu_notifier);
	if (r)
2905
		goto out_free_2;
A
Avi Kivity 已提交
2906 2907
	register_reboot_notifier(&kvm_reboot_notifier);

2908
	/* A kmem cache lets us meet the alignment requirements of fx_save. */
2909 2910 2911
	if (!vcpu_align)
		vcpu_align = __alignof__(struct kvm_vcpu);
	kvm_vcpu_cache = kmem_cache_create("kvm_vcpu", vcpu_size, vcpu_align,
J
Joe Perches 已提交
2912
					   0, NULL);
2913 2914
	if (!kvm_vcpu_cache) {
		r = -ENOMEM;
2915
		goto out_free_3;
2916 2917
	}

2918 2919 2920 2921
	r = kvm_async_pf_init();
	if (r)
		goto out_free;

A
Avi Kivity 已提交
2922
	kvm_chardev_ops.owner = module;
2923 2924
	kvm_vm_fops.owner = module;
	kvm_vcpu_fops.owner = module;
A
Avi Kivity 已提交
2925 2926 2927

	r = misc_register(&kvm_dev);
	if (r) {
M
Mike Day 已提交
2928
		printk(KERN_ERR "kvm: misc device register failed\n");
2929
		goto out_unreg;
A
Avi Kivity 已提交
2930 2931
	}

2932 2933
	register_syscore_ops(&kvm_syscore_ops);

2934 2935 2936
	kvm_preempt_ops.sched_in = kvm_sched_in;
	kvm_preempt_ops.sched_out = kvm_sched_out;

2937 2938 2939 2940 2941
	r = kvm_init_debug();
	if (r) {
		printk(KERN_ERR "kvm: create debugfs files failed\n");
		goto out_undebugfs;
	}
2942

2943
	return 0;
A
Avi Kivity 已提交
2944

2945 2946
out_undebugfs:
	unregister_syscore_ops(&kvm_syscore_ops);
2947 2948
out_unreg:
	kvm_async_pf_deinit();
A
Avi Kivity 已提交
2949
out_free:
2950
	kmem_cache_destroy(kvm_vcpu_cache);
2951
out_free_3:
A
Avi Kivity 已提交
2952
	unregister_reboot_notifier(&kvm_reboot_notifier);
A
Avi Kivity 已提交
2953
	unregister_cpu_notifier(&kvm_cpu_notifier);
2954 2955
out_free_2:
out_free_1:
2956
	kvm_arch_hardware_unsetup();
2957 2958
out_free_0a:
	free_cpumask_var(cpus_hardware_enabled);
2959
out_free_0:
2960
	kvm_arch_exit();
2961
out_fail:
A
Avi Kivity 已提交
2962 2963
	return r;
}
2964
EXPORT_SYMBOL_GPL(kvm_init);
A
Avi Kivity 已提交
2965

2966
void kvm_exit(void)
A
Avi Kivity 已提交
2967
{
2968
	kvm_exit_debug();
A
Avi Kivity 已提交
2969
	misc_deregister(&kvm_dev);
2970
	kmem_cache_destroy(kvm_vcpu_cache);
2971
	kvm_async_pf_deinit();
2972
	unregister_syscore_ops(&kvm_syscore_ops);
A
Avi Kivity 已提交
2973
	unregister_reboot_notifier(&kvm_reboot_notifier);
2974
	unregister_cpu_notifier(&kvm_cpu_notifier);
2975
	on_each_cpu(hardware_disable_nolock, NULL, 1);
2976
	kvm_arch_hardware_unsetup();
2977
	kvm_arch_exit();
2978
	free_cpumask_var(cpus_hardware_enabled);
A
Avi Kivity 已提交
2979
}
2980
EXPORT_SYMBOL_GPL(kvm_exit);