vmx.c 85.2 KB
Newer Older
A
Avi Kivity 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
/*
 * 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.
 *
 * 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.
 *
 */

18
#include "irq.h"
A
Avi Kivity 已提交
19
#include "vmx.h"
20
#include "mmu.h"
A
Avi Kivity 已提交
21

22
#include <linux/kvm_host.h>
A
Avi Kivity 已提交
23
#include <linux/module.h>
24
#include <linux/kernel.h>
A
Avi Kivity 已提交
25 26
#include <linux/mm.h>
#include <linux/highmem.h>
A
Alexey Dobriyan 已提交
27
#include <linux/sched.h>
28
#include <linux/moduleparam.h>
A
Avi Kivity 已提交
29

A
Avi Kivity 已提交
30
#include <asm/io.h>
A
Anthony Liguori 已提交
31
#include <asm/desc.h>
A
Avi Kivity 已提交
32

33 34
#define __ex(x) __kvm_handle_fault_on_reboot(x)

A
Avi Kivity 已提交
35 36 37
MODULE_AUTHOR("Qumranet");
MODULE_LICENSE("GPL");

38 39 40
static int bypass_guest_pf = 1;
module_param(bypass_guest_pf, bool, 0);

41 42 43
static int enable_vpid = 1;
module_param(enable_vpid, bool, 0);

44 45 46
static int flexpriority_enabled = 1;
module_param(flexpriority_enabled, bool, 0);

47
static int enable_ept = 1;
S
Sheng Yang 已提交
48 49
module_param(enable_ept, bool, 0);

50 51 52 53 54 55 56
struct vmcs {
	u32 revision_id;
	u32 abort;
	char data[0];
};

struct vcpu_vmx {
R
Rusty Russell 已提交
57
	struct kvm_vcpu       vcpu;
58
	struct list_head      local_vcpus_link;
59
	int                   launched;
60
	u8                    fail;
61
	u32                   idt_vectoring_info;
62 63 64 65 66 67 68 69 70 71 72 73
	struct kvm_msr_entry *guest_msrs;
	struct kvm_msr_entry *host_msrs;
	int                   nmsrs;
	int                   save_nmsrs;
	int                   msr_offset_efer;
#ifdef CONFIG_X86_64
	int                   msr_offset_kernel_gs_base;
#endif
	struct vmcs          *vmcs;
	struct {
		int           loaded;
		u16           fs_sel, gs_sel, ldt_sel;
74 75
		int           gs_ldt_reload_needed;
		int           fs_reload_needed;
76
		int           guest_efer_loaded;
M
Mike Day 已提交
77
	} host_state;
78 79 80 81 82 83 84
	struct {
		struct {
			bool pending;
			u8 vector;
			unsigned rip;
		} irq;
	} rmode;
85
	int vpid;
86 87 88 89
};

static inline struct vcpu_vmx *to_vmx(struct kvm_vcpu *vcpu)
{
R
Rusty Russell 已提交
90
	return container_of(vcpu, struct vcpu_vmx, vcpu);
91 92
}

93
static int init_rmode(struct kvm *kvm);
94
static u64 construct_eptp(unsigned long root_hpa);
95

A
Avi Kivity 已提交
96 97
static DEFINE_PER_CPU(struct vmcs *, vmxarea);
static DEFINE_PER_CPU(struct vmcs *, current_vmcs);
98
static DEFINE_PER_CPU(struct list_head, vcpus_on_cpu);
A
Avi Kivity 已提交
99

100 101
static struct page *vmx_io_bitmap_a;
static struct page *vmx_io_bitmap_b;
S
Sheng Yang 已提交
102
static struct page *vmx_msr_bitmap;
103

104 105 106
static DECLARE_BITMAP(vmx_vpid_bitmap, VMX_NR_VPIDS);
static DEFINE_SPINLOCK(vmx_vpid_lock);

107
static struct vmcs_config {
A
Avi Kivity 已提交
108 109 110
	int size;
	int order;
	u32 revision_id;
111 112
	u32 pin_based_exec_ctrl;
	u32 cpu_based_exec_ctrl;
113
	u32 cpu_based_2nd_exec_ctrl;
114 115 116
	u32 vmexit_ctrl;
	u32 vmentry_ctrl;
} vmcs_config;
A
Avi Kivity 已提交
117

S
Sheng Yang 已提交
118 119 120 121 122
struct vmx_capability {
	u32 ept;
	u32 vpid;
} vmx_capability;

A
Avi Kivity 已提交
123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146
#define VMX_SEGMENT_FIELD(seg)					\
	[VCPU_SREG_##seg] = {                                   \
		.selector = GUEST_##seg##_SELECTOR,		\
		.base = GUEST_##seg##_BASE,		   	\
		.limit = GUEST_##seg##_LIMIT,		   	\
		.ar_bytes = GUEST_##seg##_AR_BYTES,	   	\
	}

static struct kvm_vmx_segment_field {
	unsigned selector;
	unsigned base;
	unsigned limit;
	unsigned ar_bytes;
} kvm_vmx_segment_fields[] = {
	VMX_SEGMENT_FIELD(CS),
	VMX_SEGMENT_FIELD(DS),
	VMX_SEGMENT_FIELD(ES),
	VMX_SEGMENT_FIELD(FS),
	VMX_SEGMENT_FIELD(GS),
	VMX_SEGMENT_FIELD(SS),
	VMX_SEGMENT_FIELD(TR),
	VMX_SEGMENT_FIELD(LDTR),
};

147 148 149 150
/*
 * Keep MSR_K6_STAR at the end, as setup_msrs() will try to optimize it
 * away by decrementing the array size.
 */
A
Avi Kivity 已提交
151
static const u32 vmx_msr_index[] = {
152
#ifdef CONFIG_X86_64
A
Avi Kivity 已提交
153 154 155 156
	MSR_SYSCALL_MASK, MSR_LSTAR, MSR_CSTAR, MSR_KERNEL_GS_BASE,
#endif
	MSR_EFER, MSR_K6_STAR,
};
157
#define NR_VMX_MSR ARRAY_SIZE(vmx_msr_index)
A
Avi Kivity 已提交
158

159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174
static void load_msrs(struct kvm_msr_entry *e, int n)
{
	int i;

	for (i = 0; i < n; ++i)
		wrmsrl(e[i].index, e[i].data);
}

static void save_msrs(struct kvm_msr_entry *e, int n)
{
	int i;

	for (i = 0; i < n; ++i)
		rdmsrl(e[i].index, e[i].data);
}

A
Avi Kivity 已提交
175 176 177 178 179 180 181
static inline int is_page_fault(u32 intr_info)
{
	return (intr_info & (INTR_INFO_INTR_TYPE_MASK | INTR_INFO_VECTOR_MASK |
			     INTR_INFO_VALID_MASK)) ==
		(INTR_TYPE_EXCEPTION | PF_VECTOR | INTR_INFO_VALID_MASK);
}

182 183 184 185 186 187 188
static inline int is_no_device(u32 intr_info)
{
	return (intr_info & (INTR_INFO_INTR_TYPE_MASK | INTR_INFO_VECTOR_MASK |
			     INTR_INFO_VALID_MASK)) ==
		(INTR_TYPE_EXCEPTION | NM_VECTOR | INTR_INFO_VALID_MASK);
}

189 190 191 192 193 194 195
static inline int is_invalid_opcode(u32 intr_info)
{
	return (intr_info & (INTR_INFO_INTR_TYPE_MASK | INTR_INFO_VECTOR_MASK |
			     INTR_INFO_VALID_MASK)) ==
		(INTR_TYPE_EXCEPTION | UD_VECTOR | INTR_INFO_VALID_MASK);
}

A
Avi Kivity 已提交
196 197 198 199 200 201
static inline int is_external_interrupt(u32 intr_info)
{
	return (intr_info & (INTR_INFO_INTR_TYPE_MASK | INTR_INFO_VALID_MASK))
		== (INTR_TYPE_EXT_INTR | INTR_INFO_VALID_MASK);
}

S
Sheng Yang 已提交
202 203 204 205 206
static inline int cpu_has_vmx_msr_bitmap(void)
{
	return (vmcs_config.cpu_based_exec_ctrl & CPU_BASED_USE_MSR_BITMAPS);
}

207 208 209 210 211 212 213 214 215 216
static inline int cpu_has_vmx_tpr_shadow(void)
{
	return (vmcs_config.cpu_based_exec_ctrl & CPU_BASED_TPR_SHADOW);
}

static inline int vm_need_tpr_shadow(struct kvm *kvm)
{
	return ((cpu_has_vmx_tpr_shadow()) && (irqchip_in_kernel(kvm)));
}

217 218 219 220 221 222
static inline int cpu_has_secondary_exec_ctrls(void)
{
	return (vmcs_config.cpu_based_exec_ctrl &
		CPU_BASED_ACTIVATE_SECONDARY_CONTROLS);
}

223
static inline bool cpu_has_vmx_virtualize_apic_accesses(void)
224
{
225 226 227
	return flexpriority_enabled
		&& (vmcs_config.cpu_based_2nd_exec_ctrl &
		    SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES);
228 229
}

S
Sheng Yang 已提交
230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255
static inline int cpu_has_vmx_invept_individual_addr(void)
{
	return (!!(vmx_capability.ept & VMX_EPT_EXTENT_INDIVIDUAL_BIT));
}

static inline int cpu_has_vmx_invept_context(void)
{
	return (!!(vmx_capability.ept & VMX_EPT_EXTENT_CONTEXT_BIT));
}

static inline int cpu_has_vmx_invept_global(void)
{
	return (!!(vmx_capability.ept & VMX_EPT_EXTENT_GLOBAL_BIT));
}

static inline int cpu_has_vmx_ept(void)
{
	return (vmcs_config.cpu_based_2nd_exec_ctrl &
		SECONDARY_EXEC_ENABLE_EPT);
}

static inline int vm_need_ept(void)
{
	return (cpu_has_vmx_ept() && enable_ept);
}

256 257 258 259 260 261
static inline int vm_need_virtualize_apic_accesses(struct kvm *kvm)
{
	return ((cpu_has_vmx_virtualize_apic_accesses()) &&
		(irqchip_in_kernel(kvm)));
}

262 263 264 265 266 267
static inline int cpu_has_vmx_vpid(void)
{
	return (vmcs_config.cpu_based_2nd_exec_ctrl &
		SECONDARY_EXEC_ENABLE_VPID);
}

268 269 270 271 272
static inline int cpu_has_virtual_nmis(void)
{
	return vmcs_config.pin_based_exec_ctrl & PIN_BASED_VIRTUAL_NMIS;
}

R
Rusty Russell 已提交
273
static int __find_msr_index(struct vcpu_vmx *vmx, u32 msr)
274 275 276
{
	int i;

277 278
	for (i = 0; i < vmx->nmsrs; ++i)
		if (vmx->guest_msrs[i].index == msr)
279 280 281 282
			return i;
	return -1;
}

283 284 285 286 287 288 289 290
static inline void __invvpid(int ext, u16 vpid, gva_t gva)
{
    struct {
	u64 vpid : 16;
	u64 rsvd : 48;
	u64 gva;
    } operand = { vpid, 0, gva };

291
    asm volatile (__ex(ASM_VMX_INVVPID)
292 293 294 295 296
		  /* CF==1 or ZF==1 --> rc = -1 */
		  "; ja 1f ; ud2 ; 1:"
		  : : "a"(&operand), "c"(ext) : "cc", "memory");
}

297 298 299 300 301 302
static inline void __invept(int ext, u64 eptp, gpa_t gpa)
{
	struct {
		u64 eptp, gpa;
	} operand = {eptp, gpa};

303
	asm volatile (__ex(ASM_VMX_INVEPT)
304 305 306 307 308
			/* CF==1 or ZF==1 --> rc = -1 */
			"; ja 1f ; ud2 ; 1:\n"
			: : "a" (&operand), "c" (ext) : "cc", "memory");
}

R
Rusty Russell 已提交
309
static struct kvm_msr_entry *find_msr_entry(struct vcpu_vmx *vmx, u32 msr)
310 311 312
{
	int i;

R
Rusty Russell 已提交
313
	i = __find_msr_index(vmx, msr);
314
	if (i >= 0)
315
		return &vmx->guest_msrs[i];
A
Al Viro 已提交
316
	return NULL;
317 318
}

A
Avi Kivity 已提交
319 320 321 322 323
static void vmcs_clear(struct vmcs *vmcs)
{
	u64 phys_addr = __pa(vmcs);
	u8 error;

324
	asm volatile (__ex(ASM_VMX_VMCLEAR_RAX) "; setna %0"
A
Avi Kivity 已提交
325 326 327 328 329 330 331 332 333
		      : "=g"(error) : "a"(&phys_addr), "m"(phys_addr)
		      : "cc", "memory");
	if (error)
		printk(KERN_ERR "kvm: vmclear fail: %p/%llx\n",
		       vmcs, phys_addr);
}

static void __vcpu_clear(void *arg)
{
R
Rusty Russell 已提交
334
	struct vcpu_vmx *vmx = arg;
335
	int cpu = raw_smp_processor_id();
A
Avi Kivity 已提交
336

R
Rusty Russell 已提交
337
	if (vmx->vcpu.cpu == cpu)
338 339
		vmcs_clear(vmx->vmcs);
	if (per_cpu(current_vmcs, cpu) == vmx->vmcs)
A
Avi Kivity 已提交
340
		per_cpu(current_vmcs, cpu) = NULL;
341
	rdtscll(vmx->vcpu.arch.host_tsc);
342 343 344
	list_del(&vmx->local_vcpus_link);
	vmx->vcpu.cpu = -1;
	vmx->launched = 0;
A
Avi Kivity 已提交
345 346
}

R
Rusty Russell 已提交
347
static void vcpu_clear(struct vcpu_vmx *vmx)
A
Avi Kivity 已提交
348
{
349 350
	if (vmx->vcpu.cpu == -1)
		return;
351
	smp_call_function_single(vmx->vcpu.cpu, __vcpu_clear, vmx, 1);
A
Avi Kivity 已提交
352 353
}

354 355 356 357 358 359 360 361
static inline void vpid_sync_vcpu_all(struct vcpu_vmx *vmx)
{
	if (vmx->vpid == 0)
		return;

	__invvpid(VMX_VPID_EXTENT_SINGLE_CONTEXT, vmx->vpid, 0);
}

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
static inline void ept_sync_global(void)
{
	if (cpu_has_vmx_invept_global())
		__invept(VMX_EPT_EXTENT_GLOBAL, 0, 0);
}

static inline void ept_sync_context(u64 eptp)
{
	if (vm_need_ept()) {
		if (cpu_has_vmx_invept_context())
			__invept(VMX_EPT_EXTENT_CONTEXT, eptp, 0);
		else
			ept_sync_global();
	}
}

static inline void ept_sync_individual_addr(u64 eptp, gpa_t gpa)
{
	if (vm_need_ept()) {
		if (cpu_has_vmx_invept_individual_addr())
			__invept(VMX_EPT_EXTENT_INDIVIDUAL_ADDR,
					eptp, gpa);
		else
			ept_sync_context(eptp);
	}
}

A
Avi Kivity 已提交
389 390 391 392
static unsigned long vmcs_readl(unsigned long field)
{
	unsigned long value;

393
	asm volatile (__ex(ASM_VMX_VMREAD_RDX_RAX)
A
Avi Kivity 已提交
394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409
		      : "=a"(value) : "d"(field) : "cc");
	return value;
}

static u16 vmcs_read16(unsigned long field)
{
	return vmcs_readl(field);
}

static u32 vmcs_read32(unsigned long field)
{
	return vmcs_readl(field);
}

static u64 vmcs_read64(unsigned long field)
{
410
#ifdef CONFIG_X86_64
A
Avi Kivity 已提交
411 412 413 414 415 416
	return vmcs_readl(field);
#else
	return vmcs_readl(field) | ((u64)vmcs_readl(field+1) << 32);
#endif
}

417 418 419 420 421 422 423
static noinline void vmwrite_error(unsigned long field, unsigned long value)
{
	printk(KERN_ERR "vmwrite error: reg %lx value %lx (err %d)\n",
	       field, value, vmcs_read32(VM_INSTRUCTION_ERROR));
	dump_stack();
}

A
Avi Kivity 已提交
424 425 426 427
static void vmcs_writel(unsigned long field, unsigned long value)
{
	u8 error;

428
	asm volatile (__ex(ASM_VMX_VMWRITE_RAX_RDX) "; setna %0"
M
Mike Day 已提交
429
		       : "=q"(error) : "a"(value), "d"(field) : "cc");
430 431
	if (unlikely(error))
		vmwrite_error(field, value);
A
Avi Kivity 已提交
432 433 434 435 436 437 438 439 440 441 442 443 444 445 446
}

static void vmcs_write16(unsigned long field, u16 value)
{
	vmcs_writel(field, value);
}

static void vmcs_write32(unsigned long field, u32 value)
{
	vmcs_writel(field, value);
}

static void vmcs_write64(unsigned long field, u64 value)
{
	vmcs_writel(field, value);
447
#ifndef CONFIG_X86_64
A
Avi Kivity 已提交
448 449 450 451 452
	asm volatile ("");
	vmcs_writel(field+1, value >> 32);
#endif
}

453 454 455 456 457 458 459 460 461 462
static void vmcs_clear_bits(unsigned long field, u32 mask)
{
	vmcs_writel(field, vmcs_readl(field) & ~mask);
}

static void vmcs_set_bits(unsigned long field, u32 mask)
{
	vmcs_writel(field, vmcs_readl(field) | mask);
}

463 464 465 466
static void update_exception_bitmap(struct kvm_vcpu *vcpu)
{
	u32 eb;

467
	eb = (1u << PF_VECTOR) | (1u << UD_VECTOR);
468 469 470 471
	if (!vcpu->fpu_active)
		eb |= 1u << NM_VECTOR;
	if (vcpu->guest_debug.enabled)
		eb |= 1u << 1;
472
	if (vcpu->arch.rmode.active)
473
		eb = ~0;
474 475
	if (vm_need_ept())
		eb &= ~(1u << PF_VECTOR); /* bypass_guest_pf = 0 */
476 477 478
	vmcs_write32(EXCEPTION_BITMAP, eb);
}

479 480 481 482 483 484
static void reload_tss(void)
{
	/*
	 * VT restores TR but not its size.  Useless.
	 */
	struct descriptor_table gdt;
485
	struct desc_struct *descs;
486

487
	kvm_get_gdt(&gdt);
488 489 490 491 492
	descs = (void *)gdt.base;
	descs[GDT_ENTRY_TSS].type = 9; /* available TSS */
	load_TR_desc();
}

R
Rusty Russell 已提交
493
static void load_transition_efer(struct vcpu_vmx *vmx)
494
{
495
	int efer_offset = vmx->msr_offset_efer;
496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514
	u64 host_efer = vmx->host_msrs[efer_offset].data;
	u64 guest_efer = vmx->guest_msrs[efer_offset].data;
	u64 ignore_bits;

	if (efer_offset < 0)
		return;
	/*
	 * NX is emulated; LMA and LME handled by hardware; SCE meaninless
	 * outside long mode
	 */
	ignore_bits = EFER_NX | EFER_SCE;
#ifdef CONFIG_X86_64
	ignore_bits |= EFER_LMA | EFER_LME;
	/* SCE is meaningful only in long mode on Intel */
	if (guest_efer & EFER_LMA)
		ignore_bits &= ~(u64)EFER_SCE;
#endif
	if ((guest_efer & ~ignore_bits) == (host_efer & ~ignore_bits))
		return;
515

516 517 518 519
	vmx->host_state.guest_efer_loaded = 1;
	guest_efer &= ~ignore_bits;
	guest_efer |= host_efer & ignore_bits;
	wrmsrl(MSR_EFER, guest_efer);
R
Rusty Russell 已提交
520
	vmx->vcpu.stat.efer_reload++;
521 522
}

523 524 525 526 527 528 529 530
static void reload_host_efer(struct vcpu_vmx *vmx)
{
	if (vmx->host_state.guest_efer_loaded) {
		vmx->host_state.guest_efer_loaded = 0;
		load_msrs(vmx->host_msrs + vmx->msr_offset_efer, 1);
	}
}

531
static void vmx_save_host_state(struct kvm_vcpu *vcpu)
532
{
533 534
	struct vcpu_vmx *vmx = to_vmx(vcpu);

535
	if (vmx->host_state.loaded)
536 537
		return;

538
	vmx->host_state.loaded = 1;
539 540 541 542
	/*
	 * Set host fs and gs selectors.  Unfortunately, 22.2.3 does not
	 * allow segment selectors with cpl > 0 or ti == 1.
	 */
543
	vmx->host_state.ldt_sel = kvm_read_ldt();
544
	vmx->host_state.gs_ldt_reload_needed = vmx->host_state.ldt_sel;
545
	vmx->host_state.fs_sel = kvm_read_fs();
546
	if (!(vmx->host_state.fs_sel & 7)) {
547
		vmcs_write16(HOST_FS_SELECTOR, vmx->host_state.fs_sel);
548 549
		vmx->host_state.fs_reload_needed = 0;
	} else {
550
		vmcs_write16(HOST_FS_SELECTOR, 0);
551
		vmx->host_state.fs_reload_needed = 1;
552
	}
553
	vmx->host_state.gs_sel = kvm_read_gs();
554 555
	if (!(vmx->host_state.gs_sel & 7))
		vmcs_write16(HOST_GS_SELECTOR, vmx->host_state.gs_sel);
556 557
	else {
		vmcs_write16(HOST_GS_SELECTOR, 0);
558
		vmx->host_state.gs_ldt_reload_needed = 1;
559 560 561 562 563 564
	}

#ifdef CONFIG_X86_64
	vmcs_writel(HOST_FS_BASE, read_msr(MSR_FS_BASE));
	vmcs_writel(HOST_GS_BASE, read_msr(MSR_GS_BASE));
#else
565 566
	vmcs_writel(HOST_FS_BASE, segment_base(vmx->host_state.fs_sel));
	vmcs_writel(HOST_GS_BASE, segment_base(vmx->host_state.gs_sel));
567
#endif
568 569

#ifdef CONFIG_X86_64
M
Mike Day 已提交
570
	if (is_long_mode(&vmx->vcpu))
571 572
		save_msrs(vmx->host_msrs +
			  vmx->msr_offset_kernel_gs_base, 1);
M
Mike Day 已提交
573

574
#endif
575
	load_msrs(vmx->guest_msrs, vmx->save_nmsrs);
576
	load_transition_efer(vmx);
577 578
}

579
static void __vmx_load_host_state(struct vcpu_vmx *vmx)
580
{
581
	unsigned long flags;
582

583
	if (!vmx->host_state.loaded)
584 585
		return;

586
	++vmx->vcpu.stat.host_state_reload;
587
	vmx->host_state.loaded = 0;
588
	if (vmx->host_state.fs_reload_needed)
589
		kvm_load_fs(vmx->host_state.fs_sel);
590
	if (vmx->host_state.gs_ldt_reload_needed) {
591
		kvm_load_ldt(vmx->host_state.ldt_sel);
592 593 594 595
		/*
		 * If we have to reload gs, we must take care to
		 * preserve our gs base.
		 */
596
		local_irq_save(flags);
597
		kvm_load_gs(vmx->host_state.gs_sel);
598 599 600
#ifdef CONFIG_X86_64
		wrmsrl(MSR_GS_BASE, vmcs_readl(HOST_GS_BASE));
#endif
601
		local_irq_restore(flags);
602
	}
603
	reload_tss();
604 605
	save_msrs(vmx->guest_msrs, vmx->save_nmsrs);
	load_msrs(vmx->host_msrs, vmx->save_nmsrs);
606
	reload_host_efer(vmx);
607 608
}

609 610 611 612 613 614 615
static void vmx_load_host_state(struct vcpu_vmx *vmx)
{
	preempt_disable();
	__vmx_load_host_state(vmx);
	preempt_enable();
}

A
Avi Kivity 已提交
616 617 618 619
/*
 * Switches to specified vcpu, until a matching vcpu_put(), but assumes
 * vcpu mutex is already taken.
 */
620
static void vmx_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
A
Avi Kivity 已提交
621
{
622 623
	struct vcpu_vmx *vmx = to_vmx(vcpu);
	u64 phys_addr = __pa(vmx->vmcs);
624
	u64 tsc_this, delta, new_offset;
A
Avi Kivity 已提交
625

626
	if (vcpu->cpu != cpu) {
R
Rusty Russell 已提交
627
		vcpu_clear(vmx);
M
Marcelo Tosatti 已提交
628
		kvm_migrate_timers(vcpu);
629
		vpid_sync_vcpu_all(vmx);
630 631 632 633
		local_irq_disable();
		list_add(&vmx->local_vcpus_link,
			 &per_cpu(vcpus_on_cpu, cpu));
		local_irq_enable();
634
	}
A
Avi Kivity 已提交
635

636
	if (per_cpu(current_vmcs, cpu) != vmx->vmcs) {
A
Avi Kivity 已提交
637 638
		u8 error;

639
		per_cpu(current_vmcs, cpu) = vmx->vmcs;
640
		asm volatile (__ex(ASM_VMX_VMPTRLD_RAX) "; setna %0"
A
Avi Kivity 已提交
641 642 643 644
			      : "=g"(error) : "a"(&phys_addr), "m"(phys_addr)
			      : "cc");
		if (error)
			printk(KERN_ERR "kvm: vmptrld %p/%llx fail\n",
645
			       vmx->vmcs, phys_addr);
A
Avi Kivity 已提交
646 647 648 649 650 651 652 653 654 655 656
	}

	if (vcpu->cpu != cpu) {
		struct descriptor_table dt;
		unsigned long sysenter_esp;

		vcpu->cpu = cpu;
		/*
		 * Linux uses per-cpu TSS and GDT, so set these when switching
		 * processors.
		 */
657 658
		vmcs_writel(HOST_TR_BASE, kvm_read_tr_base()); /* 22.2.4 */
		kvm_get_gdt(&dt);
A
Avi Kivity 已提交
659 660 661 662
		vmcs_writel(HOST_GDTR_BASE, dt.base);   /* 22.2.4 */

		rdmsrl(MSR_IA32_SYSENTER_ESP, sysenter_esp);
		vmcs_writel(HOST_IA32_SYSENTER_ESP, sysenter_esp); /* 22.2.3 */
663 664 665 666 667

		/*
		 * Make sure the time stamp counter is monotonous.
		 */
		rdtscll(tsc_this);
668 669 670 671 672
		if (tsc_this < vcpu->arch.host_tsc) {
			delta = vcpu->arch.host_tsc - tsc_this;
			new_offset = vmcs_read64(TSC_OFFSET) + delta;
			vmcs_write64(TSC_OFFSET, new_offset);
		}
A
Avi Kivity 已提交
673 674 675 676 677
	}
}

static void vmx_vcpu_put(struct kvm_vcpu *vcpu)
{
678
	__vmx_load_host_state(to_vmx(vcpu));
A
Avi Kivity 已提交
679 680
}

681 682 683 684 685
static void vmx_fpu_activate(struct kvm_vcpu *vcpu)
{
	if (vcpu->fpu_active)
		return;
	vcpu->fpu_active = 1;
686
	vmcs_clear_bits(GUEST_CR0, X86_CR0_TS);
687
	if (vcpu->arch.cr0 & X86_CR0_TS)
688
		vmcs_set_bits(GUEST_CR0, X86_CR0_TS);
689 690 691 692 693 694 695 696
	update_exception_bitmap(vcpu);
}

static void vmx_fpu_deactivate(struct kvm_vcpu *vcpu)
{
	if (!vcpu->fpu_active)
		return;
	vcpu->fpu_active = 0;
697
	vmcs_set_bits(GUEST_CR0, X86_CR0_TS);
698 699 700
	update_exception_bitmap(vcpu);
}

A
Avi Kivity 已提交
701 702 703 704 705 706 707
static unsigned long vmx_get_rflags(struct kvm_vcpu *vcpu)
{
	return vmcs_readl(GUEST_RFLAGS);
}

static void vmx_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags)
{
708
	if (vcpu->arch.rmode.active)
709
		rflags |= X86_EFLAGS_IOPL | X86_EFLAGS_VM;
A
Avi Kivity 已提交
710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729
	vmcs_writel(GUEST_RFLAGS, rflags);
}

static void skip_emulated_instruction(struct kvm_vcpu *vcpu)
{
	unsigned long rip;
	u32 interruptibility;

	rip = vmcs_readl(GUEST_RIP);
	rip += vmcs_read32(VM_EXIT_INSTRUCTION_LEN);
	vmcs_writel(GUEST_RIP, rip);

	/*
	 * We emulated an instruction, so temporary interrupt blocking
	 * should be removed, if set.
	 */
	interruptibility = vmcs_read32(GUEST_INTERRUPTIBILITY_INFO);
	if (interruptibility & 3)
		vmcs_write32(GUEST_INTERRUPTIBILITY_INFO,
			     interruptibility & ~3);
730
	vcpu->arch.interrupt_window_open = 1;
A
Avi Kivity 已提交
731 732
}

733 734 735 736 737
static void vmx_queue_exception(struct kvm_vcpu *vcpu, unsigned nr,
				bool has_error_code, u32 error_code)
{
	vmcs_write32(VM_ENTRY_INTR_INFO_FIELD,
		     nr | INTR_TYPE_EXCEPTION
738
		     | (has_error_code ? INTR_INFO_DELIVER_CODE_MASK : 0)
739 740 741 742 743 744 745 746 747 748 749 750
		     | INTR_INFO_VALID_MASK);
	if (has_error_code)
		vmcs_write32(VM_ENTRY_EXCEPTION_ERROR_CODE, error_code);
}

static bool vmx_exception_injected(struct kvm_vcpu *vcpu)
{
	struct vcpu_vmx *vmx = to_vmx(vcpu);

	return !(vmx->idt_vectoring_info & VECTORING_INFO_VALID_MASK);
}

751 752 753
/*
 * Swap MSR entry in host/guest MSR entry array.
 */
754
#ifdef CONFIG_X86_64
R
Rusty Russell 已提交
755
static void move_msr_up(struct vcpu_vmx *vmx, int from, int to)
756
{
757 758 759 760 761 762 763 764
	struct kvm_msr_entry tmp;

	tmp = vmx->guest_msrs[to];
	vmx->guest_msrs[to] = vmx->guest_msrs[from];
	vmx->guest_msrs[from] = tmp;
	tmp = vmx->host_msrs[to];
	vmx->host_msrs[to] = vmx->host_msrs[from];
	vmx->host_msrs[from] = tmp;
765
}
766
#endif
767

768 769 770 771 772
/*
 * Set up the vmcs to automatically save and restore system
 * msrs.  Don't touch the 64-bit msrs if the guest is in legacy
 * mode, as fiddling with msrs is very expensive.
 */
R
Rusty Russell 已提交
773
static void setup_msrs(struct vcpu_vmx *vmx)
774
{
775
	int save_nmsrs;
776

777
	vmx_load_host_state(vmx);
778 779
	save_nmsrs = 0;
#ifdef CONFIG_X86_64
R
Rusty Russell 已提交
780
	if (is_long_mode(&vmx->vcpu)) {
781 782
		int index;

R
Rusty Russell 已提交
783
		index = __find_msr_index(vmx, MSR_SYSCALL_MASK);
784
		if (index >= 0)
R
Rusty Russell 已提交
785 786
			move_msr_up(vmx, index, save_nmsrs++);
		index = __find_msr_index(vmx, MSR_LSTAR);
787
		if (index >= 0)
R
Rusty Russell 已提交
788 789
			move_msr_up(vmx, index, save_nmsrs++);
		index = __find_msr_index(vmx, MSR_CSTAR);
790
		if (index >= 0)
R
Rusty Russell 已提交
791 792
			move_msr_up(vmx, index, save_nmsrs++);
		index = __find_msr_index(vmx, MSR_KERNEL_GS_BASE);
793
		if (index >= 0)
R
Rusty Russell 已提交
794
			move_msr_up(vmx, index, save_nmsrs++);
795 796 797 798
		/*
		 * MSR_K6_STAR is only needed on long mode guests, and only
		 * if efer.sce is enabled.
		 */
R
Rusty Russell 已提交
799
		index = __find_msr_index(vmx, MSR_K6_STAR);
800
		if ((index >= 0) && (vmx->vcpu.arch.shadow_efer & EFER_SCE))
R
Rusty Russell 已提交
801
			move_msr_up(vmx, index, save_nmsrs++);
802 803
	}
#endif
804
	vmx->save_nmsrs = save_nmsrs;
805

806
#ifdef CONFIG_X86_64
807
	vmx->msr_offset_kernel_gs_base =
R
Rusty Russell 已提交
808
		__find_msr_index(vmx, MSR_KERNEL_GS_BASE);
809
#endif
R
Rusty Russell 已提交
810
	vmx->msr_offset_efer = __find_msr_index(vmx, MSR_EFER);
811 812
}

A
Avi Kivity 已提交
813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845
/*
 * reads and returns guest's timestamp counter "register"
 * guest_tsc = host_tsc + tsc_offset    -- 21.3
 */
static u64 guest_read_tsc(void)
{
	u64 host_tsc, tsc_offset;

	rdtscll(host_tsc);
	tsc_offset = vmcs_read64(TSC_OFFSET);
	return host_tsc + tsc_offset;
}

/*
 * writes 'guest_tsc' into guest's timestamp counter "register"
 * guest_tsc = host_tsc + tsc_offset ==> tsc_offset = guest_tsc - host_tsc
 */
static void guest_write_tsc(u64 guest_tsc)
{
	u64 host_tsc;

	rdtscll(host_tsc);
	vmcs_write64(TSC_OFFSET, guest_tsc - host_tsc);
}

/*
 * Reads an msr value (of 'msr_index') into 'pdata'.
 * Returns 0 on success, non-0 otherwise.
 * Assumes vcpu_load() was already called.
 */
static int vmx_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata)
{
	u64 data;
846
	struct kvm_msr_entry *msr;
A
Avi Kivity 已提交
847 848 849 850 851 852 853

	if (!pdata) {
		printk(KERN_ERR "BUG: get_msr called with NULL pdata\n");
		return -EINVAL;
	}

	switch (msr_index) {
854
#ifdef CONFIG_X86_64
A
Avi Kivity 已提交
855 856 857 858 859 860 861
	case MSR_FS_BASE:
		data = vmcs_readl(GUEST_FS_BASE);
		break;
	case MSR_GS_BASE:
		data = vmcs_readl(GUEST_GS_BASE);
		break;
	case MSR_EFER:
862
		return kvm_get_msr_common(vcpu, msr_index, pdata);
A
Avi Kivity 已提交
863 864 865 866 867 868 869 870
#endif
	case MSR_IA32_TIME_STAMP_COUNTER:
		data = guest_read_tsc();
		break;
	case MSR_IA32_SYSENTER_CS:
		data = vmcs_read32(GUEST_SYSENTER_CS);
		break;
	case MSR_IA32_SYSENTER_EIP:
A
Avi Kivity 已提交
871
		data = vmcs_readl(GUEST_SYSENTER_EIP);
A
Avi Kivity 已提交
872 873
		break;
	case MSR_IA32_SYSENTER_ESP:
A
Avi Kivity 已提交
874
		data = vmcs_readl(GUEST_SYSENTER_ESP);
A
Avi Kivity 已提交
875 876
		break;
	default:
R
Rusty Russell 已提交
877
		msr = find_msr_entry(to_vmx(vcpu), msr_index);
878 879 880
		if (msr) {
			data = msr->data;
			break;
A
Avi Kivity 已提交
881
		}
882
		return kvm_get_msr_common(vcpu, msr_index, pdata);
A
Avi Kivity 已提交
883 884 885 886 887 888 889 890 891 892 893 894 895
	}

	*pdata = data;
	return 0;
}

/*
 * Writes msr value into into the appropriate "register".
 * Returns 0 on success, non-0 otherwise.
 * Assumes vcpu_load() was already called.
 */
static int vmx_set_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 data)
{
896 897
	struct vcpu_vmx *vmx = to_vmx(vcpu);
	struct kvm_msr_entry *msr;
898 899
	int ret = 0;

A
Avi Kivity 已提交
900
	switch (msr_index) {
901
#ifdef CONFIG_X86_64
902
	case MSR_EFER:
903
		vmx_load_host_state(vmx);
904 905
		ret = kvm_set_msr_common(vcpu, msr_index, data);
		break;
A
Avi Kivity 已提交
906 907 908 909 910 911 912 913 914 915 916
	case MSR_FS_BASE:
		vmcs_writel(GUEST_FS_BASE, data);
		break;
	case MSR_GS_BASE:
		vmcs_writel(GUEST_GS_BASE, data);
		break;
#endif
	case MSR_IA32_SYSENTER_CS:
		vmcs_write32(GUEST_SYSENTER_CS, data);
		break;
	case MSR_IA32_SYSENTER_EIP:
A
Avi Kivity 已提交
917
		vmcs_writel(GUEST_SYSENTER_EIP, data);
A
Avi Kivity 已提交
918 919
		break;
	case MSR_IA32_SYSENTER_ESP:
A
Avi Kivity 已提交
920
		vmcs_writel(GUEST_SYSENTER_ESP, data);
A
Avi Kivity 已提交
921
		break;
A
Avi Kivity 已提交
922
	case MSR_IA32_TIME_STAMP_COUNTER:
A
Avi Kivity 已提交
923
		guest_write_tsc(data);
924 925 926 927 928 929 930 931 932 933 934 935
		break;
	case MSR_P6_PERFCTR0:
	case MSR_P6_PERFCTR1:
	case MSR_P6_EVNTSEL0:
	case MSR_P6_EVNTSEL1:
		/*
		 * Just discard all writes to the performance counters; this
		 * should keep both older linux and windows 64-bit guests
		 * happy
		 */
		pr_unimpl(vcpu, "unimplemented perfctr wrmsr: 0x%x data 0x%llx\n", msr_index, data);

A
Avi Kivity 已提交
936 937
		break;
	default:
938
		vmx_load_host_state(vmx);
R
Rusty Russell 已提交
939
		msr = find_msr_entry(vmx, msr_index);
940 941 942
		if (msr) {
			msr->data = data;
			break;
A
Avi Kivity 已提交
943
		}
944
		ret = kvm_set_msr_common(vcpu, msr_index, data);
A
Avi Kivity 已提交
945 946
	}

947
	return ret;
A
Avi Kivity 已提交
948 949 950 951
}

/*
 * Sync the rsp and rip registers into the vcpu structure.  This allows
952
 * registers to be accessed by indexing vcpu->arch.regs.
A
Avi Kivity 已提交
953 954 955
 */
static void vcpu_load_rsp_rip(struct kvm_vcpu *vcpu)
{
956 957
	vcpu->arch.regs[VCPU_REGS_RSP] = vmcs_readl(GUEST_RSP);
	vcpu->arch.rip = vmcs_readl(GUEST_RIP);
A
Avi Kivity 已提交
958 959 960 961 962 963 964 965
}

/*
 * Syncs rsp and rip back into the vmcs.  Should be called after possible
 * modification.
 */
static void vcpu_put_rsp_rip(struct kvm_vcpu *vcpu)
{
966 967
	vmcs_writel(GUEST_RSP, vcpu->arch.regs[VCPU_REGS_RSP]);
	vmcs_writel(GUEST_RIP, vcpu->arch.rip);
A
Avi Kivity 已提交
968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990
}

static int set_guest_debug(struct kvm_vcpu *vcpu, struct kvm_debug_guest *dbg)
{
	unsigned long dr7 = 0x400;
	int old_singlestep;

	old_singlestep = vcpu->guest_debug.singlestep;

	vcpu->guest_debug.enabled = dbg->enabled;
	if (vcpu->guest_debug.enabled) {
		int i;

		dr7 |= 0x200;  /* exact */
		for (i = 0; i < 4; ++i) {
			if (!dbg->breakpoints[i].enabled)
				continue;
			vcpu->guest_debug.bp[i] = dbg->breakpoints[i].address;
			dr7 |= 2 << (i*2);    /* global enable */
			dr7 |= 0 << (i*4+16); /* execution breakpoint */
		}

		vcpu->guest_debug.singlestep = dbg->singlestep;
991
	} else
A
Avi Kivity 已提交
992 993 994 995 996 997 998 999 1000 1001
		vcpu->guest_debug.singlestep = 0;

	if (old_singlestep && !vcpu->guest_debug.singlestep) {
		unsigned long flags;

		flags = vmcs_readl(GUEST_RFLAGS);
		flags &= ~(X86_EFLAGS_TF | X86_EFLAGS_RF);
		vmcs_writel(GUEST_RFLAGS, flags);
	}

1002
	update_exception_bitmap(vcpu);
A
Avi Kivity 已提交
1003 1004 1005 1006 1007
	vmcs_writel(GUEST_DR7, dr7);

	return 0;
}

E
Eddie Dong 已提交
1008 1009
static int vmx_get_irq(struct kvm_vcpu *vcpu)
{
1010
	struct vcpu_vmx *vmx = to_vmx(vcpu);
E
Eddie Dong 已提交
1011 1012
	u32 idtv_info_field;

1013
	idtv_info_field = vmx->idt_vectoring_info;
E
Eddie Dong 已提交
1014 1015 1016 1017
	if (idtv_info_field & INTR_INFO_VALID_MASK) {
		if (is_external_interrupt(idtv_info_field))
			return idtv_info_field & VECTORING_INFO_VECTOR_MASK;
		else
M
Mike Day 已提交
1018
			printk(KERN_DEBUG "pending exception: not handled yet\n");
E
Eddie Dong 已提交
1019 1020 1021 1022
	}
	return -1;
}

A
Avi Kivity 已提交
1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033
static __init int cpu_has_kvm_support(void)
{
	unsigned long ecx = cpuid_ecx(1);
	return test_bit(5, &ecx); /* CPUID.1:ECX.VMX[bit 5] -> VT */
}

static __init int vmx_disabled_by_bios(void)
{
	u64 msr;

	rdmsrl(MSR_IA32_FEATURE_CONTROL, msr);
1034 1035 1036 1037
	return (msr & (MSR_IA32_FEATURE_CONTROL_LOCKED |
		       MSR_IA32_FEATURE_CONTROL_VMXON_ENABLED))
	    == MSR_IA32_FEATURE_CONTROL_LOCKED;
	/* locked but not enabled */
A
Avi Kivity 已提交
1038 1039
}

A
Avi Kivity 已提交
1040
static void hardware_enable(void *garbage)
A
Avi Kivity 已提交
1041 1042 1043 1044 1045
{
	int cpu = raw_smp_processor_id();
	u64 phys_addr = __pa(per_cpu(vmxarea, cpu));
	u64 old;

1046
	INIT_LIST_HEAD(&per_cpu(vcpus_on_cpu, cpu));
A
Avi Kivity 已提交
1047
	rdmsrl(MSR_IA32_FEATURE_CONTROL, old);
1048 1049 1050 1051
	if ((old & (MSR_IA32_FEATURE_CONTROL_LOCKED |
		    MSR_IA32_FEATURE_CONTROL_VMXON_ENABLED))
	    != (MSR_IA32_FEATURE_CONTROL_LOCKED |
		MSR_IA32_FEATURE_CONTROL_VMXON_ENABLED))
A
Avi Kivity 已提交
1052
		/* enable and lock */
1053 1054 1055
		wrmsrl(MSR_IA32_FEATURE_CONTROL, old |
		       MSR_IA32_FEATURE_CONTROL_LOCKED |
		       MSR_IA32_FEATURE_CONTROL_VMXON_ENABLED);
1056
	write_cr4(read_cr4() | X86_CR4_VMXE); /* FIXME: not cpu hotplug safe */
1057 1058
	asm volatile (ASM_VMX_VMXON_RAX
		      : : "a"(&phys_addr), "m"(phys_addr)
A
Avi Kivity 已提交
1059 1060 1061
		      : "memory", "cc");
}

1062 1063 1064 1065 1066 1067 1068 1069 1070 1071
static void vmclear_local_vcpus(void)
{
	int cpu = raw_smp_processor_id();
	struct vcpu_vmx *vmx, *n;

	list_for_each_entry_safe(vmx, n, &per_cpu(vcpus_on_cpu, cpu),
				 local_vcpus_link)
		__vcpu_clear(vmx);
}

A
Avi Kivity 已提交
1072 1073
static void hardware_disable(void *garbage)
{
1074
	vmclear_local_vcpus();
1075
	asm volatile (__ex(ASM_VMX_VMXOFF) : : : "cc");
1076
	write_cr4(read_cr4() & ~X86_CR4_VMXE);
A
Avi Kivity 已提交
1077 1078
}

1079
static __init int adjust_vmx_controls(u32 ctl_min, u32 ctl_opt,
M
Mike Day 已提交
1080
				      u32 msr, u32 *result)
1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091
{
	u32 vmx_msr_low, vmx_msr_high;
	u32 ctl = ctl_min | ctl_opt;

	rdmsr(msr, vmx_msr_low, vmx_msr_high);

	ctl &= vmx_msr_high; /* bit == 0 in high word ==> must be zero */
	ctl |= vmx_msr_low;  /* bit == 1 in low word  ==> must be one  */

	/* Ensure minimum (required) set of control bits are supported. */
	if (ctl_min & ~ctl)
Y
Yang, Sheng 已提交
1092
		return -EIO;
1093 1094 1095 1096 1097

	*result = ctl;
	return 0;
}

Y
Yang, Sheng 已提交
1098
static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
A
Avi Kivity 已提交
1099 1100
{
	u32 vmx_msr_low, vmx_msr_high;
S
Sheng Yang 已提交
1101
	u32 min, opt, min2, opt2;
1102 1103
	u32 _pin_based_exec_control = 0;
	u32 _cpu_based_exec_control = 0;
1104
	u32 _cpu_based_2nd_exec_control = 0;
1105 1106 1107 1108
	u32 _vmexit_control = 0;
	u32 _vmentry_control = 0;

	min = PIN_BASED_EXT_INTR_MASK | PIN_BASED_NMI_EXITING;
1109
	opt = PIN_BASED_VIRTUAL_NMIS;
1110 1111
	if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_PINBASED_CTLS,
				&_pin_based_exec_control) < 0)
Y
Yang, Sheng 已提交
1112
		return -EIO;
1113 1114 1115 1116 1117 1118

	min = CPU_BASED_HLT_EXITING |
#ifdef CONFIG_X86_64
	      CPU_BASED_CR8_LOAD_EXITING |
	      CPU_BASED_CR8_STORE_EXITING |
#endif
S
Sheng Yang 已提交
1119 1120
	      CPU_BASED_CR3_LOAD_EXITING |
	      CPU_BASED_CR3_STORE_EXITING |
1121 1122 1123
	      CPU_BASED_USE_IO_BITMAPS |
	      CPU_BASED_MOV_DR_EXITING |
	      CPU_BASED_USE_TSC_OFFSETING;
1124
	opt = CPU_BASED_TPR_SHADOW |
S
Sheng Yang 已提交
1125
	      CPU_BASED_USE_MSR_BITMAPS |
1126
	      CPU_BASED_ACTIVATE_SECONDARY_CONTROLS;
1127 1128
	if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_PROCBASED_CTLS,
				&_cpu_based_exec_control) < 0)
Y
Yang, Sheng 已提交
1129
		return -EIO;
1130 1131 1132 1133 1134
#ifdef CONFIG_X86_64
	if ((_cpu_based_exec_control & CPU_BASED_TPR_SHADOW))
		_cpu_based_exec_control &= ~CPU_BASED_CR8_LOAD_EXITING &
					   ~CPU_BASED_CR8_STORE_EXITING;
#endif
1135
	if (_cpu_based_exec_control & CPU_BASED_ACTIVATE_SECONDARY_CONTROLS) {
S
Sheng Yang 已提交
1136 1137
		min2 = 0;
		opt2 = SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES |
1138
			SECONDARY_EXEC_WBINVD_EXITING |
S
Sheng Yang 已提交
1139 1140 1141 1142
			SECONDARY_EXEC_ENABLE_VPID |
			SECONDARY_EXEC_ENABLE_EPT;
		if (adjust_vmx_controls(min2, opt2,
					MSR_IA32_VMX_PROCBASED_CTLS2,
1143 1144 1145 1146 1147 1148 1149 1150
					&_cpu_based_2nd_exec_control) < 0)
			return -EIO;
	}
#ifndef CONFIG_X86_64
	if (!(_cpu_based_2nd_exec_control &
				SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES))
		_cpu_based_exec_control &= ~CPU_BASED_TPR_SHADOW;
#endif
S
Sheng Yang 已提交
1151 1152 1153 1154 1155 1156 1157 1158 1159 1160
	if (_cpu_based_2nd_exec_control & SECONDARY_EXEC_ENABLE_EPT) {
		/* CR3 accesses don't need to cause VM Exits when EPT enabled */
		min &= ~(CPU_BASED_CR3_LOAD_EXITING |
			 CPU_BASED_CR3_STORE_EXITING);
		if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_PROCBASED_CTLS,
					&_cpu_based_exec_control) < 0)
			return -EIO;
		rdmsr(MSR_IA32_VMX_EPT_VPID_CAP,
		      vmx_capability.ept, vmx_capability.vpid);
	}
1161 1162 1163 1164 1165 1166 1167 1168

	min = 0;
#ifdef CONFIG_X86_64
	min |= VM_EXIT_HOST_ADDR_SPACE_SIZE;
#endif
	opt = 0;
	if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_EXIT_CTLS,
				&_vmexit_control) < 0)
Y
Yang, Sheng 已提交
1169
		return -EIO;
1170 1171 1172 1173

	min = opt = 0;
	if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_ENTRY_CTLS,
				&_vmentry_control) < 0)
Y
Yang, Sheng 已提交
1174
		return -EIO;
A
Avi Kivity 已提交
1175

N
Nguyen Anh Quynh 已提交
1176
	rdmsr(MSR_IA32_VMX_BASIC, vmx_msr_low, vmx_msr_high);
1177 1178 1179

	/* IA-32 SDM Vol 3B: VMCS size is never greater than 4kB. */
	if ((vmx_msr_high & 0x1fff) > PAGE_SIZE)
Y
Yang, Sheng 已提交
1180
		return -EIO;
1181 1182 1183 1184

#ifdef CONFIG_X86_64
	/* IA-32 SDM Vol 3B: 64-bit CPUs always have VMX_BASIC_MSR[48]==0. */
	if (vmx_msr_high & (1u<<16))
Y
Yang, Sheng 已提交
1185
		return -EIO;
1186 1187 1188 1189
#endif

	/* Require Write-Back (WB) memory type for VMCS accesses. */
	if (((vmx_msr_high >> 18) & 15) != 6)
Y
Yang, Sheng 已提交
1190
		return -EIO;
1191

Y
Yang, Sheng 已提交
1192 1193 1194
	vmcs_conf->size = vmx_msr_high & 0x1fff;
	vmcs_conf->order = get_order(vmcs_config.size);
	vmcs_conf->revision_id = vmx_msr_low;
1195

Y
Yang, Sheng 已提交
1196 1197
	vmcs_conf->pin_based_exec_ctrl = _pin_based_exec_control;
	vmcs_conf->cpu_based_exec_ctrl = _cpu_based_exec_control;
1198
	vmcs_conf->cpu_based_2nd_exec_ctrl = _cpu_based_2nd_exec_control;
Y
Yang, Sheng 已提交
1199 1200
	vmcs_conf->vmexit_ctrl         = _vmexit_control;
	vmcs_conf->vmentry_ctrl        = _vmentry_control;
1201 1202

	return 0;
N
Nguyen Anh Quynh 已提交
1203
}
A
Avi Kivity 已提交
1204 1205 1206 1207 1208 1209 1210

static struct vmcs *alloc_vmcs_cpu(int cpu)
{
	int node = cpu_to_node(cpu);
	struct page *pages;
	struct vmcs *vmcs;

1211
	pages = alloc_pages_node(node, GFP_KERNEL, vmcs_config.order);
A
Avi Kivity 已提交
1212 1213 1214
	if (!pages)
		return NULL;
	vmcs = page_address(pages);
1215 1216
	memset(vmcs, 0, vmcs_config.size);
	vmcs->revision_id = vmcs_config.revision_id; /* vmcs revision id */
A
Avi Kivity 已提交
1217 1218 1219 1220 1221
	return vmcs;
}

static struct vmcs *alloc_vmcs(void)
{
1222
	return alloc_vmcs_cpu(raw_smp_processor_id());
A
Avi Kivity 已提交
1223 1224 1225 1226
}

static void free_vmcs(struct vmcs *vmcs)
{
1227
	free_pages((unsigned long)vmcs, vmcs_config.order);
A
Avi Kivity 已提交
1228 1229
}

1230
static void free_kvm_area(void)
A
Avi Kivity 已提交
1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257
{
	int cpu;

	for_each_online_cpu(cpu)
		free_vmcs(per_cpu(vmxarea, cpu));
}

static __init int alloc_kvm_area(void)
{
	int cpu;

	for_each_online_cpu(cpu) {
		struct vmcs *vmcs;

		vmcs = alloc_vmcs_cpu(cpu);
		if (!vmcs) {
			free_kvm_area();
			return -ENOMEM;
		}

		per_cpu(vmxarea, cpu) = vmcs;
	}
	return 0;
}

static __init int hardware_setup(void)
{
Y
Yang, Sheng 已提交
1258 1259
	if (setup_vmcs_config(&vmcs_config) < 0)
		return -EIO;
1260 1261 1262 1263

	if (boot_cpu_has(X86_FEATURE_NX))
		kvm_enable_efer_bits(EFER_NX);

A
Avi Kivity 已提交
1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275
	return alloc_kvm_area();
}

static __exit void hardware_unsetup(void)
{
	free_kvm_area();
}

static void fix_pmode_dataseg(int seg, struct kvm_save_segment *save)
{
	struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];

1276
	if (vmcs_readl(sf->base) == save->base && (save->base & AR_S_MASK)) {
A
Avi Kivity 已提交
1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291
		vmcs_write16(sf->selector, save->selector);
		vmcs_writel(sf->base, save->base);
		vmcs_write32(sf->limit, save->limit);
		vmcs_write32(sf->ar_bytes, save->ar);
	} else {
		u32 dpl = (vmcs_read16(sf->selector) & SELECTOR_RPL_MASK)
			<< AR_DPL_SHIFT;
		vmcs_write32(sf->ar_bytes, 0x93 | dpl);
	}
}

static void enter_pmode(struct kvm_vcpu *vcpu)
{
	unsigned long flags;

1292
	vcpu->arch.rmode.active = 0;
A
Avi Kivity 已提交
1293

1294 1295 1296
	vmcs_writel(GUEST_TR_BASE, vcpu->arch.rmode.tr.base);
	vmcs_write32(GUEST_TR_LIMIT, vcpu->arch.rmode.tr.limit);
	vmcs_write32(GUEST_TR_AR_BYTES, vcpu->arch.rmode.tr.ar);
A
Avi Kivity 已提交
1297 1298

	flags = vmcs_readl(GUEST_RFLAGS);
1299
	flags &= ~(X86_EFLAGS_IOPL | X86_EFLAGS_VM);
1300
	flags |= (vcpu->arch.rmode.save_iopl << IOPL_SHIFT);
A
Avi Kivity 已提交
1301 1302
	vmcs_writel(GUEST_RFLAGS, flags);

1303 1304
	vmcs_writel(GUEST_CR4, (vmcs_readl(GUEST_CR4) & ~X86_CR4_VME) |
			(vmcs_readl(CR4_READ_SHADOW) & X86_CR4_VME));
A
Avi Kivity 已提交
1305 1306 1307

	update_exception_bitmap(vcpu);

1308 1309 1310 1311
	fix_pmode_dataseg(VCPU_SREG_ES, &vcpu->arch.rmode.es);
	fix_pmode_dataseg(VCPU_SREG_DS, &vcpu->arch.rmode.ds);
	fix_pmode_dataseg(VCPU_SREG_GS, &vcpu->arch.rmode.gs);
	fix_pmode_dataseg(VCPU_SREG_FS, &vcpu->arch.rmode.fs);
A
Avi Kivity 已提交
1312 1313 1314 1315 1316 1317 1318 1319 1320

	vmcs_write16(GUEST_SS_SELECTOR, 0);
	vmcs_write32(GUEST_SS_AR_BYTES, 0x93);

	vmcs_write16(GUEST_CS_SELECTOR,
		     vmcs_read16(GUEST_CS_SELECTOR) & ~SELECTOR_RPL_MASK);
	vmcs_write32(GUEST_CS_AR_BYTES, 0x9b);
}

M
Mike Day 已提交
1321
static gva_t rmode_tss_base(struct kvm *kvm)
A
Avi Kivity 已提交
1322
{
1323
	if (!kvm->arch.tss_addr) {
1324 1325 1326 1327
		gfn_t base_gfn = kvm->memslots[0].base_gfn +
				 kvm->memslots[0].npages - 3;
		return base_gfn << PAGE_SHIFT;
	}
1328
	return kvm->arch.tss_addr;
A
Avi Kivity 已提交
1329 1330 1331 1332 1333 1334 1335 1336 1337 1338
}

static void fix_rmode_seg(int seg, struct kvm_save_segment *save)
{
	struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];

	save->selector = vmcs_read16(sf->selector);
	save->base = vmcs_readl(sf->base);
	save->limit = vmcs_read32(sf->limit);
	save->ar = vmcs_read32(sf->ar_bytes);
1339 1340
	vmcs_write16(sf->selector, save->base >> 4);
	vmcs_write32(sf->base, save->base & 0xfffff);
A
Avi Kivity 已提交
1341 1342 1343 1344 1345 1346 1347 1348
	vmcs_write32(sf->limit, 0xffff);
	vmcs_write32(sf->ar_bytes, 0xf3);
}

static void enter_rmode(struct kvm_vcpu *vcpu)
{
	unsigned long flags;

1349
	vcpu->arch.rmode.active = 1;
A
Avi Kivity 已提交
1350

1351
	vcpu->arch.rmode.tr.base = vmcs_readl(GUEST_TR_BASE);
A
Avi Kivity 已提交
1352 1353
	vmcs_writel(GUEST_TR_BASE, rmode_tss_base(vcpu->kvm));

1354
	vcpu->arch.rmode.tr.limit = vmcs_read32(GUEST_TR_LIMIT);
A
Avi Kivity 已提交
1355 1356
	vmcs_write32(GUEST_TR_LIMIT, RMODE_TSS_SIZE - 1);

1357
	vcpu->arch.rmode.tr.ar = vmcs_read32(GUEST_TR_AR_BYTES);
A
Avi Kivity 已提交
1358 1359 1360
	vmcs_write32(GUEST_TR_AR_BYTES, 0x008b);

	flags = vmcs_readl(GUEST_RFLAGS);
1361 1362
	vcpu->arch.rmode.save_iopl
		= (flags & X86_EFLAGS_IOPL) >> IOPL_SHIFT;
A
Avi Kivity 已提交
1363

1364
	flags |= X86_EFLAGS_IOPL | X86_EFLAGS_VM;
A
Avi Kivity 已提交
1365 1366

	vmcs_writel(GUEST_RFLAGS, flags);
1367
	vmcs_writel(GUEST_CR4, vmcs_readl(GUEST_CR4) | X86_CR4_VME);
A
Avi Kivity 已提交
1368 1369 1370 1371 1372 1373 1374
	update_exception_bitmap(vcpu);

	vmcs_write16(GUEST_SS_SELECTOR, vmcs_readl(GUEST_SS_BASE) >> 4);
	vmcs_write32(GUEST_SS_LIMIT, 0xffff);
	vmcs_write32(GUEST_SS_AR_BYTES, 0xf3);

	vmcs_write32(GUEST_CS_AR_BYTES, 0xf3);
1375
	vmcs_write32(GUEST_CS_LIMIT, 0xffff);
1376 1377
	if (vmcs_readl(GUEST_CS_BASE) == 0xffff0000)
		vmcs_writel(GUEST_CS_BASE, 0xf0000);
A
Avi Kivity 已提交
1378 1379
	vmcs_write16(GUEST_CS_SELECTOR, vmcs_readl(GUEST_CS_BASE) >> 4);

1380 1381 1382 1383
	fix_rmode_seg(VCPU_SREG_ES, &vcpu->arch.rmode.es);
	fix_rmode_seg(VCPU_SREG_DS, &vcpu->arch.rmode.ds);
	fix_rmode_seg(VCPU_SREG_GS, &vcpu->arch.rmode.gs);
	fix_rmode_seg(VCPU_SREG_FS, &vcpu->arch.rmode.fs);
1384

1385
	kvm_mmu_reset_context(vcpu);
1386
	init_rmode(vcpu->kvm);
A
Avi Kivity 已提交
1387 1388
}

1389
#ifdef CONFIG_X86_64
A
Avi Kivity 已提交
1390 1391 1392 1393 1394 1395 1396 1397

static void enter_lmode(struct kvm_vcpu *vcpu)
{
	u32 guest_tr_ar;

	guest_tr_ar = vmcs_read32(GUEST_TR_AR_BYTES);
	if ((guest_tr_ar & AR_TYPE_MASK) != AR_TYPE_BUSY_64_TSS) {
		printk(KERN_DEBUG "%s: tss fixup for long mode. \n",
1398
		       __func__);
A
Avi Kivity 已提交
1399 1400 1401 1402 1403
		vmcs_write32(GUEST_TR_AR_BYTES,
			     (guest_tr_ar & ~AR_TYPE_MASK)
			     | AR_TYPE_BUSY_64_TSS);
	}

1404
	vcpu->arch.shadow_efer |= EFER_LMA;
A
Avi Kivity 已提交
1405

R
Rusty Russell 已提交
1406
	find_msr_entry(to_vmx(vcpu), MSR_EFER)->data |= EFER_LMA | EFER_LME;
A
Avi Kivity 已提交
1407 1408
	vmcs_write32(VM_ENTRY_CONTROLS,
		     vmcs_read32(VM_ENTRY_CONTROLS)
1409
		     | VM_ENTRY_IA32E_MODE);
A
Avi Kivity 已提交
1410 1411 1412 1413
}

static void exit_lmode(struct kvm_vcpu *vcpu)
{
1414
	vcpu->arch.shadow_efer &= ~EFER_LMA;
A
Avi Kivity 已提交
1415 1416 1417

	vmcs_write32(VM_ENTRY_CONTROLS,
		     vmcs_read32(VM_ENTRY_CONTROLS)
1418
		     & ~VM_ENTRY_IA32E_MODE);
A
Avi Kivity 已提交
1419 1420 1421 1422
}

#endif

1423 1424 1425
static void vmx_flush_tlb(struct kvm_vcpu *vcpu)
{
	vpid_sync_vcpu_all(to_vmx(vcpu));
1426 1427
	if (vm_need_ept())
		ept_sync_context(construct_eptp(vcpu->arch.mmu.root_hpa));
1428 1429
}

1430
static void vmx_decache_cr4_guest_bits(struct kvm_vcpu *vcpu)
1431
{
1432 1433
	vcpu->arch.cr4 &= KVM_GUEST_CR4_MASK;
	vcpu->arch.cr4 |= vmcs_readl(GUEST_CR4) & ~KVM_GUEST_CR4_MASK;
1434 1435
}

1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458
static void ept_load_pdptrs(struct kvm_vcpu *vcpu)
{
	if (is_paging(vcpu) && is_pae(vcpu) && !is_long_mode(vcpu)) {
		if (!load_pdptrs(vcpu, vcpu->arch.cr3)) {
			printk(KERN_ERR "EPT: Fail to load pdptrs!\n");
			return;
		}
		vmcs_write64(GUEST_PDPTR0, vcpu->arch.pdptrs[0]);
		vmcs_write64(GUEST_PDPTR1, vcpu->arch.pdptrs[1]);
		vmcs_write64(GUEST_PDPTR2, vcpu->arch.pdptrs[2]);
		vmcs_write64(GUEST_PDPTR3, vcpu->arch.pdptrs[3]);
	}
}

static void vmx_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);

static void ept_update_paging_mode_cr0(unsigned long *hw_cr0,
					unsigned long cr0,
					struct kvm_vcpu *vcpu)
{
	if (!(cr0 & X86_CR0_PG)) {
		/* From paging/starting to nonpaging */
		vmcs_write32(CPU_BASED_VM_EXEC_CONTROL,
1459
			     vmcs_read32(CPU_BASED_VM_EXEC_CONTROL) |
1460 1461 1462 1463 1464 1465 1466 1467 1468
			     (CPU_BASED_CR3_LOAD_EXITING |
			      CPU_BASED_CR3_STORE_EXITING));
		vcpu->arch.cr0 = cr0;
		vmx_set_cr4(vcpu, vcpu->arch.cr4);
		*hw_cr0 |= X86_CR0_PE | X86_CR0_PG;
		*hw_cr0 &= ~X86_CR0_WP;
	} else if (!is_paging(vcpu)) {
		/* From nonpaging to paging */
		vmcs_write32(CPU_BASED_VM_EXEC_CONTROL,
1469
			     vmcs_read32(CPU_BASED_VM_EXEC_CONTROL) &
1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488
			     ~(CPU_BASED_CR3_LOAD_EXITING |
			       CPU_BASED_CR3_STORE_EXITING));
		vcpu->arch.cr0 = cr0;
		vmx_set_cr4(vcpu, vcpu->arch.cr4);
		if (!(vcpu->arch.cr0 & X86_CR0_WP))
			*hw_cr0 &= ~X86_CR0_WP;
	}
}

static void ept_update_paging_mode_cr4(unsigned long *hw_cr4,
					struct kvm_vcpu *vcpu)
{
	if (!is_paging(vcpu)) {
		*hw_cr4 &= ~X86_CR4_PAE;
		*hw_cr4 |= X86_CR4_PSE;
	} else if (!(vcpu->arch.cr4 & X86_CR4_PAE))
		*hw_cr4 &= ~X86_CR4_PAE;
}

A
Avi Kivity 已提交
1489 1490
static void vmx_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
{
1491 1492 1493
	unsigned long hw_cr0 = (cr0 & ~KVM_GUEST_CR0_MASK) |
				KVM_VM_CR0_ALWAYS_ON;

1494 1495
	vmx_fpu_deactivate(vcpu);

1496
	if (vcpu->arch.rmode.active && (cr0 & X86_CR0_PE))
A
Avi Kivity 已提交
1497 1498
		enter_pmode(vcpu);

1499
	if (!vcpu->arch.rmode.active && !(cr0 & X86_CR0_PE))
A
Avi Kivity 已提交
1500 1501
		enter_rmode(vcpu);

1502
#ifdef CONFIG_X86_64
1503
	if (vcpu->arch.shadow_efer & EFER_LME) {
1504
		if (!is_paging(vcpu) && (cr0 & X86_CR0_PG))
A
Avi Kivity 已提交
1505
			enter_lmode(vcpu);
1506
		if (is_paging(vcpu) && !(cr0 & X86_CR0_PG))
A
Avi Kivity 已提交
1507 1508 1509 1510
			exit_lmode(vcpu);
	}
#endif

1511 1512 1513
	if (vm_need_ept())
		ept_update_paging_mode_cr0(&hw_cr0, cr0, vcpu);

A
Avi Kivity 已提交
1514
	vmcs_writel(CR0_READ_SHADOW, cr0);
1515
	vmcs_writel(GUEST_CR0, hw_cr0);
1516
	vcpu->arch.cr0 = cr0;
1517

1518
	if (!(cr0 & X86_CR0_TS) || !(cr0 & X86_CR0_PE))
1519
		vmx_fpu_activate(vcpu);
A
Avi Kivity 已提交
1520 1521
}

1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533
static u64 construct_eptp(unsigned long root_hpa)
{
	u64 eptp;

	/* TODO write the value reading from MSR */
	eptp = VMX_EPT_DEFAULT_MT |
		VMX_EPT_DEFAULT_GAW << VMX_EPT_GAW_EPTP_SHIFT;
	eptp |= (root_hpa & PAGE_MASK);

	return eptp;
}

A
Avi Kivity 已提交
1534 1535
static void vmx_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3)
{
1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548
	unsigned long guest_cr3;
	u64 eptp;

	guest_cr3 = cr3;
	if (vm_need_ept()) {
		eptp = construct_eptp(cr3);
		vmcs_write64(EPT_POINTER, eptp);
		ept_sync_context(eptp);
		ept_load_pdptrs(vcpu);
		guest_cr3 = is_paging(vcpu) ? vcpu->arch.cr3 :
			VMX_EPT_IDENTITY_PAGETABLE_ADDR;
	}

1549
	vmx_flush_tlb(vcpu);
1550
	vmcs_writel(GUEST_CR3, guest_cr3);
1551
	if (vcpu->arch.cr0 & X86_CR0_PE)
1552
		vmx_fpu_deactivate(vcpu);
A
Avi Kivity 已提交
1553 1554 1555 1556
}

static void vmx_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
{
1557 1558 1559
	unsigned long hw_cr4 = cr4 | (vcpu->arch.rmode.active ?
		    KVM_RMODE_VM_CR4_ALWAYS_ON : KVM_PMODE_VM_CR4_ALWAYS_ON);

1560
	vcpu->arch.cr4 = cr4;
1561 1562 1563 1564 1565
	if (vm_need_ept())
		ept_update_paging_mode_cr4(&hw_cr4, vcpu);

	vmcs_writel(CR4_READ_SHADOW, cr4);
	vmcs_writel(GUEST_CR4, hw_cr4);
A
Avi Kivity 已提交
1566 1567 1568 1569
}

static void vmx_set_efer(struct kvm_vcpu *vcpu, u64 efer)
{
R
Rusty Russell 已提交
1570 1571
	struct vcpu_vmx *vmx = to_vmx(vcpu);
	struct kvm_msr_entry *msr = find_msr_entry(vmx, MSR_EFER);
A
Avi Kivity 已提交
1572

1573
	vcpu->arch.shadow_efer = efer;
1574 1575
	if (!msr)
		return;
A
Avi Kivity 已提交
1576 1577 1578
	if (efer & EFER_LMA) {
		vmcs_write32(VM_ENTRY_CONTROLS,
				     vmcs_read32(VM_ENTRY_CONTROLS) |
1579
				     VM_ENTRY_IA32E_MODE);
A
Avi Kivity 已提交
1580 1581 1582 1583 1584
		msr->data = efer;

	} else {
		vmcs_write32(VM_ENTRY_CONTROLS,
				     vmcs_read32(VM_ENTRY_CONTROLS) &
1585
				     ~VM_ENTRY_IA32E_MODE);
A
Avi Kivity 已提交
1586 1587 1588

		msr->data = efer & ~EFER_LME;
	}
R
Rusty Russell 已提交
1589
	setup_msrs(vmx);
A
Avi Kivity 已提交
1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621
}

static u64 vmx_get_segment_base(struct kvm_vcpu *vcpu, int seg)
{
	struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];

	return vmcs_readl(sf->base);
}

static void vmx_get_segment(struct kvm_vcpu *vcpu,
			    struct kvm_segment *var, int seg)
{
	struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
	u32 ar;

	var->base = vmcs_readl(sf->base);
	var->limit = vmcs_read32(sf->limit);
	var->selector = vmcs_read16(sf->selector);
	ar = vmcs_read32(sf->ar_bytes);
	if (ar & AR_UNUSABLE_MASK)
		ar = 0;
	var->type = ar & 15;
	var->s = (ar >> 4) & 1;
	var->dpl = (ar >> 5) & 3;
	var->present = (ar >> 7) & 1;
	var->avl = (ar >> 12) & 1;
	var->l = (ar >> 13) & 1;
	var->db = (ar >> 14) & 1;
	var->g = (ar >> 15) & 1;
	var->unusable = (ar >> 16) & 1;
}

1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635
static int vmx_get_cpl(struct kvm_vcpu *vcpu)
{
	struct kvm_segment kvm_seg;

	if (!(vcpu->arch.cr0 & X86_CR0_PE)) /* if real mode */
		return 0;

	if (vmx_get_rflags(vcpu) & X86_EFLAGS_VM) /* if virtual 8086 */
		return 3;

	vmx_get_segment(vcpu, &kvm_seg, VCPU_SREG_CS);
	return kvm_seg.selector & 3;
}

1636
static u32 vmx_segment_access_rights(struct kvm_segment *var)
A
Avi Kivity 已提交
1637 1638 1639
{
	u32 ar;

1640
	if (var->unusable)
A
Avi Kivity 已提交
1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651
		ar = 1 << 16;
	else {
		ar = var->type & 15;
		ar |= (var->s & 1) << 4;
		ar |= (var->dpl & 3) << 5;
		ar |= (var->present & 1) << 7;
		ar |= (var->avl & 1) << 12;
		ar |= (var->l & 1) << 13;
		ar |= (var->db & 1) << 14;
		ar |= (var->g & 1) << 15;
	}
1652 1653
	if (ar == 0) /* a 0 value means unusable */
		ar = AR_UNUSABLE_MASK;
1654 1655 1656 1657 1658 1659 1660 1661 1662 1663

	return ar;
}

static void vmx_set_segment(struct kvm_vcpu *vcpu,
			    struct kvm_segment *var, int seg)
{
	struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
	u32 ar;

1664 1665 1666 1667 1668
	if (vcpu->arch.rmode.active && seg == VCPU_SREG_TR) {
		vcpu->arch.rmode.tr.selector = var->selector;
		vcpu->arch.rmode.tr.base = var->base;
		vcpu->arch.rmode.tr.limit = var->limit;
		vcpu->arch.rmode.tr.ar = vmx_segment_access_rights(var);
1669 1670 1671 1672 1673
		return;
	}
	vmcs_writel(sf->base, var->base);
	vmcs_write32(sf->limit, var->limit);
	vmcs_write16(sf->selector, var->selector);
1674
	if (vcpu->arch.rmode.active && var->s) {
1675 1676 1677 1678 1679 1680 1681 1682
		/*
		 * Hack real-mode segments into vm86 compatibility.
		 */
		if (var->base == 0xffff0000 && var->selector == 0xf000)
			vmcs_writel(sf->base, 0xf0000);
		ar = 0xf3;
	} else
		ar = vmx_segment_access_rights(var);
A
Avi Kivity 已提交
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 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717
	vmcs_write32(sf->ar_bytes, ar);
}

static void vmx_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l)
{
	u32 ar = vmcs_read32(GUEST_CS_AR_BYTES);

	*db = (ar >> 14) & 1;
	*l = (ar >> 13) & 1;
}

static void vmx_get_idt(struct kvm_vcpu *vcpu, struct descriptor_table *dt)
{
	dt->limit = vmcs_read32(GUEST_IDTR_LIMIT);
	dt->base = vmcs_readl(GUEST_IDTR_BASE);
}

static void vmx_set_idt(struct kvm_vcpu *vcpu, struct descriptor_table *dt)
{
	vmcs_write32(GUEST_IDTR_LIMIT, dt->limit);
	vmcs_writel(GUEST_IDTR_BASE, dt->base);
}

static void vmx_get_gdt(struct kvm_vcpu *vcpu, struct descriptor_table *dt)
{
	dt->limit = vmcs_read32(GUEST_GDTR_LIMIT);
	dt->base = vmcs_readl(GUEST_GDTR_BASE);
}

static void vmx_set_gdt(struct kvm_vcpu *vcpu, struct descriptor_table *dt)
{
	vmcs_write32(GUEST_GDTR_LIMIT, dt->limit);
	vmcs_writel(GUEST_GDTR_BASE, dt->base);
}

M
Mike Day 已提交
1718
static int init_rmode_tss(struct kvm *kvm)
A
Avi Kivity 已提交
1719 1720
{
	gfn_t fn = rmode_tss_base(kvm) >> PAGE_SHIFT;
1721
	u16 data = 0;
1722
	int ret = 0;
1723
	int r;
A
Avi Kivity 已提交
1724

1725 1726
	r = kvm_clear_guest_page(kvm, fn, 0, PAGE_SIZE);
	if (r < 0)
1727
		goto out;
1728 1729 1730
	data = TSS_BASE_SIZE + TSS_REDIRECTION_SIZE;
	r = kvm_write_guest_page(kvm, fn++, &data, 0x66, sizeof(u16));
	if (r < 0)
1731
		goto out;
1732 1733
	r = kvm_clear_guest_page(kvm, fn++, 0, PAGE_SIZE);
	if (r < 0)
1734
		goto out;
1735 1736
	r = kvm_clear_guest_page(kvm, fn, 0, PAGE_SIZE);
	if (r < 0)
1737
		goto out;
1738
	data = ~0;
1739 1740 1741
	r = kvm_write_guest_page(kvm, fn, &data,
				 RMODE_TSS_SIZE - 2 * PAGE_SIZE - 1,
				 sizeof(u8));
1742
	if (r < 0)
1743 1744 1745 1746 1747
		goto out;

	ret = 1;
out:
	return ret;
A
Avi Kivity 已提交
1748 1749
}

1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784
static int init_rmode_identity_map(struct kvm *kvm)
{
	int i, r, ret;
	pfn_t identity_map_pfn;
	u32 tmp;

	if (!vm_need_ept())
		return 1;
	if (unlikely(!kvm->arch.ept_identity_pagetable)) {
		printk(KERN_ERR "EPT: identity-mapping pagetable "
			"haven't been allocated!\n");
		return 0;
	}
	if (likely(kvm->arch.ept_identity_pagetable_done))
		return 1;
	ret = 0;
	identity_map_pfn = VMX_EPT_IDENTITY_PAGETABLE_ADDR >> PAGE_SHIFT;
	r = kvm_clear_guest_page(kvm, identity_map_pfn, 0, PAGE_SIZE);
	if (r < 0)
		goto out;
	/* Set up identity-mapping pagetable for EPT in real mode */
	for (i = 0; i < PT32_ENT_PER_PAGE; i++) {
		tmp = (i << 22) + (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER |
			_PAGE_ACCESSED | _PAGE_DIRTY | _PAGE_PSE);
		r = kvm_write_guest_page(kvm, identity_map_pfn,
				&tmp, i * sizeof(tmp), sizeof(tmp));
		if (r < 0)
			goto out;
	}
	kvm->arch.ept_identity_pagetable_done = true;
	ret = 1;
out:
	return ret;
}

A
Avi Kivity 已提交
1785 1786 1787 1788 1789 1790 1791 1792 1793 1794
static void seg_setup(int seg)
{
	struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];

	vmcs_write16(sf->selector, 0);
	vmcs_writel(sf->base, 0);
	vmcs_write32(sf->limit, 0xffff);
	vmcs_write32(sf->ar_bytes, 0x93);
}

1795 1796 1797 1798 1799
static int alloc_apic_access_page(struct kvm *kvm)
{
	struct kvm_userspace_memory_region kvm_userspace_mem;
	int r = 0;

1800
	down_write(&kvm->slots_lock);
1801
	if (kvm->arch.apic_access_page)
1802 1803 1804 1805 1806 1807 1808 1809
		goto out;
	kvm_userspace_mem.slot = APIC_ACCESS_PAGE_PRIVATE_MEMSLOT;
	kvm_userspace_mem.flags = 0;
	kvm_userspace_mem.guest_phys_addr = 0xfee00000ULL;
	kvm_userspace_mem.memory_size = PAGE_SIZE;
	r = __kvm_set_memory_region(kvm, &kvm_userspace_mem, 0);
	if (r)
		goto out;
1810 1811

	down_read(&current->mm->mmap_sem);
1812
	kvm->arch.apic_access_page = gfn_to_page(kvm, 0xfee00);
1813
	up_read(&current->mm->mmap_sem);
1814
out:
1815
	up_write(&kvm->slots_lock);
1816 1817 1818
	return r;
}

1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843
static int alloc_identity_pagetable(struct kvm *kvm)
{
	struct kvm_userspace_memory_region kvm_userspace_mem;
	int r = 0;

	down_write(&kvm->slots_lock);
	if (kvm->arch.ept_identity_pagetable)
		goto out;
	kvm_userspace_mem.slot = IDENTITY_PAGETABLE_PRIVATE_MEMSLOT;
	kvm_userspace_mem.flags = 0;
	kvm_userspace_mem.guest_phys_addr = VMX_EPT_IDENTITY_PAGETABLE_ADDR;
	kvm_userspace_mem.memory_size = PAGE_SIZE;
	r = __kvm_set_memory_region(kvm, &kvm_userspace_mem, 0);
	if (r)
		goto out;

	down_read(&current->mm->mmap_sem);
	kvm->arch.ept_identity_pagetable = gfn_to_page(kvm,
			VMX_EPT_IDENTITY_PAGETABLE_ADDR >> PAGE_SHIFT);
	up_read(&current->mm->mmap_sem);
out:
	up_write(&kvm->slots_lock);
	return r;
}

1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859
static void allocate_vpid(struct vcpu_vmx *vmx)
{
	int vpid;

	vmx->vpid = 0;
	if (!enable_vpid || !cpu_has_vmx_vpid())
		return;
	spin_lock(&vmx_vpid_lock);
	vpid = find_first_zero_bit(vmx_vpid_bitmap, VMX_NR_VPIDS);
	if (vpid < VMX_NR_VPIDS) {
		vmx->vpid = vpid;
		__set_bit(vpid, vmx_vpid_bitmap);
	}
	spin_unlock(&vmx_vpid_lock);
}

1860
static void vmx_disable_intercept_for_msr(struct page *msr_bitmap, u32 msr)
S
Sheng Yang 已提交
1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883
{
	void *va;

	if (!cpu_has_vmx_msr_bitmap())
		return;

	/*
	 * See Intel PRM Vol. 3, 20.6.9 (MSR-Bitmap Address). Early manuals
	 * have the write-low and read-high bitmap offsets the wrong way round.
	 * We can control MSRs 0x00000000-0x00001fff and 0xc0000000-0xc0001fff.
	 */
	va = kmap(msr_bitmap);
	if (msr <= 0x1fff) {
		__clear_bit(msr, va + 0x000); /* read-low */
		__clear_bit(msr, va + 0x800); /* write-low */
	} else if ((msr >= 0xc0000000) && (msr <= 0xc0001fff)) {
		msr &= 0x1fff;
		__clear_bit(msr, va + 0x400); /* read-high */
		__clear_bit(msr, va + 0xc00); /* write-high */
	}
	kunmap(msr_bitmap);
}

A
Avi Kivity 已提交
1884 1885 1886
/*
 * Sets up the vmcs for emulated real mode.
 */
R
Rusty Russell 已提交
1887
static int vmx_vcpu_setup(struct vcpu_vmx *vmx)
A
Avi Kivity 已提交
1888 1889 1890 1891 1892 1893
{
	u32 host_sysenter_cs;
	u32 junk;
	unsigned long a;
	struct descriptor_table dt;
	int i;
1894
	unsigned long kvm_vmx_return;
1895
	u32 exec_control;
A
Avi Kivity 已提交
1896 1897

	/* I/O */
1898 1899
	vmcs_write64(IO_BITMAP_A, page_to_phys(vmx_io_bitmap_a));
	vmcs_write64(IO_BITMAP_B, page_to_phys(vmx_io_bitmap_b));
A
Avi Kivity 已提交
1900

S
Sheng Yang 已提交
1901 1902 1903
	if (cpu_has_vmx_msr_bitmap())
		vmcs_write64(MSR_BITMAP, page_to_phys(vmx_msr_bitmap));

A
Avi Kivity 已提交
1904 1905 1906
	vmcs_write64(VMCS_LINK_POINTER, -1ull); /* 22.3.1.5 */

	/* Control */
1907 1908
	vmcs_write32(PIN_BASED_VM_EXEC_CONTROL,
		vmcs_config.pin_based_exec_ctrl);
1909 1910 1911 1912 1913 1914 1915 1916 1917

	exec_control = vmcs_config.cpu_based_exec_ctrl;
	if (!vm_need_tpr_shadow(vmx->vcpu.kvm)) {
		exec_control &= ~CPU_BASED_TPR_SHADOW;
#ifdef CONFIG_X86_64
		exec_control |= CPU_BASED_CR8_STORE_EXITING |
				CPU_BASED_CR8_LOAD_EXITING;
#endif
	}
S
Sheng Yang 已提交
1918 1919 1920
	if (!vm_need_ept())
		exec_control |= CPU_BASED_CR3_STORE_EXITING |
				CPU_BASED_CR3_LOAD_EXITING;
1921
	vmcs_write32(CPU_BASED_VM_EXEC_CONTROL, exec_control);
A
Avi Kivity 已提交
1922

1923 1924 1925 1926 1927
	if (cpu_has_secondary_exec_ctrls()) {
		exec_control = vmcs_config.cpu_based_2nd_exec_ctrl;
		if (!vm_need_virtualize_apic_accesses(vmx->vcpu.kvm))
			exec_control &=
				~SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES;
1928 1929
		if (vmx->vpid == 0)
			exec_control &= ~SECONDARY_EXEC_ENABLE_VPID;
S
Sheng Yang 已提交
1930 1931
		if (!vm_need_ept())
			exec_control &= ~SECONDARY_EXEC_ENABLE_EPT;
1932 1933
		vmcs_write32(SECONDARY_VM_EXEC_CONTROL, exec_control);
	}
1934

1935 1936
	vmcs_write32(PAGE_FAULT_ERROR_CODE_MASK, !!bypass_guest_pf);
	vmcs_write32(PAGE_FAULT_ERROR_CODE_MATCH, !!bypass_guest_pf);
A
Avi Kivity 已提交
1937 1938 1939 1940 1941 1942 1943 1944 1945
	vmcs_write32(CR3_TARGET_COUNT, 0);           /* 22.2.1 */

	vmcs_writel(HOST_CR0, read_cr0());  /* 22.2.3 */
	vmcs_writel(HOST_CR4, read_cr4());  /* 22.2.3, 22.2.5 */
	vmcs_writel(HOST_CR3, read_cr3());  /* 22.2.3  FIXME: shadow tables */

	vmcs_write16(HOST_CS_SELECTOR, __KERNEL_CS);  /* 22.2.4 */
	vmcs_write16(HOST_DS_SELECTOR, __KERNEL_DS);  /* 22.2.4 */
	vmcs_write16(HOST_ES_SELECTOR, __KERNEL_DS);  /* 22.2.4 */
1946 1947
	vmcs_write16(HOST_FS_SELECTOR, kvm_read_fs());    /* 22.2.4 */
	vmcs_write16(HOST_GS_SELECTOR, kvm_read_gs());    /* 22.2.4 */
A
Avi Kivity 已提交
1948
	vmcs_write16(HOST_SS_SELECTOR, __KERNEL_DS);  /* 22.2.4 */
1949
#ifdef CONFIG_X86_64
A
Avi Kivity 已提交
1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960
	rdmsrl(MSR_FS_BASE, a);
	vmcs_writel(HOST_FS_BASE, a); /* 22.2.4 */
	rdmsrl(MSR_GS_BASE, a);
	vmcs_writel(HOST_GS_BASE, a); /* 22.2.4 */
#else
	vmcs_writel(HOST_FS_BASE, 0); /* 22.2.4 */
	vmcs_writel(HOST_GS_BASE, 0); /* 22.2.4 */
#endif

	vmcs_write16(HOST_TR_SELECTOR, GDT_ENTRY_TSS*8);  /* 22.2.4 */

1961
	kvm_get_idt(&dt);
A
Avi Kivity 已提交
1962 1963
	vmcs_writel(HOST_IDTR_BASE, dt.base);   /* 22.2.4 */

M
Mike Day 已提交
1964
	asm("mov $.Lkvm_vmx_return, %0" : "=r"(kvm_vmx_return));
1965
	vmcs_writel(HOST_RIP, kvm_vmx_return); /* 22.2.5 */
1966 1967 1968
	vmcs_write32(VM_EXIT_MSR_STORE_COUNT, 0);
	vmcs_write32(VM_EXIT_MSR_LOAD_COUNT, 0);
	vmcs_write32(VM_ENTRY_MSR_LOAD_COUNT, 0);
A
Avi Kivity 已提交
1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980

	rdmsr(MSR_IA32_SYSENTER_CS, host_sysenter_cs, junk);
	vmcs_write32(HOST_IA32_SYSENTER_CS, host_sysenter_cs);
	rdmsrl(MSR_IA32_SYSENTER_ESP, a);
	vmcs_writel(HOST_IA32_SYSENTER_ESP, a);   /* 22.2.3 */
	rdmsrl(MSR_IA32_SYSENTER_EIP, a);
	vmcs_writel(HOST_IA32_SYSENTER_EIP, a);   /* 22.2.3 */

	for (i = 0; i < NR_VMX_MSR; ++i) {
		u32 index = vmx_msr_index[i];
		u32 data_low, data_high;
		u64 data;
1981
		int j = vmx->nmsrs;
A
Avi Kivity 已提交
1982 1983 1984

		if (rdmsr_safe(index, &data_low, &data_high) < 0)
			continue;
1985 1986
		if (wrmsr_safe(index, data_low, data_high) < 0)
			continue;
A
Avi Kivity 已提交
1987
		data = data_low | ((u64)data_high << 32);
1988 1989 1990 1991 1992
		vmx->host_msrs[j].index = index;
		vmx->host_msrs[j].reserved = 0;
		vmx->host_msrs[j].data = data;
		vmx->guest_msrs[j] = vmx->host_msrs[j];
		++vmx->nmsrs;
A
Avi Kivity 已提交
1993 1994
	}

1995
	vmcs_write32(VM_EXIT_CONTROLS, vmcs_config.vmexit_ctrl);
A
Avi Kivity 已提交
1996 1997

	/* 22.2.1, 20.8.1 */
1998 1999
	vmcs_write32(VM_ENTRY_CONTROLS, vmcs_config.vmentry_ctrl);

2000 2001 2002
	vmcs_writel(CR0_GUEST_HOST_MASK, ~0UL);
	vmcs_writel(CR4_GUEST_HOST_MASK, KVM_GUEST_CR4_MASK);

2003

2004 2005 2006
	return 0;
}

2007 2008 2009 2010 2011 2012 2013 2014 2015
static int init_rmode(struct kvm *kvm)
{
	if (!init_rmode_tss(kvm))
		return 0;
	if (!init_rmode_identity_map(kvm))
		return 0;
	return 1;
}

2016 2017 2018 2019 2020 2021
static int vmx_vcpu_reset(struct kvm_vcpu *vcpu)
{
	struct vcpu_vmx *vmx = to_vmx(vcpu);
	u64 msr;
	int ret;

2022
	down_read(&vcpu->kvm->slots_lock);
2023
	if (!init_rmode(vmx->vcpu.kvm)) {
2024 2025 2026 2027
		ret = -ENOMEM;
		goto out;
	}

2028
	vmx->vcpu.arch.rmode.active = 0;
2029

2030
	vmx->vcpu.arch.regs[VCPU_REGS_RDX] = get_rdx_init_val();
2031
	kvm_set_cr8(&vmx->vcpu, 0);
2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046
	msr = 0xfee00000 | MSR_IA32_APICBASE_ENABLE;
	if (vmx->vcpu.vcpu_id == 0)
		msr |= MSR_IA32_APICBASE_BSP;
	kvm_set_apic_base(&vmx->vcpu, msr);

	fx_init(&vmx->vcpu);

	/*
	 * GUEST_CS_BASE should really be 0xffff0000, but VT vm86 mode
	 * insists on having GUEST_CS_BASE == GUEST_CS_SELECTOR << 4.  Sigh.
	 */
	if (vmx->vcpu.vcpu_id == 0) {
		vmcs_write16(GUEST_CS_SELECTOR, 0xf000);
		vmcs_writel(GUEST_CS_BASE, 0x000f0000);
	} else {
2047 2048
		vmcs_write16(GUEST_CS_SELECTOR, vmx->vcpu.arch.sipi_vector << 8);
		vmcs_writel(GUEST_CS_BASE, vmx->vcpu.arch.sipi_vector << 12);
2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099
	}
	vmcs_write32(GUEST_CS_LIMIT, 0xffff);
	vmcs_write32(GUEST_CS_AR_BYTES, 0x9b);

	seg_setup(VCPU_SREG_DS);
	seg_setup(VCPU_SREG_ES);
	seg_setup(VCPU_SREG_FS);
	seg_setup(VCPU_SREG_GS);
	seg_setup(VCPU_SREG_SS);

	vmcs_write16(GUEST_TR_SELECTOR, 0);
	vmcs_writel(GUEST_TR_BASE, 0);
	vmcs_write32(GUEST_TR_LIMIT, 0xffff);
	vmcs_write32(GUEST_TR_AR_BYTES, 0x008b);

	vmcs_write16(GUEST_LDTR_SELECTOR, 0);
	vmcs_writel(GUEST_LDTR_BASE, 0);
	vmcs_write32(GUEST_LDTR_LIMIT, 0xffff);
	vmcs_write32(GUEST_LDTR_AR_BYTES, 0x00082);

	vmcs_write32(GUEST_SYSENTER_CS, 0);
	vmcs_writel(GUEST_SYSENTER_ESP, 0);
	vmcs_writel(GUEST_SYSENTER_EIP, 0);

	vmcs_writel(GUEST_RFLAGS, 0x02);
	if (vmx->vcpu.vcpu_id == 0)
		vmcs_writel(GUEST_RIP, 0xfff0);
	else
		vmcs_writel(GUEST_RIP, 0);
	vmcs_writel(GUEST_RSP, 0);

	/* todo: dr0 = dr1 = dr2 = dr3 = 0; dr6 = 0xffff0ff0 */
	vmcs_writel(GUEST_DR7, 0x400);

	vmcs_writel(GUEST_GDTR_BASE, 0);
	vmcs_write32(GUEST_GDTR_LIMIT, 0xffff);

	vmcs_writel(GUEST_IDTR_BASE, 0);
	vmcs_write32(GUEST_IDTR_LIMIT, 0xffff);

	vmcs_write32(GUEST_ACTIVITY_STATE, 0);
	vmcs_write32(GUEST_INTERRUPTIBILITY_INFO, 0);
	vmcs_write32(GUEST_PENDING_DBG_EXCEPTIONS, 0);

	guest_write_tsc(0);

	/* Special registers */
	vmcs_write64(GUEST_IA32_DEBUGCTL, 0);

	setup_msrs(vmx);

A
Avi Kivity 已提交
2100 2101
	vmcs_write32(VM_ENTRY_INTR_INFO_FIELD, 0);  /* 22.2.1 */

2102 2103 2104 2105
	if (cpu_has_vmx_tpr_shadow()) {
		vmcs_write64(VIRTUAL_APIC_PAGE_ADDR, 0);
		if (vm_need_tpr_shadow(vmx->vcpu.kvm))
			vmcs_write64(VIRTUAL_APIC_PAGE_ADDR,
2106
				page_to_phys(vmx->vcpu.arch.apic->regs_page));
2107 2108 2109 2110 2111
		vmcs_write32(TPR_THRESHOLD, 0);
	}

	if (vm_need_virtualize_apic_accesses(vmx->vcpu.kvm))
		vmcs_write64(APIC_ACCESS_ADDR,
2112
			     page_to_phys(vmx->vcpu.kvm->arch.apic_access_page));
A
Avi Kivity 已提交
2113

2114 2115 2116
	if (vmx->vpid != 0)
		vmcs_write16(VIRTUAL_PROCESSOR_ID, vmx->vpid);

2117 2118
	vmx->vcpu.arch.cr0 = 0x60000010;
	vmx_set_cr0(&vmx->vcpu, vmx->vcpu.arch.cr0); /* enter rmode */
R
Rusty Russell 已提交
2119 2120 2121 2122
	vmx_set_cr4(&vmx->vcpu, 0);
	vmx_set_efer(&vmx->vcpu, 0);
	vmx_fpu_activate(&vmx->vcpu);
	update_exception_bitmap(&vmx->vcpu);
A
Avi Kivity 已提交
2123

2124 2125
	vpid_sync_vcpu_all(vmx);

2126
	ret = 0;
A
Avi Kivity 已提交
2127 2128

out:
2129
	up_read(&vcpu->kvm->slots_lock);
A
Avi Kivity 已提交
2130 2131 2132
	return ret;
}

2133 2134
static void vmx_inject_irq(struct kvm_vcpu *vcpu, int irq)
{
2135 2136
	struct vcpu_vmx *vmx = to_vmx(vcpu);

F
Feng (Eric) Liu 已提交
2137 2138
	KVMTRACE_1D(INJ_VIRQ, vcpu, (u32)irq, handler);

2139
	if (vcpu->arch.rmode.active) {
2140 2141 2142
		vmx->rmode.irq.pending = true;
		vmx->rmode.irq.vector = irq;
		vmx->rmode.irq.rip = vmcs_readl(GUEST_RIP);
2143 2144 2145
		vmcs_write32(VM_ENTRY_INTR_INFO_FIELD,
			     irq | INTR_TYPE_SOFT_INTR | INTR_INFO_VALID_MASK);
		vmcs_write32(VM_ENTRY_INSTRUCTION_LEN, 1);
2146
		vmcs_writel(GUEST_RIP, vmx->rmode.irq.rip - 1);
2147 2148 2149 2150 2151 2152
		return;
	}
	vmcs_write32(VM_ENTRY_INTR_INFO_FIELD,
			irq | INTR_TYPE_EXT_INTR | INTR_INFO_VALID_MASK);
}

2153 2154 2155 2156 2157 2158 2159
static void vmx_inject_nmi(struct kvm_vcpu *vcpu)
{
	vmcs_write32(VM_ENTRY_INTR_INFO_FIELD,
			INTR_TYPE_NMI_INTR | INTR_INFO_VALID_MASK | NMI_VECTOR);
	vcpu->arch.nmi_pending = 0;
}

A
Avi Kivity 已提交
2160 2161
static void kvm_do_inject_irq(struct kvm_vcpu *vcpu)
{
2162 2163
	int word_index = __ffs(vcpu->arch.irq_summary);
	int bit_index = __ffs(vcpu->arch.irq_pending[word_index]);
A
Avi Kivity 已提交
2164 2165
	int irq = word_index * BITS_PER_LONG + bit_index;

2166 2167 2168
	clear_bit(bit_index, &vcpu->arch.irq_pending[word_index]);
	if (!vcpu->arch.irq_pending[word_index])
		clear_bit(word_index, &vcpu->arch.irq_summary);
2169
	vmx_inject_irq(vcpu, irq);
A
Avi Kivity 已提交
2170 2171
}

2172 2173 2174

static void do_interrupt_requests(struct kvm_vcpu *vcpu,
				       struct kvm_run *kvm_run)
A
Avi Kivity 已提交
2175
{
2176 2177
	u32 cpu_based_vm_exec_control;

2178
	vcpu->arch.interrupt_window_open =
2179 2180 2181
		((vmcs_readl(GUEST_RFLAGS) & X86_EFLAGS_IF) &&
		 (vmcs_read32(GUEST_INTERRUPTIBILITY_INFO) & 3) == 0);

2182 2183
	if (vcpu->arch.interrupt_window_open &&
	    vcpu->arch.irq_summary &&
2184
	    !(vmcs_read32(VM_ENTRY_INTR_INFO_FIELD) & INTR_INFO_VALID_MASK))
A
Avi Kivity 已提交
2185
		/*
2186
		 * If interrupts enabled, and not blocked by sti or mov ss. Good.
A
Avi Kivity 已提交
2187 2188
		 */
		kvm_do_inject_irq(vcpu);
2189 2190

	cpu_based_vm_exec_control = vmcs_read32(CPU_BASED_VM_EXEC_CONTROL);
2191 2192
	if (!vcpu->arch.interrupt_window_open &&
	    (vcpu->arch.irq_summary || kvm_run->request_interrupt_window))
A
Avi Kivity 已提交
2193 2194 2195
		/*
		 * Interrupts blocked.  Wait for unblock.
		 */
2196 2197 2198 2199
		cpu_based_vm_exec_control |= CPU_BASED_VIRTUAL_INTR_PENDING;
	else
		cpu_based_vm_exec_control &= ~CPU_BASED_VIRTUAL_INTR_PENDING;
	vmcs_write32(CPU_BASED_VM_EXEC_CONTROL, cpu_based_vm_exec_control);
A
Avi Kivity 已提交
2200 2201
}

2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214
static int vmx_set_tss_addr(struct kvm *kvm, unsigned int addr)
{
	int ret;
	struct kvm_userspace_memory_region tss_mem = {
		.slot = 8,
		.guest_phys_addr = addr,
		.memory_size = PAGE_SIZE * 3,
		.flags = 0,
	};

	ret = kvm_set_memory_region(kvm, &tss_mem, 0);
	if (ret)
		return ret;
2215
	kvm->arch.tss_addr = addr;
2216 2217 2218
	return 0;
}

A
Avi Kivity 已提交
2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239
static void kvm_guest_debug_pre(struct kvm_vcpu *vcpu)
{
	struct kvm_guest_debug *dbg = &vcpu->guest_debug;

	set_debugreg(dbg->bp[0], 0);
	set_debugreg(dbg->bp[1], 1);
	set_debugreg(dbg->bp[2], 2);
	set_debugreg(dbg->bp[3], 3);

	if (dbg->singlestep) {
		unsigned long flags;

		flags = vmcs_readl(GUEST_RFLAGS);
		flags |= X86_EFLAGS_TF | X86_EFLAGS_RF;
		vmcs_writel(GUEST_RFLAGS, flags);
	}
}

static int handle_rmode_exception(struct kvm_vcpu *vcpu,
				  int vec, u32 err_code)
{
2240
	if (!vcpu->arch.rmode.active)
A
Avi Kivity 已提交
2241 2242
		return 0;

2243 2244 2245 2246 2247
	/*
	 * Instruction with address size override prefix opcode 0x67
	 * Cause the #SS fault with 0 error code in VM86 mode.
	 */
	if (((vec == GP_VECTOR) || (vec == SS_VECTOR)) && err_code == 0)
2248
		if (emulate_instruction(vcpu, NULL, 0, 0, 0) == EMULATE_DONE)
A
Avi Kivity 已提交
2249 2250 2251 2252 2253 2254
			return 1;
	return 0;
}

static int handle_exception(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
2255
	struct vcpu_vmx *vmx = to_vmx(vcpu);
A
Avi Kivity 已提交
2256 2257 2258 2259 2260
	u32 intr_info, error_code;
	unsigned long cr2, rip;
	u32 vect_info;
	enum emulation_result er;

2261
	vect_info = vmx->idt_vectoring_info;
A
Avi Kivity 已提交
2262 2263 2264
	intr_info = vmcs_read32(VM_EXIT_INTR_INFO);

	if ((vect_info & VECTORING_INFO_VALID_MASK) &&
M
Mike Day 已提交
2265
						!is_page_fault(intr_info))
A
Avi Kivity 已提交
2266
		printk(KERN_ERR "%s: unexpected, vectoring info 0x%x "
2267
		       "intr info 0x%x\n", __func__, vect_info, intr_info);
A
Avi Kivity 已提交
2268

2269
	if (!irqchip_in_kernel(vcpu->kvm) && is_external_interrupt(vect_info)) {
A
Avi Kivity 已提交
2270
		int irq = vect_info & VECTORING_INFO_VECTOR_MASK;
2271 2272
		set_bit(irq, vcpu->arch.irq_pending);
		set_bit(irq / BITS_PER_LONG, &vcpu->arch.irq_summary);
A
Avi Kivity 已提交
2273 2274
	}

2275 2276
	if ((intr_info & INTR_INFO_INTR_TYPE_MASK) == 0x200) /* nmi */
		return 1;  /* already handled by vmx_vcpu_run() */
2277 2278

	if (is_no_device(intr_info)) {
2279
		vmx_fpu_activate(vcpu);
2280 2281 2282
		return 1;
	}

2283
	if (is_invalid_opcode(intr_info)) {
2284
		er = emulate_instruction(vcpu, kvm_run, 0, 0, EMULTYPE_TRAP_UD);
2285
		if (er != EMULATE_DONE)
2286
			kvm_queue_exception(vcpu, UD_VECTOR);
2287 2288 2289
		return 1;
	}

A
Avi Kivity 已提交
2290 2291
	error_code = 0;
	rip = vmcs_readl(GUEST_RIP);
2292
	if (intr_info & INTR_INFO_DELIVER_CODE_MASK)
A
Avi Kivity 已提交
2293 2294
		error_code = vmcs_read32(VM_EXIT_INTR_ERROR_CODE);
	if (is_page_fault(intr_info)) {
2295 2296 2297
		/* EPT won't cause page fault directly */
		if (vm_need_ept())
			BUG();
A
Avi Kivity 已提交
2298
		cr2 = vmcs_readl(EXIT_QUALIFICATION);
F
Feng (Eric) Liu 已提交
2299 2300
		KVMTRACE_3D(PAGE_FAULT, vcpu, error_code, (u32)cr2,
			    (u32)((u64)cr2 >> 32), handler);
2301 2302
		if (vect_info & VECTORING_INFO_VALID_MASK)
			kvm_mmu_unprotect_page_virt(vcpu, cr2);
2303
		return kvm_mmu_page_fault(vcpu, cr2, error_code);
A
Avi Kivity 已提交
2304 2305
	}

2306
	if (vcpu->arch.rmode.active &&
A
Avi Kivity 已提交
2307
	    handle_rmode_exception(vcpu, intr_info & INTR_INFO_VECTOR_MASK,
2308
								error_code)) {
2309 2310
		if (vcpu->arch.halt_request) {
			vcpu->arch.halt_request = 0;
2311 2312
			return kvm_emulate_halt(vcpu);
		}
A
Avi Kivity 已提交
2313
		return 1;
2314
	}
A
Avi Kivity 已提交
2315

M
Mike Day 已提交
2316 2317
	if ((intr_info & (INTR_INFO_INTR_TYPE_MASK | INTR_INFO_VECTOR_MASK)) ==
	    (INTR_TYPE_EXCEPTION | 1)) {
A
Avi Kivity 已提交
2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329
		kvm_run->exit_reason = KVM_EXIT_DEBUG;
		return 0;
	}
	kvm_run->exit_reason = KVM_EXIT_EXCEPTION;
	kvm_run->ex.exception = intr_info & INTR_INFO_VECTOR_MASK;
	kvm_run->ex.error_code = error_code;
	return 0;
}

static int handle_external_interrupt(struct kvm_vcpu *vcpu,
				     struct kvm_run *kvm_run)
{
A
Avi Kivity 已提交
2330
	++vcpu->stat.irq_exits;
F
Feng (Eric) Liu 已提交
2331
	KVMTRACE_1D(INTR, vcpu, vmcs_read32(VM_EXIT_INTR_INFO), handler);
A
Avi Kivity 已提交
2332 2333 2334
	return 1;
}

2335 2336 2337 2338 2339
static int handle_triple_fault(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
	kvm_run->exit_reason = KVM_EXIT_SHUTDOWN;
	return 0;
}
A
Avi Kivity 已提交
2340 2341 2342

static int handle_io(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
2343
	unsigned long exit_qualification;
2344 2345
	int size, down, in, string, rep;
	unsigned port;
A
Avi Kivity 已提交
2346

A
Avi Kivity 已提交
2347
	++vcpu->stat.io_exits;
2348
	exit_qualification = vmcs_readl(EXIT_QUALIFICATION);
2349
	string = (exit_qualification & 16) != 0;
2350 2351

	if (string) {
2352 2353
		if (emulate_instruction(vcpu,
					kvm_run, 0, 0, 0) == EMULATE_DO_MMIO)
2354 2355 2356 2357 2358 2359
			return 0;
		return 1;
	}

	size = (exit_qualification & 7) + 1;
	in = (exit_qualification & 8) != 0;
2360 2361 2362
	down = (vmcs_readl(GUEST_RFLAGS) & X86_EFLAGS_DF) != 0;
	rep = (exit_qualification & 32) != 0;
	port = exit_qualification >> 16;
2363

L
Laurent Vivier 已提交
2364
	return kvm_emulate_pio(vcpu, kvm_run, in, size, port);
A
Avi Kivity 已提交
2365 2366
}

I
Ingo Molnar 已提交
2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377
static void
vmx_patch_hypercall(struct kvm_vcpu *vcpu, unsigned char *hypercall)
{
	/*
	 * Patch in the VMCALL instruction:
	 */
	hypercall[0] = 0x0f;
	hypercall[1] = 0x01;
	hypercall[2] = 0xc1;
}

A
Avi Kivity 已提交
2378 2379
static int handle_cr(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
2380
	unsigned long exit_qualification;
A
Avi Kivity 已提交
2381 2382 2383
	int cr;
	int reg;

2384
	exit_qualification = vmcs_readl(EXIT_QUALIFICATION);
A
Avi Kivity 已提交
2385 2386 2387 2388
	cr = exit_qualification & 15;
	reg = (exit_qualification >> 8) & 15;
	switch ((exit_qualification >> 4) & 3) {
	case 0: /* mov to cr */
F
Feng (Eric) Liu 已提交
2389 2390
		KVMTRACE_3D(CR_WRITE, vcpu, (u32)cr, (u32)vcpu->arch.regs[reg],
			    (u32)((u64)vcpu->arch.regs[reg] >> 32), handler);
A
Avi Kivity 已提交
2391 2392 2393
		switch (cr) {
		case 0:
			vcpu_load_rsp_rip(vcpu);
2394
			kvm_set_cr0(vcpu, vcpu->arch.regs[reg]);
A
Avi Kivity 已提交
2395 2396 2397 2398
			skip_emulated_instruction(vcpu);
			return 1;
		case 3:
			vcpu_load_rsp_rip(vcpu);
2399
			kvm_set_cr3(vcpu, vcpu->arch.regs[reg]);
A
Avi Kivity 已提交
2400 2401 2402 2403
			skip_emulated_instruction(vcpu);
			return 1;
		case 4:
			vcpu_load_rsp_rip(vcpu);
2404
			kvm_set_cr4(vcpu, vcpu->arch.regs[reg]);
A
Avi Kivity 已提交
2405 2406 2407 2408
			skip_emulated_instruction(vcpu);
			return 1;
		case 8:
			vcpu_load_rsp_rip(vcpu);
2409
			kvm_set_cr8(vcpu, vcpu->arch.regs[reg]);
A
Avi Kivity 已提交
2410
			skip_emulated_instruction(vcpu);
2411 2412
			if (irqchip_in_kernel(vcpu->kvm))
				return 1;
2413 2414
			kvm_run->exit_reason = KVM_EXIT_SET_TPR;
			return 0;
A
Avi Kivity 已提交
2415 2416
		};
		break;
2417 2418
	case 2: /* clts */
		vcpu_load_rsp_rip(vcpu);
2419
		vmx_fpu_deactivate(vcpu);
2420 2421
		vcpu->arch.cr0 &= ~X86_CR0_TS;
		vmcs_writel(CR0_READ_SHADOW, vcpu->arch.cr0);
2422
		vmx_fpu_activate(vcpu);
F
Feng (Eric) Liu 已提交
2423
		KVMTRACE_0D(CLTS, vcpu, handler);
2424 2425
		skip_emulated_instruction(vcpu);
		return 1;
A
Avi Kivity 已提交
2426 2427 2428 2429
	case 1: /*mov from cr*/
		switch (cr) {
		case 3:
			vcpu_load_rsp_rip(vcpu);
2430
			vcpu->arch.regs[reg] = vcpu->arch.cr3;
A
Avi Kivity 已提交
2431
			vcpu_put_rsp_rip(vcpu);
F
Feng (Eric) Liu 已提交
2432 2433 2434 2435
			KVMTRACE_3D(CR_READ, vcpu, (u32)cr,
				    (u32)vcpu->arch.regs[reg],
				    (u32)((u64)vcpu->arch.regs[reg] >> 32),
				    handler);
A
Avi Kivity 已提交
2436 2437 2438 2439
			skip_emulated_instruction(vcpu);
			return 1;
		case 8:
			vcpu_load_rsp_rip(vcpu);
2440
			vcpu->arch.regs[reg] = kvm_get_cr8(vcpu);
A
Avi Kivity 已提交
2441
			vcpu_put_rsp_rip(vcpu);
F
Feng (Eric) Liu 已提交
2442 2443
			KVMTRACE_2D(CR_READ, vcpu, (u32)cr,
				    (u32)vcpu->arch.regs[reg], handler);
A
Avi Kivity 已提交
2444 2445 2446 2447 2448
			skip_emulated_instruction(vcpu);
			return 1;
		}
		break;
	case 3: /* lmsw */
2449
		kvm_lmsw(vcpu, (exit_qualification >> LMSW_SOURCE_DATA_SHIFT) & 0x0f);
A
Avi Kivity 已提交
2450 2451 2452 2453 2454 2455 2456

		skip_emulated_instruction(vcpu);
		return 1;
	default:
		break;
	}
	kvm_run->exit_reason = 0;
2457
	pr_unimpl(vcpu, "unhandled control register: op %d cr %d\n",
A
Avi Kivity 已提交
2458 2459 2460 2461 2462 2463
	       (int)(exit_qualification >> 4) & 3, cr);
	return 0;
}

static int handle_dr(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
2464
	unsigned long exit_qualification;
A
Avi Kivity 已提交
2465 2466 2467 2468 2469 2470 2471
	unsigned long val;
	int dr, reg;

	/*
	 * FIXME: this code assumes the host is debugging the guest.
	 *        need to deal with guest debugging itself too.
	 */
2472
	exit_qualification = vmcs_readl(EXIT_QUALIFICATION);
A
Avi Kivity 已提交
2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487
	dr = exit_qualification & 7;
	reg = (exit_qualification >> 8) & 15;
	vcpu_load_rsp_rip(vcpu);
	if (exit_qualification & 16) {
		/* mov from dr */
		switch (dr) {
		case 6:
			val = 0xffff0ff0;
			break;
		case 7:
			val = 0x400;
			break;
		default:
			val = 0;
		}
2488
		vcpu->arch.regs[reg] = val;
F
Feng (Eric) Liu 已提交
2489
		KVMTRACE_2D(DR_READ, vcpu, (u32)dr, (u32)val, handler);
A
Avi Kivity 已提交
2490 2491 2492 2493 2494 2495 2496 2497 2498 2499
	} else {
		/* mov to dr */
	}
	vcpu_put_rsp_rip(vcpu);
	skip_emulated_instruction(vcpu);
	return 1;
}

static int handle_cpuid(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
2500 2501
	kvm_emulate_cpuid(vcpu);
	return 1;
A
Avi Kivity 已提交
2502 2503 2504 2505
}

static int handle_rdmsr(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
2506
	u32 ecx = vcpu->arch.regs[VCPU_REGS_RCX];
A
Avi Kivity 已提交
2507 2508 2509
	u64 data;

	if (vmx_get_msr(vcpu, ecx, &data)) {
2510
		kvm_inject_gp(vcpu, 0);
A
Avi Kivity 已提交
2511 2512 2513
		return 1;
	}

F
Feng (Eric) Liu 已提交
2514 2515 2516
	KVMTRACE_3D(MSR_READ, vcpu, ecx, (u32)data, (u32)(data >> 32),
		    handler);

A
Avi Kivity 已提交
2517
	/* FIXME: handling of bits 32:63 of rax, rdx */
2518 2519
	vcpu->arch.regs[VCPU_REGS_RAX] = data & -1u;
	vcpu->arch.regs[VCPU_REGS_RDX] = (data >> 32) & -1u;
A
Avi Kivity 已提交
2520 2521 2522 2523 2524 2525
	skip_emulated_instruction(vcpu);
	return 1;
}

static int handle_wrmsr(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
2526 2527 2528
	u32 ecx = vcpu->arch.regs[VCPU_REGS_RCX];
	u64 data = (vcpu->arch.regs[VCPU_REGS_RAX] & -1u)
		| ((u64)(vcpu->arch.regs[VCPU_REGS_RDX] & -1u) << 32);
A
Avi Kivity 已提交
2529

F
Feng (Eric) Liu 已提交
2530 2531 2532
	KVMTRACE_3D(MSR_WRITE, vcpu, ecx, (u32)data, (u32)(data >> 32),
		    handler);

A
Avi Kivity 已提交
2533
	if (vmx_set_msr(vcpu, ecx, data) != 0) {
2534
		kvm_inject_gp(vcpu, 0);
A
Avi Kivity 已提交
2535 2536 2537 2538 2539 2540 2541
		return 1;
	}

	skip_emulated_instruction(vcpu);
	return 1;
}

2542 2543 2544 2545 2546 2547
static int handle_tpr_below_threshold(struct kvm_vcpu *vcpu,
				      struct kvm_run *kvm_run)
{
	return 1;
}

A
Avi Kivity 已提交
2548 2549 2550
static int handle_interrupt_window(struct kvm_vcpu *vcpu,
				   struct kvm_run *kvm_run)
{
2551 2552 2553 2554 2555 2556
	u32 cpu_based_vm_exec_control;

	/* clear pending irq */
	cpu_based_vm_exec_control = vmcs_read32(CPU_BASED_VM_EXEC_CONTROL);
	cpu_based_vm_exec_control &= ~CPU_BASED_VIRTUAL_INTR_PENDING;
	vmcs_write32(CPU_BASED_VM_EXEC_CONTROL, cpu_based_vm_exec_control);
F
Feng (Eric) Liu 已提交
2557 2558 2559

	KVMTRACE_0D(PEND_INTR, vcpu, handler);

2560 2561 2562 2563 2564
	/*
	 * If the user space waits to inject interrupts, exit as soon as
	 * possible
	 */
	if (kvm_run->request_interrupt_window &&
2565
	    !vcpu->arch.irq_summary) {
2566
		kvm_run->exit_reason = KVM_EXIT_IRQ_WINDOW_OPEN;
A
Avi Kivity 已提交
2567
		++vcpu->stat.irq_window_exits;
2568 2569
		return 0;
	}
A
Avi Kivity 已提交
2570 2571 2572 2573 2574 2575
	return 1;
}

static int handle_halt(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
	skip_emulated_instruction(vcpu);
2576
	return kvm_emulate_halt(vcpu);
A
Avi Kivity 已提交
2577 2578
}

2579 2580
static int handle_vmcall(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
2581
	skip_emulated_instruction(vcpu);
2582 2583
	kvm_emulate_hypercall(vcpu);
	return 1;
2584 2585
}

E
Eddie Dong 已提交
2586 2587 2588 2589 2590 2591 2592
static int handle_wbinvd(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
	skip_emulated_instruction(vcpu);
	/* TODO: Add support for VT-d/pass-through device */
	return 1;
}

2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612
static int handle_apic_access(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
	u64 exit_qualification;
	enum emulation_result er;
	unsigned long offset;

	exit_qualification = vmcs_read64(EXIT_QUALIFICATION);
	offset = exit_qualification & 0xffful;

	er = emulate_instruction(vcpu, kvm_run, 0, 0, 0);

	if (er !=  EMULATE_DONE) {
		printk(KERN_ERR
		       "Fail to handle apic access vmexit! Offset is 0x%lx\n",
		       offset);
		return -ENOTSUPP;
	}
	return 1;
}

2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626
static int handle_task_switch(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
	unsigned long exit_qualification;
	u16 tss_selector;
	int reason;

	exit_qualification = vmcs_readl(EXIT_QUALIFICATION);

	reason = (u32)exit_qualification >> 30;
	tss_selector = exit_qualification;

	return kvm_task_switch(vcpu, tss_selector, reason);
}

2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684
static int handle_ept_violation(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
	u64 exit_qualification;
	enum emulation_result er;
	gpa_t gpa;
	unsigned long hva;
	int gla_validity;
	int r;

	exit_qualification = vmcs_read64(EXIT_QUALIFICATION);

	if (exit_qualification & (1 << 6)) {
		printk(KERN_ERR "EPT: GPA exceeds GAW!\n");
		return -ENOTSUPP;
	}

	gla_validity = (exit_qualification >> 7) & 0x3;
	if (gla_validity != 0x3 && gla_validity != 0x1 && gla_validity != 0) {
		printk(KERN_ERR "EPT: Handling EPT violation failed!\n");
		printk(KERN_ERR "EPT: GPA: 0x%lx, GVA: 0x%lx\n",
			(long unsigned int)vmcs_read64(GUEST_PHYSICAL_ADDRESS),
			(long unsigned int)vmcs_read64(GUEST_LINEAR_ADDRESS));
		printk(KERN_ERR "EPT: Exit qualification is 0x%lx\n",
			(long unsigned int)exit_qualification);
		kvm_run->exit_reason = KVM_EXIT_UNKNOWN;
		kvm_run->hw.hardware_exit_reason = 0;
		return -ENOTSUPP;
	}

	gpa = vmcs_read64(GUEST_PHYSICAL_ADDRESS);
	hva = gfn_to_hva(vcpu->kvm, gpa >> PAGE_SHIFT);
	if (!kvm_is_error_hva(hva)) {
		r = kvm_mmu_page_fault(vcpu, gpa & PAGE_MASK, 0);
		if (r < 0) {
			printk(KERN_ERR "EPT: Not enough memory!\n");
			return -ENOMEM;
		}
		return 1;
	} else {
		/* must be MMIO */
		er = emulate_instruction(vcpu, kvm_run, 0, 0, 0);

		if (er == EMULATE_FAIL) {
			printk(KERN_ERR
			 "EPT: Fail to handle EPT violation vmexit!er is %d\n",
			 er);
			printk(KERN_ERR "EPT: GPA: 0x%lx, GVA: 0x%lx\n",
			 (long unsigned int)vmcs_read64(GUEST_PHYSICAL_ADDRESS),
			 (long unsigned int)vmcs_read64(GUEST_LINEAR_ADDRESS));
			printk(KERN_ERR "EPT: Exit qualification is 0x%lx\n",
				(long unsigned int)exit_qualification);
			return -ENOTSUPP;
		} else if (er == EMULATE_DO_MMIO)
			return 0;
	}
	return 1;
}

2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697
static int handle_nmi_window(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
	u32 cpu_based_vm_exec_control;

	/* clear pending NMI */
	cpu_based_vm_exec_control = vmcs_read32(CPU_BASED_VM_EXEC_CONTROL);
	cpu_based_vm_exec_control &= ~CPU_BASED_VIRTUAL_NMI_PENDING;
	vmcs_write32(CPU_BASED_VM_EXEC_CONTROL, cpu_based_vm_exec_control);
	++vcpu->stat.nmi_window_exits;

	return 1;
}

A
Avi Kivity 已提交
2698 2699 2700 2701 2702 2703 2704 2705 2706
/*
 * The exit handlers return 1 if the exit was handled fully and guest execution
 * may resume.  Otherwise they set the kvm_run parameter to indicate what needs
 * to be done to userspace and return 0.
 */
static int (*kvm_vmx_exit_handlers[])(struct kvm_vcpu *vcpu,
				      struct kvm_run *kvm_run) = {
	[EXIT_REASON_EXCEPTION_NMI]           = handle_exception,
	[EXIT_REASON_EXTERNAL_INTERRUPT]      = handle_external_interrupt,
2707
	[EXIT_REASON_TRIPLE_FAULT]            = handle_triple_fault,
2708
	[EXIT_REASON_NMI_WINDOW]	      = handle_nmi_window,
A
Avi Kivity 已提交
2709 2710 2711 2712 2713 2714 2715 2716
	[EXIT_REASON_IO_INSTRUCTION]          = handle_io,
	[EXIT_REASON_CR_ACCESS]               = handle_cr,
	[EXIT_REASON_DR_ACCESS]               = handle_dr,
	[EXIT_REASON_CPUID]                   = handle_cpuid,
	[EXIT_REASON_MSR_READ]                = handle_rdmsr,
	[EXIT_REASON_MSR_WRITE]               = handle_wrmsr,
	[EXIT_REASON_PENDING_INTERRUPT]       = handle_interrupt_window,
	[EXIT_REASON_HLT]                     = handle_halt,
2717
	[EXIT_REASON_VMCALL]                  = handle_vmcall,
2718 2719
	[EXIT_REASON_TPR_BELOW_THRESHOLD]     = handle_tpr_below_threshold,
	[EXIT_REASON_APIC_ACCESS]             = handle_apic_access,
E
Eddie Dong 已提交
2720
	[EXIT_REASON_WBINVD]                  = handle_wbinvd,
2721
	[EXIT_REASON_TASK_SWITCH]             = handle_task_switch,
2722
	[EXIT_REASON_EPT_VIOLATION]	      = handle_ept_violation,
A
Avi Kivity 已提交
2723 2724 2725
};

static const int kvm_vmx_max_exit_handlers =
2726
	ARRAY_SIZE(kvm_vmx_exit_handlers);
A
Avi Kivity 已提交
2727 2728 2729 2730 2731 2732 2733 2734

/*
 * The guest has exited.  See if we can fix it or if we need userspace
 * assistance.
 */
static int kvm_handle_exit(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu)
{
	u32 exit_reason = vmcs_read32(VM_EXIT_REASON);
2735
	struct vcpu_vmx *vmx = to_vmx(vcpu);
2736
	u32 vectoring_info = vmx->idt_vectoring_info;
2737

F
Feng (Eric) Liu 已提交
2738 2739 2740
	KVMTRACE_3D(VMEXIT, vcpu, exit_reason, (u32)vmcs_readl(GUEST_RIP),
		    (u32)((u64)vmcs_readl(GUEST_RIP) >> 32), entryexit);

2741 2742 2743 2744 2745 2746 2747
	/* Access CR3 don't cause VMExit in paging mode, so we need
	 * to sync with guest real CR3. */
	if (vm_need_ept() && is_paging(vcpu)) {
		vcpu->arch.cr3 = vmcs_readl(GUEST_CR3);
		ept_load_pdptrs(vcpu);
	}

2748 2749 2750 2751 2752 2753
	if (unlikely(vmx->fail)) {
		kvm_run->exit_reason = KVM_EXIT_FAIL_ENTRY;
		kvm_run->fail_entry.hardware_entry_failure_reason
			= vmcs_read32(VM_INSTRUCTION_ERROR);
		return 0;
	}
A
Avi Kivity 已提交
2754

M
Mike Day 已提交
2755
	if ((vectoring_info & VECTORING_INFO_VALID_MASK) &&
2756 2757
			(exit_reason != EXIT_REASON_EXCEPTION_NMI &&
			exit_reason != EXIT_REASON_EPT_VIOLATION))
A
Avi Kivity 已提交
2758
		printk(KERN_WARNING "%s: unexpected, valid vectoring info and "
2759
		       "exit reason is 0x%x\n", __func__, exit_reason);
A
Avi Kivity 已提交
2760 2761 2762 2763 2764 2765 2766 2767 2768 2769
	if (exit_reason < kvm_vmx_max_exit_handlers
	    && kvm_vmx_exit_handlers[exit_reason])
		return kvm_vmx_exit_handlers[exit_reason](vcpu, kvm_run);
	else {
		kvm_run->exit_reason = KVM_EXIT_UNKNOWN;
		kvm_run->hw.hardware_exit_reason = exit_reason;
	}
	return 0;
}

2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786
static void update_tpr_threshold(struct kvm_vcpu *vcpu)
{
	int max_irr, tpr;

	if (!vm_need_tpr_shadow(vcpu->kvm))
		return;

	if (!kvm_lapic_enabled(vcpu) ||
	    ((max_irr = kvm_lapic_find_highest_irr(vcpu)) == -1)) {
		vmcs_write32(TPR_THRESHOLD, 0);
		return;
	}

	tpr = (kvm_lapic_get_cr8(vcpu) & 0x0f) << 4;
	vmcs_write32(TPR_THRESHOLD, (max_irr > tpr) ? tpr >> 4 : max_irr >> 4);
}

2787 2788 2789 2790 2791 2792 2793 2794 2795
static void enable_irq_window(struct kvm_vcpu *vcpu)
{
	u32 cpu_based_vm_exec_control;

	cpu_based_vm_exec_control = vmcs_read32(CPU_BASED_VM_EXEC_CONTROL);
	cpu_based_vm_exec_control |= CPU_BASED_VIRTUAL_INTR_PENDING;
	vmcs_write32(CPU_BASED_VM_EXEC_CONTROL, cpu_based_vm_exec_control);
}

2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831
static void enable_nmi_window(struct kvm_vcpu *vcpu)
{
	u32 cpu_based_vm_exec_control;

	if (!cpu_has_virtual_nmis())
		return;

	cpu_based_vm_exec_control = vmcs_read32(CPU_BASED_VM_EXEC_CONTROL);
	cpu_based_vm_exec_control |= CPU_BASED_VIRTUAL_NMI_PENDING;
	vmcs_write32(CPU_BASED_VM_EXEC_CONTROL, cpu_based_vm_exec_control);
}

static int vmx_nmi_enabled(struct kvm_vcpu *vcpu)
{
	u32 guest_intr = vmcs_read32(GUEST_INTERRUPTIBILITY_INFO);
	return !(guest_intr & (GUEST_INTR_STATE_NMI |
			       GUEST_INTR_STATE_MOV_SS |
			       GUEST_INTR_STATE_STI));
}

static int vmx_irq_enabled(struct kvm_vcpu *vcpu)
{
	u32 guest_intr = vmcs_read32(GUEST_INTERRUPTIBILITY_INFO);
	return (!(guest_intr & (GUEST_INTR_STATE_MOV_SS |
			       GUEST_INTR_STATE_STI)) &&
		(vmcs_readl(GUEST_RFLAGS) & X86_EFLAGS_IF));
}

static void enable_intr_window(struct kvm_vcpu *vcpu)
{
	if (vcpu->arch.nmi_pending)
		enable_nmi_window(vcpu);
	else if (kvm_cpu_has_interrupt(vcpu))
		enable_irq_window(vcpu);
}

2832 2833
static void vmx_intr_assist(struct kvm_vcpu *vcpu)
{
2834
	struct vcpu_vmx *vmx = to_vmx(vcpu);
2835
	u32 idtv_info_field, intr_info_field, exit_intr_info_field;
2836
	int vector;
2837

2838 2839
	update_tpr_threshold(vcpu);

2840
	intr_info_field = vmcs_read32(VM_ENTRY_INTR_INFO_FIELD);
2841
	exit_intr_info_field = vmcs_read32(VM_EXIT_INTR_INFO);
2842
	idtv_info_field = vmx->idt_vectoring_info;
2843 2844 2845
	if (intr_info_field & INTR_INFO_VALID_MASK) {
		if (idtv_info_field & INTR_INFO_VALID_MASK) {
			/* TODO: fault when IDT_Vectoring */
2846 2847
			if (printk_ratelimit())
				printk(KERN_ERR "Fault when IDT_Vectoring\n");
2848
		}
2849
		enable_intr_window(vcpu);
2850 2851 2852
		return;
	}
	if (unlikely(idtv_info_field & INTR_INFO_VALID_MASK)) {
2853 2854
		if ((idtv_info_field & VECTORING_INFO_TYPE_MASK)
		    == INTR_TYPE_EXT_INTR
2855
		    && vcpu->arch.rmode.active) {
2856 2857 2858
			u8 vect = idtv_info_field & VECTORING_INFO_VECTOR_MASK;

			vmx_inject_irq(vcpu, vect);
2859
			enable_intr_window(vcpu);
2860 2861 2862
			return;
		}

F
Feng (Eric) Liu 已提交
2863 2864
		KVMTRACE_1D(REDELIVER_EVT, vcpu, idtv_info_field, handler);

2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877
		/*
		 * SDM 3: 25.7.1.2
		 * Clear bit "block by NMI" before VM entry if a NMI delivery
		 * faulted.
		 */
		if ((idtv_info_field & VECTORING_INFO_TYPE_MASK)
		    == INTR_TYPE_NMI_INTR && cpu_has_virtual_nmis())
			vmcs_write32(GUEST_INTERRUPTIBILITY_INFO,
				vmcs_read32(GUEST_INTERRUPTIBILITY_INFO) &
				~GUEST_INTR_STATE_NMI);

		vmcs_write32(VM_ENTRY_INTR_INFO_FIELD, idtv_info_field
				& ~INTR_INFO_RESVD_BITS_MASK);
2878 2879 2880
		vmcs_write32(VM_ENTRY_INSTRUCTION_LEN,
				vmcs_read32(VM_EXIT_INSTRUCTION_LEN));

2881
		if (unlikely(idtv_info_field & INTR_INFO_DELIVER_CODE_MASK))
2882 2883
			vmcs_write32(VM_ENTRY_EXCEPTION_ERROR_CODE,
				vmcs_read32(IDT_VECTORING_ERROR_CODE));
2884
		enable_intr_window(vcpu);
2885 2886
		return;
	}
2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906
	if (cpu_has_virtual_nmis()) {
		/*
		 * SDM 3: 25.7.1.2
		 * Re-set bit "block by NMI" before VM entry if vmexit caused by
		 * a guest IRET fault.
		 */
		if ((exit_intr_info_field & INTR_INFO_UNBLOCK_NMI) &&
		    (exit_intr_info_field & INTR_INFO_VECTOR_MASK) != 8)
			vmcs_write32(GUEST_INTERRUPTIBILITY_INFO,
				vmcs_read32(GUEST_INTERRUPTIBILITY_INFO) |
				GUEST_INTR_STATE_NMI);
		else if (vcpu->arch.nmi_pending) {
			if (vmx_nmi_enabled(vcpu))
				vmx_inject_nmi(vcpu);
			enable_intr_window(vcpu);
			return;
		}

	}
	if (!kvm_cpu_has_interrupt(vcpu))
2907
		return;
2908
	if (vmx_irq_enabled(vcpu)) {
2909 2910 2911 2912
		vector = kvm_cpu_get_interrupt(vcpu);
		vmx_inject_irq(vcpu, vector);
		kvm_timer_intr_post(vcpu, vector);
	} else
2913 2914 2915
		enable_irq_window(vcpu);
}

2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938
/*
 * Failure to inject an interrupt should give us the information
 * in IDT_VECTORING_INFO_FIELD.  However, if the failure occurs
 * when fetching the interrupt redirection bitmap in the real-mode
 * tss, this doesn't happen.  So we do it ourselves.
 */
static void fixup_rmode_irq(struct vcpu_vmx *vmx)
{
	vmx->rmode.irq.pending = 0;
	if (vmcs_readl(GUEST_RIP) + 1 != vmx->rmode.irq.rip)
		return;
	vmcs_writel(GUEST_RIP, vmx->rmode.irq.rip);
	if (vmx->idt_vectoring_info & VECTORING_INFO_VALID_MASK) {
		vmx->idt_vectoring_info &= ~VECTORING_INFO_TYPE_MASK;
		vmx->idt_vectoring_info |= INTR_TYPE_EXT_INTR;
		return;
	}
	vmx->idt_vectoring_info =
		VECTORING_INFO_VALID_MASK
		| INTR_TYPE_EXT_INTR
		| vmx->rmode.irq.vector;
}

2939
static void vmx_vcpu_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
A
Avi Kivity 已提交
2940
{
2941
	struct vcpu_vmx *vmx = to_vmx(vcpu);
2942
	u32 intr_info;
2943 2944 2945 2946 2947 2948

	/*
	 * Loading guest fpu may have cleared host cr0.ts
	 */
	vmcs_writel(HOST_CR0, read_cr0());

M
Mike Day 已提交
2949
	asm(
A
Avi Kivity 已提交
2950
		/* Store host registers */
2951
#ifdef CONFIG_X86_64
2952
		"push %%rdx; push %%rbp;"
A
Avi Kivity 已提交
2953 2954
		"push %%rcx \n\t"
#else
2955 2956
		"push %%edx; push %%ebp;"
		"push %%ecx \n\t"
A
Avi Kivity 已提交
2957
#endif
2958
		__ex(ASM_VMX_VMWRITE_RSP_RDX) "\n\t"
A
Avi Kivity 已提交
2959
		/* Check if vmlaunch of vmresume is needed */
2960
		"cmpl $0, %c[launched](%0) \n\t"
A
Avi Kivity 已提交
2961
		/* Load guest registers.  Don't clobber flags. */
2962
#ifdef CONFIG_X86_64
2963
		"mov %c[cr2](%0), %%rax \n\t"
A
Avi Kivity 已提交
2964
		"mov %%rax, %%cr2 \n\t"
2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979
		"mov %c[rax](%0), %%rax \n\t"
		"mov %c[rbx](%0), %%rbx \n\t"
		"mov %c[rdx](%0), %%rdx \n\t"
		"mov %c[rsi](%0), %%rsi \n\t"
		"mov %c[rdi](%0), %%rdi \n\t"
		"mov %c[rbp](%0), %%rbp \n\t"
		"mov %c[r8](%0),  %%r8  \n\t"
		"mov %c[r9](%0),  %%r9  \n\t"
		"mov %c[r10](%0), %%r10 \n\t"
		"mov %c[r11](%0), %%r11 \n\t"
		"mov %c[r12](%0), %%r12 \n\t"
		"mov %c[r13](%0), %%r13 \n\t"
		"mov %c[r14](%0), %%r14 \n\t"
		"mov %c[r15](%0), %%r15 \n\t"
		"mov %c[rcx](%0), %%rcx \n\t" /* kills %0 (rcx) */
A
Avi Kivity 已提交
2980
#else
2981
		"mov %c[cr2](%0), %%eax \n\t"
A
Avi Kivity 已提交
2982
		"mov %%eax,   %%cr2 \n\t"
2983 2984 2985 2986 2987 2988 2989
		"mov %c[rax](%0), %%eax \n\t"
		"mov %c[rbx](%0), %%ebx \n\t"
		"mov %c[rdx](%0), %%edx \n\t"
		"mov %c[rsi](%0), %%esi \n\t"
		"mov %c[rdi](%0), %%edi \n\t"
		"mov %c[rbp](%0), %%ebp \n\t"
		"mov %c[rcx](%0), %%ecx \n\t" /* kills %0 (ecx) */
A
Avi Kivity 已提交
2990 2991
#endif
		/* Enter guest mode */
2992
		"jne .Llaunched \n\t"
2993
		__ex(ASM_VMX_VMLAUNCH) "\n\t"
2994
		"jmp .Lkvm_vmx_return \n\t"
2995
		".Llaunched: " __ex(ASM_VMX_VMRESUME) "\n\t"
2996
		".Lkvm_vmx_return: "
A
Avi Kivity 已提交
2997
		/* Save guest registers, load host registers, keep flags */
2998
#ifdef CONFIG_X86_64
2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014
		"xchg %0,     (%%rsp) \n\t"
		"mov %%rax, %c[rax](%0) \n\t"
		"mov %%rbx, %c[rbx](%0) \n\t"
		"pushq (%%rsp); popq %c[rcx](%0) \n\t"
		"mov %%rdx, %c[rdx](%0) \n\t"
		"mov %%rsi, %c[rsi](%0) \n\t"
		"mov %%rdi, %c[rdi](%0) \n\t"
		"mov %%rbp, %c[rbp](%0) \n\t"
		"mov %%r8,  %c[r8](%0) \n\t"
		"mov %%r9,  %c[r9](%0) \n\t"
		"mov %%r10, %c[r10](%0) \n\t"
		"mov %%r11, %c[r11](%0) \n\t"
		"mov %%r12, %c[r12](%0) \n\t"
		"mov %%r13, %c[r13](%0) \n\t"
		"mov %%r14, %c[r14](%0) \n\t"
		"mov %%r15, %c[r15](%0) \n\t"
A
Avi Kivity 已提交
3015
		"mov %%cr2, %%rax   \n\t"
3016
		"mov %%rax, %c[cr2](%0) \n\t"
A
Avi Kivity 已提交
3017

3018
		"pop  %%rbp; pop  %%rbp; pop  %%rdx \n\t"
A
Avi Kivity 已提交
3019
#else
3020 3021 3022 3023 3024 3025 3026 3027
		"xchg %0, (%%esp) \n\t"
		"mov %%eax, %c[rax](%0) \n\t"
		"mov %%ebx, %c[rbx](%0) \n\t"
		"pushl (%%esp); popl %c[rcx](%0) \n\t"
		"mov %%edx, %c[rdx](%0) \n\t"
		"mov %%esi, %c[rsi](%0) \n\t"
		"mov %%edi, %c[rdi](%0) \n\t"
		"mov %%ebp, %c[rbp](%0) \n\t"
A
Avi Kivity 已提交
3028
		"mov %%cr2, %%eax  \n\t"
3029
		"mov %%eax, %c[cr2](%0) \n\t"
A
Avi Kivity 已提交
3030

3031
		"pop %%ebp; pop %%ebp; pop %%edx \n\t"
A
Avi Kivity 已提交
3032
#endif
3033 3034 3035 3036
		"setbe %c[fail](%0) \n\t"
	      : : "c"(vmx), "d"((unsigned long)HOST_RSP),
		[launched]"i"(offsetof(struct vcpu_vmx, launched)),
		[fail]"i"(offsetof(struct vcpu_vmx, fail)),
3037 3038 3039 3040 3041 3042 3043
		[rax]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_RAX])),
		[rbx]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_RBX])),
		[rcx]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_RCX])),
		[rdx]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_RDX])),
		[rsi]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_RSI])),
		[rdi]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_RDI])),
		[rbp]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_RBP])),
3044
#ifdef CONFIG_X86_64
3045 3046 3047 3048 3049 3050 3051 3052
		[r8]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_R8])),
		[r9]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_R9])),
		[r10]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_R10])),
		[r11]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_R11])),
		[r12]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_R12])),
		[r13]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_R13])),
		[r14]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_R14])),
		[r15]"i"(offsetof(struct vcpu_vmx, vcpu.arch.regs[VCPU_REGS_R15])),
A
Avi Kivity 已提交
3053
#endif
3054
		[cr2]"i"(offsetof(struct vcpu_vmx, vcpu.arch.cr2))
3055 3056 3057 3058
	      : "cc", "memory"
#ifdef CONFIG_X86_64
		, "rbx", "rdi", "rsi"
		, "r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15"
3059 3060
#else
		, "ebx", "edi", "rsi"
3061 3062
#endif
	      );
A
Avi Kivity 已提交
3063

3064
	vmx->idt_vectoring_info = vmcs_read32(IDT_VECTORING_INFO_FIELD);
3065 3066
	if (vmx->rmode.irq.pending)
		fixup_rmode_irq(vmx);
3067

3068
	vcpu->arch.interrupt_window_open =
3069 3070
		(vmcs_read32(GUEST_INTERRUPTIBILITY_INFO) &
		 (GUEST_INTR_STATE_STI | GUEST_INTR_STATE_MOV_SS)) == 0;
A
Avi Kivity 已提交
3071

M
Mike Day 已提交
3072
	asm("mov %0, %%ds; mov %0, %%es" : : "r"(__USER_DS));
3073
	vmx->launched = 1;
3074 3075 3076 3077

	intr_info = vmcs_read32(VM_EXIT_INTR_INFO);

	/* We need to handle NMIs before interrupts are enabled */
3078 3079
	if ((intr_info & INTR_INFO_INTR_TYPE_MASK) == 0x200 &&
	    (intr_info & INTR_INFO_VALID_MASK)) {
F
Feng (Eric) Liu 已提交
3080
		KVMTRACE_0D(NMI, vcpu, handler);
3081
		asm("int $2");
F
Feng (Eric) Liu 已提交
3082
	}
A
Avi Kivity 已提交
3083 3084 3085 3086
}

static void vmx_free_vmcs(struct kvm_vcpu *vcpu)
{
3087 3088 3089
	struct vcpu_vmx *vmx = to_vmx(vcpu);

	if (vmx->vmcs) {
3090
		vcpu_clear(vmx);
3091 3092
		free_vmcs(vmx->vmcs);
		vmx->vmcs = NULL;
A
Avi Kivity 已提交
3093 3094 3095 3096 3097
	}
}

static void vmx_free_vcpu(struct kvm_vcpu *vcpu)
{
R
Rusty Russell 已提交
3098 3099
	struct vcpu_vmx *vmx = to_vmx(vcpu);

3100 3101 3102 3103
	spin_lock(&vmx_vpid_lock);
	if (vmx->vpid != 0)
		__clear_bit(vmx->vpid, vmx_vpid_bitmap);
	spin_unlock(&vmx_vpid_lock);
A
Avi Kivity 已提交
3104
	vmx_free_vmcs(vcpu);
R
Rusty Russell 已提交
3105 3106 3107
	kfree(vmx->host_msrs);
	kfree(vmx->guest_msrs);
	kvm_vcpu_uninit(vcpu);
3108
	kmem_cache_free(kvm_vcpu_cache, vmx);
A
Avi Kivity 已提交
3109 3110
}

R
Rusty Russell 已提交
3111
static struct kvm_vcpu *vmx_create_vcpu(struct kvm *kvm, unsigned int id)
A
Avi Kivity 已提交
3112
{
R
Rusty Russell 已提交
3113
	int err;
3114
	struct vcpu_vmx *vmx = kmem_cache_zalloc(kvm_vcpu_cache, GFP_KERNEL);
3115
	int cpu;
A
Avi Kivity 已提交
3116

3117
	if (!vmx)
R
Rusty Russell 已提交
3118 3119
		return ERR_PTR(-ENOMEM);

3120
	allocate_vpid(vmx);
3121 3122 3123 3124 3125 3126 3127 3128 3129
	if (id == 0 && vm_need_ept()) {
		kvm_mmu_set_base_ptes(VMX_EPT_READABLE_MASK |
			VMX_EPT_WRITABLE_MASK |
			VMX_EPT_DEFAULT_MT << VMX_EPT_MT_EPTE_SHIFT);
		kvm_mmu_set_mask_ptes(0ull, VMX_EPT_FAKE_ACCESSED_MASK,
				VMX_EPT_FAKE_DIRTY_MASK, 0ull,
				VMX_EPT_EXECUTABLE_MASK);
		kvm_enable_tdp();
	}
3130

R
Rusty Russell 已提交
3131 3132 3133
	err = kvm_vcpu_init(&vmx->vcpu, kvm, id);
	if (err)
		goto free_vcpu;
3134

3135
	vmx->guest_msrs = kmalloc(PAGE_SIZE, GFP_KERNEL);
R
Rusty Russell 已提交
3136 3137 3138 3139
	if (!vmx->guest_msrs) {
		err = -ENOMEM;
		goto uninit_vcpu;
	}
3140

3141 3142
	vmx->host_msrs = kmalloc(PAGE_SIZE, GFP_KERNEL);
	if (!vmx->host_msrs)
R
Rusty Russell 已提交
3143
		goto free_guest_msrs;
3144

3145 3146
	vmx->vmcs = alloc_vmcs();
	if (!vmx->vmcs)
R
Rusty Russell 已提交
3147
		goto free_msrs;
3148 3149 3150

	vmcs_clear(vmx->vmcs);

3151 3152
	cpu = get_cpu();
	vmx_vcpu_load(&vmx->vcpu, cpu);
R
Rusty Russell 已提交
3153
	err = vmx_vcpu_setup(vmx);
R
Rusty Russell 已提交
3154
	vmx_vcpu_put(&vmx->vcpu);
3155
	put_cpu();
R
Rusty Russell 已提交
3156 3157
	if (err)
		goto free_vmcs;
3158 3159 3160
	if (vm_need_virtualize_apic_accesses(kvm))
		if (alloc_apic_access_page(kvm) != 0)
			goto free_vmcs;
R
Rusty Russell 已提交
3161

3162 3163 3164 3165
	if (vm_need_ept())
		if (alloc_identity_pagetable(kvm) != 0)
			goto free_vmcs;

R
Rusty Russell 已提交
3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176
	return &vmx->vcpu;

free_vmcs:
	free_vmcs(vmx->vmcs);
free_msrs:
	kfree(vmx->host_msrs);
free_guest_msrs:
	kfree(vmx->guest_msrs);
uninit_vcpu:
	kvm_vcpu_uninit(&vmx->vcpu);
free_vcpu:
3177
	kmem_cache_free(kvm_vcpu_cache, vmx);
R
Rusty Russell 已提交
3178
	return ERR_PTR(err);
A
Avi Kivity 已提交
3179 3180
}

Y
Yang, Sheng 已提交
3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194
static void __init vmx_check_processor_compat(void *rtn)
{
	struct vmcs_config vmcs_conf;

	*(int *)rtn = 0;
	if (setup_vmcs_config(&vmcs_conf) < 0)
		*(int *)rtn = -EIO;
	if (memcmp(&vmcs_config, &vmcs_conf, sizeof(struct vmcs_config)) != 0) {
		printk(KERN_ERR "kvm: CPU %d feature inconsistency!\n",
				smp_processor_id());
		*(int *)rtn = -EIO;
	}
}

3195 3196 3197 3198 3199
static int get_ept_level(void)
{
	return VMX_EPT_DEFAULT_GAW + 1;
}

3200
static struct kvm_x86_ops vmx_x86_ops = {
A
Avi Kivity 已提交
3201 3202 3203 3204
	.cpu_has_kvm_support = cpu_has_kvm_support,
	.disabled_by_bios = vmx_disabled_by_bios,
	.hardware_setup = hardware_setup,
	.hardware_unsetup = hardware_unsetup,
Y
Yang, Sheng 已提交
3205
	.check_processor_compatibility = vmx_check_processor_compat,
A
Avi Kivity 已提交
3206 3207
	.hardware_enable = hardware_enable,
	.hardware_disable = hardware_disable,
3208
	.cpu_has_accelerated_tpr = cpu_has_vmx_virtualize_apic_accesses,
A
Avi Kivity 已提交
3209 3210 3211

	.vcpu_create = vmx_create_vcpu,
	.vcpu_free = vmx_free_vcpu,
3212
	.vcpu_reset = vmx_vcpu_reset,
A
Avi Kivity 已提交
3213

3214
	.prepare_guest_switch = vmx_save_host_state,
A
Avi Kivity 已提交
3215 3216 3217 3218
	.vcpu_load = vmx_vcpu_load,
	.vcpu_put = vmx_vcpu_put,

	.set_guest_debug = set_guest_debug,
3219
	.guest_debug_pre = kvm_guest_debug_pre,
A
Avi Kivity 已提交
3220 3221 3222 3223 3224
	.get_msr = vmx_get_msr,
	.set_msr = vmx_set_msr,
	.get_segment_base = vmx_get_segment_base,
	.get_segment = vmx_get_segment,
	.set_segment = vmx_set_segment,
3225
	.get_cpl = vmx_get_cpl,
A
Avi Kivity 已提交
3226
	.get_cs_db_l_bits = vmx_get_cs_db_l_bits,
3227
	.decache_cr4_guest_bits = vmx_decache_cr4_guest_bits,
A
Avi Kivity 已提交
3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243
	.set_cr0 = vmx_set_cr0,
	.set_cr3 = vmx_set_cr3,
	.set_cr4 = vmx_set_cr4,
	.set_efer = vmx_set_efer,
	.get_idt = vmx_get_idt,
	.set_idt = vmx_set_idt,
	.get_gdt = vmx_get_gdt,
	.set_gdt = vmx_set_gdt,
	.cache_regs = vcpu_load_rsp_rip,
	.decache_regs = vcpu_put_rsp_rip,
	.get_rflags = vmx_get_rflags,
	.set_rflags = vmx_set_rflags,

	.tlb_flush = vmx_flush_tlb,

	.run = vmx_vcpu_run,
3244
	.handle_exit = kvm_handle_exit,
A
Avi Kivity 已提交
3245
	.skip_emulated_instruction = skip_emulated_instruction,
I
Ingo Molnar 已提交
3246
	.patch_hypercall = vmx_patch_hypercall,
E
Eddie Dong 已提交
3247 3248
	.get_irq = vmx_get_irq,
	.set_irq = vmx_inject_irq,
3249 3250
	.queue_exception = vmx_queue_exception,
	.exception_injected = vmx_exception_injected,
3251 3252
	.inject_pending_irq = vmx_intr_assist,
	.inject_pending_vectors = do_interrupt_requests,
3253 3254

	.set_tss_addr = vmx_set_tss_addr,
3255
	.get_tdp_level = get_ept_level,
A
Avi Kivity 已提交
3256 3257 3258 3259
};

static int __init vmx_init(void)
{
S
Sheng Yang 已提交
3260
	void *va;
3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272
	int r;

	vmx_io_bitmap_a = alloc_page(GFP_KERNEL | __GFP_HIGHMEM);
	if (!vmx_io_bitmap_a)
		return -ENOMEM;

	vmx_io_bitmap_b = alloc_page(GFP_KERNEL | __GFP_HIGHMEM);
	if (!vmx_io_bitmap_b) {
		r = -ENOMEM;
		goto out;
	}

S
Sheng Yang 已提交
3273 3274 3275 3276 3277 3278
	vmx_msr_bitmap = alloc_page(GFP_KERNEL | __GFP_HIGHMEM);
	if (!vmx_msr_bitmap) {
		r = -ENOMEM;
		goto out1;
	}

3279 3280 3281 3282
	/*
	 * Allow direct access to the PC debug port (it is often used for I/O
	 * delays, but the vmexits simply slow things down).
	 */
S
Sheng Yang 已提交
3283 3284 3285
	va = kmap(vmx_io_bitmap_a);
	memset(va, 0xff, PAGE_SIZE);
	clear_bit(0x80, va);
3286
	kunmap(vmx_io_bitmap_a);
3287

S
Sheng Yang 已提交
3288 3289
	va = kmap(vmx_io_bitmap_b);
	memset(va, 0xff, PAGE_SIZE);
3290
	kunmap(vmx_io_bitmap_b);
3291

S
Sheng Yang 已提交
3292 3293 3294 3295
	va = kmap(vmx_msr_bitmap);
	memset(va, 0xff, PAGE_SIZE);
	kunmap(vmx_msr_bitmap);

3296 3297
	set_bit(0, vmx_vpid_bitmap); /* 0 is reserved for host */

3298
	r = kvm_init(&vmx_x86_ops, sizeof(struct vcpu_vmx), THIS_MODULE);
3299
	if (r)
S
Sheng Yang 已提交
3300 3301 3302 3303 3304 3305 3306
		goto out2;

	vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_FS_BASE);
	vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_GS_BASE);
	vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_CS);
	vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_ESP);
	vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_EIP);
3307

3308
	if (vm_need_ept())
3309 3310
		bypass_guest_pf = 0;

3311 3312 3313
	if (bypass_guest_pf)
		kvm_mmu_set_nonpresent_ptes(~0xffeull, 0ull);

3314 3315
	ept_sync_global();

3316 3317
	return 0;

S
Sheng Yang 已提交
3318 3319
out2:
	__free_page(vmx_msr_bitmap);
3320 3321 3322 3323 3324
out1:
	__free_page(vmx_io_bitmap_b);
out:
	__free_page(vmx_io_bitmap_a);
	return r;
A
Avi Kivity 已提交
3325 3326 3327 3328
}

static void __exit vmx_exit(void)
{
S
Sheng Yang 已提交
3329
	__free_page(vmx_msr_bitmap);
3330 3331 3332
	__free_page(vmx_io_bitmap_b);
	__free_page(vmx_io_bitmap_a);

3333
	kvm_exit();
A
Avi Kivity 已提交
3334 3335 3336 3337
}

module_init(vmx_init)
module_exit(vmx_exit)