svm.c 192.7 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0-only
A
Avi Kivity 已提交
2 3 4 5 6 7
/*
 * Kernel-based Virtual Machine driver for Linux
 *
 * AMD SVM support
 *
 * Copyright (C) 2006 Qumranet, Inc.
N
Nicolas Kaiser 已提交
8
 * Copyright 2010 Red Hat, Inc. and/or its affiliates.
A
Avi Kivity 已提交
9 10 11 12 13
 *
 * Authors:
 *   Yaniv Kamay  <yaniv@qumranet.com>
 *   Avi Kivity   <avi@qumranet.com>
 */
14 15 16

#define pr_fmt(fmt) "SVM: " fmt

17 18
#include <linux/kvm_host.h>

19
#include "irq.h"
20
#include "mmu.h"
21
#include "kvm_cache_regs.h"
22
#include "x86.h"
23
#include "cpuid.h"
24
#include "pmu.h"
A
Avi Kivity 已提交
25

A
Avi Kivity 已提交
26
#include <linux/module.h>
27
#include <linux/mod_devicetable.h>
28
#include <linux/kernel.h>
A
Avi Kivity 已提交
29 30
#include <linux/vmalloc.h>
#include <linux/highmem.h>
A
Alexey Dobriyan 已提交
31
#include <linux/sched.h>
32
#include <linux/trace_events.h>
33
#include <linux/slab.h>
34 35
#include <linux/amd-iommu.h>
#include <linux/hashtable.h>
36
#include <linux/frame.h>
B
Brijesh Singh 已提交
37
#include <linux/psp-sev.h>
B
Brijesh Singh 已提交
38
#include <linux/file.h>
39 40
#include <linux/pagemap.h>
#include <linux/swap.h>
41
#include <linux/rwsem.h>
A
Avi Kivity 已提交
42

43
#include <asm/apic.h>
44
#include <asm/perf_event.h>
45
#include <asm/tlbflush.h>
A
Avi Kivity 已提交
46
#include <asm/desc.h>
47
#include <asm/debugreg.h>
G
Gleb Natapov 已提交
48
#include <asm/kvm_para.h>
49
#include <asm/irq_remapping.h>
50
#include <asm/spec-ctrl.h>
A
Avi Kivity 已提交
51

52
#include <asm/virtext.h>
53
#include "trace.h"
54

55 56
#define __ex(x) __kvm_handle_fault_on_reboot(x)

A
Avi Kivity 已提交
57 58 59
MODULE_AUTHOR("Qumranet");
MODULE_LICENSE("GPL");

60 61 62 63 64 65
static const struct x86_cpu_id svm_cpu_id[] = {
	X86_FEATURE_MATCH(X86_FEATURE_SVM),
	{}
};
MODULE_DEVICE_TABLE(x86cpu, svm_cpu_id);

A
Avi Kivity 已提交
66 67 68 69 70 71
#define IOPM_ALLOC_ORDER 2
#define MSRPM_ALLOC_ORDER 1

#define SEG_TYPE_LDT 2
#define SEG_TYPE_BUSY_TSS16 3

72 73
#define SVM_FEATURE_LBRV           (1 <<  1)
#define SVM_FEATURE_SVML           (1 <<  2)
74 75 76 77
#define SVM_FEATURE_TSC_RATE       (1 <<  4)
#define SVM_FEATURE_VMCB_CLEAN     (1 <<  5)
#define SVM_FEATURE_FLUSH_ASID     (1 <<  6)
#define SVM_FEATURE_DECODE_ASSIST  (1 <<  7)
78
#define SVM_FEATURE_PAUSE_FILTER   (1 << 10)
79

80 81
#define SVM_AVIC_DOORBELL	0xc001011b

82 83 84 85
#define NESTED_EXIT_HOST	0	/* Exit handled on host level */
#define NESTED_EXIT_DONE	1	/* Exit caused nested vmexit  */
#define NESTED_EXIT_CONTINUE	2	/* Further checks needed      */

86 87
#define DEBUGCTL_RESERVED_BITS (~(0x3fULL))

88
#define TSC_RATIO_RSVD          0xffffff0000000000ULL
89 90
#define TSC_RATIO_MIN		0x0000000000000001ULL
#define TSC_RATIO_MAX		0x000000ffffffffffULL
91

D
Dan Carpenter 已提交
92
#define AVIC_HPA_MASK	~((0xFFFULL << 52) | 0xFFF)
93 94 95 96 97 98 99

/*
 * 0xff is broadcast, so the max index allowed for physical APIC ID
 * table is 0xfe.  APIC IDs above 0xff are reserved.
 */
#define AVIC_MAX_PHYSICAL_ID_COUNT	255

100 101 102 103
#define AVIC_UNACCEL_ACCESS_WRITE_MASK		1
#define AVIC_UNACCEL_ACCESS_OFFSET_MASK		0xFF0
#define AVIC_UNACCEL_ACCESS_VECTOR_MASK		0xFFFFFFFF

104 105 106 107 108 109 110 111 112 113 114 115 116
/* AVIC GATAG is encoded using VM and VCPU IDs */
#define AVIC_VCPU_ID_BITS		8
#define AVIC_VCPU_ID_MASK		((1 << AVIC_VCPU_ID_BITS) - 1)

#define AVIC_VM_ID_BITS			24
#define AVIC_VM_ID_NR			(1 << AVIC_VM_ID_BITS)
#define AVIC_VM_ID_MASK			((1 << AVIC_VM_ID_BITS) - 1)

#define AVIC_GATAG(x, y)		(((x & AVIC_VM_ID_MASK) << AVIC_VCPU_ID_BITS) | \
						(y & AVIC_VCPU_ID_MASK))
#define AVIC_GATAG_TO_VMID(x)		((x >> AVIC_VCPU_ID_BITS) & AVIC_VM_ID_MASK)
#define AVIC_GATAG_TO_VCPUID(x)		(x & AVIC_VCPU_ID_MASK)

117 118
static bool erratum_383_found __read_mostly;

A
Avi Kivity 已提交
119 120 121 122 123 124
static const u32 host_save_user_msrs[] = {
#ifdef CONFIG_X86_64
	MSR_STAR, MSR_LSTAR, MSR_CSTAR, MSR_SYSCALL_MASK, MSR_KERNEL_GS_BASE,
	MSR_FS_BASE,
#endif
	MSR_IA32_SYSENTER_CS, MSR_IA32_SYSENTER_ESP, MSR_IA32_SYSENTER_EIP,
P
Paolo Bonzini 已提交
125
	MSR_TSC_AUX,
A
Avi Kivity 已提交
126 127 128 129
};

#define NR_HOST_SAVE_USER_MSRS ARRAY_SIZE(host_save_user_msrs)

130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150
struct kvm_sev_info {
	bool active;		/* SEV enabled guest */
	unsigned int asid;	/* ASID used for this guest */
	unsigned int handle;	/* SEV firmware handle */
	int fd;			/* SEV device fd */
	unsigned long pages_locked; /* Number of pages locked */
	struct list_head regions_list;  /* List of registered regions */
};

struct kvm_svm {
	struct kvm kvm;

	/* Struct members for AVIC */
	u32 avic_vm_id;
	struct page *avic_logical_id_table_page;
	struct page *avic_physical_id_table_page;
	struct hlist_node hnode;

	struct kvm_sev_info sev_info;
};

A
Avi Kivity 已提交
151 152
struct kvm_vcpu;

153 154 155
struct nested_state {
	struct vmcb *hsave;
	u64 hsave_msr;
156
	u64 vm_cr_msr;
157 158 159 160 161 162 163
	u64 vmcb;

	/* These are the merged vectors */
	u32 *msrpm;

	/* gpa pointers to the real vectors */
	u64 vmcb_msrpm;
164
	u64 vmcb_iopm;
J
Joerg Roedel 已提交
165

166 167 168
	/* A VMEXIT is required but not yet emulated */
	bool exit_required;

J
Joerg Roedel 已提交
169
	/* cache for intercepts of the guest */
170
	u32 intercept_cr;
171
	u32 intercept_dr;
J
Joerg Roedel 已提交
172 173 174
	u32 intercept_exceptions;
	u64 intercept;

175 176
	/* Nested Paging related state */
	u64 nested_cr3;
177 178
};

179 180 181
#define MSRPM_OFFSETS	16
static u32 msrpm_offsets[MSRPM_OFFSETS] __read_mostly;

182 183 184 185 186 187
/*
 * Set osvw_len to higher value when updated Revision Guides
 * are published and we know what the new status bits are
 */
static uint64_t osvw_len = 4, osvw_status;

A
Avi Kivity 已提交
188 189 190 191 192 193 194 195
struct vcpu_svm {
	struct kvm_vcpu vcpu;
	struct vmcb *vmcb;
	unsigned long vmcb_pa;
	struct svm_cpu_data *svm_data;
	uint64_t asid_generation;
	uint64_t sysenter_esp;
	uint64_t sysenter_eip;
P
Paolo Bonzini 已提交
196
	uint64_t tsc_aux;
A
Avi Kivity 已提交
197

198 199
	u64 msr_decfg;

A
Avi Kivity 已提交
200 201 202
	u64 next_rip;

	u64 host_user_msrs[NR_HOST_SAVE_USER_MSRS];
203
	struct {
204 205 206
		u16 fs;
		u16 gs;
		u16 ldt;
207 208
		u64 gs_base;
	} host;
A
Avi Kivity 已提交
209

210
	u64 spec_ctrl;
211 212 213 214 215 216
	/*
	 * Contains guest-controlled bits of VIRT_SPEC_CTRL, which will be
	 * translated into the appropriate L2_CFG bits on the host to
	 * perform speculative control.
	 */
	u64 virt_spec_ctrl;
217

A
Avi Kivity 已提交
218 219
	u32 *msrpm;

220 221
	ulong nmi_iret_rip;

222
	struct nested_state nested;
J
Jan Kiszka 已提交
223 224

	bool nmi_singlestep;
225
	u64 nmi_singlestep_guest_rflags;
226 227 228

	unsigned int3_injected;
	unsigned long int3_rip;
229

230 231
	/* cached guest cpuid flags for faster access */
	bool nrips_enabled	: 1;
232

233
	u32 ldr_reg;
234
	u32 dfr_reg;
235 236
	struct page *avic_backing_page;
	u64 *avic_physical_id_cache;
237
	bool avic_is_running;
238 239 240 241 242 243 244 245 246

	/*
	 * Per-vcpu list of struct amd_svm_iommu_ir:
	 * This is used mainly to store interrupt remapping information used
	 * when update the vcpu affinity. This avoids the need to scan for
	 * IRTE and try to match ga_tag in the IOMMU driver.
	 */
	struct list_head ir_list;
	spinlock_t ir_list_lock;
247 248 249

	/* which host CPU was used for running this vcpu */
	unsigned int last_cpu;
250 251 252 253 254 255 256 257
};

/*
 * This is a wrapper of struct amd_iommu_ir_data.
 */
struct amd_svm_iommu_ir {
	struct list_head node;	/* Used by SVM for per-vcpu ir_list */
	void *data;		/* Storing pointer to struct amd_ir_data */
A
Avi Kivity 已提交
258 259
};

260
#define AVIC_LOGICAL_ID_ENTRY_GUEST_PHYSICAL_ID_MASK	(0xFF)
261
#define AVIC_LOGICAL_ID_ENTRY_VALID_BIT			31
262 263 264 265 266 267 268
#define AVIC_LOGICAL_ID_ENTRY_VALID_MASK		(1 << 31)

#define AVIC_PHYSICAL_ID_ENTRY_HOST_PHYSICAL_ID_MASK	(0xFFULL)
#define AVIC_PHYSICAL_ID_ENTRY_BACKING_PAGE_MASK	(0xFFFFFFFFFFULL << 12)
#define AVIC_PHYSICAL_ID_ENTRY_IS_RUNNING_MASK		(1ULL << 62)
#define AVIC_PHYSICAL_ID_ENTRY_VALID_MASK		(1ULL << 63)

269 270 271
static DEFINE_PER_CPU(u64, current_tsc_ratio);
#define TSC_RATIO_DEFAULT	0x0100000000ULL

272 273
#define MSR_INVALID			0xffffffffU

274
static const struct svm_direct_access_msrs {
275 276 277
	u32 index;   /* Index of the MSR */
	bool always; /* True if intercept is always on */
} direct_access_msrs[] = {
B
Brian Gerst 已提交
278
	{ .index = MSR_STAR,				.always = true  },
279 280 281 282 283 284 285 286 287
	{ .index = MSR_IA32_SYSENTER_CS,		.always = true  },
#ifdef CONFIG_X86_64
	{ .index = MSR_GS_BASE,				.always = true  },
	{ .index = MSR_FS_BASE,				.always = true  },
	{ .index = MSR_KERNEL_GS_BASE,			.always = true  },
	{ .index = MSR_LSTAR,				.always = true  },
	{ .index = MSR_CSTAR,				.always = true  },
	{ .index = MSR_SYSCALL_MASK,			.always = true  },
#endif
288
	{ .index = MSR_IA32_SPEC_CTRL,			.always = false },
A
Ashok Raj 已提交
289
	{ .index = MSR_IA32_PRED_CMD,			.always = false },
290 291 292 293 294
	{ .index = MSR_IA32_LASTBRANCHFROMIP,		.always = false },
	{ .index = MSR_IA32_LASTBRANCHTOIP,		.always = false },
	{ .index = MSR_IA32_LASTINTFROMIP,		.always = false },
	{ .index = MSR_IA32_LASTINTTOIP,		.always = false },
	{ .index = MSR_INVALID,				.always = false },
A
Avi Kivity 已提交
295 296
};

297 298 299 300
/* enable NPT for AMD64 and X86 with PAE */
#if defined(CONFIG_X86_64) || defined(CONFIG_X86_PAE)
static bool npt_enabled = true;
#else
J
Joerg Roedel 已提交
301
static bool npt_enabled;
302
#endif
303

304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351
/*
 * These 2 parameters are used to config the controls for Pause-Loop Exiting:
 * pause_filter_count: On processors that support Pause filtering(indicated
 *	by CPUID Fn8000_000A_EDX), the VMCB provides a 16 bit pause filter
 *	count value. On VMRUN this value is loaded into an internal counter.
 *	Each time a pause instruction is executed, this counter is decremented
 *	until it reaches zero at which time a #VMEXIT is generated if pause
 *	intercept is enabled. Refer to  AMD APM Vol 2 Section 15.14.4 Pause
 *	Intercept Filtering for more details.
 *	This also indicate if ple logic enabled.
 *
 * pause_filter_thresh: In addition, some processor families support advanced
 *	pause filtering (indicated by CPUID Fn8000_000A_EDX) upper bound on
 *	the amount of time a guest is allowed to execute in a pause loop.
 *	In this mode, a 16-bit pause filter threshold field is added in the
 *	VMCB. The threshold value is a cycle count that is used to reset the
 *	pause counter. As with simple pause filtering, VMRUN loads the pause
 *	count value from VMCB into an internal counter. Then, on each pause
 *	instruction the hardware checks the elapsed number of cycles since
 *	the most recent pause instruction against the pause filter threshold.
 *	If the elapsed cycle count is greater than the pause filter threshold,
 *	then the internal pause count is reloaded from the VMCB and execution
 *	continues. If the elapsed cycle count is less than the pause filter
 *	threshold, then the internal pause count is decremented. If the count
 *	value is less than zero and PAUSE intercept is enabled, a #VMEXIT is
 *	triggered. If advanced pause filtering is supported and pause filter
 *	threshold field is set to zero, the filter will operate in the simpler,
 *	count only mode.
 */

static unsigned short pause_filter_thresh = KVM_DEFAULT_PLE_GAP;
module_param(pause_filter_thresh, ushort, 0444);

static unsigned short pause_filter_count = KVM_SVM_DEFAULT_PLE_WINDOW;
module_param(pause_filter_count, ushort, 0444);

/* Default doubles per-vcpu window every exit. */
static unsigned short pause_filter_count_grow = KVM_DEFAULT_PLE_WINDOW_GROW;
module_param(pause_filter_count_grow, ushort, 0444);

/* Default resets per-vcpu window every exit to pause_filter_count. */
static unsigned short pause_filter_count_shrink = KVM_DEFAULT_PLE_WINDOW_SHRINK;
module_param(pause_filter_count_shrink, ushort, 0444);

/* Default is to compute the maximum so we can never overflow. */
static unsigned short pause_filter_count_max = KVM_SVM_DEFAULT_PLE_WINDOW_MAX;
module_param(pause_filter_count_max, ushort, 0444);

352 353
/* allow nested paging (virtualized MMU) for all guests */
static int npt = true;
354
module_param(npt, int, S_IRUGO);
355

356 357
/* allow nested virtualization in KVM/SVM */
static int nested = true;
358 359
module_param(nested, int, S_IRUGO);

360 361
/* enable / disable AVIC */
static int avic;
362
#ifdef CONFIG_X86_LOCAL_APIC
363
module_param(avic, int, S_IRUGO);
364
#endif
365

366 367 368 369
/* enable/disable Next RIP Save */
static int nrips = true;
module_param(nrips, int, 0444);

370 371 372 373
/* enable/disable Virtual VMLOAD VMSAVE */
static int vls = true;
module_param(vls, int, 0444);

374 375 376
/* enable/disable Virtual GIF */
static int vgif = true;
module_param(vgif, int, 0444);
377

B
Brijesh Singh 已提交
378 379 380 381
/* enable/disable SEV support */
static int sev = IS_ENABLED(CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT);
module_param(sev, int, 0444);

382 383 384
static bool __read_mostly dump_invalid_vmcb = 0;
module_param(dump_invalid_vmcb, bool, 0644);

B
Brijesh Singh 已提交
385 386
static u8 rsm_ins_bytes[] = "\x0f\xaa";

387
static void svm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
388
static void svm_flush_tlb(struct kvm_vcpu *vcpu, bool invalidate_gpa);
389
static void svm_complete_interrupts(struct vcpu_svm *svm);
390
static void svm_toggle_avic_for_irq_window(struct kvm_vcpu *vcpu, bool activate);
391
static inline void avic_post_state_restore(struct kvm_vcpu *vcpu);
392

393
static int nested_svm_exit_handled(struct vcpu_svm *svm);
394
static int nested_svm_intercept(struct vcpu_svm *svm);
395 396 397 398
static int nested_svm_vmexit(struct vcpu_svm *svm);
static int nested_svm_check_exception(struct vcpu_svm *svm, unsigned nr,
				      bool has_error_code, u32 error_code);

399
enum {
400 401
	VMCB_INTERCEPTS, /* Intercept vectors, TSC offset,
			    pause filter count */
402
	VMCB_PERM_MAP,   /* IOPM Base and MSRPM Base */
403
	VMCB_ASID,	 /* ASID */
404
	VMCB_INTR,	 /* int_ctl, int_vector */
405
	VMCB_NPT,        /* npt_en, nCR3, gPAT */
406
	VMCB_CR,	 /* CR0, CR3, CR4, EFER */
407
	VMCB_DR,         /* DR6, DR7 */
408
	VMCB_DT,         /* GDT, IDT */
409
	VMCB_SEG,        /* CS, DS, SS, ES, CPL */
410
	VMCB_CR2,        /* CR2 only */
411
	VMCB_LBR,        /* DBGCTL, BR_FROM, BR_TO, LAST_EX_FROM, LAST_EX_TO */
412 413 414 415
	VMCB_AVIC,       /* AVIC APIC_BAR, AVIC APIC_BACKING_PAGE,
			  * AVIC PHYSICAL_TABLE pointer,
			  * AVIC LOGICAL_TABLE pointer
			  */
416 417 418
	VMCB_DIRTY_MAX,
};

419 420
/* TPR and CR2 are always written before VMRUN */
#define VMCB_ALWAYS_DIRTY_MASK	((1U << VMCB_INTR) | (1U << VMCB_CR2))
421

422 423
#define VMCB_AVIC_APIC_BAR_MASK		0xFFFFFFFFFF000ULL

424 425
static int sev_flush_asids(void);
static DECLARE_RWSEM(sev_deactivate_lock);
426
static DEFINE_MUTEX(sev_bitmap_lock);
427
static unsigned int max_sev_asid;
B
Brijesh Singh 已提交
428 429
static unsigned int min_sev_asid;
static unsigned long *sev_asid_bitmap;
430
static unsigned long *sev_reclaim_asid_bitmap;
431
#define __sme_page_pa(x) __sme_set(page_to_pfn(x) << PAGE_SHIFT)
B
Brijesh Singh 已提交
432

433 434 435 436 437 438 439 440
struct enc_region {
	struct list_head list;
	unsigned long npages;
	struct page **pages;
	unsigned long uaddr;
	unsigned long size;
};

441 442 443 444 445 446

static inline struct kvm_svm *to_kvm_svm(struct kvm *kvm)
{
	return container_of(kvm, struct kvm_svm, kvm);
}

B
Brijesh Singh 已提交
447 448
static inline bool svm_sev_enabled(void)
{
449
	return IS_ENABLED(CONFIG_KVM_AMD_SEV) ? max_sev_asid : 0;
B
Brijesh Singh 已提交
450 451 452 453
}

static inline bool sev_guest(struct kvm *kvm)
{
454
#ifdef CONFIG_KVM_AMD_SEV
455
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
B
Brijesh Singh 已提交
456 457

	return sev->active;
458 459 460
#else
	return false;
#endif
B
Brijesh Singh 已提交
461
}
462

463 464
static inline int sev_get_asid(struct kvm *kvm)
{
465
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
466 467 468 469

	return sev->asid;
}

470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485
static inline void mark_all_dirty(struct vmcb *vmcb)
{
	vmcb->control.clean = 0;
}

static inline void mark_all_clean(struct vmcb *vmcb)
{
	vmcb->control.clean = ((1 << VMCB_DIRTY_MAX) - 1)
			       & ~VMCB_ALWAYS_DIRTY_MASK;
}

static inline void mark_dirty(struct vmcb *vmcb, int bit)
{
	vmcb->control.clean &= ~(1 << bit);
}

486 487
static inline struct vcpu_svm *to_svm(struct kvm_vcpu *vcpu)
{
R
Rusty Russell 已提交
488
	return container_of(vcpu, struct vcpu_svm, vcpu);
489 490
}

491 492 493 494 495 496
static inline void avic_update_vapic_bar(struct vcpu_svm *svm, u64 data)
{
	svm->vmcb->control.avic_vapic_bar = data & VMCB_AVIC_APIC_BAR_MASK;
	mark_dirty(svm->vmcb, VMCB_AVIC);
}

497 498 499 500 501 502 503 504 505 506 507
static inline bool avic_vcpu_is_running(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);
	u64 *entry = svm->avic_physical_id_cache;

	if (!entry)
		return false;

	return (READ_ONCE(*entry) & AVIC_PHYSICAL_ID_ENTRY_IS_RUNNING_MASK);
}

508 509 510 511 512
static void recalc_intercepts(struct vcpu_svm *svm)
{
	struct vmcb_control_area *c, *h;
	struct nested_state *g;

513 514
	mark_dirty(svm->vmcb, VMCB_INTERCEPTS);

515 516 517 518 519 520 521
	if (!is_guest_mode(&svm->vcpu))
		return;

	c = &svm->vmcb->control;
	h = &svm->nested.hsave->control;
	g = &svm->nested;

522
	c->intercept_cr = h->intercept_cr | g->intercept_cr;
523
	c->intercept_dr = h->intercept_dr | g->intercept_dr;
524
	c->intercept_exceptions = h->intercept_exceptions | g->intercept_exceptions;
525 526 527
	c->intercept = h->intercept | g->intercept;
}

528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560
static inline struct vmcb *get_host_vmcb(struct vcpu_svm *svm)
{
	if (is_guest_mode(&svm->vcpu))
		return svm->nested.hsave;
	else
		return svm->vmcb;
}

static inline void set_cr_intercept(struct vcpu_svm *svm, int bit)
{
	struct vmcb *vmcb = get_host_vmcb(svm);

	vmcb->control.intercept_cr |= (1U << bit);

	recalc_intercepts(svm);
}

static inline void clr_cr_intercept(struct vcpu_svm *svm, int bit)
{
	struct vmcb *vmcb = get_host_vmcb(svm);

	vmcb->control.intercept_cr &= ~(1U << bit);

	recalc_intercepts(svm);
}

static inline bool is_cr_intercept(struct vcpu_svm *svm, int bit)
{
	struct vmcb *vmcb = get_host_vmcb(svm);

	return vmcb->control.intercept_cr & (1U << bit);
}

561
static inline void set_dr_intercepts(struct vcpu_svm *svm)
562 563 564
{
	struct vmcb *vmcb = get_host_vmcb(svm);

565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580
	vmcb->control.intercept_dr = (1 << INTERCEPT_DR0_READ)
		| (1 << INTERCEPT_DR1_READ)
		| (1 << INTERCEPT_DR2_READ)
		| (1 << INTERCEPT_DR3_READ)
		| (1 << INTERCEPT_DR4_READ)
		| (1 << INTERCEPT_DR5_READ)
		| (1 << INTERCEPT_DR6_READ)
		| (1 << INTERCEPT_DR7_READ)
		| (1 << INTERCEPT_DR0_WRITE)
		| (1 << INTERCEPT_DR1_WRITE)
		| (1 << INTERCEPT_DR2_WRITE)
		| (1 << INTERCEPT_DR3_WRITE)
		| (1 << INTERCEPT_DR4_WRITE)
		| (1 << INTERCEPT_DR5_WRITE)
		| (1 << INTERCEPT_DR6_WRITE)
		| (1 << INTERCEPT_DR7_WRITE);
581 582 583 584

	recalc_intercepts(svm);
}

585
static inline void clr_dr_intercepts(struct vcpu_svm *svm)
586 587 588
{
	struct vmcb *vmcb = get_host_vmcb(svm);

589
	vmcb->control.intercept_dr = 0;
590 591 592 593

	recalc_intercepts(svm);
}

594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611
static inline void set_exception_intercept(struct vcpu_svm *svm, int bit)
{
	struct vmcb *vmcb = get_host_vmcb(svm);

	vmcb->control.intercept_exceptions |= (1U << bit);

	recalc_intercepts(svm);
}

static inline void clr_exception_intercept(struct vcpu_svm *svm, int bit)
{
	struct vmcb *vmcb = get_host_vmcb(svm);

	vmcb->control.intercept_exceptions &= ~(1U << bit);

	recalc_intercepts(svm);
}

612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629
static inline void set_intercept(struct vcpu_svm *svm, int bit)
{
	struct vmcb *vmcb = get_host_vmcb(svm);

	vmcb->control.intercept |= (1ULL << bit);

	recalc_intercepts(svm);
}

static inline void clr_intercept(struct vcpu_svm *svm, int bit)
{
	struct vmcb *vmcb = get_host_vmcb(svm);

	vmcb->control.intercept &= ~(1ULL << bit);

	recalc_intercepts(svm);
}

630 631 632 633 634
static inline bool vgif_enabled(struct vcpu_svm *svm)
{
	return !!(svm->vmcb->control.int_ctl & V_GIF_ENABLE_MASK);
}

635 636
static inline void enable_gif(struct vcpu_svm *svm)
{
637 638 639 640
	if (vgif_enabled(svm))
		svm->vmcb->control.int_ctl |= V_GIF_MASK;
	else
		svm->vcpu.arch.hflags |= HF_GIF_MASK;
641 642 643 644
}

static inline void disable_gif(struct vcpu_svm *svm)
{
645 646 647 648
	if (vgif_enabled(svm))
		svm->vmcb->control.int_ctl &= ~V_GIF_MASK;
	else
		svm->vcpu.arch.hflags &= ~HF_GIF_MASK;
649 650 651 652
}

static inline bool gif_set(struct vcpu_svm *svm)
{
653 654 655 656
	if (vgif_enabled(svm))
		return !!(svm->vmcb->control.int_ctl & V_GIF_MASK);
	else
		return !!(svm->vcpu.arch.hflags & HF_GIF_MASK);
657 658
}

659
static unsigned long iopm_base;
A
Avi Kivity 已提交
660 661 662 663

struct kvm_ldttss_desc {
	u16 limit0;
	u16 base0;
J
Joerg Roedel 已提交
664 665
	unsigned base1:8, type:5, dpl:2, p:1;
	unsigned limit1:4, zero0:3, g:1, base2:8;
A
Avi Kivity 已提交
666 667 668 669 670 671 672
	u32 base3;
	u32 zero1;
} __attribute__((packed));

struct svm_cpu_data {
	int cpu;

A
Avi Kivity 已提交
673 674 675
	u64 asid_generation;
	u32 max_asid;
	u32 next_asid;
676
	u32 min_asid;
A
Avi Kivity 已提交
677 678 679
	struct kvm_ldttss_desc *tss_desc;

	struct page *save_area;
A
Ashok Raj 已提交
680
	struct vmcb *current_vmcb;
681 682 683

	/* index = sev_asid, value = vmcb pointer */
	struct vmcb **sev_vmcbs;
A
Avi Kivity 已提交
684 685 686 687
};

static DEFINE_PER_CPU(struct svm_cpu_data *, svm_data);

688
static const u32 msrpm_ranges[] = {0, 0xc0000000, 0xc0010000};
A
Avi Kivity 已提交
689

690
#define NUM_MSR_MAPS ARRAY_SIZE(msrpm_ranges)
A
Avi Kivity 已提交
691 692 693
#define MSRS_RANGE_SIZE 2048
#define MSRS_IN_RANGE (MSRS_RANGE_SIZE * 8 / 2)

694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714
static u32 svm_msrpm_offset(u32 msr)
{
	u32 offset;
	int i;

	for (i = 0; i < NUM_MSR_MAPS; i++) {
		if (msr < msrpm_ranges[i] ||
		    msr >= msrpm_ranges[i] + MSRS_IN_RANGE)
			continue;

		offset  = (msr - msrpm_ranges[i]) / 4; /* 4 msrs per u8 */
		offset += (i * MSRS_RANGE_SIZE);       /* add range offset */

		/* Now we have the u8 offset - but need the u32 offset */
		return offset / 4;
	}

	/* MSR not in any range */
	return MSR_INVALID;
}

A
Avi Kivity 已提交
715 716 717 718
#define MAX_INST_SIZE 15

static inline void clgi(void)
{
719
	asm volatile (__ex("clgi"));
A
Avi Kivity 已提交
720 721 722 723
}

static inline void stgi(void)
{
724
	asm volatile (__ex("stgi"));
A
Avi Kivity 已提交
725 726 727 728
}

static inline void invlpga(unsigned long addr, u32 asid)
{
729
	asm volatile (__ex("invlpga %1, %0") : : "c"(asid), "a"(addr));
A
Avi Kivity 已提交
730 731
}

732
static int get_npt_level(struct kvm_vcpu *vcpu)
733 734
{
#ifdef CONFIG_X86_64
735
	return PT64_ROOT_4LEVEL;
736 737 738 739 740
#else
	return PT32E_ROOT_LEVEL;
#endif
}

A
Avi Kivity 已提交
741 742
static void svm_set_efer(struct kvm_vcpu *vcpu, u64 efer)
{
743
	vcpu->arch.efer = efer;
744 745 746 747 748 749 750 751

	if (!npt_enabled) {
		/* Shadow paging assumes NX to be available.  */
		efer |= EFER_NX;

		if (!(efer & EFER_LMA))
			efer &= ~EFER_LME;
	}
A
Avi Kivity 已提交
752

753
	to_svm(vcpu)->vmcb->save.efer = efer | EFER_SVME;
754
	mark_dirty(to_svm(vcpu)->vmcb, VMCB_CR);
A
Avi Kivity 已提交
755 756 757 758 759 760 761 762
}

static int is_external_interrupt(u32 info)
{
	info &= SVM_EVTINJ_TYPE_MASK | SVM_EVTINJ_VALID;
	return info == (SVM_EVTINJ_VALID | SVM_EVTINJ_TYPE_INTR);
}

763
static u32 svm_get_interrupt_shadow(struct kvm_vcpu *vcpu)
764 765 766 767 768
{
	struct vcpu_svm *svm = to_svm(vcpu);
	u32 ret = 0;

	if (svm->vmcb->control.int_state & SVM_INTERRUPT_SHADOW_MASK)
769 770
		ret = KVM_X86_SHADOW_INT_STI | KVM_X86_SHADOW_INT_MOV_SS;
	return ret;
771 772 773 774 775 776 777 778 779 780 781 782 783
}

static void svm_set_interrupt_shadow(struct kvm_vcpu *vcpu, int mask)
{
	struct vcpu_svm *svm = to_svm(vcpu);

	if (mask == 0)
		svm->vmcb->control.int_state &= ~SVM_INTERRUPT_SHADOW_MASK;
	else
		svm->vmcb->control.int_state |= SVM_INTERRUPT_SHADOW_MASK;

}

784
static int skip_emulated_instruction(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
785
{
786 787
	struct vcpu_svm *svm = to_svm(vcpu);

788
	if (nrips && svm->vmcb->control.next_rip != 0) {
789
		WARN_ON_ONCE(!static_cpu_has(X86_FEATURE_NRIPS));
790
		svm->next_rip = svm->vmcb->control.next_rip;
791
	}
792

793 794 795 796 797 798 799 800 801
	if (!svm->next_rip) {
		if (!kvm_emulate_instruction(vcpu, EMULTYPE_SKIP))
			return 0;
	} else {
		if (svm->next_rip - kvm_rip_read(vcpu) > MAX_INST_SIZE)
			pr_err("%s: ip 0x%lx next 0x%llx\n",
			       __func__, kvm_rip_read(vcpu), svm->next_rip);
		kvm_rip_write(vcpu, svm->next_rip);
	}
802
	svm_set_interrupt_shadow(vcpu, 0);
803

804
	return 1;
A
Avi Kivity 已提交
805 806
}

807
static void svm_queue_exception(struct kvm_vcpu *vcpu)
J
Jan Kiszka 已提交
808 809
{
	struct vcpu_svm *svm = to_svm(vcpu);
810 811
	unsigned nr = vcpu->arch.exception.nr;
	bool has_error_code = vcpu->arch.exception.has_error_code;
812
	bool reinject = vcpu->arch.exception.injected;
813
	u32 error_code = vcpu->arch.exception.error_code;
J
Jan Kiszka 已提交
814

J
Joerg Roedel 已提交
815 816 817 818
	/*
	 * If we are within a nested VM we'd better #VMEXIT and let the guest
	 * handle the exception
	 */
819 820
	if (!reinject &&
	    nested_svm_check_exception(svm, nr, has_error_code, error_code))
J
Jan Kiszka 已提交
821 822
		return;

823 824
	kvm_deliver_exception_payload(&svm->vcpu);

825
	if (nr == BP_VECTOR && !nrips) {
826 827 828 829 830 831 832 833 834
		unsigned long rip, old_rip = kvm_rip_read(&svm->vcpu);

		/*
		 * For guest debugging where we have to reinject #BP if some
		 * INT3 is guest-owned:
		 * Emulate nRIP by moving RIP forward. Will fail if injection
		 * raises a fault that is not intercepted. Still better than
		 * failing in all cases.
		 */
835
		(void)skip_emulated_instruction(&svm->vcpu);
836 837 838 839 840
		rip = kvm_rip_read(&svm->vcpu);
		svm->int3_rip = rip + svm->vmcb->save.cs.base;
		svm->int3_injected = rip - old_rip;
	}

J
Jan Kiszka 已提交
841 842 843 844 845 846 847
	svm->vmcb->control.event_inj = nr
		| SVM_EVTINJ_VALID
		| (has_error_code ? SVM_EVTINJ_VALID_ERR : 0)
		| SVM_EVTINJ_TYPE_EXEPT;
	svm->vmcb->control.event_inj_err = error_code;
}

848 849 850 851 852 853
static void svm_init_erratum_383(void)
{
	u32 low, high;
	int err;
	u64 val;

854
	if (!static_cpu_has_bug(X86_BUG_AMD_TLB_MMATCH))
855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871
		return;

	/* Use _safe variants to not break nested virtualization */
	val = native_read_msr_safe(MSR_AMD64_DC_CFG, &err);
	if (err)
		return;

	val |= (1ULL << 47);

	low  = lower_32_bits(val);
	high = upper_32_bits(val);

	native_write_msr_safe(MSR_AMD64_DC_CFG, low, high);

	erratum_383_found = true;
}

872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892
static void svm_init_osvw(struct kvm_vcpu *vcpu)
{
	/*
	 * Guests should see errata 400 and 415 as fixed (assuming that
	 * HLT and IO instructions are intercepted).
	 */
	vcpu->arch.osvw.length = (osvw_len >= 3) ? (osvw_len) : 3;
	vcpu->arch.osvw.status = osvw_status & ~(6ULL);

	/*
	 * By increasing VCPU's osvw.length to 3 we are telling the guest that
	 * all osvw.status bits inside that length, including bit 0 (which is
	 * reserved for erratum 298), are valid. However, if host processor's
	 * osvw_len is 0 then osvw_status[0] carries no information. We need to
	 * be conservative here and therefore we tell the guest that erratum 298
	 * is present (because we really don't know).
	 */
	if (osvw_len == 0 && boot_cpu_data.x86 == 0x10)
		vcpu->arch.osvw.status |= 1;
}

A
Avi Kivity 已提交
893 894
static int has_svm(void)
{
895
	const char *msg;
A
Avi Kivity 已提交
896

897
	if (!cpu_has_svm(&msg)) {
J
Joe Perches 已提交
898
		printk(KERN_INFO "has_svm: %s\n", msg);
A
Avi Kivity 已提交
899 900 901 902 903 904
		return 0;
	}

	return 1;
}

905
static void svm_hardware_disable(void)
A
Avi Kivity 已提交
906
{
907 908 909 910
	/* Make sure we clean up behind us */
	if (static_cpu_has(X86_FEATURE_TSCRATEMSR))
		wrmsrl(MSR_AMD64_TSC_RATIO, TSC_RATIO_DEFAULT);

911
	cpu_svm_disable();
912 913

	amd_pmu_disable_virt();
A
Avi Kivity 已提交
914 915
}

916
static int svm_hardware_enable(void)
A
Avi Kivity 已提交
917 918
{

919
	struct svm_cpu_data *sd;
A
Avi Kivity 已提交
920 921 922 923
	uint64_t efer;
	struct desc_struct *gdt;
	int me = raw_smp_processor_id();

924 925 926 927
	rdmsrl(MSR_EFER, efer);
	if (efer & EFER_SVME)
		return -EBUSY;

A
Avi Kivity 已提交
928
	if (!has_svm()) {
929
		pr_err("%s: err EOPNOTSUPP on %d\n", __func__, me);
930
		return -EINVAL;
A
Avi Kivity 已提交
931
	}
932 933
	sd = per_cpu(svm_data, me);
	if (!sd) {
934
		pr_err("%s: svm_data is NULL on %d\n", __func__, me);
935
		return -EINVAL;
A
Avi Kivity 已提交
936 937
	}

938 939 940
	sd->asid_generation = 1;
	sd->max_asid = cpuid_ebx(SVM_CPUID_FUNC) - 1;
	sd->next_asid = sd->max_asid + 1;
941
	sd->min_asid = max_sev_asid + 1;
A
Avi Kivity 已提交
942

943
	gdt = get_current_gdt_rw();
944
	sd->tss_desc = (struct kvm_ldttss_desc *)(gdt + GDT_ENTRY_TSS);
A
Avi Kivity 已提交
945

946
	wrmsrl(MSR_EFER, efer | EFER_SVME);
A
Avi Kivity 已提交
947

948
	wrmsrl(MSR_VM_HSAVE_PA, page_to_pfn(sd->save_area) << PAGE_SHIFT);
949

950 951
	if (static_cpu_has(X86_FEATURE_TSCRATEMSR)) {
		wrmsrl(MSR_AMD64_TSC_RATIO, TSC_RATIO_DEFAULT);
952
		__this_cpu_write(current_tsc_ratio, TSC_RATIO_DEFAULT);
953 954
	}

955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984

	/*
	 * Get OSVW bits.
	 *
	 * Note that it is possible to have a system with mixed processor
	 * revisions and therefore different OSVW bits. If bits are not the same
	 * on different processors then choose the worst case (i.e. if erratum
	 * is present on one processor and not on another then assume that the
	 * erratum is present everywhere).
	 */
	if (cpu_has(&boot_cpu_data, X86_FEATURE_OSVW)) {
		uint64_t len, status = 0;
		int err;

		len = native_read_msr_safe(MSR_AMD64_OSVW_ID_LENGTH, &err);
		if (!err)
			status = native_read_msr_safe(MSR_AMD64_OSVW_STATUS,
						      &err);

		if (err)
			osvw_status = osvw_len = 0;
		else {
			if (len < osvw_len)
				osvw_len = len;
			osvw_status |= status;
			osvw_status &= (1ULL << osvw_len) - 1;
		}
	} else
		osvw_status = osvw_len = 0;

985 986
	svm_init_erratum_383();

987 988
	amd_pmu_enable_virt();

989
	return 0;
A
Avi Kivity 已提交
990 991
}

992 993
static void svm_cpu_uninit(int cpu)
{
994
	struct svm_cpu_data *sd = per_cpu(svm_data, raw_smp_processor_id());
995

996
	if (!sd)
997 998 999
		return;

	per_cpu(svm_data, raw_smp_processor_id()) = NULL;
1000
	kfree(sd->sev_vmcbs);
1001 1002
	__free_page(sd->save_area);
	kfree(sd);
1003 1004
}

A
Avi Kivity 已提交
1005 1006
static int svm_cpu_init(int cpu)
{
1007
	struct svm_cpu_data *sd;
A
Avi Kivity 已提交
1008

1009 1010
	sd = kzalloc(sizeof(struct svm_cpu_data), GFP_KERNEL);
	if (!sd)
A
Avi Kivity 已提交
1011
		return -ENOMEM;
1012
	sd->cpu = cpu;
1013
	sd->save_area = alloc_page(GFP_KERNEL);
1014
	if (!sd->save_area)
1015
		goto free_cpu_data;
A
Avi Kivity 已提交
1016

1017
	if (svm_sev_enabled()) {
1018 1019 1020
		sd->sev_vmcbs = kmalloc_array(max_sev_asid + 1,
					      sizeof(void *),
					      GFP_KERNEL);
1021
		if (!sd->sev_vmcbs)
1022
			goto free_save_area;
1023 1024
	}

1025
	per_cpu(svm_data, cpu) = sd;
A
Avi Kivity 已提交
1026 1027 1028

	return 0;

1029 1030 1031
free_save_area:
	__free_page(sd->save_area);
free_cpu_data:
1032
	kfree(sd);
1033
	return -ENOMEM;
A
Avi Kivity 已提交
1034 1035 1036

}

1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047
static bool valid_msr_intercept(u32 index)
{
	int i;

	for (i = 0; direct_access_msrs[i].index != MSR_INVALID; i++)
		if (direct_access_msrs[i].index == index)
			return true;

	return false;
}

1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066
static bool msr_write_intercepted(struct kvm_vcpu *vcpu, unsigned msr)
{
	u8 bit_write;
	unsigned long tmp;
	u32 offset;
	u32 *msrpm;

	msrpm = is_guest_mode(vcpu) ? to_svm(vcpu)->nested.msrpm:
				      to_svm(vcpu)->msrpm;

	offset    = svm_msrpm_offset(msr);
	bit_write = 2 * (msr & 0x0f) + 1;
	tmp       = msrpm[offset];

	BUG_ON(offset == MSR_INVALID);

	return !!test_bit(bit_write,  &tmp);
}

1067 1068
static void set_msr_interception(u32 *msrpm, unsigned msr,
				 int read, int write)
A
Avi Kivity 已提交
1069
{
1070 1071 1072
	u8 bit_read, bit_write;
	unsigned long tmp;
	u32 offset;
A
Avi Kivity 已提交
1073

1074 1075 1076 1077 1078 1079
	/*
	 * If this warning triggers extend the direct_access_msrs list at the
	 * beginning of the file
	 */
	WARN_ON(!valid_msr_intercept(msr));

1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090
	offset    = svm_msrpm_offset(msr);
	bit_read  = 2 * (msr & 0x0f);
	bit_write = 2 * (msr & 0x0f) + 1;
	tmp       = msrpm[offset];

	BUG_ON(offset == MSR_INVALID);

	read  ? clear_bit(bit_read,  &tmp) : set_bit(bit_read,  &tmp);
	write ? clear_bit(bit_write, &tmp) : set_bit(bit_write, &tmp);

	msrpm[offset] = tmp;
A
Avi Kivity 已提交
1091 1092
}

1093
static void svm_vcpu_init_msrpm(u32 *msrpm)
A
Avi Kivity 已提交
1094 1095 1096
{
	int i;

1097 1098
	memset(msrpm, 0xff, PAGE_SIZE * (1 << MSRPM_ALLOC_ORDER));

1099 1100 1101 1102 1103 1104
	for (i = 0; direct_access_msrs[i].index != MSR_INVALID; i++) {
		if (!direct_access_msrs[i].always)
			continue;

		set_msr_interception(msrpm, direct_access_msrs[i].index, 1, 1);
	}
1105 1106
}

1107 1108 1109 1110 1111 1112 1113 1114
static void add_msr_offset(u32 offset)
{
	int i;

	for (i = 0; i < MSRPM_OFFSETS; ++i) {

		/* Offset already in list? */
		if (msrpm_offsets[i] == offset)
1115
			return;
1116 1117 1118 1119 1120 1121 1122 1123 1124

		/* Slot used by another offset? */
		if (msrpm_offsets[i] != MSR_INVALID)
			continue;

		/* Add offset to list */
		msrpm_offsets[i] = offset;

		return;
A
Avi Kivity 已提交
1125
	}
1126 1127 1128 1129 1130

	/*
	 * If this BUG triggers the msrpm_offsets table has an overflow. Just
	 * increase MSRPM_OFFSETS in this case.
	 */
1131
	BUG();
A
Avi Kivity 已提交
1132 1133
}

1134
static void init_msrpm_offsets(void)
1135
{
1136
	int i;
1137

1138 1139 1140 1141 1142 1143 1144 1145 1146 1147
	memset(msrpm_offsets, 0xff, sizeof(msrpm_offsets));

	for (i = 0; direct_access_msrs[i].index != MSR_INVALID; i++) {
		u32 offset;

		offset = svm_msrpm_offset(direct_access_msrs[i].index);
		BUG_ON(offset == MSR_INVALID);

		add_msr_offset(offset);
	}
1148 1149
}

1150 1151 1152 1153
static void svm_enable_lbrv(struct vcpu_svm *svm)
{
	u32 *msrpm = svm->msrpm;

1154
	svm->vmcb->control.virt_ext |= LBR_CTL_ENABLE_MASK;
1155 1156 1157 1158 1159 1160 1161 1162 1163 1164
	set_msr_interception(msrpm, MSR_IA32_LASTBRANCHFROMIP, 1, 1);
	set_msr_interception(msrpm, MSR_IA32_LASTBRANCHTOIP, 1, 1);
	set_msr_interception(msrpm, MSR_IA32_LASTINTFROMIP, 1, 1);
	set_msr_interception(msrpm, MSR_IA32_LASTINTTOIP, 1, 1);
}

static void svm_disable_lbrv(struct vcpu_svm *svm)
{
	u32 *msrpm = svm->msrpm;

1165
	svm->vmcb->control.virt_ext &= ~LBR_CTL_ENABLE_MASK;
1166 1167 1168 1169 1170 1171
	set_msr_interception(msrpm, MSR_IA32_LASTBRANCHFROMIP, 0, 0);
	set_msr_interception(msrpm, MSR_IA32_LASTBRANCHTOIP, 0, 0);
	set_msr_interception(msrpm, MSR_IA32_LASTINTFROMIP, 0, 0);
	set_msr_interception(msrpm, MSR_IA32_LASTINTTOIP, 0, 0);
}

1172 1173 1174
static void disable_nmi_singlestep(struct vcpu_svm *svm)
{
	svm->nmi_singlestep = false;
1175

1176 1177 1178 1179 1180 1181 1182
	if (!(svm->vcpu.guest_debug & KVM_GUESTDBG_SINGLESTEP)) {
		/* Clear our flags if they were not set by the guest */
		if (!(svm->nmi_singlestep_guest_rflags & X86_EFLAGS_TF))
			svm->vmcb->save.rflags &= ~X86_EFLAGS_TF;
		if (!(svm->nmi_singlestep_guest_rflags & X86_EFLAGS_RF))
			svm->vmcb->save.rflags &= ~X86_EFLAGS_RF;
	}
1183 1184
}

1185
/* Note:
1186
 * This hash table is used to map VM_ID to a struct kvm_svm,
1187 1188 1189 1190
 * when handling AMD IOMMU GALOG notification to schedule in
 * a particular vCPU.
 */
#define SVM_VM_DATA_HASH_BITS	8
1191
static DEFINE_HASHTABLE(svm_vm_data_hash, SVM_VM_DATA_HASH_BITS);
1192 1193
static u32 next_vm_id = 0;
static bool next_vm_id_wrapped = 0;
1194
static DEFINE_SPINLOCK(svm_vm_data_hash_lock);
1195 1196 1197 1198 1199 1200 1201 1202

/* Note:
 * This function is called from IOMMU driver to notify
 * SVM to schedule in a particular vCPU of a particular VM.
 */
static int avic_ga_log_notifier(u32 ga_tag)
{
	unsigned long flags;
1203
	struct kvm_svm *kvm_svm;
1204 1205 1206 1207 1208 1209 1210
	struct kvm_vcpu *vcpu = NULL;
	u32 vm_id = AVIC_GATAG_TO_VMID(ga_tag);
	u32 vcpu_id = AVIC_GATAG_TO_VCPUID(ga_tag);

	pr_debug("SVM: %s: vm_id=%#x, vcpu_id=%#x\n", __func__, vm_id, vcpu_id);

	spin_lock_irqsave(&svm_vm_data_hash_lock, flags);
1211 1212
	hash_for_each_possible(svm_vm_data_hash, kvm_svm, hnode, vm_id) {
		if (kvm_svm->avic_vm_id != vm_id)
1213
			continue;
1214
		vcpu = kvm_get_vcpu_by_id(&kvm_svm->kvm, vcpu_id);
1215 1216 1217 1218 1219 1220 1221 1222 1223
		break;
	}
	spin_unlock_irqrestore(&svm_vm_data_hash_lock, flags);

	/* Note:
	 * At this point, the IOMMU should have already set the pending
	 * bit in the vAPIC backing page. So, we just need to schedule
	 * in the vcpu.
	 */
1224
	if (vcpu)
1225 1226 1227 1228 1229
		kvm_vcpu_wake_up(vcpu);

	return 0;
}

B
Brijesh Singh 已提交
1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240
static __init int sev_hardware_setup(void)
{
	struct sev_user_data_status *status;
	int rc;

	/* Maximum number of encrypted guests supported simultaneously */
	max_sev_asid = cpuid_ecx(0x8000001F);

	if (!max_sev_asid)
		return 1;

B
Brijesh Singh 已提交
1241 1242 1243
	/* Minimum ASID value that should be used for SEV guest */
	min_sev_asid = cpuid_edx(0x8000001F);

1244
	/* Initialize SEV ASID bitmaps */
1245
	sev_asid_bitmap = bitmap_zalloc(max_sev_asid, GFP_KERNEL);
B
Brijesh Singh 已提交
1246 1247 1248
	if (!sev_asid_bitmap)
		return 1;

1249 1250 1251 1252
	sev_reclaim_asid_bitmap = bitmap_zalloc(max_sev_asid, GFP_KERNEL);
	if (!sev_reclaim_asid_bitmap)
		return 1;

B
Brijesh Singh 已提交
1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274
	status = kmalloc(sizeof(*status), GFP_KERNEL);
	if (!status)
		return 1;

	/*
	 * Check SEV platform status.
	 *
	 * PLATFORM_STATUS can be called in any state, if we failed to query
	 * the PLATFORM status then either PSP firmware does not support SEV
	 * feature or SEV firmware is dead.
	 */
	rc = sev_platform_status(status, NULL);
	if (rc)
		goto err;

	pr_info("SEV supported\n");

err:
	kfree(status);
	return rc;
}

1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285
static void grow_ple_window(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);
	struct vmcb_control_area *control = &svm->vmcb->control;
	int old = control->pause_filter_count;

	control->pause_filter_count = __grow_ple_window(old,
							pause_filter_count,
							pause_filter_count_grow,
							pause_filter_count_max);

P
Peter Xu 已提交
1286
	if (control->pause_filter_count != old) {
1287
		mark_dirty(svm->vmcb, VMCB_INTERCEPTS);
P
Peter Xu 已提交
1288 1289 1290
		trace_kvm_ple_window_update(vcpu->vcpu_id,
					    control->pause_filter_count, old);
	}
1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303
}

static void shrink_ple_window(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);
	struct vmcb_control_area *control = &svm->vmcb->control;
	int old = control->pause_filter_count;

	control->pause_filter_count =
				__shrink_ple_window(old,
						    pause_filter_count,
						    pause_filter_count_shrink,
						    pause_filter_count);
P
Peter Xu 已提交
1304
	if (control->pause_filter_count != old) {
1305
		mark_dirty(svm->vmcb, VMCB_INTERCEPTS);
P
Peter Xu 已提交
1306 1307 1308
		trace_kvm_ple_window_update(vcpu->vcpu_id,
					    control->pause_filter_count, old);
	}
1309 1310
}

1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351
/*
 * The default MMIO mask is a single bit (excluding the present bit),
 * which could conflict with the memory encryption bit. Check for
 * memory encryption support and override the default MMIO mask if
 * memory encryption is enabled.
 */
static __init void svm_adjust_mmio_mask(void)
{
	unsigned int enc_bit, mask_bit;
	u64 msr, mask;

	/* If there is no memory encryption support, use existing mask */
	if (cpuid_eax(0x80000000) < 0x8000001f)
		return;

	/* If memory encryption is not enabled, use existing mask */
	rdmsrl(MSR_K8_SYSCFG, msr);
	if (!(msr & MSR_K8_SYSCFG_MEM_ENCRYPT))
		return;

	enc_bit = cpuid_ebx(0x8000001f) & 0x3f;
	mask_bit = boot_cpu_data.x86_phys_bits;

	/* Increment the mask bit if it is the same as the encryption bit */
	if (enc_bit == mask_bit)
		mask_bit++;

	/*
	 * If the mask bit location is below 52, then some bits above the
	 * physical addressing limit will always be reserved, so use the
	 * rsvd_bits() function to generate the mask. This mask, along with
	 * the present bit, will be used to generate a page fault with
	 * PFER.RSV = 1.
	 *
	 * If the mask bit location is 52 (or above), then clear the mask.
	 */
	mask = (mask_bit < 52) ? rsvd_bits(mask_bit, 51) | PT_PRESENT_MASK : 0;

	kvm_mmu_set_mmio_spte_mask(mask, mask, PT_WRITABLE_MASK | PT_USER_MASK);
}

1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369
static void svm_hardware_teardown(void)
{
	int cpu;

	if (svm_sev_enabled()) {
		bitmap_free(sev_asid_bitmap);
		bitmap_free(sev_reclaim_asid_bitmap);

		sev_flush_asids();
	}

	for_each_possible_cpu(cpu)
		svm_cpu_uninit(cpu);

	__free_pages(pfn_to_page(iopm_base >> PAGE_SHIFT), IOPM_ALLOC_ORDER);
	iopm_base = 0;
}

A
Avi Kivity 已提交
1370 1371 1372 1373
static __init int svm_hardware_setup(void)
{
	int cpu;
	struct page *iopm_pages;
1374
	void *iopm_va;
A
Avi Kivity 已提交
1375 1376 1377 1378 1379 1380
	int r;

	iopm_pages = alloc_pages(GFP_KERNEL, IOPM_ALLOC_ORDER);

	if (!iopm_pages)
		return -ENOMEM;
1381 1382 1383

	iopm_va = page_address(iopm_pages);
	memset(iopm_va, 0xff, PAGE_SIZE * (1 << IOPM_ALLOC_ORDER));
A
Avi Kivity 已提交
1384 1385
	iopm_base = page_to_pfn(iopm_pages) << PAGE_SHIFT;

1386 1387
	init_msrpm_offsets();

1388 1389 1390
	if (boot_cpu_has(X86_FEATURE_NX))
		kvm_enable_efer_bits(EFER_NX);

A
Alexander Graf 已提交
1391 1392 1393
	if (boot_cpu_has(X86_FEATURE_FXSR_OPT))
		kvm_enable_efer_bits(EFER_FFXSR);

1394 1395
	if (boot_cpu_has(X86_FEATURE_TSCRATEMSR)) {
		kvm_has_tsc_control = true;
1396 1397
		kvm_max_tsc_scaling_ratio = TSC_RATIO_MAX;
		kvm_tsc_scaling_ratio_frac_bits = 32;
1398 1399
	}

1400 1401 1402 1403 1404 1405 1406 1407
	/* Check for pause filtering support */
	if (!boot_cpu_has(X86_FEATURE_PAUSEFILTER)) {
		pause_filter_count = 0;
		pause_filter_thresh = 0;
	} else if (!boot_cpu_has(X86_FEATURE_PFTHRESHOLD)) {
		pause_filter_thresh = 0;
	}

1408 1409
	if (nested) {
		printk(KERN_INFO "kvm: Nested Virtualization enabled\n");
1410
		kvm_enable_efer_bits(EFER_SVME | EFER_LMSLE);
1411 1412
	}

B
Brijesh Singh 已提交
1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423
	if (sev) {
		if (boot_cpu_has(X86_FEATURE_SEV) &&
		    IS_ENABLED(CONFIG_KVM_AMD_SEV)) {
			r = sev_hardware_setup();
			if (r)
				sev = false;
		} else {
			sev = false;
		}
	}

1424 1425
	svm_adjust_mmio_mask();

Z
Zachary Amsden 已提交
1426
	for_each_possible_cpu(cpu) {
A
Avi Kivity 已提交
1427 1428
		r = svm_cpu_init(cpu);
		if (r)
1429
			goto err;
A
Avi Kivity 已提交
1430
	}
1431

1432
	if (!boot_cpu_has(X86_FEATURE_NPT))
1433 1434
		npt_enabled = false;

1435 1436 1437 1438 1439
	if (npt_enabled && !npt) {
		printk(KERN_INFO "kvm: Nested Paging disabled\n");
		npt_enabled = false;
	}

1440
	if (npt_enabled) {
1441
		printk(KERN_INFO "kvm: Nested Paging enabled\n");
1442
		kvm_enable_tdp();
1443 1444
	} else
		kvm_disable_tdp();
1445

1446 1447 1448 1449 1450
	if (nrips) {
		if (!boot_cpu_has(X86_FEATURE_NRIPS))
			nrips = false;
	}

1451 1452 1453
	if (avic) {
		if (!npt_enabled ||
		    !boot_cpu_has(X86_FEATURE_AVIC) ||
1454
		    !IS_ENABLED(CONFIG_X86_LOCAL_APIC)) {
1455
			avic = false;
1456
		} else {
1457
			pr_info("AVIC enabled\n");
1458 1459 1460

			amd_iommu_register_ga_log_notifier(&avic_ga_log_notifier);
		}
1461
	}
1462

1463 1464
	if (vls) {
		if (!npt_enabled ||
1465
		    !boot_cpu_has(X86_FEATURE_V_VMSAVE_VMLOAD) ||
1466 1467 1468 1469 1470 1471 1472
		    !IS_ENABLED(CONFIG_X86_64)) {
			vls = false;
		} else {
			pr_info("Virtual VMLOAD VMSAVE supported\n");
		}
	}

1473 1474 1475 1476 1477 1478 1479
	if (vgif) {
		if (!boot_cpu_has(X86_FEATURE_VGIF))
			vgif = false;
		else
			pr_info("Virtual GIF supported\n");
	}

A
Avi Kivity 已提交
1480 1481
	return 0;

1482
err:
1483
	svm_hardware_teardown();
A
Avi Kivity 已提交
1484 1485 1486 1487 1488 1489 1490
	return r;
}

static void init_seg(struct vmcb_seg *seg)
{
	seg->selector = 0;
	seg->attrib = SVM_SELECTOR_P_MASK | SVM_SELECTOR_S_MASK |
J
Joerg Roedel 已提交
1491
		      SVM_SELECTOR_WRITE_MASK; /* Read/Write Data Segment */
A
Avi Kivity 已提交
1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503
	seg->limit = 0xffff;
	seg->base = 0;
}

static void init_sys_seg(struct vmcb_seg *seg, uint32_t type)
{
	seg->selector = 0;
	seg->attrib = SVM_SELECTOR_P_MASK | type;
	seg->limit = 0xffff;
	seg->base = 0;
}

1504 1505 1506 1507 1508 1509 1510 1511 1512 1513
static u64 svm_read_l1_tsc_offset(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);

	if (is_guest_mode(vcpu))
		return svm->nested.hsave->control.tsc_offset;

	return vcpu->arch.tsc_offset;
}

1514
static u64 svm_write_l1_tsc_offset(struct kvm_vcpu *vcpu, u64 offset)
1515 1516 1517 1518
{
	struct vcpu_svm *svm = to_svm(vcpu);
	u64 g_tsc_offset = 0;

1519
	if (is_guest_mode(vcpu)) {
1520
		/* Write L1's TSC offset.  */
1521 1522 1523
		g_tsc_offset = svm->vmcb->control.tsc_offset -
			       svm->nested.hsave->control.tsc_offset;
		svm->nested.hsave->control.tsc_offset = offset;
1524 1525 1526 1527 1528
	}

	trace_kvm_write_tsc_offset(vcpu->vcpu_id,
				   svm->vmcb->control.tsc_offset - g_tsc_offset,
				   offset);
1529 1530

	svm->vmcb->control.tsc_offset = offset + g_tsc_offset;
1531 1532

	mark_dirty(svm->vmcb, VMCB_INTERCEPTS);
1533
	return svm->vmcb->control.tsc_offset;
1534 1535
}

1536 1537 1538
static void avic_init_vmcb(struct vcpu_svm *svm)
{
	struct vmcb *vmcb = svm->vmcb;
1539
	struct kvm_svm *kvm_svm = to_kvm_svm(svm->vcpu.kvm);
1540
	phys_addr_t bpa = __sme_set(page_to_phys(svm->avic_backing_page));
1541 1542
	phys_addr_t lpa = __sme_set(page_to_phys(kvm_svm->avic_logical_id_table_page));
	phys_addr_t ppa = __sme_set(page_to_phys(kvm_svm->avic_physical_id_table_page));
1543 1544 1545 1546 1547

	vmcb->control.avic_backing_page = bpa & AVIC_HPA_MASK;
	vmcb->control.avic_logical_id = lpa & AVIC_HPA_MASK;
	vmcb->control.avic_physical_id = ppa & AVIC_HPA_MASK;
	vmcb->control.avic_physical_id |= AVIC_MAX_PHYSICAL_ID_COUNT;
1548 1549 1550 1551
	if (kvm_apicv_activated(svm->vcpu.kvm))
		vmcb->control.int_ctl |= AVIC_ENABLE_MASK;
	else
		vmcb->control.int_ctl &= ~AVIC_ENABLE_MASK;
1552 1553
}

P
Paolo Bonzini 已提交
1554
static void init_vmcb(struct vcpu_svm *svm)
A
Avi Kivity 已提交
1555
{
1556 1557
	struct vmcb_control_area *control = &svm->vmcb->control;
	struct vmcb_save_area *save = &svm->vmcb->save;
A
Avi Kivity 已提交
1558

1559
	svm->vcpu.arch.hflags = 0;
1560

1561 1562 1563 1564 1565 1566
	set_cr_intercept(svm, INTERCEPT_CR0_READ);
	set_cr_intercept(svm, INTERCEPT_CR3_READ);
	set_cr_intercept(svm, INTERCEPT_CR4_READ);
	set_cr_intercept(svm, INTERCEPT_CR0_WRITE);
	set_cr_intercept(svm, INTERCEPT_CR3_WRITE);
	set_cr_intercept(svm, INTERCEPT_CR4_WRITE);
1567 1568
	if (!kvm_vcpu_apicv_active(&svm->vcpu))
		set_cr_intercept(svm, INTERCEPT_CR8_WRITE);
A
Avi Kivity 已提交
1569

1570
	set_dr_intercepts(svm);
A
Avi Kivity 已提交
1571

1572 1573 1574
	set_exception_intercept(svm, PF_VECTOR);
	set_exception_intercept(svm, UD_VECTOR);
	set_exception_intercept(svm, MC_VECTOR);
1575
	set_exception_intercept(svm, AC_VECTOR);
1576
	set_exception_intercept(svm, DB_VECTOR);
1577 1578 1579 1580 1581 1582 1583 1584
	/*
	 * Guest access to VMware backdoor ports could legitimately
	 * trigger #GP because of TSS I/O permission bitmap.
	 * We intercept those #GP and allow access to them anyway
	 * as VMware does.
	 */
	if (enable_vmware_backdoor)
		set_exception_intercept(svm, GP_VECTOR);
A
Avi Kivity 已提交
1585

1586 1587 1588 1589
	set_intercept(svm, INTERCEPT_INTR);
	set_intercept(svm, INTERCEPT_NMI);
	set_intercept(svm, INTERCEPT_SMI);
	set_intercept(svm, INTERCEPT_SELECTIVE_CR0);
A
Avi Kivity 已提交
1590
	set_intercept(svm, INTERCEPT_RDPMC);
1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606
	set_intercept(svm, INTERCEPT_CPUID);
	set_intercept(svm, INTERCEPT_INVD);
	set_intercept(svm, INTERCEPT_INVLPG);
	set_intercept(svm, INTERCEPT_INVLPGA);
	set_intercept(svm, INTERCEPT_IOIO_PROT);
	set_intercept(svm, INTERCEPT_MSR_PROT);
	set_intercept(svm, INTERCEPT_TASK_SWITCH);
	set_intercept(svm, INTERCEPT_SHUTDOWN);
	set_intercept(svm, INTERCEPT_VMRUN);
	set_intercept(svm, INTERCEPT_VMMCALL);
	set_intercept(svm, INTERCEPT_VMLOAD);
	set_intercept(svm, INTERCEPT_VMSAVE);
	set_intercept(svm, INTERCEPT_STGI);
	set_intercept(svm, INTERCEPT_CLGI);
	set_intercept(svm, INTERCEPT_SKINIT);
	set_intercept(svm, INTERCEPT_WBINVD);
J
Joerg Roedel 已提交
1607
	set_intercept(svm, INTERCEPT_XSETBV);
J
Jim Mattson 已提交
1608
	set_intercept(svm, INTERCEPT_RDPRU);
B
Brijesh Singh 已提交
1609
	set_intercept(svm, INTERCEPT_RSM);
A
Avi Kivity 已提交
1610

1611
	if (!kvm_mwait_in_guest(svm->vcpu.kvm)) {
1612 1613 1614 1615
		set_intercept(svm, INTERCEPT_MONITOR);
		set_intercept(svm, INTERCEPT_MWAIT);
	}

1616 1617 1618
	if (!kvm_hlt_in_guest(svm->vcpu.kvm))
		set_intercept(svm, INTERCEPT_HLT);

1619 1620
	control->iopm_base_pa = __sme_set(iopm_base);
	control->msrpm_base_pa = __sme_set(__pa(svm->msrpm));
A
Avi Kivity 已提交
1621 1622 1623 1624 1625 1626 1627 1628 1629
	control->int_ctl = V_INTR_MASKING_MASK;

	init_seg(&save->es);
	init_seg(&save->ss);
	init_seg(&save->ds);
	init_seg(&save->fs);
	init_seg(&save->gs);

	save->cs.selector = 0xf000;
1630
	save->cs.base = 0xffff0000;
A
Avi Kivity 已提交
1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641
	/* Executable/Readable Code Segment */
	save->cs.attrib = SVM_SELECTOR_READ_MASK | SVM_SELECTOR_P_MASK |
		SVM_SELECTOR_S_MASK | SVM_SELECTOR_CODE_MASK;
	save->cs.limit = 0xffff;

	save->gdtr.limit = 0xffff;
	save->idtr.limit = 0xffff;

	init_sys_seg(&save->ldtr, SEG_TYPE_LDT);
	init_sys_seg(&save->tr, SEG_TYPE_BUSY_TSS16);

P
Paolo Bonzini 已提交
1642
	svm_set_efer(&svm->vcpu, 0);
M
Mike Day 已提交
1643
	save->dr6 = 0xffff0ff0;
1644
	kvm_set_rflags(&svm->vcpu, 2);
A
Avi Kivity 已提交
1645
	save->rip = 0x0000fff0;
1646
	svm->vcpu.arch.regs[VCPU_REGS_RIP] = save->rip;
A
Avi Kivity 已提交
1647

J
Joerg Roedel 已提交
1648
	/*
1649
	 * svm_set_cr0() sets PG and WP and clears NW and CD on save->cr0.
1650
	 * It also updates the guest-visible cr0 value.
A
Avi Kivity 已提交
1651
	 */
1652
	svm_set_cr0(&svm->vcpu, X86_CR0_NW | X86_CR0_CD | X86_CR0_ET);
1653
	kvm_mmu_reset_context(&svm->vcpu);
1654

1655
	save->cr4 = X86_CR4_PAE;
A
Avi Kivity 已提交
1656
	/* rdx = ?? */
1657 1658 1659

	if (npt_enabled) {
		/* Setup VMCB for Nested Paging */
1660
		control->nested_ctl |= SVM_NESTED_CTL_NP_ENABLE;
1661
		clr_intercept(svm, INTERCEPT_INVLPG);
1662
		clr_exception_intercept(svm, PF_VECTOR);
1663 1664
		clr_cr_intercept(svm, INTERCEPT_CR3_READ);
		clr_cr_intercept(svm, INTERCEPT_CR3_WRITE);
1665
		save->g_pat = svm->vcpu.arch.pat;
1666 1667 1668
		save->cr3 = 0;
		save->cr4 = 0;
	}
1669
	svm->asid_generation = 0;
1670

1671
	svm->nested.vmcb = 0;
1672 1673
	svm->vcpu.arch.hflags = 0;

1674 1675 1676 1677
	if (pause_filter_count) {
		control->pause_filter_count = pause_filter_count;
		if (pause_filter_thresh)
			control->pause_filter_thresh = pause_filter_thresh;
1678
		set_intercept(svm, INTERCEPT_PAUSE);
1679 1680
	} else {
		clr_intercept(svm, INTERCEPT_PAUSE);
1681 1682
	}

1683
	if (kvm_vcpu_apicv_active(&svm->vcpu))
1684 1685
		avic_init_vmcb(svm);

1686 1687 1688 1689 1690 1691 1692 1693 1694 1695
	/*
	 * If hardware supports Virtual VMLOAD VMSAVE then enable it
	 * in VMCB and clear intercepts to avoid #VMEXIT.
	 */
	if (vls) {
		clr_intercept(svm, INTERCEPT_VMLOAD);
		clr_intercept(svm, INTERCEPT_VMSAVE);
		svm->vmcb->control.virt_ext |= VIRTUAL_VMLOAD_VMSAVE_ENABLE_MASK;
	}

1696 1697 1698 1699 1700 1701
	if (vgif) {
		clr_intercept(svm, INTERCEPT_STGI);
		clr_intercept(svm, INTERCEPT_CLGI);
		svm->vmcb->control.int_ctl |= V_GIF_ENABLE_MASK;
	}

1702
	if (sev_guest(svm->vcpu.kvm)) {
B
Brijesh Singh 已提交
1703
		svm->vmcb->control.nested_ctl |= SVM_NESTED_CTL_SEV_ENABLE;
1704 1705
		clr_exception_intercept(svm, UD_VECTOR);
	}
B
Brijesh Singh 已提交
1706

1707 1708
	mark_all_dirty(svm->vmcb);

1709
	enable_gif(svm);
1710 1711 1712

}

1713 1714
static u64 *avic_get_physical_id_entry(struct kvm_vcpu *vcpu,
				       unsigned int index)
1715 1716
{
	u64 *avic_physical_id_table;
1717
	struct kvm_svm *kvm_svm = to_kvm_svm(vcpu->kvm);
1718 1719 1720 1721

	if (index >= AVIC_MAX_PHYSICAL_ID_COUNT)
		return NULL;

1722
	avic_physical_id_table = page_address(kvm_svm->avic_physical_id_table_page);
1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734

	return &avic_physical_id_table[index];
}

/**
 * Note:
 * AVIC hardware walks the nested page table to check permissions,
 * but does not use the SPA address specified in the leaf page
 * table entry since it uses  address in the AVIC_BACKING_PAGE pointer
 * field of the VMCB. Therefore, we set up the
 * APIC_ACCESS_PAGE_PRIVATE_MEMSLOT (4KB) here.
 */
1735
static int avic_update_access_page(struct kvm *kvm, bool activate)
1736
{
1737
	int ret = 0;
1738

1739
	mutex_lock(&kvm->slots_lock);
1740 1741 1742 1743 1744 1745 1746
	/*
	 * During kvm_destroy_vm(), kvm_pit_set_reinject() could trigger
	 * APICv mode change, which update APIC_ACCESS_PAGE_PRIVATE_MEMSLOT
	 * memory region. So, we need to ensure that kvm->mm == current->mm.
	 */
	if ((kvm->arch.apic_access_page_done == activate) ||
	    (kvm->mm != current->mm))
1747
		goto out;
1748

1749 1750 1751
	ret = __x86_set_memory_region(kvm,
				      APIC_ACCESS_PAGE_PRIVATE_MEMSLOT,
				      APIC_DEFAULT_PHYS_BASE,
1752
				      activate ? PAGE_SIZE : 0);
1753
	if (ret)
1754
		goto out;
1755

1756
	kvm->arch.apic_access_page_done = activate;
1757 1758 1759
out:
	mutex_unlock(&kvm->slots_lock);
	return ret;
1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773
}

static int avic_init_backing_page(struct kvm_vcpu *vcpu)
{
	u64 *entry, new_entry;
	int id = vcpu->vcpu_id;
	struct vcpu_svm *svm = to_svm(vcpu);

	if (id >= AVIC_MAX_PHYSICAL_ID_COUNT)
		return -EINVAL;

	if (!svm->vcpu.arch.apic->regs)
		return -EINVAL;

1774 1775 1776 1777 1778 1779 1780 1781
	if (kvm_apicv_activated(vcpu->kvm)) {
		int ret;

		ret = avic_update_access_page(vcpu->kvm, true);
		if (ret)
			return ret;
	}

1782 1783 1784 1785 1786 1787 1788
	svm->avic_backing_page = virt_to_page(svm->vcpu.arch.apic->regs);

	/* Setting AVIC backing page address in the phy APIC ID table */
	entry = avic_get_physical_id_entry(vcpu, id);
	if (!entry)
		return -EINVAL;

1789 1790 1791
	new_entry = __sme_set((page_to_phys(svm->avic_backing_page) &
			      AVIC_PHYSICAL_ID_ENTRY_BACKING_PAGE_MASK) |
			      AVIC_PHYSICAL_ID_ENTRY_VALID_MASK);
1792 1793 1794 1795 1796 1797 1798
	WRITE_ONCE(*entry, new_entry);

	svm->avic_physical_id_cache = entry;

	return 0;
}

1799
static void sev_asid_free(int asid)
B
Brijesh Singh 已提交
1800
{
1801 1802
	struct svm_cpu_data *sd;
	int cpu, pos;
B
Brijesh Singh 已提交
1803

1804 1805
	mutex_lock(&sev_bitmap_lock);

B
Brijesh Singh 已提交
1806
	pos = asid - 1;
1807
	__set_bit(pos, sev_reclaim_asid_bitmap);
1808 1809 1810 1811 1812

	for_each_possible_cpu(cpu) {
		sd = per_cpu(svm_data, cpu);
		sd->sev_vmcbs[pos] = NULL;
	}
B
Brijesh Singh 已提交
1813

1814
	mutex_unlock(&sev_bitmap_lock);
B
Brijesh Singh 已提交
1815 1816
}

1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830
static void sev_unbind_asid(struct kvm *kvm, unsigned int handle)
{
	struct sev_data_decommission *decommission;
	struct sev_data_deactivate *data;

	if (!handle)
		return;

	data = kzalloc(sizeof(*data), GFP_KERNEL);
	if (!data)
		return;

	/* deactivate handle */
	data->handle = handle;
1831

1832 1833
	/* Guard DEACTIVATE against WBINVD/DF_FLUSH used in ASID recycling */
	down_read(&sev_deactivate_lock);
1834
	sev_guest_deactivate(data, NULL);
1835
	up_read(&sev_deactivate_lock);
1836

1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849
	kfree(data);

	decommission = kzalloc(sizeof(*decommission), GFP_KERNEL);
	if (!decommission)
		return;

	/* decommission handle */
	decommission->handle = handle;
	sev_guest_decommission(decommission, NULL);

	kfree(decommission);
}

1850 1851 1852 1853
static struct page **sev_pin_memory(struct kvm *kvm, unsigned long uaddr,
				    unsigned long ulen, unsigned long *n,
				    int write)
{
1854
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
1855 1856 1857
	unsigned long npages, npinned, size;
	unsigned long locked, lock_limit;
	struct page **pages;
1858 1859 1860 1861
	unsigned long first, last;

	if (ulen == 0 || uaddr + ulen < uaddr)
		return NULL;
1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877

	/* Calculate number of pages. */
	first = (uaddr & PAGE_MASK) >> PAGE_SHIFT;
	last = ((uaddr + ulen - 1) & PAGE_MASK) >> PAGE_SHIFT;
	npages = (last - first + 1);

	locked = sev->pages_locked + npages;
	lock_limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT;
	if (locked > lock_limit && !capable(CAP_IPC_LOCK)) {
		pr_err("SEV: %lu locked pages exceed the lock limit of %lu.\n", locked, lock_limit);
		return NULL;
	}

	/* Avoid using vmalloc for smaller buffers. */
	size = npages * sizeof(struct page *);
	if (size > PAGE_SIZE)
1878 1879
		pages = __vmalloc(size, GFP_KERNEL_ACCOUNT | __GFP_ZERO,
				  PAGE_KERNEL);
1880
	else
1881
		pages = kmalloc(size, GFP_KERNEL_ACCOUNT);
1882 1883 1884 1885 1886

	if (!pages)
		return NULL;

	/* Pin the user virtual address. */
1887
	npinned = get_user_pages_fast(uaddr, npages, FOLL_WRITE, pages);
1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908
	if (npinned != npages) {
		pr_err("SEV: Failure locking %lu pages.\n", npages);
		goto err;
	}

	*n = npages;
	sev->pages_locked = locked;

	return pages;

err:
	if (npinned > 0)
		release_pages(pages, npinned);

	kvfree(pages);
	return NULL;
}

static void sev_unpin_memory(struct kvm *kvm, struct page **pages,
			     unsigned long npages)
{
1909
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930

	release_pages(pages, npages);
	kvfree(pages);
	sev->pages_locked -= npages;
}

static void sev_clflush_pages(struct page *pages[], unsigned long npages)
{
	uint8_t *page_virtual;
	unsigned long i;

	if (npages == 0 || pages == NULL)
		return;

	for (i = 0; i < npages; i++) {
		page_virtual = kmap_atomic(pages[i]);
		clflush_cache_range(page_virtual, PAGE_SIZE);
		kunmap_atomic(page_virtual);
	}
}

1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946
static void __unregister_enc_region_locked(struct kvm *kvm,
					   struct enc_region *region)
{
	/*
	 * The guest may change the memory encryption attribute from C=0 -> C=1
	 * or vice versa for this memory range. Lets make sure caches are
	 * flushed to ensure that guest data gets written into memory with
	 * correct C-bit.
	 */
	sev_clflush_pages(region->pages, region->npages);

	sev_unpin_memory(kvm, region->pages, region->npages);
	list_del(&region->list);
	kfree(region);
}

B
Brijesh Singh 已提交
1947 1948
static void sev_vm_destroy(struct kvm *kvm)
{
1949
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
1950 1951
	struct list_head *head = &sev->regions_list;
	struct list_head *pos, *q;
1952

B
Brijesh Singh 已提交
1953 1954 1955
	if (!sev_guest(kvm))
		return;

1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970
	mutex_lock(&kvm->lock);

	/*
	 * if userspace was terminated before unregistering the memory regions
	 * then lets unpin all the registered memory.
	 */
	if (!list_empty(head)) {
		list_for_each_safe(pos, q, head) {
			__unregister_enc_region_locked(kvm,
				list_entry(pos, struct enc_region, list));
		}
	}

	mutex_unlock(&kvm->lock);

1971
	sev_unbind_asid(kvm, sev->handle);
1972
	sev_asid_free(sev->asid);
B
Brijesh Singh 已提交
1973 1974
}

1975 1976
static void avic_vm_destroy(struct kvm *kvm)
{
1977
	unsigned long flags;
1978
	struct kvm_svm *kvm_svm = to_kvm_svm(kvm);
1979

1980 1981 1982
	if (!avic)
		return;

1983 1984 1985 1986
	if (kvm_svm->avic_logical_id_table_page)
		__free_page(kvm_svm->avic_logical_id_table_page);
	if (kvm_svm->avic_physical_id_table_page)
		__free_page(kvm_svm->avic_physical_id_table_page);
1987 1988

	spin_lock_irqsave(&svm_vm_data_hash_lock, flags);
1989
	hash_del(&kvm_svm->hnode);
1990
	spin_unlock_irqrestore(&svm_vm_data_hash_lock, flags);
1991 1992
}

B
Brijesh Singh 已提交
1993 1994 1995 1996 1997 1998
static void svm_vm_destroy(struct kvm *kvm)
{
	avic_vm_destroy(kvm);
	sev_vm_destroy(kvm);
}

1999 2000
static int avic_vm_init(struct kvm *kvm)
{
2001
	unsigned long flags;
2002
	int err = -ENOMEM;
2003 2004
	struct kvm_svm *kvm_svm = to_kvm_svm(kvm);
	struct kvm_svm *k2;
2005 2006
	struct page *p_page;
	struct page *l_page;
2007
	u32 vm_id;
2008 2009 2010 2011 2012

	if (!avic)
		return 0;

	/* Allocating physical APIC ID table (4KB) */
2013
	p_page = alloc_page(GFP_KERNEL_ACCOUNT);
2014 2015 2016
	if (!p_page)
		goto free_avic;

2017
	kvm_svm->avic_physical_id_table_page = p_page;
2018 2019 2020
	clear_page(page_address(p_page));

	/* Allocating logical APIC ID table (4KB) */
2021
	l_page = alloc_page(GFP_KERNEL_ACCOUNT);
2022 2023 2024
	if (!l_page)
		goto free_avic;

2025
	kvm_svm->avic_logical_id_table_page = l_page;
2026 2027
	clear_page(page_address(l_page));

2028
	spin_lock_irqsave(&svm_vm_data_hash_lock, flags);
2029 2030 2031 2032 2033 2034 2035 2036
 again:
	vm_id = next_vm_id = (next_vm_id + 1) & AVIC_VM_ID_MASK;
	if (vm_id == 0) { /* id is 1-based, zero is not okay */
		next_vm_id_wrapped = 1;
		goto again;
	}
	/* Is it still in use? Only possible if wrapped at least once */
	if (next_vm_id_wrapped) {
2037 2038
		hash_for_each_possible(svm_vm_data_hash, k2, hnode, vm_id) {
			if (k2->avic_vm_id == vm_id)
2039 2040 2041
				goto again;
		}
	}
2042 2043
	kvm_svm->avic_vm_id = vm_id;
	hash_add(svm_vm_data_hash, &kvm_svm->hnode, kvm_svm->avic_vm_id);
2044 2045
	spin_unlock_irqrestore(&svm_vm_data_hash_lock, flags);

2046 2047 2048 2049 2050
	return 0;

free_avic:
	avic_vm_destroy(kvm);
	return err;
A
Avi Kivity 已提交
2051 2052
}

2053 2054 2055 2056 2057 2058 2059 2060
static int svm_vm_init(struct kvm *kvm)
{
	if (avic) {
		int ret = avic_vm_init(kvm);
		if (ret)
			return ret;
	}

2061
	kvm_apicv_init(kvm, avic);
2062 2063 2064
	return 0;
}

2065 2066
static inline int
avic_update_iommu_vcpu_affinity(struct kvm_vcpu *vcpu, int cpu, bool r)
2067
{
2068 2069 2070
	int ret = 0;
	unsigned long flags;
	struct amd_svm_iommu_ir *ir;
2071 2072
	struct vcpu_svm *svm = to_svm(vcpu);

2073 2074
	if (!kvm_arch_has_assigned_device(vcpu->kvm))
		return 0;
2075

2076 2077 2078 2079 2080
	/*
	 * Here, we go through the per-vcpu ir_list to update all existing
	 * interrupt remapping table entry targeting this vcpu.
	 */
	spin_lock_irqsave(&svm->ir_list_lock, flags);
2081

2082 2083
	if (list_empty(&svm->ir_list))
		goto out;
2084

2085 2086 2087 2088 2089 2090 2091 2092
	list_for_each_entry(ir, &svm->ir_list, node) {
		ret = amd_iommu_update_ga(cpu, r, ir->data);
		if (ret)
			break;
	}
out:
	spin_unlock_irqrestore(&svm->ir_list_lock, flags);
	return ret;
2093 2094 2095 2096 2097 2098
}

static void avic_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
{
	u64 entry;
	/* ID = 0xff (broadcast), ID > 0xff (reserved) */
2099
	int h_physical_id = kvm_cpu_get_apicid(cpu);
2100 2101 2102 2103 2104
	struct vcpu_svm *svm = to_svm(vcpu);

	if (!kvm_vcpu_apicv_active(vcpu))
		return;

2105 2106 2107 2108 2109
	/*
	 * Since the host physical APIC id is 8 bits,
	 * we can support host APIC ID upto 255.
	 */
	if (WARN_ON(h_physical_id > AVIC_PHYSICAL_ID_ENTRY_HOST_PHYSICAL_ID_MASK))
2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122
		return;

	entry = READ_ONCE(*(svm->avic_physical_id_cache));
	WARN_ON(entry & AVIC_PHYSICAL_ID_ENTRY_IS_RUNNING_MASK);

	entry &= ~AVIC_PHYSICAL_ID_ENTRY_HOST_PHYSICAL_ID_MASK;
	entry |= (h_physical_id & AVIC_PHYSICAL_ID_ENTRY_HOST_PHYSICAL_ID_MASK);

	entry &= ~AVIC_PHYSICAL_ID_ENTRY_IS_RUNNING_MASK;
	if (svm->avic_is_running)
		entry |= AVIC_PHYSICAL_ID_ENTRY_IS_RUNNING_MASK;

	WRITE_ONCE(*(svm->avic_physical_id_cache), entry);
2123 2124
	avic_update_iommu_vcpu_affinity(vcpu, h_physical_id,
					svm->avic_is_running);
2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135
}

static void avic_vcpu_put(struct kvm_vcpu *vcpu)
{
	u64 entry;
	struct vcpu_svm *svm = to_svm(vcpu);

	if (!kvm_vcpu_apicv_active(vcpu))
		return;

	entry = READ_ONCE(*(svm->avic_physical_id_cache));
2136 2137 2138
	if (entry & AVIC_PHYSICAL_ID_ENTRY_IS_RUNNING_MASK)
		avic_update_iommu_vcpu_affinity(vcpu, -1, 0);

2139 2140
	entry &= ~AVIC_PHYSICAL_ID_ENTRY_IS_RUNNING_MASK;
	WRITE_ONCE(*(svm->avic_physical_id_cache), entry);
A
Avi Kivity 已提交
2141 2142
}

2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156
/**
 * This function is called during VCPU halt/unhalt.
 */
static void avic_set_running(struct kvm_vcpu *vcpu, bool is_run)
{
	struct vcpu_svm *svm = to_svm(vcpu);

	svm->avic_is_running = is_run;
	if (is_run)
		avic_vcpu_load(vcpu, vcpu->cpu);
	else
		avic_vcpu_put(vcpu);
}

2157
static void svm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event)
2158 2159
{
	struct vcpu_svm *svm = to_svm(vcpu);
2160 2161
	u32 dummy;
	u32 eax = 1;
2162

2163
	svm->spec_ctrl = 0;
2164
	svm->virt_spec_ctrl = 0;
2165

2166 2167 2168 2169 2170 2171
	if (!init_event) {
		svm->vcpu.arch.apic_base = APIC_DEFAULT_PHYS_BASE |
					   MSR_IA32_APICBASE_ENABLE;
		if (kvm_vcpu_is_reset_bsp(&svm->vcpu))
			svm->vcpu.arch.apic_base |= MSR_IA32_APICBASE_BSP;
	}
P
Paolo Bonzini 已提交
2172
	init_vmcb(svm);
A
Avi Kivity 已提交
2173

2174
	kvm_cpuid(vcpu, &eax, &dummy, &dummy, &dummy, true);
2175
	kvm_rdx_write(vcpu, eax);
2176 2177 2178

	if (kvm_vcpu_apicv_active(vcpu) && !init_event)
		avic_update_vapic_bar(svm, APIC_DEFAULT_PHYS_BASE);
2179 2180
}

2181 2182 2183 2184
static int avic_init_vcpu(struct vcpu_svm *svm)
{
	int ret;

2185
	if (!kvm_vcpu_apicv_active(&svm->vcpu))
2186 2187 2188 2189 2190 2191 2192 2193
		return 0;

	ret = avic_init_backing_page(&svm->vcpu);
	if (ret)
		return ret;

	INIT_LIST_HEAD(&svm->ir_list);
	spin_lock_init(&svm->ir_list_lock);
2194
	svm->dfr_reg = APIC_DFR_FLAT;
2195 2196 2197 2198

	return ret;
}

2199
static int svm_create_vcpu(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
2200
{
2201
	struct vcpu_svm *svm;
A
Avi Kivity 已提交
2202
	struct page *page;
2203
	struct page *msrpm_pages;
A
Alexander Graf 已提交
2204
	struct page *hsave_page;
A
Alexander Graf 已提交
2205
	struct page *nested_msrpm_pages;
R
Rusty Russell 已提交
2206
	int err;
A
Avi Kivity 已提交
2207

2208 2209
	BUILD_BUG_ON(offsetof(struct vcpu_svm, vcpu) != 0);
	svm = to_svm(vcpu);
R
Rusty Russell 已提交
2210

2211
	err = -ENOMEM;
2212
	page = alloc_page(GFP_KERNEL_ACCOUNT);
2213
	if (!page)
2214
		goto out;
A
Avi Kivity 已提交
2215

2216
	msrpm_pages = alloc_pages(GFP_KERNEL_ACCOUNT, MSRPM_ALLOC_ORDER);
2217
	if (!msrpm_pages)
2218
		goto free_page1;
A
Alexander Graf 已提交
2219

2220
	nested_msrpm_pages = alloc_pages(GFP_KERNEL_ACCOUNT, MSRPM_ALLOC_ORDER);
A
Alexander Graf 已提交
2221
	if (!nested_msrpm_pages)
2222
		goto free_page2;
2223

2224
	hsave_page = alloc_page(GFP_KERNEL_ACCOUNT);
A
Alexander Graf 已提交
2225
	if (!hsave_page)
2226 2227
		goto free_page3;

2228 2229 2230
	err = avic_init_vcpu(svm);
	if (err)
		goto free_page4;
2231

2232 2233 2234
	/* We initialize this flag to true to make sure that the is_running
	 * bit would be set the first time the vcpu is loaded.
	 */
2235 2236
	if (irqchip_in_kernel(vcpu->kvm) && kvm_apicv_activated(vcpu->kvm))
		svm->avic_is_running = true;
2237

2238
	svm->nested.hsave = page_address(hsave_page);
A
Alexander Graf 已提交
2239

2240 2241 2242
	svm->msrpm = page_address(msrpm_pages);
	svm_vcpu_init_msrpm(svm->msrpm);

2243
	svm->nested.msrpm = page_address(nested_msrpm_pages);
2244
	svm_vcpu_init_msrpm(svm->nested.msrpm);
A
Alexander Graf 已提交
2245

2246 2247
	svm->vmcb = page_address(page);
	clear_page(svm->vmcb);
2248
	svm->vmcb_pa = __sme_set(page_to_pfn(page) << PAGE_SHIFT);
2249
	svm->asid_generation = 0;
P
Paolo Bonzini 已提交
2250
	init_vmcb(svm);
A
Avi Kivity 已提交
2251

2252
	svm_init_osvw(vcpu);
2253
	vcpu->arch.microcode_version = 0x01000065;
2254

2255
	return 0;
2256

2257 2258
free_page4:
	__free_page(hsave_page);
2259 2260 2261 2262 2263 2264
free_page3:
	__free_pages(nested_msrpm_pages, MSRPM_ALLOC_ORDER);
free_page2:
	__free_pages(msrpm_pages, MSRPM_ALLOC_ORDER);
free_page1:
	__free_page(page);
2265
out:
2266
	return err;
A
Avi Kivity 已提交
2267 2268
}

2269 2270 2271 2272 2273 2274 2275 2276
static void svm_clear_current_vmcb(struct vmcb *vmcb)
{
	int i;

	for_each_online_cpu(i)
		cmpxchg(&per_cpu(svm_data, i)->current_vmcb, vmcb, NULL);
}

A
Avi Kivity 已提交
2277 2278
static void svm_free_vcpu(struct kvm_vcpu *vcpu)
{
2279 2280
	struct vcpu_svm *svm = to_svm(vcpu);

2281 2282 2283 2284 2285 2286 2287
	/*
	 * The vmcb page can be recycled, causing a false negative in
	 * svm_vcpu_load(). So, ensure that no logical CPU has this
	 * vmcb page recorded as its current vmcb.
	 */
	svm_clear_current_vmcb(svm->vmcb);

2288
	__free_page(pfn_to_page(__sme_clr(svm->vmcb_pa) >> PAGE_SHIFT));
2289
	__free_pages(virt_to_page(svm->msrpm), MSRPM_ALLOC_ORDER);
2290 2291
	__free_page(virt_to_page(svm->nested.hsave));
	__free_pages(virt_to_page(svm->nested.msrpm), MSRPM_ALLOC_ORDER);
A
Avi Kivity 已提交
2292 2293
}

2294
static void svm_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
A
Avi Kivity 已提交
2295
{
2296
	struct vcpu_svm *svm = to_svm(vcpu);
A
Ashok Raj 已提交
2297
	struct svm_cpu_data *sd = per_cpu(svm_data, cpu);
2298
	int i;
2299 2300

	if (unlikely(cpu != vcpu->cpu)) {
2301
		svm->asid_generation = 0;
2302
		mark_all_dirty(svm->vmcb);
2303
	}
2304

2305 2306 2307
#ifdef CONFIG_X86_64
	rdmsrl(MSR_GS_BASE, to_svm(vcpu)->host.gs_base);
#endif
2308 2309 2310 2311
	savesegment(fs, svm->host.fs);
	savesegment(gs, svm->host.gs);
	svm->host.ldt = kvm_read_ldt();

2312
	for (i = 0; i < NR_HOST_SAVE_USER_MSRS; i++)
2313
		rdmsrl(host_save_user_msrs[i], svm->host_user_msrs[i]);
2314

2315 2316 2317 2318 2319 2320
	if (static_cpu_has(X86_FEATURE_TSCRATEMSR)) {
		u64 tsc_ratio = vcpu->arch.tsc_scaling_ratio;
		if (tsc_ratio != __this_cpu_read(current_tsc_ratio)) {
			__this_cpu_write(current_tsc_ratio, tsc_ratio);
			wrmsrl(MSR_AMD64_TSC_RATIO, tsc_ratio);
		}
2321
	}
P
Paolo Bonzini 已提交
2322 2323 2324
	/* This assumes that the kernel never uses MSR_TSC_AUX */
	if (static_cpu_has(X86_FEATURE_RDTSCP))
		wrmsrl(MSR_TSC_AUX, svm->tsc_aux);
2325

A
Ashok Raj 已提交
2326 2327 2328 2329
	if (sd->current_vmcb != svm->vmcb) {
		sd->current_vmcb = svm->vmcb;
		indirect_branch_prediction_barrier();
	}
2330
	avic_vcpu_load(vcpu, cpu);
A
Avi Kivity 已提交
2331 2332 2333 2334
}

static void svm_vcpu_put(struct kvm_vcpu *vcpu)
{
2335
	struct vcpu_svm *svm = to_svm(vcpu);
2336 2337
	int i;

2338 2339
	avic_vcpu_put(vcpu);

2340
	++vcpu->stat.host_state_reload;
2341 2342 2343
	kvm_load_ldt(svm->host.ldt);
#ifdef CONFIG_X86_64
	loadsegment(fs, svm->host.fs);
2344
	wrmsrl(MSR_KERNEL_GS_BASE, current->thread.gsbase);
2345
	load_gs_index(svm->host.gs);
2346
#else
2347
#ifdef CONFIG_X86_32_LAZY_GS
2348
	loadsegment(gs, svm->host.gs);
2349
#endif
2350
#endif
2351
	for (i = 0; i < NR_HOST_SAVE_USER_MSRS; i++)
2352
		wrmsrl(host_save_user_msrs[i], svm->host_user_msrs[i]);
A
Avi Kivity 已提交
2353 2354
}

2355 2356 2357 2358 2359 2360 2361
static void svm_vcpu_blocking(struct kvm_vcpu *vcpu)
{
	avic_set_running(vcpu, false);
}

static void svm_vcpu_unblocking(struct kvm_vcpu *vcpu)
{
2362 2363
	if (kvm_check_request(KVM_REQ_APICV_UPDATE, vcpu))
		kvm_vcpu_update_apicv(vcpu);
2364 2365 2366
	avic_set_running(vcpu, true);
}

A
Avi Kivity 已提交
2367 2368
static unsigned long svm_get_rflags(struct kvm_vcpu *vcpu)
{
2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379
	struct vcpu_svm *svm = to_svm(vcpu);
	unsigned long rflags = svm->vmcb->save.rflags;

	if (svm->nmi_singlestep) {
		/* Hide our flags if they were not set by the guest */
		if (!(svm->nmi_singlestep_guest_rflags & X86_EFLAGS_TF))
			rflags &= ~X86_EFLAGS_TF;
		if (!(svm->nmi_singlestep_guest_rflags & X86_EFLAGS_RF))
			rflags &= ~X86_EFLAGS_RF;
	}
	return rflags;
A
Avi Kivity 已提交
2380 2381 2382 2383
}

static void svm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags)
{
2384 2385 2386
	if (to_svm(vcpu)->nmi_singlestep)
		rflags |= (X86_EFLAGS_TF | X86_EFLAGS_RF);

P
Paolo Bonzini 已提交
2387
       /*
A
Andrea Gelmini 已提交
2388
        * Any change of EFLAGS.VM is accompanied by a reload of SS
P
Paolo Bonzini 已提交
2389 2390 2391
        * (caused by either a task switch or an inter-privilege IRET),
        * so we do not need to update the CPL here.
        */
2392
	to_svm(vcpu)->vmcb->save.rflags = rflags;
A
Avi Kivity 已提交
2393 2394
}

A
Avi Kivity 已提交
2395 2396 2397 2398 2399
static void svm_cache_reg(struct kvm_vcpu *vcpu, enum kvm_reg reg)
{
	switch (reg) {
	case VCPU_EXREG_PDPTR:
		BUG_ON(!npt_enabled);
2400
		load_pdptrs(vcpu, vcpu->arch.walk_mmu, kvm_read_cr3(vcpu));
A
Avi Kivity 已提交
2401 2402
		break;
	default:
2403
		WARN_ON_ONCE(1);
A
Avi Kivity 已提交
2404 2405 2406
	}
}

2407 2408
static void svm_set_vintr(struct vcpu_svm *svm)
{
2409
	set_intercept(svm, INTERCEPT_VINTR);
2410 2411 2412 2413
}

static void svm_clear_vintr(struct vcpu_svm *svm)
{
2414
	clr_intercept(svm, INTERCEPT_VINTR);
2415 2416
}

A
Avi Kivity 已提交
2417 2418
static struct vmcb_seg *svm_seg(struct kvm_vcpu *vcpu, int seg)
{
2419
	struct vmcb_save_area *save = &to_svm(vcpu)->vmcb->save;
A
Avi Kivity 已提交
2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431

	switch (seg) {
	case VCPU_SREG_CS: return &save->cs;
	case VCPU_SREG_DS: return &save->ds;
	case VCPU_SREG_ES: return &save->es;
	case VCPU_SREG_FS: return &save->fs;
	case VCPU_SREG_GS: return &save->gs;
	case VCPU_SREG_SS: return &save->ss;
	case VCPU_SREG_TR: return &save->tr;
	case VCPU_SREG_LDTR: return &save->ldtr;
	}
	BUG();
A
Al Viro 已提交
2432
	return NULL;
A
Avi Kivity 已提交
2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456
}

static u64 svm_get_segment_base(struct kvm_vcpu *vcpu, int seg)
{
	struct vmcb_seg *s = svm_seg(vcpu, seg);

	return s->base;
}

static void svm_get_segment(struct kvm_vcpu *vcpu,
			    struct kvm_segment *var, int seg)
{
	struct vmcb_seg *s = svm_seg(vcpu, seg);

	var->base = s->base;
	var->limit = s->limit;
	var->selector = s->selector;
	var->type = s->attrib & SVM_SELECTOR_TYPE_MASK;
	var->s = (s->attrib >> SVM_SELECTOR_S_SHIFT) & 1;
	var->dpl = (s->attrib >> SVM_SELECTOR_DPL_SHIFT) & 3;
	var->present = (s->attrib >> SVM_SELECTOR_P_SHIFT) & 1;
	var->avl = (s->attrib >> SVM_SELECTOR_AVL_SHIFT) & 1;
	var->l = (s->attrib >> SVM_SELECTOR_L_SHIFT) & 1;
	var->db = (s->attrib >> SVM_SELECTOR_DB_SHIFT) & 1;
2457 2458 2459 2460 2461 2462 2463 2464 2465 2466

	/*
	 * AMD CPUs circa 2014 track the G bit for all segments except CS.
	 * However, the SVM spec states that the G bit is not observed by the
	 * CPU, and some VMware virtual CPUs drop the G bit for all segments.
	 * So let's synthesize a legal G bit for all segments, this helps
	 * running KVM nested. It also helps cross-vendor migration, because
	 * Intel's vmentry has a check on the 'G' bit.
	 */
	var->g = s->limit > 0xfffff;
2467

J
Joerg Roedel 已提交
2468 2469
	/*
	 * AMD's VMCB does not have an explicit unusable field, so emulate it
2470 2471
	 * for cross vendor migration purposes by "not present"
	 */
2472
	var->unusable = !var->present;
2473

2474 2475 2476 2477 2478 2479
	switch (seg) {
	case VCPU_SREG_TR:
		/*
		 * Work around a bug where the busy flag in the tr selector
		 * isn't exposed
		 */
2480
		var->type |= 0x2;
2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495
		break;
	case VCPU_SREG_DS:
	case VCPU_SREG_ES:
	case VCPU_SREG_FS:
	case VCPU_SREG_GS:
		/*
		 * The accessed bit must always be set in the segment
		 * descriptor cache, although it can be cleared in the
		 * descriptor, the cached bit always remains at 1. Since
		 * Intel has a check on this, set it here to support
		 * cross-vendor migration.
		 */
		if (!var->unusable)
			var->type |= 0x1;
		break;
2496
	case VCPU_SREG_SS:
J
Joerg Roedel 已提交
2497 2498
		/*
		 * On AMD CPUs sometimes the DB bit in the segment
2499 2500 2501 2502 2503 2504
		 * descriptor is left as 1, although the whole segment has
		 * been made unusable. Clear it here to pass an Intel VMX
		 * entry check when cross vendor migrating.
		 */
		if (var->unusable)
			var->db = 0;
2505
		/* This is symmetric with svm_set_segment() */
J
Jan Kiszka 已提交
2506
		var->dpl = to_svm(vcpu)->vmcb->save.cpl;
2507
		break;
2508
	}
A
Avi Kivity 已提交
2509 2510
}

2511 2512 2513 2514 2515 2516 2517
static int svm_get_cpl(struct kvm_vcpu *vcpu)
{
	struct vmcb_save_area *save = &to_svm(vcpu)->vmcb->save;

	return save->cpl;
}

2518
static void svm_get_idt(struct kvm_vcpu *vcpu, struct desc_ptr *dt)
A
Avi Kivity 已提交
2519
{
2520 2521
	struct vcpu_svm *svm = to_svm(vcpu);

2522 2523
	dt->size = svm->vmcb->save.idtr.limit;
	dt->address = svm->vmcb->save.idtr.base;
A
Avi Kivity 已提交
2524 2525
}

2526
static void svm_set_idt(struct kvm_vcpu *vcpu, struct desc_ptr *dt)
A
Avi Kivity 已提交
2527
{
2528 2529
	struct vcpu_svm *svm = to_svm(vcpu);

2530 2531
	svm->vmcb->save.idtr.limit = dt->size;
	svm->vmcb->save.idtr.base = dt->address ;
2532
	mark_dirty(svm->vmcb, VMCB_DT);
A
Avi Kivity 已提交
2533 2534
}

2535
static void svm_get_gdt(struct kvm_vcpu *vcpu, struct desc_ptr *dt)
A
Avi Kivity 已提交
2536
{
2537 2538
	struct vcpu_svm *svm = to_svm(vcpu);

2539 2540
	dt->size = svm->vmcb->save.gdtr.limit;
	dt->address = svm->vmcb->save.gdtr.base;
A
Avi Kivity 已提交
2541 2542
}

2543
static void svm_set_gdt(struct kvm_vcpu *vcpu, struct desc_ptr *dt)
A
Avi Kivity 已提交
2544
{
2545 2546
	struct vcpu_svm *svm = to_svm(vcpu);

2547 2548
	svm->vmcb->save.gdtr.limit = dt->size;
	svm->vmcb->save.gdtr.base = dt->address ;
2549
	mark_dirty(svm->vmcb, VMCB_DT);
A
Avi Kivity 已提交
2550 2551
}

2552 2553 2554 2555
static void svm_decache_cr0_guest_bits(struct kvm_vcpu *vcpu)
{
}

2556
static void svm_decache_cr4_guest_bits(struct kvm_vcpu *vcpu)
2557 2558 2559
{
}

A
Avi Kivity 已提交
2560 2561 2562 2563 2564
static void update_cr0_intercept(struct vcpu_svm *svm)
{
	ulong gcr0 = svm->vcpu.arch.cr0;
	u64 *hcr0 = &svm->vmcb->save.cr0;

2565 2566
	*hcr0 = (*hcr0 & ~SVM_CR0_SELECTIVE_MASK)
		| (gcr0 & SVM_CR0_SELECTIVE_MASK);
A
Avi Kivity 已提交
2567

2568
	mark_dirty(svm->vmcb, VMCB_CR);
A
Avi Kivity 已提交
2569

2570
	if (gcr0 == *hcr0) {
2571 2572
		clr_cr_intercept(svm, INTERCEPT_CR0_READ);
		clr_cr_intercept(svm, INTERCEPT_CR0_WRITE);
A
Avi Kivity 已提交
2573
	} else {
2574 2575
		set_cr_intercept(svm, INTERCEPT_CR0_READ);
		set_cr_intercept(svm, INTERCEPT_CR0_WRITE);
A
Avi Kivity 已提交
2576 2577 2578
	}
}

A
Avi Kivity 已提交
2579 2580
static void svm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
{
2581 2582
	struct vcpu_svm *svm = to_svm(vcpu);

2583
#ifdef CONFIG_X86_64
2584
	if (vcpu->arch.efer & EFER_LME) {
2585
		if (!is_paging(vcpu) && (cr0 & X86_CR0_PG)) {
2586
			vcpu->arch.efer |= EFER_LMA;
2587
			svm->vmcb->save.efer |= EFER_LMA | EFER_LME;
A
Avi Kivity 已提交
2588 2589
		}

M
Mike Day 已提交
2590
		if (is_paging(vcpu) && !(cr0 & X86_CR0_PG)) {
2591
			vcpu->arch.efer &= ~EFER_LMA;
2592
			svm->vmcb->save.efer &= ~(EFER_LMA | EFER_LME);
A
Avi Kivity 已提交
2593 2594 2595
		}
	}
#endif
2596
	vcpu->arch.cr0 = cr0;
2597 2598 2599

	if (!npt_enabled)
		cr0 |= X86_CR0_PG | X86_CR0_WP;
2600

2601 2602 2603 2604 2605 2606 2607
	/*
	 * re-enable caching here because the QEMU bios
	 * does not do it - this results in some delay at
	 * reboot
	 */
	if (kvm_check_has_quirk(vcpu->kvm, KVM_X86_QUIRK_CD_NW_CLEARED))
		cr0 &= ~(X86_CR0_CD | X86_CR0_NW);
2608
	svm->vmcb->save.cr0 = cr0;
2609
	mark_dirty(svm->vmcb, VMCB_CR);
A
Avi Kivity 已提交
2610
	update_cr0_intercept(svm);
A
Avi Kivity 已提交
2611 2612
}

2613
static int svm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
A
Avi Kivity 已提交
2614
{
2615
	unsigned long host_cr4_mce = cr4_read_shadow() & X86_CR4_MCE;
2616 2617
	unsigned long old_cr4 = to_svm(vcpu)->vmcb->save.cr4;

2618 2619 2620
	if (cr4 & X86_CR4_VMXE)
		return 1;

2621
	if (npt_enabled && ((old_cr4 ^ cr4) & X86_CR4_PGE))
2622
		svm_flush_tlb(vcpu, true);
2623

2624 2625 2626
	vcpu->arch.cr4 = cr4;
	if (!npt_enabled)
		cr4 |= X86_CR4_PAE;
2627
	cr4 |= host_cr4_mce;
2628
	to_svm(vcpu)->vmcb->save.cr4 = cr4;
2629
	mark_dirty(to_svm(vcpu)->vmcb, VMCB_CR);
2630
	return 0;
A
Avi Kivity 已提交
2631 2632 2633 2634 2635
}

static void svm_set_segment(struct kvm_vcpu *vcpu,
			    struct kvm_segment *var, int seg)
{
2636
	struct vcpu_svm *svm = to_svm(vcpu);
A
Avi Kivity 已提交
2637 2638 2639 2640 2641
	struct vmcb_seg *s = svm_seg(vcpu, seg);

	s->base = var->base;
	s->limit = var->limit;
	s->selector = var->selector;
2642 2643 2644 2645 2646 2647 2648 2649
	s->attrib = (var->type & SVM_SELECTOR_TYPE_MASK);
	s->attrib |= (var->s & 1) << SVM_SELECTOR_S_SHIFT;
	s->attrib |= (var->dpl & 3) << SVM_SELECTOR_DPL_SHIFT;
	s->attrib |= ((var->present & 1) && !var->unusable) << SVM_SELECTOR_P_SHIFT;
	s->attrib |= (var->avl & 1) << SVM_SELECTOR_AVL_SHIFT;
	s->attrib |= (var->l & 1) << SVM_SELECTOR_L_SHIFT;
	s->attrib |= (var->db & 1) << SVM_SELECTOR_DB_SHIFT;
	s->attrib |= (var->g & 1) << SVM_SELECTOR_G_SHIFT;
P
Paolo Bonzini 已提交
2650 2651 2652 2653 2654 2655 2656 2657

	/*
	 * This is always accurate, except if SYSRET returned to a segment
	 * with SS.DPL != 3.  Intel does not have this quirk, and always
	 * forces SS.DPL to 3 on sysret, so we ignore that case; fixing it
	 * would entail passing the CPL to userspace and back.
	 */
	if (seg == VCPU_SREG_SS)
2658 2659
		/* This is symmetric with svm_get_segment() */
		svm->vmcb->save.cpl = (var->dpl & 3);
A
Avi Kivity 已提交
2660

2661
	mark_dirty(svm->vmcb, VMCB_SEG);
A
Avi Kivity 已提交
2662 2663
}

2664
static void update_bp_intercept(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
2665
{
J
Jan Kiszka 已提交
2666 2667
	struct vcpu_svm *svm = to_svm(vcpu);

2668
	clr_exception_intercept(svm, BP_VECTOR);
2669

J
Jan Kiszka 已提交
2670 2671
	if (vcpu->guest_debug & KVM_GUESTDBG_ENABLE) {
		if (vcpu->guest_debug & KVM_GUESTDBG_USE_SW_BP)
2672
			set_exception_intercept(svm, BP_VECTOR);
J
Jan Kiszka 已提交
2673 2674
	} else
		vcpu->guest_debug = 0;
2675 2676
}

2677
static void new_asid(struct vcpu_svm *svm, struct svm_cpu_data *sd)
A
Avi Kivity 已提交
2678
{
2679 2680
	if (sd->next_asid > sd->max_asid) {
		++sd->asid_generation;
2681
		sd->next_asid = sd->min_asid;
2682
		svm->vmcb->control.tlb_ctl = TLB_CONTROL_FLUSH_ALL_ASID;
A
Avi Kivity 已提交
2683 2684
	}

2685 2686
	svm->asid_generation = sd->asid_generation;
	svm->vmcb->control.asid = sd->next_asid++;
2687 2688

	mark_dirty(svm->vmcb, VMCB_ASID);
A
Avi Kivity 已提交
2689 2690
}

J
Jan Kiszka 已提交
2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703
static u64 svm_get_dr6(struct kvm_vcpu *vcpu)
{
	return to_svm(vcpu)->vmcb->save.dr6;
}

static void svm_set_dr6(struct kvm_vcpu *vcpu, unsigned long value)
{
	struct vcpu_svm *svm = to_svm(vcpu);

	svm->vmcb->save.dr6 = value;
	mark_dirty(svm->vmcb, VMCB_DR);
}

2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718
static void svm_sync_dirty_debug_regs(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);

	get_debugreg(vcpu->arch.db[0], 0);
	get_debugreg(vcpu->arch.db[1], 1);
	get_debugreg(vcpu->arch.db[2], 2);
	get_debugreg(vcpu->arch.db[3], 3);
	vcpu->arch.dr6 = svm_get_dr6(vcpu);
	vcpu->arch.dr7 = svm->vmcb->save.dr7;

	vcpu->arch.switch_db_regs &= ~KVM_DEBUGREG_WONT_EXIT;
	set_dr_intercepts(svm);
}

2719
static void svm_set_dr7(struct kvm_vcpu *vcpu, unsigned long value)
A
Avi Kivity 已提交
2720
{
2721 2722
	struct vcpu_svm *svm = to_svm(vcpu);

2723
	svm->vmcb->save.dr7 = value;
2724
	mark_dirty(svm->vmcb, VMCB_DR);
A
Avi Kivity 已提交
2725 2726
}

A
Avi Kivity 已提交
2727
static int pf_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
2728
{
2729
	u64 fault_address = __sme_clr(svm->vmcb->control.exit_info_2);
2730
	u64 error_code = svm->vmcb->control.exit_info_1;
A
Avi Kivity 已提交
2731

2732
	return kvm_handle_page_fault(&svm->vcpu, error_code, fault_address,
2733 2734
			static_cpu_has(X86_FEATURE_DECODEASSISTS) ?
			svm->vmcb->control.insn_bytes : NULL,
2735 2736 2737 2738 2739
			svm->vmcb->control.insn_len);
}

static int npf_interception(struct vcpu_svm *svm)
{
2740
	u64 fault_address = __sme_clr(svm->vmcb->control.exit_info_2);
2741 2742 2743 2744
	u64 error_code = svm->vmcb->control.exit_info_1;

	trace_kvm_page_fault(fault_address, error_code);
	return kvm_mmu_page_fault(&svm->vcpu, fault_address, error_code,
2745 2746
			static_cpu_has(X86_FEATURE_DECODEASSISTS) ?
			svm->vmcb->control.insn_bytes : NULL,
2747
			svm->vmcb->control.insn_len);
A
Avi Kivity 已提交
2748 2749
}

A
Avi Kivity 已提交
2750
static int db_interception(struct vcpu_svm *svm)
J
Jan Kiszka 已提交
2751
{
A
Avi Kivity 已提交
2752
	struct kvm_run *kvm_run = svm->vcpu.run;
2753
	struct kvm_vcpu *vcpu = &svm->vcpu;
A
Avi Kivity 已提交
2754

J
Jan Kiszka 已提交
2755
	if (!(svm->vcpu.guest_debug &
2756
	      (KVM_GUESTDBG_SINGLESTEP | KVM_GUESTDBG_USE_HW_BP)) &&
J
Jan Kiszka 已提交
2757
		!svm->nmi_singlestep) {
J
Jan Kiszka 已提交
2758 2759 2760
		kvm_queue_exception(&svm->vcpu, DB_VECTOR);
		return 1;
	}
2761

J
Jan Kiszka 已提交
2762
	if (svm->nmi_singlestep) {
2763
		disable_nmi_singlestep(svm);
2764 2765
		/* Make sure we check for pending NMIs upon entry */
		kvm_make_request(KVM_REQ_EVENT, vcpu);
2766 2767 2768
	}

	if (svm->vcpu.guest_debug &
J
Joerg Roedel 已提交
2769
	    (KVM_GUESTDBG_SINGLESTEP | KVM_GUESTDBG_USE_HW_BP)) {
2770 2771 2772 2773 2774 2775 2776 2777
		kvm_run->exit_reason = KVM_EXIT_DEBUG;
		kvm_run->debug.arch.pc =
			svm->vmcb->save.cs.base + svm->vmcb->save.rip;
		kvm_run->debug.arch.exception = DB_VECTOR;
		return 0;
	}

	return 1;
J
Jan Kiszka 已提交
2778 2779
}

A
Avi Kivity 已提交
2780
static int bp_interception(struct vcpu_svm *svm)
J
Jan Kiszka 已提交
2781
{
A
Avi Kivity 已提交
2782 2783
	struct kvm_run *kvm_run = svm->vcpu.run;

J
Jan Kiszka 已提交
2784 2785 2786 2787 2788 2789
	kvm_run->exit_reason = KVM_EXIT_DEBUG;
	kvm_run->debug.arch.pc = svm->vmcb->save.cs.base + svm->vmcb->save.rip;
	kvm_run->debug.arch.exception = BP_VECTOR;
	return 0;
}

A
Avi Kivity 已提交
2790
static int ud_interception(struct vcpu_svm *svm)
2791
{
W
Wanpeng Li 已提交
2792
	return handle_ud(&svm->vcpu);
2793 2794
}

2795 2796 2797 2798 2799 2800
static int ac_interception(struct vcpu_svm *svm)
{
	kvm_queue_exception_e(&svm->vcpu, AC_VECTOR, 0);
	return 1;
}

2801 2802 2803 2804 2805 2806 2807
static int gp_interception(struct vcpu_svm *svm)
{
	struct kvm_vcpu *vcpu = &svm->vcpu;
	u32 error_code = svm->vmcb->control.exit_info_1;

	WARN_ON_ONCE(!enable_vmware_backdoor);

2808 2809 2810 2811 2812 2813 2814 2815
	/*
	 * VMware backdoor emulation on #GP interception only handles IN{S},
	 * OUT{S}, and RDPMC, none of which generate a non-zero error code.
	 */
	if (error_code) {
		kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
		return 1;
	}
2816
	return kvm_emulate_instruction(vcpu, EMULTYPE_VMWARE_GP);
2817 2818
}

2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857
static bool is_erratum_383(void)
{
	int err, i;
	u64 value;

	if (!erratum_383_found)
		return false;

	value = native_read_msr_safe(MSR_IA32_MC0_STATUS, &err);
	if (err)
		return false;

	/* Bit 62 may or may not be set for this mce */
	value &= ~(1ULL << 62);

	if (value != 0xb600000000010015ULL)
		return false;

	/* Clear MCi_STATUS registers */
	for (i = 0; i < 6; ++i)
		native_write_msr_safe(MSR_IA32_MCx_STATUS(i), 0, 0);

	value = native_read_msr_safe(MSR_IA32_MCG_STATUS, &err);
	if (!err) {
		u32 low, high;

		value &= ~(1ULL << 2);
		low    = lower_32_bits(value);
		high   = upper_32_bits(value);

		native_write_msr_safe(MSR_IA32_MCG_STATUS, low, high);
	}

	/* Flush tlb to evict multi-match entries */
	__flush_tlb_all();

	return true;
}

2858
static void svm_handle_mce(struct vcpu_svm *svm)
2859
{
2860 2861 2862 2863 2864 2865 2866
	if (is_erratum_383()) {
		/*
		 * Erratum 383 triggered. Guest state is corrupt so kill the
		 * guest.
		 */
		pr_err("KVM: Guest triggered AMD Erratum 383\n");

2867
		kvm_make_request(KVM_REQ_TRIPLE_FAULT, &svm->vcpu);
2868 2869 2870 2871

		return;
	}

2872 2873 2874 2875 2876 2877 2878 2879
	/*
	 * On an #MC intercept the MCE handler is not called automatically in
	 * the host. So do it by hand here.
	 */
	asm volatile (
		"int $0x12\n");
	/* not sure if we ever come back to this point */

2880 2881 2882 2883 2884
	return;
}

static int mc_interception(struct vcpu_svm *svm)
{
2885 2886 2887
	return 1;
}

A
Avi Kivity 已提交
2888
static int shutdown_interception(struct vcpu_svm *svm)
2889
{
A
Avi Kivity 已提交
2890 2891
	struct kvm_run *kvm_run = svm->vcpu.run;

2892 2893 2894 2895
	/*
	 * VMCB is undefined after a SHUTDOWN intercept
	 * so reinitialize it.
	 */
2896
	clear_page(svm->vmcb);
P
Paolo Bonzini 已提交
2897
	init_vmcb(svm);
2898 2899 2900 2901 2902

	kvm_run->exit_reason = KVM_EXIT_SHUTDOWN;
	return 0;
}

A
Avi Kivity 已提交
2903
static int io_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
2904
{
2905
	struct kvm_vcpu *vcpu = &svm->vcpu;
M
Mike Day 已提交
2906
	u32 io_info = svm->vmcb->control.exit_info_1; /* address size bug? */
2907
	int size, in, string;
2908
	unsigned port;
A
Avi Kivity 已提交
2909

R
Rusty Russell 已提交
2910
	++svm->vcpu.stat.io_exits;
2911
	string = (io_info & SVM_IOIO_STR_MASK) != 0;
2912
	in = (io_info & SVM_IOIO_TYPE_MASK) != 0;
2913
	if (string)
2914
		return kvm_emulate_instruction(vcpu, 0);
2915

2916 2917
	port = io_info >> 16;
	size = (io_info & SVM_IOIO_SIZE_MASK) >> SVM_IOIO_SIZE_SHIFT;
2918 2919
	svm->next_rip = svm->vmcb->control.exit_info_2;

2920
	return kvm_fast_pio(&svm->vcpu, size, port, in);
A
Avi Kivity 已提交
2921 2922
}

A
Avi Kivity 已提交
2923
static int nmi_interception(struct vcpu_svm *svm)
2924 2925 2926 2927
{
	return 1;
}

A
Avi Kivity 已提交
2928
static int intr_interception(struct vcpu_svm *svm)
2929 2930 2931 2932 2933
{
	++svm->vcpu.stat.irq_exits;
	return 1;
}

A
Avi Kivity 已提交
2934
static int nop_on_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
2935 2936 2937 2938
{
	return 1;
}

A
Avi Kivity 已提交
2939
static int halt_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
2940
{
R
Rusty Russell 已提交
2941
	return kvm_emulate_halt(&svm->vcpu);
A
Avi Kivity 已提交
2942 2943
}

A
Avi Kivity 已提交
2944
static int vmmcall_interception(struct vcpu_svm *svm)
2945
{
2946
	return kvm_emulate_hypercall(&svm->vcpu);
2947 2948
}

2949 2950 2951 2952 2953 2954 2955
static unsigned long nested_svm_get_tdp_cr3(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);

	return svm->nested.nested_cr3;
}

2956 2957 2958 2959 2960 2961 2962
static u64 nested_svm_get_tdp_pdptr(struct kvm_vcpu *vcpu, int index)
{
	struct vcpu_svm *svm = to_svm(vcpu);
	u64 cr3 = svm->nested.nested_cr3;
	u64 pdpte;
	int ret;

2963
	ret = kvm_vcpu_read_guest_page(vcpu, gpa_to_gfn(__sme_clr(cr3)), &pdpte,
2964
				       offset_in_page(cr3) + index * 8, 8);
2965 2966 2967 2968 2969
	if (ret)
		return 0;
	return pdpte;
}

2970 2971 2972 2973 2974
static void nested_svm_set_tdp_cr3(struct kvm_vcpu *vcpu,
				   unsigned long root)
{
	struct vcpu_svm *svm = to_svm(vcpu);

2975
	svm->vmcb->control.nested_cr3 = __sme_set(root);
2976
	mark_dirty(svm->vmcb, VMCB_NPT);
2977 2978
}

2979 2980
static void nested_svm_inject_npf_exit(struct kvm_vcpu *vcpu,
				       struct x86_exception *fault)
2981 2982 2983
{
	struct vcpu_svm *svm = to_svm(vcpu);

2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003
	if (svm->vmcb->control.exit_code != SVM_EXIT_NPF) {
		/*
		 * TODO: track the cause of the nested page fault, and
		 * correctly fill in the high bits of exit_info_1.
		 */
		svm->vmcb->control.exit_code = SVM_EXIT_NPF;
		svm->vmcb->control.exit_code_hi = 0;
		svm->vmcb->control.exit_info_1 = (1ULL << 32);
		svm->vmcb->control.exit_info_2 = fault->address;
	}

	svm->vmcb->control.exit_info_1 &= ~0xffffffffULL;
	svm->vmcb->control.exit_info_1 |= fault->error_code;

	/*
	 * The present bit is always zero for page structure faults on real
	 * hardware.
	 */
	if (svm->vmcb->control.exit_info_1 & (2ULL << 32))
		svm->vmcb->control.exit_info_1 &= ~1;
3004 3005 3006 3007

	nested_svm_vmexit(svm);
}

3008
static void nested_svm_init_mmu_context(struct kvm_vcpu *vcpu)
3009
{
3010
	WARN_ON(mmu_is_nested(vcpu));
3011 3012

	vcpu->arch.mmu = &vcpu->arch.guest_mmu;
3013
	kvm_init_shadow_mmu(vcpu);
3014
	vcpu->arch.mmu->set_cr3           = nested_svm_set_tdp_cr3;
3015
	vcpu->arch.mmu->get_guest_pgd     = nested_svm_get_tdp_cr3;
3016 3017 3018 3019
	vcpu->arch.mmu->get_pdptr         = nested_svm_get_tdp_pdptr;
	vcpu->arch.mmu->inject_page_fault = nested_svm_inject_npf_exit;
	vcpu->arch.mmu->shadow_root_level = get_npt_level(vcpu);
	reset_shadow_zero_bits_mask(vcpu, vcpu->arch.mmu);
3020 3021 3022 3023 3024
	vcpu->arch.walk_mmu              = &vcpu->arch.nested_mmu;
}

static void nested_svm_uninit_mmu_context(struct kvm_vcpu *vcpu)
{
3025
	vcpu->arch.mmu = &vcpu->arch.root_mmu;
3026
	vcpu->arch.walk_mmu = &vcpu->arch.root_mmu;
3027 3028
}

3029 3030
static int nested_svm_check_permissions(struct vcpu_svm *svm)
{
3031 3032
	if (!(svm->vcpu.arch.efer & EFER_SVME) ||
	    !is_paging(&svm->vcpu)) {
3033 3034 3035 3036 3037 3038 3039 3040 3041
		kvm_queue_exception(&svm->vcpu, UD_VECTOR);
		return 1;
	}

	if (svm->vmcb->save.cpl) {
		kvm_inject_gp(&svm->vcpu, 0);
		return 1;
	}

3042
	return 0;
3043 3044
}

3045 3046 3047
static int nested_svm_check_exception(struct vcpu_svm *svm, unsigned nr,
				      bool has_error_code, u32 error_code)
{
3048 3049
	int vmexit;

3050
	if (!is_guest_mode(&svm->vcpu))
3051
		return 0;
3052

3053 3054 3055 3056
	vmexit = nested_svm_intercept(svm);
	if (vmexit != NESTED_EXIT_DONE)
		return 0;

3057 3058 3059
	svm->vmcb->control.exit_code = SVM_EXIT_EXCP_BASE + nr;
	svm->vmcb->control.exit_code_hi = 0;
	svm->vmcb->control.exit_info_1 = error_code;
3060 3061

	/*
3062 3063
	 * EXITINFO2 is undefined for all exception intercepts other
	 * than #PF.
3064
	 */
3065 3066
	if (svm->vcpu.arch.exception.nested_apf)
		svm->vmcb->control.exit_info_2 = svm->vcpu.arch.apf.nested_apf_token;
3067 3068
	else if (svm->vcpu.arch.exception.has_payload)
		svm->vmcb->control.exit_info_2 = svm->vcpu.arch.exception.payload;
3069 3070
	else
		svm->vmcb->control.exit_info_2 = svm->vcpu.arch.cr2;
3071

3072
	svm->nested.exit_required = true;
3073
	return vmexit;
3074 3075
}

3076 3077
/* This function returns true if it is save to enable the irq window */
static inline bool nested_svm_intr(struct vcpu_svm *svm)
3078
{
3079
	if (!is_guest_mode(&svm->vcpu))
3080
		return true;
3081

3082
	if (!(svm->vcpu.arch.hflags & HF_VINTR_MASK))
3083
		return true;
3084

3085
	if (!(svm->vcpu.arch.hflags & HF_HIF_MASK))
3086
		return false;
3087

3088 3089 3090 3091 3092 3093 3094 3095
	/*
	 * if vmexit was already requested (by intercepted exception
	 * for instance) do not overwrite it with "external interrupt"
	 * vmexit.
	 */
	if (svm->nested.exit_required)
		return false;

3096 3097 3098
	svm->vmcb->control.exit_code   = SVM_EXIT_INTR;
	svm->vmcb->control.exit_info_1 = 0;
	svm->vmcb->control.exit_info_2 = 0;
3099

3100 3101 3102
	if (svm->nested.intercept & 1ULL) {
		/*
		 * The #vmexit can't be emulated here directly because this
G
Guo Chao 已提交
3103
		 * code path runs with irqs and preemption disabled. A
3104 3105 3106 3107
		 * #vmexit emulation might sleep. Only signal request for
		 * the #vmexit here.
		 */
		svm->nested.exit_required = true;
3108
		trace_kvm_nested_intr_vmexit(svm->vmcb->save.rip);
3109
		return false;
3110 3111
	}

3112
	return true;
3113 3114
}

3115 3116 3117
/* This function returns true if it is save to enable the nmi window */
static inline bool nested_svm_nmi(struct vcpu_svm *svm)
{
3118
	if (!is_guest_mode(&svm->vcpu))
3119 3120 3121 3122 3123 3124 3125 3126 3127
		return true;

	if (!(svm->nested.intercept & (1ULL << INTERCEPT_NMI)))
		return true;

	svm->vmcb->control.exit_code = SVM_EXIT_NMI;
	svm->nested.exit_required = true;

	return false;
3128 3129
}

3130 3131
static int nested_svm_intercept_ioio(struct vcpu_svm *svm)
{
3132 3133 3134
	unsigned port, size, iopm_len;
	u16 val, mask;
	u8 start_bit;
3135
	u64 gpa;
3136

3137 3138
	if (!(svm->nested.intercept & (1ULL << INTERCEPT_IOIO_PROT)))
		return NESTED_EXIT_HOST;
3139

3140
	port = svm->vmcb->control.exit_info_1 >> 16;
3141 3142
	size = (svm->vmcb->control.exit_info_1 & SVM_IOIO_SIZE_MASK) >>
		SVM_IOIO_SIZE_SHIFT;
3143
	gpa  = svm->nested.vmcb_iopm + (port / 8);
3144 3145 3146 3147
	start_bit = port % 8;
	iopm_len = (start_bit + size > 8) ? 2 : 1;
	mask = (0xf >> (4 - size)) << start_bit;
	val = 0;
3148

3149
	if (kvm_vcpu_read_guest(&svm->vcpu, gpa, &val, iopm_len))
3150
		return NESTED_EXIT_DONE;
3151

3152
	return (val & mask) ? NESTED_EXIT_DONE : NESTED_EXIT_HOST;
3153 3154
}

3155
static int nested_svm_exit_handled_msr(struct vcpu_svm *svm)
3156
{
3157 3158
	u32 offset, msr, value;
	int write, mask;
3159

3160
	if (!(svm->nested.intercept & (1ULL << INTERCEPT_MSR_PROT)))
3161
		return NESTED_EXIT_HOST;
3162

3163 3164 3165 3166
	msr    = svm->vcpu.arch.regs[VCPU_REGS_RCX];
	offset = svm_msrpm_offset(msr);
	write  = svm->vmcb->control.exit_info_1 & 1;
	mask   = 1 << ((2 * (msr & 0xf)) + write);
3167

3168 3169
	if (offset == MSR_INVALID)
		return NESTED_EXIT_DONE;
3170

3171 3172
	/* Offset is in 32 bit units but need in 8 bit units */
	offset *= 4;
3173

3174
	if (kvm_vcpu_read_guest(&svm->vcpu, svm->nested.vmcb_msrpm + offset, &value, 4))
3175
		return NESTED_EXIT_DONE;
3176

3177
	return (value & mask) ? NESTED_EXIT_DONE : NESTED_EXIT_HOST;
3178 3179
}

3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204
/* DB exceptions for our internal use must not cause vmexit */
static int nested_svm_intercept_db(struct vcpu_svm *svm)
{
	unsigned long dr6;

	/* if we're not singlestepping, it's not ours */
	if (!svm->nmi_singlestep)
		return NESTED_EXIT_DONE;

	/* if it's not a singlestep exception, it's not ours */
	if (kvm_get_dr(&svm->vcpu, 6, &dr6))
		return NESTED_EXIT_DONE;
	if (!(dr6 & DR6_BS))
		return NESTED_EXIT_DONE;

	/* if the guest is singlestepping, it should get the vmexit */
	if (svm->nmi_singlestep_guest_rflags & X86_EFLAGS_TF) {
		disable_nmi_singlestep(svm);
		return NESTED_EXIT_DONE;
	}

	/* it's ours, the nested hypervisor must not see this one */
	return NESTED_EXIT_HOST;
}

3205
static int nested_svm_exit_special(struct vcpu_svm *svm)
3206 3207
{
	u32 exit_code = svm->vmcb->control.exit_code;
3208

3209 3210 3211
	switch (exit_code) {
	case SVM_EXIT_INTR:
	case SVM_EXIT_NMI:
3212
	case SVM_EXIT_EXCP_BASE + MC_VECTOR:
3213 3214
		return NESTED_EXIT_HOST;
	case SVM_EXIT_NPF:
J
Joerg Roedel 已提交
3215
		/* For now we are always handling NPFs when using them */
3216 3217 3218 3219
		if (npt_enabled)
			return NESTED_EXIT_HOST;
		break;
	case SVM_EXIT_EXCP_BASE + PF_VECTOR:
G
Gleb Natapov 已提交
3220
		/* When we're shadowing, trap PFs, but not async PF */
3221
		if (!npt_enabled && svm->vcpu.arch.apf.host_apf_reason == 0)
3222 3223 3224 3225
			return NESTED_EXIT_HOST;
		break;
	default:
		break;
3226 3227
	}

3228 3229 3230 3231 3232 3233
	return NESTED_EXIT_CONTINUE;
}

/*
 * If this function returns true, this #vmexit was already handled
 */
3234
static int nested_svm_intercept(struct vcpu_svm *svm)
3235 3236 3237 3238
{
	u32 exit_code = svm->vmcb->control.exit_code;
	int vmexit = NESTED_EXIT_HOST;

3239
	switch (exit_code) {
3240
	case SVM_EXIT_MSR:
3241
		vmexit = nested_svm_exit_handled_msr(svm);
3242
		break;
3243 3244 3245
	case SVM_EXIT_IOIO:
		vmexit = nested_svm_intercept_ioio(svm);
		break;
3246 3247 3248
	case SVM_EXIT_READ_CR0 ... SVM_EXIT_WRITE_CR8: {
		u32 bit = 1U << (exit_code - SVM_EXIT_READ_CR0);
		if (svm->nested.intercept_cr & bit)
3249
			vmexit = NESTED_EXIT_DONE;
3250 3251
		break;
	}
3252 3253 3254
	case SVM_EXIT_READ_DR0 ... SVM_EXIT_WRITE_DR7: {
		u32 bit = 1U << (exit_code - SVM_EXIT_READ_DR0);
		if (svm->nested.intercept_dr & bit)
3255
			vmexit = NESTED_EXIT_DONE;
3256 3257 3258 3259
		break;
	}
	case SVM_EXIT_EXCP_BASE ... SVM_EXIT_EXCP_BASE + 0x1f: {
		u32 excp_bits = 1 << (exit_code - SVM_EXIT_EXCP_BASE);
3260 3261 3262 3263 3264 3265
		if (svm->nested.intercept_exceptions & excp_bits) {
			if (exit_code == SVM_EXIT_EXCP_BASE + DB_VECTOR)
				vmexit = nested_svm_intercept_db(svm);
			else
				vmexit = NESTED_EXIT_DONE;
		}
G
Gleb Natapov 已提交
3266 3267
		/* async page fault always cause vmexit */
		else if ((exit_code == SVM_EXIT_EXCP_BASE + PF_VECTOR) &&
3268
			 svm->vcpu.arch.exception.nested_apf != 0)
G
Gleb Natapov 已提交
3269
			vmexit = NESTED_EXIT_DONE;
3270 3271
		break;
	}
3272 3273 3274 3275
	case SVM_EXIT_ERR: {
		vmexit = NESTED_EXIT_DONE;
		break;
	}
3276 3277
	default: {
		u64 exit_bits = 1ULL << (exit_code - SVM_EXIT_INTR);
J
Joerg Roedel 已提交
3278
		if (svm->nested.intercept & exit_bits)
3279
			vmexit = NESTED_EXIT_DONE;
3280 3281 3282
	}
	}

3283 3284 3285 3286 3287 3288 3289 3290 3291 3292
	return vmexit;
}

static int nested_svm_exit_handled(struct vcpu_svm *svm)
{
	int vmexit;

	vmexit = nested_svm_intercept(svm);

	if (vmexit == NESTED_EXIT_DONE)
3293 3294 3295
		nested_svm_vmexit(svm);

	return vmexit;
3296 3297
}

3298 3299 3300 3301 3302
static inline void copy_vmcb_control_area(struct vmcb *dst_vmcb, struct vmcb *from_vmcb)
{
	struct vmcb_control_area *dst  = &dst_vmcb->control;
	struct vmcb_control_area *from = &from_vmcb->control;

3303
	dst->intercept_cr         = from->intercept_cr;
3304
	dst->intercept_dr         = from->intercept_dr;
3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324
	dst->intercept_exceptions = from->intercept_exceptions;
	dst->intercept            = from->intercept;
	dst->iopm_base_pa         = from->iopm_base_pa;
	dst->msrpm_base_pa        = from->msrpm_base_pa;
	dst->tsc_offset           = from->tsc_offset;
	dst->asid                 = from->asid;
	dst->tlb_ctl              = from->tlb_ctl;
	dst->int_ctl              = from->int_ctl;
	dst->int_vector           = from->int_vector;
	dst->int_state            = from->int_state;
	dst->exit_code            = from->exit_code;
	dst->exit_code_hi         = from->exit_code_hi;
	dst->exit_info_1          = from->exit_info_1;
	dst->exit_info_2          = from->exit_info_2;
	dst->exit_int_info        = from->exit_int_info;
	dst->exit_int_info_err    = from->exit_int_info_err;
	dst->nested_ctl           = from->nested_ctl;
	dst->event_inj            = from->event_inj;
	dst->event_inj_err        = from->event_inj_err;
	dst->nested_cr3           = from->nested_cr3;
3325
	dst->virt_ext              = from->virt_ext;
3326 3327
	dst->pause_filter_count   = from->pause_filter_count;
	dst->pause_filter_thresh  = from->pause_filter_thresh;
3328 3329
}

3330
static int nested_svm_vmexit(struct vcpu_svm *svm)
3331
{
3332
	int rc;
3333
	struct vmcb *nested_vmcb;
3334
	struct vmcb *hsave = svm->nested.hsave;
J
Joerg Roedel 已提交
3335
	struct vmcb *vmcb = svm->vmcb;
3336
	struct kvm_host_map map;
3337

3338 3339 3340 3341
	trace_kvm_nested_vmexit_inject(vmcb->control.exit_code,
				       vmcb->control.exit_info_1,
				       vmcb->control.exit_info_2,
				       vmcb->control.exit_int_info,
3342 3343
				       vmcb->control.exit_int_info_err,
				       KVM_ISA_SVM);
3344

3345
	rc = kvm_vcpu_map(&svm->vcpu, gpa_to_gfn(svm->nested.vmcb), &map);
3346 3347 3348
	if (rc) {
		if (rc == -EINVAL)
			kvm_inject_gp(&svm->vcpu, 0);
3349
		return 1;
3350 3351 3352
	}

	nested_vmcb = map.hva;
3353

3354 3355
	/* Exit Guest-Mode */
	leave_guest_mode(&svm->vcpu);
3356 3357
	svm->nested.vmcb = 0;

3358
	/* Give the current vmcb to the guest */
J
Joerg Roedel 已提交
3359 3360 3361 3362 3363 3364 3365 3366
	disable_gif(svm);

	nested_vmcb->save.es     = vmcb->save.es;
	nested_vmcb->save.cs     = vmcb->save.cs;
	nested_vmcb->save.ss     = vmcb->save.ss;
	nested_vmcb->save.ds     = vmcb->save.ds;
	nested_vmcb->save.gdtr   = vmcb->save.gdtr;
	nested_vmcb->save.idtr   = vmcb->save.idtr;
3367
	nested_vmcb->save.efer   = svm->vcpu.arch.efer;
3368
	nested_vmcb->save.cr0    = kvm_read_cr0(&svm->vcpu);
3369
	nested_vmcb->save.cr3    = kvm_read_cr3(&svm->vcpu);
J
Joerg Roedel 已提交
3370
	nested_vmcb->save.cr2    = vmcb->save.cr2;
3371
	nested_vmcb->save.cr4    = svm->vcpu.arch.cr4;
3372
	nested_vmcb->save.rflags = kvm_get_rflags(&svm->vcpu);
J
Joerg Roedel 已提交
3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388
	nested_vmcb->save.rip    = vmcb->save.rip;
	nested_vmcb->save.rsp    = vmcb->save.rsp;
	nested_vmcb->save.rax    = vmcb->save.rax;
	nested_vmcb->save.dr7    = vmcb->save.dr7;
	nested_vmcb->save.dr6    = vmcb->save.dr6;
	nested_vmcb->save.cpl    = vmcb->save.cpl;

	nested_vmcb->control.int_ctl           = vmcb->control.int_ctl;
	nested_vmcb->control.int_vector        = vmcb->control.int_vector;
	nested_vmcb->control.int_state         = vmcb->control.int_state;
	nested_vmcb->control.exit_code         = vmcb->control.exit_code;
	nested_vmcb->control.exit_code_hi      = vmcb->control.exit_code_hi;
	nested_vmcb->control.exit_info_1       = vmcb->control.exit_info_1;
	nested_vmcb->control.exit_info_2       = vmcb->control.exit_info_2;
	nested_vmcb->control.exit_int_info     = vmcb->control.exit_int_info;
	nested_vmcb->control.exit_int_info_err = vmcb->control.exit_int_info_err;
3389 3390 3391

	if (svm->nrips_enabled)
		nested_vmcb->control.next_rip  = vmcb->control.next_rip;
3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407

	/*
	 * If we emulate a VMRUN/#VMEXIT in the same host #vmexit cycle we have
	 * to make sure that we do not lose injected events. So check event_inj
	 * here and copy it to exit_int_info if it is valid.
	 * Exit_int_info and event_inj can't be both valid because the case
	 * below only happens on a VMRUN instruction intercept which has
	 * no valid exit_int_info set.
	 */
	if (vmcb->control.event_inj & SVM_EVTINJ_VALID) {
		struct vmcb_control_area *nc = &nested_vmcb->control;

		nc->exit_int_info     = vmcb->control.event_inj;
		nc->exit_int_info_err = vmcb->control.event_inj_err;
	}

J
Joerg Roedel 已提交
3408 3409 3410
	nested_vmcb->control.tlb_ctl           = 0;
	nested_vmcb->control.event_inj         = 0;
	nested_vmcb->control.event_inj_err     = 0;
3411

3412 3413 3414 3415 3416
	nested_vmcb->control.pause_filter_count =
		svm->vmcb->control.pause_filter_count;
	nested_vmcb->control.pause_filter_thresh =
		svm->vmcb->control.pause_filter_thresh;

3417 3418 3419 3420 3421
	/* We always set V_INTR_MASKING and remember the old value in hflags */
	if (!(svm->vcpu.arch.hflags & HF_VINTR_MASK))
		nested_vmcb->control.int_ctl &= ~V_INTR_MASKING_MASK;

	/* Restore the original control entries */
3422
	copy_vmcb_control_area(vmcb, hsave);
3423

3424
	svm->vcpu.arch.tsc_offset = svm->vmcb->control.tsc_offset;
3425 3426
	kvm_clear_exception_queue(&svm->vcpu);
	kvm_clear_interrupt_queue(&svm->vcpu);
3427

3428 3429
	svm->nested.nested_cr3 = 0;

3430 3431 3432 3433 3434 3435 3436
	/* Restore selected save entries */
	svm->vmcb->save.es = hsave->save.es;
	svm->vmcb->save.cs = hsave->save.cs;
	svm->vmcb->save.ss = hsave->save.ss;
	svm->vmcb->save.ds = hsave->save.ds;
	svm->vmcb->save.gdtr = hsave->save.gdtr;
	svm->vmcb->save.idtr = hsave->save.idtr;
3437
	kvm_set_rflags(&svm->vcpu, hsave->save.rflags);
3438 3439 3440 3441 3442 3443 3444
	svm_set_efer(&svm->vcpu, hsave->save.efer);
	svm_set_cr0(&svm->vcpu, hsave->save.cr0 | X86_CR0_PE);
	svm_set_cr4(&svm->vcpu, hsave->save.cr4);
	if (npt_enabled) {
		svm->vmcb->save.cr3 = hsave->save.cr3;
		svm->vcpu.arch.cr3 = hsave->save.cr3;
	} else {
3445
		(void)kvm_set_cr3(&svm->vcpu, hsave->save.cr3);
3446
	}
3447
	kvm_rax_write(&svm->vcpu, hsave->save.rax);
3448 3449
	kvm_rsp_write(&svm->vcpu, hsave->save.rsp);
	kvm_rip_write(&svm->vcpu, hsave->save.rip);
3450 3451 3452 3453
	svm->vmcb->save.dr7 = 0;
	svm->vmcb->save.cpl = 0;
	svm->vmcb->control.exit_int_info = 0;

3454 3455
	mark_all_dirty(svm->vmcb);

3456
	kvm_vcpu_unmap(&svm->vcpu, &map, true);
3457

3458
	nested_svm_uninit_mmu_context(&svm->vcpu);
3459 3460 3461
	kvm_mmu_reset_context(&svm->vcpu);
	kvm_mmu_load(&svm->vcpu);

3462 3463 3464 3465 3466 3467 3468 3469
	/*
	 * Drop what we picked up for L2 via svm_complete_interrupts() so it
	 * doesn't end up in L1.
	 */
	svm->vcpu.arch.nmi_injected = false;
	kvm_clear_exception_queue(&svm->vcpu);
	kvm_clear_interrupt_queue(&svm->vcpu);

3470 3471
	return 0;
}
A
Alexander Graf 已提交
3472

3473
static bool nested_svm_vmrun_msrpm(struct vcpu_svm *svm)
A
Alexander Graf 已提交
3474
{
3475 3476
	/*
	 * This function merges the msr permission bitmaps of kvm and the
G
Guo Chao 已提交
3477
	 * nested vmcb. It is optimized in that it only merges the parts where
3478 3479
	 * the kvm msr permission bitmap may contain zero bits
	 */
A
Alexander Graf 已提交
3480
	int i;
3481

3482 3483
	if (!(svm->nested.intercept & (1ULL << INTERCEPT_MSR_PROT)))
		return true;
3484

3485 3486 3487
	for (i = 0; i < MSRPM_OFFSETS; i++) {
		u32 value, p;
		u64 offset;
3488

3489 3490
		if (msrpm_offsets[i] == 0xffffffff)
			break;
A
Alexander Graf 已提交
3491

3492 3493
		p      = msrpm_offsets[i];
		offset = svm->nested.vmcb_msrpm + (p * 4);
3494

3495
		if (kvm_vcpu_read_guest(&svm->vcpu, offset, &value, 4))
3496 3497 3498 3499
			return false;

		svm->nested.msrpm[p] = svm->msrpm[p] | value;
	}
A
Alexander Graf 已提交
3500

3501
	svm->vmcb->control.msrpm_base_pa = __sme_set(__pa(svm->nested.msrpm));
3502 3503

	return true;
A
Alexander Graf 已提交
3504 3505
}

3506 3507 3508 3509 3510
static bool nested_vmcb_checks(struct vmcb *vmcb)
{
	if ((vmcb->control.intercept & (1ULL << INTERCEPT_VMRUN)) == 0)
		return false;

3511 3512 3513
	if (vmcb->control.asid == 0)
		return false;

3514 3515
	if ((vmcb->control.nested_ctl & SVM_NESTED_CTL_NP_ENABLE) &&
	    !npt_enabled)
3516 3517
		return false;

3518 3519 3520
	return true;
}

L
Ladi Prosek 已提交
3521
static void enter_svm_guest_mode(struct vcpu_svm *svm, u64 vmcb_gpa,
3522
				 struct vmcb *nested_vmcb, struct kvm_host_map *map)
A
Alexander Graf 已提交
3523
{
3524
	if (kvm_get_rflags(&svm->vcpu) & X86_EFLAGS_IF)
A
Alexander Graf 已提交
3525 3526 3527 3528
		svm->vcpu.arch.hflags |= HF_HIF_MASK;
	else
		svm->vcpu.arch.hflags &= ~HF_HIF_MASK;

3529
	if (nested_vmcb->control.nested_ctl & SVM_NESTED_CTL_NP_ENABLE) {
3530 3531 3532 3533
		svm->nested.nested_cr3 = nested_vmcb->control.nested_cr3;
		nested_svm_init_mmu_context(&svm->vcpu);
	}

A
Alexander Graf 已提交
3534 3535 3536 3537 3538 3539 3540
	/* Load the nested guest state */
	svm->vmcb->save.es = nested_vmcb->save.es;
	svm->vmcb->save.cs = nested_vmcb->save.cs;
	svm->vmcb->save.ss = nested_vmcb->save.ss;
	svm->vmcb->save.ds = nested_vmcb->save.ds;
	svm->vmcb->save.gdtr = nested_vmcb->save.gdtr;
	svm->vmcb->save.idtr = nested_vmcb->save.idtr;
3541
	kvm_set_rflags(&svm->vcpu, nested_vmcb->save.rflags);
A
Alexander Graf 已提交
3542 3543 3544 3545 3546 3547
	svm_set_efer(&svm->vcpu, nested_vmcb->save.efer);
	svm_set_cr0(&svm->vcpu, nested_vmcb->save.cr0);
	svm_set_cr4(&svm->vcpu, nested_vmcb->save.cr4);
	if (npt_enabled) {
		svm->vmcb->save.cr3 = nested_vmcb->save.cr3;
		svm->vcpu.arch.cr3 = nested_vmcb->save.cr3;
3548
	} else
3549
		(void)kvm_set_cr3(&svm->vcpu, nested_vmcb->save.cr3);
3550 3551 3552 3553

	/* Guest paging mode is active - reset mmu */
	kvm_mmu_reset_context(&svm->vcpu);

J
Joerg Roedel 已提交
3554
	svm->vmcb->save.cr2 = svm->vcpu.arch.cr2 = nested_vmcb->save.cr2;
3555
	kvm_rax_write(&svm->vcpu, nested_vmcb->save.rax);
3556 3557
	kvm_rsp_write(&svm->vcpu, nested_vmcb->save.rsp);
	kvm_rip_write(&svm->vcpu, nested_vmcb->save.rip);
J
Joerg Roedel 已提交
3558

A
Alexander Graf 已提交
3559 3560 3561 3562 3563 3564 3565 3566
	/* In case we don't even reach vcpu_run, the fields are not updated */
	svm->vmcb->save.rax = nested_vmcb->save.rax;
	svm->vmcb->save.rsp = nested_vmcb->save.rsp;
	svm->vmcb->save.rip = nested_vmcb->save.rip;
	svm->vmcb->save.dr7 = nested_vmcb->save.dr7;
	svm->vmcb->save.dr6 = nested_vmcb->save.dr6;
	svm->vmcb->save.cpl = nested_vmcb->save.cpl;

3567
	svm->nested.vmcb_msrpm = nested_vmcb->control.msrpm_base_pa & ~0x0fffULL;
3568
	svm->nested.vmcb_iopm  = nested_vmcb->control.iopm_base_pa  & ~0x0fffULL;
A
Alexander Graf 已提交
3569

J
Joerg Roedel 已提交
3570
	/* cache intercepts */
3571
	svm->nested.intercept_cr         = nested_vmcb->control.intercept_cr;
3572
	svm->nested.intercept_dr         = nested_vmcb->control.intercept_dr;
J
Joerg Roedel 已提交
3573 3574 3575
	svm->nested.intercept_exceptions = nested_vmcb->control.intercept_exceptions;
	svm->nested.intercept            = nested_vmcb->control.intercept;

3576
	svm_flush_tlb(&svm->vcpu, true);
A
Alexander Graf 已提交
3577 3578 3579 3580 3581 3582
	svm->vmcb->control.int_ctl = nested_vmcb->control.int_ctl | V_INTR_MASKING_MASK;
	if (nested_vmcb->control.int_ctl & V_INTR_MASKING_MASK)
		svm->vcpu.arch.hflags |= HF_VINTR_MASK;
	else
		svm->vcpu.arch.hflags &= ~HF_VINTR_MASK;

3583 3584
	if (svm->vcpu.arch.hflags & HF_VINTR_MASK) {
		/* We only want the cr8 intercept bits of the guest */
3585 3586
		clr_cr_intercept(svm, INTERCEPT_CR8_READ);
		clr_cr_intercept(svm, INTERCEPT_CR8_WRITE);
3587 3588
	}

3589
	/* We don't want to see VMMCALLs from a nested guest */
3590
	clr_intercept(svm, INTERCEPT_VMMCALL);
3591

3592 3593 3594
	svm->vcpu.arch.tsc_offset += nested_vmcb->control.tsc_offset;
	svm->vmcb->control.tsc_offset = svm->vcpu.arch.tsc_offset;

3595
	svm->vmcb->control.virt_ext = nested_vmcb->control.virt_ext;
A
Alexander Graf 已提交
3596 3597 3598 3599 3600
	svm->vmcb->control.int_vector = nested_vmcb->control.int_vector;
	svm->vmcb->control.int_state = nested_vmcb->control.int_state;
	svm->vmcb->control.event_inj = nested_vmcb->control.event_inj;
	svm->vmcb->control.event_inj_err = nested_vmcb->control.event_inj_err;

3601 3602 3603 3604 3605
	svm->vmcb->control.pause_filter_count =
		nested_vmcb->control.pause_filter_count;
	svm->vmcb->control.pause_filter_thresh =
		nested_vmcb->control.pause_filter_thresh;

3606
	kvm_vcpu_unmap(&svm->vcpu, map, true);
3607

3608 3609 3610
	/* Enter Guest-Mode */
	enter_guest_mode(&svm->vcpu);

3611 3612 3613 3614 3615 3616
	/*
	 * Merge guest and host intercepts - must be called  with vcpu in
	 * guest-mode to take affect here
	 */
	recalc_intercepts(svm);

3617
	svm->nested.vmcb = vmcb_gpa;
3618

3619
	enable_gif(svm);
A
Alexander Graf 已提交
3620

3621
	mark_all_dirty(svm->vmcb);
L
Ladi Prosek 已提交
3622 3623
}

3624
static int nested_svm_vmrun(struct vcpu_svm *svm)
L
Ladi Prosek 已提交
3625
{
3626
	int ret;
L
Ladi Prosek 已提交
3627 3628 3629
	struct vmcb *nested_vmcb;
	struct vmcb *hsave = svm->nested.hsave;
	struct vmcb *vmcb = svm->vmcb;
3630
	struct kvm_host_map map;
L
Ladi Prosek 已提交
3631 3632 3633 3634
	u64 vmcb_gpa;

	vmcb_gpa = svm->vmcb->save.rax;

3635
	ret = kvm_vcpu_map(&svm->vcpu, gpa_to_gfn(vmcb_gpa), &map);
3636
	if (ret == -EINVAL) {
3637
		kvm_inject_gp(&svm->vcpu, 0);
3638
		return 1;
3639 3640
	} else if (ret) {
		return kvm_skip_emulated_instruction(&svm->vcpu);
3641 3642
	}

3643 3644
	ret = kvm_skip_emulated_instruction(&svm->vcpu);

3645
	nested_vmcb = map.hva;
L
Ladi Prosek 已提交
3646 3647 3648 3649 3650 3651 3652

	if (!nested_vmcb_checks(nested_vmcb)) {
		nested_vmcb->control.exit_code    = SVM_EXIT_ERR;
		nested_vmcb->control.exit_code_hi = 0;
		nested_vmcb->control.exit_info_1  = 0;
		nested_vmcb->control.exit_info_2  = 0;

3653
		kvm_vcpu_unmap(&svm->vcpu, &map, true);
L
Ladi Prosek 已提交
3654

3655
		return ret;
L
Ladi Prosek 已提交
3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696
	}

	trace_kvm_nested_vmrun(svm->vmcb->save.rip, vmcb_gpa,
			       nested_vmcb->save.rip,
			       nested_vmcb->control.int_ctl,
			       nested_vmcb->control.event_inj,
			       nested_vmcb->control.nested_ctl);

	trace_kvm_nested_intercepts(nested_vmcb->control.intercept_cr & 0xffff,
				    nested_vmcb->control.intercept_cr >> 16,
				    nested_vmcb->control.intercept_exceptions,
				    nested_vmcb->control.intercept);

	/* Clear internal status */
	kvm_clear_exception_queue(&svm->vcpu);
	kvm_clear_interrupt_queue(&svm->vcpu);

	/*
	 * Save the old vmcb, so we don't need to pick what we save, but can
	 * restore everything when a VMEXIT occurs
	 */
	hsave->save.es     = vmcb->save.es;
	hsave->save.cs     = vmcb->save.cs;
	hsave->save.ss     = vmcb->save.ss;
	hsave->save.ds     = vmcb->save.ds;
	hsave->save.gdtr   = vmcb->save.gdtr;
	hsave->save.idtr   = vmcb->save.idtr;
	hsave->save.efer   = svm->vcpu.arch.efer;
	hsave->save.cr0    = kvm_read_cr0(&svm->vcpu);
	hsave->save.cr4    = svm->vcpu.arch.cr4;
	hsave->save.rflags = kvm_get_rflags(&svm->vcpu);
	hsave->save.rip    = kvm_rip_read(&svm->vcpu);
	hsave->save.rsp    = vmcb->save.rsp;
	hsave->save.rax    = vmcb->save.rax;
	if (npt_enabled)
		hsave->save.cr3    = vmcb->save.cr3;
	else
		hsave->save.cr3    = kvm_read_cr3(&svm->vcpu);

	copy_vmcb_control_area(hsave, vmcb);

3697
	enter_svm_guest_mode(svm, vmcb_gpa, nested_vmcb, &map);
3698

3699 3700 3701 3702 3703 3704 3705 3706 3707
	if (!nested_svm_vmrun_msrpm(svm)) {
		svm->vmcb->control.exit_code    = SVM_EXIT_ERR;
		svm->vmcb->control.exit_code_hi = 0;
		svm->vmcb->control.exit_info_1  = 0;
		svm->vmcb->control.exit_info_2  = 0;

		nested_svm_vmexit(svm);
	}

3708
	return ret;
A
Alexander Graf 已提交
3709 3710
}

3711
static void nested_svm_vmloadsave(struct vmcb *from_vmcb, struct vmcb *to_vmcb)
3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726
{
	to_vmcb->save.fs = from_vmcb->save.fs;
	to_vmcb->save.gs = from_vmcb->save.gs;
	to_vmcb->save.tr = from_vmcb->save.tr;
	to_vmcb->save.ldtr = from_vmcb->save.ldtr;
	to_vmcb->save.kernel_gs_base = from_vmcb->save.kernel_gs_base;
	to_vmcb->save.star = from_vmcb->save.star;
	to_vmcb->save.lstar = from_vmcb->save.lstar;
	to_vmcb->save.cstar = from_vmcb->save.cstar;
	to_vmcb->save.sfmask = from_vmcb->save.sfmask;
	to_vmcb->save.sysenter_cs = from_vmcb->save.sysenter_cs;
	to_vmcb->save.sysenter_esp = from_vmcb->save.sysenter_esp;
	to_vmcb->save.sysenter_eip = from_vmcb->save.sysenter_eip;
}

A
Avi Kivity 已提交
3727
static int vmload_interception(struct vcpu_svm *svm)
3728
{
3729
	struct vmcb *nested_vmcb;
3730
	struct kvm_host_map map;
3731
	int ret;
3732

3733 3734 3735
	if (nested_svm_check_permissions(svm))
		return 1;

3736 3737 3738 3739
	ret = kvm_vcpu_map(&svm->vcpu, gpa_to_gfn(svm->vmcb->save.rax), &map);
	if (ret) {
		if (ret == -EINVAL)
			kvm_inject_gp(&svm->vcpu, 0);
3740
		return 1;
3741 3742 3743
	}

	nested_vmcb = map.hva;
3744

3745
	ret = kvm_skip_emulated_instruction(&svm->vcpu);
3746

3747
	nested_svm_vmloadsave(nested_vmcb, svm->vmcb);
3748
	kvm_vcpu_unmap(&svm->vcpu, &map, true);
3749

3750
	return ret;
3751 3752
}

A
Avi Kivity 已提交
3753
static int vmsave_interception(struct vcpu_svm *svm)
3754
{
3755
	struct vmcb *nested_vmcb;
3756
	struct kvm_host_map map;
3757
	int ret;
3758

3759 3760 3761
	if (nested_svm_check_permissions(svm))
		return 1;

3762 3763 3764 3765
	ret = kvm_vcpu_map(&svm->vcpu, gpa_to_gfn(svm->vmcb->save.rax), &map);
	if (ret) {
		if (ret == -EINVAL)
			kvm_inject_gp(&svm->vcpu, 0);
3766
		return 1;
3767 3768 3769
	}

	nested_vmcb = map.hva;
3770

3771
	ret = kvm_skip_emulated_instruction(&svm->vcpu);
3772

3773
	nested_svm_vmloadsave(svm->vmcb, nested_vmcb);
3774
	kvm_vcpu_unmap(&svm->vcpu, &map, true);
3775

3776
	return ret;
3777 3778
}

A
Avi Kivity 已提交
3779
static int vmrun_interception(struct vcpu_svm *svm)
A
Alexander Graf 已提交
3780 3781 3782 3783
{
	if (nested_svm_check_permissions(svm))
		return 1;

3784
	return nested_svm_vmrun(svm);
A
Alexander Graf 已提交
3785 3786
}

A
Avi Kivity 已提交
3787
static int stgi_interception(struct vcpu_svm *svm)
3788
{
3789 3790
	int ret;

3791 3792 3793
	if (nested_svm_check_permissions(svm))
		return 1;

3794 3795
	/*
	 * If VGIF is enabled, the STGI intercept is only added to
3796
	 * detect the opening of the SMI/NMI window; remove it now.
3797 3798 3799 3800
	 */
	if (vgif_enabled(svm))
		clr_intercept(svm, INTERCEPT_STGI);

3801
	ret = kvm_skip_emulated_instruction(&svm->vcpu);
3802
	kvm_make_request(KVM_REQ_EVENT, &svm->vcpu);
3803

3804
	enable_gif(svm);
3805

3806
	return ret;
3807 3808
}

A
Avi Kivity 已提交
3809
static int clgi_interception(struct vcpu_svm *svm)
3810
{
3811 3812
	int ret;

3813 3814 3815
	if (nested_svm_check_permissions(svm))
		return 1;

3816
	ret = kvm_skip_emulated_instruction(&svm->vcpu);
3817

3818
	disable_gif(svm);
3819 3820

	/* After a CLGI no interrupts should come */
3821 3822 3823 3824 3825
	if (!kvm_vcpu_apicv_active(&svm->vcpu)) {
		svm_clear_vintr(svm);
		svm->vmcb->control.int_ctl &= ~V_IRQ_MASK;
		mark_dirty(svm->vmcb, VMCB_INTR);
	}
3826

3827
	return ret;
3828 3829
}

A
Avi Kivity 已提交
3830
static int invlpga_interception(struct vcpu_svm *svm)
A
Alexander Graf 已提交
3831 3832 3833
{
	struct kvm_vcpu *vcpu = &svm->vcpu;

3834 3835
	trace_kvm_invlpga(svm->vmcb->save.rip, kvm_rcx_read(&svm->vcpu),
			  kvm_rax_read(&svm->vcpu));
3836

A
Alexander Graf 已提交
3837
	/* Let's treat INVLPGA the same as INVLPG (can be optimized!) */
3838
	kvm_mmu_invlpg(vcpu, kvm_rax_read(&svm->vcpu));
A
Alexander Graf 已提交
3839

3840
	return kvm_skip_emulated_instruction(&svm->vcpu);
A
Alexander Graf 已提交
3841 3842
}

3843 3844
static int skinit_interception(struct vcpu_svm *svm)
{
3845
	trace_kvm_skinit(svm->vmcb->save.rip, kvm_rax_read(&svm->vcpu));
3846 3847 3848 3849 3850

	kvm_queue_exception(&svm->vcpu, UD_VECTOR);
	return 1;
}

D
David Kaplan 已提交
3851 3852
static int wbinvd_interception(struct vcpu_svm *svm)
{
3853
	return kvm_emulate_wbinvd(&svm->vcpu);
D
David Kaplan 已提交
3854 3855
}

J
Joerg Roedel 已提交
3856 3857 3858
static int xsetbv_interception(struct vcpu_svm *svm)
{
	u64 new_bv = kvm_read_edx_eax(&svm->vcpu);
3859
	u32 index = kvm_rcx_read(&svm->vcpu);
J
Joerg Roedel 已提交
3860 3861

	if (kvm_set_xcr(&svm->vcpu, index, new_bv) == 0) {
3862
		return kvm_skip_emulated_instruction(&svm->vcpu);
J
Joerg Roedel 已提交
3863 3864 3865 3866 3867
	}

	return 1;
}

J
Jim Mattson 已提交
3868 3869 3870 3871 3872 3873
static int rdpru_interception(struct vcpu_svm *svm)
{
	kvm_queue_exception(&svm->vcpu, UD_VECTOR);
	return 1;
}

A
Avi Kivity 已提交
3874
static int task_switch_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
3875
{
3876
	u16 tss_selector;
3877 3878 3879
	int reason;
	int int_type = svm->vmcb->control.exit_int_info &
		SVM_EXITINTINFO_TYPE_MASK;
3880
	int int_vec = svm->vmcb->control.exit_int_info & SVM_EVTINJ_VEC_MASK;
3881 3882 3883 3884
	uint32_t type =
		svm->vmcb->control.exit_int_info & SVM_EXITINTINFO_TYPE_MASK;
	uint32_t idt_v =
		svm->vmcb->control.exit_int_info & SVM_EXITINTINFO_VALID;
3885 3886
	bool has_error_code = false;
	u32 error_code = 0;
3887 3888

	tss_selector = (u16)svm->vmcb->control.exit_info_1;
3889

3890 3891
	if (svm->vmcb->control.exit_info_2 &
	    (1ULL << SVM_EXITINFOSHIFT_TS_REASON_IRET))
3892 3893 3894 3895
		reason = TASK_SWITCH_IRET;
	else if (svm->vmcb->control.exit_info_2 &
		 (1ULL << SVM_EXITINFOSHIFT_TS_REASON_JMP))
		reason = TASK_SWITCH_JMP;
3896
	else if (idt_v)
3897 3898 3899 3900
		reason = TASK_SWITCH_GATE;
	else
		reason = TASK_SWITCH_CALL;

3901 3902 3903 3904 3905 3906
	if (reason == TASK_SWITCH_GATE) {
		switch (type) {
		case SVM_EXITINTINFO_TYPE_NMI:
			svm->vcpu.arch.nmi_injected = false;
			break;
		case SVM_EXITINTINFO_TYPE_EXEPT:
3907 3908 3909 3910 3911 3912
			if (svm->vmcb->control.exit_info_2 &
			    (1ULL << SVM_EXITINFOSHIFT_TS_HAS_ERROR_CODE)) {
				has_error_code = true;
				error_code =
					(u32)svm->vmcb->control.exit_info_2;
			}
3913 3914 3915 3916 3917 3918 3919 3920 3921
			kvm_clear_exception_queue(&svm->vcpu);
			break;
		case SVM_EXITINTINFO_TYPE_INTR:
			kvm_clear_interrupt_queue(&svm->vcpu);
			break;
		default:
			break;
		}
	}
3922

3923 3924 3925
	if (reason != TASK_SWITCH_GATE ||
	    int_type == SVM_EXITINTINFO_TYPE_SOFT ||
	    (int_type == SVM_EXITINTINFO_TYPE_EXEPT &&
3926
	     (int_vec == OF_VECTOR || int_vec == BP_VECTOR))) {
3927
		if (!skip_emulated_instruction(&svm->vcpu))
3928
			return 0;
3929
	}
3930

3931 3932 3933
	if (int_type != SVM_EXITINTINFO_TYPE_SOFT)
		int_vec = -1;

3934
	return kvm_task_switch(&svm->vcpu, tss_selector, int_vec, reason,
3935
			       has_error_code, error_code);
A
Avi Kivity 已提交
3936 3937
}

A
Avi Kivity 已提交
3938
static int cpuid_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
3939
{
3940
	return kvm_emulate_cpuid(&svm->vcpu);
A
Avi Kivity 已提交
3941 3942
}

A
Avi Kivity 已提交
3943
static int iret_interception(struct vcpu_svm *svm)
3944 3945
{
	++svm->vcpu.stat.nmi_window_exits;
3946
	clr_intercept(svm, INTERCEPT_IRET);
3947
	svm->vcpu.arch.hflags |= HF_IRET_MASK;
3948
	svm->nmi_iret_rip = kvm_rip_read(&svm->vcpu);
3949
	kvm_make_request(KVM_REQ_EVENT, &svm->vcpu);
3950 3951 3952
	return 1;
}

A
Avi Kivity 已提交
3953
static int invlpg_interception(struct vcpu_svm *svm)
M
Marcelo Tosatti 已提交
3954
{
3955
	if (!static_cpu_has(X86_FEATURE_DECODEASSISTS))
3956
		return kvm_emulate_instruction(&svm->vcpu, 0);
3957 3958

	kvm_mmu_invlpg(&svm->vcpu, svm->vmcb->control.exit_info_1);
3959
	return kvm_skip_emulated_instruction(&svm->vcpu);
M
Marcelo Tosatti 已提交
3960 3961
}

A
Avi Kivity 已提交
3962
static int emulate_on_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
3963
{
3964
	return kvm_emulate_instruction(&svm->vcpu, 0);
A
Avi Kivity 已提交
3965 3966
}

B
Brijesh Singh 已提交
3967 3968
static int rsm_interception(struct vcpu_svm *svm)
{
3969
	return kvm_emulate_instruction_from_buffer(&svm->vcpu, rsm_ins_bytes, 2);
B
Brijesh Singh 已提交
3970 3971
}

A
Avi Kivity 已提交
3972 3973 3974 3975
static int rdpmc_interception(struct vcpu_svm *svm)
{
	int err;

3976
	if (!nrips)
A
Avi Kivity 已提交
3977 3978 3979
		return emulate_on_interception(svm);

	err = kvm_rdpmc(&svm->vcpu);
3980
	return kvm_complete_insn_gp(&svm->vcpu, err);
A
Avi Kivity 已提交
3981 3982
}

3983 3984
static bool check_selective_cr0_intercepted(struct vcpu_svm *svm,
					    unsigned long val)
3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006
{
	unsigned long cr0 = svm->vcpu.arch.cr0;
	bool ret = false;
	u64 intercept;

	intercept = svm->nested.intercept;

	if (!is_guest_mode(&svm->vcpu) ||
	    (!(intercept & (1ULL << INTERCEPT_SELECTIVE_CR0))))
		return false;

	cr0 &= ~SVM_CR0_SELECTIVE_MASK;
	val &= ~SVM_CR0_SELECTIVE_MASK;

	if (cr0 ^ val) {
		svm->vmcb->control.exit_code = SVM_EXIT_CR0_SEL_WRITE;
		ret = (nested_svm_exit_handled(svm) == NESTED_EXIT_DONE);
	}

	return ret;
}

4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021
#define CR_VALID (1ULL << 63)

static int cr_interception(struct vcpu_svm *svm)
{
	int reg, cr;
	unsigned long val;
	int err;

	if (!static_cpu_has(X86_FEATURE_DECODEASSISTS))
		return emulate_on_interception(svm);

	if (unlikely((svm->vmcb->control.exit_info_1 & CR_VALID) == 0))
		return emulate_on_interception(svm);

	reg = svm->vmcb->control.exit_info_1 & SVM_EXITINFO_REG_MASK;
4022 4023 4024 4025
	if (svm->vmcb->control.exit_code == SVM_EXIT_CR0_SEL_WRITE)
		cr = SVM_EXIT_WRITE_CR0 - SVM_EXIT_READ_CR0;
	else
		cr = svm->vmcb->control.exit_code - SVM_EXIT_READ_CR0;
4026 4027 4028 4029 4030 4031 4032

	err = 0;
	if (cr >= 16) { /* mov to cr */
		cr -= 16;
		val = kvm_register_read(&svm->vcpu, reg);
		switch (cr) {
		case 0:
4033 4034
			if (!check_selective_cr0_intercepted(svm, val))
				err = kvm_set_cr0(&svm->vcpu, val);
4035 4036 4037
			else
				return 1;

4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061
			break;
		case 3:
			err = kvm_set_cr3(&svm->vcpu, val);
			break;
		case 4:
			err = kvm_set_cr4(&svm->vcpu, val);
			break;
		case 8:
			err = kvm_set_cr8(&svm->vcpu, val);
			break;
		default:
			WARN(1, "unhandled write to CR%d", cr);
			kvm_queue_exception(&svm->vcpu, UD_VECTOR);
			return 1;
		}
	} else { /* mov from cr */
		switch (cr) {
		case 0:
			val = kvm_read_cr0(&svm->vcpu);
			break;
		case 2:
			val = svm->vcpu.arch.cr2;
			break;
		case 3:
4062
			val = kvm_read_cr3(&svm->vcpu);
4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076
			break;
		case 4:
			val = kvm_read_cr4(&svm->vcpu);
			break;
		case 8:
			val = kvm_get_cr8(&svm->vcpu);
			break;
		default:
			WARN(1, "unhandled read from CR%d", cr);
			kvm_queue_exception(&svm->vcpu, UD_VECTOR);
			return 1;
		}
		kvm_register_write(&svm->vcpu, reg, val);
	}
4077
	return kvm_complete_insn_gp(&svm->vcpu, err);
4078 4079
}

4080 4081 4082 4083 4084
static int dr_interception(struct vcpu_svm *svm)
{
	int reg, dr;
	unsigned long val;

4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095
	if (svm->vcpu.guest_debug == 0) {
		/*
		 * No more DR vmexits; force a reload of the debug registers
		 * and reenter on this instruction.  The next vmexit will
		 * retrieve the full state of the debug registers.
		 */
		clr_dr_intercepts(svm);
		svm->vcpu.arch.switch_db_regs |= KVM_DEBUGREG_WONT_EXIT;
		return 1;
	}

4096 4097 4098 4099 4100 4101 4102
	if (!boot_cpu_has(X86_FEATURE_DECODEASSISTS))
		return emulate_on_interception(svm);

	reg = svm->vmcb->control.exit_info_1 & SVM_EXITINFO_REG_MASK;
	dr = svm->vmcb->control.exit_code - SVM_EXIT_READ_DR0;

	if (dr >= 16) { /* mov to DRn */
4103 4104
		if (!kvm_require_dr(&svm->vcpu, dr - 16))
			return 1;
4105 4106 4107
		val = kvm_register_read(&svm->vcpu, reg);
		kvm_set_dr(&svm->vcpu, dr - 16, val);
	} else {
4108 4109 4110 4111
		if (!kvm_require_dr(&svm->vcpu, dr))
			return 1;
		kvm_get_dr(&svm->vcpu, dr, &val);
		kvm_register_write(&svm->vcpu, reg, val);
4112 4113
	}

4114
	return kvm_skip_emulated_instruction(&svm->vcpu);
4115 4116
}

A
Avi Kivity 已提交
4117
static int cr8_write_interception(struct vcpu_svm *svm)
4118
{
A
Avi Kivity 已提交
4119
	struct kvm_run *kvm_run = svm->vcpu.run;
A
Andre Przywara 已提交
4120
	int r;
A
Avi Kivity 已提交
4121

4122 4123
	u8 cr8_prev = kvm_get_cr8(&svm->vcpu);
	/* instruction emulation calls kvm_set_cr8() */
4124
	r = cr_interception(svm);
4125
	if (lapic_in_kernel(&svm->vcpu))
4126
		return r;
4127
	if (cr8_prev <= kvm_get_cr8(&svm->vcpu))
4128
		return r;
4129 4130 4131 4132
	kvm_run->exit_reason = KVM_EXIT_SET_TPR;
	return 0;
}

4133 4134
static int svm_get_msr_feature(struct kvm_msr_entry *msr)
{
4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146
	msr->data = 0;

	switch (msr->index) {
	case MSR_F10H_DECFG:
		if (boot_cpu_has(X86_FEATURE_LFENCE_RDTSC))
			msr->data |= MSR_F10H_DECFG_LFENCE_SERIALIZE;
		break;
	default:
		return 1;
	}

	return 0;
4147 4148
}

4149
static int svm_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
A
Avi Kivity 已提交
4150
{
4151 4152
	struct vcpu_svm *svm = to_svm(vcpu);

4153
	switch (msr_info->index) {
B
Brian Gerst 已提交
4154
	case MSR_STAR:
4155
		msr_info->data = svm->vmcb->save.star;
A
Avi Kivity 已提交
4156
		break;
4157
#ifdef CONFIG_X86_64
A
Avi Kivity 已提交
4158
	case MSR_LSTAR:
4159
		msr_info->data = svm->vmcb->save.lstar;
A
Avi Kivity 已提交
4160 4161
		break;
	case MSR_CSTAR:
4162
		msr_info->data = svm->vmcb->save.cstar;
A
Avi Kivity 已提交
4163 4164
		break;
	case MSR_KERNEL_GS_BASE:
4165
		msr_info->data = svm->vmcb->save.kernel_gs_base;
A
Avi Kivity 已提交
4166 4167
		break;
	case MSR_SYSCALL_MASK:
4168
		msr_info->data = svm->vmcb->save.sfmask;
A
Avi Kivity 已提交
4169 4170 4171
		break;
#endif
	case MSR_IA32_SYSENTER_CS:
4172
		msr_info->data = svm->vmcb->save.sysenter_cs;
A
Avi Kivity 已提交
4173 4174
		break;
	case MSR_IA32_SYSENTER_EIP:
4175
		msr_info->data = svm->sysenter_eip;
A
Avi Kivity 已提交
4176 4177
		break;
	case MSR_IA32_SYSENTER_ESP:
4178
		msr_info->data = svm->sysenter_esp;
A
Avi Kivity 已提交
4179
		break;
P
Paolo Bonzini 已提交
4180 4181 4182 4183 4184
	case MSR_TSC_AUX:
		if (!boot_cpu_has(X86_FEATURE_RDTSCP))
			return 1;
		msr_info->data = svm->tsc_aux;
		break;
J
Joerg Roedel 已提交
4185 4186 4187 4188 4189
	/*
	 * Nobody will change the following 5 values in the VMCB so we can
	 * safely return them on rdmsr. They will always be 0 until LBRV is
	 * implemented.
	 */
4190
	case MSR_IA32_DEBUGCTLMSR:
4191
		msr_info->data = svm->vmcb->save.dbgctl;
4192 4193
		break;
	case MSR_IA32_LASTBRANCHFROMIP:
4194
		msr_info->data = svm->vmcb->save.br_from;
4195 4196
		break;
	case MSR_IA32_LASTBRANCHTOIP:
4197
		msr_info->data = svm->vmcb->save.br_to;
4198 4199
		break;
	case MSR_IA32_LASTINTFROMIP:
4200
		msr_info->data = svm->vmcb->save.last_excp_from;
4201 4202
		break;
	case MSR_IA32_LASTINTTOIP:
4203
		msr_info->data = svm->vmcb->save.last_excp_to;
4204
		break;
A
Alexander Graf 已提交
4205
	case MSR_VM_HSAVE_PA:
4206
		msr_info->data = svm->nested.hsave_msr;
A
Alexander Graf 已提交
4207
		break;
4208
	case MSR_VM_CR:
4209
		msr_info->data = svm->nested.vm_cr_msr;
4210
		break;
4211 4212
	case MSR_IA32_SPEC_CTRL:
		if (!msr_info->host_initiated &&
4213 4214
		    !guest_cpuid_has(vcpu, X86_FEATURE_SPEC_CTRL) &&
		    !guest_cpuid_has(vcpu, X86_FEATURE_AMD_STIBP) &&
4215 4216
		    !guest_cpuid_has(vcpu, X86_FEATURE_AMD_IBRS) &&
		    !guest_cpuid_has(vcpu, X86_FEATURE_AMD_SSBD))
4217 4218 4219 4220
			return 1;

		msr_info->data = svm->spec_ctrl;
		break;
4221 4222 4223 4224 4225 4226 4227
	case MSR_AMD64_VIRT_SPEC_CTRL:
		if (!msr_info->host_initiated &&
		    !guest_cpuid_has(vcpu, X86_FEATURE_VIRT_SSBD))
			return 1;

		msr_info->data = svm->virt_spec_ctrl;
		break;
4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244
	case MSR_F15H_IC_CFG: {

		int family, model;

		family = guest_cpuid_family(vcpu);
		model  = guest_cpuid_model(vcpu);

		if (family < 0 || model < 0)
			return kvm_get_msr_common(vcpu, msr_info);

		msr_info->data = 0;

		if (family == 0x15 &&
		    (model >= 0x2 && model < 0x20))
			msr_info->data = 0x1E;
		}
		break;
4245 4246 4247
	case MSR_F10H_DECFG:
		msr_info->data = svm->msr_decfg;
		break;
A
Avi Kivity 已提交
4248
	default:
4249
		return kvm_get_msr_common(vcpu, msr_info);
A
Avi Kivity 已提交
4250 4251 4252 4253
	}
	return 0;
}

A
Avi Kivity 已提交
4254
static int rdmsr_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
4255
{
4256
	return kvm_emulate_rdmsr(&svm->vcpu);
A
Avi Kivity 已提交
4257 4258
}

4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283
static int svm_set_vm_cr(struct kvm_vcpu *vcpu, u64 data)
{
	struct vcpu_svm *svm = to_svm(vcpu);
	int svm_dis, chg_mask;

	if (data & ~SVM_VM_CR_VALID_MASK)
		return 1;

	chg_mask = SVM_VM_CR_VALID_MASK;

	if (svm->nested.vm_cr_msr & SVM_VM_CR_SVM_DIS_MASK)
		chg_mask &= ~(SVM_VM_CR_SVM_LOCK_MASK | SVM_VM_CR_SVM_DIS_MASK);

	svm->nested.vm_cr_msr &= ~chg_mask;
	svm->nested.vm_cr_msr |= (data & chg_mask);

	svm_dis = svm->nested.vm_cr_msr & SVM_VM_CR_SVM_DIS_MASK;

	/* check for svm_disable while efer.svme is set */
	if (svm_dis && (vcpu->arch.efer & EFER_SVME))
		return 1;

	return 0;
}

4284
static int svm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr)
A
Avi Kivity 已提交
4285
{
4286 4287
	struct vcpu_svm *svm = to_svm(vcpu);

4288 4289
	u32 ecx = msr->index;
	u64 data = msr->data;
A
Avi Kivity 已提交
4290
	switch (ecx) {
P
Paolo Bonzini 已提交
4291 4292 4293 4294 4295 4296 4297
	case MSR_IA32_CR_PAT:
		if (!kvm_mtrr_valid(vcpu, MSR_IA32_CR_PAT, data))
			return 1;
		vcpu->arch.pat = data;
		svm->vmcb->save.g_pat = data;
		mark_dirty(svm->vmcb, VMCB_NPT);
		break;
4298 4299
	case MSR_IA32_SPEC_CTRL:
		if (!msr->host_initiated &&
4300 4301
		    !guest_cpuid_has(vcpu, X86_FEATURE_SPEC_CTRL) &&
		    !guest_cpuid_has(vcpu, X86_FEATURE_AMD_STIBP) &&
4302 4303
		    !guest_cpuid_has(vcpu, X86_FEATURE_AMD_IBRS) &&
		    !guest_cpuid_has(vcpu, X86_FEATURE_AMD_SSBD))
4304 4305
			return 1;

4306
		if (data & ~kvm_spec_ctrl_valid_bits(vcpu))
4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325
			return 1;

		svm->spec_ctrl = data;
		if (!data)
			break;

		/*
		 * For non-nested:
		 * When it's written (to non-zero) for the first time, pass
		 * it through.
		 *
		 * For nested:
		 * The handling of the MSR bitmap for L2 guests is done in
		 * nested_svm_vmrun_msrpm.
		 * We update the L1 MSR bit as well since it will end up
		 * touching the MSR anyway now.
		 */
		set_msr_interception(svm->msrpm, MSR_IA32_SPEC_CTRL, 1, 1);
		break;
A
Ashok Raj 已提交
4326 4327
	case MSR_IA32_PRED_CMD:
		if (!msr->host_initiated &&
4328
		    !guest_cpuid_has(vcpu, X86_FEATURE_AMD_IBPB))
A
Ashok Raj 已提交
4329 4330 4331 4332
			return 1;

		if (data & ~PRED_CMD_IBPB)
			return 1;
4333 4334
		if (!boot_cpu_has(X86_FEATURE_AMD_IBPB))
			return 1;
A
Ashok Raj 已提交
4335 4336 4337 4338 4339 4340
		if (!data)
			break;

		wrmsrl(MSR_IA32_PRED_CMD, PRED_CMD_IBPB);
		set_msr_interception(svm->msrpm, MSR_IA32_PRED_CMD, 0, 1);
		break;
4341 4342 4343 4344 4345 4346 4347 4348 4349 4350
	case MSR_AMD64_VIRT_SPEC_CTRL:
		if (!msr->host_initiated &&
		    !guest_cpuid_has(vcpu, X86_FEATURE_VIRT_SSBD))
			return 1;

		if (data & ~SPEC_CTRL_SSBD)
			return 1;

		svm->virt_spec_ctrl = data;
		break;
B
Brian Gerst 已提交
4351
	case MSR_STAR:
4352
		svm->vmcb->save.star = data;
A
Avi Kivity 已提交
4353
		break;
4354
#ifdef CONFIG_X86_64
A
Avi Kivity 已提交
4355
	case MSR_LSTAR:
4356
		svm->vmcb->save.lstar = data;
A
Avi Kivity 已提交
4357 4358
		break;
	case MSR_CSTAR:
4359
		svm->vmcb->save.cstar = data;
A
Avi Kivity 已提交
4360 4361
		break;
	case MSR_KERNEL_GS_BASE:
4362
		svm->vmcb->save.kernel_gs_base = data;
A
Avi Kivity 已提交
4363 4364
		break;
	case MSR_SYSCALL_MASK:
4365
		svm->vmcb->save.sfmask = data;
A
Avi Kivity 已提交
4366 4367 4368
		break;
#endif
	case MSR_IA32_SYSENTER_CS:
4369
		svm->vmcb->save.sysenter_cs = data;
A
Avi Kivity 已提交
4370 4371
		break;
	case MSR_IA32_SYSENTER_EIP:
4372
		svm->sysenter_eip = data;
4373
		svm->vmcb->save.sysenter_eip = data;
A
Avi Kivity 已提交
4374 4375
		break;
	case MSR_IA32_SYSENTER_ESP:
4376
		svm->sysenter_esp = data;
4377
		svm->vmcb->save.sysenter_esp = data;
A
Avi Kivity 已提交
4378
		break;
P
Paolo Bonzini 已提交
4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390
	case MSR_TSC_AUX:
		if (!boot_cpu_has(X86_FEATURE_RDTSCP))
			return 1;

		/*
		 * This is rare, so we update the MSR here instead of using
		 * direct_access_msrs.  Doing that would require a rdmsr in
		 * svm_vcpu_put.
		 */
		svm->tsc_aux = data;
		wrmsrl(MSR_TSC_AUX, svm->tsc_aux);
		break;
4391
	case MSR_IA32_DEBUGCTLMSR:
4392
		if (!boot_cpu_has(X86_FEATURE_LBRV)) {
4393 4394
			vcpu_unimpl(vcpu, "%s: MSR_IA32_DEBUGCTL 0x%llx, nop\n",
				    __func__, data);
4395 4396 4397 4398 4399 4400
			break;
		}
		if (data & DEBUGCTL_RESERVED_BITS)
			return 1;

		svm->vmcb->save.dbgctl = data;
4401
		mark_dirty(svm->vmcb, VMCB_LBR);
4402 4403 4404 4405
		if (data & (1ULL<<0))
			svm_enable_lbrv(svm);
		else
			svm_disable_lbrv(svm);
4406
		break;
A
Alexander Graf 已提交
4407
	case MSR_VM_HSAVE_PA:
4408
		svm->nested.hsave_msr = data;
4409
		break;
4410
	case MSR_VM_CR:
4411
		return svm_set_vm_cr(vcpu, data);
4412
	case MSR_VM_IGNNE:
4413
		vcpu_unimpl(vcpu, "unimplemented wrmsr: 0x%x data 0x%llx\n", ecx, data);
4414
		break;
4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432
	case MSR_F10H_DECFG: {
		struct kvm_msr_entry msr_entry;

		msr_entry.index = msr->index;
		if (svm_get_msr_feature(&msr_entry))
			return 1;

		/* Check the supported bits */
		if (data & ~msr_entry.data)
			return 1;

		/* Don't allow the guest to change a bit, #GP */
		if (!msr->host_initiated && (data ^ msr_entry.data))
			return 1;

		svm->msr_decfg = data;
		break;
	}
4433 4434 4435
	case MSR_IA32_APICBASE:
		if (kvm_vcpu_apicv_active(vcpu))
			avic_update_vapic_bar(to_svm(vcpu), data);
4436
		/* Fall through */
A
Avi Kivity 已提交
4437
	default:
4438
		return kvm_set_msr_common(vcpu, msr);
A
Avi Kivity 已提交
4439 4440 4441 4442
	}
	return 0;
}

A
Avi Kivity 已提交
4443
static int wrmsr_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
4444
{
4445
	return kvm_emulate_wrmsr(&svm->vcpu);
A
Avi Kivity 已提交
4446 4447
}

A
Avi Kivity 已提交
4448
static int msr_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
4449
{
R
Rusty Russell 已提交
4450
	if (svm->vmcb->control.exit_info_1)
A
Avi Kivity 已提交
4451
		return wrmsr_interception(svm);
A
Avi Kivity 已提交
4452
	else
A
Avi Kivity 已提交
4453
		return rdmsr_interception(svm);
A
Avi Kivity 已提交
4454 4455
}

A
Avi Kivity 已提交
4456
static int interrupt_window_interception(struct vcpu_svm *svm)
4457
{
4458
	kvm_make_request(KVM_REQ_EVENT, &svm->vcpu);
4459
	svm_clear_vintr(svm);
4460 4461 4462 4463 4464 4465 4466 4467

	/*
	 * For AVIC, the only reason to end up here is ExtINTs.
	 * In this case AVIC was temporarily disabled for
	 * requesting the IRQ window and we have to re-enable it.
	 */
	svm_toggle_avic_for_irq_window(&svm->vcpu, true);

4468
	svm->vmcb->control.int_ctl &= ~V_IRQ_MASK;
4469
	mark_dirty(svm->vmcb, VMCB_INTR);
4470
	++svm->vcpu.stat.irq_window_exits;
4471 4472 4473
	return 1;
}

4474 4475
static int pause_interception(struct vcpu_svm *svm)
{
4476 4477 4478
	struct kvm_vcpu *vcpu = &svm->vcpu;
	bool in_kernel = (svm_get_cpl(vcpu) == 0);

4479 4480 4481
	if (pause_filter_thresh)
		grow_ple_window(vcpu);

4482
	kvm_vcpu_on_spin(vcpu, in_kernel);
4483 4484 4485
	return 1;
}

4486 4487
static int nop_interception(struct vcpu_svm *svm)
{
4488
	return kvm_skip_emulated_instruction(&(svm->vcpu));
4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502
}

static int monitor_interception(struct vcpu_svm *svm)
{
	printk_once(KERN_WARNING "kvm: MONITOR instruction emulated as NOP!\n");
	return nop_interception(svm);
}

static int mwait_interception(struct vcpu_svm *svm)
{
	printk_once(KERN_WARNING "kvm: MWAIT instruction emulated as NOP!\n");
	return nop_interception(svm);
}

4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514
enum avic_ipi_failure_cause {
	AVIC_IPI_FAILURE_INVALID_INT_TYPE,
	AVIC_IPI_FAILURE_TARGET_NOT_RUNNING,
	AVIC_IPI_FAILURE_INVALID_TARGET,
	AVIC_IPI_FAILURE_INVALID_BACKING_PAGE,
};

static int avic_incomplete_ipi_interception(struct vcpu_svm *svm)
{
	u32 icrh = svm->vmcb->control.exit_info_1 >> 32;
	u32 icrl = svm->vmcb->control.exit_info_1;
	u32 id = svm->vmcb->control.exit_info_2 >> 32;
D
Dan Carpenter 已提交
4515
	u32 index = svm->vmcb->control.exit_info_2 & 0xFF;
4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536
	struct kvm_lapic *apic = svm->vcpu.arch.apic;

	trace_kvm_avic_incomplete_ipi(svm->vcpu.vcpu_id, icrh, icrl, id, index);

	switch (id) {
	case AVIC_IPI_FAILURE_INVALID_INT_TYPE:
		/*
		 * AVIC hardware handles the generation of
		 * IPIs when the specified Message Type is Fixed
		 * (also known as fixed delivery mode) and
		 * the Trigger Mode is edge-triggered. The hardware
		 * also supports self and broadcast delivery modes
		 * specified via the Destination Shorthand(DSH)
		 * field of the ICRL. Logical and physical APIC ID
		 * formats are supported. All other IPI types cause
		 * a #VMEXIT, which needs to emulated.
		 */
		kvm_lapic_reg_write(apic, APIC_ICR2, icrh);
		kvm_lapic_reg_write(apic, APIC_ICR, icrl);
		break;
	case AVIC_IPI_FAILURE_TARGET_NOT_RUNNING: {
4537 4538 4539
		int i;
		struct kvm_vcpu *vcpu;
		struct kvm *kvm = svm->vcpu.kvm;
4540 4541 4542
		struct kvm_lapic *apic = svm->vcpu.arch.apic;

		/*
4543 4544 4545
		 * At this point, we expect that the AVIC HW has already
		 * set the appropriate IRR bits on the valid target
		 * vcpus. So, we just need to kick the appropriate vcpu.
4546
		 */
4547 4548
		kvm_for_each_vcpu(i, vcpu, kvm) {
			bool m = kvm_apic_match_dest(vcpu, apic,
4549
						     icrl & APIC_SHORT_MASK,
4550
						     GET_APIC_DEST_FIELD(icrh),
4551
						     icrl & APIC_DEST_MASK);
4552 4553 4554 4555

			if (m && !avic_vcpu_is_running(vcpu))
				kvm_vcpu_wake_up(vcpu);
		}
4556 4557 4558
		break;
	}
	case AVIC_IPI_FAILURE_INVALID_TARGET:
4559 4560
		WARN_ONCE(1, "Invalid IPI target: index=%u, vcpu=%d, icr=%#0x:%#0x\n",
			  index, svm->vcpu.vcpu_id, icrh, icrl);
4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573
		break;
	case AVIC_IPI_FAILURE_INVALID_BACKING_PAGE:
		WARN_ONCE(1, "Invalid backing page\n");
		break;
	default:
		pr_err("Unknown IPI interception\n");
	}

	return 1;
}

static u32 *avic_get_logical_id_entry(struct kvm_vcpu *vcpu, u32 ldr, bool flat)
{
4574
	struct kvm_svm *kvm_svm = to_kvm_svm(vcpu->kvm);
4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595
	int index;
	u32 *logical_apic_id_table;
	int dlid = GET_APIC_LOGICAL_ID(ldr);

	if (!dlid)
		return NULL;

	if (flat) { /* flat */
		index = ffs(dlid) - 1;
		if (index > 7)
			return NULL;
	} else { /* cluster */
		int cluster = (dlid & 0xf0) >> 4;
		int apic = ffs(dlid & 0x0f) - 1;

		if ((apic < 0) || (apic > 7) ||
		    (cluster >= 0xf))
			return NULL;
		index = (cluster << 2) + apic;
	}

4596
	logical_apic_id_table = (u32 *) page_address(kvm_svm->avic_logical_id_table_page);
4597 4598 4599 4600

	return &logical_apic_id_table[index];
}

4601
static int avic_ldr_write(struct kvm_vcpu *vcpu, u8 g_physical_id, u32 ldr)
4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613
{
	bool flat;
	u32 *entry, new_entry;

	flat = kvm_lapic_get_reg(vcpu->arch.apic, APIC_DFR) == APIC_DFR_FLAT;
	entry = avic_get_logical_id_entry(vcpu, ldr, flat);
	if (!entry)
		return -EINVAL;

	new_entry = READ_ONCE(*entry);
	new_entry &= ~AVIC_LOGICAL_ID_ENTRY_GUEST_PHYSICAL_ID_MASK;
	new_entry |= (g_physical_id & AVIC_LOGICAL_ID_ENTRY_GUEST_PHYSICAL_ID_MASK);
4614
	new_entry |= AVIC_LOGICAL_ID_ENTRY_VALID_MASK;
4615 4616 4617 4618 4619
	WRITE_ONCE(*entry, new_entry);

	return 0;
}

4620 4621 4622 4623 4624 4625 4626
static void avic_invalidate_logical_id_entry(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);
	bool flat = svm->dfr_reg == APIC_DFR_FLAT;
	u32 *entry = avic_get_logical_id_entry(vcpu, svm->ldr_reg, flat);

	if (entry)
4627
		clear_bit(AVIC_LOGICAL_ID_ENTRY_VALID_BIT, (unsigned long *)entry);
4628 4629
}

4630 4631
static int avic_handle_ldr_update(struct kvm_vcpu *vcpu)
{
4632
	int ret = 0;
4633 4634
	struct vcpu_svm *svm = to_svm(vcpu);
	u32 ldr = kvm_lapic_get_reg(vcpu->arch.apic, APIC_LDR);
4635
	u32 id = kvm_xapic_id(vcpu->arch.apic);
4636

4637 4638
	if (ldr == svm->ldr_reg)
		return 0;
4639

4640 4641 4642
	avic_invalidate_logical_id_entry(vcpu);

	if (ldr)
4643
		ret = avic_ldr_write(vcpu, id, ldr);
4644 4645

	if (!ret)
4646
		svm->ldr_reg = ldr;
4647

4648 4649 4650 4651 4652 4653 4654
	return ret;
}

static int avic_handle_apic_id_update(struct kvm_vcpu *vcpu)
{
	u64 *old, *new;
	struct vcpu_svm *svm = to_svm(vcpu);
4655
	u32 id = kvm_xapic_id(vcpu->arch.apic);
4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679

	if (vcpu->vcpu_id == id)
		return 0;

	old = avic_get_physical_id_entry(vcpu, vcpu->vcpu_id);
	new = avic_get_physical_id_entry(vcpu, id);
	if (!new || !old)
		return 1;

	/* We need to move physical_id_entry to new offset */
	*new = *old;
	*old = 0ULL;
	to_svm(vcpu)->avic_physical_id_cache = new;

	/*
	 * Also update the guest physical APIC ID in the logical
	 * APIC ID table entry if already setup the LDR.
	 */
	if (svm->ldr_reg)
		avic_handle_ldr_update(vcpu);

	return 0;
}

4680
static void avic_handle_dfr_update(struct kvm_vcpu *vcpu)
4681 4682 4683 4684
{
	struct vcpu_svm *svm = to_svm(vcpu);
	u32 dfr = kvm_lapic_get_reg(vcpu->arch.apic, APIC_DFR);

4685 4686
	if (svm->dfr_reg == dfr)
		return;
4687

4688 4689
	avic_invalidate_logical_id_entry(vcpu);
	svm->dfr_reg = dfr;
4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766
}

static int avic_unaccel_trap_write(struct vcpu_svm *svm)
{
	struct kvm_lapic *apic = svm->vcpu.arch.apic;
	u32 offset = svm->vmcb->control.exit_info_1 &
				AVIC_UNACCEL_ACCESS_OFFSET_MASK;

	switch (offset) {
	case APIC_ID:
		if (avic_handle_apic_id_update(&svm->vcpu))
			return 0;
		break;
	case APIC_LDR:
		if (avic_handle_ldr_update(&svm->vcpu))
			return 0;
		break;
	case APIC_DFR:
		avic_handle_dfr_update(&svm->vcpu);
		break;
	default:
		break;
	}

	kvm_lapic_reg_write(apic, offset, kvm_lapic_get_reg(apic, offset));

	return 1;
}

static bool is_avic_unaccelerated_access_trap(u32 offset)
{
	bool ret = false;

	switch (offset) {
	case APIC_ID:
	case APIC_EOI:
	case APIC_RRR:
	case APIC_LDR:
	case APIC_DFR:
	case APIC_SPIV:
	case APIC_ESR:
	case APIC_ICR:
	case APIC_LVTT:
	case APIC_LVTTHMR:
	case APIC_LVTPC:
	case APIC_LVT0:
	case APIC_LVT1:
	case APIC_LVTERR:
	case APIC_TMICT:
	case APIC_TDCR:
		ret = true;
		break;
	default:
		break;
	}
	return ret;
}

static int avic_unaccelerated_access_interception(struct vcpu_svm *svm)
{
	int ret = 0;
	u32 offset = svm->vmcb->control.exit_info_1 &
		     AVIC_UNACCEL_ACCESS_OFFSET_MASK;
	u32 vector = svm->vmcb->control.exit_info_2 &
		     AVIC_UNACCEL_ACCESS_VECTOR_MASK;
	bool write = (svm->vmcb->control.exit_info_1 >> 32) &
		     AVIC_UNACCEL_ACCESS_WRITE_MASK;
	bool trap = is_avic_unaccelerated_access_trap(offset);

	trace_kvm_avic_unaccelerated_access(svm->vcpu.vcpu_id, offset,
					    trap, write, vector);
	if (trap) {
		/* Handling Trap */
		WARN_ONCE(!write, "svm: Handling trap read.\n");
		ret = avic_unaccel_trap_write(svm);
	} else {
		/* Handling Fault */
4767
		ret = kvm_emulate_instruction(&svm->vcpu, 0);
4768 4769 4770 4771 4772
	}

	return ret;
}

4773
static int (*const svm_exit_handlers[])(struct vcpu_svm *svm) = {
4774 4775 4776 4777
	[SVM_EXIT_READ_CR0]			= cr_interception,
	[SVM_EXIT_READ_CR3]			= cr_interception,
	[SVM_EXIT_READ_CR4]			= cr_interception,
	[SVM_EXIT_READ_CR8]			= cr_interception,
4778
	[SVM_EXIT_CR0_SEL_WRITE]		= cr_interception,
4779
	[SVM_EXIT_WRITE_CR0]			= cr_interception,
4780 4781
	[SVM_EXIT_WRITE_CR3]			= cr_interception,
	[SVM_EXIT_WRITE_CR4]			= cr_interception,
J
Joerg Roedel 已提交
4782
	[SVM_EXIT_WRITE_CR8]			= cr8_write_interception,
4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798
	[SVM_EXIT_READ_DR0]			= dr_interception,
	[SVM_EXIT_READ_DR1]			= dr_interception,
	[SVM_EXIT_READ_DR2]			= dr_interception,
	[SVM_EXIT_READ_DR3]			= dr_interception,
	[SVM_EXIT_READ_DR4]			= dr_interception,
	[SVM_EXIT_READ_DR5]			= dr_interception,
	[SVM_EXIT_READ_DR6]			= dr_interception,
	[SVM_EXIT_READ_DR7]			= dr_interception,
	[SVM_EXIT_WRITE_DR0]			= dr_interception,
	[SVM_EXIT_WRITE_DR1]			= dr_interception,
	[SVM_EXIT_WRITE_DR2]			= dr_interception,
	[SVM_EXIT_WRITE_DR3]			= dr_interception,
	[SVM_EXIT_WRITE_DR4]			= dr_interception,
	[SVM_EXIT_WRITE_DR5]			= dr_interception,
	[SVM_EXIT_WRITE_DR6]			= dr_interception,
	[SVM_EXIT_WRITE_DR7]			= dr_interception,
J
Jan Kiszka 已提交
4799 4800
	[SVM_EXIT_EXCP_BASE + DB_VECTOR]	= db_interception,
	[SVM_EXIT_EXCP_BASE + BP_VECTOR]	= bp_interception,
4801
	[SVM_EXIT_EXCP_BASE + UD_VECTOR]	= ud_interception,
J
Joerg Roedel 已提交
4802 4803
	[SVM_EXIT_EXCP_BASE + PF_VECTOR]	= pf_interception,
	[SVM_EXIT_EXCP_BASE + MC_VECTOR]	= mc_interception,
4804
	[SVM_EXIT_EXCP_BASE + AC_VECTOR]	= ac_interception,
4805
	[SVM_EXIT_EXCP_BASE + GP_VECTOR]	= gp_interception,
J
Joerg Roedel 已提交
4806
	[SVM_EXIT_INTR]				= intr_interception,
4807
	[SVM_EXIT_NMI]				= nmi_interception,
A
Avi Kivity 已提交
4808 4809
	[SVM_EXIT_SMI]				= nop_on_interception,
	[SVM_EXIT_INIT]				= nop_on_interception,
4810
	[SVM_EXIT_VINTR]			= interrupt_window_interception,
A
Avi Kivity 已提交
4811
	[SVM_EXIT_RDPMC]			= rdpmc_interception,
A
Avi Kivity 已提交
4812
	[SVM_EXIT_CPUID]			= cpuid_interception,
4813
	[SVM_EXIT_IRET]                         = iret_interception,
4814
	[SVM_EXIT_INVD]                         = emulate_on_interception,
4815
	[SVM_EXIT_PAUSE]			= pause_interception,
A
Avi Kivity 已提交
4816
	[SVM_EXIT_HLT]				= halt_interception,
M
Marcelo Tosatti 已提交
4817
	[SVM_EXIT_INVLPG]			= invlpg_interception,
A
Alexander Graf 已提交
4818
	[SVM_EXIT_INVLPGA]			= invlpga_interception,
J
Joerg Roedel 已提交
4819
	[SVM_EXIT_IOIO]				= io_interception,
A
Avi Kivity 已提交
4820 4821
	[SVM_EXIT_MSR]				= msr_interception,
	[SVM_EXIT_TASK_SWITCH]			= task_switch_interception,
4822
	[SVM_EXIT_SHUTDOWN]			= shutdown_interception,
A
Alexander Graf 已提交
4823
	[SVM_EXIT_VMRUN]			= vmrun_interception,
4824
	[SVM_EXIT_VMMCALL]			= vmmcall_interception,
4825 4826
	[SVM_EXIT_VMLOAD]			= vmload_interception,
	[SVM_EXIT_VMSAVE]			= vmsave_interception,
4827 4828
	[SVM_EXIT_STGI]				= stgi_interception,
	[SVM_EXIT_CLGI]				= clgi_interception,
4829
	[SVM_EXIT_SKINIT]			= skinit_interception,
D
David Kaplan 已提交
4830
	[SVM_EXIT_WBINVD]                       = wbinvd_interception,
4831 4832
	[SVM_EXIT_MONITOR]			= monitor_interception,
	[SVM_EXIT_MWAIT]			= mwait_interception,
J
Joerg Roedel 已提交
4833
	[SVM_EXIT_XSETBV]			= xsetbv_interception,
J
Jim Mattson 已提交
4834
	[SVM_EXIT_RDPRU]			= rdpru_interception,
4835
	[SVM_EXIT_NPF]				= npf_interception,
B
Brijesh Singh 已提交
4836
	[SVM_EXIT_RSM]                          = rsm_interception,
4837 4838
	[SVM_EXIT_AVIC_INCOMPLETE_IPI]		= avic_incomplete_ipi_interception,
	[SVM_EXIT_AVIC_UNACCELERATED_ACCESS]	= avic_unaccelerated_access_interception,
A
Avi Kivity 已提交
4839 4840
};

4841
static void dump_vmcb(struct kvm_vcpu *vcpu)
4842 4843 4844 4845 4846
{
	struct vcpu_svm *svm = to_svm(vcpu);
	struct vmcb_control_area *control = &svm->vmcb->control;
	struct vmcb_save_area *save = &svm->vmcb->save;

4847 4848 4849 4850 4851
	if (!dump_invalid_vmcb) {
		pr_warn_ratelimited("set kvm_amd.dump_invalid_vmcb=1 to dump internal KVM state.\n");
		return;
	}

4852
	pr_err("VMCB Control Area:\n");
4853 4854 4855 4856 4857 4858 4859
	pr_err("%-20s%04x\n", "cr_read:", control->intercept_cr & 0xffff);
	pr_err("%-20s%04x\n", "cr_write:", control->intercept_cr >> 16);
	pr_err("%-20s%04x\n", "dr_read:", control->intercept_dr & 0xffff);
	pr_err("%-20s%04x\n", "dr_write:", control->intercept_dr >> 16);
	pr_err("%-20s%08x\n", "exceptions:", control->intercept_exceptions);
	pr_err("%-20s%016llx\n", "intercepts:", control->intercept);
	pr_err("%-20s%d\n", "pause filter count:", control->pause_filter_count);
4860 4861
	pr_err("%-20s%d\n", "pause filter threshold:",
	       control->pause_filter_thresh);
4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876
	pr_err("%-20s%016llx\n", "iopm_base_pa:", control->iopm_base_pa);
	pr_err("%-20s%016llx\n", "msrpm_base_pa:", control->msrpm_base_pa);
	pr_err("%-20s%016llx\n", "tsc_offset:", control->tsc_offset);
	pr_err("%-20s%d\n", "asid:", control->asid);
	pr_err("%-20s%d\n", "tlb_ctl:", control->tlb_ctl);
	pr_err("%-20s%08x\n", "int_ctl:", control->int_ctl);
	pr_err("%-20s%08x\n", "int_vector:", control->int_vector);
	pr_err("%-20s%08x\n", "int_state:", control->int_state);
	pr_err("%-20s%08x\n", "exit_code:", control->exit_code);
	pr_err("%-20s%016llx\n", "exit_info1:", control->exit_info_1);
	pr_err("%-20s%016llx\n", "exit_info2:", control->exit_info_2);
	pr_err("%-20s%08x\n", "exit_int_info:", control->exit_int_info);
	pr_err("%-20s%08x\n", "exit_int_info_err:", control->exit_int_info_err);
	pr_err("%-20s%lld\n", "nested_ctl:", control->nested_ctl);
	pr_err("%-20s%016llx\n", "nested_cr3:", control->nested_cr3);
4877
	pr_err("%-20s%016llx\n", "avic_vapic_bar:", control->avic_vapic_bar);
4878 4879
	pr_err("%-20s%08x\n", "event_inj:", control->event_inj);
	pr_err("%-20s%08x\n", "event_inj_err:", control->event_inj_err);
4880
	pr_err("%-20s%lld\n", "virt_ext:", control->virt_ext);
4881
	pr_err("%-20s%016llx\n", "next_rip:", control->next_rip);
4882 4883 4884
	pr_err("%-20s%016llx\n", "avic_backing_page:", control->avic_backing_page);
	pr_err("%-20s%016llx\n", "avic_logical_id:", control->avic_logical_id);
	pr_err("%-20s%016llx\n", "avic_physical_id:", control->avic_physical_id);
4885
	pr_err("VMCB State Save Area:\n");
4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925
	pr_err("%-5s s: %04x a: %04x l: %08x b: %016llx\n",
	       "es:",
	       save->es.selector, save->es.attrib,
	       save->es.limit, save->es.base);
	pr_err("%-5s s: %04x a: %04x l: %08x b: %016llx\n",
	       "cs:",
	       save->cs.selector, save->cs.attrib,
	       save->cs.limit, save->cs.base);
	pr_err("%-5s s: %04x a: %04x l: %08x b: %016llx\n",
	       "ss:",
	       save->ss.selector, save->ss.attrib,
	       save->ss.limit, save->ss.base);
	pr_err("%-5s s: %04x a: %04x l: %08x b: %016llx\n",
	       "ds:",
	       save->ds.selector, save->ds.attrib,
	       save->ds.limit, save->ds.base);
	pr_err("%-5s s: %04x a: %04x l: %08x b: %016llx\n",
	       "fs:",
	       save->fs.selector, save->fs.attrib,
	       save->fs.limit, save->fs.base);
	pr_err("%-5s s: %04x a: %04x l: %08x b: %016llx\n",
	       "gs:",
	       save->gs.selector, save->gs.attrib,
	       save->gs.limit, save->gs.base);
	pr_err("%-5s s: %04x a: %04x l: %08x b: %016llx\n",
	       "gdtr:",
	       save->gdtr.selector, save->gdtr.attrib,
	       save->gdtr.limit, save->gdtr.base);
	pr_err("%-5s s: %04x a: %04x l: %08x b: %016llx\n",
	       "ldtr:",
	       save->ldtr.selector, save->ldtr.attrib,
	       save->ldtr.limit, save->ldtr.base);
	pr_err("%-5s s: %04x a: %04x l: %08x b: %016llx\n",
	       "idtr:",
	       save->idtr.selector, save->idtr.attrib,
	       save->idtr.limit, save->idtr.base);
	pr_err("%-5s s: %04x a: %04x l: %08x b: %016llx\n",
	       "tr:",
	       save->tr.selector, save->tr.attrib,
	       save->tr.limit, save->tr.base);
4926 4927
	pr_err("cpl:            %d                efer:         %016llx\n",
		save->cpl, save->efer);
4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950 4951 4952 4953 4954
	pr_err("%-15s %016llx %-13s %016llx\n",
	       "cr0:", save->cr0, "cr2:", save->cr2);
	pr_err("%-15s %016llx %-13s %016llx\n",
	       "cr3:", save->cr3, "cr4:", save->cr4);
	pr_err("%-15s %016llx %-13s %016llx\n",
	       "dr6:", save->dr6, "dr7:", save->dr7);
	pr_err("%-15s %016llx %-13s %016llx\n",
	       "rip:", save->rip, "rflags:", save->rflags);
	pr_err("%-15s %016llx %-13s %016llx\n",
	       "rsp:", save->rsp, "rax:", save->rax);
	pr_err("%-15s %016llx %-13s %016llx\n",
	       "star:", save->star, "lstar:", save->lstar);
	pr_err("%-15s %016llx %-13s %016llx\n",
	       "cstar:", save->cstar, "sfmask:", save->sfmask);
	pr_err("%-15s %016llx %-13s %016llx\n",
	       "kernel_gs_base:", save->kernel_gs_base,
	       "sysenter_cs:", save->sysenter_cs);
	pr_err("%-15s %016llx %-13s %016llx\n",
	       "sysenter_esp:", save->sysenter_esp,
	       "sysenter_eip:", save->sysenter_eip);
	pr_err("%-15s %016llx %-13s %016llx\n",
	       "gpat:", save->g_pat, "dbgctl:", save->dbgctl);
	pr_err("%-15s %016llx %-13s %016llx\n",
	       "br_from:", save->br_from, "br_to:", save->br_to);
	pr_err("%-15s %016llx %-13s %016llx\n",
	       "excp_from:", save->last_excp_from,
	       "excp_to:", save->last_excp_to);
4955 4956
}

4957 4958 4959 4960 4961 4962 4963 4964
static void svm_get_exit_info(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2)
{
	struct vmcb_control_area *control = &to_svm(vcpu)->vmcb->control;

	*info1 = control->exit_info_1;
	*info2 = control->exit_info_2;
}

4965 4966
static int handle_exit(struct kvm_vcpu *vcpu,
	enum exit_fastpath_completion exit_fastpath)
A
Avi Kivity 已提交
4967
{
4968
	struct vcpu_svm *svm = to_svm(vcpu);
A
Avi Kivity 已提交
4969
	struct kvm_run *kvm_run = vcpu->run;
4970
	u32 exit_code = svm->vmcb->control.exit_code;
A
Avi Kivity 已提交
4971

4972 4973
	trace_kvm_exit(exit_code, vcpu, KVM_ISA_SVM);

4974
	if (!is_cr_intercept(svm, INTERCEPT_CR0_WRITE))
4975 4976 4977
		vcpu->arch.cr0 = svm->vmcb->save.cr0;
	if (npt_enabled)
		vcpu->arch.cr3 = svm->vmcb->save.cr3;
4978

4979 4980 4981 4982 4983 4984 4985
	if (unlikely(svm->nested.exit_required)) {
		nested_svm_vmexit(svm);
		svm->nested.exit_required = false;

		return 1;
	}

4986
	if (is_guest_mode(vcpu)) {
4987 4988
		int vmexit;

4989 4990 4991 4992
		trace_kvm_nested_vmexit(svm->vmcb->save.rip, exit_code,
					svm->vmcb->control.exit_info_1,
					svm->vmcb->control.exit_info_2,
					svm->vmcb->control.exit_int_info,
4993 4994
					svm->vmcb->control.exit_int_info_err,
					KVM_ISA_SVM);
4995

4996 4997 4998 4999 5000 5001
		vmexit = nested_svm_exit_special(svm);

		if (vmexit == NESTED_EXIT_CONTINUE)
			vmexit = nested_svm_exit_handled(svm);

		if (vmexit == NESTED_EXIT_DONE)
5002 5003 5004
			return 1;
	}

5005 5006
	svm_complete_interrupts(svm);

5007 5008 5009 5010
	if (svm->vmcb->control.exit_code == SVM_EXIT_ERR) {
		kvm_run->exit_reason = KVM_EXIT_FAIL_ENTRY;
		kvm_run->fail_entry.hardware_entry_failure_reason
			= svm->vmcb->control.exit_code;
5011
		dump_vmcb(vcpu);
5012 5013 5014
		return 0;
	}

5015
	if (is_external_interrupt(svm->vmcb->control.exit_int_info) &&
5016
	    exit_code != SVM_EXIT_EXCP_BASE + PF_VECTOR &&
5017 5018
	    exit_code != SVM_EXIT_NPF && exit_code != SVM_EXIT_TASK_SWITCH &&
	    exit_code != SVM_EXIT_INTR && exit_code != SVM_EXIT_NMI)
5019
		printk(KERN_ERR "%s: unexpected exit_int_info 0x%x "
A
Avi Kivity 已提交
5020
		       "exit_code 0x%x\n",
5021
		       __func__, svm->vmcb->control.exit_int_info,
A
Avi Kivity 已提交
5022 5023
		       exit_code);

5024 5025 5026 5027
	if (exit_fastpath == EXIT_FASTPATH_SKIP_EMUL_INS) {
		kvm_skip_emulated_instruction(vcpu);
		return 1;
	} else if (exit_code >= ARRAY_SIZE(svm_exit_handlers)
J
Joe Perches 已提交
5028
	    || !svm_exit_handlers[exit_code]) {
5029 5030 5031 5032 5033 5034 5035 5036
		vcpu_unimpl(vcpu, "svm: unexpected exit reason 0x%x\n", exit_code);
		dump_vmcb(vcpu);
		vcpu->run->exit_reason = KVM_EXIT_INTERNAL_ERROR;
		vcpu->run->internal.suberror =
			KVM_INTERNAL_ERROR_UNEXPECTED_EXIT_REASON;
		vcpu->run->internal.ndata = 1;
		vcpu->run->internal.data[0] = exit_code;
		return 0;
A
Avi Kivity 已提交
5037 5038
	}

5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050
#ifdef CONFIG_RETPOLINE
	if (exit_code == SVM_EXIT_MSR)
		return msr_interception(svm);
	else if (exit_code == SVM_EXIT_VINTR)
		return interrupt_window_interception(svm);
	else if (exit_code == SVM_EXIT_INTR)
		return intr_interception(svm);
	else if (exit_code == SVM_EXIT_HLT)
		return halt_interception(svm);
	else if (exit_code == SVM_EXIT_NPF)
		return npf_interception(svm);
#endif
A
Avi Kivity 已提交
5051
	return svm_exit_handlers[exit_code](svm);
A
Avi Kivity 已提交
5052 5053 5054 5055 5056 5057
}

static void reload_tss(struct kvm_vcpu *vcpu)
{
	int cpu = raw_smp_processor_id();

5058 5059
	struct svm_cpu_data *sd = per_cpu(svm_data, cpu);
	sd->tss_desc->type = 9; /* available 32/64-bit TSS */
A
Avi Kivity 已提交
5060 5061 5062
	load_TR_desc();
}

5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086
static void pre_sev_run(struct vcpu_svm *svm, int cpu)
{
	struct svm_cpu_data *sd = per_cpu(svm_data, cpu);
	int asid = sev_get_asid(svm->vcpu.kvm);

	/* Assign the asid allocated with this SEV guest */
	svm->vmcb->control.asid = asid;

	/*
	 * Flush guest TLB:
	 *
	 * 1) when different VMCB for the same ASID is to be run on the same host CPU.
	 * 2) or this VMCB was executed on different host CPU in previous VMRUNs.
	 */
	if (sd->sev_vmcbs[asid] == svm->vmcb &&
	    svm->last_cpu == cpu)
		return;

	svm->last_cpu = cpu;
	sd->sev_vmcbs[asid] = svm->vmcb;
	svm->vmcb->control.tlb_ctl = TLB_CONTROL_FLUSH_ASID;
	mark_dirty(svm->vmcb, VMCB_ASID);
}

R
Rusty Russell 已提交
5087
static void pre_svm_run(struct vcpu_svm *svm)
A
Avi Kivity 已提交
5088 5089 5090
{
	int cpu = raw_smp_processor_id();

5091
	struct svm_cpu_data *sd = per_cpu(svm_data, cpu);
A
Avi Kivity 已提交
5092

5093 5094 5095
	if (sev_guest(svm->vcpu.kvm))
		return pre_sev_run(svm, cpu);

5096
	/* FIXME: handle wraparound of asid_generation */
5097 5098
	if (svm->asid_generation != sd->asid_generation)
		new_asid(svm, sd);
A
Avi Kivity 已提交
5099 5100
}

5101 5102 5103 5104 5105 5106
static void svm_inject_nmi(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);

	svm->vmcb->control.event_inj = SVM_EVTINJ_VALID | SVM_EVTINJ_TYPE_NMI;
	vcpu->arch.hflags |= HF_NMI_MASK;
5107
	set_intercept(svm, INTERCEPT_IRET);
5108 5109
	++vcpu->stat.nmi_injections;
}
A
Avi Kivity 已提交
5110

5111
static inline void svm_inject_irq(struct vcpu_svm *svm, int irq)
A
Avi Kivity 已提交
5112 5113 5114
{
	struct vmcb_control_area *control;

5115
	/* The following fields are ignored when AVIC is enabled */
R
Rusty Russell 已提交
5116
	control = &svm->vmcb->control;
5117
	control->int_vector = irq;
A
Avi Kivity 已提交
5118 5119 5120
	control->int_ctl &= ~V_INTR_PRIO_MASK;
	control->int_ctl |= V_IRQ_MASK |
		((/*control->int_vector >> 4*/ 0xf) << V_INTR_PRIO_SHIFT);
5121
	mark_dirty(svm->vmcb, VMCB_INTR);
A
Avi Kivity 已提交
5122 5123
}

5124
static void svm_set_irq(struct kvm_vcpu *vcpu)
E
Eddie Dong 已提交
5125 5126 5127
{
	struct vcpu_svm *svm = to_svm(vcpu);

5128
	BUG_ON(!(gif_set(svm)));
5129

5130 5131 5132
	trace_kvm_inj_virq(vcpu->arch.interrupt.nr);
	++vcpu->stat.irq_injections;

5133 5134
	svm->vmcb->control.event_inj = vcpu->arch.interrupt.nr |
		SVM_EVTINJ_VALID | SVM_EVTINJ_TYPE_INTR;
E
Eddie Dong 已提交
5135 5136
}

5137 5138 5139 5140 5141
static inline bool svm_nested_virtualize_tpr(struct kvm_vcpu *vcpu)
{
	return is_guest_mode(vcpu) && (vcpu->arch.hflags & HF_VINTR_MASK);
}

5142
static void update_cr8_intercept(struct kvm_vcpu *vcpu, int tpr, int irr)
5143 5144 5145
{
	struct vcpu_svm *svm = to_svm(vcpu);

5146
	if (svm_nested_virtualize_tpr(vcpu))
5147 5148
		return;

5149 5150
	clr_cr_intercept(svm, INTERCEPT_CR8_WRITE);

5151
	if (irr == -1)
5152 5153
		return;

5154
	if (tpr >= irr)
5155
		set_cr_intercept(svm, INTERCEPT_CR8_WRITE);
5156
}
5157

5158
static void svm_set_virtual_apic_mode(struct kvm_vcpu *vcpu)
5159 5160 5161 5162
{
	return;
}

5163 5164
static void svm_hwapic_irr_update(struct kvm_vcpu *vcpu, int max_irr)
{
5165 5166
}

5167
static void svm_hwapic_isr_update(struct kvm_vcpu *vcpu, int max_isr)
5168
{
5169 5170
}

5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181
static void svm_toggle_avic_for_irq_window(struct kvm_vcpu *vcpu, bool activate)
{
	if (!avic || !lapic_in_kernel(vcpu))
		return;

	srcu_read_unlock(&vcpu->kvm->srcu, vcpu->srcu_idx);
	kvm_request_apicv_update(vcpu->kvm, activate,
				 APICV_INHIBIT_REASON_IRQWIN);
	vcpu->srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
}

5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203 5204 5205 5206 5207 5208 5209 5210 5211 5212 5213
static int svm_set_pi_irte_mode(struct kvm_vcpu *vcpu, bool activate)
{
	int ret = 0;
	unsigned long flags;
	struct amd_svm_iommu_ir *ir;
	struct vcpu_svm *svm = to_svm(vcpu);

	if (!kvm_arch_has_assigned_device(vcpu->kvm))
		return 0;

	/*
	 * Here, we go through the per-vcpu ir_list to update all existing
	 * interrupt remapping table entry targeting this vcpu.
	 */
	spin_lock_irqsave(&svm->ir_list_lock, flags);

	if (list_empty(&svm->ir_list))
		goto out;

	list_for_each_entry(ir, &svm->ir_list, node) {
		if (activate)
			ret = amd_iommu_activate_guest_mode(ir->data);
		else
			ret = amd_iommu_deactivate_guest_mode(ir->data);
		if (ret)
			break;
	}
out:
	spin_unlock_irqrestore(&svm->ir_list_lock, flags);
	return ret;
}

5214
static void svm_refresh_apicv_exec_ctrl(struct kvm_vcpu *vcpu)
5215
{
5216 5217
	struct vcpu_svm *svm = to_svm(vcpu);
	struct vmcb *vmcb = svm->vmcb;
5218
	bool activated = kvm_vcpu_apicv_active(vcpu);
5219

5220 5221 5222
	if (!avic)
		return;

5223 5224 5225 5226 5227 5228 5229 5230 5231
	if (activated) {
		/**
		 * During AVIC temporary deactivation, guest could update
		 * APIC ID, DFR and LDR registers, which would not be trapped
		 * by avic_unaccelerated_access_interception(). In this case,
		 * we need to check and update the AVIC logical APIC ID table
		 * accordingly before re-activating.
		 */
		avic_post_state_restore(vcpu);
5232
		vmcb->control.int_ctl |= AVIC_ENABLE_MASK;
5233
	} else {
5234
		vmcb->control.int_ctl &= ~AVIC_ENABLE_MASK;
5235
	}
5236
	mark_dirty(vmcb, VMCB_AVIC);
5237 5238

	svm_set_pi_irte_mode(vcpu, activated);
5239 5240
}

5241
static void svm_load_eoi_exitmap(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap)
5242 5243 5244 5245
{
	return;
}

5246
static int svm_deliver_avic_intr(struct kvm_vcpu *vcpu, int vec)
5247
{
5248 5249 5250
	if (!vcpu->arch.apicv_active)
		return -1;

5251 5252 5253
	kvm_lapic_set_irr(vec, vcpu->arch.apic);
	smp_mb__after_atomic();

5254 5255 5256 5257 5258 5259 5260
	if (avic_vcpu_is_running(vcpu)) {
		int cpuid = vcpu->cpu;

		if (cpuid != get_cpu())
			wrmsrl(SVM_AVIC_DOORBELL, kvm_cpu_get_apicid(cpuid));
		put_cpu();
	} else
5261
		kvm_vcpu_wake_up(vcpu);
5262 5263

	return 0;
5264 5265
}

5266 5267 5268 5269 5270
static bool svm_dy_apicv_has_pending_interrupt(struct kvm_vcpu *vcpu)
{
	return false;
}

5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290 5291 5292 5293 5294 5295 5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316
static void svm_ir_list_del(struct vcpu_svm *svm, struct amd_iommu_pi_data *pi)
{
	unsigned long flags;
	struct amd_svm_iommu_ir *cur;

	spin_lock_irqsave(&svm->ir_list_lock, flags);
	list_for_each_entry(cur, &svm->ir_list, node) {
		if (cur->data != pi->ir_data)
			continue;
		list_del(&cur->node);
		kfree(cur);
		break;
	}
	spin_unlock_irqrestore(&svm->ir_list_lock, flags);
}

static int svm_ir_list_add(struct vcpu_svm *svm, struct amd_iommu_pi_data *pi)
{
	int ret = 0;
	unsigned long flags;
	struct amd_svm_iommu_ir *ir;

	/**
	 * In some cases, the existing irte is updaed and re-set,
	 * so we need to check here if it's already been * added
	 * to the ir_list.
	 */
	if (pi->ir_data && (pi->prev_ga_tag != 0)) {
		struct kvm *kvm = svm->vcpu.kvm;
		u32 vcpu_id = AVIC_GATAG_TO_VCPUID(pi->prev_ga_tag);
		struct kvm_vcpu *prev_vcpu = kvm_get_vcpu_by_id(kvm, vcpu_id);
		struct vcpu_svm *prev_svm;

		if (!prev_vcpu) {
			ret = -EINVAL;
			goto out;
		}

		prev_svm = to_svm(prev_vcpu);
		svm_ir_list_del(prev_svm, pi);
	}

	/**
	 * Allocating new amd_iommu_pi_data, which will get
	 * add to the per-vcpu ir_list.
	 */
5317
	ir = kzalloc(sizeof(struct amd_svm_iommu_ir), GFP_KERNEL_ACCOUNT);
5318 5319 5320 5321 5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350
	if (!ir) {
		ret = -ENOMEM;
		goto out;
	}
	ir->data = pi->ir_data;

	spin_lock_irqsave(&svm->ir_list_lock, flags);
	list_add(&ir->node, &svm->ir_list);
	spin_unlock_irqrestore(&svm->ir_list_lock, flags);
out:
	return ret;
}

/**
 * Note:
 * The HW cannot support posting multicast/broadcast
 * interrupts to a vCPU. So, we still use legacy interrupt
 * remapping for these kind of interrupts.
 *
 * For lowest-priority interrupts, we only support
 * those with single CPU as the destination, e.g. user
 * configures the interrupts via /proc/irq or uses
 * irqbalance to make the interrupts single-CPU.
 */
static int
get_pi_vcpu_info(struct kvm *kvm, struct kvm_kernel_irq_routing_entry *e,
		 struct vcpu_data *vcpu_info, struct vcpu_svm **svm)
{
	struct kvm_lapic_irq irq;
	struct kvm_vcpu *vcpu = NULL;

	kvm_set_msi_irq(kvm, e, &irq);

5351 5352
	if (!kvm_intr_is_single_vcpu(kvm, &irq, &vcpu) ||
	    !kvm_irq_is_postable(&irq)) {
5353 5354 5355 5356 5357 5358 5359 5360
		pr_debug("SVM: %s: use legacy intr remap mode for irq %u\n",
			 __func__, irq.vector);
		return -1;
	}

	pr_debug("SVM: %s: use GA mode for irq %u\n", __func__,
		 irq.vector);
	*svm = to_svm(vcpu);
5361
	vcpu_info->pi_desc_addr = __sme_set(page_to_phys((*svm)->avic_backing_page));
5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405
	vcpu_info->vector = irq.vector;

	return 0;
}

/*
 * svm_update_pi_irte - set IRTE for Posted-Interrupts
 *
 * @kvm: kvm
 * @host_irq: host irq of the interrupt
 * @guest_irq: gsi of the interrupt
 * @set: set or unset PI
 * returns 0 on success, < 0 on failure
 */
static int svm_update_pi_irte(struct kvm *kvm, unsigned int host_irq,
			      uint32_t guest_irq, bool set)
{
	struct kvm_kernel_irq_routing_entry *e;
	struct kvm_irq_routing_table *irq_rt;
	int idx, ret = -EINVAL;

	if (!kvm_arch_has_assigned_device(kvm) ||
	    !irq_remapping_cap(IRQ_POSTING_CAP))
		return 0;

	pr_debug("SVM: %s: host_irq=%#x, guest_irq=%#x, set=%#x\n",
		 __func__, host_irq, guest_irq, set);

	idx = srcu_read_lock(&kvm->irq_srcu);
	irq_rt = srcu_dereference(kvm->irq_routing, &kvm->irq_srcu);
	WARN_ON(guest_irq >= irq_rt->nr_rt_entries);

	hlist_for_each_entry(e, &irq_rt->map[guest_irq], link) {
		struct vcpu_data vcpu_info;
		struct vcpu_svm *svm = NULL;

		if (e->type != KVM_IRQ_ROUTING_MSI)
			continue;

		/**
		 * Here, we setup with legacy mode in the following cases:
		 * 1. When cannot target interrupt to a specific vcpu.
		 * 2. Unsetting posted interrupt.
		 * 3. APIC virtialization is disabled for the vcpu.
5406
		 * 4. IRQ has incompatible delivery mode (SMI, INIT, etc)
5407 5408 5409 5410 5411 5412
		 */
		if (!get_pi_vcpu_info(kvm, e, &vcpu_info, &svm) && set &&
		    kvm_vcpu_apicv_active(&svm->vcpu)) {
			struct amd_iommu_pi_data pi;

			/* Try to enable guest_mode in IRTE */
5413 5414
			pi.base = __sme_set(page_to_phys(svm->avic_backing_page) &
					    AVIC_HPA_MASK);
5415
			pi.ga_tag = AVIC_GATAG(to_kvm_svm(kvm)->avic_vm_id,
5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453 5454 5455 5456 5457 5458
						     svm->vcpu.vcpu_id);
			pi.is_guest_mode = true;
			pi.vcpu_data = &vcpu_info;
			ret = irq_set_vcpu_affinity(host_irq, &pi);

			/**
			 * Here, we successfully setting up vcpu affinity in
			 * IOMMU guest mode. Now, we need to store the posted
			 * interrupt information in a per-vcpu ir_list so that
			 * we can reference to them directly when we update vcpu
			 * scheduling information in IOMMU irte.
			 */
			if (!ret && pi.is_guest_mode)
				svm_ir_list_add(svm, &pi);
		} else {
			/* Use legacy mode in IRTE */
			struct amd_iommu_pi_data pi;

			/**
			 * Here, pi is used to:
			 * - Tell IOMMU to use legacy mode for this interrupt.
			 * - Retrieve ga_tag of prior interrupt remapping data.
			 */
			pi.is_guest_mode = false;
			ret = irq_set_vcpu_affinity(host_irq, &pi);

			/**
			 * Check if the posted interrupt was previously
			 * setup with the guest_mode by checking if the ga_tag
			 * was cached. If so, we need to clean up the per-vcpu
			 * ir_list.
			 */
			if (!ret && pi.prev_ga_tag) {
				int id = AVIC_GATAG_TO_VCPUID(pi.prev_ga_tag);
				struct kvm_vcpu *vcpu;

				vcpu = kvm_get_vcpu_by_id(kvm, id);
				if (vcpu)
					svm_ir_list_del(to_svm(vcpu), &pi);
			}
		}

		if (!ret && svm) {
5459 5460
			trace_kvm_pi_irte_update(host_irq, svm->vcpu.vcpu_id,
						 e->gsi, vcpu_info.vector,
5461 5462 5463 5464 5465 5466 5467 5468 5469 5470 5471 5472 5473 5474 5475
						 vcpu_info.pi_desc_addr, set);
		}

		if (ret < 0) {
			pr_err("%s: failed to update PI IRTE\n", __func__);
			goto out;
		}
	}

	ret = 0;
out:
	srcu_read_unlock(&kvm->irq_srcu, idx);
	return ret;
}

5476 5477 5478 5479
static int svm_nmi_allowed(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);
	struct vmcb *vmcb = svm->vmcb;
J
Joerg Roedel 已提交
5480 5481 5482 5483 5484 5485
	int ret;
	ret = !(vmcb->control.int_state & SVM_INTERRUPT_SHADOW_MASK) &&
	      !(svm->vcpu.arch.hflags & HF_NMI_MASK);
	ret = ret && gif_set(svm) && nested_svm_nmi(svm);

	return ret;
5486 5487
}

J
Jan Kiszka 已提交
5488 5489 5490 5491 5492 5493 5494 5495 5496 5497 5498 5499 5500
static bool svm_get_nmi_mask(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);

	return !!(svm->vcpu.arch.hflags & HF_NMI_MASK);
}

static void svm_set_nmi_mask(struct kvm_vcpu *vcpu, bool masked)
{
	struct vcpu_svm *svm = to_svm(vcpu);

	if (masked) {
		svm->vcpu.arch.hflags |= HF_NMI_MASK;
5501
		set_intercept(svm, INTERCEPT_IRET);
J
Jan Kiszka 已提交
5502 5503
	} else {
		svm->vcpu.arch.hflags &= ~HF_NMI_MASK;
5504
		clr_intercept(svm, INTERCEPT_IRET);
J
Jan Kiszka 已提交
5505 5506 5507
	}
}

5508 5509 5510 5511
static int svm_interrupt_allowed(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);
	struct vmcb *vmcb = svm->vmcb;
5512 5513 5514 5515 5516 5517
	int ret;

	if (!gif_set(svm) ||
	     (vmcb->control.int_state & SVM_INTERRUPT_SHADOW_MASK))
		return 0;

5518
	ret = !!(kvm_get_rflags(vcpu) & X86_EFLAGS_IF);
5519

5520
	if (is_guest_mode(vcpu))
5521 5522 5523
		return ret && !(svm->vcpu.arch.hflags & HF_VINTR_MASK);

	return ret;
5524 5525
}

5526
static void enable_irq_window(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
5527
{
5528 5529
	struct vcpu_svm *svm = to_svm(vcpu);

J
Joerg Roedel 已提交
5530 5531 5532 5533
	/*
	 * In case GIF=0 we can't rely on the CPU to tell us when GIF becomes
	 * 1, because that's a separate STGI/VMRUN intercept.  The next time we
	 * get that intercept, this function will be called again though and
5534 5535 5536
	 * we'll get the vintr intercept. However, if the vGIF feature is
	 * enabled, the STGI interception will not occur. Enable the irq
	 * window under the assumption that the hardware will set the GIF.
J
Joerg Roedel 已提交
5537
	 */
5538
	if ((vgif_enabled(svm) || gif_set(svm)) && nested_svm_intr(svm)) {
5539 5540 5541 5542 5543 5544 5545
		/*
		 * IRQ window is not needed when AVIC is enabled,
		 * unless we have pending ExtINT since it cannot be injected
		 * via AVIC. In such case, we need to temporarily disable AVIC,
		 * and fallback to injecting IRQ via V_IRQ.
		 */
		svm_toggle_avic_for_irq_window(vcpu, false);
5546 5547 5548
		svm_set_vintr(svm);
		svm_inject_irq(svm, 0x0);
	}
5549 5550
}

5551
static void enable_nmi_window(struct kvm_vcpu *vcpu)
5552
{
5553
	struct vcpu_svm *svm = to_svm(vcpu);
5554

5555 5556
	if ((svm->vcpu.arch.hflags & (HF_NMI_MASK | HF_IRET_MASK))
	    == HF_NMI_MASK)
5557
		return; /* IRET will cause a vm exit */
5558

5559 5560 5561
	if (!gif_set(svm)) {
		if (vgif_enabled(svm))
			set_intercept(svm, INTERCEPT_STGI);
5562
		return; /* STGI will cause a vm exit */
5563
	}
5564 5565 5566 5567

	if (svm->nested.exit_required)
		return; /* we're not going to run the guest yet */

J
Joerg Roedel 已提交
5568 5569 5570 5571
	/*
	 * Something prevents NMI from been injected. Single step over possible
	 * problem (IRET or exception injection or interrupt shadow)
	 */
5572
	svm->nmi_singlestep_guest_rflags = svm_get_rflags(vcpu);
J
Jan Kiszka 已提交
5573
	svm->nmi_singlestep = true;
5574
	svm->vmcb->save.rflags |= (X86_EFLAGS_TF | X86_EFLAGS_RF);
5575 5576
}

5577 5578 5579 5580 5581
static int svm_set_tss_addr(struct kvm *kvm, unsigned int addr)
{
	return 0;
}

5582 5583 5584 5585 5586
static int svm_set_identity_map_addr(struct kvm *kvm, u64 ident_addr)
{
	return 0;
}

5587
static void svm_flush_tlb(struct kvm_vcpu *vcpu, bool invalidate_gpa)
5588
{
5589 5590 5591 5592 5593 5594
	struct vcpu_svm *svm = to_svm(vcpu);

	if (static_cpu_has(X86_FEATURE_FLUSHBYASID))
		svm->vmcb->control.tlb_ctl = TLB_CONTROL_FLUSH_ASID;
	else
		svm->asid_generation--;
5595 5596
}

5597 5598 5599 5600 5601 5602 5603
static void svm_flush_tlb_gva(struct kvm_vcpu *vcpu, gva_t gva)
{
	struct vcpu_svm *svm = to_svm(vcpu);

	invlpga(gva, svm->vmcb->control.asid);
}

5604 5605 5606 5607
static void svm_prepare_guest_switch(struct kvm_vcpu *vcpu)
{
}

5608 5609 5610 5611
static inline void sync_cr8_to_lapic(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);

5612
	if (svm_nested_virtualize_tpr(vcpu))
5613 5614
		return;

5615
	if (!is_cr_intercept(svm, INTERCEPT_CR8_WRITE)) {
5616
		int cr8 = svm->vmcb->control.int_ctl & V_TPR_MASK;
5617
		kvm_set_cr8(vcpu, cr8);
5618 5619 5620
	}
}

5621 5622 5623 5624 5625
static inline void sync_lapic_to_cr8(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);
	u64 cr8;

5626 5627
	if (svm_nested_virtualize_tpr(vcpu) ||
	    kvm_vcpu_apicv_active(vcpu))
5628 5629
		return;

5630 5631 5632 5633 5634
	cr8 = kvm_get_cr8(vcpu);
	svm->vmcb->control.int_ctl &= ~V_TPR_MASK;
	svm->vmcb->control.int_ctl |= cr8 & V_TPR_MASK;
}

5635 5636 5637 5638 5639
static void svm_complete_interrupts(struct vcpu_svm *svm)
{
	u8 vector;
	int type;
	u32 exitintinfo = svm->vmcb->control.exit_int_info;
5640 5641 5642
	unsigned int3_injected = svm->int3_injected;

	svm->int3_injected = 0;
5643

5644 5645 5646 5647 5648 5649
	/*
	 * If we've made progress since setting HF_IRET_MASK, we've
	 * executed an IRET and can allow NMI injection.
	 */
	if ((svm->vcpu.arch.hflags & HF_IRET_MASK)
	    && kvm_rip_read(&svm->vcpu) != svm->nmi_iret_rip) {
5650
		svm->vcpu.arch.hflags &= ~(HF_NMI_MASK | HF_IRET_MASK);
5651 5652
		kvm_make_request(KVM_REQ_EVENT, &svm->vcpu);
	}
5653

5654 5655 5656 5657 5658 5659 5660
	svm->vcpu.arch.nmi_injected = false;
	kvm_clear_exception_queue(&svm->vcpu);
	kvm_clear_interrupt_queue(&svm->vcpu);

	if (!(exitintinfo & SVM_EXITINTINFO_VALID))
		return;

5661 5662
	kvm_make_request(KVM_REQ_EVENT, &svm->vcpu);

5663 5664 5665 5666 5667 5668 5669 5670
	vector = exitintinfo & SVM_EXITINTINFO_VEC_MASK;
	type = exitintinfo & SVM_EXITINTINFO_TYPE_MASK;

	switch (type) {
	case SVM_EXITINTINFO_TYPE_NMI:
		svm->vcpu.arch.nmi_injected = true;
		break;
	case SVM_EXITINTINFO_TYPE_EXEPT:
5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681
		/*
		 * In case of software exceptions, do not reinject the vector,
		 * but re-execute the instruction instead. Rewind RIP first
		 * if we emulated INT3 before.
		 */
		if (kvm_exception_is_soft(vector)) {
			if (vector == BP_VECTOR && int3_injected &&
			    kvm_is_linear_rip(&svm->vcpu, svm->int3_rip))
				kvm_rip_write(&svm->vcpu,
					      kvm_rip_read(&svm->vcpu) -
					      int3_injected);
5682
			break;
5683
		}
5684 5685
		if (exitintinfo & SVM_EXITINTINFO_VALID_ERR) {
			u32 err = svm->vmcb->control.exit_int_info_err;
5686
			kvm_requeue_exception_e(&svm->vcpu, vector, err);
5687 5688

		} else
5689
			kvm_requeue_exception(&svm->vcpu, vector);
5690 5691
		break;
	case SVM_EXITINTINFO_TYPE_INTR:
5692
		kvm_queue_interrupt(&svm->vcpu, vector, false);
5693 5694 5695 5696 5697 5698
		break;
	default:
		break;
	}
}

A
Avi Kivity 已提交
5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709
static void svm_cancel_injection(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);
	struct vmcb_control_area *control = &svm->vmcb->control;

	control->exit_int_info = control->event_inj;
	control->exit_int_info_err = control->event_inj_err;
	control->event_inj = 0;
	svm_complete_interrupts(svm);
}

A
Avi Kivity 已提交
5710
static void svm_vcpu_run(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
5711
{
5712
	struct vcpu_svm *svm = to_svm(vcpu);
5713

5714 5715 5716 5717
	svm->vmcb->save.rax = vcpu->arch.regs[VCPU_REGS_RAX];
	svm->vmcb->save.rsp = vcpu->arch.regs[VCPU_REGS_RSP];
	svm->vmcb->save.rip = vcpu->arch.regs[VCPU_REGS_RIP];

5718 5719 5720 5721 5722 5723 5724
	/*
	 * A vmexit emulation is required before the vcpu can be executed
	 * again.
	 */
	if (unlikely(svm->nested.exit_required))
		return;

5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735 5736 5737 5738 5739 5740
	/*
	 * Disable singlestep if we're injecting an interrupt/exception.
	 * We don't want our modified rflags to be pushed on the stack where
	 * we might not be able to easily reset them if we disabled NMI
	 * singlestep later.
	 */
	if (svm->nmi_singlestep && svm->vmcb->control.event_inj) {
		/*
		 * Event injection happens before external interrupts cause a
		 * vmexit and interrupts are disabled here, so smp_send_reschedule
		 * is enough to force an immediate vmexit.
		 */
		disable_nmi_singlestep(svm);
		smp_send_reschedule(vcpu->cpu);
	}

R
Rusty Russell 已提交
5741
	pre_svm_run(svm);
A
Avi Kivity 已提交
5742

5743 5744
	sync_lapic_to_cr8(vcpu);

5745
	svm->vmcb->save.cr2 = vcpu->arch.cr2;
A
Avi Kivity 已提交
5746

5747
	clgi();
5748
	kvm_load_guest_xsave_state(vcpu);
5749

5750 5751 5752 5753
	if (lapic_in_kernel(vcpu) &&
		vcpu->arch.apic->lapic_timer.timer_advance_ns)
		kvm_wait_lapic_expire(vcpu);

5754 5755 5756 5757 5758 5759
	/*
	 * If this vCPU has touched SPEC_CTRL, restore the guest's value if
	 * it's non-zero. Since vmentry is serialising on affected CPUs, there
	 * is no need to worry about the conditional branch over the wrmsr
	 * being speculatively taken.
	 */
5760
	x86_spec_ctrl_set_guest(svm->spec_ctrl, svm->virt_spec_ctrl);
5761

5762 5763
	local_irq_enable();

A
Avi Kivity 已提交
5764
	asm volatile (
A
Avi Kivity 已提交
5765 5766 5767 5768 5769 5770 5771
		"push %%" _ASM_BP "; \n\t"
		"mov %c[rbx](%[svm]), %%" _ASM_BX " \n\t"
		"mov %c[rcx](%[svm]), %%" _ASM_CX " \n\t"
		"mov %c[rdx](%[svm]), %%" _ASM_DX " \n\t"
		"mov %c[rsi](%[svm]), %%" _ASM_SI " \n\t"
		"mov %c[rdi](%[svm]), %%" _ASM_DI " \n\t"
		"mov %c[rbp](%[svm]), %%" _ASM_BP " \n\t"
5772
#ifdef CONFIG_X86_64
R
Rusty Russell 已提交
5773 5774 5775 5776 5777 5778 5779 5780
		"mov %c[r8](%[svm]),  %%r8  \n\t"
		"mov %c[r9](%[svm]),  %%r9  \n\t"
		"mov %c[r10](%[svm]), %%r10 \n\t"
		"mov %c[r11](%[svm]), %%r11 \n\t"
		"mov %c[r12](%[svm]), %%r12 \n\t"
		"mov %c[r13](%[svm]), %%r13 \n\t"
		"mov %c[r14](%[svm]), %%r14 \n\t"
		"mov %c[r15](%[svm]), %%r15 \n\t"
A
Avi Kivity 已提交
5781 5782 5783
#endif

		/* Enter guest mode */
A
Avi Kivity 已提交
5784 5785
		"push %%" _ASM_AX " \n\t"
		"mov %c[vmcb](%[svm]), %%" _ASM_AX " \n\t"
5786 5787 5788
		__ex("vmload %%" _ASM_AX) "\n\t"
		__ex("vmrun %%" _ASM_AX) "\n\t"
		__ex("vmsave %%" _ASM_AX) "\n\t"
A
Avi Kivity 已提交
5789
		"pop %%" _ASM_AX " \n\t"
A
Avi Kivity 已提交
5790 5791

		/* Save guest registers, load host registers */
A
Avi Kivity 已提交
5792 5793 5794 5795 5796 5797
		"mov %%" _ASM_BX ", %c[rbx](%[svm]) \n\t"
		"mov %%" _ASM_CX ", %c[rcx](%[svm]) \n\t"
		"mov %%" _ASM_DX ", %c[rdx](%[svm]) \n\t"
		"mov %%" _ASM_SI ", %c[rsi](%[svm]) \n\t"
		"mov %%" _ASM_DI ", %c[rdi](%[svm]) \n\t"
		"mov %%" _ASM_BP ", %c[rbp](%[svm]) \n\t"
5798
#ifdef CONFIG_X86_64
R
Rusty Russell 已提交
5799 5800 5801 5802 5803 5804 5805 5806
		"mov %%r8,  %c[r8](%[svm]) \n\t"
		"mov %%r9,  %c[r9](%[svm]) \n\t"
		"mov %%r10, %c[r10](%[svm]) \n\t"
		"mov %%r11, %c[r11](%[svm]) \n\t"
		"mov %%r12, %c[r12](%[svm]) \n\t"
		"mov %%r13, %c[r13](%[svm]) \n\t"
		"mov %%r14, %c[r14](%[svm]) \n\t"
		"mov %%r15, %c[r15](%[svm]) \n\t"
5807 5808 5809 5810
		/*
		* Clear host registers marked as clobbered to prevent
		* speculative use.
		*/
5811 5812 5813 5814 5815 5816 5817 5818
		"xor %%r8d, %%r8d \n\t"
		"xor %%r9d, %%r9d \n\t"
		"xor %%r10d, %%r10d \n\t"
		"xor %%r11d, %%r11d \n\t"
		"xor %%r12d, %%r12d \n\t"
		"xor %%r13d, %%r13d \n\t"
		"xor %%r14d, %%r14d \n\t"
		"xor %%r15d, %%r15d \n\t"
A
Avi Kivity 已提交
5819
#endif
5820 5821 5822 5823 5824
		"xor %%ebx, %%ebx \n\t"
		"xor %%ecx, %%ecx \n\t"
		"xor %%edx, %%edx \n\t"
		"xor %%esi, %%esi \n\t"
		"xor %%edi, %%edi \n\t"
A
Avi Kivity 已提交
5825
		"pop %%" _ASM_BP
A
Avi Kivity 已提交
5826
		:
R
Rusty Russell 已提交
5827
		: [svm]"a"(svm),
A
Avi Kivity 已提交
5828
		  [vmcb]"i"(offsetof(struct vcpu_svm, vmcb_pa)),
5829 5830 5831 5832 5833 5834
		  [rbx]"i"(offsetof(struct vcpu_svm, vcpu.arch.regs[VCPU_REGS_RBX])),
		  [rcx]"i"(offsetof(struct vcpu_svm, vcpu.arch.regs[VCPU_REGS_RCX])),
		  [rdx]"i"(offsetof(struct vcpu_svm, vcpu.arch.regs[VCPU_REGS_RDX])),
		  [rsi]"i"(offsetof(struct vcpu_svm, vcpu.arch.regs[VCPU_REGS_RSI])),
		  [rdi]"i"(offsetof(struct vcpu_svm, vcpu.arch.regs[VCPU_REGS_RDI])),
		  [rbp]"i"(offsetof(struct vcpu_svm, vcpu.arch.regs[VCPU_REGS_RBP]))
5835
#ifdef CONFIG_X86_64
5836 5837 5838 5839 5840 5841 5842 5843
		  , [r8]"i"(offsetof(struct vcpu_svm, vcpu.arch.regs[VCPU_REGS_R8])),
		  [r9]"i"(offsetof(struct vcpu_svm, vcpu.arch.regs[VCPU_REGS_R9])),
		  [r10]"i"(offsetof(struct vcpu_svm, vcpu.arch.regs[VCPU_REGS_R10])),
		  [r11]"i"(offsetof(struct vcpu_svm, vcpu.arch.regs[VCPU_REGS_R11])),
		  [r12]"i"(offsetof(struct vcpu_svm, vcpu.arch.regs[VCPU_REGS_R12])),
		  [r13]"i"(offsetof(struct vcpu_svm, vcpu.arch.regs[VCPU_REGS_R13])),
		  [r14]"i"(offsetof(struct vcpu_svm, vcpu.arch.regs[VCPU_REGS_R14])),
		  [r15]"i"(offsetof(struct vcpu_svm, vcpu.arch.regs[VCPU_REGS_R15]))
A
Avi Kivity 已提交
5844
#endif
5845 5846
		: "cc", "memory"
#ifdef CONFIG_X86_64
A
Avi Kivity 已提交
5847
		, "rbx", "rcx", "rdx", "rsi", "rdi"
5848
		, "r8", "r9", "r10", "r11" , "r12", "r13", "r14", "r15"
A
Avi Kivity 已提交
5849 5850
#else
		, "ebx", "ecx", "edx", "esi", "edi"
5851 5852
#endif
		);
A
Avi Kivity 已提交
5853

5854 5855 5856 5857 5858 5859 5860 5861 5862 5863 5864 5865
	/* Eliminate branch target predictions from guest mode */
	vmexit_fill_RSB();

#ifdef CONFIG_X86_64
	wrmsrl(MSR_GS_BASE, svm->host.gs_base);
#else
	loadsegment(fs, svm->host.fs);
#ifndef CONFIG_X86_32_LAZY_GS
	loadsegment(gs, svm->host.gs);
#endif
#endif

5866 5867 5868 5869 5870 5871 5872 5873 5874 5875 5876 5877 5878 5879 5880
	/*
	 * We do not use IBRS in the kernel. If this vCPU has used the
	 * SPEC_CTRL MSR it may have left it on; save the value and
	 * turn it off. This is much more efficient than blindly adding
	 * it to the atomic save/restore list. Especially as the former
	 * (Saving guest MSRs on vmexit) doesn't even exist in KVM.
	 *
	 * For non-nested case:
	 * If the L01 MSR bitmap does not intercept the MSR, then we need to
	 * save it.
	 *
	 * For nested case:
	 * If the L02 MSR bitmap does not intercept the MSR, then we need to
	 * save it.
	 */
5881
	if (unlikely(!msr_write_intercepted(vcpu, MSR_IA32_SPEC_CTRL)))
5882
		svm->spec_ctrl = native_read_msr(MSR_IA32_SPEC_CTRL);
5883

A
Avi Kivity 已提交
5884 5885
	reload_tss(vcpu);

5886 5887
	local_irq_disable();

5888 5889
	x86_spec_ctrl_restore_host(svm->spec_ctrl, svm->virt_spec_ctrl);

5890 5891 5892 5893 5894
	vcpu->arch.cr2 = svm->vmcb->save.cr2;
	vcpu->arch.regs[VCPU_REGS_RAX] = svm->vmcb->save.rax;
	vcpu->arch.regs[VCPU_REGS_RSP] = svm->vmcb->save.rsp;
	vcpu->arch.regs[VCPU_REGS_RIP] = svm->vmcb->save.rip;

5895
	if (unlikely(svm->vmcb->control.exit_code == SVM_EXIT_NMI))
5896
		kvm_before_interrupt(&svm->vcpu);
5897

5898
	kvm_load_host_xsave_state(vcpu);
5899 5900 5901 5902 5903
	stgi();

	/* Any pending NMI will happen here */

	if (unlikely(svm->vmcb->control.exit_code == SVM_EXIT_NMI))
5904
		kvm_after_interrupt(&svm->vcpu);
5905

5906 5907
	sync_cr8_to_lapic(vcpu);

5908
	svm->next_rip = 0;
5909

5910 5911
	svm->vmcb->control.tlb_ctl = TLB_CONTROL_DO_NOTHING;

G
Gleb Natapov 已提交
5912 5913
	/* if exit due to PF check for async PF */
	if (svm->vmcb->control.exit_code == SVM_EXIT_EXCP_BASE + PF_VECTOR)
5914
		svm->vcpu.arch.apf.host_apf_reason = kvm_read_and_reset_pf_reason();
G
Gleb Natapov 已提交
5915

A
Avi Kivity 已提交
5916 5917 5918 5919
	if (npt_enabled) {
		vcpu->arch.regs_avail &= ~(1 << VCPU_EXREG_PDPTR);
		vcpu->arch.regs_dirty &= ~(1 << VCPU_EXREG_PDPTR);
	}
5920 5921 5922 5923 5924 5925 5926 5927

	/*
	 * We need to handle MC intercepts here before the vcpu has a chance to
	 * change the physical cpu
	 */
	if (unlikely(svm->vmcb->control.exit_code ==
		     SVM_EXIT_EXCP_BASE + MC_VECTOR))
		svm_handle_mce(svm);
5928 5929

	mark_all_clean(svm->vmcb);
A
Avi Kivity 已提交
5930
}
5931
STACK_FRAME_NON_STANDARD(svm_vcpu_run);
A
Avi Kivity 已提交
5932 5933 5934

static void svm_set_cr3(struct kvm_vcpu *vcpu, unsigned long root)
{
5935 5936
	struct vcpu_svm *svm = to_svm(vcpu);

5937
	svm->vmcb->save.cr3 = __sme_set(root);
5938
	mark_dirty(svm->vmcb, VMCB_CR);
A
Avi Kivity 已提交
5939 5940
}

5941 5942 5943 5944
static void set_tdp_cr3(struct kvm_vcpu *vcpu, unsigned long root)
{
	struct vcpu_svm *svm = to_svm(vcpu);

5945
	svm->vmcb->control.nested_cr3 = __sme_set(root);
5946
	mark_dirty(svm->vmcb, VMCB_NPT);
5947 5948

	/* Also sync guest cr3 here in case we live migrate */
5949
	svm->vmcb->save.cr3 = kvm_read_cr3(vcpu);
5950
	mark_dirty(svm->vmcb, VMCB_CR);
5951 5952
}

A
Avi Kivity 已提交
5953 5954
static int is_disabled(void)
{
5955 5956 5957 5958 5959 5960
	u64 vm_cr;

	rdmsrl(MSR_VM_CR, vm_cr);
	if (vm_cr & (1 << SVM_VM_CR_SVM_DISABLE))
		return 1;

A
Avi Kivity 已提交
5961 5962 5963
	return 0;
}

I
Ingo Molnar 已提交
5964 5965 5966 5967 5968 5969 5970 5971 5972 5973 5974
static void
svm_patch_hypercall(struct kvm_vcpu *vcpu, unsigned char *hypercall)
{
	/*
	 * Patch in the VMMCALL instruction:
	 */
	hypercall[0] = 0x0f;
	hypercall[1] = 0x01;
	hypercall[2] = 0xd9;
}

5975
static int __init svm_check_processor_compat(void)
Y
Yang, Sheng 已提交
5976
{
5977
	return 0;
Y
Yang, Sheng 已提交
5978 5979
}

5980 5981 5982 5983 5984
static bool svm_cpu_has_accelerated_tpr(void)
{
	return false;
}

5985
static bool svm_has_emulated_msr(int index)
5986
{
5987 5988
	switch (index) {
	case MSR_IA32_MCG_EXT_CTL:
5989
	case MSR_IA32_VMX_BASIC ... MSR_IA32_VMX_VMFUNC:
5990 5991 5992 5993 5994
		return false;
	default:
		break;
	}

5995 5996 5997
	return true;
}

5998 5999 6000 6001 6002
static u64 svm_get_mt_mask(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio)
{
	return 0;
}

6003 6004
static void svm_cpuid_update(struct kvm_vcpu *vcpu)
{
6005 6006
	struct vcpu_svm *svm = to_svm(vcpu);

6007
	vcpu->arch.xsaves_enabled = guest_cpuid_has(vcpu, X86_FEATURE_XSAVE) &&
6008
				    boot_cpu_has(X86_FEATURE_XSAVE) &&
6009 6010
				    boot_cpu_has(X86_FEATURE_XSAVES);

6011
	/* Update nrips enabled cache */
6012
	svm->nrips_enabled = !!guest_cpuid_has(&svm->vcpu, X86_FEATURE_NRIPS);
6013 6014 6015 6016

	if (!kvm_vcpu_apicv_active(vcpu))
		return;

6017 6018 6019 6020 6021 6022 6023
	/*
	 * AVIC does not work with an x2APIC mode guest. If the X2APIC feature
	 * is exposed to the guest, disable AVIC.
	 */
	if (guest_cpuid_has(vcpu, X86_FEATURE_X2APIC))
		kvm_request_apicv_update(vcpu->kvm, false,
					 APICV_INHIBIT_REASON_X2APIC);
6024 6025 6026 6027 6028 6029 6030 6031

	/*
	 * Currently, AVIC does not work with nested virtualization.
	 * So, we disable AVIC when cpuid for SVM is set in the L1 guest.
	 */
	if (nested && guest_cpuid_has(vcpu, X86_FEATURE_SVM))
		kvm_request_apicv_update(vcpu->kvm, false,
					 APICV_INHIBIT_REASON_NESTED);
6032 6033
}

6034
#define F feature_bit
6035

6036 6037
static void svm_set_supported_cpuid(u32 func, struct kvm_cpuid_entry2 *entry)
{
6038
	switch (func) {
6039 6040 6041 6042
	case 0x80000001:
		if (nested)
			entry->ecx |= (1 << 2); /* Set SVM bit */
		break;
6043 6044 6045 6046 6047
	case 0x80000008:
		if (boot_cpu_has(X86_FEATURE_LS_CFG_SSBD) ||
		     boot_cpu_has(X86_FEATURE_AMD_SSBD))
			entry->ebx |= F(VIRT_SSBD);
		break;
6048 6049 6050 6051 6052
	case 0x8000000A:
		entry->eax = 1; /* SVM revision 1 */
		entry->ebx = 8; /* Lets support 8 ASIDs in case we add proper
				   ASID emulation to nested SVM */
		entry->ecx = 0; /* Reserved */
6053 6054 6055 6056
		entry->edx = 0; /* Per default do not support any
				   additional features */

		/* Support next_rip if host supports it */
6057
		if (boot_cpu_has(X86_FEATURE_NRIPS))
6058
			entry->edx |= F(NRIPS);
6059

6060 6061
		/* Support NPT for the guest if enabled */
		if (npt_enabled)
6062
			entry->edx |= F(NPT);
6063

6064
	}
6065 6066
}

6067
static int svm_get_lpage_level(void)
6068
{
6069
	return PT_PDPE_LEVEL;
6070 6071
}

6072 6073
static bool svm_rdtscp_supported(void)
{
P
Paolo Bonzini 已提交
6074
	return boot_cpu_has(X86_FEATURE_RDTSCP);
6075 6076
}

6077 6078 6079 6080 6081
static bool svm_invpcid_supported(void)
{
	return false;
}

6082 6083 6084 6085 6086
static bool svm_mpx_supported(void)
{
	return false;
}

6087 6088
static bool svm_xsaves_supported(void)
{
6089
	return boot_cpu_has(X86_FEATURE_XSAVES);
6090 6091
}

6092 6093 6094 6095 6096
static bool svm_umip_emulated(void)
{
	return false;
}

6097 6098 6099 6100 6101
static bool svm_pt_supported(void)
{
	return false;
}

6102 6103 6104 6105 6106
static bool svm_has_wbinvd_exit(void)
{
	return true;
}

6107 6108 6109 6110 6111
static bool svm_pku_supported(void)
{
	return false;
}

6112
#define PRE_EX(exit)  { .exit_code = (exit), \
6113
			.stage = X86_ICPT_PRE_EXCEPT, }
6114
#define POST_EX(exit) { .exit_code = (exit), \
6115
			.stage = X86_ICPT_POST_EXCEPT, }
6116
#define POST_MEM(exit) { .exit_code = (exit), \
6117
			.stage = X86_ICPT_POST_MEMACCESS, }
6118

6119
static const struct __x86_intercept {
6120 6121 6122 6123 6124 6125 6126 6127
	u32 exit_code;
	enum x86_intercept_stage stage;
} x86_intercept_map[] = {
	[x86_intercept_cr_read]		= POST_EX(SVM_EXIT_READ_CR0),
	[x86_intercept_cr_write]	= POST_EX(SVM_EXIT_WRITE_CR0),
	[x86_intercept_clts]		= POST_EX(SVM_EXIT_WRITE_CR0),
	[x86_intercept_lmsw]		= POST_EX(SVM_EXIT_WRITE_CR0),
	[x86_intercept_smsw]		= POST_EX(SVM_EXIT_READ_CR0),
6128 6129
	[x86_intercept_dr_read]		= POST_EX(SVM_EXIT_READ_DR0),
	[x86_intercept_dr_write]	= POST_EX(SVM_EXIT_WRITE_DR0),
6130 6131 6132 6133 6134 6135 6136 6137
	[x86_intercept_sldt]		= POST_EX(SVM_EXIT_LDTR_READ),
	[x86_intercept_str]		= POST_EX(SVM_EXIT_TR_READ),
	[x86_intercept_lldt]		= POST_EX(SVM_EXIT_LDTR_WRITE),
	[x86_intercept_ltr]		= POST_EX(SVM_EXIT_TR_WRITE),
	[x86_intercept_sgdt]		= POST_EX(SVM_EXIT_GDTR_READ),
	[x86_intercept_sidt]		= POST_EX(SVM_EXIT_IDTR_READ),
	[x86_intercept_lgdt]		= POST_EX(SVM_EXIT_GDTR_WRITE),
	[x86_intercept_lidt]		= POST_EX(SVM_EXIT_IDTR_WRITE),
6138 6139 6140 6141 6142 6143 6144 6145
	[x86_intercept_vmrun]		= POST_EX(SVM_EXIT_VMRUN),
	[x86_intercept_vmmcall]		= POST_EX(SVM_EXIT_VMMCALL),
	[x86_intercept_vmload]		= POST_EX(SVM_EXIT_VMLOAD),
	[x86_intercept_vmsave]		= POST_EX(SVM_EXIT_VMSAVE),
	[x86_intercept_stgi]		= POST_EX(SVM_EXIT_STGI),
	[x86_intercept_clgi]		= POST_EX(SVM_EXIT_CLGI),
	[x86_intercept_skinit]		= POST_EX(SVM_EXIT_SKINIT),
	[x86_intercept_invlpga]		= POST_EX(SVM_EXIT_INVLPGA),
6146 6147 6148
	[x86_intercept_rdtscp]		= POST_EX(SVM_EXIT_RDTSCP),
	[x86_intercept_monitor]		= POST_MEM(SVM_EXIT_MONITOR),
	[x86_intercept_mwait]		= POST_EX(SVM_EXIT_MWAIT),
6149 6150 6151 6152 6153 6154 6155 6156 6157
	[x86_intercept_invlpg]		= POST_EX(SVM_EXIT_INVLPG),
	[x86_intercept_invd]		= POST_EX(SVM_EXIT_INVD),
	[x86_intercept_wbinvd]		= POST_EX(SVM_EXIT_WBINVD),
	[x86_intercept_wrmsr]		= POST_EX(SVM_EXIT_MSR),
	[x86_intercept_rdtsc]		= POST_EX(SVM_EXIT_RDTSC),
	[x86_intercept_rdmsr]		= POST_EX(SVM_EXIT_MSR),
	[x86_intercept_rdpmc]		= POST_EX(SVM_EXIT_RDPMC),
	[x86_intercept_cpuid]		= PRE_EX(SVM_EXIT_CPUID),
	[x86_intercept_rsm]		= PRE_EX(SVM_EXIT_RSM),
6158 6159 6160 6161 6162 6163 6164
	[x86_intercept_pause]		= PRE_EX(SVM_EXIT_PAUSE),
	[x86_intercept_pushf]		= PRE_EX(SVM_EXIT_PUSHF),
	[x86_intercept_popf]		= PRE_EX(SVM_EXIT_POPF),
	[x86_intercept_intn]		= PRE_EX(SVM_EXIT_SWINT),
	[x86_intercept_iret]		= PRE_EX(SVM_EXIT_IRET),
	[x86_intercept_icebp]		= PRE_EX(SVM_EXIT_ICEBP),
	[x86_intercept_hlt]		= POST_EX(SVM_EXIT_HLT),
6165 6166 6167 6168
	[x86_intercept_in]		= POST_EX(SVM_EXIT_IOIO),
	[x86_intercept_ins]		= POST_EX(SVM_EXIT_IOIO),
	[x86_intercept_out]		= POST_EX(SVM_EXIT_IOIO),
	[x86_intercept_outs]		= POST_EX(SVM_EXIT_IOIO),
6169
	[x86_intercept_xsetbv]		= PRE_EX(SVM_EXIT_XSETBV),
6170 6171
};

6172
#undef PRE_EX
6173
#undef POST_EX
6174
#undef POST_MEM
6175

6176 6177
static int svm_check_intercept(struct kvm_vcpu *vcpu,
			       struct x86_instruction_info *info,
6178 6179
			       enum x86_intercept_stage stage,
			       struct x86_exception *exception)
6180
{
6181 6182 6183 6184 6185 6186 6187 6188 6189 6190
	struct vcpu_svm *svm = to_svm(vcpu);
	int vmexit, ret = X86EMUL_CONTINUE;
	struct __x86_intercept icpt_info;
	struct vmcb *vmcb = svm->vmcb;

	if (info->intercept >= ARRAY_SIZE(x86_intercept_map))
		goto out;

	icpt_info = x86_intercept_map[info->intercept];

6191
	if (stage != icpt_info.stage)
6192 6193 6194 6195 6196 6197 6198 6199 6200 6201 6202 6203 6204 6205
		goto out;

	switch (icpt_info.exit_code) {
	case SVM_EXIT_READ_CR0:
		if (info->intercept == x86_intercept_cr_read)
			icpt_info.exit_code += info->modrm_reg;
		break;
	case SVM_EXIT_WRITE_CR0: {
		unsigned long cr0, val;
		u64 intercept;

		if (info->intercept == x86_intercept_cr_write)
			icpt_info.exit_code += info->modrm_reg;

6206 6207
		if (icpt_info.exit_code != SVM_EXIT_WRITE_CR0 ||
		    info->intercept == x86_intercept_clts)
6208 6209 6210 6211 6212 6213 6214 6215 6216 6217 6218 6219 6220 6221 6222 6223 6224 6225 6226 6227 6228 6229 6230
			break;

		intercept = svm->nested.intercept;

		if (!(intercept & (1ULL << INTERCEPT_SELECTIVE_CR0)))
			break;

		cr0 = vcpu->arch.cr0 & ~SVM_CR0_SELECTIVE_MASK;
		val = info->src_val  & ~SVM_CR0_SELECTIVE_MASK;

		if (info->intercept == x86_intercept_lmsw) {
			cr0 &= 0xfUL;
			val &= 0xfUL;
			/* lmsw can't clear PE - catch this here */
			if (cr0 & X86_CR0_PE)
				val |= X86_CR0_PE;
		}

		if (cr0 ^ val)
			icpt_info.exit_code = SVM_EXIT_CR0_SEL_WRITE;

		break;
	}
6231 6232 6233 6234
	case SVM_EXIT_READ_DR0:
	case SVM_EXIT_WRITE_DR0:
		icpt_info.exit_code += info->modrm_reg;
		break;
6235 6236 6237 6238 6239 6240
	case SVM_EXIT_MSR:
		if (info->intercept == x86_intercept_wrmsr)
			vmcb->control.exit_info_1 = 1;
		else
			vmcb->control.exit_info_1 = 0;
		break;
6241 6242 6243 6244 6245 6246 6247
	case SVM_EXIT_PAUSE:
		/*
		 * We get this for NOP only, but pause
		 * is rep not, check this here
		 */
		if (info->rep_prefix != REPE_PREFIX)
			goto out;
6248
		break;
6249 6250 6251 6252 6253 6254
	case SVM_EXIT_IOIO: {
		u64 exit_info;
		u32 bytes;

		if (info->intercept == x86_intercept_in ||
		    info->intercept == x86_intercept_ins) {
6255 6256
			exit_info = ((info->src_val & 0xffff) << 16) |
				SVM_IOIO_TYPE_MASK;
6257
			bytes = info->dst_bytes;
6258
		} else {
6259
			exit_info = (info->dst_val & 0xffff) << 16;
6260
			bytes = info->src_bytes;
6261 6262 6263 6264 6265 6266 6267 6268 6269 6270 6271 6272 6273 6274 6275 6276 6277 6278 6279 6280
		}

		if (info->intercept == x86_intercept_outs ||
		    info->intercept == x86_intercept_ins)
			exit_info |= SVM_IOIO_STR_MASK;

		if (info->rep_prefix)
			exit_info |= SVM_IOIO_REP_MASK;

		bytes = min(bytes, 4u);

		exit_info |= bytes << SVM_IOIO_SIZE_SHIFT;

		exit_info |= (u32)info->ad_bytes << (SVM_IOIO_ASIZE_SHIFT - 1);

		vmcb->control.exit_info_1 = exit_info;
		vmcb->control.exit_info_2 = info->next_rip;

		break;
	}
6281 6282 6283 6284
	default:
		break;
	}

6285 6286 6287
	/* TODO: Advertise NRIPS to guest hypervisor unconditionally */
	if (static_cpu_has(X86_FEATURE_NRIPS))
		vmcb->control.next_rip  = info->next_rip;
6288 6289 6290 6291 6292 6293 6294 6295
	vmcb->control.exit_code = icpt_info.exit_code;
	vmexit = nested_svm_exit_handled(svm);

	ret = (vmexit == NESTED_EXIT_DONE) ? X86EMUL_INTERCEPTED
					   : X86EMUL_CONTINUE;

out:
	return ret;
6296 6297
}

6298 6299
static void svm_handle_exit_irqoff(struct kvm_vcpu *vcpu,
	enum exit_fastpath_completion *exit_fastpath)
6300
{
6301 6302 6303
	if (!is_guest_mode(vcpu) &&
		to_svm(vcpu)->vmcb->control.exit_code == EXIT_REASON_MSR_WRITE)
		*exit_fastpath = handle_fastpath_set_msr_irqoff(vcpu);
6304 6305
}

6306 6307
static void svm_sched_in(struct kvm_vcpu *vcpu, int cpu)
{
6308 6309
	if (pause_filter_thresh)
		shrink_ple_window(vcpu);
6310 6311
}

6312 6313 6314 6315
static inline void avic_post_state_restore(struct kvm_vcpu *vcpu)
{
	if (avic_handle_apic_id_update(vcpu) != 0)
		return;
6316
	avic_handle_dfr_update(vcpu);
6317 6318 6319
	avic_handle_ldr_update(vcpu);
}

6320 6321 6322 6323 6324 6325
static void svm_setup_mce(struct kvm_vcpu *vcpu)
{
	/* [63:9] are reserved. */
	vcpu->arch.mcg_cap &= 0x1ff;
}

6326 6327
static int svm_smi_allowed(struct kvm_vcpu *vcpu)
{
6328 6329 6330 6331 6332 6333 6334 6335 6336 6337 6338 6339 6340 6341
	struct vcpu_svm *svm = to_svm(vcpu);

	/* Per APM Vol.2 15.22.2 "Response to SMI" */
	if (!gif_set(svm))
		return 0;

	if (is_guest_mode(&svm->vcpu) &&
	    svm->nested.intercept & (1ULL << INTERCEPT_SMI)) {
		/* TODO: Might need to set exit_info_1 and exit_info_2 here */
		svm->vmcb->control.exit_code = SVM_EXIT_SMI;
		svm->nested.exit_required = true;
		return 0;
	}

6342 6343 6344
	return 1;
}

6345 6346
static int svm_pre_enter_smm(struct kvm_vcpu *vcpu, char *smstate)
{
6347 6348 6349 6350 6351 6352 6353 6354 6355 6356 6357 6358 6359 6360 6361 6362 6363
	struct vcpu_svm *svm = to_svm(vcpu);
	int ret;

	if (is_guest_mode(vcpu)) {
		/* FED8h - SVM Guest */
		put_smstate(u64, smstate, 0x7ed8, 1);
		/* FEE0h - SVM Guest VMCB Physical Address */
		put_smstate(u64, smstate, 0x7ee0, svm->nested.vmcb);

		svm->vmcb->save.rax = vcpu->arch.regs[VCPU_REGS_RAX];
		svm->vmcb->save.rsp = vcpu->arch.regs[VCPU_REGS_RSP];
		svm->vmcb->save.rip = vcpu->arch.regs[VCPU_REGS_RIP];

		ret = nested_svm_vmexit(svm);
		if (ret)
			return ret;
	}
6364 6365 6366
	return 0;
}

6367
static int svm_pre_leave_smm(struct kvm_vcpu *vcpu, const char *smstate)
6368
{
6369 6370
	struct vcpu_svm *svm = to_svm(vcpu);
	struct vmcb *nested_vmcb;
6371
	struct kvm_host_map map;
6372 6373
	u64 guest;
	u64 vmcb;
6374

6375 6376
	guest = GET_SMSTATE(u64, smstate, 0x7ed8);
	vmcb = GET_SMSTATE(u64, smstate, 0x7ee0);
6377

6378
	if (guest) {
6379
		if (kvm_vcpu_map(&svm->vcpu, gpa_to_gfn(vmcb), &map) == -EINVAL)
6380
			return 1;
6381 6382
		nested_vmcb = map.hva;
		enter_svm_guest_mode(svm, vmcb, nested_vmcb, &map);
6383
	}
6384
	return 0;
6385 6386
}

6387 6388 6389 6390 6391 6392 6393 6394 6395 6396 6397 6398 6399
static int enable_smi_window(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);

	if (!gif_set(svm)) {
		if (vgif_enabled(svm))
			set_intercept(svm, INTERCEPT_STGI);
		/* STGI will cause a vm exit */
		return 1;
	}
	return 0;
}

6400 6401 6402 6403 6404 6405 6406 6407 6408 6409 6410 6411 6412 6413 6414 6415 6416 6417 6418 6419 6420 6421 6422 6423 6424 6425 6426 6427 6428 6429 6430 6431 6432 6433 6434 6435 6436 6437 6438 6439 6440 6441
static int sev_flush_asids(void)
{
	int ret, error;

	/*
	 * DEACTIVATE will clear the WBINVD indicator causing DF_FLUSH to fail,
	 * so it must be guarded.
	 */
	down_write(&sev_deactivate_lock);

	wbinvd_on_all_cpus();
	ret = sev_guest_df_flush(&error);

	up_write(&sev_deactivate_lock);

	if (ret)
		pr_err("SEV: DF_FLUSH failed, ret=%d, error=%#x\n", ret, error);

	return ret;
}

/* Must be called with the sev_bitmap_lock held */
static bool __sev_recycle_asids(void)
{
	int pos;

	/* Check if there are any ASIDs to reclaim before performing a flush */
	pos = find_next_bit(sev_reclaim_asid_bitmap,
			    max_sev_asid, min_sev_asid - 1);
	if (pos >= max_sev_asid)
		return false;

	if (sev_flush_asids())
		return false;

	bitmap_xor(sev_asid_bitmap, sev_asid_bitmap, sev_reclaim_asid_bitmap,
		   max_sev_asid);
	bitmap_zero(sev_reclaim_asid_bitmap, max_sev_asid);

	return true;
}

B
Brijesh Singh 已提交
6442 6443
static int sev_asid_new(void)
{
6444
	bool retry = true;
B
Brijesh Singh 已提交
6445 6446
	int pos;

6447 6448
	mutex_lock(&sev_bitmap_lock);

B
Brijesh Singh 已提交
6449 6450 6451
	/*
	 * SEV-enabled guest must use asid from min_sev_asid to max_sev_asid.
	 */
6452
again:
B
Brijesh Singh 已提交
6453
	pos = find_next_zero_bit(sev_asid_bitmap, max_sev_asid, min_sev_asid - 1);
6454
	if (pos >= max_sev_asid) {
6455 6456 6457 6458
		if (retry && __sev_recycle_asids()) {
			retry = false;
			goto again;
		}
6459
		mutex_unlock(&sev_bitmap_lock);
B
Brijesh Singh 已提交
6460
		return -EBUSY;
6461 6462 6463 6464 6465
	}

	__set_bit(pos, sev_asid_bitmap);

	mutex_unlock(&sev_bitmap_lock);
B
Brijesh Singh 已提交
6466 6467 6468 6469 6470 6471

	return pos + 1;
}

static int sev_guest_init(struct kvm *kvm, struct kvm_sev_cmd *argp)
{
6472
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
B
Brijesh Singh 已提交
6473 6474 6475
	int asid, ret;

	ret = -EBUSY;
6476 6477 6478
	if (unlikely(sev->active))
		return ret;

B
Brijesh Singh 已提交
6479 6480 6481 6482 6483 6484 6485 6486 6487 6488
	asid = sev_asid_new();
	if (asid < 0)
		return ret;

	ret = sev_platform_init(&argp->error);
	if (ret)
		goto e_free;

	sev->active = true;
	sev->asid = asid;
6489
	INIT_LIST_HEAD(&sev->regions_list);
B
Brijesh Singh 已提交
6490 6491 6492 6493

	return 0;

e_free:
6494
	sev_asid_free(asid);
B
Brijesh Singh 已提交
6495 6496 6497
	return ret;
}

6498 6499 6500 6501 6502 6503
static int sev_bind_asid(struct kvm *kvm, unsigned int handle, int *error)
{
	struct sev_data_activate *data;
	int asid = sev_get_asid(kvm);
	int ret;

6504
	data = kzalloc(sizeof(*data), GFP_KERNEL_ACCOUNT);
6505 6506 6507 6508 6509 6510 6511 6512 6513 6514 6515 6516
	if (!data)
		return -ENOMEM;

	/* activate ASID on the given handle */
	data->handle = handle;
	data->asid   = asid;
	ret = sev_guest_activate(data, error);
	kfree(data);

	return ret;
}

6517
static int __sev_issue_cmd(int fd, int id, void *data, int *error)
6518 6519 6520 6521 6522 6523 6524 6525 6526 6527 6528 6529 6530 6531
{
	struct fd f;
	int ret;

	f = fdget(fd);
	if (!f.file)
		return -EBADF;

	ret = sev_issue_cmd_external_user(f.file, id, data, error);

	fdput(f);
	return ret;
}

6532 6533
static int sev_issue_cmd(struct kvm *kvm, int id, void *data, int *error)
{
6534
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
6535 6536 6537 6538

	return __sev_issue_cmd(sev->fd, id, data, error);
}

6539 6540
static int sev_launch_start(struct kvm *kvm, struct kvm_sev_cmd *argp)
{
6541
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
6542 6543 6544 6545 6546 6547 6548 6549 6550 6551 6552 6553
	struct sev_data_launch_start *start;
	struct kvm_sev_launch_start params;
	void *dh_blob, *session_blob;
	int *error = &argp->error;
	int ret;

	if (!sev_guest(kvm))
		return -ENOTTY;

	if (copy_from_user(&params, (void __user *)(uintptr_t)argp->data, sizeof(params)))
		return -EFAULT;

6554
	start = kzalloc(sizeof(*start), GFP_KERNEL_ACCOUNT);
6555 6556 6557 6558 6559 6560 6561 6562 6563 6564 6565 6566 6567 6568 6569 6570 6571 6572 6573 6574 6575 6576 6577 6578 6579 6580 6581 6582 6583 6584 6585
	if (!start)
		return -ENOMEM;

	dh_blob = NULL;
	if (params.dh_uaddr) {
		dh_blob = psp_copy_user_blob(params.dh_uaddr, params.dh_len);
		if (IS_ERR(dh_blob)) {
			ret = PTR_ERR(dh_blob);
			goto e_free;
		}

		start->dh_cert_address = __sme_set(__pa(dh_blob));
		start->dh_cert_len = params.dh_len;
	}

	session_blob = NULL;
	if (params.session_uaddr) {
		session_blob = psp_copy_user_blob(params.session_uaddr, params.session_len);
		if (IS_ERR(session_blob)) {
			ret = PTR_ERR(session_blob);
			goto e_free_dh;
		}

		start->session_address = __sme_set(__pa(session_blob));
		start->session_len = params.session_len;
	}

	start->handle = params.handle;
	start->policy = params.policy;

	/* create memory encryption context */
6586
	ret = __sev_issue_cmd(argp->sev_fd, SEV_CMD_LAUNCH_START, start, error);
6587 6588 6589 6590 6591 6592 6593 6594 6595 6596 6597 6598 6599 6600 6601 6602 6603 6604 6605 6606 6607 6608 6609 6610 6611 6612 6613 6614
	if (ret)
		goto e_free_session;

	/* Bind ASID to this guest */
	ret = sev_bind_asid(kvm, start->handle, error);
	if (ret)
		goto e_free_session;

	/* return handle to userspace */
	params.handle = start->handle;
	if (copy_to_user((void __user *)(uintptr_t)argp->data, &params, sizeof(params))) {
		sev_unbind_asid(kvm, start->handle);
		ret = -EFAULT;
		goto e_free_session;
	}

	sev->handle = start->handle;
	sev->fd = argp->sev_fd;

e_free_session:
	kfree(session_blob);
e_free_dh:
	kfree(dh_blob);
e_free:
	kfree(start);
	return ret;
}

6615 6616
static unsigned long get_num_contig_pages(unsigned long idx,
				struct page **inpages, unsigned long npages)
6617 6618
{
	unsigned long paddr, next_paddr;
6619
	unsigned long i = idx + 1, pages = 1;
6620 6621 6622 6623 6624 6625 6626 6627 6628 6629 6630 6631 6632 6633 6634 6635 6636 6637

	/* find the number of contiguous pages starting from idx */
	paddr = __sme_page_pa(inpages[idx]);
	while (i < npages) {
		next_paddr = __sme_page_pa(inpages[i++]);
		if ((paddr + PAGE_SIZE) == next_paddr) {
			pages++;
			paddr = next_paddr;
			continue;
		}
		break;
	}

	return pages;
}

static int sev_launch_update_data(struct kvm *kvm, struct kvm_sev_cmd *argp)
{
6638
	unsigned long vaddr, vaddr_end, next_vaddr, npages, pages, size, i;
6639
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
6640 6641 6642
	struct kvm_sev_launch_update_data params;
	struct sev_data_launch_update_data *data;
	struct page **inpages;
6643
	int ret;
6644 6645 6646 6647 6648 6649 6650

	if (!sev_guest(kvm))
		return -ENOTTY;

	if (copy_from_user(&params, (void __user *)(uintptr_t)argp->data, sizeof(params)))
		return -EFAULT;

6651
	data = kzalloc(sizeof(*data), GFP_KERNEL_ACCOUNT);
6652 6653 6654 6655 6656 6657 6658 6659 6660 6661 6662 6663 6664 6665 6666 6667 6668 6669 6670 6671 6672 6673 6674 6675 6676 6677 6678 6679 6680 6681 6682 6683 6684 6685 6686 6687 6688 6689 6690 6691 6692 6693 6694 6695 6696 6697 6698 6699 6700 6701 6702 6703 6704 6705 6706 6707 6708 6709 6710 6711
	if (!data)
		return -ENOMEM;

	vaddr = params.uaddr;
	size = params.len;
	vaddr_end = vaddr + size;

	/* Lock the user memory. */
	inpages = sev_pin_memory(kvm, vaddr, size, &npages, 1);
	if (!inpages) {
		ret = -ENOMEM;
		goto e_free;
	}

	/*
	 * The LAUNCH_UPDATE command will perform in-place encryption of the
	 * memory content (i.e it will write the same memory region with C=1).
	 * It's possible that the cache may contain the data with C=0, i.e.,
	 * unencrypted so invalidate it first.
	 */
	sev_clflush_pages(inpages, npages);

	for (i = 0; vaddr < vaddr_end; vaddr = next_vaddr, i += pages) {
		int offset, len;

		/*
		 * If the user buffer is not page-aligned, calculate the offset
		 * within the page.
		 */
		offset = vaddr & (PAGE_SIZE - 1);

		/* Calculate the number of pages that can be encrypted in one go. */
		pages = get_num_contig_pages(i, inpages, npages);

		len = min_t(size_t, ((pages * PAGE_SIZE) - offset), size);

		data->handle = sev->handle;
		data->len = len;
		data->address = __sme_page_pa(inpages[i]) + offset;
		ret = sev_issue_cmd(kvm, SEV_CMD_LAUNCH_UPDATE_DATA, data, &argp->error);
		if (ret)
			goto e_unpin;

		size -= len;
		next_vaddr = vaddr + len;
	}

e_unpin:
	/* content of memory is updated, mark pages dirty */
	for (i = 0; i < npages; i++) {
		set_page_dirty_lock(inpages[i]);
		mark_page_accessed(inpages[i]);
	}
	/* unlock the user pages */
	sev_unpin_memory(kvm, inpages, npages);
e_free:
	kfree(data);
	return ret;
}

6712 6713
static int sev_launch_measure(struct kvm *kvm, struct kvm_sev_cmd *argp)
{
6714
	void __user *measure = (void __user *)(uintptr_t)argp->data;
6715
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
6716 6717
	struct sev_data_launch_measure *data;
	struct kvm_sev_launch_measure params;
6718
	void __user *p = NULL;
6719 6720 6721 6722 6723 6724
	void *blob = NULL;
	int ret;

	if (!sev_guest(kvm))
		return -ENOTTY;

6725
	if (copy_from_user(&params, measure, sizeof(params)))
6726 6727
		return -EFAULT;

6728
	data = kzalloc(sizeof(*data), GFP_KERNEL_ACCOUNT);
6729 6730 6731 6732 6733 6734 6735
	if (!data)
		return -ENOMEM;

	/* User wants to query the blob length */
	if (!params.len)
		goto cmd;

6736 6737
	p = (void __user *)(uintptr_t)params.uaddr;
	if (p) {
6738 6739 6740 6741 6742 6743 6744 6745 6746 6747 6748 6749 6750 6751 6752 6753 6754 6755 6756 6757 6758 6759 6760 6761 6762 6763 6764 6765
		if (params.len > SEV_FW_BLOB_MAX_SIZE) {
			ret = -EINVAL;
			goto e_free;
		}

		ret = -ENOMEM;
		blob = kmalloc(params.len, GFP_KERNEL);
		if (!blob)
			goto e_free;

		data->address = __psp_pa(blob);
		data->len = params.len;
	}

cmd:
	data->handle = sev->handle;
	ret = sev_issue_cmd(kvm, SEV_CMD_LAUNCH_MEASURE, data, &argp->error);

	/*
	 * If we query the session length, FW responded with expected data.
	 */
	if (!params.len)
		goto done;

	if (ret)
		goto e_free_blob;

	if (blob) {
6766
		if (copy_to_user(p, blob, params.len))
6767 6768 6769 6770 6771
			ret = -EFAULT;
	}

done:
	params.len = data->len;
6772
	if (copy_to_user(measure, &params, sizeof(params)))
6773 6774 6775 6776 6777 6778 6779 6780
		ret = -EFAULT;
e_free_blob:
	kfree(blob);
e_free:
	kfree(data);
	return ret;
}

6781 6782
static int sev_launch_finish(struct kvm *kvm, struct kvm_sev_cmd *argp)
{
6783
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
6784 6785 6786 6787 6788 6789
	struct sev_data_launch_finish *data;
	int ret;

	if (!sev_guest(kvm))
		return -ENOTTY;

6790
	data = kzalloc(sizeof(*data), GFP_KERNEL_ACCOUNT);
6791 6792 6793 6794 6795 6796 6797 6798 6799 6800
	if (!data)
		return -ENOMEM;

	data->handle = sev->handle;
	ret = sev_issue_cmd(kvm, SEV_CMD_LAUNCH_FINISH, data, &argp->error);

	kfree(data);
	return ret;
}

6801 6802
static int sev_guest_status(struct kvm *kvm, struct kvm_sev_cmd *argp)
{
6803
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
6804 6805 6806 6807 6808 6809 6810
	struct kvm_sev_guest_status params;
	struct sev_data_guest_status *data;
	int ret;

	if (!sev_guest(kvm))
		return -ENOTTY;

6811
	data = kzalloc(sizeof(*data), GFP_KERNEL_ACCOUNT);
6812 6813 6814 6815 6816 6817 6818 6819 6820 6821 6822 6823 6824 6825 6826 6827 6828 6829 6830
	if (!data)
		return -ENOMEM;

	data->handle = sev->handle;
	ret = sev_issue_cmd(kvm, SEV_CMD_GUEST_STATUS, data, &argp->error);
	if (ret)
		goto e_free;

	params.policy = data->policy;
	params.state = data->state;
	params.handle = data->handle;

	if (copy_to_user((void __user *)(uintptr_t)argp->data, &params, sizeof(params)))
		ret = -EFAULT;
e_free:
	kfree(data);
	return ret;
}

6831 6832 6833 6834
static int __sev_issue_dbg_cmd(struct kvm *kvm, unsigned long src,
			       unsigned long dst, int size,
			       int *error, bool enc)
{
6835
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
6836 6837 6838
	struct sev_data_dbg *data;
	int ret;

6839
	data = kzalloc(sizeof(*data), GFP_KERNEL_ACCOUNT);
6840 6841 6842 6843 6844 6845 6846 6847 6848 6849 6850 6851 6852 6853 6854 6855 6856 6857 6858 6859 6860 6861 6862 6863 6864 6865 6866 6867 6868 6869 6870 6871 6872 6873 6874 6875 6876 6877 6878 6879 6880 6881 6882 6883 6884 6885 6886 6887 6888 6889 6890 6891 6892 6893 6894 6895 6896 6897 6898 6899 6900 6901 6902 6903 6904 6905 6906 6907
	if (!data)
		return -ENOMEM;

	data->handle = sev->handle;
	data->dst_addr = dst;
	data->src_addr = src;
	data->len = size;

	ret = sev_issue_cmd(kvm,
			    enc ? SEV_CMD_DBG_ENCRYPT : SEV_CMD_DBG_DECRYPT,
			    data, error);
	kfree(data);
	return ret;
}

static int __sev_dbg_decrypt(struct kvm *kvm, unsigned long src_paddr,
			     unsigned long dst_paddr, int sz, int *err)
{
	int offset;

	/*
	 * Its safe to read more than we are asked, caller should ensure that
	 * destination has enough space.
	 */
	src_paddr = round_down(src_paddr, 16);
	offset = src_paddr & 15;
	sz = round_up(sz + offset, 16);

	return __sev_issue_dbg_cmd(kvm, src_paddr, dst_paddr, sz, err, false);
}

static int __sev_dbg_decrypt_user(struct kvm *kvm, unsigned long paddr,
				  unsigned long __user dst_uaddr,
				  unsigned long dst_paddr,
				  int size, int *err)
{
	struct page *tpage = NULL;
	int ret, offset;

	/* if inputs are not 16-byte then use intermediate buffer */
	if (!IS_ALIGNED(dst_paddr, 16) ||
	    !IS_ALIGNED(paddr,     16) ||
	    !IS_ALIGNED(size,      16)) {
		tpage = (void *)alloc_page(GFP_KERNEL);
		if (!tpage)
			return -ENOMEM;

		dst_paddr = __sme_page_pa(tpage);
	}

	ret = __sev_dbg_decrypt(kvm, paddr, dst_paddr, size, err);
	if (ret)
		goto e_free;

	if (tpage) {
		offset = paddr & 15;
		if (copy_to_user((void __user *)(uintptr_t)dst_uaddr,
				 page_address(tpage) + offset, size))
			ret = -EFAULT;
	}

e_free:
	if (tpage)
		__free_page(tpage);

	return ret;
}

6908 6909 6910 6911 6912 6913 6914 6915 6916 6917 6918 6919 6920 6921 6922 6923 6924 6925 6926 6927 6928 6929 6930 6931 6932 6933 6934 6935 6936 6937 6938 6939 6940 6941 6942 6943 6944 6945 6946 6947 6948 6949 6950 6951 6952 6953 6954 6955 6956 6957 6958 6959 6960 6961 6962 6963 6964 6965 6966 6967 6968 6969 6970 6971 6972 6973 6974 6975 6976 6977 6978 6979 6980 6981 6982 6983 6984
static int __sev_dbg_encrypt_user(struct kvm *kvm, unsigned long paddr,
				  unsigned long __user vaddr,
				  unsigned long dst_paddr,
				  unsigned long __user dst_vaddr,
				  int size, int *error)
{
	struct page *src_tpage = NULL;
	struct page *dst_tpage = NULL;
	int ret, len = size;

	/* If source buffer is not aligned then use an intermediate buffer */
	if (!IS_ALIGNED(vaddr, 16)) {
		src_tpage = alloc_page(GFP_KERNEL);
		if (!src_tpage)
			return -ENOMEM;

		if (copy_from_user(page_address(src_tpage),
				(void __user *)(uintptr_t)vaddr, size)) {
			__free_page(src_tpage);
			return -EFAULT;
		}

		paddr = __sme_page_pa(src_tpage);
	}

	/*
	 *  If destination buffer or length is not aligned then do read-modify-write:
	 *   - decrypt destination in an intermediate buffer
	 *   - copy the source buffer in an intermediate buffer
	 *   - use the intermediate buffer as source buffer
	 */
	if (!IS_ALIGNED(dst_vaddr, 16) || !IS_ALIGNED(size, 16)) {
		int dst_offset;

		dst_tpage = alloc_page(GFP_KERNEL);
		if (!dst_tpage) {
			ret = -ENOMEM;
			goto e_free;
		}

		ret = __sev_dbg_decrypt(kvm, dst_paddr,
					__sme_page_pa(dst_tpage), size, error);
		if (ret)
			goto e_free;

		/*
		 *  If source is kernel buffer then use memcpy() otherwise
		 *  copy_from_user().
		 */
		dst_offset = dst_paddr & 15;

		if (src_tpage)
			memcpy(page_address(dst_tpage) + dst_offset,
			       page_address(src_tpage), size);
		else {
			if (copy_from_user(page_address(dst_tpage) + dst_offset,
					   (void __user *)(uintptr_t)vaddr, size)) {
				ret = -EFAULT;
				goto e_free;
			}
		}

		paddr = __sme_page_pa(dst_tpage);
		dst_paddr = round_down(dst_paddr, 16);
		len = round_up(size, 16);
	}

	ret = __sev_issue_dbg_cmd(kvm, paddr, dst_paddr, len, error, true);

e_free:
	if (src_tpage)
		__free_page(src_tpage);
	if (dst_tpage)
		__free_page(dst_tpage);
	return ret;
}

6985 6986 6987
static int sev_dbg_crypt(struct kvm *kvm, struct kvm_sev_cmd *argp, bool dec)
{
	unsigned long vaddr, vaddr_end, next_vaddr;
6988
	unsigned long dst_vaddr;
6989 6990 6991
	struct page **src_p, **dst_p;
	struct kvm_sev_dbg debug;
	unsigned long n;
6992 6993
	unsigned int size;
	int ret;
6994 6995 6996 6997 6998 6999 7000

	if (!sev_guest(kvm))
		return -ENOTTY;

	if (copy_from_user(&debug, (void __user *)(uintptr_t)argp->data, sizeof(debug)))
		return -EFAULT;

7001 7002 7003 7004 7005
	if (!debug.len || debug.src_uaddr + debug.len < debug.src_uaddr)
		return -EINVAL;
	if (!debug.dst_uaddr)
		return -EINVAL;

7006 7007 7008 7009 7010 7011 7012 7013 7014 7015 7016 7017 7018 7019 7020 7021 7022 7023 7024 7025 7026 7027 7028 7029 7030 7031 7032 7033 7034 7035 7036 7037 7038 7039 7040 7041
	vaddr = debug.src_uaddr;
	size = debug.len;
	vaddr_end = vaddr + size;
	dst_vaddr = debug.dst_uaddr;

	for (; vaddr < vaddr_end; vaddr = next_vaddr) {
		int len, s_off, d_off;

		/* lock userspace source and destination page */
		src_p = sev_pin_memory(kvm, vaddr & PAGE_MASK, PAGE_SIZE, &n, 0);
		if (!src_p)
			return -EFAULT;

		dst_p = sev_pin_memory(kvm, dst_vaddr & PAGE_MASK, PAGE_SIZE, &n, 1);
		if (!dst_p) {
			sev_unpin_memory(kvm, src_p, n);
			return -EFAULT;
		}

		/*
		 * The DBG_{DE,EN}CRYPT commands will perform {dec,en}cryption of the
		 * memory content (i.e it will write the same memory region with C=1).
		 * It's possible that the cache may contain the data with C=0, i.e.,
		 * unencrypted so invalidate it first.
		 */
		sev_clflush_pages(src_p, 1);
		sev_clflush_pages(dst_p, 1);

		/*
		 * Since user buffer may not be page aligned, calculate the
		 * offset within the page.
		 */
		s_off = vaddr & ~PAGE_MASK;
		d_off = dst_vaddr & ~PAGE_MASK;
		len = min_t(size_t, (PAGE_SIZE - s_off), size);

7042 7043 7044 7045 7046 7047 7048 7049 7050 7051 7052 7053 7054
		if (dec)
			ret = __sev_dbg_decrypt_user(kvm,
						     __sme_page_pa(src_p[0]) + s_off,
						     dst_vaddr,
						     __sme_page_pa(dst_p[0]) + d_off,
						     len, &argp->error);
		else
			ret = __sev_dbg_encrypt_user(kvm,
						     __sme_page_pa(src_p[0]) + s_off,
						     vaddr,
						     __sme_page_pa(dst_p[0]) + d_off,
						     dst_vaddr,
						     len, &argp->error);
7055

7056 7057
		sev_unpin_memory(kvm, src_p, n);
		sev_unpin_memory(kvm, dst_p, n);
7058 7059 7060 7061 7062 7063 7064 7065 7066 7067 7068 7069

		if (ret)
			goto err;

		next_vaddr = vaddr + len;
		dst_vaddr = dst_vaddr + len;
		size -= len;
	}
err:
	return ret;
}

7070 7071
static int sev_launch_secret(struct kvm *kvm, struct kvm_sev_cmd *argp)
{
7072
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
7073 7074 7075 7076 7077
	struct sev_data_launch_secret *data;
	struct kvm_sev_launch_secret params;
	struct page **pages;
	void *blob, *hdr;
	unsigned long n;
7078
	int ret, offset;
7079 7080 7081 7082 7083 7084 7085 7086 7087 7088 7089 7090 7091 7092 7093 7094 7095 7096 7097 7098 7099

	if (!sev_guest(kvm))
		return -ENOTTY;

	if (copy_from_user(&params, (void __user *)(uintptr_t)argp->data, sizeof(params)))
		return -EFAULT;

	pages = sev_pin_memory(kvm, params.guest_uaddr, params.guest_len, &n, 1);
	if (!pages)
		return -ENOMEM;

	/*
	 * The secret must be copied into contiguous memory region, lets verify
	 * that userspace memory pages are contiguous before we issue command.
	 */
	if (get_num_contig_pages(0, pages, n) != n) {
		ret = -EINVAL;
		goto e_unpin_memory;
	}

	ret = -ENOMEM;
7100
	data = kzalloc(sizeof(*data), GFP_KERNEL_ACCOUNT);
7101 7102 7103
	if (!data)
		goto e_unpin_memory;

7104 7105 7106 7107
	offset = params.guest_uaddr & (PAGE_SIZE - 1);
	data->guest_address = __sme_page_pa(pages[0]) + offset;
	data->guest_len = params.guest_len;

7108 7109 7110 7111 7112 7113 7114 7115 7116 7117 7118 7119 7120 7121
	blob = psp_copy_user_blob(params.trans_uaddr, params.trans_len);
	if (IS_ERR(blob)) {
		ret = PTR_ERR(blob);
		goto e_free;
	}

	data->trans_address = __psp_pa(blob);
	data->trans_len = params.trans_len;

	hdr = psp_copy_user_blob(params.hdr_uaddr, params.hdr_len);
	if (IS_ERR(hdr)) {
		ret = PTR_ERR(hdr);
		goto e_free_blob;
	}
7122 7123
	data->hdr_address = __psp_pa(hdr);
	data->hdr_len = params.hdr_len;
7124 7125 7126 7127 7128 7129 7130 7131 7132 7133 7134 7135 7136 7137 7138

	data->handle = sev->handle;
	ret = sev_issue_cmd(kvm, SEV_CMD_LAUNCH_UPDATE_SECRET, data, &argp->error);

	kfree(hdr);

e_free_blob:
	kfree(blob);
e_free:
	kfree(data);
e_unpin_memory:
	sev_unpin_memory(kvm, pages, n);
	return ret;
}

B
Brijesh Singh 已提交
7139 7140 7141 7142 7143 7144 7145 7146 7147 7148 7149 7150 7151 7152 7153 7154 7155
static int svm_mem_enc_op(struct kvm *kvm, void __user *argp)
{
	struct kvm_sev_cmd sev_cmd;
	int r;

	if (!svm_sev_enabled())
		return -ENOTTY;

	if (copy_from_user(&sev_cmd, argp, sizeof(struct kvm_sev_cmd)))
		return -EFAULT;

	mutex_lock(&kvm->lock);

	switch (sev_cmd.id) {
	case KVM_SEV_INIT:
		r = sev_guest_init(kvm, &sev_cmd);
		break;
7156 7157 7158
	case KVM_SEV_LAUNCH_START:
		r = sev_launch_start(kvm, &sev_cmd);
		break;
7159 7160 7161
	case KVM_SEV_LAUNCH_UPDATE_DATA:
		r = sev_launch_update_data(kvm, &sev_cmd);
		break;
7162 7163 7164
	case KVM_SEV_LAUNCH_MEASURE:
		r = sev_launch_measure(kvm, &sev_cmd);
		break;
7165 7166 7167
	case KVM_SEV_LAUNCH_FINISH:
		r = sev_launch_finish(kvm, &sev_cmd);
		break;
7168 7169 7170
	case KVM_SEV_GUEST_STATUS:
		r = sev_guest_status(kvm, &sev_cmd);
		break;
7171 7172 7173
	case KVM_SEV_DBG_DECRYPT:
		r = sev_dbg_crypt(kvm, &sev_cmd, true);
		break;
7174 7175 7176
	case KVM_SEV_DBG_ENCRYPT:
		r = sev_dbg_crypt(kvm, &sev_cmd, false);
		break;
7177 7178 7179
	case KVM_SEV_LAUNCH_SECRET:
		r = sev_launch_secret(kvm, &sev_cmd);
		break;
B
Brijesh Singh 已提交
7180 7181 7182 7183 7184 7185 7186 7187 7188 7189 7190 7191 7192
	default:
		r = -EINVAL;
		goto out;
	}

	if (copy_to_user(argp, &sev_cmd, sizeof(struct kvm_sev_cmd)))
		r = -EFAULT;

out:
	mutex_unlock(&kvm->lock);
	return r;
}

7193 7194 7195
static int svm_register_enc_region(struct kvm *kvm,
				   struct kvm_enc_region *range)
{
7196
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
7197 7198 7199 7200 7201 7202
	struct enc_region *region;
	int ret = 0;

	if (!sev_guest(kvm))
		return -ENOTTY;

7203 7204 7205
	if (range->addr > ULONG_MAX || range->size > ULONG_MAX)
		return -EINVAL;

7206
	region = kzalloc(sizeof(*region), GFP_KERNEL_ACCOUNT);
7207 7208 7209 7210 7211 7212 7213 7214 7215 7216 7217 7218 7219 7220 7221 7222 7223 7224 7225 7226 7227 7228 7229 7230 7231 7232 7233 7234 7235 7236 7237 7238 7239 7240
	if (!region)
		return -ENOMEM;

	region->pages = sev_pin_memory(kvm, range->addr, range->size, &region->npages, 1);
	if (!region->pages) {
		ret = -ENOMEM;
		goto e_free;
	}

	/*
	 * The guest may change the memory encryption attribute from C=0 -> C=1
	 * or vice versa for this memory range. Lets make sure caches are
	 * flushed to ensure that guest data gets written into memory with
	 * correct C-bit.
	 */
	sev_clflush_pages(region->pages, region->npages);

	region->uaddr = range->addr;
	region->size = range->size;

	mutex_lock(&kvm->lock);
	list_add_tail(&region->list, &sev->regions_list);
	mutex_unlock(&kvm->lock);

	return ret;

e_free:
	kfree(region);
	return ret;
}

static struct enc_region *
find_enc_region(struct kvm *kvm, struct kvm_enc_region *range)
{
7241
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
7242 7243 7244 7245 7246 7247 7248 7249 7250 7251 7252 7253 7254 7255 7256 7257 7258 7259 7260 7261 7262 7263 7264 7265 7266 7267 7268 7269 7270 7271 7272 7273 7274 7275 7276 7277 7278 7279 7280 7281 7282 7283
	struct list_head *head = &sev->regions_list;
	struct enc_region *i;

	list_for_each_entry(i, head, list) {
		if (i->uaddr == range->addr &&
		    i->size == range->size)
			return i;
	}

	return NULL;
}


static int svm_unregister_enc_region(struct kvm *kvm,
				     struct kvm_enc_region *range)
{
	struct enc_region *region;
	int ret;

	mutex_lock(&kvm->lock);

	if (!sev_guest(kvm)) {
		ret = -ENOTTY;
		goto failed;
	}

	region = find_enc_region(kvm, range);
	if (!region) {
		ret = -EINVAL;
		goto failed;
	}

	__unregister_enc_region_locked(kvm, region);

	mutex_unlock(&kvm->lock);
	return 0;

failed:
	mutex_unlock(&kvm->lock);
	return ret;
}

7284 7285
static bool svm_need_emulation_on_page_fault(struct kvm_vcpu *vcpu)
{
7286 7287 7288 7289
	unsigned long cr4 = kvm_read_cr4(vcpu);
	bool smep = cr4 & X86_CR4_SMEP;
	bool smap = cr4 & X86_CR4_SMAP;
	bool is_user = svm_get_cpl(vcpu) == 3;
7290 7291

	/*
7292 7293 7294 7295 7296 7297 7298 7299 7300 7301 7302 7303 7304 7305 7306 7307 7308 7309 7310 7311 7312 7313 7314 7315 7316 7317 7318 7319 7320
	 * Detect and workaround Errata 1096 Fam_17h_00_0Fh.
	 *
	 * Errata:
	 * When CPU raise #NPF on guest data access and vCPU CR4.SMAP=1, it is
	 * possible that CPU microcode implementing DecodeAssist will fail
	 * to read bytes of instruction which caused #NPF. In this case,
	 * GuestIntrBytes field of the VMCB on a VMEXIT will incorrectly
	 * return 0 instead of the correct guest instruction bytes.
	 *
	 * This happens because CPU microcode reading instruction bytes
	 * uses a special opcode which attempts to read data using CPL=0
	 * priviledges. The microcode reads CS:RIP and if it hits a SMAP
	 * fault, it gives up and returns no instruction bytes.
	 *
	 * Detection:
	 * We reach here in case CPU supports DecodeAssist, raised #NPF and
	 * returned 0 in GuestIntrBytes field of the VMCB.
	 * First, errata can only be triggered in case vCPU CR4.SMAP=1.
	 * Second, if vCPU CR4.SMEP=1, errata could only be triggered
	 * in case vCPU CPL==3 (Because otherwise guest would have triggered
	 * a SMEP fault instead of #NPF).
	 * Otherwise, vCPU CR4.SMEP=0, errata could be triggered by any vCPU CPL.
	 * As most guests enable SMAP if they have also enabled SMEP, use above
	 * logic in order to attempt minimize false-positive of detecting errata
	 * while still preserving all cases semantic correctness.
	 *
	 * Workaround:
	 * To determine what instruction the guest was executing, the hypervisor
	 * will have to decode the instruction at the instruction pointer.
7321 7322 7323 7324 7325 7326 7327 7328 7329 7330
	 *
	 * In non SEV guest, hypervisor will be able to read the guest
	 * memory to decode the instruction pointer when insn_len is zero
	 * so we return true to indicate that decoding is possible.
	 *
	 * But in the SEV guest, the guest memory is encrypted with the
	 * guest specific key and hypervisor will not be able to decode the
	 * instruction pointer so we will not able to workaround it. Lets
	 * print the error and request to kill the guest.
	 */
7331
	if (smap && (!smep || is_user)) {
7332 7333 7334
		if (!sev_guest(vcpu->kvm))
			return true;

7335
		pr_err_ratelimited("KVM: SEV Guest triggered AMD Erratum 1096\n");
7336 7337 7338 7339 7340 7341
		kvm_make_request(KVM_REQ_TRIPLE_FAULT, vcpu);
	}

	return false;
}

7342 7343 7344 7345 7346 7347 7348 7349 7350 7351 7352 7353 7354 7355 7356
static bool svm_apic_init_signal_blocked(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);

	/*
	 * TODO: Last condition latch INIT signals on vCPU when
	 * vCPU is in guest-mode and vmcb12 defines intercept on INIT.
	 * To properly emulate the INIT intercept, SVM should implement
	 * kvm_x86_ops->check_nested_events() and call nested_svm_vmexit()
	 * there if an INIT signal is pending.
	 */
	return !gif_set(svm) ||
		   (svm->vmcb->control.intercept & (1ULL << INTERCEPT_INIT));
}

7357 7358
static bool svm_check_apicv_inhibit_reasons(ulong bit)
{
7359
	ulong supported = BIT(APICV_INHIBIT_REASON_DISABLE) |
7360
			  BIT(APICV_INHIBIT_REASON_HYPERV) |
7361
			  BIT(APICV_INHIBIT_REASON_NESTED) |
7362
			  BIT(APICV_INHIBIT_REASON_IRQWIN) |
7363 7364
			  BIT(APICV_INHIBIT_REASON_PIT_REINJ) |
			  BIT(APICV_INHIBIT_REASON_X2APIC);
7365 7366 7367 7368

	return supported & BIT(bit);
}

7369 7370 7371 7372 7373
static void svm_pre_update_apicv_exec_ctrl(struct kvm *kvm, bool activate)
{
	avic_update_access_page(kvm, activate);
}

7374
static struct kvm_x86_ops svm_x86_ops __ro_after_init = {
A
Avi Kivity 已提交
7375 7376 7377
	.cpu_has_kvm_support = has_svm,
	.disabled_by_bios = is_disabled,
	.hardware_setup = svm_hardware_setup,
7378
	.hardware_unsetup = svm_hardware_teardown,
Y
Yang, Sheng 已提交
7379
	.check_processor_compatibility = svm_check_processor_compat,
A
Avi Kivity 已提交
7380 7381
	.hardware_enable = svm_hardware_enable,
	.hardware_disable = svm_hardware_disable,
7382
	.cpu_has_accelerated_tpr = svm_cpu_has_accelerated_tpr,
7383
	.has_emulated_msr = svm_has_emulated_msr,
A
Avi Kivity 已提交
7384 7385 7386

	.vcpu_create = svm_create_vcpu,
	.vcpu_free = svm_free_vcpu,
7387
	.vcpu_reset = svm_vcpu_reset,
A
Avi Kivity 已提交
7388

7389
	.vm_size = sizeof(struct kvm_svm),
7390
	.vm_init = svm_vm_init,
B
Brijesh Singh 已提交
7391
	.vm_destroy = svm_vm_destroy,
7392

7393
	.prepare_guest_switch = svm_prepare_guest_switch,
A
Avi Kivity 已提交
7394 7395
	.vcpu_load = svm_vcpu_load,
	.vcpu_put = svm_vcpu_put,
7396 7397
	.vcpu_blocking = svm_vcpu_blocking,
	.vcpu_unblocking = svm_vcpu_unblocking,
A
Avi Kivity 已提交
7398

7399
	.update_bp_intercept = update_bp_intercept,
7400
	.get_msr_feature = svm_get_msr_feature,
A
Avi Kivity 已提交
7401 7402 7403 7404 7405
	.get_msr = svm_get_msr,
	.set_msr = svm_set_msr,
	.get_segment_base = svm_get_segment_base,
	.get_segment = svm_get_segment,
	.set_segment = svm_set_segment,
7406
	.get_cpl = svm_get_cpl,
7407
	.get_cs_db_l_bits = kvm_get_cs_db_l_bits,
7408
	.decache_cr0_guest_bits = svm_decache_cr0_guest_bits,
7409
	.decache_cr4_guest_bits = svm_decache_cr4_guest_bits,
A
Avi Kivity 已提交
7410 7411 7412 7413 7414 7415 7416 7417
	.set_cr0 = svm_set_cr0,
	.set_cr3 = svm_set_cr3,
	.set_cr4 = svm_set_cr4,
	.set_efer = svm_set_efer,
	.get_idt = svm_get_idt,
	.set_idt = svm_set_idt,
	.get_gdt = svm_get_gdt,
	.set_gdt = svm_set_gdt,
J
Jan Kiszka 已提交
7418 7419
	.get_dr6 = svm_get_dr6,
	.set_dr6 = svm_set_dr6,
7420
	.set_dr7 = svm_set_dr7,
7421
	.sync_dirty_debug_regs = svm_sync_dirty_debug_regs,
A
Avi Kivity 已提交
7422
	.cache_reg = svm_cache_reg,
A
Avi Kivity 已提交
7423 7424
	.get_rflags = svm_get_rflags,
	.set_rflags = svm_set_rflags,
7425

A
Avi Kivity 已提交
7426
	.tlb_flush = svm_flush_tlb,
7427
	.tlb_flush_gva = svm_flush_tlb_gva,
A
Avi Kivity 已提交
7428 7429

	.run = svm_vcpu_run,
7430
	.handle_exit = handle_exit,
A
Avi Kivity 已提交
7431
	.skip_emulated_instruction = skip_emulated_instruction,
7432
	.update_emulated_instruction = NULL,
7433 7434
	.set_interrupt_shadow = svm_set_interrupt_shadow,
	.get_interrupt_shadow = svm_get_interrupt_shadow,
I
Ingo Molnar 已提交
7435
	.patch_hypercall = svm_patch_hypercall,
E
Eddie Dong 已提交
7436
	.set_irq = svm_set_irq,
7437
	.set_nmi = svm_inject_nmi,
7438
	.queue_exception = svm_queue_exception,
A
Avi Kivity 已提交
7439
	.cancel_injection = svm_cancel_injection,
7440
	.interrupt_allowed = svm_interrupt_allowed,
7441
	.nmi_allowed = svm_nmi_allowed,
J
Jan Kiszka 已提交
7442 7443
	.get_nmi_mask = svm_get_nmi_mask,
	.set_nmi_mask = svm_set_nmi_mask,
7444 7445 7446
	.enable_nmi_window = enable_nmi_window,
	.enable_irq_window = enable_irq_window,
	.update_cr8_intercept = update_cr8_intercept,
7447
	.set_virtual_apic_mode = svm_set_virtual_apic_mode,
7448
	.refresh_apicv_exec_ctrl = svm_refresh_apicv_exec_ctrl,
7449
	.check_apicv_inhibit_reasons = svm_check_apicv_inhibit_reasons,
7450
	.pre_update_apicv_exec_ctrl = svm_pre_update_apicv_exec_ctrl,
7451
	.load_eoi_exitmap = svm_load_eoi_exitmap,
7452 7453
	.hwapic_irr_update = svm_hwapic_irr_update,
	.hwapic_isr_update = svm_hwapic_isr_update,
7454
	.sync_pir_to_irr = kvm_lapic_find_highest_irr,
7455
	.apicv_post_state_restore = avic_post_state_restore,
7456 7457

	.set_tss_addr = svm_set_tss_addr,
7458
	.set_identity_map_addr = svm_set_identity_map_addr,
7459
	.get_tdp_level = get_npt_level,
7460
	.get_mt_mask = svm_get_mt_mask,
7461

7462 7463
	.get_exit_info = svm_get_exit_info,

7464
	.get_lpage_level = svm_get_lpage_level,
7465 7466

	.cpuid_update = svm_cpuid_update,
7467 7468

	.rdtscp_supported = svm_rdtscp_supported,
7469
	.invpcid_supported = svm_invpcid_supported,
7470
	.mpx_supported = svm_mpx_supported,
7471
	.xsaves_supported = svm_xsaves_supported,
7472
	.umip_emulated = svm_umip_emulated,
7473
	.pt_supported = svm_pt_supported,
7474
	.pku_supported = svm_pku_supported,
7475 7476

	.set_supported_cpuid = svm_set_supported_cpuid,
7477 7478

	.has_wbinvd_exit = svm_has_wbinvd_exit,
7479

7480
	.read_l1_tsc_offset = svm_read_l1_tsc_offset,
7481
	.write_l1_tsc_offset = svm_write_l1_tsc_offset,
7482 7483

	.set_tdp_cr3 = set_tdp_cr3,
7484 7485

	.check_intercept = svm_check_intercept,
7486
	.handle_exit_irqoff = svm_handle_exit_irqoff,
7487

7488 7489
	.request_immediate_exit = __kvm_request_immediate_exit,

7490
	.sched_in = svm_sched_in,
7491 7492

	.pmu_ops = &amd_pmu_ops,
7493
	.deliver_posted_interrupt = svm_deliver_avic_intr,
7494
	.dy_apicv_has_pending_interrupt = svm_dy_apicv_has_pending_interrupt,
7495
	.update_pi_irte = svm_update_pi_irte,
7496
	.setup_mce = svm_setup_mce,
7497

7498
	.smi_allowed = svm_smi_allowed,
7499 7500
	.pre_enter_smm = svm_pre_enter_smm,
	.pre_leave_smm = svm_pre_leave_smm,
7501
	.enable_smi_window = enable_smi_window,
B
Brijesh Singh 已提交
7502 7503

	.mem_enc_op = svm_mem_enc_op,
7504 7505
	.mem_enc_reg_region = svm_register_enc_region,
	.mem_enc_unreg_region = svm_unregister_enc_region,
7506

7507
	.nested_enable_evmcs = NULL,
7508
	.nested_get_evmcs_version = NULL,
7509 7510

	.need_emulation_on_page_fault = svm_need_emulation_on_page_fault,
7511 7512

	.apic_init_signal_blocked = svm_apic_init_signal_blocked,
A
Avi Kivity 已提交
7513 7514 7515 7516
};

static int __init svm_init(void)
{
7517
	return kvm_init(&svm_x86_ops, sizeof(struct vcpu_svm),
7518
			__alignof__(struct vcpu_svm), THIS_MODULE);
A
Avi Kivity 已提交
7519 7520 7521 7522
}

static void __exit svm_exit(void)
{
7523
	kvm_exit();
A
Avi Kivity 已提交
7524 7525 7526 7527
}

module_init(svm_init)
module_exit(svm_exit)