svm.c 192.2 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
	int r;

1010 1011
	sd = kzalloc(sizeof(struct svm_cpu_data), GFP_KERNEL);
	if (!sd)
A
Avi Kivity 已提交
1012
		return -ENOMEM;
1013
	sd->cpu = cpu;
A
Avi Kivity 已提交
1014
	r = -ENOMEM;
1015
	sd->save_area = alloc_page(GFP_KERNEL);
1016
	if (!sd->save_area)
A
Avi Kivity 已提交
1017 1018
		goto err_1;

1019 1020
	if (svm_sev_enabled()) {
		r = -ENOMEM;
1021 1022 1023
		sd->sev_vmcbs = kmalloc_array(max_sev_asid + 1,
					      sizeof(void *),
					      GFP_KERNEL);
1024 1025 1026 1027
		if (!sd->sev_vmcbs)
			goto err_1;
	}

1028
	per_cpu(svm_data, cpu) = sd;
A
Avi Kivity 已提交
1029 1030 1031 1032

	return 0;

err_1:
1033
	kfree(sd);
A
Avi Kivity 已提交
1034 1035 1036 1037
	return r;

}

1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048
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;
}

1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067
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);
}

1068 1069
static void set_msr_interception(u32 *msrpm, unsigned msr,
				 int read, int write)
A
Avi Kivity 已提交
1070
{
1071 1072 1073
	u8 bit_read, bit_write;
	unsigned long tmp;
	u32 offset;
A
Avi Kivity 已提交
1074

1075 1076 1077 1078 1079 1080
	/*
	 * If this warning triggers extend the direct_access_msrs list at the
	 * beginning of the file
	 */
	WARN_ON(!valid_msr_intercept(msr));

1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091
	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
1092 1093
}

1094
static void svm_vcpu_init_msrpm(u32 *msrpm)
A
Avi Kivity 已提交
1095 1096 1097
{
	int i;

1098 1099
	memset(msrpm, 0xff, PAGE_SIZE * (1 << MSRPM_ALLOC_ORDER));

1100 1101 1102 1103 1104 1105
	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);
	}
1106 1107
}

1108 1109 1110 1111 1112 1113 1114 1115
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)
1116
			return;
1117 1118 1119 1120 1121 1122 1123 1124 1125

		/* 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
1126
	}
1127 1128 1129 1130 1131

	/*
	 * If this BUG triggers the msrpm_offsets table has an overflow. Just
	 * increase MSRPM_OFFSETS in this case.
	 */
1132
	BUG();
A
Avi Kivity 已提交
1133 1134
}

1135
static void init_msrpm_offsets(void)
1136
{
1137
	int i;
1138

1139 1140 1141 1142 1143 1144 1145 1146 1147 1148
	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);
	}
1149 1150
}

1151 1152 1153 1154
static void svm_enable_lbrv(struct vcpu_svm *svm)
{
	u32 *msrpm = svm->msrpm;

1155
	svm->vmcb->control.virt_ext |= LBR_CTL_ENABLE_MASK;
1156 1157 1158 1159 1160 1161 1162 1163 1164 1165
	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;

1166
	svm->vmcb->control.virt_ext &= ~LBR_CTL_ENABLE_MASK;
1167 1168 1169 1170 1171 1172
	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);
}

1173 1174 1175
static void disable_nmi_singlestep(struct vcpu_svm *svm)
{
	svm->nmi_singlestep = false;
1176

1177 1178 1179 1180 1181 1182 1183
	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;
	}
1184 1185
}

1186
/* Note:
1187
 * This hash table is used to map VM_ID to a struct kvm_svm,
1188 1189 1190 1191
 * when handling AMD IOMMU GALOG notification to schedule in
 * a particular vCPU.
 */
#define SVM_VM_DATA_HASH_BITS	8
1192
static DEFINE_HASHTABLE(svm_vm_data_hash, SVM_VM_DATA_HASH_BITS);
1193 1194
static u32 next_vm_id = 0;
static bool next_vm_id_wrapped = 0;
1195
static DEFINE_SPINLOCK(svm_vm_data_hash_lock);
1196 1197 1198 1199 1200 1201 1202 1203

/* 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;
1204
	struct kvm_svm *kvm_svm;
1205 1206 1207 1208 1209 1210 1211
	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);
1212 1213
	hash_for_each_possible(svm_vm_data_hash, kvm_svm, hnode, vm_id) {
		if (kvm_svm->avic_vm_id != vm_id)
1214
			continue;
1215
		vcpu = kvm_get_vcpu_by_id(&kvm_svm->kvm, vcpu_id);
1216 1217 1218 1219 1220 1221 1222 1223 1224
		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.
	 */
1225
	if (vcpu)
1226 1227 1228 1229 1230
		kvm_vcpu_wake_up(vcpu);

	return 0;
}

B
Brijesh Singh 已提交
1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241
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
1242 1243 1244
	/* Minimum ASID value that should be used for SEV guest */
	min_sev_asid = cpuid_edx(0x8000001F);

1245
	/* Initialize SEV ASID bitmaps */
1246
	sev_asid_bitmap = bitmap_zalloc(max_sev_asid, GFP_KERNEL);
B
Brijesh Singh 已提交
1247 1248 1249
	if (!sev_asid_bitmap)
		return 1;

1250 1251 1252 1253
	sev_reclaim_asid_bitmap = bitmap_zalloc(max_sev_asid, GFP_KERNEL);
	if (!sev_reclaim_asid_bitmap)
		return 1;

B
Brijesh Singh 已提交
1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275
	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;
}

1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286
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
1287
	if (control->pause_filter_count != old) {
1288
		mark_dirty(svm->vmcb, VMCB_INTERCEPTS);
P
Peter Xu 已提交
1289 1290 1291
		trace_kvm_ple_window_update(vcpu->vcpu_id,
					    control->pause_filter_count, old);
	}
1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304
}

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
1305
	if (control->pause_filter_count != old) {
1306
		mark_dirty(svm->vmcb, VMCB_INTERCEPTS);
P
Peter Xu 已提交
1307 1308 1309
		trace_kvm_ple_window_update(vcpu->vcpu_id,
					    control->pause_filter_count, old);
	}
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 1352
/*
 * 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);
}

A
Avi Kivity 已提交
1353 1354 1355 1356
static __init int svm_hardware_setup(void)
{
	int cpu;
	struct page *iopm_pages;
1357
	void *iopm_va;
A
Avi Kivity 已提交
1358 1359 1360 1361 1362 1363
	int r;

	iopm_pages = alloc_pages(GFP_KERNEL, IOPM_ALLOC_ORDER);

	if (!iopm_pages)
		return -ENOMEM;
1364 1365 1366

	iopm_va = page_address(iopm_pages);
	memset(iopm_va, 0xff, PAGE_SIZE * (1 << IOPM_ALLOC_ORDER));
A
Avi Kivity 已提交
1367 1368
	iopm_base = page_to_pfn(iopm_pages) << PAGE_SHIFT;

1369 1370
	init_msrpm_offsets();

1371 1372 1373
	if (boot_cpu_has(X86_FEATURE_NX))
		kvm_enable_efer_bits(EFER_NX);

A
Alexander Graf 已提交
1374 1375 1376
	if (boot_cpu_has(X86_FEATURE_FXSR_OPT))
		kvm_enable_efer_bits(EFER_FFXSR);

1377 1378
	if (boot_cpu_has(X86_FEATURE_TSCRATEMSR)) {
		kvm_has_tsc_control = true;
1379 1380
		kvm_max_tsc_scaling_ratio = TSC_RATIO_MAX;
		kvm_tsc_scaling_ratio_frac_bits = 32;
1381 1382
	}

1383 1384 1385 1386 1387 1388 1389 1390
	/* 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;
	}

1391 1392
	if (nested) {
		printk(KERN_INFO "kvm: Nested Virtualization enabled\n");
1393
		kvm_enable_efer_bits(EFER_SVME | EFER_LMSLE);
1394 1395
	}

B
Brijesh Singh 已提交
1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406
	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;
		}
	}

1407 1408
	svm_adjust_mmio_mask();

Z
Zachary Amsden 已提交
1409
	for_each_possible_cpu(cpu) {
A
Avi Kivity 已提交
1410 1411
		r = svm_cpu_init(cpu);
		if (r)
1412
			goto err;
A
Avi Kivity 已提交
1413
	}
1414

1415
	if (!boot_cpu_has(X86_FEATURE_NPT))
1416 1417
		npt_enabled = false;

1418 1419 1420 1421 1422
	if (npt_enabled && !npt) {
		printk(KERN_INFO "kvm: Nested Paging disabled\n");
		npt_enabled = false;
	}

1423
	if (npt_enabled) {
1424
		printk(KERN_INFO "kvm: Nested Paging enabled\n");
1425
		kvm_enable_tdp();
1426 1427
	} else
		kvm_disable_tdp();
1428

1429 1430 1431 1432 1433
	if (nrips) {
		if (!boot_cpu_has(X86_FEATURE_NRIPS))
			nrips = false;
	}

1434 1435 1436
	if (avic) {
		if (!npt_enabled ||
		    !boot_cpu_has(X86_FEATURE_AVIC) ||
1437
		    !IS_ENABLED(CONFIG_X86_LOCAL_APIC)) {
1438
			avic = false;
1439
		} else {
1440
			pr_info("AVIC enabled\n");
1441 1442 1443

			amd_iommu_register_ga_log_notifier(&avic_ga_log_notifier);
		}
1444
	}
1445

1446 1447
	if (vls) {
		if (!npt_enabled ||
1448
		    !boot_cpu_has(X86_FEATURE_V_VMSAVE_VMLOAD) ||
1449 1450 1451 1452 1453 1454 1455
		    !IS_ENABLED(CONFIG_X86_64)) {
			vls = false;
		} else {
			pr_info("Virtual VMLOAD VMSAVE supported\n");
		}
	}

1456 1457 1458 1459 1460 1461 1462
	if (vgif) {
		if (!boot_cpu_has(X86_FEATURE_VGIF))
			vgif = false;
		else
			pr_info("Virtual GIF supported\n");
	}

A
Avi Kivity 已提交
1463 1464
	return 0;

1465
err:
A
Avi Kivity 已提交
1466 1467 1468 1469 1470 1471 1472
	__free_pages(iopm_pages, IOPM_ALLOC_ORDER);
	iopm_base = 0;
	return r;
}

static __exit void svm_hardware_unsetup(void)
{
1473 1474
	int cpu;

1475
	if (svm_sev_enabled()) {
1476
		bitmap_free(sev_asid_bitmap);
1477 1478 1479 1480
		bitmap_free(sev_reclaim_asid_bitmap);

		sev_flush_asids();
	}
B
Brijesh Singh 已提交
1481

Z
Zachary Amsden 已提交
1482
	for_each_possible_cpu(cpu)
1483 1484
		svm_cpu_uninit(cpu);

A
Avi Kivity 已提交
1485
	__free_pages(pfn_to_page(iopm_base >> PAGE_SHIFT), IOPM_ALLOC_ORDER);
1486
	iopm_base = 0;
A
Avi Kivity 已提交
1487 1488 1489 1490 1491 1492
}

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
1493
		      SVM_SELECTOR_WRITE_MASK; /* Read/Write Data Segment */
A
Avi Kivity 已提交
1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505
	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;
}

1506 1507 1508 1509 1510 1511 1512 1513 1514 1515
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;
}

1516
static u64 svm_write_l1_tsc_offset(struct kvm_vcpu *vcpu, u64 offset)
1517 1518 1519 1520
{
	struct vcpu_svm *svm = to_svm(vcpu);
	u64 g_tsc_offset = 0;

1521
	if (is_guest_mode(vcpu)) {
1522
		/* Write L1's TSC offset.  */
1523 1524 1525
		g_tsc_offset = svm->vmcb->control.tsc_offset -
			       svm->nested.hsave->control.tsc_offset;
		svm->nested.hsave->control.tsc_offset = offset;
1526 1527 1528 1529 1530
	}

	trace_kvm_write_tsc_offset(vcpu->vcpu_id,
				   svm->vmcb->control.tsc_offset - g_tsc_offset,
				   offset);
1531 1532

	svm->vmcb->control.tsc_offset = offset + g_tsc_offset;
1533 1534

	mark_dirty(svm->vmcb, VMCB_INTERCEPTS);
1535
	return svm->vmcb->control.tsc_offset;
1536 1537
}

1538 1539 1540
static void avic_init_vmcb(struct vcpu_svm *svm)
{
	struct vmcb *vmcb = svm->vmcb;
1541
	struct kvm_svm *kvm_svm = to_kvm_svm(svm->vcpu.kvm);
1542
	phys_addr_t bpa = __sme_set(page_to_phys(svm->avic_backing_page));
1543 1544
	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));
1545 1546 1547 1548 1549

	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;
1550 1551 1552 1553
	if (kvm_apicv_activated(svm->vcpu.kvm))
		vmcb->control.int_ctl |= AVIC_ENABLE_MASK;
	else
		vmcb->control.int_ctl &= ~AVIC_ENABLE_MASK;
1554 1555
}

P
Paolo Bonzini 已提交
1556
static void init_vmcb(struct vcpu_svm *svm)
A
Avi Kivity 已提交
1557
{
1558 1559
	struct vmcb_control_area *control = &svm->vmcb->control;
	struct vmcb_save_area *save = &svm->vmcb->save;
A
Avi Kivity 已提交
1560

1561
	svm->vcpu.arch.hflags = 0;
1562

1563 1564 1565 1566 1567 1568
	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);
1569 1570
	if (!kvm_vcpu_apicv_active(&svm->vcpu))
		set_cr_intercept(svm, INTERCEPT_CR8_WRITE);
A
Avi Kivity 已提交
1571

1572
	set_dr_intercepts(svm);
A
Avi Kivity 已提交
1573

1574 1575 1576
	set_exception_intercept(svm, PF_VECTOR);
	set_exception_intercept(svm, UD_VECTOR);
	set_exception_intercept(svm, MC_VECTOR);
1577
	set_exception_intercept(svm, AC_VECTOR);
1578
	set_exception_intercept(svm, DB_VECTOR);
1579 1580 1581 1582 1583 1584 1585 1586
	/*
	 * 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
1587

1588 1589 1590 1591
	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
1592
	set_intercept(svm, INTERCEPT_RDPMC);
1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608
	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
1609
	set_intercept(svm, INTERCEPT_XSETBV);
J
Jim Mattson 已提交
1610
	set_intercept(svm, INTERCEPT_RDPRU);
B
Brijesh Singh 已提交
1611
	set_intercept(svm, INTERCEPT_RSM);
A
Avi Kivity 已提交
1612

1613
	if (!kvm_mwait_in_guest(svm->vcpu.kvm)) {
1614 1615 1616 1617
		set_intercept(svm, INTERCEPT_MONITOR);
		set_intercept(svm, INTERCEPT_MWAIT);
	}

1618 1619 1620
	if (!kvm_hlt_in_guest(svm->vcpu.kvm))
		set_intercept(svm, INTERCEPT_HLT);

1621 1622
	control->iopm_base_pa = __sme_set(iopm_base);
	control->msrpm_base_pa = __sme_set(__pa(svm->msrpm));
A
Avi Kivity 已提交
1623 1624 1625 1626 1627 1628 1629 1630 1631
	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;
1632
	save->cs.base = 0xffff0000;
A
Avi Kivity 已提交
1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643
	/* 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
1644
	svm_set_efer(&svm->vcpu, 0);
M
Mike Day 已提交
1645
	save->dr6 = 0xffff0ff0;
1646
	kvm_set_rflags(&svm->vcpu, 2);
A
Avi Kivity 已提交
1647
	save->rip = 0x0000fff0;
1648
	svm->vcpu.arch.regs[VCPU_REGS_RIP] = save->rip;
A
Avi Kivity 已提交
1649

J
Joerg Roedel 已提交
1650
	/*
1651
	 * svm_set_cr0() sets PG and WP and clears NW and CD on save->cr0.
1652
	 * It also updates the guest-visible cr0 value.
A
Avi Kivity 已提交
1653
	 */
1654
	svm_set_cr0(&svm->vcpu, X86_CR0_NW | X86_CR0_CD | X86_CR0_ET);
1655
	kvm_mmu_reset_context(&svm->vcpu);
1656

1657
	save->cr4 = X86_CR4_PAE;
A
Avi Kivity 已提交
1658
	/* rdx = ?? */
1659 1660 1661

	if (npt_enabled) {
		/* Setup VMCB for Nested Paging */
1662
		control->nested_ctl |= SVM_NESTED_CTL_NP_ENABLE;
1663
		clr_intercept(svm, INTERCEPT_INVLPG);
1664
		clr_exception_intercept(svm, PF_VECTOR);
1665 1666
		clr_cr_intercept(svm, INTERCEPT_CR3_READ);
		clr_cr_intercept(svm, INTERCEPT_CR3_WRITE);
1667
		save->g_pat = svm->vcpu.arch.pat;
1668 1669 1670
		save->cr3 = 0;
		save->cr4 = 0;
	}
1671
	svm->asid_generation = 0;
1672

1673
	svm->nested.vmcb = 0;
1674 1675
	svm->vcpu.arch.hflags = 0;

1676 1677 1678 1679
	if (pause_filter_count) {
		control->pause_filter_count = pause_filter_count;
		if (pause_filter_thresh)
			control->pause_filter_thresh = pause_filter_thresh;
1680
		set_intercept(svm, INTERCEPT_PAUSE);
1681 1682
	} else {
		clr_intercept(svm, INTERCEPT_PAUSE);
1683 1684
	}

1685
	if (kvm_vcpu_apicv_active(&svm->vcpu))
1686 1687
		avic_init_vmcb(svm);

1688 1689 1690 1691 1692 1693 1694 1695 1696 1697
	/*
	 * 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;
	}

1698 1699 1700 1701 1702 1703
	if (vgif) {
		clr_intercept(svm, INTERCEPT_STGI);
		clr_intercept(svm, INTERCEPT_CLGI);
		svm->vmcb->control.int_ctl |= V_GIF_ENABLE_MASK;
	}

1704
	if (sev_guest(svm->vcpu.kvm)) {
B
Brijesh Singh 已提交
1705
		svm->vmcb->control.nested_ctl |= SVM_NESTED_CTL_SEV_ENABLE;
1706 1707
		clr_exception_intercept(svm, UD_VECTOR);
	}
B
Brijesh Singh 已提交
1708

1709 1710
	mark_all_dirty(svm->vmcb);

1711
	enable_gif(svm);
1712 1713 1714

}

1715 1716
static u64 *avic_get_physical_id_entry(struct kvm_vcpu *vcpu,
				       unsigned int index)
1717 1718
{
	u64 *avic_physical_id_table;
1719
	struct kvm_svm *kvm_svm = to_kvm_svm(vcpu->kvm);
1720 1721 1722 1723

	if (index >= AVIC_MAX_PHYSICAL_ID_COUNT)
		return NULL;

1724
	avic_physical_id_table = page_address(kvm_svm->avic_physical_id_table_page);
1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736

	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.
 */
1737
static int avic_update_access_page(struct kvm *kvm, bool activate)
1738
{
1739
	int ret = 0;
1740

1741
	mutex_lock(&kvm->slots_lock);
1742
	if (kvm->arch.apic_access_page_done == activate)
1743
		goto out;
1744

1745 1746 1747
	ret = __x86_set_memory_region(kvm,
				      APIC_ACCESS_PAGE_PRIVATE_MEMSLOT,
				      APIC_DEFAULT_PHYS_BASE,
1748
				      activate ? PAGE_SIZE : 0);
1749
	if (ret)
1750
		goto out;
1751

1752
	kvm->arch.apic_access_page_done = activate;
1753 1754 1755
out:
	mutex_unlock(&kvm->slots_lock);
	return ret;
1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769
}

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;

1770 1771 1772 1773 1774 1775 1776 1777
	if (kvm_apicv_activated(vcpu->kvm)) {
		int ret;

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

1778 1779 1780 1781 1782 1783 1784
	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;

1785 1786 1787
	new_entry = __sme_set((page_to_phys(svm->avic_backing_page) &
			      AVIC_PHYSICAL_ID_ENTRY_BACKING_PAGE_MASK) |
			      AVIC_PHYSICAL_ID_ENTRY_VALID_MASK);
1788 1789 1790 1791 1792 1793 1794
	WRITE_ONCE(*entry, new_entry);

	svm->avic_physical_id_cache = entry;

	return 0;
}

1795
static void sev_asid_free(int asid)
B
Brijesh Singh 已提交
1796
{
1797 1798
	struct svm_cpu_data *sd;
	int cpu, pos;
B
Brijesh Singh 已提交
1799

1800 1801
	mutex_lock(&sev_bitmap_lock);

B
Brijesh Singh 已提交
1802
	pos = asid - 1;
1803
	__set_bit(pos, sev_reclaim_asid_bitmap);
1804 1805 1806 1807 1808

	for_each_possible_cpu(cpu) {
		sd = per_cpu(svm_data, cpu);
		sd->sev_vmcbs[pos] = NULL;
	}
B
Brijesh Singh 已提交
1809

1810
	mutex_unlock(&sev_bitmap_lock);
B
Brijesh Singh 已提交
1811 1812
}

1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826
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;
1827

1828 1829
	/* Guard DEACTIVATE against WBINVD/DF_FLUSH used in ASID recycling */
	down_read(&sev_deactivate_lock);
1830
	sev_guest_deactivate(data, NULL);
1831
	up_read(&sev_deactivate_lock);
1832

1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845
	kfree(data);

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

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

	kfree(decommission);
}

1846 1847 1848 1849
static struct page **sev_pin_memory(struct kvm *kvm, unsigned long uaddr,
				    unsigned long ulen, unsigned long *n,
				    int write)
{
1850
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
1851 1852 1853
	unsigned long npages, npinned, size;
	unsigned long locked, lock_limit;
	struct page **pages;
1854 1855 1856 1857
	unsigned long first, last;

	if (ulen == 0 || uaddr + ulen < uaddr)
		return NULL;
1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873

	/* 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)
1874 1875
		pages = __vmalloc(size, GFP_KERNEL_ACCOUNT | __GFP_ZERO,
				  PAGE_KERNEL);
1876
	else
1877
		pages = kmalloc(size, GFP_KERNEL_ACCOUNT);
1878 1879 1880 1881 1882

	if (!pages)
		return NULL;

	/* Pin the user virtual address. */
1883
	npinned = get_user_pages_fast(uaddr, npages, FOLL_WRITE, pages);
1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904
	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)
{
1905
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926

	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);
	}
}

1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942
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);
}

1943 1944
static struct kvm *svm_vm_alloc(void)
{
1945 1946 1947
	struct kvm_svm *kvm_svm = __vmalloc(sizeof(struct kvm_svm),
					    GFP_KERNEL_ACCOUNT | __GFP_ZERO,
					    PAGE_KERNEL);
1948
	return &kvm_svm->kvm;
1949 1950 1951 1952
}

static void svm_vm_free(struct kvm *kvm)
{
1953
	vfree(to_kvm_svm(kvm));
1954 1955
}

B
Brijesh Singh 已提交
1956 1957
static void sev_vm_destroy(struct kvm *kvm)
{
1958
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
1959 1960
	struct list_head *head = &sev->regions_list;
	struct list_head *pos, *q;
1961

B
Brijesh Singh 已提交
1962 1963 1964
	if (!sev_guest(kvm))
		return;

1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979
	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);

1980
	sev_unbind_asid(kvm, sev->handle);
1981
	sev_asid_free(sev->asid);
B
Brijesh Singh 已提交
1982 1983
}

1984 1985
static void avic_vm_destroy(struct kvm *kvm)
{
1986
	unsigned long flags;
1987
	struct kvm_svm *kvm_svm = to_kvm_svm(kvm);
1988

1989 1990 1991
	if (!avic)
		return;

1992 1993 1994 1995
	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);
1996 1997

	spin_lock_irqsave(&svm_vm_data_hash_lock, flags);
1998
	hash_del(&kvm_svm->hnode);
1999
	spin_unlock_irqrestore(&svm_vm_data_hash_lock, flags);
2000 2001
}

B
Brijesh Singh 已提交
2002 2003 2004 2005 2006 2007
static void svm_vm_destroy(struct kvm *kvm)
{
	avic_vm_destroy(kvm);
	sev_vm_destroy(kvm);
}

2008 2009
static int avic_vm_init(struct kvm *kvm)
{
2010
	unsigned long flags;
2011
	int err = -ENOMEM;
2012 2013
	struct kvm_svm *kvm_svm = to_kvm_svm(kvm);
	struct kvm_svm *k2;
2014 2015
	struct page *p_page;
	struct page *l_page;
2016
	u32 vm_id;
2017 2018 2019 2020 2021

	if (!avic)
		return 0;

	/* Allocating physical APIC ID table (4KB) */
2022
	p_page = alloc_page(GFP_KERNEL_ACCOUNT);
2023 2024 2025
	if (!p_page)
		goto free_avic;

2026
	kvm_svm->avic_physical_id_table_page = p_page;
2027 2028 2029
	clear_page(page_address(p_page));

	/* Allocating logical APIC ID table (4KB) */
2030
	l_page = alloc_page(GFP_KERNEL_ACCOUNT);
2031 2032 2033
	if (!l_page)
		goto free_avic;

2034
	kvm_svm->avic_logical_id_table_page = l_page;
2035 2036
	clear_page(page_address(l_page));

2037
	spin_lock_irqsave(&svm_vm_data_hash_lock, flags);
2038 2039 2040 2041 2042 2043 2044 2045
 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) {
2046 2047
		hash_for_each_possible(svm_vm_data_hash, k2, hnode, vm_id) {
			if (k2->avic_vm_id == vm_id)
2048 2049 2050
				goto again;
		}
	}
2051 2052
	kvm_svm->avic_vm_id = vm_id;
	hash_add(svm_vm_data_hash, &kvm_svm->hnode, kvm_svm->avic_vm_id);
2053 2054
	spin_unlock_irqrestore(&svm_vm_data_hash_lock, flags);

2055 2056 2057 2058 2059
	return 0;

free_avic:
	avic_vm_destroy(kvm);
	return err;
A
Avi Kivity 已提交
2060 2061
}

2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073
static int svm_vm_init(struct kvm *kvm)
{
	if (avic) {
		int ret = avic_vm_init(kvm);
		if (ret)
			return ret;
	}

	kvm_apicv_init(kvm, avic && irqchip_split(kvm));
	return 0;
}

2074 2075
static inline int
avic_update_iommu_vcpu_affinity(struct kvm_vcpu *vcpu, int cpu, bool r)
2076
{
2077 2078 2079
	int ret = 0;
	unsigned long flags;
	struct amd_svm_iommu_ir *ir;
2080 2081
	struct vcpu_svm *svm = to_svm(vcpu);

2082 2083
	if (!kvm_arch_has_assigned_device(vcpu->kvm))
		return 0;
2084

2085 2086 2087 2088 2089
	/*
	 * 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);
2090

2091 2092
	if (list_empty(&svm->ir_list))
		goto out;
2093

2094 2095 2096 2097 2098 2099 2100 2101
	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;
2102 2103 2104 2105 2106 2107
}

static void avic_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
{
	u64 entry;
	/* ID = 0xff (broadcast), ID > 0xff (reserved) */
2108
	int h_physical_id = kvm_cpu_get_apicid(cpu);
2109 2110 2111 2112 2113
	struct vcpu_svm *svm = to_svm(vcpu);

	if (!kvm_vcpu_apicv_active(vcpu))
		return;

2114 2115 2116 2117 2118
	/*
	 * 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))
2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131
		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);
2132 2133
	avic_update_iommu_vcpu_affinity(vcpu, h_physical_id,
					svm->avic_is_running);
2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144
}

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));
2145 2146 2147
	if (entry & AVIC_PHYSICAL_ID_ENTRY_IS_RUNNING_MASK)
		avic_update_iommu_vcpu_affinity(vcpu, -1, 0);

2148 2149
	entry &= ~AVIC_PHYSICAL_ID_ENTRY_IS_RUNNING_MASK;
	WRITE_ONCE(*(svm->avic_physical_id_cache), entry);
A
Avi Kivity 已提交
2150 2151
}

2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165
/**
 * 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);
}

2166
static void svm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event)
2167 2168
{
	struct vcpu_svm *svm = to_svm(vcpu);
2169 2170
	u32 dummy;
	u32 eax = 1;
2171

2172
	vcpu->arch.microcode_version = 0x01000065;
2173
	svm->spec_ctrl = 0;
2174
	svm->virt_spec_ctrl = 0;
2175

2176 2177 2178 2179 2180 2181
	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
2182
	init_vmcb(svm);
A
Avi Kivity 已提交
2183

2184
	kvm_cpuid(vcpu, &eax, &dummy, &dummy, &dummy, true);
2185
	kvm_rdx_write(vcpu, eax);
2186 2187 2188

	if (kvm_vcpu_apicv_active(vcpu) && !init_event)
		avic_update_vapic_bar(svm, APIC_DEFAULT_PHYS_BASE);
2189 2190
}

2191 2192 2193 2194
static int avic_init_vcpu(struct vcpu_svm *svm)
{
	int ret;

2195
	if (!kvm_vcpu_apicv_active(&svm->vcpu))
2196 2197 2198 2199 2200 2201 2202 2203
		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);
2204
	svm->dfr_reg = APIC_DFR_FLAT;
2205 2206 2207 2208

	return ret;
}

2209
static int svm_create_vcpu(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
2210
{
2211
	struct vcpu_svm *svm;
A
Avi Kivity 已提交
2212
	struct page *page;
2213
	struct page *msrpm_pages;
A
Alexander Graf 已提交
2214
	struct page *hsave_page;
A
Alexander Graf 已提交
2215
	struct page *nested_msrpm_pages;
R
Rusty Russell 已提交
2216
	int err;
A
Avi Kivity 已提交
2217

2218 2219
	BUILD_BUG_ON(offsetof(struct vcpu_svm, vcpu) != 0);
	svm = to_svm(vcpu);
R
Rusty Russell 已提交
2220

2221
	err = -ENOMEM;
2222
	page = alloc_page(GFP_KERNEL_ACCOUNT);
2223
	if (!page)
2224
		goto out;
A
Avi Kivity 已提交
2225

2226
	msrpm_pages = alloc_pages(GFP_KERNEL_ACCOUNT, MSRPM_ALLOC_ORDER);
2227
	if (!msrpm_pages)
2228
		goto free_page1;
A
Alexander Graf 已提交
2229

2230
	nested_msrpm_pages = alloc_pages(GFP_KERNEL_ACCOUNT, MSRPM_ALLOC_ORDER);
A
Alexander Graf 已提交
2231
	if (!nested_msrpm_pages)
2232
		goto free_page2;
2233

2234
	hsave_page = alloc_page(GFP_KERNEL_ACCOUNT);
A
Alexander Graf 已提交
2235
	if (!hsave_page)
2236 2237
		goto free_page3;

2238 2239 2240
	err = avic_init_vcpu(svm);
	if (err)
		goto free_page4;
2241

2242 2243 2244
	/* We initialize this flag to true to make sure that the is_running
	 * bit would be set the first time the vcpu is loaded.
	 */
2245 2246
	if (irqchip_in_kernel(vcpu->kvm) && kvm_apicv_activated(vcpu->kvm))
		svm->avic_is_running = true;
2247

2248
	svm->nested.hsave = page_address(hsave_page);
A
Alexander Graf 已提交
2249

2250 2251 2252
	svm->msrpm = page_address(msrpm_pages);
	svm_vcpu_init_msrpm(svm->msrpm);

2253
	svm->nested.msrpm = page_address(nested_msrpm_pages);
2254
	svm_vcpu_init_msrpm(svm->nested.msrpm);
A
Alexander Graf 已提交
2255

2256 2257
	svm->vmcb = page_address(page);
	clear_page(svm->vmcb);
2258
	svm->vmcb_pa = __sme_set(page_to_pfn(page) << PAGE_SHIFT);
2259
	svm->asid_generation = 0;
P
Paolo Bonzini 已提交
2260
	init_vmcb(svm);
A
Avi Kivity 已提交
2261

2262
	svm_init_osvw(vcpu);
2263

2264
	return 0;
2265

2266 2267
free_page4:
	__free_page(hsave_page);
2268 2269 2270 2271 2272 2273
free_page3:
	__free_pages(nested_msrpm_pages, MSRPM_ALLOC_ORDER);
free_page2:
	__free_pages(msrpm_pages, MSRPM_ALLOC_ORDER);
free_page1:
	__free_page(page);
2274
out:
2275
	return err;
A
Avi Kivity 已提交
2276 2277
}

2278 2279 2280 2281 2282 2283 2284 2285
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
2286 2287
static void svm_free_vcpu(struct kvm_vcpu *vcpu)
{
2288 2289
	struct vcpu_svm *svm = to_svm(vcpu);

2290 2291 2292 2293 2294 2295 2296
	/*
	 * 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);

2297
	__free_page(pfn_to_page(__sme_clr(svm->vmcb_pa) >> PAGE_SHIFT));
2298
	__free_pages(virt_to_page(svm->msrpm), MSRPM_ALLOC_ORDER);
2299 2300
	__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
2301 2302
}

2303
static void svm_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
A
Avi Kivity 已提交
2304
{
2305
	struct vcpu_svm *svm = to_svm(vcpu);
A
Ashok Raj 已提交
2306
	struct svm_cpu_data *sd = per_cpu(svm_data, cpu);
2307
	int i;
2308 2309

	if (unlikely(cpu != vcpu->cpu)) {
2310
		svm->asid_generation = 0;
2311
		mark_all_dirty(svm->vmcb);
2312
	}
2313

2314 2315 2316
#ifdef CONFIG_X86_64
	rdmsrl(MSR_GS_BASE, to_svm(vcpu)->host.gs_base);
#endif
2317 2318 2319 2320
	savesegment(fs, svm->host.fs);
	savesegment(gs, svm->host.gs);
	svm->host.ldt = kvm_read_ldt();

2321
	for (i = 0; i < NR_HOST_SAVE_USER_MSRS; i++)
2322
		rdmsrl(host_save_user_msrs[i], svm->host_user_msrs[i]);
2323

2324 2325 2326 2327 2328 2329
	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);
		}
2330
	}
P
Paolo Bonzini 已提交
2331 2332 2333
	/* This assumes that the kernel never uses MSR_TSC_AUX */
	if (static_cpu_has(X86_FEATURE_RDTSCP))
		wrmsrl(MSR_TSC_AUX, svm->tsc_aux);
2334

A
Ashok Raj 已提交
2335 2336 2337 2338
	if (sd->current_vmcb != svm->vmcb) {
		sd->current_vmcb = svm->vmcb;
		indirect_branch_prediction_barrier();
	}
2339
	avic_vcpu_load(vcpu, cpu);
A
Avi Kivity 已提交
2340 2341 2342 2343
}

static void svm_vcpu_put(struct kvm_vcpu *vcpu)
{
2344
	struct vcpu_svm *svm = to_svm(vcpu);
2345 2346
	int i;

2347 2348
	avic_vcpu_put(vcpu);

2349
	++vcpu->stat.host_state_reload;
2350 2351 2352
	kvm_load_ldt(svm->host.ldt);
#ifdef CONFIG_X86_64
	loadsegment(fs, svm->host.fs);
2353
	wrmsrl(MSR_KERNEL_GS_BASE, current->thread.gsbase);
2354
	load_gs_index(svm->host.gs);
2355
#else
2356
#ifdef CONFIG_X86_32_LAZY_GS
2357
	loadsegment(gs, svm->host.gs);
2358
#endif
2359
#endif
2360
	for (i = 0; i < NR_HOST_SAVE_USER_MSRS; i++)
2361
		wrmsrl(host_save_user_msrs[i], svm->host_user_msrs[i]);
A
Avi Kivity 已提交
2362 2363
}

2364 2365 2366 2367 2368 2369 2370
static void svm_vcpu_blocking(struct kvm_vcpu *vcpu)
{
	avic_set_running(vcpu, false);
}

static void svm_vcpu_unblocking(struct kvm_vcpu *vcpu)
{
2371 2372
	if (kvm_check_request(KVM_REQ_APICV_UPDATE, vcpu))
		kvm_vcpu_update_apicv(vcpu);
2373 2374 2375
	avic_set_running(vcpu, true);
}

A
Avi Kivity 已提交
2376 2377
static unsigned long svm_get_rflags(struct kvm_vcpu *vcpu)
{
2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388
	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
2389 2390 2391 2392
}

static void svm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags)
{
2393 2394 2395
	if (to_svm(vcpu)->nmi_singlestep)
		rflags |= (X86_EFLAGS_TF | X86_EFLAGS_RF);

P
Paolo Bonzini 已提交
2396
       /*
A
Andrea Gelmini 已提交
2397
        * Any change of EFLAGS.VM is accompanied by a reload of SS
P
Paolo Bonzini 已提交
2398 2399 2400
        * (caused by either a task switch or an inter-privilege IRET),
        * so we do not need to update the CPL here.
        */
2401
	to_svm(vcpu)->vmcb->save.rflags = rflags;
A
Avi Kivity 已提交
2402 2403
}

A
Avi Kivity 已提交
2404 2405 2406 2407 2408
static void svm_cache_reg(struct kvm_vcpu *vcpu, enum kvm_reg reg)
{
	switch (reg) {
	case VCPU_EXREG_PDPTR:
		BUG_ON(!npt_enabled);
2409
		load_pdptrs(vcpu, vcpu->arch.walk_mmu, kvm_read_cr3(vcpu));
A
Avi Kivity 已提交
2410 2411
		break;
	default:
2412
		WARN_ON_ONCE(1);
A
Avi Kivity 已提交
2413 2414 2415
	}
}

2416 2417
static void svm_set_vintr(struct vcpu_svm *svm)
{
2418
	set_intercept(svm, INTERCEPT_VINTR);
2419 2420 2421 2422
}

static void svm_clear_vintr(struct vcpu_svm *svm)
{
2423
	clr_intercept(svm, INTERCEPT_VINTR);
2424 2425
}

A
Avi Kivity 已提交
2426 2427
static struct vmcb_seg *svm_seg(struct kvm_vcpu *vcpu, int seg)
{
2428
	struct vmcb_save_area *save = &to_svm(vcpu)->vmcb->save;
A
Avi Kivity 已提交
2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440

	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
2441
	return NULL;
A
Avi Kivity 已提交
2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465
}

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;
2466 2467 2468 2469 2470 2471 2472 2473 2474 2475

	/*
	 * 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;
2476

J
Joerg Roedel 已提交
2477 2478
	/*
	 * AMD's VMCB does not have an explicit unusable field, so emulate it
2479 2480
	 * for cross vendor migration purposes by "not present"
	 */
2481
	var->unusable = !var->present;
2482

2483 2484 2485 2486 2487 2488
	switch (seg) {
	case VCPU_SREG_TR:
		/*
		 * Work around a bug where the busy flag in the tr selector
		 * isn't exposed
		 */
2489
		var->type |= 0x2;
2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504
		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;
2505
	case VCPU_SREG_SS:
J
Joerg Roedel 已提交
2506 2507
		/*
		 * On AMD CPUs sometimes the DB bit in the segment
2508 2509 2510 2511 2512 2513
		 * 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;
2514
		/* This is symmetric with svm_set_segment() */
J
Jan Kiszka 已提交
2515
		var->dpl = to_svm(vcpu)->vmcb->save.cpl;
2516
		break;
2517
	}
A
Avi Kivity 已提交
2518 2519
}

2520 2521 2522 2523 2524 2525 2526
static int svm_get_cpl(struct kvm_vcpu *vcpu)
{
	struct vmcb_save_area *save = &to_svm(vcpu)->vmcb->save;

	return save->cpl;
}

2527
static void svm_get_idt(struct kvm_vcpu *vcpu, struct desc_ptr *dt)
A
Avi Kivity 已提交
2528
{
2529 2530
	struct vcpu_svm *svm = to_svm(vcpu);

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

2535
static void svm_set_idt(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
	svm->vmcb->save.idtr.limit = dt->size;
	svm->vmcb->save.idtr.base = dt->address ;
2541
	mark_dirty(svm->vmcb, VMCB_DT);
A
Avi Kivity 已提交
2542 2543
}

2544
static void svm_get_gdt(struct kvm_vcpu *vcpu, struct desc_ptr *dt)
A
Avi Kivity 已提交
2545
{
2546 2547
	struct vcpu_svm *svm = to_svm(vcpu);

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

2552
static void svm_set_gdt(struct kvm_vcpu *vcpu, struct desc_ptr *dt)
A
Avi Kivity 已提交
2553
{
2554 2555
	struct vcpu_svm *svm = to_svm(vcpu);

2556 2557
	svm->vmcb->save.gdtr.limit = dt->size;
	svm->vmcb->save.gdtr.base = dt->address ;
2558
	mark_dirty(svm->vmcb, VMCB_DT);
A
Avi Kivity 已提交
2559 2560
}

2561 2562 2563 2564
static void svm_decache_cr0_guest_bits(struct kvm_vcpu *vcpu)
{
}

2565
static void svm_decache_cr4_guest_bits(struct kvm_vcpu *vcpu)
2566 2567 2568
{
}

A
Avi Kivity 已提交
2569 2570 2571 2572 2573
static void update_cr0_intercept(struct vcpu_svm *svm)
{
	ulong gcr0 = svm->vcpu.arch.cr0;
	u64 *hcr0 = &svm->vmcb->save.cr0;

2574 2575
	*hcr0 = (*hcr0 & ~SVM_CR0_SELECTIVE_MASK)
		| (gcr0 & SVM_CR0_SELECTIVE_MASK);
A
Avi Kivity 已提交
2576

2577
	mark_dirty(svm->vmcb, VMCB_CR);
A
Avi Kivity 已提交
2578

2579
	if (gcr0 == *hcr0) {
2580 2581
		clr_cr_intercept(svm, INTERCEPT_CR0_READ);
		clr_cr_intercept(svm, INTERCEPT_CR0_WRITE);
A
Avi Kivity 已提交
2582
	} else {
2583 2584
		set_cr_intercept(svm, INTERCEPT_CR0_READ);
		set_cr_intercept(svm, INTERCEPT_CR0_WRITE);
A
Avi Kivity 已提交
2585 2586 2587
	}
}

A
Avi Kivity 已提交
2588 2589
static void svm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
{
2590 2591
	struct vcpu_svm *svm = to_svm(vcpu);

2592
#ifdef CONFIG_X86_64
2593
	if (vcpu->arch.efer & EFER_LME) {
2594
		if (!is_paging(vcpu) && (cr0 & X86_CR0_PG)) {
2595
			vcpu->arch.efer |= EFER_LMA;
2596
			svm->vmcb->save.efer |= EFER_LMA | EFER_LME;
A
Avi Kivity 已提交
2597 2598
		}

M
Mike Day 已提交
2599
		if (is_paging(vcpu) && !(cr0 & X86_CR0_PG)) {
2600
			vcpu->arch.efer &= ~EFER_LMA;
2601
			svm->vmcb->save.efer &= ~(EFER_LMA | EFER_LME);
A
Avi Kivity 已提交
2602 2603 2604
		}
	}
#endif
2605
	vcpu->arch.cr0 = cr0;
2606 2607 2608

	if (!npt_enabled)
		cr0 |= X86_CR0_PG | X86_CR0_WP;
2609

2610 2611 2612 2613 2614 2615 2616
	/*
	 * 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);
2617
	svm->vmcb->save.cr0 = cr0;
2618
	mark_dirty(svm->vmcb, VMCB_CR);
A
Avi Kivity 已提交
2619
	update_cr0_intercept(svm);
A
Avi Kivity 已提交
2620 2621
}

2622
static int svm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
A
Avi Kivity 已提交
2623
{
2624
	unsigned long host_cr4_mce = cr4_read_shadow() & X86_CR4_MCE;
2625 2626
	unsigned long old_cr4 = to_svm(vcpu)->vmcb->save.cr4;

2627 2628 2629
	if (cr4 & X86_CR4_VMXE)
		return 1;

2630
	if (npt_enabled && ((old_cr4 ^ cr4) & X86_CR4_PGE))
2631
		svm_flush_tlb(vcpu, true);
2632

2633 2634 2635
	vcpu->arch.cr4 = cr4;
	if (!npt_enabled)
		cr4 |= X86_CR4_PAE;
2636
	cr4 |= host_cr4_mce;
2637
	to_svm(vcpu)->vmcb->save.cr4 = cr4;
2638
	mark_dirty(to_svm(vcpu)->vmcb, VMCB_CR);
2639
	return 0;
A
Avi Kivity 已提交
2640 2641 2642 2643 2644
}

static void svm_set_segment(struct kvm_vcpu *vcpu,
			    struct kvm_segment *var, int seg)
{
2645
	struct vcpu_svm *svm = to_svm(vcpu);
A
Avi Kivity 已提交
2646 2647 2648 2649 2650
	struct vmcb_seg *s = svm_seg(vcpu, seg);

	s->base = var->base;
	s->limit = var->limit;
	s->selector = var->selector;
2651 2652 2653 2654 2655 2656 2657 2658
	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
2659 2660 2661 2662 2663 2664 2665 2666

	/*
	 * 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)
2667 2668
		/* This is symmetric with svm_get_segment() */
		svm->vmcb->save.cpl = (var->dpl & 3);
A
Avi Kivity 已提交
2669

2670
	mark_dirty(svm->vmcb, VMCB_SEG);
A
Avi Kivity 已提交
2671 2672
}

2673
static void update_bp_intercept(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
2674
{
J
Jan Kiszka 已提交
2675 2676
	struct vcpu_svm *svm = to_svm(vcpu);

2677
	clr_exception_intercept(svm, BP_VECTOR);
2678

J
Jan Kiszka 已提交
2679 2680
	if (vcpu->guest_debug & KVM_GUESTDBG_ENABLE) {
		if (vcpu->guest_debug & KVM_GUESTDBG_USE_SW_BP)
2681
			set_exception_intercept(svm, BP_VECTOR);
J
Jan Kiszka 已提交
2682 2683
	} else
		vcpu->guest_debug = 0;
2684 2685
}

2686
static void new_asid(struct vcpu_svm *svm, struct svm_cpu_data *sd)
A
Avi Kivity 已提交
2687
{
2688 2689
	if (sd->next_asid > sd->max_asid) {
		++sd->asid_generation;
2690
		sd->next_asid = sd->min_asid;
2691
		svm->vmcb->control.tlb_ctl = TLB_CONTROL_FLUSH_ALL_ASID;
A
Avi Kivity 已提交
2692 2693
	}

2694 2695
	svm->asid_generation = sd->asid_generation;
	svm->vmcb->control.asid = sd->next_asid++;
2696 2697

	mark_dirty(svm->vmcb, VMCB_ASID);
A
Avi Kivity 已提交
2698 2699
}

J
Jan Kiszka 已提交
2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712
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);
}

2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727
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);
}

2728
static void svm_set_dr7(struct kvm_vcpu *vcpu, unsigned long value)
A
Avi Kivity 已提交
2729
{
2730 2731
	struct vcpu_svm *svm = to_svm(vcpu);

2732
	svm->vmcb->save.dr7 = value;
2733
	mark_dirty(svm->vmcb, VMCB_DR);
A
Avi Kivity 已提交
2734 2735
}

A
Avi Kivity 已提交
2736
static int pf_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
2737
{
2738
	u64 fault_address = __sme_clr(svm->vmcb->control.exit_info_2);
2739
	u64 error_code = svm->vmcb->control.exit_info_1;
A
Avi Kivity 已提交
2740

2741
	return kvm_handle_page_fault(&svm->vcpu, error_code, fault_address,
2742 2743
			static_cpu_has(X86_FEATURE_DECODEASSISTS) ?
			svm->vmcb->control.insn_bytes : NULL,
2744 2745 2746 2747 2748
			svm->vmcb->control.insn_len);
}

static int npf_interception(struct vcpu_svm *svm)
{
2749
	u64 fault_address = __sme_clr(svm->vmcb->control.exit_info_2);
2750 2751 2752 2753
	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,
2754 2755
			static_cpu_has(X86_FEATURE_DECODEASSISTS) ?
			svm->vmcb->control.insn_bytes : NULL,
2756
			svm->vmcb->control.insn_len);
A
Avi Kivity 已提交
2757 2758
}

A
Avi Kivity 已提交
2759
static int db_interception(struct vcpu_svm *svm)
J
Jan Kiszka 已提交
2760
{
A
Avi Kivity 已提交
2761
	struct kvm_run *kvm_run = svm->vcpu.run;
2762
	struct kvm_vcpu *vcpu = &svm->vcpu;
A
Avi Kivity 已提交
2763

J
Jan Kiszka 已提交
2764
	if (!(svm->vcpu.guest_debug &
2765
	      (KVM_GUESTDBG_SINGLESTEP | KVM_GUESTDBG_USE_HW_BP)) &&
J
Jan Kiszka 已提交
2766
		!svm->nmi_singlestep) {
J
Jan Kiszka 已提交
2767 2768 2769
		kvm_queue_exception(&svm->vcpu, DB_VECTOR);
		return 1;
	}
2770

J
Jan Kiszka 已提交
2771
	if (svm->nmi_singlestep) {
2772
		disable_nmi_singlestep(svm);
2773 2774
		/* Make sure we check for pending NMIs upon entry */
		kvm_make_request(KVM_REQ_EVENT, vcpu);
2775 2776 2777
	}

	if (svm->vcpu.guest_debug &
J
Joerg Roedel 已提交
2778
	    (KVM_GUESTDBG_SINGLESTEP | KVM_GUESTDBG_USE_HW_BP)) {
2779 2780 2781 2782 2783 2784 2785 2786
		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
2787 2788
}

A
Avi Kivity 已提交
2789
static int bp_interception(struct vcpu_svm *svm)
J
Jan Kiszka 已提交
2790
{
A
Avi Kivity 已提交
2791 2792
	struct kvm_run *kvm_run = svm->vcpu.run;

J
Jan Kiszka 已提交
2793 2794 2795 2796 2797 2798
	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
2799
static int ud_interception(struct vcpu_svm *svm)
2800
{
W
Wanpeng Li 已提交
2801
	return handle_ud(&svm->vcpu);
2802 2803
}

2804 2805 2806 2807 2808 2809
static int ac_interception(struct vcpu_svm *svm)
{
	kvm_queue_exception_e(&svm->vcpu, AC_VECTOR, 0);
	return 1;
}

2810 2811 2812 2813 2814 2815 2816
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);

2817 2818 2819 2820 2821 2822 2823 2824
	/*
	 * 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;
	}
2825
	return kvm_emulate_instruction(vcpu, EMULTYPE_VMWARE_GP);
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 2858 2859 2860 2861 2862 2863 2864 2865 2866
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;
}

2867
static void svm_handle_mce(struct vcpu_svm *svm)
2868
{
2869 2870 2871 2872 2873 2874 2875
	if (is_erratum_383()) {
		/*
		 * Erratum 383 triggered. Guest state is corrupt so kill the
		 * guest.
		 */
		pr_err("KVM: Guest triggered AMD Erratum 383\n");

2876
		kvm_make_request(KVM_REQ_TRIPLE_FAULT, &svm->vcpu);
2877 2878 2879 2880

		return;
	}

2881 2882 2883 2884 2885 2886 2887 2888
	/*
	 * 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 */

2889 2890 2891 2892 2893
	return;
}

static int mc_interception(struct vcpu_svm *svm)
{
2894 2895 2896
	return 1;
}

A
Avi Kivity 已提交
2897
static int shutdown_interception(struct vcpu_svm *svm)
2898
{
A
Avi Kivity 已提交
2899 2900
	struct kvm_run *kvm_run = svm->vcpu.run;

2901 2902 2903 2904
	/*
	 * VMCB is undefined after a SHUTDOWN intercept
	 * so reinitialize it.
	 */
2905
	clear_page(svm->vmcb);
P
Paolo Bonzini 已提交
2906
	init_vmcb(svm);
2907 2908 2909 2910 2911

	kvm_run->exit_reason = KVM_EXIT_SHUTDOWN;
	return 0;
}

A
Avi Kivity 已提交
2912
static int io_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
2913
{
2914
	struct kvm_vcpu *vcpu = &svm->vcpu;
M
Mike Day 已提交
2915
	u32 io_info = svm->vmcb->control.exit_info_1; /* address size bug? */
2916
	int size, in, string;
2917
	unsigned port;
A
Avi Kivity 已提交
2918

R
Rusty Russell 已提交
2919
	++svm->vcpu.stat.io_exits;
2920
	string = (io_info & SVM_IOIO_STR_MASK) != 0;
2921
	in = (io_info & SVM_IOIO_TYPE_MASK) != 0;
2922
	if (string)
2923
		return kvm_emulate_instruction(vcpu, 0);
2924

2925 2926
	port = io_info >> 16;
	size = (io_info & SVM_IOIO_SIZE_MASK) >> SVM_IOIO_SIZE_SHIFT;
2927 2928
	svm->next_rip = svm->vmcb->control.exit_info_2;

2929
	return kvm_fast_pio(&svm->vcpu, size, port, in);
A
Avi Kivity 已提交
2930 2931
}

A
Avi Kivity 已提交
2932
static int nmi_interception(struct vcpu_svm *svm)
2933 2934 2935 2936
{
	return 1;
}

A
Avi Kivity 已提交
2937
static int intr_interception(struct vcpu_svm *svm)
2938 2939 2940 2941 2942
{
	++svm->vcpu.stat.irq_exits;
	return 1;
}

A
Avi Kivity 已提交
2943
static int nop_on_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
2944 2945 2946 2947
{
	return 1;
}

A
Avi Kivity 已提交
2948
static int halt_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
2949
{
R
Rusty Russell 已提交
2950
	return kvm_emulate_halt(&svm->vcpu);
A
Avi Kivity 已提交
2951 2952
}

A
Avi Kivity 已提交
2953
static int vmmcall_interception(struct vcpu_svm *svm)
2954
{
2955
	return kvm_emulate_hypercall(&svm->vcpu);
2956 2957
}

2958 2959 2960 2961 2962 2963 2964
static unsigned long nested_svm_get_tdp_cr3(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);

	return svm->nested.nested_cr3;
}

2965 2966 2967 2968 2969 2970 2971
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;

2972
	ret = kvm_vcpu_read_guest_page(vcpu, gpa_to_gfn(__sme_clr(cr3)), &pdpte,
2973
				       offset_in_page(cr3) + index * 8, 8);
2974 2975 2976 2977 2978
	if (ret)
		return 0;
	return pdpte;
}

2979 2980 2981 2982 2983
static void nested_svm_set_tdp_cr3(struct kvm_vcpu *vcpu,
				   unsigned long root)
{
	struct vcpu_svm *svm = to_svm(vcpu);

2984
	svm->vmcb->control.nested_cr3 = __sme_set(root);
2985
	mark_dirty(svm->vmcb, VMCB_NPT);
2986 2987
}

2988 2989
static void nested_svm_inject_npf_exit(struct kvm_vcpu *vcpu,
				       struct x86_exception *fault)
2990 2991 2992
{
	struct vcpu_svm *svm = to_svm(vcpu);

2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012
	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;
3013 3014 3015 3016

	nested_svm_vmexit(svm);
}

3017
static void nested_svm_init_mmu_context(struct kvm_vcpu *vcpu)
3018
{
3019
	WARN_ON(mmu_is_nested(vcpu));
3020 3021

	vcpu->arch.mmu = &vcpu->arch.guest_mmu;
3022
	kvm_init_shadow_mmu(vcpu);
3023 3024 3025 3026 3027 3028
	vcpu->arch.mmu->set_cr3           = nested_svm_set_tdp_cr3;
	vcpu->arch.mmu->get_cr3           = nested_svm_get_tdp_cr3;
	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);
3029 3030 3031 3032 3033
	vcpu->arch.walk_mmu              = &vcpu->arch.nested_mmu;
}

static void nested_svm_uninit_mmu_context(struct kvm_vcpu *vcpu)
{
3034
	vcpu->arch.mmu = &vcpu->arch.root_mmu;
3035
	vcpu->arch.walk_mmu = &vcpu->arch.root_mmu;
3036 3037
}

3038 3039
static int nested_svm_check_permissions(struct vcpu_svm *svm)
{
3040 3041
	if (!(svm->vcpu.arch.efer & EFER_SVME) ||
	    !is_paging(&svm->vcpu)) {
3042 3043 3044 3045 3046 3047 3048 3049 3050
		kvm_queue_exception(&svm->vcpu, UD_VECTOR);
		return 1;
	}

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

3051
	return 0;
3052 3053
}

3054 3055 3056
static int nested_svm_check_exception(struct vcpu_svm *svm, unsigned nr,
				      bool has_error_code, u32 error_code)
{
3057 3058
	int vmexit;

3059
	if (!is_guest_mode(&svm->vcpu))
3060
		return 0;
3061

3062 3063 3064 3065
	vmexit = nested_svm_intercept(svm);
	if (vmexit != NESTED_EXIT_DONE)
		return 0;

3066 3067 3068
	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;
3069 3070

	/*
3071 3072
	 * EXITINFO2 is undefined for all exception intercepts other
	 * than #PF.
3073
	 */
3074 3075
	if (svm->vcpu.arch.exception.nested_apf)
		svm->vmcb->control.exit_info_2 = svm->vcpu.arch.apf.nested_apf_token;
3076 3077
	else if (svm->vcpu.arch.exception.has_payload)
		svm->vmcb->control.exit_info_2 = svm->vcpu.arch.exception.payload;
3078 3079
	else
		svm->vmcb->control.exit_info_2 = svm->vcpu.arch.cr2;
3080

3081
	svm->nested.exit_required = true;
3082
	return vmexit;
3083 3084
}

3085 3086
/* This function returns true if it is save to enable the irq window */
static inline bool nested_svm_intr(struct vcpu_svm *svm)
3087
{
3088
	if (!is_guest_mode(&svm->vcpu))
3089
		return true;
3090

3091
	if (!(svm->vcpu.arch.hflags & HF_VINTR_MASK))
3092
		return true;
3093

3094
	if (!(svm->vcpu.arch.hflags & HF_HIF_MASK))
3095
		return false;
3096

3097 3098 3099 3100 3101 3102 3103 3104
	/*
	 * 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;

3105 3106 3107
	svm->vmcb->control.exit_code   = SVM_EXIT_INTR;
	svm->vmcb->control.exit_info_1 = 0;
	svm->vmcb->control.exit_info_2 = 0;
3108

3109 3110 3111
	if (svm->nested.intercept & 1ULL) {
		/*
		 * The #vmexit can't be emulated here directly because this
G
Guo Chao 已提交
3112
		 * code path runs with irqs and preemption disabled. A
3113 3114 3115 3116
		 * #vmexit emulation might sleep. Only signal request for
		 * the #vmexit here.
		 */
		svm->nested.exit_required = true;
3117
		trace_kvm_nested_intr_vmexit(svm->vmcb->save.rip);
3118
		return false;
3119 3120
	}

3121
	return true;
3122 3123
}

3124 3125 3126
/* This function returns true if it is save to enable the nmi window */
static inline bool nested_svm_nmi(struct vcpu_svm *svm)
{
3127
	if (!is_guest_mode(&svm->vcpu))
3128 3129 3130 3131 3132 3133 3134 3135 3136
		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;
3137 3138
}

3139 3140
static int nested_svm_intercept_ioio(struct vcpu_svm *svm)
{
3141 3142 3143
	unsigned port, size, iopm_len;
	u16 val, mask;
	u8 start_bit;
3144
	u64 gpa;
3145

3146 3147
	if (!(svm->nested.intercept & (1ULL << INTERCEPT_IOIO_PROT)))
		return NESTED_EXIT_HOST;
3148

3149
	port = svm->vmcb->control.exit_info_1 >> 16;
3150 3151
	size = (svm->vmcb->control.exit_info_1 & SVM_IOIO_SIZE_MASK) >>
		SVM_IOIO_SIZE_SHIFT;
3152
	gpa  = svm->nested.vmcb_iopm + (port / 8);
3153 3154 3155 3156
	start_bit = port % 8;
	iopm_len = (start_bit + size > 8) ? 2 : 1;
	mask = (0xf >> (4 - size)) << start_bit;
	val = 0;
3157

3158
	if (kvm_vcpu_read_guest(&svm->vcpu, gpa, &val, iopm_len))
3159
		return NESTED_EXIT_DONE;
3160

3161
	return (val & mask) ? NESTED_EXIT_DONE : NESTED_EXIT_HOST;
3162 3163
}

3164
static int nested_svm_exit_handled_msr(struct vcpu_svm *svm)
3165
{
3166 3167
	u32 offset, msr, value;
	int write, mask;
3168

3169
	if (!(svm->nested.intercept & (1ULL << INTERCEPT_MSR_PROT)))
3170
		return NESTED_EXIT_HOST;
3171

3172 3173 3174 3175
	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);
3176

3177 3178
	if (offset == MSR_INVALID)
		return NESTED_EXIT_DONE;
3179

3180 3181
	/* Offset is in 32 bit units but need in 8 bit units */
	offset *= 4;
3182

3183
	if (kvm_vcpu_read_guest(&svm->vcpu, svm->nested.vmcb_msrpm + offset, &value, 4))
3184
		return NESTED_EXIT_DONE;
3185

3186
	return (value & mask) ? NESTED_EXIT_DONE : NESTED_EXIT_HOST;
3187 3188
}

3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213
/* 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;
}

3214
static int nested_svm_exit_special(struct vcpu_svm *svm)
3215 3216
{
	u32 exit_code = svm->vmcb->control.exit_code;
3217

3218 3219 3220
	switch (exit_code) {
	case SVM_EXIT_INTR:
	case SVM_EXIT_NMI:
3221
	case SVM_EXIT_EXCP_BASE + MC_VECTOR:
3222 3223
		return NESTED_EXIT_HOST;
	case SVM_EXIT_NPF:
J
Joerg Roedel 已提交
3224
		/* For now we are always handling NPFs when using them */
3225 3226 3227 3228
		if (npt_enabled)
			return NESTED_EXIT_HOST;
		break;
	case SVM_EXIT_EXCP_BASE + PF_VECTOR:
G
Gleb Natapov 已提交
3229
		/* When we're shadowing, trap PFs, but not async PF */
3230
		if (!npt_enabled && svm->vcpu.arch.apf.host_apf_reason == 0)
3231 3232 3233 3234
			return NESTED_EXIT_HOST;
		break;
	default:
		break;
3235 3236
	}

3237 3238 3239 3240 3241 3242
	return NESTED_EXIT_CONTINUE;
}

/*
 * If this function returns true, this #vmexit was already handled
 */
3243
static int nested_svm_intercept(struct vcpu_svm *svm)
3244 3245 3246 3247
{
	u32 exit_code = svm->vmcb->control.exit_code;
	int vmexit = NESTED_EXIT_HOST;

3248
	switch (exit_code) {
3249
	case SVM_EXIT_MSR:
3250
		vmexit = nested_svm_exit_handled_msr(svm);
3251
		break;
3252 3253 3254
	case SVM_EXIT_IOIO:
		vmexit = nested_svm_intercept_ioio(svm);
		break;
3255 3256 3257
	case SVM_EXIT_READ_CR0 ... SVM_EXIT_WRITE_CR8: {
		u32 bit = 1U << (exit_code - SVM_EXIT_READ_CR0);
		if (svm->nested.intercept_cr & bit)
3258
			vmexit = NESTED_EXIT_DONE;
3259 3260
		break;
	}
3261 3262 3263
	case SVM_EXIT_READ_DR0 ... SVM_EXIT_WRITE_DR7: {
		u32 bit = 1U << (exit_code - SVM_EXIT_READ_DR0);
		if (svm->nested.intercept_dr & bit)
3264
			vmexit = NESTED_EXIT_DONE;
3265 3266 3267 3268
		break;
	}
	case SVM_EXIT_EXCP_BASE ... SVM_EXIT_EXCP_BASE + 0x1f: {
		u32 excp_bits = 1 << (exit_code - SVM_EXIT_EXCP_BASE);
3269 3270 3271 3272 3273 3274
		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
3275 3276
		/* async page fault always cause vmexit */
		else if ((exit_code == SVM_EXIT_EXCP_BASE + PF_VECTOR) &&
3277
			 svm->vcpu.arch.exception.nested_apf != 0)
G
Gleb Natapov 已提交
3278
			vmexit = NESTED_EXIT_DONE;
3279 3280
		break;
	}
3281 3282 3283 3284
	case SVM_EXIT_ERR: {
		vmexit = NESTED_EXIT_DONE;
		break;
	}
3285 3286
	default: {
		u64 exit_bits = 1ULL << (exit_code - SVM_EXIT_INTR);
J
Joerg Roedel 已提交
3287
		if (svm->nested.intercept & exit_bits)
3288
			vmexit = NESTED_EXIT_DONE;
3289 3290 3291
	}
	}

3292 3293 3294 3295 3296 3297 3298 3299 3300 3301
	return vmexit;
}

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

	vmexit = nested_svm_intercept(svm);

	if (vmexit == NESTED_EXIT_DONE)
3302 3303 3304
		nested_svm_vmexit(svm);

	return vmexit;
3305 3306
}

3307 3308 3309 3310 3311
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;

3312
	dst->intercept_cr         = from->intercept_cr;
3313
	dst->intercept_dr         = from->intercept_dr;
3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333
	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;
3334
	dst->virt_ext              = from->virt_ext;
3335 3336
	dst->pause_filter_count   = from->pause_filter_count;
	dst->pause_filter_thresh  = from->pause_filter_thresh;
3337 3338
}

3339
static int nested_svm_vmexit(struct vcpu_svm *svm)
3340
{
3341
	int rc;
3342
	struct vmcb *nested_vmcb;
3343
	struct vmcb *hsave = svm->nested.hsave;
J
Joerg Roedel 已提交
3344
	struct vmcb *vmcb = svm->vmcb;
3345
	struct kvm_host_map map;
3346

3347 3348 3349 3350
	trace_kvm_nested_vmexit_inject(vmcb->control.exit_code,
				       vmcb->control.exit_info_1,
				       vmcb->control.exit_info_2,
				       vmcb->control.exit_int_info,
3351 3352
				       vmcb->control.exit_int_info_err,
				       KVM_ISA_SVM);
3353

3354
	rc = kvm_vcpu_map(&svm->vcpu, gpa_to_gfn(svm->nested.vmcb), &map);
3355 3356 3357
	if (rc) {
		if (rc == -EINVAL)
			kvm_inject_gp(&svm->vcpu, 0);
3358
		return 1;
3359 3360 3361
	}

	nested_vmcb = map.hva;
3362

3363 3364
	/* Exit Guest-Mode */
	leave_guest_mode(&svm->vcpu);
3365 3366
	svm->nested.vmcb = 0;

3367
	/* Give the current vmcb to the guest */
J
Joerg Roedel 已提交
3368 3369 3370 3371 3372 3373 3374 3375
	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;
3376
	nested_vmcb->save.efer   = svm->vcpu.arch.efer;
3377
	nested_vmcb->save.cr0    = kvm_read_cr0(&svm->vcpu);
3378
	nested_vmcb->save.cr3    = kvm_read_cr3(&svm->vcpu);
J
Joerg Roedel 已提交
3379
	nested_vmcb->save.cr2    = vmcb->save.cr2;
3380
	nested_vmcb->save.cr4    = svm->vcpu.arch.cr4;
3381
	nested_vmcb->save.rflags = kvm_get_rflags(&svm->vcpu);
J
Joerg Roedel 已提交
3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397
	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;
3398 3399 3400

	if (svm->nrips_enabled)
		nested_vmcb->control.next_rip  = vmcb->control.next_rip;
3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416

	/*
	 * 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
3417 3418 3419
	nested_vmcb->control.tlb_ctl           = 0;
	nested_vmcb->control.event_inj         = 0;
	nested_vmcb->control.event_inj_err     = 0;
3420

3421 3422 3423 3424 3425
	nested_vmcb->control.pause_filter_count =
		svm->vmcb->control.pause_filter_count;
	nested_vmcb->control.pause_filter_thresh =
		svm->vmcb->control.pause_filter_thresh;

3426 3427 3428 3429 3430
	/* 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 */
3431
	copy_vmcb_control_area(vmcb, hsave);
3432

3433
	svm->vcpu.arch.tsc_offset = svm->vmcb->control.tsc_offset;
3434 3435
	kvm_clear_exception_queue(&svm->vcpu);
	kvm_clear_interrupt_queue(&svm->vcpu);
3436

3437 3438
	svm->nested.nested_cr3 = 0;

3439 3440 3441 3442 3443 3444 3445
	/* 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;
3446
	kvm_set_rflags(&svm->vcpu, hsave->save.rflags);
3447 3448 3449 3450 3451 3452 3453
	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 {
3454
		(void)kvm_set_cr3(&svm->vcpu, hsave->save.cr3);
3455
	}
3456
	kvm_rax_write(&svm->vcpu, hsave->save.rax);
3457 3458
	kvm_rsp_write(&svm->vcpu, hsave->save.rsp);
	kvm_rip_write(&svm->vcpu, hsave->save.rip);
3459 3460 3461 3462
	svm->vmcb->save.dr7 = 0;
	svm->vmcb->save.cpl = 0;
	svm->vmcb->control.exit_int_info = 0;

3463 3464
	mark_all_dirty(svm->vmcb);

3465
	kvm_vcpu_unmap(&svm->vcpu, &map, true);
3466

3467
	nested_svm_uninit_mmu_context(&svm->vcpu);
3468 3469 3470
	kvm_mmu_reset_context(&svm->vcpu);
	kvm_mmu_load(&svm->vcpu);

3471 3472 3473 3474 3475 3476 3477 3478
	/*
	 * 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);

3479 3480
	return 0;
}
A
Alexander Graf 已提交
3481

3482
static bool nested_svm_vmrun_msrpm(struct vcpu_svm *svm)
A
Alexander Graf 已提交
3483
{
3484 3485
	/*
	 * This function merges the msr permission bitmaps of kvm and the
G
Guo Chao 已提交
3486
	 * nested vmcb. It is optimized in that it only merges the parts where
3487 3488
	 * the kvm msr permission bitmap may contain zero bits
	 */
A
Alexander Graf 已提交
3489
	int i;
3490

3491 3492
	if (!(svm->nested.intercept & (1ULL << INTERCEPT_MSR_PROT)))
		return true;
3493

3494 3495 3496
	for (i = 0; i < MSRPM_OFFSETS; i++) {
		u32 value, p;
		u64 offset;
3497

3498 3499
		if (msrpm_offsets[i] == 0xffffffff)
			break;
A
Alexander Graf 已提交
3500

3501 3502
		p      = msrpm_offsets[i];
		offset = svm->nested.vmcb_msrpm + (p * 4);
3503

3504
		if (kvm_vcpu_read_guest(&svm->vcpu, offset, &value, 4))
3505 3506 3507 3508
			return false;

		svm->nested.msrpm[p] = svm->msrpm[p] | value;
	}
A
Alexander Graf 已提交
3509

3510
	svm->vmcb->control.msrpm_base_pa = __sme_set(__pa(svm->nested.msrpm));
3511 3512

	return true;
A
Alexander Graf 已提交
3513 3514
}

3515 3516 3517 3518 3519
static bool nested_vmcb_checks(struct vmcb *vmcb)
{
	if ((vmcb->control.intercept & (1ULL << INTERCEPT_VMRUN)) == 0)
		return false;

3520 3521 3522
	if (vmcb->control.asid == 0)
		return false;

3523 3524
	if ((vmcb->control.nested_ctl & SVM_NESTED_CTL_NP_ENABLE) &&
	    !npt_enabled)
3525 3526
		return false;

3527 3528 3529
	return true;
}

L
Ladi Prosek 已提交
3530
static void enter_svm_guest_mode(struct vcpu_svm *svm, u64 vmcb_gpa,
3531
				 struct vmcb *nested_vmcb, struct kvm_host_map *map)
A
Alexander Graf 已提交
3532
{
3533
	if (kvm_get_rflags(&svm->vcpu) & X86_EFLAGS_IF)
A
Alexander Graf 已提交
3534 3535 3536 3537
		svm->vcpu.arch.hflags |= HF_HIF_MASK;
	else
		svm->vcpu.arch.hflags &= ~HF_HIF_MASK;

3538
	if (nested_vmcb->control.nested_ctl & SVM_NESTED_CTL_NP_ENABLE) {
3539 3540 3541 3542
		svm->nested.nested_cr3 = nested_vmcb->control.nested_cr3;
		nested_svm_init_mmu_context(&svm->vcpu);
	}

A
Alexander Graf 已提交
3543 3544 3545 3546 3547 3548 3549
	/* 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;
3550
	kvm_set_rflags(&svm->vcpu, nested_vmcb->save.rflags);
A
Alexander Graf 已提交
3551 3552 3553 3554 3555 3556
	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;
3557
	} else
3558
		(void)kvm_set_cr3(&svm->vcpu, nested_vmcb->save.cr3);
3559 3560 3561 3562

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

J
Joerg Roedel 已提交
3563
	svm->vmcb->save.cr2 = svm->vcpu.arch.cr2 = nested_vmcb->save.cr2;
3564
	kvm_rax_write(&svm->vcpu, nested_vmcb->save.rax);
3565 3566
	kvm_rsp_write(&svm->vcpu, nested_vmcb->save.rsp);
	kvm_rip_write(&svm->vcpu, nested_vmcb->save.rip);
J
Joerg Roedel 已提交
3567

A
Alexander Graf 已提交
3568 3569 3570 3571 3572 3573 3574 3575
	/* 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;

3576
	svm->nested.vmcb_msrpm = nested_vmcb->control.msrpm_base_pa & ~0x0fffULL;
3577
	svm->nested.vmcb_iopm  = nested_vmcb->control.iopm_base_pa  & ~0x0fffULL;
A
Alexander Graf 已提交
3578

J
Joerg Roedel 已提交
3579
	/* cache intercepts */
3580
	svm->nested.intercept_cr         = nested_vmcb->control.intercept_cr;
3581
	svm->nested.intercept_dr         = nested_vmcb->control.intercept_dr;
J
Joerg Roedel 已提交
3582 3583 3584
	svm->nested.intercept_exceptions = nested_vmcb->control.intercept_exceptions;
	svm->nested.intercept            = nested_vmcb->control.intercept;

3585
	svm_flush_tlb(&svm->vcpu, true);
A
Alexander Graf 已提交
3586 3587 3588 3589 3590 3591
	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;

3592 3593
	if (svm->vcpu.arch.hflags & HF_VINTR_MASK) {
		/* We only want the cr8 intercept bits of the guest */
3594 3595
		clr_cr_intercept(svm, INTERCEPT_CR8_READ);
		clr_cr_intercept(svm, INTERCEPT_CR8_WRITE);
3596 3597
	}

3598
	/* We don't want to see VMMCALLs from a nested guest */
3599
	clr_intercept(svm, INTERCEPT_VMMCALL);
3600

3601 3602 3603
	svm->vcpu.arch.tsc_offset += nested_vmcb->control.tsc_offset;
	svm->vmcb->control.tsc_offset = svm->vcpu.arch.tsc_offset;

3604
	svm->vmcb->control.virt_ext = nested_vmcb->control.virt_ext;
A
Alexander Graf 已提交
3605 3606 3607 3608 3609
	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;

3610 3611 3612 3613 3614
	svm->vmcb->control.pause_filter_count =
		nested_vmcb->control.pause_filter_count;
	svm->vmcb->control.pause_filter_thresh =
		nested_vmcb->control.pause_filter_thresh;

3615
	kvm_vcpu_unmap(&svm->vcpu, map, true);
3616

3617 3618 3619
	/* Enter Guest-Mode */
	enter_guest_mode(&svm->vcpu);

3620 3621 3622 3623 3624 3625
	/*
	 * Merge guest and host intercepts - must be called  with vcpu in
	 * guest-mode to take affect here
	 */
	recalc_intercepts(svm);

3626
	svm->nested.vmcb = vmcb_gpa;
3627

3628
	enable_gif(svm);
A
Alexander Graf 已提交
3629

3630
	mark_all_dirty(svm->vmcb);
L
Ladi Prosek 已提交
3631 3632
}

3633
static int nested_svm_vmrun(struct vcpu_svm *svm)
L
Ladi Prosek 已提交
3634
{
3635
	int ret;
L
Ladi Prosek 已提交
3636 3637 3638
	struct vmcb *nested_vmcb;
	struct vmcb *hsave = svm->nested.hsave;
	struct vmcb *vmcb = svm->vmcb;
3639
	struct kvm_host_map map;
L
Ladi Prosek 已提交
3640 3641 3642 3643
	u64 vmcb_gpa;

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

3644
	ret = kvm_vcpu_map(&svm->vcpu, gpa_to_gfn(vmcb_gpa), &map);
3645
	if (ret == -EINVAL) {
3646
		kvm_inject_gp(&svm->vcpu, 0);
3647
		return 1;
3648 3649
	} else if (ret) {
		return kvm_skip_emulated_instruction(&svm->vcpu);
3650 3651
	}

3652 3653
	ret = kvm_skip_emulated_instruction(&svm->vcpu);

3654
	nested_vmcb = map.hva;
L
Ladi Prosek 已提交
3655 3656 3657 3658 3659 3660 3661

	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;

3662
		kvm_vcpu_unmap(&svm->vcpu, &map, true);
L
Ladi Prosek 已提交
3663

3664
		return ret;
L
Ladi Prosek 已提交
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 3697 3698 3699 3700 3701 3702 3703 3704 3705
	}

	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);

3706
	enter_svm_guest_mode(svm, vmcb_gpa, nested_vmcb, &map);
3707

3708 3709 3710 3711 3712 3713 3714 3715 3716
	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);
	}

3717
	return ret;
A
Alexander Graf 已提交
3718 3719
}

3720
static void nested_svm_vmloadsave(struct vmcb *from_vmcb, struct vmcb *to_vmcb)
3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735
{
	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
3736
static int vmload_interception(struct vcpu_svm *svm)
3737
{
3738
	struct vmcb *nested_vmcb;
3739
	struct kvm_host_map map;
3740
	int ret;
3741

3742 3743 3744
	if (nested_svm_check_permissions(svm))
		return 1;

3745 3746 3747 3748
	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);
3749
		return 1;
3750 3751 3752
	}

	nested_vmcb = map.hva;
3753

3754
	ret = kvm_skip_emulated_instruction(&svm->vcpu);
3755

3756
	nested_svm_vmloadsave(nested_vmcb, svm->vmcb);
3757
	kvm_vcpu_unmap(&svm->vcpu, &map, true);
3758

3759
	return ret;
3760 3761
}

A
Avi Kivity 已提交
3762
static int vmsave_interception(struct vcpu_svm *svm)
3763
{
3764
	struct vmcb *nested_vmcb;
3765
	struct kvm_host_map map;
3766
	int ret;
3767

3768 3769 3770
	if (nested_svm_check_permissions(svm))
		return 1;

3771 3772 3773 3774
	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);
3775
		return 1;
3776 3777 3778
	}

	nested_vmcb = map.hva;
3779

3780
	ret = kvm_skip_emulated_instruction(&svm->vcpu);
3781

3782
	nested_svm_vmloadsave(svm->vmcb, nested_vmcb);
3783
	kvm_vcpu_unmap(&svm->vcpu, &map, true);
3784

3785
	return ret;
3786 3787
}

A
Avi Kivity 已提交
3788
static int vmrun_interception(struct vcpu_svm *svm)
A
Alexander Graf 已提交
3789 3790 3791 3792
{
	if (nested_svm_check_permissions(svm))
		return 1;

3793
	return nested_svm_vmrun(svm);
A
Alexander Graf 已提交
3794 3795
}

A
Avi Kivity 已提交
3796
static int stgi_interception(struct vcpu_svm *svm)
3797
{
3798 3799
	int ret;

3800 3801 3802
	if (nested_svm_check_permissions(svm))
		return 1;

3803 3804
	/*
	 * If VGIF is enabled, the STGI intercept is only added to
3805
	 * detect the opening of the SMI/NMI window; remove it now.
3806 3807 3808 3809
	 */
	if (vgif_enabled(svm))
		clr_intercept(svm, INTERCEPT_STGI);

3810
	ret = kvm_skip_emulated_instruction(&svm->vcpu);
3811
	kvm_make_request(KVM_REQ_EVENT, &svm->vcpu);
3812

3813
	enable_gif(svm);
3814

3815
	return ret;
3816 3817
}

A
Avi Kivity 已提交
3818
static int clgi_interception(struct vcpu_svm *svm)
3819
{
3820 3821
	int ret;

3822 3823 3824
	if (nested_svm_check_permissions(svm))
		return 1;

3825
	ret = kvm_skip_emulated_instruction(&svm->vcpu);
3826

3827
	disable_gif(svm);
3828 3829

	/* After a CLGI no interrupts should come */
3830 3831 3832 3833 3834
	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);
	}
3835

3836
	return ret;
3837 3838
}

A
Avi Kivity 已提交
3839
static int invlpga_interception(struct vcpu_svm *svm)
A
Alexander Graf 已提交
3840 3841 3842
{
	struct kvm_vcpu *vcpu = &svm->vcpu;

3843 3844
	trace_kvm_invlpga(svm->vmcb->save.rip, kvm_rcx_read(&svm->vcpu),
			  kvm_rax_read(&svm->vcpu));
3845

A
Alexander Graf 已提交
3846
	/* Let's treat INVLPGA the same as INVLPG (can be optimized!) */
3847
	kvm_mmu_invlpg(vcpu, kvm_rax_read(&svm->vcpu));
A
Alexander Graf 已提交
3848

3849
	return kvm_skip_emulated_instruction(&svm->vcpu);
A
Alexander Graf 已提交
3850 3851
}

3852 3853
static int skinit_interception(struct vcpu_svm *svm)
{
3854
	trace_kvm_skinit(svm->vmcb->save.rip, kvm_rax_read(&svm->vcpu));
3855 3856 3857 3858 3859

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

D
David Kaplan 已提交
3860 3861
static int wbinvd_interception(struct vcpu_svm *svm)
{
3862
	return kvm_emulate_wbinvd(&svm->vcpu);
D
David Kaplan 已提交
3863 3864
}

J
Joerg Roedel 已提交
3865 3866 3867
static int xsetbv_interception(struct vcpu_svm *svm)
{
	u64 new_bv = kvm_read_edx_eax(&svm->vcpu);
3868
	u32 index = kvm_rcx_read(&svm->vcpu);
J
Joerg Roedel 已提交
3869 3870

	if (kvm_set_xcr(&svm->vcpu, index, new_bv) == 0) {
3871
		return kvm_skip_emulated_instruction(&svm->vcpu);
J
Joerg Roedel 已提交
3872 3873 3874 3875 3876
	}

	return 1;
}

J
Jim Mattson 已提交
3877 3878 3879 3880 3881 3882
static int rdpru_interception(struct vcpu_svm *svm)
{
	kvm_queue_exception(&svm->vcpu, UD_VECTOR);
	return 1;
}

A
Avi Kivity 已提交
3883
static int task_switch_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
3884
{
3885
	u16 tss_selector;
3886 3887 3888
	int reason;
	int int_type = svm->vmcb->control.exit_int_info &
		SVM_EXITINTINFO_TYPE_MASK;
3889
	int int_vec = svm->vmcb->control.exit_int_info & SVM_EVTINJ_VEC_MASK;
3890 3891 3892 3893
	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;
3894 3895
	bool has_error_code = false;
	u32 error_code = 0;
3896 3897

	tss_selector = (u16)svm->vmcb->control.exit_info_1;
3898

3899 3900
	if (svm->vmcb->control.exit_info_2 &
	    (1ULL << SVM_EXITINFOSHIFT_TS_REASON_IRET))
3901 3902 3903 3904
		reason = TASK_SWITCH_IRET;
	else if (svm->vmcb->control.exit_info_2 &
		 (1ULL << SVM_EXITINFOSHIFT_TS_REASON_JMP))
		reason = TASK_SWITCH_JMP;
3905
	else if (idt_v)
3906 3907 3908 3909
		reason = TASK_SWITCH_GATE;
	else
		reason = TASK_SWITCH_CALL;

3910 3911 3912 3913 3914 3915
	if (reason == TASK_SWITCH_GATE) {
		switch (type) {
		case SVM_EXITINTINFO_TYPE_NMI:
			svm->vcpu.arch.nmi_injected = false;
			break;
		case SVM_EXITINTINFO_TYPE_EXEPT:
3916 3917 3918 3919 3920 3921
			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;
			}
3922 3923 3924 3925 3926 3927 3928 3929 3930
			kvm_clear_exception_queue(&svm->vcpu);
			break;
		case SVM_EXITINTINFO_TYPE_INTR:
			kvm_clear_interrupt_queue(&svm->vcpu);
			break;
		default:
			break;
		}
	}
3931

3932 3933 3934
	if (reason != TASK_SWITCH_GATE ||
	    int_type == SVM_EXITINTINFO_TYPE_SOFT ||
	    (int_type == SVM_EXITINTINFO_TYPE_EXEPT &&
3935
	     (int_vec == OF_VECTOR || int_vec == BP_VECTOR))) {
3936
		if (!skip_emulated_instruction(&svm->vcpu))
3937
			return 0;
3938
	}
3939

3940 3941 3942
	if (int_type != SVM_EXITINTINFO_TYPE_SOFT)
		int_vec = -1;

3943
	return kvm_task_switch(&svm->vcpu, tss_selector, int_vec, reason,
3944
			       has_error_code, error_code);
A
Avi Kivity 已提交
3945 3946
}

A
Avi Kivity 已提交
3947
static int cpuid_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
3948
{
3949
	return kvm_emulate_cpuid(&svm->vcpu);
A
Avi Kivity 已提交
3950 3951
}

A
Avi Kivity 已提交
3952
static int iret_interception(struct vcpu_svm *svm)
3953 3954
{
	++svm->vcpu.stat.nmi_window_exits;
3955
	clr_intercept(svm, INTERCEPT_IRET);
3956
	svm->vcpu.arch.hflags |= HF_IRET_MASK;
3957
	svm->nmi_iret_rip = kvm_rip_read(&svm->vcpu);
3958
	kvm_make_request(KVM_REQ_EVENT, &svm->vcpu);
3959 3960 3961
	return 1;
}

A
Avi Kivity 已提交
3962
static int invlpg_interception(struct vcpu_svm *svm)
M
Marcelo Tosatti 已提交
3963
{
3964
	if (!static_cpu_has(X86_FEATURE_DECODEASSISTS))
3965
		return kvm_emulate_instruction(&svm->vcpu, 0);
3966 3967

	kvm_mmu_invlpg(&svm->vcpu, svm->vmcb->control.exit_info_1);
3968
	return kvm_skip_emulated_instruction(&svm->vcpu);
M
Marcelo Tosatti 已提交
3969 3970
}

A
Avi Kivity 已提交
3971
static int emulate_on_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
3972
{
3973
	return kvm_emulate_instruction(&svm->vcpu, 0);
A
Avi Kivity 已提交
3974 3975
}

B
Brijesh Singh 已提交
3976 3977
static int rsm_interception(struct vcpu_svm *svm)
{
3978
	return kvm_emulate_instruction_from_buffer(&svm->vcpu, rsm_ins_bytes, 2);
B
Brijesh Singh 已提交
3979 3980
}

A
Avi Kivity 已提交
3981 3982 3983 3984
static int rdpmc_interception(struct vcpu_svm *svm)
{
	int err;

3985
	if (!nrips)
A
Avi Kivity 已提交
3986 3987 3988
		return emulate_on_interception(svm);

	err = kvm_rdpmc(&svm->vcpu);
3989
	return kvm_complete_insn_gp(&svm->vcpu, err);
A
Avi Kivity 已提交
3990 3991
}

3992 3993
static bool check_selective_cr0_intercepted(struct vcpu_svm *svm,
					    unsigned long val)
3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015
{
	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;
}

4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030
#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;
4031 4032 4033 4034
	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;
4035 4036 4037 4038 4039 4040 4041

	err = 0;
	if (cr >= 16) { /* mov to cr */
		cr -= 16;
		val = kvm_register_read(&svm->vcpu, reg);
		switch (cr) {
		case 0:
4042 4043
			if (!check_selective_cr0_intercepted(svm, val))
				err = kvm_set_cr0(&svm->vcpu, val);
4044 4045 4046
			else
				return 1;

4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070
			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:
4071
			val = kvm_read_cr3(&svm->vcpu);
4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085
			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);
	}
4086
	return kvm_complete_insn_gp(&svm->vcpu, err);
4087 4088
}

4089 4090 4091 4092 4093
static int dr_interception(struct vcpu_svm *svm)
{
	int reg, dr;
	unsigned long val;

4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104
	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;
	}

4105 4106 4107 4108 4109 4110 4111
	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 */
4112 4113
		if (!kvm_require_dr(&svm->vcpu, dr - 16))
			return 1;
4114 4115 4116
		val = kvm_register_read(&svm->vcpu, reg);
		kvm_set_dr(&svm->vcpu, dr - 16, val);
	} else {
4117 4118 4119 4120
		if (!kvm_require_dr(&svm->vcpu, dr))
			return 1;
		kvm_get_dr(&svm->vcpu, dr, &val);
		kvm_register_write(&svm->vcpu, reg, val);
4121 4122
	}

4123
	return kvm_skip_emulated_instruction(&svm->vcpu);
4124 4125
}

A
Avi Kivity 已提交
4126
static int cr8_write_interception(struct vcpu_svm *svm)
4127
{
A
Avi Kivity 已提交
4128
	struct kvm_run *kvm_run = svm->vcpu.run;
A
Andre Przywara 已提交
4129
	int r;
A
Avi Kivity 已提交
4130

4131 4132
	u8 cr8_prev = kvm_get_cr8(&svm->vcpu);
	/* instruction emulation calls kvm_set_cr8() */
4133
	r = cr_interception(svm);
4134
	if (lapic_in_kernel(&svm->vcpu))
4135
		return r;
4136
	if (cr8_prev <= kvm_get_cr8(&svm->vcpu))
4137
		return r;
4138 4139 4140 4141
	kvm_run->exit_reason = KVM_EXIT_SET_TPR;
	return 0;
}

4142 4143
static int svm_get_msr_feature(struct kvm_msr_entry *msr)
{
4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155
	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;
4156 4157
}

4158
static int svm_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
A
Avi Kivity 已提交
4159
{
4160 4161
	struct vcpu_svm *svm = to_svm(vcpu);

4162
	switch (msr_info->index) {
B
Brian Gerst 已提交
4163
	case MSR_STAR:
4164
		msr_info->data = svm->vmcb->save.star;
A
Avi Kivity 已提交
4165
		break;
4166
#ifdef CONFIG_X86_64
A
Avi Kivity 已提交
4167
	case MSR_LSTAR:
4168
		msr_info->data = svm->vmcb->save.lstar;
A
Avi Kivity 已提交
4169 4170
		break;
	case MSR_CSTAR:
4171
		msr_info->data = svm->vmcb->save.cstar;
A
Avi Kivity 已提交
4172 4173
		break;
	case MSR_KERNEL_GS_BASE:
4174
		msr_info->data = svm->vmcb->save.kernel_gs_base;
A
Avi Kivity 已提交
4175 4176
		break;
	case MSR_SYSCALL_MASK:
4177
		msr_info->data = svm->vmcb->save.sfmask;
A
Avi Kivity 已提交
4178 4179 4180
		break;
#endif
	case MSR_IA32_SYSENTER_CS:
4181
		msr_info->data = svm->vmcb->save.sysenter_cs;
A
Avi Kivity 已提交
4182 4183
		break;
	case MSR_IA32_SYSENTER_EIP:
4184
		msr_info->data = svm->sysenter_eip;
A
Avi Kivity 已提交
4185 4186
		break;
	case MSR_IA32_SYSENTER_ESP:
4187
		msr_info->data = svm->sysenter_esp;
A
Avi Kivity 已提交
4188
		break;
P
Paolo Bonzini 已提交
4189 4190 4191 4192 4193
	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
4194 4195 4196 4197 4198
	/*
	 * 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.
	 */
4199
	case MSR_IA32_DEBUGCTLMSR:
4200
		msr_info->data = svm->vmcb->save.dbgctl;
4201 4202
		break;
	case MSR_IA32_LASTBRANCHFROMIP:
4203
		msr_info->data = svm->vmcb->save.br_from;
4204 4205
		break;
	case MSR_IA32_LASTBRANCHTOIP:
4206
		msr_info->data = svm->vmcb->save.br_to;
4207 4208
		break;
	case MSR_IA32_LASTINTFROMIP:
4209
		msr_info->data = svm->vmcb->save.last_excp_from;
4210 4211
		break;
	case MSR_IA32_LASTINTTOIP:
4212
		msr_info->data = svm->vmcb->save.last_excp_to;
4213
		break;
A
Alexander Graf 已提交
4214
	case MSR_VM_HSAVE_PA:
4215
		msr_info->data = svm->nested.hsave_msr;
A
Alexander Graf 已提交
4216
		break;
4217
	case MSR_VM_CR:
4218
		msr_info->data = svm->nested.vm_cr_msr;
4219
		break;
4220 4221
	case MSR_IA32_SPEC_CTRL:
		if (!msr_info->host_initiated &&
4222 4223
		    !guest_cpuid_has(vcpu, X86_FEATURE_AMD_IBRS) &&
		    !guest_cpuid_has(vcpu, X86_FEATURE_AMD_SSBD))
4224 4225 4226 4227
			return 1;

		msr_info->data = svm->spec_ctrl;
		break;
4228 4229 4230 4231 4232 4233 4234
	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;
4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251
	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;
4252 4253 4254
	case MSR_F10H_DECFG:
		msr_info->data = svm->msr_decfg;
		break;
A
Avi Kivity 已提交
4255
	default:
4256
		return kvm_get_msr_common(vcpu, msr_info);
A
Avi Kivity 已提交
4257 4258 4259 4260
	}
	return 0;
}

A
Avi Kivity 已提交
4261
static int rdmsr_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
4262
{
4263
	return kvm_emulate_rdmsr(&svm->vcpu);
A
Avi Kivity 已提交
4264 4265
}

4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290
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;
}

4291
static int svm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr)
A
Avi Kivity 已提交
4292
{
4293 4294
	struct vcpu_svm *svm = to_svm(vcpu);

4295 4296
	u32 ecx = msr->index;
	u64 data = msr->data;
A
Avi Kivity 已提交
4297
	switch (ecx) {
P
Paolo Bonzini 已提交
4298 4299 4300 4301 4302 4303 4304
	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;
4305 4306
	case MSR_IA32_SPEC_CTRL:
		if (!msr->host_initiated &&
4307 4308
		    !guest_cpuid_has(vcpu, X86_FEATURE_AMD_IBRS) &&
		    !guest_cpuid_has(vcpu, X86_FEATURE_AMD_SSBD))
4309 4310
			return 1;

4311
		if (data & ~kvm_spec_ctrl_valid_bits(vcpu))
4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330
			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
4331 4332
	case MSR_IA32_PRED_CMD:
		if (!msr->host_initiated &&
4333
		    !guest_cpuid_has(vcpu, X86_FEATURE_AMD_IBPB))
A
Ashok Raj 已提交
4334 4335 4336 4337
			return 1;

		if (data & ~PRED_CMD_IBPB)
			return 1;
4338 4339
		if (!boot_cpu_has(X86_FEATURE_AMD_IBPB))
			return 1;
A
Ashok Raj 已提交
4340 4341 4342 4343 4344 4345
		if (!data)
			break;

		wrmsrl(MSR_IA32_PRED_CMD, PRED_CMD_IBPB);
		set_msr_interception(svm->msrpm, MSR_IA32_PRED_CMD, 0, 1);
		break;
4346 4347 4348 4349 4350 4351 4352 4353 4354 4355
	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
4356
	case MSR_STAR:
4357
		svm->vmcb->save.star = data;
A
Avi Kivity 已提交
4358
		break;
4359
#ifdef CONFIG_X86_64
A
Avi Kivity 已提交
4360
	case MSR_LSTAR:
4361
		svm->vmcb->save.lstar = data;
A
Avi Kivity 已提交
4362 4363
		break;
	case MSR_CSTAR:
4364
		svm->vmcb->save.cstar = data;
A
Avi Kivity 已提交
4365 4366
		break;
	case MSR_KERNEL_GS_BASE:
4367
		svm->vmcb->save.kernel_gs_base = data;
A
Avi Kivity 已提交
4368 4369
		break;
	case MSR_SYSCALL_MASK:
4370
		svm->vmcb->save.sfmask = data;
A
Avi Kivity 已提交
4371 4372 4373
		break;
#endif
	case MSR_IA32_SYSENTER_CS:
4374
		svm->vmcb->save.sysenter_cs = data;
A
Avi Kivity 已提交
4375 4376
		break;
	case MSR_IA32_SYSENTER_EIP:
4377
		svm->sysenter_eip = data;
4378
		svm->vmcb->save.sysenter_eip = data;
A
Avi Kivity 已提交
4379 4380
		break;
	case MSR_IA32_SYSENTER_ESP:
4381
		svm->sysenter_esp = data;
4382
		svm->vmcb->save.sysenter_esp = data;
A
Avi Kivity 已提交
4383
		break;
P
Paolo Bonzini 已提交
4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395
	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;
4396
	case MSR_IA32_DEBUGCTLMSR:
4397
		if (!boot_cpu_has(X86_FEATURE_LBRV)) {
4398 4399
			vcpu_unimpl(vcpu, "%s: MSR_IA32_DEBUGCTL 0x%llx, nop\n",
				    __func__, data);
4400 4401 4402 4403 4404 4405
			break;
		}
		if (data & DEBUGCTL_RESERVED_BITS)
			return 1;

		svm->vmcb->save.dbgctl = data;
4406
		mark_dirty(svm->vmcb, VMCB_LBR);
4407 4408 4409 4410
		if (data & (1ULL<<0))
			svm_enable_lbrv(svm);
		else
			svm_disable_lbrv(svm);
4411
		break;
A
Alexander Graf 已提交
4412
	case MSR_VM_HSAVE_PA:
4413
		svm->nested.hsave_msr = data;
4414
		break;
4415
	case MSR_VM_CR:
4416
		return svm_set_vm_cr(vcpu, data);
4417
	case MSR_VM_IGNNE:
4418
		vcpu_unimpl(vcpu, "unimplemented wrmsr: 0x%x data 0x%llx\n", ecx, data);
4419
		break;
4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437
	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;
	}
4438 4439 4440
	case MSR_IA32_APICBASE:
		if (kvm_vcpu_apicv_active(vcpu))
			avic_update_vapic_bar(to_svm(vcpu), data);
4441
		/* Fall through */
A
Avi Kivity 已提交
4442
	default:
4443
		return kvm_set_msr_common(vcpu, msr);
A
Avi Kivity 已提交
4444 4445 4446 4447
	}
	return 0;
}

A
Avi Kivity 已提交
4448
static int wrmsr_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
4449
{
4450
	return kvm_emulate_wrmsr(&svm->vcpu);
A
Avi Kivity 已提交
4451 4452
}

A
Avi Kivity 已提交
4453
static int msr_interception(struct vcpu_svm *svm)
A
Avi Kivity 已提交
4454
{
R
Rusty Russell 已提交
4455
	if (svm->vmcb->control.exit_info_1)
A
Avi Kivity 已提交
4456
		return wrmsr_interception(svm);
A
Avi Kivity 已提交
4457
	else
A
Avi Kivity 已提交
4458
		return rdmsr_interception(svm);
A
Avi Kivity 已提交
4459 4460
}

A
Avi Kivity 已提交
4461
static int interrupt_window_interception(struct vcpu_svm *svm)
4462
{
4463
	kvm_make_request(KVM_REQ_EVENT, &svm->vcpu);
4464
	svm_clear_vintr(svm);
4465 4466 4467 4468 4469 4470 4471 4472

	/*
	 * 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);

4473
	svm->vmcb->control.int_ctl &= ~V_IRQ_MASK;
4474
	mark_dirty(svm->vmcb, VMCB_INTR);
4475
	++svm->vcpu.stat.irq_window_exits;
4476 4477 4478
	return 1;
}

4479 4480
static int pause_interception(struct vcpu_svm *svm)
{
4481 4482 4483
	struct kvm_vcpu *vcpu = &svm->vcpu;
	bool in_kernel = (svm_get_cpl(vcpu) == 0);

4484 4485 4486
	if (pause_filter_thresh)
		grow_ple_window(vcpu);

4487
	kvm_vcpu_on_spin(vcpu, in_kernel);
4488 4489 4490
	return 1;
}

4491 4492
static int nop_interception(struct vcpu_svm *svm)
{
4493
	return kvm_skip_emulated_instruction(&(svm->vcpu));
4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507
}

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);
}

4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519
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
4520
	u32 index = svm->vmcb->control.exit_info_2 & 0xFF;
4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541
	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: {
4542 4543 4544
		int i;
		struct kvm_vcpu *vcpu;
		struct kvm *kvm = svm->vcpu.kvm;
4545 4546 4547
		struct kvm_lapic *apic = svm->vcpu.arch.apic;

		/*
4548 4549 4550
		 * 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.
4551
		 */
4552 4553
		kvm_for_each_vcpu(i, vcpu, kvm) {
			bool m = kvm_apic_match_dest(vcpu, apic,
4554
						     icrl & APIC_SHORT_MASK,
4555
						     GET_APIC_DEST_FIELD(icrh),
4556
						     icrl & APIC_DEST_MASK);
4557 4558 4559 4560

			if (m && !avic_vcpu_is_running(vcpu))
				kvm_vcpu_wake_up(vcpu);
		}
4561 4562 4563
		break;
	}
	case AVIC_IPI_FAILURE_INVALID_TARGET:
4564 4565
		WARN_ONCE(1, "Invalid IPI target: index=%u, vcpu=%d, icr=%#0x:%#0x\n",
			  index, svm->vcpu.vcpu_id, icrh, icrl);
4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578
		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)
{
4579
	struct kvm_svm *kvm_svm = to_kvm_svm(vcpu->kvm);
4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600
	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;
	}

4601
	logical_apic_id_table = (u32 *) page_address(kvm_svm->avic_logical_id_table_page);
4602 4603 4604 4605

	return &logical_apic_id_table[index];
}

4606
static int avic_ldr_write(struct kvm_vcpu *vcpu, u8 g_physical_id, u32 ldr)
4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618
{
	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);
4619
	new_entry |= AVIC_LOGICAL_ID_ENTRY_VALID_MASK;
4620 4621 4622 4623 4624
	WRITE_ONCE(*entry, new_entry);

	return 0;
}

4625 4626 4627 4628 4629 4630 4631
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)
4632
		clear_bit(AVIC_LOGICAL_ID_ENTRY_VALID_BIT, (unsigned long *)entry);
4633 4634
}

4635 4636
static int avic_handle_ldr_update(struct kvm_vcpu *vcpu)
{
4637
	int ret = 0;
4638 4639
	struct vcpu_svm *svm = to_svm(vcpu);
	u32 ldr = kvm_lapic_get_reg(vcpu->arch.apic, APIC_LDR);
4640
	u32 id = kvm_xapic_id(vcpu->arch.apic);
4641

4642 4643
	if (ldr == svm->ldr_reg)
		return 0;
4644

4645 4646 4647
	avic_invalidate_logical_id_entry(vcpu);

	if (ldr)
4648
		ret = avic_ldr_write(vcpu, id, ldr);
4649 4650

	if (!ret)
4651
		svm->ldr_reg = ldr;
4652

4653 4654 4655 4656 4657 4658 4659
	return ret;
}

static int avic_handle_apic_id_update(struct kvm_vcpu *vcpu)
{
	u64 *old, *new;
	struct vcpu_svm *svm = to_svm(vcpu);
4660
	u32 id = kvm_xapic_id(vcpu->arch.apic);
4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684

	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;
}

4685
static void avic_handle_dfr_update(struct kvm_vcpu *vcpu)
4686 4687 4688 4689
{
	struct vcpu_svm *svm = to_svm(vcpu);
	u32 dfr = kvm_lapic_get_reg(vcpu->arch.apic, APIC_DFR);

4690 4691
	if (svm->dfr_reg == dfr)
		return;
4692

4693 4694
	avic_invalidate_logical_id_entry(vcpu);
	svm->dfr_reg = dfr;
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 4767 4768 4769 4770 4771
}

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 */
4772
		ret = kvm_emulate_instruction(&svm->vcpu, 0);
4773 4774 4775 4776 4777
	}

	return ret;
}

4778
static int (*const svm_exit_handlers[])(struct vcpu_svm *svm) = {
4779 4780 4781 4782
	[SVM_EXIT_READ_CR0]			= cr_interception,
	[SVM_EXIT_READ_CR3]			= cr_interception,
	[SVM_EXIT_READ_CR4]			= cr_interception,
	[SVM_EXIT_READ_CR8]			= cr_interception,
4783
	[SVM_EXIT_CR0_SEL_WRITE]		= cr_interception,
4784
	[SVM_EXIT_WRITE_CR0]			= cr_interception,
4785 4786
	[SVM_EXIT_WRITE_CR3]			= cr_interception,
	[SVM_EXIT_WRITE_CR4]			= cr_interception,
J
Joerg Roedel 已提交
4787
	[SVM_EXIT_WRITE_CR8]			= cr8_write_interception,
4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803
	[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
4804 4805
	[SVM_EXIT_EXCP_BASE + DB_VECTOR]	= db_interception,
	[SVM_EXIT_EXCP_BASE + BP_VECTOR]	= bp_interception,
4806
	[SVM_EXIT_EXCP_BASE + UD_VECTOR]	= ud_interception,
J
Joerg Roedel 已提交
4807 4808
	[SVM_EXIT_EXCP_BASE + PF_VECTOR]	= pf_interception,
	[SVM_EXIT_EXCP_BASE + MC_VECTOR]	= mc_interception,
4809
	[SVM_EXIT_EXCP_BASE + AC_VECTOR]	= ac_interception,
4810
	[SVM_EXIT_EXCP_BASE + GP_VECTOR]	= gp_interception,
J
Joerg Roedel 已提交
4811
	[SVM_EXIT_INTR]				= intr_interception,
4812
	[SVM_EXIT_NMI]				= nmi_interception,
A
Avi Kivity 已提交
4813 4814
	[SVM_EXIT_SMI]				= nop_on_interception,
	[SVM_EXIT_INIT]				= nop_on_interception,
4815
	[SVM_EXIT_VINTR]			= interrupt_window_interception,
A
Avi Kivity 已提交
4816
	[SVM_EXIT_RDPMC]			= rdpmc_interception,
A
Avi Kivity 已提交
4817
	[SVM_EXIT_CPUID]			= cpuid_interception,
4818
	[SVM_EXIT_IRET]                         = iret_interception,
4819
	[SVM_EXIT_INVD]                         = emulate_on_interception,
4820
	[SVM_EXIT_PAUSE]			= pause_interception,
A
Avi Kivity 已提交
4821
	[SVM_EXIT_HLT]				= halt_interception,
M
Marcelo Tosatti 已提交
4822
	[SVM_EXIT_INVLPG]			= invlpg_interception,
A
Alexander Graf 已提交
4823
	[SVM_EXIT_INVLPGA]			= invlpga_interception,
J
Joerg Roedel 已提交
4824
	[SVM_EXIT_IOIO]				= io_interception,
A
Avi Kivity 已提交
4825 4826
	[SVM_EXIT_MSR]				= msr_interception,
	[SVM_EXIT_TASK_SWITCH]			= task_switch_interception,
4827
	[SVM_EXIT_SHUTDOWN]			= shutdown_interception,
A
Alexander Graf 已提交
4828
	[SVM_EXIT_VMRUN]			= vmrun_interception,
4829
	[SVM_EXIT_VMMCALL]			= vmmcall_interception,
4830 4831
	[SVM_EXIT_VMLOAD]			= vmload_interception,
	[SVM_EXIT_VMSAVE]			= vmsave_interception,
4832 4833
	[SVM_EXIT_STGI]				= stgi_interception,
	[SVM_EXIT_CLGI]				= clgi_interception,
4834
	[SVM_EXIT_SKINIT]			= skinit_interception,
D
David Kaplan 已提交
4835
	[SVM_EXIT_WBINVD]                       = wbinvd_interception,
4836 4837
	[SVM_EXIT_MONITOR]			= monitor_interception,
	[SVM_EXIT_MWAIT]			= mwait_interception,
J
Joerg Roedel 已提交
4838
	[SVM_EXIT_XSETBV]			= xsetbv_interception,
J
Jim Mattson 已提交
4839
	[SVM_EXIT_RDPRU]			= rdpru_interception,
4840
	[SVM_EXIT_NPF]				= npf_interception,
B
Brijesh Singh 已提交
4841
	[SVM_EXIT_RSM]                          = rsm_interception,
4842 4843
	[SVM_EXIT_AVIC_INCOMPLETE_IPI]		= avic_incomplete_ipi_interception,
	[SVM_EXIT_AVIC_UNACCELERATED_ACCESS]	= avic_unaccelerated_access_interception,
A
Avi Kivity 已提交
4844 4845
};

4846
static void dump_vmcb(struct kvm_vcpu *vcpu)
4847 4848 4849 4850 4851
{
	struct vcpu_svm *svm = to_svm(vcpu);
	struct vmcb_control_area *control = &svm->vmcb->control;
	struct vmcb_save_area *save = &svm->vmcb->save;

4852 4853 4854 4855 4856
	if (!dump_invalid_vmcb) {
		pr_warn_ratelimited("set kvm_amd.dump_invalid_vmcb=1 to dump internal KVM state.\n");
		return;
	}

4857
	pr_err("VMCB Control Area:\n");
4858 4859 4860 4861 4862 4863 4864
	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);
4865 4866
	pr_err("%-20s%d\n", "pause filter threshold:",
	       control->pause_filter_thresh);
4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881
	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);
4882
	pr_err("%-20s%016llx\n", "avic_vapic_bar:", control->avic_vapic_bar);
4883 4884
	pr_err("%-20s%08x\n", "event_inj:", control->event_inj);
	pr_err("%-20s%08x\n", "event_inj_err:", control->event_inj_err);
4885
	pr_err("%-20s%lld\n", "virt_ext:", control->virt_ext);
4886
	pr_err("%-20s%016llx\n", "next_rip:", control->next_rip);
4887 4888 4889
	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);
4890
	pr_err("VMCB State Save Area:\n");
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 4926 4927 4928 4929 4930
	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);
4931 4932
	pr_err("cpl:            %d                efer:         %016llx\n",
		save->cpl, save->efer);
4933 4934 4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950 4951 4952 4953 4954 4955 4956 4957 4958 4959
	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);
4960 4961
}

4962 4963 4964 4965 4966 4967 4968 4969
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;
}

4970 4971
static int handle_exit(struct kvm_vcpu *vcpu,
	enum exit_fastpath_completion exit_fastpath)
A
Avi Kivity 已提交
4972
{
4973
	struct vcpu_svm *svm = to_svm(vcpu);
A
Avi Kivity 已提交
4974
	struct kvm_run *kvm_run = vcpu->run;
4975
	u32 exit_code = svm->vmcb->control.exit_code;
A
Avi Kivity 已提交
4976

4977 4978
	trace_kvm_exit(exit_code, vcpu, KVM_ISA_SVM);

4979
	if (!is_cr_intercept(svm, INTERCEPT_CR0_WRITE))
4980 4981 4982
		vcpu->arch.cr0 = svm->vmcb->save.cr0;
	if (npt_enabled)
		vcpu->arch.cr3 = svm->vmcb->save.cr3;
4983

4984 4985 4986 4987 4988 4989 4990
	if (unlikely(svm->nested.exit_required)) {
		nested_svm_vmexit(svm);
		svm->nested.exit_required = false;

		return 1;
	}

4991
	if (is_guest_mode(vcpu)) {
4992 4993
		int vmexit;

4994 4995 4996 4997
		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,
4998 4999
					svm->vmcb->control.exit_int_info_err,
					KVM_ISA_SVM);
5000

5001 5002 5003 5004 5005 5006
		vmexit = nested_svm_exit_special(svm);

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

		if (vmexit == NESTED_EXIT_DONE)
5007 5008 5009
			return 1;
	}

5010 5011
	svm_complete_interrupts(svm);

5012 5013 5014 5015
	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;
5016
		dump_vmcb(vcpu);
5017 5018 5019
		return 0;
	}

5020
	if (is_external_interrupt(svm->vmcb->control.exit_int_info) &&
5021
	    exit_code != SVM_EXIT_EXCP_BASE + PF_VECTOR &&
5022 5023
	    exit_code != SVM_EXIT_NPF && exit_code != SVM_EXIT_TASK_SWITCH &&
	    exit_code != SVM_EXIT_INTR && exit_code != SVM_EXIT_NMI)
5024
		printk(KERN_ERR "%s: unexpected exit_int_info 0x%x "
A
Avi Kivity 已提交
5025
		       "exit_code 0x%x\n",
5026
		       __func__, svm->vmcb->control.exit_int_info,
A
Avi Kivity 已提交
5027 5028
		       exit_code);

5029 5030 5031 5032
	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
5033
	    || !svm_exit_handlers[exit_code]) {
5034 5035 5036 5037 5038 5039 5040 5041
		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
5042 5043
	}

5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055
#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
5056
	return svm_exit_handlers[exit_code](svm);
A
Avi Kivity 已提交
5057 5058 5059 5060 5061 5062
}

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

5063 5064
	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
5065 5066 5067
	load_TR_desc();
}

5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091
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
5092
static void pre_svm_run(struct vcpu_svm *svm)
A
Avi Kivity 已提交
5093 5094 5095
{
	int cpu = raw_smp_processor_id();

5096
	struct svm_cpu_data *sd = per_cpu(svm_data, cpu);
A
Avi Kivity 已提交
5097

5098 5099 5100
	if (sev_guest(svm->vcpu.kvm))
		return pre_sev_run(svm, cpu);

5101
	/* FIXME: handle wraparound of asid_generation */
5102 5103
	if (svm->asid_generation != sd->asid_generation)
		new_asid(svm, sd);
A
Avi Kivity 已提交
5104 5105
}

5106 5107 5108 5109 5110 5111
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;
5112
	set_intercept(svm, INTERCEPT_IRET);
5113 5114
	++vcpu->stat.nmi_injections;
}
A
Avi Kivity 已提交
5115

5116
static inline void svm_inject_irq(struct vcpu_svm *svm, int irq)
A
Avi Kivity 已提交
5117 5118 5119
{
	struct vmcb_control_area *control;

5120
	/* The following fields are ignored when AVIC is enabled */
R
Rusty Russell 已提交
5121
	control = &svm->vmcb->control;
5122
	control->int_vector = irq;
A
Avi Kivity 已提交
5123 5124 5125
	control->int_ctl &= ~V_INTR_PRIO_MASK;
	control->int_ctl |= V_IRQ_MASK |
		((/*control->int_vector >> 4*/ 0xf) << V_INTR_PRIO_SHIFT);
5126
	mark_dirty(svm->vmcb, VMCB_INTR);
A
Avi Kivity 已提交
5127 5128
}

5129
static void svm_set_irq(struct kvm_vcpu *vcpu)
E
Eddie Dong 已提交
5130 5131 5132
{
	struct vcpu_svm *svm = to_svm(vcpu);

5133
	BUG_ON(!(gif_set(svm)));
5134

5135 5136 5137
	trace_kvm_inj_virq(vcpu->arch.interrupt.nr);
	++vcpu->stat.irq_injections;

5138 5139
	svm->vmcb->control.event_inj = vcpu->arch.interrupt.nr |
		SVM_EVTINJ_VALID | SVM_EVTINJ_TYPE_INTR;
E
Eddie Dong 已提交
5140 5141
}

5142 5143 5144 5145 5146
static inline bool svm_nested_virtualize_tpr(struct kvm_vcpu *vcpu)
{
	return is_guest_mode(vcpu) && (vcpu->arch.hflags & HF_VINTR_MASK);
}

5147
static void update_cr8_intercept(struct kvm_vcpu *vcpu, int tpr, int irr)
5148 5149 5150
{
	struct vcpu_svm *svm = to_svm(vcpu);

5151
	if (svm_nested_virtualize_tpr(vcpu))
5152 5153
		return;

5154 5155
	clr_cr_intercept(svm, INTERCEPT_CR8_WRITE);

5156
	if (irr == -1)
5157 5158
		return;

5159
	if (tpr >= irr)
5160
		set_cr_intercept(svm, INTERCEPT_CR8_WRITE);
5161
}
5162

5163
static void svm_set_virtual_apic_mode(struct kvm_vcpu *vcpu)
5164 5165 5166 5167
{
	return;
}

5168 5169
static void svm_hwapic_irr_update(struct kvm_vcpu *vcpu, int max_irr)
{
5170 5171
}

5172
static void svm_hwapic_isr_update(struct kvm_vcpu *vcpu, int max_isr)
5173
{
5174 5175
}

5176 5177 5178 5179 5180 5181 5182 5183 5184 5185 5186
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);
}

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 5214 5215 5216 5217 5218
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;
}

5219
static void svm_refresh_apicv_exec_ctrl(struct kvm_vcpu *vcpu)
5220
{
5221 5222
	struct vcpu_svm *svm = to_svm(vcpu);
	struct vmcb *vmcb = svm->vmcb;
5223
	bool activated = kvm_vcpu_apicv_active(vcpu);
5224

5225 5226 5227 5228 5229 5230 5231 5232 5233
	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);
5234
		vmcb->control.int_ctl |= AVIC_ENABLE_MASK;
5235
	} else {
5236
		vmcb->control.int_ctl &= ~AVIC_ENABLE_MASK;
5237
	}
5238
	mark_dirty(vmcb, VMCB_AVIC);
5239 5240

	svm_set_pi_irte_mode(vcpu, activated);
5241 5242
}

5243
static void svm_load_eoi_exitmap(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap)
5244 5245 5246 5247
{
	return;
}

5248 5249 5250 5251 5252
static void svm_deliver_avic_intr(struct kvm_vcpu *vcpu, int vec)
{
	kvm_lapic_set_irr(vec, vcpu->arch.apic);
	smp_mb__after_atomic();

5253 5254 5255 5256 5257 5258 5259
	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
5260 5261 5262
		kvm_vcpu_wake_up(vcpu);
}

5263 5264 5265 5266 5267
static bool svm_dy_apicv_has_pending_interrupt(struct kvm_vcpu *vcpu)
{
	return false;
}

5268 5269 5270 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
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.
	 */
5314
	ir = kzalloc(sizeof(struct amd_svm_iommu_ir), GFP_KERNEL_ACCOUNT);
5315 5316 5317 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
	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);

5348 5349
	if (!kvm_intr_is_single_vcpu(kvm, &irq, &vcpu) ||
	    !kvm_irq_is_postable(&irq)) {
5350 5351 5352 5353 5354 5355 5356 5357
		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);
5358
	vcpu_info->pi_desc_addr = __sme_set(page_to_phys((*svm)->avic_backing_page));
5359 5360 5361 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
	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.
5403
		 * 4. IRQ has incompatible delivery mode (SMI, INIT, etc)
5404 5405 5406 5407 5408 5409
		 */
		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 */
5410 5411
			pi.base = __sme_set(page_to_phys(svm->avic_backing_page) &
					    AVIC_HPA_MASK);
5412
			pi.ga_tag = AVIC_GATAG(to_kvm_svm(kvm)->avic_vm_id,
5413 5414 5415 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
						     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) {
5456 5457
			trace_kvm_pi_irte_update(host_irq, svm->vcpu.vcpu_id,
						 e->gsi, vcpu_info.vector,
5458 5459 5460 5461 5462 5463 5464 5465 5466 5467 5468 5469 5470 5471 5472
						 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;
}

5473 5474 5475 5476
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
5477 5478 5479 5480 5481 5482
	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;
5483 5484
}

J
Jan Kiszka 已提交
5485 5486 5487 5488 5489 5490 5491 5492 5493 5494 5495 5496 5497
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;
5498
		set_intercept(svm, INTERCEPT_IRET);
J
Jan Kiszka 已提交
5499 5500
	} else {
		svm->vcpu.arch.hflags &= ~HF_NMI_MASK;
5501
		clr_intercept(svm, INTERCEPT_IRET);
J
Jan Kiszka 已提交
5502 5503 5504
	}
}

5505 5506 5507 5508
static int svm_interrupt_allowed(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);
	struct vmcb *vmcb = svm->vmcb;
5509 5510 5511 5512 5513 5514
	int ret;

	if (!gif_set(svm) ||
	     (vmcb->control.int_state & SVM_INTERRUPT_SHADOW_MASK))
		return 0;

5515
	ret = !!(kvm_get_rflags(vcpu) & X86_EFLAGS_IF);
5516

5517
	if (is_guest_mode(vcpu))
5518 5519 5520
		return ret && !(svm->vcpu.arch.hflags & HF_VINTR_MASK);

	return ret;
5521 5522
}

5523
static void enable_irq_window(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
5524
{
5525 5526
	struct vcpu_svm *svm = to_svm(vcpu);

J
Joerg Roedel 已提交
5527 5528 5529 5530
	/*
	 * 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
5531 5532 5533
	 * 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
5534
	 */
5535
	if ((vgif_enabled(svm) || gif_set(svm)) && nested_svm_intr(svm)) {
5536 5537 5538 5539 5540 5541 5542
		/*
		 * 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);
5543 5544 5545
		svm_set_vintr(svm);
		svm_inject_irq(svm, 0x0);
	}
5546 5547
}

5548
static void enable_nmi_window(struct kvm_vcpu *vcpu)
5549
{
5550
	struct vcpu_svm *svm = to_svm(vcpu);
5551

5552 5553
	if ((svm->vcpu.arch.hflags & (HF_NMI_MASK | HF_IRET_MASK))
	    == HF_NMI_MASK)
5554
		return; /* IRET will cause a vm exit */
5555

5556 5557 5558
	if (!gif_set(svm)) {
		if (vgif_enabled(svm))
			set_intercept(svm, INTERCEPT_STGI);
5559
		return; /* STGI will cause a vm exit */
5560
	}
5561 5562 5563 5564

	if (svm->nested.exit_required)
		return; /* we're not going to run the guest yet */

J
Joerg Roedel 已提交
5565 5566 5567 5568
	/*
	 * Something prevents NMI from been injected. Single step over possible
	 * problem (IRET or exception injection or interrupt shadow)
	 */
5569
	svm->nmi_singlestep_guest_rflags = svm_get_rflags(vcpu);
J
Jan Kiszka 已提交
5570
	svm->nmi_singlestep = true;
5571
	svm->vmcb->save.rflags |= (X86_EFLAGS_TF | X86_EFLAGS_RF);
5572 5573
}

5574 5575 5576 5577 5578
static int svm_set_tss_addr(struct kvm *kvm, unsigned int addr)
{
	return 0;
}

5579 5580 5581 5582 5583
static int svm_set_identity_map_addr(struct kvm *kvm, u64 ident_addr)
{
	return 0;
}

5584
static void svm_flush_tlb(struct kvm_vcpu *vcpu, bool invalidate_gpa)
5585
{
5586 5587 5588 5589 5590 5591
	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--;
5592 5593
}

5594 5595 5596 5597 5598 5599 5600
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);
}

5601 5602 5603 5604
static void svm_prepare_guest_switch(struct kvm_vcpu *vcpu)
{
}

5605 5606 5607 5608
static inline void sync_cr8_to_lapic(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);

5609
	if (svm_nested_virtualize_tpr(vcpu))
5610 5611
		return;

5612
	if (!is_cr_intercept(svm, INTERCEPT_CR8_WRITE)) {
5613
		int cr8 = svm->vmcb->control.int_ctl & V_TPR_MASK;
5614
		kvm_set_cr8(vcpu, cr8);
5615 5616 5617
	}
}

5618 5619 5620 5621 5622
static inline void sync_lapic_to_cr8(struct kvm_vcpu *vcpu)
{
	struct vcpu_svm *svm = to_svm(vcpu);
	u64 cr8;

5623 5624
	if (svm_nested_virtualize_tpr(vcpu) ||
	    kvm_vcpu_apicv_active(vcpu))
5625 5626
		return;

5627 5628 5629 5630 5631
	cr8 = kvm_get_cr8(vcpu);
	svm->vmcb->control.int_ctl &= ~V_TPR_MASK;
	svm->vmcb->control.int_ctl |= cr8 & V_TPR_MASK;
}

5632 5633 5634 5635 5636
static void svm_complete_interrupts(struct vcpu_svm *svm)
{
	u8 vector;
	int type;
	u32 exitintinfo = svm->vmcb->control.exit_int_info;
5637 5638 5639
	unsigned int3_injected = svm->int3_injected;

	svm->int3_injected = 0;
5640

5641 5642 5643 5644 5645 5646
	/*
	 * 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) {
5647
		svm->vcpu.arch.hflags &= ~(HF_NMI_MASK | HF_IRET_MASK);
5648 5649
		kvm_make_request(KVM_REQ_EVENT, &svm->vcpu);
	}
5650

5651 5652 5653 5654 5655 5656 5657
	svm->vcpu.arch.nmi_injected = false;
	kvm_clear_exception_queue(&svm->vcpu);
	kvm_clear_interrupt_queue(&svm->vcpu);

	if (!(exitintinfo & SVM_EXITINTINFO_VALID))
		return;

5658 5659
	kvm_make_request(KVM_REQ_EVENT, &svm->vcpu);

5660 5661 5662 5663 5664 5665 5666 5667
	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:
5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678
		/*
		 * 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);
5679
			break;
5680
		}
5681 5682
		if (exitintinfo & SVM_EXITINTINFO_VALID_ERR) {
			u32 err = svm->vmcb->control.exit_int_info_err;
5683
			kvm_requeue_exception_e(&svm->vcpu, vector, err);
5684 5685

		} else
5686
			kvm_requeue_exception(&svm->vcpu, vector);
5687 5688
		break;
	case SVM_EXITINTINFO_TYPE_INTR:
5689
		kvm_queue_interrupt(&svm->vcpu, vector, false);
5690 5691 5692 5693 5694 5695
		break;
	default:
		break;
	}
}

A
Avi Kivity 已提交
5696 5697 5698 5699 5700 5701 5702 5703 5704 5705 5706
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
5707
static void svm_vcpu_run(struct kvm_vcpu *vcpu)
A
Avi Kivity 已提交
5708
{
5709
	struct vcpu_svm *svm = to_svm(vcpu);
5710

5711 5712 5713 5714
	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];

5715 5716 5717 5718 5719 5720 5721
	/*
	 * A vmexit emulation is required before the vcpu can be executed
	 * again.
	 */
	if (unlikely(svm->nested.exit_required))
		return;

5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735 5736 5737
	/*
	 * 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
5738
	pre_svm_run(svm);
A
Avi Kivity 已提交
5739

5740 5741
	sync_lapic_to_cr8(vcpu);

5742
	svm->vmcb->save.cr2 = vcpu->arch.cr2;
A
Avi Kivity 已提交
5743

5744
	clgi();
5745
	kvm_load_guest_xsave_state(vcpu);
5746

5747 5748 5749 5750
	if (lapic_in_kernel(vcpu) &&
		vcpu->arch.apic->lapic_timer.timer_advance_ns)
		kvm_wait_lapic_expire(vcpu);

5751 5752 5753 5754 5755 5756
	/*
	 * 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.
	 */
5757
	x86_spec_ctrl_set_guest(svm->spec_ctrl, svm->virt_spec_ctrl);
5758

5759 5760
	local_irq_enable();

A
Avi Kivity 已提交
5761
	asm volatile (
A
Avi Kivity 已提交
5762 5763 5764 5765 5766 5767 5768
		"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"
5769
#ifdef CONFIG_X86_64
R
Rusty Russell 已提交
5770 5771 5772 5773 5774 5775 5776 5777
		"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
5778 5779 5780
#endif

		/* Enter guest mode */
A
Avi Kivity 已提交
5781 5782
		"push %%" _ASM_AX " \n\t"
		"mov %c[vmcb](%[svm]), %%" _ASM_AX " \n\t"
5783 5784 5785
		__ex("vmload %%" _ASM_AX) "\n\t"
		__ex("vmrun %%" _ASM_AX) "\n\t"
		__ex("vmsave %%" _ASM_AX) "\n\t"
A
Avi Kivity 已提交
5786
		"pop %%" _ASM_AX " \n\t"
A
Avi Kivity 已提交
5787 5788

		/* Save guest registers, load host registers */
A
Avi Kivity 已提交
5789 5790 5791 5792 5793 5794
		"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"
5795
#ifdef CONFIG_X86_64
R
Rusty Russell 已提交
5796 5797 5798 5799 5800 5801 5802 5803
		"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"
5804 5805 5806 5807
		/*
		* Clear host registers marked as clobbered to prevent
		* speculative use.
		*/
5808 5809 5810 5811 5812 5813 5814 5815
		"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
5816
#endif
5817 5818 5819 5820 5821
		"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
5822
		"pop %%" _ASM_BP
A
Avi Kivity 已提交
5823
		:
R
Rusty Russell 已提交
5824
		: [svm]"a"(svm),
A
Avi Kivity 已提交
5825
		  [vmcb]"i"(offsetof(struct vcpu_svm, vmcb_pa)),
5826 5827 5828 5829 5830 5831
		  [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]))
5832
#ifdef CONFIG_X86_64
5833 5834 5835 5836 5837 5838 5839 5840
		  , [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
5841
#endif
5842 5843
		: "cc", "memory"
#ifdef CONFIG_X86_64
A
Avi Kivity 已提交
5844
		, "rbx", "rcx", "rdx", "rsi", "rdi"
5845
		, "r8", "r9", "r10", "r11" , "r12", "r13", "r14", "r15"
A
Avi Kivity 已提交
5846 5847
#else
		, "ebx", "ecx", "edx", "esi", "edi"
5848 5849
#endif
		);
A
Avi Kivity 已提交
5850

5851 5852 5853 5854 5855 5856 5857 5858 5859 5860 5861 5862
	/* 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

5863 5864 5865 5866 5867 5868 5869 5870 5871 5872 5873 5874 5875 5876 5877
	/*
	 * 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.
	 */
5878
	if (unlikely(!msr_write_intercepted(vcpu, MSR_IA32_SPEC_CTRL)))
5879
		svm->spec_ctrl = native_read_msr(MSR_IA32_SPEC_CTRL);
5880

A
Avi Kivity 已提交
5881 5882
	reload_tss(vcpu);

5883 5884
	local_irq_disable();

5885 5886
	x86_spec_ctrl_restore_host(svm->spec_ctrl, svm->virt_spec_ctrl);

5887 5888 5889 5890 5891
	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;

5892
	if (unlikely(svm->vmcb->control.exit_code == SVM_EXIT_NMI))
5893
		kvm_before_interrupt(&svm->vcpu);
5894

5895
	kvm_load_host_xsave_state(vcpu);
5896 5897 5898 5899 5900
	stgi();

	/* Any pending NMI will happen here */

	if (unlikely(svm->vmcb->control.exit_code == SVM_EXIT_NMI))
5901
		kvm_after_interrupt(&svm->vcpu);
5902

5903 5904
	sync_cr8_to_lapic(vcpu);

5905
	svm->next_rip = 0;
5906

5907 5908
	svm->vmcb->control.tlb_ctl = TLB_CONTROL_DO_NOTHING;

G
Gleb Natapov 已提交
5909 5910
	/* if exit due to PF check for async PF */
	if (svm->vmcb->control.exit_code == SVM_EXIT_EXCP_BASE + PF_VECTOR)
5911
		svm->vcpu.arch.apf.host_apf_reason = kvm_read_and_reset_pf_reason();
G
Gleb Natapov 已提交
5912

A
Avi Kivity 已提交
5913 5914 5915 5916
	if (npt_enabled) {
		vcpu->arch.regs_avail &= ~(1 << VCPU_EXREG_PDPTR);
		vcpu->arch.regs_dirty &= ~(1 << VCPU_EXREG_PDPTR);
	}
5917 5918 5919 5920 5921 5922 5923 5924

	/*
	 * 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);
5925 5926

	mark_all_clean(svm->vmcb);
A
Avi Kivity 已提交
5927
}
5928
STACK_FRAME_NON_STANDARD(svm_vcpu_run);
A
Avi Kivity 已提交
5929 5930 5931

static void svm_set_cr3(struct kvm_vcpu *vcpu, unsigned long root)
{
5932 5933
	struct vcpu_svm *svm = to_svm(vcpu);

5934
	svm->vmcb->save.cr3 = __sme_set(root);
5935
	mark_dirty(svm->vmcb, VMCB_CR);
A
Avi Kivity 已提交
5936 5937
}

5938 5939 5940 5941
static void set_tdp_cr3(struct kvm_vcpu *vcpu, unsigned long root)
{
	struct vcpu_svm *svm = to_svm(vcpu);

5942
	svm->vmcb->control.nested_cr3 = __sme_set(root);
5943
	mark_dirty(svm->vmcb, VMCB_NPT);
5944 5945

	/* Also sync guest cr3 here in case we live migrate */
5946
	svm->vmcb->save.cr3 = kvm_read_cr3(vcpu);
5947
	mark_dirty(svm->vmcb, VMCB_CR);
5948 5949
}

A
Avi Kivity 已提交
5950 5951
static int is_disabled(void)
{
5952 5953 5954 5955 5956 5957
	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
5958 5959 5960
	return 0;
}

I
Ingo Molnar 已提交
5961 5962 5963 5964 5965 5966 5967 5968 5969 5970 5971
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;
}

5972
static int __init svm_check_processor_compat(void)
Y
Yang, Sheng 已提交
5973
{
5974
	return 0;
Y
Yang, Sheng 已提交
5975 5976
}

5977 5978 5979 5980 5981
static bool svm_cpu_has_accelerated_tpr(void)
{
	return false;
}

5982
static bool svm_has_emulated_msr(int index)
5983
{
5984 5985
	switch (index) {
	case MSR_IA32_MCG_EXT_CTL:
5986
	case MSR_IA32_VMX_BASIC ... MSR_IA32_VMX_VMFUNC:
5987 5988 5989 5990 5991
		return false;
	default:
		break;
	}

5992 5993 5994
	return true;
}

5995 5996 5997 5998 5999
static u64 svm_get_mt_mask(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio)
{
	return 0;
}

6000 6001
static void svm_cpuid_update(struct kvm_vcpu *vcpu)
{
6002 6003
	struct vcpu_svm *svm = to_svm(vcpu);

6004
	vcpu->arch.xsaves_enabled = guest_cpuid_has(vcpu, X86_FEATURE_XSAVE) &&
6005
				    boot_cpu_has(X86_FEATURE_XSAVE) &&
6006 6007
				    boot_cpu_has(X86_FEATURE_XSAVES);

6008
	/* Update nrips enabled cache */
6009
	svm->nrips_enabled = !!guest_cpuid_has(&svm->vcpu, X86_FEATURE_NRIPS);
6010 6011 6012 6013

	if (!kvm_vcpu_apicv_active(vcpu))
		return;

6014
	guest_cpuid_clear(vcpu, X86_FEATURE_X2APIC);
6015 6016 6017 6018 6019 6020 6021 6022

	/*
	 * 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);
6023 6024
}

6025
#define F feature_bit
6026

6027 6028
static void svm_set_supported_cpuid(u32 func, struct kvm_cpuid_entry2 *entry)
{
6029
	switch (func) {
6030 6031
	case 0x1:
		if (avic)
6032
			entry->ecx &= ~F(X2APIC);
6033
		break;
6034 6035 6036 6037
	case 0x80000001:
		if (nested)
			entry->ecx |= (1 << 2); /* Set SVM bit */
		break;
6038 6039 6040 6041 6042
	case 0x80000008:
		if (boot_cpu_has(X86_FEATURE_LS_CFG_SSBD) ||
		     boot_cpu_has(X86_FEATURE_AMD_SSBD))
			entry->ebx |= F(VIRT_SSBD);
		break;
6043 6044 6045 6046 6047
	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 */
6048 6049 6050 6051
		entry->edx = 0; /* Per default do not support any
				   additional features */

		/* Support next_rip if host supports it */
6052
		if (boot_cpu_has(X86_FEATURE_NRIPS))
6053
			entry->edx |= F(NRIPS);
6054

6055 6056
		/* Support NPT for the guest if enabled */
		if (npt_enabled)
6057
			entry->edx |= F(NPT);
6058

6059
	}
6060 6061
}

6062
static int svm_get_lpage_level(void)
6063
{
6064
	return PT_PDPE_LEVEL;
6065 6066
}

6067 6068
static bool svm_rdtscp_supported(void)
{
P
Paolo Bonzini 已提交
6069
	return boot_cpu_has(X86_FEATURE_RDTSCP);
6070 6071
}

6072 6073 6074 6075 6076
static bool svm_invpcid_supported(void)
{
	return false;
}

6077 6078 6079 6080 6081
static bool svm_mpx_supported(void)
{
	return false;
}

6082 6083
static bool svm_xsaves_supported(void)
{
6084
	return boot_cpu_has(X86_FEATURE_XSAVES);
6085 6086
}

6087 6088 6089 6090 6091
static bool svm_umip_emulated(void)
{
	return false;
}

6092 6093 6094 6095 6096
static bool svm_pt_supported(void)
{
	return false;
}

6097 6098 6099 6100 6101
static bool svm_has_wbinvd_exit(void)
{
	return true;
}

6102 6103 6104 6105 6106
static bool svm_pku_supported(void)
{
	return false;
}

6107
#define PRE_EX(exit)  { .exit_code = (exit), \
6108
			.stage = X86_ICPT_PRE_EXCEPT, }
6109
#define POST_EX(exit) { .exit_code = (exit), \
6110
			.stage = X86_ICPT_POST_EXCEPT, }
6111
#define POST_MEM(exit) { .exit_code = (exit), \
6112
			.stage = X86_ICPT_POST_MEMACCESS, }
6113

6114
static const struct __x86_intercept {
6115 6116 6117 6118 6119 6120 6121 6122
	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),
6123 6124
	[x86_intercept_dr_read]		= POST_EX(SVM_EXIT_READ_DR0),
	[x86_intercept_dr_write]	= POST_EX(SVM_EXIT_WRITE_DR0),
6125 6126 6127 6128 6129 6130 6131 6132
	[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),
6133 6134 6135 6136 6137 6138 6139 6140
	[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),
6141 6142 6143
	[x86_intercept_rdtscp]		= POST_EX(SVM_EXIT_RDTSCP),
	[x86_intercept_monitor]		= POST_MEM(SVM_EXIT_MONITOR),
	[x86_intercept_mwait]		= POST_EX(SVM_EXIT_MWAIT),
6144 6145 6146 6147 6148 6149 6150 6151 6152
	[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),
6153 6154 6155 6156 6157 6158 6159
	[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),
6160 6161 6162 6163
	[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),
6164
	[x86_intercept_xsetbv]		= PRE_EX(SVM_EXIT_XSETBV),
6165 6166
};

6167
#undef PRE_EX
6168
#undef POST_EX
6169
#undef POST_MEM
6170

6171 6172 6173 6174
static int svm_check_intercept(struct kvm_vcpu *vcpu,
			       struct x86_instruction_info *info,
			       enum x86_intercept_stage stage)
{
6175 6176 6177 6178 6179 6180 6181 6182 6183 6184
	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];

6185
	if (stage != icpt_info.stage)
6186 6187 6188 6189 6190 6191 6192 6193 6194 6195 6196 6197 6198 6199
		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;

6200 6201
		if (icpt_info.exit_code != SVM_EXIT_WRITE_CR0 ||
		    info->intercept == x86_intercept_clts)
6202 6203 6204 6205 6206 6207 6208 6209 6210 6211 6212 6213 6214 6215 6216 6217 6218 6219 6220 6221 6222 6223 6224
			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;
	}
6225 6226 6227 6228
	case SVM_EXIT_READ_DR0:
	case SVM_EXIT_WRITE_DR0:
		icpt_info.exit_code += info->modrm_reg;
		break;
6229 6230 6231 6232 6233 6234
	case SVM_EXIT_MSR:
		if (info->intercept == x86_intercept_wrmsr)
			vmcb->control.exit_info_1 = 1;
		else
			vmcb->control.exit_info_1 = 0;
		break;
6235 6236 6237 6238 6239 6240 6241
	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;
6242
		break;
6243 6244 6245 6246 6247 6248
	case SVM_EXIT_IOIO: {
		u64 exit_info;
		u32 bytes;

		if (info->intercept == x86_intercept_in ||
		    info->intercept == x86_intercept_ins) {
6249 6250
			exit_info = ((info->src_val & 0xffff) << 16) |
				SVM_IOIO_TYPE_MASK;
6251
			bytes = info->dst_bytes;
6252
		} else {
6253
			exit_info = (info->dst_val & 0xffff) << 16;
6254
			bytes = info->src_bytes;
6255 6256 6257 6258 6259 6260 6261 6262 6263 6264 6265 6266 6267 6268 6269 6270 6271 6272 6273 6274
		}

		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;
	}
6275 6276 6277 6278
	default:
		break;
	}

6279 6280 6281
	/* TODO: Advertise NRIPS to guest hypervisor unconditionally */
	if (static_cpu_has(X86_FEATURE_NRIPS))
		vmcb->control.next_rip  = info->next_rip;
6282 6283 6284 6285 6286 6287 6288 6289
	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;
6290 6291
}

6292 6293
static void svm_handle_exit_irqoff(struct kvm_vcpu *vcpu,
	enum exit_fastpath_completion *exit_fastpath)
6294
{
6295 6296 6297
	if (!is_guest_mode(vcpu) &&
		to_svm(vcpu)->vmcb->control.exit_code == EXIT_REASON_MSR_WRITE)
		*exit_fastpath = handle_fastpath_set_msr_irqoff(vcpu);
6298 6299
}

6300 6301
static void svm_sched_in(struct kvm_vcpu *vcpu, int cpu)
{
6302 6303
	if (pause_filter_thresh)
		shrink_ple_window(vcpu);
6304 6305
}

6306 6307 6308 6309
static inline void avic_post_state_restore(struct kvm_vcpu *vcpu)
{
	if (avic_handle_apic_id_update(vcpu) != 0)
		return;
6310
	avic_handle_dfr_update(vcpu);
6311 6312 6313
	avic_handle_ldr_update(vcpu);
}

6314 6315 6316 6317 6318 6319
static void svm_setup_mce(struct kvm_vcpu *vcpu)
{
	/* [63:9] are reserved. */
	vcpu->arch.mcg_cap &= 0x1ff;
}

6320 6321
static int svm_smi_allowed(struct kvm_vcpu *vcpu)
{
6322 6323 6324 6325 6326 6327 6328 6329 6330 6331 6332 6333 6334 6335
	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;
	}

6336 6337 6338
	return 1;
}

6339 6340
static int svm_pre_enter_smm(struct kvm_vcpu *vcpu, char *smstate)
{
6341 6342 6343 6344 6345 6346 6347 6348 6349 6350 6351 6352 6353 6354 6355 6356 6357
	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;
	}
6358 6359 6360
	return 0;
}

6361
static int svm_pre_leave_smm(struct kvm_vcpu *vcpu, const char *smstate)
6362
{
6363 6364
	struct vcpu_svm *svm = to_svm(vcpu);
	struct vmcb *nested_vmcb;
6365
	struct kvm_host_map map;
6366 6367
	u64 guest;
	u64 vmcb;
6368

6369 6370
	guest = GET_SMSTATE(u64, smstate, 0x7ed8);
	vmcb = GET_SMSTATE(u64, smstate, 0x7ee0);
6371

6372
	if (guest) {
6373
		if (kvm_vcpu_map(&svm->vcpu, gpa_to_gfn(vmcb), &map) == -EINVAL)
6374
			return 1;
6375 6376
		nested_vmcb = map.hva;
		enter_svm_guest_mode(svm, vmcb, nested_vmcb, &map);
6377
	}
6378
	return 0;
6379 6380
}

6381 6382 6383 6384 6385 6386 6387 6388 6389 6390 6391 6392 6393
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;
}

6394 6395 6396 6397 6398 6399 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
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
6436 6437
static int sev_asid_new(void)
{
6438
	bool retry = true;
B
Brijesh Singh 已提交
6439 6440
	int pos;

6441 6442
	mutex_lock(&sev_bitmap_lock);

B
Brijesh Singh 已提交
6443 6444 6445
	/*
	 * SEV-enabled guest must use asid from min_sev_asid to max_sev_asid.
	 */
6446
again:
B
Brijesh Singh 已提交
6447
	pos = find_next_zero_bit(sev_asid_bitmap, max_sev_asid, min_sev_asid - 1);
6448
	if (pos >= max_sev_asid) {
6449 6450 6451 6452
		if (retry && __sev_recycle_asids()) {
			retry = false;
			goto again;
		}
6453
		mutex_unlock(&sev_bitmap_lock);
B
Brijesh Singh 已提交
6454
		return -EBUSY;
6455 6456 6457 6458 6459
	}

	__set_bit(pos, sev_asid_bitmap);

	mutex_unlock(&sev_bitmap_lock);
B
Brijesh Singh 已提交
6460 6461 6462 6463 6464 6465

	return pos + 1;
}

static int sev_guest_init(struct kvm *kvm, struct kvm_sev_cmd *argp)
{
6466
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
B
Brijesh Singh 已提交
6467 6468 6469
	int asid, ret;

	ret = -EBUSY;
6470 6471 6472
	if (unlikely(sev->active))
		return ret;

B
Brijesh Singh 已提交
6473 6474 6475 6476 6477 6478 6479 6480 6481 6482
	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;
6483
	INIT_LIST_HEAD(&sev->regions_list);
B
Brijesh Singh 已提交
6484 6485 6486 6487

	return 0;

e_free:
6488
	sev_asid_free(asid);
B
Brijesh Singh 已提交
6489 6490 6491
	return ret;
}

6492 6493 6494 6495 6496 6497
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;

6498
	data = kzalloc(sizeof(*data), GFP_KERNEL_ACCOUNT);
6499 6500 6501 6502 6503 6504 6505 6506 6507 6508 6509 6510
	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;
}

6511
static int __sev_issue_cmd(int fd, int id, void *data, int *error)
6512 6513 6514 6515 6516 6517 6518 6519 6520 6521 6522 6523 6524 6525
{
	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;
}

6526 6527
static int sev_issue_cmd(struct kvm *kvm, int id, void *data, int *error)
{
6528
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
6529 6530 6531 6532

	return __sev_issue_cmd(sev->fd, id, data, error);
}

6533 6534
static int sev_launch_start(struct kvm *kvm, struct kvm_sev_cmd *argp)
{
6535
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
6536 6537 6538 6539 6540 6541 6542 6543 6544 6545 6546 6547
	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;

6548
	start = kzalloc(sizeof(*start), GFP_KERNEL_ACCOUNT);
6549 6550 6551 6552 6553 6554 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
	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 */
6580
	ret = __sev_issue_cmd(argp->sev_fd, SEV_CMD_LAUNCH_START, start, error);
6581 6582 6583 6584 6585 6586 6587 6588 6589 6590 6591 6592 6593 6594 6595 6596 6597 6598 6599 6600 6601 6602 6603 6604 6605 6606 6607 6608
	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;
}

6609 6610
static unsigned long get_num_contig_pages(unsigned long idx,
				struct page **inpages, unsigned long npages)
6611 6612
{
	unsigned long paddr, next_paddr;
6613
	unsigned long i = idx + 1, pages = 1;
6614 6615 6616 6617 6618 6619 6620 6621 6622 6623 6624 6625 6626 6627 6628 6629 6630 6631

	/* 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)
{
6632
	unsigned long vaddr, vaddr_end, next_vaddr, npages, pages, size, i;
6633
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
6634 6635 6636
	struct kvm_sev_launch_update_data params;
	struct sev_data_launch_update_data *data;
	struct page **inpages;
6637
	int ret;
6638 6639 6640 6641 6642 6643 6644

	if (!sev_guest(kvm))
		return -ENOTTY;

	if (copy_from_user(&params, (void __user *)(uintptr_t)argp->data, sizeof(params)))
		return -EFAULT;

6645
	data = kzalloc(sizeof(*data), GFP_KERNEL_ACCOUNT);
6646 6647 6648 6649 6650 6651 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
	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;
}

6706 6707
static int sev_launch_measure(struct kvm *kvm, struct kvm_sev_cmd *argp)
{
6708
	void __user *measure = (void __user *)(uintptr_t)argp->data;
6709
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
6710 6711
	struct sev_data_launch_measure *data;
	struct kvm_sev_launch_measure params;
6712
	void __user *p = NULL;
6713 6714 6715 6716 6717 6718
	void *blob = NULL;
	int ret;

	if (!sev_guest(kvm))
		return -ENOTTY;

6719
	if (copy_from_user(&params, measure, sizeof(params)))
6720 6721
		return -EFAULT;

6722
	data = kzalloc(sizeof(*data), GFP_KERNEL_ACCOUNT);
6723 6724 6725 6726 6727 6728 6729
	if (!data)
		return -ENOMEM;

	/* User wants to query the blob length */
	if (!params.len)
		goto cmd;

6730 6731
	p = (void __user *)(uintptr_t)params.uaddr;
	if (p) {
6732 6733 6734 6735 6736 6737 6738 6739 6740 6741 6742 6743 6744 6745 6746 6747 6748 6749 6750 6751 6752 6753 6754 6755 6756 6757 6758 6759
		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) {
6760
		if (copy_to_user(p, blob, params.len))
6761 6762 6763 6764 6765
			ret = -EFAULT;
	}

done:
	params.len = data->len;
6766
	if (copy_to_user(measure, &params, sizeof(params)))
6767 6768 6769 6770 6771 6772 6773 6774
		ret = -EFAULT;
e_free_blob:
	kfree(blob);
e_free:
	kfree(data);
	return ret;
}

6775 6776
static int sev_launch_finish(struct kvm *kvm, struct kvm_sev_cmd *argp)
{
6777
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
6778 6779 6780 6781 6782 6783
	struct sev_data_launch_finish *data;
	int ret;

	if (!sev_guest(kvm))
		return -ENOTTY;

6784
	data = kzalloc(sizeof(*data), GFP_KERNEL_ACCOUNT);
6785 6786 6787 6788 6789 6790 6791 6792 6793 6794
	if (!data)
		return -ENOMEM;

	data->handle = sev->handle;
	ret = sev_issue_cmd(kvm, SEV_CMD_LAUNCH_FINISH, data, &argp->error);

	kfree(data);
	return ret;
}

6795 6796
static int sev_guest_status(struct kvm *kvm, struct kvm_sev_cmd *argp)
{
6797
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
6798 6799 6800 6801 6802 6803 6804
	struct kvm_sev_guest_status params;
	struct sev_data_guest_status *data;
	int ret;

	if (!sev_guest(kvm))
		return -ENOTTY;

6805
	data = kzalloc(sizeof(*data), GFP_KERNEL_ACCOUNT);
6806 6807 6808 6809 6810 6811 6812 6813 6814 6815 6816 6817 6818 6819 6820 6821 6822 6823 6824
	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;
}

6825 6826 6827 6828
static int __sev_issue_dbg_cmd(struct kvm *kvm, unsigned long src,
			       unsigned long dst, int size,
			       int *error, bool enc)
{
6829
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
6830 6831 6832
	struct sev_data_dbg *data;
	int ret;

6833
	data = kzalloc(sizeof(*data), GFP_KERNEL_ACCOUNT);
6834 6835 6836 6837 6838 6839 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
	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;
}

6902 6903 6904 6905 6906 6907 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
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;
}

6979 6980 6981
static int sev_dbg_crypt(struct kvm *kvm, struct kvm_sev_cmd *argp, bool dec)
{
	unsigned long vaddr, vaddr_end, next_vaddr;
6982
	unsigned long dst_vaddr;
6983 6984 6985
	struct page **src_p, **dst_p;
	struct kvm_sev_dbg debug;
	unsigned long n;
6986 6987
	unsigned int size;
	int ret;
6988 6989 6990 6991 6992 6993 6994

	if (!sev_guest(kvm))
		return -ENOTTY;

	if (copy_from_user(&debug, (void __user *)(uintptr_t)argp->data, sizeof(debug)))
		return -EFAULT;

6995 6996 6997 6998 6999
	if (!debug.len || debug.src_uaddr + debug.len < debug.src_uaddr)
		return -EINVAL;
	if (!debug.dst_uaddr)
		return -EINVAL;

7000 7001 7002 7003 7004 7005 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
	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);

7036 7037 7038 7039 7040 7041 7042 7043 7044 7045 7046 7047 7048
		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);
7049

7050 7051
		sev_unpin_memory(kvm, src_p, n);
		sev_unpin_memory(kvm, dst_p, n);
7052 7053 7054 7055 7056 7057 7058 7059 7060 7061 7062 7063

		if (ret)
			goto err;

		next_vaddr = vaddr + len;
		dst_vaddr = dst_vaddr + len;
		size -= len;
	}
err:
	return ret;
}

7064 7065
static int sev_launch_secret(struct kvm *kvm, struct kvm_sev_cmd *argp)
{
7066
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
7067 7068 7069 7070 7071
	struct sev_data_launch_secret *data;
	struct kvm_sev_launch_secret params;
	struct page **pages;
	void *blob, *hdr;
	unsigned long n;
7072
	int ret, offset;
7073 7074 7075 7076 7077 7078 7079 7080 7081 7082 7083 7084 7085 7086 7087 7088 7089 7090 7091 7092 7093

	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;
7094
	data = kzalloc(sizeof(*data), GFP_KERNEL_ACCOUNT);
7095 7096 7097
	if (!data)
		goto e_unpin_memory;

7098 7099 7100 7101
	offset = params.guest_uaddr & (PAGE_SIZE - 1);
	data->guest_address = __sme_page_pa(pages[0]) + offset;
	data->guest_len = params.guest_len;

7102 7103 7104 7105 7106 7107 7108 7109 7110 7111 7112 7113 7114 7115
	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;
	}
7116 7117
	data->hdr_address = __psp_pa(hdr);
	data->hdr_len = params.hdr_len;
7118 7119 7120 7121 7122 7123 7124 7125 7126 7127 7128 7129 7130 7131 7132

	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
7133 7134 7135 7136 7137 7138 7139 7140 7141 7142 7143 7144 7145 7146 7147 7148 7149
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;
7150 7151 7152
	case KVM_SEV_LAUNCH_START:
		r = sev_launch_start(kvm, &sev_cmd);
		break;
7153 7154 7155
	case KVM_SEV_LAUNCH_UPDATE_DATA:
		r = sev_launch_update_data(kvm, &sev_cmd);
		break;
7156 7157 7158
	case KVM_SEV_LAUNCH_MEASURE:
		r = sev_launch_measure(kvm, &sev_cmd);
		break;
7159 7160 7161
	case KVM_SEV_LAUNCH_FINISH:
		r = sev_launch_finish(kvm, &sev_cmd);
		break;
7162 7163 7164
	case KVM_SEV_GUEST_STATUS:
		r = sev_guest_status(kvm, &sev_cmd);
		break;
7165 7166 7167
	case KVM_SEV_DBG_DECRYPT:
		r = sev_dbg_crypt(kvm, &sev_cmd, true);
		break;
7168 7169 7170
	case KVM_SEV_DBG_ENCRYPT:
		r = sev_dbg_crypt(kvm, &sev_cmd, false);
		break;
7171 7172 7173
	case KVM_SEV_LAUNCH_SECRET:
		r = sev_launch_secret(kvm, &sev_cmd);
		break;
B
Brijesh Singh 已提交
7174 7175 7176 7177 7178 7179 7180 7181 7182 7183 7184 7185 7186
	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;
}

7187 7188 7189
static int svm_register_enc_region(struct kvm *kvm,
				   struct kvm_enc_region *range)
{
7190
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
7191 7192 7193 7194 7195 7196
	struct enc_region *region;
	int ret = 0;

	if (!sev_guest(kvm))
		return -ENOTTY;

7197 7198 7199
	if (range->addr > ULONG_MAX || range->size > ULONG_MAX)
		return -EINVAL;

7200
	region = kzalloc(sizeof(*region), GFP_KERNEL_ACCOUNT);
7201 7202 7203 7204 7205 7206 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
	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)
{
7235
	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
7236 7237 7238 7239 7240 7241 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
	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;
}

7278 7279
static bool svm_need_emulation_on_page_fault(struct kvm_vcpu *vcpu)
{
7280 7281 7282 7283
	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;
7284 7285

	/*
7286 7287 7288 7289 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
	 * 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.
7315 7316 7317 7318 7319 7320 7321 7322 7323 7324
	 *
	 * 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.
	 */
7325
	if (smap && (!smep || is_user)) {
7326 7327 7328
		if (!sev_guest(vcpu->kvm))
			return true;

7329
		pr_err_ratelimited("KVM: SEV Guest triggered AMD Erratum 1096\n");
7330 7331 7332 7333 7334 7335
		kvm_make_request(KVM_REQ_TRIPLE_FAULT, vcpu);
	}

	return false;
}

7336 7337 7338 7339 7340 7341 7342 7343 7344 7345 7346 7347 7348 7349 7350
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));
}

7351 7352
static bool svm_check_apicv_inhibit_reasons(ulong bit)
{
7353
	ulong supported = BIT(APICV_INHIBIT_REASON_DISABLE) |
7354
			  BIT(APICV_INHIBIT_REASON_HYPERV) |
7355 7356
			  BIT(APICV_INHIBIT_REASON_NESTED) |
			  BIT(APICV_INHIBIT_REASON_IRQWIN);
7357 7358 7359 7360

	return supported & BIT(bit);
}

7361 7362 7363 7364 7365
static void svm_pre_update_apicv_exec_ctrl(struct kvm *kvm, bool activate)
{
	avic_update_access_page(kvm, activate);
}

7366
static struct kvm_x86_ops svm_x86_ops __ro_after_init = {
A
Avi Kivity 已提交
7367 7368 7369 7370
	.cpu_has_kvm_support = has_svm,
	.disabled_by_bios = is_disabled,
	.hardware_setup = svm_hardware_setup,
	.hardware_unsetup = svm_hardware_unsetup,
Y
Yang, Sheng 已提交
7371
	.check_processor_compatibility = svm_check_processor_compat,
A
Avi Kivity 已提交
7372 7373
	.hardware_enable = svm_hardware_enable,
	.hardware_disable = svm_hardware_disable,
7374
	.cpu_has_accelerated_tpr = svm_cpu_has_accelerated_tpr,
7375
	.has_emulated_msr = svm_has_emulated_msr,
A
Avi Kivity 已提交
7376 7377 7378

	.vcpu_create = svm_create_vcpu,
	.vcpu_free = svm_free_vcpu,
7379
	.vcpu_reset = svm_vcpu_reset,
A
Avi Kivity 已提交
7380

7381 7382
	.vm_alloc = svm_vm_alloc,
	.vm_free = svm_vm_free,
7383
	.vm_init = svm_vm_init,
B
Brijesh Singh 已提交
7384
	.vm_destroy = svm_vm_destroy,
7385

7386
	.prepare_guest_switch = svm_prepare_guest_switch,
A
Avi Kivity 已提交
7387 7388
	.vcpu_load = svm_vcpu_load,
	.vcpu_put = svm_vcpu_put,
7389 7390
	.vcpu_blocking = svm_vcpu_blocking,
	.vcpu_unblocking = svm_vcpu_unblocking,
A
Avi Kivity 已提交
7391

7392
	.update_bp_intercept = update_bp_intercept,
7393
	.get_msr_feature = svm_get_msr_feature,
A
Avi Kivity 已提交
7394 7395 7396 7397 7398
	.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,
7399
	.get_cpl = svm_get_cpl,
7400
	.get_cs_db_l_bits = kvm_get_cs_db_l_bits,
7401
	.decache_cr0_guest_bits = svm_decache_cr0_guest_bits,
7402
	.decache_cr4_guest_bits = svm_decache_cr4_guest_bits,
A
Avi Kivity 已提交
7403 7404 7405 7406 7407 7408 7409 7410
	.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
7411 7412
	.get_dr6 = svm_get_dr6,
	.set_dr6 = svm_set_dr6,
7413
	.set_dr7 = svm_set_dr7,
7414
	.sync_dirty_debug_regs = svm_sync_dirty_debug_regs,
A
Avi Kivity 已提交
7415
	.cache_reg = svm_cache_reg,
A
Avi Kivity 已提交
7416 7417
	.get_rflags = svm_get_rflags,
	.set_rflags = svm_set_rflags,
7418

A
Avi Kivity 已提交
7419
	.tlb_flush = svm_flush_tlb,
7420
	.tlb_flush_gva = svm_flush_tlb_gva,
A
Avi Kivity 已提交
7421 7422

	.run = svm_vcpu_run,
7423
	.handle_exit = handle_exit,
A
Avi Kivity 已提交
7424
	.skip_emulated_instruction = skip_emulated_instruction,
7425 7426
	.set_interrupt_shadow = svm_set_interrupt_shadow,
	.get_interrupt_shadow = svm_get_interrupt_shadow,
I
Ingo Molnar 已提交
7427
	.patch_hypercall = svm_patch_hypercall,
E
Eddie Dong 已提交
7428
	.set_irq = svm_set_irq,
7429
	.set_nmi = svm_inject_nmi,
7430
	.queue_exception = svm_queue_exception,
A
Avi Kivity 已提交
7431
	.cancel_injection = svm_cancel_injection,
7432
	.interrupt_allowed = svm_interrupt_allowed,
7433
	.nmi_allowed = svm_nmi_allowed,
J
Jan Kiszka 已提交
7434 7435
	.get_nmi_mask = svm_get_nmi_mask,
	.set_nmi_mask = svm_set_nmi_mask,
7436 7437 7438
	.enable_nmi_window = enable_nmi_window,
	.enable_irq_window = enable_irq_window,
	.update_cr8_intercept = update_cr8_intercept,
7439
	.set_virtual_apic_mode = svm_set_virtual_apic_mode,
7440
	.refresh_apicv_exec_ctrl = svm_refresh_apicv_exec_ctrl,
7441
	.check_apicv_inhibit_reasons = svm_check_apicv_inhibit_reasons,
7442
	.pre_update_apicv_exec_ctrl = svm_pre_update_apicv_exec_ctrl,
7443
	.load_eoi_exitmap = svm_load_eoi_exitmap,
7444 7445
	.hwapic_irr_update = svm_hwapic_irr_update,
	.hwapic_isr_update = svm_hwapic_isr_update,
7446
	.sync_pir_to_irr = kvm_lapic_find_highest_irr,
7447
	.apicv_post_state_restore = avic_post_state_restore,
7448 7449

	.set_tss_addr = svm_set_tss_addr,
7450
	.set_identity_map_addr = svm_set_identity_map_addr,
7451
	.get_tdp_level = get_npt_level,
7452
	.get_mt_mask = svm_get_mt_mask,
7453

7454 7455
	.get_exit_info = svm_get_exit_info,

7456
	.get_lpage_level = svm_get_lpage_level,
7457 7458

	.cpuid_update = svm_cpuid_update,
7459 7460

	.rdtscp_supported = svm_rdtscp_supported,
7461
	.invpcid_supported = svm_invpcid_supported,
7462
	.mpx_supported = svm_mpx_supported,
7463
	.xsaves_supported = svm_xsaves_supported,
7464
	.umip_emulated = svm_umip_emulated,
7465
	.pt_supported = svm_pt_supported,
7466
	.pku_supported = svm_pku_supported,
7467 7468

	.set_supported_cpuid = svm_set_supported_cpuid,
7469 7470

	.has_wbinvd_exit = svm_has_wbinvd_exit,
7471

7472
	.read_l1_tsc_offset = svm_read_l1_tsc_offset,
7473
	.write_l1_tsc_offset = svm_write_l1_tsc_offset,
7474 7475

	.set_tdp_cr3 = set_tdp_cr3,
7476 7477

	.check_intercept = svm_check_intercept,
7478
	.handle_exit_irqoff = svm_handle_exit_irqoff,
7479

7480 7481
	.request_immediate_exit = __kvm_request_immediate_exit,

7482
	.sched_in = svm_sched_in,
7483 7484

	.pmu_ops = &amd_pmu_ops,
7485
	.deliver_posted_interrupt = svm_deliver_avic_intr,
7486
	.dy_apicv_has_pending_interrupt = svm_dy_apicv_has_pending_interrupt,
7487
	.update_pi_irte = svm_update_pi_irte,
7488
	.setup_mce = svm_setup_mce,
7489

7490
	.smi_allowed = svm_smi_allowed,
7491 7492
	.pre_enter_smm = svm_pre_enter_smm,
	.pre_leave_smm = svm_pre_leave_smm,
7493
	.enable_smi_window = enable_smi_window,
B
Brijesh Singh 已提交
7494 7495

	.mem_enc_op = svm_mem_enc_op,
7496 7497
	.mem_enc_reg_region = svm_register_enc_region,
	.mem_enc_unreg_region = svm_unregister_enc_region,
7498

7499
	.nested_enable_evmcs = NULL,
7500
	.nested_get_evmcs_version = NULL,
7501 7502

	.need_emulation_on_page_fault = svm_need_emulation_on_page_fault,
7503 7504

	.apic_init_signal_blocked = svm_apic_init_signal_blocked,
A
Avi Kivity 已提交
7505 7506 7507 7508
};

static int __init svm_init(void)
{
7509
	return kvm_init(&svm_x86_ops, sizeof(struct vcpu_svm),
7510
			__alignof__(struct vcpu_svm), THIS_MODULE);
A
Avi Kivity 已提交
7511 7512 7513 7514
}

static void __exit svm_exit(void)
{
7515
	kvm_exit();
A
Avi Kivity 已提交
7516 7517 7518 7519
}

module_init(svm_init)
module_exit(svm_exit)