xref: /linux/arch/x86/include/asm/kvm_host.h (revision 52338415)
1 /* SPDX-License-Identifier: GPL-2.0-only */
2 /*
3  * Kernel-based Virtual Machine driver for Linux
4  *
5  * This header defines architecture specific interfaces, x86 version
6  */
7 
8 #ifndef _ASM_X86_KVM_HOST_H
9 #define _ASM_X86_KVM_HOST_H
10 
11 #include <linux/types.h>
12 #include <linux/mm.h>
13 #include <linux/mmu_notifier.h>
14 #include <linux/tracepoint.h>
15 #include <linux/cpumask.h>
16 #include <linux/irq_work.h>
17 #include <linux/irq.h>
18 
19 #include <linux/kvm.h>
20 #include <linux/kvm_para.h>
21 #include <linux/kvm_types.h>
22 #include <linux/perf_event.h>
23 #include <linux/pvclock_gtod.h>
24 #include <linux/clocksource.h>
25 #include <linux/irqbypass.h>
26 #include <linux/hyperv.h>
27 
28 #include <asm/apic.h>
29 #include <asm/pvclock-abi.h>
30 #include <asm/desc.h>
31 #include <asm/mtrr.h>
32 #include <asm/msr-index.h>
33 #include <asm/asm.h>
34 #include <asm/kvm_page_track.h>
35 #include <asm/kvm_vcpu_regs.h>
36 #include <asm/hyperv-tlfs.h>
37 
38 #define __KVM_HAVE_ARCH_VCPU_DEBUGFS
39 
40 #define KVM_MAX_VCPUS 288
41 #define KVM_SOFT_MAX_VCPUS 240
42 #define KVM_MAX_VCPU_ID 1023
43 #define KVM_USER_MEM_SLOTS 509
44 /* memory slots that are not exposed to userspace */
45 #define KVM_PRIVATE_MEM_SLOTS 3
46 #define KVM_MEM_SLOTS_NUM (KVM_USER_MEM_SLOTS + KVM_PRIVATE_MEM_SLOTS)
47 
48 #define KVM_HALT_POLL_NS_DEFAULT 200000
49 
50 #define KVM_IRQCHIP_NUM_PINS  KVM_IOAPIC_NUM_PINS
51 
52 /* x86-specific vcpu->requests bit members */
53 #define KVM_REQ_MIGRATE_TIMER		KVM_ARCH_REQ(0)
54 #define KVM_REQ_REPORT_TPR_ACCESS	KVM_ARCH_REQ(1)
55 #define KVM_REQ_TRIPLE_FAULT		KVM_ARCH_REQ(2)
56 #define KVM_REQ_MMU_SYNC		KVM_ARCH_REQ(3)
57 #define KVM_REQ_CLOCK_UPDATE		KVM_ARCH_REQ(4)
58 #define KVM_REQ_LOAD_CR3		KVM_ARCH_REQ(5)
59 #define KVM_REQ_EVENT			KVM_ARCH_REQ(6)
60 #define KVM_REQ_APF_HALT		KVM_ARCH_REQ(7)
61 #define KVM_REQ_STEAL_UPDATE		KVM_ARCH_REQ(8)
62 #define KVM_REQ_NMI			KVM_ARCH_REQ(9)
63 #define KVM_REQ_PMU			KVM_ARCH_REQ(10)
64 #define KVM_REQ_PMI			KVM_ARCH_REQ(11)
65 #define KVM_REQ_SMI			KVM_ARCH_REQ(12)
66 #define KVM_REQ_MASTERCLOCK_UPDATE	KVM_ARCH_REQ(13)
67 #define KVM_REQ_MCLOCK_INPROGRESS \
68 	KVM_ARCH_REQ_FLAGS(14, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
69 #define KVM_REQ_SCAN_IOAPIC \
70 	KVM_ARCH_REQ_FLAGS(15, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
71 #define KVM_REQ_GLOBAL_CLOCK_UPDATE	KVM_ARCH_REQ(16)
72 #define KVM_REQ_APIC_PAGE_RELOAD \
73 	KVM_ARCH_REQ_FLAGS(17, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
74 #define KVM_REQ_HV_CRASH		KVM_ARCH_REQ(18)
75 #define KVM_REQ_IOAPIC_EOI_EXIT		KVM_ARCH_REQ(19)
76 #define KVM_REQ_HV_RESET		KVM_ARCH_REQ(20)
77 #define KVM_REQ_HV_EXIT			KVM_ARCH_REQ(21)
78 #define KVM_REQ_HV_STIMER		KVM_ARCH_REQ(22)
79 #define KVM_REQ_LOAD_EOI_EXITMAP	KVM_ARCH_REQ(23)
80 #define KVM_REQ_GET_VMCS12_PAGES	KVM_ARCH_REQ(24)
81 
82 #define CR0_RESERVED_BITS                                               \
83 	(~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
84 			  | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
85 			  | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
86 
87 #define CR4_RESERVED_BITS                                               \
88 	(~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
89 			  | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE     \
90 			  | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR | X86_CR4_PCIDE \
91 			  | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_FSGSBASE \
92 			  | X86_CR4_OSXMMEXCPT | X86_CR4_LA57 | X86_CR4_VMXE \
93 			  | X86_CR4_SMAP | X86_CR4_PKE | X86_CR4_UMIP))
94 
95 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
96 
97 
98 
99 #define INVALID_PAGE (~(hpa_t)0)
100 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
101 
102 #define UNMAPPED_GVA (~(gpa_t)0)
103 
104 /* KVM Hugepage definitions for x86 */
105 enum {
106 	PT_PAGE_TABLE_LEVEL   = 1,
107 	PT_DIRECTORY_LEVEL    = 2,
108 	PT_PDPE_LEVEL         = 3,
109 	/* set max level to the biggest one */
110 	PT_MAX_HUGEPAGE_LEVEL = PT_PDPE_LEVEL,
111 };
112 #define KVM_NR_PAGE_SIZES	(PT_MAX_HUGEPAGE_LEVEL - \
113 				 PT_PAGE_TABLE_LEVEL + 1)
114 #define KVM_HPAGE_GFN_SHIFT(x)	(((x) - 1) * 9)
115 #define KVM_HPAGE_SHIFT(x)	(PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
116 #define KVM_HPAGE_SIZE(x)	(1UL << KVM_HPAGE_SHIFT(x))
117 #define KVM_HPAGE_MASK(x)	(~(KVM_HPAGE_SIZE(x) - 1))
118 #define KVM_PAGES_PER_HPAGE(x)	(KVM_HPAGE_SIZE(x) / PAGE_SIZE)
119 
120 static inline gfn_t gfn_to_index(gfn_t gfn, gfn_t base_gfn, int level)
121 {
122 	/* KVM_HPAGE_GFN_SHIFT(PT_PAGE_TABLE_LEVEL) must be 0. */
123 	return (gfn >> KVM_HPAGE_GFN_SHIFT(level)) -
124 		(base_gfn >> KVM_HPAGE_GFN_SHIFT(level));
125 }
126 
127 #define KVM_PERMILLE_MMU_PAGES 20
128 #define KVM_MIN_ALLOC_MMU_PAGES 64UL
129 #define KVM_MMU_HASH_SHIFT 12
130 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
131 #define KVM_MIN_FREE_MMU_PAGES 5
132 #define KVM_REFILL_PAGES 25
133 #define KVM_MAX_CPUID_ENTRIES 80
134 #define KVM_NR_FIXED_MTRR_REGION 88
135 #define KVM_NR_VAR_MTRR 8
136 
137 #define ASYNC_PF_PER_VCPU 64
138 
139 enum kvm_reg {
140 	VCPU_REGS_RAX = __VCPU_REGS_RAX,
141 	VCPU_REGS_RCX = __VCPU_REGS_RCX,
142 	VCPU_REGS_RDX = __VCPU_REGS_RDX,
143 	VCPU_REGS_RBX = __VCPU_REGS_RBX,
144 	VCPU_REGS_RSP = __VCPU_REGS_RSP,
145 	VCPU_REGS_RBP = __VCPU_REGS_RBP,
146 	VCPU_REGS_RSI = __VCPU_REGS_RSI,
147 	VCPU_REGS_RDI = __VCPU_REGS_RDI,
148 #ifdef CONFIG_X86_64
149 	VCPU_REGS_R8  = __VCPU_REGS_R8,
150 	VCPU_REGS_R9  = __VCPU_REGS_R9,
151 	VCPU_REGS_R10 = __VCPU_REGS_R10,
152 	VCPU_REGS_R11 = __VCPU_REGS_R11,
153 	VCPU_REGS_R12 = __VCPU_REGS_R12,
154 	VCPU_REGS_R13 = __VCPU_REGS_R13,
155 	VCPU_REGS_R14 = __VCPU_REGS_R14,
156 	VCPU_REGS_R15 = __VCPU_REGS_R15,
157 #endif
158 	VCPU_REGS_RIP,
159 	NR_VCPU_REGS
160 };
161 
162 enum kvm_reg_ex {
163 	VCPU_EXREG_PDPTR = NR_VCPU_REGS,
164 	VCPU_EXREG_CR3,
165 	VCPU_EXREG_RFLAGS,
166 	VCPU_EXREG_SEGMENTS,
167 };
168 
169 enum {
170 	VCPU_SREG_ES,
171 	VCPU_SREG_CS,
172 	VCPU_SREG_SS,
173 	VCPU_SREG_DS,
174 	VCPU_SREG_FS,
175 	VCPU_SREG_GS,
176 	VCPU_SREG_TR,
177 	VCPU_SREG_LDTR,
178 };
179 
180 #include <asm/kvm_emulate.h>
181 
182 #define KVM_NR_MEM_OBJS 40
183 
184 #define KVM_NR_DB_REGS	4
185 
186 #define DR6_BD		(1 << 13)
187 #define DR6_BS		(1 << 14)
188 #define DR6_BT		(1 << 15)
189 #define DR6_RTM		(1 << 16)
190 #define DR6_FIXED_1	0xfffe0ff0
191 #define DR6_INIT	0xffff0ff0
192 #define DR6_VOLATILE	0x0001e00f
193 
194 #define DR7_BP_EN_MASK	0x000000ff
195 #define DR7_GE		(1 << 9)
196 #define DR7_GD		(1 << 13)
197 #define DR7_FIXED_1	0x00000400
198 #define DR7_VOLATILE	0xffff2bff
199 
200 #define PFERR_PRESENT_BIT 0
201 #define PFERR_WRITE_BIT 1
202 #define PFERR_USER_BIT 2
203 #define PFERR_RSVD_BIT 3
204 #define PFERR_FETCH_BIT 4
205 #define PFERR_PK_BIT 5
206 #define PFERR_GUEST_FINAL_BIT 32
207 #define PFERR_GUEST_PAGE_BIT 33
208 
209 #define PFERR_PRESENT_MASK (1U << PFERR_PRESENT_BIT)
210 #define PFERR_WRITE_MASK (1U << PFERR_WRITE_BIT)
211 #define PFERR_USER_MASK (1U << PFERR_USER_BIT)
212 #define PFERR_RSVD_MASK (1U << PFERR_RSVD_BIT)
213 #define PFERR_FETCH_MASK (1U << PFERR_FETCH_BIT)
214 #define PFERR_PK_MASK (1U << PFERR_PK_BIT)
215 #define PFERR_GUEST_FINAL_MASK (1ULL << PFERR_GUEST_FINAL_BIT)
216 #define PFERR_GUEST_PAGE_MASK (1ULL << PFERR_GUEST_PAGE_BIT)
217 
218 #define PFERR_NESTED_GUEST_PAGE (PFERR_GUEST_PAGE_MASK |	\
219 				 PFERR_WRITE_MASK |		\
220 				 PFERR_PRESENT_MASK)
221 
222 /* apic attention bits */
223 #define KVM_APIC_CHECK_VAPIC	0
224 /*
225  * The following bit is set with PV-EOI, unset on EOI.
226  * We detect PV-EOI changes by guest by comparing
227  * this bit with PV-EOI in guest memory.
228  * See the implementation in apic_update_pv_eoi.
229  */
230 #define KVM_APIC_PV_EOI_PENDING	1
231 
232 struct kvm_kernel_irq_routing_entry;
233 
234 /*
235  * We don't want allocation failures within the mmu code, so we preallocate
236  * enough memory for a single page fault in a cache.
237  */
238 struct kvm_mmu_memory_cache {
239 	int nobjs;
240 	void *objects[KVM_NR_MEM_OBJS];
241 };
242 
243 /*
244  * the pages used as guest page table on soft mmu are tracked by
245  * kvm_memory_slot.arch.gfn_track which is 16 bits, so the role bits used
246  * by indirect shadow page can not be more than 15 bits.
247  *
248  * Currently, we used 14 bits that are @level, @gpte_is_8_bytes, @quadrant, @access,
249  * @nxe, @cr0_wp, @smep_andnot_wp and @smap_andnot_wp.
250  */
251 union kvm_mmu_page_role {
252 	u32 word;
253 	struct {
254 		unsigned level:4;
255 		unsigned gpte_is_8_bytes:1;
256 		unsigned quadrant:2;
257 		unsigned direct:1;
258 		unsigned access:3;
259 		unsigned invalid:1;
260 		unsigned nxe:1;
261 		unsigned cr0_wp:1;
262 		unsigned smep_andnot_wp:1;
263 		unsigned smap_andnot_wp:1;
264 		unsigned ad_disabled:1;
265 		unsigned guest_mode:1;
266 		unsigned :6;
267 
268 		/*
269 		 * This is left at the top of the word so that
270 		 * kvm_memslots_for_spte_role can extract it with a
271 		 * simple shift.  While there is room, give it a whole
272 		 * byte so it is also faster to load it from memory.
273 		 */
274 		unsigned smm:8;
275 	};
276 };
277 
278 union kvm_mmu_extended_role {
279 /*
280  * This structure complements kvm_mmu_page_role caching everything needed for
281  * MMU configuration. If nothing in both these structures changed, MMU
282  * re-configuration can be skipped. @valid bit is set on first usage so we don't
283  * treat all-zero structure as valid data.
284  */
285 	u32 word;
286 	struct {
287 		unsigned int valid:1;
288 		unsigned int execonly:1;
289 		unsigned int cr0_pg:1;
290 		unsigned int cr4_pae:1;
291 		unsigned int cr4_pse:1;
292 		unsigned int cr4_pke:1;
293 		unsigned int cr4_smap:1;
294 		unsigned int cr4_smep:1;
295 		unsigned int cr4_la57:1;
296 		unsigned int maxphyaddr:6;
297 	};
298 };
299 
300 union kvm_mmu_role {
301 	u64 as_u64;
302 	struct {
303 		union kvm_mmu_page_role base;
304 		union kvm_mmu_extended_role ext;
305 	};
306 };
307 
308 struct kvm_rmap_head {
309 	unsigned long val;
310 };
311 
312 struct kvm_mmu_page {
313 	struct list_head link;
314 	struct hlist_node hash_link;
315 	bool unsync;
316 	u8 mmu_valid_gen;
317 	bool mmio_cached;
318 
319 	/*
320 	 * The following two entries are used to key the shadow page in the
321 	 * hash table.
322 	 */
323 	union kvm_mmu_page_role role;
324 	gfn_t gfn;
325 
326 	u64 *spt;
327 	/* hold the gfn of each spte inside spt */
328 	gfn_t *gfns;
329 	int root_count;          /* Currently serving as active root */
330 	unsigned int unsync_children;
331 	struct kvm_rmap_head parent_ptes; /* rmap pointers to parent sptes */
332 	DECLARE_BITMAP(unsync_child_bitmap, 512);
333 
334 #ifdef CONFIG_X86_32
335 	/*
336 	 * Used out of the mmu-lock to avoid reading spte values while an
337 	 * update is in progress; see the comments in __get_spte_lockless().
338 	 */
339 	int clear_spte_count;
340 #endif
341 
342 	/* Number of writes since the last time traversal visited this page.  */
343 	atomic_t write_flooding_count;
344 };
345 
346 struct kvm_pio_request {
347 	unsigned long linear_rip;
348 	unsigned long count;
349 	int in;
350 	int port;
351 	int size;
352 };
353 
354 #define PT64_ROOT_MAX_LEVEL 5
355 
356 struct rsvd_bits_validate {
357 	u64 rsvd_bits_mask[2][PT64_ROOT_MAX_LEVEL];
358 	u64 bad_mt_xwr;
359 };
360 
361 struct kvm_mmu_root_info {
362 	gpa_t cr3;
363 	hpa_t hpa;
364 };
365 
366 #define KVM_MMU_ROOT_INFO_INVALID \
367 	((struct kvm_mmu_root_info) { .cr3 = INVALID_PAGE, .hpa = INVALID_PAGE })
368 
369 #define KVM_MMU_NUM_PREV_ROOTS 3
370 
371 /*
372  * x86 supports 4 paging modes (5-level 64-bit, 4-level 64-bit, 3-level 32-bit,
373  * and 2-level 32-bit).  The kvm_mmu structure abstracts the details of the
374  * current mmu mode.
375  */
376 struct kvm_mmu {
377 	void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
378 	unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
379 	u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
380 	int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
381 			  bool prefault);
382 	void (*inject_page_fault)(struct kvm_vcpu *vcpu,
383 				  struct x86_exception *fault);
384 	gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
385 			    struct x86_exception *exception);
386 	gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
387 			       struct x86_exception *exception);
388 	int (*sync_page)(struct kvm_vcpu *vcpu,
389 			 struct kvm_mmu_page *sp);
390 	void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva, hpa_t root_hpa);
391 	void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
392 			   u64 *spte, const void *pte);
393 	hpa_t root_hpa;
394 	gpa_t root_cr3;
395 	union kvm_mmu_role mmu_role;
396 	u8 root_level;
397 	u8 shadow_root_level;
398 	u8 ept_ad;
399 	bool direct_map;
400 	struct kvm_mmu_root_info prev_roots[KVM_MMU_NUM_PREV_ROOTS];
401 
402 	/*
403 	 * Bitmap; bit set = permission fault
404 	 * Byte index: page fault error code [4:1]
405 	 * Bit index: pte permissions in ACC_* format
406 	 */
407 	u8 permissions[16];
408 
409 	/*
410 	* The pkru_mask indicates if protection key checks are needed.  It
411 	* consists of 16 domains indexed by page fault error code bits [4:1],
412 	* with PFEC.RSVD replaced by ACC_USER_MASK from the page tables.
413 	* Each domain has 2 bits which are ANDed with AD and WD from PKRU.
414 	*/
415 	u32 pkru_mask;
416 
417 	u64 *pae_root;
418 	u64 *lm_root;
419 
420 	/*
421 	 * check zero bits on shadow page table entries, these
422 	 * bits include not only hardware reserved bits but also
423 	 * the bits spte never used.
424 	 */
425 	struct rsvd_bits_validate shadow_zero_check;
426 
427 	struct rsvd_bits_validate guest_rsvd_check;
428 
429 	/* Can have large pages at levels 2..last_nonleaf_level-1. */
430 	u8 last_nonleaf_level;
431 
432 	bool nx;
433 
434 	u64 pdptrs[4]; /* pae */
435 };
436 
437 struct kvm_tlb_range {
438 	u64 start_gfn;
439 	u64 pages;
440 };
441 
442 enum pmc_type {
443 	KVM_PMC_GP = 0,
444 	KVM_PMC_FIXED,
445 };
446 
447 struct kvm_pmc {
448 	enum pmc_type type;
449 	u8 idx;
450 	u64 counter;
451 	u64 eventsel;
452 	struct perf_event *perf_event;
453 	struct kvm_vcpu *vcpu;
454 };
455 
456 struct kvm_pmu {
457 	unsigned nr_arch_gp_counters;
458 	unsigned nr_arch_fixed_counters;
459 	unsigned available_event_types;
460 	u64 fixed_ctr_ctrl;
461 	u64 global_ctrl;
462 	u64 global_status;
463 	u64 global_ovf_ctrl;
464 	u64 counter_bitmask[2];
465 	u64 global_ctrl_mask;
466 	u64 global_ovf_ctrl_mask;
467 	u64 reserved_bits;
468 	u8 version;
469 	struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
470 	struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
471 	struct irq_work irq_work;
472 	u64 reprogram_pmi;
473 };
474 
475 struct kvm_pmu_ops;
476 
477 enum {
478 	KVM_DEBUGREG_BP_ENABLED = 1,
479 	KVM_DEBUGREG_WONT_EXIT = 2,
480 	KVM_DEBUGREG_RELOAD = 4,
481 };
482 
483 struct kvm_mtrr_range {
484 	u64 base;
485 	u64 mask;
486 	struct list_head node;
487 };
488 
489 struct kvm_mtrr {
490 	struct kvm_mtrr_range var_ranges[KVM_NR_VAR_MTRR];
491 	mtrr_type fixed_ranges[KVM_NR_FIXED_MTRR_REGION];
492 	u64 deftype;
493 
494 	struct list_head head;
495 };
496 
497 /* Hyper-V SynIC timer */
498 struct kvm_vcpu_hv_stimer {
499 	struct hrtimer timer;
500 	int index;
501 	union hv_stimer_config config;
502 	u64 count;
503 	u64 exp_time;
504 	struct hv_message msg;
505 	bool msg_pending;
506 };
507 
508 /* Hyper-V synthetic interrupt controller (SynIC)*/
509 struct kvm_vcpu_hv_synic {
510 	u64 version;
511 	u64 control;
512 	u64 msg_page;
513 	u64 evt_page;
514 	atomic64_t sint[HV_SYNIC_SINT_COUNT];
515 	atomic_t sint_to_gsi[HV_SYNIC_SINT_COUNT];
516 	DECLARE_BITMAP(auto_eoi_bitmap, 256);
517 	DECLARE_BITMAP(vec_bitmap, 256);
518 	bool active;
519 	bool dont_zero_synic_pages;
520 };
521 
522 /* Hyper-V per vcpu emulation context */
523 struct kvm_vcpu_hv {
524 	u32 vp_index;
525 	u64 hv_vapic;
526 	s64 runtime_offset;
527 	struct kvm_vcpu_hv_synic synic;
528 	struct kvm_hyperv_exit exit;
529 	struct kvm_vcpu_hv_stimer stimer[HV_SYNIC_STIMER_COUNT];
530 	DECLARE_BITMAP(stimer_pending_bitmap, HV_SYNIC_STIMER_COUNT);
531 	cpumask_t tlb_flush;
532 };
533 
534 struct kvm_vcpu_arch {
535 	/*
536 	 * rip and regs accesses must go through
537 	 * kvm_{register,rip}_{read,write} functions.
538 	 */
539 	unsigned long regs[NR_VCPU_REGS];
540 	u32 regs_avail;
541 	u32 regs_dirty;
542 
543 	unsigned long cr0;
544 	unsigned long cr0_guest_owned_bits;
545 	unsigned long cr2;
546 	unsigned long cr3;
547 	unsigned long cr4;
548 	unsigned long cr4_guest_owned_bits;
549 	unsigned long cr8;
550 	u32 pkru;
551 	u32 hflags;
552 	u64 efer;
553 	u64 apic_base;
554 	struct kvm_lapic *apic;    /* kernel irqchip context */
555 	bool apicv_active;
556 	bool load_eoi_exitmap_pending;
557 	DECLARE_BITMAP(ioapic_handled_vectors, 256);
558 	unsigned long apic_attention;
559 	int32_t apic_arb_prio;
560 	int mp_state;
561 	u64 ia32_misc_enable_msr;
562 	u64 smbase;
563 	u64 smi_count;
564 	bool tpr_access_reporting;
565 	u64 ia32_xss;
566 	u64 microcode_version;
567 	u64 arch_capabilities;
568 
569 	/*
570 	 * Paging state of the vcpu
571 	 *
572 	 * If the vcpu runs in guest mode with two level paging this still saves
573 	 * the paging mode of the l1 guest. This context is always used to
574 	 * handle faults.
575 	 */
576 	struct kvm_mmu *mmu;
577 
578 	/* Non-nested MMU for L1 */
579 	struct kvm_mmu root_mmu;
580 
581 	/* L1 MMU when running nested */
582 	struct kvm_mmu guest_mmu;
583 
584 	/*
585 	 * Paging state of an L2 guest (used for nested npt)
586 	 *
587 	 * This context will save all necessary information to walk page tables
588 	 * of the an L2 guest. This context is only initialized for page table
589 	 * walking and not for faulting since we never handle l2 page faults on
590 	 * the host.
591 	 */
592 	struct kvm_mmu nested_mmu;
593 
594 	/*
595 	 * Pointer to the mmu context currently used for
596 	 * gva_to_gpa translations.
597 	 */
598 	struct kvm_mmu *walk_mmu;
599 
600 	struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
601 	struct kvm_mmu_memory_cache mmu_page_cache;
602 	struct kvm_mmu_memory_cache mmu_page_header_cache;
603 
604 	/*
605 	 * QEMU userspace and the guest each have their own FPU state.
606 	 * In vcpu_run, we switch between the user and guest FPU contexts.
607 	 * While running a VCPU, the VCPU thread will have the guest FPU
608 	 * context.
609 	 *
610 	 * Note that while the PKRU state lives inside the fpu registers,
611 	 * it is switched out separately at VMENTER and VMEXIT time. The
612 	 * "guest_fpu" state here contains the guest FPU context, with the
613 	 * host PRKU bits.
614 	 */
615 	struct fpu *user_fpu;
616 	struct fpu *guest_fpu;
617 
618 	u64 xcr0;
619 	u64 guest_supported_xcr0;
620 	u32 guest_xstate_size;
621 
622 	struct kvm_pio_request pio;
623 	void *pio_data;
624 
625 	u8 event_exit_inst_len;
626 
627 	struct kvm_queued_exception {
628 		bool pending;
629 		bool injected;
630 		bool has_error_code;
631 		u8 nr;
632 		u32 error_code;
633 		unsigned long payload;
634 		bool has_payload;
635 		u8 nested_apf;
636 	} exception;
637 
638 	struct kvm_queued_interrupt {
639 		bool injected;
640 		bool soft;
641 		u8 nr;
642 	} interrupt;
643 
644 	int halt_request; /* real mode on Intel only */
645 
646 	int cpuid_nent;
647 	struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
648 
649 	int maxphyaddr;
650 
651 	/* emulate context */
652 
653 	struct x86_emulate_ctxt emulate_ctxt;
654 	bool emulate_regs_need_sync_to_vcpu;
655 	bool emulate_regs_need_sync_from_vcpu;
656 	int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
657 
658 	gpa_t time;
659 	struct pvclock_vcpu_time_info hv_clock;
660 	unsigned int hw_tsc_khz;
661 	struct gfn_to_hva_cache pv_time;
662 	bool pv_time_enabled;
663 	/* set guest stopped flag in pvclock flags field */
664 	bool pvclock_set_guest_stopped_request;
665 
666 	struct {
667 		u64 msr_val;
668 		u64 last_steal;
669 		struct gfn_to_hva_cache stime;
670 		struct kvm_steal_time steal;
671 	} st;
672 
673 	u64 tsc_offset;
674 	u64 last_guest_tsc;
675 	u64 last_host_tsc;
676 	u64 tsc_offset_adjustment;
677 	u64 this_tsc_nsec;
678 	u64 this_tsc_write;
679 	u64 this_tsc_generation;
680 	bool tsc_catchup;
681 	bool tsc_always_catchup;
682 	s8 virtual_tsc_shift;
683 	u32 virtual_tsc_mult;
684 	u32 virtual_tsc_khz;
685 	s64 ia32_tsc_adjust_msr;
686 	u64 msr_ia32_power_ctl;
687 	u64 tsc_scaling_ratio;
688 
689 	atomic_t nmi_queued;  /* unprocessed asynchronous NMIs */
690 	unsigned nmi_pending; /* NMI queued after currently running handler */
691 	bool nmi_injected;    /* Trying to inject an NMI this entry */
692 	bool smi_pending;    /* SMI queued after currently running handler */
693 
694 	struct kvm_mtrr mtrr_state;
695 	u64 pat;
696 
697 	unsigned switch_db_regs;
698 	unsigned long db[KVM_NR_DB_REGS];
699 	unsigned long dr6;
700 	unsigned long dr7;
701 	unsigned long eff_db[KVM_NR_DB_REGS];
702 	unsigned long guest_debug_dr7;
703 	u64 msr_platform_info;
704 	u64 msr_misc_features_enables;
705 
706 	u64 mcg_cap;
707 	u64 mcg_status;
708 	u64 mcg_ctl;
709 	u64 mcg_ext_ctl;
710 	u64 *mce_banks;
711 
712 	/* Cache MMIO info */
713 	u64 mmio_gva;
714 	unsigned mmio_access;
715 	gfn_t mmio_gfn;
716 	u64 mmio_gen;
717 
718 	struct kvm_pmu pmu;
719 
720 	/* used for guest single stepping over the given code position */
721 	unsigned long singlestep_rip;
722 
723 	struct kvm_vcpu_hv hyperv;
724 
725 	cpumask_var_t wbinvd_dirty_mask;
726 
727 	unsigned long last_retry_eip;
728 	unsigned long last_retry_addr;
729 
730 	struct {
731 		bool halted;
732 		gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
733 		struct gfn_to_hva_cache data;
734 		u64 msr_val;
735 		u32 id;
736 		bool send_user_only;
737 		u32 host_apf_reason;
738 		unsigned long nested_apf_token;
739 		bool delivery_as_pf_vmexit;
740 	} apf;
741 
742 	/* OSVW MSRs (AMD only) */
743 	struct {
744 		u64 length;
745 		u64 status;
746 	} osvw;
747 
748 	struct {
749 		u64 msr_val;
750 		struct gfn_to_hva_cache data;
751 	} pv_eoi;
752 
753 	u64 msr_kvm_poll_control;
754 
755 	/*
756 	 * Indicate whether the access faults on its page table in guest
757 	 * which is set when fix page fault and used to detect unhandeable
758 	 * instruction.
759 	 */
760 	bool write_fault_to_shadow_pgtable;
761 
762 	/* set at EPT violation at this point */
763 	unsigned long exit_qualification;
764 
765 	/* pv related host specific info */
766 	struct {
767 		bool pv_unhalted;
768 	} pv;
769 
770 	int pending_ioapic_eoi;
771 	int pending_external_vector;
772 
773 	/* GPA available */
774 	bool gpa_available;
775 	gpa_t gpa_val;
776 
777 	/* be preempted when it's in kernel-mode(cpl=0) */
778 	bool preempted_in_kernel;
779 
780 	/* Flush the L1 Data cache for L1TF mitigation on VMENTER */
781 	bool l1tf_flush_l1d;
782 
783 	/* AMD MSRC001_0015 Hardware Configuration */
784 	u64 msr_hwcr;
785 };
786 
787 struct kvm_lpage_info {
788 	int disallow_lpage;
789 };
790 
791 struct kvm_arch_memory_slot {
792 	struct kvm_rmap_head *rmap[KVM_NR_PAGE_SIZES];
793 	struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
794 	unsigned short *gfn_track[KVM_PAGE_TRACK_MAX];
795 };
796 
797 /*
798  * We use as the mode the number of bits allocated in the LDR for the
799  * logical processor ID.  It happens that these are all powers of two.
800  * This makes it is very easy to detect cases where the APICs are
801  * configured for multiple modes; in that case, we cannot use the map and
802  * hence cannot use kvm_irq_delivery_to_apic_fast either.
803  */
804 #define KVM_APIC_MODE_XAPIC_CLUSTER          4
805 #define KVM_APIC_MODE_XAPIC_FLAT             8
806 #define KVM_APIC_MODE_X2APIC                16
807 
808 struct kvm_apic_map {
809 	struct rcu_head rcu;
810 	u8 mode;
811 	u32 max_apic_id;
812 	union {
813 		struct kvm_lapic *xapic_flat_map[8];
814 		struct kvm_lapic *xapic_cluster_map[16][4];
815 	};
816 	struct kvm_lapic *phys_map[];
817 };
818 
819 /* Hyper-V emulation context */
820 struct kvm_hv {
821 	struct mutex hv_lock;
822 	u64 hv_guest_os_id;
823 	u64 hv_hypercall;
824 	u64 hv_tsc_page;
825 
826 	/* Hyper-v based guest crash (NT kernel bugcheck) parameters */
827 	u64 hv_crash_param[HV_X64_MSR_CRASH_PARAMS];
828 	u64 hv_crash_ctl;
829 
830 	HV_REFERENCE_TSC_PAGE tsc_ref;
831 
832 	struct idr conn_to_evt;
833 
834 	u64 hv_reenlightenment_control;
835 	u64 hv_tsc_emulation_control;
836 	u64 hv_tsc_emulation_status;
837 
838 	/* How many vCPUs have VP index != vCPU index */
839 	atomic_t num_mismatched_vp_indexes;
840 
841 	struct hv_partition_assist_pg *hv_pa_pg;
842 };
843 
844 enum kvm_irqchip_mode {
845 	KVM_IRQCHIP_NONE,
846 	KVM_IRQCHIP_KERNEL,       /* created with KVM_CREATE_IRQCHIP */
847 	KVM_IRQCHIP_SPLIT,        /* created with KVM_CAP_SPLIT_IRQCHIP */
848 };
849 
850 struct kvm_arch {
851 	unsigned long n_used_mmu_pages;
852 	unsigned long n_requested_mmu_pages;
853 	unsigned long n_max_mmu_pages;
854 	unsigned int indirect_shadow_pages;
855 	u8 mmu_valid_gen;
856 	struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
857 	/*
858 	 * Hash table of struct kvm_mmu_page.
859 	 */
860 	struct list_head active_mmu_pages;
861 	struct list_head zapped_obsolete_pages;
862 	struct kvm_page_track_notifier_node mmu_sp_tracker;
863 	struct kvm_page_track_notifier_head track_notifier_head;
864 
865 	struct list_head assigned_dev_head;
866 	struct iommu_domain *iommu_domain;
867 	bool iommu_noncoherent;
868 #define __KVM_HAVE_ARCH_NONCOHERENT_DMA
869 	atomic_t noncoherent_dma_count;
870 #define __KVM_HAVE_ARCH_ASSIGNED_DEVICE
871 	atomic_t assigned_device_count;
872 	struct kvm_pic *vpic;
873 	struct kvm_ioapic *vioapic;
874 	struct kvm_pit *vpit;
875 	atomic_t vapics_in_nmi_mode;
876 	struct mutex apic_map_lock;
877 	struct kvm_apic_map *apic_map;
878 
879 	bool apic_access_page_done;
880 
881 	gpa_t wall_clock;
882 
883 	bool mwait_in_guest;
884 	bool hlt_in_guest;
885 	bool pause_in_guest;
886 	bool cstate_in_guest;
887 
888 	unsigned long irq_sources_bitmap;
889 	s64 kvmclock_offset;
890 	raw_spinlock_t tsc_write_lock;
891 	u64 last_tsc_nsec;
892 	u64 last_tsc_write;
893 	u32 last_tsc_khz;
894 	u64 cur_tsc_nsec;
895 	u64 cur_tsc_write;
896 	u64 cur_tsc_offset;
897 	u64 cur_tsc_generation;
898 	int nr_vcpus_matched_tsc;
899 
900 	spinlock_t pvclock_gtod_sync_lock;
901 	bool use_master_clock;
902 	u64 master_kernel_ns;
903 	u64 master_cycle_now;
904 	struct delayed_work kvmclock_update_work;
905 	struct delayed_work kvmclock_sync_work;
906 
907 	struct kvm_xen_hvm_config xen_hvm_config;
908 
909 	/* reads protected by irq_srcu, writes by irq_lock */
910 	struct hlist_head mask_notifier_list;
911 
912 	struct kvm_hv hyperv;
913 
914 	#ifdef CONFIG_KVM_MMU_AUDIT
915 	int audit_point;
916 	#endif
917 
918 	bool backwards_tsc_observed;
919 	bool boot_vcpu_runs_old_kvmclock;
920 	u32 bsp_vcpu_id;
921 
922 	u64 disabled_quirks;
923 
924 	enum kvm_irqchip_mode irqchip_mode;
925 	u8 nr_reserved_ioapic_pins;
926 
927 	bool disabled_lapic_found;
928 
929 	bool x2apic_format;
930 	bool x2apic_broadcast_quirk_disabled;
931 
932 	bool guest_can_read_msr_platform_info;
933 	bool exception_payload_enabled;
934 
935 	struct kvm_pmu_event_filter *pmu_event_filter;
936 };
937 
938 struct kvm_vm_stat {
939 	ulong mmu_shadow_zapped;
940 	ulong mmu_pte_write;
941 	ulong mmu_pte_updated;
942 	ulong mmu_pde_zapped;
943 	ulong mmu_flooded;
944 	ulong mmu_recycled;
945 	ulong mmu_cache_miss;
946 	ulong mmu_unsync;
947 	ulong remote_tlb_flush;
948 	ulong lpages;
949 	ulong max_mmu_page_hash_collisions;
950 };
951 
952 struct kvm_vcpu_stat {
953 	u64 pf_fixed;
954 	u64 pf_guest;
955 	u64 tlb_flush;
956 	u64 invlpg;
957 
958 	u64 exits;
959 	u64 io_exits;
960 	u64 mmio_exits;
961 	u64 signal_exits;
962 	u64 irq_window_exits;
963 	u64 nmi_window_exits;
964 	u64 l1d_flush;
965 	u64 halt_exits;
966 	u64 halt_successful_poll;
967 	u64 halt_attempted_poll;
968 	u64 halt_poll_invalid;
969 	u64 halt_wakeup;
970 	u64 request_irq_exits;
971 	u64 irq_exits;
972 	u64 host_state_reload;
973 	u64 fpu_reload;
974 	u64 insn_emulation;
975 	u64 insn_emulation_fail;
976 	u64 hypercalls;
977 	u64 irq_injections;
978 	u64 nmi_injections;
979 	u64 req_event;
980 };
981 
982 struct x86_instruction_info;
983 
984 struct msr_data {
985 	bool host_initiated;
986 	u32 index;
987 	u64 data;
988 };
989 
990 struct kvm_lapic_irq {
991 	u32 vector;
992 	u16 delivery_mode;
993 	u16 dest_mode;
994 	bool level;
995 	u16 trig_mode;
996 	u32 shorthand;
997 	u32 dest_id;
998 	bool msi_redir_hint;
999 };
1000 
1001 struct kvm_x86_ops {
1002 	int (*cpu_has_kvm_support)(void);          /* __init */
1003 	int (*disabled_by_bios)(void);             /* __init */
1004 	int (*hardware_enable)(void);
1005 	void (*hardware_disable)(void);
1006 	int (*check_processor_compatibility)(void);/* __init */
1007 	int (*hardware_setup)(void);               /* __init */
1008 	void (*hardware_unsetup)(void);            /* __exit */
1009 	bool (*cpu_has_accelerated_tpr)(void);
1010 	bool (*has_emulated_msr)(int index);
1011 	void (*cpuid_update)(struct kvm_vcpu *vcpu);
1012 
1013 	struct kvm *(*vm_alloc)(void);
1014 	void (*vm_free)(struct kvm *);
1015 	int (*vm_init)(struct kvm *kvm);
1016 	void (*vm_destroy)(struct kvm *kvm);
1017 
1018 	/* Create, but do not attach this VCPU */
1019 	struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
1020 	void (*vcpu_free)(struct kvm_vcpu *vcpu);
1021 	void (*vcpu_reset)(struct kvm_vcpu *vcpu, bool init_event);
1022 
1023 	void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
1024 	void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
1025 	void (*vcpu_put)(struct kvm_vcpu *vcpu);
1026 
1027 	void (*update_bp_intercept)(struct kvm_vcpu *vcpu);
1028 	int (*get_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
1029 	int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
1030 	u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
1031 	void (*get_segment)(struct kvm_vcpu *vcpu,
1032 			    struct kvm_segment *var, int seg);
1033 	int (*get_cpl)(struct kvm_vcpu *vcpu);
1034 	void (*set_segment)(struct kvm_vcpu *vcpu,
1035 			    struct kvm_segment *var, int seg);
1036 	void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
1037 	void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
1038 	void (*decache_cr3)(struct kvm_vcpu *vcpu);
1039 	void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
1040 	void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
1041 	void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
1042 	int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
1043 	void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
1044 	void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
1045 	void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
1046 	void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
1047 	void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
1048 	u64 (*get_dr6)(struct kvm_vcpu *vcpu);
1049 	void (*set_dr6)(struct kvm_vcpu *vcpu, unsigned long value);
1050 	void (*sync_dirty_debug_regs)(struct kvm_vcpu *vcpu);
1051 	void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
1052 	void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
1053 	unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
1054 	void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
1055 
1056 	void (*tlb_flush)(struct kvm_vcpu *vcpu, bool invalidate_gpa);
1057 	int  (*tlb_remote_flush)(struct kvm *kvm);
1058 	int  (*tlb_remote_flush_with_range)(struct kvm *kvm,
1059 			struct kvm_tlb_range *range);
1060 
1061 	/*
1062 	 * Flush any TLB entries associated with the given GVA.
1063 	 * Does not need to flush GPA->HPA mappings.
1064 	 * Can potentially get non-canonical addresses through INVLPGs, which
1065 	 * the implementation may choose to ignore if appropriate.
1066 	 */
1067 	void (*tlb_flush_gva)(struct kvm_vcpu *vcpu, gva_t addr);
1068 
1069 	void (*run)(struct kvm_vcpu *vcpu);
1070 	int (*handle_exit)(struct kvm_vcpu *vcpu);
1071 	int (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
1072 	void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
1073 	u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu);
1074 	void (*patch_hypercall)(struct kvm_vcpu *vcpu,
1075 				unsigned char *hypercall_addr);
1076 	void (*set_irq)(struct kvm_vcpu *vcpu);
1077 	void (*set_nmi)(struct kvm_vcpu *vcpu);
1078 	void (*queue_exception)(struct kvm_vcpu *vcpu);
1079 	void (*cancel_injection)(struct kvm_vcpu *vcpu);
1080 	int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
1081 	int (*nmi_allowed)(struct kvm_vcpu *vcpu);
1082 	bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
1083 	void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
1084 	void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
1085 	void (*enable_irq_window)(struct kvm_vcpu *vcpu);
1086 	void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
1087 	bool (*get_enable_apicv)(struct kvm_vcpu *vcpu);
1088 	void (*refresh_apicv_exec_ctrl)(struct kvm_vcpu *vcpu);
1089 	void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
1090 	void (*hwapic_isr_update)(struct kvm_vcpu *vcpu, int isr);
1091 	bool (*guest_apic_has_interrupt)(struct kvm_vcpu *vcpu);
1092 	void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap);
1093 	void (*set_virtual_apic_mode)(struct kvm_vcpu *vcpu);
1094 	void (*set_apic_access_page_addr)(struct kvm_vcpu *vcpu, hpa_t hpa);
1095 	void (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
1096 	int (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
1097 	int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
1098 	int (*set_identity_map_addr)(struct kvm *kvm, u64 ident_addr);
1099 	int (*get_tdp_level)(struct kvm_vcpu *vcpu);
1100 	u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
1101 	int (*get_lpage_level)(void);
1102 	bool (*rdtscp_supported)(void);
1103 	bool (*invpcid_supported)(void);
1104 
1105 	void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
1106 
1107 	void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
1108 
1109 	bool (*has_wbinvd_exit)(void);
1110 
1111 	u64 (*read_l1_tsc_offset)(struct kvm_vcpu *vcpu);
1112 	/* Returns actual tsc_offset set in active VMCS */
1113 	u64 (*write_l1_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
1114 
1115 	void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
1116 
1117 	int (*check_intercept)(struct kvm_vcpu *vcpu,
1118 			       struct x86_instruction_info *info,
1119 			       enum x86_intercept_stage stage);
1120 	void (*handle_exit_irqoff)(struct kvm_vcpu *vcpu);
1121 	bool (*mpx_supported)(void);
1122 	bool (*xsaves_supported)(void);
1123 	bool (*umip_emulated)(void);
1124 	bool (*pt_supported)(void);
1125 
1126 	int (*check_nested_events)(struct kvm_vcpu *vcpu, bool external_intr);
1127 	void (*request_immediate_exit)(struct kvm_vcpu *vcpu);
1128 
1129 	void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
1130 
1131 	/*
1132 	 * Arch-specific dirty logging hooks. These hooks are only supposed to
1133 	 * be valid if the specific arch has hardware-accelerated dirty logging
1134 	 * mechanism. Currently only for PML on VMX.
1135 	 *
1136 	 *  - slot_enable_log_dirty:
1137 	 *	called when enabling log dirty mode for the slot.
1138 	 *  - slot_disable_log_dirty:
1139 	 *	called when disabling log dirty mode for the slot.
1140 	 *	also called when slot is created with log dirty disabled.
1141 	 *  - flush_log_dirty:
1142 	 *	called before reporting dirty_bitmap to userspace.
1143 	 *  - enable_log_dirty_pt_masked:
1144 	 *	called when reenabling log dirty for the GFNs in the mask after
1145 	 *	corresponding bits are cleared in slot->dirty_bitmap.
1146 	 */
1147 	void (*slot_enable_log_dirty)(struct kvm *kvm,
1148 				      struct kvm_memory_slot *slot);
1149 	void (*slot_disable_log_dirty)(struct kvm *kvm,
1150 				       struct kvm_memory_slot *slot);
1151 	void (*flush_log_dirty)(struct kvm *kvm);
1152 	void (*enable_log_dirty_pt_masked)(struct kvm *kvm,
1153 					   struct kvm_memory_slot *slot,
1154 					   gfn_t offset, unsigned long mask);
1155 	int (*write_log_dirty)(struct kvm_vcpu *vcpu);
1156 
1157 	/* pmu operations of sub-arch */
1158 	const struct kvm_pmu_ops *pmu_ops;
1159 
1160 	/*
1161 	 * Architecture specific hooks for vCPU blocking due to
1162 	 * HLT instruction.
1163 	 * Returns for .pre_block():
1164 	 *    - 0 means continue to block the vCPU.
1165 	 *    - 1 means we cannot block the vCPU since some event
1166 	 *        happens during this period, such as, 'ON' bit in
1167 	 *        posted-interrupts descriptor is set.
1168 	 */
1169 	int (*pre_block)(struct kvm_vcpu *vcpu);
1170 	void (*post_block)(struct kvm_vcpu *vcpu);
1171 
1172 	void (*vcpu_blocking)(struct kvm_vcpu *vcpu);
1173 	void (*vcpu_unblocking)(struct kvm_vcpu *vcpu);
1174 
1175 	int (*update_pi_irte)(struct kvm *kvm, unsigned int host_irq,
1176 			      uint32_t guest_irq, bool set);
1177 	void (*apicv_post_state_restore)(struct kvm_vcpu *vcpu);
1178 	bool (*dy_apicv_has_pending_interrupt)(struct kvm_vcpu *vcpu);
1179 
1180 	int (*set_hv_timer)(struct kvm_vcpu *vcpu, u64 guest_deadline_tsc,
1181 			    bool *expired);
1182 	void (*cancel_hv_timer)(struct kvm_vcpu *vcpu);
1183 
1184 	void (*setup_mce)(struct kvm_vcpu *vcpu);
1185 
1186 	int (*get_nested_state)(struct kvm_vcpu *vcpu,
1187 				struct kvm_nested_state __user *user_kvm_nested_state,
1188 				unsigned user_data_size);
1189 	int (*set_nested_state)(struct kvm_vcpu *vcpu,
1190 				struct kvm_nested_state __user *user_kvm_nested_state,
1191 				struct kvm_nested_state *kvm_state);
1192 	bool (*get_vmcs12_pages)(struct kvm_vcpu *vcpu);
1193 
1194 	int (*smi_allowed)(struct kvm_vcpu *vcpu);
1195 	int (*pre_enter_smm)(struct kvm_vcpu *vcpu, char *smstate);
1196 	int (*pre_leave_smm)(struct kvm_vcpu *vcpu, const char *smstate);
1197 	int (*enable_smi_window)(struct kvm_vcpu *vcpu);
1198 
1199 	int (*mem_enc_op)(struct kvm *kvm, void __user *argp);
1200 	int (*mem_enc_reg_region)(struct kvm *kvm, struct kvm_enc_region *argp);
1201 	int (*mem_enc_unreg_region)(struct kvm *kvm, struct kvm_enc_region *argp);
1202 
1203 	int (*get_msr_feature)(struct kvm_msr_entry *entry);
1204 
1205 	int (*nested_enable_evmcs)(struct kvm_vcpu *vcpu,
1206 				   uint16_t *vmcs_version);
1207 	uint16_t (*nested_get_evmcs_version)(struct kvm_vcpu *vcpu);
1208 
1209 	bool (*need_emulation_on_page_fault)(struct kvm_vcpu *vcpu);
1210 
1211 	bool (*apic_init_signal_blocked)(struct kvm_vcpu *vcpu);
1212 	int (*enable_direct_tlbflush)(struct kvm_vcpu *vcpu);
1213 };
1214 
1215 struct kvm_arch_async_pf {
1216 	u32 token;
1217 	gfn_t gfn;
1218 	unsigned long cr3;
1219 	bool direct_map;
1220 };
1221 
1222 extern struct kvm_x86_ops *kvm_x86_ops;
1223 extern struct kmem_cache *x86_fpu_cache;
1224 
1225 #define __KVM_HAVE_ARCH_VM_ALLOC
1226 static inline struct kvm *kvm_arch_alloc_vm(void)
1227 {
1228 	return kvm_x86_ops->vm_alloc();
1229 }
1230 
1231 static inline void kvm_arch_free_vm(struct kvm *kvm)
1232 {
1233 	return kvm_x86_ops->vm_free(kvm);
1234 }
1235 
1236 #define __KVM_HAVE_ARCH_FLUSH_REMOTE_TLB
1237 static inline int kvm_arch_flush_remote_tlb(struct kvm *kvm)
1238 {
1239 	if (kvm_x86_ops->tlb_remote_flush &&
1240 	    !kvm_x86_ops->tlb_remote_flush(kvm))
1241 		return 0;
1242 	else
1243 		return -ENOTSUPP;
1244 }
1245 
1246 int kvm_mmu_module_init(void);
1247 void kvm_mmu_module_exit(void);
1248 
1249 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
1250 int kvm_mmu_create(struct kvm_vcpu *vcpu);
1251 void kvm_mmu_init_vm(struct kvm *kvm);
1252 void kvm_mmu_uninit_vm(struct kvm *kvm);
1253 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
1254 		u64 dirty_mask, u64 nx_mask, u64 x_mask, u64 p_mask,
1255 		u64 acc_track_mask, u64 me_mask);
1256 
1257 void kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
1258 void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
1259 				      struct kvm_memory_slot *memslot);
1260 void kvm_mmu_zap_collapsible_sptes(struct kvm *kvm,
1261 				   const struct kvm_memory_slot *memslot);
1262 void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
1263 				   struct kvm_memory_slot *memslot);
1264 void kvm_mmu_slot_largepage_remove_write_access(struct kvm *kvm,
1265 					struct kvm_memory_slot *memslot);
1266 void kvm_mmu_slot_set_dirty(struct kvm *kvm,
1267 			    struct kvm_memory_slot *memslot);
1268 void kvm_mmu_clear_dirty_pt_masked(struct kvm *kvm,
1269 				   struct kvm_memory_slot *slot,
1270 				   gfn_t gfn_offset, unsigned long mask);
1271 void kvm_mmu_zap_all(struct kvm *kvm);
1272 void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm, u64 gen);
1273 unsigned long kvm_mmu_calculate_default_mmu_pages(struct kvm *kvm);
1274 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned long kvm_nr_mmu_pages);
1275 
1276 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
1277 bool pdptrs_changed(struct kvm_vcpu *vcpu);
1278 
1279 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
1280 			  const void *val, int bytes);
1281 
1282 struct kvm_irq_mask_notifier {
1283 	void (*func)(struct kvm_irq_mask_notifier *kimn, bool masked);
1284 	int irq;
1285 	struct hlist_node link;
1286 };
1287 
1288 void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
1289 				    struct kvm_irq_mask_notifier *kimn);
1290 void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
1291 				      struct kvm_irq_mask_notifier *kimn);
1292 void kvm_fire_mask_notifiers(struct kvm *kvm, unsigned irqchip, unsigned pin,
1293 			     bool mask);
1294 
1295 extern bool tdp_enabled;
1296 
1297 u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
1298 
1299 /* control of guest tsc rate supported? */
1300 extern bool kvm_has_tsc_control;
1301 /* maximum supported tsc_khz for guests */
1302 extern u32  kvm_max_guest_tsc_khz;
1303 /* number of bits of the fractional part of the TSC scaling ratio */
1304 extern u8   kvm_tsc_scaling_ratio_frac_bits;
1305 /* maximum allowed value of TSC scaling ratio */
1306 extern u64  kvm_max_tsc_scaling_ratio;
1307 /* 1ull << kvm_tsc_scaling_ratio_frac_bits */
1308 extern u64  kvm_default_tsc_scaling_ratio;
1309 
1310 extern u64 kvm_mce_cap_supported;
1311 
1312 /*
1313  * EMULTYPE_NO_DECODE - Set when re-emulating an instruction (after completing
1314  *			userspace I/O) to indicate that the emulation context
1315  *			should be resued as is, i.e. skip initialization of
1316  *			emulation context, instruction fetch and decode.
1317  *
1318  * EMULTYPE_TRAP_UD - Set when emulating an intercepted #UD from hardware.
1319  *		      Indicates that only select instructions (tagged with
1320  *		      EmulateOnUD) should be emulated (to minimize the emulator
1321  *		      attack surface).  See also EMULTYPE_TRAP_UD_FORCED.
1322  *
1323  * EMULTYPE_SKIP - Set when emulating solely to skip an instruction, i.e. to
1324  *		   decode the instruction length.  For use *only* by
1325  *		   kvm_x86_ops->skip_emulated_instruction() implementations.
1326  *
1327  * EMULTYPE_ALLOW_RETRY - Set when the emulator should resume the guest to
1328  *			  retry native execution under certain conditions.
1329  *
1330  * EMULTYPE_TRAP_UD_FORCED - Set when emulating an intercepted #UD that was
1331  *			     triggered by KVM's magic "force emulation" prefix,
1332  *			     which is opt in via module param (off by default).
1333  *			     Bypasses EmulateOnUD restriction despite emulating
1334  *			     due to an intercepted #UD (see EMULTYPE_TRAP_UD).
1335  *			     Used to test the full emulator from userspace.
1336  *
1337  * EMULTYPE_VMWARE_GP - Set when emulating an intercepted #GP for VMware
1338  *			backdoor emulation, which is opt in via module param.
1339  *			VMware backoor emulation handles select instructions
1340  *			and reinjects the #GP for all other cases.
1341  */
1342 #define EMULTYPE_NO_DECODE	    (1 << 0)
1343 #define EMULTYPE_TRAP_UD	    (1 << 1)
1344 #define EMULTYPE_SKIP		    (1 << 2)
1345 #define EMULTYPE_ALLOW_RETRY	    (1 << 3)
1346 #define EMULTYPE_TRAP_UD_FORCED	    (1 << 4)
1347 #define EMULTYPE_VMWARE_GP	    (1 << 5)
1348 int kvm_emulate_instruction(struct kvm_vcpu *vcpu, int emulation_type);
1349 int kvm_emulate_instruction_from_buffer(struct kvm_vcpu *vcpu,
1350 					void *insn, int insn_len);
1351 
1352 void kvm_enable_efer_bits(u64);
1353 bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
1354 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 index, u64 *data);
1355 int kvm_set_msr(struct kvm_vcpu *vcpu, u32 index, u64 data);
1356 int kvm_emulate_rdmsr(struct kvm_vcpu *vcpu);
1357 int kvm_emulate_wrmsr(struct kvm_vcpu *vcpu);
1358 
1359 struct x86_emulate_ctxt;
1360 
1361 int kvm_fast_pio(struct kvm_vcpu *vcpu, int size, unsigned short port, int in);
1362 int kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
1363 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
1364 int kvm_vcpu_halt(struct kvm_vcpu *vcpu);
1365 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
1366 
1367 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
1368 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
1369 void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
1370 
1371 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
1372 		    int reason, bool has_error_code, u32 error_code);
1373 
1374 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
1375 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
1376 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
1377 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
1378 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
1379 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
1380 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
1381 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
1382 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
1383 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
1384 
1385 int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
1386 int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
1387 
1388 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
1389 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
1390 bool kvm_rdpmc(struct kvm_vcpu *vcpu);
1391 
1392 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1393 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
1394 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1395 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
1396 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
1397 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
1398 			    gfn_t gfn, void *data, int offset, int len,
1399 			    u32 access);
1400 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
1401 bool kvm_require_dr(struct kvm_vcpu *vcpu, int dr);
1402 
1403 static inline int __kvm_irq_line_state(unsigned long *irq_state,
1404 				       int irq_source_id, int level)
1405 {
1406 	/* Logical OR for level trig interrupt */
1407 	if (level)
1408 		__set_bit(irq_source_id, irq_state);
1409 	else
1410 		__clear_bit(irq_source_id, irq_state);
1411 
1412 	return !!(*irq_state);
1413 }
1414 
1415 #define KVM_MMU_ROOT_CURRENT		BIT(0)
1416 #define KVM_MMU_ROOT_PREVIOUS(i)	BIT(1+i)
1417 #define KVM_MMU_ROOTS_ALL		(~0UL)
1418 
1419 int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
1420 void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
1421 
1422 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
1423 
1424 int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
1425 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
1426 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
1427 int kvm_mmu_load(struct kvm_vcpu *vcpu);
1428 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
1429 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
1430 void kvm_mmu_free_roots(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
1431 			ulong roots_to_free);
1432 gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1433 			   struct x86_exception *exception);
1434 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
1435 			      struct x86_exception *exception);
1436 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
1437 			       struct x86_exception *exception);
1438 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
1439 			       struct x86_exception *exception);
1440 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
1441 				struct x86_exception *exception);
1442 
1443 void kvm_vcpu_deactivate_apicv(struct kvm_vcpu *vcpu);
1444 
1445 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
1446 
1447 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u64 error_code,
1448 		       void *insn, int insn_len);
1449 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
1450 void kvm_mmu_invpcid_gva(struct kvm_vcpu *vcpu, gva_t gva, unsigned long pcid);
1451 void kvm_mmu_new_cr3(struct kvm_vcpu *vcpu, gpa_t new_cr3, bool skip_tlb_flush);
1452 
1453 void kvm_enable_tdp(void);
1454 void kvm_disable_tdp(void);
1455 
1456 static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1457 				  struct x86_exception *exception)
1458 {
1459 	return gpa;
1460 }
1461 
1462 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
1463 {
1464 	struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
1465 
1466 	return (struct kvm_mmu_page *)page_private(page);
1467 }
1468 
1469 static inline u16 kvm_read_ldt(void)
1470 {
1471 	u16 ldt;
1472 	asm("sldt %0" : "=g"(ldt));
1473 	return ldt;
1474 }
1475 
1476 static inline void kvm_load_ldt(u16 sel)
1477 {
1478 	asm("lldt %0" : : "rm"(sel));
1479 }
1480 
1481 #ifdef CONFIG_X86_64
1482 static inline unsigned long read_msr(unsigned long msr)
1483 {
1484 	u64 value;
1485 
1486 	rdmsrl(msr, value);
1487 	return value;
1488 }
1489 #endif
1490 
1491 static inline u32 get_rdx_init_val(void)
1492 {
1493 	return 0x600; /* P6 family */
1494 }
1495 
1496 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
1497 {
1498 	kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
1499 }
1500 
1501 #define TSS_IOPB_BASE_OFFSET 0x66
1502 #define TSS_BASE_SIZE 0x68
1503 #define TSS_IOPB_SIZE (65536 / 8)
1504 #define TSS_REDIRECTION_SIZE (256 / 8)
1505 #define RMODE_TSS_SIZE							\
1506 	(TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
1507 
1508 enum {
1509 	TASK_SWITCH_CALL = 0,
1510 	TASK_SWITCH_IRET = 1,
1511 	TASK_SWITCH_JMP = 2,
1512 	TASK_SWITCH_GATE = 3,
1513 };
1514 
1515 #define HF_GIF_MASK		(1 << 0)
1516 #define HF_HIF_MASK		(1 << 1)
1517 #define HF_VINTR_MASK		(1 << 2)
1518 #define HF_NMI_MASK		(1 << 3)
1519 #define HF_IRET_MASK		(1 << 4)
1520 #define HF_GUEST_MASK		(1 << 5) /* VCPU is in guest-mode */
1521 #define HF_SMM_MASK		(1 << 6)
1522 #define HF_SMM_INSIDE_NMI_MASK	(1 << 7)
1523 
1524 #define __KVM_VCPU_MULTIPLE_ADDRESS_SPACE
1525 #define KVM_ADDRESS_SPACE_NUM 2
1526 
1527 #define kvm_arch_vcpu_memslots_id(vcpu) ((vcpu)->arch.hflags & HF_SMM_MASK ? 1 : 0)
1528 #define kvm_memslots_for_spte_role(kvm, role) __kvm_memslots(kvm, (role).smm)
1529 
1530 asmlinkage void kvm_spurious_fault(void);
1531 
1532 /*
1533  * Hardware virtualization extension instructions may fault if a
1534  * reboot turns off virtualization while processes are running.
1535  * Usually after catching the fault we just panic; during reboot
1536  * instead the instruction is ignored.
1537  */
1538 #define __kvm_handle_fault_on_reboot(insn)				\
1539 	"666: \n\t"							\
1540 	insn "\n\t"							\
1541 	"jmp	668f \n\t"						\
1542 	"667: \n\t"							\
1543 	"call	kvm_spurious_fault \n\t"				\
1544 	"668: \n\t"							\
1545 	_ASM_EXTABLE(666b, 667b)
1546 
1547 #define KVM_ARCH_WANT_MMU_NOTIFIER
1548 int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
1549 int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
1550 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
1551 int kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
1552 int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
1553 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1554 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
1555 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
1556 void kvm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event);
1557 void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu);
1558 
1559 int kvm_pv_send_ipi(struct kvm *kvm, unsigned long ipi_bitmap_low,
1560 		    unsigned long ipi_bitmap_high, u32 min,
1561 		    unsigned long icr, int op_64_bit);
1562 
1563 void kvm_define_shared_msr(unsigned index, u32 msr);
1564 int kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
1565 
1566 u64 kvm_scale_tsc(struct kvm_vcpu *vcpu, u64 tsc);
1567 u64 kvm_read_l1_tsc(struct kvm_vcpu *vcpu, u64 host_tsc);
1568 
1569 unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
1570 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1571 
1572 void kvm_make_mclock_inprogress_request(struct kvm *kvm);
1573 void kvm_make_scan_ioapic_request(struct kvm *kvm);
1574 
1575 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1576 				     struct kvm_async_pf *work);
1577 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1578 				 struct kvm_async_pf *work);
1579 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1580 			       struct kvm_async_pf *work);
1581 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
1582 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1583 
1584 int kvm_skip_emulated_instruction(struct kvm_vcpu *vcpu);
1585 int kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1586 void __kvm_request_immediate_exit(struct kvm_vcpu *vcpu);
1587 
1588 int kvm_is_in_guest(void);
1589 
1590 int __x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa, u32 size);
1591 int x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa, u32 size);
1592 bool kvm_vcpu_is_reset_bsp(struct kvm_vcpu *vcpu);
1593 bool kvm_vcpu_is_bsp(struct kvm_vcpu *vcpu);
1594 
1595 bool kvm_intr_is_single_vcpu(struct kvm *kvm, struct kvm_lapic_irq *irq,
1596 			     struct kvm_vcpu **dest_vcpu);
1597 
1598 void kvm_set_msi_irq(struct kvm *kvm, struct kvm_kernel_irq_routing_entry *e,
1599 		     struct kvm_lapic_irq *irq);
1600 
1601 static inline bool kvm_irq_is_postable(struct kvm_lapic_irq *irq)
1602 {
1603 	/* We can only post Fixed and LowPrio IRQs */
1604 	return (irq->delivery_mode == dest_Fixed ||
1605 		irq->delivery_mode == dest_LowestPrio);
1606 }
1607 
1608 static inline void kvm_arch_vcpu_blocking(struct kvm_vcpu *vcpu)
1609 {
1610 	if (kvm_x86_ops->vcpu_blocking)
1611 		kvm_x86_ops->vcpu_blocking(vcpu);
1612 }
1613 
1614 static inline void kvm_arch_vcpu_unblocking(struct kvm_vcpu *vcpu)
1615 {
1616 	if (kvm_x86_ops->vcpu_unblocking)
1617 		kvm_x86_ops->vcpu_unblocking(vcpu);
1618 }
1619 
1620 static inline void kvm_arch_vcpu_block_finish(struct kvm_vcpu *vcpu) {}
1621 
1622 static inline int kvm_cpu_get_apicid(int mps_cpu)
1623 {
1624 #ifdef CONFIG_X86_LOCAL_APIC
1625 	return default_cpu_present_to_apicid(mps_cpu);
1626 #else
1627 	WARN_ON_ONCE(1);
1628 	return BAD_APICID;
1629 #endif
1630 }
1631 
1632 #define put_smstate(type, buf, offset, val)                      \
1633 	*(type *)((buf) + (offset) - 0x7e00) = val
1634 
1635 #define GET_SMSTATE(type, buf, offset)		\
1636 	(*(type *)((buf) + (offset) - 0x7e00))
1637 
1638 #endif /* _ASM_X86_KVM_HOST_H */
1639