xref: /openbmc/linux/arch/x86/include/asm/kvm_host.h (revision 715f23b6)
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 	VCPU_EXREG_PDPTR = NR_VCPU_REGS,
162 	VCPU_EXREG_CR3,
163 	VCPU_EXREG_RFLAGS,
164 	VCPU_EXREG_SEGMENTS,
165 };
166 
167 enum {
168 	VCPU_SREG_ES,
169 	VCPU_SREG_CS,
170 	VCPU_SREG_SS,
171 	VCPU_SREG_DS,
172 	VCPU_SREG_FS,
173 	VCPU_SREG_GS,
174 	VCPU_SREG_TR,
175 	VCPU_SREG_LDTR,
176 };
177 
178 #include <asm/kvm_emulate.h>
179 
180 #define KVM_NR_MEM_OBJS 40
181 
182 #define KVM_NR_DB_REGS	4
183 
184 #define DR6_BD		(1 << 13)
185 #define DR6_BS		(1 << 14)
186 #define DR6_BT		(1 << 15)
187 #define DR6_RTM		(1 << 16)
188 #define DR6_FIXED_1	0xfffe0ff0
189 #define DR6_INIT	0xffff0ff0
190 #define DR6_VOLATILE	0x0001e00f
191 
192 #define DR7_BP_EN_MASK	0x000000ff
193 #define DR7_GE		(1 << 9)
194 #define DR7_GD		(1 << 13)
195 #define DR7_FIXED_1	0x00000400
196 #define DR7_VOLATILE	0xffff2bff
197 
198 #define PFERR_PRESENT_BIT 0
199 #define PFERR_WRITE_BIT 1
200 #define PFERR_USER_BIT 2
201 #define PFERR_RSVD_BIT 3
202 #define PFERR_FETCH_BIT 4
203 #define PFERR_PK_BIT 5
204 #define PFERR_GUEST_FINAL_BIT 32
205 #define PFERR_GUEST_PAGE_BIT 33
206 
207 #define PFERR_PRESENT_MASK (1U << PFERR_PRESENT_BIT)
208 #define PFERR_WRITE_MASK (1U << PFERR_WRITE_BIT)
209 #define PFERR_USER_MASK (1U << PFERR_USER_BIT)
210 #define PFERR_RSVD_MASK (1U << PFERR_RSVD_BIT)
211 #define PFERR_FETCH_MASK (1U << PFERR_FETCH_BIT)
212 #define PFERR_PK_MASK (1U << PFERR_PK_BIT)
213 #define PFERR_GUEST_FINAL_MASK (1ULL << PFERR_GUEST_FINAL_BIT)
214 #define PFERR_GUEST_PAGE_MASK (1ULL << PFERR_GUEST_PAGE_BIT)
215 
216 #define PFERR_NESTED_GUEST_PAGE (PFERR_GUEST_PAGE_MASK |	\
217 				 PFERR_WRITE_MASK |		\
218 				 PFERR_PRESENT_MASK)
219 
220 /* apic attention bits */
221 #define KVM_APIC_CHECK_VAPIC	0
222 /*
223  * The following bit is set with PV-EOI, unset on EOI.
224  * We detect PV-EOI changes by guest by comparing
225  * this bit with PV-EOI in guest memory.
226  * See the implementation in apic_update_pv_eoi.
227  */
228 #define KVM_APIC_PV_EOI_PENDING	1
229 
230 struct kvm_kernel_irq_routing_entry;
231 
232 /*
233  * We don't want allocation failures within the mmu code, so we preallocate
234  * enough memory for a single page fault in a cache.
235  */
236 struct kvm_mmu_memory_cache {
237 	int nobjs;
238 	void *objects[KVM_NR_MEM_OBJS];
239 };
240 
241 /*
242  * the pages used as guest page table on soft mmu are tracked by
243  * kvm_memory_slot.arch.gfn_track which is 16 bits, so the role bits used
244  * by indirect shadow page can not be more than 15 bits.
245  *
246  * Currently, we used 14 bits that are @level, @gpte_is_8_bytes, @quadrant, @access,
247  * @nxe, @cr0_wp, @smep_andnot_wp and @smap_andnot_wp.
248  */
249 union kvm_mmu_page_role {
250 	u32 word;
251 	struct {
252 		unsigned level:4;
253 		unsigned gpte_is_8_bytes:1;
254 		unsigned quadrant:2;
255 		unsigned direct:1;
256 		unsigned access:3;
257 		unsigned invalid:1;
258 		unsigned nxe:1;
259 		unsigned cr0_wp:1;
260 		unsigned smep_andnot_wp:1;
261 		unsigned smap_andnot_wp:1;
262 		unsigned ad_disabled:1;
263 		unsigned guest_mode:1;
264 		unsigned :6;
265 
266 		/*
267 		 * This is left at the top of the word so that
268 		 * kvm_memslots_for_spte_role can extract it with a
269 		 * simple shift.  While there is room, give it a whole
270 		 * byte so it is also faster to load it from memory.
271 		 */
272 		unsigned smm:8;
273 	};
274 };
275 
276 union kvm_mmu_extended_role {
277 /*
278  * This structure complements kvm_mmu_page_role caching everything needed for
279  * MMU configuration. If nothing in both these structures changed, MMU
280  * re-configuration can be skipped. @valid bit is set on first usage so we don't
281  * treat all-zero structure as valid data.
282  */
283 	u32 word;
284 	struct {
285 		unsigned int valid:1;
286 		unsigned int execonly:1;
287 		unsigned int cr0_pg:1;
288 		unsigned int cr4_pae:1;
289 		unsigned int cr4_pse:1;
290 		unsigned int cr4_pke:1;
291 		unsigned int cr4_smap:1;
292 		unsigned int cr4_smep:1;
293 		unsigned int cr4_la57:1;
294 		unsigned int maxphyaddr:6;
295 	};
296 };
297 
298 union kvm_mmu_role {
299 	u64 as_u64;
300 	struct {
301 		union kvm_mmu_page_role base;
302 		union kvm_mmu_extended_role ext;
303 	};
304 };
305 
306 struct kvm_rmap_head {
307 	unsigned long val;
308 };
309 
310 struct kvm_mmu_page {
311 	struct list_head link;
312 	struct hlist_node hash_link;
313 	struct list_head lpage_disallowed_link;
314 
315 	bool unsync;
316 	u8 mmu_valid_gen;
317 	bool mmio_cached;
318 	bool lpage_disallowed; /* Can't be replaced by an equiv large page */
319 
320 	/*
321 	 * The following two entries are used to key the shadow page in the
322 	 * hash table.
323 	 */
324 	union kvm_mmu_page_role role;
325 	gfn_t gfn;
326 
327 	u64 *spt;
328 	/* hold the gfn of each spte inside spt */
329 	gfn_t *gfns;
330 	int root_count;          /* Currently serving as active root */
331 	unsigned int unsync_children;
332 	struct kvm_rmap_head parent_ptes; /* rmap pointers to parent sptes */
333 	DECLARE_BITMAP(unsync_child_bitmap, 512);
334 
335 #ifdef CONFIG_X86_32
336 	/*
337 	 * Used out of the mmu-lock to avoid reading spte values while an
338 	 * update is in progress; see the comments in __get_spte_lockless().
339 	 */
340 	int clear_spte_count;
341 #endif
342 
343 	/* Number of writes since the last time traversal visited this page.  */
344 	atomic_t write_flooding_count;
345 };
346 
347 struct kvm_pio_request {
348 	unsigned long linear_rip;
349 	unsigned long count;
350 	int in;
351 	int port;
352 	int size;
353 };
354 
355 #define PT64_ROOT_MAX_LEVEL 5
356 
357 struct rsvd_bits_validate {
358 	u64 rsvd_bits_mask[2][PT64_ROOT_MAX_LEVEL];
359 	u64 bad_mt_xwr;
360 };
361 
362 struct kvm_mmu_root_info {
363 	gpa_t cr3;
364 	hpa_t hpa;
365 };
366 
367 #define KVM_MMU_ROOT_INFO_INVALID \
368 	((struct kvm_mmu_root_info) { .cr3 = INVALID_PAGE, .hpa = INVALID_PAGE })
369 
370 #define KVM_MMU_NUM_PREV_ROOTS 3
371 
372 /*
373  * x86 supports 4 paging modes (5-level 64-bit, 4-level 64-bit, 3-level 32-bit,
374  * and 2-level 32-bit).  The kvm_mmu structure abstracts the details of the
375  * current mmu mode.
376  */
377 struct kvm_mmu {
378 	void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
379 	unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
380 	u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
381 	int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
382 			  bool prefault);
383 	void (*inject_page_fault)(struct kvm_vcpu *vcpu,
384 				  struct x86_exception *fault);
385 	gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
386 			    struct x86_exception *exception);
387 	gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
388 			       struct x86_exception *exception);
389 	int (*sync_page)(struct kvm_vcpu *vcpu,
390 			 struct kvm_mmu_page *sp);
391 	void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva, hpa_t root_hpa);
392 	void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
393 			   u64 *spte, const void *pte);
394 	hpa_t root_hpa;
395 	gpa_t root_cr3;
396 	union kvm_mmu_role mmu_role;
397 	u8 root_level;
398 	u8 shadow_root_level;
399 	u8 ept_ad;
400 	bool direct_map;
401 	struct kvm_mmu_root_info prev_roots[KVM_MMU_NUM_PREV_ROOTS];
402 
403 	/*
404 	 * Bitmap; bit set = permission fault
405 	 * Byte index: page fault error code [4:1]
406 	 * Bit index: pte permissions in ACC_* format
407 	 */
408 	u8 permissions[16];
409 
410 	/*
411 	* The pkru_mask indicates if protection key checks are needed.  It
412 	* consists of 16 domains indexed by page fault error code bits [4:1],
413 	* with PFEC.RSVD replaced by ACC_USER_MASK from the page tables.
414 	* Each domain has 2 bits which are ANDed with AD and WD from PKRU.
415 	*/
416 	u32 pkru_mask;
417 
418 	u64 *pae_root;
419 	u64 *lm_root;
420 
421 	/*
422 	 * check zero bits on shadow page table entries, these
423 	 * bits include not only hardware reserved bits but also
424 	 * the bits spte never used.
425 	 */
426 	struct rsvd_bits_validate shadow_zero_check;
427 
428 	struct rsvd_bits_validate guest_rsvd_check;
429 
430 	/* Can have large pages at levels 2..last_nonleaf_level-1. */
431 	u8 last_nonleaf_level;
432 
433 	bool nx;
434 
435 	u64 pdptrs[4]; /* pae */
436 };
437 
438 struct kvm_tlb_range {
439 	u64 start_gfn;
440 	u64 pages;
441 };
442 
443 enum pmc_type {
444 	KVM_PMC_GP = 0,
445 	KVM_PMC_FIXED,
446 };
447 
448 struct kvm_pmc {
449 	enum pmc_type type;
450 	u8 idx;
451 	u64 counter;
452 	u64 eventsel;
453 	struct perf_event *perf_event;
454 	struct kvm_vcpu *vcpu;
455 	/*
456 	 * eventsel value for general purpose counters,
457 	 * ctrl value for fixed counters.
458 	 */
459 	u64 current_config;
460 };
461 
462 struct kvm_pmu {
463 	unsigned nr_arch_gp_counters;
464 	unsigned nr_arch_fixed_counters;
465 	unsigned available_event_types;
466 	u64 fixed_ctr_ctrl;
467 	u64 global_ctrl;
468 	u64 global_status;
469 	u64 global_ovf_ctrl;
470 	u64 counter_bitmask[2];
471 	u64 global_ctrl_mask;
472 	u64 global_ovf_ctrl_mask;
473 	u64 reserved_bits;
474 	u8 version;
475 	struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
476 	struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
477 	struct irq_work irq_work;
478 	DECLARE_BITMAP(reprogram_pmi, X86_PMC_IDX_MAX);
479 	DECLARE_BITMAP(all_valid_pmc_idx, X86_PMC_IDX_MAX);
480 	DECLARE_BITMAP(pmc_in_use, X86_PMC_IDX_MAX);
481 
482 	/*
483 	 * The gate to release perf_events not marked in
484 	 * pmc_in_use only once in a vcpu time slice.
485 	 */
486 	bool need_cleanup;
487 
488 	/*
489 	 * The total number of programmed perf_events and it helps to avoid
490 	 * redundant check before cleanup if guest don't use vPMU at all.
491 	 */
492 	u8 event_count;
493 };
494 
495 struct kvm_pmu_ops;
496 
497 enum {
498 	KVM_DEBUGREG_BP_ENABLED = 1,
499 	KVM_DEBUGREG_WONT_EXIT = 2,
500 	KVM_DEBUGREG_RELOAD = 4,
501 };
502 
503 struct kvm_mtrr_range {
504 	u64 base;
505 	u64 mask;
506 	struct list_head node;
507 };
508 
509 struct kvm_mtrr {
510 	struct kvm_mtrr_range var_ranges[KVM_NR_VAR_MTRR];
511 	mtrr_type fixed_ranges[KVM_NR_FIXED_MTRR_REGION];
512 	u64 deftype;
513 
514 	struct list_head head;
515 };
516 
517 /* Hyper-V SynIC timer */
518 struct kvm_vcpu_hv_stimer {
519 	struct hrtimer timer;
520 	int index;
521 	union hv_stimer_config config;
522 	u64 count;
523 	u64 exp_time;
524 	struct hv_message msg;
525 	bool msg_pending;
526 };
527 
528 /* Hyper-V synthetic interrupt controller (SynIC)*/
529 struct kvm_vcpu_hv_synic {
530 	u64 version;
531 	u64 control;
532 	u64 msg_page;
533 	u64 evt_page;
534 	atomic64_t sint[HV_SYNIC_SINT_COUNT];
535 	atomic_t sint_to_gsi[HV_SYNIC_SINT_COUNT];
536 	DECLARE_BITMAP(auto_eoi_bitmap, 256);
537 	DECLARE_BITMAP(vec_bitmap, 256);
538 	bool active;
539 	bool dont_zero_synic_pages;
540 };
541 
542 /* Hyper-V per vcpu emulation context */
543 struct kvm_vcpu_hv {
544 	u32 vp_index;
545 	u64 hv_vapic;
546 	s64 runtime_offset;
547 	struct kvm_vcpu_hv_synic synic;
548 	struct kvm_hyperv_exit exit;
549 	struct kvm_vcpu_hv_stimer stimer[HV_SYNIC_STIMER_COUNT];
550 	DECLARE_BITMAP(stimer_pending_bitmap, HV_SYNIC_STIMER_COUNT);
551 	cpumask_t tlb_flush;
552 };
553 
554 struct kvm_vcpu_arch {
555 	/*
556 	 * rip and regs accesses must go through
557 	 * kvm_{register,rip}_{read,write} functions.
558 	 */
559 	unsigned long regs[NR_VCPU_REGS];
560 	u32 regs_avail;
561 	u32 regs_dirty;
562 
563 	unsigned long cr0;
564 	unsigned long cr0_guest_owned_bits;
565 	unsigned long cr2;
566 	unsigned long cr3;
567 	unsigned long cr4;
568 	unsigned long cr4_guest_owned_bits;
569 	unsigned long cr8;
570 	u32 pkru;
571 	u32 hflags;
572 	u64 efer;
573 	u64 apic_base;
574 	struct kvm_lapic *apic;    /* kernel irqchip context */
575 	bool apicv_active;
576 	bool load_eoi_exitmap_pending;
577 	DECLARE_BITMAP(ioapic_handled_vectors, 256);
578 	unsigned long apic_attention;
579 	int32_t apic_arb_prio;
580 	int mp_state;
581 	u64 ia32_misc_enable_msr;
582 	u64 smbase;
583 	u64 smi_count;
584 	bool tpr_access_reporting;
585 	bool xsaves_enabled;
586 	u64 ia32_xss;
587 	u64 microcode_version;
588 	u64 arch_capabilities;
589 
590 	/*
591 	 * Paging state of the vcpu
592 	 *
593 	 * If the vcpu runs in guest mode with two level paging this still saves
594 	 * the paging mode of the l1 guest. This context is always used to
595 	 * handle faults.
596 	 */
597 	struct kvm_mmu *mmu;
598 
599 	/* Non-nested MMU for L1 */
600 	struct kvm_mmu root_mmu;
601 
602 	/* L1 MMU when running nested */
603 	struct kvm_mmu guest_mmu;
604 
605 	/*
606 	 * Paging state of an L2 guest (used for nested npt)
607 	 *
608 	 * This context will save all necessary information to walk page tables
609 	 * of the an L2 guest. This context is only initialized for page table
610 	 * walking and not for faulting since we never handle l2 page faults on
611 	 * the host.
612 	 */
613 	struct kvm_mmu nested_mmu;
614 
615 	/*
616 	 * Pointer to the mmu context currently used for
617 	 * gva_to_gpa translations.
618 	 */
619 	struct kvm_mmu *walk_mmu;
620 
621 	struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
622 	struct kvm_mmu_memory_cache mmu_page_cache;
623 	struct kvm_mmu_memory_cache mmu_page_header_cache;
624 
625 	/*
626 	 * QEMU userspace and the guest each have their own FPU state.
627 	 * In vcpu_run, we switch between the user and guest FPU contexts.
628 	 * While running a VCPU, the VCPU thread will have the guest FPU
629 	 * context.
630 	 *
631 	 * Note that while the PKRU state lives inside the fpu registers,
632 	 * it is switched out separately at VMENTER and VMEXIT time. The
633 	 * "guest_fpu" state here contains the guest FPU context, with the
634 	 * host PRKU bits.
635 	 */
636 	struct fpu *user_fpu;
637 	struct fpu *guest_fpu;
638 
639 	u64 xcr0;
640 	u64 guest_supported_xcr0;
641 	u32 guest_xstate_size;
642 
643 	struct kvm_pio_request pio;
644 	void *pio_data;
645 
646 	u8 event_exit_inst_len;
647 
648 	struct kvm_queued_exception {
649 		bool pending;
650 		bool injected;
651 		bool has_error_code;
652 		u8 nr;
653 		u32 error_code;
654 		unsigned long payload;
655 		bool has_payload;
656 		u8 nested_apf;
657 	} exception;
658 
659 	struct kvm_queued_interrupt {
660 		bool injected;
661 		bool soft;
662 		u8 nr;
663 	} interrupt;
664 
665 	int halt_request; /* real mode on Intel only */
666 
667 	int cpuid_nent;
668 	struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
669 
670 	int maxphyaddr;
671 
672 	/* emulate context */
673 
674 	struct x86_emulate_ctxt emulate_ctxt;
675 	bool emulate_regs_need_sync_to_vcpu;
676 	bool emulate_regs_need_sync_from_vcpu;
677 	int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
678 
679 	gpa_t time;
680 	struct pvclock_vcpu_time_info hv_clock;
681 	unsigned int hw_tsc_khz;
682 	struct gfn_to_hva_cache pv_time;
683 	bool pv_time_enabled;
684 	/* set guest stopped flag in pvclock flags field */
685 	bool pvclock_set_guest_stopped_request;
686 
687 	struct {
688 		u64 msr_val;
689 		u64 last_steal;
690 		struct gfn_to_hva_cache stime;
691 		struct kvm_steal_time steal;
692 	} st;
693 
694 	u64 tsc_offset;
695 	u64 last_guest_tsc;
696 	u64 last_host_tsc;
697 	u64 tsc_offset_adjustment;
698 	u64 this_tsc_nsec;
699 	u64 this_tsc_write;
700 	u64 this_tsc_generation;
701 	bool tsc_catchup;
702 	bool tsc_always_catchup;
703 	s8 virtual_tsc_shift;
704 	u32 virtual_tsc_mult;
705 	u32 virtual_tsc_khz;
706 	s64 ia32_tsc_adjust_msr;
707 	u64 msr_ia32_power_ctl;
708 	u64 tsc_scaling_ratio;
709 
710 	atomic_t nmi_queued;  /* unprocessed asynchronous NMIs */
711 	unsigned nmi_pending; /* NMI queued after currently running handler */
712 	bool nmi_injected;    /* Trying to inject an NMI this entry */
713 	bool smi_pending;    /* SMI queued after currently running handler */
714 
715 	struct kvm_mtrr mtrr_state;
716 	u64 pat;
717 
718 	unsigned switch_db_regs;
719 	unsigned long db[KVM_NR_DB_REGS];
720 	unsigned long dr6;
721 	unsigned long dr7;
722 	unsigned long eff_db[KVM_NR_DB_REGS];
723 	unsigned long guest_debug_dr7;
724 	u64 msr_platform_info;
725 	u64 msr_misc_features_enables;
726 
727 	u64 mcg_cap;
728 	u64 mcg_status;
729 	u64 mcg_ctl;
730 	u64 mcg_ext_ctl;
731 	u64 *mce_banks;
732 
733 	/* Cache MMIO info */
734 	u64 mmio_gva;
735 	unsigned mmio_access;
736 	gfn_t mmio_gfn;
737 	u64 mmio_gen;
738 
739 	struct kvm_pmu pmu;
740 
741 	/* used for guest single stepping over the given code position */
742 	unsigned long singlestep_rip;
743 
744 	struct kvm_vcpu_hv hyperv;
745 
746 	cpumask_var_t wbinvd_dirty_mask;
747 
748 	unsigned long last_retry_eip;
749 	unsigned long last_retry_addr;
750 
751 	struct {
752 		bool halted;
753 		gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
754 		struct gfn_to_hva_cache data;
755 		u64 msr_val;
756 		u32 id;
757 		bool send_user_only;
758 		u32 host_apf_reason;
759 		unsigned long nested_apf_token;
760 		bool delivery_as_pf_vmexit;
761 	} apf;
762 
763 	/* OSVW MSRs (AMD only) */
764 	struct {
765 		u64 length;
766 		u64 status;
767 	} osvw;
768 
769 	struct {
770 		u64 msr_val;
771 		struct gfn_to_hva_cache data;
772 	} pv_eoi;
773 
774 	u64 msr_kvm_poll_control;
775 
776 	/*
777 	 * Indicate whether the access faults on its page table in guest
778 	 * which is set when fix page fault and used to detect unhandeable
779 	 * instruction.
780 	 */
781 	bool write_fault_to_shadow_pgtable;
782 
783 	/* set at EPT violation at this point */
784 	unsigned long exit_qualification;
785 
786 	/* pv related host specific info */
787 	struct {
788 		bool pv_unhalted;
789 	} pv;
790 
791 	int pending_ioapic_eoi;
792 	int pending_external_vector;
793 
794 	/* GPA available */
795 	bool gpa_available;
796 	gpa_t gpa_val;
797 
798 	/* be preempted when it's in kernel-mode(cpl=0) */
799 	bool preempted_in_kernel;
800 
801 	/* Flush the L1 Data cache for L1TF mitigation on VMENTER */
802 	bool l1tf_flush_l1d;
803 
804 	/* AMD MSRC001_0015 Hardware Configuration */
805 	u64 msr_hwcr;
806 };
807 
808 struct kvm_lpage_info {
809 	int disallow_lpage;
810 };
811 
812 struct kvm_arch_memory_slot {
813 	struct kvm_rmap_head *rmap[KVM_NR_PAGE_SIZES];
814 	struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
815 	unsigned short *gfn_track[KVM_PAGE_TRACK_MAX];
816 };
817 
818 /*
819  * We use as the mode the number of bits allocated in the LDR for the
820  * logical processor ID.  It happens that these are all powers of two.
821  * This makes it is very easy to detect cases where the APICs are
822  * configured for multiple modes; in that case, we cannot use the map and
823  * hence cannot use kvm_irq_delivery_to_apic_fast either.
824  */
825 #define KVM_APIC_MODE_XAPIC_CLUSTER          4
826 #define KVM_APIC_MODE_XAPIC_FLAT             8
827 #define KVM_APIC_MODE_X2APIC                16
828 
829 struct kvm_apic_map {
830 	struct rcu_head rcu;
831 	u8 mode;
832 	u32 max_apic_id;
833 	union {
834 		struct kvm_lapic *xapic_flat_map[8];
835 		struct kvm_lapic *xapic_cluster_map[16][4];
836 	};
837 	struct kvm_lapic *phys_map[];
838 };
839 
840 /* Hyper-V emulation context */
841 struct kvm_hv {
842 	struct mutex hv_lock;
843 	u64 hv_guest_os_id;
844 	u64 hv_hypercall;
845 	u64 hv_tsc_page;
846 
847 	/* Hyper-v based guest crash (NT kernel bugcheck) parameters */
848 	u64 hv_crash_param[HV_X64_MSR_CRASH_PARAMS];
849 	u64 hv_crash_ctl;
850 
851 	HV_REFERENCE_TSC_PAGE tsc_ref;
852 
853 	struct idr conn_to_evt;
854 
855 	u64 hv_reenlightenment_control;
856 	u64 hv_tsc_emulation_control;
857 	u64 hv_tsc_emulation_status;
858 
859 	/* How many vCPUs have VP index != vCPU index */
860 	atomic_t num_mismatched_vp_indexes;
861 
862 	struct hv_partition_assist_pg *hv_pa_pg;
863 };
864 
865 enum kvm_irqchip_mode {
866 	KVM_IRQCHIP_NONE,
867 	KVM_IRQCHIP_KERNEL,       /* created with KVM_CREATE_IRQCHIP */
868 	KVM_IRQCHIP_SPLIT,        /* created with KVM_CAP_SPLIT_IRQCHIP */
869 };
870 
871 struct kvm_arch {
872 	unsigned long n_used_mmu_pages;
873 	unsigned long n_requested_mmu_pages;
874 	unsigned long n_max_mmu_pages;
875 	unsigned int indirect_shadow_pages;
876 	u8 mmu_valid_gen;
877 	struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
878 	/*
879 	 * Hash table of struct kvm_mmu_page.
880 	 */
881 	struct list_head active_mmu_pages;
882 	struct list_head zapped_obsolete_pages;
883 	struct list_head lpage_disallowed_mmu_pages;
884 	struct kvm_page_track_notifier_node mmu_sp_tracker;
885 	struct kvm_page_track_notifier_head track_notifier_head;
886 
887 	struct list_head assigned_dev_head;
888 	struct iommu_domain *iommu_domain;
889 	bool iommu_noncoherent;
890 #define __KVM_HAVE_ARCH_NONCOHERENT_DMA
891 	atomic_t noncoherent_dma_count;
892 #define __KVM_HAVE_ARCH_ASSIGNED_DEVICE
893 	atomic_t assigned_device_count;
894 	struct kvm_pic *vpic;
895 	struct kvm_ioapic *vioapic;
896 	struct kvm_pit *vpit;
897 	atomic_t vapics_in_nmi_mode;
898 	struct mutex apic_map_lock;
899 	struct kvm_apic_map *apic_map;
900 
901 	bool apic_access_page_done;
902 
903 	gpa_t wall_clock;
904 
905 	bool mwait_in_guest;
906 	bool hlt_in_guest;
907 	bool pause_in_guest;
908 	bool cstate_in_guest;
909 
910 	unsigned long irq_sources_bitmap;
911 	s64 kvmclock_offset;
912 	raw_spinlock_t tsc_write_lock;
913 	u64 last_tsc_nsec;
914 	u64 last_tsc_write;
915 	u32 last_tsc_khz;
916 	u64 cur_tsc_nsec;
917 	u64 cur_tsc_write;
918 	u64 cur_tsc_offset;
919 	u64 cur_tsc_generation;
920 	int nr_vcpus_matched_tsc;
921 
922 	spinlock_t pvclock_gtod_sync_lock;
923 	bool use_master_clock;
924 	u64 master_kernel_ns;
925 	u64 master_cycle_now;
926 	struct delayed_work kvmclock_update_work;
927 	struct delayed_work kvmclock_sync_work;
928 
929 	struct kvm_xen_hvm_config xen_hvm_config;
930 
931 	/* reads protected by irq_srcu, writes by irq_lock */
932 	struct hlist_head mask_notifier_list;
933 
934 	struct kvm_hv hyperv;
935 
936 	#ifdef CONFIG_KVM_MMU_AUDIT
937 	int audit_point;
938 	#endif
939 
940 	bool backwards_tsc_observed;
941 	bool boot_vcpu_runs_old_kvmclock;
942 	u32 bsp_vcpu_id;
943 
944 	u64 disabled_quirks;
945 
946 	enum kvm_irqchip_mode irqchip_mode;
947 	u8 nr_reserved_ioapic_pins;
948 
949 	bool disabled_lapic_found;
950 
951 	bool x2apic_format;
952 	bool x2apic_broadcast_quirk_disabled;
953 
954 	bool guest_can_read_msr_platform_info;
955 	bool exception_payload_enabled;
956 
957 	struct kvm_pmu_event_filter *pmu_event_filter;
958 	struct task_struct *nx_lpage_recovery_thread;
959 };
960 
961 struct kvm_vm_stat {
962 	ulong mmu_shadow_zapped;
963 	ulong mmu_pte_write;
964 	ulong mmu_pte_updated;
965 	ulong mmu_pde_zapped;
966 	ulong mmu_flooded;
967 	ulong mmu_recycled;
968 	ulong mmu_cache_miss;
969 	ulong mmu_unsync;
970 	ulong remote_tlb_flush;
971 	ulong lpages;
972 	ulong nx_lpage_splits;
973 	ulong max_mmu_page_hash_collisions;
974 };
975 
976 struct kvm_vcpu_stat {
977 	u64 pf_fixed;
978 	u64 pf_guest;
979 	u64 tlb_flush;
980 	u64 invlpg;
981 
982 	u64 exits;
983 	u64 io_exits;
984 	u64 mmio_exits;
985 	u64 signal_exits;
986 	u64 irq_window_exits;
987 	u64 nmi_window_exits;
988 	u64 l1d_flush;
989 	u64 halt_exits;
990 	u64 halt_successful_poll;
991 	u64 halt_attempted_poll;
992 	u64 halt_poll_invalid;
993 	u64 halt_wakeup;
994 	u64 request_irq_exits;
995 	u64 irq_exits;
996 	u64 host_state_reload;
997 	u64 fpu_reload;
998 	u64 insn_emulation;
999 	u64 insn_emulation_fail;
1000 	u64 hypercalls;
1001 	u64 irq_injections;
1002 	u64 nmi_injections;
1003 	u64 req_event;
1004 };
1005 
1006 struct x86_instruction_info;
1007 
1008 struct msr_data {
1009 	bool host_initiated;
1010 	u32 index;
1011 	u64 data;
1012 };
1013 
1014 struct kvm_lapic_irq {
1015 	u32 vector;
1016 	u16 delivery_mode;
1017 	u16 dest_mode;
1018 	bool level;
1019 	u16 trig_mode;
1020 	u32 shorthand;
1021 	u32 dest_id;
1022 	bool msi_redir_hint;
1023 };
1024 
1025 struct kvm_x86_ops {
1026 	int (*cpu_has_kvm_support)(void);          /* __init */
1027 	int (*disabled_by_bios)(void);             /* __init */
1028 	int (*hardware_enable)(void);
1029 	void (*hardware_disable)(void);
1030 	int (*check_processor_compatibility)(void);/* __init */
1031 	int (*hardware_setup)(void);               /* __init */
1032 	void (*hardware_unsetup)(void);            /* __exit */
1033 	bool (*cpu_has_accelerated_tpr)(void);
1034 	bool (*has_emulated_msr)(int index);
1035 	void (*cpuid_update)(struct kvm_vcpu *vcpu);
1036 
1037 	struct kvm *(*vm_alloc)(void);
1038 	void (*vm_free)(struct kvm *);
1039 	int (*vm_init)(struct kvm *kvm);
1040 	void (*vm_destroy)(struct kvm *kvm);
1041 
1042 	/* Create, but do not attach this VCPU */
1043 	struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
1044 	void (*vcpu_free)(struct kvm_vcpu *vcpu);
1045 	void (*vcpu_reset)(struct kvm_vcpu *vcpu, bool init_event);
1046 
1047 	void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
1048 	void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
1049 	void (*vcpu_put)(struct kvm_vcpu *vcpu);
1050 
1051 	void (*update_bp_intercept)(struct kvm_vcpu *vcpu);
1052 	int (*get_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
1053 	int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
1054 	u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
1055 	void (*get_segment)(struct kvm_vcpu *vcpu,
1056 			    struct kvm_segment *var, int seg);
1057 	int (*get_cpl)(struct kvm_vcpu *vcpu);
1058 	void (*set_segment)(struct kvm_vcpu *vcpu,
1059 			    struct kvm_segment *var, int seg);
1060 	void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
1061 	void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
1062 	void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
1063 	void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
1064 	void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
1065 	int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
1066 	void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
1067 	void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
1068 	void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
1069 	void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
1070 	void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
1071 	u64 (*get_dr6)(struct kvm_vcpu *vcpu);
1072 	void (*set_dr6)(struct kvm_vcpu *vcpu, unsigned long value);
1073 	void (*sync_dirty_debug_regs)(struct kvm_vcpu *vcpu);
1074 	void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
1075 	void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
1076 	unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
1077 	void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
1078 
1079 	void (*tlb_flush)(struct kvm_vcpu *vcpu, bool invalidate_gpa);
1080 	int  (*tlb_remote_flush)(struct kvm *kvm);
1081 	int  (*tlb_remote_flush_with_range)(struct kvm *kvm,
1082 			struct kvm_tlb_range *range);
1083 
1084 	/*
1085 	 * Flush any TLB entries associated with the given GVA.
1086 	 * Does not need to flush GPA->HPA mappings.
1087 	 * Can potentially get non-canonical addresses through INVLPGs, which
1088 	 * the implementation may choose to ignore if appropriate.
1089 	 */
1090 	void (*tlb_flush_gva)(struct kvm_vcpu *vcpu, gva_t addr);
1091 
1092 	void (*run)(struct kvm_vcpu *vcpu);
1093 	int (*handle_exit)(struct kvm_vcpu *vcpu);
1094 	int (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
1095 	void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
1096 	u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu);
1097 	void (*patch_hypercall)(struct kvm_vcpu *vcpu,
1098 				unsigned char *hypercall_addr);
1099 	void (*set_irq)(struct kvm_vcpu *vcpu);
1100 	void (*set_nmi)(struct kvm_vcpu *vcpu);
1101 	void (*queue_exception)(struct kvm_vcpu *vcpu);
1102 	void (*cancel_injection)(struct kvm_vcpu *vcpu);
1103 	int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
1104 	int (*nmi_allowed)(struct kvm_vcpu *vcpu);
1105 	bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
1106 	void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
1107 	void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
1108 	void (*enable_irq_window)(struct kvm_vcpu *vcpu);
1109 	void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
1110 	bool (*get_enable_apicv)(struct kvm *kvm);
1111 	void (*refresh_apicv_exec_ctrl)(struct kvm_vcpu *vcpu);
1112 	void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
1113 	void (*hwapic_isr_update)(struct kvm_vcpu *vcpu, int isr);
1114 	bool (*guest_apic_has_interrupt)(struct kvm_vcpu *vcpu);
1115 	void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap);
1116 	void (*set_virtual_apic_mode)(struct kvm_vcpu *vcpu);
1117 	void (*set_apic_access_page_addr)(struct kvm_vcpu *vcpu, hpa_t hpa);
1118 	void (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
1119 	int (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
1120 	int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
1121 	int (*set_identity_map_addr)(struct kvm *kvm, u64 ident_addr);
1122 	int (*get_tdp_level)(struct kvm_vcpu *vcpu);
1123 	u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
1124 	int (*get_lpage_level)(void);
1125 	bool (*rdtscp_supported)(void);
1126 	bool (*invpcid_supported)(void);
1127 
1128 	void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
1129 
1130 	void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
1131 
1132 	bool (*has_wbinvd_exit)(void);
1133 
1134 	u64 (*read_l1_tsc_offset)(struct kvm_vcpu *vcpu);
1135 	/* Returns actual tsc_offset set in active VMCS */
1136 	u64 (*write_l1_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
1137 
1138 	void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
1139 
1140 	int (*check_intercept)(struct kvm_vcpu *vcpu,
1141 			       struct x86_instruction_info *info,
1142 			       enum x86_intercept_stage stage);
1143 	void (*handle_exit_irqoff)(struct kvm_vcpu *vcpu);
1144 	bool (*mpx_supported)(void);
1145 	bool (*xsaves_supported)(void);
1146 	bool (*umip_emulated)(void);
1147 	bool (*pt_supported)(void);
1148 
1149 	int (*check_nested_events)(struct kvm_vcpu *vcpu, bool external_intr);
1150 	void (*request_immediate_exit)(struct kvm_vcpu *vcpu);
1151 
1152 	void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
1153 
1154 	/*
1155 	 * Arch-specific dirty logging hooks. These hooks are only supposed to
1156 	 * be valid if the specific arch has hardware-accelerated dirty logging
1157 	 * mechanism. Currently only for PML on VMX.
1158 	 *
1159 	 *  - slot_enable_log_dirty:
1160 	 *	called when enabling log dirty mode for the slot.
1161 	 *  - slot_disable_log_dirty:
1162 	 *	called when disabling log dirty mode for the slot.
1163 	 *	also called when slot is created with log dirty disabled.
1164 	 *  - flush_log_dirty:
1165 	 *	called before reporting dirty_bitmap to userspace.
1166 	 *  - enable_log_dirty_pt_masked:
1167 	 *	called when reenabling log dirty for the GFNs in the mask after
1168 	 *	corresponding bits are cleared in slot->dirty_bitmap.
1169 	 */
1170 	void (*slot_enable_log_dirty)(struct kvm *kvm,
1171 				      struct kvm_memory_slot *slot);
1172 	void (*slot_disable_log_dirty)(struct kvm *kvm,
1173 				       struct kvm_memory_slot *slot);
1174 	void (*flush_log_dirty)(struct kvm *kvm);
1175 	void (*enable_log_dirty_pt_masked)(struct kvm *kvm,
1176 					   struct kvm_memory_slot *slot,
1177 					   gfn_t offset, unsigned long mask);
1178 	int (*write_log_dirty)(struct kvm_vcpu *vcpu);
1179 
1180 	/* pmu operations of sub-arch */
1181 	const struct kvm_pmu_ops *pmu_ops;
1182 
1183 	/*
1184 	 * Architecture specific hooks for vCPU blocking due to
1185 	 * HLT instruction.
1186 	 * Returns for .pre_block():
1187 	 *    - 0 means continue to block the vCPU.
1188 	 *    - 1 means we cannot block the vCPU since some event
1189 	 *        happens during this period, such as, 'ON' bit in
1190 	 *        posted-interrupts descriptor is set.
1191 	 */
1192 	int (*pre_block)(struct kvm_vcpu *vcpu);
1193 	void (*post_block)(struct kvm_vcpu *vcpu);
1194 
1195 	void (*vcpu_blocking)(struct kvm_vcpu *vcpu);
1196 	void (*vcpu_unblocking)(struct kvm_vcpu *vcpu);
1197 
1198 	int (*update_pi_irte)(struct kvm *kvm, unsigned int host_irq,
1199 			      uint32_t guest_irq, bool set);
1200 	void (*apicv_post_state_restore)(struct kvm_vcpu *vcpu);
1201 	bool (*dy_apicv_has_pending_interrupt)(struct kvm_vcpu *vcpu);
1202 
1203 	int (*set_hv_timer)(struct kvm_vcpu *vcpu, u64 guest_deadline_tsc,
1204 			    bool *expired);
1205 	void (*cancel_hv_timer)(struct kvm_vcpu *vcpu);
1206 
1207 	void (*setup_mce)(struct kvm_vcpu *vcpu);
1208 
1209 	int (*get_nested_state)(struct kvm_vcpu *vcpu,
1210 				struct kvm_nested_state __user *user_kvm_nested_state,
1211 				unsigned user_data_size);
1212 	int (*set_nested_state)(struct kvm_vcpu *vcpu,
1213 				struct kvm_nested_state __user *user_kvm_nested_state,
1214 				struct kvm_nested_state *kvm_state);
1215 	bool (*get_vmcs12_pages)(struct kvm_vcpu *vcpu);
1216 
1217 	int (*smi_allowed)(struct kvm_vcpu *vcpu);
1218 	int (*pre_enter_smm)(struct kvm_vcpu *vcpu, char *smstate);
1219 	int (*pre_leave_smm)(struct kvm_vcpu *vcpu, const char *smstate);
1220 	int (*enable_smi_window)(struct kvm_vcpu *vcpu);
1221 
1222 	int (*mem_enc_op)(struct kvm *kvm, void __user *argp);
1223 	int (*mem_enc_reg_region)(struct kvm *kvm, struct kvm_enc_region *argp);
1224 	int (*mem_enc_unreg_region)(struct kvm *kvm, struct kvm_enc_region *argp);
1225 
1226 	int (*get_msr_feature)(struct kvm_msr_entry *entry);
1227 
1228 	int (*nested_enable_evmcs)(struct kvm_vcpu *vcpu,
1229 				   uint16_t *vmcs_version);
1230 	uint16_t (*nested_get_evmcs_version)(struct kvm_vcpu *vcpu);
1231 
1232 	bool (*need_emulation_on_page_fault)(struct kvm_vcpu *vcpu);
1233 
1234 	bool (*apic_init_signal_blocked)(struct kvm_vcpu *vcpu);
1235 	int (*enable_direct_tlbflush)(struct kvm_vcpu *vcpu);
1236 };
1237 
1238 struct kvm_arch_async_pf {
1239 	u32 token;
1240 	gfn_t gfn;
1241 	unsigned long cr3;
1242 	bool direct_map;
1243 };
1244 
1245 extern struct kvm_x86_ops *kvm_x86_ops;
1246 extern struct kmem_cache *x86_fpu_cache;
1247 
1248 #define __KVM_HAVE_ARCH_VM_ALLOC
1249 static inline struct kvm *kvm_arch_alloc_vm(void)
1250 {
1251 	return kvm_x86_ops->vm_alloc();
1252 }
1253 
1254 static inline void kvm_arch_free_vm(struct kvm *kvm)
1255 {
1256 	return kvm_x86_ops->vm_free(kvm);
1257 }
1258 
1259 #define __KVM_HAVE_ARCH_FLUSH_REMOTE_TLB
1260 static inline int kvm_arch_flush_remote_tlb(struct kvm *kvm)
1261 {
1262 	if (kvm_x86_ops->tlb_remote_flush &&
1263 	    !kvm_x86_ops->tlb_remote_flush(kvm))
1264 		return 0;
1265 	else
1266 		return -ENOTSUPP;
1267 }
1268 
1269 int kvm_mmu_module_init(void);
1270 void kvm_mmu_module_exit(void);
1271 
1272 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
1273 int kvm_mmu_create(struct kvm_vcpu *vcpu);
1274 void kvm_mmu_init_vm(struct kvm *kvm);
1275 void kvm_mmu_uninit_vm(struct kvm *kvm);
1276 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
1277 		u64 dirty_mask, u64 nx_mask, u64 x_mask, u64 p_mask,
1278 		u64 acc_track_mask, u64 me_mask);
1279 
1280 void kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
1281 void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
1282 				      struct kvm_memory_slot *memslot);
1283 void kvm_mmu_zap_collapsible_sptes(struct kvm *kvm,
1284 				   const struct kvm_memory_slot *memslot);
1285 void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
1286 				   struct kvm_memory_slot *memslot);
1287 void kvm_mmu_slot_largepage_remove_write_access(struct kvm *kvm,
1288 					struct kvm_memory_slot *memslot);
1289 void kvm_mmu_slot_set_dirty(struct kvm *kvm,
1290 			    struct kvm_memory_slot *memslot);
1291 void kvm_mmu_clear_dirty_pt_masked(struct kvm *kvm,
1292 				   struct kvm_memory_slot *slot,
1293 				   gfn_t gfn_offset, unsigned long mask);
1294 void kvm_mmu_zap_all(struct kvm *kvm);
1295 void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm, u64 gen);
1296 unsigned long kvm_mmu_calculate_default_mmu_pages(struct kvm *kvm);
1297 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned long kvm_nr_mmu_pages);
1298 
1299 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
1300 bool pdptrs_changed(struct kvm_vcpu *vcpu);
1301 
1302 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
1303 			  const void *val, int bytes);
1304 
1305 struct kvm_irq_mask_notifier {
1306 	void (*func)(struct kvm_irq_mask_notifier *kimn, bool masked);
1307 	int irq;
1308 	struct hlist_node link;
1309 };
1310 
1311 void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
1312 				    struct kvm_irq_mask_notifier *kimn);
1313 void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
1314 				      struct kvm_irq_mask_notifier *kimn);
1315 void kvm_fire_mask_notifiers(struct kvm *kvm, unsigned irqchip, unsigned pin,
1316 			     bool mask);
1317 
1318 extern bool tdp_enabled;
1319 
1320 u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
1321 
1322 /* control of guest tsc rate supported? */
1323 extern bool kvm_has_tsc_control;
1324 /* maximum supported tsc_khz for guests */
1325 extern u32  kvm_max_guest_tsc_khz;
1326 /* number of bits of the fractional part of the TSC scaling ratio */
1327 extern u8   kvm_tsc_scaling_ratio_frac_bits;
1328 /* maximum allowed value of TSC scaling ratio */
1329 extern u64  kvm_max_tsc_scaling_ratio;
1330 /* 1ull << kvm_tsc_scaling_ratio_frac_bits */
1331 extern u64  kvm_default_tsc_scaling_ratio;
1332 
1333 extern u64 kvm_mce_cap_supported;
1334 
1335 /*
1336  * EMULTYPE_NO_DECODE - Set when re-emulating an instruction (after completing
1337  *			userspace I/O) to indicate that the emulation context
1338  *			should be resued as is, i.e. skip initialization of
1339  *			emulation context, instruction fetch and decode.
1340  *
1341  * EMULTYPE_TRAP_UD - Set when emulating an intercepted #UD from hardware.
1342  *		      Indicates that only select instructions (tagged with
1343  *		      EmulateOnUD) should be emulated (to minimize the emulator
1344  *		      attack surface).  See also EMULTYPE_TRAP_UD_FORCED.
1345  *
1346  * EMULTYPE_SKIP - Set when emulating solely to skip an instruction, i.e. to
1347  *		   decode the instruction length.  For use *only* by
1348  *		   kvm_x86_ops->skip_emulated_instruction() implementations.
1349  *
1350  * EMULTYPE_ALLOW_RETRY - Set when the emulator should resume the guest to
1351  *			  retry native execution under certain conditions.
1352  *
1353  * EMULTYPE_TRAP_UD_FORCED - Set when emulating an intercepted #UD that was
1354  *			     triggered by KVM's magic "force emulation" prefix,
1355  *			     which is opt in via module param (off by default).
1356  *			     Bypasses EmulateOnUD restriction despite emulating
1357  *			     due to an intercepted #UD (see EMULTYPE_TRAP_UD).
1358  *			     Used to test the full emulator from userspace.
1359  *
1360  * EMULTYPE_VMWARE_GP - Set when emulating an intercepted #GP for VMware
1361  *			backdoor emulation, which is opt in via module param.
1362  *			VMware backoor emulation handles select instructions
1363  *			and reinjects the #GP for all other cases.
1364  */
1365 #define EMULTYPE_NO_DECODE	    (1 << 0)
1366 #define EMULTYPE_TRAP_UD	    (1 << 1)
1367 #define EMULTYPE_SKIP		    (1 << 2)
1368 #define EMULTYPE_ALLOW_RETRY	    (1 << 3)
1369 #define EMULTYPE_TRAP_UD_FORCED	    (1 << 4)
1370 #define EMULTYPE_VMWARE_GP	    (1 << 5)
1371 int kvm_emulate_instruction(struct kvm_vcpu *vcpu, int emulation_type);
1372 int kvm_emulate_instruction_from_buffer(struct kvm_vcpu *vcpu,
1373 					void *insn, int insn_len);
1374 
1375 void kvm_enable_efer_bits(u64);
1376 bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
1377 int __kvm_get_msr(struct kvm_vcpu *vcpu, u32 index, u64 *data, bool host_initiated);
1378 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 index, u64 *data);
1379 int kvm_set_msr(struct kvm_vcpu *vcpu, u32 index, u64 data);
1380 int kvm_emulate_rdmsr(struct kvm_vcpu *vcpu);
1381 int kvm_emulate_wrmsr(struct kvm_vcpu *vcpu);
1382 
1383 struct x86_emulate_ctxt;
1384 
1385 int kvm_fast_pio(struct kvm_vcpu *vcpu, int size, unsigned short port, int in);
1386 int kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
1387 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
1388 int kvm_vcpu_halt(struct kvm_vcpu *vcpu);
1389 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
1390 
1391 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
1392 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
1393 void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
1394 
1395 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
1396 		    int reason, bool has_error_code, u32 error_code);
1397 
1398 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
1399 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
1400 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
1401 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
1402 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
1403 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
1404 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
1405 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
1406 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
1407 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
1408 
1409 int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
1410 int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
1411 
1412 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
1413 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
1414 bool kvm_rdpmc(struct kvm_vcpu *vcpu);
1415 
1416 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1417 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
1418 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1419 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
1420 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
1421 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
1422 			    gfn_t gfn, void *data, int offset, int len,
1423 			    u32 access);
1424 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
1425 bool kvm_require_dr(struct kvm_vcpu *vcpu, int dr);
1426 
1427 static inline int __kvm_irq_line_state(unsigned long *irq_state,
1428 				       int irq_source_id, int level)
1429 {
1430 	/* Logical OR for level trig interrupt */
1431 	if (level)
1432 		__set_bit(irq_source_id, irq_state);
1433 	else
1434 		__clear_bit(irq_source_id, irq_state);
1435 
1436 	return !!(*irq_state);
1437 }
1438 
1439 #define KVM_MMU_ROOT_CURRENT		BIT(0)
1440 #define KVM_MMU_ROOT_PREVIOUS(i)	BIT(1+i)
1441 #define KVM_MMU_ROOTS_ALL		(~0UL)
1442 
1443 int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
1444 void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
1445 
1446 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
1447 
1448 int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
1449 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
1450 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
1451 int kvm_mmu_load(struct kvm_vcpu *vcpu);
1452 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
1453 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
1454 void kvm_mmu_free_roots(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
1455 			ulong roots_to_free);
1456 gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1457 			   struct x86_exception *exception);
1458 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
1459 			      struct x86_exception *exception);
1460 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
1461 			       struct x86_exception *exception);
1462 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
1463 			       struct x86_exception *exception);
1464 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
1465 				struct x86_exception *exception);
1466 
1467 void kvm_vcpu_deactivate_apicv(struct kvm_vcpu *vcpu);
1468 
1469 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
1470 
1471 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u64 error_code,
1472 		       void *insn, int insn_len);
1473 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
1474 void kvm_mmu_invpcid_gva(struct kvm_vcpu *vcpu, gva_t gva, unsigned long pcid);
1475 void kvm_mmu_new_cr3(struct kvm_vcpu *vcpu, gpa_t new_cr3, bool skip_tlb_flush);
1476 
1477 void kvm_enable_tdp(void);
1478 void kvm_disable_tdp(void);
1479 
1480 static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1481 				  struct x86_exception *exception)
1482 {
1483 	return gpa;
1484 }
1485 
1486 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
1487 {
1488 	struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
1489 
1490 	return (struct kvm_mmu_page *)page_private(page);
1491 }
1492 
1493 static inline u16 kvm_read_ldt(void)
1494 {
1495 	u16 ldt;
1496 	asm("sldt %0" : "=g"(ldt));
1497 	return ldt;
1498 }
1499 
1500 static inline void kvm_load_ldt(u16 sel)
1501 {
1502 	asm("lldt %0" : : "rm"(sel));
1503 }
1504 
1505 #ifdef CONFIG_X86_64
1506 static inline unsigned long read_msr(unsigned long msr)
1507 {
1508 	u64 value;
1509 
1510 	rdmsrl(msr, value);
1511 	return value;
1512 }
1513 #endif
1514 
1515 static inline u32 get_rdx_init_val(void)
1516 {
1517 	return 0x600; /* P6 family */
1518 }
1519 
1520 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
1521 {
1522 	kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
1523 }
1524 
1525 #define TSS_IOPB_BASE_OFFSET 0x66
1526 #define TSS_BASE_SIZE 0x68
1527 #define TSS_IOPB_SIZE (65536 / 8)
1528 #define TSS_REDIRECTION_SIZE (256 / 8)
1529 #define RMODE_TSS_SIZE							\
1530 	(TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
1531 
1532 enum {
1533 	TASK_SWITCH_CALL = 0,
1534 	TASK_SWITCH_IRET = 1,
1535 	TASK_SWITCH_JMP = 2,
1536 	TASK_SWITCH_GATE = 3,
1537 };
1538 
1539 #define HF_GIF_MASK		(1 << 0)
1540 #define HF_HIF_MASK		(1 << 1)
1541 #define HF_VINTR_MASK		(1 << 2)
1542 #define HF_NMI_MASK		(1 << 3)
1543 #define HF_IRET_MASK		(1 << 4)
1544 #define HF_GUEST_MASK		(1 << 5) /* VCPU is in guest-mode */
1545 #define HF_SMM_MASK		(1 << 6)
1546 #define HF_SMM_INSIDE_NMI_MASK	(1 << 7)
1547 
1548 #define __KVM_VCPU_MULTIPLE_ADDRESS_SPACE
1549 #define KVM_ADDRESS_SPACE_NUM 2
1550 
1551 #define kvm_arch_vcpu_memslots_id(vcpu) ((vcpu)->arch.hflags & HF_SMM_MASK ? 1 : 0)
1552 #define kvm_memslots_for_spte_role(kvm, role) __kvm_memslots(kvm, (role).smm)
1553 
1554 asmlinkage void kvm_spurious_fault(void);
1555 
1556 /*
1557  * Hardware virtualization extension instructions may fault if a
1558  * reboot turns off virtualization while processes are running.
1559  * Usually after catching the fault we just panic; during reboot
1560  * instead the instruction is ignored.
1561  */
1562 #define __kvm_handle_fault_on_reboot(insn)				\
1563 	"666: \n\t"							\
1564 	insn "\n\t"							\
1565 	"jmp	668f \n\t"						\
1566 	"667: \n\t"							\
1567 	"call	kvm_spurious_fault \n\t"				\
1568 	"668: \n\t"							\
1569 	_ASM_EXTABLE(666b, 667b)
1570 
1571 #define KVM_ARCH_WANT_MMU_NOTIFIER
1572 int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
1573 int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
1574 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
1575 int kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
1576 int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
1577 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1578 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
1579 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
1580 void kvm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event);
1581 void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu);
1582 
1583 int kvm_pv_send_ipi(struct kvm *kvm, unsigned long ipi_bitmap_low,
1584 		    unsigned long ipi_bitmap_high, u32 min,
1585 		    unsigned long icr, int op_64_bit);
1586 
1587 void kvm_define_shared_msr(unsigned index, u32 msr);
1588 int kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
1589 
1590 u64 kvm_scale_tsc(struct kvm_vcpu *vcpu, u64 tsc);
1591 u64 kvm_read_l1_tsc(struct kvm_vcpu *vcpu, u64 host_tsc);
1592 
1593 unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
1594 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1595 
1596 void kvm_make_mclock_inprogress_request(struct kvm *kvm);
1597 void kvm_make_scan_ioapic_request(struct kvm *kvm);
1598 void kvm_make_scan_ioapic_request_mask(struct kvm *kvm,
1599 				       unsigned long *vcpu_bitmap);
1600 
1601 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1602 				     struct kvm_async_pf *work);
1603 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1604 				 struct kvm_async_pf *work);
1605 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1606 			       struct kvm_async_pf *work);
1607 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
1608 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1609 
1610 int kvm_skip_emulated_instruction(struct kvm_vcpu *vcpu);
1611 int kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1612 void __kvm_request_immediate_exit(struct kvm_vcpu *vcpu);
1613 
1614 int kvm_is_in_guest(void);
1615 
1616 int __x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa, u32 size);
1617 int x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa, u32 size);
1618 bool kvm_vcpu_is_reset_bsp(struct kvm_vcpu *vcpu);
1619 bool kvm_vcpu_is_bsp(struct kvm_vcpu *vcpu);
1620 
1621 bool kvm_intr_is_single_vcpu(struct kvm *kvm, struct kvm_lapic_irq *irq,
1622 			     struct kvm_vcpu **dest_vcpu);
1623 
1624 void kvm_set_msi_irq(struct kvm *kvm, struct kvm_kernel_irq_routing_entry *e,
1625 		     struct kvm_lapic_irq *irq);
1626 
1627 static inline bool kvm_irq_is_postable(struct kvm_lapic_irq *irq)
1628 {
1629 	/* We can only post Fixed and LowPrio IRQs */
1630 	return (irq->delivery_mode == dest_Fixed ||
1631 		irq->delivery_mode == dest_LowestPrio);
1632 }
1633 
1634 static inline void kvm_arch_vcpu_blocking(struct kvm_vcpu *vcpu)
1635 {
1636 	if (kvm_x86_ops->vcpu_blocking)
1637 		kvm_x86_ops->vcpu_blocking(vcpu);
1638 }
1639 
1640 static inline void kvm_arch_vcpu_unblocking(struct kvm_vcpu *vcpu)
1641 {
1642 	if (kvm_x86_ops->vcpu_unblocking)
1643 		kvm_x86_ops->vcpu_unblocking(vcpu);
1644 }
1645 
1646 static inline void kvm_arch_vcpu_block_finish(struct kvm_vcpu *vcpu) {}
1647 
1648 static inline int kvm_cpu_get_apicid(int mps_cpu)
1649 {
1650 #ifdef CONFIG_X86_LOCAL_APIC
1651 	return default_cpu_present_to_apicid(mps_cpu);
1652 #else
1653 	WARN_ON_ONCE(1);
1654 	return BAD_APICID;
1655 #endif
1656 }
1657 
1658 #define put_smstate(type, buf, offset, val)                      \
1659 	*(type *)((buf) + (offset) - 0x7e00) = val
1660 
1661 #define GET_SMSTATE(type, buf, offset)		\
1662 	(*(type *)((buf) + (offset) - 0x7e00))
1663 
1664 #endif /* _ASM_X86_KVM_HOST_H */
1665