xref: /openbmc/linux/arch/x86/include/asm/kvm_host.h (revision 55fd7e02)
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 #define KVM_DIRTY_LOG_MANUAL_CAPS   (KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE | \
53 					KVM_DIRTY_LOG_INITIALLY_SET)
54 
55 /* x86-specific vcpu->requests bit members */
56 #define KVM_REQ_MIGRATE_TIMER		KVM_ARCH_REQ(0)
57 #define KVM_REQ_REPORT_TPR_ACCESS	KVM_ARCH_REQ(1)
58 #define KVM_REQ_TRIPLE_FAULT		KVM_ARCH_REQ(2)
59 #define KVM_REQ_MMU_SYNC		KVM_ARCH_REQ(3)
60 #define KVM_REQ_CLOCK_UPDATE		KVM_ARCH_REQ(4)
61 #define KVM_REQ_LOAD_MMU_PGD		KVM_ARCH_REQ(5)
62 #define KVM_REQ_EVENT			KVM_ARCH_REQ(6)
63 #define KVM_REQ_APF_HALT		KVM_ARCH_REQ(7)
64 #define KVM_REQ_STEAL_UPDATE		KVM_ARCH_REQ(8)
65 #define KVM_REQ_NMI			KVM_ARCH_REQ(9)
66 #define KVM_REQ_PMU			KVM_ARCH_REQ(10)
67 #define KVM_REQ_PMI			KVM_ARCH_REQ(11)
68 #define KVM_REQ_SMI			KVM_ARCH_REQ(12)
69 #define KVM_REQ_MASTERCLOCK_UPDATE	KVM_ARCH_REQ(13)
70 #define KVM_REQ_MCLOCK_INPROGRESS \
71 	KVM_ARCH_REQ_FLAGS(14, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
72 #define KVM_REQ_SCAN_IOAPIC \
73 	KVM_ARCH_REQ_FLAGS(15, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
74 #define KVM_REQ_GLOBAL_CLOCK_UPDATE	KVM_ARCH_REQ(16)
75 #define KVM_REQ_APIC_PAGE_RELOAD \
76 	KVM_ARCH_REQ_FLAGS(17, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
77 #define KVM_REQ_HV_CRASH		KVM_ARCH_REQ(18)
78 #define KVM_REQ_IOAPIC_EOI_EXIT		KVM_ARCH_REQ(19)
79 #define KVM_REQ_HV_RESET		KVM_ARCH_REQ(20)
80 #define KVM_REQ_HV_EXIT			KVM_ARCH_REQ(21)
81 #define KVM_REQ_HV_STIMER		KVM_ARCH_REQ(22)
82 #define KVM_REQ_LOAD_EOI_EXITMAP	KVM_ARCH_REQ(23)
83 #define KVM_REQ_GET_VMCS12_PAGES	KVM_ARCH_REQ(24)
84 #define KVM_REQ_APICV_UPDATE \
85 	KVM_ARCH_REQ_FLAGS(25, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
86 #define KVM_REQ_TLB_FLUSH_CURRENT	KVM_ARCH_REQ(26)
87 #define KVM_REQ_HV_TLB_FLUSH \
88 	KVM_ARCH_REQ_FLAGS(27, KVM_REQUEST_NO_WAKEUP)
89 #define KVM_REQ_APF_READY		KVM_ARCH_REQ(28)
90 
91 #define CR0_RESERVED_BITS                                               \
92 	(~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
93 			  | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
94 			  | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
95 
96 #define CR4_RESERVED_BITS                                               \
97 	(~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
98 			  | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE     \
99 			  | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR | X86_CR4_PCIDE \
100 			  | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_FSGSBASE \
101 			  | X86_CR4_OSXMMEXCPT | X86_CR4_LA57 | X86_CR4_VMXE \
102 			  | X86_CR4_SMAP | X86_CR4_PKE | X86_CR4_UMIP))
103 
104 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
105 
106 
107 
108 #define INVALID_PAGE (~(hpa_t)0)
109 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
110 
111 #define UNMAPPED_GVA (~(gpa_t)0)
112 
113 /* KVM Hugepage definitions for x86 */
114 #define KVM_MAX_HUGEPAGE_LEVEL	PG_LEVEL_1G
115 #define KVM_NR_PAGE_SIZES	(KVM_MAX_HUGEPAGE_LEVEL - PG_LEVEL_4K + 1)
116 #define KVM_HPAGE_GFN_SHIFT(x)	(((x) - 1) * 9)
117 #define KVM_HPAGE_SHIFT(x)	(PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
118 #define KVM_HPAGE_SIZE(x)	(1UL << KVM_HPAGE_SHIFT(x))
119 #define KVM_HPAGE_MASK(x)	(~(KVM_HPAGE_SIZE(x) - 1))
120 #define KVM_PAGES_PER_HPAGE(x)	(KVM_HPAGE_SIZE(x) / PAGE_SIZE)
121 
122 static inline gfn_t gfn_to_index(gfn_t gfn, gfn_t base_gfn, int level)
123 {
124 	/* KVM_HPAGE_GFN_SHIFT(PG_LEVEL_4K) must be 0. */
125 	return (gfn >> KVM_HPAGE_GFN_SHIFT(level)) -
126 		(base_gfn >> KVM_HPAGE_GFN_SHIFT(level));
127 }
128 
129 #define KVM_PERMILLE_MMU_PAGES 20
130 #define KVM_MIN_ALLOC_MMU_PAGES 64UL
131 #define KVM_MMU_HASH_SHIFT 12
132 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
133 #define KVM_MIN_FREE_MMU_PAGES 5
134 #define KVM_REFILL_PAGES 25
135 #define KVM_MAX_CPUID_ENTRIES 80
136 #define KVM_NR_FIXED_MTRR_REGION 88
137 #define KVM_NR_VAR_MTRR 8
138 
139 #define ASYNC_PF_PER_VCPU 64
140 
141 enum kvm_reg {
142 	VCPU_REGS_RAX = __VCPU_REGS_RAX,
143 	VCPU_REGS_RCX = __VCPU_REGS_RCX,
144 	VCPU_REGS_RDX = __VCPU_REGS_RDX,
145 	VCPU_REGS_RBX = __VCPU_REGS_RBX,
146 	VCPU_REGS_RSP = __VCPU_REGS_RSP,
147 	VCPU_REGS_RBP = __VCPU_REGS_RBP,
148 	VCPU_REGS_RSI = __VCPU_REGS_RSI,
149 	VCPU_REGS_RDI = __VCPU_REGS_RDI,
150 #ifdef CONFIG_X86_64
151 	VCPU_REGS_R8  = __VCPU_REGS_R8,
152 	VCPU_REGS_R9  = __VCPU_REGS_R9,
153 	VCPU_REGS_R10 = __VCPU_REGS_R10,
154 	VCPU_REGS_R11 = __VCPU_REGS_R11,
155 	VCPU_REGS_R12 = __VCPU_REGS_R12,
156 	VCPU_REGS_R13 = __VCPU_REGS_R13,
157 	VCPU_REGS_R14 = __VCPU_REGS_R14,
158 	VCPU_REGS_R15 = __VCPU_REGS_R15,
159 #endif
160 	VCPU_REGS_RIP,
161 	NR_VCPU_REGS,
162 
163 	VCPU_EXREG_PDPTR = NR_VCPU_REGS,
164 	VCPU_EXREG_CR0,
165 	VCPU_EXREG_CR3,
166 	VCPU_EXREG_CR4,
167 	VCPU_EXREG_RFLAGS,
168 	VCPU_EXREG_SEGMENTS,
169 	VCPU_EXREG_EXIT_INFO_1,
170 	VCPU_EXREG_EXIT_INFO_2,
171 };
172 
173 enum {
174 	VCPU_SREG_ES,
175 	VCPU_SREG_CS,
176 	VCPU_SREG_SS,
177 	VCPU_SREG_DS,
178 	VCPU_SREG_FS,
179 	VCPU_SREG_GS,
180 	VCPU_SREG_TR,
181 	VCPU_SREG_LDTR,
182 };
183 
184 enum exit_fastpath_completion {
185 	EXIT_FASTPATH_NONE,
186 	EXIT_FASTPATH_REENTER_GUEST,
187 	EXIT_FASTPATH_EXIT_HANDLED,
188 };
189 typedef enum exit_fastpath_completion fastpath_t;
190 
191 struct x86_emulate_ctxt;
192 struct x86_exception;
193 enum x86_intercept;
194 enum x86_intercept_stage;
195 
196 #define KVM_NR_MEM_OBJS 40
197 
198 #define KVM_NR_DB_REGS	4
199 
200 #define DR6_BD		(1 << 13)
201 #define DR6_BS		(1 << 14)
202 #define DR6_BT		(1 << 15)
203 #define DR6_RTM		(1 << 16)
204 #define DR6_FIXED_1	0xfffe0ff0
205 #define DR6_INIT	0xffff0ff0
206 #define DR6_VOLATILE	0x0001e00f
207 
208 #define DR7_BP_EN_MASK	0x000000ff
209 #define DR7_GE		(1 << 9)
210 #define DR7_GD		(1 << 13)
211 #define DR7_FIXED_1	0x00000400
212 #define DR7_VOLATILE	0xffff2bff
213 
214 #define PFERR_PRESENT_BIT 0
215 #define PFERR_WRITE_BIT 1
216 #define PFERR_USER_BIT 2
217 #define PFERR_RSVD_BIT 3
218 #define PFERR_FETCH_BIT 4
219 #define PFERR_PK_BIT 5
220 #define PFERR_GUEST_FINAL_BIT 32
221 #define PFERR_GUEST_PAGE_BIT 33
222 
223 #define PFERR_PRESENT_MASK (1U << PFERR_PRESENT_BIT)
224 #define PFERR_WRITE_MASK (1U << PFERR_WRITE_BIT)
225 #define PFERR_USER_MASK (1U << PFERR_USER_BIT)
226 #define PFERR_RSVD_MASK (1U << PFERR_RSVD_BIT)
227 #define PFERR_FETCH_MASK (1U << PFERR_FETCH_BIT)
228 #define PFERR_PK_MASK (1U << PFERR_PK_BIT)
229 #define PFERR_GUEST_FINAL_MASK (1ULL << PFERR_GUEST_FINAL_BIT)
230 #define PFERR_GUEST_PAGE_MASK (1ULL << PFERR_GUEST_PAGE_BIT)
231 
232 #define PFERR_NESTED_GUEST_PAGE (PFERR_GUEST_PAGE_MASK |	\
233 				 PFERR_WRITE_MASK |		\
234 				 PFERR_PRESENT_MASK)
235 
236 /* apic attention bits */
237 #define KVM_APIC_CHECK_VAPIC	0
238 /*
239  * The following bit is set with PV-EOI, unset on EOI.
240  * We detect PV-EOI changes by guest by comparing
241  * this bit with PV-EOI in guest memory.
242  * See the implementation in apic_update_pv_eoi.
243  */
244 #define KVM_APIC_PV_EOI_PENDING	1
245 
246 struct kvm_kernel_irq_routing_entry;
247 
248 /*
249  * We don't want allocation failures within the mmu code, so we preallocate
250  * enough memory for a single page fault in a cache.
251  */
252 struct kvm_mmu_memory_cache {
253 	int nobjs;
254 	void *objects[KVM_NR_MEM_OBJS];
255 };
256 
257 /*
258  * the pages used as guest page table on soft mmu are tracked by
259  * kvm_memory_slot.arch.gfn_track which is 16 bits, so the role bits used
260  * by indirect shadow page can not be more than 15 bits.
261  *
262  * Currently, we used 14 bits that are @level, @gpte_is_8_bytes, @quadrant, @access,
263  * @nxe, @cr0_wp, @smep_andnot_wp and @smap_andnot_wp.
264  */
265 union kvm_mmu_page_role {
266 	u32 word;
267 	struct {
268 		unsigned level:4;
269 		unsigned gpte_is_8_bytes:1;
270 		unsigned quadrant:2;
271 		unsigned direct:1;
272 		unsigned access:3;
273 		unsigned invalid:1;
274 		unsigned nxe:1;
275 		unsigned cr0_wp:1;
276 		unsigned smep_andnot_wp:1;
277 		unsigned smap_andnot_wp:1;
278 		unsigned ad_disabled:1;
279 		unsigned guest_mode:1;
280 		unsigned :6;
281 
282 		/*
283 		 * This is left at the top of the word so that
284 		 * kvm_memslots_for_spte_role can extract it with a
285 		 * simple shift.  While there is room, give it a whole
286 		 * byte so it is also faster to load it from memory.
287 		 */
288 		unsigned smm:8;
289 	};
290 };
291 
292 union kvm_mmu_extended_role {
293 /*
294  * This structure complements kvm_mmu_page_role caching everything needed for
295  * MMU configuration. If nothing in both these structures changed, MMU
296  * re-configuration can be skipped. @valid bit is set on first usage so we don't
297  * treat all-zero structure as valid data.
298  */
299 	u32 word;
300 	struct {
301 		unsigned int valid:1;
302 		unsigned int execonly:1;
303 		unsigned int cr0_pg:1;
304 		unsigned int cr4_pae:1;
305 		unsigned int cr4_pse:1;
306 		unsigned int cr4_pke:1;
307 		unsigned int cr4_smap:1;
308 		unsigned int cr4_smep:1;
309 		unsigned int maxphyaddr:6;
310 	};
311 };
312 
313 union kvm_mmu_role {
314 	u64 as_u64;
315 	struct {
316 		union kvm_mmu_page_role base;
317 		union kvm_mmu_extended_role ext;
318 	};
319 };
320 
321 struct kvm_rmap_head {
322 	unsigned long val;
323 };
324 
325 struct kvm_mmu_page {
326 	struct list_head link;
327 	struct hlist_node hash_link;
328 	struct list_head lpage_disallowed_link;
329 
330 	bool unsync;
331 	u8 mmu_valid_gen;
332 	bool mmio_cached;
333 	bool lpage_disallowed; /* Can't be replaced by an equiv large page */
334 
335 	/*
336 	 * The following two entries are used to key the shadow page in the
337 	 * hash table.
338 	 */
339 	union kvm_mmu_page_role role;
340 	gfn_t gfn;
341 
342 	u64 *spt;
343 	/* hold the gfn of each spte inside spt */
344 	gfn_t *gfns;
345 	int root_count;          /* Currently serving as active root */
346 	unsigned int unsync_children;
347 	struct kvm_rmap_head parent_ptes; /* rmap pointers to parent sptes */
348 	DECLARE_BITMAP(unsync_child_bitmap, 512);
349 
350 #ifdef CONFIG_X86_32
351 	/*
352 	 * Used out of the mmu-lock to avoid reading spte values while an
353 	 * update is in progress; see the comments in __get_spte_lockless().
354 	 */
355 	int clear_spte_count;
356 #endif
357 
358 	/* Number of writes since the last time traversal visited this page.  */
359 	atomic_t write_flooding_count;
360 };
361 
362 struct kvm_pio_request {
363 	unsigned long linear_rip;
364 	unsigned long count;
365 	int in;
366 	int port;
367 	int size;
368 };
369 
370 #define PT64_ROOT_MAX_LEVEL 5
371 
372 struct rsvd_bits_validate {
373 	u64 rsvd_bits_mask[2][PT64_ROOT_MAX_LEVEL];
374 	u64 bad_mt_xwr;
375 };
376 
377 struct kvm_mmu_root_info {
378 	gpa_t pgd;
379 	hpa_t hpa;
380 };
381 
382 #define KVM_MMU_ROOT_INFO_INVALID \
383 	((struct kvm_mmu_root_info) { .pgd = INVALID_PAGE, .hpa = INVALID_PAGE })
384 
385 #define KVM_MMU_NUM_PREV_ROOTS 3
386 
387 /*
388  * x86 supports 4 paging modes (5-level 64-bit, 4-level 64-bit, 3-level 32-bit,
389  * and 2-level 32-bit).  The kvm_mmu structure abstracts the details of the
390  * current mmu mode.
391  */
392 struct kvm_mmu {
393 	unsigned long (*get_guest_pgd)(struct kvm_vcpu *vcpu);
394 	u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
395 	int (*page_fault)(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa, u32 err,
396 			  bool prefault);
397 	void (*inject_page_fault)(struct kvm_vcpu *vcpu,
398 				  struct x86_exception *fault);
399 	gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gpa_t gva_or_gpa,
400 			    u32 access, struct x86_exception *exception);
401 	gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
402 			       struct x86_exception *exception);
403 	int (*sync_page)(struct kvm_vcpu *vcpu,
404 			 struct kvm_mmu_page *sp);
405 	void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva, hpa_t root_hpa);
406 	void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
407 			   u64 *spte, const void *pte);
408 	hpa_t root_hpa;
409 	gpa_t root_pgd;
410 	union kvm_mmu_role mmu_role;
411 	u8 root_level;
412 	u8 shadow_root_level;
413 	u8 ept_ad;
414 	bool direct_map;
415 	struct kvm_mmu_root_info prev_roots[KVM_MMU_NUM_PREV_ROOTS];
416 
417 	/*
418 	 * Bitmap; bit set = permission fault
419 	 * Byte index: page fault error code [4:1]
420 	 * Bit index: pte permissions in ACC_* format
421 	 */
422 	u8 permissions[16];
423 
424 	/*
425 	* The pkru_mask indicates if protection key checks are needed.  It
426 	* consists of 16 domains indexed by page fault error code bits [4:1],
427 	* with PFEC.RSVD replaced by ACC_USER_MASK from the page tables.
428 	* Each domain has 2 bits which are ANDed with AD and WD from PKRU.
429 	*/
430 	u32 pkru_mask;
431 
432 	u64 *pae_root;
433 	u64 *lm_root;
434 
435 	/*
436 	 * check zero bits on shadow page table entries, these
437 	 * bits include not only hardware reserved bits but also
438 	 * the bits spte never used.
439 	 */
440 	struct rsvd_bits_validate shadow_zero_check;
441 
442 	struct rsvd_bits_validate guest_rsvd_check;
443 
444 	/* Can have large pages at levels 2..last_nonleaf_level-1. */
445 	u8 last_nonleaf_level;
446 
447 	bool nx;
448 
449 	u64 pdptrs[4]; /* pae */
450 };
451 
452 struct kvm_tlb_range {
453 	u64 start_gfn;
454 	u64 pages;
455 };
456 
457 enum pmc_type {
458 	KVM_PMC_GP = 0,
459 	KVM_PMC_FIXED,
460 };
461 
462 struct kvm_pmc {
463 	enum pmc_type type;
464 	u8 idx;
465 	u64 counter;
466 	u64 eventsel;
467 	struct perf_event *perf_event;
468 	struct kvm_vcpu *vcpu;
469 	/*
470 	 * eventsel value for general purpose counters,
471 	 * ctrl value for fixed counters.
472 	 */
473 	u64 current_config;
474 };
475 
476 struct kvm_pmu {
477 	unsigned nr_arch_gp_counters;
478 	unsigned nr_arch_fixed_counters;
479 	unsigned available_event_types;
480 	u64 fixed_ctr_ctrl;
481 	u64 global_ctrl;
482 	u64 global_status;
483 	u64 global_ovf_ctrl;
484 	u64 counter_bitmask[2];
485 	u64 global_ctrl_mask;
486 	u64 global_ovf_ctrl_mask;
487 	u64 reserved_bits;
488 	u8 version;
489 	struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
490 	struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
491 	struct irq_work irq_work;
492 	DECLARE_BITMAP(reprogram_pmi, X86_PMC_IDX_MAX);
493 	DECLARE_BITMAP(all_valid_pmc_idx, X86_PMC_IDX_MAX);
494 	DECLARE_BITMAP(pmc_in_use, X86_PMC_IDX_MAX);
495 
496 	/*
497 	 * The gate to release perf_events not marked in
498 	 * pmc_in_use only once in a vcpu time slice.
499 	 */
500 	bool need_cleanup;
501 
502 	/*
503 	 * The total number of programmed perf_events and it helps to avoid
504 	 * redundant check before cleanup if guest don't use vPMU at all.
505 	 */
506 	u8 event_count;
507 };
508 
509 struct kvm_pmu_ops;
510 
511 enum {
512 	KVM_DEBUGREG_BP_ENABLED = 1,
513 	KVM_DEBUGREG_WONT_EXIT = 2,
514 	KVM_DEBUGREG_RELOAD = 4,
515 };
516 
517 struct kvm_mtrr_range {
518 	u64 base;
519 	u64 mask;
520 	struct list_head node;
521 };
522 
523 struct kvm_mtrr {
524 	struct kvm_mtrr_range var_ranges[KVM_NR_VAR_MTRR];
525 	mtrr_type fixed_ranges[KVM_NR_FIXED_MTRR_REGION];
526 	u64 deftype;
527 
528 	struct list_head head;
529 };
530 
531 /* Hyper-V SynIC timer */
532 struct kvm_vcpu_hv_stimer {
533 	struct hrtimer timer;
534 	int index;
535 	union hv_stimer_config config;
536 	u64 count;
537 	u64 exp_time;
538 	struct hv_message msg;
539 	bool msg_pending;
540 };
541 
542 /* Hyper-V synthetic interrupt controller (SynIC)*/
543 struct kvm_vcpu_hv_synic {
544 	u64 version;
545 	u64 control;
546 	u64 msg_page;
547 	u64 evt_page;
548 	atomic64_t sint[HV_SYNIC_SINT_COUNT];
549 	atomic_t sint_to_gsi[HV_SYNIC_SINT_COUNT];
550 	DECLARE_BITMAP(auto_eoi_bitmap, 256);
551 	DECLARE_BITMAP(vec_bitmap, 256);
552 	bool active;
553 	bool dont_zero_synic_pages;
554 };
555 
556 /* Hyper-V per vcpu emulation context */
557 struct kvm_vcpu_hv {
558 	u32 vp_index;
559 	u64 hv_vapic;
560 	s64 runtime_offset;
561 	struct kvm_vcpu_hv_synic synic;
562 	struct kvm_hyperv_exit exit;
563 	struct kvm_vcpu_hv_stimer stimer[HV_SYNIC_STIMER_COUNT];
564 	DECLARE_BITMAP(stimer_pending_bitmap, HV_SYNIC_STIMER_COUNT);
565 	cpumask_t tlb_flush;
566 };
567 
568 struct kvm_vcpu_arch {
569 	/*
570 	 * rip and regs accesses must go through
571 	 * kvm_{register,rip}_{read,write} functions.
572 	 */
573 	unsigned long regs[NR_VCPU_REGS];
574 	u32 regs_avail;
575 	u32 regs_dirty;
576 
577 	unsigned long cr0;
578 	unsigned long cr0_guest_owned_bits;
579 	unsigned long cr2;
580 	unsigned long cr3;
581 	unsigned long cr4;
582 	unsigned long cr4_guest_owned_bits;
583 	unsigned long cr8;
584 	u32 host_pkru;
585 	u32 pkru;
586 	u32 hflags;
587 	u64 efer;
588 	u64 apic_base;
589 	struct kvm_lapic *apic;    /* kernel irqchip context */
590 	bool apicv_active;
591 	bool load_eoi_exitmap_pending;
592 	DECLARE_BITMAP(ioapic_handled_vectors, 256);
593 	unsigned long apic_attention;
594 	int32_t apic_arb_prio;
595 	int mp_state;
596 	u64 ia32_misc_enable_msr;
597 	u64 smbase;
598 	u64 smi_count;
599 	bool tpr_access_reporting;
600 	bool xsaves_enabled;
601 	u64 ia32_xss;
602 	u64 microcode_version;
603 	u64 arch_capabilities;
604 	u64 perf_capabilities;
605 
606 	/*
607 	 * Paging state of the vcpu
608 	 *
609 	 * If the vcpu runs in guest mode with two level paging this still saves
610 	 * the paging mode of the l1 guest. This context is always used to
611 	 * handle faults.
612 	 */
613 	struct kvm_mmu *mmu;
614 
615 	/* Non-nested MMU for L1 */
616 	struct kvm_mmu root_mmu;
617 
618 	/* L1 MMU when running nested */
619 	struct kvm_mmu guest_mmu;
620 
621 	/*
622 	 * Paging state of an L2 guest (used for nested npt)
623 	 *
624 	 * This context will save all necessary information to walk page tables
625 	 * of an L2 guest. This context is only initialized for page table
626 	 * walking and not for faulting since we never handle l2 page faults on
627 	 * the host.
628 	 */
629 	struct kvm_mmu nested_mmu;
630 
631 	/*
632 	 * Pointer to the mmu context currently used for
633 	 * gva_to_gpa translations.
634 	 */
635 	struct kvm_mmu *walk_mmu;
636 
637 	struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
638 	struct kvm_mmu_memory_cache mmu_page_cache;
639 	struct kvm_mmu_memory_cache mmu_page_header_cache;
640 
641 	/*
642 	 * QEMU userspace and the guest each have their own FPU state.
643 	 * In vcpu_run, we switch between the user and guest FPU contexts.
644 	 * While running a VCPU, the VCPU thread will have the guest FPU
645 	 * context.
646 	 *
647 	 * Note that while the PKRU state lives inside the fpu registers,
648 	 * it is switched out separately at VMENTER and VMEXIT time. The
649 	 * "guest_fpu" state here contains the guest FPU context, with the
650 	 * host PRKU bits.
651 	 */
652 	struct fpu *user_fpu;
653 	struct fpu *guest_fpu;
654 
655 	u64 xcr0;
656 	u64 guest_supported_xcr0;
657 
658 	struct kvm_pio_request pio;
659 	void *pio_data;
660 
661 	u8 event_exit_inst_len;
662 
663 	struct kvm_queued_exception {
664 		bool pending;
665 		bool injected;
666 		bool has_error_code;
667 		u8 nr;
668 		u32 error_code;
669 		unsigned long payload;
670 		bool has_payload;
671 		u8 nested_apf;
672 	} exception;
673 
674 	struct kvm_queued_interrupt {
675 		bool injected;
676 		bool soft;
677 		u8 nr;
678 	} interrupt;
679 
680 	int halt_request; /* real mode on Intel only */
681 
682 	int cpuid_nent;
683 	struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
684 
685 	int maxphyaddr;
686 	int tdp_level;
687 
688 	/* emulate context */
689 
690 	struct x86_emulate_ctxt *emulate_ctxt;
691 	bool emulate_regs_need_sync_to_vcpu;
692 	bool emulate_regs_need_sync_from_vcpu;
693 	int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
694 
695 	gpa_t time;
696 	struct pvclock_vcpu_time_info hv_clock;
697 	unsigned int hw_tsc_khz;
698 	struct gfn_to_hva_cache pv_time;
699 	bool pv_time_enabled;
700 	/* set guest stopped flag in pvclock flags field */
701 	bool pvclock_set_guest_stopped_request;
702 
703 	struct {
704 		u8 preempted;
705 		u64 msr_val;
706 		u64 last_steal;
707 		struct gfn_to_pfn_cache cache;
708 	} st;
709 
710 	u64 l1_tsc_offset;
711 	u64 tsc_offset;
712 	u64 last_guest_tsc;
713 	u64 last_host_tsc;
714 	u64 tsc_offset_adjustment;
715 	u64 this_tsc_nsec;
716 	u64 this_tsc_write;
717 	u64 this_tsc_generation;
718 	bool tsc_catchup;
719 	bool tsc_always_catchup;
720 	s8 virtual_tsc_shift;
721 	u32 virtual_tsc_mult;
722 	u32 virtual_tsc_khz;
723 	s64 ia32_tsc_adjust_msr;
724 	u64 msr_ia32_power_ctl;
725 	u64 tsc_scaling_ratio;
726 
727 	atomic_t nmi_queued;  /* unprocessed asynchronous NMIs */
728 	unsigned nmi_pending; /* NMI queued after currently running handler */
729 	bool nmi_injected;    /* Trying to inject an NMI this entry */
730 	bool smi_pending;    /* SMI queued after currently running handler */
731 
732 	struct kvm_mtrr mtrr_state;
733 	u64 pat;
734 
735 	unsigned switch_db_regs;
736 	unsigned long db[KVM_NR_DB_REGS];
737 	unsigned long dr6;
738 	unsigned long dr7;
739 	unsigned long eff_db[KVM_NR_DB_REGS];
740 	unsigned long guest_debug_dr7;
741 	u64 msr_platform_info;
742 	u64 msr_misc_features_enables;
743 
744 	u64 mcg_cap;
745 	u64 mcg_status;
746 	u64 mcg_ctl;
747 	u64 mcg_ext_ctl;
748 	u64 *mce_banks;
749 
750 	/* Cache MMIO info */
751 	u64 mmio_gva;
752 	unsigned mmio_access;
753 	gfn_t mmio_gfn;
754 	u64 mmio_gen;
755 
756 	struct kvm_pmu pmu;
757 
758 	/* used for guest single stepping over the given code position */
759 	unsigned long singlestep_rip;
760 
761 	struct kvm_vcpu_hv hyperv;
762 
763 	cpumask_var_t wbinvd_dirty_mask;
764 
765 	unsigned long last_retry_eip;
766 	unsigned long last_retry_addr;
767 
768 	struct {
769 		bool halted;
770 		gfn_t gfns[ASYNC_PF_PER_VCPU];
771 		struct gfn_to_hva_cache data;
772 		u64 msr_en_val; /* MSR_KVM_ASYNC_PF_EN */
773 		u64 msr_int_val; /* MSR_KVM_ASYNC_PF_INT */
774 		u16 vec;
775 		u32 id;
776 		bool send_user_only;
777 		u32 host_apf_flags;
778 		unsigned long nested_apf_token;
779 		bool delivery_as_pf_vmexit;
780 		bool pageready_pending;
781 	} apf;
782 
783 	/* OSVW MSRs (AMD only) */
784 	struct {
785 		u64 length;
786 		u64 status;
787 	} osvw;
788 
789 	struct {
790 		u64 msr_val;
791 		struct gfn_to_hva_cache data;
792 	} pv_eoi;
793 
794 	u64 msr_kvm_poll_control;
795 
796 	/*
797 	 * Indicates the guest is trying to write a gfn that contains one or
798 	 * more of the PTEs used to translate the write itself, i.e. the access
799 	 * is changing its own translation in the guest page tables.  KVM exits
800 	 * to userspace if emulation of the faulting instruction fails and this
801 	 * flag is set, as KVM cannot make forward progress.
802 	 *
803 	 * If emulation fails for a write to guest page tables, KVM unprotects
804 	 * (zaps) the shadow page for the target gfn and resumes the guest to
805 	 * retry the non-emulatable instruction (on hardware).  Unprotecting the
806 	 * gfn doesn't allow forward progress for a self-changing access because
807 	 * doing so also zaps the translation for the gfn, i.e. retrying the
808 	 * instruction will hit a !PRESENT fault, which results in a new shadow
809 	 * page and sends KVM back to square one.
810 	 */
811 	bool write_fault_to_shadow_pgtable;
812 
813 	/* set at EPT violation at this point */
814 	unsigned long exit_qualification;
815 
816 	/* pv related host specific info */
817 	struct {
818 		bool pv_unhalted;
819 	} pv;
820 
821 	int pending_ioapic_eoi;
822 	int pending_external_vector;
823 
824 	/* be preempted when it's in kernel-mode(cpl=0) */
825 	bool preempted_in_kernel;
826 
827 	/* Flush the L1 Data cache for L1TF mitigation on VMENTER */
828 	bool l1tf_flush_l1d;
829 
830 	/* AMD MSRC001_0015 Hardware Configuration */
831 	u64 msr_hwcr;
832 };
833 
834 struct kvm_lpage_info {
835 	int disallow_lpage;
836 };
837 
838 struct kvm_arch_memory_slot {
839 	struct kvm_rmap_head *rmap[KVM_NR_PAGE_SIZES];
840 	struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
841 	unsigned short *gfn_track[KVM_PAGE_TRACK_MAX];
842 };
843 
844 /*
845  * We use as the mode the number of bits allocated in the LDR for the
846  * logical processor ID.  It happens that these are all powers of two.
847  * This makes it is very easy to detect cases where the APICs are
848  * configured for multiple modes; in that case, we cannot use the map and
849  * hence cannot use kvm_irq_delivery_to_apic_fast either.
850  */
851 #define KVM_APIC_MODE_XAPIC_CLUSTER          4
852 #define KVM_APIC_MODE_XAPIC_FLAT             8
853 #define KVM_APIC_MODE_X2APIC                16
854 
855 struct kvm_apic_map {
856 	struct rcu_head rcu;
857 	u8 mode;
858 	u32 max_apic_id;
859 	union {
860 		struct kvm_lapic *xapic_flat_map[8];
861 		struct kvm_lapic *xapic_cluster_map[16][4];
862 	};
863 	struct kvm_lapic *phys_map[];
864 };
865 
866 /* Hyper-V synthetic debugger (SynDbg)*/
867 struct kvm_hv_syndbg {
868 	struct {
869 		u64 control;
870 		u64 status;
871 		u64 send_page;
872 		u64 recv_page;
873 		u64 pending_page;
874 	} control;
875 	u64 options;
876 };
877 
878 /* Hyper-V emulation context */
879 struct kvm_hv {
880 	struct mutex hv_lock;
881 	u64 hv_guest_os_id;
882 	u64 hv_hypercall;
883 	u64 hv_tsc_page;
884 
885 	/* Hyper-v based guest crash (NT kernel bugcheck) parameters */
886 	u64 hv_crash_param[HV_X64_MSR_CRASH_PARAMS];
887 	u64 hv_crash_ctl;
888 
889 	struct ms_hyperv_tsc_page tsc_ref;
890 
891 	struct idr conn_to_evt;
892 
893 	u64 hv_reenlightenment_control;
894 	u64 hv_tsc_emulation_control;
895 	u64 hv_tsc_emulation_status;
896 
897 	/* How many vCPUs have VP index != vCPU index */
898 	atomic_t num_mismatched_vp_indexes;
899 
900 	struct hv_partition_assist_pg *hv_pa_pg;
901 	struct kvm_hv_syndbg hv_syndbg;
902 };
903 
904 enum kvm_irqchip_mode {
905 	KVM_IRQCHIP_NONE,
906 	KVM_IRQCHIP_KERNEL,       /* created with KVM_CREATE_IRQCHIP */
907 	KVM_IRQCHIP_SPLIT,        /* created with KVM_CAP_SPLIT_IRQCHIP */
908 };
909 
910 #define APICV_INHIBIT_REASON_DISABLE    0
911 #define APICV_INHIBIT_REASON_HYPERV     1
912 #define APICV_INHIBIT_REASON_NESTED     2
913 #define APICV_INHIBIT_REASON_IRQWIN     3
914 #define APICV_INHIBIT_REASON_PIT_REINJ  4
915 #define APICV_INHIBIT_REASON_X2APIC	5
916 
917 struct kvm_arch {
918 	unsigned long n_used_mmu_pages;
919 	unsigned long n_requested_mmu_pages;
920 	unsigned long n_max_mmu_pages;
921 	unsigned int indirect_shadow_pages;
922 	u8 mmu_valid_gen;
923 	struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
924 	/*
925 	 * Hash table of struct kvm_mmu_page.
926 	 */
927 	struct list_head active_mmu_pages;
928 	struct list_head zapped_obsolete_pages;
929 	struct list_head lpage_disallowed_mmu_pages;
930 	struct kvm_page_track_notifier_node mmu_sp_tracker;
931 	struct kvm_page_track_notifier_head track_notifier_head;
932 
933 	struct list_head assigned_dev_head;
934 	struct iommu_domain *iommu_domain;
935 	bool iommu_noncoherent;
936 #define __KVM_HAVE_ARCH_NONCOHERENT_DMA
937 	atomic_t noncoherent_dma_count;
938 #define __KVM_HAVE_ARCH_ASSIGNED_DEVICE
939 	atomic_t assigned_device_count;
940 	struct kvm_pic *vpic;
941 	struct kvm_ioapic *vioapic;
942 	struct kvm_pit *vpit;
943 	atomic_t vapics_in_nmi_mode;
944 	struct mutex apic_map_lock;
945 	struct kvm_apic_map *apic_map;
946 	atomic_t apic_map_dirty;
947 
948 	bool apic_access_page_done;
949 	unsigned long apicv_inhibit_reasons;
950 
951 	gpa_t wall_clock;
952 
953 	bool mwait_in_guest;
954 	bool hlt_in_guest;
955 	bool pause_in_guest;
956 	bool cstate_in_guest;
957 
958 	unsigned long irq_sources_bitmap;
959 	s64 kvmclock_offset;
960 	raw_spinlock_t tsc_write_lock;
961 	u64 last_tsc_nsec;
962 	u64 last_tsc_write;
963 	u32 last_tsc_khz;
964 	u64 cur_tsc_nsec;
965 	u64 cur_tsc_write;
966 	u64 cur_tsc_offset;
967 	u64 cur_tsc_generation;
968 	int nr_vcpus_matched_tsc;
969 
970 	spinlock_t pvclock_gtod_sync_lock;
971 	bool use_master_clock;
972 	u64 master_kernel_ns;
973 	u64 master_cycle_now;
974 	struct delayed_work kvmclock_update_work;
975 	struct delayed_work kvmclock_sync_work;
976 
977 	struct kvm_xen_hvm_config xen_hvm_config;
978 
979 	/* reads protected by irq_srcu, writes by irq_lock */
980 	struct hlist_head mask_notifier_list;
981 
982 	struct kvm_hv hyperv;
983 
984 	#ifdef CONFIG_KVM_MMU_AUDIT
985 	int audit_point;
986 	#endif
987 
988 	bool backwards_tsc_observed;
989 	bool boot_vcpu_runs_old_kvmclock;
990 	u32 bsp_vcpu_id;
991 
992 	u64 disabled_quirks;
993 
994 	enum kvm_irqchip_mode irqchip_mode;
995 	u8 nr_reserved_ioapic_pins;
996 
997 	bool disabled_lapic_found;
998 
999 	bool x2apic_format;
1000 	bool x2apic_broadcast_quirk_disabled;
1001 
1002 	bool guest_can_read_msr_platform_info;
1003 	bool exception_payload_enabled;
1004 
1005 	struct kvm_pmu_event_filter *pmu_event_filter;
1006 	struct task_struct *nx_lpage_recovery_thread;
1007 };
1008 
1009 struct kvm_vm_stat {
1010 	ulong mmu_shadow_zapped;
1011 	ulong mmu_pte_write;
1012 	ulong mmu_pte_updated;
1013 	ulong mmu_pde_zapped;
1014 	ulong mmu_flooded;
1015 	ulong mmu_recycled;
1016 	ulong mmu_cache_miss;
1017 	ulong mmu_unsync;
1018 	ulong remote_tlb_flush;
1019 	ulong lpages;
1020 	ulong nx_lpage_splits;
1021 	ulong max_mmu_page_hash_collisions;
1022 };
1023 
1024 struct kvm_vcpu_stat {
1025 	u64 pf_fixed;
1026 	u64 pf_guest;
1027 	u64 tlb_flush;
1028 	u64 invlpg;
1029 
1030 	u64 exits;
1031 	u64 io_exits;
1032 	u64 mmio_exits;
1033 	u64 signal_exits;
1034 	u64 irq_window_exits;
1035 	u64 nmi_window_exits;
1036 	u64 l1d_flush;
1037 	u64 halt_exits;
1038 	u64 halt_successful_poll;
1039 	u64 halt_attempted_poll;
1040 	u64 halt_poll_invalid;
1041 	u64 halt_wakeup;
1042 	u64 request_irq_exits;
1043 	u64 irq_exits;
1044 	u64 host_state_reload;
1045 	u64 fpu_reload;
1046 	u64 insn_emulation;
1047 	u64 insn_emulation_fail;
1048 	u64 hypercalls;
1049 	u64 irq_injections;
1050 	u64 nmi_injections;
1051 	u64 req_event;
1052 	u64 halt_poll_success_ns;
1053 	u64 halt_poll_fail_ns;
1054 };
1055 
1056 struct x86_instruction_info;
1057 
1058 struct msr_data {
1059 	bool host_initiated;
1060 	u32 index;
1061 	u64 data;
1062 };
1063 
1064 struct kvm_lapic_irq {
1065 	u32 vector;
1066 	u16 delivery_mode;
1067 	u16 dest_mode;
1068 	bool level;
1069 	u16 trig_mode;
1070 	u32 shorthand;
1071 	u32 dest_id;
1072 	bool msi_redir_hint;
1073 };
1074 
1075 static inline u16 kvm_lapic_irq_dest_mode(bool dest_mode_logical)
1076 {
1077 	return dest_mode_logical ? APIC_DEST_LOGICAL : APIC_DEST_PHYSICAL;
1078 }
1079 
1080 struct kvm_x86_ops {
1081 	int (*hardware_enable)(void);
1082 	void (*hardware_disable)(void);
1083 	void (*hardware_unsetup)(void);
1084 	bool (*cpu_has_accelerated_tpr)(void);
1085 	bool (*has_emulated_msr)(u32 index);
1086 	void (*cpuid_update)(struct kvm_vcpu *vcpu);
1087 
1088 	unsigned int vm_size;
1089 	int (*vm_init)(struct kvm *kvm);
1090 	void (*vm_destroy)(struct kvm *kvm);
1091 
1092 	/* Create, but do not attach this VCPU */
1093 	int (*vcpu_create)(struct kvm_vcpu *vcpu);
1094 	void (*vcpu_free)(struct kvm_vcpu *vcpu);
1095 	void (*vcpu_reset)(struct kvm_vcpu *vcpu, bool init_event);
1096 
1097 	void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
1098 	void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
1099 	void (*vcpu_put)(struct kvm_vcpu *vcpu);
1100 
1101 	void (*update_bp_intercept)(struct kvm_vcpu *vcpu);
1102 	int (*get_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
1103 	int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
1104 	u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
1105 	void (*get_segment)(struct kvm_vcpu *vcpu,
1106 			    struct kvm_segment *var, int seg);
1107 	int (*get_cpl)(struct kvm_vcpu *vcpu);
1108 	void (*set_segment)(struct kvm_vcpu *vcpu,
1109 			    struct kvm_segment *var, int seg);
1110 	void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
1111 	void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
1112 	int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
1113 	void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
1114 	void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
1115 	void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
1116 	void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
1117 	void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
1118 	void (*sync_dirty_debug_regs)(struct kvm_vcpu *vcpu);
1119 	void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
1120 	void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
1121 	unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
1122 	void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
1123 
1124 	void (*tlb_flush_all)(struct kvm_vcpu *vcpu);
1125 	void (*tlb_flush_current)(struct kvm_vcpu *vcpu);
1126 	int  (*tlb_remote_flush)(struct kvm *kvm);
1127 	int  (*tlb_remote_flush_with_range)(struct kvm *kvm,
1128 			struct kvm_tlb_range *range);
1129 
1130 	/*
1131 	 * Flush any TLB entries associated with the given GVA.
1132 	 * Does not need to flush GPA->HPA mappings.
1133 	 * Can potentially get non-canonical addresses through INVLPGs, which
1134 	 * the implementation may choose to ignore if appropriate.
1135 	 */
1136 	void (*tlb_flush_gva)(struct kvm_vcpu *vcpu, gva_t addr);
1137 
1138 	/*
1139 	 * Flush any TLB entries created by the guest.  Like tlb_flush_gva(),
1140 	 * does not need to flush GPA->HPA mappings.
1141 	 */
1142 	void (*tlb_flush_guest)(struct kvm_vcpu *vcpu);
1143 
1144 	enum exit_fastpath_completion (*run)(struct kvm_vcpu *vcpu);
1145 	int (*handle_exit)(struct kvm_vcpu *vcpu,
1146 		enum exit_fastpath_completion exit_fastpath);
1147 	int (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
1148 	void (*update_emulated_instruction)(struct kvm_vcpu *vcpu);
1149 	void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
1150 	u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu);
1151 	void (*patch_hypercall)(struct kvm_vcpu *vcpu,
1152 				unsigned char *hypercall_addr);
1153 	void (*set_irq)(struct kvm_vcpu *vcpu);
1154 	void (*set_nmi)(struct kvm_vcpu *vcpu);
1155 	void (*queue_exception)(struct kvm_vcpu *vcpu);
1156 	void (*cancel_injection)(struct kvm_vcpu *vcpu);
1157 	int (*interrupt_allowed)(struct kvm_vcpu *vcpu, bool for_injection);
1158 	int (*nmi_allowed)(struct kvm_vcpu *vcpu, bool for_injection);
1159 	bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
1160 	void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
1161 	void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
1162 	void (*enable_irq_window)(struct kvm_vcpu *vcpu);
1163 	void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
1164 	bool (*check_apicv_inhibit_reasons)(ulong bit);
1165 	void (*pre_update_apicv_exec_ctrl)(struct kvm *kvm, bool activate);
1166 	void (*refresh_apicv_exec_ctrl)(struct kvm_vcpu *vcpu);
1167 	void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
1168 	void (*hwapic_isr_update)(struct kvm_vcpu *vcpu, int isr);
1169 	bool (*guest_apic_has_interrupt)(struct kvm_vcpu *vcpu);
1170 	void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap);
1171 	void (*set_virtual_apic_mode)(struct kvm_vcpu *vcpu);
1172 	void (*set_apic_access_page_addr)(struct kvm_vcpu *vcpu);
1173 	int (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
1174 	int (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
1175 	int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
1176 	int (*set_identity_map_addr)(struct kvm *kvm, u64 ident_addr);
1177 	int (*get_tdp_level)(struct kvm_vcpu *vcpu);
1178 	u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
1179 
1180 	void (*load_mmu_pgd)(struct kvm_vcpu *vcpu, unsigned long cr3);
1181 
1182 	bool (*has_wbinvd_exit)(void);
1183 
1184 	/* Returns actual tsc_offset set in active VMCS */
1185 	u64 (*write_l1_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
1186 
1187 	void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
1188 
1189 	int (*check_intercept)(struct kvm_vcpu *vcpu,
1190 			       struct x86_instruction_info *info,
1191 			       enum x86_intercept_stage stage,
1192 			       struct x86_exception *exception);
1193 	void (*handle_exit_irqoff)(struct kvm_vcpu *vcpu);
1194 
1195 	void (*request_immediate_exit)(struct kvm_vcpu *vcpu);
1196 
1197 	void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
1198 
1199 	/*
1200 	 * Arch-specific dirty logging hooks. These hooks are only supposed to
1201 	 * be valid if the specific arch has hardware-accelerated dirty logging
1202 	 * mechanism. Currently only for PML on VMX.
1203 	 *
1204 	 *  - slot_enable_log_dirty:
1205 	 *	called when enabling log dirty mode for the slot.
1206 	 *  - slot_disable_log_dirty:
1207 	 *	called when disabling log dirty mode for the slot.
1208 	 *	also called when slot is created with log dirty disabled.
1209 	 *  - flush_log_dirty:
1210 	 *	called before reporting dirty_bitmap to userspace.
1211 	 *  - enable_log_dirty_pt_masked:
1212 	 *	called when reenabling log dirty for the GFNs in the mask after
1213 	 *	corresponding bits are cleared in slot->dirty_bitmap.
1214 	 */
1215 	void (*slot_enable_log_dirty)(struct kvm *kvm,
1216 				      struct kvm_memory_slot *slot);
1217 	void (*slot_disable_log_dirty)(struct kvm *kvm,
1218 				       struct kvm_memory_slot *slot);
1219 	void (*flush_log_dirty)(struct kvm *kvm);
1220 	void (*enable_log_dirty_pt_masked)(struct kvm *kvm,
1221 					   struct kvm_memory_slot *slot,
1222 					   gfn_t offset, unsigned long mask);
1223 	int (*write_log_dirty)(struct kvm_vcpu *vcpu, gpa_t l2_gpa);
1224 
1225 	/* pmu operations of sub-arch */
1226 	const struct kvm_pmu_ops *pmu_ops;
1227 	const struct kvm_x86_nested_ops *nested_ops;
1228 
1229 	/*
1230 	 * Architecture specific hooks for vCPU blocking due to
1231 	 * HLT instruction.
1232 	 * Returns for .pre_block():
1233 	 *    - 0 means continue to block the vCPU.
1234 	 *    - 1 means we cannot block the vCPU since some event
1235 	 *        happens during this period, such as, 'ON' bit in
1236 	 *        posted-interrupts descriptor is set.
1237 	 */
1238 	int (*pre_block)(struct kvm_vcpu *vcpu);
1239 	void (*post_block)(struct kvm_vcpu *vcpu);
1240 
1241 	void (*vcpu_blocking)(struct kvm_vcpu *vcpu);
1242 	void (*vcpu_unblocking)(struct kvm_vcpu *vcpu);
1243 
1244 	int (*update_pi_irte)(struct kvm *kvm, unsigned int host_irq,
1245 			      uint32_t guest_irq, bool set);
1246 	void (*apicv_post_state_restore)(struct kvm_vcpu *vcpu);
1247 	bool (*dy_apicv_has_pending_interrupt)(struct kvm_vcpu *vcpu);
1248 
1249 	int (*set_hv_timer)(struct kvm_vcpu *vcpu, u64 guest_deadline_tsc,
1250 			    bool *expired);
1251 	void (*cancel_hv_timer)(struct kvm_vcpu *vcpu);
1252 
1253 	void (*setup_mce)(struct kvm_vcpu *vcpu);
1254 
1255 	int (*smi_allowed)(struct kvm_vcpu *vcpu, bool for_injection);
1256 	int (*pre_enter_smm)(struct kvm_vcpu *vcpu, char *smstate);
1257 	int (*pre_leave_smm)(struct kvm_vcpu *vcpu, const char *smstate);
1258 	void (*enable_smi_window)(struct kvm_vcpu *vcpu);
1259 
1260 	int (*mem_enc_op)(struct kvm *kvm, void __user *argp);
1261 	int (*mem_enc_reg_region)(struct kvm *kvm, struct kvm_enc_region *argp);
1262 	int (*mem_enc_unreg_region)(struct kvm *kvm, struct kvm_enc_region *argp);
1263 
1264 	int (*get_msr_feature)(struct kvm_msr_entry *entry);
1265 
1266 	bool (*need_emulation_on_page_fault)(struct kvm_vcpu *vcpu);
1267 
1268 	bool (*apic_init_signal_blocked)(struct kvm_vcpu *vcpu);
1269 	int (*enable_direct_tlbflush)(struct kvm_vcpu *vcpu);
1270 
1271 	void (*migrate_timers)(struct kvm_vcpu *vcpu);
1272 };
1273 
1274 struct kvm_x86_nested_ops {
1275 	int (*check_events)(struct kvm_vcpu *vcpu);
1276 	bool (*hv_timer_pending)(struct kvm_vcpu *vcpu);
1277 	int (*get_state)(struct kvm_vcpu *vcpu,
1278 			 struct kvm_nested_state __user *user_kvm_nested_state,
1279 			 unsigned user_data_size);
1280 	int (*set_state)(struct kvm_vcpu *vcpu,
1281 			 struct kvm_nested_state __user *user_kvm_nested_state,
1282 			 struct kvm_nested_state *kvm_state);
1283 	bool (*get_vmcs12_pages)(struct kvm_vcpu *vcpu);
1284 
1285 	int (*enable_evmcs)(struct kvm_vcpu *vcpu,
1286 			    uint16_t *vmcs_version);
1287 	uint16_t (*get_evmcs_version)(struct kvm_vcpu *vcpu);
1288 };
1289 
1290 struct kvm_x86_init_ops {
1291 	int (*cpu_has_kvm_support)(void);
1292 	int (*disabled_by_bios)(void);
1293 	int (*check_processor_compatibility)(void);
1294 	int (*hardware_setup)(void);
1295 
1296 	struct kvm_x86_ops *runtime_ops;
1297 };
1298 
1299 struct kvm_arch_async_pf {
1300 	u32 token;
1301 	gfn_t gfn;
1302 	unsigned long cr3;
1303 	bool direct_map;
1304 };
1305 
1306 extern u64 __read_mostly host_efer;
1307 
1308 extern struct kvm_x86_ops kvm_x86_ops;
1309 
1310 #define __KVM_HAVE_ARCH_VM_ALLOC
1311 static inline struct kvm *kvm_arch_alloc_vm(void)
1312 {
1313 	return __vmalloc(kvm_x86_ops.vm_size, GFP_KERNEL_ACCOUNT | __GFP_ZERO);
1314 }
1315 void kvm_arch_free_vm(struct kvm *kvm);
1316 
1317 #define __KVM_HAVE_ARCH_FLUSH_REMOTE_TLB
1318 static inline int kvm_arch_flush_remote_tlb(struct kvm *kvm)
1319 {
1320 	if (kvm_x86_ops.tlb_remote_flush &&
1321 	    !kvm_x86_ops.tlb_remote_flush(kvm))
1322 		return 0;
1323 	else
1324 		return -ENOTSUPP;
1325 }
1326 
1327 int kvm_mmu_module_init(void);
1328 void kvm_mmu_module_exit(void);
1329 
1330 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
1331 int kvm_mmu_create(struct kvm_vcpu *vcpu);
1332 void kvm_mmu_init_vm(struct kvm *kvm);
1333 void kvm_mmu_uninit_vm(struct kvm *kvm);
1334 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
1335 		u64 dirty_mask, u64 nx_mask, u64 x_mask, u64 p_mask,
1336 		u64 acc_track_mask, u64 me_mask);
1337 
1338 void kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
1339 void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
1340 				      struct kvm_memory_slot *memslot,
1341 				      int start_level);
1342 void kvm_mmu_zap_collapsible_sptes(struct kvm *kvm,
1343 				   const struct kvm_memory_slot *memslot);
1344 void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
1345 				   struct kvm_memory_slot *memslot);
1346 void kvm_mmu_slot_largepage_remove_write_access(struct kvm *kvm,
1347 					struct kvm_memory_slot *memslot);
1348 void kvm_mmu_slot_set_dirty(struct kvm *kvm,
1349 			    struct kvm_memory_slot *memslot);
1350 void kvm_mmu_clear_dirty_pt_masked(struct kvm *kvm,
1351 				   struct kvm_memory_slot *slot,
1352 				   gfn_t gfn_offset, unsigned long mask);
1353 void kvm_mmu_zap_all(struct kvm *kvm);
1354 void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm, u64 gen);
1355 unsigned long kvm_mmu_calculate_default_mmu_pages(struct kvm *kvm);
1356 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned long kvm_nr_mmu_pages);
1357 
1358 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
1359 bool pdptrs_changed(struct kvm_vcpu *vcpu);
1360 
1361 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
1362 			  const void *val, int bytes);
1363 
1364 struct kvm_irq_mask_notifier {
1365 	void (*func)(struct kvm_irq_mask_notifier *kimn, bool masked);
1366 	int irq;
1367 	struct hlist_node link;
1368 };
1369 
1370 void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
1371 				    struct kvm_irq_mask_notifier *kimn);
1372 void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
1373 				      struct kvm_irq_mask_notifier *kimn);
1374 void kvm_fire_mask_notifiers(struct kvm *kvm, unsigned irqchip, unsigned pin,
1375 			     bool mask);
1376 
1377 extern bool tdp_enabled;
1378 
1379 u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
1380 
1381 /* control of guest tsc rate supported? */
1382 extern bool kvm_has_tsc_control;
1383 /* maximum supported tsc_khz for guests */
1384 extern u32  kvm_max_guest_tsc_khz;
1385 /* number of bits of the fractional part of the TSC scaling ratio */
1386 extern u8   kvm_tsc_scaling_ratio_frac_bits;
1387 /* maximum allowed value of TSC scaling ratio */
1388 extern u64  kvm_max_tsc_scaling_ratio;
1389 /* 1ull << kvm_tsc_scaling_ratio_frac_bits */
1390 extern u64  kvm_default_tsc_scaling_ratio;
1391 
1392 extern u64 kvm_mce_cap_supported;
1393 
1394 /*
1395  * EMULTYPE_NO_DECODE - Set when re-emulating an instruction (after completing
1396  *			userspace I/O) to indicate that the emulation context
1397  *			should be resued as is, i.e. skip initialization of
1398  *			emulation context, instruction fetch and decode.
1399  *
1400  * EMULTYPE_TRAP_UD - Set when emulating an intercepted #UD from hardware.
1401  *		      Indicates that only select instructions (tagged with
1402  *		      EmulateOnUD) should be emulated (to minimize the emulator
1403  *		      attack surface).  See also EMULTYPE_TRAP_UD_FORCED.
1404  *
1405  * EMULTYPE_SKIP - Set when emulating solely to skip an instruction, i.e. to
1406  *		   decode the instruction length.  For use *only* by
1407  *		   kvm_x86_ops.skip_emulated_instruction() implementations.
1408  *
1409  * EMULTYPE_ALLOW_RETRY_PF - Set when the emulator should resume the guest to
1410  *			     retry native execution under certain conditions,
1411  *			     Can only be set in conjunction with EMULTYPE_PF.
1412  *
1413  * EMULTYPE_TRAP_UD_FORCED - Set when emulating an intercepted #UD that was
1414  *			     triggered by KVM's magic "force emulation" prefix,
1415  *			     which is opt in via module param (off by default).
1416  *			     Bypasses EmulateOnUD restriction despite emulating
1417  *			     due to an intercepted #UD (see EMULTYPE_TRAP_UD).
1418  *			     Used to test the full emulator from userspace.
1419  *
1420  * EMULTYPE_VMWARE_GP - Set when emulating an intercepted #GP for VMware
1421  *			backdoor emulation, which is opt in via module param.
1422  *			VMware backoor emulation handles select instructions
1423  *			and reinjects the #GP for all other cases.
1424  *
1425  * EMULTYPE_PF - Set when emulating MMIO by way of an intercepted #PF, in which
1426  *		 case the CR2/GPA value pass on the stack is valid.
1427  */
1428 #define EMULTYPE_NO_DECODE	    (1 << 0)
1429 #define EMULTYPE_TRAP_UD	    (1 << 1)
1430 #define EMULTYPE_SKIP		    (1 << 2)
1431 #define EMULTYPE_ALLOW_RETRY_PF	    (1 << 3)
1432 #define EMULTYPE_TRAP_UD_FORCED	    (1 << 4)
1433 #define EMULTYPE_VMWARE_GP	    (1 << 5)
1434 #define EMULTYPE_PF		    (1 << 6)
1435 
1436 int kvm_emulate_instruction(struct kvm_vcpu *vcpu, int emulation_type);
1437 int kvm_emulate_instruction_from_buffer(struct kvm_vcpu *vcpu,
1438 					void *insn, int insn_len);
1439 
1440 void kvm_enable_efer_bits(u64);
1441 bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
1442 int __kvm_get_msr(struct kvm_vcpu *vcpu, u32 index, u64 *data, bool host_initiated);
1443 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 index, u64 *data);
1444 int kvm_set_msr(struct kvm_vcpu *vcpu, u32 index, u64 data);
1445 int kvm_emulate_rdmsr(struct kvm_vcpu *vcpu);
1446 int kvm_emulate_wrmsr(struct kvm_vcpu *vcpu);
1447 
1448 int kvm_fast_pio(struct kvm_vcpu *vcpu, int size, unsigned short port, int in);
1449 int kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
1450 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
1451 int kvm_vcpu_halt(struct kvm_vcpu *vcpu);
1452 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
1453 
1454 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
1455 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
1456 void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
1457 
1458 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
1459 		    int reason, bool has_error_code, u32 error_code);
1460 
1461 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
1462 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
1463 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
1464 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
1465 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
1466 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
1467 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
1468 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
1469 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
1470 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
1471 
1472 int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
1473 int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
1474 
1475 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
1476 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
1477 bool kvm_rdpmc(struct kvm_vcpu *vcpu);
1478 
1479 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1480 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
1481 void kvm_queue_exception_p(struct kvm_vcpu *vcpu, unsigned nr, unsigned long payload);
1482 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1483 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
1484 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
1485 bool kvm_inject_emulated_page_fault(struct kvm_vcpu *vcpu,
1486 				    struct x86_exception *fault);
1487 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
1488 			    gfn_t gfn, void *data, int offset, int len,
1489 			    u32 access);
1490 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
1491 bool kvm_require_dr(struct kvm_vcpu *vcpu, int dr);
1492 
1493 static inline int __kvm_irq_line_state(unsigned long *irq_state,
1494 				       int irq_source_id, int level)
1495 {
1496 	/* Logical OR for level trig interrupt */
1497 	if (level)
1498 		__set_bit(irq_source_id, irq_state);
1499 	else
1500 		__clear_bit(irq_source_id, irq_state);
1501 
1502 	return !!(*irq_state);
1503 }
1504 
1505 #define KVM_MMU_ROOT_CURRENT		BIT(0)
1506 #define KVM_MMU_ROOT_PREVIOUS(i)	BIT(1+i)
1507 #define KVM_MMU_ROOTS_ALL		(~0UL)
1508 
1509 int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
1510 void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
1511 
1512 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
1513 
1514 void kvm_update_dr7(struct kvm_vcpu *vcpu);
1515 
1516 int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
1517 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
1518 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
1519 int kvm_mmu_load(struct kvm_vcpu *vcpu);
1520 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
1521 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
1522 void kvm_mmu_free_roots(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
1523 			ulong roots_to_free);
1524 gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1525 			   struct x86_exception *exception);
1526 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
1527 			      struct x86_exception *exception);
1528 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
1529 			       struct x86_exception *exception);
1530 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
1531 			       struct x86_exception *exception);
1532 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
1533 				struct x86_exception *exception);
1534 
1535 bool kvm_apicv_activated(struct kvm *kvm);
1536 void kvm_apicv_init(struct kvm *kvm, bool enable);
1537 void kvm_vcpu_update_apicv(struct kvm_vcpu *vcpu);
1538 void kvm_request_apicv_update(struct kvm *kvm, bool activate,
1539 			      unsigned long bit);
1540 
1541 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
1542 
1543 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa, u64 error_code,
1544 		       void *insn, int insn_len);
1545 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
1546 void kvm_mmu_invalidate_gva(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
1547 			    gva_t gva, hpa_t root_hpa);
1548 void kvm_mmu_invpcid_gva(struct kvm_vcpu *vcpu, gva_t gva, unsigned long pcid);
1549 void kvm_mmu_new_pgd(struct kvm_vcpu *vcpu, gpa_t new_pgd, bool skip_tlb_flush,
1550 		     bool skip_mmu_sync);
1551 
1552 void kvm_configure_mmu(bool enable_tdp, int tdp_page_level);
1553 
1554 static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1555 				  struct x86_exception *exception)
1556 {
1557 	return gpa;
1558 }
1559 
1560 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
1561 {
1562 	struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
1563 
1564 	return (struct kvm_mmu_page *)page_private(page);
1565 }
1566 
1567 static inline u16 kvm_read_ldt(void)
1568 {
1569 	u16 ldt;
1570 	asm("sldt %0" : "=g"(ldt));
1571 	return ldt;
1572 }
1573 
1574 static inline void kvm_load_ldt(u16 sel)
1575 {
1576 	asm("lldt %0" : : "rm"(sel));
1577 }
1578 
1579 #ifdef CONFIG_X86_64
1580 static inline unsigned long read_msr(unsigned long msr)
1581 {
1582 	u64 value;
1583 
1584 	rdmsrl(msr, value);
1585 	return value;
1586 }
1587 #endif
1588 
1589 static inline u32 get_rdx_init_val(void)
1590 {
1591 	return 0x600; /* P6 family */
1592 }
1593 
1594 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
1595 {
1596 	kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
1597 }
1598 
1599 #define TSS_IOPB_BASE_OFFSET 0x66
1600 #define TSS_BASE_SIZE 0x68
1601 #define TSS_IOPB_SIZE (65536 / 8)
1602 #define TSS_REDIRECTION_SIZE (256 / 8)
1603 #define RMODE_TSS_SIZE							\
1604 	(TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
1605 
1606 enum {
1607 	TASK_SWITCH_CALL = 0,
1608 	TASK_SWITCH_IRET = 1,
1609 	TASK_SWITCH_JMP = 2,
1610 	TASK_SWITCH_GATE = 3,
1611 };
1612 
1613 #define HF_GIF_MASK		(1 << 0)
1614 #define HF_NMI_MASK		(1 << 3)
1615 #define HF_IRET_MASK		(1 << 4)
1616 #define HF_GUEST_MASK		(1 << 5) /* VCPU is in guest-mode */
1617 #define HF_SMM_MASK		(1 << 6)
1618 #define HF_SMM_INSIDE_NMI_MASK	(1 << 7)
1619 
1620 #define __KVM_VCPU_MULTIPLE_ADDRESS_SPACE
1621 #define KVM_ADDRESS_SPACE_NUM 2
1622 
1623 #define kvm_arch_vcpu_memslots_id(vcpu) ((vcpu)->arch.hflags & HF_SMM_MASK ? 1 : 0)
1624 #define kvm_memslots_for_spte_role(kvm, role) __kvm_memslots(kvm, (role).smm)
1625 
1626 asmlinkage void kvm_spurious_fault(void);
1627 
1628 /*
1629  * Hardware virtualization extension instructions may fault if a
1630  * reboot turns off virtualization while processes are running.
1631  * Usually after catching the fault we just panic; during reboot
1632  * instead the instruction is ignored.
1633  */
1634 #define __kvm_handle_fault_on_reboot(insn)				\
1635 	"666: \n\t"							\
1636 	insn "\n\t"							\
1637 	"jmp	668f \n\t"						\
1638 	"667: \n\t"							\
1639 	"call	kvm_spurious_fault \n\t"				\
1640 	"668: \n\t"							\
1641 	_ASM_EXTABLE(666b, 667b)
1642 
1643 #define KVM_ARCH_WANT_MMU_NOTIFIER
1644 int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
1645 int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
1646 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
1647 int kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
1648 int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
1649 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1650 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
1651 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
1652 void kvm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event);
1653 void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu);
1654 
1655 int kvm_pv_send_ipi(struct kvm *kvm, unsigned long ipi_bitmap_low,
1656 		    unsigned long ipi_bitmap_high, u32 min,
1657 		    unsigned long icr, int op_64_bit);
1658 
1659 void kvm_define_shared_msr(unsigned index, u32 msr);
1660 int kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
1661 
1662 u64 kvm_scale_tsc(struct kvm_vcpu *vcpu, u64 tsc);
1663 u64 kvm_read_l1_tsc(struct kvm_vcpu *vcpu, u64 host_tsc);
1664 
1665 unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
1666 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1667 
1668 void kvm_make_mclock_inprogress_request(struct kvm *kvm);
1669 void kvm_make_scan_ioapic_request(struct kvm *kvm);
1670 void kvm_make_scan_ioapic_request_mask(struct kvm *kvm,
1671 				       unsigned long *vcpu_bitmap);
1672 
1673 bool kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1674 				     struct kvm_async_pf *work);
1675 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1676 				 struct kvm_async_pf *work);
1677 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1678 			       struct kvm_async_pf *work);
1679 void kvm_arch_async_page_present_queued(struct kvm_vcpu *vcpu);
1680 bool kvm_arch_can_dequeue_async_page_present(struct kvm_vcpu *vcpu);
1681 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1682 
1683 int kvm_skip_emulated_instruction(struct kvm_vcpu *vcpu);
1684 int kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1685 void __kvm_request_immediate_exit(struct kvm_vcpu *vcpu);
1686 
1687 int kvm_is_in_guest(void);
1688 
1689 int __x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa, u32 size);
1690 bool kvm_vcpu_is_reset_bsp(struct kvm_vcpu *vcpu);
1691 bool kvm_vcpu_is_bsp(struct kvm_vcpu *vcpu);
1692 
1693 bool kvm_intr_is_single_vcpu(struct kvm *kvm, struct kvm_lapic_irq *irq,
1694 			     struct kvm_vcpu **dest_vcpu);
1695 
1696 void kvm_set_msi_irq(struct kvm *kvm, struct kvm_kernel_irq_routing_entry *e,
1697 		     struct kvm_lapic_irq *irq);
1698 
1699 static inline bool kvm_irq_is_postable(struct kvm_lapic_irq *irq)
1700 {
1701 	/* We can only post Fixed and LowPrio IRQs */
1702 	return (irq->delivery_mode == APIC_DM_FIXED ||
1703 		irq->delivery_mode == APIC_DM_LOWEST);
1704 }
1705 
1706 static inline void kvm_arch_vcpu_blocking(struct kvm_vcpu *vcpu)
1707 {
1708 	if (kvm_x86_ops.vcpu_blocking)
1709 		kvm_x86_ops.vcpu_blocking(vcpu);
1710 }
1711 
1712 static inline void kvm_arch_vcpu_unblocking(struct kvm_vcpu *vcpu)
1713 {
1714 	if (kvm_x86_ops.vcpu_unblocking)
1715 		kvm_x86_ops.vcpu_unblocking(vcpu);
1716 }
1717 
1718 static inline void kvm_arch_vcpu_block_finish(struct kvm_vcpu *vcpu) {}
1719 
1720 static inline int kvm_cpu_get_apicid(int mps_cpu)
1721 {
1722 #ifdef CONFIG_X86_LOCAL_APIC
1723 	return default_cpu_present_to_apicid(mps_cpu);
1724 #else
1725 	WARN_ON_ONCE(1);
1726 	return BAD_APICID;
1727 #endif
1728 }
1729 
1730 #define put_smstate(type, buf, offset, val)                      \
1731 	*(type *)((buf) + (offset) - 0x7e00) = val
1732 
1733 #define GET_SMSTATE(type, buf, offset)		\
1734 	(*(type *)((buf) + (offset) - 0x7e00))
1735 
1736 #endif /* _ASM_X86_KVM_HOST_H */
1737