xref: /linux/arch/x86/include/asm/kvm_host.h (revision e0bf6c5ca2d3281f231c5f0c9bf145e9513644de)
1 /*
2  * Kernel-based Virtual Machine driver for Linux
3  *
4  * This header defines architecture specific interfaces, x86 version
5  *
6  * This work is licensed under the terms of the GNU GPL, version 2.  See
7  * the COPYING file in the top-level directory.
8  *
9  */
10 
11 #ifndef _ASM_X86_KVM_HOST_H
12 #define _ASM_X86_KVM_HOST_H
13 
14 #include <linux/types.h>
15 #include <linux/mm.h>
16 #include <linux/mmu_notifier.h>
17 #include <linux/tracepoint.h>
18 #include <linux/cpumask.h>
19 #include <linux/irq_work.h>
20 
21 #include <linux/kvm.h>
22 #include <linux/kvm_para.h>
23 #include <linux/kvm_types.h>
24 #include <linux/perf_event.h>
25 #include <linux/pvclock_gtod.h>
26 #include <linux/clocksource.h>
27 
28 #include <asm/pvclock-abi.h>
29 #include <asm/desc.h>
30 #include <asm/mtrr.h>
31 #include <asm/msr-index.h>
32 #include <asm/asm.h>
33 
34 #define KVM_MAX_VCPUS 255
35 #define KVM_SOFT_MAX_VCPUS 160
36 #define KVM_USER_MEM_SLOTS 509
37 /* memory slots that are not exposed to userspace */
38 #define KVM_PRIVATE_MEM_SLOTS 3
39 #define KVM_MEM_SLOTS_NUM (KVM_USER_MEM_SLOTS + KVM_PRIVATE_MEM_SLOTS)
40 
41 #define KVM_PIO_PAGE_OFFSET 1
42 #define KVM_COALESCED_MMIO_PAGE_OFFSET 2
43 
44 #define KVM_IRQCHIP_NUM_PINS  KVM_IOAPIC_NUM_PINS
45 
46 #define CR0_RESERVED_BITS                                               \
47 	(~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
48 			  | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
49 			  | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
50 
51 #define CR3_L_MODE_RESERVED_BITS 0xFFFFFF0000000000ULL
52 #define CR3_PCID_INVD		 BIT_64(63)
53 #define CR4_RESERVED_BITS                                               \
54 	(~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
55 			  | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE     \
56 			  | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR | X86_CR4_PCIDE \
57 			  | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_FSGSBASE \
58 			  | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE | X86_CR4_SMAP))
59 
60 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
61 
62 
63 
64 #define INVALID_PAGE (~(hpa_t)0)
65 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
66 
67 #define UNMAPPED_GVA (~(gpa_t)0)
68 
69 /* KVM Hugepage definitions for x86 */
70 #define KVM_NR_PAGE_SIZES	3
71 #define KVM_HPAGE_GFN_SHIFT(x)	(((x) - 1) * 9)
72 #define KVM_HPAGE_SHIFT(x)	(PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
73 #define KVM_HPAGE_SIZE(x)	(1UL << KVM_HPAGE_SHIFT(x))
74 #define KVM_HPAGE_MASK(x)	(~(KVM_HPAGE_SIZE(x) - 1))
75 #define KVM_PAGES_PER_HPAGE(x)	(KVM_HPAGE_SIZE(x) / PAGE_SIZE)
76 
77 static inline gfn_t gfn_to_index(gfn_t gfn, gfn_t base_gfn, int level)
78 {
79 	/* KVM_HPAGE_GFN_SHIFT(PT_PAGE_TABLE_LEVEL) must be 0. */
80 	return (gfn >> KVM_HPAGE_GFN_SHIFT(level)) -
81 		(base_gfn >> KVM_HPAGE_GFN_SHIFT(level));
82 }
83 
84 #define SELECTOR_TI_MASK (1 << 2)
85 #define SELECTOR_RPL_MASK 0x03
86 
87 #define IOPL_SHIFT 12
88 
89 #define KVM_PERMILLE_MMU_PAGES 20
90 #define KVM_MIN_ALLOC_MMU_PAGES 64
91 #define KVM_MMU_HASH_SHIFT 10
92 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
93 #define KVM_MIN_FREE_MMU_PAGES 5
94 #define KVM_REFILL_PAGES 25
95 #define KVM_MAX_CPUID_ENTRIES 80
96 #define KVM_NR_FIXED_MTRR_REGION 88
97 #define KVM_NR_VAR_MTRR 8
98 
99 #define ASYNC_PF_PER_VCPU 64
100 
101 enum kvm_reg {
102 	VCPU_REGS_RAX = 0,
103 	VCPU_REGS_RCX = 1,
104 	VCPU_REGS_RDX = 2,
105 	VCPU_REGS_RBX = 3,
106 	VCPU_REGS_RSP = 4,
107 	VCPU_REGS_RBP = 5,
108 	VCPU_REGS_RSI = 6,
109 	VCPU_REGS_RDI = 7,
110 #ifdef CONFIG_X86_64
111 	VCPU_REGS_R8 = 8,
112 	VCPU_REGS_R9 = 9,
113 	VCPU_REGS_R10 = 10,
114 	VCPU_REGS_R11 = 11,
115 	VCPU_REGS_R12 = 12,
116 	VCPU_REGS_R13 = 13,
117 	VCPU_REGS_R14 = 14,
118 	VCPU_REGS_R15 = 15,
119 #endif
120 	VCPU_REGS_RIP,
121 	NR_VCPU_REGS
122 };
123 
124 enum kvm_reg_ex {
125 	VCPU_EXREG_PDPTR = NR_VCPU_REGS,
126 	VCPU_EXREG_CR3,
127 	VCPU_EXREG_RFLAGS,
128 	VCPU_EXREG_SEGMENTS,
129 };
130 
131 enum {
132 	VCPU_SREG_ES,
133 	VCPU_SREG_CS,
134 	VCPU_SREG_SS,
135 	VCPU_SREG_DS,
136 	VCPU_SREG_FS,
137 	VCPU_SREG_GS,
138 	VCPU_SREG_TR,
139 	VCPU_SREG_LDTR,
140 };
141 
142 #include <asm/kvm_emulate.h>
143 
144 #define KVM_NR_MEM_OBJS 40
145 
146 #define KVM_NR_DB_REGS	4
147 
148 #define DR6_BD		(1 << 13)
149 #define DR6_BS		(1 << 14)
150 #define DR6_RTM		(1 << 16)
151 #define DR6_FIXED_1	0xfffe0ff0
152 #define DR6_INIT	0xffff0ff0
153 #define DR6_VOLATILE	0x0001e00f
154 
155 #define DR7_BP_EN_MASK	0x000000ff
156 #define DR7_GE		(1 << 9)
157 #define DR7_GD		(1 << 13)
158 #define DR7_FIXED_1	0x00000400
159 #define DR7_VOLATILE	0xffff2bff
160 
161 #define PFERR_PRESENT_BIT 0
162 #define PFERR_WRITE_BIT 1
163 #define PFERR_USER_BIT 2
164 #define PFERR_RSVD_BIT 3
165 #define PFERR_FETCH_BIT 4
166 
167 #define PFERR_PRESENT_MASK (1U << PFERR_PRESENT_BIT)
168 #define PFERR_WRITE_MASK (1U << PFERR_WRITE_BIT)
169 #define PFERR_USER_MASK (1U << PFERR_USER_BIT)
170 #define PFERR_RSVD_MASK (1U << PFERR_RSVD_BIT)
171 #define PFERR_FETCH_MASK (1U << PFERR_FETCH_BIT)
172 
173 /* apic attention bits */
174 #define KVM_APIC_CHECK_VAPIC	0
175 /*
176  * The following bit is set with PV-EOI, unset on EOI.
177  * We detect PV-EOI changes by guest by comparing
178  * this bit with PV-EOI in guest memory.
179  * See the implementation in apic_update_pv_eoi.
180  */
181 #define KVM_APIC_PV_EOI_PENDING	1
182 
183 /*
184  * We don't want allocation failures within the mmu code, so we preallocate
185  * enough memory for a single page fault in a cache.
186  */
187 struct kvm_mmu_memory_cache {
188 	int nobjs;
189 	void *objects[KVM_NR_MEM_OBJS];
190 };
191 
192 /*
193  * kvm_mmu_page_role, below, is defined as:
194  *
195  *   bits 0:3 - total guest paging levels (2-4, or zero for real mode)
196  *   bits 4:7 - page table level for this shadow (1-4)
197  *   bits 8:9 - page table quadrant for 2-level guests
198  *   bit   16 - direct mapping of virtual to physical mapping at gfn
199  *              used for real mode and two-dimensional paging
200  *   bits 17:19 - common access permissions for all ptes in this shadow page
201  */
202 union kvm_mmu_page_role {
203 	unsigned word;
204 	struct {
205 		unsigned level:4;
206 		unsigned cr4_pae:1;
207 		unsigned quadrant:2;
208 		unsigned pad_for_nice_hex_output:6;
209 		unsigned direct:1;
210 		unsigned access:3;
211 		unsigned invalid:1;
212 		unsigned nxe:1;
213 		unsigned cr0_wp:1;
214 		unsigned smep_andnot_wp:1;
215 	};
216 };
217 
218 struct kvm_mmu_page {
219 	struct list_head link;
220 	struct hlist_node hash_link;
221 
222 	/*
223 	 * The following two entries are used to key the shadow page in the
224 	 * hash table.
225 	 */
226 	gfn_t gfn;
227 	union kvm_mmu_page_role role;
228 
229 	u64 *spt;
230 	/* hold the gfn of each spte inside spt */
231 	gfn_t *gfns;
232 	bool unsync;
233 	int root_count;          /* Currently serving as active root */
234 	unsigned int unsync_children;
235 	unsigned long parent_ptes;	/* Reverse mapping for parent_pte */
236 
237 	/* The page is obsolete if mmu_valid_gen != kvm->arch.mmu_valid_gen.  */
238 	unsigned long mmu_valid_gen;
239 
240 	DECLARE_BITMAP(unsync_child_bitmap, 512);
241 
242 #ifdef CONFIG_X86_32
243 	/*
244 	 * Used out of the mmu-lock to avoid reading spte values while an
245 	 * update is in progress; see the comments in __get_spte_lockless().
246 	 */
247 	int clear_spte_count;
248 #endif
249 
250 	/* Number of writes since the last time traversal visited this page.  */
251 	int write_flooding_count;
252 };
253 
254 struct kvm_pio_request {
255 	unsigned long count;
256 	int in;
257 	int port;
258 	int size;
259 };
260 
261 /*
262  * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
263  * 32-bit).  The kvm_mmu structure abstracts the details of the current mmu
264  * mode.
265  */
266 struct kvm_mmu {
267 	void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
268 	unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
269 	u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
270 	int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
271 			  bool prefault);
272 	void (*inject_page_fault)(struct kvm_vcpu *vcpu,
273 				  struct x86_exception *fault);
274 	gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
275 			    struct x86_exception *exception);
276 	gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
277 			       struct x86_exception *exception);
278 	int (*sync_page)(struct kvm_vcpu *vcpu,
279 			 struct kvm_mmu_page *sp);
280 	void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
281 	void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
282 			   u64 *spte, const void *pte);
283 	hpa_t root_hpa;
284 	int root_level;
285 	int shadow_root_level;
286 	union kvm_mmu_page_role base_role;
287 	bool direct_map;
288 
289 	/*
290 	 * Bitmap; bit set = permission fault
291 	 * Byte index: page fault error code [4:1]
292 	 * Bit index: pte permissions in ACC_* format
293 	 */
294 	u8 permissions[16];
295 
296 	u64 *pae_root;
297 	u64 *lm_root;
298 	u64 rsvd_bits_mask[2][4];
299 	u64 bad_mt_xwr;
300 
301 	/*
302 	 * Bitmap: bit set = last pte in walk
303 	 * index[0:1]: level (zero-based)
304 	 * index[2]: pte.ps
305 	 */
306 	u8 last_pte_bitmap;
307 
308 	bool nx;
309 
310 	u64 pdptrs[4]; /* pae */
311 };
312 
313 enum pmc_type {
314 	KVM_PMC_GP = 0,
315 	KVM_PMC_FIXED,
316 };
317 
318 struct kvm_pmc {
319 	enum pmc_type type;
320 	u8 idx;
321 	u64 counter;
322 	u64 eventsel;
323 	struct perf_event *perf_event;
324 	struct kvm_vcpu *vcpu;
325 };
326 
327 struct kvm_pmu {
328 	unsigned nr_arch_gp_counters;
329 	unsigned nr_arch_fixed_counters;
330 	unsigned available_event_types;
331 	u64 fixed_ctr_ctrl;
332 	u64 global_ctrl;
333 	u64 global_status;
334 	u64 global_ovf_ctrl;
335 	u64 counter_bitmask[2];
336 	u64 global_ctrl_mask;
337 	u64 reserved_bits;
338 	u8 version;
339 	struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
340 	struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
341 	struct irq_work irq_work;
342 	u64 reprogram_pmi;
343 };
344 
345 enum {
346 	KVM_DEBUGREG_BP_ENABLED = 1,
347 	KVM_DEBUGREG_WONT_EXIT = 2,
348 };
349 
350 struct kvm_vcpu_arch {
351 	/*
352 	 * rip and regs accesses must go through
353 	 * kvm_{register,rip}_{read,write} functions.
354 	 */
355 	unsigned long regs[NR_VCPU_REGS];
356 	u32 regs_avail;
357 	u32 regs_dirty;
358 
359 	unsigned long cr0;
360 	unsigned long cr0_guest_owned_bits;
361 	unsigned long cr2;
362 	unsigned long cr3;
363 	unsigned long cr4;
364 	unsigned long cr4_guest_owned_bits;
365 	unsigned long cr8;
366 	u32 hflags;
367 	u64 efer;
368 	u64 apic_base;
369 	struct kvm_lapic *apic;    /* kernel irqchip context */
370 	unsigned long apic_attention;
371 	int32_t apic_arb_prio;
372 	int mp_state;
373 	u64 ia32_misc_enable_msr;
374 	bool tpr_access_reporting;
375 	u64 ia32_xss;
376 
377 	/*
378 	 * Paging state of the vcpu
379 	 *
380 	 * If the vcpu runs in guest mode with two level paging this still saves
381 	 * the paging mode of the l1 guest. This context is always used to
382 	 * handle faults.
383 	 */
384 	struct kvm_mmu mmu;
385 
386 	/*
387 	 * Paging state of an L2 guest (used for nested npt)
388 	 *
389 	 * This context will save all necessary information to walk page tables
390 	 * of the an L2 guest. This context is only initialized for page table
391 	 * walking and not for faulting since we never handle l2 page faults on
392 	 * the host.
393 	 */
394 	struct kvm_mmu nested_mmu;
395 
396 	/*
397 	 * Pointer to the mmu context currently used for
398 	 * gva_to_gpa translations.
399 	 */
400 	struct kvm_mmu *walk_mmu;
401 
402 	struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
403 	struct kvm_mmu_memory_cache mmu_page_cache;
404 	struct kvm_mmu_memory_cache mmu_page_header_cache;
405 
406 	struct fpu guest_fpu;
407 	u64 xcr0;
408 	u64 guest_supported_xcr0;
409 	u32 guest_xstate_size;
410 
411 	struct kvm_pio_request pio;
412 	void *pio_data;
413 
414 	u8 event_exit_inst_len;
415 
416 	struct kvm_queued_exception {
417 		bool pending;
418 		bool has_error_code;
419 		bool reinject;
420 		u8 nr;
421 		u32 error_code;
422 	} exception;
423 
424 	struct kvm_queued_interrupt {
425 		bool pending;
426 		bool soft;
427 		u8 nr;
428 	} interrupt;
429 
430 	int halt_request; /* real mode on Intel only */
431 
432 	int cpuid_nent;
433 	struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
434 	/* emulate context */
435 
436 	struct x86_emulate_ctxt emulate_ctxt;
437 	bool emulate_regs_need_sync_to_vcpu;
438 	bool emulate_regs_need_sync_from_vcpu;
439 	int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
440 
441 	gpa_t time;
442 	struct pvclock_vcpu_time_info hv_clock;
443 	unsigned int hw_tsc_khz;
444 	struct gfn_to_hva_cache pv_time;
445 	bool pv_time_enabled;
446 	/* set guest stopped flag in pvclock flags field */
447 	bool pvclock_set_guest_stopped_request;
448 
449 	struct {
450 		u64 msr_val;
451 		u64 last_steal;
452 		u64 accum_steal;
453 		struct gfn_to_hva_cache stime;
454 		struct kvm_steal_time steal;
455 	} st;
456 
457 	u64 last_guest_tsc;
458 	u64 last_host_tsc;
459 	u64 tsc_offset_adjustment;
460 	u64 this_tsc_nsec;
461 	u64 this_tsc_write;
462 	u64 this_tsc_generation;
463 	bool tsc_catchup;
464 	bool tsc_always_catchup;
465 	s8 virtual_tsc_shift;
466 	u32 virtual_tsc_mult;
467 	u32 virtual_tsc_khz;
468 	s64 ia32_tsc_adjust_msr;
469 
470 	atomic_t nmi_queued;  /* unprocessed asynchronous NMIs */
471 	unsigned nmi_pending; /* NMI queued after currently running handler */
472 	bool nmi_injected;    /* Trying to inject an NMI this entry */
473 
474 	struct mtrr_state_type mtrr_state;
475 	u64 pat;
476 
477 	unsigned switch_db_regs;
478 	unsigned long db[KVM_NR_DB_REGS];
479 	unsigned long dr6;
480 	unsigned long dr7;
481 	unsigned long eff_db[KVM_NR_DB_REGS];
482 	unsigned long guest_debug_dr7;
483 
484 	u64 mcg_cap;
485 	u64 mcg_status;
486 	u64 mcg_ctl;
487 	u64 *mce_banks;
488 
489 	/* Cache MMIO info */
490 	u64 mmio_gva;
491 	unsigned access;
492 	gfn_t mmio_gfn;
493 	u64 mmio_gen;
494 
495 	struct kvm_pmu pmu;
496 
497 	/* used for guest single stepping over the given code position */
498 	unsigned long singlestep_rip;
499 
500 	/* fields used by HYPER-V emulation */
501 	u64 hv_vapic;
502 
503 	cpumask_var_t wbinvd_dirty_mask;
504 
505 	unsigned long last_retry_eip;
506 	unsigned long last_retry_addr;
507 
508 	struct {
509 		bool halted;
510 		gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
511 		struct gfn_to_hva_cache data;
512 		u64 msr_val;
513 		u32 id;
514 		bool send_user_only;
515 	} apf;
516 
517 	/* OSVW MSRs (AMD only) */
518 	struct {
519 		u64 length;
520 		u64 status;
521 	} osvw;
522 
523 	struct {
524 		u64 msr_val;
525 		struct gfn_to_hva_cache data;
526 	} pv_eoi;
527 
528 	/*
529 	 * Indicate whether the access faults on its page table in guest
530 	 * which is set when fix page fault and used to detect unhandeable
531 	 * instruction.
532 	 */
533 	bool write_fault_to_shadow_pgtable;
534 
535 	/* set at EPT violation at this point */
536 	unsigned long exit_qualification;
537 
538 	/* pv related host specific info */
539 	struct {
540 		bool pv_unhalted;
541 	} pv;
542 };
543 
544 struct kvm_lpage_info {
545 	int write_count;
546 };
547 
548 struct kvm_arch_memory_slot {
549 	unsigned long *rmap[KVM_NR_PAGE_SIZES];
550 	struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
551 };
552 
553 struct kvm_apic_map {
554 	struct rcu_head rcu;
555 	u8 ldr_bits;
556 	/* fields bellow are used to decode ldr values in different modes */
557 	u32 cid_shift, cid_mask, lid_mask, broadcast;
558 	struct kvm_lapic *phys_map[256];
559 	/* first index is cluster id second is cpu id in a cluster */
560 	struct kvm_lapic *logical_map[16][16];
561 };
562 
563 struct kvm_arch {
564 	unsigned int n_used_mmu_pages;
565 	unsigned int n_requested_mmu_pages;
566 	unsigned int n_max_mmu_pages;
567 	unsigned int indirect_shadow_pages;
568 	unsigned long mmu_valid_gen;
569 	struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
570 	/*
571 	 * Hash table of struct kvm_mmu_page.
572 	 */
573 	struct list_head active_mmu_pages;
574 	struct list_head zapped_obsolete_pages;
575 
576 	struct list_head assigned_dev_head;
577 	struct iommu_domain *iommu_domain;
578 	bool iommu_noncoherent;
579 #define __KVM_HAVE_ARCH_NONCOHERENT_DMA
580 	atomic_t noncoherent_dma_count;
581 	struct kvm_pic *vpic;
582 	struct kvm_ioapic *vioapic;
583 	struct kvm_pit *vpit;
584 	int vapics_in_nmi_mode;
585 	struct mutex apic_map_lock;
586 	struct kvm_apic_map *apic_map;
587 
588 	unsigned int tss_addr;
589 	bool apic_access_page_done;
590 
591 	gpa_t wall_clock;
592 
593 	bool ept_identity_pagetable_done;
594 	gpa_t ept_identity_map_addr;
595 
596 	unsigned long irq_sources_bitmap;
597 	s64 kvmclock_offset;
598 	raw_spinlock_t tsc_write_lock;
599 	u64 last_tsc_nsec;
600 	u64 last_tsc_write;
601 	u32 last_tsc_khz;
602 	u64 cur_tsc_nsec;
603 	u64 cur_tsc_write;
604 	u64 cur_tsc_offset;
605 	u64 cur_tsc_generation;
606 	int nr_vcpus_matched_tsc;
607 
608 	spinlock_t pvclock_gtod_sync_lock;
609 	bool use_master_clock;
610 	u64 master_kernel_ns;
611 	cycle_t master_cycle_now;
612 	struct delayed_work kvmclock_update_work;
613 	struct delayed_work kvmclock_sync_work;
614 
615 	struct kvm_xen_hvm_config xen_hvm_config;
616 
617 	/* reads protected by irq_srcu, writes by irq_lock */
618 	struct hlist_head mask_notifier_list;
619 
620 	/* fields used by HYPER-V emulation */
621 	u64 hv_guest_os_id;
622 	u64 hv_hypercall;
623 	u64 hv_tsc_page;
624 
625 	#ifdef CONFIG_KVM_MMU_AUDIT
626 	int audit_point;
627 	#endif
628 
629 	bool boot_vcpu_runs_old_kvmclock;
630 };
631 
632 struct kvm_vm_stat {
633 	u32 mmu_shadow_zapped;
634 	u32 mmu_pte_write;
635 	u32 mmu_pte_updated;
636 	u32 mmu_pde_zapped;
637 	u32 mmu_flooded;
638 	u32 mmu_recycled;
639 	u32 mmu_cache_miss;
640 	u32 mmu_unsync;
641 	u32 remote_tlb_flush;
642 	u32 lpages;
643 };
644 
645 struct kvm_vcpu_stat {
646 	u32 pf_fixed;
647 	u32 pf_guest;
648 	u32 tlb_flush;
649 	u32 invlpg;
650 
651 	u32 exits;
652 	u32 io_exits;
653 	u32 mmio_exits;
654 	u32 signal_exits;
655 	u32 irq_window_exits;
656 	u32 nmi_window_exits;
657 	u32 halt_exits;
658 	u32 halt_successful_poll;
659 	u32 halt_wakeup;
660 	u32 request_irq_exits;
661 	u32 irq_exits;
662 	u32 host_state_reload;
663 	u32 efer_reload;
664 	u32 fpu_reload;
665 	u32 insn_emulation;
666 	u32 insn_emulation_fail;
667 	u32 hypercalls;
668 	u32 irq_injections;
669 	u32 nmi_injections;
670 };
671 
672 struct x86_instruction_info;
673 
674 struct msr_data {
675 	bool host_initiated;
676 	u32 index;
677 	u64 data;
678 };
679 
680 struct kvm_lapic_irq {
681 	u32 vector;
682 	u32 delivery_mode;
683 	u32 dest_mode;
684 	u32 level;
685 	u32 trig_mode;
686 	u32 shorthand;
687 	u32 dest_id;
688 };
689 
690 struct kvm_x86_ops {
691 	int (*cpu_has_kvm_support)(void);          /* __init */
692 	int (*disabled_by_bios)(void);             /* __init */
693 	int (*hardware_enable)(void);
694 	void (*hardware_disable)(void);
695 	void (*check_processor_compatibility)(void *rtn);
696 	int (*hardware_setup)(void);               /* __init */
697 	void (*hardware_unsetup)(void);            /* __exit */
698 	bool (*cpu_has_accelerated_tpr)(void);
699 	void (*cpuid_update)(struct kvm_vcpu *vcpu);
700 
701 	/* Create, but do not attach this VCPU */
702 	struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
703 	void (*vcpu_free)(struct kvm_vcpu *vcpu);
704 	void (*vcpu_reset)(struct kvm_vcpu *vcpu);
705 
706 	void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
707 	void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
708 	void (*vcpu_put)(struct kvm_vcpu *vcpu);
709 
710 	void (*update_db_bp_intercept)(struct kvm_vcpu *vcpu);
711 	int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
712 	int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
713 	u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
714 	void (*get_segment)(struct kvm_vcpu *vcpu,
715 			    struct kvm_segment *var, int seg);
716 	int (*get_cpl)(struct kvm_vcpu *vcpu);
717 	void (*set_segment)(struct kvm_vcpu *vcpu,
718 			    struct kvm_segment *var, int seg);
719 	void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
720 	void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
721 	void (*decache_cr3)(struct kvm_vcpu *vcpu);
722 	void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
723 	void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
724 	void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
725 	int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
726 	void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
727 	void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
728 	void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
729 	void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
730 	void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
731 	u64 (*get_dr6)(struct kvm_vcpu *vcpu);
732 	void (*set_dr6)(struct kvm_vcpu *vcpu, unsigned long value);
733 	void (*sync_dirty_debug_regs)(struct kvm_vcpu *vcpu);
734 	void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
735 	void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
736 	unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
737 	void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
738 	void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
739 
740 	void (*tlb_flush)(struct kvm_vcpu *vcpu);
741 
742 	void (*run)(struct kvm_vcpu *vcpu);
743 	int (*handle_exit)(struct kvm_vcpu *vcpu);
744 	void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
745 	void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
746 	u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu);
747 	void (*patch_hypercall)(struct kvm_vcpu *vcpu,
748 				unsigned char *hypercall_addr);
749 	void (*set_irq)(struct kvm_vcpu *vcpu);
750 	void (*set_nmi)(struct kvm_vcpu *vcpu);
751 	void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
752 				bool has_error_code, u32 error_code,
753 				bool reinject);
754 	void (*cancel_injection)(struct kvm_vcpu *vcpu);
755 	int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
756 	int (*nmi_allowed)(struct kvm_vcpu *vcpu);
757 	bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
758 	void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
759 	void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
760 	void (*enable_irq_window)(struct kvm_vcpu *vcpu);
761 	void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
762 	int (*vm_has_apicv)(struct kvm *kvm);
763 	void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
764 	void (*hwapic_isr_update)(struct kvm *kvm, int isr);
765 	void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap);
766 	void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu, bool set);
767 	void (*set_apic_access_page_addr)(struct kvm_vcpu *vcpu, hpa_t hpa);
768 	void (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
769 	void (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
770 	int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
771 	int (*get_tdp_level)(void);
772 	u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
773 	int (*get_lpage_level)(void);
774 	bool (*rdtscp_supported)(void);
775 	bool (*invpcid_supported)(void);
776 	void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment, bool host);
777 
778 	void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
779 
780 	void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
781 
782 	bool (*has_wbinvd_exit)(void);
783 
784 	void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale);
785 	u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu);
786 	void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
787 
788 	u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc);
789 	u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu, u64 host_tsc);
790 
791 	void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
792 
793 	int (*check_intercept)(struct kvm_vcpu *vcpu,
794 			       struct x86_instruction_info *info,
795 			       enum x86_intercept_stage stage);
796 	void (*handle_external_intr)(struct kvm_vcpu *vcpu);
797 	bool (*mpx_supported)(void);
798 	bool (*xsaves_supported)(void);
799 
800 	int (*check_nested_events)(struct kvm_vcpu *vcpu, bool external_intr);
801 
802 	void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
803 
804 	/*
805 	 * Arch-specific dirty logging hooks. These hooks are only supposed to
806 	 * be valid if the specific arch has hardware-accelerated dirty logging
807 	 * mechanism. Currently only for PML on VMX.
808 	 *
809 	 *  - slot_enable_log_dirty:
810 	 *	called when enabling log dirty mode for the slot.
811 	 *  - slot_disable_log_dirty:
812 	 *	called when disabling log dirty mode for the slot.
813 	 *	also called when slot is created with log dirty disabled.
814 	 *  - flush_log_dirty:
815 	 *	called before reporting dirty_bitmap to userspace.
816 	 *  - enable_log_dirty_pt_masked:
817 	 *	called when reenabling log dirty for the GFNs in the mask after
818 	 *	corresponding bits are cleared in slot->dirty_bitmap.
819 	 */
820 	void (*slot_enable_log_dirty)(struct kvm *kvm,
821 				      struct kvm_memory_slot *slot);
822 	void (*slot_disable_log_dirty)(struct kvm *kvm,
823 				       struct kvm_memory_slot *slot);
824 	void (*flush_log_dirty)(struct kvm *kvm);
825 	void (*enable_log_dirty_pt_masked)(struct kvm *kvm,
826 					   struct kvm_memory_slot *slot,
827 					   gfn_t offset, unsigned long mask);
828 };
829 
830 struct kvm_arch_async_pf {
831 	u32 token;
832 	gfn_t gfn;
833 	unsigned long cr3;
834 	bool direct_map;
835 };
836 
837 extern struct kvm_x86_ops *kvm_x86_ops;
838 
839 static inline void adjust_tsc_offset_guest(struct kvm_vcpu *vcpu,
840 					   s64 adjustment)
841 {
842 	kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, false);
843 }
844 
845 static inline void adjust_tsc_offset_host(struct kvm_vcpu *vcpu, s64 adjustment)
846 {
847 	kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, true);
848 }
849 
850 int kvm_mmu_module_init(void);
851 void kvm_mmu_module_exit(void);
852 
853 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
854 int kvm_mmu_create(struct kvm_vcpu *vcpu);
855 void kvm_mmu_setup(struct kvm_vcpu *vcpu);
856 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
857 		u64 dirty_mask, u64 nx_mask, u64 x_mask);
858 
859 void kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
860 void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
861 				      struct kvm_memory_slot *memslot);
862 void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
863 				   struct kvm_memory_slot *memslot);
864 void kvm_mmu_slot_largepage_remove_write_access(struct kvm *kvm,
865 					struct kvm_memory_slot *memslot);
866 void kvm_mmu_slot_set_dirty(struct kvm *kvm,
867 			    struct kvm_memory_slot *memslot);
868 void kvm_mmu_clear_dirty_pt_masked(struct kvm *kvm,
869 				   struct kvm_memory_slot *slot,
870 				   gfn_t gfn_offset, unsigned long mask);
871 void kvm_mmu_zap_all(struct kvm *kvm);
872 void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm);
873 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
874 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
875 
876 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
877 
878 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
879 			  const void *val, int bytes);
880 u8 kvm_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn);
881 
882 struct kvm_irq_mask_notifier {
883 	void (*func)(struct kvm_irq_mask_notifier *kimn, bool masked);
884 	int irq;
885 	struct hlist_node link;
886 };
887 
888 void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
889 				    struct kvm_irq_mask_notifier *kimn);
890 void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
891 				      struct kvm_irq_mask_notifier *kimn);
892 void kvm_fire_mask_notifiers(struct kvm *kvm, unsigned irqchip, unsigned pin,
893 			     bool mask);
894 
895 extern bool tdp_enabled;
896 
897 u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
898 
899 /* control of guest tsc rate supported? */
900 extern bool kvm_has_tsc_control;
901 /* minimum supported tsc_khz for guests */
902 extern u32  kvm_min_guest_tsc_khz;
903 /* maximum supported tsc_khz for guests */
904 extern u32  kvm_max_guest_tsc_khz;
905 
906 enum emulation_result {
907 	EMULATE_DONE,         /* no further processing */
908 	EMULATE_USER_EXIT,    /* kvm_run ready for userspace exit */
909 	EMULATE_FAIL,         /* can't emulate this instruction */
910 };
911 
912 #define EMULTYPE_NO_DECODE	    (1 << 0)
913 #define EMULTYPE_TRAP_UD	    (1 << 1)
914 #define EMULTYPE_SKIP		    (1 << 2)
915 #define EMULTYPE_RETRY		    (1 << 3)
916 #define EMULTYPE_NO_REEXECUTE	    (1 << 4)
917 int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
918 			    int emulation_type, void *insn, int insn_len);
919 
920 static inline int emulate_instruction(struct kvm_vcpu *vcpu,
921 			int emulation_type)
922 {
923 	return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
924 }
925 
926 void kvm_enable_efer_bits(u64);
927 bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
928 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
929 int kvm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
930 
931 struct x86_emulate_ctxt;
932 
933 int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
934 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
935 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
936 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
937 
938 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
939 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
940 void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
941 
942 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
943 		    int reason, bool has_error_code, u32 error_code);
944 
945 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
946 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
947 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
948 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
949 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
950 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
951 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
952 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
953 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
954 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
955 
956 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
957 int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
958 
959 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
960 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
961 bool kvm_rdpmc(struct kvm_vcpu *vcpu);
962 
963 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
964 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
965 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
966 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
967 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
968 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
969 			    gfn_t gfn, void *data, int offset, int len,
970 			    u32 access);
971 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
972 bool kvm_require_dr(struct kvm_vcpu *vcpu, int dr);
973 
974 static inline int __kvm_irq_line_state(unsigned long *irq_state,
975 				       int irq_source_id, int level)
976 {
977 	/* Logical OR for level trig interrupt */
978 	if (level)
979 		__set_bit(irq_source_id, irq_state);
980 	else
981 		__clear_bit(irq_source_id, irq_state);
982 
983 	return !!(*irq_state);
984 }
985 
986 int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
987 void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
988 
989 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
990 
991 int fx_init(struct kvm_vcpu *vcpu);
992 
993 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
994 		       const u8 *new, int bytes);
995 int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
996 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
997 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
998 int kvm_mmu_load(struct kvm_vcpu *vcpu);
999 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
1000 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
1001 gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1002 			   struct x86_exception *exception);
1003 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
1004 			      struct x86_exception *exception);
1005 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
1006 			       struct x86_exception *exception);
1007 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
1008 			       struct x86_exception *exception);
1009 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
1010 				struct x86_exception *exception);
1011 
1012 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
1013 
1014 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
1015 		       void *insn, int insn_len);
1016 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
1017 void kvm_mmu_new_cr3(struct kvm_vcpu *vcpu);
1018 
1019 void kvm_enable_tdp(void);
1020 void kvm_disable_tdp(void);
1021 
1022 static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1023 				  struct x86_exception *exception)
1024 {
1025 	return gpa;
1026 }
1027 
1028 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
1029 {
1030 	struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
1031 
1032 	return (struct kvm_mmu_page *)page_private(page);
1033 }
1034 
1035 static inline u16 kvm_read_ldt(void)
1036 {
1037 	u16 ldt;
1038 	asm("sldt %0" : "=g"(ldt));
1039 	return ldt;
1040 }
1041 
1042 static inline void kvm_load_ldt(u16 sel)
1043 {
1044 	asm("lldt %0" : : "rm"(sel));
1045 }
1046 
1047 #ifdef CONFIG_X86_64
1048 static inline unsigned long read_msr(unsigned long msr)
1049 {
1050 	u64 value;
1051 
1052 	rdmsrl(msr, value);
1053 	return value;
1054 }
1055 #endif
1056 
1057 static inline u32 get_rdx_init_val(void)
1058 {
1059 	return 0x600; /* P6 family */
1060 }
1061 
1062 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
1063 {
1064 	kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
1065 }
1066 
1067 static inline u64 get_canonical(u64 la)
1068 {
1069 	return ((int64_t)la << 16) >> 16;
1070 }
1071 
1072 static inline bool is_noncanonical_address(u64 la)
1073 {
1074 #ifdef CONFIG_X86_64
1075 	return get_canonical(la) != la;
1076 #else
1077 	return false;
1078 #endif
1079 }
1080 
1081 #define TSS_IOPB_BASE_OFFSET 0x66
1082 #define TSS_BASE_SIZE 0x68
1083 #define TSS_IOPB_SIZE (65536 / 8)
1084 #define TSS_REDIRECTION_SIZE (256 / 8)
1085 #define RMODE_TSS_SIZE							\
1086 	(TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
1087 
1088 enum {
1089 	TASK_SWITCH_CALL = 0,
1090 	TASK_SWITCH_IRET = 1,
1091 	TASK_SWITCH_JMP = 2,
1092 	TASK_SWITCH_GATE = 3,
1093 };
1094 
1095 #define HF_GIF_MASK		(1 << 0)
1096 #define HF_HIF_MASK		(1 << 1)
1097 #define HF_VINTR_MASK		(1 << 2)
1098 #define HF_NMI_MASK		(1 << 3)
1099 #define HF_IRET_MASK		(1 << 4)
1100 #define HF_GUEST_MASK		(1 << 5) /* VCPU is in guest-mode */
1101 
1102 /*
1103  * Hardware virtualization extension instructions may fault if a
1104  * reboot turns off virtualization while processes are running.
1105  * Trap the fault and ignore the instruction if that happens.
1106  */
1107 asmlinkage void kvm_spurious_fault(void);
1108 
1109 #define ____kvm_handle_fault_on_reboot(insn, cleanup_insn)	\
1110 	"666: " insn "\n\t" \
1111 	"668: \n\t"                           \
1112 	".pushsection .fixup, \"ax\" \n" \
1113 	"667: \n\t" \
1114 	cleanup_insn "\n\t"		      \
1115 	"cmpb $0, kvm_rebooting \n\t"	      \
1116 	"jne 668b \n\t"      		      \
1117 	__ASM_SIZE(push) " $666b \n\t"	      \
1118 	"call kvm_spurious_fault \n\t"	      \
1119 	".popsection \n\t" \
1120 	_ASM_EXTABLE(666b, 667b)
1121 
1122 #define __kvm_handle_fault_on_reboot(insn)		\
1123 	____kvm_handle_fault_on_reboot(insn, "")
1124 
1125 #define KVM_ARCH_WANT_MMU_NOTIFIER
1126 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
1127 int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
1128 int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
1129 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
1130 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
1131 int cpuid_maxphyaddr(struct kvm_vcpu *vcpu);
1132 int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
1133 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1134 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
1135 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
1136 void kvm_vcpu_reset(struct kvm_vcpu *vcpu);
1137 void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu);
1138 void kvm_arch_mmu_notifier_invalidate_page(struct kvm *kvm,
1139 					   unsigned long address);
1140 
1141 void kvm_define_shared_msr(unsigned index, u32 msr);
1142 int kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
1143 
1144 unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
1145 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1146 
1147 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1148 				     struct kvm_async_pf *work);
1149 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1150 				 struct kvm_async_pf *work);
1151 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1152 			       struct kvm_async_pf *work);
1153 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
1154 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1155 
1156 void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1157 
1158 int kvm_is_in_guest(void);
1159 
1160 void kvm_pmu_init(struct kvm_vcpu *vcpu);
1161 void kvm_pmu_destroy(struct kvm_vcpu *vcpu);
1162 void kvm_pmu_reset(struct kvm_vcpu *vcpu);
1163 void kvm_pmu_cpuid_update(struct kvm_vcpu *vcpu);
1164 bool kvm_pmu_msr(struct kvm_vcpu *vcpu, u32 msr);
1165 int kvm_pmu_get_msr(struct kvm_vcpu *vcpu, u32 msr, u64 *data);
1166 int kvm_pmu_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info);
1167 int kvm_pmu_check_pmc(struct kvm_vcpu *vcpu, unsigned pmc);
1168 int kvm_pmu_read_pmc(struct kvm_vcpu *vcpu, unsigned pmc, u64 *data);
1169 void kvm_handle_pmu_event(struct kvm_vcpu *vcpu);
1170 void kvm_deliver_pmi(struct kvm_vcpu *vcpu);
1171 
1172 #endif /* _ASM_X86_KVM_HOST_H */
1173