xref: /openbmc/linux/arch/x86/include/asm/kvm_host.h (revision 3d3337de)
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 KVM_PERMILLE_MMU_PAGES 20
85 #define KVM_MIN_ALLOC_MMU_PAGES 64
86 #define KVM_MMU_HASH_SHIFT 10
87 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
88 #define KVM_MIN_FREE_MMU_PAGES 5
89 #define KVM_REFILL_PAGES 25
90 #define KVM_MAX_CPUID_ENTRIES 80
91 #define KVM_NR_FIXED_MTRR_REGION 88
92 #define KVM_NR_VAR_MTRR 8
93 
94 #define ASYNC_PF_PER_VCPU 64
95 
96 enum kvm_reg {
97 	VCPU_REGS_RAX = 0,
98 	VCPU_REGS_RCX = 1,
99 	VCPU_REGS_RDX = 2,
100 	VCPU_REGS_RBX = 3,
101 	VCPU_REGS_RSP = 4,
102 	VCPU_REGS_RBP = 5,
103 	VCPU_REGS_RSI = 6,
104 	VCPU_REGS_RDI = 7,
105 #ifdef CONFIG_X86_64
106 	VCPU_REGS_R8 = 8,
107 	VCPU_REGS_R9 = 9,
108 	VCPU_REGS_R10 = 10,
109 	VCPU_REGS_R11 = 11,
110 	VCPU_REGS_R12 = 12,
111 	VCPU_REGS_R13 = 13,
112 	VCPU_REGS_R14 = 14,
113 	VCPU_REGS_R15 = 15,
114 #endif
115 	VCPU_REGS_RIP,
116 	NR_VCPU_REGS
117 };
118 
119 enum kvm_reg_ex {
120 	VCPU_EXREG_PDPTR = NR_VCPU_REGS,
121 	VCPU_EXREG_CR3,
122 	VCPU_EXREG_RFLAGS,
123 	VCPU_EXREG_SEGMENTS,
124 };
125 
126 enum {
127 	VCPU_SREG_ES,
128 	VCPU_SREG_CS,
129 	VCPU_SREG_SS,
130 	VCPU_SREG_DS,
131 	VCPU_SREG_FS,
132 	VCPU_SREG_GS,
133 	VCPU_SREG_TR,
134 	VCPU_SREG_LDTR,
135 };
136 
137 #include <asm/kvm_emulate.h>
138 
139 #define KVM_NR_MEM_OBJS 40
140 
141 #define KVM_NR_DB_REGS	4
142 
143 #define DR6_BD		(1 << 13)
144 #define DR6_BS		(1 << 14)
145 #define DR6_RTM		(1 << 16)
146 #define DR6_FIXED_1	0xfffe0ff0
147 #define DR6_INIT	0xffff0ff0
148 #define DR6_VOLATILE	0x0001e00f
149 
150 #define DR7_BP_EN_MASK	0x000000ff
151 #define DR7_GE		(1 << 9)
152 #define DR7_GD		(1 << 13)
153 #define DR7_FIXED_1	0x00000400
154 #define DR7_VOLATILE	0xffff2bff
155 
156 #define PFERR_PRESENT_BIT 0
157 #define PFERR_WRITE_BIT 1
158 #define PFERR_USER_BIT 2
159 #define PFERR_RSVD_BIT 3
160 #define PFERR_FETCH_BIT 4
161 
162 #define PFERR_PRESENT_MASK (1U << PFERR_PRESENT_BIT)
163 #define PFERR_WRITE_MASK (1U << PFERR_WRITE_BIT)
164 #define PFERR_USER_MASK (1U << PFERR_USER_BIT)
165 #define PFERR_RSVD_MASK (1U << PFERR_RSVD_BIT)
166 #define PFERR_FETCH_MASK (1U << PFERR_FETCH_BIT)
167 
168 /* apic attention bits */
169 #define KVM_APIC_CHECK_VAPIC	0
170 /*
171  * The following bit is set with PV-EOI, unset on EOI.
172  * We detect PV-EOI changes by guest by comparing
173  * this bit with PV-EOI in guest memory.
174  * See the implementation in apic_update_pv_eoi.
175  */
176 #define KVM_APIC_PV_EOI_PENDING	1
177 
178 /*
179  * We don't want allocation failures within the mmu code, so we preallocate
180  * enough memory for a single page fault in a cache.
181  */
182 struct kvm_mmu_memory_cache {
183 	int nobjs;
184 	void *objects[KVM_NR_MEM_OBJS];
185 };
186 
187 /*
188  * kvm_mmu_page_role, below, is defined as:
189  *
190  *   bits 0:3 - total guest paging levels (2-4, or zero for real mode)
191  *   bits 4:7 - page table level for this shadow (1-4)
192  *   bits 8:9 - page table quadrant for 2-level guests
193  *   bit   16 - direct mapping of virtual to physical mapping at gfn
194  *              used for real mode and two-dimensional paging
195  *   bits 17:19 - common access permissions for all ptes in this shadow page
196  */
197 union kvm_mmu_page_role {
198 	unsigned word;
199 	struct {
200 		unsigned level:4;
201 		unsigned cr4_pae:1;
202 		unsigned quadrant:2;
203 		unsigned pad_for_nice_hex_output:6;
204 		unsigned direct:1;
205 		unsigned access:3;
206 		unsigned invalid:1;
207 		unsigned nxe:1;
208 		unsigned cr0_wp:1;
209 		unsigned smep_andnot_wp:1;
210 	};
211 };
212 
213 struct kvm_mmu_page {
214 	struct list_head link;
215 	struct hlist_node hash_link;
216 
217 	/*
218 	 * The following two entries are used to key the shadow page in the
219 	 * hash table.
220 	 */
221 	gfn_t gfn;
222 	union kvm_mmu_page_role role;
223 
224 	u64 *spt;
225 	/* hold the gfn of each spte inside spt */
226 	gfn_t *gfns;
227 	bool unsync;
228 	int root_count;          /* Currently serving as active root */
229 	unsigned int unsync_children;
230 	unsigned long parent_ptes;	/* Reverse mapping for parent_pte */
231 
232 	/* The page is obsolete if mmu_valid_gen != kvm->arch.mmu_valid_gen.  */
233 	unsigned long mmu_valid_gen;
234 
235 	DECLARE_BITMAP(unsync_child_bitmap, 512);
236 
237 #ifdef CONFIG_X86_32
238 	/*
239 	 * Used out of the mmu-lock to avoid reading spte values while an
240 	 * update is in progress; see the comments in __get_spte_lockless().
241 	 */
242 	int clear_spte_count;
243 #endif
244 
245 	/* Number of writes since the last time traversal visited this page.  */
246 	int write_flooding_count;
247 };
248 
249 struct kvm_pio_request {
250 	unsigned long count;
251 	int in;
252 	int port;
253 	int size;
254 };
255 
256 /*
257  * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
258  * 32-bit).  The kvm_mmu structure abstracts the details of the current mmu
259  * mode.
260  */
261 struct kvm_mmu {
262 	void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
263 	unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
264 	u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
265 	int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
266 			  bool prefault);
267 	void (*inject_page_fault)(struct kvm_vcpu *vcpu,
268 				  struct x86_exception *fault);
269 	gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
270 			    struct x86_exception *exception);
271 	gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
272 			       struct x86_exception *exception);
273 	int (*sync_page)(struct kvm_vcpu *vcpu,
274 			 struct kvm_mmu_page *sp);
275 	void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
276 	void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
277 			   u64 *spte, const void *pte);
278 	hpa_t root_hpa;
279 	int root_level;
280 	int shadow_root_level;
281 	union kvm_mmu_page_role base_role;
282 	bool direct_map;
283 
284 	/*
285 	 * Bitmap; bit set = permission fault
286 	 * Byte index: page fault error code [4:1]
287 	 * Bit index: pte permissions in ACC_* format
288 	 */
289 	u8 permissions[16];
290 
291 	u64 *pae_root;
292 	u64 *lm_root;
293 	u64 rsvd_bits_mask[2][4];
294 	u64 bad_mt_xwr;
295 
296 	/*
297 	 * Bitmap: bit set = last pte in walk
298 	 * index[0:1]: level (zero-based)
299 	 * index[2]: pte.ps
300 	 */
301 	u8 last_pte_bitmap;
302 
303 	bool nx;
304 
305 	u64 pdptrs[4]; /* pae */
306 };
307 
308 enum pmc_type {
309 	KVM_PMC_GP = 0,
310 	KVM_PMC_FIXED,
311 };
312 
313 struct kvm_pmc {
314 	enum pmc_type type;
315 	u8 idx;
316 	u64 counter;
317 	u64 eventsel;
318 	struct perf_event *perf_event;
319 	struct kvm_vcpu *vcpu;
320 };
321 
322 struct kvm_pmu {
323 	unsigned nr_arch_gp_counters;
324 	unsigned nr_arch_fixed_counters;
325 	unsigned available_event_types;
326 	u64 fixed_ctr_ctrl;
327 	u64 global_ctrl;
328 	u64 global_status;
329 	u64 global_ovf_ctrl;
330 	u64 counter_bitmask[2];
331 	u64 global_ctrl_mask;
332 	u64 reserved_bits;
333 	u8 version;
334 	struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
335 	struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
336 	struct irq_work irq_work;
337 	u64 reprogram_pmi;
338 };
339 
340 enum {
341 	KVM_DEBUGREG_BP_ENABLED = 1,
342 	KVM_DEBUGREG_WONT_EXIT = 2,
343 	KVM_DEBUGREG_RELOAD = 4,
344 };
345 
346 struct kvm_vcpu_arch {
347 	/*
348 	 * rip and regs accesses must go through
349 	 * kvm_{register,rip}_{read,write} functions.
350 	 */
351 	unsigned long regs[NR_VCPU_REGS];
352 	u32 regs_avail;
353 	u32 regs_dirty;
354 
355 	unsigned long cr0;
356 	unsigned long cr0_guest_owned_bits;
357 	unsigned long cr2;
358 	unsigned long cr3;
359 	unsigned long cr4;
360 	unsigned long cr4_guest_owned_bits;
361 	unsigned long cr8;
362 	u32 hflags;
363 	u64 efer;
364 	u64 apic_base;
365 	struct kvm_lapic *apic;    /* kernel irqchip context */
366 	unsigned long apic_attention;
367 	int32_t apic_arb_prio;
368 	int mp_state;
369 	u64 ia32_misc_enable_msr;
370 	bool tpr_access_reporting;
371 	u64 ia32_xss;
372 
373 	/*
374 	 * Paging state of the vcpu
375 	 *
376 	 * If the vcpu runs in guest mode with two level paging this still saves
377 	 * the paging mode of the l1 guest. This context is always used to
378 	 * handle faults.
379 	 */
380 	struct kvm_mmu mmu;
381 
382 	/*
383 	 * Paging state of an L2 guest (used for nested npt)
384 	 *
385 	 * This context will save all necessary information to walk page tables
386 	 * of the an L2 guest. This context is only initialized for page table
387 	 * walking and not for faulting since we never handle l2 page faults on
388 	 * the host.
389 	 */
390 	struct kvm_mmu nested_mmu;
391 
392 	/*
393 	 * Pointer to the mmu context currently used for
394 	 * gva_to_gpa translations.
395 	 */
396 	struct kvm_mmu *walk_mmu;
397 
398 	struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
399 	struct kvm_mmu_memory_cache mmu_page_cache;
400 	struct kvm_mmu_memory_cache mmu_page_header_cache;
401 
402 	struct fpu guest_fpu;
403 	u64 xcr0;
404 	u64 guest_supported_xcr0;
405 	u32 guest_xstate_size;
406 
407 	struct kvm_pio_request pio;
408 	void *pio_data;
409 
410 	u8 event_exit_inst_len;
411 
412 	struct kvm_queued_exception {
413 		bool pending;
414 		bool has_error_code;
415 		bool reinject;
416 		u8 nr;
417 		u32 error_code;
418 	} exception;
419 
420 	struct kvm_queued_interrupt {
421 		bool pending;
422 		bool soft;
423 		u8 nr;
424 	} interrupt;
425 
426 	int halt_request; /* real mode on Intel only */
427 
428 	int cpuid_nent;
429 	struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
430 
431 	int maxphyaddr;
432 
433 	/* emulate context */
434 
435 	struct x86_emulate_ctxt emulate_ctxt;
436 	bool emulate_regs_need_sync_to_vcpu;
437 	bool emulate_regs_need_sync_from_vcpu;
438 	int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
439 
440 	gpa_t time;
441 	struct pvclock_vcpu_time_info hv_clock;
442 	unsigned int hw_tsc_khz;
443 	struct gfn_to_hva_cache pv_time;
444 	bool pv_time_enabled;
445 	/* set guest stopped flag in pvclock flags field */
446 	bool pvclock_set_guest_stopped_request;
447 
448 	struct {
449 		u64 msr_val;
450 		u64 last_steal;
451 		u64 accum_steal;
452 		struct gfn_to_hva_cache stime;
453 		struct kvm_steal_time steal;
454 	} st;
455 
456 	u64 last_guest_tsc;
457 	u64 last_host_tsc;
458 	u64 tsc_offset_adjustment;
459 	u64 this_tsc_nsec;
460 	u64 this_tsc_write;
461 	u64 this_tsc_generation;
462 	bool tsc_catchup;
463 	bool tsc_always_catchup;
464 	s8 virtual_tsc_shift;
465 	u32 virtual_tsc_mult;
466 	u32 virtual_tsc_khz;
467 	s64 ia32_tsc_adjust_msr;
468 
469 	atomic_t nmi_queued;  /* unprocessed asynchronous NMIs */
470 	unsigned nmi_pending; /* NMI queued after currently running handler */
471 	bool nmi_injected;    /* Trying to inject an NMI this entry */
472 
473 	struct mtrr_state_type mtrr_state;
474 	u64 pat;
475 
476 	unsigned switch_db_regs;
477 	unsigned long db[KVM_NR_DB_REGS];
478 	unsigned long dr6;
479 	unsigned long dr7;
480 	unsigned long eff_db[KVM_NR_DB_REGS];
481 	unsigned long guest_debug_dr7;
482 
483 	u64 mcg_cap;
484 	u64 mcg_status;
485 	u64 mcg_ctl;
486 	u64 *mce_banks;
487 
488 	/* Cache MMIO info */
489 	u64 mmio_gva;
490 	unsigned access;
491 	gfn_t mmio_gfn;
492 	u64 mmio_gen;
493 
494 	struct kvm_pmu pmu;
495 
496 	/* used for guest single stepping over the given code position */
497 	unsigned long singlestep_rip;
498 
499 	/* fields used by HYPER-V emulation */
500 	u64 hv_vapic;
501 
502 	cpumask_var_t wbinvd_dirty_mask;
503 
504 	unsigned long last_retry_eip;
505 	unsigned long last_retry_addr;
506 
507 	struct {
508 		bool halted;
509 		gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
510 		struct gfn_to_hva_cache data;
511 		u64 msr_val;
512 		u32 id;
513 		bool send_user_only;
514 	} apf;
515 
516 	/* OSVW MSRs (AMD only) */
517 	struct {
518 		u64 length;
519 		u64 status;
520 	} osvw;
521 
522 	struct {
523 		u64 msr_val;
524 		struct gfn_to_hva_cache data;
525 	} pv_eoi;
526 
527 	/*
528 	 * Indicate whether the access faults on its page table in guest
529 	 * which is set when fix page fault and used to detect unhandeable
530 	 * instruction.
531 	 */
532 	bool write_fault_to_shadow_pgtable;
533 
534 	/* set at EPT violation at this point */
535 	unsigned long exit_qualification;
536 
537 	/* pv related host specific info */
538 	struct {
539 		bool pv_unhalted;
540 	} pv;
541 };
542 
543 struct kvm_lpage_info {
544 	int write_count;
545 };
546 
547 struct kvm_arch_memory_slot {
548 	unsigned long *rmap[KVM_NR_PAGE_SIZES];
549 	struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
550 };
551 
552 /*
553  * We use as the mode the number of bits allocated in the LDR for the
554  * logical processor ID.  It happens that these are all powers of two.
555  * This makes it is very easy to detect cases where the APICs are
556  * configured for multiple modes; in that case, we cannot use the map and
557  * hence cannot use kvm_irq_delivery_to_apic_fast either.
558  */
559 #define KVM_APIC_MODE_XAPIC_CLUSTER          4
560 #define KVM_APIC_MODE_XAPIC_FLAT             8
561 #define KVM_APIC_MODE_X2APIC                16
562 
563 struct kvm_apic_map {
564 	struct rcu_head rcu;
565 	u8 mode;
566 	struct kvm_lapic *phys_map[256];
567 	/* first index is cluster id second is cpu id in a cluster */
568 	struct kvm_lapic *logical_map[16][16];
569 };
570 
571 struct kvm_arch {
572 	unsigned int n_used_mmu_pages;
573 	unsigned int n_requested_mmu_pages;
574 	unsigned int n_max_mmu_pages;
575 	unsigned int indirect_shadow_pages;
576 	unsigned long mmu_valid_gen;
577 	struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
578 	/*
579 	 * Hash table of struct kvm_mmu_page.
580 	 */
581 	struct list_head active_mmu_pages;
582 	struct list_head zapped_obsolete_pages;
583 
584 	struct list_head assigned_dev_head;
585 	struct iommu_domain *iommu_domain;
586 	bool iommu_noncoherent;
587 #define __KVM_HAVE_ARCH_NONCOHERENT_DMA
588 	atomic_t noncoherent_dma_count;
589 	struct kvm_pic *vpic;
590 	struct kvm_ioapic *vioapic;
591 	struct kvm_pit *vpit;
592 	int vapics_in_nmi_mode;
593 	struct mutex apic_map_lock;
594 	struct kvm_apic_map *apic_map;
595 
596 	unsigned int tss_addr;
597 	bool apic_access_page_done;
598 
599 	gpa_t wall_clock;
600 
601 	bool ept_identity_pagetable_done;
602 	gpa_t ept_identity_map_addr;
603 
604 	unsigned long irq_sources_bitmap;
605 	s64 kvmclock_offset;
606 	raw_spinlock_t tsc_write_lock;
607 	u64 last_tsc_nsec;
608 	u64 last_tsc_write;
609 	u32 last_tsc_khz;
610 	u64 cur_tsc_nsec;
611 	u64 cur_tsc_write;
612 	u64 cur_tsc_offset;
613 	u64 cur_tsc_generation;
614 	int nr_vcpus_matched_tsc;
615 
616 	spinlock_t pvclock_gtod_sync_lock;
617 	bool use_master_clock;
618 	u64 master_kernel_ns;
619 	cycle_t master_cycle_now;
620 	struct delayed_work kvmclock_update_work;
621 	struct delayed_work kvmclock_sync_work;
622 
623 	struct kvm_xen_hvm_config xen_hvm_config;
624 
625 	/* reads protected by irq_srcu, writes by irq_lock */
626 	struct hlist_head mask_notifier_list;
627 
628 	/* fields used by HYPER-V emulation */
629 	u64 hv_guest_os_id;
630 	u64 hv_hypercall;
631 	u64 hv_tsc_page;
632 
633 	#ifdef CONFIG_KVM_MMU_AUDIT
634 	int audit_point;
635 	#endif
636 
637 	bool boot_vcpu_runs_old_kvmclock;
638 };
639 
640 struct kvm_vm_stat {
641 	u32 mmu_shadow_zapped;
642 	u32 mmu_pte_write;
643 	u32 mmu_pte_updated;
644 	u32 mmu_pde_zapped;
645 	u32 mmu_flooded;
646 	u32 mmu_recycled;
647 	u32 mmu_cache_miss;
648 	u32 mmu_unsync;
649 	u32 remote_tlb_flush;
650 	u32 lpages;
651 };
652 
653 struct kvm_vcpu_stat {
654 	u32 pf_fixed;
655 	u32 pf_guest;
656 	u32 tlb_flush;
657 	u32 invlpg;
658 
659 	u32 exits;
660 	u32 io_exits;
661 	u32 mmio_exits;
662 	u32 signal_exits;
663 	u32 irq_window_exits;
664 	u32 nmi_window_exits;
665 	u32 halt_exits;
666 	u32 halt_successful_poll;
667 	u32 halt_wakeup;
668 	u32 request_irq_exits;
669 	u32 irq_exits;
670 	u32 host_state_reload;
671 	u32 efer_reload;
672 	u32 fpu_reload;
673 	u32 insn_emulation;
674 	u32 insn_emulation_fail;
675 	u32 hypercalls;
676 	u32 irq_injections;
677 	u32 nmi_injections;
678 };
679 
680 struct x86_instruction_info;
681 
682 struct msr_data {
683 	bool host_initiated;
684 	u32 index;
685 	u64 data;
686 };
687 
688 struct kvm_lapic_irq {
689 	u32 vector;
690 	u32 delivery_mode;
691 	u32 dest_mode;
692 	u32 level;
693 	u32 trig_mode;
694 	u32 shorthand;
695 	u32 dest_id;
696 };
697 
698 struct kvm_x86_ops {
699 	int (*cpu_has_kvm_support)(void);          /* __init */
700 	int (*disabled_by_bios)(void);             /* __init */
701 	int (*hardware_enable)(void);
702 	void (*hardware_disable)(void);
703 	void (*check_processor_compatibility)(void *rtn);
704 	int (*hardware_setup)(void);               /* __init */
705 	void (*hardware_unsetup)(void);            /* __exit */
706 	bool (*cpu_has_accelerated_tpr)(void);
707 	void (*cpuid_update)(struct kvm_vcpu *vcpu);
708 
709 	/* Create, but do not attach this VCPU */
710 	struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
711 	void (*vcpu_free)(struct kvm_vcpu *vcpu);
712 	void (*vcpu_reset)(struct kvm_vcpu *vcpu);
713 
714 	void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
715 	void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
716 	void (*vcpu_put)(struct kvm_vcpu *vcpu);
717 
718 	void (*update_db_bp_intercept)(struct kvm_vcpu *vcpu);
719 	int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
720 	int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
721 	u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
722 	void (*get_segment)(struct kvm_vcpu *vcpu,
723 			    struct kvm_segment *var, int seg);
724 	int (*get_cpl)(struct kvm_vcpu *vcpu);
725 	void (*set_segment)(struct kvm_vcpu *vcpu,
726 			    struct kvm_segment *var, int seg);
727 	void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
728 	void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
729 	void (*decache_cr3)(struct kvm_vcpu *vcpu);
730 	void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
731 	void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
732 	void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
733 	int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
734 	void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
735 	void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
736 	void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
737 	void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
738 	void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
739 	u64 (*get_dr6)(struct kvm_vcpu *vcpu);
740 	void (*set_dr6)(struct kvm_vcpu *vcpu, unsigned long value);
741 	void (*sync_dirty_debug_regs)(struct kvm_vcpu *vcpu);
742 	void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
743 	void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
744 	unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
745 	void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
746 	void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
747 
748 	void (*tlb_flush)(struct kvm_vcpu *vcpu);
749 
750 	void (*run)(struct kvm_vcpu *vcpu);
751 	int (*handle_exit)(struct kvm_vcpu *vcpu);
752 	void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
753 	void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
754 	u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu);
755 	void (*patch_hypercall)(struct kvm_vcpu *vcpu,
756 				unsigned char *hypercall_addr);
757 	void (*set_irq)(struct kvm_vcpu *vcpu);
758 	void (*set_nmi)(struct kvm_vcpu *vcpu);
759 	void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
760 				bool has_error_code, u32 error_code,
761 				bool reinject);
762 	void (*cancel_injection)(struct kvm_vcpu *vcpu);
763 	int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
764 	int (*nmi_allowed)(struct kvm_vcpu *vcpu);
765 	bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
766 	void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
767 	void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
768 	void (*enable_irq_window)(struct kvm_vcpu *vcpu);
769 	void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
770 	int (*vm_has_apicv)(struct kvm *kvm);
771 	void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
772 	void (*hwapic_isr_update)(struct kvm *kvm, int isr);
773 	void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap);
774 	void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu, bool set);
775 	void (*set_apic_access_page_addr)(struct kvm_vcpu *vcpu, hpa_t hpa);
776 	void (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
777 	void (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
778 	int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
779 	int (*get_tdp_level)(void);
780 	u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
781 	int (*get_lpage_level)(void);
782 	bool (*rdtscp_supported)(void);
783 	bool (*invpcid_supported)(void);
784 	void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment, bool host);
785 
786 	void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
787 
788 	void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
789 
790 	bool (*has_wbinvd_exit)(void);
791 
792 	void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale);
793 	u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu);
794 	void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
795 
796 	u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc);
797 	u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu, u64 host_tsc);
798 
799 	void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
800 
801 	int (*check_intercept)(struct kvm_vcpu *vcpu,
802 			       struct x86_instruction_info *info,
803 			       enum x86_intercept_stage stage);
804 	void (*handle_external_intr)(struct kvm_vcpu *vcpu);
805 	bool (*mpx_supported)(void);
806 	bool (*xsaves_supported)(void);
807 
808 	int (*check_nested_events)(struct kvm_vcpu *vcpu, bool external_intr);
809 
810 	void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
811 
812 	/*
813 	 * Arch-specific dirty logging hooks. These hooks are only supposed to
814 	 * be valid if the specific arch has hardware-accelerated dirty logging
815 	 * mechanism. Currently only for PML on VMX.
816 	 *
817 	 *  - slot_enable_log_dirty:
818 	 *	called when enabling log dirty mode for the slot.
819 	 *  - slot_disable_log_dirty:
820 	 *	called when disabling log dirty mode for the slot.
821 	 *	also called when slot is created with log dirty disabled.
822 	 *  - flush_log_dirty:
823 	 *	called before reporting dirty_bitmap to userspace.
824 	 *  - enable_log_dirty_pt_masked:
825 	 *	called when reenabling log dirty for the GFNs in the mask after
826 	 *	corresponding bits are cleared in slot->dirty_bitmap.
827 	 */
828 	void (*slot_enable_log_dirty)(struct kvm *kvm,
829 				      struct kvm_memory_slot *slot);
830 	void (*slot_disable_log_dirty)(struct kvm *kvm,
831 				       struct kvm_memory_slot *slot);
832 	void (*flush_log_dirty)(struct kvm *kvm);
833 	void (*enable_log_dirty_pt_masked)(struct kvm *kvm,
834 					   struct kvm_memory_slot *slot,
835 					   gfn_t offset, unsigned long mask);
836 };
837 
838 struct kvm_arch_async_pf {
839 	u32 token;
840 	gfn_t gfn;
841 	unsigned long cr3;
842 	bool direct_map;
843 };
844 
845 extern struct kvm_x86_ops *kvm_x86_ops;
846 
847 static inline void adjust_tsc_offset_guest(struct kvm_vcpu *vcpu,
848 					   s64 adjustment)
849 {
850 	kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, false);
851 }
852 
853 static inline void adjust_tsc_offset_host(struct kvm_vcpu *vcpu, s64 adjustment)
854 {
855 	kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, true);
856 }
857 
858 int kvm_mmu_module_init(void);
859 void kvm_mmu_module_exit(void);
860 
861 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
862 int kvm_mmu_create(struct kvm_vcpu *vcpu);
863 void kvm_mmu_setup(struct kvm_vcpu *vcpu);
864 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
865 		u64 dirty_mask, u64 nx_mask, u64 x_mask);
866 
867 void kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
868 void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
869 				      struct kvm_memory_slot *memslot);
870 void kvm_mmu_zap_collapsible_sptes(struct kvm *kvm,
871 					struct kvm_memory_slot *memslot);
872 void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
873 				   struct kvm_memory_slot *memslot);
874 void kvm_mmu_slot_largepage_remove_write_access(struct kvm *kvm,
875 					struct kvm_memory_slot *memslot);
876 void kvm_mmu_slot_set_dirty(struct kvm *kvm,
877 			    struct kvm_memory_slot *memslot);
878 void kvm_mmu_clear_dirty_pt_masked(struct kvm *kvm,
879 				   struct kvm_memory_slot *slot,
880 				   gfn_t gfn_offset, unsigned long mask);
881 void kvm_mmu_zap_all(struct kvm *kvm);
882 void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm);
883 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
884 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
885 
886 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
887 
888 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
889 			  const void *val, int bytes);
890 u8 kvm_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn);
891 
892 struct kvm_irq_mask_notifier {
893 	void (*func)(struct kvm_irq_mask_notifier *kimn, bool masked);
894 	int irq;
895 	struct hlist_node link;
896 };
897 
898 void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
899 				    struct kvm_irq_mask_notifier *kimn);
900 void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
901 				      struct kvm_irq_mask_notifier *kimn);
902 void kvm_fire_mask_notifiers(struct kvm *kvm, unsigned irqchip, unsigned pin,
903 			     bool mask);
904 
905 extern bool tdp_enabled;
906 
907 u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
908 
909 /* control of guest tsc rate supported? */
910 extern bool kvm_has_tsc_control;
911 /* minimum supported tsc_khz for guests */
912 extern u32  kvm_min_guest_tsc_khz;
913 /* maximum supported tsc_khz for guests */
914 extern u32  kvm_max_guest_tsc_khz;
915 
916 enum emulation_result {
917 	EMULATE_DONE,         /* no further processing */
918 	EMULATE_USER_EXIT,    /* kvm_run ready for userspace exit */
919 	EMULATE_FAIL,         /* can't emulate this instruction */
920 };
921 
922 #define EMULTYPE_NO_DECODE	    (1 << 0)
923 #define EMULTYPE_TRAP_UD	    (1 << 1)
924 #define EMULTYPE_SKIP		    (1 << 2)
925 #define EMULTYPE_RETRY		    (1 << 3)
926 #define EMULTYPE_NO_REEXECUTE	    (1 << 4)
927 int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
928 			    int emulation_type, void *insn, int insn_len);
929 
930 static inline int emulate_instruction(struct kvm_vcpu *vcpu,
931 			int emulation_type)
932 {
933 	return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
934 }
935 
936 void kvm_enable_efer_bits(u64);
937 bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
938 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
939 int kvm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
940 
941 struct x86_emulate_ctxt;
942 
943 int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
944 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
945 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
946 int kvm_vcpu_halt(struct kvm_vcpu *vcpu);
947 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
948 
949 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
950 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
951 void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
952 
953 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
954 		    int reason, bool has_error_code, u32 error_code);
955 
956 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
957 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
958 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
959 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
960 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
961 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
962 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
963 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
964 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
965 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
966 
967 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
968 int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
969 
970 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
971 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
972 bool kvm_rdpmc(struct kvm_vcpu *vcpu);
973 
974 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
975 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
976 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
977 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
978 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
979 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
980 			    gfn_t gfn, void *data, int offset, int len,
981 			    u32 access);
982 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
983 bool kvm_require_dr(struct kvm_vcpu *vcpu, int dr);
984 
985 static inline int __kvm_irq_line_state(unsigned long *irq_state,
986 				       int irq_source_id, int level)
987 {
988 	/* Logical OR for level trig interrupt */
989 	if (level)
990 		__set_bit(irq_source_id, irq_state);
991 	else
992 		__clear_bit(irq_source_id, irq_state);
993 
994 	return !!(*irq_state);
995 }
996 
997 int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
998 void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
999 
1000 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
1001 
1002 int fx_init(struct kvm_vcpu *vcpu);
1003 
1004 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
1005 		       const u8 *new, int bytes);
1006 int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
1007 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
1008 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
1009 int kvm_mmu_load(struct kvm_vcpu *vcpu);
1010 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
1011 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
1012 gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1013 			   struct x86_exception *exception);
1014 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
1015 			      struct x86_exception *exception);
1016 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
1017 			       struct x86_exception *exception);
1018 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
1019 			       struct x86_exception *exception);
1020 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
1021 				struct x86_exception *exception);
1022 
1023 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
1024 
1025 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
1026 		       void *insn, int insn_len);
1027 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
1028 void kvm_mmu_new_cr3(struct kvm_vcpu *vcpu);
1029 
1030 void kvm_enable_tdp(void);
1031 void kvm_disable_tdp(void);
1032 
1033 static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1034 				  struct x86_exception *exception)
1035 {
1036 	return gpa;
1037 }
1038 
1039 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
1040 {
1041 	struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
1042 
1043 	return (struct kvm_mmu_page *)page_private(page);
1044 }
1045 
1046 static inline u16 kvm_read_ldt(void)
1047 {
1048 	u16 ldt;
1049 	asm("sldt %0" : "=g"(ldt));
1050 	return ldt;
1051 }
1052 
1053 static inline void kvm_load_ldt(u16 sel)
1054 {
1055 	asm("lldt %0" : : "rm"(sel));
1056 }
1057 
1058 #ifdef CONFIG_X86_64
1059 static inline unsigned long read_msr(unsigned long msr)
1060 {
1061 	u64 value;
1062 
1063 	rdmsrl(msr, value);
1064 	return value;
1065 }
1066 #endif
1067 
1068 static inline u32 get_rdx_init_val(void)
1069 {
1070 	return 0x600; /* P6 family */
1071 }
1072 
1073 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
1074 {
1075 	kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
1076 }
1077 
1078 static inline u64 get_canonical(u64 la)
1079 {
1080 	return ((int64_t)la << 16) >> 16;
1081 }
1082 
1083 static inline bool is_noncanonical_address(u64 la)
1084 {
1085 #ifdef CONFIG_X86_64
1086 	return get_canonical(la) != la;
1087 #else
1088 	return false;
1089 #endif
1090 }
1091 
1092 #define TSS_IOPB_BASE_OFFSET 0x66
1093 #define TSS_BASE_SIZE 0x68
1094 #define TSS_IOPB_SIZE (65536 / 8)
1095 #define TSS_REDIRECTION_SIZE (256 / 8)
1096 #define RMODE_TSS_SIZE							\
1097 	(TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
1098 
1099 enum {
1100 	TASK_SWITCH_CALL = 0,
1101 	TASK_SWITCH_IRET = 1,
1102 	TASK_SWITCH_JMP = 2,
1103 	TASK_SWITCH_GATE = 3,
1104 };
1105 
1106 #define HF_GIF_MASK		(1 << 0)
1107 #define HF_HIF_MASK		(1 << 1)
1108 #define HF_VINTR_MASK		(1 << 2)
1109 #define HF_NMI_MASK		(1 << 3)
1110 #define HF_IRET_MASK		(1 << 4)
1111 #define HF_GUEST_MASK		(1 << 5) /* VCPU is in guest-mode */
1112 
1113 /*
1114  * Hardware virtualization extension instructions may fault if a
1115  * reboot turns off virtualization while processes are running.
1116  * Trap the fault and ignore the instruction if that happens.
1117  */
1118 asmlinkage void kvm_spurious_fault(void);
1119 
1120 #define ____kvm_handle_fault_on_reboot(insn, cleanup_insn)	\
1121 	"666: " insn "\n\t" \
1122 	"668: \n\t"                           \
1123 	".pushsection .fixup, \"ax\" \n" \
1124 	"667: \n\t" \
1125 	cleanup_insn "\n\t"		      \
1126 	"cmpb $0, kvm_rebooting \n\t"	      \
1127 	"jne 668b \n\t"      		      \
1128 	__ASM_SIZE(push) " $666b \n\t"	      \
1129 	"call kvm_spurious_fault \n\t"	      \
1130 	".popsection \n\t" \
1131 	_ASM_EXTABLE(666b, 667b)
1132 
1133 #define __kvm_handle_fault_on_reboot(insn)		\
1134 	____kvm_handle_fault_on_reboot(insn, "")
1135 
1136 #define KVM_ARCH_WANT_MMU_NOTIFIER
1137 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
1138 int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
1139 int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
1140 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
1141 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
1142 int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
1143 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1144 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
1145 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
1146 void kvm_vcpu_reset(struct kvm_vcpu *vcpu);
1147 void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu);
1148 void kvm_arch_mmu_notifier_invalidate_page(struct kvm *kvm,
1149 					   unsigned long address);
1150 
1151 void kvm_define_shared_msr(unsigned index, u32 msr);
1152 int kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
1153 
1154 unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
1155 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1156 
1157 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1158 				     struct kvm_async_pf *work);
1159 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1160 				 struct kvm_async_pf *work);
1161 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1162 			       struct kvm_async_pf *work);
1163 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
1164 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1165 
1166 void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1167 
1168 int kvm_is_in_guest(void);
1169 
1170 void kvm_pmu_init(struct kvm_vcpu *vcpu);
1171 void kvm_pmu_destroy(struct kvm_vcpu *vcpu);
1172 void kvm_pmu_reset(struct kvm_vcpu *vcpu);
1173 void kvm_pmu_cpuid_update(struct kvm_vcpu *vcpu);
1174 bool kvm_pmu_msr(struct kvm_vcpu *vcpu, u32 msr);
1175 int kvm_pmu_get_msr(struct kvm_vcpu *vcpu, u32 msr, u64 *data);
1176 int kvm_pmu_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info);
1177 int kvm_pmu_check_pmc(struct kvm_vcpu *vcpu, unsigned pmc);
1178 int kvm_pmu_read_pmc(struct kvm_vcpu *vcpu, unsigned pmc, u64 *data);
1179 void kvm_handle_pmu_event(struct kvm_vcpu *vcpu);
1180 void kvm_deliver_pmi(struct kvm_vcpu *vcpu);
1181 
1182 #endif /* _ASM_X86_KVM_HOST_H */
1183