xref: /linux/arch/x86/include/asm/kvm_host.h (revision 12871a0bd67dd4db4418e1daafcd46e9d329ef10)
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 
20 #include <linux/kvm.h>
21 #include <linux/kvm_para.h>
22 #include <linux/kvm_types.h>
23 
24 #include <asm/pvclock-abi.h>
25 #include <asm/desc.h>
26 #include <asm/mtrr.h>
27 #include <asm/msr-index.h>
28 
29 #define KVM_MAX_VCPUS 64
30 #define KVM_MEMORY_SLOTS 32
31 /* memory slots that does not exposed to userspace */
32 #define KVM_PRIVATE_MEM_SLOTS 4
33 #define KVM_MMIO_SIZE 16
34 
35 #define KVM_PIO_PAGE_OFFSET 1
36 #define KVM_COALESCED_MMIO_PAGE_OFFSET 2
37 
38 #define CR0_RESERVED_BITS                                               \
39 	(~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
40 			  | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
41 			  | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
42 
43 #define CR3_PAE_RESERVED_BITS ((X86_CR3_PWT | X86_CR3_PCD) - 1)
44 #define CR3_NONPAE_RESERVED_BITS ((PAGE_SIZE-1) & ~(X86_CR3_PWT | X86_CR3_PCD))
45 #define CR3_L_MODE_RESERVED_BITS (CR3_NONPAE_RESERVED_BITS |	\
46 				  0xFFFFFF0000000000ULL)
47 #define CR4_RESERVED_BITS                                               \
48 	(~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
49 			  | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE     \
50 			  | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR  \
51 			  | X86_CR4_OSXSAVE \
52 			  | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE))
53 
54 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
55 
56 
57 
58 #define INVALID_PAGE (~(hpa_t)0)
59 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
60 
61 #define UNMAPPED_GVA (~(gpa_t)0)
62 
63 /* KVM Hugepage definitions for x86 */
64 #define KVM_NR_PAGE_SIZES	3
65 #define KVM_HPAGE_GFN_SHIFT(x)	(((x) - 1) * 9)
66 #define KVM_HPAGE_SHIFT(x)	(PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
67 #define KVM_HPAGE_SIZE(x)	(1UL << KVM_HPAGE_SHIFT(x))
68 #define KVM_HPAGE_MASK(x)	(~(KVM_HPAGE_SIZE(x) - 1))
69 #define KVM_PAGES_PER_HPAGE(x)	(KVM_HPAGE_SIZE(x) / PAGE_SIZE)
70 
71 #define DE_VECTOR 0
72 #define DB_VECTOR 1
73 #define BP_VECTOR 3
74 #define OF_VECTOR 4
75 #define BR_VECTOR 5
76 #define UD_VECTOR 6
77 #define NM_VECTOR 7
78 #define DF_VECTOR 8
79 #define TS_VECTOR 10
80 #define NP_VECTOR 11
81 #define SS_VECTOR 12
82 #define GP_VECTOR 13
83 #define PF_VECTOR 14
84 #define MF_VECTOR 16
85 #define MC_VECTOR 18
86 
87 #define SELECTOR_TI_MASK (1 << 2)
88 #define SELECTOR_RPL_MASK 0x03
89 
90 #define IOPL_SHIFT 12
91 
92 #define KVM_PERMILLE_MMU_PAGES 20
93 #define KVM_MIN_ALLOC_MMU_PAGES 64
94 #define KVM_MMU_HASH_SHIFT 10
95 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
96 #define KVM_MIN_FREE_MMU_PAGES 5
97 #define KVM_REFILL_PAGES 25
98 #define KVM_MAX_CPUID_ENTRIES 80
99 #define KVM_NR_FIXED_MTRR_REGION 88
100 #define KVM_NR_VAR_MTRR 8
101 
102 #define ASYNC_PF_PER_VCPU 64
103 
104 extern raw_spinlock_t kvm_lock;
105 extern struct list_head vm_list;
106 
107 struct kvm_vcpu;
108 struct kvm;
109 struct kvm_async_pf;
110 
111 enum kvm_reg {
112 	VCPU_REGS_RAX = 0,
113 	VCPU_REGS_RCX = 1,
114 	VCPU_REGS_RDX = 2,
115 	VCPU_REGS_RBX = 3,
116 	VCPU_REGS_RSP = 4,
117 	VCPU_REGS_RBP = 5,
118 	VCPU_REGS_RSI = 6,
119 	VCPU_REGS_RDI = 7,
120 #ifdef CONFIG_X86_64
121 	VCPU_REGS_R8 = 8,
122 	VCPU_REGS_R9 = 9,
123 	VCPU_REGS_R10 = 10,
124 	VCPU_REGS_R11 = 11,
125 	VCPU_REGS_R12 = 12,
126 	VCPU_REGS_R13 = 13,
127 	VCPU_REGS_R14 = 14,
128 	VCPU_REGS_R15 = 15,
129 #endif
130 	VCPU_REGS_RIP,
131 	NR_VCPU_REGS
132 };
133 
134 enum kvm_reg_ex {
135 	VCPU_EXREG_PDPTR = NR_VCPU_REGS,
136 	VCPU_EXREG_CR3,
137 	VCPU_EXREG_RFLAGS,
138 	VCPU_EXREG_CPL,
139 	VCPU_EXREG_SEGMENTS,
140 };
141 
142 enum {
143 	VCPU_SREG_ES,
144 	VCPU_SREG_CS,
145 	VCPU_SREG_SS,
146 	VCPU_SREG_DS,
147 	VCPU_SREG_FS,
148 	VCPU_SREG_GS,
149 	VCPU_SREG_TR,
150 	VCPU_SREG_LDTR,
151 };
152 
153 #include <asm/kvm_emulate.h>
154 
155 #define KVM_NR_MEM_OBJS 40
156 
157 #define KVM_NR_DB_REGS	4
158 
159 #define DR6_BD		(1 << 13)
160 #define DR6_BS		(1 << 14)
161 #define DR6_FIXED_1	0xffff0ff0
162 #define DR6_VOLATILE	0x0000e00f
163 
164 #define DR7_BP_EN_MASK	0x000000ff
165 #define DR7_GE		(1 << 9)
166 #define DR7_GD		(1 << 13)
167 #define DR7_FIXED_1	0x00000400
168 #define DR7_VOLATILE	0xffff23ff
169 
170 /*
171  * We don't want allocation failures within the mmu code, so we preallocate
172  * enough memory for a single page fault in a cache.
173  */
174 struct kvm_mmu_memory_cache {
175 	int nobjs;
176 	void *objects[KVM_NR_MEM_OBJS];
177 };
178 
179 #define NR_PTE_CHAIN_ENTRIES 5
180 
181 struct kvm_pte_chain {
182 	u64 *parent_ptes[NR_PTE_CHAIN_ENTRIES];
183 	struct hlist_node link;
184 };
185 
186 /*
187  * kvm_mmu_page_role, below, is defined as:
188  *
189  *   bits 0:3 - total guest paging levels (2-4, or zero for real mode)
190  *   bits 4:7 - page table level for this shadow (1-4)
191  *   bits 8:9 - page table quadrant for 2-level guests
192  *   bit   16 - direct mapping of virtual to physical mapping at gfn
193  *              used for real mode and two-dimensional paging
194  *   bits 17:19 - common access permissions for all ptes in this shadow page
195  */
196 union kvm_mmu_page_role {
197 	unsigned word;
198 	struct {
199 		unsigned level:4;
200 		unsigned cr4_pae:1;
201 		unsigned quadrant:2;
202 		unsigned pad_for_nice_hex_output:6;
203 		unsigned direct:1;
204 		unsigned access:3;
205 		unsigned invalid:1;
206 		unsigned nxe:1;
207 		unsigned cr0_wp:1;
208 	};
209 };
210 
211 struct kvm_mmu_page {
212 	struct list_head link;
213 	struct hlist_node hash_link;
214 
215 	/*
216 	 * The following two entries are used to key the shadow page in the
217 	 * hash table.
218 	 */
219 	gfn_t gfn;
220 	union kvm_mmu_page_role role;
221 
222 	u64 *spt;
223 	/* hold the gfn of each spte inside spt */
224 	gfn_t *gfns;
225 	/*
226 	 * One bit set per slot which has memory
227 	 * in this shadow page.
228 	 */
229 	DECLARE_BITMAP(slot_bitmap, KVM_MEMORY_SLOTS + KVM_PRIVATE_MEM_SLOTS);
230 	bool multimapped;         /* More than one parent_pte? */
231 	bool unsync;
232 	int root_count;          /* Currently serving as active root */
233 	unsigned int unsync_children;
234 	union {
235 		u64 *parent_pte;               /* !multimapped */
236 		struct hlist_head parent_ptes; /* multimapped, kvm_pte_chain */
237 	};
238 	DECLARE_BITMAP(unsync_child_bitmap, 512);
239 };
240 
241 struct kvm_pv_mmu_op_buffer {
242 	void *ptr;
243 	unsigned len;
244 	unsigned processed;
245 	char buf[512] __aligned(sizeof(long));
246 };
247 
248 struct kvm_pio_request {
249 	unsigned long count;
250 	int in;
251 	int port;
252 	int size;
253 };
254 
255 /*
256  * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
257  * 32-bit).  The kvm_mmu structure abstracts the details of the current mmu
258  * mode.
259  */
260 struct kvm_mmu {
261 	void (*new_cr3)(struct kvm_vcpu *vcpu);
262 	void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
263 	unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
264 	int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
265 			  bool prefault);
266 	void (*inject_page_fault)(struct kvm_vcpu *vcpu,
267 				  struct x86_exception *fault);
268 	void (*free)(struct kvm_vcpu *vcpu);
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 	void (*prefetch_page)(struct kvm_vcpu *vcpu,
273 			      struct kvm_mmu_page *page);
274 	int (*sync_page)(struct kvm_vcpu *vcpu,
275 			 struct kvm_mmu_page *sp);
276 	void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
277 	void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
278 			   u64 *spte, const void *pte);
279 	hpa_t root_hpa;
280 	int root_level;
281 	int shadow_root_level;
282 	union kvm_mmu_page_role base_role;
283 	bool direct_map;
284 
285 	u64 *pae_root;
286 	u64 *lm_root;
287 	u64 rsvd_bits_mask[2][4];
288 
289 	bool nx;
290 
291 	u64 pdptrs[4]; /* pae */
292 };
293 
294 struct kvm_vcpu_arch {
295 	/*
296 	 * rip and regs accesses must go through
297 	 * kvm_{register,rip}_{read,write} functions.
298 	 */
299 	unsigned long regs[NR_VCPU_REGS];
300 	u32 regs_avail;
301 	u32 regs_dirty;
302 
303 	unsigned long cr0;
304 	unsigned long cr0_guest_owned_bits;
305 	unsigned long cr2;
306 	unsigned long cr3;
307 	unsigned long cr4;
308 	unsigned long cr4_guest_owned_bits;
309 	unsigned long cr8;
310 	u32 hflags;
311 	u64 efer;
312 	u64 apic_base;
313 	struct kvm_lapic *apic;    /* kernel irqchip context */
314 	int32_t apic_arb_prio;
315 	int mp_state;
316 	int sipi_vector;
317 	u64 ia32_misc_enable_msr;
318 	bool tpr_access_reporting;
319 
320 	/*
321 	 * Paging state of the vcpu
322 	 *
323 	 * If the vcpu runs in guest mode with two level paging this still saves
324 	 * the paging mode of the l1 guest. This context is always used to
325 	 * handle faults.
326 	 */
327 	struct kvm_mmu mmu;
328 
329 	/*
330 	 * Paging state of an L2 guest (used for nested npt)
331 	 *
332 	 * This context will save all necessary information to walk page tables
333 	 * of the an L2 guest. This context is only initialized for page table
334 	 * walking and not for faulting since we never handle l2 page faults on
335 	 * the host.
336 	 */
337 	struct kvm_mmu nested_mmu;
338 
339 	/*
340 	 * Pointer to the mmu context currently used for
341 	 * gva_to_gpa translations.
342 	 */
343 	struct kvm_mmu *walk_mmu;
344 
345 	/* only needed in kvm_pv_mmu_op() path, but it's hot so
346 	 * put it here to avoid allocation */
347 	struct kvm_pv_mmu_op_buffer mmu_op_buffer;
348 
349 	struct kvm_mmu_memory_cache mmu_pte_chain_cache;
350 	struct kvm_mmu_memory_cache mmu_rmap_desc_cache;
351 	struct kvm_mmu_memory_cache mmu_page_cache;
352 	struct kvm_mmu_memory_cache mmu_page_header_cache;
353 
354 	gfn_t last_pt_write_gfn;
355 	int   last_pt_write_count;
356 	u64  *last_pte_updated;
357 	gfn_t last_pte_gfn;
358 
359 	struct fpu guest_fpu;
360 	u64 xcr0;
361 
362 	struct kvm_pio_request pio;
363 	void *pio_data;
364 
365 	u8 event_exit_inst_len;
366 
367 	struct kvm_queued_exception {
368 		bool pending;
369 		bool has_error_code;
370 		bool reinject;
371 		u8 nr;
372 		u32 error_code;
373 	} exception;
374 
375 	struct kvm_queued_interrupt {
376 		bool pending;
377 		bool soft;
378 		u8 nr;
379 	} interrupt;
380 
381 	int halt_request; /* real mode on Intel only */
382 
383 	int cpuid_nent;
384 	struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
385 	/* emulate context */
386 
387 	struct x86_emulate_ctxt emulate_ctxt;
388 	bool emulate_regs_need_sync_to_vcpu;
389 	bool emulate_regs_need_sync_from_vcpu;
390 
391 	gpa_t time;
392 	struct pvclock_vcpu_time_info hv_clock;
393 	unsigned int hw_tsc_khz;
394 	unsigned int time_offset;
395 	struct page *time_page;
396 	u64 last_guest_tsc;
397 	u64 last_kernel_ns;
398 	u64 last_tsc_nsec;
399 	u64 last_tsc_write;
400 	u32 virtual_tsc_khz;
401 	bool tsc_catchup;
402 	u32  tsc_catchup_mult;
403 	s8   tsc_catchup_shift;
404 
405 	bool nmi_pending;
406 	bool nmi_injected;
407 
408 	struct mtrr_state_type mtrr_state;
409 	u32 pat;
410 
411 	int switch_db_regs;
412 	unsigned long db[KVM_NR_DB_REGS];
413 	unsigned long dr6;
414 	unsigned long dr7;
415 	unsigned long eff_db[KVM_NR_DB_REGS];
416 
417 	u64 mcg_cap;
418 	u64 mcg_status;
419 	u64 mcg_ctl;
420 	u64 *mce_banks;
421 
422 	/* used for guest single stepping over the given code position */
423 	unsigned long singlestep_rip;
424 
425 	/* fields used by HYPER-V emulation */
426 	u64 hv_vapic;
427 
428 	cpumask_var_t wbinvd_dirty_mask;
429 
430 	struct {
431 		bool halted;
432 		gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
433 		struct gfn_to_hva_cache data;
434 		u64 msr_val;
435 		u32 id;
436 		bool send_user_only;
437 	} apf;
438 };
439 
440 struct kvm_arch {
441 	unsigned int n_used_mmu_pages;
442 	unsigned int n_requested_mmu_pages;
443 	unsigned int n_max_mmu_pages;
444 	atomic_t invlpg_counter;
445 	struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
446 	/*
447 	 * Hash table of struct kvm_mmu_page.
448 	 */
449 	struct list_head active_mmu_pages;
450 	struct list_head assigned_dev_head;
451 	struct iommu_domain *iommu_domain;
452 	int iommu_flags;
453 	struct kvm_pic *vpic;
454 	struct kvm_ioapic *vioapic;
455 	struct kvm_pit *vpit;
456 	int vapics_in_nmi_mode;
457 
458 	unsigned int tss_addr;
459 	struct page *apic_access_page;
460 
461 	gpa_t wall_clock;
462 
463 	struct page *ept_identity_pagetable;
464 	bool ept_identity_pagetable_done;
465 	gpa_t ept_identity_map_addr;
466 
467 	unsigned long irq_sources_bitmap;
468 	s64 kvmclock_offset;
469 	raw_spinlock_t tsc_write_lock;
470 	u64 last_tsc_nsec;
471 	u64 last_tsc_offset;
472 	u64 last_tsc_write;
473 
474 	struct kvm_xen_hvm_config xen_hvm_config;
475 
476 	/* fields used by HYPER-V emulation */
477 	u64 hv_guest_os_id;
478 	u64 hv_hypercall;
479 
480 	#ifdef CONFIG_KVM_MMU_AUDIT
481 	int audit_point;
482 	#endif
483 };
484 
485 struct kvm_vm_stat {
486 	u32 mmu_shadow_zapped;
487 	u32 mmu_pte_write;
488 	u32 mmu_pte_updated;
489 	u32 mmu_pde_zapped;
490 	u32 mmu_flooded;
491 	u32 mmu_recycled;
492 	u32 mmu_cache_miss;
493 	u32 mmu_unsync;
494 	u32 remote_tlb_flush;
495 	u32 lpages;
496 };
497 
498 struct kvm_vcpu_stat {
499 	u32 pf_fixed;
500 	u32 pf_guest;
501 	u32 tlb_flush;
502 	u32 invlpg;
503 
504 	u32 exits;
505 	u32 io_exits;
506 	u32 mmio_exits;
507 	u32 signal_exits;
508 	u32 irq_window_exits;
509 	u32 nmi_window_exits;
510 	u32 halt_exits;
511 	u32 halt_wakeup;
512 	u32 request_irq_exits;
513 	u32 irq_exits;
514 	u32 host_state_reload;
515 	u32 efer_reload;
516 	u32 fpu_reload;
517 	u32 insn_emulation;
518 	u32 insn_emulation_fail;
519 	u32 hypercalls;
520 	u32 irq_injections;
521 	u32 nmi_injections;
522 };
523 
524 struct x86_instruction_info;
525 
526 struct kvm_x86_ops {
527 	int (*cpu_has_kvm_support)(void);          /* __init */
528 	int (*disabled_by_bios)(void);             /* __init */
529 	int (*hardware_enable)(void *dummy);
530 	void (*hardware_disable)(void *dummy);
531 	void (*check_processor_compatibility)(void *rtn);
532 	int (*hardware_setup)(void);               /* __init */
533 	void (*hardware_unsetup)(void);            /* __exit */
534 	bool (*cpu_has_accelerated_tpr)(void);
535 	void (*cpuid_update)(struct kvm_vcpu *vcpu);
536 
537 	/* Create, but do not attach this VCPU */
538 	struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
539 	void (*vcpu_free)(struct kvm_vcpu *vcpu);
540 	int (*vcpu_reset)(struct kvm_vcpu *vcpu);
541 
542 	void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
543 	void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
544 	void (*vcpu_put)(struct kvm_vcpu *vcpu);
545 
546 	void (*set_guest_debug)(struct kvm_vcpu *vcpu,
547 				struct kvm_guest_debug *dbg);
548 	int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
549 	int (*set_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
550 	u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
551 	void (*get_segment)(struct kvm_vcpu *vcpu,
552 			    struct kvm_segment *var, int seg);
553 	int (*get_cpl)(struct kvm_vcpu *vcpu);
554 	void (*set_segment)(struct kvm_vcpu *vcpu,
555 			    struct kvm_segment *var, int seg);
556 	void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
557 	void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
558 	void (*decache_cr3)(struct kvm_vcpu *vcpu);
559 	void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
560 	void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
561 	void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
562 	void (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
563 	void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
564 	void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
565 	void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
566 	void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
567 	void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
568 	void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
569 	void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
570 	unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
571 	void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
572 	void (*fpu_activate)(struct kvm_vcpu *vcpu);
573 	void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
574 
575 	void (*tlb_flush)(struct kvm_vcpu *vcpu);
576 
577 	void (*run)(struct kvm_vcpu *vcpu);
578 	int (*handle_exit)(struct kvm_vcpu *vcpu);
579 	void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
580 	void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
581 	u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
582 	void (*patch_hypercall)(struct kvm_vcpu *vcpu,
583 				unsigned char *hypercall_addr);
584 	void (*set_irq)(struct kvm_vcpu *vcpu);
585 	void (*set_nmi)(struct kvm_vcpu *vcpu);
586 	void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
587 				bool has_error_code, u32 error_code,
588 				bool reinject);
589 	void (*cancel_injection)(struct kvm_vcpu *vcpu);
590 	int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
591 	int (*nmi_allowed)(struct kvm_vcpu *vcpu);
592 	bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
593 	void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
594 	void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
595 	void (*enable_irq_window)(struct kvm_vcpu *vcpu);
596 	void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
597 	int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
598 	int (*get_tdp_level)(void);
599 	u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
600 	int (*get_lpage_level)(void);
601 	bool (*rdtscp_supported)(void);
602 	void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment);
603 
604 	void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
605 
606 	void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
607 
608 	bool (*has_wbinvd_exit)(void);
609 
610 	void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz);
611 	void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
612 
613 	u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc);
614 
615 	void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
616 
617 	int (*check_intercept)(struct kvm_vcpu *vcpu,
618 			       struct x86_instruction_info *info,
619 			       enum x86_intercept_stage stage);
620 
621 	const struct trace_print_flags *exit_reasons_str;
622 };
623 
624 struct kvm_arch_async_pf {
625 	u32 token;
626 	gfn_t gfn;
627 	unsigned long cr3;
628 	bool direct_map;
629 };
630 
631 extern struct kvm_x86_ops *kvm_x86_ops;
632 
633 int kvm_mmu_module_init(void);
634 void kvm_mmu_module_exit(void);
635 
636 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
637 int kvm_mmu_create(struct kvm_vcpu *vcpu);
638 int kvm_mmu_setup(struct kvm_vcpu *vcpu);
639 void kvm_mmu_set_nonpresent_ptes(u64 trap_pte, u64 notrap_pte);
640 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
641 		u64 dirty_mask, u64 nx_mask, u64 x_mask);
642 
643 int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
644 void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
645 void kvm_mmu_zap_all(struct kvm *kvm);
646 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
647 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
648 
649 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
650 
651 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
652 			  const void *val, int bytes);
653 int kvm_pv_mmu_op(struct kvm_vcpu *vcpu, unsigned long bytes,
654 		  gpa_t addr, unsigned long *ret);
655 u8 kvm_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn);
656 
657 extern bool tdp_enabled;
658 
659 /* control of guest tsc rate supported? */
660 extern bool kvm_has_tsc_control;
661 /* minimum supported tsc_khz for guests */
662 extern u32  kvm_min_guest_tsc_khz;
663 /* maximum supported tsc_khz for guests */
664 extern u32  kvm_max_guest_tsc_khz;
665 
666 enum emulation_result {
667 	EMULATE_DONE,       /* no further processing */
668 	EMULATE_DO_MMIO,      /* kvm_run filled with mmio request */
669 	EMULATE_FAIL,         /* can't emulate this instruction */
670 };
671 
672 #define EMULTYPE_NO_DECODE	    (1 << 0)
673 #define EMULTYPE_TRAP_UD	    (1 << 1)
674 #define EMULTYPE_SKIP		    (1 << 2)
675 int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
676 			    int emulation_type, void *insn, int insn_len);
677 
678 static inline int emulate_instruction(struct kvm_vcpu *vcpu,
679 			int emulation_type)
680 {
681 	return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
682 }
683 
684 void kvm_enable_efer_bits(u64);
685 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
686 int kvm_set_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
687 
688 struct x86_emulate_ctxt;
689 
690 int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
691 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
692 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
693 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
694 
695 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
696 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
697 
698 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int reason,
699 		    bool has_error_code, u32 error_code);
700 
701 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
702 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
703 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
704 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
705 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
706 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
707 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
708 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
709 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
710 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
711 
712 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
713 int kvm_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data);
714 
715 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
716 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
717 
718 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
719 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
720 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
721 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
722 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
723 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
724 			    gfn_t gfn, void *data, int offset, int len,
725 			    u32 access);
726 void kvm_propagate_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
727 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
728 
729 int kvm_pic_set_irq(void *opaque, int irq, int level);
730 
731 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
732 
733 int fx_init(struct kvm_vcpu *vcpu);
734 
735 void kvm_mmu_flush_tlb(struct kvm_vcpu *vcpu);
736 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
737 		       const u8 *new, int bytes,
738 		       bool guest_initiated);
739 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
740 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
741 int kvm_mmu_load(struct kvm_vcpu *vcpu);
742 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
743 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
744 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
745 			      struct x86_exception *exception);
746 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
747 			       struct x86_exception *exception);
748 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
749 			       struct x86_exception *exception);
750 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
751 				struct x86_exception *exception);
752 
753 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
754 
755 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
756 		       void *insn, int insn_len);
757 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
758 
759 void kvm_enable_tdp(void);
760 void kvm_disable_tdp(void);
761 
762 int complete_pio(struct kvm_vcpu *vcpu);
763 bool kvm_check_iopl(struct kvm_vcpu *vcpu);
764 
765 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
766 {
767 	struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
768 
769 	return (struct kvm_mmu_page *)page_private(page);
770 }
771 
772 static inline u16 kvm_read_ldt(void)
773 {
774 	u16 ldt;
775 	asm("sldt %0" : "=g"(ldt));
776 	return ldt;
777 }
778 
779 static inline void kvm_load_ldt(u16 sel)
780 {
781 	asm("lldt %0" : : "rm"(sel));
782 }
783 
784 #ifdef CONFIG_X86_64
785 static inline unsigned long read_msr(unsigned long msr)
786 {
787 	u64 value;
788 
789 	rdmsrl(msr, value);
790 	return value;
791 }
792 #endif
793 
794 static inline u32 get_rdx_init_val(void)
795 {
796 	return 0x600; /* P6 family */
797 }
798 
799 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
800 {
801 	kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
802 }
803 
804 #define TSS_IOPB_BASE_OFFSET 0x66
805 #define TSS_BASE_SIZE 0x68
806 #define TSS_IOPB_SIZE (65536 / 8)
807 #define TSS_REDIRECTION_SIZE (256 / 8)
808 #define RMODE_TSS_SIZE							\
809 	(TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
810 
811 enum {
812 	TASK_SWITCH_CALL = 0,
813 	TASK_SWITCH_IRET = 1,
814 	TASK_SWITCH_JMP = 2,
815 	TASK_SWITCH_GATE = 3,
816 };
817 
818 #define HF_GIF_MASK		(1 << 0)
819 #define HF_HIF_MASK		(1 << 1)
820 #define HF_VINTR_MASK		(1 << 2)
821 #define HF_NMI_MASK		(1 << 3)
822 #define HF_IRET_MASK		(1 << 4)
823 #define HF_GUEST_MASK		(1 << 5) /* VCPU is in guest-mode */
824 
825 /*
826  * Hardware virtualization extension instructions may fault if a
827  * reboot turns off virtualization while processes are running.
828  * Trap the fault and ignore the instruction if that happens.
829  */
830 asmlinkage void kvm_spurious_fault(void);
831 extern bool kvm_rebooting;
832 
833 #define __kvm_handle_fault_on_reboot(insn) \
834 	"666: " insn "\n\t" \
835 	"668: \n\t"                           \
836 	".pushsection .fixup, \"ax\" \n" \
837 	"667: \n\t" \
838 	"cmpb $0, kvm_rebooting \n\t"	      \
839 	"jne 668b \n\t"      		      \
840 	__ASM_SIZE(push) " $666b \n\t"	      \
841 	"call kvm_spurious_fault \n\t"	      \
842 	".popsection \n\t" \
843 	".pushsection __ex_table, \"a\" \n\t" \
844 	_ASM_PTR " 666b, 667b \n\t" \
845 	".popsection"
846 
847 #define KVM_ARCH_WANT_MMU_NOTIFIER
848 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
849 int kvm_age_hva(struct kvm *kvm, unsigned long hva);
850 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
851 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
852 int cpuid_maxphyaddr(struct kvm_vcpu *vcpu);
853 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
854 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
855 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
856 
857 void kvm_define_shared_msr(unsigned index, u32 msr);
858 void kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
859 
860 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
861 
862 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
863 				     struct kvm_async_pf *work);
864 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
865 				 struct kvm_async_pf *work);
866 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
867 			       struct kvm_async_pf *work);
868 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
869 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
870 
871 void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
872 
873 #endif /* _ASM_X86_KVM_HOST_H */
874