xref: /linux/arch/x86/include/asm/kvm_host.h (revision 4aad8f51d0672f1c95e2cf0e1bc7b9ab42d8e1ea)
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 
34 #define KVM_PIO_PAGE_OFFSET 1
35 #define KVM_COALESCED_MMIO_PAGE_OFFSET 2
36 
37 #define CR3_PAE_RESERVED_BITS ((X86_CR3_PWT | X86_CR3_PCD) - 1)
38 #define CR3_NONPAE_RESERVED_BITS ((PAGE_SIZE-1) & ~(X86_CR3_PWT | X86_CR3_PCD))
39 #define CR3_L_MODE_RESERVED_BITS (CR3_NONPAE_RESERVED_BITS |	\
40 				  0xFFFFFF0000000000ULL)
41 
42 #define INVALID_PAGE (~(hpa_t)0)
43 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
44 
45 #define UNMAPPED_GVA (~(gpa_t)0)
46 
47 /* KVM Hugepage definitions for x86 */
48 #define KVM_NR_PAGE_SIZES	3
49 #define KVM_HPAGE_GFN_SHIFT(x)	(((x) - 1) * 9)
50 #define KVM_HPAGE_SHIFT(x)	(PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
51 #define KVM_HPAGE_SIZE(x)	(1UL << KVM_HPAGE_SHIFT(x))
52 #define KVM_HPAGE_MASK(x)	(~(KVM_HPAGE_SIZE(x) - 1))
53 #define KVM_PAGES_PER_HPAGE(x)	(KVM_HPAGE_SIZE(x) / PAGE_SIZE)
54 
55 #define DE_VECTOR 0
56 #define DB_VECTOR 1
57 #define BP_VECTOR 3
58 #define OF_VECTOR 4
59 #define BR_VECTOR 5
60 #define UD_VECTOR 6
61 #define NM_VECTOR 7
62 #define DF_VECTOR 8
63 #define TS_VECTOR 10
64 #define NP_VECTOR 11
65 #define SS_VECTOR 12
66 #define GP_VECTOR 13
67 #define PF_VECTOR 14
68 #define MF_VECTOR 16
69 #define MC_VECTOR 18
70 
71 #define SELECTOR_TI_MASK (1 << 2)
72 #define SELECTOR_RPL_MASK 0x03
73 
74 #define IOPL_SHIFT 12
75 
76 #define KVM_PERMILLE_MMU_PAGES 20
77 #define KVM_MIN_ALLOC_MMU_PAGES 64
78 #define KVM_MMU_HASH_SHIFT 10
79 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
80 #define KVM_MIN_FREE_MMU_PAGES 5
81 #define KVM_REFILL_PAGES 25
82 #define KVM_MAX_CPUID_ENTRIES 40
83 #define KVM_NR_FIXED_MTRR_REGION 88
84 #define KVM_NR_VAR_MTRR 8
85 
86 extern spinlock_t kvm_lock;
87 extern struct list_head vm_list;
88 
89 struct kvm_vcpu;
90 struct kvm;
91 
92 enum kvm_reg {
93 	VCPU_REGS_RAX = 0,
94 	VCPU_REGS_RCX = 1,
95 	VCPU_REGS_RDX = 2,
96 	VCPU_REGS_RBX = 3,
97 	VCPU_REGS_RSP = 4,
98 	VCPU_REGS_RBP = 5,
99 	VCPU_REGS_RSI = 6,
100 	VCPU_REGS_RDI = 7,
101 #ifdef CONFIG_X86_64
102 	VCPU_REGS_R8 = 8,
103 	VCPU_REGS_R9 = 9,
104 	VCPU_REGS_R10 = 10,
105 	VCPU_REGS_R11 = 11,
106 	VCPU_REGS_R12 = 12,
107 	VCPU_REGS_R13 = 13,
108 	VCPU_REGS_R14 = 14,
109 	VCPU_REGS_R15 = 15,
110 #endif
111 	VCPU_REGS_RIP,
112 	NR_VCPU_REGS
113 };
114 
115 enum kvm_reg_ex {
116 	VCPU_EXREG_PDPTR = NR_VCPU_REGS,
117 };
118 
119 enum {
120 	VCPU_SREG_ES,
121 	VCPU_SREG_CS,
122 	VCPU_SREG_SS,
123 	VCPU_SREG_DS,
124 	VCPU_SREG_FS,
125 	VCPU_SREG_GS,
126 	VCPU_SREG_TR,
127 	VCPU_SREG_LDTR,
128 };
129 
130 #include <asm/kvm_emulate.h>
131 
132 #define KVM_NR_MEM_OBJS 40
133 
134 #define KVM_NR_DB_REGS	4
135 
136 #define DR6_BD		(1 << 13)
137 #define DR6_BS		(1 << 14)
138 #define DR6_FIXED_1	0xffff0ff0
139 #define DR6_VOLATILE	0x0000e00f
140 
141 #define DR7_BP_EN_MASK	0x000000ff
142 #define DR7_GE		(1 << 9)
143 #define DR7_GD		(1 << 13)
144 #define DR7_FIXED_1	0x00000400
145 #define DR7_VOLATILE	0xffff23ff
146 
147 /*
148  * We don't want allocation failures within the mmu code, so we preallocate
149  * enough memory for a single page fault in a cache.
150  */
151 struct kvm_mmu_memory_cache {
152 	int nobjs;
153 	void *objects[KVM_NR_MEM_OBJS];
154 };
155 
156 #define NR_PTE_CHAIN_ENTRIES 5
157 
158 struct kvm_pte_chain {
159 	u64 *parent_ptes[NR_PTE_CHAIN_ENTRIES];
160 	struct hlist_node link;
161 };
162 
163 /*
164  * kvm_mmu_page_role, below, is defined as:
165  *
166  *   bits 0:3 - total guest paging levels (2-4, or zero for real mode)
167  *   bits 4:7 - page table level for this shadow (1-4)
168  *   bits 8:9 - page table quadrant for 2-level guests
169  *   bit   16 - direct mapping of virtual to physical mapping at gfn
170  *              used for real mode and two-dimensional paging
171  *   bits 17:19 - common access permissions for all ptes in this shadow page
172  */
173 union kvm_mmu_page_role {
174 	unsigned word;
175 	struct {
176 		unsigned level:4;
177 		unsigned cr4_pae:1;
178 		unsigned quadrant:2;
179 		unsigned pad_for_nice_hex_output:6;
180 		unsigned direct:1;
181 		unsigned access:3;
182 		unsigned invalid:1;
183 		unsigned nxe:1;
184 		unsigned cr0_wp:1;
185 	};
186 };
187 
188 struct kvm_mmu_page {
189 	struct list_head link;
190 	struct hlist_node hash_link;
191 
192 	/*
193 	 * The following two entries are used to key the shadow page in the
194 	 * hash table.
195 	 */
196 	gfn_t gfn;
197 	union kvm_mmu_page_role role;
198 
199 	u64 *spt;
200 	/* hold the gfn of each spte inside spt */
201 	gfn_t *gfns;
202 	/*
203 	 * One bit set per slot which has memory
204 	 * in this shadow page.
205 	 */
206 	DECLARE_BITMAP(slot_bitmap, KVM_MEMORY_SLOTS + KVM_PRIVATE_MEM_SLOTS);
207 	bool multimapped;         /* More than one parent_pte? */
208 	bool unsync;
209 	int root_count;          /* Currently serving as active root */
210 	unsigned int unsync_children;
211 	union {
212 		u64 *parent_pte;               /* !multimapped */
213 		struct hlist_head parent_ptes; /* multimapped, kvm_pte_chain */
214 	};
215 	DECLARE_BITMAP(unsync_child_bitmap, 512);
216 };
217 
218 struct kvm_pv_mmu_op_buffer {
219 	void *ptr;
220 	unsigned len;
221 	unsigned processed;
222 	char buf[512] __aligned(sizeof(long));
223 };
224 
225 struct kvm_pio_request {
226 	unsigned long count;
227 	int in;
228 	int port;
229 	int size;
230 };
231 
232 /*
233  * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
234  * 32-bit).  The kvm_mmu structure abstracts the details of the current mmu
235  * mode.
236  */
237 struct kvm_mmu {
238 	void (*new_cr3)(struct kvm_vcpu *vcpu);
239 	void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
240 	unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
241 	int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err);
242 	void (*inject_page_fault)(struct kvm_vcpu *vcpu);
243 	void (*free)(struct kvm_vcpu *vcpu);
244 	gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
245 			    u32 *error);
246 	gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access);
247 	void (*prefetch_page)(struct kvm_vcpu *vcpu,
248 			      struct kvm_mmu_page *page);
249 	int (*sync_page)(struct kvm_vcpu *vcpu,
250 			 struct kvm_mmu_page *sp, bool clear_unsync);
251 	void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
252 	hpa_t root_hpa;
253 	int root_level;
254 	int shadow_root_level;
255 	union kvm_mmu_page_role base_role;
256 	bool direct_map;
257 
258 	u64 *pae_root;
259 	u64 *lm_root;
260 	u64 rsvd_bits_mask[2][4];
261 
262 	bool nx;
263 
264 	u64 pdptrs[4]; /* pae */
265 };
266 
267 struct kvm_vcpu_arch {
268 	/*
269 	 * rip and regs accesses must go through
270 	 * kvm_{register,rip}_{read,write} functions.
271 	 */
272 	unsigned long regs[NR_VCPU_REGS];
273 	u32 regs_avail;
274 	u32 regs_dirty;
275 
276 	unsigned long cr0;
277 	unsigned long cr0_guest_owned_bits;
278 	unsigned long cr2;
279 	unsigned long cr3;
280 	unsigned long cr4;
281 	unsigned long cr4_guest_owned_bits;
282 	unsigned long cr8;
283 	u32 hflags;
284 	u64 efer;
285 	u64 apic_base;
286 	struct kvm_lapic *apic;    /* kernel irqchip context */
287 	int32_t apic_arb_prio;
288 	int mp_state;
289 	int sipi_vector;
290 	u64 ia32_misc_enable_msr;
291 	bool tpr_access_reporting;
292 
293 	/*
294 	 * Paging state of the vcpu
295 	 *
296 	 * If the vcpu runs in guest mode with two level paging this still saves
297 	 * the paging mode of the l1 guest. This context is always used to
298 	 * handle faults.
299 	 */
300 	struct kvm_mmu mmu;
301 
302 	/*
303 	 * Paging state of an L2 guest (used for nested npt)
304 	 *
305 	 * This context will save all necessary information to walk page tables
306 	 * of the an L2 guest. This context is only initialized for page table
307 	 * walking and not for faulting since we never handle l2 page faults on
308 	 * the host.
309 	 */
310 	struct kvm_mmu nested_mmu;
311 
312 	/*
313 	 * Pointer to the mmu context currently used for
314 	 * gva_to_gpa translations.
315 	 */
316 	struct kvm_mmu *walk_mmu;
317 
318 	/*
319 	 * This struct is filled with the necessary information to propagate a
320 	 * page fault into the guest
321 	 */
322 	struct {
323 		u64      address;
324 		unsigned error_code;
325 		bool     nested;
326 	} fault;
327 
328 	/* only needed in kvm_pv_mmu_op() path, but it's hot so
329 	 * put it here to avoid allocation */
330 	struct kvm_pv_mmu_op_buffer mmu_op_buffer;
331 
332 	struct kvm_mmu_memory_cache mmu_pte_chain_cache;
333 	struct kvm_mmu_memory_cache mmu_rmap_desc_cache;
334 	struct kvm_mmu_memory_cache mmu_page_cache;
335 	struct kvm_mmu_memory_cache mmu_page_header_cache;
336 
337 	gfn_t last_pt_write_gfn;
338 	int   last_pt_write_count;
339 	u64  *last_pte_updated;
340 	gfn_t last_pte_gfn;
341 
342 	struct {
343 		gfn_t gfn;	/* presumed gfn during guest pte update */
344 		pfn_t pfn;	/* pfn corresponding to that gfn */
345 		unsigned long mmu_seq;
346 	} update_pte;
347 
348 	struct fpu guest_fpu;
349 	u64 xcr0;
350 
351 	gva_t mmio_fault_cr2;
352 	struct kvm_pio_request pio;
353 	void *pio_data;
354 
355 	u8 event_exit_inst_len;
356 
357 	struct kvm_queued_exception {
358 		bool pending;
359 		bool has_error_code;
360 		bool reinject;
361 		u8 nr;
362 		u32 error_code;
363 	} exception;
364 
365 	struct kvm_queued_interrupt {
366 		bool pending;
367 		bool soft;
368 		u8 nr;
369 	} interrupt;
370 
371 	int halt_request; /* real mode on Intel only */
372 
373 	int cpuid_nent;
374 	struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
375 	/* emulate context */
376 
377 	struct x86_emulate_ctxt emulate_ctxt;
378 
379 	gpa_t time;
380 	struct pvclock_vcpu_time_info hv_clock;
381 	unsigned int hw_tsc_khz;
382 	unsigned int time_offset;
383 	struct page *time_page;
384 	u64 last_host_tsc;
385 	u64 last_guest_tsc;
386 	u64 last_kernel_ns;
387 	u64 last_tsc_nsec;
388 	u64 last_tsc_write;
389 	bool tsc_catchup;
390 
391 	bool nmi_pending;
392 	bool nmi_injected;
393 
394 	struct mtrr_state_type mtrr_state;
395 	u32 pat;
396 
397 	int switch_db_regs;
398 	unsigned long db[KVM_NR_DB_REGS];
399 	unsigned long dr6;
400 	unsigned long dr7;
401 	unsigned long eff_db[KVM_NR_DB_REGS];
402 
403 	u64 mcg_cap;
404 	u64 mcg_status;
405 	u64 mcg_ctl;
406 	u64 *mce_banks;
407 
408 	/* used for guest single stepping over the given code position */
409 	unsigned long singlestep_rip;
410 
411 	/* fields used by HYPER-V emulation */
412 	u64 hv_vapic;
413 
414 	cpumask_var_t wbinvd_dirty_mask;
415 };
416 
417 struct kvm_arch {
418 	unsigned int n_used_mmu_pages;
419 	unsigned int n_requested_mmu_pages;
420 	unsigned int n_max_mmu_pages;
421 	atomic_t invlpg_counter;
422 	struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
423 	/*
424 	 * Hash table of struct kvm_mmu_page.
425 	 */
426 	struct list_head active_mmu_pages;
427 	struct list_head assigned_dev_head;
428 	struct iommu_domain *iommu_domain;
429 	int iommu_flags;
430 	struct kvm_pic *vpic;
431 	struct kvm_ioapic *vioapic;
432 	struct kvm_pit *vpit;
433 	int vapics_in_nmi_mode;
434 
435 	unsigned int tss_addr;
436 	struct page *apic_access_page;
437 
438 	gpa_t wall_clock;
439 
440 	struct page *ept_identity_pagetable;
441 	bool ept_identity_pagetable_done;
442 	gpa_t ept_identity_map_addr;
443 
444 	unsigned long irq_sources_bitmap;
445 	s64 kvmclock_offset;
446 	spinlock_t tsc_write_lock;
447 	u64 last_tsc_nsec;
448 	u64 last_tsc_offset;
449 	u64 last_tsc_write;
450 	u32 virtual_tsc_khz;
451 	u32 virtual_tsc_mult;
452 	s8 virtual_tsc_shift;
453 
454 	struct kvm_xen_hvm_config xen_hvm_config;
455 
456 	/* fields used by HYPER-V emulation */
457 	u64 hv_guest_os_id;
458 	u64 hv_hypercall;
459 };
460 
461 struct kvm_vm_stat {
462 	u32 mmu_shadow_zapped;
463 	u32 mmu_pte_write;
464 	u32 mmu_pte_updated;
465 	u32 mmu_pde_zapped;
466 	u32 mmu_flooded;
467 	u32 mmu_recycled;
468 	u32 mmu_cache_miss;
469 	u32 mmu_unsync;
470 	u32 remote_tlb_flush;
471 	u32 lpages;
472 };
473 
474 struct kvm_vcpu_stat {
475 	u32 pf_fixed;
476 	u32 pf_guest;
477 	u32 tlb_flush;
478 	u32 invlpg;
479 
480 	u32 exits;
481 	u32 io_exits;
482 	u32 mmio_exits;
483 	u32 signal_exits;
484 	u32 irq_window_exits;
485 	u32 nmi_window_exits;
486 	u32 halt_exits;
487 	u32 halt_wakeup;
488 	u32 request_irq_exits;
489 	u32 irq_exits;
490 	u32 host_state_reload;
491 	u32 efer_reload;
492 	u32 fpu_reload;
493 	u32 insn_emulation;
494 	u32 insn_emulation_fail;
495 	u32 hypercalls;
496 	u32 irq_injections;
497 	u32 nmi_injections;
498 };
499 
500 struct kvm_x86_ops {
501 	int (*cpu_has_kvm_support)(void);          /* __init */
502 	int (*disabled_by_bios)(void);             /* __init */
503 	int (*hardware_enable)(void *dummy);
504 	void (*hardware_disable)(void *dummy);
505 	void (*check_processor_compatibility)(void *rtn);
506 	int (*hardware_setup)(void);               /* __init */
507 	void (*hardware_unsetup)(void);            /* __exit */
508 	bool (*cpu_has_accelerated_tpr)(void);
509 	void (*cpuid_update)(struct kvm_vcpu *vcpu);
510 
511 	/* Create, but do not attach this VCPU */
512 	struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
513 	void (*vcpu_free)(struct kvm_vcpu *vcpu);
514 	int (*vcpu_reset)(struct kvm_vcpu *vcpu);
515 
516 	void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
517 	void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
518 	void (*vcpu_put)(struct kvm_vcpu *vcpu);
519 
520 	void (*set_guest_debug)(struct kvm_vcpu *vcpu,
521 				struct kvm_guest_debug *dbg);
522 	int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
523 	int (*set_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
524 	u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
525 	void (*get_segment)(struct kvm_vcpu *vcpu,
526 			    struct kvm_segment *var, int seg);
527 	int (*get_cpl)(struct kvm_vcpu *vcpu);
528 	void (*set_segment)(struct kvm_vcpu *vcpu,
529 			    struct kvm_segment *var, int seg);
530 	void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
531 	void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
532 	void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
533 	void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
534 	void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
535 	void (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
536 	void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
537 	void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
538 	void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
539 	void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
540 	void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
541 	void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
542 	void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
543 	unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
544 	void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
545 	void (*fpu_activate)(struct kvm_vcpu *vcpu);
546 	void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
547 
548 	void (*tlb_flush)(struct kvm_vcpu *vcpu);
549 
550 	void (*run)(struct kvm_vcpu *vcpu);
551 	int (*handle_exit)(struct kvm_vcpu *vcpu);
552 	void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
553 	void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
554 	u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
555 	void (*patch_hypercall)(struct kvm_vcpu *vcpu,
556 				unsigned char *hypercall_addr);
557 	void (*set_irq)(struct kvm_vcpu *vcpu);
558 	void (*set_nmi)(struct kvm_vcpu *vcpu);
559 	void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
560 				bool has_error_code, u32 error_code,
561 				bool reinject);
562 	void (*cancel_injection)(struct kvm_vcpu *vcpu);
563 	int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
564 	int (*nmi_allowed)(struct kvm_vcpu *vcpu);
565 	bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
566 	void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
567 	void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
568 	void (*enable_irq_window)(struct kvm_vcpu *vcpu);
569 	void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
570 	int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
571 	int (*get_tdp_level)(void);
572 	u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
573 	int (*get_lpage_level)(void);
574 	bool (*rdtscp_supported)(void);
575 	void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment);
576 
577 	void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
578 
579 	void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
580 
581 	bool (*has_wbinvd_exit)(void);
582 
583 	void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
584 
585 	const struct trace_print_flags *exit_reasons_str;
586 };
587 
588 extern struct kvm_x86_ops *kvm_x86_ops;
589 
590 int kvm_mmu_module_init(void);
591 void kvm_mmu_module_exit(void);
592 
593 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
594 int kvm_mmu_create(struct kvm_vcpu *vcpu);
595 int kvm_mmu_setup(struct kvm_vcpu *vcpu);
596 void kvm_mmu_set_nonpresent_ptes(u64 trap_pte, u64 notrap_pte);
597 void kvm_mmu_set_base_ptes(u64 base_pte);
598 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
599 		u64 dirty_mask, u64 nx_mask, u64 x_mask);
600 
601 int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
602 void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
603 void kvm_mmu_zap_all(struct kvm *kvm);
604 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
605 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
606 
607 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
608 
609 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
610 			  const void *val, int bytes);
611 int kvm_pv_mmu_op(struct kvm_vcpu *vcpu, unsigned long bytes,
612 		  gpa_t addr, unsigned long *ret);
613 u8 kvm_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn);
614 
615 extern bool tdp_enabled;
616 
617 enum emulation_result {
618 	EMULATE_DONE,       /* no further processing */
619 	EMULATE_DO_MMIO,      /* kvm_run filled with mmio request */
620 	EMULATE_FAIL,         /* can't emulate this instruction */
621 };
622 
623 #define EMULTYPE_NO_DECODE	    (1 << 0)
624 #define EMULTYPE_TRAP_UD	    (1 << 1)
625 #define EMULTYPE_SKIP		    (1 << 2)
626 int emulate_instruction(struct kvm_vcpu *vcpu,
627 			unsigned long cr2, u16 error_code, int emulation_type);
628 void realmode_lgdt(struct kvm_vcpu *vcpu, u16 size, unsigned long address);
629 void realmode_lidt(struct kvm_vcpu *vcpu, u16 size, unsigned long address);
630 
631 void kvm_enable_efer_bits(u64);
632 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
633 int kvm_set_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
634 
635 struct x86_emulate_ctxt;
636 
637 int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
638 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
639 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
640 int emulate_invlpg(struct kvm_vcpu *vcpu, gva_t address);
641 int emulate_clts(struct kvm_vcpu *vcpu);
642 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
643 
644 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
645 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
646 
647 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int reason,
648 		    bool has_error_code, u32 error_code);
649 
650 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
651 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
652 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
653 void kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
654 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
655 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
656 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
657 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
658 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
659 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
660 
661 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
662 int kvm_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data);
663 
664 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
665 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
666 
667 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
668 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
669 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
670 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
671 void kvm_inject_page_fault(struct kvm_vcpu *vcpu);
672 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
673 			    gfn_t gfn, void *data, int offset, int len,
674 			    u32 access);
675 void kvm_propagate_fault(struct kvm_vcpu *vcpu);
676 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
677 
678 int kvm_pic_set_irq(void *opaque, int irq, int level);
679 
680 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
681 
682 int fx_init(struct kvm_vcpu *vcpu);
683 
684 void kvm_mmu_flush_tlb(struct kvm_vcpu *vcpu);
685 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
686 		       const u8 *new, int bytes,
687 		       bool guest_initiated);
688 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
689 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
690 int kvm_mmu_load(struct kvm_vcpu *vcpu);
691 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
692 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
693 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva, u32 *error);
694 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva, u32 *error);
695 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva, u32 *error);
696 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva, u32 *error);
697 
698 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
699 
700 int kvm_fix_hypercall(struct kvm_vcpu *vcpu);
701 
702 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code);
703 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
704 
705 void kvm_enable_tdp(void);
706 void kvm_disable_tdp(void);
707 
708 int complete_pio(struct kvm_vcpu *vcpu);
709 bool kvm_check_iopl(struct kvm_vcpu *vcpu);
710 
711 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
712 {
713 	struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
714 
715 	return (struct kvm_mmu_page *)page_private(page);
716 }
717 
718 static inline u16 kvm_read_ldt(void)
719 {
720 	u16 ldt;
721 	asm("sldt %0" : "=g"(ldt));
722 	return ldt;
723 }
724 
725 static inline void kvm_load_ldt(u16 sel)
726 {
727 	asm("lldt %0" : : "rm"(sel));
728 }
729 
730 #ifdef CONFIG_X86_64
731 static inline unsigned long read_msr(unsigned long msr)
732 {
733 	u64 value;
734 
735 	rdmsrl(msr, value);
736 	return value;
737 }
738 #endif
739 
740 static inline u32 get_rdx_init_val(void)
741 {
742 	return 0x600; /* P6 family */
743 }
744 
745 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
746 {
747 	kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
748 }
749 
750 #define TSS_IOPB_BASE_OFFSET 0x66
751 #define TSS_BASE_SIZE 0x68
752 #define TSS_IOPB_SIZE (65536 / 8)
753 #define TSS_REDIRECTION_SIZE (256 / 8)
754 #define RMODE_TSS_SIZE							\
755 	(TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
756 
757 enum {
758 	TASK_SWITCH_CALL = 0,
759 	TASK_SWITCH_IRET = 1,
760 	TASK_SWITCH_JMP = 2,
761 	TASK_SWITCH_GATE = 3,
762 };
763 
764 #define HF_GIF_MASK		(1 << 0)
765 #define HF_HIF_MASK		(1 << 1)
766 #define HF_VINTR_MASK		(1 << 2)
767 #define HF_NMI_MASK		(1 << 3)
768 #define HF_IRET_MASK		(1 << 4)
769 
770 /*
771  * Hardware virtualization extension instructions may fault if a
772  * reboot turns off virtualization while processes are running.
773  * Trap the fault and ignore the instruction if that happens.
774  */
775 asmlinkage void kvm_handle_fault_on_reboot(void);
776 
777 #define __kvm_handle_fault_on_reboot(insn) \
778 	"666: " insn "\n\t" \
779 	".pushsection .fixup, \"ax\" \n" \
780 	"667: \n\t" \
781 	__ASM_SIZE(push) " $666b \n\t"	      \
782 	"jmp kvm_handle_fault_on_reboot \n\t" \
783 	".popsection \n\t" \
784 	".pushsection __ex_table, \"a\" \n\t" \
785 	_ASM_PTR " 666b, 667b \n\t" \
786 	".popsection"
787 
788 #define KVM_ARCH_WANT_MMU_NOTIFIER
789 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
790 int kvm_age_hva(struct kvm *kvm, unsigned long hva);
791 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
792 int cpuid_maxphyaddr(struct kvm_vcpu *vcpu);
793 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
794 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
795 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
796 
797 void kvm_define_shared_msr(unsigned index, u32 msr);
798 void kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
799 
800 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
801 
802 #endif /* _ASM_X86_KVM_HOST_H */
803