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 80
83 #define KVM_NR_FIXED_MTRR_REGION 88
84 #define KVM_NR_VAR_MTRR 8
85 
86 #define ASYNC_PF_PER_VCPU 64
87 
88 extern raw_spinlock_t kvm_lock;
89 extern struct list_head vm_list;
90 
91 struct kvm_vcpu;
92 struct kvm;
93 struct kvm_async_pf;
94 
95 enum kvm_reg {
96 	VCPU_REGS_RAX = 0,
97 	VCPU_REGS_RCX = 1,
98 	VCPU_REGS_RDX = 2,
99 	VCPU_REGS_RBX = 3,
100 	VCPU_REGS_RSP = 4,
101 	VCPU_REGS_RBP = 5,
102 	VCPU_REGS_RSI = 6,
103 	VCPU_REGS_RDI = 7,
104 #ifdef CONFIG_X86_64
105 	VCPU_REGS_R8 = 8,
106 	VCPU_REGS_R9 = 9,
107 	VCPU_REGS_R10 = 10,
108 	VCPU_REGS_R11 = 11,
109 	VCPU_REGS_R12 = 12,
110 	VCPU_REGS_R13 = 13,
111 	VCPU_REGS_R14 = 14,
112 	VCPU_REGS_R15 = 15,
113 #endif
114 	VCPU_REGS_RIP,
115 	NR_VCPU_REGS
116 };
117 
118 enum kvm_reg_ex {
119 	VCPU_EXREG_PDPTR = NR_VCPU_REGS,
120 	VCPU_EXREG_CR3,
121 };
122 
123 enum {
124 	VCPU_SREG_ES,
125 	VCPU_SREG_CS,
126 	VCPU_SREG_SS,
127 	VCPU_SREG_DS,
128 	VCPU_SREG_FS,
129 	VCPU_SREG_GS,
130 	VCPU_SREG_TR,
131 	VCPU_SREG_LDTR,
132 };
133 
134 #include <asm/kvm_emulate.h>
135 
136 #define KVM_NR_MEM_OBJS 40
137 
138 #define KVM_NR_DB_REGS	4
139 
140 #define DR6_BD		(1 << 13)
141 #define DR6_BS		(1 << 14)
142 #define DR6_FIXED_1	0xffff0ff0
143 #define DR6_VOLATILE	0x0000e00f
144 
145 #define DR7_BP_EN_MASK	0x000000ff
146 #define DR7_GE		(1 << 9)
147 #define DR7_GD		(1 << 13)
148 #define DR7_FIXED_1	0x00000400
149 #define DR7_VOLATILE	0xffff23ff
150 
151 /*
152  * We don't want allocation failures within the mmu code, so we preallocate
153  * enough memory for a single page fault in a cache.
154  */
155 struct kvm_mmu_memory_cache {
156 	int nobjs;
157 	void *objects[KVM_NR_MEM_OBJS];
158 };
159 
160 #define NR_PTE_CHAIN_ENTRIES 5
161 
162 struct kvm_pte_chain {
163 	u64 *parent_ptes[NR_PTE_CHAIN_ENTRIES];
164 	struct hlist_node link;
165 };
166 
167 /*
168  * kvm_mmu_page_role, below, is defined as:
169  *
170  *   bits 0:3 - total guest paging levels (2-4, or zero for real mode)
171  *   bits 4:7 - page table level for this shadow (1-4)
172  *   bits 8:9 - page table quadrant for 2-level guests
173  *   bit   16 - direct mapping of virtual to physical mapping at gfn
174  *              used for real mode and two-dimensional paging
175  *   bits 17:19 - common access permissions for all ptes in this shadow page
176  */
177 union kvm_mmu_page_role {
178 	unsigned word;
179 	struct {
180 		unsigned level:4;
181 		unsigned cr4_pae:1;
182 		unsigned quadrant:2;
183 		unsigned pad_for_nice_hex_output:6;
184 		unsigned direct:1;
185 		unsigned access:3;
186 		unsigned invalid:1;
187 		unsigned nxe:1;
188 		unsigned cr0_wp:1;
189 	};
190 };
191 
192 struct kvm_mmu_page {
193 	struct list_head link;
194 	struct hlist_node hash_link;
195 
196 	/*
197 	 * The following two entries are used to key the shadow page in the
198 	 * hash table.
199 	 */
200 	gfn_t gfn;
201 	union kvm_mmu_page_role role;
202 
203 	u64 *spt;
204 	/* hold the gfn of each spte inside spt */
205 	gfn_t *gfns;
206 	/*
207 	 * One bit set per slot which has memory
208 	 * in this shadow page.
209 	 */
210 	DECLARE_BITMAP(slot_bitmap, KVM_MEMORY_SLOTS + KVM_PRIVATE_MEM_SLOTS);
211 	bool multimapped;         /* More than one parent_pte? */
212 	bool unsync;
213 	int root_count;          /* Currently serving as active root */
214 	unsigned int unsync_children;
215 	union {
216 		u64 *parent_pte;               /* !multimapped */
217 		struct hlist_head parent_ptes; /* multimapped, kvm_pte_chain */
218 	};
219 	DECLARE_BITMAP(unsync_child_bitmap, 512);
220 };
221 
222 struct kvm_pv_mmu_op_buffer {
223 	void *ptr;
224 	unsigned len;
225 	unsigned processed;
226 	char buf[512] __aligned(sizeof(long));
227 };
228 
229 struct kvm_pio_request {
230 	unsigned long count;
231 	int in;
232 	int port;
233 	int size;
234 };
235 
236 /*
237  * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
238  * 32-bit).  The kvm_mmu structure abstracts the details of the current mmu
239  * mode.
240  */
241 struct kvm_mmu {
242 	void (*new_cr3)(struct kvm_vcpu *vcpu);
243 	void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
244 	unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
245 	int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
246 			  bool prefault);
247 	void (*inject_page_fault)(struct kvm_vcpu *vcpu,
248 				  struct x86_exception *fault);
249 	void (*free)(struct kvm_vcpu *vcpu);
250 	gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
251 			    struct x86_exception *exception);
252 	gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access);
253 	void (*prefetch_page)(struct kvm_vcpu *vcpu,
254 			      struct kvm_mmu_page *page);
255 	int (*sync_page)(struct kvm_vcpu *vcpu,
256 			 struct kvm_mmu_page *sp);
257 	void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
258 	void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
259 			u64 *spte, const void *pte, unsigned long mmu_seq);
260 	hpa_t root_hpa;
261 	int root_level;
262 	int shadow_root_level;
263 	union kvm_mmu_page_role base_role;
264 	bool direct_map;
265 
266 	u64 *pae_root;
267 	u64 *lm_root;
268 	u64 rsvd_bits_mask[2][4];
269 
270 	bool nx;
271 
272 	u64 pdptrs[4]; /* pae */
273 };
274 
275 struct kvm_vcpu_arch {
276 	/*
277 	 * rip and regs accesses must go through
278 	 * kvm_{register,rip}_{read,write} functions.
279 	 */
280 	unsigned long regs[NR_VCPU_REGS];
281 	u32 regs_avail;
282 	u32 regs_dirty;
283 
284 	unsigned long cr0;
285 	unsigned long cr0_guest_owned_bits;
286 	unsigned long cr2;
287 	unsigned long cr3;
288 	unsigned long cr4;
289 	unsigned long cr4_guest_owned_bits;
290 	unsigned long cr8;
291 	u32 hflags;
292 	u64 efer;
293 	u64 apic_base;
294 	struct kvm_lapic *apic;    /* kernel irqchip context */
295 	int32_t apic_arb_prio;
296 	int mp_state;
297 	int sipi_vector;
298 	u64 ia32_misc_enable_msr;
299 	bool tpr_access_reporting;
300 
301 	/*
302 	 * Paging state of the vcpu
303 	 *
304 	 * If the vcpu runs in guest mode with two level paging this still saves
305 	 * the paging mode of the l1 guest. This context is always used to
306 	 * handle faults.
307 	 */
308 	struct kvm_mmu mmu;
309 
310 	/*
311 	 * Paging state of an L2 guest (used for nested npt)
312 	 *
313 	 * This context will save all necessary information to walk page tables
314 	 * of the an L2 guest. This context is only initialized for page table
315 	 * walking and not for faulting since we never handle l2 page faults on
316 	 * the host.
317 	 */
318 	struct kvm_mmu nested_mmu;
319 
320 	/*
321 	 * Pointer to the mmu context currently used for
322 	 * gva_to_gpa translations.
323 	 */
324 	struct kvm_mmu *walk_mmu;
325 
326 	/* only needed in kvm_pv_mmu_op() path, but it's hot so
327 	 * put it here to avoid allocation */
328 	struct kvm_pv_mmu_op_buffer mmu_op_buffer;
329 
330 	struct kvm_mmu_memory_cache mmu_pte_chain_cache;
331 	struct kvm_mmu_memory_cache mmu_rmap_desc_cache;
332 	struct kvm_mmu_memory_cache mmu_page_cache;
333 	struct kvm_mmu_memory_cache mmu_page_header_cache;
334 
335 	gfn_t last_pt_write_gfn;
336 	int   last_pt_write_count;
337 	u64  *last_pte_updated;
338 	gfn_t last_pte_gfn;
339 
340 	struct fpu guest_fpu;
341 	u64 xcr0;
342 
343 	gva_t mmio_fault_cr2;
344 	struct kvm_pio_request pio;
345 	void *pio_data;
346 
347 	u8 event_exit_inst_len;
348 
349 	struct kvm_queued_exception {
350 		bool pending;
351 		bool has_error_code;
352 		bool reinject;
353 		u8 nr;
354 		u32 error_code;
355 	} exception;
356 
357 	struct kvm_queued_interrupt {
358 		bool pending;
359 		bool soft;
360 		u8 nr;
361 	} interrupt;
362 
363 	int halt_request; /* real mode on Intel only */
364 
365 	int cpuid_nent;
366 	struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
367 	/* emulate context */
368 
369 	struct x86_emulate_ctxt emulate_ctxt;
370 
371 	gpa_t time;
372 	struct pvclock_vcpu_time_info hv_clock;
373 	unsigned int hw_tsc_khz;
374 	unsigned int time_offset;
375 	struct page *time_page;
376 	u64 last_host_tsc;
377 	u64 last_guest_tsc;
378 	u64 last_kernel_ns;
379 	u64 last_tsc_nsec;
380 	u64 last_tsc_write;
381 	bool tsc_catchup;
382 
383 	bool nmi_pending;
384 	bool nmi_injected;
385 
386 	struct mtrr_state_type mtrr_state;
387 	u32 pat;
388 
389 	int switch_db_regs;
390 	unsigned long db[KVM_NR_DB_REGS];
391 	unsigned long dr6;
392 	unsigned long dr7;
393 	unsigned long eff_db[KVM_NR_DB_REGS];
394 
395 	u64 mcg_cap;
396 	u64 mcg_status;
397 	u64 mcg_ctl;
398 	u64 *mce_banks;
399 
400 	/* used for guest single stepping over the given code position */
401 	unsigned long singlestep_rip;
402 
403 	/* fields used by HYPER-V emulation */
404 	u64 hv_vapic;
405 
406 	cpumask_var_t wbinvd_dirty_mask;
407 
408 	struct {
409 		bool halted;
410 		gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
411 		struct gfn_to_hva_cache data;
412 		u64 msr_val;
413 		u32 id;
414 		bool send_user_only;
415 	} apf;
416 };
417 
418 struct kvm_arch {
419 	unsigned int n_used_mmu_pages;
420 	unsigned int n_requested_mmu_pages;
421 	unsigned int n_max_mmu_pages;
422 	atomic_t invlpg_counter;
423 	struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
424 	/*
425 	 * Hash table of struct kvm_mmu_page.
426 	 */
427 	struct list_head active_mmu_pages;
428 	struct list_head assigned_dev_head;
429 	struct iommu_domain *iommu_domain;
430 	int iommu_flags;
431 	struct kvm_pic *vpic;
432 	struct kvm_ioapic *vioapic;
433 	struct kvm_pit *vpit;
434 	int vapics_in_nmi_mode;
435 
436 	unsigned int tss_addr;
437 	struct page *apic_access_page;
438 
439 	gpa_t wall_clock;
440 
441 	struct page *ept_identity_pagetable;
442 	bool ept_identity_pagetable_done;
443 	gpa_t ept_identity_map_addr;
444 
445 	unsigned long irq_sources_bitmap;
446 	s64 kvmclock_offset;
447 	raw_spinlock_t tsc_write_lock;
448 	u64 last_tsc_nsec;
449 	u64 last_tsc_offset;
450 	u64 last_tsc_write;
451 	u32 virtual_tsc_khz;
452 	u32 virtual_tsc_mult;
453 	s8 virtual_tsc_shift;
454 
455 	struct kvm_xen_hvm_config xen_hvm_config;
456 
457 	/* fields used by HYPER-V emulation */
458 	u64 hv_guest_os_id;
459 	u64 hv_hypercall;
460 
461 	#ifdef CONFIG_KVM_MMU_AUDIT
462 	int audit_point;
463 	#endif
464 };
465 
466 struct kvm_vm_stat {
467 	u32 mmu_shadow_zapped;
468 	u32 mmu_pte_write;
469 	u32 mmu_pte_updated;
470 	u32 mmu_pde_zapped;
471 	u32 mmu_flooded;
472 	u32 mmu_recycled;
473 	u32 mmu_cache_miss;
474 	u32 mmu_unsync;
475 	u32 remote_tlb_flush;
476 	u32 lpages;
477 };
478 
479 struct kvm_vcpu_stat {
480 	u32 pf_fixed;
481 	u32 pf_guest;
482 	u32 tlb_flush;
483 	u32 invlpg;
484 
485 	u32 exits;
486 	u32 io_exits;
487 	u32 mmio_exits;
488 	u32 signal_exits;
489 	u32 irq_window_exits;
490 	u32 nmi_window_exits;
491 	u32 halt_exits;
492 	u32 halt_wakeup;
493 	u32 request_irq_exits;
494 	u32 irq_exits;
495 	u32 host_state_reload;
496 	u32 efer_reload;
497 	u32 fpu_reload;
498 	u32 insn_emulation;
499 	u32 insn_emulation_fail;
500 	u32 hypercalls;
501 	u32 irq_injections;
502 	u32 nmi_injections;
503 };
504 
505 struct kvm_x86_ops {
506 	int (*cpu_has_kvm_support)(void);          /* __init */
507 	int (*disabled_by_bios)(void);             /* __init */
508 	int (*hardware_enable)(void *dummy);
509 	void (*hardware_disable)(void *dummy);
510 	void (*check_processor_compatibility)(void *rtn);
511 	int (*hardware_setup)(void);               /* __init */
512 	void (*hardware_unsetup)(void);            /* __exit */
513 	bool (*cpu_has_accelerated_tpr)(void);
514 	void (*cpuid_update)(struct kvm_vcpu *vcpu);
515 
516 	/* Create, but do not attach this VCPU */
517 	struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
518 	void (*vcpu_free)(struct kvm_vcpu *vcpu);
519 	int (*vcpu_reset)(struct kvm_vcpu *vcpu);
520 
521 	void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
522 	void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
523 	void (*vcpu_put)(struct kvm_vcpu *vcpu);
524 
525 	void (*set_guest_debug)(struct kvm_vcpu *vcpu,
526 				struct kvm_guest_debug *dbg);
527 	int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
528 	int (*set_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
529 	u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
530 	void (*get_segment)(struct kvm_vcpu *vcpu,
531 			    struct kvm_segment *var, int seg);
532 	int (*get_cpl)(struct kvm_vcpu *vcpu);
533 	void (*set_segment)(struct kvm_vcpu *vcpu,
534 			    struct kvm_segment *var, int seg);
535 	void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
536 	void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
537 	void (*decache_cr3)(struct kvm_vcpu *vcpu);
538 	void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
539 	void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
540 	void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
541 	void (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
542 	void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
543 	void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
544 	void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
545 	void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
546 	void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
547 	void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
548 	void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
549 	unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
550 	void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
551 	void (*fpu_activate)(struct kvm_vcpu *vcpu);
552 	void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
553 
554 	void (*tlb_flush)(struct kvm_vcpu *vcpu);
555 
556 	void (*run)(struct kvm_vcpu *vcpu);
557 	int (*handle_exit)(struct kvm_vcpu *vcpu);
558 	void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
559 	void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
560 	u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
561 	void (*patch_hypercall)(struct kvm_vcpu *vcpu,
562 				unsigned char *hypercall_addr);
563 	void (*set_irq)(struct kvm_vcpu *vcpu);
564 	void (*set_nmi)(struct kvm_vcpu *vcpu);
565 	void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
566 				bool has_error_code, u32 error_code,
567 				bool reinject);
568 	void (*cancel_injection)(struct kvm_vcpu *vcpu);
569 	int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
570 	int (*nmi_allowed)(struct kvm_vcpu *vcpu);
571 	bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
572 	void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
573 	void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
574 	void (*enable_irq_window)(struct kvm_vcpu *vcpu);
575 	void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
576 	int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
577 	int (*get_tdp_level)(void);
578 	u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
579 	int (*get_lpage_level)(void);
580 	bool (*rdtscp_supported)(void);
581 	void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment);
582 
583 	void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
584 
585 	void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
586 
587 	bool (*has_wbinvd_exit)(void);
588 
589 	void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
590 
591 	void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
592 	const struct trace_print_flags *exit_reasons_str;
593 };
594 
595 struct kvm_arch_async_pf {
596 	u32 token;
597 	gfn_t gfn;
598 	unsigned long cr3;
599 	bool direct_map;
600 };
601 
602 extern struct kvm_x86_ops *kvm_x86_ops;
603 
604 int kvm_mmu_module_init(void);
605 void kvm_mmu_module_exit(void);
606 
607 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
608 int kvm_mmu_create(struct kvm_vcpu *vcpu);
609 int kvm_mmu_setup(struct kvm_vcpu *vcpu);
610 void kvm_mmu_set_nonpresent_ptes(u64 trap_pte, u64 notrap_pte);
611 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
612 		u64 dirty_mask, u64 nx_mask, u64 x_mask);
613 
614 int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
615 void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
616 void kvm_mmu_zap_all(struct kvm *kvm);
617 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
618 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
619 
620 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
621 
622 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
623 			  const void *val, int bytes);
624 int kvm_pv_mmu_op(struct kvm_vcpu *vcpu, unsigned long bytes,
625 		  gpa_t addr, unsigned long *ret);
626 u8 kvm_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn);
627 
628 extern bool tdp_enabled;
629 
630 enum emulation_result {
631 	EMULATE_DONE,       /* no further processing */
632 	EMULATE_DO_MMIO,      /* kvm_run filled with mmio request */
633 	EMULATE_FAIL,         /* can't emulate this instruction */
634 };
635 
636 #define EMULTYPE_NO_DECODE	    (1 << 0)
637 #define EMULTYPE_TRAP_UD	    (1 << 1)
638 #define EMULTYPE_SKIP		    (1 << 2)
639 int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
640 			    int emulation_type, void *insn, int insn_len);
641 
emulate_instruction(struct kvm_vcpu * vcpu,int emulation_type)642 static inline int emulate_instruction(struct kvm_vcpu *vcpu,
643 			int emulation_type)
644 {
645 	return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
646 }
647 
648 void realmode_lgdt(struct kvm_vcpu *vcpu, u16 size, unsigned long address);
649 void realmode_lidt(struct kvm_vcpu *vcpu, u16 size, unsigned long address);
650 
651 void kvm_enable_efer_bits(u64);
652 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
653 int kvm_set_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
654 
655 struct x86_emulate_ctxt;
656 
657 int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
658 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
659 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
660 int emulate_invlpg(struct kvm_vcpu *vcpu, gva_t address);
661 int emulate_clts(struct kvm_vcpu *vcpu);
662 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
663 
664 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
665 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
666 
667 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int reason,
668 		    bool has_error_code, u32 error_code);
669 
670 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
671 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
672 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
673 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
674 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
675 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
676 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
677 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
678 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
679 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
680 
681 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
682 int kvm_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data);
683 
684 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
685 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
686 
687 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
688 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
689 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
690 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
691 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
692 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
693 			    gfn_t gfn, void *data, int offset, int len,
694 			    u32 access);
695 void kvm_propagate_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
696 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
697 
698 int kvm_pic_set_irq(void *opaque, int irq, int level);
699 
700 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
701 
702 int fx_init(struct kvm_vcpu *vcpu);
703 
704 void kvm_mmu_flush_tlb(struct kvm_vcpu *vcpu);
705 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
706 		       const u8 *new, int bytes,
707 		       bool guest_initiated);
708 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
709 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
710 int kvm_mmu_load(struct kvm_vcpu *vcpu);
711 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
712 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
713 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
714 			      struct x86_exception *exception);
715 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
716 			       struct x86_exception *exception);
717 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
718 			       struct x86_exception *exception);
719 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
720 				struct x86_exception *exception);
721 
722 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
723 
724 int kvm_fix_hypercall(struct kvm_vcpu *vcpu);
725 
726 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
727 		       void *insn, int insn_len);
728 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
729 
730 void kvm_enable_tdp(void);
731 void kvm_disable_tdp(void);
732 
733 int complete_pio(struct kvm_vcpu *vcpu);
734 bool kvm_check_iopl(struct kvm_vcpu *vcpu);
735 
page_header(hpa_t shadow_page)736 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
737 {
738 	struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
739 
740 	return (struct kvm_mmu_page *)page_private(page);
741 }
742 
kvm_read_ldt(void)743 static inline u16 kvm_read_ldt(void)
744 {
745 	u16 ldt;
746 	asm("sldt %0" : "=g"(ldt));
747 	return ldt;
748 }
749 
kvm_load_ldt(u16 sel)750 static inline void kvm_load_ldt(u16 sel)
751 {
752 	asm("lldt %0" : : "rm"(sel));
753 }
754 
755 #ifdef CONFIG_X86_64
read_msr(unsigned long msr)756 static inline unsigned long read_msr(unsigned long msr)
757 {
758 	u64 value;
759 
760 	rdmsrl(msr, value);
761 	return value;
762 }
763 #endif
764 
get_rdx_init_val(void)765 static inline u32 get_rdx_init_val(void)
766 {
767 	return 0x600; /* P6 family */
768 }
769 
kvm_inject_gp(struct kvm_vcpu * vcpu,u32 error_code)770 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
771 {
772 	kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
773 }
774 
775 #define TSS_IOPB_BASE_OFFSET 0x66
776 #define TSS_BASE_SIZE 0x68
777 #define TSS_IOPB_SIZE (65536 / 8)
778 #define TSS_REDIRECTION_SIZE (256 / 8)
779 #define RMODE_TSS_SIZE							\
780 	(TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
781 
782 enum {
783 	TASK_SWITCH_CALL = 0,
784 	TASK_SWITCH_IRET = 1,
785 	TASK_SWITCH_JMP = 2,
786 	TASK_SWITCH_GATE = 3,
787 };
788 
789 #define HF_GIF_MASK		(1 << 0)
790 #define HF_HIF_MASK		(1 << 1)
791 #define HF_VINTR_MASK		(1 << 2)
792 #define HF_NMI_MASK		(1 << 3)
793 #define HF_IRET_MASK		(1 << 4)
794 #define HF_GUEST_MASK		(1 << 5) /* VCPU is in guest-mode */
795 
796 /*
797  * Hardware virtualization extension instructions may fault if a
798  * reboot turns off virtualization while processes are running.
799  * Trap the fault and ignore the instruction if that happens.
800  */
801 asmlinkage void kvm_spurious_fault(void);
802 extern bool kvm_rebooting;
803 
804 #define __kvm_handle_fault_on_reboot(insn) \
805 	"666: " insn "\n\t" \
806 	"668: \n\t"                           \
807 	".pushsection .fixup, \"ax\" \n" \
808 	"667: \n\t" \
809 	"cmpb $0, kvm_rebooting \n\t"	      \
810 	"jne 668b \n\t"      		      \
811 	__ASM_SIZE(push) " $666b \n\t"	      \
812 	"call kvm_spurious_fault \n\t"	      \
813 	".popsection \n\t" \
814 	".pushsection __ex_table, \"a\" \n\t" \
815 	_ASM_PTR " 666b, 667b \n\t" \
816 	".popsection"
817 
818 #define KVM_ARCH_WANT_MMU_NOTIFIER
819 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
820 int kvm_age_hva(struct kvm *kvm, unsigned long hva);
821 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
822 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
823 int cpuid_maxphyaddr(struct kvm_vcpu *vcpu);
824 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
825 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
826 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
827 
828 void kvm_define_shared_msr(unsigned index, u32 msr);
829 void kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
830 
831 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
832 
833 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
834 				     struct kvm_async_pf *work);
835 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
836 				 struct kvm_async_pf *work);
837 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
838 			       struct kvm_async_pf *work);
839 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
840 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
841 
842 void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
843 
844 #endif /* _ASM_X86_KVM_HOST_H */
845