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
26 #include <asm/pvclock-abi.h>
27 #include <asm/desc.h>
28 #include <asm/mtrr.h>
29 #include <asm/msr-index.h>
30
31 #define KVM_MAX_VCPUS 254
32 #define KVM_SOFT_MAX_VCPUS 160
33 #define KVM_MEMORY_SLOTS 32
34 /* memory slots that does not exposed to userspace */
35 #define KVM_PRIVATE_MEM_SLOTS 4
36 #define KVM_MEM_SLOTS_NUM (KVM_MEMORY_SLOTS + KVM_PRIVATE_MEM_SLOTS)
37
38 #define KVM_MMIO_SIZE 16
39
40 #define KVM_PIO_PAGE_OFFSET 1
41 #define KVM_COALESCED_MMIO_PAGE_OFFSET 2
42
43 #define CR0_RESERVED_BITS \
44 (~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
45 | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
46 | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
47
48 #define CR3_PAE_RESERVED_BITS ((X86_CR3_PWT | X86_CR3_PCD) - 1)
49 #define CR3_NONPAE_RESERVED_BITS ((PAGE_SIZE-1) & ~(X86_CR3_PWT | X86_CR3_PCD))
50 #define CR3_L_MODE_RESERVED_BITS (CR3_NONPAE_RESERVED_BITS | \
51 0xFFFFFF0000000000ULL)
52 #define CR4_RESERVED_BITS \
53 (~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
54 | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE \
55 | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR \
56 | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_RDWRGSFS \
57 | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE))
58
59 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
60
61
62
63 #define INVALID_PAGE (~(hpa_t)0)
64 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
65
66 #define UNMAPPED_GVA (~(gpa_t)0)
67
68 /* KVM Hugepage definitions for x86 */
69 #define KVM_NR_PAGE_SIZES 3
70 #define KVM_HPAGE_GFN_SHIFT(x) (((x) - 1) * 9)
71 #define KVM_HPAGE_SHIFT(x) (PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
72 #define KVM_HPAGE_SIZE(x) (1UL << KVM_HPAGE_SHIFT(x))
73 #define KVM_HPAGE_MASK(x) (~(KVM_HPAGE_SIZE(x) - 1))
74 #define KVM_PAGES_PER_HPAGE(x) (KVM_HPAGE_SIZE(x) / PAGE_SIZE)
75
76 #define DE_VECTOR 0
77 #define DB_VECTOR 1
78 #define BP_VECTOR 3
79 #define OF_VECTOR 4
80 #define BR_VECTOR 5
81 #define UD_VECTOR 6
82 #define NM_VECTOR 7
83 #define DF_VECTOR 8
84 #define TS_VECTOR 10
85 #define NP_VECTOR 11
86 #define SS_VECTOR 12
87 #define GP_VECTOR 13
88 #define PF_VECTOR 14
89 #define MF_VECTOR 16
90 #define MC_VECTOR 18
91
92 #define SELECTOR_TI_MASK (1 << 2)
93 #define SELECTOR_RPL_MASK 0x03
94
95 #define IOPL_SHIFT 12
96
97 #define KVM_PERMILLE_MMU_PAGES 20
98 #define KVM_MIN_ALLOC_MMU_PAGES 64
99 #define KVM_MMU_HASH_SHIFT 10
100 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
101 #define KVM_MIN_FREE_MMU_PAGES 5
102 #define KVM_REFILL_PAGES 25
103 #define KVM_MAX_CPUID_ENTRIES 80
104 #define KVM_NR_FIXED_MTRR_REGION 88
105 #define KVM_NR_VAR_MTRR 8
106
107 #define ASYNC_PF_PER_VCPU 64
108
109 extern raw_spinlock_t kvm_lock;
110 extern struct list_head vm_list;
111
112 struct kvm_vcpu;
113 struct kvm;
114 struct kvm_async_pf;
115
116 enum kvm_reg {
117 VCPU_REGS_RAX = 0,
118 VCPU_REGS_RCX = 1,
119 VCPU_REGS_RDX = 2,
120 VCPU_REGS_RBX = 3,
121 VCPU_REGS_RSP = 4,
122 VCPU_REGS_RBP = 5,
123 VCPU_REGS_RSI = 6,
124 VCPU_REGS_RDI = 7,
125 #ifdef CONFIG_X86_64
126 VCPU_REGS_R8 = 8,
127 VCPU_REGS_R9 = 9,
128 VCPU_REGS_R10 = 10,
129 VCPU_REGS_R11 = 11,
130 VCPU_REGS_R12 = 12,
131 VCPU_REGS_R13 = 13,
132 VCPU_REGS_R14 = 14,
133 VCPU_REGS_R15 = 15,
134 #endif
135 VCPU_REGS_RIP,
136 NR_VCPU_REGS
137 };
138
139 enum kvm_reg_ex {
140 VCPU_EXREG_PDPTR = NR_VCPU_REGS,
141 VCPU_EXREG_CR3,
142 VCPU_EXREG_RFLAGS,
143 VCPU_EXREG_CPL,
144 VCPU_EXREG_SEGMENTS,
145 };
146
147 enum {
148 VCPU_SREG_ES,
149 VCPU_SREG_CS,
150 VCPU_SREG_SS,
151 VCPU_SREG_DS,
152 VCPU_SREG_FS,
153 VCPU_SREG_GS,
154 VCPU_SREG_TR,
155 VCPU_SREG_LDTR,
156 };
157
158 #include <asm/kvm_emulate.h>
159
160 #define KVM_NR_MEM_OBJS 40
161
162 #define KVM_NR_DB_REGS 4
163
164 #define DR6_BD (1 << 13)
165 #define DR6_BS (1 << 14)
166 #define DR6_FIXED_1 0xffff0ff0
167 #define DR6_VOLATILE 0x0000e00f
168
169 #define DR7_BP_EN_MASK 0x000000ff
170 #define DR7_GE (1 << 9)
171 #define DR7_GD (1 << 13)
172 #define DR7_FIXED_1 0x00000400
173 #define DR7_VOLATILE 0xffff23ff
174
175 /*
176 * We don't want allocation failures within the mmu code, so we preallocate
177 * enough memory for a single page fault in a cache.
178 */
179 struct kvm_mmu_memory_cache {
180 int nobjs;
181 void *objects[KVM_NR_MEM_OBJS];
182 };
183
184 /*
185 * kvm_mmu_page_role, below, is defined as:
186 *
187 * bits 0:3 - total guest paging levels (2-4, or zero for real mode)
188 * bits 4:7 - page table level for this shadow (1-4)
189 * bits 8:9 - page table quadrant for 2-level guests
190 * bit 16 - direct mapping of virtual to physical mapping at gfn
191 * used for real mode and two-dimensional paging
192 * bits 17:19 - common access permissions for all ptes in this shadow page
193 */
194 union kvm_mmu_page_role {
195 unsigned word;
196 struct {
197 unsigned level:4;
198 unsigned cr4_pae:1;
199 unsigned quadrant:2;
200 unsigned pad_for_nice_hex_output:6;
201 unsigned direct:1;
202 unsigned access:3;
203 unsigned invalid:1;
204 unsigned nxe:1;
205 unsigned cr0_wp:1;
206 unsigned smep_andnot_wp:1;
207 };
208 };
209
210 struct kvm_mmu_page {
211 struct list_head link;
212 struct hlist_node hash_link;
213
214 /*
215 * The following two entries are used to key the shadow page in the
216 * hash table.
217 */
218 gfn_t gfn;
219 union kvm_mmu_page_role role;
220
221 u64 *spt;
222 /* hold the gfn of each spte inside spt */
223 gfn_t *gfns;
224 /*
225 * One bit set per slot which has memory
226 * in this shadow page.
227 */
228 DECLARE_BITMAP(slot_bitmap, KVM_MEM_SLOTS_NUM);
229 bool unsync;
230 int root_count; /* Currently serving as active root */
231 unsigned int unsync_children;
232 unsigned long parent_ptes; /* Reverse mapping for parent_pte */
233 DECLARE_BITMAP(unsync_child_bitmap, 512);
234
235 #ifdef CONFIG_X86_32
236 int clear_spte_count;
237 #endif
238
239 int write_flooding_count;
240
241 struct rcu_head rcu;
242 };
243
244 struct kvm_pio_request {
245 unsigned long count;
246 int in;
247 int port;
248 int size;
249 };
250
251 /*
252 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
253 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
254 * mode.
255 */
256 struct kvm_mmu {
257 void (*new_cr3)(struct kvm_vcpu *vcpu);
258 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
259 unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
260 u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
261 int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
262 bool prefault);
263 void (*inject_page_fault)(struct kvm_vcpu *vcpu,
264 struct x86_exception *fault);
265 void (*free)(struct kvm_vcpu *vcpu);
266 gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
267 struct x86_exception *exception);
268 gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access);
269 int (*sync_page)(struct kvm_vcpu *vcpu,
270 struct kvm_mmu_page *sp);
271 void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
272 void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
273 u64 *spte, const void *pte);
274 hpa_t root_hpa;
275 int root_level;
276 int shadow_root_level;
277 union kvm_mmu_page_role base_role;
278 bool direct_map;
279
280 u64 *pae_root;
281 u64 *lm_root;
282 u64 rsvd_bits_mask[2][4];
283
284 bool nx;
285
286 u64 pdptrs[4]; /* pae */
287 };
288
289 enum pmc_type {
290 KVM_PMC_GP = 0,
291 KVM_PMC_FIXED,
292 };
293
294 struct kvm_pmc {
295 enum pmc_type type;
296 u8 idx;
297 u64 counter;
298 u64 eventsel;
299 struct perf_event *perf_event;
300 struct kvm_vcpu *vcpu;
301 };
302
303 struct kvm_pmu {
304 unsigned nr_arch_gp_counters;
305 unsigned nr_arch_fixed_counters;
306 unsigned available_event_types;
307 u64 fixed_ctr_ctrl;
308 u64 global_ctrl;
309 u64 global_status;
310 u64 global_ovf_ctrl;
311 u64 counter_bitmask[2];
312 u64 global_ctrl_mask;
313 u8 version;
314 struct kvm_pmc gp_counters[X86_PMC_MAX_GENERIC];
315 struct kvm_pmc fixed_counters[X86_PMC_MAX_FIXED];
316 struct irq_work irq_work;
317 u64 reprogram_pmi;
318 };
319
320 struct kvm_vcpu_arch {
321 /*
322 * rip and regs accesses must go through
323 * kvm_{register,rip}_{read,write} functions.
324 */
325 unsigned long regs[NR_VCPU_REGS];
326 u32 regs_avail;
327 u32 regs_dirty;
328
329 unsigned long cr0;
330 unsigned long cr0_guest_owned_bits;
331 unsigned long cr2;
332 unsigned long cr3;
333 unsigned long cr4;
334 unsigned long cr4_guest_owned_bits;
335 unsigned long cr8;
336 u32 hflags;
337 u64 efer;
338 u64 apic_base;
339 struct kvm_lapic *apic; /* kernel irqchip context */
340 int32_t apic_arb_prio;
341 int mp_state;
342 int sipi_vector;
343 u64 ia32_misc_enable_msr;
344 bool tpr_access_reporting;
345
346 /*
347 * Paging state of the vcpu
348 *
349 * If the vcpu runs in guest mode with two level paging this still saves
350 * the paging mode of the l1 guest. This context is always used to
351 * handle faults.
352 */
353 struct kvm_mmu mmu;
354
355 /*
356 * Paging state of an L2 guest (used for nested npt)
357 *
358 * This context will save all necessary information to walk page tables
359 * of the an L2 guest. This context is only initialized for page table
360 * walking and not for faulting since we never handle l2 page faults on
361 * the host.
362 */
363 struct kvm_mmu nested_mmu;
364
365 /*
366 * Pointer to the mmu context currently used for
367 * gva_to_gpa translations.
368 */
369 struct kvm_mmu *walk_mmu;
370
371 struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
372 struct kvm_mmu_memory_cache mmu_page_cache;
373 struct kvm_mmu_memory_cache mmu_page_header_cache;
374
375 struct fpu guest_fpu;
376 u64 xcr0;
377
378 struct kvm_pio_request pio;
379 void *pio_data;
380
381 u8 event_exit_inst_len;
382
383 struct kvm_queued_exception {
384 bool pending;
385 bool has_error_code;
386 bool reinject;
387 u8 nr;
388 u32 error_code;
389 } exception;
390
391 struct kvm_queued_interrupt {
392 bool pending;
393 bool soft;
394 u8 nr;
395 } interrupt;
396
397 int halt_request; /* real mode on Intel only */
398
399 int cpuid_nent;
400 struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
401 /* emulate context */
402
403 struct x86_emulate_ctxt emulate_ctxt;
404 bool emulate_regs_need_sync_to_vcpu;
405 bool emulate_regs_need_sync_from_vcpu;
406
407 gpa_t time;
408 struct pvclock_vcpu_time_info hv_clock;
409 unsigned int hw_tsc_khz;
410 struct gfn_to_hva_cache pv_time;
411 bool pv_time_enabled;
412
413 struct {
414 u64 msr_val;
415 u64 last_steal;
416 u64 accum_steal;
417 struct gfn_to_hva_cache stime;
418 struct kvm_steal_time steal;
419 } st;
420
421 u64 last_guest_tsc;
422 u64 last_kernel_ns;
423 u64 last_host_tsc;
424 u64 tsc_offset_adjustment;
425 u64 this_tsc_nsec;
426 u64 this_tsc_write;
427 u8 this_tsc_generation;
428 bool tsc_catchup;
429 bool tsc_always_catchup;
430 s8 virtual_tsc_shift;
431 u32 virtual_tsc_mult;
432 u32 virtual_tsc_khz;
433
434 atomic_t nmi_queued; /* unprocessed asynchronous NMIs */
435 unsigned nmi_pending; /* NMI queued after currently running handler */
436 bool nmi_injected; /* Trying to inject an NMI this entry */
437
438 struct mtrr_state_type mtrr_state;
439 u32 pat;
440
441 int switch_db_regs;
442 unsigned long db[KVM_NR_DB_REGS];
443 unsigned long dr6;
444 unsigned long dr7;
445 unsigned long eff_db[KVM_NR_DB_REGS];
446
447 u64 mcg_cap;
448 u64 mcg_status;
449 u64 mcg_ctl;
450 u64 *mce_banks;
451
452 /* Cache MMIO info */
453 u64 mmio_gva;
454 unsigned access;
455 gfn_t mmio_gfn;
456
457 struct kvm_pmu pmu;
458
459 /* used for guest single stepping over the given code position */
460 unsigned long singlestep_rip;
461
462 /* fields used by HYPER-V emulation */
463 u64 hv_vapic;
464
465 cpumask_var_t wbinvd_dirty_mask;
466
467 unsigned long last_retry_eip;
468 unsigned long last_retry_addr;
469
470 struct {
471 bool halted;
472 gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
473 struct gfn_to_hva_cache data;
474 u64 msr_val;
475 u32 id;
476 bool send_user_only;
477 } apf;
478
479 /* OSVW MSRs (AMD only) */
480 struct {
481 u64 length;
482 u64 status;
483 } osvw;
484 };
485
486 struct kvm_lpage_info {
487 unsigned long rmap_pde;
488 int write_count;
489 };
490
491 struct kvm_arch_memory_slot {
492 struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
493 };
494
495 struct kvm_arch {
496 unsigned int n_used_mmu_pages;
497 unsigned int n_requested_mmu_pages;
498 unsigned int n_max_mmu_pages;
499 unsigned int indirect_shadow_pages;
500 struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
501 /*
502 * Hash table of struct kvm_mmu_page.
503 */
504 struct list_head active_mmu_pages;
505 struct list_head assigned_dev_head;
506 struct iommu_domain *iommu_domain;
507 int iommu_flags;
508 struct kvm_pic *vpic;
509 struct kvm_ioapic *vioapic;
510 struct kvm_pit *vpit;
511 int vapics_in_nmi_mode;
512
513 unsigned int tss_addr;
514 struct page *apic_access_page;
515
516 gpa_t wall_clock;
517
518 struct page *ept_identity_pagetable;
519 bool ept_identity_pagetable_done;
520 gpa_t ept_identity_map_addr;
521
522 unsigned long irq_sources_bitmap;
523 s64 kvmclock_offset;
524 raw_spinlock_t tsc_write_lock;
525 u64 last_tsc_nsec;
526 u64 last_tsc_write;
527 u32 last_tsc_khz;
528 u64 cur_tsc_nsec;
529 u64 cur_tsc_write;
530 u64 cur_tsc_offset;
531 u8 cur_tsc_generation;
532
533 struct kvm_xen_hvm_config xen_hvm_config;
534
535 /* fields used by HYPER-V emulation */
536 u64 hv_guest_os_id;
537 u64 hv_hypercall;
538
539 atomic_t reader_counter;
540
541 #ifdef CONFIG_KVM_MMU_AUDIT
542 int audit_point;
543 #endif
544 };
545
546 struct kvm_vm_stat {
547 u32 mmu_shadow_zapped;
548 u32 mmu_pte_write;
549 u32 mmu_pte_updated;
550 u32 mmu_pde_zapped;
551 u32 mmu_flooded;
552 u32 mmu_recycled;
553 u32 mmu_cache_miss;
554 u32 mmu_unsync;
555 u32 remote_tlb_flush;
556 u32 lpages;
557 };
558
559 struct kvm_vcpu_stat {
560 u32 pf_fixed;
561 u32 pf_guest;
562 u32 tlb_flush;
563 u32 invlpg;
564
565 u32 exits;
566 u32 io_exits;
567 u32 mmio_exits;
568 u32 signal_exits;
569 u32 irq_window_exits;
570 u32 nmi_window_exits;
571 u32 halt_exits;
572 u32 halt_wakeup;
573 u32 request_irq_exits;
574 u32 irq_exits;
575 u32 host_state_reload;
576 u32 efer_reload;
577 u32 fpu_reload;
578 u32 insn_emulation;
579 u32 insn_emulation_fail;
580 u32 hypercalls;
581 u32 irq_injections;
582 u32 nmi_injections;
583 };
584
585 struct x86_instruction_info;
586
587 struct kvm_x86_ops {
588 int (*cpu_has_kvm_support)(void); /* __init */
589 int (*disabled_by_bios)(void); /* __init */
590 int (*hardware_enable)(void *dummy);
591 void (*hardware_disable)(void *dummy);
592 void (*check_processor_compatibility)(void *rtn);
593 int (*hardware_setup)(void); /* __init */
594 void (*hardware_unsetup)(void); /* __exit */
595 bool (*cpu_has_accelerated_tpr)(void);
596 void (*cpuid_update)(struct kvm_vcpu *vcpu);
597
598 /* Create, but do not attach this VCPU */
599 struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
600 void (*vcpu_free)(struct kvm_vcpu *vcpu);
601 int (*vcpu_reset)(struct kvm_vcpu *vcpu);
602
603 void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
604 void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
605 void (*vcpu_put)(struct kvm_vcpu *vcpu);
606
607 void (*set_guest_debug)(struct kvm_vcpu *vcpu,
608 struct kvm_guest_debug *dbg);
609 int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
610 int (*set_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
611 u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
612 void (*get_segment)(struct kvm_vcpu *vcpu,
613 struct kvm_segment *var, int seg);
614 int (*get_cpl)(struct kvm_vcpu *vcpu);
615 void (*set_segment)(struct kvm_vcpu *vcpu,
616 struct kvm_segment *var, int seg);
617 void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
618 void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
619 void (*decache_cr3)(struct kvm_vcpu *vcpu);
620 void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
621 void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
622 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
623 int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
624 void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
625 void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
626 void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
627 void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
628 void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
629 void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
630 void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
631 unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
632 void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
633 void (*fpu_activate)(struct kvm_vcpu *vcpu);
634 void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
635
636 void (*tlb_flush)(struct kvm_vcpu *vcpu);
637
638 void (*run)(struct kvm_vcpu *vcpu);
639 int (*handle_exit)(struct kvm_vcpu *vcpu);
640 void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
641 void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
642 u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
643 void (*patch_hypercall)(struct kvm_vcpu *vcpu,
644 unsigned char *hypercall_addr);
645 void (*set_irq)(struct kvm_vcpu *vcpu);
646 void (*set_nmi)(struct kvm_vcpu *vcpu);
647 void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
648 bool has_error_code, u32 error_code,
649 bool reinject);
650 void (*cancel_injection)(struct kvm_vcpu *vcpu);
651 int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
652 int (*nmi_allowed)(struct kvm_vcpu *vcpu);
653 bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
654 void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
655 void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
656 void (*enable_irq_window)(struct kvm_vcpu *vcpu);
657 void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
658 int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
659 int (*get_tdp_level)(void);
660 u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
661 int (*get_lpage_level)(void);
662 bool (*rdtscp_supported)(void);
663 void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment, bool host);
664
665 void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
666
667 void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
668
669 bool (*has_wbinvd_exit)(void);
670
671 void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale);
672 void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
673
674 u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc);
675 u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu);
676
677 void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
678
679 int (*check_intercept)(struct kvm_vcpu *vcpu,
680 struct x86_instruction_info *info,
681 enum x86_intercept_stage stage);
682 };
683
684 struct kvm_arch_async_pf {
685 u32 token;
686 gfn_t gfn;
687 unsigned long cr3;
688 bool direct_map;
689 };
690
691 extern struct kvm_x86_ops *kvm_x86_ops;
692
adjust_tsc_offset_guest(struct kvm_vcpu * vcpu,s64 adjustment)693 static inline void adjust_tsc_offset_guest(struct kvm_vcpu *vcpu,
694 s64 adjustment)
695 {
696 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, false);
697 }
698
adjust_tsc_offset_host(struct kvm_vcpu * vcpu,s64 adjustment)699 static inline void adjust_tsc_offset_host(struct kvm_vcpu *vcpu, s64 adjustment)
700 {
701 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, true);
702 }
703
704 int kvm_mmu_module_init(void);
705 void kvm_mmu_module_exit(void);
706
707 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
708 int kvm_mmu_create(struct kvm_vcpu *vcpu);
709 int kvm_mmu_setup(struct kvm_vcpu *vcpu);
710 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
711 u64 dirty_mask, u64 nx_mask, u64 x_mask);
712
713 int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
714 void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
715 int kvm_mmu_rmap_write_protect(struct kvm *kvm, u64 gfn,
716 struct kvm_memory_slot *slot);
717 void kvm_mmu_zap_all(struct kvm *kvm);
718 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
719 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
720
721 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
722
723 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
724 const void *val, int bytes);
725 u8 kvm_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn);
726
727 extern bool tdp_enabled;
728
729 u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
730
731 /* control of guest tsc rate supported? */
732 extern bool kvm_has_tsc_control;
733 /* minimum supported tsc_khz for guests */
734 extern u32 kvm_min_guest_tsc_khz;
735 /* maximum supported tsc_khz for guests */
736 extern u32 kvm_max_guest_tsc_khz;
737
738 enum emulation_result {
739 EMULATE_DONE, /* no further processing */
740 EMULATE_DO_MMIO, /* kvm_run filled with mmio request */
741 EMULATE_FAIL, /* can't emulate this instruction */
742 };
743
744 #define EMULTYPE_NO_DECODE (1 << 0)
745 #define EMULTYPE_TRAP_UD (1 << 1)
746 #define EMULTYPE_SKIP (1 << 2)
747 #define EMULTYPE_RETRY (1 << 3)
748 int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
749 int emulation_type, void *insn, int insn_len);
750
emulate_instruction(struct kvm_vcpu * vcpu,int emulation_type)751 static inline int emulate_instruction(struct kvm_vcpu *vcpu,
752 int emulation_type)
753 {
754 return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
755 }
756
757 void kvm_enable_efer_bits(u64);
758 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
759 int kvm_set_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
760
761 struct x86_emulate_ctxt;
762
763 int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
764 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
765 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
766 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
767
768 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
769 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
770
771 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
772 int reason, bool has_error_code, u32 error_code);
773
774 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
775 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
776 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
777 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
778 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
779 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
780 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
781 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
782 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
783 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
784
785 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
786 int kvm_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data);
787
788 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
789 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
790 bool kvm_rdpmc(struct kvm_vcpu *vcpu);
791
792 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
793 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
794 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
795 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
796 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
797 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
798 gfn_t gfn, void *data, int offset, int len,
799 u32 access);
800 void kvm_propagate_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
801 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
802
803 int kvm_pic_set_irq(void *opaque, int irq, int level);
804
805 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
806
807 int fx_init(struct kvm_vcpu *vcpu);
808
809 void kvm_mmu_flush_tlb(struct kvm_vcpu *vcpu);
810 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
811 const u8 *new, int bytes);
812 int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
813 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
814 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
815 int kvm_mmu_load(struct kvm_vcpu *vcpu);
816 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
817 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
818 gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access);
819 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
820 struct x86_exception *exception);
821 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
822 struct x86_exception *exception);
823 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
824 struct x86_exception *exception);
825 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
826 struct x86_exception *exception);
827
828 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
829
830 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
831 void *insn, int insn_len);
832 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
833
834 void kvm_enable_tdp(void);
835 void kvm_disable_tdp(void);
836
837 int complete_pio(struct kvm_vcpu *vcpu);
838 bool kvm_check_iopl(struct kvm_vcpu *vcpu);
839
translate_gpa(struct kvm_vcpu * vcpu,gpa_t gpa,u32 access)840 static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access)
841 {
842 return gpa;
843 }
844
page_header(hpa_t shadow_page)845 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
846 {
847 struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
848
849 return (struct kvm_mmu_page *)page_private(page);
850 }
851
kvm_read_ldt(void)852 static inline u16 kvm_read_ldt(void)
853 {
854 u16 ldt;
855 asm("sldt %0" : "=g"(ldt));
856 return ldt;
857 }
858
kvm_load_ldt(u16 sel)859 static inline void kvm_load_ldt(u16 sel)
860 {
861 asm("lldt %0" : : "rm"(sel));
862 }
863
864 #ifdef CONFIG_X86_64
read_msr(unsigned long msr)865 static inline unsigned long read_msr(unsigned long msr)
866 {
867 u64 value;
868
869 rdmsrl(msr, value);
870 return value;
871 }
872 #endif
873
get_rdx_init_val(void)874 static inline u32 get_rdx_init_val(void)
875 {
876 return 0x600; /* P6 family */
877 }
878
kvm_inject_gp(struct kvm_vcpu * vcpu,u32 error_code)879 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
880 {
881 kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
882 }
883
884 #define TSS_IOPB_BASE_OFFSET 0x66
885 #define TSS_BASE_SIZE 0x68
886 #define TSS_IOPB_SIZE (65536 / 8)
887 #define TSS_REDIRECTION_SIZE (256 / 8)
888 #define RMODE_TSS_SIZE \
889 (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
890
891 enum {
892 TASK_SWITCH_CALL = 0,
893 TASK_SWITCH_IRET = 1,
894 TASK_SWITCH_JMP = 2,
895 TASK_SWITCH_GATE = 3,
896 };
897
898 #define HF_GIF_MASK (1 << 0)
899 #define HF_HIF_MASK (1 << 1)
900 #define HF_VINTR_MASK (1 << 2)
901 #define HF_NMI_MASK (1 << 3)
902 #define HF_IRET_MASK (1 << 4)
903 #define HF_GUEST_MASK (1 << 5) /* VCPU is in guest-mode */
904
905 /*
906 * Hardware virtualization extension instructions may fault if a
907 * reboot turns off virtualization while processes are running.
908 * Trap the fault and ignore the instruction if that happens.
909 */
910 asmlinkage void kvm_spurious_fault(void);
911 extern bool kvm_rebooting;
912
913 #define ____kvm_handle_fault_on_reboot(insn, cleanup_insn) \
914 "666: " insn "\n\t" \
915 "668: \n\t" \
916 ".pushsection .fixup, \"ax\" \n" \
917 "667: \n\t" \
918 cleanup_insn "\n\t" \
919 "cmpb $0, kvm_rebooting \n\t" \
920 "jne 668b \n\t" \
921 __ASM_SIZE(push) " $666b \n\t" \
922 "call kvm_spurious_fault \n\t" \
923 ".popsection \n\t" \
924 ".pushsection __ex_table, \"a\" \n\t" \
925 _ASM_PTR " 666b, 667b \n\t" \
926 ".popsection"
927
928 #define __kvm_handle_fault_on_reboot(insn) \
929 ____kvm_handle_fault_on_reboot(insn, "")
930
931 #define KVM_ARCH_WANT_MMU_NOTIFIER
932 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
933 int kvm_age_hva(struct kvm *kvm, unsigned long hva);
934 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
935 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
936 int cpuid_maxphyaddr(struct kvm_vcpu *vcpu);
937 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
938 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
939 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
940
941 void kvm_define_shared_msr(unsigned index, u32 msr);
942 void kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
943
944 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
945
946 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
947 struct kvm_async_pf *work);
948 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
949 struct kvm_async_pf *work);
950 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
951 struct kvm_async_pf *work);
952 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
953 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
954
955 void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
956
957 int kvm_is_in_guest(void);
958
959 void kvm_pmu_init(struct kvm_vcpu *vcpu);
960 void kvm_pmu_destroy(struct kvm_vcpu *vcpu);
961 void kvm_pmu_reset(struct kvm_vcpu *vcpu);
962 void kvm_pmu_cpuid_update(struct kvm_vcpu *vcpu);
963 bool kvm_pmu_msr(struct kvm_vcpu *vcpu, u32 msr);
964 int kvm_pmu_get_msr(struct kvm_vcpu *vcpu, u32 msr, u64 *data);
965 int kvm_pmu_set_msr(struct kvm_vcpu *vcpu, u32 msr, u64 data);
966 int kvm_pmu_read_pmc(struct kvm_vcpu *vcpu, unsigned pmc, u64 *data);
967 void kvm_handle_pmu_event(struct kvm_vcpu *vcpu);
968 void kvm_deliver_pmi(struct kvm_vcpu *vcpu);
969
970 #endif /* _ASM_X86_KVM_HOST_H */
971