1 /*
2 * Kernel-based Virtual Machine driver for Linux
3 *
4 * This header defines architecture specific interfaces, x86 version
5 *
6 * This work is licensed under the terms of the GNU GPL, version 2. See
7 * the COPYING file in the top-level directory.
8 *
9 */
10
11 #ifndef _ASM_X86_KVM_HOST_H
12 #define _ASM_X86_KVM_HOST_H
13
14 #include <linux/types.h>
15 #include <linux/mm.h>
16 #include <linux/mmu_notifier.h>
17 #include <linux/tracepoint.h>
18 #include <linux/cpumask.h>
19 #include <linux/irq_work.h>
20
21 #include <linux/kvm.h>
22 #include <linux/kvm_para.h>
23 #include <linux/kvm_types.h>
24 #include <linux/perf_event.h>
25 #include <linux/pvclock_gtod.h>
26 #include <linux/clocksource.h>
27
28 #include <asm/pvclock-abi.h>
29 #include <asm/desc.h>
30 #include <asm/mtrr.h>
31 #include <asm/msr-index.h>
32 #include <asm/asm.h>
33
34 #define KVM_MAX_VCPUS 255
35 #define KVM_SOFT_MAX_VCPUS 160
36 #define KVM_USER_MEM_SLOTS 125
37 /* memory slots that are not exposed to userspace */
38 #define KVM_PRIVATE_MEM_SLOTS 3
39 #define KVM_MEM_SLOTS_NUM (KVM_USER_MEM_SLOTS + KVM_PRIVATE_MEM_SLOTS)
40
41 #define KVM_MMIO_SIZE 16
42
43 #define KVM_PIO_PAGE_OFFSET 1
44 #define KVM_COALESCED_MMIO_PAGE_OFFSET 2
45
46 #define KVM_IRQCHIP_NUM_PINS KVM_IOAPIC_NUM_PINS
47
48 #define CR0_RESERVED_BITS \
49 (~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
50 | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
51 | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
52
53 #define CR3_PAE_RESERVED_BITS ((X86_CR3_PWT | X86_CR3_PCD) - 1)
54 #define CR3_NONPAE_RESERVED_BITS ((PAGE_SIZE-1) & ~(X86_CR3_PWT | X86_CR3_PCD))
55 #define CR3_PCID_ENABLED_RESERVED_BITS 0xFFFFFF0000000000ULL
56 #define CR3_L_MODE_RESERVED_BITS (CR3_NONPAE_RESERVED_BITS | \
57 0xFFFFFF0000000000ULL)
58 #define CR4_RESERVED_BITS \
59 (~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
60 | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE \
61 | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR | X86_CR4_PCIDE \
62 | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_RDWRGSFS \
63 | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE))
64
65 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
66
67
68
69 #define INVALID_PAGE (~(hpa_t)0)
70 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
71
72 #define UNMAPPED_GVA (~(gpa_t)0)
73
74 /* KVM Hugepage definitions for x86 */
75 #define KVM_NR_PAGE_SIZES 3
76 #define KVM_HPAGE_GFN_SHIFT(x) (((x) - 1) * 9)
77 #define KVM_HPAGE_SHIFT(x) (PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
78 #define KVM_HPAGE_SIZE(x) (1UL << KVM_HPAGE_SHIFT(x))
79 #define KVM_HPAGE_MASK(x) (~(KVM_HPAGE_SIZE(x) - 1))
80 #define KVM_PAGES_PER_HPAGE(x) (KVM_HPAGE_SIZE(x) / PAGE_SIZE)
81
82 #define SELECTOR_TI_MASK (1 << 2)
83 #define SELECTOR_RPL_MASK 0x03
84
85 #define IOPL_SHIFT 12
86
87 #define KVM_PERMILLE_MMU_PAGES 20
88 #define KVM_MIN_ALLOC_MMU_PAGES 64
89 #define KVM_MMU_HASH_SHIFT 10
90 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
91 #define KVM_MIN_FREE_MMU_PAGES 5
92 #define KVM_REFILL_PAGES 25
93 #define KVM_MAX_CPUID_ENTRIES 80
94 #define KVM_NR_FIXED_MTRR_REGION 88
95 #define KVM_NR_VAR_MTRR 8
96
97 #define ASYNC_PF_PER_VCPU 64
98
99 struct kvm_vcpu;
100 struct kvm;
101 struct kvm_async_pf;
102
103 enum kvm_reg {
104 VCPU_REGS_RAX = 0,
105 VCPU_REGS_RCX = 1,
106 VCPU_REGS_RDX = 2,
107 VCPU_REGS_RBX = 3,
108 VCPU_REGS_RSP = 4,
109 VCPU_REGS_RBP = 5,
110 VCPU_REGS_RSI = 6,
111 VCPU_REGS_RDI = 7,
112 #ifdef CONFIG_X86_64
113 VCPU_REGS_R8 = 8,
114 VCPU_REGS_R9 = 9,
115 VCPU_REGS_R10 = 10,
116 VCPU_REGS_R11 = 11,
117 VCPU_REGS_R12 = 12,
118 VCPU_REGS_R13 = 13,
119 VCPU_REGS_R14 = 14,
120 VCPU_REGS_R15 = 15,
121 #endif
122 VCPU_REGS_RIP,
123 NR_VCPU_REGS
124 };
125
126 enum kvm_reg_ex {
127 VCPU_EXREG_PDPTR = NR_VCPU_REGS,
128 VCPU_EXREG_CR3,
129 VCPU_EXREG_RFLAGS,
130 VCPU_EXREG_CPL,
131 VCPU_EXREG_SEGMENTS,
132 };
133
134 enum {
135 VCPU_SREG_ES,
136 VCPU_SREG_CS,
137 VCPU_SREG_SS,
138 VCPU_SREG_DS,
139 VCPU_SREG_FS,
140 VCPU_SREG_GS,
141 VCPU_SREG_TR,
142 VCPU_SREG_LDTR,
143 };
144
145 #include <asm/kvm_emulate.h>
146
147 #define KVM_NR_MEM_OBJS 40
148
149 #define KVM_NR_DB_REGS 4
150
151 #define DR6_BD (1 << 13)
152 #define DR6_BS (1 << 14)
153 #define DR6_FIXED_1 0xffff0ff0
154 #define DR6_VOLATILE 0x0000e00f
155
156 #define DR7_BP_EN_MASK 0x000000ff
157 #define DR7_GE (1 << 9)
158 #define DR7_GD (1 << 13)
159 #define DR7_FIXED_1 0x00000400
160 #define DR7_VOLATILE 0xffff23ff
161
162 /* apic attention bits */
163 #define KVM_APIC_CHECK_VAPIC 0
164 /*
165 * The following bit is set with PV-EOI, unset on EOI.
166 * We detect PV-EOI changes by guest by comparing
167 * this bit with PV-EOI in guest memory.
168 * See the implementation in apic_update_pv_eoi.
169 */
170 #define KVM_APIC_PV_EOI_PENDING 1
171
172 /*
173 * We don't want allocation failures within the mmu code, so we preallocate
174 * enough memory for a single page fault in a cache.
175 */
176 struct kvm_mmu_memory_cache {
177 int nobjs;
178 void *objects[KVM_NR_MEM_OBJS];
179 };
180
181 /*
182 * kvm_mmu_page_role, below, is defined as:
183 *
184 * bits 0:3 - total guest paging levels (2-4, or zero for real mode)
185 * bits 4:7 - page table level for this shadow (1-4)
186 * bits 8:9 - page table quadrant for 2-level guests
187 * bit 16 - direct mapping of virtual to physical mapping at gfn
188 * used for real mode and two-dimensional paging
189 * bits 17:19 - common access permissions for all ptes in this shadow page
190 */
191 union kvm_mmu_page_role {
192 unsigned word;
193 struct {
194 unsigned level:4;
195 unsigned cr4_pae:1;
196 unsigned quadrant:2;
197 unsigned pad_for_nice_hex_output:6;
198 unsigned direct:1;
199 unsigned access:3;
200 unsigned invalid:1;
201 unsigned nxe:1;
202 unsigned cr0_wp:1;
203 unsigned smep_andnot_wp:1;
204 };
205 };
206
207 struct kvm_mmu_page {
208 struct list_head link;
209 struct hlist_node hash_link;
210
211 /*
212 * The following two entries are used to key the shadow page in the
213 * hash table.
214 */
215 gfn_t gfn;
216 union kvm_mmu_page_role role;
217
218 u64 *spt;
219 /* hold the gfn of each spte inside spt */
220 gfn_t *gfns;
221 bool unsync;
222 int root_count; /* Currently serving as active root */
223 unsigned int unsync_children;
224 unsigned long parent_ptes; /* Reverse mapping for parent_pte */
225 DECLARE_BITMAP(unsync_child_bitmap, 512);
226
227 #ifdef CONFIG_X86_32
228 int clear_spte_count;
229 #endif
230
231 int write_flooding_count;
232 bool mmio_cached;
233 };
234
235 struct kvm_pio_request {
236 unsigned long count;
237 int in;
238 int port;
239 int size;
240 };
241
242 /*
243 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
244 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
245 * mode.
246 */
247 struct kvm_mmu {
248 void (*new_cr3)(struct kvm_vcpu *vcpu);
249 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
250 unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
251 u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
252 int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
253 bool prefault);
254 void (*inject_page_fault)(struct kvm_vcpu *vcpu,
255 struct x86_exception *fault);
256 void (*free)(struct kvm_vcpu *vcpu);
257 gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
258 struct x86_exception *exception);
259 gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access);
260 int (*sync_page)(struct kvm_vcpu *vcpu,
261 struct kvm_mmu_page *sp);
262 void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
263 void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
264 u64 *spte, const void *pte);
265 hpa_t root_hpa;
266 int root_level;
267 int shadow_root_level;
268 union kvm_mmu_page_role base_role;
269 bool direct_map;
270
271 /*
272 * Bitmap; bit set = permission fault
273 * Byte index: page fault error code [4:1]
274 * Bit index: pte permissions in ACC_* format
275 */
276 u8 permissions[16];
277
278 u64 *pae_root;
279 u64 *lm_root;
280 u64 rsvd_bits_mask[2][4];
281
282 /*
283 * Bitmap: bit set = last pte in walk
284 * index[0:1]: level (zero-based)
285 * index[2]: pte.ps
286 */
287 u8 last_pte_bitmap;
288
289 bool nx;
290
291 u64 pdptrs[4]; /* pae */
292 };
293
294 enum pmc_type {
295 KVM_PMC_GP = 0,
296 KVM_PMC_FIXED,
297 };
298
299 struct kvm_pmc {
300 enum pmc_type type;
301 u8 idx;
302 u64 counter;
303 u64 eventsel;
304 struct perf_event *perf_event;
305 struct kvm_vcpu *vcpu;
306 };
307
308 struct kvm_pmu {
309 unsigned nr_arch_gp_counters;
310 unsigned nr_arch_fixed_counters;
311 unsigned available_event_types;
312 u64 fixed_ctr_ctrl;
313 u64 global_ctrl;
314 u64 global_status;
315 u64 global_ovf_ctrl;
316 u64 counter_bitmask[2];
317 u64 global_ctrl_mask;
318 u8 version;
319 struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
320 struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
321 struct irq_work irq_work;
322 u64 reprogram_pmi;
323 };
324
325 struct kvm_vcpu_arch {
326 /*
327 * rip and regs accesses must go through
328 * kvm_{register,rip}_{read,write} functions.
329 */
330 unsigned long regs[NR_VCPU_REGS];
331 u32 regs_avail;
332 u32 regs_dirty;
333
334 unsigned long cr0;
335 unsigned long cr0_guest_owned_bits;
336 unsigned long cr2;
337 unsigned long cr3;
338 unsigned long cr4;
339 unsigned long cr4_guest_owned_bits;
340 unsigned long cr8;
341 u32 hflags;
342 u64 efer;
343 u64 apic_base;
344 struct kvm_lapic *apic; /* kernel irqchip context */
345 unsigned long apic_attention;
346 int32_t apic_arb_prio;
347 int mp_state;
348 u64 ia32_misc_enable_msr;
349 bool tpr_access_reporting;
350
351 /*
352 * Paging state of the vcpu
353 *
354 * If the vcpu runs in guest mode with two level paging this still saves
355 * the paging mode of the l1 guest. This context is always used to
356 * handle faults.
357 */
358 struct kvm_mmu mmu;
359
360 /*
361 * Paging state of an L2 guest (used for nested npt)
362 *
363 * This context will save all necessary information to walk page tables
364 * of the an L2 guest. This context is only initialized for page table
365 * walking and not for faulting since we never handle l2 page faults on
366 * the host.
367 */
368 struct kvm_mmu nested_mmu;
369
370 /*
371 * Pointer to the mmu context currently used for
372 * gva_to_gpa translations.
373 */
374 struct kvm_mmu *walk_mmu;
375
376 struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
377 struct kvm_mmu_memory_cache mmu_page_cache;
378 struct kvm_mmu_memory_cache mmu_page_header_cache;
379
380 struct fpu guest_fpu;
381 u64 xcr0;
382
383 struct kvm_pio_request pio;
384 void *pio_data;
385
386 u8 event_exit_inst_len;
387
388 struct kvm_queued_exception {
389 bool pending;
390 bool has_error_code;
391 bool reinject;
392 u8 nr;
393 u32 error_code;
394 } exception;
395
396 struct kvm_queued_interrupt {
397 bool pending;
398 bool soft;
399 u8 nr;
400 } interrupt;
401
402 int halt_request; /* real mode on Intel only */
403
404 int cpuid_nent;
405 struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
406 /* emulate context */
407
408 struct x86_emulate_ctxt emulate_ctxt;
409 bool emulate_regs_need_sync_to_vcpu;
410 bool emulate_regs_need_sync_from_vcpu;
411 int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
412
413 gpa_t time;
414 struct pvclock_vcpu_time_info hv_clock;
415 unsigned int hw_tsc_khz;
416 struct gfn_to_hva_cache pv_time;
417 bool pv_time_enabled;
418 /* set guest stopped flag in pvclock flags field */
419 bool pvclock_set_guest_stopped_request;
420
421 struct {
422 u64 msr_val;
423 u64 last_steal;
424 u64 accum_steal;
425 struct gfn_to_hva_cache stime;
426 struct kvm_steal_time steal;
427 } st;
428
429 u64 last_guest_tsc;
430 u64 last_kernel_ns;
431 u64 last_host_tsc;
432 u64 tsc_offset_adjustment;
433 u64 this_tsc_nsec;
434 u64 this_tsc_write;
435 u8 this_tsc_generation;
436 bool tsc_catchup;
437 bool tsc_always_catchup;
438 s8 virtual_tsc_shift;
439 u32 virtual_tsc_mult;
440 u32 virtual_tsc_khz;
441 s64 ia32_tsc_adjust_msr;
442
443 atomic_t nmi_queued; /* unprocessed asynchronous NMIs */
444 unsigned nmi_pending; /* NMI queued after currently running handler */
445 bool nmi_injected; /* Trying to inject an NMI this entry */
446
447 struct mtrr_state_type mtrr_state;
448 u32 pat;
449
450 int switch_db_regs;
451 unsigned long db[KVM_NR_DB_REGS];
452 unsigned long dr6;
453 unsigned long dr7;
454 unsigned long eff_db[KVM_NR_DB_REGS];
455 unsigned long guest_debug_dr7;
456
457 u64 mcg_cap;
458 u64 mcg_status;
459 u64 mcg_ctl;
460 u64 *mce_banks;
461
462 /* Cache MMIO info */
463 u64 mmio_gva;
464 unsigned access;
465 gfn_t mmio_gfn;
466
467 struct kvm_pmu pmu;
468
469 /* used for guest single stepping over the given code position */
470 unsigned long singlestep_rip;
471
472 /* fields used by HYPER-V emulation */
473 u64 hv_vapic;
474
475 cpumask_var_t wbinvd_dirty_mask;
476
477 unsigned long last_retry_eip;
478 unsigned long last_retry_addr;
479
480 struct {
481 bool halted;
482 gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
483 struct gfn_to_hva_cache data;
484 u64 msr_val;
485 u32 id;
486 bool send_user_only;
487 } apf;
488
489 /* OSVW MSRs (AMD only) */
490 struct {
491 u64 length;
492 u64 status;
493 } osvw;
494
495 struct {
496 u64 msr_val;
497 struct gfn_to_hva_cache data;
498 } pv_eoi;
499
500 /*
501 * Indicate whether the access faults on its page table in guest
502 * which is set when fix page fault and used to detect unhandeable
503 * instruction.
504 */
505 bool write_fault_to_shadow_pgtable;
506 };
507
508 struct kvm_lpage_info {
509 int write_count;
510 };
511
512 struct kvm_arch_memory_slot {
513 unsigned long *rmap[KVM_NR_PAGE_SIZES];
514 struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
515 };
516
517 struct kvm_apic_map {
518 struct rcu_head rcu;
519 u8 ldr_bits;
520 /* fields bellow are used to decode ldr values in different modes */
521 u32 cid_shift, cid_mask, lid_mask;
522 struct kvm_lapic *phys_map[256];
523 /* first index is cluster id second is cpu id in a cluster */
524 struct kvm_lapic *logical_map[16][16];
525 };
526
527 struct kvm_arch {
528 unsigned int n_used_mmu_pages;
529 unsigned int n_requested_mmu_pages;
530 unsigned int n_max_mmu_pages;
531 unsigned int indirect_shadow_pages;
532 struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
533 /*
534 * Hash table of struct kvm_mmu_page.
535 */
536 struct list_head active_mmu_pages;
537 struct list_head assigned_dev_head;
538 struct iommu_domain *iommu_domain;
539 int iommu_flags;
540 struct kvm_pic *vpic;
541 struct kvm_ioapic *vioapic;
542 struct kvm_pit *vpit;
543 int vapics_in_nmi_mode;
544 struct mutex apic_map_lock;
545 struct kvm_apic_map *apic_map;
546
547 unsigned int tss_addr;
548 struct page *apic_access_page;
549
550 gpa_t wall_clock;
551
552 struct page *ept_identity_pagetable;
553 bool ept_identity_pagetable_done;
554 gpa_t ept_identity_map_addr;
555
556 unsigned long irq_sources_bitmap;
557 s64 kvmclock_offset;
558 raw_spinlock_t tsc_write_lock;
559 u64 last_tsc_nsec;
560 u64 last_tsc_write;
561 u32 last_tsc_khz;
562 u64 cur_tsc_nsec;
563 u64 cur_tsc_write;
564 u64 cur_tsc_offset;
565 u8 cur_tsc_generation;
566 int nr_vcpus_matched_tsc;
567
568 spinlock_t pvclock_gtod_sync_lock;
569 bool use_master_clock;
570 u64 master_kernel_ns;
571 cycle_t master_cycle_now;
572
573 struct kvm_xen_hvm_config xen_hvm_config;
574
575 /* fields used by HYPER-V emulation */
576 u64 hv_guest_os_id;
577 u64 hv_hypercall;
578
579 #ifdef CONFIG_KVM_MMU_AUDIT
580 int audit_point;
581 #endif
582 };
583
584 struct kvm_vm_stat {
585 u32 mmu_shadow_zapped;
586 u32 mmu_pte_write;
587 u32 mmu_pte_updated;
588 u32 mmu_pde_zapped;
589 u32 mmu_flooded;
590 u32 mmu_recycled;
591 u32 mmu_cache_miss;
592 u32 mmu_unsync;
593 u32 remote_tlb_flush;
594 u32 lpages;
595 };
596
597 struct kvm_vcpu_stat {
598 u32 pf_fixed;
599 u32 pf_guest;
600 u32 tlb_flush;
601 u32 invlpg;
602
603 u32 exits;
604 u32 io_exits;
605 u32 mmio_exits;
606 u32 signal_exits;
607 u32 irq_window_exits;
608 u32 nmi_window_exits;
609 u32 halt_exits;
610 u32 halt_wakeup;
611 u32 request_irq_exits;
612 u32 irq_exits;
613 u32 host_state_reload;
614 u32 efer_reload;
615 u32 fpu_reload;
616 u32 insn_emulation;
617 u32 insn_emulation_fail;
618 u32 hypercalls;
619 u32 irq_injections;
620 u32 nmi_injections;
621 };
622
623 struct x86_instruction_info;
624
625 struct msr_data {
626 bool host_initiated;
627 u32 index;
628 u64 data;
629 };
630
631 struct kvm_x86_ops {
632 int (*cpu_has_kvm_support)(void); /* __init */
633 int (*disabled_by_bios)(void); /* __init */
634 int (*hardware_enable)(void *dummy);
635 void (*hardware_disable)(void *dummy);
636 void (*check_processor_compatibility)(void *rtn);
637 int (*hardware_setup)(void); /* __init */
638 void (*hardware_unsetup)(void); /* __exit */
639 bool (*cpu_has_accelerated_tpr)(void);
640 void (*cpuid_update)(struct kvm_vcpu *vcpu);
641
642 /* Create, but do not attach this VCPU */
643 struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
644 void (*vcpu_free)(struct kvm_vcpu *vcpu);
645 void (*vcpu_reset)(struct kvm_vcpu *vcpu);
646
647 void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
648 void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
649 void (*vcpu_put)(struct kvm_vcpu *vcpu);
650
651 void (*update_db_bp_intercept)(struct kvm_vcpu *vcpu);
652 int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
653 int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
654 u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
655 void (*get_segment)(struct kvm_vcpu *vcpu,
656 struct kvm_segment *var, int seg);
657 int (*get_cpl)(struct kvm_vcpu *vcpu);
658 void (*set_segment)(struct kvm_vcpu *vcpu,
659 struct kvm_segment *var, int seg);
660 void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
661 void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
662 void (*decache_cr3)(struct kvm_vcpu *vcpu);
663 void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
664 void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
665 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
666 int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
667 void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
668 void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
669 void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
670 void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
671 void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
672 void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
673 void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
674 unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
675 void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
676 void (*fpu_activate)(struct kvm_vcpu *vcpu);
677 void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
678
679 void (*tlb_flush)(struct kvm_vcpu *vcpu);
680
681 void (*run)(struct kvm_vcpu *vcpu);
682 int (*handle_exit)(struct kvm_vcpu *vcpu);
683 void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
684 void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
685 u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
686 void (*patch_hypercall)(struct kvm_vcpu *vcpu,
687 unsigned char *hypercall_addr);
688 void (*set_irq)(struct kvm_vcpu *vcpu);
689 void (*set_nmi)(struct kvm_vcpu *vcpu);
690 void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
691 bool has_error_code, u32 error_code,
692 bool reinject);
693 void (*cancel_injection)(struct kvm_vcpu *vcpu);
694 int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
695 int (*nmi_allowed)(struct kvm_vcpu *vcpu);
696 bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
697 void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
698 int (*enable_nmi_window)(struct kvm_vcpu *vcpu);
699 int (*enable_irq_window)(struct kvm_vcpu *vcpu);
700 void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
701 int (*vm_has_apicv)(struct kvm *kvm);
702 void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
703 void (*hwapic_isr_update)(struct kvm *kvm, int isr);
704 void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap);
705 void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu, bool set);
706 void (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
707 void (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
708 int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
709 int (*get_tdp_level)(void);
710 u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
711 int (*get_lpage_level)(void);
712 bool (*rdtscp_supported)(void);
713 bool (*invpcid_supported)(void);
714 void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment, bool host);
715
716 void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
717
718 void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
719
720 bool (*has_wbinvd_exit)(void);
721
722 void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale);
723 u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu);
724 void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
725
726 u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc);
727 u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu, u64 host_tsc);
728
729 void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
730
731 int (*check_intercept)(struct kvm_vcpu *vcpu,
732 struct x86_instruction_info *info,
733 enum x86_intercept_stage stage);
734 void (*handle_external_intr)(struct kvm_vcpu *vcpu);
735 };
736
737 struct kvm_arch_async_pf {
738 u32 token;
739 gfn_t gfn;
740 unsigned long cr3;
741 bool direct_map;
742 };
743
744 extern struct kvm_x86_ops *kvm_x86_ops;
745
adjust_tsc_offset_guest(struct kvm_vcpu * vcpu,s64 adjustment)746 static inline void adjust_tsc_offset_guest(struct kvm_vcpu *vcpu,
747 s64 adjustment)
748 {
749 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, false);
750 }
751
adjust_tsc_offset_host(struct kvm_vcpu * vcpu,s64 adjustment)752 static inline void adjust_tsc_offset_host(struct kvm_vcpu *vcpu, s64 adjustment)
753 {
754 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, true);
755 }
756
757 int kvm_mmu_module_init(void);
758 void kvm_mmu_module_exit(void);
759
760 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
761 int kvm_mmu_create(struct kvm_vcpu *vcpu);
762 int kvm_mmu_setup(struct kvm_vcpu *vcpu);
763 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
764 u64 dirty_mask, u64 nx_mask, u64 x_mask);
765
766 int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
767 void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
768 void kvm_mmu_write_protect_pt_masked(struct kvm *kvm,
769 struct kvm_memory_slot *slot,
770 gfn_t gfn_offset, unsigned long mask);
771 void kvm_mmu_zap_all(struct kvm *kvm);
772 void kvm_mmu_zap_mmio_sptes(struct kvm *kvm);
773 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
774 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
775
776 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
777
778 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
779 const void *val, int bytes);
780 u8 kvm_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn);
781
782 extern bool tdp_enabled;
783
784 u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
785
786 /* control of guest tsc rate supported? */
787 extern bool kvm_has_tsc_control;
788 /* minimum supported tsc_khz for guests */
789 extern u32 kvm_min_guest_tsc_khz;
790 /* maximum supported tsc_khz for guests */
791 extern u32 kvm_max_guest_tsc_khz;
792
793 enum emulation_result {
794 EMULATE_DONE, /* no further processing */
795 EMULATE_DO_MMIO, /* kvm_run filled with mmio request */
796 EMULATE_FAIL, /* can't emulate this instruction */
797 };
798
799 #define EMULTYPE_NO_DECODE (1 << 0)
800 #define EMULTYPE_TRAP_UD (1 << 1)
801 #define EMULTYPE_SKIP (1 << 2)
802 #define EMULTYPE_RETRY (1 << 3)
803 #define EMULTYPE_NO_REEXECUTE (1 << 4)
804 int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
805 int emulation_type, void *insn, int insn_len);
806
emulate_instruction(struct kvm_vcpu * vcpu,int emulation_type)807 static inline int emulate_instruction(struct kvm_vcpu *vcpu,
808 int emulation_type)
809 {
810 return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
811 }
812
813 void kvm_enable_efer_bits(u64);
814 bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
815 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
816 int kvm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
817
818 struct x86_emulate_ctxt;
819
820 int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
821 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
822 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
823 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
824
825 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
826 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
827 void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, unsigned int vector);
828
829 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
830 int reason, bool has_error_code, u32 error_code);
831
832 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
833 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
834 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
835 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
836 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
837 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
838 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
839 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
840 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
841 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
842
843 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
844 int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
845
846 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
847 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
848 bool kvm_rdpmc(struct kvm_vcpu *vcpu);
849
850 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
851 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
852 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
853 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
854 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
855 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
856 gfn_t gfn, void *data, int offset, int len,
857 u32 access);
858 void kvm_propagate_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
859 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
860
__kvm_irq_line_state(unsigned long * irq_state,int irq_source_id,int level)861 static inline int __kvm_irq_line_state(unsigned long *irq_state,
862 int irq_source_id, int level)
863 {
864 /* Logical OR for level trig interrupt */
865 if (level)
866 __set_bit(irq_source_id, irq_state);
867 else
868 __clear_bit(irq_source_id, irq_state);
869
870 return !!(*irq_state);
871 }
872
873 int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
874 void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
875
876 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
877
878 int fx_init(struct kvm_vcpu *vcpu);
879
880 void kvm_mmu_flush_tlb(struct kvm_vcpu *vcpu);
881 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
882 const u8 *new, int bytes);
883 int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
884 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
885 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
886 int kvm_mmu_load(struct kvm_vcpu *vcpu);
887 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
888 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
889 gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access);
890 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
891 struct x86_exception *exception);
892 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
893 struct x86_exception *exception);
894 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
895 struct x86_exception *exception);
896 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
897 struct x86_exception *exception);
898
899 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
900
901 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
902 void *insn, int insn_len);
903 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
904
905 void kvm_enable_tdp(void);
906 void kvm_disable_tdp(void);
907
908 int complete_pio(struct kvm_vcpu *vcpu);
909 bool kvm_check_iopl(struct kvm_vcpu *vcpu);
910
translate_gpa(struct kvm_vcpu * vcpu,gpa_t gpa,u32 access)911 static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access)
912 {
913 return gpa;
914 }
915
page_header(hpa_t shadow_page)916 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
917 {
918 struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
919
920 return (struct kvm_mmu_page *)page_private(page);
921 }
922
kvm_read_ldt(void)923 static inline u16 kvm_read_ldt(void)
924 {
925 u16 ldt;
926 asm("sldt %0" : "=g"(ldt));
927 return ldt;
928 }
929
kvm_load_ldt(u16 sel)930 static inline void kvm_load_ldt(u16 sel)
931 {
932 asm("lldt %0" : : "rm"(sel));
933 }
934
935 #ifdef CONFIG_X86_64
read_msr(unsigned long msr)936 static inline unsigned long read_msr(unsigned long msr)
937 {
938 u64 value;
939
940 rdmsrl(msr, value);
941 return value;
942 }
943 #endif
944
get_rdx_init_val(void)945 static inline u32 get_rdx_init_val(void)
946 {
947 return 0x600; /* P6 family */
948 }
949
kvm_inject_gp(struct kvm_vcpu * vcpu,u32 error_code)950 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
951 {
952 kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
953 }
954
955 #define TSS_IOPB_BASE_OFFSET 0x66
956 #define TSS_BASE_SIZE 0x68
957 #define TSS_IOPB_SIZE (65536 / 8)
958 #define TSS_REDIRECTION_SIZE (256 / 8)
959 #define RMODE_TSS_SIZE \
960 (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
961
962 enum {
963 TASK_SWITCH_CALL = 0,
964 TASK_SWITCH_IRET = 1,
965 TASK_SWITCH_JMP = 2,
966 TASK_SWITCH_GATE = 3,
967 };
968
969 #define HF_GIF_MASK (1 << 0)
970 #define HF_HIF_MASK (1 << 1)
971 #define HF_VINTR_MASK (1 << 2)
972 #define HF_NMI_MASK (1 << 3)
973 #define HF_IRET_MASK (1 << 4)
974 #define HF_GUEST_MASK (1 << 5) /* VCPU is in guest-mode */
975
976 /*
977 * Hardware virtualization extension instructions may fault if a
978 * reboot turns off virtualization while processes are running.
979 * Trap the fault and ignore the instruction if that happens.
980 */
981 asmlinkage void kvm_spurious_fault(void);
982
983 #define ____kvm_handle_fault_on_reboot(insn, cleanup_insn) \
984 "666: " insn "\n\t" \
985 "668: \n\t" \
986 ".pushsection .fixup, \"ax\" \n" \
987 "667: \n\t" \
988 cleanup_insn "\n\t" \
989 "cmpb $0, kvm_rebooting \n\t" \
990 "jne 668b \n\t" \
991 __ASM_SIZE(push) " $666b \n\t" \
992 "call kvm_spurious_fault \n\t" \
993 ".popsection \n\t" \
994 _ASM_EXTABLE(666b, 667b)
995
996 #define __kvm_handle_fault_on_reboot(insn) \
997 ____kvm_handle_fault_on_reboot(insn, "")
998
999 #define KVM_ARCH_WANT_MMU_NOTIFIER
1000 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
1001 int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
1002 int kvm_age_hva(struct kvm *kvm, unsigned long hva);
1003 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
1004 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
1005 int cpuid_maxphyaddr(struct kvm_vcpu *vcpu);
1006 int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
1007 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1008 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
1009 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
1010 void kvm_vcpu_reset(struct kvm_vcpu *vcpu);
1011
1012 void kvm_define_shared_msr(unsigned index, u32 msr);
1013 void kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
1014
1015 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1016
1017 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1018 struct kvm_async_pf *work);
1019 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1020 struct kvm_async_pf *work);
1021 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1022 struct kvm_async_pf *work);
1023 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
1024 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1025
1026 void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1027
1028 int kvm_is_in_guest(void);
1029
1030 void kvm_pmu_init(struct kvm_vcpu *vcpu);
1031 void kvm_pmu_destroy(struct kvm_vcpu *vcpu);
1032 void kvm_pmu_reset(struct kvm_vcpu *vcpu);
1033 void kvm_pmu_cpuid_update(struct kvm_vcpu *vcpu);
1034 bool kvm_pmu_msr(struct kvm_vcpu *vcpu, u32 msr);
1035 int kvm_pmu_get_msr(struct kvm_vcpu *vcpu, u32 msr, u64 *data);
1036 int kvm_pmu_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info);
1037 int kvm_pmu_read_pmc(struct kvm_vcpu *vcpu, unsigned pmc, u64 *data);
1038 void kvm_handle_pmu_event(struct kvm_vcpu *vcpu);
1039 void kvm_deliver_pmi(struct kvm_vcpu *vcpu);
1040
1041 #endif /* _ASM_X86_KVM_HOST_H */
1042