arch/x86/include/asm/kvm_host.h

   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 #include <asm/asm.h>
  31
  32 #define KVM_MAX_VCPUS 254
  33 #define KVM_SOFT_MAX_VCPUS 160
  34 #define KVM_MEMORY_SLOTS 32
  35 /* memory slots that does not exposed to userspace */
  36 #define KVM_PRIVATE_MEM_SLOTS 4
  37 #define KVM_MEM_SLOTS_NUM (KVM_MEMORY_SLOTS + KVM_PRIVATE_MEM_SLOTS)
  38
  39 #define KVM_MMIO_SIZE 16
  40
  41 #define KVM_PIO_PAGE_OFFSET 1
  42 #define KVM_COALESCED_MMIO_PAGE_OFFSET 2
  43
  44 #define CR0_RESERVED_BITS                                               \
  45         (~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
  46                           | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
  47                           | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
  48
  49 #define CR3_PAE_RESERVED_BITS ((X86_CR3_PWT | X86_CR3_PCD) - 1)
  50 #define CR3_NONPAE_RESERVED_BITS ((PAGE_SIZE-1) & ~(X86_CR3_PWT | X86_CR3_PCD))
  51 #define CR3_L_MODE_RESERVED_BITS (CR3_NONPAE_RESERVED_BITS |    \
  52                                   0xFFFFFF0000000000ULL)
  53 #define CR4_RESERVED_BITS                                               \
  54         (~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
  55                           | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE     \
  56                           | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR  \
  57                           | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_RDWRGSFS \
  58                           | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE))
  59
  60 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
  61
  62
  63
  64 #define INVALID_PAGE (~(hpa_t)0)
  65 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
  66
  67 #define UNMAPPED_GVA (~(gpa_t)0)
  68
  69 /* KVM Hugepage definitions for x86 */
  70 #define KVM_NR_PAGE_SIZES       3
  71 #define KVM_HPAGE_GFN_SHIFT(x)  (((x) - 1) * 9)
  72 #define KVM_HPAGE_SHIFT(x)      (PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
  73 #define KVM_HPAGE_SIZE(x)       (1UL << KVM_HPAGE_SHIFT(x))
  74 #define KVM_HPAGE_MASK(x)       (~(KVM_HPAGE_SIZE(x) - 1))
  75 #define KVM_PAGES_PER_HPAGE(x)  (KVM_HPAGE_SIZE(x) / PAGE_SIZE)
  76
  77 #define DE_VECTOR 0
  78 #define DB_VECTOR 1
  79 #define BP_VECTOR 3
  80 #define OF_VECTOR 4
  81 #define BR_VECTOR 5
  82 #define UD_VECTOR 6
  83 #define NM_VECTOR 7
  84 #define DF_VECTOR 8
  85 #define TS_VECTOR 10
  86 #define NP_VECTOR 11
  87 #define SS_VECTOR 12
  88 #define GP_VECTOR 13
  89 #define PF_VECTOR 14
  90 #define MF_VECTOR 16
  91 #define MC_VECTOR 18
  92
  93 #define SELECTOR_TI_MASK (1 << 2)
  94 #define SELECTOR_RPL_MASK 0x03
  95
  96 #define IOPL_SHIFT 12
  97
  98 #define KVM_PERMILLE_MMU_PAGES 20
  99 #define KVM_MIN_ALLOC_MMU_PAGES 64
 100 #define KVM_MMU_HASH_SHIFT 10
 101 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
 102 #define KVM_MIN_FREE_MMU_PAGES 5
 103 #define KVM_REFILL_PAGES 25
 104 #define KVM_MAX_CPUID_ENTRIES 80
 105 #define KVM_NR_FIXED_MTRR_REGION 88
 106 #define KVM_NR_VAR_MTRR 8
 107
 108 #define ASYNC_PF_PER_VCPU 64
 109
 110 extern raw_spinlock_t kvm_lock;
 111 extern struct list_head vm_list;
 112
 113 struct kvm_vcpu;
 114 struct kvm;
 115 struct kvm_async_pf;
 116
 117 enum kvm_reg {
 118         VCPU_REGS_RAX = 0,
 119         VCPU_REGS_RCX = 1,
 120         VCPU_REGS_RDX = 2,
 121         VCPU_REGS_RBX = 3,
 122         VCPU_REGS_RSP = 4,
 123         VCPU_REGS_RBP = 5,
 124         VCPU_REGS_RSI = 6,
 125         VCPU_REGS_RDI = 7,
 126 #ifdef CONFIG_X86_64
 127         VCPU_REGS_R8 = 8,
 128         VCPU_REGS_R9 = 9,
 129         VCPU_REGS_R10 = 10,
 130         VCPU_REGS_R11 = 11,
 131         VCPU_REGS_R12 = 12,
 132         VCPU_REGS_R13 = 13,
 133         VCPU_REGS_R14 = 14,
 134         VCPU_REGS_R15 = 15,
 135 #endif
 136         VCPU_REGS_RIP,
 137         NR_VCPU_REGS
 138 };
 139
 140 enum kvm_reg_ex {
 141         VCPU_EXREG_PDPTR = NR_VCPU_REGS,
 142         VCPU_EXREG_CR3,
 143         VCPU_EXREG_RFLAGS,
 144         VCPU_EXREG_CPL,
 145         VCPU_EXREG_SEGMENTS,
 146 };
 147
 148 enum {
 149         VCPU_SREG_ES,
 150         VCPU_SREG_CS,
 151         VCPU_SREG_SS,
 152         VCPU_SREG_DS,
 153         VCPU_SREG_FS,
 154         VCPU_SREG_GS,
 155         VCPU_SREG_TR,
 156         VCPU_SREG_LDTR,
 157 };
 158
 159 #include <asm/kvm_emulate.h>
 160
 161 #define KVM_NR_MEM_OBJS 40
 162
 163 #define KVM_NR_DB_REGS  4
 164
 165 #define DR6_BD          (1 << 13)
 166 #define DR6_BS          (1 << 14)
 167 #define DR6_FIXED_1     0xffff0ff0
 168 #define DR6_VOLATILE    0x0000e00f
 169
 170 #define DR7_BP_EN_MASK  0x000000ff
 171 #define DR7_GE          (1 << 9)
 172 #define DR7_GD          (1 << 13)
 173 #define DR7_FIXED_1     0x00000400
 174 #define DR7_VOLATILE    0xffff23ff
 175
 176 /* apic attention bits */
 177 #define KVM_APIC_CHECK_VAPIC    0
 178 /*
 179  * The following bit is set with PV-EOI, unset on EOI.
 180  * We detect PV-EOI changes by guest by comparing
 181  * this bit with PV-EOI in guest memory.
 182  * See the implementation in apic_update_pv_eoi.
 183  */
 184 #define KVM_APIC_PV_EOI_PENDING 1
 185
 186 /*
 187  * We don't want allocation failures within the mmu code, so we preallocate
 188  * enough memory for a single page fault in a cache.
 189  */
 190 struct kvm_mmu_memory_cache {
 191         int nobjs;
 192         void *objects[KVM_NR_MEM_OBJS];
 193 };
 194
 195 /*
 196  * kvm_mmu_page_role, below, is defined as:
 197  *
 198  *   bits 0:3 - total guest paging levels (2-4, or zero for real mode)
 199  *   bits 4:7 - page table level for this shadow (1-4)
 200  *   bits 8:9 - page table quadrant for 2-level guests
 201  *   bit   16 - direct mapping of virtual to physical mapping at gfn
 202  *              used for real mode and two-dimensional paging
 203  *   bits 17:19 - common access permissions for all ptes in this shadow page
 204  */
 205 union kvm_mmu_page_role {
 206         unsigned word;
 207         struct {
 208                 unsigned level:4;
 209                 unsigned cr4_pae:1;
 210                 unsigned quadrant:2;
 211                 unsigned pad_for_nice_hex_output:6;
 212                 unsigned direct:1;
 213                 unsigned access:3;
 214                 unsigned invalid:1;
 215                 unsigned nxe:1;
 216                 unsigned cr0_wp:1;
 217                 unsigned smep_andnot_wp:1;
 218         };
 219 };
 220
 221 struct kvm_mmu_page {
 222         struct list_head link;
 223         struct hlist_node hash_link;
 224
 225         /*
 226          * The following two entries are used to key the shadow page in the
 227          * hash table.
 228          */
 229         gfn_t gfn;
 230         union kvm_mmu_page_role role;
 231
 232         u64 *spt;
 233         /* hold the gfn of each spte inside spt */
 234         gfn_t *gfns;
 235         /*
 236          * One bit set per slot which has memory
 237          * in this shadow page.
 238          */
 239         DECLARE_BITMAP(slot_bitmap, KVM_MEM_SLOTS_NUM);
 240         bool unsync;
 241         int root_count;          /* Currently serving as active root */
 242         unsigned int unsync_children;
 243         unsigned long parent_ptes;      /* Reverse mapping for parent_pte */
 244         DECLARE_BITMAP(unsync_child_bitmap, 512);
 245
 246 #ifdef CONFIG_X86_32
 247         int clear_spte_count;
 248 #endif
 249
 250         int write_flooding_count;
 251 };
 252
 253 struct kvm_pio_request {
 254         unsigned long count;
 255         int in;
 256         int port;
 257         int size;
 258 };
 259
 260 /*
 261  * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
 262  * 32-bit).  The kvm_mmu structure abstracts the details of the current mmu
 263  * mode.
 264  */
 265 struct kvm_mmu {
 266         void (*new_cr3)(struct kvm_vcpu *vcpu);
 267         void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
 268         unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
 269         u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
 270         int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
 271                           bool prefault);
 272         void (*inject_page_fault)(struct kvm_vcpu *vcpu,
 273                                   struct x86_exception *fault);
 274         void (*free)(struct kvm_vcpu *vcpu);
 275         gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
 276                             struct x86_exception *exception);
 277         gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access);
 278         int (*sync_page)(struct kvm_vcpu *vcpu,
 279                          struct kvm_mmu_page *sp);
 280         void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
 281         void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
 282                            u64 *spte, const void *pte);
 283         hpa_t root_hpa;
 284         int root_level;
 285         int shadow_root_level;
 286         union kvm_mmu_page_role base_role;
 287         bool direct_map;
 288
 289         u64 *pae_root;
 290         u64 *lm_root;
 291         u64 rsvd_bits_mask[2][4];
 292
 293         bool nx;
 294
 295         u64 pdptrs[4]; /* pae */
 296 };
 297
 298 enum pmc_type {
 299         KVM_PMC_GP = 0,
 300         KVM_PMC_FIXED,
 301 };
 302
 303 struct kvm_pmc {
 304         enum pmc_type type;
 305         u8 idx;
 306         u64 counter;
 307         u64 eventsel;
 308         struct perf_event *perf_event;
 309         struct kvm_vcpu *vcpu;
 310 };
 311
 312 struct kvm_pmu {
 313         unsigned nr_arch_gp_counters;
 314         unsigned nr_arch_fixed_counters;
 315         unsigned available_event_types;
 316         u64 fixed_ctr_ctrl;
 317         u64 global_ctrl;
 318         u64 global_status;
 319         u64 global_ovf_ctrl;
 320         u64 counter_bitmask[2];
 321         u64 global_ctrl_mask;
 322         u8 version;
 323         struct kvm_pmc gp_counters[X86_PMC_MAX_GENERIC];
 324         struct kvm_pmc fixed_counters[X86_PMC_MAX_FIXED];
 325         struct irq_work irq_work;
 326         u64 reprogram_pmi;
 327 };
 328
 329 struct kvm_vcpu_arch {
 330         /*
 331          * rip and regs accesses must go through
 332          * kvm_{register,rip}_{read,write} functions.
 333          */
 334         unsigned long regs[NR_VCPU_REGS];
 335         u32 regs_avail;
 336         u32 regs_dirty;
 337
 338         unsigned long cr0;
 339         unsigned long cr0_guest_owned_bits;
 340         unsigned long cr2;
 341         unsigned long cr3;
 342         unsigned long cr4;
 343         unsigned long cr4_guest_owned_bits;
 344         unsigned long cr8;
 345         u32 hflags;
 346         u64 efer;
 347         u64 apic_base;
 348         struct kvm_lapic *apic;    /* kernel irqchip context */
 349         unsigned long apic_attention;
 350         int32_t apic_arb_prio;
 351         int mp_state;
 352         int sipi_vector;
 353         u64 ia32_misc_enable_msr;
 354         bool tpr_access_reporting;
 355
 356         /*
 357          * Paging state of the vcpu
 358          *
 359          * If the vcpu runs in guest mode with two level paging this still saves
 360          * the paging mode of the l1 guest. This context is always used to
 361          * handle faults.
 362          */
 363         struct kvm_mmu mmu;
 364
 365         /*
 366          * Paging state of an L2 guest (used for nested npt)
 367          *
 368          * This context will save all necessary information to walk page tables
 369          * of the an L2 guest. This context is only initialized for page table
 370          * walking and not for faulting since we never handle l2 page faults on
 371          * the host.
 372          */
 373         struct kvm_mmu nested_mmu;
 374
 375         /*
 376          * Pointer to the mmu context currently used for
 377          * gva_to_gpa translations.
 378          */
 379         struct kvm_mmu *walk_mmu;
 380
 381         struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
 382         struct kvm_mmu_memory_cache mmu_page_cache;
 383         struct kvm_mmu_memory_cache mmu_page_header_cache;
 384
 385         struct fpu guest_fpu;
 386         u64 xcr0;
 387
 388         struct kvm_pio_request pio;
 389         void *pio_data;
 390
 391         u8 event_exit_inst_len;
 392
 393         struct kvm_queued_exception {
 394                 bool pending;
 395                 bool has_error_code;
 396                 bool reinject;
 397                 u8 nr;
 398                 u32 error_code;
 399         } exception;
 400
 401         struct kvm_queued_interrupt {
 402                 bool pending;
 403                 bool soft;
 404                 u8 nr;
 405         } interrupt;
 406
 407         int halt_request; /* real mode on Intel only */
 408
 409         int cpuid_nent;
 410         struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
 411         /* emulate context */
 412
 413         struct x86_emulate_ctxt emulate_ctxt;
 414         bool emulate_regs_need_sync_to_vcpu;
 415         bool emulate_regs_need_sync_from_vcpu;
 416
 417         gpa_t time;
 418         struct pvclock_vcpu_time_info hv_clock;
 419         unsigned int hw_tsc_khz;
 420         unsigned int time_offset;
 421         struct page *time_page;
 422
 423         struct {
 424                 u64 msr_val;
 425                 u64 last_steal;
 426                 u64 accum_steal;
 427                 struct gfn_to_hva_cache stime;
 428                 struct kvm_steal_time steal;
 429         } st;
 430
 431         u64 last_guest_tsc;
 432         u64 last_kernel_ns;
 433         u64 last_host_tsc;
 434         u64 tsc_offset_adjustment;
 435         u64 this_tsc_nsec;
 436         u64 this_tsc_write;
 437         u8  this_tsc_generation;
 438         bool tsc_catchup;
 439         bool tsc_always_catchup;
 440         s8 virtual_tsc_shift;
 441         u32 virtual_tsc_mult;
 442         u32 virtual_tsc_khz;
 443
 444         atomic_t nmi_queued;  /* unprocessed asynchronous NMIs */
 445         unsigned nmi_pending; /* NMI queued after currently running handler */
 446         bool nmi_injected;    /* Trying to inject an NMI this entry */
 447
 448         struct mtrr_state_type mtrr_state;
 449         u32 pat;
 450
 451         int switch_db_regs;
 452         unsigned long db[KVM_NR_DB_REGS];
 453         unsigned long dr6;
 454         unsigned long dr7;
 455         unsigned long eff_db[KVM_NR_DB_REGS];
 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 struct kvm_lpage_info {
 502         unsigned long rmap_pde;
 503         int write_count;
 504 };
 505
 506 struct kvm_arch_memory_slot {
 507         struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
 508 };
 509
 510 struct kvm_arch {
 511         unsigned int n_used_mmu_pages;
 512         unsigned int n_requested_mmu_pages;
 513         unsigned int n_max_mmu_pages;
 514         unsigned int indirect_shadow_pages;
 515         struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
 516         /*
 517          * Hash table of struct kvm_mmu_page.
 518          */
 519         struct list_head active_mmu_pages;
 520         struct list_head assigned_dev_head;
 521         struct iommu_domain *iommu_domain;
 522         int iommu_flags;
 523         struct kvm_pic *vpic;
 524         struct kvm_ioapic *vioapic;
 525         struct kvm_pit *vpit;
 526         int vapics_in_nmi_mode;
 527
 528         unsigned int tss_addr;
 529         struct page *apic_access_page;
 530
 531         gpa_t wall_clock;
 532
 533         struct page *ept_identity_pagetable;
 534         bool ept_identity_pagetable_done;
 535         gpa_t ept_identity_map_addr;
 536
 537         unsigned long irq_sources_bitmap;
 538         s64 kvmclock_offset;
 539         raw_spinlock_t tsc_write_lock;
 540         u64 last_tsc_nsec;
 541         u64 last_tsc_write;
 542         u32 last_tsc_khz;
 543         u64 cur_tsc_nsec;
 544         u64 cur_tsc_write;
 545         u64 cur_tsc_offset;
 546         u8  cur_tsc_generation;
 547
 548         struct kvm_xen_hvm_config xen_hvm_config;
 549
 550         /* fields used by HYPER-V emulation */
 551         u64 hv_guest_os_id;
 552         u64 hv_hypercall;
 553
 554         #ifdef CONFIG_KVM_MMU_AUDIT
 555         int audit_point;
 556         #endif
 557 };
 558
 559 struct kvm_vm_stat {
 560         u32 mmu_shadow_zapped;
 561         u32 mmu_pte_write;
 562         u32 mmu_pte_updated;
 563         u32 mmu_pde_zapped;
 564         u32 mmu_flooded;
 565         u32 mmu_recycled;
 566         u32 mmu_cache_miss;
 567         u32 mmu_unsync;
 568         u32 remote_tlb_flush;
 569         u32 lpages;
 570 };
 571
 572 struct kvm_vcpu_stat {
 573         u32 pf_fixed;
 574         u32 pf_guest;
 575         u32 tlb_flush;
 576         u32 invlpg;
 577
 578         u32 exits;
 579         u32 io_exits;
 580         u32 mmio_exits;
 581         u32 signal_exits;
 582         u32 irq_window_exits;
 583         u32 nmi_window_exits;
 584         u32 halt_exits;
 585         u32 halt_wakeup;
 586         u32 request_irq_exits;
 587         u32 irq_exits;
 588         u32 host_state_reload;
 589         u32 efer_reload;
 590         u32 fpu_reload;
 591         u32 insn_emulation;
 592         u32 insn_emulation_fail;
 593         u32 hypercalls;
 594         u32 irq_injections;
 595         u32 nmi_injections;
 596 };
 597
 598 struct x86_instruction_info;
 599
 600 struct kvm_x86_ops {
 601         int (*cpu_has_kvm_support)(void);          /* __init */
 602         int (*disabled_by_bios)(void);             /* __init */
 603         int (*hardware_enable)(void *dummy);
 604         void (*hardware_disable)(void *dummy);
 605         void (*check_processor_compatibility)(void *rtn);
 606         int (*hardware_setup)(void);               /* __init */
 607         void (*hardware_unsetup)(void);            /* __exit */
 608         bool (*cpu_has_accelerated_tpr)(void);
 609         void (*cpuid_update)(struct kvm_vcpu *vcpu);
 610
 611         /* Create, but do not attach this VCPU */
 612         struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
 613         void (*vcpu_free)(struct kvm_vcpu *vcpu);
 614         int (*vcpu_reset)(struct kvm_vcpu *vcpu);
 615
 616         void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
 617         void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
 618         void (*vcpu_put)(struct kvm_vcpu *vcpu);
 619
 620         void (*set_guest_debug)(struct kvm_vcpu *vcpu,
 621                                 struct kvm_guest_debug *dbg);
 622         int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
 623         int (*set_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
 624         u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
 625         void (*get_segment)(struct kvm_vcpu *vcpu,
 626                             struct kvm_segment *var, int seg);
 627         int (*get_cpl)(struct kvm_vcpu *vcpu);
 628         void (*set_segment)(struct kvm_vcpu *vcpu,
 629                             struct kvm_segment *var, int seg);
 630         void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
 631         void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
 632         void (*decache_cr3)(struct kvm_vcpu *vcpu);
 633         void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
 634         void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
 635         void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
 636         int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
 637         void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
 638         void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
 639         void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
 640         void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
 641         void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
 642         void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
 643         void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
 644         unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
 645         void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
 646         void (*fpu_activate)(struct kvm_vcpu *vcpu);
 647         void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
 648
 649         void (*tlb_flush)(struct kvm_vcpu *vcpu);
 650
 651         void (*run)(struct kvm_vcpu *vcpu);
 652         int (*handle_exit)(struct kvm_vcpu *vcpu);
 653         void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
 654         void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
 655         u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
 656         void (*patch_hypercall)(struct kvm_vcpu *vcpu,
 657                                 unsigned char *hypercall_addr);
 658         void (*set_irq)(struct kvm_vcpu *vcpu);
 659         void (*set_nmi)(struct kvm_vcpu *vcpu);
 660         void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
 661                                 bool has_error_code, u32 error_code,
 662                                 bool reinject);
 663         void (*cancel_injection)(struct kvm_vcpu *vcpu);
 664         int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
 665         int (*nmi_allowed)(struct kvm_vcpu *vcpu);
 666         bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
 667         void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
 668         void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
 669         void (*enable_irq_window)(struct kvm_vcpu *vcpu);
 670         void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
 671         int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
 672         int (*get_tdp_level)(void);
 673         u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
 674         int (*get_lpage_level)(void);
 675         bool (*rdtscp_supported)(void);
 676         void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment, bool host);
 677
 678         void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
 679
 680         void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
 681
 682         bool (*has_wbinvd_exit)(void);
 683
 684         void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale);
 685         void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
 686
 687         u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc);
 688         u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu);
 689
 690         void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
 691
 692         int (*check_intercept)(struct kvm_vcpu *vcpu,
 693                                struct x86_instruction_info *info,
 694                                enum x86_intercept_stage stage);
 695 };
 696
 697 struct kvm_arch_async_pf {
 698         u32 token;
 699         gfn_t gfn;
 700         unsigned long cr3;
 701         bool direct_map;
 702 };
 703
 704 extern struct kvm_x86_ops *kvm_x86_ops;
 705
 706 static inline void adjust_tsc_offset_guest(struct kvm_vcpu *vcpu,
 707                                            s64 adjustment)
 708 {
 709         kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, false);
 710 }
 711
 712 static inline void adjust_tsc_offset_host(struct kvm_vcpu *vcpu, s64 adjustment)
 713 {
 714         kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, true);
 715 }
 716
 717 int kvm_mmu_module_init(void);
 718 void kvm_mmu_module_exit(void);
 719
 720 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
 721 int kvm_mmu_create(struct kvm_vcpu *vcpu);
 722 int kvm_mmu_setup(struct kvm_vcpu *vcpu);
 723 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
 724                 u64 dirty_mask, u64 nx_mask, u64 x_mask);
 725
 726 int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
 727 void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
 728 void kvm_mmu_write_protect_pt_masked(struct kvm *kvm,
 729                                      struct kvm_memory_slot *slot,
 730                                      gfn_t gfn_offset, unsigned long mask);
 731 void kvm_mmu_zap_all(struct kvm *kvm);
 732 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
 733 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
 734
 735 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
 736
 737 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
 738                           const void *val, int bytes);
 739 u8 kvm_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn);
 740
 741 extern bool tdp_enabled;
 742
 743 u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
 744
 745 /* control of guest tsc rate supported? */
 746 extern bool kvm_has_tsc_control;
 747 /* minimum supported tsc_khz for guests */
 748 extern u32  kvm_min_guest_tsc_khz;
 749 /* maximum supported tsc_khz for guests */
 750 extern u32  kvm_max_guest_tsc_khz;
 751
 752 enum emulation_result {
 753         EMULATE_DONE,       /* no further processing */
 754         EMULATE_DO_MMIO,      /* kvm_run filled with mmio request */
 755         EMULATE_FAIL,         /* can't emulate this instruction */
 756 };
 757
 758 #define EMULTYPE_NO_DECODE          (1 << 0)
 759 #define EMULTYPE_TRAP_UD            (1 << 1)
 760 #define EMULTYPE_SKIP               (1 << 2)
 761 #define EMULTYPE_RETRY              (1 << 3)
 762 int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
 763                             int emulation_type, void *insn, int insn_len);
 764
 765 static inline int emulate_instruction(struct kvm_vcpu *vcpu,
 766                         int emulation_type)
 767 {
 768         return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
 769 }
 770
 771 void kvm_enable_efer_bits(u64);
 772 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
 773 int kvm_set_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
 774
 775 struct x86_emulate_ctxt;
 776
 777 int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
 778 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
 779 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
 780 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
 781
 782 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
 783 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
 784
 785 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
 786                     int reason, bool has_error_code, u32 error_code);
 787
 788 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
 789 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
 790 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
 791 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
 792 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
 793 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
 794 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
 795 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
 796 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
 797 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
 798
 799 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
 800 int kvm_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data);
 801
 802 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
 803 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
 804 bool kvm_rdpmc(struct kvm_vcpu *vcpu);
 805
 806 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
 807 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
 808 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
 809 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
 810 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
 811 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
 812                             gfn_t gfn, void *data, int offset, int len,
 813                             u32 access);
 814 void kvm_propagate_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
 815 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
 816
 817 int kvm_pic_set_irq(void *opaque, int irq, int level);
 818
 819 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
 820
 821 int fx_init(struct kvm_vcpu *vcpu);
 822
 823 void kvm_mmu_flush_tlb(struct kvm_vcpu *vcpu);
 824 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
 825                        const u8 *new, int bytes);
 826 int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
 827 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
 828 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
 829 int kvm_mmu_load(struct kvm_vcpu *vcpu);
 830 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
 831 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
 832 gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access);
 833 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
 834                               struct x86_exception *exception);
 835 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
 836                                struct x86_exception *exception);
 837 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
 838                                struct x86_exception *exception);
 839 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
 840                                 struct x86_exception *exception);
 841
 842 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
 843
 844 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
 845                        void *insn, int insn_len);
 846 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
 847
 848 void kvm_enable_tdp(void);
 849 void kvm_disable_tdp(void);
 850
 851 int complete_pio(struct kvm_vcpu *vcpu);
 852 bool kvm_check_iopl(struct kvm_vcpu *vcpu);
 853
 854 static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access)
 855 {
 856         return gpa;
 857 }
 858
 859 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
 860 {
 861         struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
 862
 863         return (struct kvm_mmu_page *)page_private(page);
 864 }
 865
 866 static inline u16 kvm_read_ldt(void)
 867 {
 868         u16 ldt;
 869         asm("sldt %0" : "=g"(ldt));
 870         return ldt;
 871 }
 872
 873 static inline void kvm_load_ldt(u16 sel)
 874 {
 875         asm("lldt %0" : : "rm"(sel));
 876 }
 877
 878 #ifdef CONFIG_X86_64
 879 static inline unsigned long read_msr(unsigned long msr)
 880 {
 881         u64 value;
 882
 883         rdmsrl(msr, value);
 884         return value;
 885 }
 886 #endif
 887
 888 static inline u32 get_rdx_init_val(void)
 889 {
 890         return 0x600; /* P6 family */
 891 }
 892
 893 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
 894 {
 895         kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
 896 }
 897
 898 #define TSS_IOPB_BASE_OFFSET 0x66
 899 #define TSS_BASE_SIZE 0x68
 900 #define TSS_IOPB_SIZE (65536 / 8)
 901 #define TSS_REDIRECTION_SIZE (256 / 8)
 902 #define RMODE_TSS_SIZE                                                  \
 903         (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
 904
 905 enum {
 906         TASK_SWITCH_CALL = 0,
 907         TASK_SWITCH_IRET = 1,
 908         TASK_SWITCH_JMP = 2,
 909         TASK_SWITCH_GATE = 3,
 910 };
 911
 912 #define HF_GIF_MASK             (1 << 0)
 913 #define HF_HIF_MASK             (1 << 1)
 914 #define HF_VINTR_MASK           (1 << 2)
 915 #define HF_NMI_MASK             (1 << 3)
 916 #define HF_IRET_MASK            (1 << 4)
 917 #define HF_GUEST_MASK           (1 << 5) /* VCPU is in guest-mode */
 918
 919 /*
 920  * Hardware virtualization extension instructions may fault if a
 921  * reboot turns off virtualization while processes are running.
 922  * Trap the fault and ignore the instruction if that happens.
 923  */
 924 asmlinkage void kvm_spurious_fault(void);
 925 extern bool kvm_rebooting;
 926
 927 #define ____kvm_handle_fault_on_reboot(insn, cleanup_insn)      \
 928         "666: " insn "\n\t" \
 929         "668: \n\t"                           \
 930         ".pushsection .fixup, \"ax\" \n" \
 931         "667: \n\t" \
 932         cleanup_insn "\n\t"                   \
 933         "cmpb $0, kvm_rebooting \n\t"         \
 934         "jne 668b \n\t"                       \
 935         __ASM_SIZE(push) " $666b \n\t"        \
 936         "call kvm_spurious_fault \n\t"        \
 937         ".popsection \n\t" \
 938         _ASM_EXTABLE(666b, 667b)
 939
 940 #define __kvm_handle_fault_on_reboot(insn)              \
 941         ____kvm_handle_fault_on_reboot(insn, "")
 942
 943 #define KVM_ARCH_WANT_MMU_NOTIFIER
 944 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
 945 int kvm_age_hva(struct kvm *kvm, unsigned long hva);
 946 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
 947 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
 948 int cpuid_maxphyaddr(struct kvm_vcpu *vcpu);
 949 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
 950 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
 951 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
 952
 953 void kvm_define_shared_msr(unsigned index, u32 msr);
 954 void kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
 955
 956 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
 957
 958 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
 959                                      struct kvm_async_pf *work);
 960 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
 961                                  struct kvm_async_pf *work);
 962 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
 963                                struct kvm_async_pf *work);
 964 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
 965 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
 966
 967 void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
 968
 969 int kvm_is_in_guest(void);
 970
 971 void kvm_pmu_init(struct kvm_vcpu *vcpu);
 972 void kvm_pmu_destroy(struct kvm_vcpu *vcpu);
 973 void kvm_pmu_reset(struct kvm_vcpu *vcpu);
 974 void kvm_pmu_cpuid_update(struct kvm_vcpu *vcpu);
 975 bool kvm_pmu_msr(struct kvm_vcpu *vcpu, u32 msr);
 976 int kvm_pmu_get_msr(struct kvm_vcpu *vcpu, u32 msr, u64 *data);
 977 int kvm_pmu_set_msr(struct kvm_vcpu *vcpu, u32 msr, u64 data);
 978 int kvm_pmu_read_pmc(struct kvm_vcpu *vcpu, unsigned pmc, u64 *data);
 979 void kvm_handle_pmu_event(struct kvm_vcpu *vcpu);
 980 void kvm_deliver_pmi(struct kvm_vcpu *vcpu);
 981
 982 #endif /* _ASM_X86_KVM_HOST_H */