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