drivers/kvm/kvm.h

   1 #ifndef __KVM_H
   2 #define __KVM_H
   3
   4 /*
   5  * This work is licensed under the terms of the GNU GPL, version 2.  See
   6  * the COPYING file in the top-level directory.
   7  */
   8
   9 #include <linux/types.h>
  10 #include <linux/list.h>
  11 #include <linux/mutex.h>
  12 #include <linux/spinlock.h>
  13 #include <linux/signal.h>
  14 #include <linux/sched.h>
  15 #include <linux/mm.h>
  16 #include <asm/signal.h>
  17
  18 #include "vmx.h"
  19 #include <linux/kvm.h>
  20 #include <linux/kvm_para.h>
  21
  22 #define CR3_WPT_MASK (1ULL << 3)
  23 #define CR3_PCD_MASK (1ULL << 4)
  24
  25 #define CR3_RESEVED_BITS 0x07ULL
  26 #define CR3_L_MODE_RESEVED_BITS (~((1ULL << 40) - 1) | 0x0fe7ULL)
  27 #define CR3_FLAGS_MASK ((1ULL << 5) - 1)
  28
  29 #define CR4_VME_MASK (1ULL << 0)
  30 #define CR4_PSE_MASK (1ULL << 4)
  31 #define CR4_PAE_MASK (1ULL << 5)
  32 #define CR4_PGE_MASK (1ULL << 7)
  33 #define CR4_VMXE_MASK (1ULL << 13)
  34
  35 #define KVM_GUEST_CR0_MASK \
  36         (X86_CR0_PG | X86_CR0_PE | X86_CR0_WP | X86_CR0_NE \
  37          | X86_CR0_NW | X86_CR0_CD)
  38 #define KVM_VM_CR0_ALWAYS_ON \
  39         (X86_CR0_PG | X86_CR0_PE | X86_CR0_WP | X86_CR0_NE | X86_CR0_TS \
  40          | X86_CR0_MP)
  41 #define KVM_GUEST_CR4_MASK \
  42         (CR4_PSE_MASK | CR4_PAE_MASK | CR4_PGE_MASK | CR4_VMXE_MASK | CR4_VME_MASK)
  43 #define KVM_PMODE_VM_CR4_ALWAYS_ON (CR4_VMXE_MASK | CR4_PAE_MASK)
  44 #define KVM_RMODE_VM_CR4_ALWAYS_ON (CR4_VMXE_MASK | CR4_PAE_MASK | CR4_VME_MASK)
  45
  46 #define INVALID_PAGE (~(hpa_t)0)
  47 #define UNMAPPED_GVA (~(gpa_t)0)
  48
  49 #define KVM_MAX_VCPUS 4
  50 #define KVM_ALIAS_SLOTS 4
  51 #define KVM_MEMORY_SLOTS 4
  52 #define KVM_NUM_MMU_PAGES 1024
  53 #define KVM_MIN_FREE_MMU_PAGES 5
  54 #define KVM_REFILL_PAGES 25
  55 #define KVM_MAX_CPUID_ENTRIES 40
  56
  57 #define FX_IMAGE_SIZE 512
  58 #define FX_IMAGE_ALIGN 16
  59 #define FX_BUF_SIZE (2 * FX_IMAGE_SIZE + FX_IMAGE_ALIGN)
  60
  61 #define DE_VECTOR 0
  62 #define NM_VECTOR 7
  63 #define DF_VECTOR 8
  64 #define TS_VECTOR 10
  65 #define NP_VECTOR 11
  66 #define SS_VECTOR 12
  67 #define GP_VECTOR 13
  68 #define PF_VECTOR 14
  69
  70 #define SELECTOR_TI_MASK (1 << 2)
  71 #define SELECTOR_RPL_MASK 0x03
  72
  73 #define IOPL_SHIFT 12
  74
  75 #define KVM_PIO_PAGE_OFFSET 1
  76
  77 /*
  78  * vcpu->requests bit members
  79  */
  80 #define KVM_TLB_FLUSH 0
  81
  82 /*
  83  * Address types:
  84  *
  85  *  gva - guest virtual address
  86  *  gpa - guest physical address
  87  *  gfn - guest frame number
  88  *  hva - host virtual address
  89  *  hpa - host physical address
  90  *  hfn - host frame number
  91  */
  92
  93 typedef unsigned long  gva_t;
  94 typedef u64            gpa_t;
  95 typedef unsigned long  gfn_t;
  96
  97 typedef unsigned long  hva_t;
  98 typedef u64            hpa_t;
  99 typedef unsigned long  hfn_t;
 100
 101 #define NR_PTE_CHAIN_ENTRIES 5
 102
 103 struct kvm_pte_chain {
 104         u64 *parent_ptes[NR_PTE_CHAIN_ENTRIES];
 105         struct hlist_node link;
 106 };
 107
 108 /*
 109  * kvm_mmu_page_role, below, is defined as:
 110  *
 111  *   bits 0:3 - total guest paging levels (2-4, or zero for real mode)
 112  *   bits 4:7 - page table level for this shadow (1-4)
 113  *   bits 8:9 - page table quadrant for 2-level guests
 114  *   bit   16 - "metaphysical" - gfn is not a real page (huge page/real mode)
 115  *   bits 17:19 - "access" - the user, writable, and nx bits of a huge page pde
 116  */
 117 union kvm_mmu_page_role {
 118         unsigned word;
 119         struct {
 120                 unsigned glevels : 4;
 121                 unsigned level : 4;
 122                 unsigned quadrant : 2;
 123                 unsigned pad_for_nice_hex_output : 6;
 124                 unsigned metaphysical : 1;
 125                 unsigned hugepage_access : 3;
 126         };
 127 };
 128
 129 struct kvm_mmu_page {
 130         struct list_head link;
 131         struct hlist_node hash_link;
 132
 133         /*
 134          * The following two entries are used to key the shadow page in the
 135          * hash table.
 136          */
 137         gfn_t gfn;
 138         union kvm_mmu_page_role role;
 139
 140         u64 *spt;
 141         unsigned long slot_bitmap; /* One bit set per slot which has memory
 142                                     * in this shadow page.
 143                                     */
 144         int multimapped;         /* More than one parent_pte? */
 145         int root_count;          /* Currently serving as active root */
 146         union {
 147                 u64 *parent_pte;               /* !multimapped */
 148                 struct hlist_head parent_ptes; /* multimapped, kvm_pte_chain */
 149         };
 150 };
 151
 152 struct vmcs {
 153         u32 revision_id;
 154         u32 abort;
 155         char data[0];
 156 };
 157
 158 #define vmx_msr_entry kvm_msr_entry
 159
 160 struct kvm_vcpu;
 161
 162 /*
 163  * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
 164  * 32-bit).  The kvm_mmu structure abstracts the details of the current mmu
 165  * mode.
 166  */
 167 struct kvm_mmu {
 168         void (*new_cr3)(struct kvm_vcpu *vcpu);
 169         int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err);
 170         void (*free)(struct kvm_vcpu *vcpu);
 171         gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva);
 172         hpa_t root_hpa;
 173         int root_level;
 174         int shadow_root_level;
 175
 176         u64 *pae_root;
 177 };
 178
 179 #define KVM_NR_MEM_OBJS 20
 180
 181 struct kvm_mmu_memory_cache {
 182         int nobjs;
 183         void *objects[KVM_NR_MEM_OBJS];
 184 };
 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_guest_debug {
 191         int enabled;
 192         unsigned long bp[4];
 193         int singlestep;
 194 };
 195
 196 enum {
 197         VCPU_REGS_RAX = 0,
 198         VCPU_REGS_RCX = 1,
 199         VCPU_REGS_RDX = 2,
 200         VCPU_REGS_RBX = 3,
 201         VCPU_REGS_RSP = 4,
 202         VCPU_REGS_RBP = 5,
 203         VCPU_REGS_RSI = 6,
 204         VCPU_REGS_RDI = 7,
 205 #ifdef CONFIG_X86_64
 206         VCPU_REGS_R8 = 8,
 207         VCPU_REGS_R9 = 9,
 208         VCPU_REGS_R10 = 10,
 209         VCPU_REGS_R11 = 11,
 210         VCPU_REGS_R12 = 12,
 211         VCPU_REGS_R13 = 13,
 212         VCPU_REGS_R14 = 14,
 213         VCPU_REGS_R15 = 15,
 214 #endif
 215         NR_VCPU_REGS
 216 };
 217
 218 enum {
 219         VCPU_SREG_CS,
 220         VCPU_SREG_DS,
 221         VCPU_SREG_ES,
 222         VCPU_SREG_FS,
 223         VCPU_SREG_GS,
 224         VCPU_SREG_SS,
 225         VCPU_SREG_TR,
 226         VCPU_SREG_LDTR,
 227 };
 228
 229 struct kvm_pio_request {
 230         unsigned long count;
 231         int cur_count;
 232         struct page *guest_pages[2];
 233         unsigned guest_page_offset;
 234         int in;
 235         int port;
 236         int size;
 237         int string;
 238         int down;
 239         int rep;
 240 };
 241
 242 struct kvm_stat {
 243         u32 pf_fixed;
 244         u32 pf_guest;
 245         u32 tlb_flush;
 246         u32 invlpg;
 247
 248         u32 exits;
 249         u32 io_exits;
 250         u32 mmio_exits;
 251         u32 signal_exits;
 252         u32 irq_window_exits;
 253         u32 halt_exits;
 254         u32 request_irq_exits;
 255         u32 irq_exits;
 256         u32 light_exits;
 257         u32 efer_reload;
 258 };
 259
 260 struct kvm_io_device {
 261         void (*read)(struct kvm_io_device *this,
 262                      gpa_t addr,
 263                      int len,
 264                      void *val);
 265         void (*write)(struct kvm_io_device *this,
 266                       gpa_t addr,
 267                       int len,
 268                       const void *val);
 269         int (*in_range)(struct kvm_io_device *this, gpa_t addr);
 270         void (*destructor)(struct kvm_io_device *this);
 271
 272         void             *private;
 273 };
 274
 275 static inline void kvm_iodevice_read(struct kvm_io_device *dev,
 276                                      gpa_t addr,
 277                                      int len,
 278                                      void *val)
 279 {
 280         dev->read(dev, addr, len, val);
 281 }
 282
 283 static inline void kvm_iodevice_write(struct kvm_io_device *dev,
 284                                       gpa_t addr,
 285                                       int len,
 286                                       const void *val)
 287 {
 288         dev->write(dev, addr, len, val);
 289 }
 290
 291 static inline int kvm_iodevice_inrange(struct kvm_io_device *dev, gpa_t addr)
 292 {
 293         return dev->in_range(dev, addr);
 294 }
 295
 296 static inline void kvm_iodevice_destructor(struct kvm_io_device *dev)
 297 {
 298         if (dev->destructor)
 299                 dev->destructor(dev);
 300 }
 301
 302 /*
 303  * It would be nice to use something smarter than a linear search, TBD...
 304  * Thankfully we dont expect many devices to register (famous last words :),
 305  * so until then it will suffice.  At least its abstracted so we can change
 306  * in one place.
 307  */
 308 struct kvm_io_bus {
 309         int                   dev_count;
 310 #define NR_IOBUS_DEVS 6
 311         struct kvm_io_device *devs[NR_IOBUS_DEVS];
 312 };
 313
 314 void kvm_io_bus_init(struct kvm_io_bus *bus);
 315 void kvm_io_bus_destroy(struct kvm_io_bus *bus);
 316 struct kvm_io_device *kvm_io_bus_find_dev(struct kvm_io_bus *bus, gpa_t addr);
 317 void kvm_io_bus_register_dev(struct kvm_io_bus *bus,
 318                              struct kvm_io_device *dev);
 319
 320 struct kvm_vcpu {
 321         struct kvm *kvm;
 322         int vcpu_id;
 323         union {
 324                 struct vmcs *vmcs;
 325                 struct vcpu_svm *svm;
 326         };
 327         struct mutex mutex;
 328         int   cpu;
 329         int   launched;
 330         u64 host_tsc;
 331         struct kvm_run *run;
 332         int interrupt_window_open;
 333         int guest_mode;
 334         unsigned long requests;
 335         unsigned long irq_summary; /* bit vector: 1 per word in irq_pending */
 336 #define NR_IRQ_WORDS KVM_IRQ_BITMAP_SIZE(unsigned long)
 337         unsigned long irq_pending[NR_IRQ_WORDS];
 338         unsigned long regs[NR_VCPU_REGS]; /* for rsp: vcpu_load_rsp_rip() */
 339         unsigned long rip;      /* needs vcpu_load_rsp_rip() */
 340
 341         unsigned long cr0;
 342         unsigned long cr2;
 343         unsigned long cr3;
 344         gpa_t para_state_gpa;
 345         struct page *para_state_page;
 346         gpa_t hypercall_gpa;
 347         unsigned long cr4;
 348         unsigned long cr8;
 349         u64 pdptrs[4]; /* pae */
 350         u64 shadow_efer;
 351         u64 apic_base;
 352         u64 ia32_misc_enable_msr;
 353         int nmsrs;
 354         int save_nmsrs;
 355         int msr_offset_efer;
 356 #ifdef CONFIG_X86_64
 357         int msr_offset_kernel_gs_base;
 358 #endif
 359         struct vmx_msr_entry *guest_msrs;
 360         struct vmx_msr_entry *host_msrs;
 361
 362         struct kvm_mmu mmu;
 363
 364         struct kvm_mmu_memory_cache mmu_pte_chain_cache;
 365         struct kvm_mmu_memory_cache mmu_rmap_desc_cache;
 366         struct kvm_mmu_memory_cache mmu_page_cache;
 367         struct kvm_mmu_memory_cache mmu_page_header_cache;
 368
 369         gfn_t last_pt_write_gfn;
 370         int   last_pt_write_count;
 371
 372         struct kvm_guest_debug guest_debug;
 373
 374         char fx_buf[FX_BUF_SIZE];
 375         char *host_fx_image;
 376         char *guest_fx_image;
 377         int fpu_active;
 378         int guest_fpu_loaded;
 379         struct vmx_host_state {
 380                 int loaded;
 381                 u16 fs_sel, gs_sel, ldt_sel;
 382                 int fs_gs_ldt_reload_needed;
 383         } vmx_host_state;
 384
 385         int mmio_needed;
 386         int mmio_read_completed;
 387         int mmio_is_write;
 388         int mmio_size;
 389         unsigned char mmio_data[8];
 390         gpa_t mmio_phys_addr;
 391         gva_t mmio_fault_cr2;
 392         struct kvm_pio_request pio;
 393         void *pio_data;
 394
 395         int sigset_active;
 396         sigset_t sigset;
 397
 398         struct kvm_stat stat;
 399
 400         struct {
 401                 int active;
 402                 u8 save_iopl;
 403                 struct kvm_save_segment {
 404                         u16 selector;
 405                         unsigned long base;
 406                         u32 limit;
 407                         u32 ar;
 408                 } tr, es, ds, fs, gs;
 409         } rmode;
 410         int halt_request; /* real mode on Intel only */
 411
 412         int cpuid_nent;
 413         struct kvm_cpuid_entry cpuid_entries[KVM_MAX_CPUID_ENTRIES];
 414 };
 415
 416 struct kvm_mem_alias {
 417         gfn_t base_gfn;
 418         unsigned long npages;
 419         gfn_t target_gfn;
 420 };
 421
 422 struct kvm_memory_slot {
 423         gfn_t base_gfn;
 424         unsigned long npages;
 425         unsigned long flags;
 426         struct page **phys_mem;
 427         unsigned long *dirty_bitmap;
 428 };
 429
 430 struct kvm {
 431         spinlock_t lock; /* protects everything except vcpus */
 432         int naliases;
 433         struct kvm_mem_alias aliases[KVM_ALIAS_SLOTS];
 434         int nmemslots;
 435         struct kvm_memory_slot memslots[KVM_MEMORY_SLOTS];
 436         /*
 437          * Hash table of struct kvm_mmu_page.
 438          */
 439         struct list_head active_mmu_pages;
 440         int n_free_mmu_pages;
 441         struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
 442         int nvcpus;
 443         struct kvm_vcpu vcpus[KVM_MAX_VCPUS];
 444         int memory_config_version;
 445         int busy;
 446         unsigned long rmap_overflow;
 447         struct list_head vm_list;
 448         struct file *filp;
 449         struct kvm_io_bus mmio_bus;
 450         struct kvm_io_bus pio_bus;
 451 };
 452
 453 struct descriptor_table {
 454         u16 limit;
 455         unsigned long base;
 456 } __attribute__((packed));
 457
 458 struct kvm_arch_ops {
 459         int (*cpu_has_kvm_support)(void);          /* __init */
 460         int (*disabled_by_bios)(void);             /* __init */
 461         void (*hardware_enable)(void *dummy);      /* __init */
 462         void (*hardware_disable)(void *dummy);
 463         int (*hardware_setup)(void);               /* __init */
 464         void (*hardware_unsetup)(void);            /* __exit */
 465
 466         int (*vcpu_create)(struct kvm_vcpu *vcpu);
 467         void (*vcpu_free)(struct kvm_vcpu *vcpu);
 468
 469         void (*vcpu_load)(struct kvm_vcpu *vcpu);
 470         void (*vcpu_put)(struct kvm_vcpu *vcpu);
 471         void (*vcpu_decache)(struct kvm_vcpu *vcpu);
 472
 473         int (*set_guest_debug)(struct kvm_vcpu *vcpu,
 474                                struct kvm_debug_guest *dbg);
 475         int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
 476         int (*set_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
 477         u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
 478         void (*get_segment)(struct kvm_vcpu *vcpu,
 479                             struct kvm_segment *var, int seg);
 480         void (*set_segment)(struct kvm_vcpu *vcpu,
 481                             struct kvm_segment *var, int seg);
 482         void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
 483         void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
 484         void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
 485         void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
 486         void (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
 487         void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
 488         void (*get_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
 489         void (*set_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
 490         void (*get_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
 491         void (*set_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
 492         unsigned long (*get_dr)(struct kvm_vcpu *vcpu, int dr);
 493         void (*set_dr)(struct kvm_vcpu *vcpu, int dr, unsigned long value,
 494                        int *exception);
 495         void (*cache_regs)(struct kvm_vcpu *vcpu);
 496         void (*decache_regs)(struct kvm_vcpu *vcpu);
 497         unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
 498         void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
 499
 500         void (*invlpg)(struct kvm_vcpu *vcpu, gva_t addr);
 501         void (*tlb_flush)(struct kvm_vcpu *vcpu);
 502         void (*inject_page_fault)(struct kvm_vcpu *vcpu,
 503                                   unsigned long addr, u32 err_code);
 504
 505         void (*inject_gp)(struct kvm_vcpu *vcpu, unsigned err_code);
 506
 507         int (*run)(struct kvm_vcpu *vcpu, struct kvm_run *run);
 508         int (*vcpu_setup)(struct kvm_vcpu *vcpu);
 509         void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
 510         void (*patch_hypercall)(struct kvm_vcpu *vcpu,
 511                                 unsigned char *hypercall_addr);
 512 };
 513
 514 extern struct kvm_arch_ops *kvm_arch_ops;
 515
 516 #define kvm_printf(kvm, fmt ...) printk(KERN_DEBUG fmt)
 517 #define vcpu_printf(vcpu, fmt...) kvm_printf(vcpu->kvm, fmt)
 518
 519 int kvm_init_arch(struct kvm_arch_ops *ops, struct module *module);
 520 void kvm_exit_arch(void);
 521
 522 int kvm_mmu_module_init(void);
 523 void kvm_mmu_module_exit(void);
 524
 525 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
 526 int kvm_mmu_create(struct kvm_vcpu *vcpu);
 527 int kvm_mmu_setup(struct kvm_vcpu *vcpu);
 528
 529 int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
 530 void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
 531 void kvm_mmu_zap_all(struct kvm *kvm);
 532
 533 hpa_t gpa_to_hpa(struct kvm_vcpu *vcpu, gpa_t gpa);
 534 #define HPA_MSB ((sizeof(hpa_t) * 8) - 1)
 535 #define HPA_ERR_MASK ((hpa_t)1 << HPA_MSB)
 536 static inline int is_error_hpa(hpa_t hpa) { return hpa >> HPA_MSB; }
 537 hpa_t gva_to_hpa(struct kvm_vcpu *vcpu, gva_t gva);
 538 struct page *gva_to_page(struct kvm_vcpu *vcpu, gva_t gva);
 539
 540 void kvm_emulator_want_group7_invlpg(void);
 541
 542 extern hpa_t bad_page_address;
 543
 544 struct page *gfn_to_page(struct kvm *kvm, gfn_t gfn);
 545 struct kvm_memory_slot *gfn_to_memslot(struct kvm *kvm, gfn_t gfn);
 546 void mark_page_dirty(struct kvm *kvm, gfn_t gfn);
 547
 548 enum emulation_result {
 549         EMULATE_DONE,       /* no further processing */
 550         EMULATE_DO_MMIO,      /* kvm_run filled with mmio request */
 551         EMULATE_FAIL,         /* can't emulate this instruction */
 552 };
 553
 554 int emulate_instruction(struct kvm_vcpu *vcpu, struct kvm_run *run,
 555                         unsigned long cr2, u16 error_code);
 556 void realmode_lgdt(struct kvm_vcpu *vcpu, u16 size, unsigned long address);
 557 void realmode_lidt(struct kvm_vcpu *vcpu, u16 size, unsigned long address);
 558 void realmode_lmsw(struct kvm_vcpu *vcpu, unsigned long msw,
 559                    unsigned long *rflags);
 560
 561 unsigned long realmode_get_cr(struct kvm_vcpu *vcpu, int cr);
 562 void realmode_set_cr(struct kvm_vcpu *vcpu, int cr, unsigned long value,
 563                      unsigned long *rflags);
 564 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
 565 int kvm_set_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
 566
 567 struct x86_emulate_ctxt;
 568
 569 int kvm_setup_pio(struct kvm_vcpu *vcpu, struct kvm_run *run, int in,
 570                   int size, unsigned long count, int string, int down,
 571                   gva_t address, int rep, unsigned port);
 572 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
 573 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
 574 int emulate_invlpg(struct kvm_vcpu *vcpu, gva_t address);
 575 int emulate_clts(struct kvm_vcpu *vcpu);
 576 int emulator_get_dr(struct x86_emulate_ctxt* ctxt, int dr,
 577                     unsigned long *dest);
 578 int emulator_set_dr(struct x86_emulate_ctxt *ctxt, int dr,
 579                     unsigned long value);
 580
 581 void set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
 582 void set_cr3(struct kvm_vcpu *vcpu, unsigned long cr0);
 583 void set_cr4(struct kvm_vcpu *vcpu, unsigned long cr0);
 584 void set_cr8(struct kvm_vcpu *vcpu, unsigned long cr0);
 585 void lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
 586
 587 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
 588 int kvm_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data);
 589
 590 void fx_init(struct kvm_vcpu *vcpu);
 591
 592 void load_msrs(struct vmx_msr_entry *e, int n);
 593 void save_msrs(struct vmx_msr_entry *e, int n);
 594 void kvm_resched(struct kvm_vcpu *vcpu);
 595 void kvm_load_guest_fpu(struct kvm_vcpu *vcpu);
 596 void kvm_put_guest_fpu(struct kvm_vcpu *vcpu);
 597 void kvm_flush_remote_tlbs(struct kvm *kvm);
 598
 599 int kvm_read_guest(struct kvm_vcpu *vcpu,
 600                gva_t addr,
 601                unsigned long size,
 602                void *dest);
 603
 604 int kvm_write_guest(struct kvm_vcpu *vcpu,
 605                 gva_t addr,
 606                 unsigned long size,
 607                 void *data);
 608
 609 unsigned long segment_base(u16 selector);
 610
 611 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
 612                        const u8 *old, const u8 *new, int bytes);
 613 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
 614 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
 615 int kvm_mmu_load(struct kvm_vcpu *vcpu);
 616 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
 617
 618 int kvm_hypercall(struct kvm_vcpu *vcpu, struct kvm_run *run);
 619
 620 static inline int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva,
 621                                      u32 error_code)
 622 {
 623         return vcpu->mmu.page_fault(vcpu, gva, error_code);
 624 }
 625
 626 static inline void kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu)
 627 {
 628         if (unlikely(vcpu->kvm->n_free_mmu_pages < KVM_MIN_FREE_MMU_PAGES))
 629                 __kvm_mmu_free_some_pages(vcpu);
 630 }
 631
 632 static inline int kvm_mmu_reload(struct kvm_vcpu *vcpu)
 633 {
 634         if (likely(vcpu->mmu.root_hpa != INVALID_PAGE))
 635                 return 0;
 636
 637         return kvm_mmu_load(vcpu);
 638 }
 639
 640 static inline int is_long_mode(struct kvm_vcpu *vcpu)
 641 {
 642 #ifdef CONFIG_X86_64
 643         return vcpu->shadow_efer & EFER_LME;
 644 #else
 645         return 0;
 646 #endif
 647 }
 648
 649 static inline int is_pae(struct kvm_vcpu *vcpu)
 650 {
 651         return vcpu->cr4 & CR4_PAE_MASK;
 652 }
 653
 654 static inline int is_pse(struct kvm_vcpu *vcpu)
 655 {
 656         return vcpu->cr4 & CR4_PSE_MASK;
 657 }
 658
 659 static inline int is_paging(struct kvm_vcpu *vcpu)
 660 {
 661         return vcpu->cr0 & X86_CR0_PG;
 662 }
 663
 664 static inline int memslot_id(struct kvm *kvm, struct kvm_memory_slot *slot)
 665 {
 666         return slot - kvm->memslots;
 667 }
 668
 669 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
 670 {
 671         struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
 672
 673         return (struct kvm_mmu_page *)page_private(page);
 674 }
 675
 676 static inline u16 read_fs(void)
 677 {
 678         u16 seg;
 679         asm ("mov %%fs, %0" : "=g"(seg));
 680         return seg;
 681 }
 682
 683 static inline u16 read_gs(void)
 684 {
 685         u16 seg;
 686         asm ("mov %%gs, %0" : "=g"(seg));
 687         return seg;
 688 }
 689
 690 static inline u16 read_ldt(void)
 691 {
 692         u16 ldt;
 693         asm ("sldt %0" : "=g"(ldt));
 694         return ldt;
 695 }
 696
 697 static inline void load_fs(u16 sel)
 698 {
 699         asm ("mov %0, %%fs" : : "rm"(sel));
 700 }
 701
 702 static inline void load_gs(u16 sel)
 703 {
 704         asm ("mov %0, %%gs" : : "rm"(sel));
 705 }
 706
 707 #ifndef load_ldt
 708 static inline void load_ldt(u16 sel)
 709 {
 710         asm ("lldt %0" : : "rm"(sel));
 711 }
 712 #endif
 713
 714 static inline void get_idt(struct descriptor_table *table)
 715 {
 716         asm ("sidt %0" : "=m"(*table));
 717 }
 718
 719 static inline void get_gdt(struct descriptor_table *table)
 720 {
 721         asm ("sgdt %0" : "=m"(*table));
 722 }
 723
 724 static inline unsigned long read_tr_base(void)
 725 {
 726         u16 tr;
 727         asm ("str %0" : "=g"(tr));
 728         return segment_base(tr);
 729 }
 730
 731 #ifdef CONFIG_X86_64
 732 static inline unsigned long read_msr(unsigned long msr)
 733 {
 734         u64 value;
 735
 736         rdmsrl(msr, value);
 737         return value;
 738 }
 739 #endif
 740
 741 static inline void fx_save(void *image)
 742 {
 743         asm ("fxsave (%0)":: "r" (image));
 744 }
 745
 746 static inline void fx_restore(void *image)
 747 {
 748         asm ("fxrstor (%0)":: "r" (image));
 749 }
 750
 751 static inline void fpu_init(void)
 752 {
 753         asm ("finit");
 754 }
 755
 756 static inline u32 get_rdx_init_val(void)
 757 {
 758         return 0x600; /* P6 family */
 759 }
 760
 761 #define ASM_VMX_VMCLEAR_RAX       ".byte 0x66, 0x0f, 0xc7, 0x30"
 762 #define ASM_VMX_VMLAUNCH          ".byte 0x0f, 0x01, 0xc2"
 763 #define ASM_VMX_VMRESUME          ".byte 0x0f, 0x01, 0xc3"
 764 #define ASM_VMX_VMPTRLD_RAX       ".byte 0x0f, 0xc7, 0x30"
 765 #define ASM_VMX_VMREAD_RDX_RAX    ".byte 0x0f, 0x78, 0xd0"
 766 #define ASM_VMX_VMWRITE_RAX_RDX   ".byte 0x0f, 0x79, 0xd0"
 767 #define ASM_VMX_VMWRITE_RSP_RDX   ".byte 0x0f, 0x79, 0xd4"
 768 #define ASM_VMX_VMXOFF            ".byte 0x0f, 0x01, 0xc4"
 769 #define ASM_VMX_VMXON_RAX         ".byte 0xf3, 0x0f, 0xc7, 0x30"
 770
 771 #define MSR_IA32_TIME_STAMP_COUNTER             0x010
 772
 773 #define TSS_IOPB_BASE_OFFSET 0x66
 774 #define TSS_BASE_SIZE 0x68
 775 #define TSS_IOPB_SIZE (65536 / 8)
 776 #define TSS_REDIRECTION_SIZE (256 / 8)
 777 #define RMODE_TSS_SIZE (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
 778
 779 #endif