kvm_host.h 26 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943
  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. #ifndef _ASM_X86_KVM_HOST_H
  11. #define _ASM_X86_KVM_HOST_H
  12. #include <linux/types.h>
  13. #include <linux/mm.h>
  14. #include <linux/mmu_notifier.h>
  15. #include <linux/tracepoint.h>
  16. #include <linux/cpumask.h>
  17. #include <linux/irq_work.h>
  18. #include <linux/kvm.h>
  19. #include <linux/kvm_para.h>
  20. #include <linux/kvm_types.h>
  21. #include <linux/perf_event.h>
  22. #include <asm/pvclock-abi.h>
  23. #include <asm/desc.h>
  24. #include <asm/mtrr.h>
  25. #include <asm/msr-index.h>
  26. #define KVM_MAX_VCPUS 254
  27. #define KVM_SOFT_MAX_VCPUS 160
  28. #define KVM_MEMORY_SLOTS 32
  29. /* memory slots that does not exposed to userspace */
  30. #define KVM_PRIVATE_MEM_SLOTS 4
  31. #define KVM_MEM_SLOTS_NUM (KVM_MEMORY_SLOTS + KVM_PRIVATE_MEM_SLOTS)
  32. #define KVM_MMIO_SIZE 16
  33. #define KVM_PIO_PAGE_OFFSET 1
  34. #define KVM_COALESCED_MMIO_PAGE_OFFSET 2
  35. #define CR0_RESERVED_BITS \
  36. (~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
  37. | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
  38. | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
  39. #define CR3_PAE_RESERVED_BITS ((X86_CR3_PWT | X86_CR3_PCD) - 1)
  40. #define CR3_NONPAE_RESERVED_BITS ((PAGE_SIZE-1) & ~(X86_CR3_PWT | X86_CR3_PCD))
  41. #define CR3_L_MODE_RESERVED_BITS (CR3_NONPAE_RESERVED_BITS | \
  42. 0xFFFFFF0000000000ULL)
  43. #define CR4_RESERVED_BITS \
  44. (~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
  45. | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE \
  46. | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR \
  47. | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_RDWRGSFS \
  48. | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE))
  49. #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
  50. #define INVALID_PAGE (~(hpa_t)0)
  51. #define VALID_PAGE(x) ((x) != INVALID_PAGE)
  52. #define UNMAPPED_GVA (~(gpa_t)0)
  53. /* KVM Hugepage definitions for x86 */
  54. #define KVM_NR_PAGE_SIZES 3
  55. #define KVM_HPAGE_GFN_SHIFT(x) (((x) - 1) * 9)
  56. #define KVM_HPAGE_SHIFT(x) (PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
  57. #define KVM_HPAGE_SIZE(x) (1UL << KVM_HPAGE_SHIFT(x))
  58. #define KVM_HPAGE_MASK(x) (~(KVM_HPAGE_SIZE(x) - 1))
  59. #define KVM_PAGES_PER_HPAGE(x) (KVM_HPAGE_SIZE(x) / PAGE_SIZE)
  60. #define DE_VECTOR 0
  61. #define DB_VECTOR 1
  62. #define BP_VECTOR 3
  63. #define OF_VECTOR 4
  64. #define BR_VECTOR 5
  65. #define UD_VECTOR 6
  66. #define NM_VECTOR 7
  67. #define DF_VECTOR 8
  68. #define TS_VECTOR 10
  69. #define NP_VECTOR 11
  70. #define SS_VECTOR 12
  71. #define GP_VECTOR 13
  72. #define PF_VECTOR 14
  73. #define MF_VECTOR 16
  74. #define MC_VECTOR 18
  75. #define SELECTOR_TI_MASK (1 << 2)
  76. #define SELECTOR_RPL_MASK 0x03
  77. #define IOPL_SHIFT 12
  78. #define KVM_PERMILLE_MMU_PAGES 20
  79. #define KVM_MIN_ALLOC_MMU_PAGES 64
  80. #define KVM_MMU_HASH_SHIFT 10
  81. #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
  82. #define KVM_MIN_FREE_MMU_PAGES 5
  83. #define KVM_REFILL_PAGES 25
  84. #define KVM_MAX_CPUID_ENTRIES 80
  85. #define KVM_NR_FIXED_MTRR_REGION 88
  86. #define KVM_NR_VAR_MTRR 8
  87. #define ASYNC_PF_PER_VCPU 64
  88. extern raw_spinlock_t kvm_lock;
  89. extern struct list_head vm_list;
  90. struct kvm_vcpu;
  91. struct kvm;
  92. struct kvm_async_pf;
  93. enum kvm_reg {
  94. VCPU_REGS_RAX = 0,
  95. VCPU_REGS_RCX = 1,
  96. VCPU_REGS_RDX = 2,
  97. VCPU_REGS_RBX = 3,
  98. VCPU_REGS_RSP = 4,
  99. VCPU_REGS_RBP = 5,
  100. VCPU_REGS_RSI = 6,
  101. VCPU_REGS_RDI = 7,
  102. #ifdef CONFIG_X86_64
  103. VCPU_REGS_R8 = 8,
  104. VCPU_REGS_R9 = 9,
  105. VCPU_REGS_R10 = 10,
  106. VCPU_REGS_R11 = 11,
  107. VCPU_REGS_R12 = 12,
  108. VCPU_REGS_R13 = 13,
  109. VCPU_REGS_R14 = 14,
  110. VCPU_REGS_R15 = 15,
  111. #endif
  112. VCPU_REGS_RIP,
  113. NR_VCPU_REGS
  114. };
  115. enum kvm_reg_ex {
  116. VCPU_EXREG_PDPTR = NR_VCPU_REGS,
  117. VCPU_EXREG_CR3,
  118. VCPU_EXREG_RFLAGS,
  119. VCPU_EXREG_CPL,
  120. VCPU_EXREG_SEGMENTS,
  121. };
  122. enum {
  123. VCPU_SREG_ES,
  124. VCPU_SREG_CS,
  125. VCPU_SREG_SS,
  126. VCPU_SREG_DS,
  127. VCPU_SREG_FS,
  128. VCPU_SREG_GS,
  129. VCPU_SREG_TR,
  130. VCPU_SREG_LDTR,
  131. };
  132. #include <asm/kvm_emulate.h>
  133. #define KVM_NR_MEM_OBJS 40
  134. #define KVM_NR_DB_REGS 4
  135. #define DR6_BD (1 << 13)
  136. #define DR6_BS (1 << 14)
  137. #define DR6_FIXED_1 0xffff0ff0
  138. #define DR6_VOLATILE 0x0000e00f
  139. #define DR7_BP_EN_MASK 0x000000ff
  140. #define DR7_GE (1 << 9)
  141. #define DR7_GD (1 << 13)
  142. #define DR7_FIXED_1 0x00000400
  143. #define DR7_VOLATILE 0xffff23ff
  144. /*
  145. * We don't want allocation failures within the mmu code, so we preallocate
  146. * enough memory for a single page fault in a cache.
  147. */
  148. struct kvm_mmu_memory_cache {
  149. int nobjs;
  150. void *objects[KVM_NR_MEM_OBJS];
  151. };
  152. /*
  153. * kvm_mmu_page_role, below, is defined as:
  154. *
  155. * bits 0:3 - total guest paging levels (2-4, or zero for real mode)
  156. * bits 4:7 - page table level for this shadow (1-4)
  157. * bits 8:9 - page table quadrant for 2-level guests
  158. * bit 16 - direct mapping of virtual to physical mapping at gfn
  159. * used for real mode and two-dimensional paging
  160. * bits 17:19 - common access permissions for all ptes in this shadow page
  161. */
  162. union kvm_mmu_page_role {
  163. unsigned word;
  164. struct {
  165. unsigned level:4;
  166. unsigned cr4_pae:1;
  167. unsigned quadrant:2;
  168. unsigned pad_for_nice_hex_output:6;
  169. unsigned direct:1;
  170. unsigned access:3;
  171. unsigned invalid:1;
  172. unsigned nxe:1;
  173. unsigned cr0_wp:1;
  174. unsigned smep_andnot_wp:1;
  175. };
  176. };
  177. struct kvm_mmu_page {
  178. struct list_head link;
  179. struct hlist_node hash_link;
  180. /*
  181. * The following two entries are used to key the shadow page in the
  182. * hash table.
  183. */
  184. gfn_t gfn;
  185. union kvm_mmu_page_role role;
  186. u64 *spt;
  187. /* hold the gfn of each spte inside spt */
  188. gfn_t *gfns;
  189. /*
  190. * One bit set per slot which has memory
  191. * in this shadow page.
  192. */
  193. DECLARE_BITMAP(slot_bitmap, KVM_MEM_SLOTS_NUM);
  194. bool unsync;
  195. int root_count; /* Currently serving as active root */
  196. unsigned int unsync_children;
  197. unsigned long parent_ptes; /* Reverse mapping for parent_pte */
  198. DECLARE_BITMAP(unsync_child_bitmap, 512);
  199. #ifdef CONFIG_X86_32
  200. int clear_spte_count;
  201. #endif
  202. int write_flooding_count;
  203. struct rcu_head rcu;
  204. };
  205. struct kvm_pio_request {
  206. unsigned long count;
  207. int in;
  208. int port;
  209. int size;
  210. };
  211. /*
  212. * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
  213. * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
  214. * mode.
  215. */
  216. struct kvm_mmu {
  217. void (*new_cr3)(struct kvm_vcpu *vcpu);
  218. void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
  219. unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
  220. u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
  221. int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
  222. bool prefault);
  223. void (*inject_page_fault)(struct kvm_vcpu *vcpu,
  224. struct x86_exception *fault);
  225. void (*free)(struct kvm_vcpu *vcpu);
  226. gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
  227. struct x86_exception *exception);
  228. gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access);
  229. int (*sync_page)(struct kvm_vcpu *vcpu,
  230. struct kvm_mmu_page *sp);
  231. void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
  232. void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
  233. u64 *spte, const void *pte);
  234. hpa_t root_hpa;
  235. int root_level;
  236. int shadow_root_level;
  237. union kvm_mmu_page_role base_role;
  238. bool direct_map;
  239. u64 *pae_root;
  240. u64 *lm_root;
  241. u64 rsvd_bits_mask[2][4];
  242. bool nx;
  243. u64 pdptrs[4]; /* pae */
  244. };
  245. enum pmc_type {
  246. KVM_PMC_GP = 0,
  247. KVM_PMC_FIXED,
  248. };
  249. struct kvm_pmc {
  250. enum pmc_type type;
  251. u8 idx;
  252. u64 counter;
  253. u64 eventsel;
  254. struct perf_event *perf_event;
  255. struct kvm_vcpu *vcpu;
  256. };
  257. struct kvm_pmu {
  258. unsigned nr_arch_gp_counters;
  259. unsigned nr_arch_fixed_counters;
  260. unsigned available_event_types;
  261. u64 fixed_ctr_ctrl;
  262. u64 global_ctrl;
  263. u64 global_status;
  264. u64 global_ovf_ctrl;
  265. u64 counter_bitmask[2];
  266. u64 global_ctrl_mask;
  267. u8 version;
  268. struct kvm_pmc gp_counters[X86_PMC_MAX_GENERIC];
  269. struct kvm_pmc fixed_counters[X86_PMC_MAX_FIXED];
  270. struct irq_work irq_work;
  271. u64 reprogram_pmi;
  272. };
  273. struct kvm_vcpu_arch {
  274. /*
  275. * rip and regs accesses must go through
  276. * kvm_{register,rip}_{read,write} functions.
  277. */
  278. unsigned long regs[NR_VCPU_REGS];
  279. u32 regs_avail;
  280. u32 regs_dirty;
  281. unsigned long cr0;
  282. unsigned long cr0_guest_owned_bits;
  283. unsigned long cr2;
  284. unsigned long cr3;
  285. unsigned long cr4;
  286. unsigned long cr4_guest_owned_bits;
  287. unsigned long cr8;
  288. u32 hflags;
  289. u64 efer;
  290. u64 apic_base;
  291. struct kvm_lapic *apic; /* kernel irqchip context */
  292. int32_t apic_arb_prio;
  293. int mp_state;
  294. int sipi_vector;
  295. u64 ia32_misc_enable_msr;
  296. bool tpr_access_reporting;
  297. /*
  298. * Paging state of the vcpu
  299. *
  300. * If the vcpu runs in guest mode with two level paging this still saves
  301. * the paging mode of the l1 guest. This context is always used to
  302. * handle faults.
  303. */
  304. struct kvm_mmu mmu;
  305. /*
  306. * Paging state of an L2 guest (used for nested npt)
  307. *
  308. * This context will save all necessary information to walk page tables
  309. * of the an L2 guest. This context is only initialized for page table
  310. * walking and not for faulting since we never handle l2 page faults on
  311. * the host.
  312. */
  313. struct kvm_mmu nested_mmu;
  314. /*
  315. * Pointer to the mmu context currently used for
  316. * gva_to_gpa translations.
  317. */
  318. struct kvm_mmu *walk_mmu;
  319. struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
  320. struct kvm_mmu_memory_cache mmu_page_cache;
  321. struct kvm_mmu_memory_cache mmu_page_header_cache;
  322. struct fpu guest_fpu;
  323. u64 xcr0;
  324. struct kvm_pio_request pio;
  325. void *pio_data;
  326. u8 event_exit_inst_len;
  327. struct kvm_queued_exception {
  328. bool pending;
  329. bool has_error_code;
  330. bool reinject;
  331. u8 nr;
  332. u32 error_code;
  333. } exception;
  334. struct kvm_queued_interrupt {
  335. bool pending;
  336. bool soft;
  337. u8 nr;
  338. } interrupt;
  339. int halt_request; /* real mode on Intel only */
  340. int cpuid_nent;
  341. struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
  342. /* emulate context */
  343. struct x86_emulate_ctxt emulate_ctxt;
  344. bool emulate_regs_need_sync_to_vcpu;
  345. bool emulate_regs_need_sync_from_vcpu;
  346. gpa_t time;
  347. struct pvclock_vcpu_time_info hv_clock;
  348. unsigned int hw_tsc_khz;
  349. unsigned int time_offset;
  350. struct page *time_page;
  351. struct {
  352. u64 msr_val;
  353. u64 last_steal;
  354. u64 accum_steal;
  355. struct gfn_to_hva_cache stime;
  356. struct kvm_steal_time steal;
  357. } st;
  358. u64 last_guest_tsc;
  359. u64 last_kernel_ns;
  360. u64 last_tsc_nsec;
  361. u64 last_tsc_write;
  362. bool tsc_catchup;
  363. bool tsc_always_catchup;
  364. s8 virtual_tsc_shift;
  365. u32 virtual_tsc_mult;
  366. u32 virtual_tsc_khz;
  367. atomic_t nmi_queued; /* unprocessed asynchronous NMIs */
  368. unsigned nmi_pending; /* NMI queued after currently running handler */
  369. bool nmi_injected; /* Trying to inject an NMI this entry */
  370. struct mtrr_state_type mtrr_state;
  371. u32 pat;
  372. int switch_db_regs;
  373. unsigned long db[KVM_NR_DB_REGS];
  374. unsigned long dr6;
  375. unsigned long dr7;
  376. unsigned long eff_db[KVM_NR_DB_REGS];
  377. u64 mcg_cap;
  378. u64 mcg_status;
  379. u64 mcg_ctl;
  380. u64 *mce_banks;
  381. /* Cache MMIO info */
  382. u64 mmio_gva;
  383. unsigned access;
  384. gfn_t mmio_gfn;
  385. struct kvm_pmu pmu;
  386. /* used for guest single stepping over the given code position */
  387. unsigned long singlestep_rip;
  388. /* fields used by HYPER-V emulation */
  389. u64 hv_vapic;
  390. cpumask_var_t wbinvd_dirty_mask;
  391. unsigned long last_retry_eip;
  392. unsigned long last_retry_addr;
  393. struct {
  394. bool halted;
  395. gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
  396. struct gfn_to_hva_cache data;
  397. u64 msr_val;
  398. u32 id;
  399. bool send_user_only;
  400. } apf;
  401. /* OSVW MSRs (AMD only) */
  402. struct {
  403. u64 length;
  404. u64 status;
  405. } osvw;
  406. };
  407. struct kvm_arch {
  408. unsigned int n_used_mmu_pages;
  409. unsigned int n_requested_mmu_pages;
  410. unsigned int n_max_mmu_pages;
  411. unsigned int indirect_shadow_pages;
  412. struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
  413. /*
  414. * Hash table of struct kvm_mmu_page.
  415. */
  416. struct list_head active_mmu_pages;
  417. struct list_head assigned_dev_head;
  418. struct iommu_domain *iommu_domain;
  419. int iommu_flags;
  420. struct kvm_pic *vpic;
  421. struct kvm_ioapic *vioapic;
  422. struct kvm_pit *vpit;
  423. int vapics_in_nmi_mode;
  424. unsigned int tss_addr;
  425. struct page *apic_access_page;
  426. gpa_t wall_clock;
  427. struct page *ept_identity_pagetable;
  428. bool ept_identity_pagetable_done;
  429. gpa_t ept_identity_map_addr;
  430. unsigned long irq_sources_bitmap;
  431. s64 kvmclock_offset;
  432. raw_spinlock_t tsc_write_lock;
  433. u64 last_tsc_nsec;
  434. u64 last_tsc_offset;
  435. u64 last_tsc_write;
  436. struct kvm_xen_hvm_config xen_hvm_config;
  437. /* fields used by HYPER-V emulation */
  438. u64 hv_guest_os_id;
  439. u64 hv_hypercall;
  440. atomic_t reader_counter;
  441. #ifdef CONFIG_KVM_MMU_AUDIT
  442. int audit_point;
  443. #endif
  444. };
  445. struct kvm_vm_stat {
  446. u32 mmu_shadow_zapped;
  447. u32 mmu_pte_write;
  448. u32 mmu_pte_updated;
  449. u32 mmu_pde_zapped;
  450. u32 mmu_flooded;
  451. u32 mmu_recycled;
  452. u32 mmu_cache_miss;
  453. u32 mmu_unsync;
  454. u32 remote_tlb_flush;
  455. u32 lpages;
  456. };
  457. struct kvm_vcpu_stat {
  458. u32 pf_fixed;
  459. u32 pf_guest;
  460. u32 tlb_flush;
  461. u32 invlpg;
  462. u32 exits;
  463. u32 io_exits;
  464. u32 mmio_exits;
  465. u32 signal_exits;
  466. u32 irq_window_exits;
  467. u32 nmi_window_exits;
  468. u32 halt_exits;
  469. u32 halt_wakeup;
  470. u32 request_irq_exits;
  471. u32 irq_exits;
  472. u32 host_state_reload;
  473. u32 efer_reload;
  474. u32 fpu_reload;
  475. u32 insn_emulation;
  476. u32 insn_emulation_fail;
  477. u32 hypercalls;
  478. u32 irq_injections;
  479. u32 nmi_injections;
  480. };
  481. struct x86_instruction_info;
  482. struct kvm_x86_ops {
  483. int (*cpu_has_kvm_support)(void); /* __init */
  484. int (*disabled_by_bios)(void); /* __init */
  485. int (*hardware_enable)(void *dummy);
  486. void (*hardware_disable)(void *dummy);
  487. void (*check_processor_compatibility)(void *rtn);
  488. int (*hardware_setup)(void); /* __init */
  489. void (*hardware_unsetup)(void); /* __exit */
  490. bool (*cpu_has_accelerated_tpr)(void);
  491. void (*cpuid_update)(struct kvm_vcpu *vcpu);
  492. /* Create, but do not attach this VCPU */
  493. struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
  494. void (*vcpu_free)(struct kvm_vcpu *vcpu);
  495. int (*vcpu_reset)(struct kvm_vcpu *vcpu);
  496. void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
  497. void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
  498. void (*vcpu_put)(struct kvm_vcpu *vcpu);
  499. void (*set_guest_debug)(struct kvm_vcpu *vcpu,
  500. struct kvm_guest_debug *dbg);
  501. int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
  502. int (*set_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
  503. u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
  504. void (*get_segment)(struct kvm_vcpu *vcpu,
  505. struct kvm_segment *var, int seg);
  506. int (*get_cpl)(struct kvm_vcpu *vcpu);
  507. void (*set_segment)(struct kvm_vcpu *vcpu,
  508. struct kvm_segment *var, int seg);
  509. void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
  510. void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
  511. void (*decache_cr3)(struct kvm_vcpu *vcpu);
  512. void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
  513. void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
  514. void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
  515. int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
  516. void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
  517. void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
  518. void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
  519. void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
  520. void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
  521. void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
  522. void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
  523. unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
  524. void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
  525. void (*fpu_activate)(struct kvm_vcpu *vcpu);
  526. void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
  527. void (*tlb_flush)(struct kvm_vcpu *vcpu);
  528. void (*run)(struct kvm_vcpu *vcpu);
  529. int (*handle_exit)(struct kvm_vcpu *vcpu);
  530. void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
  531. void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
  532. u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
  533. void (*patch_hypercall)(struct kvm_vcpu *vcpu,
  534. unsigned char *hypercall_addr);
  535. void (*set_irq)(struct kvm_vcpu *vcpu);
  536. void (*set_nmi)(struct kvm_vcpu *vcpu);
  537. void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
  538. bool has_error_code, u32 error_code,
  539. bool reinject);
  540. void (*cancel_injection)(struct kvm_vcpu *vcpu);
  541. int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
  542. int (*nmi_allowed)(struct kvm_vcpu *vcpu);
  543. bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
  544. void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
  545. void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
  546. void (*enable_irq_window)(struct kvm_vcpu *vcpu);
  547. void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
  548. int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
  549. int (*get_tdp_level)(void);
  550. u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
  551. int (*get_lpage_level)(void);
  552. bool (*rdtscp_supported)(void);
  553. void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment);
  554. void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
  555. void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
  556. bool (*has_wbinvd_exit)(void);
  557. void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale);
  558. void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
  559. u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc);
  560. u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu);
  561. void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
  562. int (*check_intercept)(struct kvm_vcpu *vcpu,
  563. struct x86_instruction_info *info,
  564. enum x86_intercept_stage stage);
  565. };
  566. struct kvm_arch_async_pf {
  567. u32 token;
  568. gfn_t gfn;
  569. unsigned long cr3;
  570. bool direct_map;
  571. };
  572. extern struct kvm_x86_ops *kvm_x86_ops;
  573. int kvm_mmu_module_init(void);
  574. void kvm_mmu_module_exit(void);
  575. void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
  576. int kvm_mmu_create(struct kvm_vcpu *vcpu);
  577. int kvm_mmu_setup(struct kvm_vcpu *vcpu);
  578. void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
  579. u64 dirty_mask, u64 nx_mask, u64 x_mask);
  580. int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
  581. void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
  582. int kvm_mmu_rmap_write_protect(struct kvm *kvm, u64 gfn,
  583. struct kvm_memory_slot *slot);
  584. void kvm_mmu_zap_all(struct kvm *kvm);
  585. unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
  586. void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
  587. int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
  588. int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
  589. const void *val, int bytes);
  590. u8 kvm_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn);
  591. extern bool tdp_enabled;
  592. u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
  593. /* control of guest tsc rate supported? */
  594. extern bool kvm_has_tsc_control;
  595. /* minimum supported tsc_khz for guests */
  596. extern u32 kvm_min_guest_tsc_khz;
  597. /* maximum supported tsc_khz for guests */
  598. extern u32 kvm_max_guest_tsc_khz;
  599. enum emulation_result {
  600. EMULATE_DONE, /* no further processing */
  601. EMULATE_DO_MMIO, /* kvm_run filled with mmio request */
  602. EMULATE_FAIL, /* can't emulate this instruction */
  603. };
  604. #define EMULTYPE_NO_DECODE (1 << 0)
  605. #define EMULTYPE_TRAP_UD (1 << 1)
  606. #define EMULTYPE_SKIP (1 << 2)
  607. #define EMULTYPE_RETRY (1 << 3)
  608. int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
  609. int emulation_type, void *insn, int insn_len);
  610. static inline int emulate_instruction(struct kvm_vcpu *vcpu,
  611. int emulation_type)
  612. {
  613. return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
  614. }
  615. void kvm_enable_efer_bits(u64);
  616. int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
  617. int kvm_set_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
  618. struct x86_emulate_ctxt;
  619. int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
  620. void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
  621. int kvm_emulate_halt(struct kvm_vcpu *vcpu);
  622. int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
  623. void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
  624. int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
  625. int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int reason,
  626. bool has_error_code, u32 error_code);
  627. int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
  628. int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
  629. int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
  630. int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
  631. int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
  632. int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
  633. unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
  634. void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
  635. void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
  636. int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
  637. int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
  638. int kvm_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data);
  639. unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
  640. void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
  641. bool kvm_rdpmc(struct kvm_vcpu *vcpu);
  642. void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
  643. void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
  644. void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
  645. void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
  646. void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
  647. int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
  648. gfn_t gfn, void *data, int offset, int len,
  649. u32 access);
  650. void kvm_propagate_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
  651. bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
  652. int kvm_pic_set_irq(void *opaque, int irq, int level);
  653. void kvm_inject_nmi(struct kvm_vcpu *vcpu);
  654. int fx_init(struct kvm_vcpu *vcpu);
  655. void kvm_mmu_flush_tlb(struct kvm_vcpu *vcpu);
  656. void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
  657. const u8 *new, int bytes);
  658. int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
  659. int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
  660. void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
  661. int kvm_mmu_load(struct kvm_vcpu *vcpu);
  662. void kvm_mmu_unload(struct kvm_vcpu *vcpu);
  663. void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
  664. gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access);
  665. gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
  666. struct x86_exception *exception);
  667. gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
  668. struct x86_exception *exception);
  669. gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
  670. struct x86_exception *exception);
  671. gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
  672. struct x86_exception *exception);
  673. int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
  674. int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
  675. void *insn, int insn_len);
  676. void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
  677. void kvm_enable_tdp(void);
  678. void kvm_disable_tdp(void);
  679. int complete_pio(struct kvm_vcpu *vcpu);
  680. bool kvm_check_iopl(struct kvm_vcpu *vcpu);
  681. static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access)
  682. {
  683. return gpa;
  684. }
  685. static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
  686. {
  687. struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
  688. return (struct kvm_mmu_page *)page_private(page);
  689. }
  690. static inline u16 kvm_read_ldt(void)
  691. {
  692. u16 ldt;
  693. asm("sldt %0" : "=g"(ldt));
  694. return ldt;
  695. }
  696. static inline void kvm_load_ldt(u16 sel)
  697. {
  698. asm("lldt %0" : : "rm"(sel));
  699. }
  700. #ifdef CONFIG_X86_64
  701. static inline unsigned long read_msr(unsigned long msr)
  702. {
  703. u64 value;
  704. rdmsrl(msr, value);
  705. return value;
  706. }
  707. #endif
  708. static inline u32 get_rdx_init_val(void)
  709. {
  710. return 0x600; /* P6 family */
  711. }
  712. static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
  713. {
  714. kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
  715. }
  716. #define TSS_IOPB_BASE_OFFSET 0x66
  717. #define TSS_BASE_SIZE 0x68
  718. #define TSS_IOPB_SIZE (65536 / 8)
  719. #define TSS_REDIRECTION_SIZE (256 / 8)
  720. #define RMODE_TSS_SIZE \
  721. (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
  722. enum {
  723. TASK_SWITCH_CALL = 0,
  724. TASK_SWITCH_IRET = 1,
  725. TASK_SWITCH_JMP = 2,
  726. TASK_SWITCH_GATE = 3,
  727. };
  728. #define HF_GIF_MASK (1 << 0)
  729. #define HF_HIF_MASK (1 << 1)
  730. #define HF_VINTR_MASK (1 << 2)
  731. #define HF_NMI_MASK (1 << 3)
  732. #define HF_IRET_MASK (1 << 4)
  733. #define HF_GUEST_MASK (1 << 5) /* VCPU is in guest-mode */
  734. /*
  735. * Hardware virtualization extension instructions may fault if a
  736. * reboot turns off virtualization while processes are running.
  737. * Trap the fault and ignore the instruction if that happens.
  738. */
  739. asmlinkage void kvm_spurious_fault(void);
  740. extern bool kvm_rebooting;
  741. #define ____kvm_handle_fault_on_reboot(insn, cleanup_insn) \
  742. "666: " insn "\n\t" \
  743. "668: \n\t" \
  744. ".pushsection .fixup, \"ax\" \n" \
  745. "667: \n\t" \
  746. cleanup_insn "\n\t" \
  747. "cmpb $0, kvm_rebooting \n\t" \
  748. "jne 668b \n\t" \
  749. __ASM_SIZE(push) " $666b \n\t" \
  750. "call kvm_spurious_fault \n\t" \
  751. ".popsection \n\t" \
  752. ".pushsection __ex_table, \"a\" \n\t" \
  753. _ASM_PTR " 666b, 667b \n\t" \
  754. ".popsection"
  755. #define __kvm_handle_fault_on_reboot(insn) \
  756. ____kvm_handle_fault_on_reboot(insn, "")
  757. #define KVM_ARCH_WANT_MMU_NOTIFIER
  758. int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
  759. int kvm_age_hva(struct kvm *kvm, unsigned long hva);
  760. int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
  761. void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
  762. int cpuid_maxphyaddr(struct kvm_vcpu *vcpu);
  763. int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
  764. int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
  765. int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
  766. void kvm_define_shared_msr(unsigned index, u32 msr);
  767. void kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
  768. bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
  769. void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
  770. struct kvm_async_pf *work);
  771. void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
  772. struct kvm_async_pf *work);
  773. void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
  774. struct kvm_async_pf *work);
  775. bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
  776. extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
  777. void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
  778. int kvm_is_in_guest(void);
  779. void kvm_pmu_init(struct kvm_vcpu *vcpu);
  780. void kvm_pmu_destroy(struct kvm_vcpu *vcpu);
  781. void kvm_pmu_reset(struct kvm_vcpu *vcpu);
  782. void kvm_pmu_cpuid_update(struct kvm_vcpu *vcpu);
  783. bool kvm_pmu_msr(struct kvm_vcpu *vcpu, u32 msr);
  784. int kvm_pmu_get_msr(struct kvm_vcpu *vcpu, u32 msr, u64 *data);
  785. int kvm_pmu_set_msr(struct kvm_vcpu *vcpu, u32 msr, u64 data);
  786. int kvm_pmu_read_pmc(struct kvm_vcpu *vcpu, unsigned pmc, u64 *data);
  787. void kvm_handle_pmu_event(struct kvm_vcpu *vcpu);
  788. void kvm_deliver_pmi(struct kvm_vcpu *vcpu);
  789. #endif /* _ASM_X86_KVM_HOST_H */