2 * Kernel-based Virtual Machine driver for Linux
4 * This header defines architecture specific interfaces, x86 version
6 * This work is licensed under the terms of the GNU GPL, version 2. See
7 * the COPYING file in the top-level directory.
11 #ifndef _ASM_X86_KVM_HOST_H
12 #define _ASM_X86_KVM_HOST_H
14 #include <linux/types.h>
16 #include <linux/mmu_notifier.h>
17 #include <linux/tracepoint.h>
18 #include <linux/cpumask.h>
19 #include <linux/irq_work.h>
21 #include <linux/kvm.h>
22 #include <linux/kvm_para.h>
23 #include <linux/kvm_types.h>
24 #include <linux/perf_event.h>
25 #include <linux/pvclock_gtod.h>
26 #include <linux/clocksource.h>
28 #include <asm/pvclock-abi.h>
31 #include <asm/msr-index.h>
34 #define KVM_MAX_VCPUS 255
35 #define KVM_SOFT_MAX_VCPUS 160
36 #define KVM_USER_MEM_SLOTS 509
37 /* memory slots that are not exposed to userspace */
38 #define KVM_PRIVATE_MEM_SLOTS 3
39 #define KVM_MEM_SLOTS_NUM (KVM_USER_MEM_SLOTS + KVM_PRIVATE_MEM_SLOTS)
41 #define KVM_MMIO_SIZE 16
43 #define KVM_PIO_PAGE_OFFSET 1
44 #define KVM_COALESCED_MMIO_PAGE_OFFSET 2
46 #define KVM_IRQCHIP_NUM_PINS KVM_IOAPIC_NUM_PINS
48 #define CR0_RESERVED_BITS \
49 (~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
50 | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
51 | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
53 #define CR3_L_MODE_RESERVED_BITS 0xFFFFFF0000000000ULL
54 #define CR3_PCID_INVD (1UL << 63)
55 #define CR4_RESERVED_BITS \
56 (~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
57 | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE \
58 | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR | X86_CR4_PCIDE \
59 | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_FSGSBASE \
60 | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE | X86_CR4_SMAP))
62 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
66 #define INVALID_PAGE (~(hpa_t)0)
67 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
69 #define UNMAPPED_GVA (~(gpa_t)0)
71 /* KVM Hugepage definitions for x86 */
72 #define KVM_NR_PAGE_SIZES 3
73 #define KVM_HPAGE_GFN_SHIFT(x) (((x) - 1) * 9)
74 #define KVM_HPAGE_SHIFT(x) (PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
75 #define KVM_HPAGE_SIZE(x) (1UL << KVM_HPAGE_SHIFT(x))
76 #define KVM_HPAGE_MASK(x) (~(KVM_HPAGE_SIZE(x) - 1))
77 #define KVM_PAGES_PER_HPAGE(x) (KVM_HPAGE_SIZE(x) / PAGE_SIZE)
79 static inline gfn_t gfn_to_index(gfn_t gfn, gfn_t base_gfn, int level)
81 /* KVM_HPAGE_GFN_SHIFT(PT_PAGE_TABLE_LEVEL) must be 0. */
82 return (gfn >> KVM_HPAGE_GFN_SHIFT(level)) -
83 (base_gfn >> KVM_HPAGE_GFN_SHIFT(level));
86 #define SELECTOR_TI_MASK (1 << 2)
87 #define SELECTOR_RPL_MASK 0x03
91 #define KVM_PERMILLE_MMU_PAGES 20
92 #define KVM_MIN_ALLOC_MMU_PAGES 64
93 #define KVM_MMU_HASH_SHIFT 10
94 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
95 #define KVM_MIN_FREE_MMU_PAGES 5
96 #define KVM_REFILL_PAGES 25
97 #define KVM_MAX_CPUID_ENTRIES 80
98 #define KVM_NR_FIXED_MTRR_REGION 88
99 #define KVM_NR_VAR_MTRR 8
101 #define ASYNC_PF_PER_VCPU 64
127 VCPU_EXREG_PDPTR = NR_VCPU_REGS,
144 #include <asm/kvm_emulate.h>
146 #define KVM_NR_MEM_OBJS 40
148 #define KVM_NR_DB_REGS 4
150 #define DR6_BD (1 << 13)
151 #define DR6_BS (1 << 14)
152 #define DR6_RTM (1 << 16)
153 #define DR6_FIXED_1 0xfffe0ff0
154 #define DR6_INIT 0xffff0ff0
155 #define DR6_VOLATILE 0x0001e00f
157 #define DR7_BP_EN_MASK 0x000000ff
158 #define DR7_GE (1 << 9)
159 #define DR7_GD (1 << 13)
160 #define DR7_FIXED_1 0x00000400
161 #define DR7_VOLATILE 0xffff2bff
163 /* apic attention bits */
164 #define KVM_APIC_CHECK_VAPIC 0
166 * The following bit is set with PV-EOI, unset on EOI.
167 * We detect PV-EOI changes by guest by comparing
168 * this bit with PV-EOI in guest memory.
169 * See the implementation in apic_update_pv_eoi.
171 #define KVM_APIC_PV_EOI_PENDING 1
174 * We don't want allocation failures within the mmu code, so we preallocate
175 * enough memory for a single page fault in a cache.
177 struct kvm_mmu_memory_cache {
179 void *objects[KVM_NR_MEM_OBJS];
183 * kvm_mmu_page_role, below, is defined as:
185 * bits 0:3 - total guest paging levels (2-4, or zero for real mode)
186 * bits 4:7 - page table level for this shadow (1-4)
187 * bits 8:9 - page table quadrant for 2-level guests
188 * bit 16 - direct mapping of virtual to physical mapping at gfn
189 * used for real mode and two-dimensional paging
190 * bits 17:19 - common access permissions for all ptes in this shadow page
192 union kvm_mmu_page_role {
198 unsigned pad_for_nice_hex_output:6;
204 unsigned smep_andnot_wp:1;
208 struct kvm_mmu_page {
209 struct list_head link;
210 struct hlist_node hash_link;
213 * The following two entries are used to key the shadow page in the
217 union kvm_mmu_page_role role;
220 /* hold the gfn of each spte inside spt */
223 int root_count; /* Currently serving as active root */
224 unsigned int unsync_children;
225 unsigned long parent_ptes; /* Reverse mapping for parent_pte */
227 /* The page is obsolete if mmu_valid_gen != kvm->arch.mmu_valid_gen. */
228 unsigned long mmu_valid_gen;
230 DECLARE_BITMAP(unsync_child_bitmap, 512);
234 * Used out of the mmu-lock to avoid reading spte values while an
235 * update is in progress; see the comments in __get_spte_lockless().
237 int clear_spte_count;
240 /* Number of writes since the last time traversal visited this page. */
241 int write_flooding_count;
244 struct kvm_pio_request {
252 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
253 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
257 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
258 unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
259 u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
260 int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
262 void (*inject_page_fault)(struct kvm_vcpu *vcpu,
263 struct x86_exception *fault);
264 gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
265 struct x86_exception *exception);
266 gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
267 struct x86_exception *exception);
268 int (*sync_page)(struct kvm_vcpu *vcpu,
269 struct kvm_mmu_page *sp);
270 void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
271 void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
272 u64 *spte, const void *pte);
275 int shadow_root_level;
276 union kvm_mmu_page_role base_role;
280 * Bitmap; bit set = permission fault
281 * Byte index: page fault error code [4:1]
282 * Bit index: pte permissions in ACC_* format
288 u64 rsvd_bits_mask[2][4];
292 * Bitmap: bit set = last pte in walk
293 * index[0:1]: level (zero-based)
300 u64 pdptrs[4]; /* pae */
313 struct perf_event *perf_event;
314 struct kvm_vcpu *vcpu;
318 unsigned nr_arch_gp_counters;
319 unsigned nr_arch_fixed_counters;
320 unsigned available_event_types;
325 u64 counter_bitmask[2];
326 u64 global_ctrl_mask;
329 struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
330 struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
331 struct irq_work irq_work;
336 KVM_DEBUGREG_BP_ENABLED = 1,
337 KVM_DEBUGREG_WONT_EXIT = 2,
340 struct kvm_vcpu_arch {
342 * rip and regs accesses must go through
343 * kvm_{register,rip}_{read,write} functions.
345 unsigned long regs[NR_VCPU_REGS];
350 unsigned long cr0_guest_owned_bits;
354 unsigned long cr4_guest_owned_bits;
359 struct kvm_lapic *apic; /* kernel irqchip context */
360 unsigned long apic_attention;
361 int32_t apic_arb_prio;
363 u64 ia32_misc_enable_msr;
364 bool tpr_access_reporting;
368 * Paging state of the vcpu
370 * If the vcpu runs in guest mode with two level paging this still saves
371 * the paging mode of the l1 guest. This context is always used to
377 * Paging state of an L2 guest (used for nested npt)
379 * This context will save all necessary information to walk page tables
380 * of the an L2 guest. This context is only initialized for page table
381 * walking and not for faulting since we never handle l2 page faults on
384 struct kvm_mmu nested_mmu;
387 * Pointer to the mmu context currently used for
388 * gva_to_gpa translations.
390 struct kvm_mmu *walk_mmu;
392 struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
393 struct kvm_mmu_memory_cache mmu_page_cache;
394 struct kvm_mmu_memory_cache mmu_page_header_cache;
396 struct fpu guest_fpu;
398 u64 guest_supported_xcr0;
399 u32 guest_xstate_size;
401 struct kvm_pio_request pio;
404 u8 event_exit_inst_len;
406 struct kvm_queued_exception {
414 struct kvm_queued_interrupt {
420 int halt_request; /* real mode on Intel only */
423 struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
424 /* emulate context */
426 struct x86_emulate_ctxt emulate_ctxt;
427 bool emulate_regs_need_sync_to_vcpu;
428 bool emulate_regs_need_sync_from_vcpu;
429 int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
432 struct pvclock_vcpu_time_info hv_clock;
433 unsigned int hw_tsc_khz;
434 struct gfn_to_hva_cache pv_time;
435 bool pv_time_enabled;
436 /* set guest stopped flag in pvclock flags field */
437 bool pvclock_set_guest_stopped_request;
443 struct gfn_to_hva_cache stime;
444 struct kvm_steal_time steal;
449 u64 tsc_offset_adjustment;
452 u64 this_tsc_generation;
454 bool tsc_always_catchup;
455 s8 virtual_tsc_shift;
456 u32 virtual_tsc_mult;
458 s64 ia32_tsc_adjust_msr;
460 atomic_t nmi_queued; /* unprocessed asynchronous NMIs */
461 unsigned nmi_pending; /* NMI queued after currently running handler */
462 bool nmi_injected; /* Trying to inject an NMI this entry */
464 struct mtrr_state_type mtrr_state;
467 unsigned switch_db_regs;
468 unsigned long db[KVM_NR_DB_REGS];
471 unsigned long eff_db[KVM_NR_DB_REGS];
472 unsigned long guest_debug_dr7;
479 /* Cache MMIO info */
487 /* used for guest single stepping over the given code position */
488 unsigned long singlestep_rip;
490 /* fields used by HYPER-V emulation */
493 cpumask_var_t wbinvd_dirty_mask;
495 unsigned long last_retry_eip;
496 unsigned long last_retry_addr;
500 gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
501 struct gfn_to_hva_cache data;
507 /* OSVW MSRs (AMD only) */
515 struct gfn_to_hva_cache data;
519 * Indicate whether the access faults on its page table in guest
520 * which is set when fix page fault and used to detect unhandeable
523 bool write_fault_to_shadow_pgtable;
525 /* set at EPT violation at this point */
526 unsigned long exit_qualification;
528 /* pv related host specific info */
534 struct kvm_lpage_info {
538 struct kvm_arch_memory_slot {
539 unsigned long *rmap[KVM_NR_PAGE_SIZES];
540 struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
543 struct kvm_apic_map {
546 /* fields bellow are used to decode ldr values in different modes */
547 u32 cid_shift, cid_mask, lid_mask, broadcast;
548 struct kvm_lapic *phys_map[256];
549 /* first index is cluster id second is cpu id in a cluster */
550 struct kvm_lapic *logical_map[16][16];
554 unsigned int n_used_mmu_pages;
555 unsigned int n_requested_mmu_pages;
556 unsigned int n_max_mmu_pages;
557 unsigned int indirect_shadow_pages;
558 unsigned long mmu_valid_gen;
559 struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
561 * Hash table of struct kvm_mmu_page.
563 struct list_head active_mmu_pages;
564 struct list_head zapped_obsolete_pages;
566 struct list_head assigned_dev_head;
567 struct iommu_domain *iommu_domain;
568 bool iommu_noncoherent;
569 #define __KVM_HAVE_ARCH_NONCOHERENT_DMA
570 atomic_t noncoherent_dma_count;
571 struct kvm_pic *vpic;
572 struct kvm_ioapic *vioapic;
573 struct kvm_pit *vpit;
574 int vapics_in_nmi_mode;
575 struct mutex apic_map_lock;
576 struct kvm_apic_map *apic_map;
578 unsigned int tss_addr;
579 bool apic_access_page_done;
583 bool ept_identity_pagetable_done;
584 gpa_t ept_identity_map_addr;
586 unsigned long irq_sources_bitmap;
588 raw_spinlock_t tsc_write_lock;
595 u64 cur_tsc_generation;
596 int nr_vcpus_matched_tsc;
598 spinlock_t pvclock_gtod_sync_lock;
599 bool use_master_clock;
600 u64 master_kernel_ns;
601 cycle_t master_cycle_now;
602 struct delayed_work kvmclock_update_work;
603 struct delayed_work kvmclock_sync_work;
605 struct kvm_xen_hvm_config xen_hvm_config;
607 /* reads protected by irq_srcu, writes by irq_lock */
608 struct hlist_head mask_notifier_list;
610 /* fields used by HYPER-V emulation */
615 #ifdef CONFIG_KVM_MMU_AUDIT
621 u32 mmu_shadow_zapped;
629 u32 remote_tlb_flush;
633 struct kvm_vcpu_stat {
643 u32 irq_window_exits;
644 u32 nmi_window_exits;
647 u32 request_irq_exits;
649 u32 host_state_reload;
653 u32 insn_emulation_fail;
659 struct x86_instruction_info;
667 struct kvm_lapic_irq {
678 int (*cpu_has_kvm_support)(void); /* __init */
679 int (*disabled_by_bios)(void); /* __init */
680 int (*hardware_enable)(void);
681 void (*hardware_disable)(void);
682 void (*check_processor_compatibility)(void *rtn);
683 int (*hardware_setup)(void); /* __init */
684 void (*hardware_unsetup)(void); /* __exit */
685 bool (*cpu_has_accelerated_tpr)(void);
686 void (*cpuid_update)(struct kvm_vcpu *vcpu);
688 /* Create, but do not attach this VCPU */
689 struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
690 void (*vcpu_free)(struct kvm_vcpu *vcpu);
691 void (*vcpu_reset)(struct kvm_vcpu *vcpu);
693 void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
694 void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
695 void (*vcpu_put)(struct kvm_vcpu *vcpu);
697 void (*update_db_bp_intercept)(struct kvm_vcpu *vcpu);
698 int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
699 int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
700 u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
701 void (*get_segment)(struct kvm_vcpu *vcpu,
702 struct kvm_segment *var, int seg);
703 int (*get_cpl)(struct kvm_vcpu *vcpu);
704 void (*set_segment)(struct kvm_vcpu *vcpu,
705 struct kvm_segment *var, int seg);
706 void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
707 void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
708 void (*decache_cr3)(struct kvm_vcpu *vcpu);
709 void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
710 void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
711 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
712 int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
713 void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
714 void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
715 void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
716 void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
717 void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
718 u64 (*get_dr6)(struct kvm_vcpu *vcpu);
719 void (*set_dr6)(struct kvm_vcpu *vcpu, unsigned long value);
720 void (*sync_dirty_debug_regs)(struct kvm_vcpu *vcpu);
721 void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
722 void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
723 unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
724 void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
725 void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
727 void (*tlb_flush)(struct kvm_vcpu *vcpu);
729 void (*run)(struct kvm_vcpu *vcpu);
730 int (*handle_exit)(struct kvm_vcpu *vcpu);
731 void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
732 void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
733 u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu);
734 void (*patch_hypercall)(struct kvm_vcpu *vcpu,
735 unsigned char *hypercall_addr);
736 void (*set_irq)(struct kvm_vcpu *vcpu);
737 void (*set_nmi)(struct kvm_vcpu *vcpu);
738 void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
739 bool has_error_code, u32 error_code,
741 void (*cancel_injection)(struct kvm_vcpu *vcpu);
742 int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
743 int (*nmi_allowed)(struct kvm_vcpu *vcpu);
744 bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
745 void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
746 void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
747 void (*enable_irq_window)(struct kvm_vcpu *vcpu);
748 void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
749 int (*vm_has_apicv)(struct kvm *kvm);
750 void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
751 void (*hwapic_isr_update)(struct kvm *kvm, int isr);
752 void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap);
753 void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu, bool set);
754 void (*set_apic_access_page_addr)(struct kvm_vcpu *vcpu, hpa_t hpa);
755 void (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
756 void (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
757 int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
758 int (*get_tdp_level)(void);
759 u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
760 int (*get_lpage_level)(void);
761 bool (*rdtscp_supported)(void);
762 bool (*invpcid_supported)(void);
763 void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment, bool host);
765 void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
767 void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
769 bool (*has_wbinvd_exit)(void);
771 void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale);
772 u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu);
773 void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
775 u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc);
776 u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu, u64 host_tsc);
778 void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
780 int (*check_intercept)(struct kvm_vcpu *vcpu,
781 struct x86_instruction_info *info,
782 enum x86_intercept_stage stage);
783 void (*handle_external_intr)(struct kvm_vcpu *vcpu);
784 bool (*mpx_supported)(void);
785 bool (*xsaves_supported)(void);
787 int (*check_nested_events)(struct kvm_vcpu *vcpu, bool external_intr);
789 void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
792 struct kvm_arch_async_pf {
799 extern struct kvm_x86_ops *kvm_x86_ops;
801 static inline void adjust_tsc_offset_guest(struct kvm_vcpu *vcpu,
804 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, false);
807 static inline void adjust_tsc_offset_host(struct kvm_vcpu *vcpu, s64 adjustment)
809 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, true);
812 int kvm_mmu_module_init(void);
813 void kvm_mmu_module_exit(void);
815 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
816 int kvm_mmu_create(struct kvm_vcpu *vcpu);
817 void kvm_mmu_setup(struct kvm_vcpu *vcpu);
818 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
819 u64 dirty_mask, u64 nx_mask, u64 x_mask);
821 void kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
822 void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
823 void kvm_mmu_write_protect_pt_masked(struct kvm *kvm,
824 struct kvm_memory_slot *slot,
825 gfn_t gfn_offset, unsigned long mask);
826 void kvm_mmu_zap_all(struct kvm *kvm);
827 void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm);
828 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
829 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
831 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
833 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
834 const void *val, int bytes);
835 u8 kvm_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn);
837 struct kvm_irq_mask_notifier {
838 void (*func)(struct kvm_irq_mask_notifier *kimn, bool masked);
840 struct hlist_node link;
843 void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
844 struct kvm_irq_mask_notifier *kimn);
845 void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
846 struct kvm_irq_mask_notifier *kimn);
847 void kvm_fire_mask_notifiers(struct kvm *kvm, unsigned irqchip, unsigned pin,
850 extern bool tdp_enabled;
852 u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
854 /* control of guest tsc rate supported? */
855 extern bool kvm_has_tsc_control;
856 /* minimum supported tsc_khz for guests */
857 extern u32 kvm_min_guest_tsc_khz;
858 /* maximum supported tsc_khz for guests */
859 extern u32 kvm_max_guest_tsc_khz;
861 enum emulation_result {
862 EMULATE_DONE, /* no further processing */
863 EMULATE_USER_EXIT, /* kvm_run ready for userspace exit */
864 EMULATE_FAIL, /* can't emulate this instruction */
867 #define EMULTYPE_NO_DECODE (1 << 0)
868 #define EMULTYPE_TRAP_UD (1 << 1)
869 #define EMULTYPE_SKIP (1 << 2)
870 #define EMULTYPE_RETRY (1 << 3)
871 #define EMULTYPE_NO_REEXECUTE (1 << 4)
872 int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
873 int emulation_type, void *insn, int insn_len);
875 static inline int emulate_instruction(struct kvm_vcpu *vcpu,
878 return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
881 void kvm_enable_efer_bits(u64);
882 bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
883 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
884 int kvm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
886 struct x86_emulate_ctxt;
888 int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
889 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
890 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
891 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
893 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
894 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
895 void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
897 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
898 int reason, bool has_error_code, u32 error_code);
900 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
901 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
902 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
903 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
904 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
905 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
906 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
907 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
908 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
909 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
911 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
912 int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
914 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
915 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
916 bool kvm_rdpmc(struct kvm_vcpu *vcpu);
918 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
919 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
920 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
921 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
922 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
923 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
924 gfn_t gfn, void *data, int offset, int len,
926 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
927 bool kvm_require_dr(struct kvm_vcpu *vcpu, int dr);
929 static inline int __kvm_irq_line_state(unsigned long *irq_state,
930 int irq_source_id, int level)
932 /* Logical OR for level trig interrupt */
934 __set_bit(irq_source_id, irq_state);
936 __clear_bit(irq_source_id, irq_state);
938 return !!(*irq_state);
941 int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
942 void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
944 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
946 int fx_init(struct kvm_vcpu *vcpu);
948 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
949 const u8 *new, int bytes);
950 int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
951 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
952 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
953 int kvm_mmu_load(struct kvm_vcpu *vcpu);
954 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
955 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
956 gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
957 struct x86_exception *exception);
958 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
959 struct x86_exception *exception);
960 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
961 struct x86_exception *exception);
962 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
963 struct x86_exception *exception);
964 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
965 struct x86_exception *exception);
967 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
969 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
970 void *insn, int insn_len);
971 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
972 void kvm_mmu_new_cr3(struct kvm_vcpu *vcpu);
974 void kvm_enable_tdp(void);
975 void kvm_disable_tdp(void);
977 static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
978 struct x86_exception *exception)
983 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
985 struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
987 return (struct kvm_mmu_page *)page_private(page);
990 static inline u16 kvm_read_ldt(void)
993 asm("sldt %0" : "=g"(ldt));
997 static inline void kvm_load_ldt(u16 sel)
999 asm("lldt %0" : : "rm"(sel));
1002 #ifdef CONFIG_X86_64
1003 static inline unsigned long read_msr(unsigned long msr)
1012 static inline u32 get_rdx_init_val(void)
1014 return 0x600; /* P6 family */
1017 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
1019 kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
1022 static inline u64 get_canonical(u64 la)
1024 return ((int64_t)la << 16) >> 16;
1027 static inline bool is_noncanonical_address(u64 la)
1029 #ifdef CONFIG_X86_64
1030 return get_canonical(la) != la;
1036 #define TSS_IOPB_BASE_OFFSET 0x66
1037 #define TSS_BASE_SIZE 0x68
1038 #define TSS_IOPB_SIZE (65536 / 8)
1039 #define TSS_REDIRECTION_SIZE (256 / 8)
1040 #define RMODE_TSS_SIZE \
1041 (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
1044 TASK_SWITCH_CALL = 0,
1045 TASK_SWITCH_IRET = 1,
1046 TASK_SWITCH_JMP = 2,
1047 TASK_SWITCH_GATE = 3,
1050 #define HF_GIF_MASK (1 << 0)
1051 #define HF_HIF_MASK (1 << 1)
1052 #define HF_VINTR_MASK (1 << 2)
1053 #define HF_NMI_MASK (1 << 3)
1054 #define HF_IRET_MASK (1 << 4)
1055 #define HF_GUEST_MASK (1 << 5) /* VCPU is in guest-mode */
1058 * Hardware virtualization extension instructions may fault if a
1059 * reboot turns off virtualization while processes are running.
1060 * Trap the fault and ignore the instruction if that happens.
1062 asmlinkage void kvm_spurious_fault(void);
1064 #define ____kvm_handle_fault_on_reboot(insn, cleanup_insn) \
1065 "666: " insn "\n\t" \
1067 ".pushsection .fixup, \"ax\" \n" \
1069 cleanup_insn "\n\t" \
1070 "cmpb $0, kvm_rebooting \n\t" \
1072 __ASM_SIZE(push) " $666b \n\t" \
1073 "call kvm_spurious_fault \n\t" \
1074 ".popsection \n\t" \
1075 _ASM_EXTABLE(666b, 667b)
1077 #define __kvm_handle_fault_on_reboot(insn) \
1078 ____kvm_handle_fault_on_reboot(insn, "")
1080 #define KVM_ARCH_WANT_MMU_NOTIFIER
1081 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
1082 int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
1083 int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
1084 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
1085 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
1086 int cpuid_maxphyaddr(struct kvm_vcpu *vcpu);
1087 int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
1088 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1089 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
1090 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
1091 void kvm_vcpu_reset(struct kvm_vcpu *vcpu);
1092 void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu);
1093 void kvm_arch_mmu_notifier_invalidate_page(struct kvm *kvm,
1094 unsigned long address);
1096 void kvm_define_shared_msr(unsigned index, u32 msr);
1097 int kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
1099 unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
1100 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1102 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1103 struct kvm_async_pf *work);
1104 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1105 struct kvm_async_pf *work);
1106 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1107 struct kvm_async_pf *work);
1108 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
1109 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1111 void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1113 int kvm_is_in_guest(void);
1115 void kvm_pmu_init(struct kvm_vcpu *vcpu);
1116 void kvm_pmu_destroy(struct kvm_vcpu *vcpu);
1117 void kvm_pmu_reset(struct kvm_vcpu *vcpu);
1118 void kvm_pmu_cpuid_update(struct kvm_vcpu *vcpu);
1119 bool kvm_pmu_msr(struct kvm_vcpu *vcpu, u32 msr);
1120 int kvm_pmu_get_msr(struct kvm_vcpu *vcpu, u32 msr, u64 *data);
1121 int kvm_pmu_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info);
1122 int kvm_pmu_check_pmc(struct kvm_vcpu *vcpu, unsigned pmc);
1123 int kvm_pmu_read_pmc(struct kvm_vcpu *vcpu, unsigned pmc, u64 *data);
1124 void kvm_handle_pmu_event(struct kvm_vcpu *vcpu);
1125 void kvm_deliver_pmi(struct kvm_vcpu *vcpu);
1127 #endif /* _ASM_X86_KVM_HOST_H */