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_PIO_PAGE_OFFSET 1
42 #define KVM_COALESCED_MMIO_PAGE_OFFSET 2
44 #define KVM_IRQCHIP_NUM_PINS KVM_IOAPIC_NUM_PINS
46 #define CR0_RESERVED_BITS \
47 (~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
48 | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
49 | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
51 #define CR3_L_MODE_RESERVED_BITS 0xFFFFFF0000000000ULL
52 #define CR3_PCID_INVD BIT_64(63)
53 #define CR4_RESERVED_BITS \
54 (~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
55 | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE \
56 | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR | X86_CR4_PCIDE \
57 | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_FSGSBASE \
58 | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE | X86_CR4_SMAP))
60 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
64 #define INVALID_PAGE (~(hpa_t)0)
65 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
67 #define UNMAPPED_GVA (~(gpa_t)0)
69 /* KVM Hugepage definitions for x86 */
70 #define KVM_NR_PAGE_SIZES 3
71 #define KVM_HPAGE_GFN_SHIFT(x) (((x) - 1) * 9)
72 #define KVM_HPAGE_SHIFT(x) (PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
73 #define KVM_HPAGE_SIZE(x) (1UL << KVM_HPAGE_SHIFT(x))
74 #define KVM_HPAGE_MASK(x) (~(KVM_HPAGE_SIZE(x) - 1))
75 #define KVM_PAGES_PER_HPAGE(x) (KVM_HPAGE_SIZE(x) / PAGE_SIZE)
77 static inline gfn_t gfn_to_index(gfn_t gfn, gfn_t base_gfn, int level)
79 /* KVM_HPAGE_GFN_SHIFT(PT_PAGE_TABLE_LEVEL) must be 0. */
80 return (gfn >> KVM_HPAGE_GFN_SHIFT(level)) -
81 (base_gfn >> KVM_HPAGE_GFN_SHIFT(level));
84 #define KVM_PERMILLE_MMU_PAGES 20
85 #define KVM_MIN_ALLOC_MMU_PAGES 64
86 #define KVM_MMU_HASH_SHIFT 10
87 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
88 #define KVM_MIN_FREE_MMU_PAGES 5
89 #define KVM_REFILL_PAGES 25
90 #define KVM_MAX_CPUID_ENTRIES 80
91 #define KVM_NR_FIXED_MTRR_REGION 88
92 #define KVM_NR_VAR_MTRR 8
94 #define ASYNC_PF_PER_VCPU 64
120 VCPU_EXREG_PDPTR = NR_VCPU_REGS,
137 #include <asm/kvm_emulate.h>
139 #define KVM_NR_MEM_OBJS 40
141 #define KVM_NR_DB_REGS 4
143 #define DR6_BD (1 << 13)
144 #define DR6_BS (1 << 14)
145 #define DR6_RTM (1 << 16)
146 #define DR6_FIXED_1 0xfffe0ff0
147 #define DR6_INIT 0xffff0ff0
148 #define DR6_VOLATILE 0x0001e00f
150 #define DR7_BP_EN_MASK 0x000000ff
151 #define DR7_GE (1 << 9)
152 #define DR7_GD (1 << 13)
153 #define DR7_FIXED_1 0x00000400
154 #define DR7_VOLATILE 0xffff2bff
156 #define PFERR_PRESENT_BIT 0
157 #define PFERR_WRITE_BIT 1
158 #define PFERR_USER_BIT 2
159 #define PFERR_RSVD_BIT 3
160 #define PFERR_FETCH_BIT 4
162 #define PFERR_PRESENT_MASK (1U << PFERR_PRESENT_BIT)
163 #define PFERR_WRITE_MASK (1U << PFERR_WRITE_BIT)
164 #define PFERR_USER_MASK (1U << PFERR_USER_BIT)
165 #define PFERR_RSVD_MASK (1U << PFERR_RSVD_BIT)
166 #define PFERR_FETCH_MASK (1U << PFERR_FETCH_BIT)
168 /* apic attention bits */
169 #define KVM_APIC_CHECK_VAPIC 0
171 * The following bit is set with PV-EOI, unset on EOI.
172 * We detect PV-EOI changes by guest by comparing
173 * this bit with PV-EOI in guest memory.
174 * See the implementation in apic_update_pv_eoi.
176 #define KVM_APIC_PV_EOI_PENDING 1
179 * We don't want allocation failures within the mmu code, so we preallocate
180 * enough memory for a single page fault in a cache.
182 struct kvm_mmu_memory_cache {
184 void *objects[KVM_NR_MEM_OBJS];
187 union kvm_mmu_page_role {
198 unsigned smep_andnot_wp:1;
199 unsigned smap_andnot_wp:1;
203 * This is left at the top of the word so that
204 * kvm_memslots_for_spte_role can extract it with a
205 * simple shift. While there is room, give it a whole
206 * byte so it is also faster to load it from memory.
212 struct kvm_mmu_page {
213 struct list_head link;
214 struct hlist_node hash_link;
217 * The following two entries are used to key the shadow page in the
221 union kvm_mmu_page_role role;
224 /* hold the gfn of each spte inside spt */
227 int root_count; /* Currently serving as active root */
228 unsigned int unsync_children;
229 unsigned long parent_ptes; /* Reverse mapping for parent_pte */
231 /* The page is obsolete if mmu_valid_gen != kvm->arch.mmu_valid_gen. */
232 unsigned long mmu_valid_gen;
234 DECLARE_BITMAP(unsync_child_bitmap, 512);
238 * Used out of the mmu-lock to avoid reading spte values while an
239 * update is in progress; see the comments in __get_spte_lockless().
241 int clear_spte_count;
244 /* Number of writes since the last time traversal visited this page. */
245 int write_flooding_count;
248 struct kvm_pio_request {
256 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
257 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
261 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
262 unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
263 u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
264 int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
266 void (*inject_page_fault)(struct kvm_vcpu *vcpu,
267 struct x86_exception *fault);
268 gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
269 struct x86_exception *exception);
270 gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
271 struct x86_exception *exception);
272 int (*sync_page)(struct kvm_vcpu *vcpu,
273 struct kvm_mmu_page *sp);
274 void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
275 void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
276 u64 *spte, const void *pte);
279 int shadow_root_level;
280 union kvm_mmu_page_role base_role;
284 * Bitmap; bit set = permission fault
285 * Byte index: page fault error code [4:1]
286 * Bit index: pte permissions in ACC_* format
292 u64 rsvd_bits_mask[2][4];
296 * Bitmap: bit set = last pte in walk
297 * index[0:1]: level (zero-based)
304 u64 pdptrs[4]; /* pae */
317 struct perf_event *perf_event;
318 struct kvm_vcpu *vcpu;
322 unsigned nr_arch_gp_counters;
323 unsigned nr_arch_fixed_counters;
324 unsigned available_event_types;
329 u64 counter_bitmask[2];
330 u64 global_ctrl_mask;
333 struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
334 struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
335 struct irq_work irq_work;
340 KVM_DEBUGREG_BP_ENABLED = 1,
341 KVM_DEBUGREG_WONT_EXIT = 2,
342 KVM_DEBUGREG_RELOAD = 4,
346 struct mtrr_var_range var_ranges[KVM_NR_VAR_MTRR];
347 mtrr_type fixed_ranges[KVM_NR_FIXED_MTRR_REGION];
348 unsigned char enabled;
352 struct kvm_vcpu_arch {
354 * rip and regs accesses must go through
355 * kvm_{register,rip}_{read,write} functions.
357 unsigned long regs[NR_VCPU_REGS];
362 unsigned long cr0_guest_owned_bits;
366 unsigned long cr4_guest_owned_bits;
371 struct kvm_lapic *apic; /* kernel irqchip context */
372 unsigned long apic_attention;
373 int32_t apic_arb_prio;
375 u64 ia32_misc_enable_msr;
377 bool tpr_access_reporting;
381 * Paging state of the vcpu
383 * If the vcpu runs in guest mode with two level paging this still saves
384 * the paging mode of the l1 guest. This context is always used to
390 * Paging state of an L2 guest (used for nested npt)
392 * This context will save all necessary information to walk page tables
393 * of the an L2 guest. This context is only initialized for page table
394 * walking and not for faulting since we never handle l2 page faults on
397 struct kvm_mmu nested_mmu;
400 * Pointer to the mmu context currently used for
401 * gva_to_gpa translations.
403 struct kvm_mmu *walk_mmu;
405 struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
406 struct kvm_mmu_memory_cache mmu_page_cache;
407 struct kvm_mmu_memory_cache mmu_page_header_cache;
409 struct fpu guest_fpu;
412 u64 guest_supported_xcr0;
413 u32 guest_xstate_size;
415 struct kvm_pio_request pio;
418 u8 event_exit_inst_len;
420 struct kvm_queued_exception {
428 struct kvm_queued_interrupt {
434 int halt_request; /* real mode on Intel only */
437 struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
441 /* emulate context */
443 struct x86_emulate_ctxt emulate_ctxt;
444 bool emulate_regs_need_sync_to_vcpu;
445 bool emulate_regs_need_sync_from_vcpu;
446 int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
449 struct pvclock_vcpu_time_info hv_clock;
450 unsigned int hw_tsc_khz;
451 struct gfn_to_hva_cache pv_time;
452 bool pv_time_enabled;
453 /* set guest stopped flag in pvclock flags field */
454 bool pvclock_set_guest_stopped_request;
460 struct gfn_to_hva_cache stime;
461 struct kvm_steal_time steal;
466 u64 tsc_offset_adjustment;
469 u64 this_tsc_generation;
471 bool tsc_always_catchup;
472 s8 virtual_tsc_shift;
473 u32 virtual_tsc_mult;
475 s64 ia32_tsc_adjust_msr;
477 atomic_t nmi_queued; /* unprocessed asynchronous NMIs */
478 unsigned nmi_pending; /* NMI queued after currently running handler */
479 bool nmi_injected; /* Trying to inject an NMI this entry */
480 bool smi_pending; /* SMI queued after currently running handler */
482 struct kvm_mtrr mtrr_state;
485 unsigned switch_db_regs;
486 unsigned long db[KVM_NR_DB_REGS];
489 unsigned long eff_db[KVM_NR_DB_REGS];
490 unsigned long guest_debug_dr7;
497 /* Cache MMIO info */
505 /* used for guest single stepping over the given code position */
506 unsigned long singlestep_rip;
508 /* fields used by HYPER-V emulation */
511 cpumask_var_t wbinvd_dirty_mask;
513 unsigned long last_retry_eip;
514 unsigned long last_retry_addr;
518 gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
519 struct gfn_to_hva_cache data;
525 /* OSVW MSRs (AMD only) */
533 struct gfn_to_hva_cache data;
537 * Indicate whether the access faults on its page table in guest
538 * which is set when fix page fault and used to detect unhandeable
541 bool write_fault_to_shadow_pgtable;
543 /* set at EPT violation at this point */
544 unsigned long exit_qualification;
546 /* pv related host specific info */
552 struct kvm_lpage_info {
556 struct kvm_arch_memory_slot {
557 unsigned long *rmap[KVM_NR_PAGE_SIZES];
558 struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
562 * We use as the mode the number of bits allocated in the LDR for the
563 * logical processor ID. It happens that these are all powers of two.
564 * This makes it is very easy to detect cases where the APICs are
565 * configured for multiple modes; in that case, we cannot use the map and
566 * hence cannot use kvm_irq_delivery_to_apic_fast either.
568 #define KVM_APIC_MODE_XAPIC_CLUSTER 4
569 #define KVM_APIC_MODE_XAPIC_FLAT 8
570 #define KVM_APIC_MODE_X2APIC 16
572 struct kvm_apic_map {
575 struct kvm_lapic *phys_map[256];
576 /* first index is cluster id second is cpu id in a cluster */
577 struct kvm_lapic *logical_map[16][16];
581 unsigned int n_used_mmu_pages;
582 unsigned int n_requested_mmu_pages;
583 unsigned int n_max_mmu_pages;
584 unsigned int indirect_shadow_pages;
585 unsigned long mmu_valid_gen;
586 struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
588 * Hash table of struct kvm_mmu_page.
590 struct list_head active_mmu_pages;
591 struct list_head zapped_obsolete_pages;
593 struct list_head assigned_dev_head;
594 struct iommu_domain *iommu_domain;
595 bool iommu_noncoherent;
596 #define __KVM_HAVE_ARCH_NONCOHERENT_DMA
597 atomic_t noncoherent_dma_count;
598 struct kvm_pic *vpic;
599 struct kvm_ioapic *vioapic;
600 struct kvm_pit *vpit;
601 int vapics_in_nmi_mode;
602 struct mutex apic_map_lock;
603 struct kvm_apic_map *apic_map;
605 unsigned int tss_addr;
606 bool apic_access_page_done;
610 bool ept_identity_pagetable_done;
611 gpa_t ept_identity_map_addr;
613 unsigned long irq_sources_bitmap;
615 raw_spinlock_t tsc_write_lock;
622 u64 cur_tsc_generation;
623 int nr_vcpus_matched_tsc;
625 spinlock_t pvclock_gtod_sync_lock;
626 bool use_master_clock;
627 u64 master_kernel_ns;
628 cycle_t master_cycle_now;
629 struct delayed_work kvmclock_update_work;
630 struct delayed_work kvmclock_sync_work;
632 struct kvm_xen_hvm_config xen_hvm_config;
634 /* reads protected by irq_srcu, writes by irq_lock */
635 struct hlist_head mask_notifier_list;
637 /* fields used by HYPER-V emulation */
642 #ifdef CONFIG_KVM_MMU_AUDIT
646 bool boot_vcpu_runs_old_kvmclock;
652 u32 mmu_shadow_zapped;
660 u32 remote_tlb_flush;
664 struct kvm_vcpu_stat {
674 u32 irq_window_exits;
675 u32 nmi_window_exits;
677 u32 halt_successful_poll;
679 u32 request_irq_exits;
681 u32 host_state_reload;
685 u32 insn_emulation_fail;
691 struct x86_instruction_info;
699 struct kvm_lapic_irq {
711 int (*cpu_has_kvm_support)(void); /* __init */
712 int (*disabled_by_bios)(void); /* __init */
713 int (*hardware_enable)(void);
714 void (*hardware_disable)(void);
715 void (*check_processor_compatibility)(void *rtn);
716 int (*hardware_setup)(void); /* __init */
717 void (*hardware_unsetup)(void); /* __exit */
718 bool (*cpu_has_accelerated_tpr)(void);
719 bool (*cpu_has_high_real_mode_segbase)(void);
720 void (*cpuid_update)(struct kvm_vcpu *vcpu);
722 /* Create, but do not attach this VCPU */
723 struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
724 void (*vcpu_free)(struct kvm_vcpu *vcpu);
725 void (*vcpu_reset)(struct kvm_vcpu *vcpu, bool init_event);
727 void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
728 void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
729 void (*vcpu_put)(struct kvm_vcpu *vcpu);
731 void (*update_db_bp_intercept)(struct kvm_vcpu *vcpu);
732 int (*get_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
733 int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
734 u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
735 void (*get_segment)(struct kvm_vcpu *vcpu,
736 struct kvm_segment *var, int seg);
737 int (*get_cpl)(struct kvm_vcpu *vcpu);
738 void (*set_segment)(struct kvm_vcpu *vcpu,
739 struct kvm_segment *var, int seg);
740 void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
741 void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
742 void (*decache_cr3)(struct kvm_vcpu *vcpu);
743 void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
744 void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
745 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
746 int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
747 void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
748 void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
749 void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
750 void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
751 void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
752 u64 (*get_dr6)(struct kvm_vcpu *vcpu);
753 void (*set_dr6)(struct kvm_vcpu *vcpu, unsigned long value);
754 void (*sync_dirty_debug_regs)(struct kvm_vcpu *vcpu);
755 void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
756 void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
757 unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
758 void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
759 void (*fpu_activate)(struct kvm_vcpu *vcpu);
760 void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
762 void (*tlb_flush)(struct kvm_vcpu *vcpu);
764 void (*run)(struct kvm_vcpu *vcpu);
765 int (*handle_exit)(struct kvm_vcpu *vcpu);
766 void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
767 void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
768 u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu);
769 void (*patch_hypercall)(struct kvm_vcpu *vcpu,
770 unsigned char *hypercall_addr);
771 void (*set_irq)(struct kvm_vcpu *vcpu);
772 void (*set_nmi)(struct kvm_vcpu *vcpu);
773 void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
774 bool has_error_code, u32 error_code,
776 void (*cancel_injection)(struct kvm_vcpu *vcpu);
777 int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
778 int (*nmi_allowed)(struct kvm_vcpu *vcpu);
779 bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
780 void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
781 void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
782 void (*enable_irq_window)(struct kvm_vcpu *vcpu);
783 void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
784 int (*vm_has_apicv)(struct kvm *kvm);
785 void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
786 void (*hwapic_isr_update)(struct kvm *kvm, int isr);
787 void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap);
788 void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu, bool set);
789 void (*set_apic_access_page_addr)(struct kvm_vcpu *vcpu, hpa_t hpa);
790 void (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
791 void (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
792 int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
793 int (*get_tdp_level)(void);
794 u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
795 int (*get_lpage_level)(void);
796 bool (*rdtscp_supported)(void);
797 bool (*invpcid_supported)(void);
798 void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment, bool host);
800 void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
802 void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
804 bool (*has_wbinvd_exit)(void);
806 void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale);
807 u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu);
808 void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
810 u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc);
811 u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu, u64 host_tsc);
813 void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
815 int (*check_intercept)(struct kvm_vcpu *vcpu,
816 struct x86_instruction_info *info,
817 enum x86_intercept_stage stage);
818 void (*handle_external_intr)(struct kvm_vcpu *vcpu);
819 bool (*mpx_supported)(void);
820 bool (*xsaves_supported)(void);
822 int (*check_nested_events)(struct kvm_vcpu *vcpu, bool external_intr);
824 void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
827 * Arch-specific dirty logging hooks. These hooks are only supposed to
828 * be valid if the specific arch has hardware-accelerated dirty logging
829 * mechanism. Currently only for PML on VMX.
831 * - slot_enable_log_dirty:
832 * called when enabling log dirty mode for the slot.
833 * - slot_disable_log_dirty:
834 * called when disabling log dirty mode for the slot.
835 * also called when slot is created with log dirty disabled.
837 * called before reporting dirty_bitmap to userspace.
838 * - enable_log_dirty_pt_masked:
839 * called when reenabling log dirty for the GFNs in the mask after
840 * corresponding bits are cleared in slot->dirty_bitmap.
842 void (*slot_enable_log_dirty)(struct kvm *kvm,
843 struct kvm_memory_slot *slot);
844 void (*slot_disable_log_dirty)(struct kvm *kvm,
845 struct kvm_memory_slot *slot);
846 void (*flush_log_dirty)(struct kvm *kvm);
847 void (*enable_log_dirty_pt_masked)(struct kvm *kvm,
848 struct kvm_memory_slot *slot,
849 gfn_t offset, unsigned long mask);
852 struct kvm_arch_async_pf {
859 extern struct kvm_x86_ops *kvm_x86_ops;
861 static inline void adjust_tsc_offset_guest(struct kvm_vcpu *vcpu,
864 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, false);
867 static inline void adjust_tsc_offset_host(struct kvm_vcpu *vcpu, s64 adjustment)
869 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, true);
872 int kvm_mmu_module_init(void);
873 void kvm_mmu_module_exit(void);
875 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
876 int kvm_mmu_create(struct kvm_vcpu *vcpu);
877 void kvm_mmu_setup(struct kvm_vcpu *vcpu);
878 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
879 u64 dirty_mask, u64 nx_mask, u64 x_mask);
881 void kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
882 void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
883 struct kvm_memory_slot *memslot);
884 void kvm_mmu_zap_collapsible_sptes(struct kvm *kvm,
885 const struct kvm_memory_slot *memslot);
886 void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
887 struct kvm_memory_slot *memslot);
888 void kvm_mmu_slot_largepage_remove_write_access(struct kvm *kvm,
889 struct kvm_memory_slot *memslot);
890 void kvm_mmu_slot_set_dirty(struct kvm *kvm,
891 struct kvm_memory_slot *memslot);
892 void kvm_mmu_clear_dirty_pt_masked(struct kvm *kvm,
893 struct kvm_memory_slot *slot,
894 gfn_t gfn_offset, unsigned long mask);
895 void kvm_mmu_zap_all(struct kvm *kvm);
896 void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm, struct kvm_memslots *slots);
897 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
898 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
900 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
902 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
903 const void *val, int bytes);
905 struct kvm_irq_mask_notifier {
906 void (*func)(struct kvm_irq_mask_notifier *kimn, bool masked);
908 struct hlist_node link;
911 void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
912 struct kvm_irq_mask_notifier *kimn);
913 void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
914 struct kvm_irq_mask_notifier *kimn);
915 void kvm_fire_mask_notifiers(struct kvm *kvm, unsigned irqchip, unsigned pin,
918 extern bool tdp_enabled;
920 u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
922 /* control of guest tsc rate supported? */
923 extern bool kvm_has_tsc_control;
924 /* minimum supported tsc_khz for guests */
925 extern u32 kvm_min_guest_tsc_khz;
926 /* maximum supported tsc_khz for guests */
927 extern u32 kvm_max_guest_tsc_khz;
929 enum emulation_result {
930 EMULATE_DONE, /* no further processing */
931 EMULATE_USER_EXIT, /* kvm_run ready for userspace exit */
932 EMULATE_FAIL, /* can't emulate this instruction */
935 #define EMULTYPE_NO_DECODE (1 << 0)
936 #define EMULTYPE_TRAP_UD (1 << 1)
937 #define EMULTYPE_SKIP (1 << 2)
938 #define EMULTYPE_RETRY (1 << 3)
939 #define EMULTYPE_NO_REEXECUTE (1 << 4)
940 int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
941 int emulation_type, void *insn, int insn_len);
943 static inline int emulate_instruction(struct kvm_vcpu *vcpu,
946 return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
949 void kvm_enable_efer_bits(u64);
950 bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
951 int kvm_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
952 int kvm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
954 struct x86_emulate_ctxt;
956 int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
957 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
958 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
959 int kvm_vcpu_halt(struct kvm_vcpu *vcpu);
960 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
962 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
963 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
964 void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
966 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
967 int reason, bool has_error_code, u32 error_code);
969 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
970 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
971 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
972 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
973 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
974 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
975 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
976 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
977 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
978 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
980 int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
981 int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
983 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
984 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
985 bool kvm_rdpmc(struct kvm_vcpu *vcpu);
987 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
988 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
989 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
990 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
991 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
992 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
993 gfn_t gfn, void *data, int offset, int len,
995 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
996 bool kvm_require_dr(struct kvm_vcpu *vcpu, int dr);
998 static inline int __kvm_irq_line_state(unsigned long *irq_state,
999 int irq_source_id, int level)
1001 /* Logical OR for level trig interrupt */
1003 __set_bit(irq_source_id, irq_state);
1005 __clear_bit(irq_source_id, irq_state);
1007 return !!(*irq_state);
1010 int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
1011 void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
1013 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
1015 int fx_init(struct kvm_vcpu *vcpu, bool init_event);
1017 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
1018 const u8 *new, int bytes);
1019 int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
1020 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
1021 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
1022 int kvm_mmu_load(struct kvm_vcpu *vcpu);
1023 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
1024 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
1025 gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1026 struct x86_exception *exception);
1027 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
1028 struct x86_exception *exception);
1029 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
1030 struct x86_exception *exception);
1031 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
1032 struct x86_exception *exception);
1033 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
1034 struct x86_exception *exception);
1036 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
1038 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
1039 void *insn, int insn_len);
1040 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
1041 void kvm_mmu_new_cr3(struct kvm_vcpu *vcpu);
1043 void kvm_enable_tdp(void);
1044 void kvm_disable_tdp(void);
1046 static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1047 struct x86_exception *exception)
1052 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
1054 struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
1056 return (struct kvm_mmu_page *)page_private(page);
1059 static inline u16 kvm_read_ldt(void)
1062 asm("sldt %0" : "=g"(ldt));
1066 static inline void kvm_load_ldt(u16 sel)
1068 asm("lldt %0" : : "rm"(sel));
1071 #ifdef CONFIG_X86_64
1072 static inline unsigned long read_msr(unsigned long msr)
1081 static inline u32 get_rdx_init_val(void)
1083 return 0x600; /* P6 family */
1086 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
1088 kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
1091 static inline u64 get_canonical(u64 la)
1093 return ((int64_t)la << 16) >> 16;
1096 static inline bool is_noncanonical_address(u64 la)
1098 #ifdef CONFIG_X86_64
1099 return get_canonical(la) != la;
1105 #define TSS_IOPB_BASE_OFFSET 0x66
1106 #define TSS_BASE_SIZE 0x68
1107 #define TSS_IOPB_SIZE (65536 / 8)
1108 #define TSS_REDIRECTION_SIZE (256 / 8)
1109 #define RMODE_TSS_SIZE \
1110 (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
1113 TASK_SWITCH_CALL = 0,
1114 TASK_SWITCH_IRET = 1,
1115 TASK_SWITCH_JMP = 2,
1116 TASK_SWITCH_GATE = 3,
1119 #define HF_GIF_MASK (1 << 0)
1120 #define HF_HIF_MASK (1 << 1)
1121 #define HF_VINTR_MASK (1 << 2)
1122 #define HF_NMI_MASK (1 << 3)
1123 #define HF_IRET_MASK (1 << 4)
1124 #define HF_GUEST_MASK (1 << 5) /* VCPU is in guest-mode */
1125 #define HF_SMM_MASK (1 << 6)
1126 #define HF_SMM_INSIDE_NMI_MASK (1 << 7)
1128 #define __KVM_VCPU_MULTIPLE_ADDRESS_SPACE
1129 #define KVM_ADDRESS_SPACE_NUM 2
1131 #define kvm_arch_vcpu_memslots_id(vcpu) ((vcpu)->arch.hflags & HF_SMM_MASK ? 1 : 0)
1132 #define kvm_memslots_for_spte_role(kvm, role) __kvm_memslots(kvm, (role).smm)
1135 * Hardware virtualization extension instructions may fault if a
1136 * reboot turns off virtualization while processes are running.
1137 * Trap the fault and ignore the instruction if that happens.
1139 asmlinkage void kvm_spurious_fault(void);
1141 #define ____kvm_handle_fault_on_reboot(insn, cleanup_insn) \
1142 "666: " insn "\n\t" \
1144 ".pushsection .fixup, \"ax\" \n" \
1146 cleanup_insn "\n\t" \
1147 "cmpb $0, kvm_rebooting \n\t" \
1149 __ASM_SIZE(push) " $666b \n\t" \
1150 "call kvm_spurious_fault \n\t" \
1151 ".popsection \n\t" \
1152 _ASM_EXTABLE(666b, 667b)
1154 #define __kvm_handle_fault_on_reboot(insn) \
1155 ____kvm_handle_fault_on_reboot(insn, "")
1157 #define KVM_ARCH_WANT_MMU_NOTIFIER
1158 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
1159 int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
1160 int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
1161 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
1162 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
1163 int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
1164 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1165 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
1166 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
1167 void kvm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event);
1168 void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu);
1169 void kvm_arch_mmu_notifier_invalidate_page(struct kvm *kvm,
1170 unsigned long address);
1172 void kvm_define_shared_msr(unsigned index, u32 msr);
1173 int kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
1175 unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
1176 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1178 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1179 struct kvm_async_pf *work);
1180 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1181 struct kvm_async_pf *work);
1182 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1183 struct kvm_async_pf *work);
1184 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
1185 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1187 void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1189 int kvm_is_in_guest(void);
1191 void kvm_pmu_init(struct kvm_vcpu *vcpu);
1192 void kvm_pmu_destroy(struct kvm_vcpu *vcpu);
1193 void kvm_pmu_reset(struct kvm_vcpu *vcpu);
1194 void kvm_pmu_cpuid_update(struct kvm_vcpu *vcpu);
1195 bool kvm_pmu_msr(struct kvm_vcpu *vcpu, u32 msr);
1196 int kvm_pmu_get_msr(struct kvm_vcpu *vcpu, u32 msr, u64 *data);
1197 int kvm_pmu_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info);
1198 int kvm_pmu_check_pmc(struct kvm_vcpu *vcpu, unsigned pmc);
1199 int kvm_pmu_read_pmc(struct kvm_vcpu *vcpu, unsigned pmc, u64 *data);
1200 void kvm_handle_pmu_event(struct kvm_vcpu *vcpu);
1201 void kvm_deliver_pmi(struct kvm_vcpu *vcpu);
1203 int __x86_set_memory_region(struct kvm *kvm,
1204 const struct kvm_userspace_memory_region *mem);
1205 int x86_set_memory_region(struct kvm *kvm,
1206 const struct kvm_userspace_memory_region *mem);
1208 #endif /* _ASM_X86_KVM_HOST_H */