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
43 #define KVM_HALT_POLL_NS_DEFAULT 500000
45 #define KVM_IRQCHIP_NUM_PINS KVM_IOAPIC_NUM_PINS
47 #define CR0_RESERVED_BITS \
48 (~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
49 | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
50 | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
52 #define CR3_L_MODE_RESERVED_BITS 0xFFFFFF0000000000ULL
53 #define CR3_PCID_INVD BIT_64(63)
54 #define CR4_RESERVED_BITS \
55 (~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
56 | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE \
57 | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR | X86_CR4_PCIDE \
58 | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_FSGSBASE \
59 | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE | X86_CR4_SMAP))
61 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
65 #define INVALID_PAGE (~(hpa_t)0)
66 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
68 #define UNMAPPED_GVA (~(gpa_t)0)
70 /* KVM Hugepage definitions for x86 */
71 #define KVM_NR_PAGE_SIZES 3
72 #define KVM_HPAGE_GFN_SHIFT(x) (((x) - 1) * 9)
73 #define KVM_HPAGE_SHIFT(x) (PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
74 #define KVM_HPAGE_SIZE(x) (1UL << KVM_HPAGE_SHIFT(x))
75 #define KVM_HPAGE_MASK(x) (~(KVM_HPAGE_SIZE(x) - 1))
76 #define KVM_PAGES_PER_HPAGE(x) (KVM_HPAGE_SIZE(x) / PAGE_SIZE)
78 static inline gfn_t gfn_to_index(gfn_t gfn, gfn_t base_gfn, int level)
80 /* KVM_HPAGE_GFN_SHIFT(PT_PAGE_TABLE_LEVEL) must be 0. */
81 return (gfn >> KVM_HPAGE_GFN_SHIFT(level)) -
82 (base_gfn >> KVM_HPAGE_GFN_SHIFT(level));
85 #define KVM_PERMILLE_MMU_PAGES 20
86 #define KVM_MIN_ALLOC_MMU_PAGES 64
87 #define KVM_MMU_HASH_SHIFT 10
88 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
89 #define KVM_MIN_FREE_MMU_PAGES 5
90 #define KVM_REFILL_PAGES 25
91 #define KVM_MAX_CPUID_ENTRIES 80
92 #define KVM_NR_FIXED_MTRR_REGION 88
93 #define KVM_NR_VAR_MTRR 8
95 #define ASYNC_PF_PER_VCPU 64
121 VCPU_EXREG_PDPTR = NR_VCPU_REGS,
138 #include <asm/kvm_emulate.h>
140 #define KVM_NR_MEM_OBJS 40
142 #define KVM_NR_DB_REGS 4
144 #define DR6_BD (1 << 13)
145 #define DR6_BS (1 << 14)
146 #define DR6_RTM (1 << 16)
147 #define DR6_FIXED_1 0xfffe0ff0
148 #define DR6_INIT 0xffff0ff0
149 #define DR6_VOLATILE 0x0001e00f
151 #define DR7_BP_EN_MASK 0x000000ff
152 #define DR7_GE (1 << 9)
153 #define DR7_GD (1 << 13)
154 #define DR7_FIXED_1 0x00000400
155 #define DR7_VOLATILE 0xffff2bff
157 #define PFERR_PRESENT_BIT 0
158 #define PFERR_WRITE_BIT 1
159 #define PFERR_USER_BIT 2
160 #define PFERR_RSVD_BIT 3
161 #define PFERR_FETCH_BIT 4
163 #define PFERR_PRESENT_MASK (1U << PFERR_PRESENT_BIT)
164 #define PFERR_WRITE_MASK (1U << PFERR_WRITE_BIT)
165 #define PFERR_USER_MASK (1U << PFERR_USER_BIT)
166 #define PFERR_RSVD_MASK (1U << PFERR_RSVD_BIT)
167 #define PFERR_FETCH_MASK (1U << PFERR_FETCH_BIT)
169 /* apic attention bits */
170 #define KVM_APIC_CHECK_VAPIC 0
172 * The following bit is set with PV-EOI, unset on EOI.
173 * We detect PV-EOI changes by guest by comparing
174 * this bit with PV-EOI in guest memory.
175 * See the implementation in apic_update_pv_eoi.
177 #define KVM_APIC_PV_EOI_PENDING 1
180 * We don't want allocation failures within the mmu code, so we preallocate
181 * enough memory for a single page fault in a cache.
183 struct kvm_mmu_memory_cache {
185 void *objects[KVM_NR_MEM_OBJS];
188 union kvm_mmu_page_role {
199 unsigned smep_andnot_wp:1;
200 unsigned smap_andnot_wp:1;
204 * This is left at the top of the word so that
205 * kvm_memslots_for_spte_role can extract it with a
206 * simple shift. While there is room, give it a whole
207 * byte so it is also faster to load it from memory.
213 struct kvm_mmu_page {
214 struct list_head link;
215 struct hlist_node hash_link;
218 * The following two entries are used to key the shadow page in the
222 union kvm_mmu_page_role role;
225 /* hold the gfn of each spte inside spt */
228 int root_count; /* Currently serving as active root */
229 unsigned int unsync_children;
230 unsigned long parent_ptes; /* Reverse mapping for parent_pte */
232 /* The page is obsolete if mmu_valid_gen != kvm->arch.mmu_valid_gen. */
233 unsigned long mmu_valid_gen;
235 DECLARE_BITMAP(unsync_child_bitmap, 512);
239 * Used out of the mmu-lock to avoid reading spte values while an
240 * update is in progress; see the comments in __get_spte_lockless().
242 int clear_spte_count;
245 /* Number of writes since the last time traversal visited this page. */
246 int write_flooding_count;
249 struct kvm_pio_request {
256 struct rsvd_bits_validate {
257 u64 rsvd_bits_mask[2][4];
262 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
263 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
267 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
268 unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
269 u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
270 int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
272 void (*inject_page_fault)(struct kvm_vcpu *vcpu,
273 struct x86_exception *fault);
274 gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
275 struct x86_exception *exception);
276 gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
277 struct x86_exception *exception);
278 int (*sync_page)(struct kvm_vcpu *vcpu,
279 struct kvm_mmu_page *sp);
280 void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
281 void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
282 u64 *spte, const void *pte);
285 int shadow_root_level;
286 union kvm_mmu_page_role base_role;
290 * Bitmap; bit set = permission fault
291 * Byte index: page fault error code [4:1]
292 * Bit index: pte permissions in ACC_* format
300 * check zero bits on shadow page table entries, these
301 * bits include not only hardware reserved bits but also
302 * the bits spte never used.
304 struct rsvd_bits_validate shadow_zero_check;
306 struct rsvd_bits_validate guest_rsvd_check;
309 * Bitmap: bit set = last pte in walk
310 * index[0:1]: level (zero-based)
317 u64 pdptrs[4]; /* pae */
330 struct perf_event *perf_event;
331 struct kvm_vcpu *vcpu;
335 unsigned nr_arch_gp_counters;
336 unsigned nr_arch_fixed_counters;
337 unsigned available_event_types;
342 u64 counter_bitmask[2];
343 u64 global_ctrl_mask;
346 struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
347 struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
348 struct irq_work irq_work;
355 KVM_DEBUGREG_BP_ENABLED = 1,
356 KVM_DEBUGREG_WONT_EXIT = 2,
357 KVM_DEBUGREG_RELOAD = 4,
360 struct kvm_mtrr_range {
363 struct list_head node;
367 struct kvm_mtrr_range var_ranges[KVM_NR_VAR_MTRR];
368 mtrr_type fixed_ranges[KVM_NR_FIXED_MTRR_REGION];
371 struct list_head head;
374 /* Hyper-V per vcpu emulation context */
379 struct kvm_vcpu_arch {
381 * rip and regs accesses must go through
382 * kvm_{register,rip}_{read,write} functions.
384 unsigned long regs[NR_VCPU_REGS];
389 unsigned long cr0_guest_owned_bits;
393 unsigned long cr4_guest_owned_bits;
398 struct kvm_lapic *apic; /* kernel irqchip context */
399 unsigned long apic_attention;
400 int32_t apic_arb_prio;
402 u64 ia32_misc_enable_msr;
404 bool tpr_access_reporting;
408 * Paging state of the vcpu
410 * If the vcpu runs in guest mode with two level paging this still saves
411 * the paging mode of the l1 guest. This context is always used to
417 * Paging state of an L2 guest (used for nested npt)
419 * This context will save all necessary information to walk page tables
420 * of the an L2 guest. This context is only initialized for page table
421 * walking and not for faulting since we never handle l2 page faults on
424 struct kvm_mmu nested_mmu;
427 * Pointer to the mmu context currently used for
428 * gva_to_gpa translations.
430 struct kvm_mmu *walk_mmu;
432 struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
433 struct kvm_mmu_memory_cache mmu_page_cache;
434 struct kvm_mmu_memory_cache mmu_page_header_cache;
436 struct fpu guest_fpu;
439 u64 guest_supported_xcr0;
440 u32 guest_xstate_size;
442 struct kvm_pio_request pio;
445 u8 event_exit_inst_len;
447 struct kvm_queued_exception {
455 struct kvm_queued_interrupt {
461 int halt_request; /* real mode on Intel only */
464 struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
468 /* emulate context */
470 struct x86_emulate_ctxt emulate_ctxt;
471 bool emulate_regs_need_sync_to_vcpu;
472 bool emulate_regs_need_sync_from_vcpu;
473 int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
476 struct pvclock_vcpu_time_info hv_clock;
477 unsigned int hw_tsc_khz;
478 struct gfn_to_hva_cache pv_time;
479 bool pv_time_enabled;
480 /* set guest stopped flag in pvclock flags field */
481 bool pvclock_set_guest_stopped_request;
487 struct gfn_to_hva_cache stime;
488 struct kvm_steal_time steal;
493 u64 tsc_offset_adjustment;
496 u64 this_tsc_generation;
498 bool tsc_always_catchup;
499 s8 virtual_tsc_shift;
500 u32 virtual_tsc_mult;
502 s64 ia32_tsc_adjust_msr;
504 atomic_t nmi_queued; /* unprocessed asynchronous NMIs */
505 unsigned nmi_pending; /* NMI queued after currently running handler */
506 bool nmi_injected; /* Trying to inject an NMI this entry */
507 bool smi_pending; /* SMI queued after currently running handler */
509 struct kvm_mtrr mtrr_state;
512 unsigned switch_db_regs;
513 unsigned long db[KVM_NR_DB_REGS];
516 unsigned long eff_db[KVM_NR_DB_REGS];
517 unsigned long guest_debug_dr7;
524 /* Cache MMIO info */
532 /* used for guest single stepping over the given code position */
533 unsigned long singlestep_rip;
535 struct kvm_vcpu_hv hyperv;
537 cpumask_var_t wbinvd_dirty_mask;
539 unsigned long last_retry_eip;
540 unsigned long last_retry_addr;
544 gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
545 struct gfn_to_hva_cache data;
551 /* OSVW MSRs (AMD only) */
559 struct gfn_to_hva_cache data;
563 * Indicate whether the access faults on its page table in guest
564 * which is set when fix page fault and used to detect unhandeable
567 bool write_fault_to_shadow_pgtable;
569 /* set at EPT violation at this point */
570 unsigned long exit_qualification;
572 /* pv related host specific info */
578 struct kvm_lpage_info {
582 struct kvm_arch_memory_slot {
583 unsigned long *rmap[KVM_NR_PAGE_SIZES];
584 struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
588 * We use as the mode the number of bits allocated in the LDR for the
589 * logical processor ID. It happens that these are all powers of two.
590 * This makes it is very easy to detect cases where the APICs are
591 * configured for multiple modes; in that case, we cannot use the map and
592 * hence cannot use kvm_irq_delivery_to_apic_fast either.
594 #define KVM_APIC_MODE_XAPIC_CLUSTER 4
595 #define KVM_APIC_MODE_XAPIC_FLAT 8
596 #define KVM_APIC_MODE_X2APIC 16
598 struct kvm_apic_map {
601 struct kvm_lapic *phys_map[256];
602 /* first index is cluster id second is cpu id in a cluster */
603 struct kvm_lapic *logical_map[16][16];
606 /* Hyper-V emulation context */
612 /* Hyper-v based guest crash (NT kernel bugcheck) parameters */
613 u64 hv_crash_param[HV_X64_MSR_CRASH_PARAMS];
618 unsigned int n_used_mmu_pages;
619 unsigned int n_requested_mmu_pages;
620 unsigned int n_max_mmu_pages;
621 unsigned int indirect_shadow_pages;
622 unsigned long mmu_valid_gen;
623 struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
625 * Hash table of struct kvm_mmu_page.
627 struct list_head active_mmu_pages;
628 struct list_head zapped_obsolete_pages;
630 struct list_head assigned_dev_head;
631 struct iommu_domain *iommu_domain;
632 bool iommu_noncoherent;
633 #define __KVM_HAVE_ARCH_NONCOHERENT_DMA
634 atomic_t noncoherent_dma_count;
635 #define __KVM_HAVE_ARCH_ASSIGNED_DEVICE
636 atomic_t assigned_device_count;
637 struct kvm_pic *vpic;
638 struct kvm_ioapic *vioapic;
639 struct kvm_pit *vpit;
640 atomic_t vapics_in_nmi_mode;
641 struct mutex apic_map_lock;
642 struct kvm_apic_map *apic_map;
644 unsigned int tss_addr;
645 bool apic_access_page_done;
649 bool ept_identity_pagetable_done;
650 gpa_t ept_identity_map_addr;
652 unsigned long irq_sources_bitmap;
654 raw_spinlock_t tsc_write_lock;
661 u64 cur_tsc_generation;
662 int nr_vcpus_matched_tsc;
664 spinlock_t pvclock_gtod_sync_lock;
665 bool use_master_clock;
666 u64 master_kernel_ns;
667 cycle_t master_cycle_now;
668 struct delayed_work kvmclock_update_work;
669 struct delayed_work kvmclock_sync_work;
671 struct kvm_xen_hvm_config xen_hvm_config;
673 /* reads protected by irq_srcu, writes by irq_lock */
674 struct hlist_head mask_notifier_list;
676 struct kvm_hv hyperv;
678 #ifdef CONFIG_KVM_MMU_AUDIT
682 bool boot_vcpu_runs_old_kvmclock;
689 u32 mmu_shadow_zapped;
697 u32 remote_tlb_flush;
701 struct kvm_vcpu_stat {
711 u32 irq_window_exits;
712 u32 nmi_window_exits;
714 u32 halt_successful_poll;
715 u32 halt_attempted_poll;
717 u32 request_irq_exits;
719 u32 host_state_reload;
723 u32 insn_emulation_fail;
729 struct x86_instruction_info;
737 struct kvm_lapic_irq {
749 int (*cpu_has_kvm_support)(void); /* __init */
750 int (*disabled_by_bios)(void); /* __init */
751 int (*hardware_enable)(void);
752 void (*hardware_disable)(void);
753 void (*check_processor_compatibility)(void *rtn);
754 int (*hardware_setup)(void); /* __init */
755 void (*hardware_unsetup)(void); /* __exit */
756 bool (*cpu_has_accelerated_tpr)(void);
757 bool (*cpu_has_high_real_mode_segbase)(void);
758 void (*cpuid_update)(struct kvm_vcpu *vcpu);
760 /* Create, but do not attach this VCPU */
761 struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
762 void (*vcpu_free)(struct kvm_vcpu *vcpu);
763 void (*vcpu_reset)(struct kvm_vcpu *vcpu, bool init_event);
765 void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
766 void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
767 void (*vcpu_put)(struct kvm_vcpu *vcpu);
769 void (*update_db_bp_intercept)(struct kvm_vcpu *vcpu);
770 int (*get_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
771 int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
772 u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
773 void (*get_segment)(struct kvm_vcpu *vcpu,
774 struct kvm_segment *var, int seg);
775 int (*get_cpl)(struct kvm_vcpu *vcpu);
776 void (*set_segment)(struct kvm_vcpu *vcpu,
777 struct kvm_segment *var, int seg);
778 void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
779 void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
780 void (*decache_cr3)(struct kvm_vcpu *vcpu);
781 void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
782 void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
783 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
784 int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
785 void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
786 void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
787 void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
788 void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
789 void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
790 u64 (*get_dr6)(struct kvm_vcpu *vcpu);
791 void (*set_dr6)(struct kvm_vcpu *vcpu, unsigned long value);
792 void (*sync_dirty_debug_regs)(struct kvm_vcpu *vcpu);
793 void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
794 void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
795 unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
796 void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
797 void (*fpu_activate)(struct kvm_vcpu *vcpu);
798 void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
800 void (*tlb_flush)(struct kvm_vcpu *vcpu);
802 void (*run)(struct kvm_vcpu *vcpu);
803 int (*handle_exit)(struct kvm_vcpu *vcpu);
804 void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
805 void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
806 u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu);
807 void (*patch_hypercall)(struct kvm_vcpu *vcpu,
808 unsigned char *hypercall_addr);
809 void (*set_irq)(struct kvm_vcpu *vcpu);
810 void (*set_nmi)(struct kvm_vcpu *vcpu);
811 void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
812 bool has_error_code, u32 error_code,
814 void (*cancel_injection)(struct kvm_vcpu *vcpu);
815 int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
816 int (*nmi_allowed)(struct kvm_vcpu *vcpu);
817 bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
818 void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
819 void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
820 void (*enable_irq_window)(struct kvm_vcpu *vcpu);
821 void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
822 int (*vm_has_apicv)(struct kvm *kvm);
823 void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
824 void (*hwapic_isr_update)(struct kvm *kvm, int isr);
825 void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap);
826 void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu, bool set);
827 void (*set_apic_access_page_addr)(struct kvm_vcpu *vcpu, hpa_t hpa);
828 void (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
829 void (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
830 int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
831 int (*get_tdp_level)(void);
832 u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
833 int (*get_lpage_level)(void);
834 bool (*rdtscp_supported)(void);
835 bool (*invpcid_supported)(void);
836 void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment, bool host);
838 void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
840 void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
842 bool (*has_wbinvd_exit)(void);
844 void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale);
845 u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu);
846 void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
848 u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc);
849 u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu, u64 host_tsc);
851 void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
853 int (*check_intercept)(struct kvm_vcpu *vcpu,
854 struct x86_instruction_info *info,
855 enum x86_intercept_stage stage);
856 void (*handle_external_intr)(struct kvm_vcpu *vcpu);
857 bool (*mpx_supported)(void);
858 bool (*xsaves_supported)(void);
860 int (*check_nested_events)(struct kvm_vcpu *vcpu, bool external_intr);
862 void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
865 * Arch-specific dirty logging hooks. These hooks are only supposed to
866 * be valid if the specific arch has hardware-accelerated dirty logging
867 * mechanism. Currently only for PML on VMX.
869 * - slot_enable_log_dirty:
870 * called when enabling log dirty mode for the slot.
871 * - slot_disable_log_dirty:
872 * called when disabling log dirty mode for the slot.
873 * also called when slot is created with log dirty disabled.
875 * called before reporting dirty_bitmap to userspace.
876 * - enable_log_dirty_pt_masked:
877 * called when reenabling log dirty for the GFNs in the mask after
878 * corresponding bits are cleared in slot->dirty_bitmap.
880 void (*slot_enable_log_dirty)(struct kvm *kvm,
881 struct kvm_memory_slot *slot);
882 void (*slot_disable_log_dirty)(struct kvm *kvm,
883 struct kvm_memory_slot *slot);
884 void (*flush_log_dirty)(struct kvm *kvm);
885 void (*enable_log_dirty_pt_masked)(struct kvm *kvm,
886 struct kvm_memory_slot *slot,
887 gfn_t offset, unsigned long mask);
888 /* pmu operations of sub-arch */
889 const struct kvm_pmu_ops *pmu_ops;
892 struct kvm_arch_async_pf {
899 extern struct kvm_x86_ops *kvm_x86_ops;
901 static inline void adjust_tsc_offset_guest(struct kvm_vcpu *vcpu,
904 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, false);
907 static inline void adjust_tsc_offset_host(struct kvm_vcpu *vcpu, s64 adjustment)
909 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, true);
912 int kvm_mmu_module_init(void);
913 void kvm_mmu_module_exit(void);
915 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
916 int kvm_mmu_create(struct kvm_vcpu *vcpu);
917 void kvm_mmu_setup(struct kvm_vcpu *vcpu);
918 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
919 u64 dirty_mask, u64 nx_mask, u64 x_mask);
921 void kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
922 void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
923 struct kvm_memory_slot *memslot);
924 void kvm_mmu_zap_collapsible_sptes(struct kvm *kvm,
925 const struct kvm_memory_slot *memslot);
926 void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
927 struct kvm_memory_slot *memslot);
928 void kvm_mmu_slot_largepage_remove_write_access(struct kvm *kvm,
929 struct kvm_memory_slot *memslot);
930 void kvm_mmu_slot_set_dirty(struct kvm *kvm,
931 struct kvm_memory_slot *memslot);
932 void kvm_mmu_clear_dirty_pt_masked(struct kvm *kvm,
933 struct kvm_memory_slot *slot,
934 gfn_t gfn_offset, unsigned long mask);
935 void kvm_mmu_zap_all(struct kvm *kvm);
936 void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm, struct kvm_memslots *slots);
937 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
938 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
940 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
942 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
943 const void *val, int bytes);
945 struct kvm_irq_mask_notifier {
946 void (*func)(struct kvm_irq_mask_notifier *kimn, bool masked);
948 struct hlist_node link;
951 void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
952 struct kvm_irq_mask_notifier *kimn);
953 void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
954 struct kvm_irq_mask_notifier *kimn);
955 void kvm_fire_mask_notifiers(struct kvm *kvm, unsigned irqchip, unsigned pin,
958 extern bool tdp_enabled;
960 u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
962 /* control of guest tsc rate supported? */
963 extern bool kvm_has_tsc_control;
964 /* minimum supported tsc_khz for guests */
965 extern u32 kvm_min_guest_tsc_khz;
966 /* maximum supported tsc_khz for guests */
967 extern u32 kvm_max_guest_tsc_khz;
969 enum emulation_result {
970 EMULATE_DONE, /* no further processing */
971 EMULATE_USER_EXIT, /* kvm_run ready for userspace exit */
972 EMULATE_FAIL, /* can't emulate this instruction */
975 #define EMULTYPE_NO_DECODE (1 << 0)
976 #define EMULTYPE_TRAP_UD (1 << 1)
977 #define EMULTYPE_SKIP (1 << 2)
978 #define EMULTYPE_RETRY (1 << 3)
979 #define EMULTYPE_NO_REEXECUTE (1 << 4)
980 int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
981 int emulation_type, void *insn, int insn_len);
983 static inline int emulate_instruction(struct kvm_vcpu *vcpu,
986 return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
989 void kvm_enable_efer_bits(u64);
990 bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
991 int kvm_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
992 int kvm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
994 struct x86_emulate_ctxt;
996 int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
997 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
998 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
999 int kvm_vcpu_halt(struct kvm_vcpu *vcpu);
1000 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
1002 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
1003 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
1004 void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
1006 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
1007 int reason, bool has_error_code, u32 error_code);
1009 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
1010 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
1011 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
1012 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
1013 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
1014 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
1015 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
1016 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
1017 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
1018 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
1020 int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
1021 int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
1023 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
1024 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
1025 bool kvm_rdpmc(struct kvm_vcpu *vcpu);
1027 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1028 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
1029 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1030 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
1031 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
1032 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
1033 gfn_t gfn, void *data, int offset, int len,
1035 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
1036 bool kvm_require_dr(struct kvm_vcpu *vcpu, int dr);
1038 static inline int __kvm_irq_line_state(unsigned long *irq_state,
1039 int irq_source_id, int level)
1041 /* Logical OR for level trig interrupt */
1043 __set_bit(irq_source_id, irq_state);
1045 __clear_bit(irq_source_id, irq_state);
1047 return !!(*irq_state);
1050 int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
1051 void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
1053 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
1055 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
1056 const u8 *new, int bytes);
1057 int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
1058 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
1059 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
1060 int kvm_mmu_load(struct kvm_vcpu *vcpu);
1061 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
1062 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
1063 gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1064 struct x86_exception *exception);
1065 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
1066 struct x86_exception *exception);
1067 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
1068 struct x86_exception *exception);
1069 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
1070 struct x86_exception *exception);
1071 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
1072 struct x86_exception *exception);
1074 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
1076 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
1077 void *insn, int insn_len);
1078 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
1079 void kvm_mmu_new_cr3(struct kvm_vcpu *vcpu);
1081 void kvm_enable_tdp(void);
1082 void kvm_disable_tdp(void);
1084 static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1085 struct x86_exception *exception)
1090 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
1092 struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
1094 return (struct kvm_mmu_page *)page_private(page);
1097 static inline u16 kvm_read_ldt(void)
1100 asm("sldt %0" : "=g"(ldt));
1104 static inline void kvm_load_ldt(u16 sel)
1106 asm("lldt %0" : : "rm"(sel));
1109 #ifdef CONFIG_X86_64
1110 static inline unsigned long read_msr(unsigned long msr)
1119 static inline u32 get_rdx_init_val(void)
1121 return 0x600; /* P6 family */
1124 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
1126 kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
1129 static inline u64 get_canonical(u64 la)
1131 return ((int64_t)la << 16) >> 16;
1134 static inline bool is_noncanonical_address(u64 la)
1136 #ifdef CONFIG_X86_64
1137 return get_canonical(la) != la;
1143 #define TSS_IOPB_BASE_OFFSET 0x66
1144 #define TSS_BASE_SIZE 0x68
1145 #define TSS_IOPB_SIZE (65536 / 8)
1146 #define TSS_REDIRECTION_SIZE (256 / 8)
1147 #define RMODE_TSS_SIZE \
1148 (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
1151 TASK_SWITCH_CALL = 0,
1152 TASK_SWITCH_IRET = 1,
1153 TASK_SWITCH_JMP = 2,
1154 TASK_SWITCH_GATE = 3,
1157 #define HF_GIF_MASK (1 << 0)
1158 #define HF_HIF_MASK (1 << 1)
1159 #define HF_VINTR_MASK (1 << 2)
1160 #define HF_NMI_MASK (1 << 3)
1161 #define HF_IRET_MASK (1 << 4)
1162 #define HF_GUEST_MASK (1 << 5) /* VCPU is in guest-mode */
1163 #define HF_SMM_MASK (1 << 6)
1164 #define HF_SMM_INSIDE_NMI_MASK (1 << 7)
1166 #define __KVM_VCPU_MULTIPLE_ADDRESS_SPACE
1167 #define KVM_ADDRESS_SPACE_NUM 2
1169 #define kvm_arch_vcpu_memslots_id(vcpu) ((vcpu)->arch.hflags & HF_SMM_MASK ? 1 : 0)
1170 #define kvm_memslots_for_spte_role(kvm, role) __kvm_memslots(kvm, (role).smm)
1173 * Hardware virtualization extension instructions may fault if a
1174 * reboot turns off virtualization while processes are running.
1175 * Trap the fault and ignore the instruction if that happens.
1177 asmlinkage void kvm_spurious_fault(void);
1179 #define ____kvm_handle_fault_on_reboot(insn, cleanup_insn) \
1180 "666: " insn "\n\t" \
1182 ".pushsection .fixup, \"ax\" \n" \
1184 cleanup_insn "\n\t" \
1185 "cmpb $0, kvm_rebooting \n\t" \
1187 __ASM_SIZE(push) " $666b \n\t" \
1188 "call kvm_spurious_fault \n\t" \
1189 ".popsection \n\t" \
1190 _ASM_EXTABLE(666b, 667b)
1192 #define __kvm_handle_fault_on_reboot(insn) \
1193 ____kvm_handle_fault_on_reboot(insn, "")
1195 #define KVM_ARCH_WANT_MMU_NOTIFIER
1196 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
1197 int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
1198 int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
1199 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
1200 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
1201 int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
1202 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1203 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
1204 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
1205 void kvm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event);
1206 void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu);
1207 void kvm_arch_mmu_notifier_invalidate_page(struct kvm *kvm,
1208 unsigned long address);
1210 void kvm_define_shared_msr(unsigned index, u32 msr);
1211 int kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
1213 unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
1214 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1216 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1217 struct kvm_async_pf *work);
1218 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1219 struct kvm_async_pf *work);
1220 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1221 struct kvm_async_pf *work);
1222 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
1223 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1225 void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1227 int kvm_is_in_guest(void);
1229 int __x86_set_memory_region(struct kvm *kvm,
1230 const struct kvm_userspace_memory_region *mem);
1231 int x86_set_memory_region(struct kvm *kvm,
1232 const struct kvm_userspace_memory_region *mem);
1233 bool kvm_vcpu_is_reset_bsp(struct kvm_vcpu *vcpu);
1234 bool kvm_vcpu_is_bsp(struct kvm_vcpu *vcpu);
1236 #endif /* _ASM_X86_KVM_HOST_H */