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 SELECTOR_TI_MASK (1 << 2)
85 #define SELECTOR_RPL_MASK 0x03
87 #define KVM_PERMILLE_MMU_PAGES 20
88 #define KVM_MIN_ALLOC_MMU_PAGES 64
89 #define KVM_MMU_HASH_SHIFT 10
90 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
91 #define KVM_MIN_FREE_MMU_PAGES 5
92 #define KVM_REFILL_PAGES 25
93 #define KVM_MAX_CPUID_ENTRIES 80
94 #define KVM_NR_FIXED_MTRR_REGION 88
95 #define KVM_NR_VAR_MTRR 8
97 #define ASYNC_PF_PER_VCPU 64
123 VCPU_EXREG_PDPTR = NR_VCPU_REGS,
140 #include <asm/kvm_emulate.h>
142 #define KVM_NR_MEM_OBJS 40
144 #define KVM_NR_DB_REGS 4
146 #define DR6_BD (1 << 13)
147 #define DR6_BS (1 << 14)
148 #define DR6_RTM (1 << 16)
149 #define DR6_FIXED_1 0xfffe0ff0
150 #define DR6_INIT 0xffff0ff0
151 #define DR6_VOLATILE 0x0001e00f
153 #define DR7_BP_EN_MASK 0x000000ff
154 #define DR7_GE (1 << 9)
155 #define DR7_GD (1 << 13)
156 #define DR7_FIXED_1 0x00000400
157 #define DR7_VOLATILE 0xffff2bff
159 #define PFERR_PRESENT_BIT 0
160 #define PFERR_WRITE_BIT 1
161 #define PFERR_USER_BIT 2
162 #define PFERR_RSVD_BIT 3
163 #define PFERR_FETCH_BIT 4
165 #define PFERR_PRESENT_MASK (1U << PFERR_PRESENT_BIT)
166 #define PFERR_WRITE_MASK (1U << PFERR_WRITE_BIT)
167 #define PFERR_USER_MASK (1U << PFERR_USER_BIT)
168 #define PFERR_RSVD_MASK (1U << PFERR_RSVD_BIT)
169 #define PFERR_FETCH_MASK (1U << PFERR_FETCH_BIT)
171 /* apic attention bits */
172 #define KVM_APIC_CHECK_VAPIC 0
174 * The following bit is set with PV-EOI, unset on EOI.
175 * We detect PV-EOI changes by guest by comparing
176 * this bit with PV-EOI in guest memory.
177 * See the implementation in apic_update_pv_eoi.
179 #define KVM_APIC_PV_EOI_PENDING 1
182 * We don't want allocation failures within the mmu code, so we preallocate
183 * enough memory for a single page fault in a cache.
185 struct kvm_mmu_memory_cache {
187 void *objects[KVM_NR_MEM_OBJS];
191 * kvm_mmu_page_role, below, is defined as:
193 * bits 0:3 - total guest paging levels (2-4, or zero for real mode)
194 * bits 4:7 - page table level for this shadow (1-4)
195 * bits 8:9 - page table quadrant for 2-level guests
196 * bit 16 - direct mapping of virtual to physical mapping at gfn
197 * used for real mode and two-dimensional paging
198 * bits 17:19 - common access permissions for all ptes in this shadow page
200 union kvm_mmu_page_role {
206 unsigned pad_for_nice_hex_output:6;
212 unsigned smep_andnot_wp:1;
216 struct kvm_mmu_page {
217 struct list_head link;
218 struct hlist_node hash_link;
221 * The following two entries are used to key the shadow page in the
225 union kvm_mmu_page_role role;
228 /* hold the gfn of each spte inside spt */
231 int root_count; /* Currently serving as active root */
232 unsigned int unsync_children;
233 unsigned long parent_ptes; /* Reverse mapping for parent_pte */
235 /* The page is obsolete if mmu_valid_gen != kvm->arch.mmu_valid_gen. */
236 unsigned long mmu_valid_gen;
238 DECLARE_BITMAP(unsync_child_bitmap, 512);
242 * Used out of the mmu-lock to avoid reading spte values while an
243 * update is in progress; see the comments in __get_spte_lockless().
245 int clear_spte_count;
248 /* Number of writes since the last time traversal visited this page. */
249 int write_flooding_count;
252 struct kvm_pio_request {
260 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
261 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
265 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
266 unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
267 u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
268 int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
270 void (*inject_page_fault)(struct kvm_vcpu *vcpu,
271 struct x86_exception *fault);
272 gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
273 struct x86_exception *exception);
274 gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
275 struct x86_exception *exception);
276 int (*sync_page)(struct kvm_vcpu *vcpu,
277 struct kvm_mmu_page *sp);
278 void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
279 void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
280 u64 *spte, const void *pte);
283 int shadow_root_level;
284 union kvm_mmu_page_role base_role;
288 * Bitmap; bit set = permission fault
289 * Byte index: page fault error code [4:1]
290 * Bit index: pte permissions in ACC_* format
296 u64 rsvd_bits_mask[2][4];
300 * Bitmap: bit set = last pte in walk
301 * index[0:1]: level (zero-based)
308 u64 pdptrs[4]; /* pae */
321 struct perf_event *perf_event;
322 struct kvm_vcpu *vcpu;
326 unsigned nr_arch_gp_counters;
327 unsigned nr_arch_fixed_counters;
328 unsigned available_event_types;
333 u64 counter_bitmask[2];
334 u64 global_ctrl_mask;
337 struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
338 struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
339 struct irq_work irq_work;
344 KVM_DEBUGREG_BP_ENABLED = 1,
345 KVM_DEBUGREG_WONT_EXIT = 2,
348 struct kvm_vcpu_arch {
350 * rip and regs accesses must go through
351 * kvm_{register,rip}_{read,write} functions.
353 unsigned long regs[NR_VCPU_REGS];
358 unsigned long cr0_guest_owned_bits;
362 unsigned long cr4_guest_owned_bits;
367 struct kvm_lapic *apic; /* kernel irqchip context */
368 unsigned long apic_attention;
369 int32_t apic_arb_prio;
371 u64 ia32_misc_enable_msr;
372 bool tpr_access_reporting;
376 * Paging state of the vcpu
378 * If the vcpu runs in guest mode with two level paging this still saves
379 * the paging mode of the l1 guest. This context is always used to
385 * Paging state of an L2 guest (used for nested npt)
387 * This context will save all necessary information to walk page tables
388 * of the an L2 guest. This context is only initialized for page table
389 * walking and not for faulting since we never handle l2 page faults on
392 struct kvm_mmu nested_mmu;
395 * Pointer to the mmu context currently used for
396 * gva_to_gpa translations.
398 struct kvm_mmu *walk_mmu;
400 struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
401 struct kvm_mmu_memory_cache mmu_page_cache;
402 struct kvm_mmu_memory_cache mmu_page_header_cache;
404 struct fpu guest_fpu;
406 u64 guest_supported_xcr0;
407 u32 guest_xstate_size;
409 struct kvm_pio_request pio;
412 u8 event_exit_inst_len;
414 struct kvm_queued_exception {
422 struct kvm_queued_interrupt {
428 int halt_request; /* real mode on Intel only */
431 struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
432 /* emulate context */
434 struct x86_emulate_ctxt emulate_ctxt;
435 bool emulate_regs_need_sync_to_vcpu;
436 bool emulate_regs_need_sync_from_vcpu;
437 int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
440 struct pvclock_vcpu_time_info hv_clock;
441 unsigned int hw_tsc_khz;
442 struct gfn_to_hva_cache pv_time;
443 bool pv_time_enabled;
444 /* set guest stopped flag in pvclock flags field */
445 bool pvclock_set_guest_stopped_request;
451 struct gfn_to_hva_cache stime;
452 struct kvm_steal_time steal;
457 u64 tsc_offset_adjustment;
460 u64 this_tsc_generation;
462 bool tsc_always_catchup;
463 s8 virtual_tsc_shift;
464 u32 virtual_tsc_mult;
466 s64 ia32_tsc_adjust_msr;
468 atomic_t nmi_queued; /* unprocessed asynchronous NMIs */
469 unsigned nmi_pending; /* NMI queued after currently running handler */
470 bool nmi_injected; /* Trying to inject an NMI this entry */
472 struct mtrr_state_type mtrr_state;
475 unsigned switch_db_regs;
476 unsigned long db[KVM_NR_DB_REGS];
479 unsigned long eff_db[KVM_NR_DB_REGS];
480 unsigned long guest_debug_dr7;
487 /* Cache MMIO info */
495 /* used for guest single stepping over the given code position */
496 unsigned long singlestep_rip;
498 /* fields used by HYPER-V emulation */
501 cpumask_var_t wbinvd_dirty_mask;
503 unsigned long last_retry_eip;
504 unsigned long last_retry_addr;
508 gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
509 struct gfn_to_hva_cache data;
515 /* OSVW MSRs (AMD only) */
523 struct gfn_to_hva_cache data;
527 * Indicate whether the access faults on its page table in guest
528 * which is set when fix page fault and used to detect unhandeable
531 bool write_fault_to_shadow_pgtable;
533 /* set at EPT violation at this point */
534 unsigned long exit_qualification;
536 /* pv related host specific info */
542 struct kvm_lpage_info {
546 struct kvm_arch_memory_slot {
547 unsigned long *rmap[KVM_NR_PAGE_SIZES];
548 struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
551 struct kvm_apic_map {
554 /* fields bellow are used to decode ldr values in different modes */
555 u32 cid_shift, cid_mask, lid_mask, broadcast;
556 struct kvm_lapic *phys_map[256];
557 /* first index is cluster id second is cpu id in a cluster */
558 struct kvm_lapic *logical_map[16][16];
562 unsigned int n_used_mmu_pages;
563 unsigned int n_requested_mmu_pages;
564 unsigned int n_max_mmu_pages;
565 unsigned int indirect_shadow_pages;
566 unsigned long mmu_valid_gen;
567 struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
569 * Hash table of struct kvm_mmu_page.
571 struct list_head active_mmu_pages;
572 struct list_head zapped_obsolete_pages;
574 struct list_head assigned_dev_head;
575 struct iommu_domain *iommu_domain;
576 bool iommu_noncoherent;
577 #define __KVM_HAVE_ARCH_NONCOHERENT_DMA
578 atomic_t noncoherent_dma_count;
579 struct kvm_pic *vpic;
580 struct kvm_ioapic *vioapic;
581 struct kvm_pit *vpit;
582 int vapics_in_nmi_mode;
583 struct mutex apic_map_lock;
584 struct kvm_apic_map *apic_map;
586 unsigned int tss_addr;
587 bool apic_access_page_done;
591 bool ept_identity_pagetable_done;
592 gpa_t ept_identity_map_addr;
594 unsigned long irq_sources_bitmap;
596 raw_spinlock_t tsc_write_lock;
603 u64 cur_tsc_generation;
604 int nr_vcpus_matched_tsc;
606 spinlock_t pvclock_gtod_sync_lock;
607 bool use_master_clock;
608 u64 master_kernel_ns;
609 cycle_t master_cycle_now;
610 struct delayed_work kvmclock_update_work;
611 struct delayed_work kvmclock_sync_work;
613 struct kvm_xen_hvm_config xen_hvm_config;
615 /* reads protected by irq_srcu, writes by irq_lock */
616 struct hlist_head mask_notifier_list;
618 /* fields used by HYPER-V emulation */
623 #ifdef CONFIG_KVM_MMU_AUDIT
627 bool boot_vcpu_runs_old_kvmclock;
631 u32 mmu_shadow_zapped;
639 u32 remote_tlb_flush;
643 struct kvm_vcpu_stat {
653 u32 irq_window_exits;
654 u32 nmi_window_exits;
656 u32 halt_successful_poll;
658 u32 request_irq_exits;
660 u32 host_state_reload;
664 u32 insn_emulation_fail;
670 struct x86_instruction_info;
678 struct kvm_lapic_irq {
689 int (*cpu_has_kvm_support)(void); /* __init */
690 int (*disabled_by_bios)(void); /* __init */
691 int (*hardware_enable)(void);
692 void (*hardware_disable)(void);
693 void (*check_processor_compatibility)(void *rtn);
694 int (*hardware_setup)(void); /* __init */
695 void (*hardware_unsetup)(void); /* __exit */
696 bool (*cpu_has_accelerated_tpr)(void);
697 void (*cpuid_update)(struct kvm_vcpu *vcpu);
699 /* Create, but do not attach this VCPU */
700 struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
701 void (*vcpu_free)(struct kvm_vcpu *vcpu);
702 void (*vcpu_reset)(struct kvm_vcpu *vcpu);
704 void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
705 void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
706 void (*vcpu_put)(struct kvm_vcpu *vcpu);
708 void (*update_db_bp_intercept)(struct kvm_vcpu *vcpu);
709 int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
710 int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
711 u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
712 void (*get_segment)(struct kvm_vcpu *vcpu,
713 struct kvm_segment *var, int seg);
714 int (*get_cpl)(struct kvm_vcpu *vcpu);
715 void (*set_segment)(struct kvm_vcpu *vcpu,
716 struct kvm_segment *var, int seg);
717 void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
718 void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
719 void (*decache_cr3)(struct kvm_vcpu *vcpu);
720 void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
721 void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
722 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
723 int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
724 void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
725 void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
726 void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
727 void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
728 void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
729 u64 (*get_dr6)(struct kvm_vcpu *vcpu);
730 void (*set_dr6)(struct kvm_vcpu *vcpu, unsigned long value);
731 void (*sync_dirty_debug_regs)(struct kvm_vcpu *vcpu);
732 void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
733 void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
734 unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
735 void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
736 void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
738 void (*tlb_flush)(struct kvm_vcpu *vcpu);
740 void (*run)(struct kvm_vcpu *vcpu);
741 int (*handle_exit)(struct kvm_vcpu *vcpu);
742 void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
743 void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
744 u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu);
745 void (*patch_hypercall)(struct kvm_vcpu *vcpu,
746 unsigned char *hypercall_addr);
747 void (*set_irq)(struct kvm_vcpu *vcpu);
748 void (*set_nmi)(struct kvm_vcpu *vcpu);
749 void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
750 bool has_error_code, u32 error_code,
752 void (*cancel_injection)(struct kvm_vcpu *vcpu);
753 int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
754 int (*nmi_allowed)(struct kvm_vcpu *vcpu);
755 bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
756 void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
757 void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
758 void (*enable_irq_window)(struct kvm_vcpu *vcpu);
759 void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
760 int (*vm_has_apicv)(struct kvm *kvm);
761 void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
762 void (*hwapic_isr_update)(struct kvm *kvm, int isr);
763 void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap);
764 void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu, bool set);
765 void (*set_apic_access_page_addr)(struct kvm_vcpu *vcpu, hpa_t hpa);
766 void (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
767 void (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
768 int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
769 int (*get_tdp_level)(void);
770 u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
771 int (*get_lpage_level)(void);
772 bool (*rdtscp_supported)(void);
773 bool (*invpcid_supported)(void);
774 void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment, bool host);
776 void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
778 void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
780 bool (*has_wbinvd_exit)(void);
782 void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale);
783 u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu);
784 void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
786 u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc);
787 u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu, u64 host_tsc);
789 void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
791 int (*check_intercept)(struct kvm_vcpu *vcpu,
792 struct x86_instruction_info *info,
793 enum x86_intercept_stage stage);
794 void (*handle_external_intr)(struct kvm_vcpu *vcpu);
795 bool (*mpx_supported)(void);
796 bool (*xsaves_supported)(void);
798 int (*check_nested_events)(struct kvm_vcpu *vcpu, bool external_intr);
800 void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
803 * Arch-specific dirty logging hooks. These hooks are only supposed to
804 * be valid if the specific arch has hardware-accelerated dirty logging
805 * mechanism. Currently only for PML on VMX.
807 * - slot_enable_log_dirty:
808 * called when enabling log dirty mode for the slot.
809 * - slot_disable_log_dirty:
810 * called when disabling log dirty mode for the slot.
811 * also called when slot is created with log dirty disabled.
813 * called before reporting dirty_bitmap to userspace.
814 * - enable_log_dirty_pt_masked:
815 * called when reenabling log dirty for the GFNs in the mask after
816 * corresponding bits are cleared in slot->dirty_bitmap.
818 void (*slot_enable_log_dirty)(struct kvm *kvm,
819 struct kvm_memory_slot *slot);
820 void (*slot_disable_log_dirty)(struct kvm *kvm,
821 struct kvm_memory_slot *slot);
822 void (*flush_log_dirty)(struct kvm *kvm);
823 void (*enable_log_dirty_pt_masked)(struct kvm *kvm,
824 struct kvm_memory_slot *slot,
825 gfn_t offset, unsigned long mask);
828 struct kvm_arch_async_pf {
835 extern struct kvm_x86_ops *kvm_x86_ops;
837 static inline void adjust_tsc_offset_guest(struct kvm_vcpu *vcpu,
840 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, false);
843 static inline void adjust_tsc_offset_host(struct kvm_vcpu *vcpu, s64 adjustment)
845 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, true);
848 int kvm_mmu_module_init(void);
849 void kvm_mmu_module_exit(void);
851 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
852 int kvm_mmu_create(struct kvm_vcpu *vcpu);
853 void kvm_mmu_setup(struct kvm_vcpu *vcpu);
854 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
855 u64 dirty_mask, u64 nx_mask, u64 x_mask);
857 void kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
858 void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
859 struct kvm_memory_slot *memslot);
860 void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
861 struct kvm_memory_slot *memslot);
862 void kvm_mmu_slot_largepage_remove_write_access(struct kvm *kvm,
863 struct kvm_memory_slot *memslot);
864 void kvm_mmu_slot_set_dirty(struct kvm *kvm,
865 struct kvm_memory_slot *memslot);
866 void kvm_mmu_clear_dirty_pt_masked(struct kvm *kvm,
867 struct kvm_memory_slot *slot,
868 gfn_t gfn_offset, unsigned long mask);
869 void kvm_mmu_zap_all(struct kvm *kvm);
870 void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm);
871 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
872 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
874 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
876 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
877 const void *val, int bytes);
878 u8 kvm_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn);
880 struct kvm_irq_mask_notifier {
881 void (*func)(struct kvm_irq_mask_notifier *kimn, bool masked);
883 struct hlist_node link;
886 void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
887 struct kvm_irq_mask_notifier *kimn);
888 void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
889 struct kvm_irq_mask_notifier *kimn);
890 void kvm_fire_mask_notifiers(struct kvm *kvm, unsigned irqchip, unsigned pin,
893 extern bool tdp_enabled;
895 u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
897 /* control of guest tsc rate supported? */
898 extern bool kvm_has_tsc_control;
899 /* minimum supported tsc_khz for guests */
900 extern u32 kvm_min_guest_tsc_khz;
901 /* maximum supported tsc_khz for guests */
902 extern u32 kvm_max_guest_tsc_khz;
904 enum emulation_result {
905 EMULATE_DONE, /* no further processing */
906 EMULATE_USER_EXIT, /* kvm_run ready for userspace exit */
907 EMULATE_FAIL, /* can't emulate this instruction */
910 #define EMULTYPE_NO_DECODE (1 << 0)
911 #define EMULTYPE_TRAP_UD (1 << 1)
912 #define EMULTYPE_SKIP (1 << 2)
913 #define EMULTYPE_RETRY (1 << 3)
914 #define EMULTYPE_NO_REEXECUTE (1 << 4)
915 int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
916 int emulation_type, void *insn, int insn_len);
918 static inline int emulate_instruction(struct kvm_vcpu *vcpu,
921 return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
924 void kvm_enable_efer_bits(u64);
925 bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
926 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
927 int kvm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
929 struct x86_emulate_ctxt;
931 int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
932 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
933 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
934 int kvm_vcpu_halt(struct kvm_vcpu *vcpu);
935 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
937 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
938 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
939 void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
941 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
942 int reason, bool has_error_code, u32 error_code);
944 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
945 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
946 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
947 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
948 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
949 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
950 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
951 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
952 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
953 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
955 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
956 int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
958 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
959 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
960 bool kvm_rdpmc(struct kvm_vcpu *vcpu);
962 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
963 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
964 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
965 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
966 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
967 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
968 gfn_t gfn, void *data, int offset, int len,
970 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
971 bool kvm_require_dr(struct kvm_vcpu *vcpu, int dr);
973 static inline int __kvm_irq_line_state(unsigned long *irq_state,
974 int irq_source_id, int level)
976 /* Logical OR for level trig interrupt */
978 __set_bit(irq_source_id, irq_state);
980 __clear_bit(irq_source_id, irq_state);
982 return !!(*irq_state);
985 int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
986 void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
988 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
990 int fx_init(struct kvm_vcpu *vcpu);
992 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
993 const u8 *new, int bytes);
994 int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
995 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
996 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
997 int kvm_mmu_load(struct kvm_vcpu *vcpu);
998 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
999 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
1000 gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1001 struct x86_exception *exception);
1002 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
1003 struct x86_exception *exception);
1004 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
1005 struct x86_exception *exception);
1006 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
1007 struct x86_exception *exception);
1008 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
1009 struct x86_exception *exception);
1011 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
1013 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
1014 void *insn, int insn_len);
1015 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
1016 void kvm_mmu_new_cr3(struct kvm_vcpu *vcpu);
1018 void kvm_enable_tdp(void);
1019 void kvm_disable_tdp(void);
1021 static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1022 struct x86_exception *exception)
1027 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
1029 struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
1031 return (struct kvm_mmu_page *)page_private(page);
1034 static inline u16 kvm_read_ldt(void)
1037 asm("sldt %0" : "=g"(ldt));
1041 static inline void kvm_load_ldt(u16 sel)
1043 asm("lldt %0" : : "rm"(sel));
1046 #ifdef CONFIG_X86_64
1047 static inline unsigned long read_msr(unsigned long msr)
1056 static inline u32 get_rdx_init_val(void)
1058 return 0x600; /* P6 family */
1061 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
1063 kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
1066 static inline u64 get_canonical(u64 la)
1068 return ((int64_t)la << 16) >> 16;
1071 static inline bool is_noncanonical_address(u64 la)
1073 #ifdef CONFIG_X86_64
1074 return get_canonical(la) != la;
1080 #define TSS_IOPB_BASE_OFFSET 0x66
1081 #define TSS_BASE_SIZE 0x68
1082 #define TSS_IOPB_SIZE (65536 / 8)
1083 #define TSS_REDIRECTION_SIZE (256 / 8)
1084 #define RMODE_TSS_SIZE \
1085 (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
1088 TASK_SWITCH_CALL = 0,
1089 TASK_SWITCH_IRET = 1,
1090 TASK_SWITCH_JMP = 2,
1091 TASK_SWITCH_GATE = 3,
1094 #define HF_GIF_MASK (1 << 0)
1095 #define HF_HIF_MASK (1 << 1)
1096 #define HF_VINTR_MASK (1 << 2)
1097 #define HF_NMI_MASK (1 << 3)
1098 #define HF_IRET_MASK (1 << 4)
1099 #define HF_GUEST_MASK (1 << 5) /* VCPU is in guest-mode */
1102 * Hardware virtualization extension instructions may fault if a
1103 * reboot turns off virtualization while processes are running.
1104 * Trap the fault and ignore the instruction if that happens.
1106 asmlinkage void kvm_spurious_fault(void);
1108 #define ____kvm_handle_fault_on_reboot(insn, cleanup_insn) \
1109 "666: " insn "\n\t" \
1111 ".pushsection .fixup, \"ax\" \n" \
1113 cleanup_insn "\n\t" \
1114 "cmpb $0, kvm_rebooting \n\t" \
1116 __ASM_SIZE(push) " $666b \n\t" \
1117 "call kvm_spurious_fault \n\t" \
1118 ".popsection \n\t" \
1119 _ASM_EXTABLE(666b, 667b)
1121 #define __kvm_handle_fault_on_reboot(insn) \
1122 ____kvm_handle_fault_on_reboot(insn, "")
1124 #define KVM_ARCH_WANT_MMU_NOTIFIER
1125 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
1126 int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
1127 int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
1128 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
1129 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
1130 int cpuid_maxphyaddr(struct kvm_vcpu *vcpu);
1131 int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
1132 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1133 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
1134 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
1135 void kvm_vcpu_reset(struct kvm_vcpu *vcpu);
1136 void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu);
1137 void kvm_arch_mmu_notifier_invalidate_page(struct kvm *kvm,
1138 unsigned long address);
1140 void kvm_define_shared_msr(unsigned index, u32 msr);
1141 int kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
1143 unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
1144 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1146 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1147 struct kvm_async_pf *work);
1148 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1149 struct kvm_async_pf *work);
1150 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1151 struct kvm_async_pf *work);
1152 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
1153 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1155 void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1157 int kvm_is_in_guest(void);
1159 void kvm_pmu_init(struct kvm_vcpu *vcpu);
1160 void kvm_pmu_destroy(struct kvm_vcpu *vcpu);
1161 void kvm_pmu_reset(struct kvm_vcpu *vcpu);
1162 void kvm_pmu_cpuid_update(struct kvm_vcpu *vcpu);
1163 bool kvm_pmu_msr(struct kvm_vcpu *vcpu, u32 msr);
1164 int kvm_pmu_get_msr(struct kvm_vcpu *vcpu, u32 msr, u64 *data);
1165 int kvm_pmu_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info);
1166 int kvm_pmu_check_pmc(struct kvm_vcpu *vcpu, unsigned pmc);
1167 int kvm_pmu_read_pmc(struct kvm_vcpu *vcpu, unsigned pmc, u64 *data);
1168 void kvm_handle_pmu_event(struct kvm_vcpu *vcpu);
1169 void kvm_deliver_pmi(struct kvm_vcpu *vcpu);
1171 #endif /* _ASM_X86_KVM_HOST_H */