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>
27 #include <linux/irqbypass.h>
28 #include <linux/hyperv.h>
31 #include <asm/pvclock-abi.h>
34 #include <asm/msr-index.h>
36 #include <asm/kvm_page_track.h>
37 #include <asm/hyperv-tlfs.h>
39 #define KVM_MAX_VCPUS 288
40 #define KVM_SOFT_MAX_VCPUS 240
41 #define KVM_MAX_VCPU_ID 1023
42 #define KVM_USER_MEM_SLOTS 509
43 /* memory slots that are not exposed to userspace */
44 #define KVM_PRIVATE_MEM_SLOTS 3
45 #define KVM_MEM_SLOTS_NUM (KVM_USER_MEM_SLOTS + KVM_PRIVATE_MEM_SLOTS)
47 #define KVM_HALT_POLL_NS_DEFAULT 200000
49 #define KVM_IRQCHIP_NUM_PINS KVM_IOAPIC_NUM_PINS
51 /* x86-specific vcpu->requests bit members */
52 #define KVM_REQ_MIGRATE_TIMER KVM_ARCH_REQ(0)
53 #define KVM_REQ_REPORT_TPR_ACCESS KVM_ARCH_REQ(1)
54 #define KVM_REQ_TRIPLE_FAULT KVM_ARCH_REQ(2)
55 #define KVM_REQ_MMU_SYNC KVM_ARCH_REQ(3)
56 #define KVM_REQ_CLOCK_UPDATE KVM_ARCH_REQ(4)
57 #define KVM_REQ_LOAD_CR3 KVM_ARCH_REQ(5)
58 #define KVM_REQ_EVENT KVM_ARCH_REQ(6)
59 #define KVM_REQ_APF_HALT KVM_ARCH_REQ(7)
60 #define KVM_REQ_STEAL_UPDATE KVM_ARCH_REQ(8)
61 #define KVM_REQ_NMI KVM_ARCH_REQ(9)
62 #define KVM_REQ_PMU KVM_ARCH_REQ(10)
63 #define KVM_REQ_PMI KVM_ARCH_REQ(11)
64 #define KVM_REQ_SMI KVM_ARCH_REQ(12)
65 #define KVM_REQ_MASTERCLOCK_UPDATE KVM_ARCH_REQ(13)
66 #define KVM_REQ_MCLOCK_INPROGRESS \
67 KVM_ARCH_REQ_FLAGS(14, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
68 #define KVM_REQ_SCAN_IOAPIC \
69 KVM_ARCH_REQ_FLAGS(15, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
70 #define KVM_REQ_GLOBAL_CLOCK_UPDATE KVM_ARCH_REQ(16)
71 #define KVM_REQ_APIC_PAGE_RELOAD \
72 KVM_ARCH_REQ_FLAGS(17, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
73 #define KVM_REQ_HV_CRASH KVM_ARCH_REQ(18)
74 #define KVM_REQ_IOAPIC_EOI_EXIT KVM_ARCH_REQ(19)
75 #define KVM_REQ_HV_RESET KVM_ARCH_REQ(20)
76 #define KVM_REQ_HV_EXIT KVM_ARCH_REQ(21)
77 #define KVM_REQ_HV_STIMER KVM_ARCH_REQ(22)
78 #define KVM_REQ_LOAD_EOI_EXITMAP KVM_ARCH_REQ(23)
79 #define KVM_REQ_GET_VMCS12_PAGES KVM_ARCH_REQ(24)
81 #define CR0_RESERVED_BITS \
82 (~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
83 | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
84 | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
86 #define CR4_RESERVED_BITS \
87 (~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
88 | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE \
89 | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR | X86_CR4_PCIDE \
90 | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_FSGSBASE \
91 | X86_CR4_OSXMMEXCPT | X86_CR4_LA57 | X86_CR4_VMXE \
92 | X86_CR4_SMAP | X86_CR4_PKE | X86_CR4_UMIP))
94 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
98 #define INVALID_PAGE (~(hpa_t)0)
99 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
101 #define UNMAPPED_GVA (~(gpa_t)0)
103 /* KVM Hugepage definitions for x86 */
104 #define KVM_NR_PAGE_SIZES 3
105 #define KVM_HPAGE_GFN_SHIFT(x) (((x) - 1) * 9)
106 #define KVM_HPAGE_SHIFT(x) (PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
107 #define KVM_HPAGE_SIZE(x) (1UL << KVM_HPAGE_SHIFT(x))
108 #define KVM_HPAGE_MASK(x) (~(KVM_HPAGE_SIZE(x) - 1))
109 #define KVM_PAGES_PER_HPAGE(x) (KVM_HPAGE_SIZE(x) / PAGE_SIZE)
111 static inline gfn_t gfn_to_index(gfn_t gfn, gfn_t base_gfn, int level)
113 /* KVM_HPAGE_GFN_SHIFT(PT_PAGE_TABLE_LEVEL) must be 0. */
114 return (gfn >> KVM_HPAGE_GFN_SHIFT(level)) -
115 (base_gfn >> KVM_HPAGE_GFN_SHIFT(level));
118 #define KVM_PERMILLE_MMU_PAGES 20
119 #define KVM_MIN_ALLOC_MMU_PAGES 64
120 #define KVM_MMU_HASH_SHIFT 12
121 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
122 #define KVM_MIN_FREE_MMU_PAGES 5
123 #define KVM_REFILL_PAGES 25
124 #define KVM_MAX_CPUID_ENTRIES 80
125 #define KVM_NR_FIXED_MTRR_REGION 88
126 #define KVM_NR_VAR_MTRR 8
128 #define ASYNC_PF_PER_VCPU 64
154 VCPU_EXREG_PDPTR = NR_VCPU_REGS,
171 #include <asm/kvm_emulate.h>
173 #define KVM_NR_MEM_OBJS 40
175 #define KVM_NR_DB_REGS 4
177 #define DR6_BD (1 << 13)
178 #define DR6_BS (1 << 14)
179 #define DR6_RTM (1 << 16)
180 #define DR6_FIXED_1 0xfffe0ff0
181 #define DR6_INIT 0xffff0ff0
182 #define DR6_VOLATILE 0x0001e00f
184 #define DR7_BP_EN_MASK 0x000000ff
185 #define DR7_GE (1 << 9)
186 #define DR7_GD (1 << 13)
187 #define DR7_FIXED_1 0x00000400
188 #define DR7_VOLATILE 0xffff2bff
190 #define PFERR_PRESENT_BIT 0
191 #define PFERR_WRITE_BIT 1
192 #define PFERR_USER_BIT 2
193 #define PFERR_RSVD_BIT 3
194 #define PFERR_FETCH_BIT 4
195 #define PFERR_PK_BIT 5
196 #define PFERR_GUEST_FINAL_BIT 32
197 #define PFERR_GUEST_PAGE_BIT 33
199 #define PFERR_PRESENT_MASK (1U << PFERR_PRESENT_BIT)
200 #define PFERR_WRITE_MASK (1U << PFERR_WRITE_BIT)
201 #define PFERR_USER_MASK (1U << PFERR_USER_BIT)
202 #define PFERR_RSVD_MASK (1U << PFERR_RSVD_BIT)
203 #define PFERR_FETCH_MASK (1U << PFERR_FETCH_BIT)
204 #define PFERR_PK_MASK (1U << PFERR_PK_BIT)
205 #define PFERR_GUEST_FINAL_MASK (1ULL << PFERR_GUEST_FINAL_BIT)
206 #define PFERR_GUEST_PAGE_MASK (1ULL << PFERR_GUEST_PAGE_BIT)
208 #define PFERR_NESTED_GUEST_PAGE (PFERR_GUEST_PAGE_MASK | \
213 * The mask used to denote special SPTEs, which can be either MMIO SPTEs or
214 * Access Tracking SPTEs. We use bit 62 instead of bit 63 to avoid conflicting
215 * with the SVE bit in EPT PTEs.
217 #define SPTE_SPECIAL_MASK (1ULL << 62)
219 /* apic attention bits */
220 #define KVM_APIC_CHECK_VAPIC 0
222 * The following bit is set with PV-EOI, unset on EOI.
223 * We detect PV-EOI changes by guest by comparing
224 * this bit with PV-EOI in guest memory.
225 * See the implementation in apic_update_pv_eoi.
227 #define KVM_APIC_PV_EOI_PENDING 1
229 struct kvm_kernel_irq_routing_entry;
232 * We don't want allocation failures within the mmu code, so we preallocate
233 * enough memory for a single page fault in a cache.
235 struct kvm_mmu_memory_cache {
237 void *objects[KVM_NR_MEM_OBJS];
241 * the pages used as guest page table on soft mmu are tracked by
242 * kvm_memory_slot.arch.gfn_track which is 16 bits, so the role bits used
243 * by indirect shadow page can not be more than 15 bits.
245 * Currently, we used 14 bits that are @level, @cr4_pae, @quadrant, @access,
246 * @nxe, @cr0_wp, @smep_andnot_wp and @smap_andnot_wp.
248 union kvm_mmu_page_role {
259 unsigned smep_andnot_wp:1;
260 unsigned smap_andnot_wp:1;
261 unsigned ad_disabled:1;
262 unsigned guest_mode:1;
266 * This is left at the top of the word so that
267 * kvm_memslots_for_spte_role can extract it with a
268 * simple shift. While there is room, give it a whole
269 * byte so it is also faster to load it from memory.
275 struct kvm_rmap_head {
279 struct kvm_mmu_page {
280 struct list_head link;
281 struct hlist_node hash_link;
284 * The following two entries are used to key the shadow page in the
288 union kvm_mmu_page_role role;
291 /* hold the gfn of each spte inside spt */
294 int root_count; /* Currently serving as active root */
295 unsigned int unsync_children;
296 struct kvm_rmap_head parent_ptes; /* rmap pointers to parent sptes */
298 /* The page is obsolete if mmu_valid_gen != kvm->arch.mmu_valid_gen. */
299 unsigned long mmu_valid_gen;
301 DECLARE_BITMAP(unsync_child_bitmap, 512);
305 * Used out of the mmu-lock to avoid reading spte values while an
306 * update is in progress; see the comments in __get_spte_lockless().
308 int clear_spte_count;
311 /* Number of writes since the last time traversal visited this page. */
312 atomic_t write_flooding_count;
315 struct kvm_pio_request {
322 #define PT64_ROOT_MAX_LEVEL 5
324 struct rsvd_bits_validate {
325 u64 rsvd_bits_mask[2][PT64_ROOT_MAX_LEVEL];
329 struct kvm_mmu_root_info {
334 #define KVM_MMU_ROOT_INFO_INVALID \
335 ((struct kvm_mmu_root_info) { .cr3 = INVALID_PAGE, .hpa = INVALID_PAGE })
337 #define KVM_MMU_NUM_PREV_ROOTS 3
340 * x86 supports 4 paging modes (5-level 64-bit, 4-level 64-bit, 3-level 32-bit,
341 * and 2-level 32-bit). The kvm_mmu structure abstracts the details of the
345 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
346 unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
347 u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
348 int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
350 void (*inject_page_fault)(struct kvm_vcpu *vcpu,
351 struct x86_exception *fault);
352 gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
353 struct x86_exception *exception);
354 gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
355 struct x86_exception *exception);
356 int (*sync_page)(struct kvm_vcpu *vcpu,
357 struct kvm_mmu_page *sp);
358 void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva, hpa_t root_hpa);
359 void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
360 u64 *spte, const void *pte);
362 union kvm_mmu_page_role base_role;
364 u8 shadow_root_level;
367 struct kvm_mmu_root_info prev_roots[KVM_MMU_NUM_PREV_ROOTS];
370 * Bitmap; bit set = permission fault
371 * Byte index: page fault error code [4:1]
372 * Bit index: pte permissions in ACC_* format
377 * The pkru_mask indicates if protection key checks are needed. It
378 * consists of 16 domains indexed by page fault error code bits [4:1],
379 * with PFEC.RSVD replaced by ACC_USER_MASK from the page tables.
380 * Each domain has 2 bits which are ANDed with AD and WD from PKRU.
388 * check zero bits on shadow page table entries, these
389 * bits include not only hardware reserved bits but also
390 * the bits spte never used.
392 struct rsvd_bits_validate shadow_zero_check;
394 struct rsvd_bits_validate guest_rsvd_check;
396 /* Can have large pages at levels 2..last_nonleaf_level-1. */
397 u8 last_nonleaf_level;
401 u64 pdptrs[4]; /* pae */
414 struct perf_event *perf_event;
415 struct kvm_vcpu *vcpu;
419 unsigned nr_arch_gp_counters;
420 unsigned nr_arch_fixed_counters;
421 unsigned available_event_types;
426 u64 counter_bitmask[2];
427 u64 global_ctrl_mask;
430 struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
431 struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
432 struct irq_work irq_work;
439 KVM_DEBUGREG_BP_ENABLED = 1,
440 KVM_DEBUGREG_WONT_EXIT = 2,
441 KVM_DEBUGREG_RELOAD = 4,
444 struct kvm_mtrr_range {
447 struct list_head node;
451 struct kvm_mtrr_range var_ranges[KVM_NR_VAR_MTRR];
452 mtrr_type fixed_ranges[KVM_NR_FIXED_MTRR_REGION];
455 struct list_head head;
458 /* Hyper-V SynIC timer */
459 struct kvm_vcpu_hv_stimer {
460 struct hrtimer timer;
465 struct hv_message msg;
469 /* Hyper-V synthetic interrupt controller (SynIC)*/
470 struct kvm_vcpu_hv_synic {
475 atomic64_t sint[HV_SYNIC_SINT_COUNT];
476 atomic_t sint_to_gsi[HV_SYNIC_SINT_COUNT];
477 DECLARE_BITMAP(auto_eoi_bitmap, 256);
478 DECLARE_BITMAP(vec_bitmap, 256);
480 bool dont_zero_synic_pages;
483 /* Hyper-V per vcpu emulation context */
488 struct kvm_vcpu_hv_synic synic;
489 struct kvm_hyperv_exit exit;
490 struct kvm_vcpu_hv_stimer stimer[HV_SYNIC_STIMER_COUNT];
491 DECLARE_BITMAP(stimer_pending_bitmap, HV_SYNIC_STIMER_COUNT);
495 struct kvm_vcpu_arch {
497 * rip and regs accesses must go through
498 * kvm_{register,rip}_{read,write} functions.
500 unsigned long regs[NR_VCPU_REGS];
505 unsigned long cr0_guest_owned_bits;
509 unsigned long cr4_guest_owned_bits;
515 struct kvm_lapic *apic; /* kernel irqchip context */
517 bool load_eoi_exitmap_pending;
518 DECLARE_BITMAP(ioapic_handled_vectors, 256);
519 unsigned long apic_attention;
520 int32_t apic_arb_prio;
522 u64 ia32_misc_enable_msr;
525 bool tpr_access_reporting;
527 u64 microcode_version;
530 * Paging state of the vcpu
532 * If the vcpu runs in guest mode with two level paging this still saves
533 * the paging mode of the l1 guest. This context is always used to
539 * Paging state of an L2 guest (used for nested npt)
541 * This context will save all necessary information to walk page tables
542 * of the an L2 guest. This context is only initialized for page table
543 * walking and not for faulting since we never handle l2 page faults on
546 struct kvm_mmu nested_mmu;
549 * Pointer to the mmu context currently used for
550 * gva_to_gpa translations.
552 struct kvm_mmu *walk_mmu;
554 struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
555 struct kvm_mmu_memory_cache mmu_page_cache;
556 struct kvm_mmu_memory_cache mmu_page_header_cache;
559 * QEMU userspace and the guest each have their own FPU state.
560 * In vcpu_run, we switch between the user and guest FPU contexts.
561 * While running a VCPU, the VCPU thread will have the guest FPU
564 * Note that while the PKRU state lives inside the fpu registers,
565 * it is switched out separately at VMENTER and VMEXIT time. The
566 * "guest_fpu" state here contains the guest FPU context, with the
570 struct fpu guest_fpu;
573 u64 guest_supported_xcr0;
574 u32 guest_xstate_size;
576 struct kvm_pio_request pio;
579 u8 event_exit_inst_len;
581 struct kvm_queued_exception {
590 struct kvm_queued_interrupt {
596 int halt_request; /* real mode on Intel only */
599 struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
603 /* emulate context */
605 struct x86_emulate_ctxt emulate_ctxt;
606 bool emulate_regs_need_sync_to_vcpu;
607 bool emulate_regs_need_sync_from_vcpu;
608 int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
611 struct pvclock_vcpu_time_info hv_clock;
612 unsigned int hw_tsc_khz;
613 struct gfn_to_hva_cache pv_time;
614 bool pv_time_enabled;
615 /* set guest stopped flag in pvclock flags field */
616 bool pvclock_set_guest_stopped_request;
621 struct gfn_to_hva_cache stime;
622 struct kvm_steal_time steal;
628 u64 tsc_offset_adjustment;
631 u64 this_tsc_generation;
633 bool tsc_always_catchup;
634 s8 virtual_tsc_shift;
635 u32 virtual_tsc_mult;
637 s64 ia32_tsc_adjust_msr;
638 u64 tsc_scaling_ratio;
640 atomic_t nmi_queued; /* unprocessed asynchronous NMIs */
641 unsigned nmi_pending; /* NMI queued after currently running handler */
642 bool nmi_injected; /* Trying to inject an NMI this entry */
643 bool smi_pending; /* SMI queued after currently running handler */
645 struct kvm_mtrr mtrr_state;
648 unsigned switch_db_regs;
649 unsigned long db[KVM_NR_DB_REGS];
652 unsigned long eff_db[KVM_NR_DB_REGS];
653 unsigned long guest_debug_dr7;
654 u64 msr_platform_info;
655 u64 msr_misc_features_enables;
663 /* Cache MMIO info */
671 /* used for guest single stepping over the given code position */
672 unsigned long singlestep_rip;
674 struct kvm_vcpu_hv hyperv;
676 cpumask_var_t wbinvd_dirty_mask;
678 unsigned long last_retry_eip;
679 unsigned long last_retry_addr;
683 gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
684 struct gfn_to_hva_cache data;
689 unsigned long nested_apf_token;
690 bool delivery_as_pf_vmexit;
693 /* OSVW MSRs (AMD only) */
701 struct gfn_to_hva_cache data;
705 * Indicate whether the access faults on its page table in guest
706 * which is set when fix page fault and used to detect unhandeable
709 bool write_fault_to_shadow_pgtable;
711 /* set at EPT violation at this point */
712 unsigned long exit_qualification;
714 /* pv related host specific info */
719 int pending_ioapic_eoi;
720 int pending_external_vector;
726 /* be preempted when it's in kernel-mode(cpl=0) */
727 bool preempted_in_kernel;
730 struct kvm_lpage_info {
734 struct kvm_arch_memory_slot {
735 struct kvm_rmap_head *rmap[KVM_NR_PAGE_SIZES];
736 struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
737 unsigned short *gfn_track[KVM_PAGE_TRACK_MAX];
741 * We use as the mode the number of bits allocated in the LDR for the
742 * logical processor ID. It happens that these are all powers of two.
743 * This makes it is very easy to detect cases where the APICs are
744 * configured for multiple modes; in that case, we cannot use the map and
745 * hence cannot use kvm_irq_delivery_to_apic_fast either.
747 #define KVM_APIC_MODE_XAPIC_CLUSTER 4
748 #define KVM_APIC_MODE_XAPIC_FLAT 8
749 #define KVM_APIC_MODE_X2APIC 16
751 struct kvm_apic_map {
756 struct kvm_lapic *xapic_flat_map[8];
757 struct kvm_lapic *xapic_cluster_map[16][4];
759 struct kvm_lapic *phys_map[];
762 /* Hyper-V emulation context */
764 struct mutex hv_lock;
769 /* Hyper-v based guest crash (NT kernel bugcheck) parameters */
770 u64 hv_crash_param[HV_X64_MSR_CRASH_PARAMS];
773 HV_REFERENCE_TSC_PAGE tsc_ref;
775 struct idr conn_to_evt;
777 u64 hv_reenlightenment_control;
778 u64 hv_tsc_emulation_control;
779 u64 hv_tsc_emulation_status;
782 enum kvm_irqchip_mode {
784 KVM_IRQCHIP_KERNEL, /* created with KVM_CREATE_IRQCHIP */
785 KVM_IRQCHIP_SPLIT, /* created with KVM_CAP_SPLIT_IRQCHIP */
789 unsigned int n_used_mmu_pages;
790 unsigned int n_requested_mmu_pages;
791 unsigned int n_max_mmu_pages;
792 unsigned int indirect_shadow_pages;
793 unsigned long mmu_valid_gen;
794 struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
796 * Hash table of struct kvm_mmu_page.
798 struct list_head active_mmu_pages;
799 struct list_head zapped_obsolete_pages;
800 struct kvm_page_track_notifier_node mmu_sp_tracker;
801 struct kvm_page_track_notifier_head track_notifier_head;
803 struct list_head assigned_dev_head;
804 struct iommu_domain *iommu_domain;
805 bool iommu_noncoherent;
806 #define __KVM_HAVE_ARCH_NONCOHERENT_DMA
807 atomic_t noncoherent_dma_count;
808 #define __KVM_HAVE_ARCH_ASSIGNED_DEVICE
809 atomic_t assigned_device_count;
810 struct kvm_pic *vpic;
811 struct kvm_ioapic *vioapic;
812 struct kvm_pit *vpit;
813 atomic_t vapics_in_nmi_mode;
814 struct mutex apic_map_lock;
815 struct kvm_apic_map *apic_map;
817 bool apic_access_page_done;
825 unsigned long irq_sources_bitmap;
827 raw_spinlock_t tsc_write_lock;
834 u64 cur_tsc_generation;
835 int nr_vcpus_matched_tsc;
837 spinlock_t pvclock_gtod_sync_lock;
838 bool use_master_clock;
839 u64 master_kernel_ns;
840 u64 master_cycle_now;
841 struct delayed_work kvmclock_update_work;
842 struct delayed_work kvmclock_sync_work;
844 struct kvm_xen_hvm_config xen_hvm_config;
846 /* reads protected by irq_srcu, writes by irq_lock */
847 struct hlist_head mask_notifier_list;
849 struct kvm_hv hyperv;
851 #ifdef CONFIG_KVM_MMU_AUDIT
855 bool backwards_tsc_observed;
856 bool boot_vcpu_runs_old_kvmclock;
861 enum kvm_irqchip_mode irqchip_mode;
862 u8 nr_reserved_ioapic_pins;
864 bool disabled_lapic_found;
867 bool x2apic_broadcast_quirk_disabled;
871 ulong mmu_shadow_zapped;
873 ulong mmu_pte_updated;
874 ulong mmu_pde_zapped;
877 ulong mmu_cache_miss;
879 ulong remote_tlb_flush;
881 ulong max_mmu_page_hash_collisions;
884 struct kvm_vcpu_stat {
894 u64 irq_window_exits;
895 u64 nmi_window_exits;
897 u64 halt_successful_poll;
898 u64 halt_attempted_poll;
899 u64 halt_poll_invalid;
901 u64 request_irq_exits;
903 u64 host_state_reload;
906 u64 insn_emulation_fail;
913 struct x86_instruction_info;
921 struct kvm_lapic_irq {
933 int (*cpu_has_kvm_support)(void); /* __init */
934 int (*disabled_by_bios)(void); /* __init */
935 int (*hardware_enable)(void);
936 void (*hardware_disable)(void);
937 void (*check_processor_compatibility)(void *rtn);
938 int (*hardware_setup)(void); /* __init */
939 void (*hardware_unsetup)(void); /* __exit */
940 bool (*cpu_has_accelerated_tpr)(void);
941 bool (*has_emulated_msr)(int index);
942 void (*cpuid_update)(struct kvm_vcpu *vcpu);
944 struct kvm *(*vm_alloc)(void);
945 void (*vm_free)(struct kvm *);
946 int (*vm_init)(struct kvm *kvm);
947 void (*vm_destroy)(struct kvm *kvm);
949 /* Create, but do not attach this VCPU */
950 struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
951 void (*vcpu_free)(struct kvm_vcpu *vcpu);
952 void (*vcpu_reset)(struct kvm_vcpu *vcpu, bool init_event);
954 void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
955 void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
956 void (*vcpu_put)(struct kvm_vcpu *vcpu);
958 void (*update_bp_intercept)(struct kvm_vcpu *vcpu);
959 int (*get_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
960 int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
961 u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
962 void (*get_segment)(struct kvm_vcpu *vcpu,
963 struct kvm_segment *var, int seg);
964 int (*get_cpl)(struct kvm_vcpu *vcpu);
965 void (*set_segment)(struct kvm_vcpu *vcpu,
966 struct kvm_segment *var, int seg);
967 void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
968 void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
969 void (*decache_cr3)(struct kvm_vcpu *vcpu);
970 void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
971 void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
972 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
973 int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
974 void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
975 void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
976 void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
977 void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
978 void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
979 u64 (*get_dr6)(struct kvm_vcpu *vcpu);
980 void (*set_dr6)(struct kvm_vcpu *vcpu, unsigned long value);
981 void (*sync_dirty_debug_regs)(struct kvm_vcpu *vcpu);
982 void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
983 void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
984 unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
985 void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
987 void (*tlb_flush)(struct kvm_vcpu *vcpu, bool invalidate_gpa);
988 int (*tlb_remote_flush)(struct kvm *kvm);
991 * Flush any TLB entries associated with the given GVA.
992 * Does not need to flush GPA->HPA mappings.
993 * Can potentially get non-canonical addresses through INVLPGs, which
994 * the implementation may choose to ignore if appropriate.
996 void (*tlb_flush_gva)(struct kvm_vcpu *vcpu, gva_t addr);
998 void (*run)(struct kvm_vcpu *vcpu);
999 int (*handle_exit)(struct kvm_vcpu *vcpu);
1000 void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
1001 void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
1002 u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu);
1003 void (*patch_hypercall)(struct kvm_vcpu *vcpu,
1004 unsigned char *hypercall_addr);
1005 void (*set_irq)(struct kvm_vcpu *vcpu);
1006 void (*set_nmi)(struct kvm_vcpu *vcpu);
1007 void (*queue_exception)(struct kvm_vcpu *vcpu);
1008 void (*cancel_injection)(struct kvm_vcpu *vcpu);
1009 int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
1010 int (*nmi_allowed)(struct kvm_vcpu *vcpu);
1011 bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
1012 void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
1013 void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
1014 void (*enable_irq_window)(struct kvm_vcpu *vcpu);
1015 void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
1016 bool (*get_enable_apicv)(struct kvm_vcpu *vcpu);
1017 void (*refresh_apicv_exec_ctrl)(struct kvm_vcpu *vcpu);
1018 void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
1019 void (*hwapic_isr_update)(struct kvm_vcpu *vcpu, int isr);
1020 void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap);
1021 void (*set_virtual_apic_mode)(struct kvm_vcpu *vcpu);
1022 void (*set_apic_access_page_addr)(struct kvm_vcpu *vcpu, hpa_t hpa);
1023 void (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
1024 int (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
1025 int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
1026 int (*set_identity_map_addr)(struct kvm *kvm, u64 ident_addr);
1027 int (*get_tdp_level)(struct kvm_vcpu *vcpu);
1028 u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
1029 int (*get_lpage_level)(void);
1030 bool (*rdtscp_supported)(void);
1031 bool (*invpcid_supported)(void);
1033 void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
1035 void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
1037 bool (*has_wbinvd_exit)(void);
1039 u64 (*read_l1_tsc_offset)(struct kvm_vcpu *vcpu);
1040 void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
1042 void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
1044 int (*check_intercept)(struct kvm_vcpu *vcpu,
1045 struct x86_instruction_info *info,
1046 enum x86_intercept_stage stage);
1047 void (*handle_external_intr)(struct kvm_vcpu *vcpu);
1048 bool (*mpx_supported)(void);
1049 bool (*xsaves_supported)(void);
1050 bool (*umip_emulated)(void);
1052 int (*check_nested_events)(struct kvm_vcpu *vcpu, bool external_intr);
1054 void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
1057 * Arch-specific dirty logging hooks. These hooks are only supposed to
1058 * be valid if the specific arch has hardware-accelerated dirty logging
1059 * mechanism. Currently only for PML on VMX.
1061 * - slot_enable_log_dirty:
1062 * called when enabling log dirty mode for the slot.
1063 * - slot_disable_log_dirty:
1064 * called when disabling log dirty mode for the slot.
1065 * also called when slot is created with log dirty disabled.
1066 * - flush_log_dirty:
1067 * called before reporting dirty_bitmap to userspace.
1068 * - enable_log_dirty_pt_masked:
1069 * called when reenabling log dirty for the GFNs in the mask after
1070 * corresponding bits are cleared in slot->dirty_bitmap.
1072 void (*slot_enable_log_dirty)(struct kvm *kvm,
1073 struct kvm_memory_slot *slot);
1074 void (*slot_disable_log_dirty)(struct kvm *kvm,
1075 struct kvm_memory_slot *slot);
1076 void (*flush_log_dirty)(struct kvm *kvm);
1077 void (*enable_log_dirty_pt_masked)(struct kvm *kvm,
1078 struct kvm_memory_slot *slot,
1079 gfn_t offset, unsigned long mask);
1080 int (*write_log_dirty)(struct kvm_vcpu *vcpu);
1082 /* pmu operations of sub-arch */
1083 const struct kvm_pmu_ops *pmu_ops;
1086 * Architecture specific hooks for vCPU blocking due to
1088 * Returns for .pre_block():
1089 * - 0 means continue to block the vCPU.
1090 * - 1 means we cannot block the vCPU since some event
1091 * happens during this period, such as, 'ON' bit in
1092 * posted-interrupts descriptor is set.
1094 int (*pre_block)(struct kvm_vcpu *vcpu);
1095 void (*post_block)(struct kvm_vcpu *vcpu);
1097 void (*vcpu_blocking)(struct kvm_vcpu *vcpu);
1098 void (*vcpu_unblocking)(struct kvm_vcpu *vcpu);
1100 int (*update_pi_irte)(struct kvm *kvm, unsigned int host_irq,
1101 uint32_t guest_irq, bool set);
1102 void (*apicv_post_state_restore)(struct kvm_vcpu *vcpu);
1104 int (*set_hv_timer)(struct kvm_vcpu *vcpu, u64 guest_deadline_tsc);
1105 void (*cancel_hv_timer)(struct kvm_vcpu *vcpu);
1107 void (*setup_mce)(struct kvm_vcpu *vcpu);
1109 int (*get_nested_state)(struct kvm_vcpu *vcpu,
1110 struct kvm_nested_state __user *user_kvm_nested_state,
1111 unsigned user_data_size);
1112 int (*set_nested_state)(struct kvm_vcpu *vcpu,
1113 struct kvm_nested_state __user *user_kvm_nested_state,
1114 struct kvm_nested_state *kvm_state);
1115 void (*get_vmcs12_pages)(struct kvm_vcpu *vcpu);
1117 int (*smi_allowed)(struct kvm_vcpu *vcpu);
1118 int (*pre_enter_smm)(struct kvm_vcpu *vcpu, char *smstate);
1119 int (*pre_leave_smm)(struct kvm_vcpu *vcpu, u64 smbase);
1120 int (*enable_smi_window)(struct kvm_vcpu *vcpu);
1122 int (*mem_enc_op)(struct kvm *kvm, void __user *argp);
1123 int (*mem_enc_reg_region)(struct kvm *kvm, struct kvm_enc_region *argp);
1124 int (*mem_enc_unreg_region)(struct kvm *kvm, struct kvm_enc_region *argp);
1126 int (*get_msr_feature)(struct kvm_msr_entry *entry);
1129 struct kvm_arch_async_pf {
1136 extern struct kvm_x86_ops *kvm_x86_ops;
1138 #define __KVM_HAVE_ARCH_VM_ALLOC
1139 static inline struct kvm *kvm_arch_alloc_vm(void)
1141 return kvm_x86_ops->vm_alloc();
1144 static inline void kvm_arch_free_vm(struct kvm *kvm)
1146 return kvm_x86_ops->vm_free(kvm);
1149 #define __KVM_HAVE_ARCH_FLUSH_REMOTE_TLB
1150 static inline int kvm_arch_flush_remote_tlb(struct kvm *kvm)
1152 if (kvm_x86_ops->tlb_remote_flush &&
1153 !kvm_x86_ops->tlb_remote_flush(kvm))
1159 int kvm_mmu_module_init(void);
1160 void kvm_mmu_module_exit(void);
1162 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
1163 int kvm_mmu_create(struct kvm_vcpu *vcpu);
1164 void kvm_mmu_setup(struct kvm_vcpu *vcpu);
1165 void kvm_mmu_init_vm(struct kvm *kvm);
1166 void kvm_mmu_uninit_vm(struct kvm *kvm);
1167 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
1168 u64 dirty_mask, u64 nx_mask, u64 x_mask, u64 p_mask,
1169 u64 acc_track_mask, u64 me_mask);
1171 void kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
1172 void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
1173 struct kvm_memory_slot *memslot);
1174 void kvm_mmu_zap_collapsible_sptes(struct kvm *kvm,
1175 const struct kvm_memory_slot *memslot);
1176 void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
1177 struct kvm_memory_slot *memslot);
1178 void kvm_mmu_slot_largepage_remove_write_access(struct kvm *kvm,
1179 struct kvm_memory_slot *memslot);
1180 void kvm_mmu_slot_set_dirty(struct kvm *kvm,
1181 struct kvm_memory_slot *memslot);
1182 void kvm_mmu_clear_dirty_pt_masked(struct kvm *kvm,
1183 struct kvm_memory_slot *slot,
1184 gfn_t gfn_offset, unsigned long mask);
1185 void kvm_mmu_zap_all(struct kvm *kvm);
1186 void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm, struct kvm_memslots *slots);
1187 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
1188 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
1190 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
1191 bool pdptrs_changed(struct kvm_vcpu *vcpu);
1193 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
1194 const void *val, int bytes);
1196 struct kvm_irq_mask_notifier {
1197 void (*func)(struct kvm_irq_mask_notifier *kimn, bool masked);
1199 struct hlist_node link;
1202 void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
1203 struct kvm_irq_mask_notifier *kimn);
1204 void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
1205 struct kvm_irq_mask_notifier *kimn);
1206 void kvm_fire_mask_notifiers(struct kvm *kvm, unsigned irqchip, unsigned pin,
1209 extern bool tdp_enabled;
1211 u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
1213 /* control of guest tsc rate supported? */
1214 extern bool kvm_has_tsc_control;
1215 /* maximum supported tsc_khz for guests */
1216 extern u32 kvm_max_guest_tsc_khz;
1217 /* number of bits of the fractional part of the TSC scaling ratio */
1218 extern u8 kvm_tsc_scaling_ratio_frac_bits;
1219 /* maximum allowed value of TSC scaling ratio */
1220 extern u64 kvm_max_tsc_scaling_ratio;
1221 /* 1ull << kvm_tsc_scaling_ratio_frac_bits */
1222 extern u64 kvm_default_tsc_scaling_ratio;
1224 extern u64 kvm_mce_cap_supported;
1226 enum emulation_result {
1227 EMULATE_DONE, /* no further processing */
1228 EMULATE_USER_EXIT, /* kvm_run ready for userspace exit */
1229 EMULATE_FAIL, /* can't emulate this instruction */
1232 #define EMULTYPE_NO_DECODE (1 << 0)
1233 #define EMULTYPE_TRAP_UD (1 << 1)
1234 #define EMULTYPE_SKIP (1 << 2)
1235 #define EMULTYPE_RETRY (1 << 3)
1236 #define EMULTYPE_NO_REEXECUTE (1 << 4)
1237 #define EMULTYPE_NO_UD_ON_FAIL (1 << 5)
1238 #define EMULTYPE_VMWARE (1 << 6)
1239 int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
1240 int emulation_type, void *insn, int insn_len);
1242 static inline int emulate_instruction(struct kvm_vcpu *vcpu,
1245 return x86_emulate_instruction(vcpu, 0,
1246 emulation_type | EMULTYPE_NO_REEXECUTE, NULL, 0);
1249 void kvm_enable_efer_bits(u64);
1250 bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
1251 int kvm_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
1252 int kvm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
1254 struct x86_emulate_ctxt;
1256 int kvm_fast_pio(struct kvm_vcpu *vcpu, int size, unsigned short port, int in);
1257 int kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
1258 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
1259 int kvm_vcpu_halt(struct kvm_vcpu *vcpu);
1260 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
1262 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
1263 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
1264 void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
1266 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
1267 int reason, bool has_error_code, u32 error_code);
1269 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
1270 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
1271 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
1272 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
1273 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
1274 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
1275 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
1276 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
1277 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
1278 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
1280 int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
1281 int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
1283 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
1284 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
1285 bool kvm_rdpmc(struct kvm_vcpu *vcpu);
1287 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1288 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
1289 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1290 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
1291 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
1292 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
1293 gfn_t gfn, void *data, int offset, int len,
1295 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
1296 bool kvm_require_dr(struct kvm_vcpu *vcpu, int dr);
1298 static inline int __kvm_irq_line_state(unsigned long *irq_state,
1299 int irq_source_id, int level)
1301 /* Logical OR for level trig interrupt */
1303 __set_bit(irq_source_id, irq_state);
1305 __clear_bit(irq_source_id, irq_state);
1307 return !!(*irq_state);
1310 #define KVM_MMU_ROOT_CURRENT BIT(0)
1311 #define KVM_MMU_ROOT_PREVIOUS(i) BIT(1+i)
1312 #define KVM_MMU_ROOTS_ALL (~0UL)
1314 int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
1315 void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
1317 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
1319 int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
1320 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
1321 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
1322 int kvm_mmu_load(struct kvm_vcpu *vcpu);
1323 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
1324 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
1325 void kvm_mmu_free_roots(struct kvm_vcpu *vcpu, ulong roots_to_free);
1326 gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1327 struct x86_exception *exception);
1328 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
1329 struct x86_exception *exception);
1330 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
1331 struct x86_exception *exception);
1332 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
1333 struct x86_exception *exception);
1334 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
1335 struct x86_exception *exception);
1337 void kvm_vcpu_deactivate_apicv(struct kvm_vcpu *vcpu);
1339 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
1341 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u64 error_code,
1342 void *insn, int insn_len);
1343 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
1344 void kvm_mmu_invpcid_gva(struct kvm_vcpu *vcpu, gva_t gva, unsigned long pcid);
1345 void kvm_mmu_new_cr3(struct kvm_vcpu *vcpu, gpa_t new_cr3, bool skip_tlb_flush);
1347 void kvm_enable_tdp(void);
1348 void kvm_disable_tdp(void);
1350 static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1351 struct x86_exception *exception)
1356 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
1358 struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
1360 return (struct kvm_mmu_page *)page_private(page);
1363 static inline u16 kvm_read_ldt(void)
1366 asm("sldt %0" : "=g"(ldt));
1370 static inline void kvm_load_ldt(u16 sel)
1372 asm("lldt %0" : : "rm"(sel));
1375 #ifdef CONFIG_X86_64
1376 static inline unsigned long read_msr(unsigned long msr)
1385 static inline u32 get_rdx_init_val(void)
1387 return 0x600; /* P6 family */
1390 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
1392 kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
1395 #define TSS_IOPB_BASE_OFFSET 0x66
1396 #define TSS_BASE_SIZE 0x68
1397 #define TSS_IOPB_SIZE (65536 / 8)
1398 #define TSS_REDIRECTION_SIZE (256 / 8)
1399 #define RMODE_TSS_SIZE \
1400 (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
1403 TASK_SWITCH_CALL = 0,
1404 TASK_SWITCH_IRET = 1,
1405 TASK_SWITCH_JMP = 2,
1406 TASK_SWITCH_GATE = 3,
1409 #define HF_GIF_MASK (1 << 0)
1410 #define HF_HIF_MASK (1 << 1)
1411 #define HF_VINTR_MASK (1 << 2)
1412 #define HF_NMI_MASK (1 << 3)
1413 #define HF_IRET_MASK (1 << 4)
1414 #define HF_GUEST_MASK (1 << 5) /* VCPU is in guest-mode */
1415 #define HF_SMM_MASK (1 << 6)
1416 #define HF_SMM_INSIDE_NMI_MASK (1 << 7)
1418 #define __KVM_VCPU_MULTIPLE_ADDRESS_SPACE
1419 #define KVM_ADDRESS_SPACE_NUM 2
1421 #define kvm_arch_vcpu_memslots_id(vcpu) ((vcpu)->arch.hflags & HF_SMM_MASK ? 1 : 0)
1422 #define kvm_memslots_for_spte_role(kvm, role) __kvm_memslots(kvm, (role).smm)
1425 * Hardware virtualization extension instructions may fault if a
1426 * reboot turns off virtualization while processes are running.
1427 * Trap the fault and ignore the instruction if that happens.
1429 asmlinkage void kvm_spurious_fault(void);
1431 #define ____kvm_handle_fault_on_reboot(insn, cleanup_insn) \
1432 "666: " insn "\n\t" \
1434 ".pushsection .fixup, \"ax\" \n" \
1436 cleanup_insn "\n\t" \
1437 "cmpb $0, kvm_rebooting \n\t" \
1439 __ASM_SIZE(push) " $666b \n\t" \
1440 "call kvm_spurious_fault \n\t" \
1441 ".popsection \n\t" \
1442 _ASM_EXTABLE(666b, 667b)
1444 #define __kvm_handle_fault_on_reboot(insn) \
1445 ____kvm_handle_fault_on_reboot(insn, "")
1447 #define KVM_ARCH_WANT_MMU_NOTIFIER
1448 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
1449 int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
1450 int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
1451 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
1452 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
1453 int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
1454 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1455 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
1456 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
1457 void kvm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event);
1458 void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu);
1460 int kvm_pv_send_ipi(struct kvm *kvm, unsigned long ipi_bitmap_low,
1461 unsigned long ipi_bitmap_high, int min,
1462 unsigned long icr, int op_64_bit);
1464 void kvm_define_shared_msr(unsigned index, u32 msr);
1465 int kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
1467 u64 kvm_scale_tsc(struct kvm_vcpu *vcpu, u64 tsc);
1468 u64 kvm_read_l1_tsc(struct kvm_vcpu *vcpu, u64 host_tsc);
1470 unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
1471 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1473 void kvm_make_mclock_inprogress_request(struct kvm *kvm);
1474 void kvm_make_scan_ioapic_request(struct kvm *kvm);
1476 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1477 struct kvm_async_pf *work);
1478 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1479 struct kvm_async_pf *work);
1480 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1481 struct kvm_async_pf *work);
1482 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
1483 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1485 int kvm_skip_emulated_instruction(struct kvm_vcpu *vcpu);
1486 int kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1488 int kvm_is_in_guest(void);
1490 int __x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa, u32 size);
1491 int x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa, u32 size);
1492 bool kvm_vcpu_is_reset_bsp(struct kvm_vcpu *vcpu);
1493 bool kvm_vcpu_is_bsp(struct kvm_vcpu *vcpu);
1495 bool kvm_intr_is_single_vcpu(struct kvm *kvm, struct kvm_lapic_irq *irq,
1496 struct kvm_vcpu **dest_vcpu);
1498 void kvm_set_msi_irq(struct kvm *kvm, struct kvm_kernel_irq_routing_entry *e,
1499 struct kvm_lapic_irq *irq);
1501 static inline void kvm_arch_vcpu_blocking(struct kvm_vcpu *vcpu)
1503 if (kvm_x86_ops->vcpu_blocking)
1504 kvm_x86_ops->vcpu_blocking(vcpu);
1507 static inline void kvm_arch_vcpu_unblocking(struct kvm_vcpu *vcpu)
1509 if (kvm_x86_ops->vcpu_unblocking)
1510 kvm_x86_ops->vcpu_unblocking(vcpu);
1513 static inline void kvm_arch_vcpu_block_finish(struct kvm_vcpu *vcpu) {}
1515 static inline int kvm_cpu_get_apicid(int mps_cpu)
1517 #ifdef CONFIG_X86_LOCAL_APIC
1518 return default_cpu_present_to_apicid(mps_cpu);
1525 #define put_smstate(type, buf, offset, val) \
1526 *(type *)((buf) + (offset) - 0x7e00) = val
1528 #endif /* _ASM_X86_KVM_HOST_H */