2 * Kernel-based Virtual Machine driver for Linux
4 * This header defines architecture specific interfaces, x86 version
6 * This work is licensed under the terms of the GNU GPL, version 2. See
7 * the COPYING file in the top-level directory.
11 #ifndef _ASM_X86_KVM_HOST_H
12 #define _ASM_X86_KVM_HOST_H
14 #include <linux/types.h>
16 #include <linux/mmu_notifier.h>
17 #include <linux/tracepoint.h>
18 #include <linux/cpumask.h>
19 #include <linux/irq_work.h>
21 #include <linux/kvm.h>
22 #include <linux/kvm_para.h>
23 #include <linux/kvm_types.h>
24 #include <linux/perf_event.h>
25 #include <linux/pvclock_gtod.h>
26 #include <linux/clocksource.h>
28 #include <asm/pvclock-abi.h>
31 #include <asm/msr-index.h>
34 #define KVM_MAX_VCPUS 255
35 #define KVM_SOFT_MAX_VCPUS 160
36 #define KVM_USER_MEM_SLOTS 509
37 /* memory slots that are not exposed to userspace */
38 #define KVM_PRIVATE_MEM_SLOTS 3
39 #define KVM_MEM_SLOTS_NUM (KVM_USER_MEM_SLOTS + KVM_PRIVATE_MEM_SLOTS)
41 #define KVM_PIO_PAGE_OFFSET 1
42 #define KVM_COALESCED_MMIO_PAGE_OFFSET 2
44 #define KVM_IRQCHIP_NUM_PINS KVM_IOAPIC_NUM_PINS
46 #define CR0_RESERVED_BITS \
47 (~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
48 | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
49 | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
51 #define CR3_L_MODE_RESERVED_BITS 0xFFFFFF0000000000ULL
52 #define CR3_PCID_INVD BIT_64(63)
53 #define CR4_RESERVED_BITS \
54 (~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
55 | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE \
56 | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR | X86_CR4_PCIDE \
57 | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_FSGSBASE \
58 | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE | X86_CR4_SMAP))
60 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
64 #define INVALID_PAGE (~(hpa_t)0)
65 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
67 #define UNMAPPED_GVA (~(gpa_t)0)
69 /* KVM Hugepage definitions for x86 */
70 #define KVM_NR_PAGE_SIZES 3
71 #define KVM_HPAGE_GFN_SHIFT(x) (((x) - 1) * 9)
72 #define KVM_HPAGE_SHIFT(x) (PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
73 #define KVM_HPAGE_SIZE(x) (1UL << KVM_HPAGE_SHIFT(x))
74 #define KVM_HPAGE_MASK(x) (~(KVM_HPAGE_SIZE(x) - 1))
75 #define KVM_PAGES_PER_HPAGE(x) (KVM_HPAGE_SIZE(x) / PAGE_SIZE)
77 static inline gfn_t
gfn_to_index(gfn_t gfn
, gfn_t base_gfn
, int level
)
79 /* KVM_HPAGE_GFN_SHIFT(PT_PAGE_TABLE_LEVEL) must be 0. */
80 return (gfn
>> KVM_HPAGE_GFN_SHIFT(level
)) -
81 (base_gfn
>> KVM_HPAGE_GFN_SHIFT(level
));
84 #define KVM_PERMILLE_MMU_PAGES 20
85 #define KVM_MIN_ALLOC_MMU_PAGES 64
86 #define KVM_MMU_HASH_SHIFT 10
87 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
88 #define KVM_MIN_FREE_MMU_PAGES 5
89 #define KVM_REFILL_PAGES 25
90 #define KVM_MAX_CPUID_ENTRIES 80
91 #define KVM_NR_FIXED_MTRR_REGION 88
92 #define KVM_NR_VAR_MTRR 8
94 #define ASYNC_PF_PER_VCPU 64
120 VCPU_EXREG_PDPTR
= NR_VCPU_REGS
,
137 #include <asm/kvm_emulate.h>
139 #define KVM_NR_MEM_OBJS 40
141 #define KVM_NR_DB_REGS 4
143 #define DR6_BD (1 << 13)
144 #define DR6_BS (1 << 14)
145 #define DR6_RTM (1 << 16)
146 #define DR6_FIXED_1 0xfffe0ff0
147 #define DR6_INIT 0xffff0ff0
148 #define DR6_VOLATILE 0x0001e00f
150 #define DR7_BP_EN_MASK 0x000000ff
151 #define DR7_GE (1 << 9)
152 #define DR7_GD (1 << 13)
153 #define DR7_FIXED_1 0x00000400
154 #define DR7_VOLATILE 0xffff2bff
156 #define PFERR_PRESENT_BIT 0
157 #define PFERR_WRITE_BIT 1
158 #define PFERR_USER_BIT 2
159 #define PFERR_RSVD_BIT 3
160 #define PFERR_FETCH_BIT 4
162 #define PFERR_PRESENT_MASK (1U << PFERR_PRESENT_BIT)
163 #define PFERR_WRITE_MASK (1U << PFERR_WRITE_BIT)
164 #define PFERR_USER_MASK (1U << PFERR_USER_BIT)
165 #define PFERR_RSVD_MASK (1U << PFERR_RSVD_BIT)
166 #define PFERR_FETCH_MASK (1U << PFERR_FETCH_BIT)
168 /* apic attention bits */
169 #define KVM_APIC_CHECK_VAPIC 0
171 * The following bit is set with PV-EOI, unset on EOI.
172 * We detect PV-EOI changes by guest by comparing
173 * this bit with PV-EOI in guest memory.
174 * See the implementation in apic_update_pv_eoi.
176 #define KVM_APIC_PV_EOI_PENDING 1
179 * We don't want allocation failures within the mmu code, so we preallocate
180 * enough memory for a single page fault in a cache.
182 struct kvm_mmu_memory_cache
{
184 void *objects
[KVM_NR_MEM_OBJS
];
187 union kvm_mmu_page_role
{
198 unsigned smep_andnot_wp
:1;
199 unsigned smap_andnot_wp
:1;
203 * This is left at the top of the word so that
204 * kvm_memslots_for_spte_role can extract it with a
205 * simple shift. While there is room, give it a whole
206 * byte so it is also faster to load it from memory.
212 struct kvm_mmu_page
{
213 struct list_head link
;
214 struct hlist_node hash_link
;
217 * The following two entries are used to key the shadow page in the
221 union kvm_mmu_page_role role
;
224 /* hold the gfn of each spte inside spt */
227 int root_count
; /* Currently serving as active root */
228 unsigned int unsync_children
;
229 unsigned long parent_ptes
; /* Reverse mapping for parent_pte */
231 /* The page is obsolete if mmu_valid_gen != kvm->arch.mmu_valid_gen. */
232 unsigned long mmu_valid_gen
;
234 DECLARE_BITMAP(unsync_child_bitmap
, 512);
238 * Used out of the mmu-lock to avoid reading spte values while an
239 * update is in progress; see the comments in __get_spte_lockless().
241 int clear_spte_count
;
244 /* Number of writes since the last time traversal visited this page. */
245 int write_flooding_count
;
248 struct kvm_pio_request
{
256 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
257 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
261 void (*set_cr3
)(struct kvm_vcpu
*vcpu
, unsigned long root
);
262 unsigned long (*get_cr3
)(struct kvm_vcpu
*vcpu
);
263 u64 (*get_pdptr
)(struct kvm_vcpu
*vcpu
, int index
);
264 int (*page_fault
)(struct kvm_vcpu
*vcpu
, gva_t gva
, u32 err
,
266 void (*inject_page_fault
)(struct kvm_vcpu
*vcpu
,
267 struct x86_exception
*fault
);
268 gpa_t (*gva_to_gpa
)(struct kvm_vcpu
*vcpu
, gva_t gva
, u32 access
,
269 struct x86_exception
*exception
);
270 gpa_t (*translate_gpa
)(struct kvm_vcpu
*vcpu
, gpa_t gpa
, u32 access
,
271 struct x86_exception
*exception
);
272 int (*sync_page
)(struct kvm_vcpu
*vcpu
,
273 struct kvm_mmu_page
*sp
);
274 void (*invlpg
)(struct kvm_vcpu
*vcpu
, gva_t gva
);
275 void (*update_pte
)(struct kvm_vcpu
*vcpu
, struct kvm_mmu_page
*sp
,
276 u64
*spte
, const void *pte
);
279 int shadow_root_level
;
280 union kvm_mmu_page_role base_role
;
284 * Bitmap; bit set = permission fault
285 * Byte index: page fault error code [4:1]
286 * Bit index: pte permissions in ACC_* format
292 u64 rsvd_bits_mask
[2][4];
296 * Bitmap: bit set = last pte in walk
297 * index[0:1]: level (zero-based)
304 u64 pdptrs
[4]; /* pae */
317 struct perf_event
*perf_event
;
318 struct kvm_vcpu
*vcpu
;
322 unsigned nr_arch_gp_counters
;
323 unsigned nr_arch_fixed_counters
;
324 unsigned available_event_types
;
329 u64 counter_bitmask
[2];
330 u64 global_ctrl_mask
;
333 struct kvm_pmc gp_counters
[INTEL_PMC_MAX_GENERIC
];
334 struct kvm_pmc fixed_counters
[INTEL_PMC_MAX_FIXED
];
335 struct irq_work irq_work
;
342 KVM_DEBUGREG_BP_ENABLED
= 1,
343 KVM_DEBUGREG_WONT_EXIT
= 2,
344 KVM_DEBUGREG_RELOAD
= 4,
347 struct kvm_mtrr_range
{
350 struct list_head node
;
354 struct kvm_mtrr_range var_ranges
[KVM_NR_VAR_MTRR
];
355 mtrr_type fixed_ranges
[KVM_NR_FIXED_MTRR_REGION
];
358 struct list_head head
;
361 struct kvm_vcpu_arch
{
363 * rip and regs accesses must go through
364 * kvm_{register,rip}_{read,write} functions.
366 unsigned long regs
[NR_VCPU_REGS
];
371 unsigned long cr0_guest_owned_bits
;
375 unsigned long cr4_guest_owned_bits
;
380 struct kvm_lapic
*apic
; /* kernel irqchip context */
381 unsigned long apic_attention
;
382 int32_t apic_arb_prio
;
384 u64 ia32_misc_enable_msr
;
386 bool tpr_access_reporting
;
390 * Paging state of the vcpu
392 * If the vcpu runs in guest mode with two level paging this still saves
393 * the paging mode of the l1 guest. This context is always used to
399 * Paging state of an L2 guest (used for nested npt)
401 * This context will save all necessary information to walk page tables
402 * of the an L2 guest. This context is only initialized for page table
403 * walking and not for faulting since we never handle l2 page faults on
406 struct kvm_mmu nested_mmu
;
409 * Pointer to the mmu context currently used for
410 * gva_to_gpa translations.
412 struct kvm_mmu
*walk_mmu
;
414 struct kvm_mmu_memory_cache mmu_pte_list_desc_cache
;
415 struct kvm_mmu_memory_cache mmu_page_cache
;
416 struct kvm_mmu_memory_cache mmu_page_header_cache
;
418 struct fpu guest_fpu
;
421 u64 guest_supported_xcr0
;
422 u32 guest_xstate_size
;
424 struct kvm_pio_request pio
;
427 u8 event_exit_inst_len
;
429 struct kvm_queued_exception
{
437 struct kvm_queued_interrupt
{
443 int halt_request
; /* real mode on Intel only */
446 struct kvm_cpuid_entry2 cpuid_entries
[KVM_MAX_CPUID_ENTRIES
];
450 /* emulate context */
452 struct x86_emulate_ctxt emulate_ctxt
;
453 bool emulate_regs_need_sync_to_vcpu
;
454 bool emulate_regs_need_sync_from_vcpu
;
455 int (*complete_userspace_io
)(struct kvm_vcpu
*vcpu
);
458 struct pvclock_vcpu_time_info hv_clock
;
459 unsigned int hw_tsc_khz
;
460 struct gfn_to_hva_cache pv_time
;
461 bool pv_time_enabled
;
462 /* set guest stopped flag in pvclock flags field */
463 bool pvclock_set_guest_stopped_request
;
469 struct gfn_to_hva_cache stime
;
470 struct kvm_steal_time steal
;
475 u64 tsc_offset_adjustment
;
478 u64 this_tsc_generation
;
480 bool tsc_always_catchup
;
481 s8 virtual_tsc_shift
;
482 u32 virtual_tsc_mult
;
484 s64 ia32_tsc_adjust_msr
;
486 atomic_t nmi_queued
; /* unprocessed asynchronous NMIs */
487 unsigned nmi_pending
; /* NMI queued after currently running handler */
488 bool nmi_injected
; /* Trying to inject an NMI this entry */
489 bool smi_pending
; /* SMI queued after currently running handler */
491 struct kvm_mtrr mtrr_state
;
494 unsigned switch_db_regs
;
495 unsigned long db
[KVM_NR_DB_REGS
];
498 unsigned long eff_db
[KVM_NR_DB_REGS
];
499 unsigned long guest_debug_dr7
;
506 /* Cache MMIO info */
514 /* used for guest single stepping over the given code position */
515 unsigned long singlestep_rip
;
517 /* fields used by HYPER-V emulation */
520 cpumask_var_t wbinvd_dirty_mask
;
522 unsigned long last_retry_eip
;
523 unsigned long last_retry_addr
;
527 gfn_t gfns
[roundup_pow_of_two(ASYNC_PF_PER_VCPU
)];
528 struct gfn_to_hva_cache data
;
534 /* OSVW MSRs (AMD only) */
542 struct gfn_to_hva_cache data
;
546 * Indicate whether the access faults on its page table in guest
547 * which is set when fix page fault and used to detect unhandeable
550 bool write_fault_to_shadow_pgtable
;
552 /* set at EPT violation at this point */
553 unsigned long exit_qualification
;
555 /* pv related host specific info */
561 struct kvm_lpage_info
{
565 struct kvm_arch_memory_slot
{
566 unsigned long *rmap
[KVM_NR_PAGE_SIZES
];
567 struct kvm_lpage_info
*lpage_info
[KVM_NR_PAGE_SIZES
- 1];
571 * We use as the mode the number of bits allocated in the LDR for the
572 * logical processor ID. It happens that these are all powers of two.
573 * This makes it is very easy to detect cases where the APICs are
574 * configured for multiple modes; in that case, we cannot use the map and
575 * hence cannot use kvm_irq_delivery_to_apic_fast either.
577 #define KVM_APIC_MODE_XAPIC_CLUSTER 4
578 #define KVM_APIC_MODE_XAPIC_FLAT 8
579 #define KVM_APIC_MODE_X2APIC 16
581 struct kvm_apic_map
{
584 struct kvm_lapic
*phys_map
[256];
585 /* first index is cluster id second is cpu id in a cluster */
586 struct kvm_lapic
*logical_map
[16][16];
590 unsigned int n_used_mmu_pages
;
591 unsigned int n_requested_mmu_pages
;
592 unsigned int n_max_mmu_pages
;
593 unsigned int indirect_shadow_pages
;
594 unsigned long mmu_valid_gen
;
595 struct hlist_head mmu_page_hash
[KVM_NUM_MMU_PAGES
];
597 * Hash table of struct kvm_mmu_page.
599 struct list_head active_mmu_pages
;
600 struct list_head zapped_obsolete_pages
;
602 struct list_head assigned_dev_head
;
603 struct iommu_domain
*iommu_domain
;
604 bool iommu_noncoherent
;
605 #define __KVM_HAVE_ARCH_NONCOHERENT_DMA
606 atomic_t noncoherent_dma_count
;
607 struct kvm_pic
*vpic
;
608 struct kvm_ioapic
*vioapic
;
609 struct kvm_pit
*vpit
;
610 int vapics_in_nmi_mode
;
611 struct mutex apic_map_lock
;
612 struct kvm_apic_map
*apic_map
;
614 unsigned int tss_addr
;
615 bool apic_access_page_done
;
619 bool ept_identity_pagetable_done
;
620 gpa_t ept_identity_map_addr
;
622 unsigned long irq_sources_bitmap
;
624 raw_spinlock_t tsc_write_lock
;
631 u64 cur_tsc_generation
;
632 int nr_vcpus_matched_tsc
;
634 spinlock_t pvclock_gtod_sync_lock
;
635 bool use_master_clock
;
636 u64 master_kernel_ns
;
637 cycle_t master_cycle_now
;
638 struct delayed_work kvmclock_update_work
;
639 struct delayed_work kvmclock_sync_work
;
641 struct kvm_xen_hvm_config xen_hvm_config
;
643 /* reads protected by irq_srcu, writes by irq_lock */
644 struct hlist_head mask_notifier_list
;
646 /* fields used by HYPER-V emulation */
651 #ifdef CONFIG_KVM_MMU_AUDIT
655 bool boot_vcpu_runs_old_kvmclock
;
661 u32 mmu_shadow_zapped
;
669 u32 remote_tlb_flush
;
673 struct kvm_vcpu_stat
{
683 u32 irq_window_exits
;
684 u32 nmi_window_exits
;
686 u32 halt_successful_poll
;
688 u32 request_irq_exits
;
690 u32 host_state_reload
;
694 u32 insn_emulation_fail
;
700 struct x86_instruction_info
;
708 struct kvm_lapic_irq
{
720 int (*cpu_has_kvm_support
)(void); /* __init */
721 int (*disabled_by_bios
)(void); /* __init */
722 int (*hardware_enable
)(void);
723 void (*hardware_disable
)(void);
724 void (*check_processor_compatibility
)(void *rtn
);
725 int (*hardware_setup
)(void); /* __init */
726 void (*hardware_unsetup
)(void); /* __exit */
727 bool (*cpu_has_accelerated_tpr
)(void);
728 bool (*cpu_has_high_real_mode_segbase
)(void);
729 void (*cpuid_update
)(struct kvm_vcpu
*vcpu
);
731 /* Create, but do not attach this VCPU */
732 struct kvm_vcpu
*(*vcpu_create
)(struct kvm
*kvm
, unsigned id
);
733 void (*vcpu_free
)(struct kvm_vcpu
*vcpu
);
734 void (*vcpu_reset
)(struct kvm_vcpu
*vcpu
, bool init_event
);
736 void (*prepare_guest_switch
)(struct kvm_vcpu
*vcpu
);
737 void (*vcpu_load
)(struct kvm_vcpu
*vcpu
, int cpu
);
738 void (*vcpu_put
)(struct kvm_vcpu
*vcpu
);
740 void (*update_db_bp_intercept
)(struct kvm_vcpu
*vcpu
);
741 int (*get_msr
)(struct kvm_vcpu
*vcpu
, struct msr_data
*msr
);
742 int (*set_msr
)(struct kvm_vcpu
*vcpu
, struct msr_data
*msr
);
743 u64 (*get_segment_base
)(struct kvm_vcpu
*vcpu
, int seg
);
744 void (*get_segment
)(struct kvm_vcpu
*vcpu
,
745 struct kvm_segment
*var
, int seg
);
746 int (*get_cpl
)(struct kvm_vcpu
*vcpu
);
747 void (*set_segment
)(struct kvm_vcpu
*vcpu
,
748 struct kvm_segment
*var
, int seg
);
749 void (*get_cs_db_l_bits
)(struct kvm_vcpu
*vcpu
, int *db
, int *l
);
750 void (*decache_cr0_guest_bits
)(struct kvm_vcpu
*vcpu
);
751 void (*decache_cr3
)(struct kvm_vcpu
*vcpu
);
752 void (*decache_cr4_guest_bits
)(struct kvm_vcpu
*vcpu
);
753 void (*set_cr0
)(struct kvm_vcpu
*vcpu
, unsigned long cr0
);
754 void (*set_cr3
)(struct kvm_vcpu
*vcpu
, unsigned long cr3
);
755 int (*set_cr4
)(struct kvm_vcpu
*vcpu
, unsigned long cr4
);
756 void (*set_efer
)(struct kvm_vcpu
*vcpu
, u64 efer
);
757 void (*get_idt
)(struct kvm_vcpu
*vcpu
, struct desc_ptr
*dt
);
758 void (*set_idt
)(struct kvm_vcpu
*vcpu
, struct desc_ptr
*dt
);
759 void (*get_gdt
)(struct kvm_vcpu
*vcpu
, struct desc_ptr
*dt
);
760 void (*set_gdt
)(struct kvm_vcpu
*vcpu
, struct desc_ptr
*dt
);
761 u64 (*get_dr6
)(struct kvm_vcpu
*vcpu
);
762 void (*set_dr6
)(struct kvm_vcpu
*vcpu
, unsigned long value
);
763 void (*sync_dirty_debug_regs
)(struct kvm_vcpu
*vcpu
);
764 void (*set_dr7
)(struct kvm_vcpu
*vcpu
, unsigned long value
);
765 void (*cache_reg
)(struct kvm_vcpu
*vcpu
, enum kvm_reg reg
);
766 unsigned long (*get_rflags
)(struct kvm_vcpu
*vcpu
);
767 void (*set_rflags
)(struct kvm_vcpu
*vcpu
, unsigned long rflags
);
768 void (*fpu_activate
)(struct kvm_vcpu
*vcpu
);
769 void (*fpu_deactivate
)(struct kvm_vcpu
*vcpu
);
771 void (*tlb_flush
)(struct kvm_vcpu
*vcpu
);
773 void (*run
)(struct kvm_vcpu
*vcpu
);
774 int (*handle_exit
)(struct kvm_vcpu
*vcpu
);
775 void (*skip_emulated_instruction
)(struct kvm_vcpu
*vcpu
);
776 void (*set_interrupt_shadow
)(struct kvm_vcpu
*vcpu
, int mask
);
777 u32 (*get_interrupt_shadow
)(struct kvm_vcpu
*vcpu
);
778 void (*patch_hypercall
)(struct kvm_vcpu
*vcpu
,
779 unsigned char *hypercall_addr
);
780 void (*set_irq
)(struct kvm_vcpu
*vcpu
);
781 void (*set_nmi
)(struct kvm_vcpu
*vcpu
);
782 void (*queue_exception
)(struct kvm_vcpu
*vcpu
, unsigned nr
,
783 bool has_error_code
, u32 error_code
,
785 void (*cancel_injection
)(struct kvm_vcpu
*vcpu
);
786 int (*interrupt_allowed
)(struct kvm_vcpu
*vcpu
);
787 int (*nmi_allowed
)(struct kvm_vcpu
*vcpu
);
788 bool (*get_nmi_mask
)(struct kvm_vcpu
*vcpu
);
789 void (*set_nmi_mask
)(struct kvm_vcpu
*vcpu
, bool masked
);
790 void (*enable_nmi_window
)(struct kvm_vcpu
*vcpu
);
791 void (*enable_irq_window
)(struct kvm_vcpu
*vcpu
);
792 void (*update_cr8_intercept
)(struct kvm_vcpu
*vcpu
, int tpr
, int irr
);
793 int (*vm_has_apicv
)(struct kvm
*kvm
);
794 void (*hwapic_irr_update
)(struct kvm_vcpu
*vcpu
, int max_irr
);
795 void (*hwapic_isr_update
)(struct kvm
*kvm
, int isr
);
796 void (*load_eoi_exitmap
)(struct kvm_vcpu
*vcpu
, u64
*eoi_exit_bitmap
);
797 void (*set_virtual_x2apic_mode
)(struct kvm_vcpu
*vcpu
, bool set
);
798 void (*set_apic_access_page_addr
)(struct kvm_vcpu
*vcpu
, hpa_t hpa
);
799 void (*deliver_posted_interrupt
)(struct kvm_vcpu
*vcpu
, int vector
);
800 void (*sync_pir_to_irr
)(struct kvm_vcpu
*vcpu
);
801 int (*set_tss_addr
)(struct kvm
*kvm
, unsigned int addr
);
802 int (*get_tdp_level
)(void);
803 u64 (*get_mt_mask
)(struct kvm_vcpu
*vcpu
, gfn_t gfn
, bool is_mmio
);
804 int (*get_lpage_level
)(void);
805 bool (*rdtscp_supported
)(void);
806 bool (*invpcid_supported
)(void);
807 void (*adjust_tsc_offset
)(struct kvm_vcpu
*vcpu
, s64 adjustment
, bool host
);
809 void (*set_tdp_cr3
)(struct kvm_vcpu
*vcpu
, unsigned long cr3
);
811 void (*set_supported_cpuid
)(u32 func
, struct kvm_cpuid_entry2
*entry
);
813 bool (*has_wbinvd_exit
)(void);
815 void (*set_tsc_khz
)(struct kvm_vcpu
*vcpu
, u32 user_tsc_khz
, bool scale
);
816 u64 (*read_tsc_offset
)(struct kvm_vcpu
*vcpu
);
817 void (*write_tsc_offset
)(struct kvm_vcpu
*vcpu
, u64 offset
);
819 u64 (*compute_tsc_offset
)(struct kvm_vcpu
*vcpu
, u64 target_tsc
);
820 u64 (*read_l1_tsc
)(struct kvm_vcpu
*vcpu
, u64 host_tsc
);
822 void (*get_exit_info
)(struct kvm_vcpu
*vcpu
, u64
*info1
, u64
*info2
);
824 int (*check_intercept
)(struct kvm_vcpu
*vcpu
,
825 struct x86_instruction_info
*info
,
826 enum x86_intercept_stage stage
);
827 void (*handle_external_intr
)(struct kvm_vcpu
*vcpu
);
828 bool (*mpx_supported
)(void);
829 bool (*xsaves_supported
)(void);
831 int (*check_nested_events
)(struct kvm_vcpu
*vcpu
, bool external_intr
);
833 void (*sched_in
)(struct kvm_vcpu
*kvm
, int cpu
);
836 * Arch-specific dirty logging hooks. These hooks are only supposed to
837 * be valid if the specific arch has hardware-accelerated dirty logging
838 * mechanism. Currently only for PML on VMX.
840 * - slot_enable_log_dirty:
841 * called when enabling log dirty mode for the slot.
842 * - slot_disable_log_dirty:
843 * called when disabling log dirty mode for the slot.
844 * also called when slot is created with log dirty disabled.
846 * called before reporting dirty_bitmap to userspace.
847 * - enable_log_dirty_pt_masked:
848 * called when reenabling log dirty for the GFNs in the mask after
849 * corresponding bits are cleared in slot->dirty_bitmap.
851 void (*slot_enable_log_dirty
)(struct kvm
*kvm
,
852 struct kvm_memory_slot
*slot
);
853 void (*slot_disable_log_dirty
)(struct kvm
*kvm
,
854 struct kvm_memory_slot
*slot
);
855 void (*flush_log_dirty
)(struct kvm
*kvm
);
856 void (*enable_log_dirty_pt_masked
)(struct kvm
*kvm
,
857 struct kvm_memory_slot
*slot
,
858 gfn_t offset
, unsigned long mask
);
859 /* pmu operations of sub-arch */
860 const struct kvm_pmu_ops
*pmu_ops
;
863 struct kvm_arch_async_pf
{
870 extern struct kvm_x86_ops
*kvm_x86_ops
;
872 static inline void adjust_tsc_offset_guest(struct kvm_vcpu
*vcpu
,
875 kvm_x86_ops
->adjust_tsc_offset(vcpu
, adjustment
, false);
878 static inline void adjust_tsc_offset_host(struct kvm_vcpu
*vcpu
, s64 adjustment
)
880 kvm_x86_ops
->adjust_tsc_offset(vcpu
, adjustment
, true);
883 int kvm_mmu_module_init(void);
884 void kvm_mmu_module_exit(void);
886 void kvm_mmu_destroy(struct kvm_vcpu
*vcpu
);
887 int kvm_mmu_create(struct kvm_vcpu
*vcpu
);
888 void kvm_mmu_setup(struct kvm_vcpu
*vcpu
);
889 void kvm_mmu_set_mask_ptes(u64 user_mask
, u64 accessed_mask
,
890 u64 dirty_mask
, u64 nx_mask
, u64 x_mask
);
892 void kvm_mmu_reset_context(struct kvm_vcpu
*vcpu
);
893 void kvm_mmu_slot_remove_write_access(struct kvm
*kvm
,
894 struct kvm_memory_slot
*memslot
);
895 void kvm_mmu_zap_collapsible_sptes(struct kvm
*kvm
,
896 const struct kvm_memory_slot
*memslot
);
897 void kvm_mmu_slot_leaf_clear_dirty(struct kvm
*kvm
,
898 struct kvm_memory_slot
*memslot
);
899 void kvm_mmu_slot_largepage_remove_write_access(struct kvm
*kvm
,
900 struct kvm_memory_slot
*memslot
);
901 void kvm_mmu_slot_set_dirty(struct kvm
*kvm
,
902 struct kvm_memory_slot
*memslot
);
903 void kvm_mmu_clear_dirty_pt_masked(struct kvm
*kvm
,
904 struct kvm_memory_slot
*slot
,
905 gfn_t gfn_offset
, unsigned long mask
);
906 void kvm_mmu_zap_all(struct kvm
*kvm
);
907 void kvm_mmu_invalidate_mmio_sptes(struct kvm
*kvm
, struct kvm_memslots
*slots
);
908 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm
*kvm
);
909 void kvm_mmu_change_mmu_pages(struct kvm
*kvm
, unsigned int kvm_nr_mmu_pages
);
911 int load_pdptrs(struct kvm_vcpu
*vcpu
, struct kvm_mmu
*mmu
, unsigned long cr3
);
913 int emulator_write_phys(struct kvm_vcpu
*vcpu
, gpa_t gpa
,
914 const void *val
, int bytes
);
916 struct kvm_irq_mask_notifier
{
917 void (*func
)(struct kvm_irq_mask_notifier
*kimn
, bool masked
);
919 struct hlist_node link
;
922 void kvm_register_irq_mask_notifier(struct kvm
*kvm
, int irq
,
923 struct kvm_irq_mask_notifier
*kimn
);
924 void kvm_unregister_irq_mask_notifier(struct kvm
*kvm
, int irq
,
925 struct kvm_irq_mask_notifier
*kimn
);
926 void kvm_fire_mask_notifiers(struct kvm
*kvm
, unsigned irqchip
, unsigned pin
,
929 extern bool tdp_enabled
;
931 u64
vcpu_tsc_khz(struct kvm_vcpu
*vcpu
);
933 /* control of guest tsc rate supported? */
934 extern bool kvm_has_tsc_control
;
935 /* minimum supported tsc_khz for guests */
936 extern u32 kvm_min_guest_tsc_khz
;
937 /* maximum supported tsc_khz for guests */
938 extern u32 kvm_max_guest_tsc_khz
;
940 enum emulation_result
{
941 EMULATE_DONE
, /* no further processing */
942 EMULATE_USER_EXIT
, /* kvm_run ready for userspace exit */
943 EMULATE_FAIL
, /* can't emulate this instruction */
946 #define EMULTYPE_NO_DECODE (1 << 0)
947 #define EMULTYPE_TRAP_UD (1 << 1)
948 #define EMULTYPE_SKIP (1 << 2)
949 #define EMULTYPE_RETRY (1 << 3)
950 #define EMULTYPE_NO_REEXECUTE (1 << 4)
951 int x86_emulate_instruction(struct kvm_vcpu
*vcpu
, unsigned long cr2
,
952 int emulation_type
, void *insn
, int insn_len
);
954 static inline int emulate_instruction(struct kvm_vcpu
*vcpu
,
957 return x86_emulate_instruction(vcpu
, 0, emulation_type
, NULL
, 0);
960 void kvm_enable_efer_bits(u64
);
961 bool kvm_valid_efer(struct kvm_vcpu
*vcpu
, u64 efer
);
962 int kvm_get_msr(struct kvm_vcpu
*vcpu
, struct msr_data
*msr
);
963 int kvm_set_msr(struct kvm_vcpu
*vcpu
, struct msr_data
*msr
);
965 struct x86_emulate_ctxt
;
967 int kvm_fast_pio_out(struct kvm_vcpu
*vcpu
, int size
, unsigned short port
);
968 void kvm_emulate_cpuid(struct kvm_vcpu
*vcpu
);
969 int kvm_emulate_halt(struct kvm_vcpu
*vcpu
);
970 int kvm_vcpu_halt(struct kvm_vcpu
*vcpu
);
971 int kvm_emulate_wbinvd(struct kvm_vcpu
*vcpu
);
973 void kvm_get_segment(struct kvm_vcpu
*vcpu
, struct kvm_segment
*var
, int seg
);
974 int kvm_load_segment_descriptor(struct kvm_vcpu
*vcpu
, u16 selector
, int seg
);
975 void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu
*vcpu
, u8 vector
);
977 int kvm_task_switch(struct kvm_vcpu
*vcpu
, u16 tss_selector
, int idt_index
,
978 int reason
, bool has_error_code
, u32 error_code
);
980 int kvm_set_cr0(struct kvm_vcpu
*vcpu
, unsigned long cr0
);
981 int kvm_set_cr3(struct kvm_vcpu
*vcpu
, unsigned long cr3
);
982 int kvm_set_cr4(struct kvm_vcpu
*vcpu
, unsigned long cr4
);
983 int kvm_set_cr8(struct kvm_vcpu
*vcpu
, unsigned long cr8
);
984 int kvm_set_dr(struct kvm_vcpu
*vcpu
, int dr
, unsigned long val
);
985 int kvm_get_dr(struct kvm_vcpu
*vcpu
, int dr
, unsigned long *val
);
986 unsigned long kvm_get_cr8(struct kvm_vcpu
*vcpu
);
987 void kvm_lmsw(struct kvm_vcpu
*vcpu
, unsigned long msw
);
988 void kvm_get_cs_db_l_bits(struct kvm_vcpu
*vcpu
, int *db
, int *l
);
989 int kvm_set_xcr(struct kvm_vcpu
*vcpu
, u32 index
, u64 xcr
);
991 int kvm_get_msr_common(struct kvm_vcpu
*vcpu
, struct msr_data
*msr
);
992 int kvm_set_msr_common(struct kvm_vcpu
*vcpu
, struct msr_data
*msr
);
994 unsigned long kvm_get_rflags(struct kvm_vcpu
*vcpu
);
995 void kvm_set_rflags(struct kvm_vcpu
*vcpu
, unsigned long rflags
);
996 bool kvm_rdpmc(struct kvm_vcpu
*vcpu
);
998 void kvm_queue_exception(struct kvm_vcpu
*vcpu
, unsigned nr
);
999 void kvm_queue_exception_e(struct kvm_vcpu
*vcpu
, unsigned nr
, u32 error_code
);
1000 void kvm_requeue_exception(struct kvm_vcpu
*vcpu
, unsigned nr
);
1001 void kvm_requeue_exception_e(struct kvm_vcpu
*vcpu
, unsigned nr
, u32 error_code
);
1002 void kvm_inject_page_fault(struct kvm_vcpu
*vcpu
, struct x86_exception
*fault
);
1003 int kvm_read_guest_page_mmu(struct kvm_vcpu
*vcpu
, struct kvm_mmu
*mmu
,
1004 gfn_t gfn
, void *data
, int offset
, int len
,
1006 bool kvm_require_cpl(struct kvm_vcpu
*vcpu
, int required_cpl
);
1007 bool kvm_require_dr(struct kvm_vcpu
*vcpu
, int dr
);
1009 static inline int __kvm_irq_line_state(unsigned long *irq_state
,
1010 int irq_source_id
, int level
)
1012 /* Logical OR for level trig interrupt */
1014 __set_bit(irq_source_id
, irq_state
);
1016 __clear_bit(irq_source_id
, irq_state
);
1018 return !!(*irq_state
);
1021 int kvm_pic_set_irq(struct kvm_pic
*pic
, int irq
, int irq_source_id
, int level
);
1022 void kvm_pic_clear_all(struct kvm_pic
*pic
, int irq_source_id
);
1024 void kvm_inject_nmi(struct kvm_vcpu
*vcpu
);
1026 void kvm_mmu_pte_write(struct kvm_vcpu
*vcpu
, gpa_t gpa
,
1027 const u8
*new, int bytes
);
1028 int kvm_mmu_unprotect_page(struct kvm
*kvm
, gfn_t gfn
);
1029 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu
*vcpu
, gva_t gva
);
1030 void __kvm_mmu_free_some_pages(struct kvm_vcpu
*vcpu
);
1031 int kvm_mmu_load(struct kvm_vcpu
*vcpu
);
1032 void kvm_mmu_unload(struct kvm_vcpu
*vcpu
);
1033 void kvm_mmu_sync_roots(struct kvm_vcpu
*vcpu
);
1034 gpa_t
translate_nested_gpa(struct kvm_vcpu
*vcpu
, gpa_t gpa
, u32 access
,
1035 struct x86_exception
*exception
);
1036 gpa_t
kvm_mmu_gva_to_gpa_read(struct kvm_vcpu
*vcpu
, gva_t gva
,
1037 struct x86_exception
*exception
);
1038 gpa_t
kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu
*vcpu
, gva_t gva
,
1039 struct x86_exception
*exception
);
1040 gpa_t
kvm_mmu_gva_to_gpa_write(struct kvm_vcpu
*vcpu
, gva_t gva
,
1041 struct x86_exception
*exception
);
1042 gpa_t
kvm_mmu_gva_to_gpa_system(struct kvm_vcpu
*vcpu
, gva_t gva
,
1043 struct x86_exception
*exception
);
1045 int kvm_emulate_hypercall(struct kvm_vcpu
*vcpu
);
1047 int kvm_mmu_page_fault(struct kvm_vcpu
*vcpu
, gva_t gva
, u32 error_code
,
1048 void *insn
, int insn_len
);
1049 void kvm_mmu_invlpg(struct kvm_vcpu
*vcpu
, gva_t gva
);
1050 void kvm_mmu_new_cr3(struct kvm_vcpu
*vcpu
);
1052 void kvm_enable_tdp(void);
1053 void kvm_disable_tdp(void);
1055 static inline gpa_t
translate_gpa(struct kvm_vcpu
*vcpu
, gpa_t gpa
, u32 access
,
1056 struct x86_exception
*exception
)
1061 static inline struct kvm_mmu_page
*page_header(hpa_t shadow_page
)
1063 struct page
*page
= pfn_to_page(shadow_page
>> PAGE_SHIFT
);
1065 return (struct kvm_mmu_page
*)page_private(page
);
1068 static inline u16
kvm_read_ldt(void)
1071 asm("sldt %0" : "=g"(ldt
));
1075 static inline void kvm_load_ldt(u16 sel
)
1077 asm("lldt %0" : : "rm"(sel
));
1080 #ifdef CONFIG_X86_64
1081 static inline unsigned long read_msr(unsigned long msr
)
1090 static inline u32
get_rdx_init_val(void)
1092 return 0x600; /* P6 family */
1095 static inline void kvm_inject_gp(struct kvm_vcpu
*vcpu
, u32 error_code
)
1097 kvm_queue_exception_e(vcpu
, GP_VECTOR
, error_code
);
1100 static inline u64
get_canonical(u64 la
)
1102 return ((int64_t)la
<< 16) >> 16;
1105 static inline bool is_noncanonical_address(u64 la
)
1107 #ifdef CONFIG_X86_64
1108 return get_canonical(la
) != la
;
1114 #define TSS_IOPB_BASE_OFFSET 0x66
1115 #define TSS_BASE_SIZE 0x68
1116 #define TSS_IOPB_SIZE (65536 / 8)
1117 #define TSS_REDIRECTION_SIZE (256 / 8)
1118 #define RMODE_TSS_SIZE \
1119 (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
1122 TASK_SWITCH_CALL
= 0,
1123 TASK_SWITCH_IRET
= 1,
1124 TASK_SWITCH_JMP
= 2,
1125 TASK_SWITCH_GATE
= 3,
1128 #define HF_GIF_MASK (1 << 0)
1129 #define HF_HIF_MASK (1 << 1)
1130 #define HF_VINTR_MASK (1 << 2)
1131 #define HF_NMI_MASK (1 << 3)
1132 #define HF_IRET_MASK (1 << 4)
1133 #define HF_GUEST_MASK (1 << 5) /* VCPU is in guest-mode */
1134 #define HF_SMM_MASK (1 << 6)
1135 #define HF_SMM_INSIDE_NMI_MASK (1 << 7)
1137 #define __KVM_VCPU_MULTIPLE_ADDRESS_SPACE
1138 #define KVM_ADDRESS_SPACE_NUM 2
1140 #define kvm_arch_vcpu_memslots_id(vcpu) ((vcpu)->arch.hflags & HF_SMM_MASK ? 1 : 0)
1141 #define kvm_memslots_for_spte_role(kvm, role) __kvm_memslots(kvm, (role).smm)
1144 * Hardware virtualization extension instructions may fault if a
1145 * reboot turns off virtualization while processes are running.
1146 * Trap the fault and ignore the instruction if that happens.
1148 asmlinkage
void kvm_spurious_fault(void);
1150 #define ____kvm_handle_fault_on_reboot(insn, cleanup_insn) \
1151 "666: " insn "\n\t" \
1153 ".pushsection .fixup, \"ax\" \n" \
1155 cleanup_insn "\n\t" \
1156 "cmpb $0, kvm_rebooting \n\t" \
1158 __ASM_SIZE(push) " $666b \n\t" \
1159 "call kvm_spurious_fault \n\t" \
1160 ".popsection \n\t" \
1161 _ASM_EXTABLE(666b, 667b)
1163 #define __kvm_handle_fault_on_reboot(insn) \
1164 ____kvm_handle_fault_on_reboot(insn, "")
1166 #define KVM_ARCH_WANT_MMU_NOTIFIER
1167 int kvm_unmap_hva(struct kvm
*kvm
, unsigned long hva
);
1168 int kvm_unmap_hva_range(struct kvm
*kvm
, unsigned long start
, unsigned long end
);
1169 int kvm_age_hva(struct kvm
*kvm
, unsigned long start
, unsigned long end
);
1170 int kvm_test_age_hva(struct kvm
*kvm
, unsigned long hva
);
1171 void kvm_set_spte_hva(struct kvm
*kvm
, unsigned long hva
, pte_t pte
);
1172 int kvm_cpu_has_injectable_intr(struct kvm_vcpu
*v
);
1173 int kvm_cpu_has_interrupt(struct kvm_vcpu
*vcpu
);
1174 int kvm_arch_interrupt_allowed(struct kvm_vcpu
*vcpu
);
1175 int kvm_cpu_get_interrupt(struct kvm_vcpu
*v
);
1176 void kvm_vcpu_reset(struct kvm_vcpu
*vcpu
, bool init_event
);
1177 void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu
*vcpu
);
1178 void kvm_arch_mmu_notifier_invalidate_page(struct kvm
*kvm
,
1179 unsigned long address
);
1181 void kvm_define_shared_msr(unsigned index
, u32 msr
);
1182 int kvm_set_shared_msr(unsigned index
, u64 val
, u64 mask
);
1184 unsigned long kvm_get_linear_rip(struct kvm_vcpu
*vcpu
);
1185 bool kvm_is_linear_rip(struct kvm_vcpu
*vcpu
, unsigned long linear_rip
);
1187 void kvm_arch_async_page_not_present(struct kvm_vcpu
*vcpu
,
1188 struct kvm_async_pf
*work
);
1189 void kvm_arch_async_page_present(struct kvm_vcpu
*vcpu
,
1190 struct kvm_async_pf
*work
);
1191 void kvm_arch_async_page_ready(struct kvm_vcpu
*vcpu
,
1192 struct kvm_async_pf
*work
);
1193 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu
*vcpu
);
1194 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu
*vcpu
, gfn_t gfn
);
1196 void kvm_complete_insn_gp(struct kvm_vcpu
*vcpu
, int err
);
1198 int kvm_is_in_guest(void);
1200 int __x86_set_memory_region(struct kvm
*kvm
,
1201 const struct kvm_userspace_memory_region
*mem
);
1202 int x86_set_memory_region(struct kvm
*kvm
,
1203 const struct kvm_userspace_memory_region
*mem
);
1205 #endif /* _ASM_X86_KVM_HOST_H */