1 /* SPDX-License-Identifier: GPL-2.0-only */
3 * Kernel-based Virtual Machine driver for Linux
5 * This header defines architecture specific interfaces, x86 version
8 #ifndef _ASM_X86_KVM_HOST_H
9 #define _ASM_X86_KVM_HOST_H
11 #include <linux/types.h>
13 #include <linux/mmu_notifier.h>
14 #include <linux/tracepoint.h>
15 #include <linux/cpumask.h>
16 #include <linux/irq_work.h>
17 #include <linux/irq.h>
19 #include <linux/kvm.h>
20 #include <linux/kvm_para.h>
21 #include <linux/kvm_types.h>
22 #include <linux/perf_event.h>
23 #include <linux/pvclock_gtod.h>
24 #include <linux/clocksource.h>
25 #include <linux/irqbypass.h>
26 #include <linux/hyperv.h>
29 #include <asm/pvclock-abi.h>
32 #include <asm/msr-index.h>
34 #include <asm/kvm_page_track.h>
35 #include <asm/kvm_vcpu_regs.h>
36 #include <asm/hyperv-tlfs.h>
38 #define __KVM_HAVE_ARCH_VCPU_DEBUGFS
40 #define KVM_MAX_VCPUS 288
41 #define KVM_SOFT_MAX_VCPUS 240
42 #define KVM_MAX_VCPU_ID 1023
43 #define KVM_USER_MEM_SLOTS 509
44 /* memory slots that are not exposed to userspace */
45 #define KVM_PRIVATE_MEM_SLOTS 3
46 #define KVM_MEM_SLOTS_NUM (KVM_USER_MEM_SLOTS + KVM_PRIVATE_MEM_SLOTS)
48 #define KVM_HALT_POLL_NS_DEFAULT 200000
50 #define KVM_IRQCHIP_NUM_PINS KVM_IOAPIC_NUM_PINS
52 /* x86-specific vcpu->requests bit members */
53 #define KVM_REQ_MIGRATE_TIMER KVM_ARCH_REQ(0)
54 #define KVM_REQ_REPORT_TPR_ACCESS KVM_ARCH_REQ(1)
55 #define KVM_REQ_TRIPLE_FAULT KVM_ARCH_REQ(2)
56 #define KVM_REQ_MMU_SYNC KVM_ARCH_REQ(3)
57 #define KVM_REQ_CLOCK_UPDATE KVM_ARCH_REQ(4)
58 #define KVM_REQ_LOAD_CR3 KVM_ARCH_REQ(5)
59 #define KVM_REQ_EVENT KVM_ARCH_REQ(6)
60 #define KVM_REQ_APF_HALT KVM_ARCH_REQ(7)
61 #define KVM_REQ_STEAL_UPDATE KVM_ARCH_REQ(8)
62 #define KVM_REQ_NMI KVM_ARCH_REQ(9)
63 #define KVM_REQ_PMU KVM_ARCH_REQ(10)
64 #define KVM_REQ_PMI KVM_ARCH_REQ(11)
65 #define KVM_REQ_SMI KVM_ARCH_REQ(12)
66 #define KVM_REQ_MASTERCLOCK_UPDATE KVM_ARCH_REQ(13)
67 #define KVM_REQ_MCLOCK_INPROGRESS \
68 KVM_ARCH_REQ_FLAGS(14, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
69 #define KVM_REQ_SCAN_IOAPIC \
70 KVM_ARCH_REQ_FLAGS(15, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
71 #define KVM_REQ_GLOBAL_CLOCK_UPDATE KVM_ARCH_REQ(16)
72 #define KVM_REQ_APIC_PAGE_RELOAD \
73 KVM_ARCH_REQ_FLAGS(17, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
74 #define KVM_REQ_HV_CRASH KVM_ARCH_REQ(18)
75 #define KVM_REQ_IOAPIC_EOI_EXIT KVM_ARCH_REQ(19)
76 #define KVM_REQ_HV_RESET KVM_ARCH_REQ(20)
77 #define KVM_REQ_HV_EXIT KVM_ARCH_REQ(21)
78 #define KVM_REQ_HV_STIMER KVM_ARCH_REQ(22)
79 #define KVM_REQ_LOAD_EOI_EXITMAP KVM_ARCH_REQ(23)
80 #define KVM_REQ_GET_VMCS12_PAGES KVM_ARCH_REQ(24)
82 #define CR0_RESERVED_BITS \
83 (~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
84 | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
85 | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
87 #define CR4_RESERVED_BITS \
88 (~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
89 | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE \
90 | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR | X86_CR4_PCIDE \
91 | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_FSGSBASE \
92 | X86_CR4_OSXMMEXCPT | X86_CR4_LA57 | X86_CR4_VMXE \
93 | X86_CR4_SMAP | X86_CR4_PKE | X86_CR4_UMIP))
95 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
99 #define INVALID_PAGE (~(hpa_t)0)
100 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
102 #define UNMAPPED_GVA (~(gpa_t)0)
104 /* KVM Hugepage definitions for x86 */
106 PT_PAGE_TABLE_LEVEL
= 1,
107 PT_DIRECTORY_LEVEL
= 2,
109 /* set max level to the biggest one */
110 PT_MAX_HUGEPAGE_LEVEL
= PT_PDPE_LEVEL
,
112 #define KVM_NR_PAGE_SIZES (PT_MAX_HUGEPAGE_LEVEL - \
113 PT_PAGE_TABLE_LEVEL + 1)
114 #define KVM_HPAGE_GFN_SHIFT(x) (((x) - 1) * 9)
115 #define KVM_HPAGE_SHIFT(x) (PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
116 #define KVM_HPAGE_SIZE(x) (1UL << KVM_HPAGE_SHIFT(x))
117 #define KVM_HPAGE_MASK(x) (~(KVM_HPAGE_SIZE(x) - 1))
118 #define KVM_PAGES_PER_HPAGE(x) (KVM_HPAGE_SIZE(x) / PAGE_SIZE)
120 static inline gfn_t
gfn_to_index(gfn_t gfn
, gfn_t base_gfn
, int level
)
122 /* KVM_HPAGE_GFN_SHIFT(PT_PAGE_TABLE_LEVEL) must be 0. */
123 return (gfn
>> KVM_HPAGE_GFN_SHIFT(level
)) -
124 (base_gfn
>> KVM_HPAGE_GFN_SHIFT(level
));
127 #define KVM_PERMILLE_MMU_PAGES 20
128 #define KVM_MIN_ALLOC_MMU_PAGES 64UL
129 #define KVM_MMU_HASH_SHIFT 12
130 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
131 #define KVM_MIN_FREE_MMU_PAGES 5
132 #define KVM_REFILL_PAGES 25
133 #define KVM_MAX_CPUID_ENTRIES 80
134 #define KVM_NR_FIXED_MTRR_REGION 88
135 #define KVM_NR_VAR_MTRR 8
137 #define ASYNC_PF_PER_VCPU 64
140 VCPU_REGS_RAX
= __VCPU_REGS_RAX
,
141 VCPU_REGS_RCX
= __VCPU_REGS_RCX
,
142 VCPU_REGS_RDX
= __VCPU_REGS_RDX
,
143 VCPU_REGS_RBX
= __VCPU_REGS_RBX
,
144 VCPU_REGS_RSP
= __VCPU_REGS_RSP
,
145 VCPU_REGS_RBP
= __VCPU_REGS_RBP
,
146 VCPU_REGS_RSI
= __VCPU_REGS_RSI
,
147 VCPU_REGS_RDI
= __VCPU_REGS_RDI
,
149 VCPU_REGS_R8
= __VCPU_REGS_R8
,
150 VCPU_REGS_R9
= __VCPU_REGS_R9
,
151 VCPU_REGS_R10
= __VCPU_REGS_R10
,
152 VCPU_REGS_R11
= __VCPU_REGS_R11
,
153 VCPU_REGS_R12
= __VCPU_REGS_R12
,
154 VCPU_REGS_R13
= __VCPU_REGS_R13
,
155 VCPU_REGS_R14
= __VCPU_REGS_R14
,
156 VCPU_REGS_R15
= __VCPU_REGS_R15
,
163 VCPU_EXREG_PDPTR
= NR_VCPU_REGS
,
180 #include <asm/kvm_emulate.h>
182 #define KVM_NR_MEM_OBJS 40
184 #define KVM_NR_DB_REGS 4
186 #define DR6_BD (1 << 13)
187 #define DR6_BS (1 << 14)
188 #define DR6_BT (1 << 15)
189 #define DR6_RTM (1 << 16)
190 #define DR6_FIXED_1 0xfffe0ff0
191 #define DR6_INIT 0xffff0ff0
192 #define DR6_VOLATILE 0x0001e00f
194 #define DR7_BP_EN_MASK 0x000000ff
195 #define DR7_GE (1 << 9)
196 #define DR7_GD (1 << 13)
197 #define DR7_FIXED_1 0x00000400
198 #define DR7_VOLATILE 0xffff2bff
200 #define PFERR_PRESENT_BIT 0
201 #define PFERR_WRITE_BIT 1
202 #define PFERR_USER_BIT 2
203 #define PFERR_RSVD_BIT 3
204 #define PFERR_FETCH_BIT 4
205 #define PFERR_PK_BIT 5
206 #define PFERR_GUEST_FINAL_BIT 32
207 #define PFERR_GUEST_PAGE_BIT 33
209 #define PFERR_PRESENT_MASK (1U << PFERR_PRESENT_BIT)
210 #define PFERR_WRITE_MASK (1U << PFERR_WRITE_BIT)
211 #define PFERR_USER_MASK (1U << PFERR_USER_BIT)
212 #define PFERR_RSVD_MASK (1U << PFERR_RSVD_BIT)
213 #define PFERR_FETCH_MASK (1U << PFERR_FETCH_BIT)
214 #define PFERR_PK_MASK (1U << PFERR_PK_BIT)
215 #define PFERR_GUEST_FINAL_MASK (1ULL << PFERR_GUEST_FINAL_BIT)
216 #define PFERR_GUEST_PAGE_MASK (1ULL << PFERR_GUEST_PAGE_BIT)
218 #define PFERR_NESTED_GUEST_PAGE (PFERR_GUEST_PAGE_MASK | \
223 * The mask used to denote special SPTEs, which can be either MMIO SPTEs or
224 * Access Tracking SPTEs. We use bit 62 instead of bit 63 to avoid conflicting
225 * with the SVE bit in EPT PTEs.
227 #define SPTE_SPECIAL_MASK (1ULL << 62)
229 /* apic attention bits */
230 #define KVM_APIC_CHECK_VAPIC 0
232 * The following bit is set with PV-EOI, unset on EOI.
233 * We detect PV-EOI changes by guest by comparing
234 * this bit with PV-EOI in guest memory.
235 * See the implementation in apic_update_pv_eoi.
237 #define KVM_APIC_PV_EOI_PENDING 1
239 struct kvm_kernel_irq_routing_entry
;
242 * We don't want allocation failures within the mmu code, so we preallocate
243 * enough memory for a single page fault in a cache.
245 struct kvm_mmu_memory_cache
{
247 void *objects
[KVM_NR_MEM_OBJS
];
251 * the pages used as guest page table on soft mmu are tracked by
252 * kvm_memory_slot.arch.gfn_track which is 16 bits, so the role bits used
253 * by indirect shadow page can not be more than 15 bits.
255 * Currently, we used 14 bits that are @level, @gpte_is_8_bytes, @quadrant, @access,
256 * @nxe, @cr0_wp, @smep_andnot_wp and @smap_andnot_wp.
258 union kvm_mmu_page_role
{
262 unsigned gpte_is_8_bytes
:1;
269 unsigned smep_andnot_wp
:1;
270 unsigned smap_andnot_wp
:1;
271 unsigned ad_disabled
:1;
272 unsigned guest_mode
:1;
276 * This is left at the top of the word so that
277 * kvm_memslots_for_spte_role can extract it with a
278 * simple shift. While there is room, give it a whole
279 * byte so it is also faster to load it from memory.
285 union kvm_mmu_extended_role
{
287 * This structure complements kvm_mmu_page_role caching everything needed for
288 * MMU configuration. If nothing in both these structures changed, MMU
289 * re-configuration can be skipped. @valid bit is set on first usage so we don't
290 * treat all-zero structure as valid data.
294 unsigned int valid
:1;
295 unsigned int execonly
:1;
296 unsigned int cr0_pg
:1;
297 unsigned int cr4_pae
:1;
298 unsigned int cr4_pse
:1;
299 unsigned int cr4_pke
:1;
300 unsigned int cr4_smap
:1;
301 unsigned int cr4_smep
:1;
302 unsigned int cr4_la57
:1;
303 unsigned int maxphyaddr
:6;
310 union kvm_mmu_page_role base
;
311 union kvm_mmu_extended_role ext
;
315 struct kvm_rmap_head
{
319 struct kvm_mmu_page
{
320 struct list_head link
;
321 struct hlist_node hash_link
;
324 bool lpage_disallowed
; /* Can't be replaced by an equiv large page */
327 * The following two entries are used to key the shadow page in the
330 union kvm_mmu_page_role role
;
334 /* hold the gfn of each spte inside spt */
336 int root_count
; /* Currently serving as active root */
337 unsigned int unsync_children
;
338 struct kvm_rmap_head parent_ptes
; /* rmap pointers to parent sptes */
339 unsigned long mmu_valid_gen
;
340 DECLARE_BITMAP(unsync_child_bitmap
, 512);
344 * Used out of the mmu-lock to avoid reading spte values while an
345 * update is in progress; see the comments in __get_spte_lockless().
347 int clear_spte_count
;
350 /* Number of writes since the last time traversal visited this page. */
351 atomic_t write_flooding_count
;
354 struct kvm_pio_request
{
355 unsigned long linear_rip
;
362 #define PT64_ROOT_MAX_LEVEL 5
364 struct rsvd_bits_validate
{
365 u64 rsvd_bits_mask
[2][PT64_ROOT_MAX_LEVEL
];
369 struct kvm_mmu_root_info
{
374 #define KVM_MMU_ROOT_INFO_INVALID \
375 ((struct kvm_mmu_root_info) { .cr3 = INVALID_PAGE, .hpa = INVALID_PAGE })
377 #define KVM_MMU_NUM_PREV_ROOTS 3
380 * x86 supports 4 paging modes (5-level 64-bit, 4-level 64-bit, 3-level 32-bit,
381 * and 2-level 32-bit). The kvm_mmu structure abstracts the details of the
385 void (*set_cr3
)(struct kvm_vcpu
*vcpu
, unsigned long root
);
386 unsigned long (*get_cr3
)(struct kvm_vcpu
*vcpu
);
387 u64 (*get_pdptr
)(struct kvm_vcpu
*vcpu
, int index
);
388 int (*page_fault
)(struct kvm_vcpu
*vcpu
, gva_t gva
, u32 err
,
390 void (*inject_page_fault
)(struct kvm_vcpu
*vcpu
,
391 struct x86_exception
*fault
);
392 gpa_t (*gva_to_gpa
)(struct kvm_vcpu
*vcpu
, gva_t gva
, u32 access
,
393 struct x86_exception
*exception
);
394 gpa_t (*translate_gpa
)(struct kvm_vcpu
*vcpu
, gpa_t gpa
, u32 access
,
395 struct x86_exception
*exception
);
396 int (*sync_page
)(struct kvm_vcpu
*vcpu
,
397 struct kvm_mmu_page
*sp
);
398 void (*invlpg
)(struct kvm_vcpu
*vcpu
, gva_t gva
, hpa_t root_hpa
);
399 void (*update_pte
)(struct kvm_vcpu
*vcpu
, struct kvm_mmu_page
*sp
,
400 u64
*spte
, const void *pte
);
403 union kvm_mmu_role mmu_role
;
405 u8 shadow_root_level
;
408 struct kvm_mmu_root_info prev_roots
[KVM_MMU_NUM_PREV_ROOTS
];
411 * Bitmap; bit set = permission fault
412 * Byte index: page fault error code [4:1]
413 * Bit index: pte permissions in ACC_* format
418 * The pkru_mask indicates if protection key checks are needed. It
419 * consists of 16 domains indexed by page fault error code bits [4:1],
420 * with PFEC.RSVD replaced by ACC_USER_MASK from the page tables.
421 * Each domain has 2 bits which are ANDed with AD and WD from PKRU.
429 * check zero bits on shadow page table entries, these
430 * bits include not only hardware reserved bits but also
431 * the bits spte never used.
433 struct rsvd_bits_validate shadow_zero_check
;
435 struct rsvd_bits_validate guest_rsvd_check
;
437 /* Can have large pages at levels 2..last_nonleaf_level-1. */
438 u8 last_nonleaf_level
;
442 u64 pdptrs
[4]; /* pae */
445 struct kvm_tlb_range
{
460 struct perf_event
*perf_event
;
461 struct kvm_vcpu
*vcpu
;
465 unsigned nr_arch_gp_counters
;
466 unsigned nr_arch_fixed_counters
;
467 unsigned available_event_types
;
472 u64 counter_bitmask
[2];
473 u64 global_ctrl_mask
;
474 u64 global_ovf_ctrl_mask
;
477 struct kvm_pmc gp_counters
[INTEL_PMC_MAX_GENERIC
];
478 struct kvm_pmc fixed_counters
[INTEL_PMC_MAX_FIXED
];
479 struct irq_work irq_work
;
486 KVM_DEBUGREG_BP_ENABLED
= 1,
487 KVM_DEBUGREG_WONT_EXIT
= 2,
488 KVM_DEBUGREG_RELOAD
= 4,
491 struct kvm_mtrr_range
{
494 struct list_head node
;
498 struct kvm_mtrr_range var_ranges
[KVM_NR_VAR_MTRR
];
499 mtrr_type fixed_ranges
[KVM_NR_FIXED_MTRR_REGION
];
502 struct list_head head
;
505 /* Hyper-V SynIC timer */
506 struct kvm_vcpu_hv_stimer
{
507 struct hrtimer timer
;
509 union hv_stimer_config config
;
512 struct hv_message msg
;
516 /* Hyper-V synthetic interrupt controller (SynIC)*/
517 struct kvm_vcpu_hv_synic
{
522 atomic64_t sint
[HV_SYNIC_SINT_COUNT
];
523 atomic_t sint_to_gsi
[HV_SYNIC_SINT_COUNT
];
524 DECLARE_BITMAP(auto_eoi_bitmap
, 256);
525 DECLARE_BITMAP(vec_bitmap
, 256);
527 bool dont_zero_synic_pages
;
530 /* Hyper-V per vcpu emulation context */
535 struct kvm_vcpu_hv_synic synic
;
536 struct kvm_hyperv_exit exit
;
537 struct kvm_vcpu_hv_stimer stimer
[HV_SYNIC_STIMER_COUNT
];
538 DECLARE_BITMAP(stimer_pending_bitmap
, HV_SYNIC_STIMER_COUNT
);
542 struct kvm_vcpu_arch
{
544 * rip and regs accesses must go through
545 * kvm_{register,rip}_{read,write} functions.
547 unsigned long regs
[NR_VCPU_REGS
];
552 unsigned long cr0_guest_owned_bits
;
556 unsigned long cr4_guest_owned_bits
;
562 struct kvm_lapic
*apic
; /* kernel irqchip context */
564 bool load_eoi_exitmap_pending
;
565 DECLARE_BITMAP(ioapic_handled_vectors
, 256);
566 unsigned long apic_attention
;
567 int32_t apic_arb_prio
;
569 u64 ia32_misc_enable_msr
;
572 bool tpr_access_reporting
;
574 u64 microcode_version
;
575 u64 arch_capabilities
;
578 * Paging state of the vcpu
580 * If the vcpu runs in guest mode with two level paging this still saves
581 * the paging mode of the l1 guest. This context is always used to
586 /* Non-nested MMU for L1 */
587 struct kvm_mmu root_mmu
;
589 /* L1 MMU when running nested */
590 struct kvm_mmu guest_mmu
;
593 * Paging state of an L2 guest (used for nested npt)
595 * This context will save all necessary information to walk page tables
596 * of the an L2 guest. This context is only initialized for page table
597 * walking and not for faulting since we never handle l2 page faults on
600 struct kvm_mmu nested_mmu
;
603 * Pointer to the mmu context currently used for
604 * gva_to_gpa translations.
606 struct kvm_mmu
*walk_mmu
;
608 struct kvm_mmu_memory_cache mmu_pte_list_desc_cache
;
609 struct kvm_mmu_memory_cache mmu_page_cache
;
610 struct kvm_mmu_memory_cache mmu_page_header_cache
;
613 * QEMU userspace and the guest each have their own FPU state.
614 * In vcpu_run, we switch between the user and guest FPU contexts.
615 * While running a VCPU, the VCPU thread will have the guest FPU
618 * Note that while the PKRU state lives inside the fpu registers,
619 * it is switched out separately at VMENTER and VMEXIT time. The
620 * "guest_fpu" state here contains the guest FPU context, with the
623 struct fpu
*user_fpu
;
624 struct fpu
*guest_fpu
;
627 u64 guest_supported_xcr0
;
628 u32 guest_xstate_size
;
630 struct kvm_pio_request pio
;
633 u8 event_exit_inst_len
;
635 struct kvm_queued_exception
{
641 unsigned long payload
;
646 struct kvm_queued_interrupt
{
652 int halt_request
; /* real mode on Intel only */
655 struct kvm_cpuid_entry2 cpuid_entries
[KVM_MAX_CPUID_ENTRIES
];
659 /* emulate context */
661 struct x86_emulate_ctxt emulate_ctxt
;
662 bool emulate_regs_need_sync_to_vcpu
;
663 bool emulate_regs_need_sync_from_vcpu
;
664 int (*complete_userspace_io
)(struct kvm_vcpu
*vcpu
);
667 struct pvclock_vcpu_time_info hv_clock
;
668 unsigned int hw_tsc_khz
;
669 struct gfn_to_hva_cache pv_time
;
670 bool pv_time_enabled
;
671 /* set guest stopped flag in pvclock flags field */
672 bool pvclock_set_guest_stopped_request
;
677 struct gfn_to_hva_cache stime
;
678 struct kvm_steal_time steal
;
684 u64 tsc_offset_adjustment
;
687 u64 this_tsc_generation
;
689 bool tsc_always_catchup
;
690 s8 virtual_tsc_shift
;
691 u32 virtual_tsc_mult
;
693 s64 ia32_tsc_adjust_msr
;
694 u64 msr_ia32_power_ctl
;
695 u64 tsc_scaling_ratio
;
697 atomic_t nmi_queued
; /* unprocessed asynchronous NMIs */
698 unsigned nmi_pending
; /* NMI queued after currently running handler */
699 bool nmi_injected
; /* Trying to inject an NMI this entry */
700 bool smi_pending
; /* SMI queued after currently running handler */
702 struct kvm_mtrr mtrr_state
;
705 unsigned switch_db_regs
;
706 unsigned long db
[KVM_NR_DB_REGS
];
709 unsigned long eff_db
[KVM_NR_DB_REGS
];
710 unsigned long guest_debug_dr7
;
711 u64 msr_platform_info
;
712 u64 msr_misc_features_enables
;
720 /* Cache MMIO info */
728 /* used for guest single stepping over the given code position */
729 unsigned long singlestep_rip
;
731 struct kvm_vcpu_hv hyperv
;
733 cpumask_var_t wbinvd_dirty_mask
;
735 unsigned long last_retry_eip
;
736 unsigned long last_retry_addr
;
740 gfn_t gfns
[roundup_pow_of_two(ASYNC_PF_PER_VCPU
)];
741 struct gfn_to_hva_cache data
;
746 unsigned long nested_apf_token
;
747 bool delivery_as_pf_vmexit
;
750 /* OSVW MSRs (AMD only) */
758 struct gfn_to_hva_cache data
;
761 u64 msr_kvm_poll_control
;
764 * Indicate whether the access faults on its page table in guest
765 * which is set when fix page fault and used to detect unhandeable
768 bool write_fault_to_shadow_pgtable
;
770 /* set at EPT violation at this point */
771 unsigned long exit_qualification
;
773 /* pv related host specific info */
778 int pending_ioapic_eoi
;
779 int pending_external_vector
;
785 /* be preempted when it's in kernel-mode(cpl=0) */
786 bool preempted_in_kernel
;
788 /* Flush the L1 Data cache for L1TF mitigation on VMENTER */
791 /* AMD MSRC001_0015 Hardware Configuration */
795 struct kvm_lpage_info
{
799 struct kvm_arch_memory_slot
{
800 struct kvm_rmap_head
*rmap
[KVM_NR_PAGE_SIZES
];
801 struct kvm_lpage_info
*lpage_info
[KVM_NR_PAGE_SIZES
- 1];
802 unsigned short *gfn_track
[KVM_PAGE_TRACK_MAX
];
806 * We use as the mode the number of bits allocated in the LDR for the
807 * logical processor ID. It happens that these are all powers of two.
808 * This makes it is very easy to detect cases where the APICs are
809 * configured for multiple modes; in that case, we cannot use the map and
810 * hence cannot use kvm_irq_delivery_to_apic_fast either.
812 #define KVM_APIC_MODE_XAPIC_CLUSTER 4
813 #define KVM_APIC_MODE_XAPIC_FLAT 8
814 #define KVM_APIC_MODE_X2APIC 16
816 struct kvm_apic_map
{
821 struct kvm_lapic
*xapic_flat_map
[8];
822 struct kvm_lapic
*xapic_cluster_map
[16][4];
824 struct kvm_lapic
*phys_map
[];
827 /* Hyper-V emulation context */
829 struct mutex hv_lock
;
834 /* Hyper-v based guest crash (NT kernel bugcheck) parameters */
835 u64 hv_crash_param
[HV_X64_MSR_CRASH_PARAMS
];
838 HV_REFERENCE_TSC_PAGE tsc_ref
;
840 struct idr conn_to_evt
;
842 u64 hv_reenlightenment_control
;
843 u64 hv_tsc_emulation_control
;
844 u64 hv_tsc_emulation_status
;
846 /* How many vCPUs have VP index != vCPU index */
847 atomic_t num_mismatched_vp_indexes
;
850 enum kvm_irqchip_mode
{
852 KVM_IRQCHIP_KERNEL
, /* created with KVM_CREATE_IRQCHIP */
853 KVM_IRQCHIP_SPLIT
, /* created with KVM_CAP_SPLIT_IRQCHIP */
857 unsigned long n_used_mmu_pages
;
858 unsigned long n_requested_mmu_pages
;
859 unsigned long n_max_mmu_pages
;
860 unsigned int indirect_shadow_pages
;
861 unsigned long mmu_valid_gen
;
862 struct hlist_head mmu_page_hash
[KVM_NUM_MMU_PAGES
];
864 * Hash table of struct kvm_mmu_page.
866 struct list_head active_mmu_pages
;
867 struct kvm_page_track_notifier_node mmu_sp_tracker
;
868 struct kvm_page_track_notifier_head track_notifier_head
;
870 struct list_head assigned_dev_head
;
871 struct iommu_domain
*iommu_domain
;
872 bool iommu_noncoherent
;
873 #define __KVM_HAVE_ARCH_NONCOHERENT_DMA
874 atomic_t noncoherent_dma_count
;
875 #define __KVM_HAVE_ARCH_ASSIGNED_DEVICE
876 atomic_t assigned_device_count
;
877 struct kvm_pic
*vpic
;
878 struct kvm_ioapic
*vioapic
;
879 struct kvm_pit
*vpit
;
880 atomic_t vapics_in_nmi_mode
;
881 struct mutex apic_map_lock
;
882 struct kvm_apic_map
*apic_map
;
884 bool apic_access_page_done
;
891 bool cstate_in_guest
;
893 unsigned long irq_sources_bitmap
;
895 raw_spinlock_t tsc_write_lock
;
902 u64 cur_tsc_generation
;
903 int nr_vcpus_matched_tsc
;
905 spinlock_t pvclock_gtod_sync_lock
;
906 bool use_master_clock
;
907 u64 master_kernel_ns
;
908 u64 master_cycle_now
;
909 struct delayed_work kvmclock_update_work
;
910 struct delayed_work kvmclock_sync_work
;
912 struct kvm_xen_hvm_config xen_hvm_config
;
914 /* reads protected by irq_srcu, writes by irq_lock */
915 struct hlist_head mask_notifier_list
;
917 struct kvm_hv hyperv
;
919 #ifdef CONFIG_KVM_MMU_AUDIT
923 bool backwards_tsc_observed
;
924 bool boot_vcpu_runs_old_kvmclock
;
929 enum kvm_irqchip_mode irqchip_mode
;
930 u8 nr_reserved_ioapic_pins
;
932 bool disabled_lapic_found
;
935 bool x2apic_broadcast_quirk_disabled
;
937 bool guest_can_read_msr_platform_info
;
938 bool exception_payload_enabled
;
940 struct kvm_pmu_event_filter
*pmu_event_filter
;
944 ulong mmu_shadow_zapped
;
946 ulong mmu_pte_updated
;
947 ulong mmu_pde_zapped
;
950 ulong mmu_cache_miss
;
952 ulong remote_tlb_flush
;
954 ulong nx_lpage_splits
;
955 ulong max_mmu_page_hash_collisions
;
958 struct kvm_vcpu_stat
{
968 u64 irq_window_exits
;
969 u64 nmi_window_exits
;
972 u64 halt_successful_poll
;
973 u64 halt_attempted_poll
;
974 u64 halt_poll_invalid
;
976 u64 request_irq_exits
;
978 u64 host_state_reload
;
981 u64 insn_emulation_fail
;
988 struct x86_instruction_info
;
996 struct kvm_lapic_irq
{
1004 bool msi_redir_hint
;
1007 struct kvm_x86_ops
{
1008 int (*cpu_has_kvm_support
)(void); /* __init */
1009 int (*disabled_by_bios
)(void); /* __init */
1010 int (*hardware_enable
)(void);
1011 void (*hardware_disable
)(void);
1012 int (*check_processor_compatibility
)(void);/* __init */
1013 int (*hardware_setup
)(void); /* __init */
1014 void (*hardware_unsetup
)(void); /* __exit */
1015 bool (*cpu_has_accelerated_tpr
)(void);
1016 bool (*has_emulated_msr
)(int index
);
1017 void (*cpuid_update
)(struct kvm_vcpu
*vcpu
);
1019 struct kvm
*(*vm_alloc
)(void);
1020 void (*vm_free
)(struct kvm
*);
1021 int (*vm_init
)(struct kvm
*kvm
);
1022 void (*vm_destroy
)(struct kvm
*kvm
);
1024 /* Create, but do not attach this VCPU */
1025 struct kvm_vcpu
*(*vcpu_create
)(struct kvm
*kvm
, unsigned id
);
1026 void (*vcpu_free
)(struct kvm_vcpu
*vcpu
);
1027 void (*vcpu_reset
)(struct kvm_vcpu
*vcpu
, bool init_event
);
1029 void (*prepare_guest_switch
)(struct kvm_vcpu
*vcpu
);
1030 void (*vcpu_load
)(struct kvm_vcpu
*vcpu
, int cpu
);
1031 void (*vcpu_put
)(struct kvm_vcpu
*vcpu
);
1033 void (*update_bp_intercept
)(struct kvm_vcpu
*vcpu
);
1034 int (*get_msr
)(struct kvm_vcpu
*vcpu
, struct msr_data
*msr
);
1035 int (*set_msr
)(struct kvm_vcpu
*vcpu
, struct msr_data
*msr
);
1036 u64 (*get_segment_base
)(struct kvm_vcpu
*vcpu
, int seg
);
1037 void (*get_segment
)(struct kvm_vcpu
*vcpu
,
1038 struct kvm_segment
*var
, int seg
);
1039 int (*get_cpl
)(struct kvm_vcpu
*vcpu
);
1040 void (*set_segment
)(struct kvm_vcpu
*vcpu
,
1041 struct kvm_segment
*var
, int seg
);
1042 void (*get_cs_db_l_bits
)(struct kvm_vcpu
*vcpu
, int *db
, int *l
);
1043 void (*decache_cr0_guest_bits
)(struct kvm_vcpu
*vcpu
);
1044 void (*decache_cr3
)(struct kvm_vcpu
*vcpu
);
1045 void (*decache_cr4_guest_bits
)(struct kvm_vcpu
*vcpu
);
1046 void (*set_cr0
)(struct kvm_vcpu
*vcpu
, unsigned long cr0
);
1047 void (*set_cr3
)(struct kvm_vcpu
*vcpu
, unsigned long cr3
);
1048 int (*set_cr4
)(struct kvm_vcpu
*vcpu
, unsigned long cr4
);
1049 void (*set_efer
)(struct kvm_vcpu
*vcpu
, u64 efer
);
1050 void (*get_idt
)(struct kvm_vcpu
*vcpu
, struct desc_ptr
*dt
);
1051 void (*set_idt
)(struct kvm_vcpu
*vcpu
, struct desc_ptr
*dt
);
1052 void (*get_gdt
)(struct kvm_vcpu
*vcpu
, struct desc_ptr
*dt
);
1053 void (*set_gdt
)(struct kvm_vcpu
*vcpu
, struct desc_ptr
*dt
);
1054 u64 (*get_dr6
)(struct kvm_vcpu
*vcpu
);
1055 void (*set_dr6
)(struct kvm_vcpu
*vcpu
, unsigned long value
);
1056 void (*sync_dirty_debug_regs
)(struct kvm_vcpu
*vcpu
);
1057 void (*set_dr7
)(struct kvm_vcpu
*vcpu
, unsigned long value
);
1058 void (*cache_reg
)(struct kvm_vcpu
*vcpu
, enum kvm_reg reg
);
1059 unsigned long (*get_rflags
)(struct kvm_vcpu
*vcpu
);
1060 void (*set_rflags
)(struct kvm_vcpu
*vcpu
, unsigned long rflags
);
1062 void (*tlb_flush
)(struct kvm_vcpu
*vcpu
, bool invalidate_gpa
);
1063 int (*tlb_remote_flush
)(struct kvm
*kvm
);
1064 int (*tlb_remote_flush_with_range
)(struct kvm
*kvm
,
1065 struct kvm_tlb_range
*range
);
1068 * Flush any TLB entries associated with the given GVA.
1069 * Does not need to flush GPA->HPA mappings.
1070 * Can potentially get non-canonical addresses through INVLPGs, which
1071 * the implementation may choose to ignore if appropriate.
1073 void (*tlb_flush_gva
)(struct kvm_vcpu
*vcpu
, gva_t addr
);
1075 void (*run
)(struct kvm_vcpu
*vcpu
);
1076 int (*handle_exit
)(struct kvm_vcpu
*vcpu
);
1077 void (*skip_emulated_instruction
)(struct kvm_vcpu
*vcpu
);
1078 void (*set_interrupt_shadow
)(struct kvm_vcpu
*vcpu
, int mask
);
1079 u32 (*get_interrupt_shadow
)(struct kvm_vcpu
*vcpu
);
1080 void (*patch_hypercall
)(struct kvm_vcpu
*vcpu
,
1081 unsigned char *hypercall_addr
);
1082 void (*set_irq
)(struct kvm_vcpu
*vcpu
);
1083 void (*set_nmi
)(struct kvm_vcpu
*vcpu
);
1084 void (*queue_exception
)(struct kvm_vcpu
*vcpu
);
1085 void (*cancel_injection
)(struct kvm_vcpu
*vcpu
);
1086 int (*interrupt_allowed
)(struct kvm_vcpu
*vcpu
);
1087 int (*nmi_allowed
)(struct kvm_vcpu
*vcpu
);
1088 bool (*get_nmi_mask
)(struct kvm_vcpu
*vcpu
);
1089 void (*set_nmi_mask
)(struct kvm_vcpu
*vcpu
, bool masked
);
1090 void (*enable_nmi_window
)(struct kvm_vcpu
*vcpu
);
1091 void (*enable_irq_window
)(struct kvm_vcpu
*vcpu
);
1092 void (*update_cr8_intercept
)(struct kvm_vcpu
*vcpu
, int tpr
, int irr
);
1093 bool (*get_enable_apicv
)(struct kvm_vcpu
*vcpu
);
1094 void (*refresh_apicv_exec_ctrl
)(struct kvm_vcpu
*vcpu
);
1095 void (*hwapic_irr_update
)(struct kvm_vcpu
*vcpu
, int max_irr
);
1096 void (*hwapic_isr_update
)(struct kvm_vcpu
*vcpu
, int isr
);
1097 bool (*guest_apic_has_interrupt
)(struct kvm_vcpu
*vcpu
);
1098 void (*load_eoi_exitmap
)(struct kvm_vcpu
*vcpu
, u64
*eoi_exit_bitmap
);
1099 void (*set_virtual_apic_mode
)(struct kvm_vcpu
*vcpu
);
1100 void (*set_apic_access_page_addr
)(struct kvm_vcpu
*vcpu
, hpa_t hpa
);
1101 void (*deliver_posted_interrupt
)(struct kvm_vcpu
*vcpu
, int vector
);
1102 int (*sync_pir_to_irr
)(struct kvm_vcpu
*vcpu
);
1103 int (*set_tss_addr
)(struct kvm
*kvm
, unsigned int addr
);
1104 int (*set_identity_map_addr
)(struct kvm
*kvm
, u64 ident_addr
);
1105 int (*get_tdp_level
)(struct kvm_vcpu
*vcpu
);
1106 u64 (*get_mt_mask
)(struct kvm_vcpu
*vcpu
, gfn_t gfn
, bool is_mmio
);
1107 int (*get_lpage_level
)(void);
1108 bool (*rdtscp_supported
)(void);
1109 bool (*invpcid_supported
)(void);
1111 void (*set_tdp_cr3
)(struct kvm_vcpu
*vcpu
, unsigned long cr3
);
1113 void (*set_supported_cpuid
)(u32 func
, struct kvm_cpuid_entry2
*entry
);
1115 bool (*has_wbinvd_exit
)(void);
1117 u64 (*read_l1_tsc_offset
)(struct kvm_vcpu
*vcpu
);
1118 /* Returns actual tsc_offset set in active VMCS */
1119 u64 (*write_l1_tsc_offset
)(struct kvm_vcpu
*vcpu
, u64 offset
);
1121 void (*get_exit_info
)(struct kvm_vcpu
*vcpu
, u64
*info1
, u64
*info2
);
1123 int (*check_intercept
)(struct kvm_vcpu
*vcpu
,
1124 struct x86_instruction_info
*info
,
1125 enum x86_intercept_stage stage
);
1126 void (*handle_exit_irqoff
)(struct kvm_vcpu
*vcpu
);
1127 bool (*mpx_supported
)(void);
1128 bool (*xsaves_supported
)(void);
1129 bool (*umip_emulated
)(void);
1130 bool (*pt_supported
)(void);
1132 int (*check_nested_events
)(struct kvm_vcpu
*vcpu
, bool external_intr
);
1133 void (*request_immediate_exit
)(struct kvm_vcpu
*vcpu
);
1135 void (*sched_in
)(struct kvm_vcpu
*kvm
, int cpu
);
1138 * Arch-specific dirty logging hooks. These hooks are only supposed to
1139 * be valid if the specific arch has hardware-accelerated dirty logging
1140 * mechanism. Currently only for PML on VMX.
1142 * - slot_enable_log_dirty:
1143 * called when enabling log dirty mode for the slot.
1144 * - slot_disable_log_dirty:
1145 * called when disabling log dirty mode for the slot.
1146 * also called when slot is created with log dirty disabled.
1147 * - flush_log_dirty:
1148 * called before reporting dirty_bitmap to userspace.
1149 * - enable_log_dirty_pt_masked:
1150 * called when reenabling log dirty for the GFNs in the mask after
1151 * corresponding bits are cleared in slot->dirty_bitmap.
1153 void (*slot_enable_log_dirty
)(struct kvm
*kvm
,
1154 struct kvm_memory_slot
*slot
);
1155 void (*slot_disable_log_dirty
)(struct kvm
*kvm
,
1156 struct kvm_memory_slot
*slot
);
1157 void (*flush_log_dirty
)(struct kvm
*kvm
);
1158 void (*enable_log_dirty_pt_masked
)(struct kvm
*kvm
,
1159 struct kvm_memory_slot
*slot
,
1160 gfn_t offset
, unsigned long mask
);
1161 int (*write_log_dirty
)(struct kvm_vcpu
*vcpu
);
1163 /* pmu operations of sub-arch */
1164 const struct kvm_pmu_ops
*pmu_ops
;
1167 * Architecture specific hooks for vCPU blocking due to
1169 * Returns for .pre_block():
1170 * - 0 means continue to block the vCPU.
1171 * - 1 means we cannot block the vCPU since some event
1172 * happens during this period, such as, 'ON' bit in
1173 * posted-interrupts descriptor is set.
1175 int (*pre_block
)(struct kvm_vcpu
*vcpu
);
1176 void (*post_block
)(struct kvm_vcpu
*vcpu
);
1178 void (*vcpu_blocking
)(struct kvm_vcpu
*vcpu
);
1179 void (*vcpu_unblocking
)(struct kvm_vcpu
*vcpu
);
1181 int (*update_pi_irte
)(struct kvm
*kvm
, unsigned int host_irq
,
1182 uint32_t guest_irq
, bool set
);
1183 void (*apicv_post_state_restore
)(struct kvm_vcpu
*vcpu
);
1184 bool (*dy_apicv_has_pending_interrupt
)(struct kvm_vcpu
*vcpu
);
1186 int (*set_hv_timer
)(struct kvm_vcpu
*vcpu
, u64 guest_deadline_tsc
,
1188 void (*cancel_hv_timer
)(struct kvm_vcpu
*vcpu
);
1190 void (*setup_mce
)(struct kvm_vcpu
*vcpu
);
1192 int (*get_nested_state
)(struct kvm_vcpu
*vcpu
,
1193 struct kvm_nested_state __user
*user_kvm_nested_state
,
1194 unsigned user_data_size
);
1195 int (*set_nested_state
)(struct kvm_vcpu
*vcpu
,
1196 struct kvm_nested_state __user
*user_kvm_nested_state
,
1197 struct kvm_nested_state
*kvm_state
);
1198 void (*get_vmcs12_pages
)(struct kvm_vcpu
*vcpu
);
1200 int (*smi_allowed
)(struct kvm_vcpu
*vcpu
);
1201 int (*pre_enter_smm
)(struct kvm_vcpu
*vcpu
, char *smstate
);
1202 int (*pre_leave_smm
)(struct kvm_vcpu
*vcpu
, const char *smstate
);
1203 int (*enable_smi_window
)(struct kvm_vcpu
*vcpu
);
1205 int (*mem_enc_op
)(struct kvm
*kvm
, void __user
*argp
);
1206 int (*mem_enc_reg_region
)(struct kvm
*kvm
, struct kvm_enc_region
*argp
);
1207 int (*mem_enc_unreg_region
)(struct kvm
*kvm
, struct kvm_enc_region
*argp
);
1209 int (*get_msr_feature
)(struct kvm_msr_entry
*entry
);
1211 int (*nested_enable_evmcs
)(struct kvm_vcpu
*vcpu
,
1212 uint16_t *vmcs_version
);
1213 uint16_t (*nested_get_evmcs_version
)(struct kvm_vcpu
*vcpu
);
1215 bool (*need_emulation_on_page_fault
)(struct kvm_vcpu
*vcpu
);
1218 struct kvm_arch_async_pf
{
1225 extern struct kvm_x86_ops
*kvm_x86_ops
;
1226 extern struct kmem_cache
*x86_fpu_cache
;
1228 #define __KVM_HAVE_ARCH_VM_ALLOC
1229 static inline struct kvm
*kvm_arch_alloc_vm(void)
1231 return kvm_x86_ops
->vm_alloc();
1234 static inline void kvm_arch_free_vm(struct kvm
*kvm
)
1236 return kvm_x86_ops
->vm_free(kvm
);
1239 #define __KVM_HAVE_ARCH_FLUSH_REMOTE_TLB
1240 static inline int kvm_arch_flush_remote_tlb(struct kvm
*kvm
)
1242 if (kvm_x86_ops
->tlb_remote_flush
&&
1243 !kvm_x86_ops
->tlb_remote_flush(kvm
))
1249 int kvm_mmu_module_init(void);
1250 void kvm_mmu_module_exit(void);
1252 void kvm_mmu_destroy(struct kvm_vcpu
*vcpu
);
1253 int kvm_mmu_create(struct kvm_vcpu
*vcpu
);
1254 void kvm_mmu_init_vm(struct kvm
*kvm
);
1255 void kvm_mmu_uninit_vm(struct kvm
*kvm
);
1256 void kvm_mmu_set_mask_ptes(u64 user_mask
, u64 accessed_mask
,
1257 u64 dirty_mask
, u64 nx_mask
, u64 x_mask
, u64 p_mask
,
1258 u64 acc_track_mask
, u64 me_mask
);
1260 void kvm_mmu_reset_context(struct kvm_vcpu
*vcpu
);
1261 void kvm_mmu_slot_remove_write_access(struct kvm
*kvm
,
1262 struct kvm_memory_slot
*memslot
);
1263 void kvm_mmu_zap_collapsible_sptes(struct kvm
*kvm
,
1264 const struct kvm_memory_slot
*memslot
);
1265 void kvm_mmu_slot_leaf_clear_dirty(struct kvm
*kvm
,
1266 struct kvm_memory_slot
*memslot
);
1267 void kvm_mmu_slot_largepage_remove_write_access(struct kvm
*kvm
,
1268 struct kvm_memory_slot
*memslot
);
1269 void kvm_mmu_slot_set_dirty(struct kvm
*kvm
,
1270 struct kvm_memory_slot
*memslot
);
1271 void kvm_mmu_clear_dirty_pt_masked(struct kvm
*kvm
,
1272 struct kvm_memory_slot
*slot
,
1273 gfn_t gfn_offset
, unsigned long mask
);
1274 void kvm_mmu_zap_all(struct kvm
*kvm
);
1275 void kvm_mmu_invalidate_mmio_sptes(struct kvm
*kvm
, u64 gen
);
1276 unsigned long kvm_mmu_calculate_default_mmu_pages(struct kvm
*kvm
);
1277 void kvm_mmu_change_mmu_pages(struct kvm
*kvm
, unsigned long kvm_nr_mmu_pages
);
1279 int load_pdptrs(struct kvm_vcpu
*vcpu
, struct kvm_mmu
*mmu
, unsigned long cr3
);
1280 bool pdptrs_changed(struct kvm_vcpu
*vcpu
);
1282 int emulator_write_phys(struct kvm_vcpu
*vcpu
, gpa_t gpa
,
1283 const void *val
, int bytes
);
1285 struct kvm_irq_mask_notifier
{
1286 void (*func
)(struct kvm_irq_mask_notifier
*kimn
, bool masked
);
1288 struct hlist_node link
;
1291 void kvm_register_irq_mask_notifier(struct kvm
*kvm
, int irq
,
1292 struct kvm_irq_mask_notifier
*kimn
);
1293 void kvm_unregister_irq_mask_notifier(struct kvm
*kvm
, int irq
,
1294 struct kvm_irq_mask_notifier
*kimn
);
1295 void kvm_fire_mask_notifiers(struct kvm
*kvm
, unsigned irqchip
, unsigned pin
,
1298 extern bool tdp_enabled
;
1300 u64
vcpu_tsc_khz(struct kvm_vcpu
*vcpu
);
1302 /* control of guest tsc rate supported? */
1303 extern bool kvm_has_tsc_control
;
1304 /* maximum supported tsc_khz for guests */
1305 extern u32 kvm_max_guest_tsc_khz
;
1306 /* number of bits of the fractional part of the TSC scaling ratio */
1307 extern u8 kvm_tsc_scaling_ratio_frac_bits
;
1308 /* maximum allowed value of TSC scaling ratio */
1309 extern u64 kvm_max_tsc_scaling_ratio
;
1310 /* 1ull << kvm_tsc_scaling_ratio_frac_bits */
1311 extern u64 kvm_default_tsc_scaling_ratio
;
1313 extern u64 kvm_mce_cap_supported
;
1315 enum emulation_result
{
1316 EMULATE_DONE
, /* no further processing */
1317 EMULATE_USER_EXIT
, /* kvm_run ready for userspace exit */
1318 EMULATE_FAIL
, /* can't emulate this instruction */
1321 #define EMULTYPE_NO_DECODE (1 << 0)
1322 #define EMULTYPE_TRAP_UD (1 << 1)
1323 #define EMULTYPE_SKIP (1 << 2)
1324 #define EMULTYPE_ALLOW_RETRY (1 << 3)
1325 #define EMULTYPE_NO_UD_ON_FAIL (1 << 4)
1326 #define EMULTYPE_VMWARE (1 << 5)
1327 int kvm_emulate_instruction(struct kvm_vcpu
*vcpu
, int emulation_type
);
1328 int kvm_emulate_instruction_from_buffer(struct kvm_vcpu
*vcpu
,
1329 void *insn
, int insn_len
);
1331 void kvm_enable_efer_bits(u64
);
1332 bool kvm_valid_efer(struct kvm_vcpu
*vcpu
, u64 efer
);
1333 int kvm_get_msr(struct kvm_vcpu
*vcpu
, struct msr_data
*msr
);
1334 int kvm_set_msr(struct kvm_vcpu
*vcpu
, struct msr_data
*msr
);
1336 struct x86_emulate_ctxt
;
1338 int kvm_fast_pio(struct kvm_vcpu
*vcpu
, int size
, unsigned short port
, int in
);
1339 int kvm_emulate_cpuid(struct kvm_vcpu
*vcpu
);
1340 int kvm_emulate_halt(struct kvm_vcpu
*vcpu
);
1341 int kvm_vcpu_halt(struct kvm_vcpu
*vcpu
);
1342 int kvm_emulate_wbinvd(struct kvm_vcpu
*vcpu
);
1344 void kvm_get_segment(struct kvm_vcpu
*vcpu
, struct kvm_segment
*var
, int seg
);
1345 int kvm_load_segment_descriptor(struct kvm_vcpu
*vcpu
, u16 selector
, int seg
);
1346 void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu
*vcpu
, u8 vector
);
1348 int kvm_task_switch(struct kvm_vcpu
*vcpu
, u16 tss_selector
, int idt_index
,
1349 int reason
, bool has_error_code
, u32 error_code
);
1351 int kvm_set_cr0(struct kvm_vcpu
*vcpu
, unsigned long cr0
);
1352 int kvm_set_cr3(struct kvm_vcpu
*vcpu
, unsigned long cr3
);
1353 int kvm_set_cr4(struct kvm_vcpu
*vcpu
, unsigned long cr4
);
1354 int kvm_set_cr8(struct kvm_vcpu
*vcpu
, unsigned long cr8
);
1355 int kvm_set_dr(struct kvm_vcpu
*vcpu
, int dr
, unsigned long val
);
1356 int kvm_get_dr(struct kvm_vcpu
*vcpu
, int dr
, unsigned long *val
);
1357 unsigned long kvm_get_cr8(struct kvm_vcpu
*vcpu
);
1358 void kvm_lmsw(struct kvm_vcpu
*vcpu
, unsigned long msw
);
1359 void kvm_get_cs_db_l_bits(struct kvm_vcpu
*vcpu
, int *db
, int *l
);
1360 int kvm_set_xcr(struct kvm_vcpu
*vcpu
, u32 index
, u64 xcr
);
1362 int kvm_get_msr_common(struct kvm_vcpu
*vcpu
, struct msr_data
*msr
);
1363 int kvm_set_msr_common(struct kvm_vcpu
*vcpu
, struct msr_data
*msr
);
1365 unsigned long kvm_get_rflags(struct kvm_vcpu
*vcpu
);
1366 void kvm_set_rflags(struct kvm_vcpu
*vcpu
, unsigned long rflags
);
1367 bool kvm_rdpmc(struct kvm_vcpu
*vcpu
);
1369 void kvm_queue_exception(struct kvm_vcpu
*vcpu
, unsigned nr
);
1370 void kvm_queue_exception_e(struct kvm_vcpu
*vcpu
, unsigned nr
, u32 error_code
);
1371 void kvm_requeue_exception(struct kvm_vcpu
*vcpu
, unsigned nr
);
1372 void kvm_requeue_exception_e(struct kvm_vcpu
*vcpu
, unsigned nr
, u32 error_code
);
1373 void kvm_inject_page_fault(struct kvm_vcpu
*vcpu
, struct x86_exception
*fault
);
1374 int kvm_read_guest_page_mmu(struct kvm_vcpu
*vcpu
, struct kvm_mmu
*mmu
,
1375 gfn_t gfn
, void *data
, int offset
, int len
,
1377 bool kvm_require_cpl(struct kvm_vcpu
*vcpu
, int required_cpl
);
1378 bool kvm_require_dr(struct kvm_vcpu
*vcpu
, int dr
);
1380 static inline int __kvm_irq_line_state(unsigned long *irq_state
,
1381 int irq_source_id
, int level
)
1383 /* Logical OR for level trig interrupt */
1385 __set_bit(irq_source_id
, irq_state
);
1387 __clear_bit(irq_source_id
, irq_state
);
1389 return !!(*irq_state
);
1392 #define KVM_MMU_ROOT_CURRENT BIT(0)
1393 #define KVM_MMU_ROOT_PREVIOUS(i) BIT(1+i)
1394 #define KVM_MMU_ROOTS_ALL (~0UL)
1396 int kvm_pic_set_irq(struct kvm_pic
*pic
, int irq
, int irq_source_id
, int level
);
1397 void kvm_pic_clear_all(struct kvm_pic
*pic
, int irq_source_id
);
1399 void kvm_inject_nmi(struct kvm_vcpu
*vcpu
);
1401 int kvm_mmu_unprotect_page(struct kvm
*kvm
, gfn_t gfn
);
1402 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu
*vcpu
, gva_t gva
);
1403 void __kvm_mmu_free_some_pages(struct kvm_vcpu
*vcpu
);
1404 int kvm_mmu_load(struct kvm_vcpu
*vcpu
);
1405 void kvm_mmu_unload(struct kvm_vcpu
*vcpu
);
1406 void kvm_mmu_sync_roots(struct kvm_vcpu
*vcpu
);
1407 void kvm_mmu_free_roots(struct kvm_vcpu
*vcpu
, struct kvm_mmu
*mmu
,
1408 ulong roots_to_free
);
1409 gpa_t
translate_nested_gpa(struct kvm_vcpu
*vcpu
, gpa_t gpa
, u32 access
,
1410 struct x86_exception
*exception
);
1411 gpa_t
kvm_mmu_gva_to_gpa_read(struct kvm_vcpu
*vcpu
, gva_t gva
,
1412 struct x86_exception
*exception
);
1413 gpa_t
kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu
*vcpu
, gva_t gva
,
1414 struct x86_exception
*exception
);
1415 gpa_t
kvm_mmu_gva_to_gpa_write(struct kvm_vcpu
*vcpu
, gva_t gva
,
1416 struct x86_exception
*exception
);
1417 gpa_t
kvm_mmu_gva_to_gpa_system(struct kvm_vcpu
*vcpu
, gva_t gva
,
1418 struct x86_exception
*exception
);
1420 void kvm_vcpu_deactivate_apicv(struct kvm_vcpu
*vcpu
);
1422 int kvm_emulate_hypercall(struct kvm_vcpu
*vcpu
);
1424 int kvm_mmu_page_fault(struct kvm_vcpu
*vcpu
, gva_t gva
, u64 error_code
,
1425 void *insn
, int insn_len
);
1426 void kvm_mmu_invlpg(struct kvm_vcpu
*vcpu
, gva_t gva
);
1427 void kvm_mmu_invpcid_gva(struct kvm_vcpu
*vcpu
, gva_t gva
, unsigned long pcid
);
1428 void kvm_mmu_new_cr3(struct kvm_vcpu
*vcpu
, gpa_t new_cr3
, bool skip_tlb_flush
);
1430 void kvm_enable_tdp(void);
1431 void kvm_disable_tdp(void);
1433 static inline gpa_t
translate_gpa(struct kvm_vcpu
*vcpu
, gpa_t gpa
, u32 access
,
1434 struct x86_exception
*exception
)
1439 static inline struct kvm_mmu_page
*page_header(hpa_t shadow_page
)
1441 struct page
*page
= pfn_to_page(shadow_page
>> PAGE_SHIFT
);
1443 return (struct kvm_mmu_page
*)page_private(page
);
1446 static inline u16
kvm_read_ldt(void)
1449 asm("sldt %0" : "=g"(ldt
));
1453 static inline void kvm_load_ldt(u16 sel
)
1455 asm("lldt %0" : : "rm"(sel
));
1458 #ifdef CONFIG_X86_64
1459 static inline unsigned long read_msr(unsigned long msr
)
1468 static inline u32
get_rdx_init_val(void)
1470 return 0x600; /* P6 family */
1473 static inline void kvm_inject_gp(struct kvm_vcpu
*vcpu
, u32 error_code
)
1475 kvm_queue_exception_e(vcpu
, GP_VECTOR
, error_code
);
1478 #define TSS_IOPB_BASE_OFFSET 0x66
1479 #define TSS_BASE_SIZE 0x68
1480 #define TSS_IOPB_SIZE (65536 / 8)
1481 #define TSS_REDIRECTION_SIZE (256 / 8)
1482 #define RMODE_TSS_SIZE \
1483 (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
1486 TASK_SWITCH_CALL
= 0,
1487 TASK_SWITCH_IRET
= 1,
1488 TASK_SWITCH_JMP
= 2,
1489 TASK_SWITCH_GATE
= 3,
1492 #define HF_GIF_MASK (1 << 0)
1493 #define HF_HIF_MASK (1 << 1)
1494 #define HF_VINTR_MASK (1 << 2)
1495 #define HF_NMI_MASK (1 << 3)
1496 #define HF_IRET_MASK (1 << 4)
1497 #define HF_GUEST_MASK (1 << 5) /* VCPU is in guest-mode */
1498 #define HF_SMM_MASK (1 << 6)
1499 #define HF_SMM_INSIDE_NMI_MASK (1 << 7)
1501 #define __KVM_VCPU_MULTIPLE_ADDRESS_SPACE
1502 #define KVM_ADDRESS_SPACE_NUM 2
1504 #define kvm_arch_vcpu_memslots_id(vcpu) ((vcpu)->arch.hflags & HF_SMM_MASK ? 1 : 0)
1505 #define kvm_memslots_for_spte_role(kvm, role) __kvm_memslots(kvm, (role).smm)
1507 asmlinkage
void __noreturn
kvm_spurious_fault(void);
1510 * Hardware virtualization extension instructions may fault if a
1511 * reboot turns off virtualization while processes are running.
1512 * Usually after catching the fault we just panic; during reboot
1513 * instead the instruction is ignored.
1515 #define ____kvm_handle_fault_on_reboot(insn, cleanup_insn) \
1520 "call kvm_spurious_fault \n\t" \
1522 ".pushsection .fixup, \"ax\" \n\t" \
1524 cleanup_insn "\n\t" \
1525 "cmpb $0, kvm_rebooting\n\t" \
1528 ".popsection \n\t" \
1529 _ASM_EXTABLE(666b, 700b)
1531 #define __kvm_handle_fault_on_reboot(insn) \
1532 ____kvm_handle_fault_on_reboot(insn, "")
1534 #define KVM_ARCH_WANT_MMU_NOTIFIER
1535 int kvm_unmap_hva_range(struct kvm
*kvm
, unsigned long start
, unsigned long end
);
1536 int kvm_age_hva(struct kvm
*kvm
, unsigned long start
, unsigned long end
);
1537 int kvm_test_age_hva(struct kvm
*kvm
, unsigned long hva
);
1538 int kvm_set_spte_hva(struct kvm
*kvm
, unsigned long hva
, pte_t pte
);
1539 int kvm_cpu_has_injectable_intr(struct kvm_vcpu
*v
);
1540 int kvm_cpu_has_interrupt(struct kvm_vcpu
*vcpu
);
1541 int kvm_arch_interrupt_allowed(struct kvm_vcpu
*vcpu
);
1542 int kvm_cpu_get_interrupt(struct kvm_vcpu
*v
);
1543 void kvm_vcpu_reset(struct kvm_vcpu
*vcpu
, bool init_event
);
1544 void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu
*vcpu
);
1546 int kvm_pv_send_ipi(struct kvm
*kvm
, unsigned long ipi_bitmap_low
,
1547 unsigned long ipi_bitmap_high
, u32 min
,
1548 unsigned long icr
, int op_64_bit
);
1550 void kvm_define_shared_msr(unsigned index
, u32 msr
);
1551 int kvm_set_shared_msr(unsigned index
, u64 val
, u64 mask
);
1553 u64
kvm_scale_tsc(struct kvm_vcpu
*vcpu
, u64 tsc
);
1554 u64
kvm_read_l1_tsc(struct kvm_vcpu
*vcpu
, u64 host_tsc
);
1556 unsigned long kvm_get_linear_rip(struct kvm_vcpu
*vcpu
);
1557 bool kvm_is_linear_rip(struct kvm_vcpu
*vcpu
, unsigned long linear_rip
);
1559 void kvm_make_mclock_inprogress_request(struct kvm
*kvm
);
1560 void kvm_make_scan_ioapic_request(struct kvm
*kvm
);
1562 void kvm_arch_async_page_not_present(struct kvm_vcpu
*vcpu
,
1563 struct kvm_async_pf
*work
);
1564 void kvm_arch_async_page_present(struct kvm_vcpu
*vcpu
,
1565 struct kvm_async_pf
*work
);
1566 void kvm_arch_async_page_ready(struct kvm_vcpu
*vcpu
,
1567 struct kvm_async_pf
*work
);
1568 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu
*vcpu
);
1569 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu
*vcpu
, gfn_t gfn
);
1571 int kvm_skip_emulated_instruction(struct kvm_vcpu
*vcpu
);
1572 int kvm_complete_insn_gp(struct kvm_vcpu
*vcpu
, int err
);
1573 void __kvm_request_immediate_exit(struct kvm_vcpu
*vcpu
);
1575 int kvm_is_in_guest(void);
1577 int __x86_set_memory_region(struct kvm
*kvm
, int id
, gpa_t gpa
, u32 size
);
1578 int x86_set_memory_region(struct kvm
*kvm
, int id
, gpa_t gpa
, u32 size
);
1579 bool kvm_vcpu_is_reset_bsp(struct kvm_vcpu
*vcpu
);
1580 bool kvm_vcpu_is_bsp(struct kvm_vcpu
*vcpu
);
1582 bool kvm_intr_is_single_vcpu(struct kvm
*kvm
, struct kvm_lapic_irq
*irq
,
1583 struct kvm_vcpu
**dest_vcpu
);
1585 void kvm_set_msi_irq(struct kvm
*kvm
, struct kvm_kernel_irq_routing_entry
*e
,
1586 struct kvm_lapic_irq
*irq
);
1588 static inline bool kvm_irq_is_postable(struct kvm_lapic_irq
*irq
)
1590 /* We can only post Fixed and LowPrio IRQs */
1591 return (irq
->delivery_mode
== dest_Fixed
||
1592 irq
->delivery_mode
== dest_LowestPrio
);
1595 static inline void kvm_arch_vcpu_blocking(struct kvm_vcpu
*vcpu
)
1597 if (kvm_x86_ops
->vcpu_blocking
)
1598 kvm_x86_ops
->vcpu_blocking(vcpu
);
1601 static inline void kvm_arch_vcpu_unblocking(struct kvm_vcpu
*vcpu
)
1603 if (kvm_x86_ops
->vcpu_unblocking
)
1604 kvm_x86_ops
->vcpu_unblocking(vcpu
);
1607 static inline void kvm_arch_vcpu_block_finish(struct kvm_vcpu
*vcpu
) {}
1609 static inline int kvm_cpu_get_apicid(int mps_cpu
)
1611 #ifdef CONFIG_X86_LOCAL_APIC
1612 return default_cpu_present_to_apicid(mps_cpu
);
1619 #define put_smstate(type, buf, offset, val) \
1620 *(type *)((buf) + (offset) - 0x7e00) = val
1622 #define GET_SMSTATE(type, buf, offset) \
1623 (*(type *)((buf) + (offset) - 0x7e00))
1625 #endif /* _ASM_X86_KVM_HOST_H */