]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blame - arch/x86/include/asm/kvm_host.h
kvm/x86: Hyper-V synthetic interrupt controller
[mirror_ubuntu-bionic-kernel.git] / arch / x86 / include / asm / kvm_host.h
CommitLineData
a656c8ef 1/*
043405e1
CO
2 * Kernel-based Virtual Machine driver for Linux
3 *
4 * This header defines architecture specific interfaces, x86 version
5 *
6 * This work is licensed under the terms of the GNU GPL, version 2. See
7 * the COPYING file in the top-level directory.
8 *
9 */
10
1965aae3
PA
11#ifndef _ASM_X86_KVM_HOST_H
12#define _ASM_X86_KVM_HOST_H
043405e1 13
34c16eec
ZX
14#include <linux/types.h>
15#include <linux/mm.h>
e930bffe 16#include <linux/mmu_notifier.h>
229456fc 17#include <linux/tracepoint.h>
f5f48ee1 18#include <linux/cpumask.h>
f5132b01 19#include <linux/irq_work.h>
34c16eec
ZX
20
21#include <linux/kvm.h>
22#include <linux/kvm_para.h>
edf88417 23#include <linux/kvm_types.h>
f5132b01 24#include <linux/perf_event.h>
d828199e
MT
25#include <linux/pvclock_gtod.h>
26#include <linux/clocksource.h>
87276880 27#include <linux/irqbypass.h>
5c919412 28#include <linux/hyperv.h>
34c16eec 29
50d0a0f9 30#include <asm/pvclock-abi.h>
e01a1b57 31#include <asm/desc.h>
0bed3b56 32#include <asm/mtrr.h>
9962d032 33#include <asm/msr-index.h>
3ee89722 34#include <asm/asm.h>
e01a1b57 35
cbf64358 36#define KVM_MAX_VCPUS 255
a59cb29e 37#define KVM_SOFT_MAX_VCPUS 160
1d4e7e3c 38#define KVM_USER_MEM_SLOTS 509
0743247f
AW
39/* memory slots that are not exposed to userspace */
40#define KVM_PRIVATE_MEM_SLOTS 3
bbacc0c1 41#define KVM_MEM_SLOTS_NUM (KVM_USER_MEM_SLOTS + KVM_PRIVATE_MEM_SLOTS)
93a5cef0 42
69a9f69b 43#define KVM_PIO_PAGE_OFFSET 1
542472b5 44#define KVM_COALESCED_MMIO_PAGE_OFFSET 2
920552b2 45#define KVM_HALT_POLL_NS_DEFAULT 500000
69a9f69b 46
8175e5b7
AG
47#define KVM_IRQCHIP_NUM_PINS KVM_IOAPIC_NUM_PINS
48
cfec82cb
JR
49#define CR0_RESERVED_BITS \
50 (~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
51 | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
52 | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
53
346874c9 54#define CR3_L_MODE_RESERVED_BITS 0xFFFFFF0000000000ULL
cfaa790a 55#define CR3_PCID_INVD BIT_64(63)
cfec82cb
JR
56#define CR4_RESERVED_BITS \
57 (~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
58 | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE \
ad756a16 59 | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR | X86_CR4_PCIDE \
afcbf13f 60 | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_FSGSBASE \
56d6efc2 61 | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE | X86_CR4_SMAP))
cfec82cb
JR
62
63#define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
64
65
cd6e8f87 66
cd6e8f87 67#define INVALID_PAGE (~(hpa_t)0)
dd180b3e
XG
68#define VALID_PAGE(x) ((x) != INVALID_PAGE)
69
cd6e8f87
ZX
70#define UNMAPPED_GVA (~(gpa_t)0)
71
ec04b260 72/* KVM Hugepage definitions for x86 */
04326caa 73#define KVM_NR_PAGE_SIZES 3
82855413
JR
74#define KVM_HPAGE_GFN_SHIFT(x) (((x) - 1) * 9)
75#define KVM_HPAGE_SHIFT(x) (PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
ec04b260
JR
76#define KVM_HPAGE_SIZE(x) (1UL << KVM_HPAGE_SHIFT(x))
77#define KVM_HPAGE_MASK(x) (~(KVM_HPAGE_SIZE(x) - 1))
78#define KVM_PAGES_PER_HPAGE(x) (KVM_HPAGE_SIZE(x) / PAGE_SIZE)
05da4558 79
6d9d41e5
CD
80static inline gfn_t gfn_to_index(gfn_t gfn, gfn_t base_gfn, int level)
81{
82 /* KVM_HPAGE_GFN_SHIFT(PT_PAGE_TABLE_LEVEL) must be 0. */
83 return (gfn >> KVM_HPAGE_GFN_SHIFT(level)) -
84 (base_gfn >> KVM_HPAGE_GFN_SHIFT(level));
85}
86
d657a98e
ZX
87#define KVM_PERMILLE_MMU_PAGES 20
88#define KVM_MIN_ALLOC_MMU_PAGES 64
1ae0a13d
DE
89#define KVM_MMU_HASH_SHIFT 10
90#define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
d657a98e
ZX
91#define KVM_MIN_FREE_MMU_PAGES 5
92#define KVM_REFILL_PAGES 25
73c1160c 93#define KVM_MAX_CPUID_ENTRIES 80
0bed3b56 94#define KVM_NR_FIXED_MTRR_REGION 88
0d234daf 95#define KVM_NR_VAR_MTRR 8
d657a98e 96
af585b92
GN
97#define ASYNC_PF_PER_VCPU 64
98
5fdbf976 99enum kvm_reg {
2b3ccfa0
ZX
100 VCPU_REGS_RAX = 0,
101 VCPU_REGS_RCX = 1,
102 VCPU_REGS_RDX = 2,
103 VCPU_REGS_RBX = 3,
104 VCPU_REGS_RSP = 4,
105 VCPU_REGS_RBP = 5,
106 VCPU_REGS_RSI = 6,
107 VCPU_REGS_RDI = 7,
108#ifdef CONFIG_X86_64
109 VCPU_REGS_R8 = 8,
110 VCPU_REGS_R9 = 9,
111 VCPU_REGS_R10 = 10,
112 VCPU_REGS_R11 = 11,
113 VCPU_REGS_R12 = 12,
114 VCPU_REGS_R13 = 13,
115 VCPU_REGS_R14 = 14,
116 VCPU_REGS_R15 = 15,
117#endif
5fdbf976 118 VCPU_REGS_RIP,
2b3ccfa0
ZX
119 NR_VCPU_REGS
120};
121
6de4f3ad
AK
122enum kvm_reg_ex {
123 VCPU_EXREG_PDPTR = NR_VCPU_REGS,
aff48baa 124 VCPU_EXREG_CR3,
6de12732 125 VCPU_EXREG_RFLAGS,
2fb92db1 126 VCPU_EXREG_SEGMENTS,
6de4f3ad
AK
127};
128
2b3ccfa0 129enum {
81609e3e 130 VCPU_SREG_ES,
2b3ccfa0 131 VCPU_SREG_CS,
81609e3e 132 VCPU_SREG_SS,
2b3ccfa0 133 VCPU_SREG_DS,
2b3ccfa0
ZX
134 VCPU_SREG_FS,
135 VCPU_SREG_GS,
2b3ccfa0
ZX
136 VCPU_SREG_TR,
137 VCPU_SREG_LDTR,
138};
139
56e82318 140#include <asm/kvm_emulate.h>
2b3ccfa0 141
d657a98e
ZX
142#define KVM_NR_MEM_OBJS 40
143
42dbaa5a
JK
144#define KVM_NR_DB_REGS 4
145
146#define DR6_BD (1 << 13)
147#define DR6_BS (1 << 14)
6f43ed01
NA
148#define DR6_RTM (1 << 16)
149#define DR6_FIXED_1 0xfffe0ff0
150#define DR6_INIT 0xffff0ff0
151#define DR6_VOLATILE 0x0001e00f
42dbaa5a
JK
152
153#define DR7_BP_EN_MASK 0x000000ff
154#define DR7_GE (1 << 9)
155#define DR7_GD (1 << 13)
156#define DR7_FIXED_1 0x00000400
6f43ed01 157#define DR7_VOLATILE 0xffff2bff
42dbaa5a 158
c205fb7d
NA
159#define PFERR_PRESENT_BIT 0
160#define PFERR_WRITE_BIT 1
161#define PFERR_USER_BIT 2
162#define PFERR_RSVD_BIT 3
163#define PFERR_FETCH_BIT 4
164
165#define PFERR_PRESENT_MASK (1U << PFERR_PRESENT_BIT)
166#define PFERR_WRITE_MASK (1U << PFERR_WRITE_BIT)
167#define PFERR_USER_MASK (1U << PFERR_USER_BIT)
168#define PFERR_RSVD_MASK (1U << PFERR_RSVD_BIT)
169#define PFERR_FETCH_MASK (1U << PFERR_FETCH_BIT)
170
41383771
GN
171/* apic attention bits */
172#define KVM_APIC_CHECK_VAPIC 0
ae7a2a3f
MT
173/*
174 * The following bit is set with PV-EOI, unset on EOI.
175 * We detect PV-EOI changes by guest by comparing
176 * this bit with PV-EOI in guest memory.
177 * See the implementation in apic_update_pv_eoi.
178 */
179#define KVM_APIC_PV_EOI_PENDING 1
41383771 180
d84f1e07
FW
181struct kvm_kernel_irq_routing_entry;
182
d657a98e
ZX
183/*
184 * We don't want allocation failures within the mmu code, so we preallocate
185 * enough memory for a single page fault in a cache.
186 */
187struct kvm_mmu_memory_cache {
188 int nobjs;
189 void *objects[KVM_NR_MEM_OBJS];
190};
191
d657a98e
ZX
192union kvm_mmu_page_role {
193 unsigned word;
194 struct {
7d76b4d3 195 unsigned level:4;
5b7e0102 196 unsigned cr4_pae:1;
7d76b4d3 197 unsigned quadrant:2;
f6e2c02b 198 unsigned direct:1;
7d76b4d3 199 unsigned access:3;
2e53d63a 200 unsigned invalid:1;
9645bb56 201 unsigned nxe:1;
3dbe1415 202 unsigned cr0_wp:1;
411c588d 203 unsigned smep_andnot_wp:1;
0be0226f 204 unsigned smap_andnot_wp:1;
699023e2
PB
205 unsigned :8;
206
207 /*
208 * This is left at the top of the word so that
209 * kvm_memslots_for_spte_role can extract it with a
210 * simple shift. While there is room, give it a whole
211 * byte so it is also faster to load it from memory.
212 */
213 unsigned smm:8;
d657a98e
ZX
214 };
215};
216
217struct kvm_mmu_page {
218 struct list_head link;
219 struct hlist_node hash_link;
220
221 /*
222 * The following two entries are used to key the shadow page in the
223 * hash table.
224 */
225 gfn_t gfn;
226 union kvm_mmu_page_role role;
227
228 u64 *spt;
229 /* hold the gfn of each spte inside spt */
230 gfn_t *gfns;
4731d4c7 231 bool unsync;
0571d366 232 int root_count; /* Currently serving as active root */
60c8aec6 233 unsigned int unsync_children;
67052b35 234 unsigned long parent_ptes; /* Reverse mapping for parent_pte */
f6f8adee
XG
235
236 /* The page is obsolete if mmu_valid_gen != kvm->arch.mmu_valid_gen. */
5304b8d3 237 unsigned long mmu_valid_gen;
f6f8adee 238
0074ff63 239 DECLARE_BITMAP(unsync_child_bitmap, 512);
c2a2ac2b
XG
240
241#ifdef CONFIG_X86_32
accaefe0
XG
242 /*
243 * Used out of the mmu-lock to avoid reading spte values while an
244 * update is in progress; see the comments in __get_spte_lockless().
245 */
c2a2ac2b
XG
246 int clear_spte_count;
247#endif
248
0cbf8e43 249 /* Number of writes since the last time traversal visited this page. */
a30f47cb 250 int write_flooding_count;
d657a98e
ZX
251};
252
1c08364c
AK
253struct kvm_pio_request {
254 unsigned long count;
1c08364c
AK
255 int in;
256 int port;
257 int size;
1c08364c
AK
258};
259
a0a64f50
XG
260struct rsvd_bits_validate {
261 u64 rsvd_bits_mask[2][4];
262 u64 bad_mt_xwr;
263};
264
d657a98e
ZX
265/*
266 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
267 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
268 * mode.
269 */
270struct kvm_mmu {
f43addd4 271 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
5777ed34 272 unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
e4e517b4 273 u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
78b2c54a
XG
274 int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
275 bool prefault);
6389ee94
AK
276 void (*inject_page_fault)(struct kvm_vcpu *vcpu,
277 struct x86_exception *fault);
1871c602 278 gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
ab9ae313 279 struct x86_exception *exception);
54987b7a
PB
280 gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
281 struct x86_exception *exception);
e8bc217a 282 int (*sync_page)(struct kvm_vcpu *vcpu,
a4a8e6f7 283 struct kvm_mmu_page *sp);
a7052897 284 void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
0f53b5b1 285 void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
7c562522 286 u64 *spte, const void *pte);
d657a98e
ZX
287 hpa_t root_hpa;
288 int root_level;
289 int shadow_root_level;
a770f6f2 290 union kvm_mmu_page_role base_role;
c5a78f2b 291 bool direct_map;
d657a98e 292
97d64b78
AK
293 /*
294 * Bitmap; bit set = permission fault
295 * Byte index: page fault error code [4:1]
296 * Bit index: pte permissions in ACC_* format
297 */
298 u8 permissions[16];
299
d657a98e 300 u64 *pae_root;
81407ca5 301 u64 *lm_root;
c258b62b
XG
302
303 /*
304 * check zero bits on shadow page table entries, these
305 * bits include not only hardware reserved bits but also
306 * the bits spte never used.
307 */
308 struct rsvd_bits_validate shadow_zero_check;
309
a0a64f50 310 struct rsvd_bits_validate guest_rsvd_check;
ff03a073 311
6fd01b71
AK
312 /*
313 * Bitmap: bit set = last pte in walk
314 * index[0:1]: level (zero-based)
315 * index[2]: pte.ps
316 */
317 u8 last_pte_bitmap;
318
2d48a985
JR
319 bool nx;
320
ff03a073 321 u64 pdptrs[4]; /* pae */
d657a98e
ZX
322};
323
f5132b01
GN
324enum pmc_type {
325 KVM_PMC_GP = 0,
326 KVM_PMC_FIXED,
327};
328
329struct kvm_pmc {
330 enum pmc_type type;
331 u8 idx;
332 u64 counter;
333 u64 eventsel;
334 struct perf_event *perf_event;
335 struct kvm_vcpu *vcpu;
336};
337
338struct kvm_pmu {
339 unsigned nr_arch_gp_counters;
340 unsigned nr_arch_fixed_counters;
341 unsigned available_event_types;
342 u64 fixed_ctr_ctrl;
343 u64 global_ctrl;
344 u64 global_status;
345 u64 global_ovf_ctrl;
346 u64 counter_bitmask[2];
347 u64 global_ctrl_mask;
103af0a9 348 u64 reserved_bits;
f5132b01 349 u8 version;
15c7ad51
RR
350 struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
351 struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
f5132b01
GN
352 struct irq_work irq_work;
353 u64 reprogram_pmi;
354};
355
25462f7f
WH
356struct kvm_pmu_ops;
357
360b948d
PB
358enum {
359 KVM_DEBUGREG_BP_ENABLED = 1,
c77fb5fe 360 KVM_DEBUGREG_WONT_EXIT = 2,
ae561ede 361 KVM_DEBUGREG_RELOAD = 4,
360b948d
PB
362};
363
86fd5270
XG
364struct kvm_mtrr_range {
365 u64 base;
366 u64 mask;
19efffa2 367 struct list_head node;
86fd5270
XG
368};
369
70109e7d 370struct kvm_mtrr {
86fd5270 371 struct kvm_mtrr_range var_ranges[KVM_NR_VAR_MTRR];
70109e7d 372 mtrr_type fixed_ranges[KVM_NR_FIXED_MTRR_REGION];
10fac2dc 373 u64 deftype;
19efffa2
XG
374
375 struct list_head head;
70109e7d
XG
376};
377
5c919412
AS
378/* Hyper-V synthetic interrupt controller (SynIC)*/
379struct kvm_vcpu_hv_synic {
380 u64 version;
381 u64 control;
382 u64 msg_page;
383 u64 evt_page;
384 atomic64_t sint[HV_SYNIC_SINT_COUNT];
385 atomic_t sint_to_gsi[HV_SYNIC_SINT_COUNT];
386 DECLARE_BITMAP(auto_eoi_bitmap, 256);
387 DECLARE_BITMAP(vec_bitmap, 256);
388 bool active;
389};
390
e83d5887
AS
391/* Hyper-V per vcpu emulation context */
392struct kvm_vcpu_hv {
393 u64 hv_vapic;
9eec50b8 394 s64 runtime_offset;
5c919412 395 struct kvm_vcpu_hv_synic synic;
e83d5887
AS
396};
397
ad312c7c 398struct kvm_vcpu_arch {
5fdbf976
MT
399 /*
400 * rip and regs accesses must go through
401 * kvm_{register,rip}_{read,write} functions.
402 */
403 unsigned long regs[NR_VCPU_REGS];
404 u32 regs_avail;
405 u32 regs_dirty;
34c16eec
ZX
406
407 unsigned long cr0;
e8467fda 408 unsigned long cr0_guest_owned_bits;
34c16eec
ZX
409 unsigned long cr2;
410 unsigned long cr3;
411 unsigned long cr4;
fc78f519 412 unsigned long cr4_guest_owned_bits;
34c16eec 413 unsigned long cr8;
1371d904 414 u32 hflags;
f6801dff 415 u64 efer;
34c16eec
ZX
416 u64 apic_base;
417 struct kvm_lapic *apic; /* kernel irqchip context */
d62caabb 418 bool apicv_active;
6308630b 419 DECLARE_BITMAP(ioapic_handled_vectors, 256);
41383771 420 unsigned long apic_attention;
e1035715 421 int32_t apic_arb_prio;
34c16eec 422 int mp_state;
34c16eec 423 u64 ia32_misc_enable_msr;
64d60670 424 u64 smbase;
b209749f 425 bool tpr_access_reporting;
20300099 426 u64 ia32_xss;
34c16eec 427
14dfe855
JR
428 /*
429 * Paging state of the vcpu
430 *
431 * If the vcpu runs in guest mode with two level paging this still saves
432 * the paging mode of the l1 guest. This context is always used to
433 * handle faults.
434 */
34c16eec 435 struct kvm_mmu mmu;
8df25a32 436
6539e738
JR
437 /*
438 * Paging state of an L2 guest (used for nested npt)
439 *
440 * This context will save all necessary information to walk page tables
441 * of the an L2 guest. This context is only initialized for page table
442 * walking and not for faulting since we never handle l2 page faults on
443 * the host.
444 */
445 struct kvm_mmu nested_mmu;
446
14dfe855
JR
447 /*
448 * Pointer to the mmu context currently used for
449 * gva_to_gpa translations.
450 */
451 struct kvm_mmu *walk_mmu;
452
53c07b18 453 struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
34c16eec
ZX
454 struct kvm_mmu_memory_cache mmu_page_cache;
455 struct kvm_mmu_memory_cache mmu_page_header_cache;
456
98918833 457 struct fpu guest_fpu;
c447e76b 458 bool eager_fpu;
2acf923e 459 u64 xcr0;
d7876f1b 460 u64 guest_supported_xcr0;
4344ee98 461 u32 guest_xstate_size;
34c16eec 462
34c16eec
ZX
463 struct kvm_pio_request pio;
464 void *pio_data;
465
66fd3f7f
GN
466 u8 event_exit_inst_len;
467
298101da
AK
468 struct kvm_queued_exception {
469 bool pending;
470 bool has_error_code;
ce7ddec4 471 bool reinject;
298101da
AK
472 u8 nr;
473 u32 error_code;
474 } exception;
475
937a7eae
AK
476 struct kvm_queued_interrupt {
477 bool pending;
66fd3f7f 478 bool soft;
937a7eae
AK
479 u8 nr;
480 } interrupt;
481
34c16eec
ZX
482 int halt_request; /* real mode on Intel only */
483
484 int cpuid_nent;
07716717 485 struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
5a4f55cd
EK
486
487 int maxphyaddr;
488
34c16eec
ZX
489 /* emulate context */
490
491 struct x86_emulate_ctxt emulate_ctxt;
7ae441ea
GN
492 bool emulate_regs_need_sync_to_vcpu;
493 bool emulate_regs_need_sync_from_vcpu;
716d51ab 494 int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
18068523
GOC
495
496 gpa_t time;
50d0a0f9 497 struct pvclock_vcpu_time_info hv_clock;
e48672fa 498 unsigned int hw_tsc_khz;
0b79459b
AH
499 struct gfn_to_hva_cache pv_time;
500 bool pv_time_enabled;
51d59c6b
MT
501 /* set guest stopped flag in pvclock flags field */
502 bool pvclock_set_guest_stopped_request;
c9aaa895
GC
503
504 struct {
505 u64 msr_val;
506 u64 last_steal;
507 u64 accum_steal;
508 struct gfn_to_hva_cache stime;
509 struct kvm_steal_time steal;
510 } st;
511
1d5f066e 512 u64 last_guest_tsc;
6f526ec5 513 u64 last_host_tsc;
0dd6a6ed 514 u64 tsc_offset_adjustment;
e26101b1
ZA
515 u64 this_tsc_nsec;
516 u64 this_tsc_write;
0d3da0d2 517 u64 this_tsc_generation;
c285545f 518 bool tsc_catchup;
cc578287
ZA
519 bool tsc_always_catchup;
520 s8 virtual_tsc_shift;
521 u32 virtual_tsc_mult;
522 u32 virtual_tsc_khz;
ba904635 523 s64 ia32_tsc_adjust_msr;
ad721883 524 u64 tsc_scaling_ratio;
3419ffc8 525
7460fb4a
AK
526 atomic_t nmi_queued; /* unprocessed asynchronous NMIs */
527 unsigned nmi_pending; /* NMI queued after currently running handler */
528 bool nmi_injected; /* Trying to inject an NMI this entry */
f077825a 529 bool smi_pending; /* SMI queued after currently running handler */
9ba075a6 530
70109e7d 531 struct kvm_mtrr mtrr_state;
7cb060a9 532 u64 pat;
42dbaa5a 533
360b948d 534 unsigned switch_db_regs;
42dbaa5a
JK
535 unsigned long db[KVM_NR_DB_REGS];
536 unsigned long dr6;
537 unsigned long dr7;
538 unsigned long eff_db[KVM_NR_DB_REGS];
c8639010 539 unsigned long guest_debug_dr7;
890ca9ae
HY
540
541 u64 mcg_cap;
542 u64 mcg_status;
543 u64 mcg_ctl;
544 u64 *mce_banks;
94fe45da 545
bebb106a
XG
546 /* Cache MMIO info */
547 u64 mmio_gva;
548 unsigned access;
549 gfn_t mmio_gfn;
56f17dd3 550 u64 mmio_gen;
bebb106a 551
f5132b01
GN
552 struct kvm_pmu pmu;
553
94fe45da 554 /* used for guest single stepping over the given code position */
94fe45da 555 unsigned long singlestep_rip;
f92653ee 556
e83d5887 557 struct kvm_vcpu_hv hyperv;
f5f48ee1
SY
558
559 cpumask_var_t wbinvd_dirty_mask;
af585b92 560
1cb3f3ae
XG
561 unsigned long last_retry_eip;
562 unsigned long last_retry_addr;
563
af585b92
GN
564 struct {
565 bool halted;
566 gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
344d9588
GN
567 struct gfn_to_hva_cache data;
568 u64 msr_val;
7c90705b 569 u32 id;
6adba527 570 bool send_user_only;
af585b92 571 } apf;
2b036c6b
BO
572
573 /* OSVW MSRs (AMD only) */
574 struct {
575 u64 length;
576 u64 status;
577 } osvw;
ae7a2a3f
MT
578
579 struct {
580 u64 msr_val;
581 struct gfn_to_hva_cache data;
582 } pv_eoi;
93c05d3e
XG
583
584 /*
585 * Indicate whether the access faults on its page table in guest
586 * which is set when fix page fault and used to detect unhandeable
587 * instruction.
588 */
589 bool write_fault_to_shadow_pgtable;
25d92081
YZ
590
591 /* set at EPT violation at this point */
592 unsigned long exit_qualification;
6aef266c
SV
593
594 /* pv related host specific info */
595 struct {
596 bool pv_unhalted;
597 } pv;
7543a635
SR
598
599 int pending_ioapic_eoi;
1c1a9ce9 600 int pending_external_vector;
34c16eec
ZX
601};
602
db3fe4eb 603struct kvm_lpage_info {
db3fe4eb
TY
604 int write_count;
605};
606
607struct kvm_arch_memory_slot {
d89cc617 608 unsigned long *rmap[KVM_NR_PAGE_SIZES];
db3fe4eb
TY
609 struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
610};
611
3548a259
RK
612/*
613 * We use as the mode the number of bits allocated in the LDR for the
614 * logical processor ID. It happens that these are all powers of two.
615 * This makes it is very easy to detect cases where the APICs are
616 * configured for multiple modes; in that case, we cannot use the map and
617 * hence cannot use kvm_irq_delivery_to_apic_fast either.
618 */
619#define KVM_APIC_MODE_XAPIC_CLUSTER 4
620#define KVM_APIC_MODE_XAPIC_FLAT 8
621#define KVM_APIC_MODE_X2APIC 16
622
1e08ec4a
GN
623struct kvm_apic_map {
624 struct rcu_head rcu;
3548a259 625 u8 mode;
1e08ec4a
GN
626 struct kvm_lapic *phys_map[256];
627 /* first index is cluster id second is cpu id in a cluster */
628 struct kvm_lapic *logical_map[16][16];
629};
630
e83d5887
AS
631/* Hyper-V emulation context */
632struct kvm_hv {
633 u64 hv_guest_os_id;
634 u64 hv_hypercall;
635 u64 hv_tsc_page;
e7d9513b
AS
636
637 /* Hyper-v based guest crash (NT kernel bugcheck) parameters */
638 u64 hv_crash_param[HV_X64_MSR_CRASH_PARAMS];
639 u64 hv_crash_ctl;
e83d5887
AS
640};
641
fef9cce0 642struct kvm_arch {
49d5ca26 643 unsigned int n_used_mmu_pages;
f05e70ac 644 unsigned int n_requested_mmu_pages;
39de71ec 645 unsigned int n_max_mmu_pages;
332b207d 646 unsigned int indirect_shadow_pages;
5304b8d3 647 unsigned long mmu_valid_gen;
f05e70ac
ZX
648 struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
649 /*
650 * Hash table of struct kvm_mmu_page.
651 */
652 struct list_head active_mmu_pages;
365c8868
XG
653 struct list_head zapped_obsolete_pages;
654
4d5c5d0f 655 struct list_head assigned_dev_head;
19de40a8 656 struct iommu_domain *iommu_domain;
d96eb2c6 657 bool iommu_noncoherent;
e0f0bbc5
AW
658#define __KVM_HAVE_ARCH_NONCOHERENT_DMA
659 atomic_t noncoherent_dma_count;
5544eb9b
PB
660#define __KVM_HAVE_ARCH_ASSIGNED_DEVICE
661 atomic_t assigned_device_count;
d7deeeb0
ZX
662 struct kvm_pic *vpic;
663 struct kvm_ioapic *vioapic;
7837699f 664 struct kvm_pit *vpit;
42720138 665 atomic_t vapics_in_nmi_mode;
1e08ec4a
GN
666 struct mutex apic_map_lock;
667 struct kvm_apic_map *apic_map;
bfc6d222 668
bfc6d222 669 unsigned int tss_addr;
c24ae0dc 670 bool apic_access_page_done;
18068523
GOC
671
672 gpa_t wall_clock;
b7ebfb05 673
b7ebfb05 674 bool ept_identity_pagetable_done;
b927a3ce 675 gpa_t ept_identity_map_addr;
5550af4d
SY
676
677 unsigned long irq_sources_bitmap;
afbcf7ab 678 s64 kvmclock_offset;
038f8c11 679 raw_spinlock_t tsc_write_lock;
f38e098f 680 u64 last_tsc_nsec;
f38e098f 681 u64 last_tsc_write;
5d3cb0f6 682 u32 last_tsc_khz;
e26101b1
ZA
683 u64 cur_tsc_nsec;
684 u64 cur_tsc_write;
685 u64 cur_tsc_offset;
0d3da0d2 686 u64 cur_tsc_generation;
b48aa97e 687 int nr_vcpus_matched_tsc;
ffde22ac 688
d828199e
MT
689 spinlock_t pvclock_gtod_sync_lock;
690 bool use_master_clock;
691 u64 master_kernel_ns;
692 cycle_t master_cycle_now;
7e44e449 693 struct delayed_work kvmclock_update_work;
332967a3 694 struct delayed_work kvmclock_sync_work;
d828199e 695
ffde22ac 696 struct kvm_xen_hvm_config xen_hvm_config;
55cd8e5a 697
6ef768fa
PB
698 /* reads protected by irq_srcu, writes by irq_lock */
699 struct hlist_head mask_notifier_list;
700
e83d5887 701 struct kvm_hv hyperv;
b034cf01
XG
702
703 #ifdef CONFIG_KVM_MMU_AUDIT
704 int audit_point;
705 #endif
54750f2c
MT
706
707 bool boot_vcpu_runs_old_kvmclock;
d71ba788 708 u32 bsp_vcpu_id;
90de4a18
NA
709
710 u64 disabled_quirks;
49df6397
SR
711
712 bool irqchip_split;
b053b2ae 713 u8 nr_reserved_ioapic_pins;
d69fb81f
ZX
714};
715
0711456c
ZX
716struct kvm_vm_stat {
717 u32 mmu_shadow_zapped;
718 u32 mmu_pte_write;
719 u32 mmu_pte_updated;
720 u32 mmu_pde_zapped;
721 u32 mmu_flooded;
722 u32 mmu_recycled;
dfc5aa00 723 u32 mmu_cache_miss;
4731d4c7 724 u32 mmu_unsync;
0711456c 725 u32 remote_tlb_flush;
05da4558 726 u32 lpages;
0711456c
ZX
727};
728
77b4c255
ZX
729struct kvm_vcpu_stat {
730 u32 pf_fixed;
731 u32 pf_guest;
732 u32 tlb_flush;
733 u32 invlpg;
734
735 u32 exits;
736 u32 io_exits;
737 u32 mmio_exits;
738 u32 signal_exits;
739 u32 irq_window_exits;
f08864b4 740 u32 nmi_window_exits;
77b4c255 741 u32 halt_exits;
f7819512 742 u32 halt_successful_poll;
62bea5bf 743 u32 halt_attempted_poll;
77b4c255
ZX
744 u32 halt_wakeup;
745 u32 request_irq_exits;
746 u32 irq_exits;
747 u32 host_state_reload;
748 u32 efer_reload;
749 u32 fpu_reload;
750 u32 insn_emulation;
751 u32 insn_emulation_fail;
f11c3a8d 752 u32 hypercalls;
fa89a817 753 u32 irq_injections;
c4abb7c9 754 u32 nmi_injections;
77b4c255 755};
ad312c7c 756
8a76d7f2
JR
757struct x86_instruction_info;
758
8fe8ab46
WA
759struct msr_data {
760 bool host_initiated;
761 u32 index;
762 u64 data;
763};
764
cb5281a5
PB
765struct kvm_lapic_irq {
766 u32 vector;
b7cb2231
PB
767 u16 delivery_mode;
768 u16 dest_mode;
769 bool level;
770 u16 trig_mode;
cb5281a5
PB
771 u32 shorthand;
772 u32 dest_id;
93bbf0b8 773 bool msi_redir_hint;
cb5281a5
PB
774};
775
ea4a5ff8
ZX
776struct kvm_x86_ops {
777 int (*cpu_has_kvm_support)(void); /* __init */
778 int (*disabled_by_bios)(void); /* __init */
13a34e06
RK
779 int (*hardware_enable)(void);
780 void (*hardware_disable)(void);
ea4a5ff8
ZX
781 void (*check_processor_compatibility)(void *rtn);
782 int (*hardware_setup)(void); /* __init */
783 void (*hardware_unsetup)(void); /* __exit */
774ead3a 784 bool (*cpu_has_accelerated_tpr)(void);
6d396b55 785 bool (*cpu_has_high_real_mode_segbase)(void);
0e851880 786 void (*cpuid_update)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
787
788 /* Create, but do not attach this VCPU */
789 struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
790 void (*vcpu_free)(struct kvm_vcpu *vcpu);
d28bc9dd 791 void (*vcpu_reset)(struct kvm_vcpu *vcpu, bool init_event);
ea4a5ff8
ZX
792
793 void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
794 void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
795 void (*vcpu_put)(struct kvm_vcpu *vcpu);
ea4a5ff8 796
a96036b8 797 void (*update_bp_intercept)(struct kvm_vcpu *vcpu);
609e36d3 798 int (*get_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
8fe8ab46 799 int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
ea4a5ff8
ZX
800 u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
801 void (*get_segment)(struct kvm_vcpu *vcpu,
802 struct kvm_segment *var, int seg);
2e4d2653 803 int (*get_cpl)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
804 void (*set_segment)(struct kvm_vcpu *vcpu,
805 struct kvm_segment *var, int seg);
806 void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
e8467fda 807 void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
aff48baa 808 void (*decache_cr3)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
809 void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
810 void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
811 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
5e1746d6 812 int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
ea4a5ff8 813 void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
89a27f4d
GN
814 void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
815 void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
816 void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
817 void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
73aaf249
JK
818 u64 (*get_dr6)(struct kvm_vcpu *vcpu);
819 void (*set_dr6)(struct kvm_vcpu *vcpu, unsigned long value);
c77fb5fe 820 void (*sync_dirty_debug_regs)(struct kvm_vcpu *vcpu);
020df079 821 void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
5fdbf976 822 void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
ea4a5ff8
ZX
823 unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
824 void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
0fdd74f7 825 void (*fpu_activate)(struct kvm_vcpu *vcpu);
02daab21 826 void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
827
828 void (*tlb_flush)(struct kvm_vcpu *vcpu);
ea4a5ff8 829
851ba692
AK
830 void (*run)(struct kvm_vcpu *vcpu);
831 int (*handle_exit)(struct kvm_vcpu *vcpu);
ea4a5ff8 832 void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
2809f5d2 833 void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
37ccdcbe 834 u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
835 void (*patch_hypercall)(struct kvm_vcpu *vcpu,
836 unsigned char *hypercall_addr);
66fd3f7f 837 void (*set_irq)(struct kvm_vcpu *vcpu);
95ba8273 838 void (*set_nmi)(struct kvm_vcpu *vcpu);
298101da 839 void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
ce7ddec4
JR
840 bool has_error_code, u32 error_code,
841 bool reinject);
b463a6f7 842 void (*cancel_injection)(struct kvm_vcpu *vcpu);
78646121 843 int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
95ba8273 844 int (*nmi_allowed)(struct kvm_vcpu *vcpu);
3cfc3092
JK
845 bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
846 void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
c9a7953f
JK
847 void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
848 void (*enable_irq_window)(struct kvm_vcpu *vcpu);
95ba8273 849 void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
d62caabb
AS
850 bool (*get_enable_apicv)(void);
851 void (*refresh_apicv_exec_ctrl)(struct kvm_vcpu *vcpu);
c7c9c56c
YZ
852 void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
853 void (*hwapic_isr_update)(struct kvm *kvm, int isr);
6308630b 854 void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap);
8d14695f 855 void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu, bool set);
4256f43f 856 void (*set_apic_access_page_addr)(struct kvm_vcpu *vcpu, hpa_t hpa);
a20ed54d
YZ
857 void (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
858 void (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
ea4a5ff8 859 int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
67253af5 860 int (*get_tdp_level)(void);
4b12f0de 861 u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
17cc3935 862 int (*get_lpage_level)(void);
4e47c7a6 863 bool (*rdtscp_supported)(void);
ad756a16 864 bool (*invpcid_supported)(void);
58ea6767 865 void (*adjust_tsc_offset_guest)(struct kvm_vcpu *vcpu, s64 adjustment);
344f414f 866
1c97f0a0
JR
867 void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
868
d4330ef2
JR
869 void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
870
f5f48ee1
SY
871 bool (*has_wbinvd_exit)(void);
872
ba904635 873 u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu);
99e3e30a
ZA
874 void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
875
886b470c 876 u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu, u64 host_tsc);
857e4099 877
586f9607 878 void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
8a76d7f2
JR
879
880 int (*check_intercept)(struct kvm_vcpu *vcpu,
881 struct x86_instruction_info *info,
882 enum x86_intercept_stage stage);
a547c6db 883 void (*handle_external_intr)(struct kvm_vcpu *vcpu);
da8999d3 884 bool (*mpx_supported)(void);
55412b2e 885 bool (*xsaves_supported)(void);
b6b8a145
JK
886
887 int (*check_nested_events)(struct kvm_vcpu *vcpu, bool external_intr);
ae97a3b8
RK
888
889 void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
88178fd4
KH
890
891 /*
892 * Arch-specific dirty logging hooks. These hooks are only supposed to
893 * be valid if the specific arch has hardware-accelerated dirty logging
894 * mechanism. Currently only for PML on VMX.
895 *
896 * - slot_enable_log_dirty:
897 * called when enabling log dirty mode for the slot.
898 * - slot_disable_log_dirty:
899 * called when disabling log dirty mode for the slot.
900 * also called when slot is created with log dirty disabled.
901 * - flush_log_dirty:
902 * called before reporting dirty_bitmap to userspace.
903 * - enable_log_dirty_pt_masked:
904 * called when reenabling log dirty for the GFNs in the mask after
905 * corresponding bits are cleared in slot->dirty_bitmap.
906 */
907 void (*slot_enable_log_dirty)(struct kvm *kvm,
908 struct kvm_memory_slot *slot);
909 void (*slot_disable_log_dirty)(struct kvm *kvm,
910 struct kvm_memory_slot *slot);
911 void (*flush_log_dirty)(struct kvm *kvm);
912 void (*enable_log_dirty_pt_masked)(struct kvm *kvm,
913 struct kvm_memory_slot *slot,
914 gfn_t offset, unsigned long mask);
25462f7f
WH
915 /* pmu operations of sub-arch */
916 const struct kvm_pmu_ops *pmu_ops;
efc64404 917
bf9f6ac8
FW
918 /*
919 * Architecture specific hooks for vCPU blocking due to
920 * HLT instruction.
921 * Returns for .pre_block():
922 * - 0 means continue to block the vCPU.
923 * - 1 means we cannot block the vCPU since some event
924 * happens during this period, such as, 'ON' bit in
925 * posted-interrupts descriptor is set.
926 */
927 int (*pre_block)(struct kvm_vcpu *vcpu);
928 void (*post_block)(struct kvm_vcpu *vcpu);
efc64404
FW
929 int (*update_pi_irte)(struct kvm *kvm, unsigned int host_irq,
930 uint32_t guest_irq, bool set);
ea4a5ff8
ZX
931};
932
af585b92 933struct kvm_arch_async_pf {
7c90705b 934 u32 token;
af585b92 935 gfn_t gfn;
fb67e14f 936 unsigned long cr3;
c4806acd 937 bool direct_map;
af585b92
GN
938};
939
97896d04
ZX
940extern struct kvm_x86_ops *kvm_x86_ops;
941
54f1585a
ZX
942int kvm_mmu_module_init(void);
943void kvm_mmu_module_exit(void);
944
945void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
946int kvm_mmu_create(struct kvm_vcpu *vcpu);
8a3c1a33 947void kvm_mmu_setup(struct kvm_vcpu *vcpu);
7b52345e 948void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
4b12f0de 949 u64 dirty_mask, u64 nx_mask, u64 x_mask);
54f1585a 950
8a3c1a33 951void kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
1c91cad4
KH
952void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
953 struct kvm_memory_slot *memslot);
3ea3b7fa 954void kvm_mmu_zap_collapsible_sptes(struct kvm *kvm,
f36f3f28 955 const struct kvm_memory_slot *memslot);
f4b4b180
KH
956void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
957 struct kvm_memory_slot *memslot);
958void kvm_mmu_slot_largepage_remove_write_access(struct kvm *kvm,
959 struct kvm_memory_slot *memslot);
960void kvm_mmu_slot_set_dirty(struct kvm *kvm,
961 struct kvm_memory_slot *memslot);
962void kvm_mmu_clear_dirty_pt_masked(struct kvm *kvm,
963 struct kvm_memory_slot *slot,
964 gfn_t gfn_offset, unsigned long mask);
54f1585a 965void kvm_mmu_zap_all(struct kvm *kvm);
54bf36aa 966void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm, struct kvm_memslots *slots);
3ad82a7e 967unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
54f1585a
ZX
968void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
969
ff03a073 970int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
cc4b6871 971
3200f405 972int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
9f811285 973 const void *val, int bytes);
2f333bcb 974
6ef768fa
PB
975struct kvm_irq_mask_notifier {
976 void (*func)(struct kvm_irq_mask_notifier *kimn, bool masked);
977 int irq;
978 struct hlist_node link;
979};
980
981void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
982 struct kvm_irq_mask_notifier *kimn);
983void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
984 struct kvm_irq_mask_notifier *kimn);
985void kvm_fire_mask_notifiers(struct kvm *kvm, unsigned irqchip, unsigned pin,
986 bool mask);
987
2f333bcb 988extern bool tdp_enabled;
9f811285 989
a3e06bbe
LJ
990u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
991
92a1f12d
JR
992/* control of guest tsc rate supported? */
993extern bool kvm_has_tsc_control;
92a1f12d
JR
994/* maximum supported tsc_khz for guests */
995extern u32 kvm_max_guest_tsc_khz;
bc9b961b
HZ
996/* number of bits of the fractional part of the TSC scaling ratio */
997extern u8 kvm_tsc_scaling_ratio_frac_bits;
998/* maximum allowed value of TSC scaling ratio */
999extern u64 kvm_max_tsc_scaling_ratio;
92a1f12d 1000
54f1585a 1001enum emulation_result {
ac0a48c3
PB
1002 EMULATE_DONE, /* no further processing */
1003 EMULATE_USER_EXIT, /* kvm_run ready for userspace exit */
54f1585a
ZX
1004 EMULATE_FAIL, /* can't emulate this instruction */
1005};
1006
571008da
SY
1007#define EMULTYPE_NO_DECODE (1 << 0)
1008#define EMULTYPE_TRAP_UD (1 << 1)
ba8afb6b 1009#define EMULTYPE_SKIP (1 << 2)
1cb3f3ae 1010#define EMULTYPE_RETRY (1 << 3)
991eebf9 1011#define EMULTYPE_NO_REEXECUTE (1 << 4)
dc25e89e
AP
1012int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
1013 int emulation_type, void *insn, int insn_len);
51d8b661
AP
1014
1015static inline int emulate_instruction(struct kvm_vcpu *vcpu,
1016 int emulation_type)
1017{
dc25e89e 1018 return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
51d8b661
AP
1019}
1020
f2b4b7dd 1021void kvm_enable_efer_bits(u64);
384bb783 1022bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
609e36d3 1023int kvm_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
8fe8ab46 1024int kvm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
54f1585a
ZX
1025
1026struct x86_emulate_ctxt;
1027
cf8f70bf 1028int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
54f1585a
ZX
1029void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
1030int kvm_emulate_halt(struct kvm_vcpu *vcpu);
5cb56059 1031int kvm_vcpu_halt(struct kvm_vcpu *vcpu);
f5f48ee1 1032int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
54f1585a 1033
3e6e0aab 1034void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
c697518a 1035int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
2b4a273b 1036void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
3e6e0aab 1037
7f3d35fd
KW
1038int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
1039 int reason, bool has_error_code, u32 error_code);
37817f29 1040
49a9b07e 1041int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
2390218b 1042int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
a83b29c6 1043int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
eea1cff9 1044int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
020df079
GN
1045int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
1046int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
2d3ad1f4
AK
1047unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
1048void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
54f1585a 1049void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
2acf923e 1050int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
54f1585a 1051
609e36d3 1052int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
8fe8ab46 1053int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
54f1585a 1054
91586a3b
JK
1055unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
1056void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
022cd0e8 1057bool kvm_rdpmc(struct kvm_vcpu *vcpu);
91586a3b 1058
298101da
AK
1059void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1060void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
ce7ddec4
JR
1061void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1062void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
6389ee94 1063void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
ec92fe44
JR
1064int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
1065 gfn_t gfn, void *data, int offset, int len,
1066 u32 access);
0a79b009 1067bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
16f8a6f9 1068bool kvm_require_dr(struct kvm_vcpu *vcpu, int dr);
298101da 1069
1a577b72
MT
1070static inline int __kvm_irq_line_state(unsigned long *irq_state,
1071 int irq_source_id, int level)
1072{
1073 /* Logical OR for level trig interrupt */
1074 if (level)
1075 __set_bit(irq_source_id, irq_state);
1076 else
1077 __clear_bit(irq_source_id, irq_state);
1078
1079 return !!(*irq_state);
1080}
1081
1082int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
1083void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
3de42dc0 1084
3419ffc8
SY
1085void kvm_inject_nmi(struct kvm_vcpu *vcpu);
1086
54f1585a 1087void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
f57f2ef5 1088 const u8 *new, int bytes);
1cb3f3ae 1089int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
54f1585a
ZX
1090int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
1091void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
1092int kvm_mmu_load(struct kvm_vcpu *vcpu);
1093void kvm_mmu_unload(struct kvm_vcpu *vcpu);
0ba73cda 1094void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
54987b7a
PB
1095gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1096 struct x86_exception *exception);
ab9ae313
AK
1097gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
1098 struct x86_exception *exception);
1099gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
1100 struct x86_exception *exception);
1101gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
1102 struct x86_exception *exception);
1103gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
1104 struct x86_exception *exception);
54f1585a 1105
d62caabb
AS
1106void kvm_vcpu_deactivate_apicv(struct kvm_vcpu *vcpu);
1107
54f1585a
ZX
1108int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
1109
dc25e89e
AP
1110int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
1111 void *insn, int insn_len);
a7052897 1112void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
d8d173da 1113void kvm_mmu_new_cr3(struct kvm_vcpu *vcpu);
34c16eec 1114
18552672 1115void kvm_enable_tdp(void);
5f4cb662 1116void kvm_disable_tdp(void);
18552672 1117
54987b7a
PB
1118static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1119 struct x86_exception *exception)
e459e322
XG
1120{
1121 return gpa;
1122}
1123
ec6d273d
ZX
1124static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
1125{
1126 struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
1127
1128 return (struct kvm_mmu_page *)page_private(page);
1129}
1130
d6e88aec 1131static inline u16 kvm_read_ldt(void)
ec6d273d
ZX
1132{
1133 u16 ldt;
1134 asm("sldt %0" : "=g"(ldt));
1135 return ldt;
1136}
1137
d6e88aec 1138static inline void kvm_load_ldt(u16 sel)
ec6d273d
ZX
1139{
1140 asm("lldt %0" : : "rm"(sel));
1141}
ec6d273d 1142
ec6d273d
ZX
1143#ifdef CONFIG_X86_64
1144static inline unsigned long read_msr(unsigned long msr)
1145{
1146 u64 value;
1147
1148 rdmsrl(msr, value);
1149 return value;
1150}
1151#endif
1152
ec6d273d
ZX
1153static inline u32 get_rdx_init_val(void)
1154{
1155 return 0x600; /* P6 family */
1156}
1157
c1a5d4f9
AK
1158static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
1159{
1160 kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
1161}
1162
854e8bb1
NA
1163static inline u64 get_canonical(u64 la)
1164{
1165 return ((int64_t)la << 16) >> 16;
1166}
1167
1168static inline bool is_noncanonical_address(u64 la)
1169{
1170#ifdef CONFIG_X86_64
1171 return get_canonical(la) != la;
1172#else
1173 return false;
1174#endif
1175}
1176
ec6d273d
ZX
1177#define TSS_IOPB_BASE_OFFSET 0x66
1178#define TSS_BASE_SIZE 0x68
1179#define TSS_IOPB_SIZE (65536 / 8)
1180#define TSS_REDIRECTION_SIZE (256 / 8)
7d76b4d3
JP
1181#define RMODE_TSS_SIZE \
1182 (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
53e0aa7b 1183
37817f29
IE
1184enum {
1185 TASK_SWITCH_CALL = 0,
1186 TASK_SWITCH_IRET = 1,
1187 TASK_SWITCH_JMP = 2,
1188 TASK_SWITCH_GATE = 3,
1189};
1190
1371d904 1191#define HF_GIF_MASK (1 << 0)
3d6368ef
AG
1192#define HF_HIF_MASK (1 << 1)
1193#define HF_VINTR_MASK (1 << 2)
95ba8273 1194#define HF_NMI_MASK (1 << 3)
44c11430 1195#define HF_IRET_MASK (1 << 4)
ec9e60b2 1196#define HF_GUEST_MASK (1 << 5) /* VCPU is in guest-mode */
f077825a
PB
1197#define HF_SMM_MASK (1 << 6)
1198#define HF_SMM_INSIDE_NMI_MASK (1 << 7)
1371d904 1199
699023e2
PB
1200#define __KVM_VCPU_MULTIPLE_ADDRESS_SPACE
1201#define KVM_ADDRESS_SPACE_NUM 2
1202
1203#define kvm_arch_vcpu_memslots_id(vcpu) ((vcpu)->arch.hflags & HF_SMM_MASK ? 1 : 0)
1204#define kvm_memslots_for_spte_role(kvm, role) __kvm_memslots(kvm, (role).smm)
1371d904 1205
4ecac3fd
AK
1206/*
1207 * Hardware virtualization extension instructions may fault if a
1208 * reboot turns off virtualization while processes are running.
1209 * Trap the fault and ignore the instruction if that happens.
1210 */
b7c4145b 1211asmlinkage void kvm_spurious_fault(void);
4ecac3fd 1212
5e520e62 1213#define ____kvm_handle_fault_on_reboot(insn, cleanup_insn) \
4ecac3fd 1214 "666: " insn "\n\t" \
b7c4145b 1215 "668: \n\t" \
18b13e54 1216 ".pushsection .fixup, \"ax\" \n" \
4ecac3fd 1217 "667: \n\t" \
5e520e62 1218 cleanup_insn "\n\t" \
b7c4145b
AK
1219 "cmpb $0, kvm_rebooting \n\t" \
1220 "jne 668b \n\t" \
8ceed347 1221 __ASM_SIZE(push) " $666b \n\t" \
b7c4145b 1222 "call kvm_spurious_fault \n\t" \
4ecac3fd 1223 ".popsection \n\t" \
3ee89722 1224 _ASM_EXTABLE(666b, 667b)
4ecac3fd 1225
5e520e62
AK
1226#define __kvm_handle_fault_on_reboot(insn) \
1227 ____kvm_handle_fault_on_reboot(insn, "")
1228
e930bffe
AA
1229#define KVM_ARCH_WANT_MMU_NOTIFIER
1230int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
b3ae2096 1231int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
57128468 1232int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
8ee53820 1233int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
3da0dd43 1234void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
c7c9c56c 1235int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
a1b37100
GN
1236int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1237int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
0b71785d 1238int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
d28bc9dd 1239void kvm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event);
4256f43f 1240void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu);
fe71557a
TC
1241void kvm_arch_mmu_notifier_invalidate_page(struct kvm *kvm,
1242 unsigned long address);
e930bffe 1243
18863bdd 1244void kvm_define_shared_msr(unsigned index, u32 msr);
8b3c3104 1245int kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
18863bdd 1246
35181e86 1247u64 kvm_scale_tsc(struct kvm_vcpu *vcpu, u64 tsc);
4ba76538 1248u64 kvm_read_l1_tsc(struct kvm_vcpu *vcpu, u64 host_tsc);
35181e86 1249
82b32774 1250unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
f92653ee
JK
1251bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1252
af585b92
GN
1253void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1254 struct kvm_async_pf *work);
1255void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1256 struct kvm_async_pf *work);
56028d08
GN
1257void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1258 struct kvm_async_pf *work);
7c90705b 1259bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
af585b92
GN
1260extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1261
db8fcefa
AP
1262void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1263
f5132b01
GN
1264int kvm_is_in_guest(void);
1265
1d8007bd
PB
1266int __x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa, u32 size);
1267int x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa, u32 size);
d71ba788
PB
1268bool kvm_vcpu_is_reset_bsp(struct kvm_vcpu *vcpu);
1269bool kvm_vcpu_is_bsp(struct kvm_vcpu *vcpu);
f5132b01 1270
8feb4a04
FW
1271bool kvm_intr_is_single_vcpu(struct kvm *kvm, struct kvm_lapic_irq *irq,
1272 struct kvm_vcpu **dest_vcpu);
1273
d84f1e07
FW
1274void kvm_set_msi_irq(struct kvm_kernel_irq_routing_entry *e,
1275 struct kvm_lapic_irq *irq);
197a4f4b 1276
3217f7c2
CD
1277static inline void kvm_arch_vcpu_blocking(struct kvm_vcpu *vcpu) {}
1278static inline void kvm_arch_vcpu_unblocking(struct kvm_vcpu *vcpu) {}
1279
1965aae3 1280#endif /* _ASM_X86_KVM_HOST_H */