]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blame - arch/x86/include/asm/kvm_host.h
KVM: x86: MMU: Encapsulate the type of rmap-chain head in a new struct
[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
018aabb5
TY
217struct kvm_rmap_head {
218 unsigned long val;
219};
220
d657a98e
ZX
221struct kvm_mmu_page {
222 struct list_head link;
223 struct hlist_node hash_link;
224
225 /*
226 * The following two entries are used to key the shadow page in the
227 * hash table.
228 */
229 gfn_t gfn;
230 union kvm_mmu_page_role role;
231
232 u64 *spt;
233 /* hold the gfn of each spte inside spt */
234 gfn_t *gfns;
4731d4c7 235 bool unsync;
0571d366 236 int root_count; /* Currently serving as active root */
60c8aec6 237 unsigned int unsync_children;
018aabb5 238 struct kvm_rmap_head parent_ptes; /* rmap pointers to parent sptes */
f6f8adee
XG
239
240 /* The page is obsolete if mmu_valid_gen != kvm->arch.mmu_valid_gen. */
5304b8d3 241 unsigned long mmu_valid_gen;
f6f8adee 242
0074ff63 243 DECLARE_BITMAP(unsync_child_bitmap, 512);
c2a2ac2b
XG
244
245#ifdef CONFIG_X86_32
accaefe0
XG
246 /*
247 * Used out of the mmu-lock to avoid reading spte values while an
248 * update is in progress; see the comments in __get_spte_lockless().
249 */
c2a2ac2b
XG
250 int clear_spte_count;
251#endif
252
0cbf8e43 253 /* Number of writes since the last time traversal visited this page. */
a30f47cb 254 int write_flooding_count;
d657a98e
ZX
255};
256
1c08364c
AK
257struct kvm_pio_request {
258 unsigned long count;
1c08364c
AK
259 int in;
260 int port;
261 int size;
1c08364c
AK
262};
263
a0a64f50
XG
264struct rsvd_bits_validate {
265 u64 rsvd_bits_mask[2][4];
266 u64 bad_mt_xwr;
267};
268
d657a98e
ZX
269/*
270 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
271 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
272 * mode.
273 */
274struct kvm_mmu {
f43addd4 275 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
5777ed34 276 unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
e4e517b4 277 u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
78b2c54a
XG
278 int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
279 bool prefault);
6389ee94
AK
280 void (*inject_page_fault)(struct kvm_vcpu *vcpu,
281 struct x86_exception *fault);
1871c602 282 gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
ab9ae313 283 struct x86_exception *exception);
54987b7a
PB
284 gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
285 struct x86_exception *exception);
e8bc217a 286 int (*sync_page)(struct kvm_vcpu *vcpu,
a4a8e6f7 287 struct kvm_mmu_page *sp);
a7052897 288 void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
0f53b5b1 289 void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
7c562522 290 u64 *spte, const void *pte);
d657a98e
ZX
291 hpa_t root_hpa;
292 int root_level;
293 int shadow_root_level;
a770f6f2 294 union kvm_mmu_page_role base_role;
c5a78f2b 295 bool direct_map;
d657a98e 296
97d64b78
AK
297 /*
298 * Bitmap; bit set = permission fault
299 * Byte index: page fault error code [4:1]
300 * Bit index: pte permissions in ACC_* format
301 */
302 u8 permissions[16];
303
d657a98e 304 u64 *pae_root;
81407ca5 305 u64 *lm_root;
c258b62b
XG
306
307 /*
308 * check zero bits on shadow page table entries, these
309 * bits include not only hardware reserved bits but also
310 * the bits spte never used.
311 */
312 struct rsvd_bits_validate shadow_zero_check;
313
a0a64f50 314 struct rsvd_bits_validate guest_rsvd_check;
ff03a073 315
6fd01b71
AK
316 /*
317 * Bitmap: bit set = last pte in walk
318 * index[0:1]: level (zero-based)
319 * index[2]: pte.ps
320 */
321 u8 last_pte_bitmap;
322
2d48a985
JR
323 bool nx;
324
ff03a073 325 u64 pdptrs[4]; /* pae */
d657a98e
ZX
326};
327
f5132b01
GN
328enum pmc_type {
329 KVM_PMC_GP = 0,
330 KVM_PMC_FIXED,
331};
332
333struct kvm_pmc {
334 enum pmc_type type;
335 u8 idx;
336 u64 counter;
337 u64 eventsel;
338 struct perf_event *perf_event;
339 struct kvm_vcpu *vcpu;
340};
341
342struct kvm_pmu {
343 unsigned nr_arch_gp_counters;
344 unsigned nr_arch_fixed_counters;
345 unsigned available_event_types;
346 u64 fixed_ctr_ctrl;
347 u64 global_ctrl;
348 u64 global_status;
349 u64 global_ovf_ctrl;
350 u64 counter_bitmask[2];
351 u64 global_ctrl_mask;
103af0a9 352 u64 reserved_bits;
f5132b01 353 u8 version;
15c7ad51
RR
354 struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
355 struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
f5132b01
GN
356 struct irq_work irq_work;
357 u64 reprogram_pmi;
358};
359
25462f7f
WH
360struct kvm_pmu_ops;
361
360b948d
PB
362enum {
363 KVM_DEBUGREG_BP_ENABLED = 1,
c77fb5fe 364 KVM_DEBUGREG_WONT_EXIT = 2,
ae561ede 365 KVM_DEBUGREG_RELOAD = 4,
360b948d
PB
366};
367
86fd5270
XG
368struct kvm_mtrr_range {
369 u64 base;
370 u64 mask;
19efffa2 371 struct list_head node;
86fd5270
XG
372};
373
70109e7d 374struct kvm_mtrr {
86fd5270 375 struct kvm_mtrr_range var_ranges[KVM_NR_VAR_MTRR];
70109e7d 376 mtrr_type fixed_ranges[KVM_NR_FIXED_MTRR_REGION];
10fac2dc 377 u64 deftype;
19efffa2
XG
378
379 struct list_head head;
70109e7d
XG
380};
381
5c919412
AS
382/* Hyper-V synthetic interrupt controller (SynIC)*/
383struct kvm_vcpu_hv_synic {
384 u64 version;
385 u64 control;
386 u64 msg_page;
387 u64 evt_page;
388 atomic64_t sint[HV_SYNIC_SINT_COUNT];
389 atomic_t sint_to_gsi[HV_SYNIC_SINT_COUNT];
390 DECLARE_BITMAP(auto_eoi_bitmap, 256);
391 DECLARE_BITMAP(vec_bitmap, 256);
392 bool active;
393};
394
e83d5887
AS
395/* Hyper-V per vcpu emulation context */
396struct kvm_vcpu_hv {
397 u64 hv_vapic;
9eec50b8 398 s64 runtime_offset;
5c919412 399 struct kvm_vcpu_hv_synic synic;
db397571 400 struct kvm_hyperv_exit exit;
e83d5887
AS
401};
402
ad312c7c 403struct kvm_vcpu_arch {
5fdbf976
MT
404 /*
405 * rip and regs accesses must go through
406 * kvm_{register,rip}_{read,write} functions.
407 */
408 unsigned long regs[NR_VCPU_REGS];
409 u32 regs_avail;
410 u32 regs_dirty;
34c16eec
ZX
411
412 unsigned long cr0;
e8467fda 413 unsigned long cr0_guest_owned_bits;
34c16eec
ZX
414 unsigned long cr2;
415 unsigned long cr3;
416 unsigned long cr4;
fc78f519 417 unsigned long cr4_guest_owned_bits;
34c16eec 418 unsigned long cr8;
1371d904 419 u32 hflags;
f6801dff 420 u64 efer;
34c16eec
ZX
421 u64 apic_base;
422 struct kvm_lapic *apic; /* kernel irqchip context */
d62caabb 423 bool apicv_active;
6308630b 424 DECLARE_BITMAP(ioapic_handled_vectors, 256);
41383771 425 unsigned long apic_attention;
e1035715 426 int32_t apic_arb_prio;
34c16eec 427 int mp_state;
34c16eec 428 u64 ia32_misc_enable_msr;
64d60670 429 u64 smbase;
b209749f 430 bool tpr_access_reporting;
20300099 431 u64 ia32_xss;
34c16eec 432
14dfe855
JR
433 /*
434 * Paging state of the vcpu
435 *
436 * If the vcpu runs in guest mode with two level paging this still saves
437 * the paging mode of the l1 guest. This context is always used to
438 * handle faults.
439 */
34c16eec 440 struct kvm_mmu mmu;
8df25a32 441
6539e738
JR
442 /*
443 * Paging state of an L2 guest (used for nested npt)
444 *
445 * This context will save all necessary information to walk page tables
446 * of the an L2 guest. This context is only initialized for page table
447 * walking and not for faulting since we never handle l2 page faults on
448 * the host.
449 */
450 struct kvm_mmu nested_mmu;
451
14dfe855
JR
452 /*
453 * Pointer to the mmu context currently used for
454 * gva_to_gpa translations.
455 */
456 struct kvm_mmu *walk_mmu;
457
53c07b18 458 struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
34c16eec
ZX
459 struct kvm_mmu_memory_cache mmu_page_cache;
460 struct kvm_mmu_memory_cache mmu_page_header_cache;
461
98918833 462 struct fpu guest_fpu;
c447e76b 463 bool eager_fpu;
2acf923e 464 u64 xcr0;
d7876f1b 465 u64 guest_supported_xcr0;
4344ee98 466 u32 guest_xstate_size;
34c16eec 467
34c16eec
ZX
468 struct kvm_pio_request pio;
469 void *pio_data;
470
66fd3f7f
GN
471 u8 event_exit_inst_len;
472
298101da
AK
473 struct kvm_queued_exception {
474 bool pending;
475 bool has_error_code;
ce7ddec4 476 bool reinject;
298101da
AK
477 u8 nr;
478 u32 error_code;
479 } exception;
480
937a7eae
AK
481 struct kvm_queued_interrupt {
482 bool pending;
66fd3f7f 483 bool soft;
937a7eae
AK
484 u8 nr;
485 } interrupt;
486
34c16eec
ZX
487 int halt_request; /* real mode on Intel only */
488
489 int cpuid_nent;
07716717 490 struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
5a4f55cd
EK
491
492 int maxphyaddr;
493
34c16eec
ZX
494 /* emulate context */
495
496 struct x86_emulate_ctxt emulate_ctxt;
7ae441ea
GN
497 bool emulate_regs_need_sync_to_vcpu;
498 bool emulate_regs_need_sync_from_vcpu;
716d51ab 499 int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
18068523
GOC
500
501 gpa_t time;
50d0a0f9 502 struct pvclock_vcpu_time_info hv_clock;
e48672fa 503 unsigned int hw_tsc_khz;
0b79459b
AH
504 struct gfn_to_hva_cache pv_time;
505 bool pv_time_enabled;
51d59c6b
MT
506 /* set guest stopped flag in pvclock flags field */
507 bool pvclock_set_guest_stopped_request;
c9aaa895
GC
508
509 struct {
510 u64 msr_val;
511 u64 last_steal;
512 u64 accum_steal;
513 struct gfn_to_hva_cache stime;
514 struct kvm_steal_time steal;
515 } st;
516
1d5f066e 517 u64 last_guest_tsc;
6f526ec5 518 u64 last_host_tsc;
0dd6a6ed 519 u64 tsc_offset_adjustment;
e26101b1
ZA
520 u64 this_tsc_nsec;
521 u64 this_tsc_write;
0d3da0d2 522 u64 this_tsc_generation;
c285545f 523 bool tsc_catchup;
cc578287
ZA
524 bool tsc_always_catchup;
525 s8 virtual_tsc_shift;
526 u32 virtual_tsc_mult;
527 u32 virtual_tsc_khz;
ba904635 528 s64 ia32_tsc_adjust_msr;
ad721883 529 u64 tsc_scaling_ratio;
3419ffc8 530
7460fb4a
AK
531 atomic_t nmi_queued; /* unprocessed asynchronous NMIs */
532 unsigned nmi_pending; /* NMI queued after currently running handler */
533 bool nmi_injected; /* Trying to inject an NMI this entry */
f077825a 534 bool smi_pending; /* SMI queued after currently running handler */
9ba075a6 535
70109e7d 536 struct kvm_mtrr mtrr_state;
7cb060a9 537 u64 pat;
42dbaa5a 538
360b948d 539 unsigned switch_db_regs;
42dbaa5a
JK
540 unsigned long db[KVM_NR_DB_REGS];
541 unsigned long dr6;
542 unsigned long dr7;
543 unsigned long eff_db[KVM_NR_DB_REGS];
c8639010 544 unsigned long guest_debug_dr7;
890ca9ae
HY
545
546 u64 mcg_cap;
547 u64 mcg_status;
548 u64 mcg_ctl;
549 u64 *mce_banks;
94fe45da 550
bebb106a
XG
551 /* Cache MMIO info */
552 u64 mmio_gva;
553 unsigned access;
554 gfn_t mmio_gfn;
56f17dd3 555 u64 mmio_gen;
bebb106a 556
f5132b01
GN
557 struct kvm_pmu pmu;
558
94fe45da 559 /* used for guest single stepping over the given code position */
94fe45da 560 unsigned long singlestep_rip;
f92653ee 561
e83d5887 562 struct kvm_vcpu_hv hyperv;
f5f48ee1
SY
563
564 cpumask_var_t wbinvd_dirty_mask;
af585b92 565
1cb3f3ae
XG
566 unsigned long last_retry_eip;
567 unsigned long last_retry_addr;
568
af585b92
GN
569 struct {
570 bool halted;
571 gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
344d9588
GN
572 struct gfn_to_hva_cache data;
573 u64 msr_val;
7c90705b 574 u32 id;
6adba527 575 bool send_user_only;
af585b92 576 } apf;
2b036c6b
BO
577
578 /* OSVW MSRs (AMD only) */
579 struct {
580 u64 length;
581 u64 status;
582 } osvw;
ae7a2a3f
MT
583
584 struct {
585 u64 msr_val;
586 struct gfn_to_hva_cache data;
587 } pv_eoi;
93c05d3e
XG
588
589 /*
590 * Indicate whether the access faults on its page table in guest
591 * which is set when fix page fault and used to detect unhandeable
592 * instruction.
593 */
594 bool write_fault_to_shadow_pgtable;
25d92081
YZ
595
596 /* set at EPT violation at this point */
597 unsigned long exit_qualification;
6aef266c
SV
598
599 /* pv related host specific info */
600 struct {
601 bool pv_unhalted;
602 } pv;
7543a635
SR
603
604 int pending_ioapic_eoi;
1c1a9ce9 605 int pending_external_vector;
34c16eec
ZX
606};
607
db3fe4eb 608struct kvm_lpage_info {
db3fe4eb
TY
609 int write_count;
610};
611
612struct kvm_arch_memory_slot {
018aabb5 613 struct kvm_rmap_head *rmap[KVM_NR_PAGE_SIZES];
db3fe4eb
TY
614 struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
615};
616
3548a259
RK
617/*
618 * We use as the mode the number of bits allocated in the LDR for the
619 * logical processor ID. It happens that these are all powers of two.
620 * This makes it is very easy to detect cases where the APICs are
621 * configured for multiple modes; in that case, we cannot use the map and
622 * hence cannot use kvm_irq_delivery_to_apic_fast either.
623 */
624#define KVM_APIC_MODE_XAPIC_CLUSTER 4
625#define KVM_APIC_MODE_XAPIC_FLAT 8
626#define KVM_APIC_MODE_X2APIC 16
627
1e08ec4a
GN
628struct kvm_apic_map {
629 struct rcu_head rcu;
3548a259 630 u8 mode;
1e08ec4a
GN
631 struct kvm_lapic *phys_map[256];
632 /* first index is cluster id second is cpu id in a cluster */
633 struct kvm_lapic *logical_map[16][16];
634};
635
e83d5887
AS
636/* Hyper-V emulation context */
637struct kvm_hv {
638 u64 hv_guest_os_id;
639 u64 hv_hypercall;
640 u64 hv_tsc_page;
e7d9513b
AS
641
642 /* Hyper-v based guest crash (NT kernel bugcheck) parameters */
643 u64 hv_crash_param[HV_X64_MSR_CRASH_PARAMS];
644 u64 hv_crash_ctl;
e83d5887
AS
645};
646
fef9cce0 647struct kvm_arch {
49d5ca26 648 unsigned int n_used_mmu_pages;
f05e70ac 649 unsigned int n_requested_mmu_pages;
39de71ec 650 unsigned int n_max_mmu_pages;
332b207d 651 unsigned int indirect_shadow_pages;
5304b8d3 652 unsigned long mmu_valid_gen;
f05e70ac
ZX
653 struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
654 /*
655 * Hash table of struct kvm_mmu_page.
656 */
657 struct list_head active_mmu_pages;
365c8868
XG
658 struct list_head zapped_obsolete_pages;
659
4d5c5d0f 660 struct list_head assigned_dev_head;
19de40a8 661 struct iommu_domain *iommu_domain;
d96eb2c6 662 bool iommu_noncoherent;
e0f0bbc5
AW
663#define __KVM_HAVE_ARCH_NONCOHERENT_DMA
664 atomic_t noncoherent_dma_count;
5544eb9b
PB
665#define __KVM_HAVE_ARCH_ASSIGNED_DEVICE
666 atomic_t assigned_device_count;
d7deeeb0
ZX
667 struct kvm_pic *vpic;
668 struct kvm_ioapic *vioapic;
7837699f 669 struct kvm_pit *vpit;
42720138 670 atomic_t vapics_in_nmi_mode;
1e08ec4a
GN
671 struct mutex apic_map_lock;
672 struct kvm_apic_map *apic_map;
bfc6d222 673
bfc6d222 674 unsigned int tss_addr;
c24ae0dc 675 bool apic_access_page_done;
18068523
GOC
676
677 gpa_t wall_clock;
b7ebfb05 678
b7ebfb05 679 bool ept_identity_pagetable_done;
b927a3ce 680 gpa_t ept_identity_map_addr;
5550af4d
SY
681
682 unsigned long irq_sources_bitmap;
afbcf7ab 683 s64 kvmclock_offset;
038f8c11 684 raw_spinlock_t tsc_write_lock;
f38e098f 685 u64 last_tsc_nsec;
f38e098f 686 u64 last_tsc_write;
5d3cb0f6 687 u32 last_tsc_khz;
e26101b1
ZA
688 u64 cur_tsc_nsec;
689 u64 cur_tsc_write;
690 u64 cur_tsc_offset;
0d3da0d2 691 u64 cur_tsc_generation;
b48aa97e 692 int nr_vcpus_matched_tsc;
ffde22ac 693
d828199e
MT
694 spinlock_t pvclock_gtod_sync_lock;
695 bool use_master_clock;
696 u64 master_kernel_ns;
697 cycle_t master_cycle_now;
7e44e449 698 struct delayed_work kvmclock_update_work;
332967a3 699 struct delayed_work kvmclock_sync_work;
d828199e 700
ffde22ac 701 struct kvm_xen_hvm_config xen_hvm_config;
55cd8e5a 702
6ef768fa
PB
703 /* reads protected by irq_srcu, writes by irq_lock */
704 struct hlist_head mask_notifier_list;
705
e83d5887 706 struct kvm_hv hyperv;
b034cf01
XG
707
708 #ifdef CONFIG_KVM_MMU_AUDIT
709 int audit_point;
710 #endif
54750f2c
MT
711
712 bool boot_vcpu_runs_old_kvmclock;
d71ba788 713 u32 bsp_vcpu_id;
90de4a18
NA
714
715 u64 disabled_quirks;
49df6397
SR
716
717 bool irqchip_split;
b053b2ae 718 u8 nr_reserved_ioapic_pins;
d69fb81f
ZX
719};
720
0711456c
ZX
721struct kvm_vm_stat {
722 u32 mmu_shadow_zapped;
723 u32 mmu_pte_write;
724 u32 mmu_pte_updated;
725 u32 mmu_pde_zapped;
726 u32 mmu_flooded;
727 u32 mmu_recycled;
dfc5aa00 728 u32 mmu_cache_miss;
4731d4c7 729 u32 mmu_unsync;
0711456c 730 u32 remote_tlb_flush;
05da4558 731 u32 lpages;
0711456c
ZX
732};
733
77b4c255
ZX
734struct kvm_vcpu_stat {
735 u32 pf_fixed;
736 u32 pf_guest;
737 u32 tlb_flush;
738 u32 invlpg;
739
740 u32 exits;
741 u32 io_exits;
742 u32 mmio_exits;
743 u32 signal_exits;
744 u32 irq_window_exits;
f08864b4 745 u32 nmi_window_exits;
77b4c255 746 u32 halt_exits;
f7819512 747 u32 halt_successful_poll;
62bea5bf 748 u32 halt_attempted_poll;
77b4c255
ZX
749 u32 halt_wakeup;
750 u32 request_irq_exits;
751 u32 irq_exits;
752 u32 host_state_reload;
753 u32 efer_reload;
754 u32 fpu_reload;
755 u32 insn_emulation;
756 u32 insn_emulation_fail;
f11c3a8d 757 u32 hypercalls;
fa89a817 758 u32 irq_injections;
c4abb7c9 759 u32 nmi_injections;
77b4c255 760};
ad312c7c 761
8a76d7f2
JR
762struct x86_instruction_info;
763
8fe8ab46
WA
764struct msr_data {
765 bool host_initiated;
766 u32 index;
767 u64 data;
768};
769
cb5281a5
PB
770struct kvm_lapic_irq {
771 u32 vector;
b7cb2231
PB
772 u16 delivery_mode;
773 u16 dest_mode;
774 bool level;
775 u16 trig_mode;
cb5281a5
PB
776 u32 shorthand;
777 u32 dest_id;
93bbf0b8 778 bool msi_redir_hint;
cb5281a5
PB
779};
780
ea4a5ff8
ZX
781struct kvm_x86_ops {
782 int (*cpu_has_kvm_support)(void); /* __init */
783 int (*disabled_by_bios)(void); /* __init */
13a34e06
RK
784 int (*hardware_enable)(void);
785 void (*hardware_disable)(void);
ea4a5ff8
ZX
786 void (*check_processor_compatibility)(void *rtn);
787 int (*hardware_setup)(void); /* __init */
788 void (*hardware_unsetup)(void); /* __exit */
774ead3a 789 bool (*cpu_has_accelerated_tpr)(void);
6d396b55 790 bool (*cpu_has_high_real_mode_segbase)(void);
0e851880 791 void (*cpuid_update)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
792
793 /* Create, but do not attach this VCPU */
794 struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
795 void (*vcpu_free)(struct kvm_vcpu *vcpu);
d28bc9dd 796 void (*vcpu_reset)(struct kvm_vcpu *vcpu, bool init_event);
ea4a5ff8
ZX
797
798 void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
799 void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
800 void (*vcpu_put)(struct kvm_vcpu *vcpu);
ea4a5ff8 801
a96036b8 802 void (*update_bp_intercept)(struct kvm_vcpu *vcpu);
609e36d3 803 int (*get_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
8fe8ab46 804 int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
ea4a5ff8
ZX
805 u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
806 void (*get_segment)(struct kvm_vcpu *vcpu,
807 struct kvm_segment *var, int seg);
2e4d2653 808 int (*get_cpl)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
809 void (*set_segment)(struct kvm_vcpu *vcpu,
810 struct kvm_segment *var, int seg);
811 void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
e8467fda 812 void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
aff48baa 813 void (*decache_cr3)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
814 void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
815 void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
816 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
5e1746d6 817 int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
ea4a5ff8 818 void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
89a27f4d
GN
819 void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
820 void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
821 void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
822 void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
73aaf249
JK
823 u64 (*get_dr6)(struct kvm_vcpu *vcpu);
824 void (*set_dr6)(struct kvm_vcpu *vcpu, unsigned long value);
c77fb5fe 825 void (*sync_dirty_debug_regs)(struct kvm_vcpu *vcpu);
020df079 826 void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
5fdbf976 827 void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
ea4a5ff8
ZX
828 unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
829 void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
0fdd74f7 830 void (*fpu_activate)(struct kvm_vcpu *vcpu);
02daab21 831 void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
832
833 void (*tlb_flush)(struct kvm_vcpu *vcpu);
ea4a5ff8 834
851ba692
AK
835 void (*run)(struct kvm_vcpu *vcpu);
836 int (*handle_exit)(struct kvm_vcpu *vcpu);
ea4a5ff8 837 void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
2809f5d2 838 void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
37ccdcbe 839 u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
840 void (*patch_hypercall)(struct kvm_vcpu *vcpu,
841 unsigned char *hypercall_addr);
66fd3f7f 842 void (*set_irq)(struct kvm_vcpu *vcpu);
95ba8273 843 void (*set_nmi)(struct kvm_vcpu *vcpu);
298101da 844 void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
ce7ddec4
JR
845 bool has_error_code, u32 error_code,
846 bool reinject);
b463a6f7 847 void (*cancel_injection)(struct kvm_vcpu *vcpu);
78646121 848 int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
95ba8273 849 int (*nmi_allowed)(struct kvm_vcpu *vcpu);
3cfc3092
JK
850 bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
851 void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
c9a7953f
JK
852 void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
853 void (*enable_irq_window)(struct kvm_vcpu *vcpu);
95ba8273 854 void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
d62caabb
AS
855 bool (*get_enable_apicv)(void);
856 void (*refresh_apicv_exec_ctrl)(struct kvm_vcpu *vcpu);
c7c9c56c
YZ
857 void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
858 void (*hwapic_isr_update)(struct kvm *kvm, int isr);
6308630b 859 void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap);
8d14695f 860 void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu, bool set);
4256f43f 861 void (*set_apic_access_page_addr)(struct kvm_vcpu *vcpu, hpa_t hpa);
a20ed54d
YZ
862 void (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
863 void (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
ea4a5ff8 864 int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
67253af5 865 int (*get_tdp_level)(void);
4b12f0de 866 u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
17cc3935 867 int (*get_lpage_level)(void);
4e47c7a6 868 bool (*rdtscp_supported)(void);
ad756a16 869 bool (*invpcid_supported)(void);
58ea6767 870 void (*adjust_tsc_offset_guest)(struct kvm_vcpu *vcpu, s64 adjustment);
344f414f 871
1c97f0a0
JR
872 void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
873
d4330ef2
JR
874 void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
875
f5f48ee1
SY
876 bool (*has_wbinvd_exit)(void);
877
ba904635 878 u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu);
99e3e30a
ZA
879 void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
880
886b470c 881 u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu, u64 host_tsc);
857e4099 882
586f9607 883 void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
8a76d7f2
JR
884
885 int (*check_intercept)(struct kvm_vcpu *vcpu,
886 struct x86_instruction_info *info,
887 enum x86_intercept_stage stage);
a547c6db 888 void (*handle_external_intr)(struct kvm_vcpu *vcpu);
da8999d3 889 bool (*mpx_supported)(void);
55412b2e 890 bool (*xsaves_supported)(void);
b6b8a145
JK
891
892 int (*check_nested_events)(struct kvm_vcpu *vcpu, bool external_intr);
ae97a3b8
RK
893
894 void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
88178fd4
KH
895
896 /*
897 * Arch-specific dirty logging hooks. These hooks are only supposed to
898 * be valid if the specific arch has hardware-accelerated dirty logging
899 * mechanism. Currently only for PML on VMX.
900 *
901 * - slot_enable_log_dirty:
902 * called when enabling log dirty mode for the slot.
903 * - slot_disable_log_dirty:
904 * called when disabling log dirty mode for the slot.
905 * also called when slot is created with log dirty disabled.
906 * - flush_log_dirty:
907 * called before reporting dirty_bitmap to userspace.
908 * - enable_log_dirty_pt_masked:
909 * called when reenabling log dirty for the GFNs in the mask after
910 * corresponding bits are cleared in slot->dirty_bitmap.
911 */
912 void (*slot_enable_log_dirty)(struct kvm *kvm,
913 struct kvm_memory_slot *slot);
914 void (*slot_disable_log_dirty)(struct kvm *kvm,
915 struct kvm_memory_slot *slot);
916 void (*flush_log_dirty)(struct kvm *kvm);
917 void (*enable_log_dirty_pt_masked)(struct kvm *kvm,
918 struct kvm_memory_slot *slot,
919 gfn_t offset, unsigned long mask);
25462f7f
WH
920 /* pmu operations of sub-arch */
921 const struct kvm_pmu_ops *pmu_ops;
efc64404 922
bf9f6ac8
FW
923 /*
924 * Architecture specific hooks for vCPU blocking due to
925 * HLT instruction.
926 * Returns for .pre_block():
927 * - 0 means continue to block the vCPU.
928 * - 1 means we cannot block the vCPU since some event
929 * happens during this period, such as, 'ON' bit in
930 * posted-interrupts descriptor is set.
931 */
932 int (*pre_block)(struct kvm_vcpu *vcpu);
933 void (*post_block)(struct kvm_vcpu *vcpu);
efc64404
FW
934 int (*update_pi_irte)(struct kvm *kvm, unsigned int host_irq,
935 uint32_t guest_irq, bool set);
ea4a5ff8
ZX
936};
937
af585b92 938struct kvm_arch_async_pf {
7c90705b 939 u32 token;
af585b92 940 gfn_t gfn;
fb67e14f 941 unsigned long cr3;
c4806acd 942 bool direct_map;
af585b92
GN
943};
944
97896d04
ZX
945extern struct kvm_x86_ops *kvm_x86_ops;
946
54f1585a
ZX
947int kvm_mmu_module_init(void);
948void kvm_mmu_module_exit(void);
949
950void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
951int kvm_mmu_create(struct kvm_vcpu *vcpu);
8a3c1a33 952void kvm_mmu_setup(struct kvm_vcpu *vcpu);
7b52345e 953void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
4b12f0de 954 u64 dirty_mask, u64 nx_mask, u64 x_mask);
54f1585a 955
8a3c1a33 956void kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
1c91cad4
KH
957void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
958 struct kvm_memory_slot *memslot);
3ea3b7fa 959void kvm_mmu_zap_collapsible_sptes(struct kvm *kvm,
f36f3f28 960 const struct kvm_memory_slot *memslot);
f4b4b180
KH
961void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
962 struct kvm_memory_slot *memslot);
963void kvm_mmu_slot_largepage_remove_write_access(struct kvm *kvm,
964 struct kvm_memory_slot *memslot);
965void kvm_mmu_slot_set_dirty(struct kvm *kvm,
966 struct kvm_memory_slot *memslot);
967void kvm_mmu_clear_dirty_pt_masked(struct kvm *kvm,
968 struct kvm_memory_slot *slot,
969 gfn_t gfn_offset, unsigned long mask);
54f1585a 970void kvm_mmu_zap_all(struct kvm *kvm);
54bf36aa 971void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm, struct kvm_memslots *slots);
3ad82a7e 972unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
54f1585a
ZX
973void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
974
ff03a073 975int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
cc4b6871 976
3200f405 977int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
9f811285 978 const void *val, int bytes);
2f333bcb 979
6ef768fa
PB
980struct kvm_irq_mask_notifier {
981 void (*func)(struct kvm_irq_mask_notifier *kimn, bool masked);
982 int irq;
983 struct hlist_node link;
984};
985
986void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
987 struct kvm_irq_mask_notifier *kimn);
988void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
989 struct kvm_irq_mask_notifier *kimn);
990void kvm_fire_mask_notifiers(struct kvm *kvm, unsigned irqchip, unsigned pin,
991 bool mask);
992
2f333bcb 993extern bool tdp_enabled;
9f811285 994
a3e06bbe
LJ
995u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
996
92a1f12d
JR
997/* control of guest tsc rate supported? */
998extern bool kvm_has_tsc_control;
92a1f12d
JR
999/* maximum supported tsc_khz for guests */
1000extern u32 kvm_max_guest_tsc_khz;
bc9b961b
HZ
1001/* number of bits of the fractional part of the TSC scaling ratio */
1002extern u8 kvm_tsc_scaling_ratio_frac_bits;
1003/* maximum allowed value of TSC scaling ratio */
1004extern u64 kvm_max_tsc_scaling_ratio;
92a1f12d 1005
54f1585a 1006enum emulation_result {
ac0a48c3
PB
1007 EMULATE_DONE, /* no further processing */
1008 EMULATE_USER_EXIT, /* kvm_run ready for userspace exit */
54f1585a
ZX
1009 EMULATE_FAIL, /* can't emulate this instruction */
1010};
1011
571008da
SY
1012#define EMULTYPE_NO_DECODE (1 << 0)
1013#define EMULTYPE_TRAP_UD (1 << 1)
ba8afb6b 1014#define EMULTYPE_SKIP (1 << 2)
1cb3f3ae 1015#define EMULTYPE_RETRY (1 << 3)
991eebf9 1016#define EMULTYPE_NO_REEXECUTE (1 << 4)
dc25e89e
AP
1017int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
1018 int emulation_type, void *insn, int insn_len);
51d8b661
AP
1019
1020static inline int emulate_instruction(struct kvm_vcpu *vcpu,
1021 int emulation_type)
1022{
dc25e89e 1023 return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
51d8b661
AP
1024}
1025
f2b4b7dd 1026void kvm_enable_efer_bits(u64);
384bb783 1027bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
609e36d3 1028int kvm_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
8fe8ab46 1029int kvm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
54f1585a
ZX
1030
1031struct x86_emulate_ctxt;
1032
cf8f70bf 1033int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
54f1585a
ZX
1034void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
1035int kvm_emulate_halt(struct kvm_vcpu *vcpu);
5cb56059 1036int kvm_vcpu_halt(struct kvm_vcpu *vcpu);
f5f48ee1 1037int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
54f1585a 1038
3e6e0aab 1039void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
c697518a 1040int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
2b4a273b 1041void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
3e6e0aab 1042
7f3d35fd
KW
1043int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
1044 int reason, bool has_error_code, u32 error_code);
37817f29 1045
49a9b07e 1046int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
2390218b 1047int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
a83b29c6 1048int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
eea1cff9 1049int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
020df079
GN
1050int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
1051int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
2d3ad1f4
AK
1052unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
1053void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
54f1585a 1054void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
2acf923e 1055int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
54f1585a 1056
609e36d3 1057int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
8fe8ab46 1058int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
54f1585a 1059
91586a3b
JK
1060unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
1061void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
022cd0e8 1062bool kvm_rdpmc(struct kvm_vcpu *vcpu);
91586a3b 1063
298101da
AK
1064void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1065void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
ce7ddec4
JR
1066void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1067void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
6389ee94 1068void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
ec92fe44
JR
1069int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
1070 gfn_t gfn, void *data, int offset, int len,
1071 u32 access);
0a79b009 1072bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
16f8a6f9 1073bool kvm_require_dr(struct kvm_vcpu *vcpu, int dr);
298101da 1074
1a577b72
MT
1075static inline int __kvm_irq_line_state(unsigned long *irq_state,
1076 int irq_source_id, int level)
1077{
1078 /* Logical OR for level trig interrupt */
1079 if (level)
1080 __set_bit(irq_source_id, irq_state);
1081 else
1082 __clear_bit(irq_source_id, irq_state);
1083
1084 return !!(*irq_state);
1085}
1086
1087int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
1088void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
3de42dc0 1089
3419ffc8
SY
1090void kvm_inject_nmi(struct kvm_vcpu *vcpu);
1091
54f1585a 1092void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
f57f2ef5 1093 const u8 *new, int bytes);
1cb3f3ae 1094int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
54f1585a
ZX
1095int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
1096void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
1097int kvm_mmu_load(struct kvm_vcpu *vcpu);
1098void kvm_mmu_unload(struct kvm_vcpu *vcpu);
0ba73cda 1099void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
54987b7a
PB
1100gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1101 struct x86_exception *exception);
ab9ae313
AK
1102gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
1103 struct x86_exception *exception);
1104gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
1105 struct x86_exception *exception);
1106gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
1107 struct x86_exception *exception);
1108gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
1109 struct x86_exception *exception);
54f1585a 1110
d62caabb
AS
1111void kvm_vcpu_deactivate_apicv(struct kvm_vcpu *vcpu);
1112
54f1585a
ZX
1113int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
1114
dc25e89e
AP
1115int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
1116 void *insn, int insn_len);
a7052897 1117void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
d8d173da 1118void kvm_mmu_new_cr3(struct kvm_vcpu *vcpu);
34c16eec 1119
18552672 1120void kvm_enable_tdp(void);
5f4cb662 1121void kvm_disable_tdp(void);
18552672 1122
54987b7a
PB
1123static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1124 struct x86_exception *exception)
e459e322
XG
1125{
1126 return gpa;
1127}
1128
ec6d273d
ZX
1129static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
1130{
1131 struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
1132
1133 return (struct kvm_mmu_page *)page_private(page);
1134}
1135
d6e88aec 1136static inline u16 kvm_read_ldt(void)
ec6d273d
ZX
1137{
1138 u16 ldt;
1139 asm("sldt %0" : "=g"(ldt));
1140 return ldt;
1141}
1142
d6e88aec 1143static inline void kvm_load_ldt(u16 sel)
ec6d273d
ZX
1144{
1145 asm("lldt %0" : : "rm"(sel));
1146}
ec6d273d 1147
ec6d273d
ZX
1148#ifdef CONFIG_X86_64
1149static inline unsigned long read_msr(unsigned long msr)
1150{
1151 u64 value;
1152
1153 rdmsrl(msr, value);
1154 return value;
1155}
1156#endif
1157
ec6d273d
ZX
1158static inline u32 get_rdx_init_val(void)
1159{
1160 return 0x600; /* P6 family */
1161}
1162
c1a5d4f9
AK
1163static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
1164{
1165 kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
1166}
1167
854e8bb1
NA
1168static inline u64 get_canonical(u64 la)
1169{
1170 return ((int64_t)la << 16) >> 16;
1171}
1172
1173static inline bool is_noncanonical_address(u64 la)
1174{
1175#ifdef CONFIG_X86_64
1176 return get_canonical(la) != la;
1177#else
1178 return false;
1179#endif
1180}
1181
ec6d273d
ZX
1182#define TSS_IOPB_BASE_OFFSET 0x66
1183#define TSS_BASE_SIZE 0x68
1184#define TSS_IOPB_SIZE (65536 / 8)
1185#define TSS_REDIRECTION_SIZE (256 / 8)
7d76b4d3
JP
1186#define RMODE_TSS_SIZE \
1187 (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
53e0aa7b 1188
37817f29
IE
1189enum {
1190 TASK_SWITCH_CALL = 0,
1191 TASK_SWITCH_IRET = 1,
1192 TASK_SWITCH_JMP = 2,
1193 TASK_SWITCH_GATE = 3,
1194};
1195
1371d904 1196#define HF_GIF_MASK (1 << 0)
3d6368ef
AG
1197#define HF_HIF_MASK (1 << 1)
1198#define HF_VINTR_MASK (1 << 2)
95ba8273 1199#define HF_NMI_MASK (1 << 3)
44c11430 1200#define HF_IRET_MASK (1 << 4)
ec9e60b2 1201#define HF_GUEST_MASK (1 << 5) /* VCPU is in guest-mode */
f077825a
PB
1202#define HF_SMM_MASK (1 << 6)
1203#define HF_SMM_INSIDE_NMI_MASK (1 << 7)
1371d904 1204
699023e2
PB
1205#define __KVM_VCPU_MULTIPLE_ADDRESS_SPACE
1206#define KVM_ADDRESS_SPACE_NUM 2
1207
1208#define kvm_arch_vcpu_memslots_id(vcpu) ((vcpu)->arch.hflags & HF_SMM_MASK ? 1 : 0)
1209#define kvm_memslots_for_spte_role(kvm, role) __kvm_memslots(kvm, (role).smm)
1371d904 1210
4ecac3fd
AK
1211/*
1212 * Hardware virtualization extension instructions may fault if a
1213 * reboot turns off virtualization while processes are running.
1214 * Trap the fault and ignore the instruction if that happens.
1215 */
b7c4145b 1216asmlinkage void kvm_spurious_fault(void);
4ecac3fd 1217
5e520e62 1218#define ____kvm_handle_fault_on_reboot(insn, cleanup_insn) \
4ecac3fd 1219 "666: " insn "\n\t" \
b7c4145b 1220 "668: \n\t" \
18b13e54 1221 ".pushsection .fixup, \"ax\" \n" \
4ecac3fd 1222 "667: \n\t" \
5e520e62 1223 cleanup_insn "\n\t" \
b7c4145b
AK
1224 "cmpb $0, kvm_rebooting \n\t" \
1225 "jne 668b \n\t" \
8ceed347 1226 __ASM_SIZE(push) " $666b \n\t" \
b7c4145b 1227 "call kvm_spurious_fault \n\t" \
4ecac3fd 1228 ".popsection \n\t" \
3ee89722 1229 _ASM_EXTABLE(666b, 667b)
4ecac3fd 1230
5e520e62
AK
1231#define __kvm_handle_fault_on_reboot(insn) \
1232 ____kvm_handle_fault_on_reboot(insn, "")
1233
e930bffe
AA
1234#define KVM_ARCH_WANT_MMU_NOTIFIER
1235int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
b3ae2096 1236int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
57128468 1237int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
8ee53820 1238int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
3da0dd43 1239void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
c7c9c56c 1240int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
a1b37100
GN
1241int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1242int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
0b71785d 1243int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
d28bc9dd 1244void kvm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event);
4256f43f 1245void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu);
fe71557a
TC
1246void kvm_arch_mmu_notifier_invalidate_page(struct kvm *kvm,
1247 unsigned long address);
e930bffe 1248
18863bdd 1249void kvm_define_shared_msr(unsigned index, u32 msr);
8b3c3104 1250int kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
18863bdd 1251
35181e86 1252u64 kvm_scale_tsc(struct kvm_vcpu *vcpu, u64 tsc);
4ba76538 1253u64 kvm_read_l1_tsc(struct kvm_vcpu *vcpu, u64 host_tsc);
35181e86 1254
82b32774 1255unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
f92653ee
JK
1256bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1257
af585b92
GN
1258void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1259 struct kvm_async_pf *work);
1260void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1261 struct kvm_async_pf *work);
56028d08
GN
1262void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1263 struct kvm_async_pf *work);
7c90705b 1264bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
af585b92
GN
1265extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1266
db8fcefa
AP
1267void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1268
f5132b01
GN
1269int kvm_is_in_guest(void);
1270
1d8007bd
PB
1271int __x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa, u32 size);
1272int x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa, u32 size);
d71ba788
PB
1273bool kvm_vcpu_is_reset_bsp(struct kvm_vcpu *vcpu);
1274bool kvm_vcpu_is_bsp(struct kvm_vcpu *vcpu);
f5132b01 1275
8feb4a04
FW
1276bool kvm_intr_is_single_vcpu(struct kvm *kvm, struct kvm_lapic_irq *irq,
1277 struct kvm_vcpu **dest_vcpu);
1278
d84f1e07
FW
1279void kvm_set_msi_irq(struct kvm_kernel_irq_routing_entry *e,
1280 struct kvm_lapic_irq *irq);
197a4f4b 1281
3217f7c2
CD
1282static inline void kvm_arch_vcpu_blocking(struct kvm_vcpu *vcpu) {}
1283static inline void kvm_arch_vcpu_unblocking(struct kvm_vcpu *vcpu) {}
1284
1965aae3 1285#endif /* _ASM_X86_KVM_HOST_H */