]> git.proxmox.com Git - mirror_ubuntu-disco-kernel.git/blame - arch/x86/include/asm/kvm_host.h
KVM: Add some helper functions for Posted-Interrupts
[mirror_ubuntu-disco-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>
34c16eec 27
50d0a0f9 28#include <asm/pvclock-abi.h>
e01a1b57 29#include <asm/desc.h>
0bed3b56 30#include <asm/mtrr.h>
9962d032 31#include <asm/msr-index.h>
3ee89722 32#include <asm/asm.h>
e01a1b57 33
cbf64358 34#define KVM_MAX_VCPUS 255
a59cb29e 35#define KVM_SOFT_MAX_VCPUS 160
1d4e7e3c 36#define KVM_USER_MEM_SLOTS 509
0743247f
AW
37/* memory slots that are not exposed to userspace */
38#define KVM_PRIVATE_MEM_SLOTS 3
bbacc0c1 39#define KVM_MEM_SLOTS_NUM (KVM_USER_MEM_SLOTS + KVM_PRIVATE_MEM_SLOTS)
93a5cef0 40
69a9f69b 41#define KVM_PIO_PAGE_OFFSET 1
542472b5 42#define KVM_COALESCED_MMIO_PAGE_OFFSET 2
920552b2 43#define KVM_HALT_POLL_NS_DEFAULT 500000
69a9f69b 44
8175e5b7
AG
45#define KVM_IRQCHIP_NUM_PINS KVM_IOAPIC_NUM_PINS
46
cfec82cb
JR
47#define CR0_RESERVED_BITS \
48 (~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
49 | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
50 | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
51
346874c9 52#define CR3_L_MODE_RESERVED_BITS 0xFFFFFF0000000000ULL
cfaa790a 53#define CR3_PCID_INVD BIT_64(63)
cfec82cb
JR
54#define CR4_RESERVED_BITS \
55 (~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
56 | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE \
ad756a16 57 | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR | X86_CR4_PCIDE \
afcbf13f 58 | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_FSGSBASE \
56d6efc2 59 | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE | X86_CR4_SMAP))
cfec82cb
JR
60
61#define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
62
63
cd6e8f87 64
cd6e8f87 65#define INVALID_PAGE (~(hpa_t)0)
dd180b3e
XG
66#define VALID_PAGE(x) ((x) != INVALID_PAGE)
67
cd6e8f87
ZX
68#define UNMAPPED_GVA (~(gpa_t)0)
69
ec04b260 70/* KVM Hugepage definitions for x86 */
04326caa 71#define KVM_NR_PAGE_SIZES 3
82855413
JR
72#define KVM_HPAGE_GFN_SHIFT(x) (((x) - 1) * 9)
73#define KVM_HPAGE_SHIFT(x) (PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
ec04b260
JR
74#define KVM_HPAGE_SIZE(x) (1UL << KVM_HPAGE_SHIFT(x))
75#define KVM_HPAGE_MASK(x) (~(KVM_HPAGE_SIZE(x) - 1))
76#define KVM_PAGES_PER_HPAGE(x) (KVM_HPAGE_SIZE(x) / PAGE_SIZE)
05da4558 77
6d9d41e5
CD
78static inline gfn_t gfn_to_index(gfn_t gfn, gfn_t base_gfn, int level)
79{
80 /* KVM_HPAGE_GFN_SHIFT(PT_PAGE_TABLE_LEVEL) must be 0. */
81 return (gfn >> KVM_HPAGE_GFN_SHIFT(level)) -
82 (base_gfn >> KVM_HPAGE_GFN_SHIFT(level));
83}
84
d657a98e
ZX
85#define KVM_PERMILLE_MMU_PAGES 20
86#define KVM_MIN_ALLOC_MMU_PAGES 64
1ae0a13d
DE
87#define KVM_MMU_HASH_SHIFT 10
88#define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
d657a98e
ZX
89#define KVM_MIN_FREE_MMU_PAGES 5
90#define KVM_REFILL_PAGES 25
73c1160c 91#define KVM_MAX_CPUID_ENTRIES 80
0bed3b56 92#define KVM_NR_FIXED_MTRR_REGION 88
0d234daf 93#define KVM_NR_VAR_MTRR 8
d657a98e 94
af585b92
GN
95#define ASYNC_PF_PER_VCPU 64
96
5fdbf976 97enum kvm_reg {
2b3ccfa0
ZX
98 VCPU_REGS_RAX = 0,
99 VCPU_REGS_RCX = 1,
100 VCPU_REGS_RDX = 2,
101 VCPU_REGS_RBX = 3,
102 VCPU_REGS_RSP = 4,
103 VCPU_REGS_RBP = 5,
104 VCPU_REGS_RSI = 6,
105 VCPU_REGS_RDI = 7,
106#ifdef CONFIG_X86_64
107 VCPU_REGS_R8 = 8,
108 VCPU_REGS_R9 = 9,
109 VCPU_REGS_R10 = 10,
110 VCPU_REGS_R11 = 11,
111 VCPU_REGS_R12 = 12,
112 VCPU_REGS_R13 = 13,
113 VCPU_REGS_R14 = 14,
114 VCPU_REGS_R15 = 15,
115#endif
5fdbf976 116 VCPU_REGS_RIP,
2b3ccfa0
ZX
117 NR_VCPU_REGS
118};
119
6de4f3ad
AK
120enum kvm_reg_ex {
121 VCPU_EXREG_PDPTR = NR_VCPU_REGS,
aff48baa 122 VCPU_EXREG_CR3,
6de12732 123 VCPU_EXREG_RFLAGS,
2fb92db1 124 VCPU_EXREG_SEGMENTS,
6de4f3ad
AK
125};
126
2b3ccfa0 127enum {
81609e3e 128 VCPU_SREG_ES,
2b3ccfa0 129 VCPU_SREG_CS,
81609e3e 130 VCPU_SREG_SS,
2b3ccfa0 131 VCPU_SREG_DS,
2b3ccfa0
ZX
132 VCPU_SREG_FS,
133 VCPU_SREG_GS,
2b3ccfa0
ZX
134 VCPU_SREG_TR,
135 VCPU_SREG_LDTR,
136};
137
56e82318 138#include <asm/kvm_emulate.h>
2b3ccfa0 139
d657a98e
ZX
140#define KVM_NR_MEM_OBJS 40
141
42dbaa5a
JK
142#define KVM_NR_DB_REGS 4
143
144#define DR6_BD (1 << 13)
145#define DR6_BS (1 << 14)
6f43ed01
NA
146#define DR6_RTM (1 << 16)
147#define DR6_FIXED_1 0xfffe0ff0
148#define DR6_INIT 0xffff0ff0
149#define DR6_VOLATILE 0x0001e00f
42dbaa5a
JK
150
151#define DR7_BP_EN_MASK 0x000000ff
152#define DR7_GE (1 << 9)
153#define DR7_GD (1 << 13)
154#define DR7_FIXED_1 0x00000400
6f43ed01 155#define DR7_VOLATILE 0xffff2bff
42dbaa5a 156
c205fb7d
NA
157#define PFERR_PRESENT_BIT 0
158#define PFERR_WRITE_BIT 1
159#define PFERR_USER_BIT 2
160#define PFERR_RSVD_BIT 3
161#define PFERR_FETCH_BIT 4
162
163#define PFERR_PRESENT_MASK (1U << PFERR_PRESENT_BIT)
164#define PFERR_WRITE_MASK (1U << PFERR_WRITE_BIT)
165#define PFERR_USER_MASK (1U << PFERR_USER_BIT)
166#define PFERR_RSVD_MASK (1U << PFERR_RSVD_BIT)
167#define PFERR_FETCH_MASK (1U << PFERR_FETCH_BIT)
168
41383771
GN
169/* apic attention bits */
170#define KVM_APIC_CHECK_VAPIC 0
ae7a2a3f
MT
171/*
172 * The following bit is set with PV-EOI, unset on EOI.
173 * We detect PV-EOI changes by guest by comparing
174 * this bit with PV-EOI in guest memory.
175 * See the implementation in apic_update_pv_eoi.
176 */
177#define KVM_APIC_PV_EOI_PENDING 1
41383771 178
d657a98e
ZX
179/*
180 * We don't want allocation failures within the mmu code, so we preallocate
181 * enough memory for a single page fault in a cache.
182 */
183struct kvm_mmu_memory_cache {
184 int nobjs;
185 void *objects[KVM_NR_MEM_OBJS];
186};
187
d657a98e
ZX
188union kvm_mmu_page_role {
189 unsigned word;
190 struct {
7d76b4d3 191 unsigned level:4;
5b7e0102 192 unsigned cr4_pae:1;
7d76b4d3 193 unsigned quadrant:2;
f6e2c02b 194 unsigned direct:1;
7d76b4d3 195 unsigned access:3;
2e53d63a 196 unsigned invalid:1;
9645bb56 197 unsigned nxe:1;
3dbe1415 198 unsigned cr0_wp:1;
411c588d 199 unsigned smep_andnot_wp:1;
0be0226f 200 unsigned smap_andnot_wp:1;
699023e2
PB
201 unsigned :8;
202
203 /*
204 * This is left at the top of the word so that
205 * kvm_memslots_for_spte_role can extract it with a
206 * simple shift. While there is room, give it a whole
207 * byte so it is also faster to load it from memory.
208 */
209 unsigned smm:8;
d657a98e
ZX
210 };
211};
212
213struct kvm_mmu_page {
214 struct list_head link;
215 struct hlist_node hash_link;
216
217 /*
218 * The following two entries are used to key the shadow page in the
219 * hash table.
220 */
221 gfn_t gfn;
222 union kvm_mmu_page_role role;
223
224 u64 *spt;
225 /* hold the gfn of each spte inside spt */
226 gfn_t *gfns;
4731d4c7 227 bool unsync;
0571d366 228 int root_count; /* Currently serving as active root */
60c8aec6 229 unsigned int unsync_children;
67052b35 230 unsigned long parent_ptes; /* Reverse mapping for parent_pte */
f6f8adee
XG
231
232 /* The page is obsolete if mmu_valid_gen != kvm->arch.mmu_valid_gen. */
5304b8d3 233 unsigned long mmu_valid_gen;
f6f8adee 234
0074ff63 235 DECLARE_BITMAP(unsync_child_bitmap, 512);
c2a2ac2b
XG
236
237#ifdef CONFIG_X86_32
accaefe0
XG
238 /*
239 * Used out of the mmu-lock to avoid reading spte values while an
240 * update is in progress; see the comments in __get_spte_lockless().
241 */
c2a2ac2b
XG
242 int clear_spte_count;
243#endif
244
0cbf8e43 245 /* Number of writes since the last time traversal visited this page. */
a30f47cb 246 int write_flooding_count;
d657a98e
ZX
247};
248
1c08364c
AK
249struct kvm_pio_request {
250 unsigned long count;
1c08364c
AK
251 int in;
252 int port;
253 int size;
1c08364c
AK
254};
255
a0a64f50
XG
256struct rsvd_bits_validate {
257 u64 rsvd_bits_mask[2][4];
258 u64 bad_mt_xwr;
259};
260
d657a98e
ZX
261/*
262 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
263 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
264 * mode.
265 */
266struct kvm_mmu {
f43addd4 267 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
5777ed34 268 unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
e4e517b4 269 u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
78b2c54a
XG
270 int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
271 bool prefault);
6389ee94
AK
272 void (*inject_page_fault)(struct kvm_vcpu *vcpu,
273 struct x86_exception *fault);
1871c602 274 gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
ab9ae313 275 struct x86_exception *exception);
54987b7a
PB
276 gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
277 struct x86_exception *exception);
e8bc217a 278 int (*sync_page)(struct kvm_vcpu *vcpu,
a4a8e6f7 279 struct kvm_mmu_page *sp);
a7052897 280 void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
0f53b5b1 281 void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
7c562522 282 u64 *spte, const void *pte);
d657a98e
ZX
283 hpa_t root_hpa;
284 int root_level;
285 int shadow_root_level;
a770f6f2 286 union kvm_mmu_page_role base_role;
c5a78f2b 287 bool direct_map;
d657a98e 288
97d64b78
AK
289 /*
290 * Bitmap; bit set = permission fault
291 * Byte index: page fault error code [4:1]
292 * Bit index: pte permissions in ACC_* format
293 */
294 u8 permissions[16];
295
d657a98e 296 u64 *pae_root;
81407ca5 297 u64 *lm_root;
c258b62b
XG
298
299 /*
300 * check zero bits on shadow page table entries, these
301 * bits include not only hardware reserved bits but also
302 * the bits spte never used.
303 */
304 struct rsvd_bits_validate shadow_zero_check;
305
a0a64f50 306 struct rsvd_bits_validate guest_rsvd_check;
ff03a073 307
6fd01b71
AK
308 /*
309 * Bitmap: bit set = last pte in walk
310 * index[0:1]: level (zero-based)
311 * index[2]: pte.ps
312 */
313 u8 last_pte_bitmap;
314
2d48a985
JR
315 bool nx;
316
ff03a073 317 u64 pdptrs[4]; /* pae */
d657a98e
ZX
318};
319
f5132b01
GN
320enum pmc_type {
321 KVM_PMC_GP = 0,
322 KVM_PMC_FIXED,
323};
324
325struct kvm_pmc {
326 enum pmc_type type;
327 u8 idx;
328 u64 counter;
329 u64 eventsel;
330 struct perf_event *perf_event;
331 struct kvm_vcpu *vcpu;
332};
333
334struct kvm_pmu {
335 unsigned nr_arch_gp_counters;
336 unsigned nr_arch_fixed_counters;
337 unsigned available_event_types;
338 u64 fixed_ctr_ctrl;
339 u64 global_ctrl;
340 u64 global_status;
341 u64 global_ovf_ctrl;
342 u64 counter_bitmask[2];
343 u64 global_ctrl_mask;
103af0a9 344 u64 reserved_bits;
f5132b01 345 u8 version;
15c7ad51
RR
346 struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
347 struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
f5132b01
GN
348 struct irq_work irq_work;
349 u64 reprogram_pmi;
350};
351
25462f7f
WH
352struct kvm_pmu_ops;
353
360b948d
PB
354enum {
355 KVM_DEBUGREG_BP_ENABLED = 1,
c77fb5fe 356 KVM_DEBUGREG_WONT_EXIT = 2,
ae561ede 357 KVM_DEBUGREG_RELOAD = 4,
360b948d
PB
358};
359
86fd5270
XG
360struct kvm_mtrr_range {
361 u64 base;
362 u64 mask;
19efffa2 363 struct list_head node;
86fd5270
XG
364};
365
70109e7d 366struct kvm_mtrr {
86fd5270 367 struct kvm_mtrr_range var_ranges[KVM_NR_VAR_MTRR];
70109e7d 368 mtrr_type fixed_ranges[KVM_NR_FIXED_MTRR_REGION];
10fac2dc 369 u64 deftype;
19efffa2
XG
370
371 struct list_head head;
70109e7d
XG
372};
373
e83d5887
AS
374/* Hyper-V per vcpu emulation context */
375struct kvm_vcpu_hv {
376 u64 hv_vapic;
9eec50b8 377 s64 runtime_offset;
e83d5887
AS
378};
379
ad312c7c 380struct kvm_vcpu_arch {
5fdbf976
MT
381 /*
382 * rip and regs accesses must go through
383 * kvm_{register,rip}_{read,write} functions.
384 */
385 unsigned long regs[NR_VCPU_REGS];
386 u32 regs_avail;
387 u32 regs_dirty;
34c16eec
ZX
388
389 unsigned long cr0;
e8467fda 390 unsigned long cr0_guest_owned_bits;
34c16eec
ZX
391 unsigned long cr2;
392 unsigned long cr3;
393 unsigned long cr4;
fc78f519 394 unsigned long cr4_guest_owned_bits;
34c16eec 395 unsigned long cr8;
1371d904 396 u32 hflags;
f6801dff 397 u64 efer;
34c16eec
ZX
398 u64 apic_base;
399 struct kvm_lapic *apic; /* kernel irqchip context */
3bb345f3 400 u64 eoi_exit_bitmap[4];
41383771 401 unsigned long apic_attention;
e1035715 402 int32_t apic_arb_prio;
34c16eec 403 int mp_state;
34c16eec 404 u64 ia32_misc_enable_msr;
64d60670 405 u64 smbase;
b209749f 406 bool tpr_access_reporting;
20300099 407 u64 ia32_xss;
34c16eec 408
14dfe855
JR
409 /*
410 * Paging state of the vcpu
411 *
412 * If the vcpu runs in guest mode with two level paging this still saves
413 * the paging mode of the l1 guest. This context is always used to
414 * handle faults.
415 */
34c16eec 416 struct kvm_mmu mmu;
8df25a32 417
6539e738
JR
418 /*
419 * Paging state of an L2 guest (used for nested npt)
420 *
421 * This context will save all necessary information to walk page tables
422 * of the an L2 guest. This context is only initialized for page table
423 * walking and not for faulting since we never handle l2 page faults on
424 * the host.
425 */
426 struct kvm_mmu nested_mmu;
427
14dfe855
JR
428 /*
429 * Pointer to the mmu context currently used for
430 * gva_to_gpa translations.
431 */
432 struct kvm_mmu *walk_mmu;
433
53c07b18 434 struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
34c16eec
ZX
435 struct kvm_mmu_memory_cache mmu_page_cache;
436 struct kvm_mmu_memory_cache mmu_page_header_cache;
437
98918833 438 struct fpu guest_fpu;
c447e76b 439 bool eager_fpu;
2acf923e 440 u64 xcr0;
d7876f1b 441 u64 guest_supported_xcr0;
4344ee98 442 u32 guest_xstate_size;
34c16eec 443
34c16eec
ZX
444 struct kvm_pio_request pio;
445 void *pio_data;
446
66fd3f7f
GN
447 u8 event_exit_inst_len;
448
298101da
AK
449 struct kvm_queued_exception {
450 bool pending;
451 bool has_error_code;
ce7ddec4 452 bool reinject;
298101da
AK
453 u8 nr;
454 u32 error_code;
455 } exception;
456
937a7eae
AK
457 struct kvm_queued_interrupt {
458 bool pending;
66fd3f7f 459 bool soft;
937a7eae
AK
460 u8 nr;
461 } interrupt;
462
34c16eec
ZX
463 int halt_request; /* real mode on Intel only */
464
465 int cpuid_nent;
07716717 466 struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
5a4f55cd
EK
467
468 int maxphyaddr;
469
34c16eec
ZX
470 /* emulate context */
471
472 struct x86_emulate_ctxt emulate_ctxt;
7ae441ea
GN
473 bool emulate_regs_need_sync_to_vcpu;
474 bool emulate_regs_need_sync_from_vcpu;
716d51ab 475 int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
18068523
GOC
476
477 gpa_t time;
50d0a0f9 478 struct pvclock_vcpu_time_info hv_clock;
e48672fa 479 unsigned int hw_tsc_khz;
0b79459b
AH
480 struct gfn_to_hva_cache pv_time;
481 bool pv_time_enabled;
51d59c6b
MT
482 /* set guest stopped flag in pvclock flags field */
483 bool pvclock_set_guest_stopped_request;
c9aaa895
GC
484
485 struct {
486 u64 msr_val;
487 u64 last_steal;
488 u64 accum_steal;
489 struct gfn_to_hva_cache stime;
490 struct kvm_steal_time steal;
491 } st;
492
1d5f066e 493 u64 last_guest_tsc;
6f526ec5 494 u64 last_host_tsc;
0dd6a6ed 495 u64 tsc_offset_adjustment;
e26101b1
ZA
496 u64 this_tsc_nsec;
497 u64 this_tsc_write;
0d3da0d2 498 u64 this_tsc_generation;
c285545f 499 bool tsc_catchup;
cc578287
ZA
500 bool tsc_always_catchup;
501 s8 virtual_tsc_shift;
502 u32 virtual_tsc_mult;
503 u32 virtual_tsc_khz;
ba904635 504 s64 ia32_tsc_adjust_msr;
3419ffc8 505
7460fb4a
AK
506 atomic_t nmi_queued; /* unprocessed asynchronous NMIs */
507 unsigned nmi_pending; /* NMI queued after currently running handler */
508 bool nmi_injected; /* Trying to inject an NMI this entry */
f077825a 509 bool smi_pending; /* SMI queued after currently running handler */
9ba075a6 510
70109e7d 511 struct kvm_mtrr mtrr_state;
7cb060a9 512 u64 pat;
42dbaa5a 513
360b948d 514 unsigned switch_db_regs;
42dbaa5a
JK
515 unsigned long db[KVM_NR_DB_REGS];
516 unsigned long dr6;
517 unsigned long dr7;
518 unsigned long eff_db[KVM_NR_DB_REGS];
c8639010 519 unsigned long guest_debug_dr7;
890ca9ae
HY
520
521 u64 mcg_cap;
522 u64 mcg_status;
523 u64 mcg_ctl;
524 u64 *mce_banks;
94fe45da 525
bebb106a
XG
526 /* Cache MMIO info */
527 u64 mmio_gva;
528 unsigned access;
529 gfn_t mmio_gfn;
56f17dd3 530 u64 mmio_gen;
bebb106a 531
f5132b01
GN
532 struct kvm_pmu pmu;
533
94fe45da 534 /* used for guest single stepping over the given code position */
94fe45da 535 unsigned long singlestep_rip;
f92653ee 536
e83d5887 537 struct kvm_vcpu_hv hyperv;
f5f48ee1
SY
538
539 cpumask_var_t wbinvd_dirty_mask;
af585b92 540
1cb3f3ae
XG
541 unsigned long last_retry_eip;
542 unsigned long last_retry_addr;
543
af585b92
GN
544 struct {
545 bool halted;
546 gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
344d9588
GN
547 struct gfn_to_hva_cache data;
548 u64 msr_val;
7c90705b 549 u32 id;
6adba527 550 bool send_user_only;
af585b92 551 } apf;
2b036c6b
BO
552
553 /* OSVW MSRs (AMD only) */
554 struct {
555 u64 length;
556 u64 status;
557 } osvw;
ae7a2a3f
MT
558
559 struct {
560 u64 msr_val;
561 struct gfn_to_hva_cache data;
562 } pv_eoi;
93c05d3e
XG
563
564 /*
565 * Indicate whether the access faults on its page table in guest
566 * which is set when fix page fault and used to detect unhandeable
567 * instruction.
568 */
569 bool write_fault_to_shadow_pgtable;
25d92081
YZ
570
571 /* set at EPT violation at this point */
572 unsigned long exit_qualification;
6aef266c
SV
573
574 /* pv related host specific info */
575 struct {
576 bool pv_unhalted;
577 } pv;
7543a635
SR
578
579 int pending_ioapic_eoi;
1c1a9ce9 580 int pending_external_vector;
34c16eec
ZX
581};
582
db3fe4eb 583struct kvm_lpage_info {
db3fe4eb
TY
584 int write_count;
585};
586
587struct kvm_arch_memory_slot {
d89cc617 588 unsigned long *rmap[KVM_NR_PAGE_SIZES];
db3fe4eb
TY
589 struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
590};
591
3548a259
RK
592/*
593 * We use as the mode the number of bits allocated in the LDR for the
594 * logical processor ID. It happens that these are all powers of two.
595 * This makes it is very easy to detect cases where the APICs are
596 * configured for multiple modes; in that case, we cannot use the map and
597 * hence cannot use kvm_irq_delivery_to_apic_fast either.
598 */
599#define KVM_APIC_MODE_XAPIC_CLUSTER 4
600#define KVM_APIC_MODE_XAPIC_FLAT 8
601#define KVM_APIC_MODE_X2APIC 16
602
1e08ec4a
GN
603struct kvm_apic_map {
604 struct rcu_head rcu;
3548a259 605 u8 mode;
1e08ec4a
GN
606 struct kvm_lapic *phys_map[256];
607 /* first index is cluster id second is cpu id in a cluster */
608 struct kvm_lapic *logical_map[16][16];
609};
610
e83d5887
AS
611/* Hyper-V emulation context */
612struct kvm_hv {
613 u64 hv_guest_os_id;
614 u64 hv_hypercall;
615 u64 hv_tsc_page;
e7d9513b
AS
616
617 /* Hyper-v based guest crash (NT kernel bugcheck) parameters */
618 u64 hv_crash_param[HV_X64_MSR_CRASH_PARAMS];
619 u64 hv_crash_ctl;
e83d5887
AS
620};
621
fef9cce0 622struct kvm_arch {
49d5ca26 623 unsigned int n_used_mmu_pages;
f05e70ac 624 unsigned int n_requested_mmu_pages;
39de71ec 625 unsigned int n_max_mmu_pages;
332b207d 626 unsigned int indirect_shadow_pages;
5304b8d3 627 unsigned long mmu_valid_gen;
f05e70ac
ZX
628 struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
629 /*
630 * Hash table of struct kvm_mmu_page.
631 */
632 struct list_head active_mmu_pages;
365c8868
XG
633 struct list_head zapped_obsolete_pages;
634
4d5c5d0f 635 struct list_head assigned_dev_head;
19de40a8 636 struct iommu_domain *iommu_domain;
d96eb2c6 637 bool iommu_noncoherent;
e0f0bbc5
AW
638#define __KVM_HAVE_ARCH_NONCOHERENT_DMA
639 atomic_t noncoherent_dma_count;
5544eb9b
PB
640#define __KVM_HAVE_ARCH_ASSIGNED_DEVICE
641 atomic_t assigned_device_count;
d7deeeb0
ZX
642 struct kvm_pic *vpic;
643 struct kvm_ioapic *vioapic;
7837699f 644 struct kvm_pit *vpit;
42720138 645 atomic_t vapics_in_nmi_mode;
1e08ec4a
GN
646 struct mutex apic_map_lock;
647 struct kvm_apic_map *apic_map;
bfc6d222 648
bfc6d222 649 unsigned int tss_addr;
c24ae0dc 650 bool apic_access_page_done;
18068523
GOC
651
652 gpa_t wall_clock;
b7ebfb05 653
b7ebfb05 654 bool ept_identity_pagetable_done;
b927a3ce 655 gpa_t ept_identity_map_addr;
5550af4d
SY
656
657 unsigned long irq_sources_bitmap;
afbcf7ab 658 s64 kvmclock_offset;
038f8c11 659 raw_spinlock_t tsc_write_lock;
f38e098f 660 u64 last_tsc_nsec;
f38e098f 661 u64 last_tsc_write;
5d3cb0f6 662 u32 last_tsc_khz;
e26101b1
ZA
663 u64 cur_tsc_nsec;
664 u64 cur_tsc_write;
665 u64 cur_tsc_offset;
0d3da0d2 666 u64 cur_tsc_generation;
b48aa97e 667 int nr_vcpus_matched_tsc;
ffde22ac 668
d828199e
MT
669 spinlock_t pvclock_gtod_sync_lock;
670 bool use_master_clock;
671 u64 master_kernel_ns;
672 cycle_t master_cycle_now;
7e44e449 673 struct delayed_work kvmclock_update_work;
332967a3 674 struct delayed_work kvmclock_sync_work;
d828199e 675
ffde22ac 676 struct kvm_xen_hvm_config xen_hvm_config;
55cd8e5a 677
6ef768fa
PB
678 /* reads protected by irq_srcu, writes by irq_lock */
679 struct hlist_head mask_notifier_list;
680
e83d5887 681 struct kvm_hv hyperv;
b034cf01
XG
682
683 #ifdef CONFIG_KVM_MMU_AUDIT
684 int audit_point;
685 #endif
54750f2c
MT
686
687 bool boot_vcpu_runs_old_kvmclock;
d71ba788 688 u32 bsp_vcpu_id;
90de4a18
NA
689
690 u64 disabled_quirks;
49df6397
SR
691
692 bool irqchip_split;
b053b2ae 693 u8 nr_reserved_ioapic_pins;
d69fb81f
ZX
694};
695
0711456c
ZX
696struct kvm_vm_stat {
697 u32 mmu_shadow_zapped;
698 u32 mmu_pte_write;
699 u32 mmu_pte_updated;
700 u32 mmu_pde_zapped;
701 u32 mmu_flooded;
702 u32 mmu_recycled;
dfc5aa00 703 u32 mmu_cache_miss;
4731d4c7 704 u32 mmu_unsync;
0711456c 705 u32 remote_tlb_flush;
05da4558 706 u32 lpages;
0711456c
ZX
707};
708
77b4c255
ZX
709struct kvm_vcpu_stat {
710 u32 pf_fixed;
711 u32 pf_guest;
712 u32 tlb_flush;
713 u32 invlpg;
714
715 u32 exits;
716 u32 io_exits;
717 u32 mmio_exits;
718 u32 signal_exits;
719 u32 irq_window_exits;
f08864b4 720 u32 nmi_window_exits;
77b4c255 721 u32 halt_exits;
f7819512 722 u32 halt_successful_poll;
62bea5bf 723 u32 halt_attempted_poll;
77b4c255
ZX
724 u32 halt_wakeup;
725 u32 request_irq_exits;
726 u32 irq_exits;
727 u32 host_state_reload;
728 u32 efer_reload;
729 u32 fpu_reload;
730 u32 insn_emulation;
731 u32 insn_emulation_fail;
f11c3a8d 732 u32 hypercalls;
fa89a817 733 u32 irq_injections;
c4abb7c9 734 u32 nmi_injections;
77b4c255 735};
ad312c7c 736
8a76d7f2
JR
737struct x86_instruction_info;
738
8fe8ab46
WA
739struct msr_data {
740 bool host_initiated;
741 u32 index;
742 u64 data;
743};
744
cb5281a5
PB
745struct kvm_lapic_irq {
746 u32 vector;
b7cb2231
PB
747 u16 delivery_mode;
748 u16 dest_mode;
749 bool level;
750 u16 trig_mode;
cb5281a5
PB
751 u32 shorthand;
752 u32 dest_id;
93bbf0b8 753 bool msi_redir_hint;
cb5281a5
PB
754};
755
ea4a5ff8
ZX
756struct kvm_x86_ops {
757 int (*cpu_has_kvm_support)(void); /* __init */
758 int (*disabled_by_bios)(void); /* __init */
13a34e06
RK
759 int (*hardware_enable)(void);
760 void (*hardware_disable)(void);
ea4a5ff8
ZX
761 void (*check_processor_compatibility)(void *rtn);
762 int (*hardware_setup)(void); /* __init */
763 void (*hardware_unsetup)(void); /* __exit */
774ead3a 764 bool (*cpu_has_accelerated_tpr)(void);
6d396b55 765 bool (*cpu_has_high_real_mode_segbase)(void);
0e851880 766 void (*cpuid_update)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
767
768 /* Create, but do not attach this VCPU */
769 struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
770 void (*vcpu_free)(struct kvm_vcpu *vcpu);
d28bc9dd 771 void (*vcpu_reset)(struct kvm_vcpu *vcpu, bool init_event);
ea4a5ff8
ZX
772
773 void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
774 void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
775 void (*vcpu_put)(struct kvm_vcpu *vcpu);
ea4a5ff8 776
c8639010 777 void (*update_db_bp_intercept)(struct kvm_vcpu *vcpu);
609e36d3 778 int (*get_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
8fe8ab46 779 int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
ea4a5ff8
ZX
780 u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
781 void (*get_segment)(struct kvm_vcpu *vcpu,
782 struct kvm_segment *var, int seg);
2e4d2653 783 int (*get_cpl)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
784 void (*set_segment)(struct kvm_vcpu *vcpu,
785 struct kvm_segment *var, int seg);
786 void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
e8467fda 787 void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
aff48baa 788 void (*decache_cr3)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
789 void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
790 void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
791 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
5e1746d6 792 int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
ea4a5ff8 793 void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
89a27f4d
GN
794 void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
795 void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
796 void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
797 void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
73aaf249
JK
798 u64 (*get_dr6)(struct kvm_vcpu *vcpu);
799 void (*set_dr6)(struct kvm_vcpu *vcpu, unsigned long value);
c77fb5fe 800 void (*sync_dirty_debug_regs)(struct kvm_vcpu *vcpu);
020df079 801 void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
5fdbf976 802 void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
ea4a5ff8
ZX
803 unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
804 void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
0fdd74f7 805 void (*fpu_activate)(struct kvm_vcpu *vcpu);
02daab21 806 void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
807
808 void (*tlb_flush)(struct kvm_vcpu *vcpu);
ea4a5ff8 809
851ba692
AK
810 void (*run)(struct kvm_vcpu *vcpu);
811 int (*handle_exit)(struct kvm_vcpu *vcpu);
ea4a5ff8 812 void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
2809f5d2 813 void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
37ccdcbe 814 u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
815 void (*patch_hypercall)(struct kvm_vcpu *vcpu,
816 unsigned char *hypercall_addr);
66fd3f7f 817 void (*set_irq)(struct kvm_vcpu *vcpu);
95ba8273 818 void (*set_nmi)(struct kvm_vcpu *vcpu);
298101da 819 void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
ce7ddec4
JR
820 bool has_error_code, u32 error_code,
821 bool reinject);
b463a6f7 822 void (*cancel_injection)(struct kvm_vcpu *vcpu);
78646121 823 int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
95ba8273 824 int (*nmi_allowed)(struct kvm_vcpu *vcpu);
3cfc3092
JK
825 bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
826 void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
c9a7953f
JK
827 void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
828 void (*enable_irq_window)(struct kvm_vcpu *vcpu);
95ba8273 829 void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
d50ab6c1 830 int (*cpu_uses_apicv)(struct kvm_vcpu *vcpu);
c7c9c56c
YZ
831 void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
832 void (*hwapic_isr_update)(struct kvm *kvm, int isr);
3bb345f3 833 void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu);
8d14695f 834 void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu, bool set);
4256f43f 835 void (*set_apic_access_page_addr)(struct kvm_vcpu *vcpu, hpa_t hpa);
a20ed54d
YZ
836 void (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
837 void (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
ea4a5ff8 838 int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
67253af5 839 int (*get_tdp_level)(void);
4b12f0de 840 u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
17cc3935 841 int (*get_lpage_level)(void);
4e47c7a6 842 bool (*rdtscp_supported)(void);
ad756a16 843 bool (*invpcid_supported)(void);
f1e2b260 844 void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment, bool host);
344f414f 845
1c97f0a0
JR
846 void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
847
d4330ef2
JR
848 void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
849
f5f48ee1
SY
850 bool (*has_wbinvd_exit)(void);
851
cc578287 852 void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale);
ba904635 853 u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu);
99e3e30a
ZA
854 void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
855
857e4099 856 u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc);
886b470c 857 u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu, u64 host_tsc);
857e4099 858
586f9607 859 void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
8a76d7f2
JR
860
861 int (*check_intercept)(struct kvm_vcpu *vcpu,
862 struct x86_instruction_info *info,
863 enum x86_intercept_stage stage);
a547c6db 864 void (*handle_external_intr)(struct kvm_vcpu *vcpu);
da8999d3 865 bool (*mpx_supported)(void);
55412b2e 866 bool (*xsaves_supported)(void);
b6b8a145
JK
867
868 int (*check_nested_events)(struct kvm_vcpu *vcpu, bool external_intr);
ae97a3b8
RK
869
870 void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
88178fd4
KH
871
872 /*
873 * Arch-specific dirty logging hooks. These hooks are only supposed to
874 * be valid if the specific arch has hardware-accelerated dirty logging
875 * mechanism. Currently only for PML on VMX.
876 *
877 * - slot_enable_log_dirty:
878 * called when enabling log dirty mode for the slot.
879 * - slot_disable_log_dirty:
880 * called when disabling log dirty mode for the slot.
881 * also called when slot is created with log dirty disabled.
882 * - flush_log_dirty:
883 * called before reporting dirty_bitmap to userspace.
884 * - enable_log_dirty_pt_masked:
885 * called when reenabling log dirty for the GFNs in the mask after
886 * corresponding bits are cleared in slot->dirty_bitmap.
887 */
888 void (*slot_enable_log_dirty)(struct kvm *kvm,
889 struct kvm_memory_slot *slot);
890 void (*slot_disable_log_dirty)(struct kvm *kvm,
891 struct kvm_memory_slot *slot);
892 void (*flush_log_dirty)(struct kvm *kvm);
893 void (*enable_log_dirty_pt_masked)(struct kvm *kvm,
894 struct kvm_memory_slot *slot,
895 gfn_t offset, unsigned long mask);
25462f7f
WH
896 /* pmu operations of sub-arch */
897 const struct kvm_pmu_ops *pmu_ops;
ea4a5ff8
ZX
898};
899
af585b92 900struct kvm_arch_async_pf {
7c90705b 901 u32 token;
af585b92 902 gfn_t gfn;
fb67e14f 903 unsigned long cr3;
c4806acd 904 bool direct_map;
af585b92
GN
905};
906
97896d04
ZX
907extern struct kvm_x86_ops *kvm_x86_ops;
908
f1e2b260
MT
909static inline void adjust_tsc_offset_guest(struct kvm_vcpu *vcpu,
910 s64 adjustment)
911{
912 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, false);
913}
914
915static inline void adjust_tsc_offset_host(struct kvm_vcpu *vcpu, s64 adjustment)
916{
917 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, true);
918}
919
54f1585a
ZX
920int kvm_mmu_module_init(void);
921void kvm_mmu_module_exit(void);
922
923void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
924int kvm_mmu_create(struct kvm_vcpu *vcpu);
8a3c1a33 925void kvm_mmu_setup(struct kvm_vcpu *vcpu);
7b52345e 926void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
4b12f0de 927 u64 dirty_mask, u64 nx_mask, u64 x_mask);
54f1585a 928
8a3c1a33 929void kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
1c91cad4
KH
930void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
931 struct kvm_memory_slot *memslot);
3ea3b7fa 932void kvm_mmu_zap_collapsible_sptes(struct kvm *kvm,
f36f3f28 933 const struct kvm_memory_slot *memslot);
f4b4b180
KH
934void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
935 struct kvm_memory_slot *memslot);
936void kvm_mmu_slot_largepage_remove_write_access(struct kvm *kvm,
937 struct kvm_memory_slot *memslot);
938void kvm_mmu_slot_set_dirty(struct kvm *kvm,
939 struct kvm_memory_slot *memslot);
940void kvm_mmu_clear_dirty_pt_masked(struct kvm *kvm,
941 struct kvm_memory_slot *slot,
942 gfn_t gfn_offset, unsigned long mask);
54f1585a 943void kvm_mmu_zap_all(struct kvm *kvm);
54bf36aa 944void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm, struct kvm_memslots *slots);
3ad82a7e 945unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
54f1585a
ZX
946void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
947
ff03a073 948int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
cc4b6871 949
3200f405 950int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
9f811285 951 const void *val, int bytes);
2f333bcb 952
6ef768fa
PB
953struct kvm_irq_mask_notifier {
954 void (*func)(struct kvm_irq_mask_notifier *kimn, bool masked);
955 int irq;
956 struct hlist_node link;
957};
958
959void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
960 struct kvm_irq_mask_notifier *kimn);
961void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
962 struct kvm_irq_mask_notifier *kimn);
963void kvm_fire_mask_notifiers(struct kvm *kvm, unsigned irqchip, unsigned pin,
964 bool mask);
965
2f333bcb 966extern bool tdp_enabled;
9f811285 967
a3e06bbe
LJ
968u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
969
92a1f12d
JR
970/* control of guest tsc rate supported? */
971extern bool kvm_has_tsc_control;
972/* minimum supported tsc_khz for guests */
973extern u32 kvm_min_guest_tsc_khz;
974/* maximum supported tsc_khz for guests */
975extern u32 kvm_max_guest_tsc_khz;
976
54f1585a 977enum emulation_result {
ac0a48c3
PB
978 EMULATE_DONE, /* no further processing */
979 EMULATE_USER_EXIT, /* kvm_run ready for userspace exit */
54f1585a
ZX
980 EMULATE_FAIL, /* can't emulate this instruction */
981};
982
571008da
SY
983#define EMULTYPE_NO_DECODE (1 << 0)
984#define EMULTYPE_TRAP_UD (1 << 1)
ba8afb6b 985#define EMULTYPE_SKIP (1 << 2)
1cb3f3ae 986#define EMULTYPE_RETRY (1 << 3)
991eebf9 987#define EMULTYPE_NO_REEXECUTE (1 << 4)
dc25e89e
AP
988int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
989 int emulation_type, void *insn, int insn_len);
51d8b661
AP
990
991static inline int emulate_instruction(struct kvm_vcpu *vcpu,
992 int emulation_type)
993{
dc25e89e 994 return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
51d8b661
AP
995}
996
f2b4b7dd 997void kvm_enable_efer_bits(u64);
384bb783 998bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
609e36d3 999int kvm_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
8fe8ab46 1000int kvm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
54f1585a
ZX
1001
1002struct x86_emulate_ctxt;
1003
cf8f70bf 1004int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
54f1585a
ZX
1005void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
1006int kvm_emulate_halt(struct kvm_vcpu *vcpu);
5cb56059 1007int kvm_vcpu_halt(struct kvm_vcpu *vcpu);
f5f48ee1 1008int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
54f1585a 1009
3e6e0aab 1010void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
c697518a 1011int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
2b4a273b 1012void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
3e6e0aab 1013
7f3d35fd
KW
1014int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
1015 int reason, bool has_error_code, u32 error_code);
37817f29 1016
49a9b07e 1017int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
2390218b 1018int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
a83b29c6 1019int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
eea1cff9 1020int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
020df079
GN
1021int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
1022int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
2d3ad1f4
AK
1023unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
1024void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
54f1585a 1025void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
2acf923e 1026int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
54f1585a 1027
609e36d3 1028int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
8fe8ab46 1029int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
54f1585a 1030
91586a3b
JK
1031unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
1032void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
022cd0e8 1033bool kvm_rdpmc(struct kvm_vcpu *vcpu);
91586a3b 1034
298101da
AK
1035void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1036void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
ce7ddec4
JR
1037void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1038void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
6389ee94 1039void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
ec92fe44
JR
1040int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
1041 gfn_t gfn, void *data, int offset, int len,
1042 u32 access);
0a79b009 1043bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
16f8a6f9 1044bool kvm_require_dr(struct kvm_vcpu *vcpu, int dr);
298101da 1045
1a577b72
MT
1046static inline int __kvm_irq_line_state(unsigned long *irq_state,
1047 int irq_source_id, int level)
1048{
1049 /* Logical OR for level trig interrupt */
1050 if (level)
1051 __set_bit(irq_source_id, irq_state);
1052 else
1053 __clear_bit(irq_source_id, irq_state);
1054
1055 return !!(*irq_state);
1056}
1057
1058int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
1059void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
3de42dc0 1060
3419ffc8
SY
1061void kvm_inject_nmi(struct kvm_vcpu *vcpu);
1062
54f1585a 1063void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
f57f2ef5 1064 const u8 *new, int bytes);
1cb3f3ae 1065int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
54f1585a
ZX
1066int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
1067void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
1068int kvm_mmu_load(struct kvm_vcpu *vcpu);
1069void kvm_mmu_unload(struct kvm_vcpu *vcpu);
0ba73cda 1070void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
54987b7a
PB
1071gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1072 struct x86_exception *exception);
ab9ae313
AK
1073gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
1074 struct x86_exception *exception);
1075gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
1076 struct x86_exception *exception);
1077gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
1078 struct x86_exception *exception);
1079gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
1080 struct x86_exception *exception);
54f1585a
ZX
1081
1082int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
1083
dc25e89e
AP
1084int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
1085 void *insn, int insn_len);
a7052897 1086void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
d8d173da 1087void kvm_mmu_new_cr3(struct kvm_vcpu *vcpu);
34c16eec 1088
18552672 1089void kvm_enable_tdp(void);
5f4cb662 1090void kvm_disable_tdp(void);
18552672 1091
54987b7a
PB
1092static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1093 struct x86_exception *exception)
e459e322
XG
1094{
1095 return gpa;
1096}
1097
ec6d273d
ZX
1098static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
1099{
1100 struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
1101
1102 return (struct kvm_mmu_page *)page_private(page);
1103}
1104
d6e88aec 1105static inline u16 kvm_read_ldt(void)
ec6d273d
ZX
1106{
1107 u16 ldt;
1108 asm("sldt %0" : "=g"(ldt));
1109 return ldt;
1110}
1111
d6e88aec 1112static inline void kvm_load_ldt(u16 sel)
ec6d273d
ZX
1113{
1114 asm("lldt %0" : : "rm"(sel));
1115}
ec6d273d 1116
ec6d273d
ZX
1117#ifdef CONFIG_X86_64
1118static inline unsigned long read_msr(unsigned long msr)
1119{
1120 u64 value;
1121
1122 rdmsrl(msr, value);
1123 return value;
1124}
1125#endif
1126
ec6d273d
ZX
1127static inline u32 get_rdx_init_val(void)
1128{
1129 return 0x600; /* P6 family */
1130}
1131
c1a5d4f9
AK
1132static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
1133{
1134 kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
1135}
1136
854e8bb1
NA
1137static inline u64 get_canonical(u64 la)
1138{
1139 return ((int64_t)la << 16) >> 16;
1140}
1141
1142static inline bool is_noncanonical_address(u64 la)
1143{
1144#ifdef CONFIG_X86_64
1145 return get_canonical(la) != la;
1146#else
1147 return false;
1148#endif
1149}
1150
ec6d273d
ZX
1151#define TSS_IOPB_BASE_OFFSET 0x66
1152#define TSS_BASE_SIZE 0x68
1153#define TSS_IOPB_SIZE (65536 / 8)
1154#define TSS_REDIRECTION_SIZE (256 / 8)
7d76b4d3
JP
1155#define RMODE_TSS_SIZE \
1156 (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
53e0aa7b 1157
37817f29
IE
1158enum {
1159 TASK_SWITCH_CALL = 0,
1160 TASK_SWITCH_IRET = 1,
1161 TASK_SWITCH_JMP = 2,
1162 TASK_SWITCH_GATE = 3,
1163};
1164
1371d904 1165#define HF_GIF_MASK (1 << 0)
3d6368ef
AG
1166#define HF_HIF_MASK (1 << 1)
1167#define HF_VINTR_MASK (1 << 2)
95ba8273 1168#define HF_NMI_MASK (1 << 3)
44c11430 1169#define HF_IRET_MASK (1 << 4)
ec9e60b2 1170#define HF_GUEST_MASK (1 << 5) /* VCPU is in guest-mode */
f077825a
PB
1171#define HF_SMM_MASK (1 << 6)
1172#define HF_SMM_INSIDE_NMI_MASK (1 << 7)
1371d904 1173
699023e2
PB
1174#define __KVM_VCPU_MULTIPLE_ADDRESS_SPACE
1175#define KVM_ADDRESS_SPACE_NUM 2
1176
1177#define kvm_arch_vcpu_memslots_id(vcpu) ((vcpu)->arch.hflags & HF_SMM_MASK ? 1 : 0)
1178#define kvm_memslots_for_spte_role(kvm, role) __kvm_memslots(kvm, (role).smm)
1371d904 1179
4ecac3fd
AK
1180/*
1181 * Hardware virtualization extension instructions may fault if a
1182 * reboot turns off virtualization while processes are running.
1183 * Trap the fault and ignore the instruction if that happens.
1184 */
b7c4145b 1185asmlinkage void kvm_spurious_fault(void);
4ecac3fd 1186
5e520e62 1187#define ____kvm_handle_fault_on_reboot(insn, cleanup_insn) \
4ecac3fd 1188 "666: " insn "\n\t" \
b7c4145b 1189 "668: \n\t" \
18b13e54 1190 ".pushsection .fixup, \"ax\" \n" \
4ecac3fd 1191 "667: \n\t" \
5e520e62 1192 cleanup_insn "\n\t" \
b7c4145b
AK
1193 "cmpb $0, kvm_rebooting \n\t" \
1194 "jne 668b \n\t" \
8ceed347 1195 __ASM_SIZE(push) " $666b \n\t" \
b7c4145b 1196 "call kvm_spurious_fault \n\t" \
4ecac3fd 1197 ".popsection \n\t" \
3ee89722 1198 _ASM_EXTABLE(666b, 667b)
4ecac3fd 1199
5e520e62
AK
1200#define __kvm_handle_fault_on_reboot(insn) \
1201 ____kvm_handle_fault_on_reboot(insn, "")
1202
e930bffe
AA
1203#define KVM_ARCH_WANT_MMU_NOTIFIER
1204int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
b3ae2096 1205int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
57128468 1206int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
8ee53820 1207int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
3da0dd43 1208void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
c7c9c56c 1209int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
a1b37100
GN
1210int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1211int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
0b71785d 1212int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
d28bc9dd 1213void kvm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event);
4256f43f 1214void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu);
fe71557a
TC
1215void kvm_arch_mmu_notifier_invalidate_page(struct kvm *kvm,
1216 unsigned long address);
e930bffe 1217
18863bdd 1218void kvm_define_shared_msr(unsigned index, u32 msr);
8b3c3104 1219int kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
18863bdd 1220
82b32774 1221unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
f92653ee
JK
1222bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1223
af585b92
GN
1224void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1225 struct kvm_async_pf *work);
1226void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1227 struct kvm_async_pf *work);
56028d08
GN
1228void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1229 struct kvm_async_pf *work);
7c90705b 1230bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
af585b92
GN
1231extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1232
db8fcefa
AP
1233void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1234
f5132b01
GN
1235int kvm_is_in_guest(void);
1236
9da0e4d5
PB
1237int __x86_set_memory_region(struct kvm *kvm,
1238 const struct kvm_userspace_memory_region *mem);
1239int x86_set_memory_region(struct kvm *kvm,
1240 const struct kvm_userspace_memory_region *mem);
d71ba788
PB
1241bool kvm_vcpu_is_reset_bsp(struct kvm_vcpu *vcpu);
1242bool kvm_vcpu_is_bsp(struct kvm_vcpu *vcpu);
f5132b01 1243
1965aae3 1244#endif /* _ASM_X86_KVM_HOST_H */