]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blame - arch/x86/include/asm/kvm_host.h
Merge branch 'kvm-ppc-next' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus...
[mirror_ubuntu-jammy-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>
34c16eec 28
50d0a0f9 29#include <asm/pvclock-abi.h>
e01a1b57 30#include <asm/desc.h>
0bed3b56 31#include <asm/mtrr.h>
9962d032 32#include <asm/msr-index.h>
3ee89722 33#include <asm/asm.h>
e01a1b57 34
cbf64358 35#define KVM_MAX_VCPUS 255
a59cb29e 36#define KVM_SOFT_MAX_VCPUS 160
1d4e7e3c 37#define KVM_USER_MEM_SLOTS 509
0743247f
AW
38/* memory slots that are not exposed to userspace */
39#define KVM_PRIVATE_MEM_SLOTS 3
bbacc0c1 40#define KVM_MEM_SLOTS_NUM (KVM_USER_MEM_SLOTS + KVM_PRIVATE_MEM_SLOTS)
93a5cef0 41
69a9f69b 42#define KVM_PIO_PAGE_OFFSET 1
542472b5 43#define KVM_COALESCED_MMIO_PAGE_OFFSET 2
920552b2 44#define KVM_HALT_POLL_NS_DEFAULT 500000
69a9f69b 45
8175e5b7
AG
46#define KVM_IRQCHIP_NUM_PINS KVM_IOAPIC_NUM_PINS
47
cfec82cb
JR
48#define CR0_RESERVED_BITS \
49 (~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
50 | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
51 | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
52
346874c9 53#define CR3_L_MODE_RESERVED_BITS 0xFFFFFF0000000000ULL
cfaa790a 54#define CR3_PCID_INVD BIT_64(63)
cfec82cb
JR
55#define CR4_RESERVED_BITS \
56 (~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
57 | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE \
ad756a16 58 | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR | X86_CR4_PCIDE \
afcbf13f 59 | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_FSGSBASE \
56d6efc2 60 | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE | X86_CR4_SMAP))
cfec82cb
JR
61
62#define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
63
64
cd6e8f87 65
cd6e8f87 66#define INVALID_PAGE (~(hpa_t)0)
dd180b3e
XG
67#define VALID_PAGE(x) ((x) != INVALID_PAGE)
68
cd6e8f87
ZX
69#define UNMAPPED_GVA (~(gpa_t)0)
70
ec04b260 71/* KVM Hugepage definitions for x86 */
04326caa 72#define KVM_NR_PAGE_SIZES 3
82855413
JR
73#define KVM_HPAGE_GFN_SHIFT(x) (((x) - 1) * 9)
74#define KVM_HPAGE_SHIFT(x) (PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
ec04b260
JR
75#define KVM_HPAGE_SIZE(x) (1UL << KVM_HPAGE_SHIFT(x))
76#define KVM_HPAGE_MASK(x) (~(KVM_HPAGE_SIZE(x) - 1))
77#define KVM_PAGES_PER_HPAGE(x) (KVM_HPAGE_SIZE(x) / PAGE_SIZE)
05da4558 78
6d9d41e5
CD
79static inline gfn_t gfn_to_index(gfn_t gfn, gfn_t base_gfn, int level)
80{
81 /* KVM_HPAGE_GFN_SHIFT(PT_PAGE_TABLE_LEVEL) must be 0. */
82 return (gfn >> KVM_HPAGE_GFN_SHIFT(level)) -
83 (base_gfn >> KVM_HPAGE_GFN_SHIFT(level));
84}
85
d657a98e
ZX
86#define KVM_PERMILLE_MMU_PAGES 20
87#define KVM_MIN_ALLOC_MMU_PAGES 64
1ae0a13d
DE
88#define KVM_MMU_HASH_SHIFT 10
89#define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
d657a98e
ZX
90#define KVM_MIN_FREE_MMU_PAGES 5
91#define KVM_REFILL_PAGES 25
73c1160c 92#define KVM_MAX_CPUID_ENTRIES 80
0bed3b56 93#define KVM_NR_FIXED_MTRR_REGION 88
0d234daf 94#define KVM_NR_VAR_MTRR 8
d657a98e 95
af585b92
GN
96#define ASYNC_PF_PER_VCPU 64
97
5fdbf976 98enum kvm_reg {
2b3ccfa0
ZX
99 VCPU_REGS_RAX = 0,
100 VCPU_REGS_RCX = 1,
101 VCPU_REGS_RDX = 2,
102 VCPU_REGS_RBX = 3,
103 VCPU_REGS_RSP = 4,
104 VCPU_REGS_RBP = 5,
105 VCPU_REGS_RSI = 6,
106 VCPU_REGS_RDI = 7,
107#ifdef CONFIG_X86_64
108 VCPU_REGS_R8 = 8,
109 VCPU_REGS_R9 = 9,
110 VCPU_REGS_R10 = 10,
111 VCPU_REGS_R11 = 11,
112 VCPU_REGS_R12 = 12,
113 VCPU_REGS_R13 = 13,
114 VCPU_REGS_R14 = 14,
115 VCPU_REGS_R15 = 15,
116#endif
5fdbf976 117 VCPU_REGS_RIP,
2b3ccfa0
ZX
118 NR_VCPU_REGS
119};
120
6de4f3ad
AK
121enum kvm_reg_ex {
122 VCPU_EXREG_PDPTR = NR_VCPU_REGS,
aff48baa 123 VCPU_EXREG_CR3,
6de12732 124 VCPU_EXREG_RFLAGS,
2fb92db1 125 VCPU_EXREG_SEGMENTS,
6de4f3ad
AK
126};
127
2b3ccfa0 128enum {
81609e3e 129 VCPU_SREG_ES,
2b3ccfa0 130 VCPU_SREG_CS,
81609e3e 131 VCPU_SREG_SS,
2b3ccfa0 132 VCPU_SREG_DS,
2b3ccfa0
ZX
133 VCPU_SREG_FS,
134 VCPU_SREG_GS,
2b3ccfa0
ZX
135 VCPU_SREG_TR,
136 VCPU_SREG_LDTR,
137};
138
56e82318 139#include <asm/kvm_emulate.h>
2b3ccfa0 140
d657a98e
ZX
141#define KVM_NR_MEM_OBJS 40
142
42dbaa5a
JK
143#define KVM_NR_DB_REGS 4
144
145#define DR6_BD (1 << 13)
146#define DR6_BS (1 << 14)
6f43ed01
NA
147#define DR6_RTM (1 << 16)
148#define DR6_FIXED_1 0xfffe0ff0
149#define DR6_INIT 0xffff0ff0
150#define DR6_VOLATILE 0x0001e00f
42dbaa5a
JK
151
152#define DR7_BP_EN_MASK 0x000000ff
153#define DR7_GE (1 << 9)
154#define DR7_GD (1 << 13)
155#define DR7_FIXED_1 0x00000400
6f43ed01 156#define DR7_VOLATILE 0xffff2bff
42dbaa5a 157
c205fb7d
NA
158#define PFERR_PRESENT_BIT 0
159#define PFERR_WRITE_BIT 1
160#define PFERR_USER_BIT 2
161#define PFERR_RSVD_BIT 3
162#define PFERR_FETCH_BIT 4
163
164#define PFERR_PRESENT_MASK (1U << PFERR_PRESENT_BIT)
165#define PFERR_WRITE_MASK (1U << PFERR_WRITE_BIT)
166#define PFERR_USER_MASK (1U << PFERR_USER_BIT)
167#define PFERR_RSVD_MASK (1U << PFERR_RSVD_BIT)
168#define PFERR_FETCH_MASK (1U << PFERR_FETCH_BIT)
169
41383771
GN
170/* apic attention bits */
171#define KVM_APIC_CHECK_VAPIC 0
ae7a2a3f
MT
172/*
173 * The following bit is set with PV-EOI, unset on EOI.
174 * We detect PV-EOI changes by guest by comparing
175 * this bit with PV-EOI in guest memory.
176 * See the implementation in apic_update_pv_eoi.
177 */
178#define KVM_APIC_PV_EOI_PENDING 1
41383771 179
d84f1e07
FW
180struct kvm_kernel_irq_routing_entry;
181
d657a98e
ZX
182/*
183 * We don't want allocation failures within the mmu code, so we preallocate
184 * enough memory for a single page fault in a cache.
185 */
186struct kvm_mmu_memory_cache {
187 int nobjs;
188 void *objects[KVM_NR_MEM_OBJS];
189};
190
d657a98e
ZX
191union kvm_mmu_page_role {
192 unsigned word;
193 struct {
7d76b4d3 194 unsigned level:4;
5b7e0102 195 unsigned cr4_pae:1;
7d76b4d3 196 unsigned quadrant:2;
f6e2c02b 197 unsigned direct:1;
7d76b4d3 198 unsigned access:3;
2e53d63a 199 unsigned invalid:1;
9645bb56 200 unsigned nxe:1;
3dbe1415 201 unsigned cr0_wp:1;
411c588d 202 unsigned smep_andnot_wp:1;
0be0226f 203 unsigned smap_andnot_wp:1;
699023e2
PB
204 unsigned :8;
205
206 /*
207 * This is left at the top of the word so that
208 * kvm_memslots_for_spte_role can extract it with a
209 * simple shift. While there is room, give it a whole
210 * byte so it is also faster to load it from memory.
211 */
212 unsigned smm:8;
d657a98e
ZX
213 };
214};
215
216struct kvm_mmu_page {
217 struct list_head link;
218 struct hlist_node hash_link;
219
220 /*
221 * The following two entries are used to key the shadow page in the
222 * hash table.
223 */
224 gfn_t gfn;
225 union kvm_mmu_page_role role;
226
227 u64 *spt;
228 /* hold the gfn of each spte inside spt */
229 gfn_t *gfns;
4731d4c7 230 bool unsync;
0571d366 231 int root_count; /* Currently serving as active root */
60c8aec6 232 unsigned int unsync_children;
67052b35 233 unsigned long parent_ptes; /* Reverse mapping for parent_pte */
f6f8adee
XG
234
235 /* The page is obsolete if mmu_valid_gen != kvm->arch.mmu_valid_gen. */
5304b8d3 236 unsigned long mmu_valid_gen;
f6f8adee 237
0074ff63 238 DECLARE_BITMAP(unsync_child_bitmap, 512);
c2a2ac2b
XG
239
240#ifdef CONFIG_X86_32
accaefe0
XG
241 /*
242 * Used out of the mmu-lock to avoid reading spte values while an
243 * update is in progress; see the comments in __get_spte_lockless().
244 */
c2a2ac2b
XG
245 int clear_spte_count;
246#endif
247
0cbf8e43 248 /* Number of writes since the last time traversal visited this page. */
a30f47cb 249 int write_flooding_count;
d657a98e
ZX
250};
251
1c08364c
AK
252struct kvm_pio_request {
253 unsigned long count;
1c08364c
AK
254 int in;
255 int port;
256 int size;
1c08364c
AK
257};
258
a0a64f50
XG
259struct rsvd_bits_validate {
260 u64 rsvd_bits_mask[2][4];
261 u64 bad_mt_xwr;
262};
263
d657a98e
ZX
264/*
265 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
266 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
267 * mode.
268 */
269struct kvm_mmu {
f43addd4 270 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
5777ed34 271 unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
e4e517b4 272 u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
78b2c54a
XG
273 int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
274 bool prefault);
6389ee94
AK
275 void (*inject_page_fault)(struct kvm_vcpu *vcpu,
276 struct x86_exception *fault);
1871c602 277 gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
ab9ae313 278 struct x86_exception *exception);
54987b7a
PB
279 gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
280 struct x86_exception *exception);
e8bc217a 281 int (*sync_page)(struct kvm_vcpu *vcpu,
a4a8e6f7 282 struct kvm_mmu_page *sp);
a7052897 283 void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
0f53b5b1 284 void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
7c562522 285 u64 *spte, const void *pte);
d657a98e
ZX
286 hpa_t root_hpa;
287 int root_level;
288 int shadow_root_level;
a770f6f2 289 union kvm_mmu_page_role base_role;
c5a78f2b 290 bool direct_map;
d657a98e 291
97d64b78
AK
292 /*
293 * Bitmap; bit set = permission fault
294 * Byte index: page fault error code [4:1]
295 * Bit index: pte permissions in ACC_* format
296 */
297 u8 permissions[16];
298
d657a98e 299 u64 *pae_root;
81407ca5 300 u64 *lm_root;
c258b62b
XG
301
302 /*
303 * check zero bits on shadow page table entries, these
304 * bits include not only hardware reserved bits but also
305 * the bits spte never used.
306 */
307 struct rsvd_bits_validate shadow_zero_check;
308
a0a64f50 309 struct rsvd_bits_validate guest_rsvd_check;
ff03a073 310
6fd01b71
AK
311 /*
312 * Bitmap: bit set = last pte in walk
313 * index[0:1]: level (zero-based)
314 * index[2]: pte.ps
315 */
316 u8 last_pte_bitmap;
317
2d48a985
JR
318 bool nx;
319
ff03a073 320 u64 pdptrs[4]; /* pae */
d657a98e
ZX
321};
322
f5132b01
GN
323enum pmc_type {
324 KVM_PMC_GP = 0,
325 KVM_PMC_FIXED,
326};
327
328struct kvm_pmc {
329 enum pmc_type type;
330 u8 idx;
331 u64 counter;
332 u64 eventsel;
333 struct perf_event *perf_event;
334 struct kvm_vcpu *vcpu;
335};
336
337struct kvm_pmu {
338 unsigned nr_arch_gp_counters;
339 unsigned nr_arch_fixed_counters;
340 unsigned available_event_types;
341 u64 fixed_ctr_ctrl;
342 u64 global_ctrl;
343 u64 global_status;
344 u64 global_ovf_ctrl;
345 u64 counter_bitmask[2];
346 u64 global_ctrl_mask;
103af0a9 347 u64 reserved_bits;
f5132b01 348 u8 version;
15c7ad51
RR
349 struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
350 struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
f5132b01
GN
351 struct irq_work irq_work;
352 u64 reprogram_pmi;
353};
354
25462f7f
WH
355struct kvm_pmu_ops;
356
360b948d
PB
357enum {
358 KVM_DEBUGREG_BP_ENABLED = 1,
c77fb5fe 359 KVM_DEBUGREG_WONT_EXIT = 2,
ae561ede 360 KVM_DEBUGREG_RELOAD = 4,
360b948d
PB
361};
362
86fd5270
XG
363struct kvm_mtrr_range {
364 u64 base;
365 u64 mask;
19efffa2 366 struct list_head node;
86fd5270
XG
367};
368
70109e7d 369struct kvm_mtrr {
86fd5270 370 struct kvm_mtrr_range var_ranges[KVM_NR_VAR_MTRR];
70109e7d 371 mtrr_type fixed_ranges[KVM_NR_FIXED_MTRR_REGION];
10fac2dc 372 u64 deftype;
19efffa2
XG
373
374 struct list_head head;
70109e7d
XG
375};
376
e83d5887
AS
377/* Hyper-V per vcpu emulation context */
378struct kvm_vcpu_hv {
379 u64 hv_vapic;
9eec50b8 380 s64 runtime_offset;
e83d5887
AS
381};
382
ad312c7c 383struct kvm_vcpu_arch {
5fdbf976
MT
384 /*
385 * rip and regs accesses must go through
386 * kvm_{register,rip}_{read,write} functions.
387 */
388 unsigned long regs[NR_VCPU_REGS];
389 u32 regs_avail;
390 u32 regs_dirty;
34c16eec
ZX
391
392 unsigned long cr0;
e8467fda 393 unsigned long cr0_guest_owned_bits;
34c16eec
ZX
394 unsigned long cr2;
395 unsigned long cr3;
396 unsigned long cr4;
fc78f519 397 unsigned long cr4_guest_owned_bits;
34c16eec 398 unsigned long cr8;
1371d904 399 u32 hflags;
f6801dff 400 u64 efer;
34c16eec
ZX
401 u64 apic_base;
402 struct kvm_lapic *apic; /* kernel irqchip context */
3bb345f3 403 u64 eoi_exit_bitmap[4];
41383771 404 unsigned long apic_attention;
e1035715 405 int32_t apic_arb_prio;
34c16eec 406 int mp_state;
34c16eec 407 u64 ia32_misc_enable_msr;
64d60670 408 u64 smbase;
b209749f 409 bool tpr_access_reporting;
20300099 410 u64 ia32_xss;
34c16eec 411
14dfe855
JR
412 /*
413 * Paging state of the vcpu
414 *
415 * If the vcpu runs in guest mode with two level paging this still saves
416 * the paging mode of the l1 guest. This context is always used to
417 * handle faults.
418 */
34c16eec 419 struct kvm_mmu mmu;
8df25a32 420
6539e738
JR
421 /*
422 * Paging state of an L2 guest (used for nested npt)
423 *
424 * This context will save all necessary information to walk page tables
425 * of the an L2 guest. This context is only initialized for page table
426 * walking and not for faulting since we never handle l2 page faults on
427 * the host.
428 */
429 struct kvm_mmu nested_mmu;
430
14dfe855
JR
431 /*
432 * Pointer to the mmu context currently used for
433 * gva_to_gpa translations.
434 */
435 struct kvm_mmu *walk_mmu;
436
53c07b18 437 struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
34c16eec
ZX
438 struct kvm_mmu_memory_cache mmu_page_cache;
439 struct kvm_mmu_memory_cache mmu_page_header_cache;
440
98918833 441 struct fpu guest_fpu;
c447e76b 442 bool eager_fpu;
2acf923e 443 u64 xcr0;
d7876f1b 444 u64 guest_supported_xcr0;
4344ee98 445 u32 guest_xstate_size;
34c16eec 446
34c16eec
ZX
447 struct kvm_pio_request pio;
448 void *pio_data;
449
66fd3f7f
GN
450 u8 event_exit_inst_len;
451
298101da
AK
452 struct kvm_queued_exception {
453 bool pending;
454 bool has_error_code;
ce7ddec4 455 bool reinject;
298101da
AK
456 u8 nr;
457 u32 error_code;
458 } exception;
459
937a7eae
AK
460 struct kvm_queued_interrupt {
461 bool pending;
66fd3f7f 462 bool soft;
937a7eae
AK
463 u8 nr;
464 } interrupt;
465
34c16eec
ZX
466 int halt_request; /* real mode on Intel only */
467
468 int cpuid_nent;
07716717 469 struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
5a4f55cd
EK
470
471 int maxphyaddr;
472
34c16eec
ZX
473 /* emulate context */
474
475 struct x86_emulate_ctxt emulate_ctxt;
7ae441ea
GN
476 bool emulate_regs_need_sync_to_vcpu;
477 bool emulate_regs_need_sync_from_vcpu;
716d51ab 478 int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
18068523
GOC
479
480 gpa_t time;
50d0a0f9 481 struct pvclock_vcpu_time_info hv_clock;
e48672fa 482 unsigned int hw_tsc_khz;
0b79459b
AH
483 struct gfn_to_hva_cache pv_time;
484 bool pv_time_enabled;
51d59c6b
MT
485 /* set guest stopped flag in pvclock flags field */
486 bool pvclock_set_guest_stopped_request;
c9aaa895
GC
487
488 struct {
489 u64 msr_val;
490 u64 last_steal;
491 u64 accum_steal;
492 struct gfn_to_hva_cache stime;
493 struct kvm_steal_time steal;
494 } st;
495
1d5f066e 496 u64 last_guest_tsc;
6f526ec5 497 u64 last_host_tsc;
0dd6a6ed 498 u64 tsc_offset_adjustment;
e26101b1
ZA
499 u64 this_tsc_nsec;
500 u64 this_tsc_write;
0d3da0d2 501 u64 this_tsc_generation;
c285545f 502 bool tsc_catchup;
cc578287
ZA
503 bool tsc_always_catchup;
504 s8 virtual_tsc_shift;
505 u32 virtual_tsc_mult;
506 u32 virtual_tsc_khz;
ba904635 507 s64 ia32_tsc_adjust_msr;
3419ffc8 508
7460fb4a
AK
509 atomic_t nmi_queued; /* unprocessed asynchronous NMIs */
510 unsigned nmi_pending; /* NMI queued after currently running handler */
511 bool nmi_injected; /* Trying to inject an NMI this entry */
f077825a 512 bool smi_pending; /* SMI queued after currently running handler */
9ba075a6 513
70109e7d 514 struct kvm_mtrr mtrr_state;
7cb060a9 515 u64 pat;
42dbaa5a 516
360b948d 517 unsigned switch_db_regs;
42dbaa5a
JK
518 unsigned long db[KVM_NR_DB_REGS];
519 unsigned long dr6;
520 unsigned long dr7;
521 unsigned long eff_db[KVM_NR_DB_REGS];
c8639010 522 unsigned long guest_debug_dr7;
890ca9ae
HY
523
524 u64 mcg_cap;
525 u64 mcg_status;
526 u64 mcg_ctl;
527 u64 *mce_banks;
94fe45da 528
bebb106a
XG
529 /* Cache MMIO info */
530 u64 mmio_gva;
531 unsigned access;
532 gfn_t mmio_gfn;
56f17dd3 533 u64 mmio_gen;
bebb106a 534
f5132b01
GN
535 struct kvm_pmu pmu;
536
94fe45da 537 /* used for guest single stepping over the given code position */
94fe45da 538 unsigned long singlestep_rip;
f92653ee 539
e83d5887 540 struct kvm_vcpu_hv hyperv;
f5f48ee1
SY
541
542 cpumask_var_t wbinvd_dirty_mask;
af585b92 543
1cb3f3ae
XG
544 unsigned long last_retry_eip;
545 unsigned long last_retry_addr;
546
af585b92
GN
547 struct {
548 bool halted;
549 gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
344d9588
GN
550 struct gfn_to_hva_cache data;
551 u64 msr_val;
7c90705b 552 u32 id;
6adba527 553 bool send_user_only;
af585b92 554 } apf;
2b036c6b
BO
555
556 /* OSVW MSRs (AMD only) */
557 struct {
558 u64 length;
559 u64 status;
560 } osvw;
ae7a2a3f
MT
561
562 struct {
563 u64 msr_val;
564 struct gfn_to_hva_cache data;
565 } pv_eoi;
93c05d3e
XG
566
567 /*
568 * Indicate whether the access faults on its page table in guest
569 * which is set when fix page fault and used to detect unhandeable
570 * instruction.
571 */
572 bool write_fault_to_shadow_pgtable;
25d92081
YZ
573
574 /* set at EPT violation at this point */
575 unsigned long exit_qualification;
6aef266c
SV
576
577 /* pv related host specific info */
578 struct {
579 bool pv_unhalted;
580 } pv;
7543a635
SR
581
582 int pending_ioapic_eoi;
1c1a9ce9 583 int pending_external_vector;
34c16eec
ZX
584};
585
db3fe4eb 586struct kvm_lpage_info {
db3fe4eb
TY
587 int write_count;
588};
589
590struct kvm_arch_memory_slot {
d89cc617 591 unsigned long *rmap[KVM_NR_PAGE_SIZES];
db3fe4eb
TY
592 struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
593};
594
3548a259
RK
595/*
596 * We use as the mode the number of bits allocated in the LDR for the
597 * logical processor ID. It happens that these are all powers of two.
598 * This makes it is very easy to detect cases where the APICs are
599 * configured for multiple modes; in that case, we cannot use the map and
600 * hence cannot use kvm_irq_delivery_to_apic_fast either.
601 */
602#define KVM_APIC_MODE_XAPIC_CLUSTER 4
603#define KVM_APIC_MODE_XAPIC_FLAT 8
604#define KVM_APIC_MODE_X2APIC 16
605
1e08ec4a
GN
606struct kvm_apic_map {
607 struct rcu_head rcu;
3548a259 608 u8 mode;
1e08ec4a
GN
609 struct kvm_lapic *phys_map[256];
610 /* first index is cluster id second is cpu id in a cluster */
611 struct kvm_lapic *logical_map[16][16];
612};
613
e83d5887
AS
614/* Hyper-V emulation context */
615struct kvm_hv {
616 u64 hv_guest_os_id;
617 u64 hv_hypercall;
618 u64 hv_tsc_page;
e7d9513b
AS
619
620 /* Hyper-v based guest crash (NT kernel bugcheck) parameters */
621 u64 hv_crash_param[HV_X64_MSR_CRASH_PARAMS];
622 u64 hv_crash_ctl;
e83d5887
AS
623};
624
fef9cce0 625struct kvm_arch {
49d5ca26 626 unsigned int n_used_mmu_pages;
f05e70ac 627 unsigned int n_requested_mmu_pages;
39de71ec 628 unsigned int n_max_mmu_pages;
332b207d 629 unsigned int indirect_shadow_pages;
5304b8d3 630 unsigned long mmu_valid_gen;
f05e70ac
ZX
631 struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
632 /*
633 * Hash table of struct kvm_mmu_page.
634 */
635 struct list_head active_mmu_pages;
365c8868
XG
636 struct list_head zapped_obsolete_pages;
637
4d5c5d0f 638 struct list_head assigned_dev_head;
19de40a8 639 struct iommu_domain *iommu_domain;
d96eb2c6 640 bool iommu_noncoherent;
e0f0bbc5
AW
641#define __KVM_HAVE_ARCH_NONCOHERENT_DMA
642 atomic_t noncoherent_dma_count;
5544eb9b
PB
643#define __KVM_HAVE_ARCH_ASSIGNED_DEVICE
644 atomic_t assigned_device_count;
d7deeeb0
ZX
645 struct kvm_pic *vpic;
646 struct kvm_ioapic *vioapic;
7837699f 647 struct kvm_pit *vpit;
42720138 648 atomic_t vapics_in_nmi_mode;
1e08ec4a
GN
649 struct mutex apic_map_lock;
650 struct kvm_apic_map *apic_map;
bfc6d222 651
bfc6d222 652 unsigned int tss_addr;
c24ae0dc 653 bool apic_access_page_done;
18068523
GOC
654
655 gpa_t wall_clock;
b7ebfb05 656
b7ebfb05 657 bool ept_identity_pagetable_done;
b927a3ce 658 gpa_t ept_identity_map_addr;
5550af4d
SY
659
660 unsigned long irq_sources_bitmap;
afbcf7ab 661 s64 kvmclock_offset;
038f8c11 662 raw_spinlock_t tsc_write_lock;
f38e098f 663 u64 last_tsc_nsec;
f38e098f 664 u64 last_tsc_write;
5d3cb0f6 665 u32 last_tsc_khz;
e26101b1
ZA
666 u64 cur_tsc_nsec;
667 u64 cur_tsc_write;
668 u64 cur_tsc_offset;
0d3da0d2 669 u64 cur_tsc_generation;
b48aa97e 670 int nr_vcpus_matched_tsc;
ffde22ac 671
d828199e
MT
672 spinlock_t pvclock_gtod_sync_lock;
673 bool use_master_clock;
674 u64 master_kernel_ns;
675 cycle_t master_cycle_now;
7e44e449 676 struct delayed_work kvmclock_update_work;
332967a3 677 struct delayed_work kvmclock_sync_work;
d828199e 678
ffde22ac 679 struct kvm_xen_hvm_config xen_hvm_config;
55cd8e5a 680
6ef768fa
PB
681 /* reads protected by irq_srcu, writes by irq_lock */
682 struct hlist_head mask_notifier_list;
683
e83d5887 684 struct kvm_hv hyperv;
b034cf01
XG
685
686 #ifdef CONFIG_KVM_MMU_AUDIT
687 int audit_point;
688 #endif
54750f2c
MT
689
690 bool boot_vcpu_runs_old_kvmclock;
d71ba788 691 u32 bsp_vcpu_id;
90de4a18
NA
692
693 u64 disabled_quirks;
49df6397
SR
694
695 bool irqchip_split;
b053b2ae 696 u8 nr_reserved_ioapic_pins;
d69fb81f
ZX
697};
698
0711456c
ZX
699struct kvm_vm_stat {
700 u32 mmu_shadow_zapped;
701 u32 mmu_pte_write;
702 u32 mmu_pte_updated;
703 u32 mmu_pde_zapped;
704 u32 mmu_flooded;
705 u32 mmu_recycled;
dfc5aa00 706 u32 mmu_cache_miss;
4731d4c7 707 u32 mmu_unsync;
0711456c 708 u32 remote_tlb_flush;
05da4558 709 u32 lpages;
0711456c
ZX
710};
711
77b4c255
ZX
712struct kvm_vcpu_stat {
713 u32 pf_fixed;
714 u32 pf_guest;
715 u32 tlb_flush;
716 u32 invlpg;
717
718 u32 exits;
719 u32 io_exits;
720 u32 mmio_exits;
721 u32 signal_exits;
722 u32 irq_window_exits;
f08864b4 723 u32 nmi_window_exits;
77b4c255 724 u32 halt_exits;
f7819512 725 u32 halt_successful_poll;
62bea5bf 726 u32 halt_attempted_poll;
77b4c255
ZX
727 u32 halt_wakeup;
728 u32 request_irq_exits;
729 u32 irq_exits;
730 u32 host_state_reload;
731 u32 efer_reload;
732 u32 fpu_reload;
733 u32 insn_emulation;
734 u32 insn_emulation_fail;
f11c3a8d 735 u32 hypercalls;
fa89a817 736 u32 irq_injections;
c4abb7c9 737 u32 nmi_injections;
77b4c255 738};
ad312c7c 739
8a76d7f2
JR
740struct x86_instruction_info;
741
8fe8ab46
WA
742struct msr_data {
743 bool host_initiated;
744 u32 index;
745 u64 data;
746};
747
cb5281a5
PB
748struct kvm_lapic_irq {
749 u32 vector;
b7cb2231
PB
750 u16 delivery_mode;
751 u16 dest_mode;
752 bool level;
753 u16 trig_mode;
cb5281a5
PB
754 u32 shorthand;
755 u32 dest_id;
93bbf0b8 756 bool msi_redir_hint;
cb5281a5
PB
757};
758
ea4a5ff8
ZX
759struct kvm_x86_ops {
760 int (*cpu_has_kvm_support)(void); /* __init */
761 int (*disabled_by_bios)(void); /* __init */
13a34e06
RK
762 int (*hardware_enable)(void);
763 void (*hardware_disable)(void);
ea4a5ff8
ZX
764 void (*check_processor_compatibility)(void *rtn);
765 int (*hardware_setup)(void); /* __init */
766 void (*hardware_unsetup)(void); /* __exit */
774ead3a 767 bool (*cpu_has_accelerated_tpr)(void);
6d396b55 768 bool (*cpu_has_high_real_mode_segbase)(void);
0e851880 769 void (*cpuid_update)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
770
771 /* Create, but do not attach this VCPU */
772 struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
773 void (*vcpu_free)(struct kvm_vcpu *vcpu);
d28bc9dd 774 void (*vcpu_reset)(struct kvm_vcpu *vcpu, bool init_event);
ea4a5ff8
ZX
775
776 void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
777 void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
778 void (*vcpu_put)(struct kvm_vcpu *vcpu);
ea4a5ff8 779
c8639010 780 void (*update_db_bp_intercept)(struct kvm_vcpu *vcpu);
609e36d3 781 int (*get_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
8fe8ab46 782 int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
ea4a5ff8
ZX
783 u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
784 void (*get_segment)(struct kvm_vcpu *vcpu,
785 struct kvm_segment *var, int seg);
2e4d2653 786 int (*get_cpl)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
787 void (*set_segment)(struct kvm_vcpu *vcpu,
788 struct kvm_segment *var, int seg);
789 void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
e8467fda 790 void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
aff48baa 791 void (*decache_cr3)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
792 void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
793 void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
794 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
5e1746d6 795 int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
ea4a5ff8 796 void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
89a27f4d
GN
797 void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
798 void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
799 void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
800 void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
73aaf249
JK
801 u64 (*get_dr6)(struct kvm_vcpu *vcpu);
802 void (*set_dr6)(struct kvm_vcpu *vcpu, unsigned long value);
c77fb5fe 803 void (*sync_dirty_debug_regs)(struct kvm_vcpu *vcpu);
020df079 804 void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
5fdbf976 805 void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
ea4a5ff8
ZX
806 unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
807 void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
0fdd74f7 808 void (*fpu_activate)(struct kvm_vcpu *vcpu);
02daab21 809 void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
810
811 void (*tlb_flush)(struct kvm_vcpu *vcpu);
ea4a5ff8 812
851ba692
AK
813 void (*run)(struct kvm_vcpu *vcpu);
814 int (*handle_exit)(struct kvm_vcpu *vcpu);
ea4a5ff8 815 void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
2809f5d2 816 void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
37ccdcbe 817 u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu);
ea4a5ff8
ZX
818 void (*patch_hypercall)(struct kvm_vcpu *vcpu,
819 unsigned char *hypercall_addr);
66fd3f7f 820 void (*set_irq)(struct kvm_vcpu *vcpu);
95ba8273 821 void (*set_nmi)(struct kvm_vcpu *vcpu);
298101da 822 void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
ce7ddec4
JR
823 bool has_error_code, u32 error_code,
824 bool reinject);
b463a6f7 825 void (*cancel_injection)(struct kvm_vcpu *vcpu);
78646121 826 int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
95ba8273 827 int (*nmi_allowed)(struct kvm_vcpu *vcpu);
3cfc3092
JK
828 bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
829 void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
c9a7953f
JK
830 void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
831 void (*enable_irq_window)(struct kvm_vcpu *vcpu);
95ba8273 832 void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
d50ab6c1 833 int (*cpu_uses_apicv)(struct kvm_vcpu *vcpu);
c7c9c56c
YZ
834 void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
835 void (*hwapic_isr_update)(struct kvm *kvm, int isr);
3bb345f3 836 void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu);
8d14695f 837 void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu, bool set);
4256f43f 838 void (*set_apic_access_page_addr)(struct kvm_vcpu *vcpu, hpa_t hpa);
a20ed54d
YZ
839 void (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu, int vector);
840 void (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);
ea4a5ff8 841 int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
67253af5 842 int (*get_tdp_level)(void);
4b12f0de 843 u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
17cc3935 844 int (*get_lpage_level)(void);
4e47c7a6 845 bool (*rdtscp_supported)(void);
ad756a16 846 bool (*invpcid_supported)(void);
f1e2b260 847 void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment, bool host);
344f414f 848
1c97f0a0
JR
849 void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
850
d4330ef2
JR
851 void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
852
f5f48ee1
SY
853 bool (*has_wbinvd_exit)(void);
854
cc578287 855 void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale);
ba904635 856 u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu);
99e3e30a
ZA
857 void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
858
857e4099 859 u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc);
886b470c 860 u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu, u64 host_tsc);
857e4099 861
586f9607 862 void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
8a76d7f2
JR
863
864 int (*check_intercept)(struct kvm_vcpu *vcpu,
865 struct x86_instruction_info *info,
866 enum x86_intercept_stage stage);
a547c6db 867 void (*handle_external_intr)(struct kvm_vcpu *vcpu);
da8999d3 868 bool (*mpx_supported)(void);
55412b2e 869 bool (*xsaves_supported)(void);
b6b8a145
JK
870
871 int (*check_nested_events)(struct kvm_vcpu *vcpu, bool external_intr);
ae97a3b8
RK
872
873 void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
88178fd4
KH
874
875 /*
876 * Arch-specific dirty logging hooks. These hooks are only supposed to
877 * be valid if the specific arch has hardware-accelerated dirty logging
878 * mechanism. Currently only for PML on VMX.
879 *
880 * - slot_enable_log_dirty:
881 * called when enabling log dirty mode for the slot.
882 * - slot_disable_log_dirty:
883 * called when disabling log dirty mode for the slot.
884 * also called when slot is created with log dirty disabled.
885 * - flush_log_dirty:
886 * called before reporting dirty_bitmap to userspace.
887 * - enable_log_dirty_pt_masked:
888 * called when reenabling log dirty for the GFNs in the mask after
889 * corresponding bits are cleared in slot->dirty_bitmap.
890 */
891 void (*slot_enable_log_dirty)(struct kvm *kvm,
892 struct kvm_memory_slot *slot);
893 void (*slot_disable_log_dirty)(struct kvm *kvm,
894 struct kvm_memory_slot *slot);
895 void (*flush_log_dirty)(struct kvm *kvm);
896 void (*enable_log_dirty_pt_masked)(struct kvm *kvm,
897 struct kvm_memory_slot *slot,
898 gfn_t offset, unsigned long mask);
25462f7f
WH
899 /* pmu operations of sub-arch */
900 const struct kvm_pmu_ops *pmu_ops;
efc64404 901
bf9f6ac8
FW
902 /*
903 * Architecture specific hooks for vCPU blocking due to
904 * HLT instruction.
905 * Returns for .pre_block():
906 * - 0 means continue to block the vCPU.
907 * - 1 means we cannot block the vCPU since some event
908 * happens during this period, such as, 'ON' bit in
909 * posted-interrupts descriptor is set.
910 */
911 int (*pre_block)(struct kvm_vcpu *vcpu);
912 void (*post_block)(struct kvm_vcpu *vcpu);
efc64404
FW
913 int (*update_pi_irte)(struct kvm *kvm, unsigned int host_irq,
914 uint32_t guest_irq, bool set);
ea4a5ff8
ZX
915};
916
af585b92 917struct kvm_arch_async_pf {
7c90705b 918 u32 token;
af585b92 919 gfn_t gfn;
fb67e14f 920 unsigned long cr3;
c4806acd 921 bool direct_map;
af585b92
GN
922};
923
97896d04
ZX
924extern struct kvm_x86_ops *kvm_x86_ops;
925
f1e2b260
MT
926static inline void adjust_tsc_offset_guest(struct kvm_vcpu *vcpu,
927 s64 adjustment)
928{
929 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, false);
930}
931
932static inline void adjust_tsc_offset_host(struct kvm_vcpu *vcpu, s64 adjustment)
933{
934 kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, true);
935}
936
54f1585a
ZX
937int kvm_mmu_module_init(void);
938void kvm_mmu_module_exit(void);
939
940void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
941int kvm_mmu_create(struct kvm_vcpu *vcpu);
8a3c1a33 942void kvm_mmu_setup(struct kvm_vcpu *vcpu);
7b52345e 943void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
4b12f0de 944 u64 dirty_mask, u64 nx_mask, u64 x_mask);
54f1585a 945
8a3c1a33 946void kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
1c91cad4
KH
947void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
948 struct kvm_memory_slot *memslot);
3ea3b7fa 949void kvm_mmu_zap_collapsible_sptes(struct kvm *kvm,
f36f3f28 950 const struct kvm_memory_slot *memslot);
f4b4b180
KH
951void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
952 struct kvm_memory_slot *memslot);
953void kvm_mmu_slot_largepage_remove_write_access(struct kvm *kvm,
954 struct kvm_memory_slot *memslot);
955void kvm_mmu_slot_set_dirty(struct kvm *kvm,
956 struct kvm_memory_slot *memslot);
957void kvm_mmu_clear_dirty_pt_masked(struct kvm *kvm,
958 struct kvm_memory_slot *slot,
959 gfn_t gfn_offset, unsigned long mask);
54f1585a 960void kvm_mmu_zap_all(struct kvm *kvm);
54bf36aa 961void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm, struct kvm_memslots *slots);
3ad82a7e 962unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
54f1585a
ZX
963void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
964
ff03a073 965int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
cc4b6871 966
3200f405 967int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
9f811285 968 const void *val, int bytes);
2f333bcb 969
6ef768fa
PB
970struct kvm_irq_mask_notifier {
971 void (*func)(struct kvm_irq_mask_notifier *kimn, bool masked);
972 int irq;
973 struct hlist_node link;
974};
975
976void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
977 struct kvm_irq_mask_notifier *kimn);
978void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
979 struct kvm_irq_mask_notifier *kimn);
980void kvm_fire_mask_notifiers(struct kvm *kvm, unsigned irqchip, unsigned pin,
981 bool mask);
982
2f333bcb 983extern bool tdp_enabled;
9f811285 984
a3e06bbe
LJ
985u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
986
92a1f12d
JR
987/* control of guest tsc rate supported? */
988extern bool kvm_has_tsc_control;
989/* minimum supported tsc_khz for guests */
990extern u32 kvm_min_guest_tsc_khz;
991/* maximum supported tsc_khz for guests */
992extern u32 kvm_max_guest_tsc_khz;
993
54f1585a 994enum emulation_result {
ac0a48c3
PB
995 EMULATE_DONE, /* no further processing */
996 EMULATE_USER_EXIT, /* kvm_run ready for userspace exit */
54f1585a
ZX
997 EMULATE_FAIL, /* can't emulate this instruction */
998};
999
571008da
SY
1000#define EMULTYPE_NO_DECODE (1 << 0)
1001#define EMULTYPE_TRAP_UD (1 << 1)
ba8afb6b 1002#define EMULTYPE_SKIP (1 << 2)
1cb3f3ae 1003#define EMULTYPE_RETRY (1 << 3)
991eebf9 1004#define EMULTYPE_NO_REEXECUTE (1 << 4)
dc25e89e
AP
1005int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
1006 int emulation_type, void *insn, int insn_len);
51d8b661
AP
1007
1008static inline int emulate_instruction(struct kvm_vcpu *vcpu,
1009 int emulation_type)
1010{
dc25e89e 1011 return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
51d8b661
AP
1012}
1013
f2b4b7dd 1014void kvm_enable_efer_bits(u64);
384bb783 1015bool kvm_valid_efer(struct kvm_vcpu *vcpu, u64 efer);
609e36d3 1016int kvm_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
8fe8ab46 1017int kvm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
54f1585a
ZX
1018
1019struct x86_emulate_ctxt;
1020
cf8f70bf 1021int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
54f1585a
ZX
1022void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
1023int kvm_emulate_halt(struct kvm_vcpu *vcpu);
5cb56059 1024int kvm_vcpu_halt(struct kvm_vcpu *vcpu);
f5f48ee1 1025int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
54f1585a 1026
3e6e0aab 1027void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
c697518a 1028int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
2b4a273b 1029void kvm_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
3e6e0aab 1030
7f3d35fd
KW
1031int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
1032 int reason, bool has_error_code, u32 error_code);
37817f29 1033
49a9b07e 1034int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
2390218b 1035int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
a83b29c6 1036int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
eea1cff9 1037int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
020df079
GN
1038int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
1039int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
2d3ad1f4
AK
1040unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
1041void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
54f1585a 1042void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
2acf923e 1043int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
54f1585a 1044
609e36d3 1045int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
8fe8ab46 1046int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
54f1585a 1047
91586a3b
JK
1048unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
1049void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
022cd0e8 1050bool kvm_rdpmc(struct kvm_vcpu *vcpu);
91586a3b 1051
298101da
AK
1052void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1053void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
ce7ddec4
JR
1054void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
1055void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
6389ee94 1056void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
ec92fe44
JR
1057int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
1058 gfn_t gfn, void *data, int offset, int len,
1059 u32 access);
0a79b009 1060bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
16f8a6f9 1061bool kvm_require_dr(struct kvm_vcpu *vcpu, int dr);
298101da 1062
1a577b72
MT
1063static inline int __kvm_irq_line_state(unsigned long *irq_state,
1064 int irq_source_id, int level)
1065{
1066 /* Logical OR for level trig interrupt */
1067 if (level)
1068 __set_bit(irq_source_id, irq_state);
1069 else
1070 __clear_bit(irq_source_id, irq_state);
1071
1072 return !!(*irq_state);
1073}
1074
1075int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
1076void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
3de42dc0 1077
3419ffc8
SY
1078void kvm_inject_nmi(struct kvm_vcpu *vcpu);
1079
54f1585a 1080void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
f57f2ef5 1081 const u8 *new, int bytes);
1cb3f3ae 1082int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
54f1585a
ZX
1083int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
1084void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
1085int kvm_mmu_load(struct kvm_vcpu *vcpu);
1086void kvm_mmu_unload(struct kvm_vcpu *vcpu);
0ba73cda 1087void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
54987b7a
PB
1088gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1089 struct x86_exception *exception);
ab9ae313
AK
1090gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
1091 struct x86_exception *exception);
1092gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
1093 struct x86_exception *exception);
1094gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
1095 struct x86_exception *exception);
1096gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
1097 struct x86_exception *exception);
54f1585a
ZX
1098
1099int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
1100
dc25e89e
AP
1101int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
1102 void *insn, int insn_len);
a7052897 1103void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
d8d173da 1104void kvm_mmu_new_cr3(struct kvm_vcpu *vcpu);
34c16eec 1105
18552672 1106void kvm_enable_tdp(void);
5f4cb662 1107void kvm_disable_tdp(void);
18552672 1108
54987b7a
PB
1109static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access,
1110 struct x86_exception *exception)
e459e322
XG
1111{
1112 return gpa;
1113}
1114
ec6d273d
ZX
1115static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
1116{
1117 struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
1118
1119 return (struct kvm_mmu_page *)page_private(page);
1120}
1121
d6e88aec 1122static inline u16 kvm_read_ldt(void)
ec6d273d
ZX
1123{
1124 u16 ldt;
1125 asm("sldt %0" : "=g"(ldt));
1126 return ldt;
1127}
1128
d6e88aec 1129static inline void kvm_load_ldt(u16 sel)
ec6d273d
ZX
1130{
1131 asm("lldt %0" : : "rm"(sel));
1132}
ec6d273d 1133
ec6d273d
ZX
1134#ifdef CONFIG_X86_64
1135static inline unsigned long read_msr(unsigned long msr)
1136{
1137 u64 value;
1138
1139 rdmsrl(msr, value);
1140 return value;
1141}
1142#endif
1143
ec6d273d
ZX
1144static inline u32 get_rdx_init_val(void)
1145{
1146 return 0x600; /* P6 family */
1147}
1148
c1a5d4f9
AK
1149static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
1150{
1151 kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
1152}
1153
854e8bb1
NA
1154static inline u64 get_canonical(u64 la)
1155{
1156 return ((int64_t)la << 16) >> 16;
1157}
1158
1159static inline bool is_noncanonical_address(u64 la)
1160{
1161#ifdef CONFIG_X86_64
1162 return get_canonical(la) != la;
1163#else
1164 return false;
1165#endif
1166}
1167
ec6d273d
ZX
1168#define TSS_IOPB_BASE_OFFSET 0x66
1169#define TSS_BASE_SIZE 0x68
1170#define TSS_IOPB_SIZE (65536 / 8)
1171#define TSS_REDIRECTION_SIZE (256 / 8)
7d76b4d3
JP
1172#define RMODE_TSS_SIZE \
1173 (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
53e0aa7b 1174
37817f29
IE
1175enum {
1176 TASK_SWITCH_CALL = 0,
1177 TASK_SWITCH_IRET = 1,
1178 TASK_SWITCH_JMP = 2,
1179 TASK_SWITCH_GATE = 3,
1180};
1181
1371d904 1182#define HF_GIF_MASK (1 << 0)
3d6368ef
AG
1183#define HF_HIF_MASK (1 << 1)
1184#define HF_VINTR_MASK (1 << 2)
95ba8273 1185#define HF_NMI_MASK (1 << 3)
44c11430 1186#define HF_IRET_MASK (1 << 4)
ec9e60b2 1187#define HF_GUEST_MASK (1 << 5) /* VCPU is in guest-mode */
f077825a
PB
1188#define HF_SMM_MASK (1 << 6)
1189#define HF_SMM_INSIDE_NMI_MASK (1 << 7)
1371d904 1190
699023e2
PB
1191#define __KVM_VCPU_MULTIPLE_ADDRESS_SPACE
1192#define KVM_ADDRESS_SPACE_NUM 2
1193
1194#define kvm_arch_vcpu_memslots_id(vcpu) ((vcpu)->arch.hflags & HF_SMM_MASK ? 1 : 0)
1195#define kvm_memslots_for_spte_role(kvm, role) __kvm_memslots(kvm, (role).smm)
1371d904 1196
4ecac3fd
AK
1197/*
1198 * Hardware virtualization extension instructions may fault if a
1199 * reboot turns off virtualization while processes are running.
1200 * Trap the fault and ignore the instruction if that happens.
1201 */
b7c4145b 1202asmlinkage void kvm_spurious_fault(void);
4ecac3fd 1203
5e520e62 1204#define ____kvm_handle_fault_on_reboot(insn, cleanup_insn) \
4ecac3fd 1205 "666: " insn "\n\t" \
b7c4145b 1206 "668: \n\t" \
18b13e54 1207 ".pushsection .fixup, \"ax\" \n" \
4ecac3fd 1208 "667: \n\t" \
5e520e62 1209 cleanup_insn "\n\t" \
b7c4145b
AK
1210 "cmpb $0, kvm_rebooting \n\t" \
1211 "jne 668b \n\t" \
8ceed347 1212 __ASM_SIZE(push) " $666b \n\t" \
b7c4145b 1213 "call kvm_spurious_fault \n\t" \
4ecac3fd 1214 ".popsection \n\t" \
3ee89722 1215 _ASM_EXTABLE(666b, 667b)
4ecac3fd 1216
5e520e62
AK
1217#define __kvm_handle_fault_on_reboot(insn) \
1218 ____kvm_handle_fault_on_reboot(insn, "")
1219
e930bffe
AA
1220#define KVM_ARCH_WANT_MMU_NOTIFIER
1221int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
b3ae2096 1222int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
57128468 1223int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
8ee53820 1224int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
3da0dd43 1225void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
c7c9c56c 1226int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
a1b37100
GN
1227int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1228int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
0b71785d 1229int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
d28bc9dd 1230void kvm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event);
4256f43f 1231void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu);
fe71557a
TC
1232void kvm_arch_mmu_notifier_invalidate_page(struct kvm *kvm,
1233 unsigned long address);
e930bffe 1234
18863bdd 1235void kvm_define_shared_msr(unsigned index, u32 msr);
8b3c3104 1236int kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
18863bdd 1237
82b32774 1238unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
f92653ee
JK
1239bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1240
af585b92
GN
1241void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1242 struct kvm_async_pf *work);
1243void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1244 struct kvm_async_pf *work);
56028d08
GN
1245void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1246 struct kvm_async_pf *work);
7c90705b 1247bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
af585b92
GN
1248extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1249
db8fcefa
AP
1250void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1251
f5132b01
GN
1252int kvm_is_in_guest(void);
1253
1d8007bd
PB
1254int __x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa, u32 size);
1255int x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa, u32 size);
d71ba788
PB
1256bool kvm_vcpu_is_reset_bsp(struct kvm_vcpu *vcpu);
1257bool kvm_vcpu_is_bsp(struct kvm_vcpu *vcpu);
f5132b01 1258
8feb4a04
FW
1259bool kvm_intr_is_single_vcpu(struct kvm *kvm, struct kvm_lapic_irq *irq,
1260 struct kvm_vcpu **dest_vcpu);
1261
d84f1e07
FW
1262void kvm_set_msi_irq(struct kvm_kernel_irq_routing_entry *e,
1263 struct kvm_lapic_irq *irq);
1965aae3 1264#endif /* _ASM_X86_KVM_HOST_H */