-
Notifications
You must be signed in to change notification settings - Fork 0
/
exit.c
2760 lines (2391 loc) · 76.9 KB
/
exit.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/*
* ksm - a really simple and fast x64 hypervisor
* Copyright (C) 2016, 2017 Ahmed Samy <[email protected]>
*
* This file handles a VM-exit from guest, if any error occurs,
* it either:
* 1) crashes the system
* 2) injects an exception into guest
* Otherwise it returns execution to guest.
*
* This program is free software; you can redistribute it and/or modify it
* under the terms and conditions of the GNU General Public License,
* version 2, as published by the Free Software Foundation.
*
* This program is distributed in the hope it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
* more details.
*
* You should have received a copy of the GNU General Public License along with
* this program; If not, see <http://www.gnu.org/licenses/>.
*/
#ifdef __linux__
#include <linux/kernel.h>
#else
#include <ntddk.h>
#include <intrin.h>
#endif
#include "ksm.h"
#ifdef NESTED_VMX
/* FIXME: Support these! */
static const u32 nested_unsupported_primary = CPU_BASED_MOV_DR_EXITING;
static const u32 nested_unsupported_secondary = SECONDARY_EXEC_ENABLE_VMFUNC | SECONDARY_EXEC_DESC_TABLE_EXITING;
static inline bool __nested_vmcs_write(uintptr_t vmcs, u32 field, u64 value)
{
u64 *s = (u64 *)vmcs;
u16 off = field_offset(field);
u64 v = s[off];
switch (field_width(field)) {
case FIELD_U16:
v = value & 0xFFFF;
break;
case FIELD_U32:
v = value & 0xFFFFFFFF;
break;
case FIELD_U64:
if (field & 1) { /* _HIGH */
v &= 0xFFFFFFFF;
v |= value << 32;
break;
}
/* fallthrough */
default:
v = value;
break;
}
s[off] = v;
return true;
}
static inline bool __nested_vmcs_write64(uintptr_t vmcs, u32 field, u64 value)
{
#if 0
return __nested_vmcs_write(vmcs, field, (u32)value) &&
__nested_vmcs_write(vmcs, field + 1, (u32)(value >> 32));
#else
return __nested_vmcs_write(vmcs, field, value);
#endif
}
static inline bool nested_vmcs_write(uintptr_t vmcs, u32 field, u64 value)
{
if (!field_supported(field))
return false;
switch (field) {
case CPU_BASED_VM_EXEC_CONTROL:
if (value & nested_unsupported_primary)
return false;
break;
case SECONDARY_VM_EXEC_CONTROL:
if (value & nested_unsupported_secondary)
return false;
break;
}
return __nested_vmcs_write(vmcs, field, value);
}
static inline u64 __nested_vmcs_read(uintptr_t vmcs, u32 field)
{
u64 *s = (u64 *)vmcs;
u64 v = s[field_offset(field)];
switch (field_width(field)) {
case FIELD_U16:
v &= 0xFFFF;
break;
case FIELD_U32:
v &= 0xFFFFFFFF;
break;
case FIELD_U64:
if (field & 1) /* _HIGH */
v >>= 32;
break;
case FIELD_NATURAL:
default:
break;
}
return v;
}
static inline u64 __nested_vmcs_read64(uintptr_t vmcs, u32 field)
{
#if 0
return __nested_vmcs_read(vmcs, field) |
__nested_vmcs_read(vmcs, field + 1) << 32;
#else
return __nested_vmcs_read(vmcs, field);
#endif
}
static inline u32 __nested_vmcs_read32(uintptr_t vmcs, u32 field)
{
return (u32)__nested_vmcs_read(vmcs, field);
}
static inline u16 __nested_vmcs_read16(uintptr_t vmcs, u32 field)
{
return (u16)__nested_vmcs_read(vmcs, field);
}
static inline bool nested_vmcs_read(uintptr_t vmcs, u32 field, u64 *val)
{
if (!field_supported(field))
return false;
*val = __nested_vmcs_read(vmcs, field);
return true;
}
#endif
static inline int vcpu_read_cpl(void)
{
u32 ar = vmcs_read32(GUEST_SS_AR_BYTES);
return VMX_AR_DPL(ar);
}
static inline bool vcpu_probe_cpl(int required)
{
return vcpu_read_cpl() <= required;
}
typedef enum {
EXCEPTION_BENIGN,
EXCEPTION_CONTRIBUTORY,
EXCEPTION_PAGE_FAULT,
} except_class_t;
static inline except_class_t exception_class(u8 vec)
{
switch (vec) {
case X86_TRAP_PF:
return EXCEPTION_PAGE_FAULT;
case X86_TRAP_DE:
case X86_TRAP_TS:
case X86_TRAP_NP:
case X86_TRAP_SS:
case X86_TRAP_GP:
return EXCEPTION_CONTRIBUTORY;
}
return EXCEPTION_BENIGN;
}
static inline void vcpu_pack_irq(struct pending_irq *pirq, u32 instr_len, u16 intr_type,
u8 vector, bool has_err, u32 ec)
{
u32 irq = vector | intr_type | INTR_INFO_VALID_MASK;
if (has_err)
irq |= INTR_INFO_DELIVER_CODE_MASK;
pirq->pending = true;
pirq->err = ec;
pirq->instr_len = instr_len;
pirq->bits = irq & ~INTR_INFO_RESVD_BITS_MASK;
}
static inline void vcpu_inject_irq(struct vcpu *vcpu, u32 instr_len, u16 intr_type,
u8 vector, bool has_err, u32 ec)
{
/*
* Queue the IRQ, no injection happens here.
* In case we have contributory exceptions that follow, then
* we overwrite the previous with the appropriate IRQ.
*/
struct pending_irq *pirq = &vcpu->irq;
if (pirq->pending) {
u8 prev_vec = (u8)pirq->bits;
BREAK_ON(prev_vec == X86_TRAP_DF); /* FIXME: Triple fault */
except_class_t lhs = exception_class(prev_vec);
except_class_t rhs = exception_class(vector);
if ((lhs == EXCEPTION_CONTRIBUTORY && rhs == EXCEPTION_CONTRIBUTORY) ||
(lhs == EXCEPTION_PAGE_FAULT && rhs != EXCEPTION_BENIGN))
return vcpu_pack_irq(pirq, instr_len, INTR_TYPE_HARD_EXCEPTION,
X86_TRAP_DF, true, 0);
}
return vcpu_pack_irq(pirq, instr_len, intr_type, vector, has_err, ec);
}
static inline void vcpu_inject_hardirq_noerr(struct vcpu *vcpu, u8 vector)
{
return vcpu_inject_irq(vcpu, vmcs_read32(VM_EXIT_INSTRUCTION_LEN),
INTR_TYPE_HARD_EXCEPTION, vector, false, 0);
}
static inline void vcpu_inject_hardirq(struct vcpu *vcpu, u8 vector, u32 err)
{
return vcpu_inject_irq(vcpu, vmcs_read32(VM_EXIT_INSTRUCTION_LEN),
INTR_TYPE_HARD_EXCEPTION, vector, true, err);
}
static inline void vcpu_inject_pf(struct vcpu *vcpu, u64 gla, u32 ec)
{
__writecr2(gla);
return vcpu_inject_irq(vcpu, 0, INTR_TYPE_HARD_EXCEPTION,
X86_TRAP_PF, true, ec);
}
static inline bool vcpu_inject_gp_if(struct vcpu *vcpu, bool cond)
{
if (cond)
vcpu_inject_hardirq(vcpu, X86_TRAP_GP, 0);
return cond;
}
static inline void vcpu_advance_rip(struct vcpu *vcpu)
{
if (vcpu->eflags & X86_EFLAGS_TF) {
vcpu_inject_hardirq_noerr(vcpu, X86_TRAP_DB);
if (vcpu_probe_cpl(0)) {
__writedr(6, __readdr(6) | DR6_BS | DR6_RTM);
__writedr(7, __readdr(7) & ~DR7_GD);
u64 dbg = vmcs_read64(GUEST_IA32_DEBUGCTL);
vmcs_write64(GUEST_IA32_DEBUGCTL, dbg & ~DEBUGCTLMSR_LBR);
}
}
u32 instr_len = vmcs_read32(VM_EXIT_INSTRUCTION_LEN);
vmcs_write(GUEST_RIP, vcpu->ip + instr_len);
size_t interruptibility = vmcs_read32(GUEST_INTERRUPTIBILITY_INFO);
vmcs_write32(GUEST_INTERRUPTIBILITY_INFO,
interruptibility & ~(GUEST_INTR_STATE_MOV_SS | GUEST_INTR_STATE_STI));
}
#ifdef NESTED_VMX
static inline bool nested_inject_ve(struct vcpu *vcpu)
{
struct nested_vcpu *nested = &vcpu->nested_vcpu;
uintptr_t vmcs = nested->vmcs;
/*
* First see if nested opt'd in for #VE handling, then
* write the required fields to whatever ve info address points
* to then inject #VE as a normal IDT injection.
*/
u32 secondary_ctl = __nested_vmcs_read32(vmcs, SECONDARY_VM_EXEC_CONTROL);
if (!(secondary_ctl & SECONDARY_EXEC_ENABLE_VE))
return false;
u64 ve_info_addr = __nested_vmcs_read64(vmcs, VE_INFO_ADDRESS);
if (!page_aligned(ve_info_addr))
return false;
u64 hpa;
if (!gpa_to_hpa(vcpu, ve_info_addr, &hpa))
return false;
struct ve_except_info *info = mm_remap(hpa, PAGE_SIZE);
if (!info)
return false;
if (info->except_mask == 0) {
KSM_DEBUG("Trying to inject #VE but guest opted-out.\n");
mm_unmap(info, PAGE_SIZE);
return false;
}
/* Set appropriate data in VE structure */
info->eptp = __nested_vmcs_read16(vmcs, EPTP_INDEX);
info->except_mask = (u32)~0UL;
info->reason = EXIT_REASON_EPT_VIOLATION;
info->gpa = vmcs_read64(GUEST_PHYSICAL_ADDRESS);
info->gla = vmcs_read(GUEST_LINEAR_ADDRESS);
info->exit = vmcs_read(EXIT_QUALIFICATION);
mm_unmap(info, PAGE_SIZE);
vcpu_inject_hardirq_noerr(vcpu, X86_TRAP_VE);
return true;
}
#endif
static inline bool vcpu_parse_vmx_addr(struct vcpu *vcpu, u64 disp, u64 inst, u64 *out)
{
/*
* Register access is handled before this call or not
* supported at all.
*
* Register access is only valid in those cases:
* 1) vmwrite
* 2) vmread
*
* Other cases such as vmptrld, vmptrst, vmclear, vmon, etc, must
* be passed through a memory reference, e.g. stack, something
* like:
* pushq phys_add
* vmon 0(%rsp)
*
* or even:
* vmon %cs:some_global_phys_addr
*
* C:
* u64 phys = __pa(vmon);
* __vmxon(&phys);
*
* See also vcpu.c.
*
* So we need to first get the address, then dereference to get the
* actual physical address, note that dereferencing does not happen
* here, here we only validate the address and return the virtual address.
*
* Dereferencing happens in:
* vcpu_read_vmx_addr()
* vcpu_write_vmx_addr()
*/
if ((inst >> 10) & 1) {
vcpu_inject_hardirq_noerr(vcpu, X86_TRAP_UD);
return false;
}
u64 seg_offset = (inst >> 15) & 7;
if (vcpu_inject_gp_if(vcpu, seg_offset > 5))
return false;
uintptr_t base = 0;
if (!((inst >> 27) & 1))
base = ksm_read_reg(vcpu, (inst >> 23) & 15);
uintptr_t index = 0;
if (!((inst >> 22) & 1))
index = ksm_read_reg(vcpu, (inst >> 18) & 15) << (inst & 3);
uintptr_t gva = vmcs_read(GUEST_ES_BASE + (seg_offset << 1)) +
base + index + disp;
if (((inst >> 7) & 7) == 1)
gva &= 0xFFFFFFFF;
if (!is_canonical_addr(gva)) {
vcpu_inject_hardirq(vcpu, seg_offset == 2 ? X86_TRAP_SS : X86_TRAP_GP, 0);
return false;
}
*out = gva;
return true;
}
static bool vcpu_nop(struct vcpu *vcpu)
{
VCPU_TRACER_START();
KSM_DEBUG_RAW("you need to handle the corresponding VM-exit for the handler you set.\n");
KSM_PANIC(KSM_PANIC_CODE, VCPU_BUG_UNHANDLED, vcpu->curr_handler, vcpu->prev_handler);
return false;
}
static bool vcpu_handle_except_nmi(struct vcpu *vcpu)
{
VCPU_TRACER_START();
u32 intr_info = vmcs_read32(VM_EXIT_INTR_INFO);
u16 intr_type = intr_info & INTR_INFO_INTR_TYPE_MASK;
u8 vector = intr_info & INTR_INFO_VECTOR_MASK;
u32 instr_len = 0;
if (intr_type & INTR_TYPE_HARD_EXCEPTION && vector == X86_TRAP_PF)
__writecr2(vmcs_read(EXIT_QUALIFICATION));
else
instr_len = vmcs_read32(VM_EXIT_INSTRUCTION_LEN);
bool has_err = intr_info & INTR_INFO_DELIVER_CODE_MASK;
u32 err = vmcs_read32(IDT_VECTORING_ERROR_CODE);
vcpu_inject_irq(vcpu, instr_len, intr_type, vector, has_err, err);
VCPU_TRACER_END();
return true;
}
static bool vcpu_handle_triplefault(struct vcpu *vcpu)
{
/* A triple fault occured during handling of a double fault in guest, bug check. */
VCPU_TRACER_START();
KSM_PANIC(KSM_PANIC_CODE, VCPU_TRIPLEFAULT, vcpu->curr_handler, vcpu->prev_handler);
VCPU_TRACER_END();
return false;
}
static bool vcpu_handle_taskswitch(struct vcpu *vcpu)
{
/* Not really called */
VCPU_TRACER_START();
uintptr_t exit = vmcs_read(EXIT_QUALIFICATION);
u16 selector = (u16)exit;
u8 src = (exit >> 30) & 3;
const char *name;
switch (src) {
case 0:
name = "call";
break;
case 1:
name = "iret";
break;
case 2:
name = "jmp";
break;
case 3:
default:
name = "task gate";
break;
}
const char *table = "gdt";
if (selector & 4)
table = "ldt";
KSM_DEBUG("switching through %s (selector: %d => table: %s index: %d)\n",
name, selector, table, selector >> 3);
VCPU_TRACER_END();
return true;
}
static bool vcpu_handle_cpuid(struct vcpu *vcpu)
{
VCPU_TRACER_START();
int cpuid[4];
int func = ksm_read_reg32(vcpu, STACK_REG_AX);
int subf = ksm_read_reg32(vcpu, STACK_REG_CX);
__cpuidex(cpuid, func, subf);
#ifndef NESTED_VMX
if (func == 1)
cpuid[2] &= ~(1 << (X86_FEATURE_VMX & 31));
#endif
ksm_write_reg32(vcpu, STACK_REG_AX, cpuid[0]);
ksm_write_reg32(vcpu, STACK_REG_BX, cpuid[1]);
ksm_write_reg32(vcpu, STACK_REG_CX, cpuid[2]);
ksm_write_reg32(vcpu, STACK_REG_DX, cpuid[3]);
vcpu_advance_rip(vcpu);
VCPU_TRACER_END();
return true;
}
static bool vcpu_handle_hlt(struct vcpu *vcpu)
{
VCPU_TRACER_START();
__halt();
vcpu_advance_rip(vcpu);
VCPU_TRACER_END();
return true;
}
static bool vcpu_handle_invd(struct vcpu *vcpu)
{
VCPU_TRACER_START();
__invd();
vcpu_advance_rip(vcpu);
VCPU_TRACER_END();
return true;
}
static bool vcpu_handle_invlpg(struct vcpu *vcpu)
{
VCPU_TRACER_START();
uintptr_t addr = vmcs_read(EXIT_QUALIFICATION);
__invlpg((void *)addr);
__invvpid_addr(vpid_nr(), addr);
vcpu_advance_rip(vcpu);
VCPU_TRACER_END();
return true;
}
static bool vcpu_handle_rdtsc(struct vcpu *vcpu)
{
VCPU_TRACER_START();
u64 tsc = __rdtsc();
ksm_write_reg32(vcpu, STACK_REG_AX, (u32)tsc);
ksm_write_reg32(vcpu, STACK_REG_DX, tsc >> 32);
vcpu_advance_rip(vcpu);
VCPU_TRACER_END();
return true;
}
static bool vcpu_handle_vmfunc(struct vcpu *vcpu)
{
/*
* VM functions do not cause VM exit unless:
* 1) funciton is not supported
* 2) EPTP index is too high.
*/
VCPU_TRACER_START();
KSM_DEBUG("vmfunc caused VM-exit! func is %d eptp index is %d\n",
ksm_read_reg32(vcpu, STACK_REG_AX), ksm_read_reg32(vcpu, STACK_REG_CX));
vcpu_inject_hardirq_noerr(vcpu, X86_TRAP_UD);
vcpu_advance_rip(vcpu);
VCPU_TRACER_END()
return true;
}
#ifdef ENABLE_PML
static bool vcpu_dump_pml(struct vcpu *vcpu)
{
/* CPU _decrements_ PML index (i.e. from 511 to 0 then overflows to FFFF),
* make sure we don't have an empty table... */
u16 pml_index = vmcs_read16(GUEST_PML_INDEX);
if (pml_index == PML_MAX_ENTRIES - 1)
return false;
/* PML index always points to next available PML entry. */
if (pml_index >= PML_MAX_ENTRIES)
pml_index = 0;
else
pml_index++;
/* Dump it... */
struct ept *ept = &vcpu->ept;
u16 eptp = vcpu_eptp_idx(vcpu);
for (; pml_index < PML_MAX_ENTRIES; ++pml_index) {
/* CPU guarantees that the lower 12 bits (the offset) are always 0. */
u64 gpa = *((u64 *)vcpu->pml + pml_index);
u64 gva = (u64)__va(gpa);
KSM_DEBUG("On PML %d: GPA %p GVA %p\n", pml_index, gpa, gva);
/* Reset AD bits now otherwise we probably won't get this page again */
u64 *epte = ept_pte(EPT4(ept, eptp), gpa);
*epte &= ~(EPT_ACCESSED | EPT_DIRTY);
}
/* Reset the PML index now... */
vmcs_write16(GUEST_PML_INDEX, pml_index);
/* We're done here */
KSM_DEBUG_RAW("PML dump done\n");
/* We definitely modified AD bits */
__invept_all();
return true;
}
#endif
static bool vcpu_handle_pml_full(struct vcpu *vcpu)
{
#ifdef ENABLE_PML
/* Page Modification Log is now full, dump it. */
KSM_DEBUG_RAW("PML full\n");
return vcpu_dump_pml(vcpu);
#else
KSM_PANIC(KSM_PANIC_CODE, VCPU_BUG_UNHANDLED, 0xDEAFDEAF, 0xBAADF00D);
return false;
#endif
}
static inline void vcpu_vm_succeed(struct vcpu *vcpu)
{
vcpu->eflags &= ~(X86_EFLAGS_CF | X86_EFLAGS_PF | X86_EFLAGS_AF |
X86_EFLAGS_ZF | X86_EFLAGS_SF | X86_EFLAGS_OF);
}
static inline void vcpu_vm_fail_invalid(struct vcpu *vcpu)
{
vcpu->eflags |= X86_EFLAGS_CF;
vcpu->eflags &= ~(X86_EFLAGS_PF | X86_EFLAGS_AF | X86_EFLAGS_ZF | X86_EFLAGS_SF | X86_EFLAGS_OF);
}
#ifdef NESTED_VMX
static inline void vcpu_vm_fail_valid(struct vcpu *vcpu, size_t err)
{
struct nested_vcpu *nested = &vcpu->nested_vcpu;
if (nested_has_vmcs(nested))
__nested_vmcs_write(nested->vmcs, VM_INSTRUCTION_ERROR, err);
vcpu->eflags |= X86_EFLAGS_ZF;
vcpu->eflags &= ~(X86_EFLAGS_CF | X86_EFLAGS_PF | X86_EFLAGS_AF | X86_EFLAGS_SF | X86_EFLAGS_OF);
}
#endif
static inline void vcpu_adjust_rflags(struct vcpu *vcpu, bool success)
{
if (success)
return vcpu_vm_succeed(vcpu);
return vcpu_vm_fail_invalid(vcpu);
}
static inline void vcpu_do_exit(struct vcpu *vcpu)
{
/* Fix GDT */
struct gdtr gdt = {
.limit = (u16)vmcs_read32(GUEST_GDTR_LIMIT),
.base = vmcs_read(GUEST_GDTR_BASE),
};
__lgdt(&gdt);
/* Fix IDT (restore whatever guest last loaded...) */
__lidt(&vcpu->g_idt);
uintptr_t ret = vcpu->ip + vmcs_read32(VM_EXIT_INSTRUCTION_LEN);
vcpu_vm_succeed(vcpu);
uintptr_t cr3 = vmcs_read(GUEST_CR3);
__writecr3(cr3);
/* See __vmx_entrypoint in assembly on how this is used. */
ksm_write_reg(vcpu, STACK_REG_CX, ret);
ksm_write_reg(vcpu, STACK_REG_DX, ksm_read_reg(vcpu, STACK_REG_SP));
ksm_write_reg(vcpu, STACK_REG_AX, vcpu->eflags);
}
#ifdef EPAGE_HOOK
static bool vcpu_handle_hook(struct vcpu *vcpu, struct epage_info *h)
{
KSM_DEBUG("page hook request for %p => %p (%p)\n", h->dpa, h->cpa, h->c_va);
ksm_handle_epage(vcpu, h);
return true;
}
static inline bool vcpu_handle_unhook(struct vcpu *vcpu, uintptr_t dpa)
{
struct ept *ept = &vcpu->ept;
int mt = ept_memory_type(vcpu_to_ksm(vcpu), dpa);
KSM_DEBUG("unhook page %p\n", dpa);
for_each_eptp(ept, i)
ept_alloc_page(EPT4(ept, i), EPT_ACCESS_ALL, mt, dpa, dpa);
__invept_all();
return true;
}
#endif
static inline void vcpu_flush_idt(struct vcpu *vcpu)
{
vmcs_write32(GUEST_IDTR_LIMIT, vcpu->idt.limit);
vmcs_write(GUEST_IDTR_BASE, vcpu->idt.base);
}
static inline bool vcpu_hook_idte(struct vcpu *vcpu, struct shadow_idt_entry *h)
{
u16 cs = vmcs_read16(GUEST_CS_SELECTOR);
vcpu_put_idt(vcpu, cs, h->n, h->h);
vcpu_flush_idt(vcpu);
return true;
}
static inline bool vcpu_unhook_idte(struct vcpu *vcpu, struct shadow_idt_entry *h)
{
struct kidt_entry64 *entry = &vcpu->shadow_idt[h->n];
if (!idte_present(entry))
return false;
put_entry(vcpu->idt.base, h->n, entry);
vcpu_flush_idt(vcpu);
entry->e32.p = 0;
return true;
}
static inline bool vcpu_emulate_vmfunc(struct vcpu *vcpu, struct h_vmfunc *vmfunc)
{
/* Emulate a VMFUNC due it to not being supported natively. */
struct ept *ept = &vcpu->ept;
if (vmfunc->func >= 64 || !(vcpu->vm_func_ctl & (1ULL << vmfunc->func)) ||
(vmfunc->func == 0 && !test_bit(vmfunc->eptp,
(volatile const unsigned long *)&ept->ptr_list[0]))) {
vcpu_inject_hardirq_noerr(vcpu, X86_TRAP_UD);
return false;
}
vcpu_switch_root_eptp(vcpu, (u16)vmfunc->eptp);
return true;
}
#ifdef NESTED_VMX
static inline bool nested_has_pin(const struct nested_vcpu *nested, u32 bits)
{
return (__nested_vmcs_read32(nested->vmcs, PIN_BASED_VM_EXEC_CONTROL) & bits) == bits;
}
static inline bool nested_has_primary(const struct nested_vcpu *nested, u32 bits)
{
return (__nested_vmcs_read32(nested->vmcs, CPU_BASED_VM_EXEC_CONTROL) & bits) == bits;
}
static inline bool nested_has_secondary(const struct nested_vcpu *nested, u32 bits)
{
return (__nested_vmcs_read32(nested->vmcs, SECONDARY_VM_EXEC_CONTROL) & bits) == bits;
}
static inline u32 nested_build_ar_bytes(u32 type, u32 s, u32 dpl, u32 present,
u32 avl, u32 l, u32 db, u32 g)
{
return type | (s << 4) | (dpl << 5) | (present << 7) |
(avl << 12) | (l << 13) | (db << 14) | (g << 15);
}
static inline void nested_save(uintptr_t vmcs, u32 field)
{
BREAK_ON(!__nested_vmcs_write(vmcs, field, vmcs_read(field)));
}
static inline void nested_save16(uintptr_t vmcs, u32 field)
{
BREAK_ON(!__nested_vmcs_write(vmcs, field, vmcs_read16(field)));
}
static inline void nested_save32(uintptr_t vmcs, u32 field)
{
BREAK_ON(!__nested_vmcs_write(vmcs, field, vmcs_read32(field)));
}
static inline void nested_save64(uintptr_t vmcs, u32 field)
{
#if 0
BREAK_ON(!__nested_vmcs_write(vmcs, field, vmcs_read32(field)) ||
!__nested_vmcs_write(vmcs, field + 1, vmcs_read32(field + 1)));
#else
BREAK_ON(!__nested_vmcs_write(vmcs, field, vmcs_read(field)));
#endif
}
static inline void nested_save_guest_state(struct nested_vcpu *nested)
{
uintptr_t vmcs = nested->vmcs;
nested_save(vmcs, GUEST_RSP);
nested_save(vmcs, GUEST_RIP);
nested_save(vmcs, GUEST_RFLAGS);
nested_save(vmcs, GUEST_CR0);
nested_save(vmcs, GUEST_CR3);
nested_save(vmcs, GUEST_CR4);
nested_save64(vmcs, GUEST_PDPTR0);
nested_save64(vmcs, GUEST_PDPTR1);
nested_save64(vmcs, GUEST_PDPTR2);
nested_save64(vmcs, GUEST_PDPTR3);
nested_save16(vmcs, GUEST_ES_SELECTOR);
nested_save16(vmcs, GUEST_FS_SELECTOR);
nested_save16(vmcs, GUEST_CS_SELECTOR);
nested_save16(vmcs, GUEST_SS_SELECTOR);
nested_save16(vmcs, GUEST_GS_SELECTOR);
nested_save16(vmcs, GUEST_DS_SELECTOR);
nested_save16(vmcs, GUEST_LDTR_SELECTOR);
nested_save16(vmcs, GUEST_TR_SELECTOR);
nested_save32(vmcs, GUEST_ES_LIMIT);
nested_save32(vmcs, GUEST_FS_LIMIT);
nested_save32(vmcs, GUEST_CS_LIMIT);
nested_save32(vmcs, GUEST_SS_LIMIT);
nested_save32(vmcs, GUEST_GS_LIMIT);
nested_save32(vmcs, GUEST_DS_LIMIT);
nested_save32(vmcs, GUEST_LDTR_LIMIT);
nested_save32(vmcs, GUEST_TR_LIMIT);
nested_save32(vmcs, GUEST_IDTR_LIMIT);
nested_save32(vmcs, GUEST_GDTR_LIMIT);
nested_save32(vmcs, GUEST_ES_AR_BYTES);
nested_save32(vmcs, GUEST_FS_AR_BYTES);
nested_save32(vmcs, GUEST_CS_AR_BYTES);
nested_save32(vmcs, GUEST_GS_AR_BYTES);
nested_save32(vmcs, GUEST_SS_AR_BYTES);
nested_save32(vmcs, GUEST_LDTR_AR_BYTES);
nested_save32(vmcs, GUEST_TR_AR_BYTES);
nested_save(vmcs, GUEST_ES_BASE);
nested_save(vmcs, GUEST_FS_BASE);
nested_save(vmcs, GUEST_CS_BASE);
nested_save(vmcs, GUEST_SS_BASE);
nested_save(vmcs, GUEST_GS_BASE);
nested_save(vmcs, GUEST_LDTR_BASE);
nested_save(vmcs, GUEST_TR_BASE);
nested_save(vmcs, GUEST_IDTR_BASE);
nested_save(vmcs, GUEST_GDTR_BASE);
nested_save32(vmcs, GUEST_INTERRUPTIBILITY_INFO);
nested_save(vmcs, GUEST_PENDING_DBG_EXCEPTIONS);
nested_save(vmcs, GUEST_SYSENTER_CS);
nested_save(vmcs, GUEST_SYSENTER_EIP);
nested_save(vmcs, GUEST_SYSENTER_ESP);
nested_save64(vmcs, GUEST_BNDCFGS);
u32 exit = __nested_vmcs_read32(vmcs, VM_EXIT_CONTROLS);
if (exit & VM_EXIT_SAVE_DEBUG_CONTROLS) {
nested_save(vmcs, GUEST_DR7);
nested_save64(vmcs, GUEST_IA32_DEBUGCTL);
}
if (exit & VM_EXIT_LOAD_IA32_PAT)
nested_save64(vmcs, GUEST_IA32_PAT);
if (exit & VM_ENTRY_LOAD_IA32_EFER)
nested_save64(vmcs, GUEST_IA32_EFER);
if (nested_has_primary(nested, CPU_BASED_ACTIVATE_SECONDARY_CONTROLS) &&
nested_has_secondary(nested, SECONDARY_EXEC_VIRTUAL_INTR_DELIVERY))
nested_save16(vmcs, GUEST_INTR_STATUS);
}
static inline bool nested_prepare_hypervisor(struct vcpu *vcpu, uintptr_t vmcs)
{
struct ksm *k = vcpu_to_ksm(vcpu);
u8 err = 0;
err |= vmcs_write(GUEST_RIP, __nested_vmcs_read(vmcs, HOST_RIP));
err |= vmcs_write(GUEST_RSP, __nested_vmcs_read(vmcs, HOST_RSP));
err |= vmcs_write(GUEST_RFLAGS, X86_EFLAGS_FIXED);
err |= vmcs_write(CR0_GUEST_HOST_MASK, vcpu->cr0_guest_host_mask);
err |= vmcs_write(CR4_GUEST_HOST_MASK, vcpu->cr4_guest_host_mask);
err |= vmcs_write(CR0_READ_SHADOW, __nested_vmcs_read(vmcs, HOST_CR0) & ~vcpu->cr0_guest_host_mask);
err |= vmcs_write(CR4_READ_SHADOW, __nested_vmcs_read(vmcs, HOST_CR4) & ~vcpu->cr4_guest_host_mask);
err |= vmcs_write(GUEST_CR0, __nested_vmcs_read(vmcs, HOST_CR0));
err |= vmcs_write(GUEST_CR4, __nested_vmcs_read(vmcs, HOST_CR4));
err |= vmcs_write(GUEST_CR3, __nested_vmcs_read(vmcs, HOST_CR3));
err |= vmcs_write(GUEST_SYSENTER_CS, __nested_vmcs_read(vmcs, HOST_IA32_SYSENTER_CS));
err |= vmcs_write(GUEST_SYSENTER_EIP, __nested_vmcs_read(vmcs, HOST_IA32_SYSENTER_EIP));
err |= vmcs_write(GUEST_SYSENTER_ESP, __nested_vmcs_read(vmcs, HOST_IA32_SYSENTER_ESP));
err |= vmcs_write16(GUEST_ES_SELECTOR, __nested_vmcs_read16(vmcs, HOST_ES_SELECTOR));
err |= vmcs_write16(GUEST_CS_SELECTOR, __nested_vmcs_read16(vmcs, HOST_CS_SELECTOR));
err |= vmcs_write16(GUEST_SS_SELECTOR, __nested_vmcs_read16(vmcs, HOST_SS_SELECTOR));
err |= vmcs_write16(GUEST_DS_SELECTOR, __nested_vmcs_read16(vmcs, HOST_DS_SELECTOR));
err |= vmcs_write16(GUEST_FS_SELECTOR, __nested_vmcs_read16(vmcs, HOST_FS_SELECTOR));
err |= vmcs_write16(GUEST_GS_SELECTOR, __nested_vmcs_read16(vmcs, HOST_GS_SELECTOR));
err |= vmcs_write16(GUEST_DS_SELECTOR, __nested_vmcs_read16(vmcs, HOST_DS_SELECTOR));
err |= vmcs_write16(GUEST_TR_SELECTOR, __nested_vmcs_read16(vmcs, HOST_TR_SELECTOR));
err |= vmcs_write(GUEST_IDTR_BASE, __nested_vmcs_read(vmcs, HOST_IDTR_BASE));
err |= vmcs_write(GUEST_GDTR_BASE, __nested_vmcs_read(vmcs, HOST_GDTR_BASE));
err |= vmcs_write(GUEST_CS_BASE, 0);
err |= vmcs_write32(GUEST_CS_LIMIT, 0xFFFFFFFF);
if (__nested_vmcs_read(vmcs, VM_EXIT_CONTROLS) & VM_EXIT_HOST_ADDR_SPACE_SIZE)
err |= vmcs_write32(GUEST_CS_AR_BYTES, nested_build_ar_bytes(11, 1, 0, 1, 0, 1, 0, 1));
else
err |= vmcs_write32(GUEST_CS_AR_BYTES, nested_build_ar_bytes(11, 1, 0, 1, 0, 0, 1, 0));
const u32 ar = nested_build_ar_bytes(3, 1, 0, 1, 0, 0, 1, 1);
err |= vmcs_write32(GUEST_ES_LIMIT, 0xFFFFFFFF);
err |= vmcs_write(GUEST_ES_BASE, 0);
err |= vmcs_write32(GUEST_ES_AR_BYTES, ar);
err |= vmcs_write32(GUEST_DS_LIMIT, 0xFFFFFFFF);
err |= vmcs_write(GUEST_DS_BASE, 0);
err |= vmcs_write32(GUEST_DS_AR_BYTES, ar);
err |= vmcs_write32(GUEST_SS_LIMIT, 0xFFFFFFFF);
err |= vmcs_write(GUEST_SS_BASE, 0);
err |= vmcs_write32(GUEST_SS_AR_BYTES, ar);
err |= vmcs_write32(GUEST_FS_LIMIT, 0xFFFFFFFF);
err |= vmcs_write(GUEST_FS_BASE, __nested_vmcs_read(vmcs, HOST_FS_BASE));
err |= vmcs_write32(GUEST_FS_AR_BYTES, ar);
err |= vmcs_write32(GUEST_GS_LIMIT, 0xFFFFFFFF);
err |= vmcs_write(GUEST_GS_BASE, __nested_vmcs_read(vmcs, HOST_GS_BASE));
err |= vmcs_write32(GUEST_GS_AR_BYTES, ar);
const u32 tar = nested_build_ar_bytes(11, 0, 0, 1, 0, 0, 0, 0);
err |= vmcs_write32(GUEST_TR_AR_BYTES, tar);
err |= vmcs_write(GUEST_TR_BASE, __nested_vmcs_read(vmcs, HOST_TR_BASE));
err |= vmcs_write32(GUEST_TR_LIMIT, 0x67);
err |= vmcs_write(GUEST_DR7, DR7_FIXED_1);
err |= vmcs_write64(GUEST_IA32_DEBUGCTL, 0);
err |= vmcs_write64(VMCS_LINK_POINTER, ~0ULL);
err |= vmcs_write32(VM_ENTRY_CONTROLS, vcpu->entry_ctl);
err |= vmcs_write32(VM_EXIT_CONTROLS, vcpu->exit_ctl);
err |= vmcs_write32(PIN_BASED_VM_EXEC_CONTROL, vcpu->pin_ctl);
err |= vmcs_write32(CPU_BASED_VM_EXEC_CONTROL, vcpu->cpu_ctl);
err |= vmcs_write32(SECONDARY_VM_EXEC_CONTROL, vcpu->secondary_ctl);
err |= vmcs_write64(MSR_BITMAP, __pa(k->msr_bitmap));
err |= vmcs_write64(IO_BITMAP_A, __pa(k->io_bitmap_a));
err |= vmcs_write64(IO_BITMAP_B, __pa(k->io_bitmap_b));
err |= vmcs_write16(VIRTUAL_PROCESSOR_ID, vpid_nr());
vcpu_switch_root_eptp(vcpu, vcpu_eptp_idx(vcpu));
if (err == 0)
__invvpid_all();
return err == 0;
}
static inline bool vcpu_enter_nested_hypervisor(struct vcpu *vcpu, u32 exit_reason)
{
/*
* Here we came from the nested hypervisor's guest, we have received an
* event that we can't help ourselves, so we need to throw it back to the
* nested hypervisor to handle it appropriately.
*
* Do so by setting the appropriate _nested_ VMCS fields and then setting
* "guest's RIP" to that of the nested hypervisor's RIP (Host RIP from their
* VMCS).
*/
struct nested_vcpu *nested = &vcpu->nested_vcpu;
uintptr_t vmcs = nested->vmcs;
/*
* Mark it as left the nested hypervisor' guest, so we can know if the next
* vm-exit came from it and not from it's guest.
*/
nested_leave(nested);
nested_save_guest_state(nested);
if (!nested_prepare_hypervisor(vcpu, vmcs))
return false;
u16 handler = (u16)exit_reason;
if (lapic_in_kernel() && handler == EXIT_REASON_EXTERNAL_INTERRUPT &&
__nested_vmcs_read(vmcs, VM_EXIT_CONTROLS) & VM_EXIT_ACK_INTR_ON_EXIT)
;/* FIXME */
const u32 intr_mask = INTR_INFO_DELIVER_CODE_MASK | INTR_INFO_VALID_MASK;
u32 intr_info = vmcs_read32(VM_EXIT_INTR_INFO);
if ((intr_info & intr_mask) == intr_mask)
nested_save(vmcs, VM_EXIT_INTR_ERROR_CODE);
__nested_vmcs_write(vmcs, VM_EXIT_REASON, (u16)exit_reason);
__nested_vmcs_write(vmcs, VM_EXIT_INTR_INFO, intr_info);
__nested_vmcs_write(vmcs, EXIT_QUALIFICATION, vmcs_read(EXIT_QUALIFICATION));
__nested_vmcs_write(vmcs, VM_EXIT_INSTRUCTION_LEN, vmcs_read32(VM_EXIT_INSTRUCTION_LEN));
if (handler == EXIT_REASON_GDT_IDT_ACCESS || handler == EXIT_REASON_LDT_TR_ACCESS ||
(handler >= EXIT_REASON_VMCLEAR && handler <= EXIT_REASON_VMON))
__nested_vmcs_write(vmcs, VMX_INSTRUCTION_INFO, vmcs_read(VMX_INSTRUCTION_INFO));
__nested_vmcs_write(vmcs, GUEST_LINEAR_ADDRESS, vmcs_read(GUEST_LINEAR_ADDRESS));
__nested_vmcs_write64(vmcs, GUEST_PHYSICAL_ADDRESS, vmcs_read64(GUEST_PHYSICAL_ADDRESS));
return true;
}
#endif
static bool vcpu_handle_vmcall(struct vcpu *vcpu)
{
VCPU_TRACER_START();
/* VMFUNC does not have CPL checks, so emulator shouldn't have too... */
u32 nr = ksm_read_reg32(vcpu, STACK_REG_CX);
if (nr != HCALL_VMFUNC && vcpu_inject_gp_if(vcpu, !vcpu_probe_cpl(0)))
goto out;
uintptr_t arg = ksm_read_reg(vcpu, STACK_REG_DX);
switch (nr) {
case HCALL_STOP:
vcpu_do_exit(vcpu);
VCPU_TRACER_END();
return false;
case HCALL_IDT:
vcpu_adjust_rflags(vcpu, vcpu_hook_idte(vcpu, (struct shadow_idt_entry *)arg));
break;
case HCALL_UIDT:
vcpu_adjust_rflags(vcpu, vcpu_unhook_idte(vcpu, (struct shadow_idt_entry *)arg));
break;
#ifdef EPAGE_HOOK
case HCALL_HOOK:
vcpu_adjust_rflags(vcpu, vcpu_handle_hook(vcpu, (struct epage_info *)arg));
break;
case HCALL_UNHOOK:
vcpu_adjust_rflags(vcpu, vcpu_handle_unhook(vcpu, arg));
break;
#endif
case HCALL_VMFUNC:
vcpu_adjust_rflags(vcpu, vcpu_emulate_vmfunc(vcpu, (struct h_vmfunc *)arg));
break;
#ifdef PMEM_SANDBOX
case HCALL_SA_TASK:
vcpu_adjust_rflags(vcpu, ksm_sandbox_handle_vmcall(vcpu, arg));