-
Notifications
You must be signed in to change notification settings - Fork 8
/
tlbsplit4.10.patch
938 lines (907 loc) · 32 KB
/
tlbsplit4.10.patch
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
diff --git a/arch/x86/kvm/Makefile b/arch/x86/kvm/Makefile
index 3bff207..752e419 100644
--- a/arch/x86/kvm/Makefile
+++ b/arch/x86/kvm/Makefile
@@ -14,6 +14,7 @@ kvm-$(CONFIG_KVM_ASYNC_PF) += $(KVM)/async_pf.o
kvm-y += x86.o mmu.o emulate.o i8259.o irq.o lapic.o \
i8254.o ioapic.o irq_comm.o cpuid.o pmu.o mtrr.o \
hyperv.o page_track.o debugfs.o
+kvm-y += tlbsplit.o
kvm-$(CONFIG_KVM_DEVICE_ASSIGNMENT) += assigned-dev.o iommu.o
diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index 7012de4..2d31a1c 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -43,6 +43,7 @@
#include <asm/io.h>
#include <asm/vmx.h>
#include <asm/kvm_page_track.h>
+#include "tlbsplit.h"
/*
* When setting this variable to true it enables Two-Dimensional-Paging
@@ -307,7 +308,7 @@ static int is_nx(struct kvm_vcpu *vcpu)
static int is_shadow_present_pte(u64 pte)
{
- return (pte & 0xFFFFFFFFull) && !is_mmio_spte(pte);
+ return ((pte & 0xFFFFFFFFull) && !is_mmio_spte(pte)) || COULD_BE_SPLIT_PAGE(pte); //splittlb
}
static int is_large_pte(u64 pte)
@@ -1170,6 +1171,11 @@ static u64 *rmap_get_next(struct rmap_iterator *iter)
static void drop_spte(struct kvm *kvm, u64 *sptep)
{
+ if (COULD_BE_SPLIT_PAGE(*sptep)) {
+ //tlbs debug
+ printk(KERN_WARNING "drop_spte: got something that looks like split page in setter spte:0x%llx checkerfunc:%d\n",*sptep,split_tlb_has_split_page(kvm,sptep));
+ WARN_ON(1);
+ }
if (mmu_spte_clear_track_bits(sptep))
rmap_remove(kvm, sptep);
}
@@ -1387,6 +1393,9 @@ static bool kvm_zap_rmapp(struct kvm *kvm, struct kvm_rmap_head *rmap_head)
bool flush = false;
while ((sptep = rmap_get_first(rmap_head, &iter))) {
+ if (COULD_BE_SPLIT_PAGE(*sptep) && split_tlb_has_split_page(kvm,sptep)) { //tlbsplit
+ printk(KERN_WARNING "kvm_zap_rmapp: flipped page to exec 0x%llx\n",*sptep); //tlbsplit
+ } //tlbsplit
rmap_printk("%s: spte %p %llx.\n", __func__, sptep, *sptep);
drop_spte(kvm, sptep);
@@ -2283,6 +2292,11 @@ static bool mmu_page_zap_pte(struct kvm *kvm, struct kvm_mmu_page *sp,
pte = *spte;
if (is_shadow_present_pte(pte)) {
+ if (COULD_BE_SPLIT_PAGE(pte)&&split_tlb_has_split_page(kvm,spte)) {
+ printk(KERN_WARNING "mmu_page_zap_pte: zapping split page in read mode, flipped it to code 0x%llx\n", pte);
+ //split_tlb_flip_to_code(kvm,sp->gfn,spte);
+ pte = *spte;
+ }
if (is_last_spte(pte, sp->role.level)) {
drop_spte(kvm, spte);
if (is_large_pte(pte))
@@ -2512,6 +2526,14 @@ static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep,
u64 spte = 0;
int ret = 0;
+ struct kvm_splitpage* page = split_tlb_findpage(vcpu->kvm, gfn<<PAGE_SHIFT);
+
+ if (COULD_BE_SPLIT_PAGE(*sptep)) {
+ //tlbs debug
+ printk(KERN_WARNING "set_spte: got something that looks like active split page in setter spte:0x%llx\n",*sptep);
+ WARN_ON(1);
+ }
+
if (set_mmio_spte(vcpu, sptep, gfn, pfn, pte_access))
return 0;
@@ -2584,6 +2606,10 @@ static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep,
}
set_pte:
+ if (page&&page->active) {
+ printk(KERN_WARNING "set_spte: adjusting spte to execute only :0x%llx\n",spte);
+ spte&=~(VMX_EPT_WRITABLE_MASK|VMX_EPT_READABLE_MASK);
+ }
if (mmu_spte_update(sptep, spte))
kvm_flush_remote_tlbs(vcpu->kvm);
done:
@@ -2601,6 +2627,12 @@ static bool mmu_set_spte(struct kvm_vcpu *vcpu, u64 *sptep, unsigned pte_access,
pgprintk("%s: spte %llx write_fault %d gfn %llx\n", __func__,
*sptep, write_fault, gfn);
+ if (COULD_BE_SPLIT_PAGE(*sptep)) {
+ //tlbs debug
+ printk(KERN_WARNING "mmu_set_spte: got something that looks like split page in setter spte:0x%llx\n",*sptep);
+ WARN_ON(1);
+ }
+
if (is_shadow_present_pte(*sptep)) {
/*
* If we overwrite a PTE page pointer with a 2MB PMD, unlink
@@ -2735,6 +2767,30 @@ static void direct_pte_prefetch(struct kvm_vcpu *vcpu, u64 *sptep)
__direct_pte_prefetch(vcpu, sp, sptep);
}
+u64* split_tlb_findspte(struct kvm_vcpu *vcpu,gfn_t gfn) {
+
+ struct kvm_shadow_walk_iterator iterator;
+
+ for_each_shadow_entry(vcpu, gfn << PAGE_SHIFT, iterator) {
+ int last = is_last_spte(*iterator.sptep, iterator.level);
+ int large = is_large_pte(*iterator.sptep);
+ if (last && !large) {
+ return iterator.sptep;
+ }
+ if (last && large) {
+ struct kvm_memory_slot *slot;
+ printk(KERN_WARNING "Large page found for 0x%llx spte:0x%llx level:%d\n",gfn << PAGE_SHIFT,*iterator.sptep,iterator.level);
+ //return NULL;
+ slot = kvm_vcpu_gfn_to_memslot(vcpu, gfn);
+ kvm_mmu_gfn_disallow_lpage(slot, gfn);
+ //drop_large_spte(vcpu,iterator.sptep);
+ last = is_last_spte(*iterator.sptep, iterator.level);
+ printk(KERN_WARNING "For page for 0x%llx spte:0x%llx level:%d last:%d\n",gfn << PAGE_SHIFT,*iterator.sptep,iterator.level,last);
+ }
+ }
+ return NULL;
+}
+
static int __direct_map(struct kvm_vcpu *vcpu, int write, int map_writable,
int level, gfn_t gfn, kvm_pfn_t pfn, bool prefault)
{
diff --git a/arch/x86/kvm/mmu.h b/arch/x86/kvm/mmu.h
index ddc56e9..e64858a 100644
--- a/arch/x86/kvm/mmu.h
+++ b/arch/x86/kvm/mmu.h
@@ -195,6 +195,7 @@ static inline u8 permission_fault(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
}
void kvm_mmu_invalidate_zap_all_pages(struct kvm *kvm);
+u64* split_tlb_findspte(struct kvm_vcpu *vcpu,gfn_t gfn); //splittlb
void kvm_zap_gfn_range(struct kvm *kvm, gfn_t gfn_start, gfn_t gfn_end);
void kvm_mmu_gfn_disallow_lpage(struct kvm_memory_slot *slot, gfn_t gfn);
diff --git a/arch/x86/kvm/tlbsplit.c b/arch/x86/kvm/tlbsplit.c
new file mode 100644
index 0000000..95373e2
--- /dev/null
+++ b/arch/x86/kvm/tlbsplit.c
@@ -0,0 +1,623 @@
+/*
+ * tlbsplit.c
+ *
+ * Created on: Dec 28, 2015
+ * Author: nick
+ */
+
+#include "tlbsplit.h"
+#include <asm/vmx.h>
+#include <linux/debugfs.h>
+#include <linux/kvm_host.h>
+//#include <linux/gfp.h>
+#include "mmu.h"
+
+#define PT64_BASE_ADDR_MASK (((1ULL << 52) - 1) & ~(u64)(PAGE_SIZE-1))
+
+#define PTE_WRITE (1<<1)
+#define PTE_READ (1<<0)
+#define PTE_EXECUTE (1<<2)
+
+#define KVM_MAX_TRACKER 0x10000
+
+struct kvm_ept_violation_tracker_entry {
+ u32 counter;
+ u16 read;
+ u16 vmnumber;
+ u64 gva;
+ u64 rip;
+ u64 cr3;
+} __attribute__( ( packed ) ) ;
+
+atomic_t split_tracker_next_write;
+struct kvm_ept_violation_tracker {
+ int max_number_of_entries;
+ struct kvm_ept_violation_tracker_entry entries[KVM_MAX_TRACKER];
+} __attribute__( ( packed ) ) split_tracker;
+
+static struct dentry *split_dentry;
+
+static int next_vm;
+
+/* read file operation */
+static ssize_t split_counter_reader(struct file *fp, char __user *user_buffer,
+ size_t count, loff_t *position)
+{
+ return simple_read_from_buffer(user_buffer, count, position, &split_tracker, sizeof split_tracker);
+}
+
+static const struct file_operations split_debug = {
+ .read = split_counter_reader,
+};
+
+void split_init_debugfs(void) {
+ atomic_set(&split_tracker_next_write,0);
+ split_tracker.max_number_of_entries = KVM_MAX_TRACKER;
+ split_dentry = debugfs_create_file("tlb_split", 0444, kvm_debugfs_dir, NULL, &split_debug);
+ printk(KERN_INFO "tlb_split_init:debugfs_create_file returned 0%lx\n",(unsigned long)split_dentry);
+ next_vm = 0;
+}
+
+void _register_ept_flip(gva_t gva,gva_t rip,unsigned long cr3,int vmnumber,bool read) {
+ int counter = atomic_inc_return(&split_tracker_next_write);
+ int nextRow = (counter - 1) % KVM_MAX_TRACKER;
+ split_tracker.entries[nextRow].gva = gva;
+ split_tracker.entries[nextRow].rip = rip;
+ split_tracker.entries[nextRow].cr3 = cr3;
+ split_tracker.entries[nextRow].vmnumber = vmnumber;
+ split_tracker.entries[nextRow].read = read;
+ split_tracker.entries[nextRow].counter = counter;
+}
+
+void split_shutdown_debugfs(void) {
+ debugfs_remove(split_dentry);
+}
+
+bool tlb_split_init(struct kvm *kvm) {
+ kvm->splitpages = kzalloc(sizeof(struct kvm_splitpages), GFP_KERNEL);
+ if (kvm->splitpages!=NULL) {
+ kvm->splitpages->vmcounter = next_vm++;
+ return true;
+ }
+ else
+ return false;
+}
+
+void kvm_split_tlb_freepage(struct kvm_splitpage *page)
+{
+ page->cr3 = 0;
+ page->gpa = 0;
+ page->active = 0;
+ page->gva = 0;
+ page->codeaddr = 0;
+ if (page->dataaddr) {
+ kfree(page->dataaddr);
+ page->dataaddr = NULL;
+ }
+}
+EXPORT_SYMBOL_GPL(kvm_split_tlb_freepage);
+
+void kvm_split_tlb_deactivateall(struct kvm *kvm) {
+ struct kvm_splitpages *spages = kvm->splitpages;
+ int i;
+ for (i = 0; i < KVM_MAX_SPLIT_PAGES; i++)
+ kvm_split_tlb_freepage(&spages->pages[i]);
+ kfree(kvm->splitpages);
+}
+EXPORT_SYMBOL_GPL(kvm_split_tlb_deactivateall);
+
+static struct kvm_splitpage* _split_tlb_findpage(struct kvm *kvms,gpa_t gpa) {
+ int i;
+ struct kvm_splitpage* found;
+ gpa_t pagestart;
+ pagestart = gpa&PAGE_MASK;
+ for (i=0; i<KVM_MAX_SPLIT_PAGES; i++) {
+ found = kvms->splitpages->pages+i;
+ if (found->gpa == pagestart)
+ return found;
+ }
+ return NULL;
+}
+
+struct kvm_splitpage* split_tlb_findpage(struct kvm *kvms,gpa_t gpa) {
+ if (gpa&PAGE_MASK)
+ return _split_tlb_findpage(kvms,gpa);
+ else
+ return NULL;
+}
+EXPORT_SYMBOL_GPL(split_tlb_findpage);
+
+struct kvm_splitpage* split_tlb_findpage_gva_cr3(struct kvm *kvms, gva_t gva, ulong cr3) {
+ struct kvm_splitpage* found;
+ gva_t pagestart;
+ int i;
+ pagestart = gva&PAGE_MASK;
+ for (i=0; i<KVM_MAX_SPLIT_PAGES; i++) {
+ found = kvms->splitpages->pages+i;
+ if (found->gva == pagestart && found->cr3 == cr3)
+ return found;
+ }
+ return NULL;
+}
+
+
+int split_tlb_setdatapage(struct kvm_vcpu *vcpu, gva_t gva, gva_t datagva, ulong cr3) {
+ gpa_t gpa;
+ u32 access;
+ struct kvm_splitpage* page;
+ struct x86_exception exception;
+ gpa_t translated;
+ int r;
+ access = (kvm_x86_ops->get_cpl(vcpu) == 3) ? PFERR_USER_MASK : 0;
+ gpa = vcpu->arch.walk_mmu->gva_to_gpa(vcpu, gva, access, &exception);
+ if (gpa == UNMAPPED_GVA) {
+ printk(KERN_WARNING "split:tlb_setdatapage gva:0x%lx gpa not found %d\n",gva,exception.error_code);
+ gpa = 0;
+ }
+ printk(KERN_WARNING "split:tlb_setdatapage cr3:0x%lx gva:0x%lx gpa:0x%llx\n",cr3,gva,gpa);
+ if (gpa!=0)
+ page = split_tlb_findpage(vcpu->kvm,gpa);
+ else
+ page = split_tlb_findpage_gva_cr3(vcpu->kvm,gva,cr3);
+ if (page == NULL) {
+ page = _split_tlb_findpage(vcpu->kvm,0);
+ if (page == NULL) {
+ printk(KERN_WARNING "No more slots in the split page table\n");
+ return 0;
+ }
+ page->cr3 = cr3;
+ page->gpa = gpa&PAGE_MASK;
+ page->gva = gva&PAGE_MASK;
+ page->dataaddr = kmalloc(4096,GFP_KERNEL);
+ BUG_ON(((long unsigned int)page->dataaddr&~PAGE_MASK)!=0);
+ translated = vcpu->arch.walk_mmu->gva_to_gpa(vcpu, datagva&PAGE_MASK, access, &exception);
+ if (translated == UNMAPPED_GVA) {
+ printk(KERN_WARNING "split:tlb_setdatapage gva:0x%lx gpa not found for data %d\n",datagva,exception.error_code);
+ return 0;
+ }
+ r = kvm_read_guest(vcpu->kvm,translated,page->dataaddr,4096);
+ printk(KERN_WARNING "split:tlb_setdatapage cr3:0x%lx gva:0x%lx gpa:0x%llx allocated:0x%llx copy result:%d\n",cr3,gva,gpa,(u64)page->dataaddr,r);
+ } else {
+ printk(KERN_WARNING "Already a page for: gpa:0x%llx with cr3:0x%lx and gva=0x%lx\n",gpa,page->cr3,page->gva);
+ return 0;
+ }
+ return 1;
+}
+//EXPORT_SYMBOL_GPL(split_tlb_setdatapage);
+
+
+int split_tlb_activatepage(struct kvm_vcpu *vcpu, gva_t gva, ulong cr3) {
+ gpa_t gpa;
+ u32 access;
+ struct kvm_splitpage* page;
+ struct x86_exception exception;
+ u64* sptep;
+ //struct kvm_shadow_walk_iterator iterator;
+ gfn_t gfn;
+
+ access = (kvm_x86_ops->get_cpl(vcpu) == 3) ? PFERR_USER_MASK : 0;
+ gpa = vcpu->arch.walk_mmu->gva_to_gpa(vcpu, gva, access, &exception);
+ if (gpa == UNMAPPED_GVA) {
+ printk(KERN_WARNING "split:split_tlb_activatepage gva:0x%lx gpa not found %d\n",gva,exception.error_code);
+ return 0;
+ }
+ page = split_tlb_findpage_gva_cr3(vcpu->kvm,gva,cr3);
+ if (page == NULL) {
+ printk(KERN_WARNING "split:tlb_activatepage page not foundcr3:0x%lx gva:0x%lx translated gpa:0x%llx \n",cr3,gva,gpa);
+ return 0;
+ }
+ printk(KERN_WARNING "split_tlb_activatepage found page cr3:0x%lx gva:0x%lx gpa:0x%llx page_gpa:0x%llx\n",cr3,gva,gpa,page->gpa);
+ if (page->gpa != (gpa&PAGE_MASK) ) {
+ printk(KERN_WARNING "split:tlb_activatepage gpa changed 0x%llx->0x%llx, adjusting\n",page->gpa,gpa&PAGE_MASK);
+ page->gpa = gpa&PAGE_MASK;
+ }
+
+ gfn = gpa >> PAGE_SHIFT;
+ sptep = split_tlb_findspte(vcpu,gfn);
+ if (sptep!=NULL) {
+ u64 newspte = *sptep & ~(VMX_EPT_READABLE_MASK|VMX_EPT_WRITABLE_MASK);
+ //newspte = 0L;
+ printk(KERN_WARNING "split_tlb_activatepage: spte=0x%llx->newspte=0x%llx ,sptep=x%llx\n",*sptep,newspte,(u64)sptep);
+ *sptep = newspte;
+ page->active = true;
+ kvm_flush_remote_tlbs(vcpu->kvm);
+ return 1;
+ } else
+ printk(KERN_WARNING "split_tlb_activatepage: spte not found 0x%llx\n",gpa);
+ return 0;
+}
+//EXPORT_SYMBOL_GPL(split_tlb_activatepage);
+
+int split_tlb_copymem(struct kvm_vcpu *vcpu, gva_t from, gva_t to, u64 count) {
+ if (count>MAX_PATCH_SIZE)
+ return 0;
+ else {
+ int r;
+ char buf[count];
+ struct x86_exception exception;
+ u32 access = (kvm_x86_ops->get_cpl(vcpu) == 3) ? PFERR_USER_MASK : 0;
+ gpa_t from_gpa = vcpu->arch.walk_mmu->gva_to_gpa(vcpu, from, access, &exception);
+ gpa_t to_gpa = vcpu->arch.walk_mmu->gva_to_gpa(vcpu, to, access, &exception);
+ if (from_gpa == UNMAPPED_GVA) {
+ printk(KERN_WARNING "split_tlb_copymem: from gva:0x%lx gpa not found %d\n",from,exception.error_code);
+ return 0;
+ }
+ if (to_gpa == UNMAPPED_GVA) {
+ printk(KERN_WARNING "split_tlb_copymem: to gva:0x%lx gpa not found %d\n",to,exception.error_code);
+ return 0;
+ }
+ r = kvm_read_guest(vcpu->kvm,from_gpa,buf,count);
+ if (r != 0) {
+ printk(KERN_WARNING "split_tlb_copymem: read gva:0x%lx gpa:0x%llx failed with the result %d\n",from,from_gpa,r);
+ return 0;
+ }
+ r = kvm_write_guest(vcpu->kvm,to_gpa,buf,count);
+ if (r != 0) {
+ printk(KERN_WARNING "split_tlb_copymem: write gva:0x%lx gpa:0x%llx failed with the result %d\n",to,to_gpa,r);
+ return 0;
+ }
+ return 1;
+ }
+}
+
+int split_tlb_restore_spte_atomic(struct kvm *kvms,gfn_t gfn,u64* sptep,hpa_t stepaddr) {
+ if (sptep!=NULL) {
+ u64 newspte = *sptep;
+ if ((newspte&VMX_EPT_READABLE_MASK)==0||(newspte&VMX_EPT_EXECUTABLE_MASK)==0||(newspte&VMX_EPT_WRITABLE_MASK)==0) {
+ newspte|=VMX_EPT_READABLE_MASK|VMX_EPT_WRITABLE_MASK|VMX_EPT_EXECUTABLE_MASK;
+ newspte&=~PT64_BASE_ADDR_MASK;
+ newspte|=stepaddr & PT64_BASE_ADDR_MASK;
+ printk(KERN_WARNING "split_tlb_restore_spte_atomic: fixing spte 0%llx->0%llx for 0%llx\n", *sptep, newspte, gfn<<PAGE_SHIFT);
+ *sptep = newspte;
+ } else
+ printk(KERN_WARNING "split_tlb_restore_spte_atomic: spte for 0%llx seems untouched: 0%llx\n", gfn<<PAGE_SHIFT, *sptep);
+ return 1;
+ } else {
+ printk(KERN_WARNING "split_tlb_restore_spte_atomic: spte not found for 0x%llx\n", gfn<<PAGE_SHIFT);
+ return 0;
+ }
+}
+
+hpa_t ts_gfn_to_pfa(struct kvm_vcpu *vcpu,gfn_t gfn) {
+struct kvm_memory_slot *slot;
+bool async,writable;
+kvm_pfn_t pfn;
+
+ slot = kvm_vcpu_gfn_to_memslot(vcpu, gfn);
+ async = false;
+ pfn = __gfn_to_pfn_memslot(slot, gfn, false, &async, false, &writable);
+ if (async || !writable) {
+ printk(KERN_WARNING "ts_gfn_to_pfn: unexpected async:%d writable%d\n", async, writable);
+ WARN_ON(1);
+ }
+ return pfn << PAGE_SHIFT;
+
+}
+
+int split_tlb_restore_spte(struct kvm_vcpu *vcpu,gfn_t gfn) {
+ int result;
+ u64* sptep;
+ hpa_t stepaddr = ts_gfn_to_pfa(vcpu,gfn) ;
+// if (async || !writable)
+// printk(KERN_WARNING "split_tlb_restore_spte: unexpected async:%d writable%d gpa:0%llx hfn:0%llx\n", async, writable, gfn<<PAGE_SHIFT,stepaddr);
+ spin_lock(&vcpu->kvm->mmu_lock);
+ sptep = split_tlb_findspte(vcpu,gfn);
+ if (sptep!=NULL && *sptep==0) {
+ spin_unlock(&vcpu->kvm->mmu_lock);
+ printk(KERN_WARNING "split_tlb_restore_spte: zero spte, falling back to default handler gpa:0%llx\n", gfn<<PAGE_SHIFT);
+ return 0;
+ }
+ result = split_tlb_restore_spte_atomic(vcpu->kvm,gfn,sptep,stepaddr);
+ spin_unlock(&vcpu->kvm->mmu_lock);
+ return result;
+}
+
+/*
+int split_tlb_restore_spte_base(struct kvm *kvms,gfn_t gfn,u64* sptep) {
+ bool async,writable;
+ hpa_t stepaddr = gfn_to_pfn_async(kvms,gfn,&async,false,&writable);
+ if (async || !writable)
+ printk(KERN_WARNING "split_tlb_restore_spte_base: unexpected async:%d writable%d gpa:0%llx\n", async, writable, gfn<<PAGE_SHIFT);
+ return split_tlb_restore_spte_atomic(kvms,gfn,sptep,stepaddr);
+}
+*/
+
+int split_tlb_flip_to_code(struct kvm *kvms,hpa_t hpa,u64* sptep) {
+ if (sptep!=NULL) {
+ u64 newspte = *sptep;
+ if ((newspte&VMX_EPT_READABLE_MASK)!=0||(newspte&VMX_EPT_EXECUTABLE_MASK)==0||(newspte&VMX_EPT_WRITABLE_MASK)==0) {
+ WARN_ON(hpa==0);
+ newspte&=~(VMX_EPT_WRITABLE_MASK|VMX_EPT_READABLE_MASK);
+ newspte|=VMX_EPT_EXECUTABLE_MASK;
+ newspte&=~PT64_BASE_ADDR_MASK;
+ newspte|=hpa&PT64_BASE_ADDR_MASK;
+ printk(KERN_WARNING "split_tlb_flip_to_code: fixing spte 0%llx->0%llx for 0%llx\n", *sptep, newspte, hpa);
+ *sptep = newspte;
+ } else
+ printk(KERN_WARNING "split_tlb_flip_to_code: spte for 0%llx seems untouched: 0%llx\n", hpa, *sptep);
+ return 1;
+ } else {
+ printk(KERN_WARNING "split_tlb_flip_to_code: spte not found for hpa 0x%llx\n", hpa);
+ return 0;
+ }
+}
+
+
+int split_tlb_freepage_by_gpa(struct kvm_vcpu *vcpu, gpa_t gpa) {
+ gfn_t gfn;
+ struct kvm_splitpage* page;
+ page = split_tlb_findpage(vcpu->kvm,gpa);
+ if (page!=NULL) {
+ if (page->active) {
+ int rc = kvm_write_guest(vcpu->kvm,gpa&PAGE_MASK,page->dataaddr,4096);
+ gfn = gpa >> PAGE_SHIFT;
+ split_tlb_restore_spte(vcpu,gfn);
+ printk(KERN_WARNING "split_tlb_freepage_by_gpa: copying data cr3:0x%lx gva:0x%lx gpa:0x%llx cpyrc:%d\n",page->cr3,page->gva,page->gpa,rc);
+ } else {
+ printk(KERN_WARNING "split_tlb_freepage_by_gpa: inactive page cr3:0x%lx gva:0x%lx gpa:0x%llx\n",page->cr3,page->gva,page->gpa);
+ }
+ kvm_split_tlb_freepage(page);
+ return 1;
+ } else
+ printk(KERN_WARNING "split_tlb_freepage_by_gpa: page not found gpa:0x%llx\n",gpa);
+ return 0;
+}
+
+int split_tlb_freepage(struct kvm_vcpu *vcpu, gva_t gva) {
+ gpa_t gpa;
+ u32 access;
+ struct x86_exception exception;
+
+ access = (kvm_x86_ops->get_cpl(vcpu) == 3) ? PFERR_USER_MASK : 0;
+ gpa = vcpu->arch.walk_mmu->gva_to_gpa(vcpu, gva, access, &exception);
+ if (gpa == UNMAPPED_GVA) {
+ printk(KERN_WARNING "split:tlb_freepage gva:0x%lx gpa not found %d\n",gva,exception.error_code);
+ return 0;
+ }
+ return split_tlb_freepage_by_gpa(vcpu,gpa);
+}
+
+int split_tlb_flip_page(struct kvm_vcpu *vcpu, gpa_t gpa, struct kvm_splitpage* splitpage, unsigned long exit_qualification)
+{
+ gfn_t gfn = gpa >> PAGE_SHIFT;
+ unsigned long rip = kvm_rip_read(vcpu);
+ unsigned long cr3 = kvm_read_cr3(vcpu);
+ phys_addr_t detouraddr = virt_to_phys(splitpage->dataaddr);
+
+
+ if (exit_qualification & PTE_WRITE) //write
+ {
+ int rc;
+ //newspte|=VMX_EPT_READABLE_MASK|VMX_EPT_WRITABLE_MASK|VMX_EPT_EXECUTABLE_MASK;
+ //newspte&=~PT64_BASE_ADDR_MASK;
+ //newspte|=(stepaddr<<PAGE_SHIFT)&PT64_BASE_ADDR_MASK;
+ printk(KERN_WARNING "split_tlb_flip_page: WRITE EPT fault at 0x%llx. detourpa:0x%llx rip:0x%lx\n vcpuid:%d Removing the page\n",gpa,detouraddr,rip,vcpu->vcpu_id);
+ if (split_tlb_restore_spte(vcpu,gfn)==0)
+ return 0;
+ rc = kvm_write_guest(vcpu->kvm,gpa&PAGE_MASK,splitpage->dataaddr,4096);
+ kvm_split_tlb_freepage(splitpage);
+ printk(KERN_WARNING "split_tlb_flip_page: WRITE EPT fault at 0x%llx data copied rc:%d\n",gpa,rc);
+ } else if (exit_qualification & PTE_READ) //read
+ {
+ u64* sptep;
+ hpa_t stepaddr = ts_gfn_to_pfa(vcpu,gfn);
+ //if (async || !writable)
+ // printk(KERN_WARNING "split_tlb_flip_page: unexpected async:%d writable%d\n", async, writable);
+ spin_lock(&vcpu->kvm->mmu_lock);
+ sptep = split_tlb_findspte(vcpu,gfn);
+ if (exit_qualification & PTE_EXECUTE) //TODO handle execute&read, not sure if needed
+ {
+ printk(KERN_ERR "split_tlb_flip_page: read&execute EPT fault at 0x%llx. Need to handle it properly \n",gpa);
+ }
+ if (sptep!=NULL) {
+ u64 newspte = *sptep;
+ if (newspte==0) {
+ printk(KERN_INFO "split_tlb_flip_page: fallback to default handler(READ):0x%llx \n",gpa);
+ spin_unlock(&vcpu->kvm->mmu_lock);
+ return 0;
+ }
+ if ((newspte&(VMX_EPT_WRITABLE_MASK|VMX_EPT_EXECUTABLE_MASK|VMX_EPT_READABLE_MASK))==0) {
+ printk(KERN_INFO "split_tlb_flip_page: sptep last 3 bits are 0 for gpa:0x%llx \n",gpa);
+ }
+ splitpage->codeaddr = stepaddr;
+ newspte&=~(VMX_EPT_WRITABLE_MASK|VMX_EPT_EXECUTABLE_MASK);
+ newspte|=VMX_EPT_READABLE_MASK;
+ newspte&=~PT64_BASE_ADDR_MASK;
+ newspte|=detouraddr&PT64_BASE_ADDR_MASK;
+ //printk(KERN_WARNING "split_tlb_flip_page: read EPT fault at 0x%llx/0x%llx -> 0x%llx detourpa:0x%llx rip:0x%lx\n vcpuid:%d\n",gpa,*sptep,newspte,detouraddr,rip,vcpu->vcpu_id);
+ *sptep = newspte;
+ } else
+ printk(KERN_ERR "split_tlb_flip_page: sptep not found for 0x%llx \n",gpa);
+ spin_unlock(&vcpu->kvm->mmu_lock);
+ _register_ept_flip(splitpage->gva,rip,cr3,vcpu->kvm->splitpages->vmcounter,true);
+ } else if (exit_qualification & PTE_EXECUTE) //execute
+ {
+ u64* sptep;
+ hpa_t stepaddr = ts_gfn_to_pfa(vcpu,gfn);
+// if (async || !writable)
+// printk(KERN_WARNING "split_tlb_flip_page: unexpected async:%d writable%d\n", async, writable);
+ spin_lock(&vcpu->kvm->mmu_lock);
+ sptep = split_tlb_findspte(vcpu,gfn);
+ if (sptep!=NULL) {
+ u64 newspte = *sptep;
+ if (newspte==0) {
+ printk(KERN_INFO "split_tlb_flip_page: fallback to default handler (EXEC):0x%llx \n",gpa);
+ spin_unlock(&vcpu->kvm->mmu_lock);
+ return 0;
+ }
+ if ((newspte&(VMX_EPT_WRITABLE_MASK|VMX_EPT_EXECUTABLE_MASK|VMX_EPT_READABLE_MASK))==0) {
+ printk(KERN_INFO "split_tlb_flip_page: sptep last 3 bits are 0 for gpa:0x%llx \n",gpa);
+ }
+ newspte&=~(VMX_EPT_WRITABLE_MASK|VMX_EPT_READABLE_MASK);
+ newspte|=VMX_EPT_EXECUTABLE_MASK;
+ newspte&=~PT64_BASE_ADDR_MASK;
+ newspte|=stepaddr&PT64_BASE_ADDR_MASK;
+ //printk(KERN_WARNING "split_tlb_flip_page: execute EPT fault at 0x%llx/0x%llx -> 0x%llx detourpa:0x%llx rip:0x%lx\n vcpuid:%d\n",gpa,*sptep,newspte,detouraddr,rip,vcpu->vcpu_id);
+ *sptep = newspte;
+ } else
+ printk(KERN_ERR "split_tlb_flip_page: sptep not found for 0x%llx \n",gpa);
+ spin_unlock(&vcpu->kvm->mmu_lock);
+ _register_ept_flip(splitpage->gva,rip,cr3,vcpu->kvm->splitpages->vmcounter,false);
+ } else
+ printk(KERN_ERR "split_tlb_flip_page: unexpected EPT fault at 0x%llx \n",gpa);
+ return 1;
+}
+EXPORT_SYMBOL_GPL(split_tlb_flip_page);
+
+int deactivateAllPages(struct kvm_vcpu *vcpu) {
+ struct kvm_splitpages *spages = vcpu->kvm->splitpages;
+ int i;
+ for (i = 0; i < KVM_MAX_SPLIT_PAGES; i++) {
+ gva_t gva = spages->pages[i].gva;
+ gpa_t gpa = spages->pages[i].gpa;
+ if (gva) {
+ if (split_tlb_freepage_by_gpa(vcpu,gpa)==0) {
+ printk(KERN_WARNING "deactivateAllPages: split_tlb_freepage failed for gva=%lx/gpa=%llx attempting to fix and free it based on saved gpa\n",gva,gpa);
+ split_tlb_restore_spte(vcpu,gpa >> PAGE_SHIFT);
+
+ kvm_split_tlb_freepage(spages->pages+i);
+ }
+ }
+ }
+ return 1;
+}
+
+int isPageSplit(struct kvm_vcpu *vcpu, gva_t addr ) {
+ u32 access = (kvm_x86_ops->get_cpl(vcpu) == 3) ? PFERR_USER_MASK : 0;
+ struct kvm_splitpage* page;
+ struct x86_exception exception;
+ gpa_t addr_gpa = vcpu->arch.walk_mmu->gva_to_gpa(vcpu, addr, access, &exception);
+ if (addr_gpa == UNMAPPED_GVA)
+ return 0;
+// printk(KERN_WARNING "isPageSplit: address translated gva=%lx to gpa=0x%llx\n",addr,addr_gpa);
+ page = split_tlb_findpage(vcpu->kvm,addr_gpa);
+ if (page != NULL)
+ return 1;
+ else
+ return 0;
+}
+
+
+/*
+ * rax low word - opcode, upper 3 words will be used for secret code
+ *
+ * 0x0001: Set data for page.
+ * rbx - guest virtual address for page
+ * rdx - guest virtual address for page data
+ *
+ * 0x0002: Activate page.
+ * rbx - guest virtual address for page
+ *
+ * 0x0003: Write code for page. Only usable after page is active
+ * rcx - guest virtual address for data
+ * rdx - guest virtual address for destination
+ * r8 - number of bytes
+ *
+ * 0x0004: Deactivate page.
+ * rbx - guest virtual address for page
+ *
+ * 0x0005: check if support is present
+ *
+ * 0x0006: deactivate all
+ *
+ * return ax = 1 - success
+ * ax = 0 - failure
+ *
+ * 0x0007: page present
+ * rcx - guest virtual address for data
+ *
+ * return ax = 1 - present
+ * ax = 0 - not present
+ */
+
+int split_tlb_vmcall_dispatch(struct kvm_vcpu *vcpu)
+{
+ unsigned long rip,cr3,rax,rbx,rdx,rcx,r8;
+ int result;
+
+ rip = kvm_rip_read(vcpu);
+ cr3 = kvm_read_cr3(vcpu);
+ rax = kvm_register_read(vcpu, VCPU_REGS_RAX);
+ rbx = kvm_register_read(vcpu, VCPU_REGS_RBX);
+ rcx = kvm_register_read(vcpu, VCPU_REGS_RCX);
+ rdx = kvm_register_read(vcpu, VCPU_REGS_RDX);
+ r8 = kvm_register_read(vcpu, VCPU_REGS_R8);
+ //printk(KERN_DEBUG "VMCALL: rip:0x%lx cr3:0x%lx rax:0x%lx rbx:0x%lx rcx:0x%lx rdx:0x%lx r8:0x%lx\n",rip,cr3,rax,rbx,rcx,rdx,r8);
+
+ switch (rax&0xFFFF) {
+ case 0x0001:
+ result = split_tlb_setdatapage(vcpu,rbx,rdx,cr3);
+ break;
+ case 0x0002:
+ result = split_tlb_activatepage(vcpu,rbx,cr3);
+ break;
+ case 0x0003:
+ result = split_tlb_copymem(vcpu,rcx,rdx,r8);
+ break;
+ case 0x0004:
+ result = split_tlb_freepage(vcpu,rbx);
+ break;
+ case 0x0005:
+ result = 1;
+ break;
+ case 0x0006:
+ result = deactivateAllPages(vcpu);
+ break;
+ case 0x0007:
+ result = isPageSplit(vcpu,rcx);
+ break;
+ default:
+ result = 0;
+ printk(KERN_WARNING "VMCALL: invalid operation 0x%x \n",(unsigned short)(rax&0xFFFF));
+ }
+ kvm_register_write(vcpu, VCPU_REGS_RAX, result);
+ return 1;
+}
+EXPORT_SYMBOL_GPL(split_tlb_vmcall_dispatch);
+
+int split_tlb_has_split_page(struct kvm *kvms, u64* sptep) {
+ struct kvm_splitpage* found;
+ int i;
+ phys_addr_t pagehpa = *sptep & PT64_BASE_ADDR_MASK;
+ for (i=0; i<KVM_MAX_SPLIT_PAGES; i++) {
+ found = kvms->splitpages->pages+i;
+ if (found->active) {
+ phys_addr_t detouraddr = virt_to_phys(found->dataaddr);
+ printk(KERN_WARNING "split_tlb_has_split_page: comparing pagehpa:0x%llx with detouraddr:0x%llx\n",pagehpa,detouraddr);
+ if (pagehpa == detouraddr) {
+ printk(KERN_WARNING "split_tlb_has_split_page: found page gva:0x%lx reverting to code\n",found->gva);
+ split_tlb_flip_to_code(kvms,found->codeaddr,sptep);
+ return 1;
+ }
+ }
+ }
+ printk(KERN_WARNING "split_tlb_has_split_page: did not find split page spte:0x%llx\n",*sptep);
+ return 0;
+}
+
+int split_tlb_handle_ept_violation(struct kvm_vcpu *vcpu,gpa_t gpa,unsigned long exit_qualification,int* splitresult) {
+ struct kvm_splitpage* splitpage;
+
+ splitpage = split_tlb_findpage(vcpu->kvm,gpa);
+ if (splitpage!=NULL) {
+ //printk(KERN_DEBUG "handle_ept_violation on split page: 0x%llx exitqualification:%lx\n",gpa,exit_qualification);
+ if (split_tlb_flip_page(vcpu,gpa,splitpage,exit_qualification)){
+ int emulation_type = EMULTYPE_RETRY;
+ enum emulation_result er;
+ er = x86_emulate_instruction(vcpu, gpa, emulation_type, NULL, 0);
+ if (er==EMULATE_DONE) {
+ //printk(KERN_DEBUG "handle_ept_violation on split page after emulation EMULATE_DONE\n");
+ *splitresult = 1;
+ } else {
+ printk(KERN_WARNING "handle_ept_violation on split page after emulation %s\n",er==EMULATE_FAIL?"EMULATE_FAIL":"EMULATE_USER_EXIT or smth");
+ *splitresult = 0;
+ }
+ } else {
+ printk(KERN_WARNING "handle_ept_violation split_tlb_flip_page returned 0 page: 0x%llx",gpa);
+ return 0;
+ }
+ return 1;
+ }
+ return 0;
+}
+EXPORT_SYMBOL_GPL(split_tlb_handle_ept_violation);
diff --git a/arch/x86/kvm/tlbsplit.h b/arch/x86/kvm/tlbsplit.h
new file mode 100644
index 0000000..527372c
--- /dev/null
+++ b/arch/x86/kvm/tlbsplit.h
@@ -0,0 +1,52 @@
+/*
+ * tlbsplit.h
+ *
+ * Created on: Dec 28, 2015
+ * Author: nick
+ */
+
+#ifndef ARCH_X86_KVM_TLBSPLIT_H_
+#define ARCH_X86_KVM_TLBSPLIT_H_
+#include <linux/types.h>
+#include <linux/export.h>
+#include <linux/kvm_host.h>
+
+#define KVM_MAX_SPLIT_PAGES 100
+#define MAX_PATCH_SIZE 1024
+
+struct kvm_splitpage {
+ gpa_t gpa;
+ gva_t gva;
+ unsigned long cr3;
+ void * dataaddr;
+ hpa_t codeaddr;
+ bool active;
+};
+
+struct kvm_splitpages {
+ struct kvm_splitpage pages[KVM_MAX_SPLIT_PAGES];
+ int vmcounter;
+};
+
+bool tlb_split_init(struct kvm *kvm);
+void kvm_split_tlb_freepage(struct kvm_splitpage *page);
+void kvm_split_tlb_deactivateall(struct kvm *kvm);
+void split_init_debugfs(void);
+void split_shutdown_debugfs(void);
+
+struct kvm_splitpage* split_tlb_findpage(struct kvm *kvms,gpa_t gpa);
+int split_tlb_activatepage(struct kvm_vcpu *vcpu, gva_t gva, ulong cr3);
+int split_tlb_setdatapage(struct kvm_vcpu *vcpu, gva_t gva, gva_t datagva, ulong cr3);
+int split_tlb_flip_page(struct kvm_vcpu *vcpu, gpa_t gpa, struct kvm_splitpage* splitpage, unsigned long exit_qualification);
+int split_tlb_freepage(struct kvm_vcpu *vcpu, gva_t gva);
+int split_tlb_vmcall_dispatch(struct kvm_vcpu *vcpu);
+int split_tlb_handle_ept_violation(struct kvm_vcpu *vcpu,gpa_t gpa,unsigned long exit_qualification,int* splitresult);
+int split_tlb_has_split_page(struct kvm *kvms, u64* sptep);
+int split_tlb_restore_spte(struct kvm_vcpu *vcpu,gfn_t gfn);
+//int split_tlb_restore_spte_base(struct kvm *kvms,gfn_t gfn,u64* sptep);
+int split_tlb_flip_to_code(struct kvm *kvms,hpa_t hpa,u64* sptep);
+
+#define COULD_BE_SPLIT_PAGE(spte) ( (spte&VMX_EPT_WRITABLE_MASK)==0 && (spte&(VMX_EPT_READABLE_MASK|VMX_EPT_EXECUTABLE_MASK))!=0 \
+&& ( spte&(VMX_EPT_READABLE_MASK|VMX_EPT_EXECUTABLE_MASK))!=(VMX_EPT_READABLE_MASK|VMX_EPT_EXECUTABLE_MASK) )
+
+#endif /* ARCH_X86_KVM_TLBSPLIT_H_ */
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index a236dec..7800700 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -49,6 +49,8 @@
#include <asm/apic.h>
#include <asm/irq_remapping.h>
+#include "tlbsplit.h"
+
#include "trace.h"
#include "pmu.h"
@@ -6178,6 +6180,10 @@ static int handle_halt(struct kvm_vcpu *vcpu)
static int handle_vmcall(struct kvm_vcpu *vcpu)
{
+ if (split_tlb_vmcall_dispatch(vcpu)) {
+ kvm_skip_emulated_instruction(vcpu);
+ return 1;
+ }
return kvm_emulate_hypercall(vcpu);
}
@@ -6344,6 +6350,7 @@ static int handle_ept_violation(struct kvm_vcpu *vcpu)
gpa_t gpa;
u32 error_code;
int gla_validity;
+ int splitresult;
exit_qualification = vmcs_readl(EXIT_QUALIFICATION);
@@ -6385,7 +6392,10 @@ static int handle_ept_violation(struct kvm_vcpu *vcpu)
vcpu->arch.exit_qualification = exit_qualification;
- return kvm_mmu_page_fault(vcpu, gpa, error_code, NULL, 0);
+ if (split_tlb_handle_ept_violation(vcpu,gpa,exit_qualification,&splitresult))
+ return splitresult;
+ else
+ return kvm_mmu_page_fault(vcpu, gpa, error_code, NULL, 0);
}
static int handle_ept_misconfig(struct kvm_vcpu *vcpu)
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index 1c5190d..6665afb 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -376,6 +376,7 @@ struct kvm {
struct mutex slots_lock;
struct mm_struct *mm; /* userspace tied to this vm */
struct kvm_memslots *memslots[KVM_ADDRESS_SPACE_NUM];
+ struct kvm_splitpages *splitpages; //splittlb
struct srcu_struct srcu;
struct srcu_struct irq_srcu;
struct kvm_vcpu *vcpus[KVM_MAX_VCPUS];
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 482612b..6b9cd77 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -62,6 +62,7 @@
#define CREATE_TRACE_POINTS
#include <trace/events/kvm.h>
+#include "tlbsplit.h"
/* Worst case buffer size needed for holding an integer. */
#define ITOA_MAX_LEN 12
@@ -646,6 +647,9 @@ static struct kvm *kvm_create_vm(unsigned long type)
goto out_err_no_srcu;
}
+ if (!tlb_split_init(kvm))
+ goto out_err_no_srcu;
+
if (init_srcu_struct(&kvm->srcu))
goto out_err_no_srcu;
if (init_srcu_struct(&kvm->irq_srcu))
@@ -735,6 +739,7 @@ static void kvm_destroy_vm(struct kvm *kvm)
kvm_destroy_devices(kvm);
for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++)
kvm_free_memslots(kvm, kvm->memslots[i]);
+ kvm_split_tlb_deactivateall(kvm); //splittbl
cleanup_srcu_struct(&kvm->irq_srcu);
cleanup_srcu_struct(&kvm->srcu);
kvm_arch_free_vm(kvm);
@@ -3850,6 +3855,7 @@ static int kvm_init_debug(void)
goto out_dir;
}
+ split_init_debugfs();
return 0;
out_dir:
@@ -4016,6 +4022,7 @@ EXPORT_SYMBOL_GPL(kvm_init);
void kvm_exit(void)
{
+ split_shutdown_debugfs(); //splittlb
debugfs_remove_recursive(kvm_debugfs_dir);
misc_deregister(&kvm_dev);
kmem_cache_destroy(kvm_vcpu_cache);