-
Notifications
You must be signed in to change notification settings - Fork 12
/
uhyve-x86_64.c
1444 lines (1214 loc) · 42.2 KB
/
uhyve-x86_64.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
/*
* Copyright (c) 2018, Stefan Lankes, RWTH Aachen University
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* * Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this
* software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
* DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE FOR ANY
* DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
* LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
* ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
* SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifdef __x86_64__
#define _GNU_SOURCE
#include <elf.h>
#include <err.h>
#include <errno.h>
#include <fcntl.h>
#include <limits.h>
#include <linux/const.h>
#include <linux/kvm.h>
#include <poll.h>
#include <pthread.h>
#include <sched.h>
#include <semaphore.h>
#include <signal.h>
#include <stdbool.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/eventfd.h>
#include <sys/ioctl.h>
#include <sys/mman.h>
#include <sys/stat.h>
#include <sys/time.h>
#include <sys/types.h>
#include <sys/wait.h>
#include <unistd.h>
#ifdef HAVE_MSR_INDEX_H
#include <asm/msr-index.h>
#else
/* x86-64 specific MSRs */
#define MSR_EFER 0xc0000080 /* extended feature register */
#define MSR_STAR 0xc0000081 /* legacy mode SYSCALL target */
#define MSR_LSTAR 0xc0000082 /* long mode SYSCALL target */
#define MSR_CSTAR 0xc0000083 /* compat mode SYSCALL target */
#define MSR_SYSCALL_MASK 0xc0000084 /* EFLAGS mask for syscall */
#define MSR_FS_BASE 0xc0000100 /* 64bit FS base */
#define MSR_GS_BASE 0xc0000101 /* 64bit GS base */
#define MSR_KERNEL_GS_BASE 0xc0000102 /* SwapGS GS shadow */
#define MSR_TSC_AUX 0xc0000103 /* Auxiliary TSC */
#define MSR_IA32_CR_PAT 0x00000277
#define MSR_PEBS_FRONTEND 0x000003f7
#define MSR_IA32_POWER_CTL 0x000001fc
#define MSR_IA32_MC0_CTL 0x00000400
#define MSR_IA32_MC0_STATUS 0x00000401
#define MSR_IA32_MC0_ADDR 0x00000402
#define MSR_IA32_MC0_MISC 0x00000403
#define MSR_IA32_SYSENTER_CS 0x00000174
#define MSR_IA32_SYSENTER_ESP 0x00000175
#define MSR_IA32_SYSENTER_EIP 0x00000176
#define MSR_IA32_APICBASE 0x0000001b
#define MSR_IA32_APICBASE_BSP (1<<8)
#define MSR_IA32_APICBASE_ENABLE (1<<11)
#define MSR_IA32_APICBASE_BASE (0xfffff<<12)
#define MSR_IA32_MISC_ENABLE 0x000001a0
#define MSR_IA32_TSC 0x00000010
/* EFER bits: */
#define _EFER_SCE 0 /* SYSCALL/SYSRET */
#define _EFER_LME 8 /* Long mode enable */
#define _EFER_LMA 10 /* Long mode active (read-only) */
#define _EFER_NX 11 /* No execute enable */
#define _EFER_SVME 12 /* Enable virtualization */
#define _EFER_LMSLE 13 /* Long Mode Segment Limit Enable */
#define _EFER_FFXSR 14 /* Enable Fast FXSAVE/FXRSTOR */
#define EFER_SCE (1<<_EFER_SCE)
#define EFER_LME (1<<_EFER_LME)
#define EFER_LMA (1<<_EFER_LMA)
#define EFER_NX (1<<_EFER_NX)
#define EFER_SVME (1<<_EFER_SVME)
#define EFER_LMSLE (1<<_EFER_LMSLE)
#define EFER_FFXSR (1<<_EFER_FFXSR)
#endif
#include <asm/mman.h>
#include "uhyve-common.h"
#include "uhyve-gdb.h"
#include "uhyve-migration.h"
#include "uhyve-net.h"
#include "uhyve-syscalls.h"
#include "uhyve-x86_64.h"
#include "uhyve.h"
// define this macro to create checkpoints with KVM's dirty log
//#define USE_DIRTY_LOG
#define MAX_FNAME 256
#define GUEST_OFFSET 0x0
#define CPUID_FUNC_PERFMON 0x0A
#define GUEST_PAGE_SIZE 0x200000 /* 2 MB pages in guest */
#define BOOT_GDT 0x1000
#define BOOT_INFO 0x2000
#define BOOT_PML4 0x10000
#define BOOT_PDPTE 0x11000
#define BOOT_PDE 0x12000
#define BOOT_GDT_NULL 0
#define BOOT_GDT_CODE 1
#define BOOT_GDT_DATA 2
#define BOOT_GDT_MAX 3
#define KVM_32BIT_MAX_MEM_SIZE (1ULL << 32)
#define KVM_32BIT_GAP_SIZE (768 << 20)
#define KVM_32BIT_GAP_START (KVM_32BIT_MAX_MEM_SIZE - KVM_32BIT_GAP_SIZE)
#define KVM_32BIT_GAP_END (KVM_32BIT_GAP_START + KVM_32BIT_GAP_SIZE)
/// Page offset bits
#define PAGE_BITS 12
#define PAGE_2M_BITS 21
#define PAGE_SIZE (1L << PAGE_BITS)
/// Mask the page address without page map flags and XD flag
#if 0
#define PAGE_MASK ((~0L) << PAGE_BITS)
#define PAGE_2M_MASK (~0L) << PAGE_2M_BITS)
#else
#define PAGE_MASK (((~0UL) << PAGE_BITS) & ~PG_XD)
#define PAGE_2M_MASK (((~0UL) << PAGE_2M_BITS) & ~PG_XD)
#endif
// Page is present
#define PG_PRESENT (1 << 0)
// Page is read- and writable
#define PG_RW (1 << 1)
// Page is addressable from userspace
#define PG_USER (1 << 2)
// Page write through is activated
#define PG_PWT (1 << 3)
// Page cache is disabled
#define PG_PCD (1 << 4)
// Page was recently accessed (set by CPU)
#define PG_ACCESSED (1 << 5)
// Page is dirty due to recent write-access (set by CPU)
#define PG_DIRTY (1 << 6)
// Huge page: 4MB (or 2MB, 1GB)
#define PG_PSE (1 << 7)
// Page attribute table
#define PG_PAT PG_PSE
#if 1
/* @brief Global TLB entry (Pentium Pro and later)
*
* HermitCore is a single-address space operating system
* => CR3 never changed => The flag isn't required for HermitCore
*/
#define PG_GLOBAL 0
#else
#define PG_GLOBAL (1 << 8)
#endif
// This table is a self-reference and should skipped by page_map_copy()
#define PG_SELF (1 << 9)
/// Disable execution for this page
#define PG_XD (1L << 63)
#define BITS 64
#define PHYS_BITS 52
#define VIRT_BITS 48
#define PAGE_MAP_BITS 9
#define PAGE_MAP_MASK 0x1FF
#define PAGE_LEVELS 4
#define IOAPIC_DEFAULT_BASE 0xfec00000
#define APIC_DEFAULT_BASE 0xfee00000
static bool cap_tsc_deadline = false;
static bool cap_irqchip = false;
static bool cap_adjust_clock_stable = false;
static bool cap_irqfd = false;
static bool cap_vapic = false;
FILE *chk_file = NULL;
extern size_t guest_size;
extern pthread_barrier_t barrier;
extern pthread_barrier_t migration_barrier;
extern pthread_t* vcpu_threads;
extern uint64_t elf_entry;
extern uint8_t* klog;
extern bool verbose;
extern bool full_checkpoint;
extern uint32_t no_checkpoint;
extern uint32_t ncores;
extern uint8_t* guest_mem;
extern size_t guest_size;
extern int kvm, vmfd, netfd, efd, mig_efd;
extern volatile kernel_header_t* kheader;
extern __thread struct kvm_run *run;
extern __thread int vcpufd;
extern __thread uint32_t cpuid;
extern vcpu_state_t *vcpu_thread_states;
extern mem_mappings_t mem_mappings;
extern mem_mappings_t guest_physical_memory;
static inline void show_dtable(const char *name, struct kvm_dtable *dtable)
{
fprintf(stderr, " %s %016zx %08hx\n", name, (size_t) dtable->base, (uint16_t) dtable->limit);
}
static inline void show_segment(const char *name, struct kvm_segment *seg)
{
fprintf(stderr, " %s %04hx %016zx %08x %02hhx %x %x %x %x %x %x %x\n",
name, (uint16_t) seg->selector, (size_t) seg->base, (uint32_t) seg->limit,
(uint8_t) seg->type, seg->present, seg->dpl, seg->db, seg->s, seg->l, seg->g, seg->avl);
}
static void show_registers(int id, struct kvm_regs* regs, struct kvm_sregs* sregs)
{
size_t cr0, cr2, cr3;
size_t cr4, cr8;
size_t rax, rbx, rcx;
size_t rdx, rsi, rdi;
size_t rbp, r8, r9;
size_t r10, r11, r12;
size_t r13, r14, r15;
size_t rip, rsp;
size_t rflags;
int i;
rflags = regs->rflags;
rip = regs->rip; rsp = regs->rsp;
rax = regs->rax; rbx = regs->rbx; rcx = regs->rcx;
rdx = regs->rdx; rsi = regs->rsi; rdi = regs->rdi;
rbp = regs->rbp; r8 = regs->r8; r9 = regs->r9;
r10 = regs->r10; r11 = regs->r11; r12 = regs->r12;
r13 = regs->r13; r14 = regs->r14; r15 = regs->r15;
fprintf(stderr, "\n Dump state of CPU %d\n", id);
fprintf(stderr, "\n Registers:\n");
fprintf(stderr, " ----------\n");
fprintf(stderr, " rip: %016zx rsp: %016zx flags: %016zx\n", rip, rsp, rflags);
fprintf(stderr, " rax: %016zx rbx: %016zx rcx: %016zx\n", rax, rbx, rcx);
fprintf(stderr, " rdx: %016zx rsi: %016zx rdi: %016zx\n", rdx, rsi, rdi);
fprintf(stderr, " rbp: %016zx r8: %016zx r9: %016zx\n", rbp, r8, r9);
fprintf(stderr, " r10: %016zx r11: %016zx r12: %016zx\n", r10, r11, r12);
fprintf(stderr, " r13: %016zx r14: %016zx r15: %016zx\n", r13, r14, r15);
cr0 = sregs->cr0; cr2 = sregs->cr2; cr3 = sregs->cr3;
cr4 = sregs->cr4; cr8 = sregs->cr8;
fprintf(stderr, " cr0: %016zx cr2: %016zx cr3: %016zx\n", cr0, cr2, cr3);
fprintf(stderr, " cr4: %016zx cr8: %016zx\n", cr4, cr8);
fprintf(stderr, "\n Segment registers:\n");
fprintf(stderr, " ------------------\n");
fprintf(stderr, " register selector base limit type p dpl db s l g avl\n");
show_segment("cs ", &sregs->cs);
show_segment("ss ", &sregs->ss);
show_segment("ds ", &sregs->ds);
show_segment("es ", &sregs->es);
show_segment("fs ", &sregs->fs);
show_segment("gs ", &sregs->gs);
show_segment("tr ", &sregs->tr);
show_segment("ldt", &sregs->ldt);
show_dtable("gdt", &sregs->gdt);
show_dtable("idt", &sregs->idt);
fprintf(stderr, "\n APIC:\n");
fprintf(stderr, " -----\n");
fprintf(stderr, " efer: %016zx apic base: %016zx\n",
(size_t) sregs->efer, (size_t) sregs->apic_base);
fprintf(stderr, "\n Interrupt bitmap:\n");
fprintf(stderr, " -----------------\n");
for (i = 0; i < (KVM_NR_INTERRUPTS + 63) / 64; i++)
fprintf(stderr, " %016zx", (size_t) sregs->interrupt_bitmap[i]);
fprintf(stderr, "\n");
}
void print_registers(void)
{
struct kvm_regs regs;
struct kvm_sregs sregs;
kvm_ioctl(vcpufd, KVM_GET_SREGS, &sregs);
kvm_ioctl(vcpufd, KVM_GET_REGS, ®s);
show_registers(cpuid, ®s, &sregs);
}
/// Filter CPUID functions that are not supported by the hypervisor and enable
/// features according to our needs.
static void filter_cpuid(struct kvm_cpuid2 *kvm_cpuid)
{
for (uint32_t i = 0; i < kvm_cpuid->nent; i++) {
struct kvm_cpuid_entry2 *entry = &kvm_cpuid->entries[i];
switch (entry->function) {
case 1:
// CPUID to define basic cpu features
entry->ecx |= (1U << 31); // propagate that we are running on a hypervisor
if (cap_tsc_deadline)
entry->ecx |= (1U << 24); // enable TSC deadline feature
entry->edx |= (1U << 5); // enable msr support
break;
case CPUID_FUNC_PERFMON:
// disable it
entry->eax = 0x00;
break;
default:
// Keep the CPUID function as-is
break;
};
}
}
static void setup_system_64bit(struct kvm_sregs *sregs)
{
sregs->cr0 |= X86_CR0_PE;
sregs->cr4 |= X86_CR4_PAE;
sregs->efer |= EFER_LME|EFER_LMA|EFER_NX;
}
static void setup_system_page_tables(struct kvm_sregs *sregs, uint8_t *mem)
{
uint64_t *pml4 = (uint64_t *) (mem + BOOT_PML4);
uint64_t *pdpte = (uint64_t *) (mem + BOOT_PDPTE);
uint64_t *pde = (uint64_t *) (mem + BOOT_PDE);
uint64_t paddr, i;
/*
* For simplicity we currently use 2MB pages and only a single
* PML4/PDPTE/PDE.
*/
memset(pml4, 0x00, 4096);
memset(pdpte, 0x00, 4096);
memset(pde, 0x00, 4096);
pml4[0] = BOOT_PDPTE | (X86_PDPT_P | X86_PDPT_RW);
pml4[511] = BOOT_PML4 | (X86_PDPT_P | X86_PDPT_RW);
pdpte[0] = BOOT_PDE | (X86_PDPT_P | X86_PDPT_RW);
for (paddr = 0, i = 0; i < 512; paddr += GUEST_PAGE_SIZE, i++)
pde[i] = paddr | (X86_PDPT_P | X86_PDPT_RW | X86_PDPT_PS);
sregs->cr3 = BOOT_PML4;
sregs->cr4 |= X86_CR4_PAE;
sregs->cr0 |= X86_CR0_PG;
}
static void setup_system_gdt(struct kvm_sregs *sregs,
uint8_t *mem,
uint64_t off)
{
uint64_t *gdt = (uint64_t *) (mem + off);
struct kvm_segment data_seg, code_seg;
/* flags, base, limit */
gdt[BOOT_GDT_NULL] = GDT_ENTRY(0, 0, 0);
gdt[BOOT_GDT_CODE] = GDT_ENTRY(0xA09B, 0, 0xFFFFF);
gdt[BOOT_GDT_DATA] = GDT_ENTRY(0xC093, 0, 0xFFFFF);
sregs->gdt.base = off;
sregs->gdt.limit = (sizeof(uint64_t) * BOOT_GDT_MAX) - 1;
GDT_TO_KVM_SEGMENT(code_seg, gdt, BOOT_GDT_CODE);
GDT_TO_KVM_SEGMENT(data_seg, gdt, BOOT_GDT_DATA);
sregs->cs = code_seg;
sregs->ds = data_seg;
sregs->es = data_seg;
sregs->fs = data_seg;
sregs->gs = data_seg;
sregs->ss = data_seg;
}
static void setup_system(int vcpufd, uint8_t *mem, uint32_t id)
{
static struct kvm_sregs sregs;
// all cores use the same startup code
// => all cores use the same sregs
// => only the boot processor has to initialize sregs
if (id == 0) {
kvm_ioctl(vcpufd, KVM_GET_SREGS, &sregs);
/* Set all cpu/mem system structures */
setup_system_gdt(&sregs, mem, BOOT_GDT);
setup_system_page_tables(&sregs, mem);
setup_system_64bit(&sregs);
}
kvm_ioctl(vcpufd, KVM_SET_SREGS, &sregs);
}
static void setup_cpuid(int kvm, int vcpufd)
{
struct kvm_cpuid2 *kvm_cpuid;
unsigned int max_entries = 100;
// allocate space for cpuid we get from KVM
kvm_cpuid = calloc(1, sizeof(*kvm_cpuid) + (max_entries * sizeof(kvm_cpuid->entries[0])));
kvm_cpuid->nent = max_entries;
kvm_ioctl(kvm, KVM_GET_SUPPORTED_CPUID, kvm_cpuid);
// set features
filter_cpuid(kvm_cpuid);
kvm_ioctl(vcpufd, KVM_SET_CPUID2, kvm_cpuid);
free(kvm_cpuid);
}
/**
* \brief Fills mem_mappings with chunks of guest-physical memory
*/
static inline void
determine_guest_physical_memory_regions(mem_mappings_t *mem_mappings)
{
if (guest_size < KVM_32BIT_GAP_START) {
mem_mappings->count = 1;
mem_mappings->mem_chunks = (mem_chunk_t*)malloc(sizeof(mem_chunk_t)*mem_mappings->count);
mem_mappings->mem_chunks[0].ptr = 0;
mem_mappings->mem_chunks[0].size = guest_size;
} else {
mem_mappings->count = 2;
mem_mappings->mem_chunks = (mem_chunk_t*)malloc(sizeof(mem_chunk_t)*mem_mappings->count);
mem_mappings->mem_chunks[0].ptr = 0;
mem_mappings->mem_chunks[0].size = KVM_32BIT_GAP_START;
mem_mappings->mem_chunks[1].ptr = (uint8_t*)((uint64_t)(KVM_32BIT_GAP_START + KVM_32BIT_GAP_SIZE));
mem_mappings->mem_chunks[1].size = (uint64_t)guest_size - (KVM_32BIT_GAP_START + KVM_32BIT_GAP_SIZE);
}
return;
}
static void convert_to_host_virt(mem_mappings_t *mem_mappings)
{
size_t i = 0;
/* convert guest-physical addrees to host-virtual */
fprintf(stderr, "[INFO] We have %zu memory chunks\n", mem_mappings->count);
fprintf(stderr, "[INFO] ");
for (i=0; i<mem_mappings->count; ++i) {
fprintf(stderr, "(ADDR[%lu]: 0x%p, SIZE[%lu]: 0x%lx); ",
i,
mem_mappings->mem_chunks[i].ptr,
i,
mem_mappings->mem_chunks[i].size);
}
printf("\b\b\n");
for (i=0; i<mem_mappings->count; ++i) {
uint8_t *cur_ptr = mem_mappings->mem_chunks[i].ptr;
mem_mappings->mem_chunks[i].ptr = (uint8_t*)(guest_mem+(size_t)cur_ptr);
}
}
/* checks wether the mem_mappings.mem_chunks[cur_slot] contains the IO gap */
static inline void check_for_io_gap(size_t *cur_slot)
{
size_t cur_start = (size_t)mem_mappings.mem_chunks[*cur_slot].ptr;
size_t cur_end = (size_t)mem_mappings.mem_chunks[*cur_slot].ptr + (size_t)mem_mappings.mem_chunks[*cur_slot].size;
if (cur_start == KVM_32BIT_GAP_START && cur_end == KVM_32BIT_GAP_END) {
/* remove the chunk */
*cur_slot -= 1;
mem_mappings.count--;
} else if (cur_end == KVM_32BIT_GAP_END) {
/* shrink element */
mem_mappings.mem_chunks[*cur_slot].size -= KVM_32BIT_GAP_SIZE;
} else if (cur_start == KVM_32BIT_GAP_START) {
/* shrink element and move start pointer*/
mem_mappings.mem_chunks[*cur_slot].size -= KVM_32BIT_GAP_SIZE;
mem_mappings.mem_chunks[*cur_slot].ptr += KVM_32BIT_GAP_SIZE;
} else if ((cur_start < KVM_32BIT_GAP_START) &&
(cur_end > KVM_32BIT_GAP_END)) {
/* split element */
mem_mappings.mem_chunks[*cur_slot].size = KVM_32BIT_GAP_START - cur_start;
*cur_slot += 1;
mem_mappings.mem_chunks[*cur_slot].ptr = (uint8_t*)KVM_32BIT_GAP_END;
mem_mappings.mem_chunks[*cur_slot].size = cur_end - KVM_32BIT_GAP_END;
mem_mappings.count++;
}
return;
}
size_t determine_dest_offset(size_t src_addr)
{
size_t ret = 0;
if (src_addr & PG_PSE) {
ret = src_addr & PAGE_2M_MASK;
} else {
ret = src_addr & PAGE_MASK;
}
return ret;
}
void init_cpu_state(uint64_t elf_entry)
{
struct kvm_regs regs = {
.rip = elf_entry, // entry point to HermitCore
.rdi = BOOT_INFO_ADDR, // location of the boot info
.rflags = 0x2, // POR value required by x86 architecture
};
struct kvm_mp_state mp_state = { KVM_MP_STATE_RUNNABLE };
struct {
struct kvm_msrs info;
struct kvm_msr_entry entries[MAX_MSR_ENTRIES];
} msr_data;
struct kvm_msr_entry *msrs = msr_data.entries;
run->apic_base = APIC_DEFAULT_BASE;
setup_cpuid(kvm, vcpufd);
// be sure that the multiprocessor is runable
kvm_ioctl(vcpufd, KVM_SET_MP_STATE, &mp_state);
// enable fast string operations
msrs[0].index = MSR_IA32_MISC_ENABLE;
msrs[0].data = 1;
msr_data.info.nmsrs = 1;
kvm_ioctl(vcpufd, KVM_SET_MSRS, &msr_data);
// only one core is able to enter startup code
// => the wait for the predecessor core
while (kheader->cpu_online < cpuid)
pthread_yield();
kheader->current_boot_id = cpuid;
/* Setup registers and memory. */
setup_system(vcpufd, guest_mem, cpuid);
kvm_ioctl(vcpufd, KVM_SET_REGS, ®s);
}
vcpu_state_t read_cpu_state(void)
{
vcpu_state_t cpu_state;
char fname[MAX_FNAME];
snprintf(fname, MAX_FNAME, "checkpoint/chk%u_core%u.dat", no_checkpoint, cpuid);
FILE* f = fopen(fname, "r");
if (f == NULL)
err(1, "fopen: unable to open file");
if (fread(&cpu_state, sizeof(cpu_state), 1, f) != 1)
err(1, "fread failed\n");
fclose(f);
return cpu_state;
}
void restore_cpu_state(vcpu_state_t cpu_state)
{
cpu_state.mp_state.mp_state = KVM_MP_STATE_RUNNABLE;
run->apic_base = APIC_DEFAULT_BASE;
setup_cpuid(kvm, vcpufd);
kvm_ioctl(vcpufd, KVM_SET_SREGS, &cpu_state.sregs);
kvm_ioctl(vcpufd, KVM_SET_REGS, &cpu_state.regs);
kvm_ioctl(vcpufd, KVM_SET_MSRS, &cpu_state.msr_data);
kvm_ioctl(vcpufd, KVM_SET_XCRS, &cpu_state.xcrs);
kvm_ioctl(vcpufd, KVM_SET_MP_STATE, &cpu_state.mp_state);
kvm_ioctl(vcpufd, KVM_SET_LAPIC, &cpu_state.lapic);
kvm_ioctl(vcpufd, KVM_SET_FPU, &cpu_state.fpu);
kvm_ioctl(vcpufd, KVM_SET_XSAVE, &cpu_state.xsave);
kvm_ioctl(vcpufd, KVM_SET_VCPU_EVENTS, &cpu_state.events);
}
vcpu_state_t save_cpu_state(void)
{
int n = 0;
vcpu_state_t cpu_state;
/* define the list of required MSRs */
cpu_state.msr_data.entries[n++].index = MSR_IA32_APICBASE;
cpu_state.msr_data.entries[n++].index = MSR_IA32_SYSENTER_CS;
cpu_state.msr_data.entries[n++].index = MSR_IA32_SYSENTER_ESP;
cpu_state.msr_data.entries[n++].index = MSR_IA32_SYSENTER_EIP;
cpu_state.msr_data.entries[n++].index = MSR_IA32_CR_PAT;
cpu_state.msr_data.entries[n++].index = MSR_IA32_MISC_ENABLE;
cpu_state.msr_data.entries[n++].index = MSR_IA32_TSC;
cpu_state.msr_data.entries[n++].index = MSR_CSTAR;
cpu_state.msr_data.entries[n++].index = MSR_STAR;
cpu_state.msr_data.entries[n++].index = MSR_EFER;
cpu_state.msr_data.entries[n++].index = MSR_LSTAR;
cpu_state.msr_data.entries[n++].index = MSR_GS_BASE;
cpu_state.msr_data.entries[n++].index = MSR_FS_BASE;
cpu_state.msr_data.entries[n++].index = MSR_KERNEL_GS_BASE;
//msrs[n++].index = MSR_IA32_FEATURE_CONTROL;
cpu_state.msr_data.info.nmsrs = n;
kvm_ioctl(vcpufd, KVM_GET_SREGS, &cpu_state.sregs);
kvm_ioctl(vcpufd, KVM_GET_REGS, &cpu_state.regs);
kvm_ioctl(vcpufd, KVM_GET_MSRS, &cpu_state.msr_data);
kvm_ioctl(vcpufd, KVM_GET_XCRS, &cpu_state.xcrs);
kvm_ioctl(vcpufd, KVM_GET_LAPIC, &cpu_state.lapic);
kvm_ioctl(vcpufd, KVM_GET_FPU, &cpu_state.fpu);
kvm_ioctl(vcpufd, KVM_GET_XSAVE, &cpu_state.xsave);
kvm_ioctl(vcpufd, KVM_GET_VCPU_EVENTS, &cpu_state.events);
kvm_ioctl(vcpufd, KVM_GET_MP_STATE, &cpu_state.mp_state);
return cpu_state;
}
void write_cpu_state(void)
{
vcpu_state_t cpu_state = save_cpu_state();
char fname[MAX_FNAME];
snprintf(fname, MAX_FNAME, "checkpoint/chk%u_core%u.dat", no_checkpoint, cpuid);
FILE* f = fopen(fname, "w");
if (f == NULL) {
err(1, "fopen: unable to open file\n");
}
if (fwrite(&cpu_state, sizeof(cpu_state), 1, f) != 1)
err(1, "fwrite failed\n");
fclose(f);
}
void scan_dirty_log(void (*save_page)(void*, size_t, void*, size_t))
{
size_t slot_offset = 0;
static struct kvm_dirty_log dlog = {
.slot = 0,
.dirty_bitmap = NULL
};
size_t dirty_log_size = (guest_size >> PAGE_BITS) / sizeof(size_t);
// do we create our first checkpoint
if (dlog.dirty_bitmap == NULL) {
// besure that all paddings are zero
memset(&dlog, 0x00, sizeof(dlog));
dlog.dirty_bitmap = malloc(dirty_log_size * sizeof(size_t));
if (dlog.dirty_bitmap == NULL)
err(1, "malloc failed!\n");
}
memset(dlog.dirty_bitmap, 0x00, dirty_log_size * sizeof(size_t));
dlog.slot = 0;
nextslot:
kvm_ioctl(vmfd, KVM_GET_DIRTY_LOG, &dlog);
for(size_t i=0; i<dirty_log_size; i++) {
size_t value = ((size_t*) dlog.dirty_bitmap)[i];
if (value) {
for(size_t j=0; j<sizeof(size_t)*8; j++) {
size_t test = 1ULL << j;
if ((value & test) == test) {
size_t addr = (i*sizeof(size_t)*8+j)*PAGE_SIZE + slot_offset;
save_page(&addr, sizeof(size_t), (void*)((uint64_t)guest_mem+(uint64_t)addr), PAGE_SIZE);
}
}
}
}
// do we have to check the second slot?
if ((dlog.slot == 0) && (guest_size > KVM_32BIT_GAP_START - GUEST_OFFSET)) {
slot_offset = KVM_32BIT_MAX_MEM_SIZE;
dlog.slot = 1;
memset(dlog.dirty_bitmap, 0x00, dirty_log_size * sizeof(size_t));
goto nextslot;
}
}
void scan_page_tables(void (*save_page)(void*, size_t, void*, size_t))
{
const size_t flag = (!full_checkpoint && (no_checkpoint > 0)) ? PG_DIRTY : PG_ACCESSED;
size_t* pml4 = (size_t*) (guest_mem+BOOT_PML4);
for(size_t i=0; i<(1 << PAGE_MAP_BITS); i++) {
if ((pml4[i] & PG_PRESENT) != PG_PRESENT)
continue;
//printf("pml[%zd] 0x%zx\n", i, pml4[i]);
size_t* pdpt = (size_t*) (guest_mem+(pml4[i] & PAGE_MASK));
for(size_t j=0; j<(1 << PAGE_MAP_BITS); j++) {
if ((pdpt[j] & PG_PRESENT) != PG_PRESENT)
continue;
//printf("\tpdpt[%zd] 0x%zx\n", j, pdpt[j]);
size_t* pgd = (size_t*) (guest_mem+(pdpt[j] & PAGE_MASK));
for(size_t k=0; k<(1 << PAGE_MAP_BITS); k++) {
if ((pgd[k] & PG_PRESENT) != PG_PRESENT)
continue;
//printf("\t\tpgd[%zd] 0x%zx\n", k, pgd[k] & ~PG_XD);
if ((pgd[k] & PG_PSE) != PG_PSE) {
size_t* pgt = (size_t*) (guest_mem+(pgd[k] & PAGE_MASK));
for(size_t l=0; l<(1 << PAGE_MAP_BITS); l++) {
if ((pgt[l] & (PG_PRESENT|flag)) == (PG_PRESENT|flag)) {
//printf("\t\t\t*pgt[%zd] 0x%zx, 4KB\n", l, pgt[l] & ~PG_XD);
if (!full_checkpoint)
pgt[l] = pgt[l] & ~(PG_DIRTY|PG_ACCESSED);
size_t pgt_entry = pgt[l] & ~PG_PSE; // because PAT use the same bit as PSE
save_page(&pgt_entry, sizeof(size_t), (void*) (guest_mem + (pgt[l] & PAGE_MASK)), (1UL << PAGE_BITS));
}
}
} else if ((pgd[k] & flag) == flag) {
//printf("\t\t*pgd[%zd] 0x%zx, 2MB\n", k, pgd[k] & ~PG_XD);
if (!full_checkpoint)
pgd[k] = pgd[k] & ~(PG_DIRTY|PG_ACCESSED);
save_page(pgd+k, sizeof(size_t), (void*) (guest_mem + (pgd[k] & PAGE_2M_MASK)), (1UL << PAGE_2M_BITS));
}
}
}
}
}
void open_chk_file(char *fname)
{
chk_file = fopen(fname, "w");
if (chk_file == NULL) {
err(1, "fopen: unable to open file");
}
}
/* determine guests memory mappings based on its free list
*
* The memory mappings are stored in terms of guest-physical pointers
*/
void determine_mem_mappings(free_list_t *free_list)
{
/* determine list length */
size_t free_list_length = 0;
free_list_t *cur = free_list;
for (free_list_length=0; (size_t)cur != (size_t)guest_mem; ++free_list_length) {
cur = virt_to_phys_with_offset(cur->next);
}
/* create mem_mapppings structur
* +1: alloc is possible one element longer
* +1: the IO_GAP might result in an additional element
*/
size_t mem_mappings_cnt_max = free_list_length+2;
mem_mappings.mem_chunks = (mem_chunk_t*)malloc(mem_mappings_cnt_max*sizeof(mem_chunk_t));
mem_mappings.count = mem_mappings_cnt_max-2;
/* allocation prior to first free region */
size_t cur_slot = 0;
mem_mappings.mem_chunks[0].ptr = (uint8_t*)0x800000;
mem_mappings.mem_chunks[0].size = (uint8_t*)free_list->start - mem_mappings.mem_chunks[0].ptr;
check_for_io_gap(&cur_slot);
/* iterate the free list */
cur = free_list;
free_list_t *next = virt_to_phys_with_offset(cur->next);
for (++cur_slot; (uint8_t*)next != guest_mem; ++cur_slot) {
mem_mappings.mem_chunks[cur_slot].ptr = (uint8_t*)cur->end;
mem_mappings.mem_chunks[cur_slot].size = next->start - cur->end;
/* did we inlcude the IO gap? */
check_for_io_gap(&cur_slot);
cur = virt_to_phys_with_offset(cur->next);
next = virt_to_phys_with_offset(cur->next);
}
/* allocation behind last free region */
if (guest_size != cur->end) {
mem_mappings.mem_chunks[cur_slot].ptr = (uint8_t*)cur->end;
mem_mappings.mem_chunks[cur_slot].size = guest_size - cur->end;
check_for_io_gap(&cur_slot);
}
return;
}
void close_chk_file(void)
{
fclose(chk_file);
}
void write_chk_file(void *addr, size_t bytes)
{
if (fwrite(addr, bytes, 1, chk_file) != 1) {
err(1, "fwrite failed");
}
}
void write_mem_page_to_chk_file(void *entry, size_t entry_size, void *page, size_t page_size)
{
write_chk_file(entry, entry_size);
write_chk_file(page, page_size);
}
void determine_dirty_pages(void (*save_page_handler)(void*, size_t, void*, size_t))
{
#ifdef USE_DIRTY_LOG
scan_dirty_log(save_page_handler);
#else
scan_page_tables(save_page_handler);
#endif
}
void timer_handler(int signum)
{
struct stat st = {0};
char fname[MAX_FNAME];
struct timeval begin, end;
if (verbose)
gettimeofday(&begin, NULL);
if (stat("checkpoint", &st) == -1)
mkdir("checkpoint", 0700);
for(size_t i = 0; i < ncores; i++)
if (vcpu_threads[i] != pthread_self())
pthread_kill(vcpu_threads[i], SIGTHRCHKP);
pthread_barrier_wait(&barrier);
write_cpu_state();
snprintf(fname, MAX_FNAME, "checkpoint/chk%u_mem.dat", no_checkpoint);
open_chk_file(fname);
/*struct kvm_irqchip irqchip = {};
if (cap_irqchip)
kvm_ioctl(vmfd, KVM_GET_IRQCHIP, &irqchip);
else
memset(&irqchip, 0x00, sizeof(irqchip));
if (fwrite(&irqchip, sizeof(irqchip), 1, f) != 1)
err(1, "fwrite failed");*/
struct kvm_clock_data clock = {};
kvm_ioctl(vmfd, KVM_GET_CLOCK, &clock);
write_chk_file(&clock, sizeof(clock));
#if 0
if (fwrite(guest_mem, guest_size, 1, f) != 1)
err(1, "fwrite failed");
#else
determine_dirty_pages(write_mem_page_to_chk_file);
#endif
close_chk_file();
pthread_barrier_wait(&barrier);
// update configuration file
FILE *f = fopen("checkpoint/chk_config.txt", "w");
if (f == NULL) {
err(1, "fopen: unable to open file");
}
fprintf(f, "number of cores: %u\n", ncores);
fprintf(f, "memory size: 0x%zx\n", guest_size);
fprintf(f, "checkpoint number: %u\n", no_checkpoint);
fprintf(f, "entry point: 0x%zx\n", elf_entry);
if (full_checkpoint)
fprintf(f, "full checkpoint: 1");
else
fprintf(f, "full checkpoint: 0");
fclose(f);
if (verbose) {
gettimeofday(&end, NULL);
size_t msec = (end.tv_sec - begin.tv_sec) * 1000;
msec += (end.tv_usec - begin.tv_usec) / 1000;
fprintf(stderr, "Create checkpoint %u in %zd ms\n", no_checkpoint, msec);
}
no_checkpoint++;
}
void *migration_handler(void *arg)
{
sigset_t *signal_mask = (sigset_t *)arg;
int res = 0;
size_t i = 0;
int sig_caught;
/* wait for a migration request and connect to the migration server*/
while (1) {
sigwait(signal_mask, &sig_caught);
if (connect_to_server() < 0) {
fprintf(stderr, "[ERROR] Could not connect to the "
"destination. Abort!\n");
} else {
break;
}
}
/* send metadata */
migration_metadata_t metadata = {
ncores,
guest_size,
0, /* no_checkpoint */
elf_entry,
full_checkpoint};
/* the guest size is recalculated at the destination */
if ((guest_size-KVM_32BIT_GAP_SIZE) >= KVM_32BIT_GAP_START) {
metadata.guest_size -= KVM_32BIT_GAP_SIZE;
}
res = send_data(&metadata, sizeof(migration_metadata_t));
fprintf(stderr, "Metadata sent! (%d bytes)\n", res);
/* determine guest-physical and guest-allocated memory regions */
determine_guest_allocations();
determine_guest_physical_memory_regions(&guest_physical_memory);
/* send to the destination */
send_mem_regions(guest_physical_memory, mem_mappings);
/* we can now determine host-local VAs */
convert_to_host_virt(&guest_physical_memory);
convert_to_host_virt(&mem_mappings);
/* pre-copy phase */
precopy_phase(guest_physical_memory, mem_mappings);
/* synchronize VCPU threads */
assert(vcpu_thread_states == NULL);
vcpu_thread_states = (vcpu_state_t*)calloc(ncores, sizeof(vcpu_state_t));
for(i = 0; i < ncores; i++)
pthread_kill(vcpu_threads[i], SIGTHRMIG);
pthread_barrier_wait(&migration_barrier);
/* send the final dump */
stop_and_copy_phase();
fprintf(stderr, "Memory sent! (Guest size: %zu bytes)\n", guest_size);
/* free mem_mappings and guest_physical_memory info */
free(mem_mappings.mem_chunks);
mem_mappings.count = 0;
free(guest_physical_memory.mem_chunks);
guest_physical_memory.count = 0;
/* send CPU state and cleanup */
res = send_data(vcpu_thread_states, sizeof(vcpu_state_t)*ncores);
fprintf(stderr, "CPU state sent! (%d bytes)\n", res);
free(vcpu_thread_states);
vcpu_thread_states = NULL;
/* send clock */
if (cap_adjust_clock_stable) {
struct kvm_clock_data clock = {};
kvm_ioctl(vmfd, KVM_GET_CLOCK, &clock);
res = send_data(&clock, sizeof(clock));
fprintf(stderr, "Clock sent! (%d bytes)\n", res);
}
/* close socket */
close_migration_channel();
exit(EXIT_SUCCESS);
}
void load_migration_data(uint8_t* mem)
{
size_t paddr = elf_entry;
int res = 0;
if (!klog)
klog = mem+paddr+0x5000-GUEST_OFFSET;
if (!kheader)