-
Notifications
You must be signed in to change notification settings - Fork 583
/
AddressSpace.cc
2562 lines (2319 loc) · 87 KB
/
AddressSpace.cc
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
/* -*- Mode: C++; tab-width: 8; c-basic-offset: 2; indent-tabs-mode: nil; -*- */
#include "AddressSpace.h"
#include <limits.h>
#include <linux/kdev_t.h>
#include <linux/prctl.h>
#include <sys/stat.h>
#include <unistd.h>
#include <limits>
#include "rr/rr.h"
#include "preload/preload_interface.h"
#include "AutoRemoteSyscalls.h"
#include "MonitoredSharedMemory.h"
#include "RecordSession.h"
#include "RecordTask.h"
#include "Session.h"
#include "Task.h"
#include "core.h"
#include "log.h"
using namespace std;
namespace rr {
static const uint8_t x86_breakpoint_insn[] = { 0xcc }; // int $3
static const uint8_t arm64_breakpoint_insn[4] = {0x0, 0x0, 0x20, 0xd4}; // brk #0
static const uint8_t *breakpoint_insn(SupportedArch arch) {
switch (arch) {
case x86:
case x86_64:
return x86_breakpoint_insn;
case aarch64:
return arm64_breakpoint_insn;
default:
DEBUG_ASSERT(0 && "Must define breakpoint insn for this architecture");
return nullptr;
}
}
/**
* Advance *str to skip leading blank characters.
*/
static const char* trim_leading_blanks(const char* str) {
const char* trimmed = str;
while (isblank(*trimmed)) {
++trimmed;
}
return trimmed;
}
/**
* Returns true if a task in t's thread-group other than t is doing an exec.
*/
static bool thread_group_in_exec(Task* t) {
if (!t->session().is_recording()) {
return false;
}
for (Task* tt : t->thread_group()->task_set()) {
if (tt == t || t->already_exited()) {
continue;
}
RecordTask* rt = static_cast<RecordTask*>(tt);
Event& ev = rt->ev();
if (ev.is_syscall_event() && ev.Syscall().is_exec()) {
return true;
}
}
return false;
}
KernelMapIterator::KernelMapIterator(Task* t, bool* ok)
: tid(t->tid) {
// See https://lkml.org/lkml/2016/9/21/423
ASSERT(t, !thread_group_in_exec(t)) << "Task-group in execve, so reading "
"/proc/.../maps may trigger kernel "
"deadlock!";
init(ok);
}
KernelMapIterator::~KernelMapIterator() {
if (maps_file) {
fclose(maps_file);
}
}
void KernelMapIterator::init(bool* ok) {
char maps_path[PATH_MAX];
sprintf(maps_path, "/proc/%d/maps", tid);
if (ok) {
*ok = true;
}
if (!(maps_file = fopen(maps_path, "r"))) {
if (ok) {
*ok = false;
} else {
FATAL() << "Failed to open " << maps_path;
}
}
++*this;
}
void KernelMapIterator::operator++() {
char line[PATH_MAX * 2];
if (!fgets(line, sizeof(line), maps_file)) {
fclose(maps_file);
maps_file = nullptr;
return;
}
uint64_t start, end, offset, inode;
int dev_major, dev_minor;
char flags[32];
int chars_scanned;
int nparsed = sscanf(line, "%" SCNx64 "-%" SCNx64 " %31s %" SCNx64
" %x:%x %" SCNu64 " %n",
&start, &end, flags, &offset, &dev_major, &dev_minor,
&inode, &chars_scanned);
DEBUG_ASSERT(8 /*number of info fields*/ == nparsed ||
7 /*num fields if name is blank*/ == nparsed);
// trim trailing newline, if any
int last_char = strlen(line) - 1;
if (line[last_char] == '\n') {
line[last_char] = 0;
}
raw_line = line;
const char* name = trim_leading_blanks(line + chars_scanned);
#if defined(__i386__)
if (start > numeric_limits<uint32_t>::max() ||
end > numeric_limits<uint32_t>::max() ||
strcmp(name, "[vsyscall]") == 0) {
// We manually read the exe link here because
// this helper is used to set
// |t->vm()->exe_image()|, so we can't rely on
// that being correct yet.
char proc_exe[PATH_MAX];
char exe[PATH_MAX];
snprintf(proc_exe, sizeof(proc_exe), "/proc/%d/exe", tid);
ssize_t size = readlink(proc_exe, exe, sizeof(exe));
if (size < 0) {
FATAL() << "readlink failed";
}
FATAL() << "Sorry, tracee " << tid << " has x86-64 image " << exe
<< " and that's not supported with a 32-bit rr.";
}
#endif
int prot = (strchr(flags, 'r') ? PROT_READ : 0) |
(strchr(flags, 'w') ? PROT_WRITE : 0) |
(strchr(flags, 'x') ? PROT_EXEC : 0);
int f = (strchr(flags, 'p') ? MAP_PRIVATE : 0) |
(strchr(flags, 's') ? MAP_SHARED : 0);
string tmp_name;
if (strchr(name, '\\')) {
// Unescape any '\012' sequences
while (*name) {
if (strncmp(name, "\\012", 4) == 0) {
tmp_name.push_back('\n');
name += 4;
} else {
tmp_name.push_back(*name);
++name;
}
}
name = tmp_name.c_str();
}
km = KernelMapping(start, end, name, MKDEV(dev_major, dev_minor), inode, prot,
f, offset);
}
static KernelMapping read_kernel_mapping(pid_t tid, remote_ptr<void> addr) {
MemoryRange range(addr, 1);
bool ok;
KernelMapIterator it(tid, &ok);
if (!ok) {
return KernelMapping();
}
for (; !it.at_end(); ++it) {
const KernelMapping& km = it.current();
if (km.contains(range)) {
return km;
}
}
return KernelMapping();
}
KernelMapping AddressSpace::read_kernel_mapping(Task* t,
remote_ptr<void> addr) {
return rr::read_kernel_mapping(t->tid, addr);
}
KernelMapping AddressSpace::read_local_kernel_mapping(uint8_t* addr) {
return rr::read_kernel_mapping(getpid(), remote_ptr<void>((uintptr_t)addr));
}
/**
* Cat the /proc/[t->tid]/maps file to stdout, line by line.
*/
void AddressSpace::print_process_maps(Task* t) {
for (KernelMapIterator it(t); !it.at_end(); ++it) {
string line;
it.current(&line);
cerr << line << '\n';
}
}
AddressSpace::Mapping::Mapping(const KernelMapping& map,
const KernelMapping& recorded_map,
EmuFile::shr_ptr emu_file,
std::unique_ptr<struct stat> mapped_file_stat,
void* local_addr,
shared_ptr<MonitoredSharedMemory>&& monitored)
: map(map),
recorded_map(recorded_map),
emu_file(emu_file),
mapped_file_stat(std::move(mapped_file_stat)),
local_addr(static_cast<uint8_t*>(local_addr)),
monitored_shared_memory(std::move(monitored)),
flags(FLAG_NONE) {}
static unique_ptr<struct stat> clone_stat(
const unique_ptr<struct stat>& other) {
return other ? unique_ptr<struct stat>(new struct stat(*other)) : nullptr;
}
AddressSpace::Mapping::Mapping(const Mapping& other)
: map(other.map),
recorded_map(other.recorded_map),
emu_file(other.emu_file),
mapped_file_stat(clone_stat(other.mapped_file_stat)),
local_addr(other.local_addr),
monitored_shared_memory(other.monitored_shared_memory),
flags(other.flags) {}
AddressSpace::Mapping::~Mapping() {}
AddressSpace::Mapping AddressSpace::Mapping::subrange(MemoryRange range,
std::function<KernelMapping(const KernelMapping&)> f) const {
Mapping mapping(
f(map.subrange(range.start(), range.end())),
f(recorded_map.subrange(range.start(), range.end())),
emu_file, clone_stat(mapped_file_stat),
local_addr ? local_addr + (range.start() - map.start()) : 0,
monitored_shared_memory
? monitored_shared_memory->subrange(range.start() - map.start(),
range.size())
: nullptr);
mapping.flags = flags;
return mapping;
}
AddressSpace::~AddressSpace() {
for (auto& m : mem) {
if (m.second.local_addr) {
int ret = munmap(m.second.local_addr, m.second.map.size());
if (ret < 0) {
FATAL() << "Can't munmap";
}
}
}
session_->on_destroy(this);
}
static uint32_t find_offset_of_syscall_instruction_in(SupportedArch arch,
uint8_t* vdso_data,
size_t vdso_len) {
auto instruction = syscall_instruction(arch);
for (uint32_t i = 1; i < vdso_len - instruction.size(); ++i) {
if (memcmp(vdso_data + i, instruction.data(), instruction.size()) == 0) {
return i;
}
}
return 0;
}
uint32_t AddressSpace::offset_to_syscall_in_vdso[SupportedArch_MAX + 1];
remote_code_ptr AddressSpace::find_syscall_instruction_in_vdso(Task* t) {
SupportedArch arch = t->arch();
if (!offset_to_syscall_in_vdso[arch]) {
if (!vdso_start_addr) {
return remote_code_ptr();
}
auto vdso_data = t->read_mem(vdso().start().cast<uint8_t>(), vdso().size());
offset_to_syscall_in_vdso[arch] = find_offset_of_syscall_instruction_in(
arch, vdso_data.data(), vdso_data.size());
ASSERT(t, offset_to_syscall_in_vdso[arch])
<< "No syscall instruction found in VDSO";
}
return remote_code_ptr(
(vdso().start().cast<uint8_t>() + offset_to_syscall_in_vdso[arch])
.as_int());
}
static string rr_page_file_name(SupportedArch arch, const char** fname_out) {
string file_name;
const char *fname = nullptr;
switch (arch) {
case x86_64:
case aarch64:
fname = RRPAGE_LIB_FILENAME;
break;
case x86:
#if defined(__x86_64__)
fname = RRPAGE_LIB_FILENAME_32;
#else
fname = RRPAGE_LIB_FILENAME;
#endif
break;
}
*fname_out = fname;
string path = find_helper_library(fname);
if (path.empty()) {
return path;
}
path += fname;
return path;
}
void AddressSpace::map_rr_page(AutoRemoteSyscalls& remote) {
int prot = PROT_EXEC | PROT_READ;
int flags = MAP_PRIVATE | MAP_FIXED;
Task* t = remote.task();
SupportedArch arch = t->arch();
const char* fname;
string path = rr_page_file_name(arch, &fname);
if (path.empty()) {
FATAL() << "Failed to locate " << fname << "; needed by "
<< t->exe_path() << " (" << arch_name(arch) << ")";
}
size_t offset_pages = t->session().is_recording() ?
RRPAGE_RECORD_PAGE_OFFSET : RRPAGE_REPLAY_PAGE_OFFSET;
size_t offset_bytes = offset_pages * PRELOAD_LIBRARY_PAGE_SIZE;
{
ScopedFd page(path.c_str(), O_RDONLY);
ASSERT(t, page.is_open()) << "Failed to open rrpage library " << path;
int child_fd = remote.infallible_send_fd_if_alive(page);
if (child_fd >= 0) {
if (t->session().is_recording()) {
remote.infallible_mmap_syscall_if_alive(rr_page_start() - offset_bytes, offset_bytes, prot, flags,
child_fd, 0);
}
remote.infallible_mmap_syscall_if_alive(rr_page_start(), PRELOAD_LIBRARY_PAGE_SIZE, prot, flags,
child_fd, offset_bytes);
struct stat fstat = t->stat_fd(child_fd);
string file_name = t->file_name_of_fd(child_fd);
remote.infallible_close_syscall_if_alive(child_fd);
map(t, rr_page_start(), PRELOAD_LIBRARY_PAGE_SIZE, prot, flags,
offset_bytes, file_name,
fstat.st_dev, fstat.st_ino);
mapping_flags_of(rr_page_start()) = Mapping::IS_RR_PAGE;
if (t->session().is_recording()) {
map(t, rr_page_start() - offset_bytes, offset_bytes, prot, flags,
0, file_name,
fstat.st_dev, fstat.st_ino);
mapping_flags_of(rr_page_start() - offset_bytes) = Mapping::IS_RR_VDSO_PAGE;
}
}
}
if (t->session().is_recording()) {
// brk() will not have been called yet so the brk area is empty.
brk_start = brk_end =
remote.infallible_syscall(syscall_number_for_brk(arch), 0);
ASSERT(t, !brk_end.is_null());
}
}
vector<uint8_t> AddressSpace::read_rr_page_for_recording(SupportedArch arch) {
const char* fname;
string path = rr_page_file_name(arch, &fname);
if (path.empty()) {
FATAL() << "Failed to locate " << fname;
}
ScopedFd page(path.c_str(), O_RDONLY);
vector<uint8_t> result;
result.resize(PRELOAD_LIBRARY_PAGE_SIZE);
ssize_t ret = read_to_end(page,
RRPAGE_RECORD_PAGE_OFFSET * PRELOAD_LIBRARY_PAGE_SIZE, result.data(),
result.size());
if (ret != static_cast<ssize_t>(result.size())) {
FATAL() << "Failed to read full page from " << path;
}
return result;
}
void AddressSpace::unmap_all_but_rr_mappings(AutoRemoteSyscalls& remote,
UnmapOptions options) {
vector<MemoryRange> unmaps;
for (const auto& m : maps()) {
// Do not attempt to unmap [vsyscall] --- it doesn't work.
if (m.map.start() != AddressSpace::rr_page_start() &&
m.map.start() != AddressSpace::preload_thread_locals_start() &&
!m.map.is_vsyscall() &&
(!options.exclude_vdso_vvar || (!m.map.is_vdso() && m.map.is_vvar()))) {
unmaps.push_back(m.map);
}
}
for (auto& m : unmaps) {
remote.infallible_syscall(syscall_number_for_munmap(remote.task()->arch()),
m.start(), m.size());
unmap(remote.task(), m.start(), m.size());
}
}
/**
* Must match generate_rr_page.py
*/
static const AddressSpace::SyscallType entry_points[] = {
{ AddressSpace::TRACED, AddressSpace::UNPRIVILEGED,
AddressSpace::RECORDING_AND_REPLAY },
{ AddressSpace::TRACED, AddressSpace::PRIVILEGED,
AddressSpace::RECORDING_AND_REPLAY },
{ AddressSpace::UNTRACED, AddressSpace::UNPRIVILEGED,
AddressSpace::RECORDING_AND_REPLAY },
{ AddressSpace::UNTRACED, AddressSpace::UNPRIVILEGED,
AddressSpace::REPLAY_ONLY },
{ AddressSpace::UNTRACED, AddressSpace::UNPRIVILEGED,
AddressSpace::RECORDING_ONLY },
{ AddressSpace::UNTRACED, AddressSpace::PRIVILEGED,
AddressSpace::RECORDING_AND_REPLAY },
{ AddressSpace::UNTRACED, AddressSpace::PRIVILEGED,
AddressSpace::REPLAY_ONLY },
{ AddressSpace::UNTRACED, AddressSpace::PRIVILEGED,
AddressSpace::RECORDING_ONLY },
{ AddressSpace::UNTRACED, AddressSpace::UNPRIVILEGED,
AddressSpace::REPLAY_ASSIST },
};
static int rr_page_syscall_stub_size(SupportedArch arch) {
int val = 0;
switch (arch) {
case x86:
case x86_64:
val = 3;
break;
case aarch64:
val = 8;
break;
default:
FATAL() << "Syscall stub size not defined for this architecture";
}
if (arch == NativeArch::arch()) {
DEBUG_ASSERT(val == RR_PAGE_SYSCALL_STUB_SIZE);
}
return val;
}
static int rr_page_syscall_instruction_end(SupportedArch arch) {
int val = 0;
switch (arch) {
case x86:
case x86_64:
val = 2;
break;
case aarch64:
val = 4;
break;
default:
FATAL() << "Syscall stub size not defined for this architecture";
}
if (arch == NativeArch::arch()) {
DEBUG_ASSERT(val == RR_PAGE_SYSCALL_INSTRUCTION_END);
}
return val;
}
static remote_code_ptr entry_ip_from_index(SupportedArch arch, size_t i) {
return remote_code_ptr(RR_PAGE_ADDR + rr_page_syscall_stub_size(arch) * i);
}
static remote_code_ptr exit_ip_from_index(SupportedArch arch, size_t i) {
return remote_code_ptr(RR_PAGE_ADDR + rr_page_syscall_stub_size(arch) * i +
rr_page_syscall_instruction_end(arch));
}
remote_code_ptr AddressSpace::rr_page_syscall_exit_point(Traced traced,
Privileged privileged,
Enabled enabled,
SupportedArch arch) {
for (auto& e : entry_points) {
if (e.traced == traced && e.privileged == privileged &&
e.enabled == enabled) {
return exit_ip_from_index(arch, &e - entry_points);
}
}
return nullptr;
}
remote_code_ptr AddressSpace::rr_page_syscall_entry_point(Traced traced,
Privileged privileged,
Enabled enabled,
SupportedArch arch) {
for (auto& e : entry_points) {
if (e.traced == traced && e.privileged == privileged &&
e.enabled == enabled) {
return entry_ip_from_index(arch, &e - entry_points);
}
}
return nullptr;
}
const AddressSpace::SyscallType* AddressSpace::rr_page_syscall_from_exit_point(
SupportedArch arch, remote_code_ptr ip) {
for (size_t i = 0; i < array_length(entry_points); ++i) {
if (exit_ip_from_index(arch, i) == ip) {
return &entry_points[i];
}
}
return nullptr;
}
const AddressSpace::SyscallType* AddressSpace::rr_page_syscall_from_entry_point(
SupportedArch arch, remote_code_ptr ip) {
for (size_t i = 0; i < array_length(entry_points); ++i) {
if (entry_ip_from_index(arch, i) == ip) {
return &entry_points[i];
}
}
return nullptr;
}
vector<AddressSpace::SyscallType> AddressSpace::rr_page_syscalls() {
vector<SyscallType> result;
for (auto& e : entry_points) {
result.push_back(e);
}
return result;
}
void AddressSpace::save_auxv(Task* t) {
saved_auxv_ = read_auxv(t);
save_interpreter_base(t, saved_auxv());
}
void AddressSpace::save_interpreter_base(Task* t, std::vector<uint8_t> auxv) {
saved_interpreter_base_ = read_interpreter_base(auxv);
save_ld_path(t, saved_interpreter_base());
}
void AddressSpace::save_ld_path(Task* t, remote_ptr<void> interpreter_base) {
saved_ld_path_ = read_ld_path(t, interpreter_base);
}
void AddressSpace::read_mm_map(Task* t, NativeArch::prctl_mm_map* map) {
char buf[PATH_MAX+1024];
{
string proc_stat = t->proc_stat_path();
ScopedFd fd(proc_stat.c_str(), O_RDONLY);
memset(buf, 0, sizeof(buf));
int err = read_to_end(fd, 0, buf, sizeof(buf)-1);
if (err < 0) {
FATAL() << "Failed to read /proc/<pid>/stat";
}
}
// The last close-paren indicates the end of the comm and the
// start of the fixed-width area
char* fixed = strrchr(buf, ')');
// We don't change /proc/pid/exe, since we're unlikely to have CAP_SYS_ADMIN
map->exe_fd = -1;
// auxv is restored separately
map->auxv.val = 0;
map->auxv_size = 0;
// All of these fields of /proc/pid/stat, we don't use (currently)
char state;
pid_t ppid;
pid_t pgrp;
int session;
int tty_nr;
int tpgid;
unsigned int flags;
unsigned long minflt, cminflt, majflt, cmajflt, utime, stime;
long cutime, cstime, priority, nice, num_threads, itrealvalue;
unsigned long long starttime;
unsigned long vsize;
long rss;
unsigned long rsslim, kstkesp, kstskip, signal;
unsigned long blocked, sigignore, sigcatch, wchan, nswap, cnswap;
int exit_signal, processor;
unsigned int rt_priority, policy;
unsigned long long delayacct_blkio_ticks;
unsigned long guest_time;
long cguest_time;
int exit_code;
// See the proc(5) man page for the correct scan codes for these
size_t n = sscanf(fixed + 1,
// state ppid pgrp session tty_nr tpgid
" %c %d %d %d %d %d"
// flags minflt cminflt majflt cmajflt utime stime cutime cstime
" %u %lu %lu %lu %lu %lu %lu %ld %ld"
// priority nice num_threads itrealvalue starttime vsize rss
" %ld %ld %ld %ld %llu %lu %ld"
// rsslim startcode endcode startstack kstkesp kstskip signal
" %lu %lu %lu %lu %lu %lu %lu"
// blocked sigignore sigcatch wchan nswap cnswap exit_signal
" %lu %lu %lu %lu %lu %lu %d"
// processor rt_priority policy delayacct_blkio_ticks guest_time cguest_time
" %d %u %u %llu %lu %ld "
// start_data end_data start_brk arg_start arg_end env_start env_end exit_code
" %lu %lu %lu %lu %lu %lu %lu %d",
&state, &ppid, &pgrp, &session, &tty_nr, &tpgid,
&flags, &minflt, &cminflt, &majflt, &cmajflt, &utime, &stime, &cutime, &cstime,
&priority, &nice, &num_threads, &itrealvalue, &starttime, &vsize, &rss,
&rsslim, (unsigned long *)&map->start_code, (unsigned long *)&map->end_code,
(unsigned long *)&map->start_stack, &kstkesp, &kstskip, &signal,
&blocked, &sigignore, &sigcatch, &wchan, &nswap, &cnswap, &exit_signal,
&processor, &rt_priority, &policy, &delayacct_blkio_ticks, &guest_time,
&cguest_time, (unsigned long *)&map->start_data, (unsigned long *)&map->end_data,
(unsigned long *)&map->start_brk, (unsigned long *)&map->arg_start,
(unsigned long *)&map->arg_end, (unsigned long *)&map->env_start,
(unsigned long *)&map->env_end, &exit_code);
ASSERT(t, n == 50);
// Fill in brk end
ASSERT(t, map->start_brk == this->brk_start.as_int());
map->brk = this->brk_end.as_int();
}
void AddressSpace::post_exec_syscall(Task* t) {
// First locate a syscall instruction we can use for remote syscalls.
traced_syscall_ip_ = find_syscall_instruction_in_vdso(t);
privileged_traced_syscall_ip_ = nullptr;
do_breakpoint_fault_addr_ = nullptr;
stopping_breakpoint_table_ = nullptr;
stopping_breakpoint_table_entry_size_ = 0;
// Now remote syscalls work, we can open_mem_fd.
t->open_mem_fd();
// Set up AutoRemoteSyscalls again now that the mem-fd is open.
AutoRemoteSyscalls remote(t);
// Now we can set up the "rr page" at its fixed address. This gives
// us traced and untraced syscall instructions at known, fixed addresses.
map_rr_page(remote);
// Set up the preload_thread_locals shared area.
t->session().create_shared_mmap(remote, PRELOAD_THREAD_LOCALS_SIZE,
preload_thread_locals_start(),
"preload_thread_locals");
mapping_flags_of(preload_thread_locals_start()) |=
AddressSpace::Mapping::IS_THREAD_LOCALS;
}
void AddressSpace::brk(Task* t, remote_ptr<void> addr, int prot) {
LOG(debug) << "brk(" << addr << ")";
remote_ptr<void> old_brk = ceil_page_size(brk_end);
remote_ptr<void> new_brk = ceil_page_size(addr);
if (old_brk < new_brk) {
map(t, old_brk, new_brk - old_brk, prot, MAP_ANONYMOUS | MAP_PRIVATE, 0,
"[heap]");
} else {
unmap(t, new_brk, old_brk - new_brk);
}
brk_end = addr;
}
static const char* stringify_flags(int flags) {
switch (flags) {
case AddressSpace::Mapping::FLAG_NONE:
return "";
case AddressSpace::Mapping::IS_SYSCALLBUF:
return " [syscallbuf]";
case AddressSpace::Mapping::IS_THREAD_LOCALS:
return " [thread_locals]";
case AddressSpace::Mapping::IS_PATCH_STUBS:
return " [patch_stubs]";
case AddressSpace::Mapping::IS_RR_PAGE:
return " [rr_page]";
case AddressSpace::Mapping::IS_RR_VDSO_PAGE:
return " [rr_vdso_page]";
default:
return " [unknown_flags]";
}
}
void AddressSpace::dump() const {
fprintf(stderr, " (heap: %p-%p)\n", (void*)brk_start.as_int(),
(void*)brk_end.as_int());
for (auto it = mem.begin(); it != mem.end(); ++it) {
const KernelMapping& m = it->second.map;
fprintf(stderr, "%s%s\n", m.str().c_str(),
stringify_flags(it->second.flags));
}
}
SupportedArch AddressSpace::arch() const {
return (*task_set().begin())->arch();
}
BreakpointType AddressSpace::get_breakpoint_type_for_retired_insn(
remote_code_ptr ip) {
remote_code_ptr addr = ip.undo_executed_bkpt(arch());
return get_breakpoint_type_at_addr(addr);
}
BreakpointType AddressSpace::get_breakpoint_type_at_addr(remote_code_ptr addr) {
auto it = breakpoints.find(addr);
return it == breakpoints.end() ? BKPT_NONE : it->second.type();
}
bool AddressSpace::is_exec_watchpoint(remote_code_ptr addr) {
for (auto& kv : watchpoints) {
if (kv.first.contains(addr.to_data_ptr<void>()) &&
(kv.second.watched_bits() & EXEC_BIT)) {
return true;
}
}
return false;
}
bool AddressSpace::is_breakpoint_in_private_read_only_memory(
remote_code_ptr addr) {
for (const auto& m : maps_containing_or_after(addr.to_data_ptr<void>())) {
if (m.map.start() >=
addr.increment_by_bkpt_insn_length(arch()).to_data_ptr<void>()) {
break;
}
if ((m.map.prot() & PROT_WRITE) || (m.map.flags() & MAP_SHARED)) {
return false;
}
}
return true;
}
void AddressSpace::replace_breakpoints_with_original_values(
uint8_t* dest, size_t length, remote_ptr<uint8_t> addr) {
for (auto& it : breakpoints) {
replace_in_buffer(MemoryRange(it.first.to_data_ptr<void>(), bkpt_instruction_length(arch())),
it.second.original_data(),
MemoryRange(addr, length), dest);
}
}
bool AddressSpace::is_breakpoint_instruction(Task* t, remote_code_ptr ip) {
bool ok = true;
uint8_t data[MAX_BKPT_INSTRUCTION_LENGTH];
t->read_bytes_helper(ip.to_data_ptr<uint8_t>(),
bkpt_instruction_length(t->arch()), data, &ok);
return memcmp(data, breakpoint_insn(t->arch()),
bkpt_instruction_length(t->arch())) == 0 && ok;
}
static void remove_range(set<MemoryRange>& ranges, const MemoryRange& range) {
if (ranges.empty()) {
return;
}
auto start = ranges.lower_bound(range);
// An earlier range might extend into range, so check for that.
if (start != ranges.begin()) {
--start;
if (start->end() <= range.start()) {
++start;
}
}
auto end = start;
auto prev_end = start;
while (end != ranges.end() && end->start() < range.end()) {
prev_end = end;
++end;
}
if (start == end) {
return;
}
MemoryRange start_range = *start;
MemoryRange end_range = *prev_end;
ranges.erase(start, end);
if (start_range.start() < range.start()) {
ranges.insert(MemoryRange(start_range.start(), range.start()));
}
if (range.end() < end_range.end()) {
ranges.insert(MemoryRange(range.end(), end_range.end()));
}
}
static void add_range(set<MemoryRange>& ranges, const MemoryRange& range) {
// Remove overlapping ranges
remove_range(ranges, range);
ranges.insert(range);
// We could coalesce adjacent ranges, but there's probably no need.
}
KernelMapping AddressSpace::map(Task* t, remote_ptr<void> addr,
size_t num_bytes, int prot, int flags,
off64_t offset_bytes, const string& fsname,
dev_t device, ino_t inode,
unique_ptr<struct stat> mapped_file_stat,
const KernelMapping* recorded_map,
EmuFile::shr_ptr emu_file, void* local_addr,
shared_ptr<MonitoredSharedMemory> monitored) {
LOG(debug) << "mmap(" << addr << ", " << num_bytes << ", " << HEX(prot)
<< ", " << HEX(flags) << ", " << HEX(offset_bytes) << ")";
num_bytes = ceil_page_size(num_bytes);
KernelMapping m(addr, addr + num_bytes, fsname, device, inode, prot, flags,
offset_bytes);
if (!num_bytes) {
return m;
}
remove_range(dont_fork, MemoryRange(addr, num_bytes));
remove_range(wipe_on_fork, MemoryRange(addr, num_bytes));
// The mmap() man page doesn't specifically describe
// what should happen if an existing map is
// "overwritten" by a new map (of the same resource).
// In testing, the behavior seems to be as if the
// overlapping region is unmapped and then remapped
// per the arguments to the second call.
unmap_internal(t, addr, num_bytes);
const KernelMapping& actual_recorded_map = recorded_map ? *recorded_map : m;
map_and_coalesce(t, m, actual_recorded_map, emu_file,
std::move(mapped_file_stat),
std::move(local_addr),
std::move(monitored));
// During an emulated exec, we will explicitly map in a (copy of) the VDSO
// at the recorded address.
if (actual_recorded_map.is_vdso()) {
vdso_start_addr = addr;
}
if (MemoryRange(addr, num_bytes).contains(RR_PAGE_ADDR)) {
update_syscall_ips(t);
}
return m;
}
template <typename Arch> void AddressSpace::at_preload_init_arch(Task* t) {
auto params = t->read_mem(
remote_ptr<rrcall_init_preload_params<Arch>>(t->regs().orig_arg1()));
if (t->session().is_recording()) {
ASSERT(t,
t->session().as_record()->use_syscall_buffer() ==
params.syscallbuf_enabled)
<< "Tracee thinks syscallbuf is "
<< (params.syscallbuf_enabled ? "en" : "dis")
<< "abled, but tracer thinks "
<< (t->session().as_record()->use_syscall_buffer() ? "en" : "dis")
<< "abled";
} else {
if (params.breakpoint_table_entry_size == -1) {
do_breakpoint_fault_addr_ = params.breakpoint_instr_addr.rptr().as_int();
} else {
stopping_breakpoint_table_ = params.breakpoint_table.rptr().as_int();
stopping_breakpoint_table_entry_size_ =
params.breakpoint_table_entry_size;
}
}
if (!params.syscallbuf_enabled) {
return;
}
syscallbuf_enabled_ = true;
if (t->session().is_recording()) {
monkeypatch_state->patch_at_preload_init(static_cast<RecordTask*>(t));
}
}
void AddressSpace::at_preload_init(Task* t) {
RR_ARCH_FUNCTION(at_preload_init_arch, t->arch(), t);
}
const AddressSpace::Mapping& AddressSpace::mapping_of(
remote_ptr<void> addr) const {
MemoryRange range(floor_page_size(addr), 1);
auto it = mem.find(range);
DEBUG_ASSERT(it != mem.end());
DEBUG_ASSERT(it->second.map.contains(range));
return it->second;
}
uint32_t& AddressSpace::mapping_flags_of(remote_ptr<void> addr) {
return const_cast<AddressSpace::Mapping&>(
static_cast<const AddressSpace*>(this)->mapping_of(addr))
.flags;
}
uint8_t* AddressSpace::local_mapping(remote_ptr<void> addr, size_t size) {
MemoryRange range(floor_page_size(addr), 1);
auto it = mem.find(range);
if (it == mem.end()) {
return nullptr;
}
DEBUG_ASSERT(it->second.map.contains(range));
const Mapping& map = it->second;
// Fall back to the slow path if we can't get the entire region
if (size > static_cast<size_t>(map.map.end() - addr)) {
return nullptr;
}
if (map.local_addr != nullptr) {
size_t offset = addr - map.map.start();
return static_cast<uint8_t*>(map.local_addr) + offset;
}
return nullptr;
}
void* AddressSpace::detach_local_mapping(remote_ptr<void> addr) {
auto m = const_cast<AddressSpace::Mapping&>(mapping_of(addr));
void* p = m.local_addr;
m.local_addr = nullptr;
return p;
}
bool AddressSpace::has_mapping(remote_ptr<void> addr) const {
if (addr + page_size() < addr) {
// Assume the last byte in the address space is never mapped; avoid overflow
return false;
}
MemoryRange m(floor_page_size(addr), 1);
auto it = mem.find(m);
return it != mem.end() && it->first.contains(m);
}
bool AddressSpace::has_rr_page() const {
MemoryRange m(RR_PAGE_ADDR, 1);
auto it = mem.find(m);
return it != mem.end() && (it->second.flags & Mapping::IS_RR_PAGE);
}
void AddressSpace::protect(Task* t, remote_ptr<void> addr, size_t num_bytes,
int prot) {
LOG(debug) << "mprotect(" << addr << ", " << num_bytes << ", " << HEX(prot)
<< ")";
MemoryRange last_overlap;
auto protector = [this, prot, &last_overlap](Mapping m,
MemoryRange rem) {
LOG(debug) << " protecting (" << rem << ") ...";
remove_from_map(m.map);
// PROT_GROWSDOWN means that if this is a grows-down segment
// (which for us means "stack") then the change should be
// extended to the start of the segment.
// We don't try to handle the analogous PROT_GROWSUP, because we
// don't understand the idea of a grows-up segment.
remote_ptr<void> new_start;
if ((m.map.start() < rem.start()) && (prot & PROT_GROWSDOWN)) {
new_start = m.map.start();
LOG(debug) << " PROT_GROWSDOWN: expanded region down to " << new_start;
} else {
new_start = rem.start();
}
LOG(debug) << " erased (" << m.map << ")";
// If the first segment we protect underflows the
// region, remap the underflow region with previous
// prot.
if (m.map.start() < new_start) {
Mapping underflow = m.subrange(MemoryRange(m.map.start(), new_start),
[](const KernelMapping& km) { return km; });
add_to_map(underflow);
}
// Remap the overlapping region with the new prot.
remote_ptr<void> new_end = min(rem.end(), m.map.end());
int new_prot = prot & (PROT_READ | PROT_WRITE | PROT_EXEC);
Mapping overlap = m.subrange(MemoryRange(new_start, new_end),
[new_prot](const KernelMapping& km) { return km.set_prot(new_prot); });
add_to_map(overlap);
last_overlap = overlap.map;
// If the last segment we protect overflows the
// region, remap the overflow region with previous
// prot.
if (new_end < m.map.end()) {
Mapping overflow = m.subrange(MemoryRange(new_end, m.map.end()),
[](const KernelMapping& km) { return km; });
add_to_map(overflow);
}
};
for_each_in_range(addr, num_bytes, protector, ITERATE_CONTIGUOUS);
if (last_overlap.size()) {
// All mappings that we altered which might need coalescing
// are adjacent to |last_overlap|.
coalesce_around(t, mem.find(last_overlap));
}
}