-
Notifications
You must be signed in to change notification settings - Fork 54
/
Classifier.cpp
1614 lines (1314 loc) · 49.8 KB
/
Classifier.cpp
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; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
#include "Classifier.h"
#include "LookupCacheV4.h"
#include "nsIPrefBranch.h"
#include "nsIPrefService.h"
#include "nsISimpleEnumerator.h"
#include "nsIRandomGenerator.h"
#include "nsIInputStream.h"
#include "nsISeekableStream.h"
#include "nsIFile.h"
#include "nsNetCID.h"
#include "nsPrintfCString.h"
#include "nsThreadUtils.h"
#include "mozilla/Components.h"
#include "mozilla/EndianUtils.h"
#include "mozilla/Telemetry.h"
#include "mozilla/IntegerPrintfMacros.h"
#include "mozilla/Logging.h"
#include "mozilla/SyncRunnable.h"
#include "mozilla/Base64.h"
#include "mozilla/Unused.h"
#include "mozilla/UniquePtr.h"
#include "nsUrlClassifierDBService.h"
#include "nsUrlClassifierUtils.h"
// MOZ_LOG=UrlClassifierDbService:5
extern mozilla::LazyLogModule gUrlClassifierDbServiceLog;
#define LOG(args) \
MOZ_LOG(gUrlClassifierDbServiceLog, mozilla::LogLevel::Debug, args)
#define LOG_ENABLED() \
MOZ_LOG_TEST(gUrlClassifierDbServiceLog, mozilla::LogLevel::Debug)
#define STORE_DIRECTORY NS_LITERAL_CSTRING("safebrowsing")
#define TO_DELETE_DIR_SUFFIX NS_LITERAL_CSTRING("-to_delete")
#define BACKUP_DIR_SUFFIX NS_LITERAL_CSTRING("-backup")
#define UPDATING_DIR_SUFFIX NS_LITERAL_CSTRING("-updating")
#define METADATA_SUFFIX NS_LITERAL_CSTRING(".metadata")
namespace mozilla {
namespace safebrowsing {
void Classifier::SplitTables(const nsACString& str,
nsTArray<nsCString>& tables) {
tables.Clear();
nsACString::const_iterator begin, iter, end;
str.BeginReading(begin);
str.EndReading(end);
while (begin != end) {
iter = begin;
FindCharInReadable(',', iter, end);
nsDependentCSubstring table = Substring(begin, iter);
if (!table.IsEmpty()) {
tables.AppendElement(Substring(begin, iter));
}
begin = iter;
if (begin != end) {
begin++;
}
}
// Remove duplicates
tables.Sort();
const auto newEnd = std::unique(tables.begin(), tables.end());
tables.TruncateLength(std::distance(tables.begin(), newEnd));
}
nsresult Classifier::GetPrivateStoreDirectory(
nsIFile* aRootStoreDirectory, const nsACString& aTableName,
const nsACString& aProvider, nsIFile** aPrivateStoreDirectory) {
NS_ENSURE_ARG_POINTER(aPrivateStoreDirectory);
if (!StringEndsWith(aTableName, NS_LITERAL_CSTRING("-proto"))) {
// Only V4 table names (ends with '-proto') would be stored
// to per-provider sub-directory.
nsCOMPtr<nsIFile>(aRootStoreDirectory).forget(aPrivateStoreDirectory);
return NS_OK;
}
if (aProvider.IsEmpty()) {
// When failing to get provider, just store in the root folder.
nsCOMPtr<nsIFile>(aRootStoreDirectory).forget(aPrivateStoreDirectory);
return NS_OK;
}
nsCOMPtr<nsIFile> providerDirectory;
// Clone first since we are gonna create a new directory.
nsresult rv = aRootStoreDirectory->Clone(getter_AddRefs(providerDirectory));
NS_ENSURE_SUCCESS(rv, rv);
// Append the provider name to the root store directory.
rv = providerDirectory->AppendNative(aProvider);
NS_ENSURE_SUCCESS(rv, rv);
// Ensure existence of the provider directory.
bool dirExists;
rv = providerDirectory->Exists(&dirExists);
NS_ENSURE_SUCCESS(rv, rv);
if (!dirExists) {
LOG(("Creating private directory for %s", nsCString(aTableName).get()));
rv = providerDirectory->Create(nsIFile::DIRECTORY_TYPE, 0755);
NS_ENSURE_SUCCESS(rv, rv);
providerDirectory.forget(aPrivateStoreDirectory);
return rv;
}
// Store directory exists. Check if it's a directory.
bool isDir;
rv = providerDirectory->IsDirectory(&isDir);
NS_ENSURE_SUCCESS(rv, rv);
if (!isDir) {
return NS_ERROR_FILE_DESTINATION_NOT_DIR;
}
providerDirectory.forget(aPrivateStoreDirectory);
return NS_OK;
}
Classifier::Classifier()
: mIsTableRequestResultOutdated(true),
mUpdateInterrupted(true),
mIsClosed(false) {
NS_NewNamedThread(NS_LITERAL_CSTRING("Classifier Update"),
getter_AddRefs(mUpdateThread));
}
Classifier::~Classifier() { Close(); }
nsresult Classifier::SetupPathNames() {
// Get the root directory where to store all the databases.
nsresult rv = mCacheDirectory->Clone(getter_AddRefs(mRootStoreDirectory));
NS_ENSURE_SUCCESS(rv, rv);
rv = mRootStoreDirectory->AppendNative(STORE_DIRECTORY);
NS_ENSURE_SUCCESS(rv, rv);
// Make sure LookupCaches (which are persistent and survive updates)
// are reading/writing in the right place. We will be moving their
// files "underneath" them during backup/restore.
for (uint32_t i = 0; i < mLookupCaches.Length(); i++) {
mLookupCaches[i]->UpdateRootDirHandle(mRootStoreDirectory);
}
// Directory where to move a backup before an update.
rv = mCacheDirectory->Clone(getter_AddRefs(mBackupDirectory));
NS_ENSURE_SUCCESS(rv, rv);
rv = mBackupDirectory->AppendNative(STORE_DIRECTORY + BACKUP_DIR_SUFFIX);
NS_ENSURE_SUCCESS(rv, rv);
// Directory where to be working on the update.
rv = mCacheDirectory->Clone(getter_AddRefs(mUpdatingDirectory));
NS_ENSURE_SUCCESS(rv, rv);
rv = mUpdatingDirectory->AppendNative(STORE_DIRECTORY + UPDATING_DIR_SUFFIX);
NS_ENSURE_SUCCESS(rv, rv);
// Directory where to move the backup so we can atomically
// delete (really move) it.
rv = mCacheDirectory->Clone(getter_AddRefs(mToDeleteDirectory));
NS_ENSURE_SUCCESS(rv, rv);
rv = mToDeleteDirectory->AppendNative(STORE_DIRECTORY + TO_DELETE_DIR_SUFFIX);
NS_ENSURE_SUCCESS(rv, rv);
return NS_OK;
}
nsresult Classifier::CreateStoreDirectory() {
if (ShouldAbort()) {
return NS_OK; // nothing to do, the classifier is done
}
// Ensure the safebrowsing directory exists.
bool storeExists;
nsresult rv = mRootStoreDirectory->Exists(&storeExists);
NS_ENSURE_SUCCESS(rv, rv);
if (!storeExists) {
rv = mRootStoreDirectory->Create(nsIFile::DIRECTORY_TYPE, 0755);
NS_ENSURE_SUCCESS(rv, rv);
} else {
bool storeIsDir;
rv = mRootStoreDirectory->IsDirectory(&storeIsDir);
NS_ENSURE_SUCCESS(rv, rv);
if (!storeIsDir) return NS_ERROR_FILE_DESTINATION_NOT_DIR;
}
return NS_OK;
}
nsresult Classifier::Open(nsIFile& aCacheDirectory) {
// Remember the Local profile directory.
nsresult rv = aCacheDirectory.Clone(getter_AddRefs(mCacheDirectory));
NS_ENSURE_SUCCESS(rv, rv);
// Create the handles to the update and backup directories.
rv = SetupPathNames();
NS_ENSURE_SUCCESS(rv, rv);
// Clean up any to-delete directories that haven't been deleted yet.
// This is still required for backward compatibility.
rv = CleanToDelete();
NS_ENSURE_SUCCESS(rv, rv);
// If we met a crash during the previous update, "safebrowsing-updating"
// directory will exist and let's remove it.
rv = mUpdatingDirectory->Remove(true);
if (NS_SUCCEEDED(rv)) {
// If the "safebrowsing-updating" exists, it implies a crash occurred
// in the previous update.
LOG(("We may have hit a crash in the previous update."));
}
// Check whether we have an incomplete update and recover from the
// backup if so.
rv = RecoverBackups();
NS_ENSURE_SUCCESS(rv, rv);
// Make sure the main store directory exists.
rv = CreateStoreDirectory();
NS_ENSURE_SUCCESS(rv, rv);
// Build the list of know urlclassifier lists
// XXX: Disk IO potentially on the main thread during startup
RegenActiveTables();
return NS_OK;
}
void Classifier::Close() {
// Close will be called by PreShutdown, so it is important to note that
// things put here should not affect an ongoing update thread.
mIsClosed = true;
DropStores();
}
void Classifier::Reset() {
MOZ_ASSERT(NS_GetCurrentThread() != mUpdateThread,
"Reset() MUST NOT be called on update thread");
LOG(("Reset() is called so we interrupt the update."));
mUpdateInterrupted = true;
RefPtr<Classifier> self = this;
auto resetFunc = [self] {
if (self->mIsClosed) {
return; // too late to reset, bail
}
self->DropStores();
self->mRootStoreDirectory->Remove(true);
self->mBackupDirectory->Remove(true);
self->mUpdatingDirectory->Remove(true);
self->mToDeleteDirectory->Remove(true);
self->CreateStoreDirectory();
self->RegenActiveTables();
};
if (!mUpdateThread) {
LOG(("Async update has been disabled. Just Reset() on worker thread."));
resetFunc();
return;
}
nsCOMPtr<nsIRunnable> r =
NS_NewRunnableFunction("safebrowsing::Classifier::Reset", resetFunc);
SyncRunnable::DispatchToThread(mUpdateThread, r);
}
void Classifier::ResetTables(ClearType aType,
const nsTArray<nsCString>& aTables) {
for (uint32_t i = 0; i < aTables.Length(); i++) {
LOG(("Resetting table: %s", aTables[i].get()));
RefPtr<LookupCache> cache = GetLookupCache(aTables[i]);
if (cache) {
// Remove any cached Completes for this table if clear type is Clear_Cache
if (aType == Clear_Cache) {
cache->ClearCache();
} else {
cache->ClearAll();
}
}
}
// Clear on-disk database if clear type is Clear_All
if (aType == Clear_All) {
DeleteTables(mRootStoreDirectory, aTables);
RegenActiveTables();
}
}
void Classifier::DeleteTables(nsIFile* aDirectory,
const nsTArray<nsCString>& aTables) {
nsCOMPtr<nsIDirectoryEnumerator> entries;
nsresult rv = aDirectory->GetDirectoryEntries(getter_AddRefs(entries));
NS_ENSURE_SUCCESS_VOID(rv);
nsCOMPtr<nsIFile> file;
while (NS_SUCCEEDED(rv = entries->GetNextFile(getter_AddRefs(file))) &&
file) {
// If |file| is a directory, recurse to find its entries as well.
bool isDirectory;
if (NS_FAILED(file->IsDirectory(&isDirectory))) {
continue;
}
if (isDirectory) {
DeleteTables(file, aTables);
continue;
}
nsCString leafName;
rv = file->GetNativeLeafName(leafName);
NS_ENSURE_SUCCESS_VOID(rv);
// Remove file extension if there's one.
int32_t dotPosition = leafName.RFind(".");
if (dotPosition >= 0) {
leafName.Truncate(dotPosition);
}
if (!leafName.IsEmpty() && aTables.Contains(leafName)) {
if (NS_FAILED(file->Remove(false))) {
NS_WARNING(nsPrintfCString("Fail to remove file %s from the disk",
leafName.get())
.get());
}
}
}
NS_ENSURE_SUCCESS_VOID(rv);
}
void Classifier::TableRequest(nsACString& aResult) {
MOZ_ASSERT(!NS_IsMainThread(),
"TableRequest must be called on the classifier worker thread.");
// This function and all disk I/O are guaranteed to occur
// on the same thread so we don't need to add a lock around.
if (!mIsTableRequestResultOutdated) {
aResult = mTableRequestResult;
return;
}
// Generating v2 table info.
nsTArray<nsCString> tables;
ActiveTables(tables);
for (uint32_t i = 0; i < tables.Length(); i++) {
HashStore store(tables[i], GetProvider(tables[i]), mRootStoreDirectory);
nsresult rv = store.Open();
if (NS_FAILED(rv)) {
continue;
}
ChunkSet& adds = store.AddChunks();
ChunkSet& subs = store.SubChunks();
// Open HashStore will always succeed even that is not a v2 table.
// So skip tables without add and sub chunks.
if (adds.Length() == 0 && subs.Length() == 0) {
continue;
}
aResult.Append(store.TableName());
aResult.Append(';');
if (adds.Length() > 0) {
aResult.AppendLiteral("a:");
nsAutoCString addList;
adds.Serialize(addList);
aResult.Append(addList);
}
if (subs.Length() > 0) {
if (adds.Length() > 0) aResult.Append(':');
aResult.AppendLiteral("s:");
nsAutoCString subList;
subs.Serialize(subList);
aResult.Append(subList);
}
aResult.Append('\n');
}
// Load meta data from *.metadata files in the root directory.
// Specifically for v4 tables.
nsCString metadata;
nsresult rv = LoadMetadata(mRootStoreDirectory, metadata);
if (NS_SUCCEEDED(rv)) {
aResult.Append(metadata);
}
// Update the TableRequest result in-memory cache.
mTableRequestResult = aResult;
mIsTableRequestResultOutdated = false;
}
nsresult Classifier::CheckURIFragments(
const nsTArray<nsCString>& aSpecFragments, const nsACString& aTable,
LookupResultArray& aResults) {
// A URL can form up to 30 different fragments
MOZ_ASSERT(aSpecFragments.Length() <=
(MAX_HOST_COMPONENTS * (MAX_PATH_COMPONENTS + 2)));
LOG(("Checking table %s", aTable.BeginReading()));
RefPtr<LookupCache> cache = GetLookupCache(aTable);
if (NS_WARN_IF(!cache)) {
return NS_ERROR_FAILURE;
}
// Now check each lookup fragment against the entries in the DB.
for (uint32_t i = 0; i < aSpecFragments.Length(); i++) {
Completion lookupHash;
lookupHash.FromPlaintext(aSpecFragments[i]);
if (LOG_ENABLED()) {
nsAutoCString checking;
lookupHash.ToHexString(checking);
LOG(("Checking fragment %s, hash %s (%X)", aSpecFragments[i].get(),
checking.get(), lookupHash.ToUint32()));
}
bool has, confirmed;
uint32_t matchLength;
nsresult rv = cache->Has(lookupHash, &has, &matchLength, &confirmed);
NS_ENSURE_SUCCESS(rv, rv);
if (has) {
RefPtr<LookupResult> result = new LookupResult;
aResults.AppendElement(result);
LOG(("Found a result in %s: %s", cache->TableName().get(),
confirmed ? "confirmed." : "Not confirmed."));
result->hash.complete = lookupHash;
result->mConfirmed = confirmed;
result->mTableName.Assign(cache->TableName());
result->mPartialHashLength = confirmed ? COMPLETE_SIZE : matchLength;
result->mProtocolV2 = LookupCache::Cast<LookupCacheV2>(cache);
}
}
return NS_OK;
}
static nsresult SwapDirectoryContent(nsIFile* aDir1, nsIFile* aDir2,
nsIFile* aParentDir, nsIFile* aTempDir) {
// Pre-condition: |aDir1| and |aDir2| are directory and their parent
// are both |aParentDir|.
//
// Post-condition: The locations where aDir1 and aDir2 point to will not
// change but their contents will be exchanged. If we failed
// to swap their content, everything will be rolled back.
nsAutoCString tempDirName;
aTempDir->GetNativeLeafName(tempDirName);
nsresult rv;
nsAutoCString dirName1, dirName2;
aDir1->GetNativeLeafName(dirName1);
aDir2->GetNativeLeafName(dirName2);
LOG(("Swapping directories %s and %s...", dirName1.get(), dirName2.get()));
// 1. Rename "dirName1" to "temp"
rv = aDir1->RenameToNative(nullptr, tempDirName);
if (NS_FAILED(rv)) {
LOG(("Unable to rename %s to %s", dirName1.get(), tempDirName.get()));
return rv; // Nothing to roll back.
}
// 1.1. Create a handle for temp directory. This is required since
// |nsIFile.rename| will not change the location where the
// object points to.
nsCOMPtr<nsIFile> tempDirectory;
rv = aParentDir->Clone(getter_AddRefs(tempDirectory));
rv = tempDirectory->AppendNative(tempDirName);
// 2. Rename "dirName2" to "dirName1".
rv = aDir2->RenameToNative(nullptr, dirName1);
if (NS_FAILED(rv)) {
LOG(("Failed to rename %s to %s. Rename temp directory back to %s",
dirName2.get(), dirName1.get(), dirName1.get()));
nsresult rbrv = tempDirectory->RenameToNative(nullptr, dirName1);
NS_ENSURE_SUCCESS(rbrv, rbrv);
return rv;
}
// 3. Rename "temp" to "dirName2".
rv = tempDirectory->RenameToNative(nullptr, dirName2);
if (NS_FAILED(rv)) {
LOG(("Failed to rename temp directory to %s. ", dirName2.get()));
// We've done (1) renaming "dir1 to temp" and
// (2) renaming "dir2 to dir1"
// so the rollback is
// (1) renaming "dir1 to dir2" and
// (2) renaming "temp to dir1"
nsresult rbrv; // rollback result
rbrv = aDir1->RenameToNative(nullptr, dirName2);
NS_ENSURE_SUCCESS(rbrv, rbrv);
rbrv = tempDirectory->RenameToNative(nullptr, dirName1);
NS_ENSURE_SUCCESS(rbrv, rbrv);
return rv;
}
return rv;
}
void Classifier::RemoveUpdateIntermediaries() {
// Remove old LookupCaches.
mNewLookupCaches.Clear();
// Remove the "old" directory. (despite its looking-new name)
if (NS_FAILED(mUpdatingDirectory->Remove(true))) {
// If the directory is locked from removal for some reason,
// we will fail here and it doesn't matter until the next
// update. (the next udpate will fail due to the removable
// "safebrowsing-udpating" directory.)
LOG(("Failed to remove updating directory."));
}
}
void Classifier::CopyAndInvalidateFullHashCache() {
MOZ_ASSERT(NS_GetCurrentThread() != mUpdateThread,
"CopyAndInvalidateFullHashCache cannot be called on update thread "
"since it mutates mLookupCaches which is only safe on "
"worker thread.");
// New lookup caches are built from disk, data likes cache which is
// generated online won't exist. We have to manually copy cache from
// old LookupCache to new LookupCache.
for (auto& newCache : mNewLookupCaches) {
for (auto& oldCache : mLookupCaches) {
if (oldCache->TableName() == newCache->TableName()) {
newCache->CopyFullHashCache(oldCache);
break;
}
}
}
// Clear cache when update.
// Invalidate cache entries in CopyAndInvalidateFullHashCache because only
// at this point we will have cache data in LookupCache.
for (auto& newCache : mNewLookupCaches) {
newCache->InvalidateExpiredCacheEntries();
}
}
void Classifier::MergeNewLookupCaches() {
MOZ_ASSERT(NS_GetCurrentThread() != mUpdateThread,
"MergeNewLookupCaches cannot be called on update thread "
"since it mutates mLookupCaches which is only safe on "
"worker thread.");
for (auto& newCache : mNewLookupCaches) {
// For each element in mNewLookCaches, it will be swapped with
// - An old cache in mLookupCache with the same table name or
// - nullptr (mLookupCache will be expaned) otherwise.
size_t swapIndex = 0;
for (; swapIndex < mLookupCaches.Length(); swapIndex++) {
if (mLookupCaches[swapIndex]->TableName() == newCache->TableName()) {
break;
}
}
if (swapIndex == mLookupCaches.Length()) {
mLookupCaches.AppendElement(nullptr);
}
Swap(mLookupCaches[swapIndex], newCache);
mLookupCaches[swapIndex]->UpdateRootDirHandle(mRootStoreDirectory);
}
// At this point, mNewLookupCaches's length remains the same but
// will contain either old cache (override) or nullptr (append).
}
nsresult Classifier::SwapInNewTablesAndCleanup() {
nsresult rv;
// Step 1. Swap in on-disk tables. The idea of using "safebrowsing-backup"
// as the intermediary directory is we can get databases recovered if
// crash occurred in any step of the swap. (We will recover from
// "safebrowsing-backup" in OpenDb().)
rv = SwapDirectoryContent(mUpdatingDirectory, // contains new tables
mRootStoreDirectory, // contains old tables
mCacheDirectory, // common parent dir
mBackupDirectory); // intermediary dir for swap
if (NS_FAILED(rv)) {
LOG(("Failed to swap in on-disk tables."));
RemoveUpdateIntermediaries();
return rv;
}
// Step 2. Merge mNewLookupCaches into mLookupCaches. The outdated
// LookupCaches will be stored in mNewLookupCaches and be cleaned
// up later.
MergeNewLookupCaches();
// Step 3. Re-generate active tables based on on-disk tables.
rv = RegenActiveTables();
if (NS_FAILED(rv)) {
LOG(("Failed to re-generate active tables!"));
}
// Step 4. Clean up intermediaries for update.
RemoveUpdateIntermediaries();
// Step 5. Invalidate cached tableRequest request.
mIsTableRequestResultOutdated = true;
LOG(("Done swap in updated tables."));
return rv;
}
void Classifier::FlushAndDisableAsyncUpdate() {
LOG(("Classifier::FlushAndDisableAsyncUpdate [%p, %p]", this,
mUpdateThread.get()));
if (!mUpdateThread) {
LOG(("Async update has been disabled."));
return;
}
mUpdateThread->Shutdown();
mUpdateThread = nullptr;
}
nsresult Classifier::AsyncApplyUpdates(const TableUpdateArray& aUpdates,
const AsyncUpdateCallback& aCallback) {
LOG(("Classifier::AsyncApplyUpdates"));
if (!mUpdateThread) {
LOG(("Async update has already been disabled."));
return NS_ERROR_FAILURE;
}
// Caller thread | Update thread
// --------------------------------------------------------
// | ApplyUpdatesBackground
// (processing other task) | (bg-update done. ping back to caller
// thread) (processing other task) | idle... ApplyUpdatesForeground |
// callback |
MOZ_ASSERT(mNewLookupCaches.IsEmpty(),
"There should be no leftovers from a previous update.");
mUpdateInterrupted = false;
nsresult rv =
mRootStoreDirectory->Clone(getter_AddRefs(mRootStoreDirectoryForUpdate));
if (NS_FAILED(rv)) {
LOG(("Failed to clone mRootStoreDirectory for update."));
return rv;
}
nsCOMPtr<nsIThread> callerThread = NS_GetCurrentThread();
MOZ_ASSERT(callerThread != mUpdateThread);
RefPtr<Classifier> self = this;
nsCOMPtr<nsIRunnable> bgRunnable = NS_NewRunnableFunction(
"safebrowsing::Classifier::AsyncApplyUpdates",
[self, aUpdates, aCallback, callerThread] {
MOZ_ASSERT(NS_GetCurrentThread() == self->mUpdateThread,
"MUST be on update thread");
nsresult bgRv;
nsCString failedTableName;
TableUpdateArray updates;
// Make a copy of the array since we'll be removing entries as
// we process them on the background thread.
if (updates.AppendElements(aUpdates, fallible)) {
LOG(("Step 1. ApplyUpdatesBackground on update thread."));
bgRv = self->ApplyUpdatesBackground(updates, failedTableName);
} else {
LOG(
("Step 1. Not enough memory to run ApplyUpdatesBackground on "
"update thread."));
bgRv = NS_ERROR_OUT_OF_MEMORY;
}
nsCOMPtr<nsIRunnable> fgRunnable = NS_NewRunnableFunction(
"safebrowsing::Classifier::AsyncApplyUpdates",
[self, aCallback, bgRv, failedTableName, callerThread] {
MOZ_ASSERT(NS_GetCurrentThread() == callerThread,
"MUST be on caller thread");
LOG(("Step 2. ApplyUpdatesForeground on caller thread"));
nsresult rv = self->ApplyUpdatesForeground(bgRv, failedTableName);
LOG(("Step 3. Updates applied! Fire callback."));
aCallback(rv);
});
callerThread->Dispatch(fgRunnable, NS_DISPATCH_NORMAL);
});
return mUpdateThread->Dispatch(bgRunnable, NS_DISPATCH_NORMAL);
}
nsresult Classifier::ApplyUpdatesBackground(TableUpdateArray& aUpdates,
nsACString& aFailedTableName) {
// |mUpdateInterrupted| is guaranteed to have been unset.
// If |mUpdateInterrupted| is set at any point, Reset() must have
// been called then we need to interrupt the update process.
// We only add checkpoints for non-trivial tasks.
if (aUpdates.IsEmpty()) {
return NS_OK;
}
nsUrlClassifierUtils* urlUtil = nsUrlClassifierUtils::GetInstance();
if (NS_WARN_IF(!urlUtil)) {
return NS_ERROR_FAILURE;
}
nsCString provider;
// Assume all TableUpdate objects should have the same provider.
urlUtil->GetTelemetryProvider(aUpdates[0]->TableName(), provider);
Telemetry::AutoTimer<Telemetry::URLCLASSIFIER_CL_KEYED_UPDATE_TIME>
keyedTimer(provider);
PRIntervalTime clockStart = 0;
if (LOG_ENABLED()) {
clockStart = PR_IntervalNow();
}
nsresult rv;
// Check point 1: Copying files takes time so we check ShouldAbort()
// inside CopyInUseDirForUpdate().
rv = CopyInUseDirForUpdate(); // i.e. mUpdatingDirectory will be setup.
if (NS_FAILED(rv)) {
LOG(("Failed to copy in-use directory for update."));
return (rv == NS_ERROR_ABORT) ? NS_OK : rv;
}
LOG(("Applying %zu table updates.", aUpdates.Length()));
for (uint32_t i = 0; i < aUpdates.Length(); i++) {
RefPtr<const TableUpdate> update = aUpdates[i];
if (!update) {
// Previous UpdateHashStore() may have consumed this update..
continue;
}
// Run all updates for one table
nsAutoCString updateTable(update->TableName());
// Check point 2: Processing downloaded data takes time.
if (ShouldAbort()) {
LOG(("Update is interrupted. Stop building new tables."));
return NS_OK;
}
// Will update the mirrored in-memory and on-disk databases.
if (TableUpdate::Cast<TableUpdateV2>(update)) {
rv = UpdateHashStore(aUpdates, updateTable);
} else {
rv = UpdateTableV4(aUpdates, updateTable);
}
if (NS_FAILED(rv)) {
aFailedTableName = updateTable;
RemoveUpdateIntermediaries();
return rv;
}
}
if (LOG_ENABLED()) {
PRIntervalTime clockEnd = PR_IntervalNow();
LOG(("update took %dms\n",
PR_IntervalToMilliseconds(clockEnd - clockStart)));
}
return rv;
}
nsresult Classifier::ApplyUpdatesForeground(
nsresult aBackgroundRv, const nsACString& aFailedTableName) {
if (ShouldAbort()) {
LOG(("Update is interrupted! Just remove update intermediaries."));
RemoveUpdateIntermediaries();
return NS_OK;
}
if (NS_SUCCEEDED(aBackgroundRv)) {
// Copy and Invalidate fullhash cache here because this call requires
// mLookupCaches which is only available on work-thread
CopyAndInvalidateFullHashCache();
return SwapInNewTablesAndCleanup();
}
if (NS_ERROR_OUT_OF_MEMORY != aBackgroundRv) {
ResetTables(Clear_All, nsTArray<nsCString>{nsCString(aFailedTableName)});
}
return aBackgroundRv;
}
nsresult Classifier::ApplyFullHashes(ConstTableUpdateArray& aUpdates) {
MOZ_ASSERT(NS_GetCurrentThread() != mUpdateThread,
"ApplyFullHashes() MUST NOT be called on update thread");
MOZ_ASSERT(
!NS_IsMainThread(),
"ApplyFullHashes() must be called on the classifier worker thread.");
LOG(("Applying %zu table gethashes.", aUpdates.Length()));
for (uint32_t i = 0; i < aUpdates.Length(); i++) {
nsresult rv = UpdateCache(aUpdates[i]);
NS_ENSURE_SUCCESS(rv, rv);
aUpdates[i] = nullptr;
}
return NS_OK;
}
void Classifier::GetCacheInfo(const nsACString& aTable,
nsIUrlClassifierCacheInfo** aCache) {
RefPtr<const LookupCache> lookupCache = GetLookupCache(aTable);
if (!lookupCache) {
return;
}
lookupCache->GetCacheInfo(aCache);
}
void Classifier::DropStores() {
// See the comment in Classifier::Close() before adding anything here.
mLookupCaches.Clear();
}
nsresult Classifier::RegenActiveTables() {
if (ShouldAbort()) {
return NS_OK; // nothing to do, the classifier is done
}
mActiveTablesCache.Clear();
nsTArray<nsCString> foundTables;
ScanStoreDir(mRootStoreDirectory, foundTables);
for (uint32_t i = 0; i < foundTables.Length(); i++) {
nsCString table(foundTables[i]);
RefPtr<const LookupCache> lookupCache = GetLookupCache(table);
if (!lookupCache) {
LOG(("Inactive table (no cache): %s", table.get()));
continue;
}
if (!lookupCache->IsPrimed()) {
LOG(("Inactive table (cache not primed): %s", table.get()));
continue;
}
if (LookupCache::Cast<const LookupCacheV4>(lookupCache)) {
LOG(("Active v4 table: %s", table.get()));
} else {
HashStore store(table, GetProvider(table), mRootStoreDirectory);
nsresult rv = store.Open();
if (NS_FAILED(rv)) {
continue;
}
const ChunkSet& adds = store.AddChunks();
const ChunkSet& subs = store.SubChunks();
if (adds.Length() == 0 && subs.Length() == 0) {
continue;
}
LOG(("Active v2 table: %s", store.TableName().get()));
}
mActiveTablesCache.AppendElement(table);
}
return NS_OK;
}
nsresult Classifier::ScanStoreDir(nsIFile* aDirectory,
nsTArray<nsCString>& aTables) {
nsCOMPtr<nsIDirectoryEnumerator> entries;
nsresult rv = aDirectory->GetDirectoryEntries(getter_AddRefs(entries));
NS_ENSURE_SUCCESS(rv, rv);
nsCOMPtr<nsIFile> file;
while (NS_SUCCEEDED(rv = entries->GetNextFile(getter_AddRefs(file))) &&
file) {
// If |file| is a directory, recurse to find its entries as well.
bool isDirectory;
if (NS_FAILED(file->IsDirectory(&isDirectory))) {
continue;
}
if (isDirectory) {
ScanStoreDir(file, aTables);
continue;
}
nsCString leafName;
rv = file->GetNativeLeafName(leafName);
NS_ENSURE_SUCCESS(rv, rv);
// Both v2 and v4 contain .pset file
nsCString suffix(NS_LITERAL_CSTRING(".pset"));
int32_t dot = leafName.RFind(suffix);
if (dot != -1) {
leafName.Cut(dot, suffix.Length());
aTables.AppendElement(leafName);
}
}
NS_ENSURE_SUCCESS(rv, rv);
return NS_OK;
}
nsresult Classifier::ActiveTables(nsTArray<nsCString>& aTables) const {
aTables = mActiveTablesCache;
return NS_OK;
}
nsresult Classifier::CleanToDelete() {
bool exists;
nsresult rv = mToDeleteDirectory->Exists(&exists);
NS_ENSURE_SUCCESS(rv, rv);
if (exists) {
rv = mToDeleteDirectory->Remove(true);
NS_ENSURE_SUCCESS(rv, rv);
}
return NS_OK;
}
#ifdef MOZ_SAFEBROWSING_DUMP_FAILED_UPDATES
already_AddRefed<nsIFile> Classifier::GetFailedUpdateDirectroy() {
nsCString failedUpdatekDirName = STORE_DIRECTORY + nsCString("-failedupdate");
nsCOMPtr<nsIFile> failedUpdatekDirectory;
if (NS_FAILED(
mCacheDirectory->Clone(getter_AddRefs(failedUpdatekDirectory))) ||
NS_FAILED(failedUpdatekDirectory->AppendNative(failedUpdatekDirName))) {
LOG(("Failed to init failedUpdatekDirectory."));
return nullptr;
}
return failedUpdatekDirectory.forget();
}
nsresult Classifier::DumpRawTableUpdates(const nsACString& aRawUpdates) {
LOG(("Dumping raw table updates..."));
DumpFailedUpdate();
nsCOMPtr<nsIFile> failedUpdatekDirectory = GetFailedUpdateDirectroy();
// Create tableupdate.bin and dump raw table update data.
nsCOMPtr<nsIFile> rawTableUpdatesFile;
nsCOMPtr<nsIOutputStream> outputStream;
if (NS_FAILED(
failedUpdatekDirectory->Clone(getter_AddRefs(rawTableUpdatesFile))) ||
NS_FAILED(
rawTableUpdatesFile->AppendNative(nsCString("tableupdates.bin"))) ||
NS_FAILED(NS_NewLocalFileOutputStream(
getter_AddRefs(outputStream), rawTableUpdatesFile,
PR_WRONLY | PR_TRUNCATE | PR_CREATE_FILE))) {
LOG(("Failed to create file to dump raw table updates."));
return NS_ERROR_FAILURE;
}
// Write out the data.
uint32_t written;
nsresult rv = outputStream->Write(aRawUpdates.BeginReading(),
aRawUpdates.Length(), &written);
NS_ENSURE_SUCCESS(rv, rv);
NS_ENSURE_TRUE(written == aRawUpdates.Length(), NS_ERROR_FAILURE);
return rv;
}
nsresult Classifier::DumpFailedUpdate() {
LOG(("Dumping failed update..."));