-
Notifications
You must be signed in to change notification settings - Fork 6
Expand file tree
/
Copy pathcompress.c
More file actions
1625 lines (1456 loc) · 43.6 KB
/
compress.c
File metadata and controls
1625 lines (1456 loc) · 43.6 KB
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
// SPDX-License-Identifier: GPL-2.0-or-later
/*
* NTFS kernel compressed attributes handling.
*
* Copyright (c) 2001-2004 Anton Altaparmakov
* Copyright (c) 2002 Richard Russon
* Copyright (c) 2025 LG Electronics Co., Ltd.
*
* Part of this file is based on code from the NTFS-3G.
* and is copyrighted by the respective authors below:
* Copyright (c) 2004-2005 Anton Altaparmakov
* Copyright (c) 2004-2006 Szabolcs Szakacsits
* Copyright (c) 2005 Yura Pakhuchiy
* Copyright (c) 2009-2014 Jean-Pierre Andre
* Copyright (c) 2014 Eric Biggers
*/
#include <linux/fs.h>
#include <linux/blkdev.h>
#include <linux/vmalloc.h>
#include <linux/slab.h>
#include "attrib.h"
#include "inode.h"
#include "debug.h"
#include "ntfs.h"
#include "lcnalloc.h"
#include "mft.h"
/*
* Constants used in the compression code
*/
enum {
/* Token types and access mask. */
NTFS_SYMBOL_TOKEN = 0,
NTFS_PHRASE_TOKEN = 1,
NTFS_TOKEN_MASK = 1,
/* Compression sub-block constants. */
NTFS_SB_SIZE_MASK = 0x0fff,
NTFS_SB_SIZE = 0x1000,
NTFS_SB_IS_COMPRESSED = 0x8000,
/*
* The maximum compression block size is by definition 16 * the cluster
* size, with the maximum supported cluster size being 4kiB. Thus the
* maximum compression buffer size is 64kiB, so we use this when
* initializing the compression buffer.
*/
NTFS_MAX_CB_SIZE = 64 * 1024,
};
/*
* ntfs_compression_buffer - one buffer for the decompression engine
*/
static u8 *ntfs_compression_buffer;
/*
* ntfs_cb_lock - mutex lock which protects ntfs_compression_buffer
*/
static DEFINE_MUTEX(ntfs_cb_lock);
/*
* allocate_compression_buffers - allocate the decompression buffers
*
* Caller has to hold the ntfs_lock mutex.
*
* Return 0 on success or -ENOMEM if the allocations failed.
*/
int allocate_compression_buffers(void)
{
if (ntfs_compression_buffer)
return 0;
ntfs_compression_buffer = vmalloc(NTFS_MAX_CB_SIZE);
if (!ntfs_compression_buffer)
return -ENOMEM;
return 0;
}
/*
* free_compression_buffers - free the decompression buffers
*
* Caller has to hold the ntfs_lock mutex.
*/
void free_compression_buffers(void)
{
mutex_lock(&ntfs_cb_lock);
if (!ntfs_compression_buffer) {
mutex_unlock(&ntfs_cb_lock);
return;
}
vfree(ntfs_compression_buffer);
ntfs_compression_buffer = NULL;
mutex_unlock(&ntfs_cb_lock);
}
/*
* zero_partial_compressed_page - zero out of bounds compressed page region
* @page: page to zero
* @initialized_size: initialized size of the attribute
*/
static void zero_partial_compressed_page(struct page *page,
const s64 initialized_size)
{
u8 *kp = page_address(page);
unsigned int kp_ofs;
ntfs_debug("Zeroing page region outside initialized size.");
#if LINUX_VERSION_CODE >= KERNEL_VERSION(6, 16, 0)
if (((s64)page->__folio_index << PAGE_SHIFT) >= initialized_size) {
clear_page(kp);
return;
}
#else
if (((s64)page->index << PAGE_SHIFT) >= initialized_size) {
clear_page(kp);
return;
}
#endif
kp_ofs = initialized_size & ~PAGE_MASK;
memset(kp + kp_ofs, 0, PAGE_SIZE - kp_ofs);
}
/*
* handle_bounds_compressed_page - test for&handle out of bounds compressed page
* @page: page to check and handle
* @i_size: file size
* @initialized_size: initialized size of the attribute
*/
static inline void handle_bounds_compressed_page(struct page *page,
const loff_t i_size, const s64 initialized_size)
{
#if LINUX_VERSION_CODE >= KERNEL_VERSION(6, 16, 0)
if ((page->__folio_index >= (initialized_size >> PAGE_SHIFT)) &&
(initialized_size < i_size))
zero_partial_compressed_page(page, initialized_size);
#else
if ((page->index >= (initialized_size >> PAGE_SHIFT)) &&
(initialized_size < i_size))
zero_partial_compressed_page(page, initialized_size);
#endif
}
/*
* ntfs_decompress - decompress a compression block into an array of pages
* @dest_pages: destination array of pages
* @completed_pages: scratch space to track completed pages
* @dest_index: current index into @dest_pages (IN/OUT)
* @dest_ofs: current offset within @dest_pages[@dest_index] (IN/OUT)
* @dest_max_index: maximum index into @dest_pages (IN)
* @dest_max_ofs: maximum offset within @dest_pages[@dest_max_index] (IN)
* @xpage: the target page (-1 if none) (IN)
* @xpage_done: set to 1 if xpage was completed successfully (IN/OUT)
* @cb_start: compression block to decompress (IN)
* @cb_size: size of compression block @cb_start in bytes (IN)
* @i_size: file size when we started the read (IN)
* @initialized_size: initialized file size when we started the read (IN)
*
* The caller must have disabled preemption. ntfs_decompress() reenables it when
* the critical section is finished.
*
* This decompresses the compression block @cb_start into the array of
* destination pages @dest_pages starting at index @dest_index into @dest_pages
* and at offset @dest_pos into the page @dest_pages[@dest_index].
*
* When the page @dest_pages[@xpage] is completed, @xpage_done is set to 1.
* If xpage is -1 or @xpage has not been completed, @xpage_done is not modified.
*
* @cb_start is a pointer to the compression block which needs decompressing
* and @cb_size is the size of @cb_start in bytes (8-64kiB).
*
* Return 0 if success or -EOVERFLOW on error in the compressed stream.
* @xpage_done indicates whether the target page (@dest_pages[@xpage]) was
* completed during the decompression of the compression block (@cb_start).
*
* Warning: This function *REQUIRES* PAGE_SIZE >= 4096 or it will blow up
* unpredicatbly! You have been warned!
*
* Note to hackers: This function may not sleep until it has finished accessing
* the compression block @cb_start as it is a per-CPU buffer.
*/
static int ntfs_decompress(struct page *dest_pages[], int completed_pages[],
int *dest_index, int *dest_ofs, const int dest_max_index,
const int dest_max_ofs, const int xpage, char *xpage_done,
u8 *const cb_start, const u32 cb_size, const loff_t i_size,
const s64 initialized_size)
{
/*
* Pointers into the compressed data, i.e. the compression block (cb),
* and the therein contained sub-blocks (sb).
*/
u8 *cb_end = cb_start + cb_size; /* End of cb. */
u8 *cb = cb_start; /* Current position in cb. */
u8 *cb_sb_start = cb; /* Beginning of the current sb in the cb. */
u8 *cb_sb_end; /* End of current sb / beginning of next sb. */
/* Variables for uncompressed data / destination. */
struct page *dp; /* Current destination page being worked on. */
u8 *dp_addr; /* Current pointer into dp. */
u8 *dp_sb_start; /* Start of current sub-block in dp. */
u8 *dp_sb_end; /* End of current sb in dp (dp_sb_start + NTFS_SB_SIZE). */
u16 do_sb_start; /* @dest_ofs when starting this sub-block. */
u16 do_sb_end; /* @dest_ofs of end of this sb (do_sb_start + NTFS_SB_SIZE). */
/* Variables for tag and token parsing. */
u8 tag; /* Current tag. */
int token; /* Loop counter for the eight tokens in tag. */
int nr_completed_pages = 0;
/* Default error code. */
int err = -EOVERFLOW;
ntfs_debug("Entering, cb_size = 0x%x.", cb_size);
do_next_sb:
ntfs_debug("Beginning sub-block at offset = 0x%zx in the cb.",
cb - cb_start);
/*
* Have we reached the end of the compression block or the end of the
* decompressed data? The latter can happen for example if the current
* position in the compression block is one byte before its end so the
* first two checks do not detect it.
*/
if (cb == cb_end || !le16_to_cpup((__le16 *)cb) ||
(*dest_index == dest_max_index &&
*dest_ofs == dest_max_ofs)) {
int i;
ntfs_debug("Completed. Returning success (0).");
err = 0;
return_error:
/* We can sleep from now on, so we drop lock. */
mutex_unlock(&ntfs_cb_lock);
/* Second stage: finalize completed pages. */
if (nr_completed_pages > 0) {
for (i = 0; i < nr_completed_pages; i++) {
int di = completed_pages[i];
dp = dest_pages[di];
/*
* If we are outside the initialized size, zero
* the out of bounds page range.
*/
handle_bounds_compressed_page(dp, i_size,
initialized_size);
flush_dcache_page(dp);
kunmap_local(page_address(dp));
SetPageUptodate(dp);
unlock_page(dp);
if (di == xpage)
*xpage_done = 1;
else
put_page(dp);
dest_pages[di] = NULL;
}
}
return err;
}
/* Setup offsets for the current sub-block destination. */
do_sb_start = *dest_ofs;
do_sb_end = do_sb_start + NTFS_SB_SIZE;
/* Check that we are still within allowed boundaries. */
if (*dest_index == dest_max_index && do_sb_end > dest_max_ofs)
goto return_overflow;
/* Does the minimum size of a compressed sb overflow valid range? */
if (cb + 6 > cb_end)
goto return_overflow;
/* Setup the current sub-block source pointers and validate range. */
cb_sb_start = cb;
cb_sb_end = cb_sb_start + (le16_to_cpup((__le16 *)cb) & NTFS_SB_SIZE_MASK)
+ 3;
if (cb_sb_end > cb_end)
goto return_overflow;
/* Get the current destination page. */
dp = dest_pages[*dest_index];
if (!dp) {
/* No page present. Skip decompression of this sub-block. */
cb = cb_sb_end;
/* Advance destination position to next sub-block. */
*dest_ofs = (*dest_ofs + NTFS_SB_SIZE) & ~PAGE_MASK;
if (!*dest_ofs && (++*dest_index > dest_max_index))
goto return_overflow;
goto do_next_sb;
}
/* We have a valid destination page. Setup the destination pointers. */
dp_addr = (u8 *)page_address(dp) + do_sb_start;
/* Now, we are ready to process the current sub-block (sb). */
if (!(le16_to_cpup((__le16 *)cb) & NTFS_SB_IS_COMPRESSED)) {
ntfs_debug("Found uncompressed sub-block.");
/* This sb is not compressed, just copy it into destination. */
/* Advance source position to first data byte. */
cb += 2;
/* An uncompressed sb must be full size. */
if (cb_sb_end - cb != NTFS_SB_SIZE)
goto return_overflow;
/* Copy the block and advance the source position. */
memcpy(dp_addr, cb, NTFS_SB_SIZE);
cb += NTFS_SB_SIZE;
/* Advance destination position to next sub-block. */
*dest_ofs += NTFS_SB_SIZE;
*dest_ofs &= ~PAGE_MASK;
if (!(*dest_ofs)) {
finalize_page:
/*
* First stage: add current page index to array of
* completed pages.
*/
completed_pages[nr_completed_pages++] = *dest_index;
if (++*dest_index > dest_max_index)
goto return_overflow;
}
goto do_next_sb;
}
ntfs_debug("Found compressed sub-block.");
/* This sb is compressed, decompress it into destination. */
/* Setup destination pointers. */
dp_sb_start = dp_addr;
dp_sb_end = dp_sb_start + NTFS_SB_SIZE;
/* Forward to the first tag in the sub-block. */
cb += 2;
do_next_tag:
if (cb == cb_sb_end) {
/* Check if the decompressed sub-block was not full-length. */
if (dp_addr < dp_sb_end) {
int nr_bytes = do_sb_end - *dest_ofs;
ntfs_debug("Filling incomplete sub-block with zeroes.");
/* Zero remainder and update destination position. */
memset(dp_addr, 0, nr_bytes);
*dest_ofs += nr_bytes;
}
/* We have finished the current sub-block. */
*dest_ofs &= ~PAGE_MASK;
if (!(*dest_ofs))
goto finalize_page;
goto do_next_sb;
}
/* Check we are still in range. */
if (cb > cb_sb_end || dp_addr > dp_sb_end)
goto return_overflow;
/* Get the next tag and advance to first token. */
tag = *cb++;
/* Parse the eight tokens described by the tag. */
for (token = 0; token < 8; token++, tag >>= 1) {
register u16 i;
u16 lg, pt, length, max_non_overlap;
u8 *dp_back_addr;
/* Check if we are done / still in range. */
if (cb >= cb_sb_end || dp_addr > dp_sb_end)
break;
/* Determine token type and parse appropriately.*/
if ((tag & NTFS_TOKEN_MASK) == NTFS_SYMBOL_TOKEN) {
/*
* We have a symbol token, copy the symbol across, and
* advance the source and destination positions.
*/
*dp_addr++ = *cb++;
++*dest_ofs;
/* Continue with the next token. */
continue;
}
/*
* We have a phrase token. Make sure it is not the first tag in
* the sb as this is illegal and would confuse the code below.
*/
if (dp_addr == dp_sb_start)
goto return_overflow;
/*
* Determine the number of bytes to go back (p) and the number
* of bytes to copy (l). We use an optimized algorithm in which
* we first calculate log2(current destination position in sb),
* which allows determination of l and p in O(1) rather than
* O(n). We just need an arch-optimized log2() function now.
*/
lg = 0;
for (i = *dest_ofs - do_sb_start - 1; i >= 0x10; i >>= 1)
lg++;
/* Get the phrase token into i. */
pt = le16_to_cpup((__le16 *)cb);
/*
* Calculate starting position of the byte sequence in
* the destination using the fact that p = (pt >> (12 - lg)) + 1
* and make sure we don't go too far back.
*/
dp_back_addr = dp_addr - (pt >> (12 - lg)) - 1;
if (dp_back_addr < dp_sb_start)
goto return_overflow;
/* Now calculate the length of the byte sequence. */
length = (pt & (0xfff >> lg)) + 3;
/* Advance destination position and verify it is in range. */
*dest_ofs += length;
if (*dest_ofs > do_sb_end)
goto return_overflow;
/* The number of non-overlapping bytes. */
max_non_overlap = dp_addr - dp_back_addr;
if (length <= max_non_overlap) {
/* The byte sequence doesn't overlap, just copy it. */
memcpy(dp_addr, dp_back_addr, length);
/* Advance destination pointer. */
dp_addr += length;
} else {
/*
* The byte sequence does overlap, copy non-overlapping
* part and then do a slow byte by byte copy for the
* overlapping part. Also, advance the destination
* pointer.
*/
memcpy(dp_addr, dp_back_addr, max_non_overlap);
dp_addr += max_non_overlap;
dp_back_addr += max_non_overlap;
length -= max_non_overlap;
while (length--)
*dp_addr++ = *dp_back_addr++;
}
/* Advance source position and continue with the next token. */
cb += 2;
}
/* No tokens left in the current tag. Continue with the next tag. */
goto do_next_tag;
return_overflow:
ntfs_error(NULL, "Failed. Returning -EOVERFLOW.");
goto return_error;
}
/*
* ntfs_read_compressed_block - read a compressed block into the page cache
* @folio: locked folio in the compression block(s) we need to read
*
* When we are called the page has already been verified to be locked and the
* attribute is known to be non-resident, not encrypted, but compressed.
*
* 1. Determine which compression block(s) @page is in.
* 2. Get hold of all pages corresponding to this/these compression block(s).
* 3. Read the (first) compression block.
* 4. Decompress it into the corresponding pages.
* 5. Throw the compressed data away and proceed to 3. for the next compression
* block or return success if no more compression blocks left.
*
* Warning: We have to be careful what we do about existing pages. They might
* have been written to so that we would lose data if we were to just overwrite
* them with the out-of-date uncompressed data.
*/
#if LINUX_VERSION_CODE >= KERNEL_VERSION(6, 6, 0)
int ntfs_read_compressed_block(struct folio *folio)
{
struct page *page = &folio->page;
#else
int ntfs_read_compressed_block(struct page *page)
{
#endif
loff_t i_size;
s64 initialized_size;
struct address_space *mapping = page->mapping;
struct ntfs_inode *ni = NTFS_I(mapping->host);
struct ntfs_volume *vol = ni->vol;
struct super_block *sb = vol->sb;
struct runlist_element *rl;
unsigned long flags;
u8 *cb, *cb_pos, *cb_end;
#if LINUX_VERSION_CODE >= KERNEL_VERSION(6, 16, 0)
unsigned long offset, index = page->__folio_index;
#else
unsigned long offset, index = page->index;
#endif
u32 cb_size = ni->itype.compressed.block_size;
u64 cb_size_mask = cb_size - 1UL;
s64 vcn;
s64 lcn;
/* The first wanted vcn (minimum alignment is PAGE_SIZE). */
s64 start_vcn = (((s64)index << PAGE_SHIFT) & ~cb_size_mask) >>
vol->cluster_size_bits;
/*
* The first vcn after the last wanted vcn (minimum alignment is again
* PAGE_SIZE.
*/
s64 end_vcn = ((((s64)(index + 1UL) << PAGE_SHIFT) + cb_size - 1)
& ~cb_size_mask) >> vol->cluster_size_bits;
/* Number of compression blocks (cbs) in the wanted vcn range. */
unsigned int nr_cbs = ntfs_cluster_to_bytes(vol, end_vcn - start_vcn) >>
ni->itype.compressed.block_size_bits;
/*
* Number of pages required to store the uncompressed data from all
* compression blocks (cbs) overlapping @page. Due to alignment
* guarantees of start_vcn and end_vcn, no need to round up here.
*/
unsigned int nr_pages = ntfs_cluster_to_pidx(vol, end_vcn - start_vcn);
unsigned int xpage, max_page, cur_page, cur_ofs, i, page_ofs, page_index;
unsigned int cb_clusters, cb_max_ofs;
int cb_max_page, err = 0;
struct page **pages;
int *completed_pages;
unsigned char xpage_done = 0;
struct page *lpage;
ntfs_debug("Entering, page->index = 0x%lx, cb_size = 0x%x, nr_pages = %i.",
index, cb_size, nr_pages);
/*
* Bad things happen if we get here for anything that is not an
* unnamed $DATA attribute.
*/
if (ni->type != AT_DATA || ni->name_len) {
unlock_page(page);
return -EIO;
}
pages = kmalloc_array(nr_pages, sizeof(struct page *), GFP_NOFS);
completed_pages = kmalloc_array(nr_pages + 1, sizeof(int), GFP_NOFS);
if (unlikely(!pages || !completed_pages)) {
kfree(pages);
kfree(completed_pages);
unlock_page(page);
ntfs_error(vol->sb, "Failed to allocate internal buffers.");
return -ENOMEM;
}
/*
* We have already been given one page, this is the one we must do.
* Once again, the alignment guarantees keep it simple.
*/
offset = ntfs_cluster_to_pidx(vol, start_vcn);
xpage = index - offset;
pages[xpage] = page;
/*
* The remaining pages need to be allocated and inserted into the page
* cache, alignment guarantees keep all the below much simpler. (-8
*/
read_lock_irqsave(&ni->size_lock, flags);
i_size = i_size_read(VFS_I(ni));
initialized_size = ni->initialized_size;
read_unlock_irqrestore(&ni->size_lock, flags);
max_page = ((i_size + PAGE_SIZE - 1) >> PAGE_SHIFT) -
offset;
/* Is the page fully outside i_size? (truncate in progress) */
if (xpage >= max_page) {
kfree(pages);
kfree(completed_pages);
#if LINUX_VERSION_CODE >= KERNEL_VERSION(6, 17, 0)
zero_user_segments(page, 0, PAGE_SIZE, 0, 0);
#else
zero_user(page, 0, PAGE_SIZE);
#endif
ntfs_debug("Compressed read outside i_size - truncated?");
SetPageUptodate(page);
unlock_page(page);
return 0;
}
if (nr_pages < max_page)
max_page = nr_pages;
for (i = 0; i < max_page; i++, offset++) {
if (i != xpage)
pages[i] = grab_cache_page_nowait(mapping, offset);
page = pages[i];
if (page) {
/*
* We only (re)read the page if it isn't already read
* in and/or dirty or we would be losing data or at
* least wasting our time.
*/
if (!PageDirty(page) && (!PageUptodate(page))) {
kmap_local_page(page);
continue;
}
unlock_page(page);
put_page(page);
pages[i] = NULL;
}
}
/*
* We have the runlist, and all the destination pages we need to fill.
* Now read the first compression block.
*/
cur_page = 0;
cur_ofs = 0;
cb_clusters = ni->itype.compressed.block_clusters;
do_next_cb:
nr_cbs--;
mutex_lock(&ntfs_cb_lock);
if (!ntfs_compression_buffer)
if (allocate_compression_buffers()) {
mutex_unlock(&ntfs_cb_lock);
goto err_out;
}
cb = ntfs_compression_buffer;
cb_pos = cb;
cb_end = cb + cb_size;
rl = NULL;
for (vcn = start_vcn, start_vcn += cb_clusters; vcn < start_vcn;
vcn++) {
bool is_retry = false;
if (!rl) {
lock_retry_remap:
down_read(&ni->runlist.lock);
rl = ni->runlist.rl;
}
if (likely(rl != NULL)) {
/* Seek to element containing target vcn. */
while (rl->length && rl[1].vcn <= vcn)
rl++;
lcn = ntfs_rl_vcn_to_lcn(rl, vcn);
} else
lcn = LCN_RL_NOT_MAPPED;
ntfs_debug("Reading vcn = 0x%llx, lcn = 0x%llx.",
(unsigned long long)vcn,
(unsigned long long)lcn);
if (lcn < 0) {
/*
* When we reach the first sparse cluster we have
* finished with the cb.
*/
if (lcn == LCN_HOLE)
break;
if (is_retry || lcn != LCN_RL_NOT_MAPPED) {
mutex_unlock(&ntfs_cb_lock);
goto rl_err;
}
is_retry = true;
/*
* Attempt to map runlist, dropping lock for the
* duration.
*/
up_read(&ni->runlist.lock);
if (!ntfs_map_runlist(ni, vcn))
goto lock_retry_remap;
mutex_unlock(&ntfs_cb_lock);
goto map_rl_err;
}
page_ofs = ntfs_cluster_to_poff(vol, lcn);
page_index = ntfs_cluster_to_pidx(vol, lcn);
#if LINUX_VERSION_CODE >= KERNEL_VERSION(6, 11, 0)
lpage = read_mapping_page(sb->s_bdev->bd_mapping,
page_index, NULL);
#else
lpage = read_mapping_page(sb->s_bdev->bd_inode->i_mapping,
page_index, NULL);
#endif
if (IS_ERR(lpage)) {
err = PTR_ERR(lpage);
mutex_unlock(&ntfs_cb_lock);
goto read_err;
}
lock_page(lpage);
memcpy(cb_pos, page_address(lpage) + page_ofs,
vol->cluster_size);
unlock_page(lpage);
put_page(lpage);
cb_pos += vol->cluster_size;
}
/* Release the lock if we took it. */
if (rl)
up_read(&ni->runlist.lock);
/* Just a precaution. */
if (cb_pos + 2 <= cb + cb_size)
*(u16 *)cb_pos = 0;
/* Reset cb_pos back to the beginning. */
cb_pos = cb;
/* We now have both source (if present) and destination. */
ntfs_debug("Successfully read the compression block.");
/* The last page and maximum offset within it for the current cb. */
cb_max_page = (cur_page << PAGE_SHIFT) + cur_ofs + cb_size;
cb_max_ofs = cb_max_page & ~PAGE_MASK;
cb_max_page >>= PAGE_SHIFT;
/* Catch end of file inside a compression block. */
if (cb_max_page > max_page)
cb_max_page = max_page;
if (vcn == start_vcn - cb_clusters) {
/* Sparse cb, zero out page range overlapping the cb. */
ntfs_debug("Found sparse compression block.");
/* We can sleep from now on, so we drop lock. */
mutex_unlock(&ntfs_cb_lock);
if (cb_max_ofs)
cb_max_page--;
for (; cur_page < cb_max_page; cur_page++) {
page = pages[cur_page];
if (page) {
if (likely(!cur_ofs))
clear_page(page_address(page));
else
memset(page_address(page) + cur_ofs, 0,
PAGE_SIZE -
cur_ofs);
flush_dcache_page(page);
kunmap_local(page_address(page));
SetPageUptodate(page);
unlock_page(page);
if (cur_page == xpage)
xpage_done = 1;
else
put_page(page);
pages[cur_page] = NULL;
}
cb_pos += PAGE_SIZE - cur_ofs;
cur_ofs = 0;
if (cb_pos >= cb_end)
break;
}
/* If we have a partial final page, deal with it now. */
if (cb_max_ofs && cb_pos < cb_end) {
page = pages[cur_page];
if (page)
memset(page_address(page) + cur_ofs, 0,
cb_max_ofs - cur_ofs);
/*
* No need to update cb_pos at this stage:
* cb_pos += cb_max_ofs - cur_ofs;
*/
cur_ofs = cb_max_ofs;
}
} else if (vcn == start_vcn) {
/* We can't sleep so we need two stages. */
unsigned int cur2_page = cur_page;
unsigned int cur_ofs2 = cur_ofs;
u8 *cb_pos2 = cb_pos;
ntfs_debug("Found uncompressed compression block.");
/* Uncompressed cb, copy it to the destination pages. */
if (cb_max_ofs)
cb_max_page--;
/* First stage: copy data into destination pages. */
for (; cur_page < cb_max_page; cur_page++) {
page = pages[cur_page];
if (page)
memcpy(page_address(page) + cur_ofs, cb_pos,
PAGE_SIZE - cur_ofs);
cb_pos += PAGE_SIZE - cur_ofs;
cur_ofs = 0;
if (cb_pos >= cb_end)
break;
}
/* If we have a partial final page, deal with it now. */
if (cb_max_ofs && cb_pos < cb_end) {
page = pages[cur_page];
if (page)
memcpy(page_address(page) + cur_ofs, cb_pos,
cb_max_ofs - cur_ofs);
cb_pos += cb_max_ofs - cur_ofs;
cur_ofs = cb_max_ofs;
}
/* We can sleep from now on, so drop lock. */
mutex_unlock(&ntfs_cb_lock);
/* Second stage: finalize pages. */
for (; cur2_page < cb_max_page; cur2_page++) {
page = pages[cur2_page];
if (page) {
/*
* If we are outside the initialized size, zero
* the out of bounds page range.
*/
handle_bounds_compressed_page(page, i_size,
initialized_size);
flush_dcache_page(page);
kunmap_local(page_address(page));
SetPageUptodate(page);
unlock_page(page);
if (cur2_page == xpage)
xpage_done = 1;
else
put_page(page);
pages[cur2_page] = NULL;
}
cb_pos2 += PAGE_SIZE - cur_ofs2;
cur_ofs2 = 0;
if (cb_pos2 >= cb_end)
break;
}
} else {
/* Compressed cb, decompress it into the destination page(s). */
unsigned int prev_cur_page = cur_page;
ntfs_debug("Found compressed compression block.");
err = ntfs_decompress(pages, completed_pages, &cur_page,
&cur_ofs, cb_max_page, cb_max_ofs, xpage,
&xpage_done, cb_pos, cb_size - (cb_pos - cb),
i_size, initialized_size);
/*
* We can sleep from now on, lock already dropped by
* ntfs_decompress().
*/
if (err) {
ntfs_error(vol->sb,
"ntfs_decompress() failed in inode 0x%lx with error code %i. Skipping this compression block.",
ni->mft_no, -err);
/* Release the unfinished pages. */
for (; prev_cur_page < cur_page; prev_cur_page++) {
page = pages[prev_cur_page];
if (page) {
flush_dcache_page(page);
kunmap_local(page_address(page));
unlock_page(page);
if (prev_cur_page != xpage)
put_page(page);
pages[prev_cur_page] = NULL;
}
}
}
}
/* Do we have more work to do? */
if (nr_cbs)
goto do_next_cb;
/* Clean up if we have any pages left. Should never happen. */
for (cur_page = 0; cur_page < max_page; cur_page++) {
page = pages[cur_page];
if (page) {
ntfs_error(vol->sb,
"Still have pages left! Terminating them with extreme prejudice. Inode 0x%lx, page index 0x%lx.",
#if LINUX_VERSION_CODE >= KERNEL_VERSION(6, 16, 0)
ni->mft_no, page->__folio_index);
#else
ni->mft_no, page->index);
#endif
flush_dcache_page(page);
kunmap_local(page_address(page));
unlock_page(page);
if (cur_page != xpage)
put_page(page);
pages[cur_page] = NULL;
}
}
/* We no longer need the list of pages. */
kfree(pages);
kfree(completed_pages);
/* If we have completed the requested page, we return success. */
if (likely(xpage_done))
return 0;
ntfs_debug("Failed. Returning error code %s.", err == -EOVERFLOW ?
"EOVERFLOW" : (!err ? "EIO" : "unknown error"));
return err < 0 ? err : -EIO;
map_rl_err:
ntfs_error(vol->sb, "ntfs_map_runlist() failed. Cannot read compression block.");
goto err_out;
rl_err:
up_read(&ni->runlist.lock);
ntfs_error(vol->sb, "ntfs_rl_vcn_to_lcn() failed. Cannot read compression block.");
goto err_out;
read_err:
up_read(&ni->runlist.lock);
ntfs_error(vol->sb, "IO error while reading compressed data.");
err_out:
for (i = cur_page; i < max_page; i++) {
page = pages[i];
if (page) {
flush_dcache_page(page);
kunmap_local(page_address(page));
unlock_page(page);
if (i != xpage)
put_page(page);
}
}
kfree(pages);
kfree(completed_pages);
return -EIO;
}
/*
* Match length at or above which ntfs_best_match() will stop searching for
* longer matches.
*/
#define NICE_MATCH_LEN 18
/*
* Maximum number of potential matches that ntfs_best_match() will consider at
* each position.
*/
#define MAX_SEARCH_DEPTH 24
/* log base 2 of the number of entries in the hash table for match-finding. */
#define HASH_SHIFT 14
/*
* Constant for the multiplicative hash function. These hashing constants
* are used solely for the match-finding algorithm during compression.
* They are NOT part of the on-disk format. The decompressor does not
* utilize this hash.
*/
#define HASH_MULTIPLIER 0x1E35A7BD
struct compress_context {
const unsigned char *inbuf;
int bufsize;
int size;
int rel;
int mxsz;
s16 head[1 << HASH_SHIFT];
s16 prev[NTFS_SB_SIZE];
};
/*
* Hash the next 3-byte sequence in the input buffer
*/
static inline unsigned int ntfs_hash(const u8 *p)
{
u32 str;
u32 hash;
/*
* Unaligned access allowed, and little endian CPU.
* Callers ensure that at least 4 (not 3) bytes are remaining.
*/
str = *(const u32 *)p & 0xFFFFFF;
hash = str * HASH_MULTIPLIER;
/* High bits are more random than the low bits. */
return hash >> (32 - HASH_SHIFT);
}
/*
* Search for the longest sequence matching current position
*
* A hash table, each entry of which points to a chain of sequence
* positions sharing the corresponding hash code, is maintained to speed up
* searching for matches. To maintain the hash table, either
* ntfs_best_match() or ntfs_skip_position() has to be called for each
* consecutive position.
*
* This function is heavily used; it has to be optimized carefully.
*
* This function sets pctx->size and pctx->rel to the length and offset,
* respectively, of the longest match found.
*
* The minimum match length is assumed to be 3, and the maximum match
* length is assumed to be pctx->mxsz. If this function produces
* pctx->size < 3, then no match was found.
*
* Note: for the following reasons, this function is not guaranteed to find
* *the* longest match up to pctx->mxsz:
*
* (1) If this function finds a match of NICE_MATCH_LEN bytes or greater,
* it ends early because a match this long is good enough and it's not
* worth spending more time searching.
*
* (2) If this function considers MAX_SEARCH_DEPTH matches with a single
* position, it ends early and returns the longest match found so far.
* This saves a lot of time on degenerate inputs.
*/
static void ntfs_best_match(struct compress_context *pctx, const int i,
int best_len)
{
const u8 * const inbuf = pctx->inbuf;
const u8 * const strptr = &inbuf[i]; /* String we're matching against */
s16 * const prev = pctx->prev;
const int max_len = min(pctx->bufsize - i, pctx->mxsz);