2 * Copyright (c) 2020 Ori Bernstein
3 * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
5 * Permission to use, copy, modify, and distribute this software for any
6 * purpose with or without fee is hereby granted, provided that the above
7 * copyright notice and this permission notice appear in all copies.
9 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18 #include <sys/types.h>
19 #include <sys/queue.h>
30 #include "got_error.h"
32 #include "got_lib_deltify.h"
33 #include "murmurhash2.h"
36 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
40 * The algorihm used here is FastCDC (Fast Content-Defined Chunking)
41 * https://www.usenix.org/conference/atc16/technical-sessions/presentation/xia
44 static uint32_t geartab[256] = {
45 0x67ed26b7, 0x32da500c, 0x53d0fee0, 0xce387dc7, 0xcd406d90, 0x2e83a4d4,
46 0x9fc9a38d, 0xb67259dc, 0xca6b1722, 0x6d2ea08c, 0x235cea2e, 0x3149bb5f,
47 0x1beda787, 0x2a6b77d5, 0x2f22d9ac, 0x91fc0544, 0xe413acfa, 0x5a30ff7a,
48 0xad6fdde0, 0x444fd0f5, 0x7ad87864, 0x58c5ff05, 0x8d2ec336, 0x2371f853,
49 0x550f8572, 0x6aa448dd, 0x7c9ddbcf, 0x95221e14, 0x2a82ec33, 0xcbec5a78,
50 0xc6795a0d, 0x243995b7, 0x1c909a2f, 0x4fded51c, 0x635d334b, 0x0e2b9999,
51 0x2702968d, 0x856de1d5, 0x3325d60e, 0xeb6a7502, 0xec2a9844, 0x0905835a,
52 0xa1820375, 0xa4be5cab, 0x96a6c058, 0x2c2ccd70, 0xba40fce3, 0xd794c46b,
53 0x8fbae83e, 0xc3aa7899, 0x3d3ff8ed, 0xa0d42b5b, 0x571c0c97, 0xd2811516,
54 0xf7e7b96c, 0x4fd2fcbd, 0xe2fdec94, 0x282cc436, 0x78e8e95c, 0x80a3b613,
55 0xcfbee20c, 0xd4a32d1c, 0x2a12ff13, 0x6af82936, 0xe5630258, 0x8efa6a98,
56 0x294fb2d1, 0xdeb57086, 0x5f0fddb3, 0xeceda7ce, 0x4c87305f, 0x3a6d3307,
57 0xe22d2942, 0x9d060217, 0x1e42ed02, 0xb6f63b52, 0x4367f39f, 0x055cf262,
58 0x03a461b2, 0x5ef9e382, 0x386bc03a, 0x2a1e79c7, 0xf1a0058b, 0xd4d2dea9,
59 0x56baf37d, 0x5daff6cc, 0xf03a951d, 0xaef7de45, 0xa8f4581e, 0x3960b555,
60 0xffbfff6d, 0xbe702a23, 0x8f5b6d6f, 0x061739fb, 0x98696f47, 0x3fd596d4,
61 0x151eac6b, 0xa9fcc4f5, 0x69181a12, 0x3ac5a107, 0xb5198fe7, 0x96bcb1da,
62 0x1b5ddf8e, 0xc757d650, 0x65865c3a, 0x8fc0a41a, 0x87435536, 0x99eda6f2,
63 0x41874794, 0x29cff4e8, 0xb70efd9a, 0x3103f6e7, 0x84d2453b, 0x15a450bd,
64 0x74f49af1, 0x60f664b1, 0xa1c86935, 0xfdafbce1, 0xe36353e3, 0x5d9ba739,
65 0xbc0559ba, 0x708b0054, 0xd41d808c, 0xb2f31723, 0x9027c41f, 0xf136d165,
66 0xb5374b12, 0x9420a6ac, 0x273958b6, 0xe6c2fad0, 0xebdc1f21, 0xfb33af8b,
67 0xc71c25cd, 0xe9a2d8e5, 0xbeb38a50, 0xbceb7cc2, 0x4e4e73f0, 0xcd6c251d,
68 0xde4c032c, 0x4b04ac30, 0x725b8b21, 0x4eb8c33b, 0x20d07b75, 0x0567aa63,
69 0xb56b2bb7, 0xc1f5fd3a, 0xcafd35ca, 0x470dd4da, 0xfe4f94cd, 0xfb8de424,
70 0xe8dbcf40, 0xfe50a37a, 0x62db5b5d, 0xf32f4ab6, 0x2c4a8a51, 0x18473dc0,
71 0xfe0cbb6e, 0xfe399efd, 0xdf34ecc9, 0x6ccd5055, 0x46097073, 0x139135c2,
72 0x721c76f6, 0x1c6a94b4, 0x6eee014d, 0x8a508e02, 0x3da538f5, 0x280d394f,
73 0x5248a0c4, 0x3ce94c6c, 0x9a71ad3a, 0x8493dd05, 0xe43f0ab6, 0x18e4ed42,
74 0x6c5c0e09, 0x42b06ec9, 0x8d330343, 0xa45b6f59, 0x2a573c0c, 0xd7fd3de6,
75 0xeedeab68, 0x5c84dafc, 0xbbd1b1a8, 0xa3ce1ad1, 0x85b70bed, 0xb6add07f,
76 0xa531309c, 0x8f8ab852, 0x564de332, 0xeac9ed0c, 0x73da402c, 0x3ec52761,
77 0x43af2f4d, 0xd6ff45c8, 0x4c367462, 0xd553bd6a, 0x44724855, 0x3b2aa728,
78 0x56e5eb65, 0xeaf16173, 0x33fa42ff, 0xd714bb5d, 0xfbd0a3b9, 0xaf517134,
79 0x9416c8cd, 0x534cf94f, 0x548947c2, 0x34193569, 0x32f4389a, 0xfe7028bc,
80 0xed73b1ed, 0x9db95770, 0x468e3922, 0x0440c3cd, 0x60059a62, 0x33504562,
81 0x2b229fbd, 0x5174dca5, 0xf7028752, 0xd63c6aa8, 0x31276f38, 0x0646721c,
82 0xb0191da8, 0xe00e6de0, 0x9eac1a6e, 0x9f7628a5, 0xed6c06ea, 0x0bb8af15,
83 0xf119fb12, 0x38693c1c, 0x732bc0fe, 0x84953275, 0xb82ec888, 0x33a4f1b3,
84 0x3099835e, 0x028a8782, 0x5fdd51d7, 0xc6c717b3, 0xb06caf71, 0x17c8c111,
85 0x61bad754, 0x9fd03061, 0xe09df1af, 0x3bc9eb73, 0x85878413, 0x9889aaf2,
86 0x3f5a9e46, 0x42c9f01f, 0x9984a4f4, 0xd5de43cc, 0xd294daed, 0xbecba2d2,
87 0xf1f6e72c, 0x5551128a, 0x83af87e2, 0x6f0342ba,
91 hashblk(const unsigned char *p, off_t n)
93 return murmurhash2(p, n, 0x1d7c5ac3);
96 static const struct got_error *
97 addblk(struct got_delta_table *dt, FILE *f, off_t file_offset0, off_t len,
98 off_t offset, uint32_t h)
100 const struct got_error *err = NULL;
102 uint8_t buf[GOT_DELTIFY_MAXCHUNK];
103 uint8_t buf2[GOT_DELTIFY_MAXCHUNK];
110 while (dt->blocks[i].len != 0) {
112 * Avoid adding duplicate blocks.
113 * NB: A matching hash is insufficient for detecting equality.
114 * The hash can only detect inequality.
116 if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
118 if (fseeko(f, file_offset0 + offset, SEEK_SET) == -1)
119 return got_error_from_errno("fseeko");
120 r = fread(buf, 1, len, f);
124 return got_ferror(f, GOT_ERR_IO);
127 if (fseeko(f, file_offset0 + dt->blocks[i].offset,
129 return got_error_from_errno("fseeko");
130 if (fread(buf2, 1, len, f) != len)
131 return got_ferror(f, GOT_ERR_IO);
132 if (memcmp(buf, buf2, len) == 0)
136 i = (i + 1) % dt->nalloc;
138 assert(dt->blocks[i].len == 0);
139 dt->blocks[i].len = len;
140 dt->blocks[i].offset = offset;
141 dt->blocks[i].hash = h;
143 if (dt->nalloc < dt->nblocks + 64) {
144 struct got_delta_block *db;
145 size_t old_size = dt->nalloc;
147 dt->blocks = calloc(dt->nalloc + 64,
148 sizeof(struct got_delta_block));
149 if (dt->blocks == NULL) {
150 err = got_error_from_errno("calloc");
156 * Recompute all block positions. Hash-based indices of blocks
157 * in the array depend on the allocated length of the array.
160 for (i = 0; i < old_size; i++) {
163 err = addblk(dt, f, file_offset0, db[i].len,
164 db[i].offset, db[i].hash);
174 static const struct got_error *
175 addblk_mem(struct got_delta_table *dt, uint8_t *data, off_t file_offset0,
176 off_t len, off_t offset, uint32_t h)
178 const struct got_error *err = NULL;
187 while (dt->blocks[i].len != 0) {
189 * Avoid adding duplicate blocks.
190 * NB: A matching hash is insufficient for detecting equality.
191 * The hash can only detect inequality.
193 if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
194 block1 = data + file_offset0 + dt->blocks[i].offset;
195 block2 = data + file_offset0 + offset;
196 if (memcmp(block1, block2, len) == 0)
200 i = (i + 1) % dt->nalloc;
202 assert(dt->blocks[i].len == 0);
203 dt->blocks[i].len = len;
204 dt->blocks[i].offset = offset;
205 dt->blocks[i].hash = h;
207 if (dt->nalloc < dt->nblocks + 64) {
208 struct got_delta_block *db;
209 size_t old_size = dt->nalloc;
211 dt->blocks = calloc(dt->nalloc + 64,
212 sizeof(struct got_delta_block));
213 if (dt->blocks == NULL) {
214 err = got_error_from_errno("calloc");
220 * Recompute all block positions. Hash-based indices of blocks
221 * in the array depend on the allocated length of the array.
224 for (i = 0; i < old_size; i++) {
227 err = addblk_mem(dt, data, file_offset0, db[i].len,
228 db[i].offset, db[i].hash);
238 static const struct got_error *
239 lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
240 unsigned char *p, off_t len, FILE *basefile, off_t basefile_offset0)
244 uint8_t buf[GOT_DELTIFY_MAXCHUNK];
250 for (i = h % dt->nalloc; dt->blocks[i].len != 0;
251 i = (i + 1) % dt->nalloc) {
252 if (dt->blocks[i].hash != h ||
253 dt->blocks[i].len != len)
255 if (fseeko(basefile, basefile_offset0 + dt->blocks[i].offset,
257 return got_error_from_errno("fseeko");
258 r = fread(buf, 1, len, basefile);
260 return got_ferror(basefile, GOT_ERR_IO);
261 if (memcmp(p, buf, len) == 0) {
262 *block = &dt->blocks[i];
269 static const struct got_error *
270 lookupblk_mem(struct got_delta_block **block, struct got_delta_table *dt,
271 unsigned char *p, off_t len, uint8_t *basedata, off_t basefile_offset0)
280 for (i = h % dt->nalloc; dt->blocks[i].len != 0;
281 i = (i + 1) % dt->nalloc) {
282 if (dt->blocks[i].hash != h ||
283 dt->blocks[i].len != len)
285 b = basedata + basefile_offset0 + dt->blocks[i].offset;
286 if (memcmp(p, b, len) == 0) {
287 *block = &dt->blocks[i];
294 static const struct got_error *
295 nextblk(uint8_t *buf, off_t *blocklen, FILE *f)
298 const unsigned char *p;
300 off_t pos = ftello(f);
304 r = fread(buf, 1, GOT_DELTIFY_MAXCHUNK, f);
305 if (r == 0 && ferror(f))
306 return got_ferror(f, GOT_ERR_IO);
307 if (r < GOT_DELTIFY_MINCHUNK)
308 return NULL; /* no more delta-worthy blocks left */
310 /* Got a deltifiable block. Find the split-point where it ends. */
311 p = buf + GOT_DELTIFY_MINCHUNK;
313 while (p != buf + r) {
314 gh = (gh << 1) + geartab[*p++];
315 if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
319 *blocklen = (p - buf);
320 if (fseeko(f, pos + *blocklen, SEEK_SET) == -1)
321 return got_error_from_errno("fseeko");
326 static const struct got_error *
327 nextblk_mem(off_t *blocklen, uint8_t *data, off_t fileoffset, off_t filesize)
330 const unsigned char *p;
334 if (fileoffset >= filesize ||
335 filesize - fileoffset < GOT_DELTIFY_MINCHUNK)
336 return NULL; /* no more delta-worthy blocks left */
338 /* Got a deltifiable block. Find the split-point where it ends. */
339 p = data + fileoffset + GOT_DELTIFY_MINCHUNK;
341 while (p != data + MIN(fileoffset + GOT_DELTIFY_MAXCHUNK, filesize)) {
342 gh = (gh << 1) + geartab[*p++];
343 if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
347 *blocklen = (p - (data + fileoffset));
351 const struct got_error *
352 got_deltify_init(struct got_delta_table **dt, FILE *f, off_t fileoffset,
355 const struct got_error *err = NULL;
357 const off_t offset0 = fileoffset;
359 *dt = calloc(1, sizeof(**dt));
361 return got_error_from_errno("calloc");
365 (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
366 if ((*dt)->blocks == NULL) {
367 err = got_error_from_errno("calloc");
371 if (fseeko(f, fileoffset, SEEK_SET) == -1)
372 return got_error_from_errno("fseeko");
374 while (fileoffset < filesize) {
375 uint8_t buf[GOT_DELTIFY_MAXCHUNK];
377 err = nextblk(buf, &blocklen, f);
382 h = hashblk(buf, blocklen);
383 err = addblk(*dt, f, offset0, blocklen,
384 fileoffset - offset0, h);
387 fileoffset += blocklen;
388 if (fseeko(f, fileoffset, SEEK_SET) == -1)
389 return got_error_from_errno("fseeko");
401 const struct got_error *
402 got_deltify_init_mem(struct got_delta_table **dt, uint8_t *data,
403 off_t fileoffset, off_t filesize)
405 const struct got_error *err = NULL;
407 const off_t offset0 = fileoffset;
409 *dt = calloc(1, sizeof(**dt));
411 return got_error_from_errno("calloc");
415 (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
416 if ((*dt)->blocks == NULL) {
417 err = got_error_from_errno("calloc");
421 while (fileoffset < filesize) {
423 err = nextblk_mem(&blocklen, data, fileoffset, filesize);
428 h = hashblk(data + fileoffset, blocklen);
429 err = addblk_mem(*dt, data, offset0, blocklen,
430 fileoffset - offset0, h);
433 fileoffset += blocklen;
446 got_deltify_free(struct got_delta_table *dt)
454 static const struct got_error *
455 emitdelta(struct got_delta_instruction **deltas, size_t *nalloc, int *ndeltas,
456 const size_t alloc_chunk_size, int copy, off_t offset, off_t len)
458 struct got_delta_instruction *d, *p;
460 if (*nalloc < *ndeltas + alloc_chunk_size) {
461 p = reallocarray(*deltas, *nalloc + alloc_chunk_size,
462 sizeof(struct got_delta_instruction));
464 return got_error_from_errno("reallocarray");
466 *nalloc += alloc_chunk_size;
469 d = &(*deltas)[*ndeltas - 1];
476 static const struct got_error *
477 stretchblk(FILE *basefile, off_t base_offset0, struct got_delta_block *block,
478 FILE *f, off_t filesize, off_t *blocklen)
480 uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
484 if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
486 return got_error_from_errno("fseeko");
488 while (buf_equal && *blocklen < (1 << 24) - 1) {
489 base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
491 if (ferror(basefile))
492 return got_ferror(basefile, GOT_ERR_IO);
495 r = fread(buf, 1, sizeof(buf), f);
498 return got_ferror(f, GOT_ERR_IO);
501 for (i = 0; i < MIN(base_r, r); i++) {
502 if (buf[i] != basebuf[i]) {
513 static const struct got_error *
514 stretchblk_file_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
515 struct got_delta_block *block, FILE *f, off_t filesize, off_t *blocklen)
517 uint8_t buf[GOT_DELTIFY_MAXCHUNK];
520 off_t base_offset = base_offset0 + block->offset + *blocklen;
522 if (base_offset > basefile_size) {
523 return got_error_fmt(GOT_ERR_RANGE,
524 "read beyond the size of delta base at offset %llu",
528 while (buf_equal && *blocklen < (1 << 24) - 1) {
529 if (base_offset + *blocklen >= basefile_size)
531 r = fread(buf, 1, sizeof(buf), f);
534 return got_ferror(f, GOT_ERR_IO);
537 for (i = 0; i < MIN(basefile_size - base_offset, r); i++) {
538 if (buf[i] != *(basedata + base_offset + i)) {
549 static const struct got_error *
550 stretchblk_mem_file(FILE *basefile, off_t base_offset0,
551 struct got_delta_block *block, uint8_t *data, off_t fileoffset,
552 off_t filesize, off_t *blocklen)
554 uint8_t basebuf[GOT_DELTIFY_MAXCHUNK];
558 if (fileoffset > filesize) {
559 return got_error_fmt(GOT_ERR_RANGE,
560 "read beyond the size of deltify file at offset %llu",
564 if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
566 return got_error_from_errno("fseeko");
568 while (buf_equal && *blocklen < (1 << 24) - 1) {
569 if (fileoffset + *blocklen >= filesize)
571 base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
573 if (ferror(basefile))
574 return got_ferror(basefile, GOT_ERR_IO);
577 for (i = 0; i < MIN(base_r, filesize - fileoffset); i++) {
578 if (*(data + fileoffset + i) != basebuf[i]) {
589 static const struct got_error *
590 stretchblk_mem_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
591 struct got_delta_block *block, uint8_t *data, off_t fileoffset,
592 off_t filesize, off_t *blocklen)
595 off_t base_offset = base_offset0 + block->offset + *blocklen;
598 if (base_offset > basefile_size) {
599 return got_error_fmt(GOT_ERR_RANGE,
600 "read beyond the size of delta base at offset %llu",
604 if (fileoffset > filesize) {
605 return got_error_fmt(GOT_ERR_RANGE,
606 "read beyond the size of deltify file at offset %llu",
610 p = data + fileoffset;
611 q = basedata + base_offset;
612 maxlen = MIN(basefile_size - base_offset, filesize - fileoffset);
613 for (i = 0; i < maxlen && *blocklen < (1 << 24) - 1; i++) {
622 const struct got_error *
623 got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
624 FILE *f, off_t fileoffset, off_t filesize,
625 struct got_delta_table *dt, FILE *basefile,
626 off_t basefile_offset0, off_t basefile_size)
628 const struct got_error *err = NULL;
629 const off_t offset0 = fileoffset;
631 const size_t alloc_chunk_size = 64;
637 * offset0 indicates where data to be deltified begins.
638 * For example, we want to avoid deltifying a Git object header at
639 * the beginning of the file.
641 if (fseeko(f, offset0, SEEK_SET) == -1)
642 return got_error_from_errno("fseeko");
644 *deltas = reallocarray(NULL, alloc_chunk_size,
645 sizeof(struct got_delta_instruction));
647 return got_error_from_errno("reallocarray");
648 nalloc = alloc_chunk_size;
650 while (fileoffset < filesize) {
651 uint8_t buf[GOT_DELTIFY_MAXCHUNK];
653 struct got_delta_block *block;
654 err = nextblk(buf, &blocklen, f);
658 /* Source remainder from the file itself. */
659 if (fileoffset < filesize) {
660 err = emitdelta(deltas, &nalloc, ndeltas,
661 alloc_chunk_size, 0, fileoffset - offset0,
662 filesize - fileoffset);
666 err = lookupblk(&block, dt, buf, blocklen, basefile,
672 * We have found a matching block in the delta base.
673 * Attempt to stretch the block as far as possible and
674 * generate a copy instruction.
676 err = stretchblk(basefile, basefile_offset0, block,
677 f, filesize, &blocklen);
680 err = emitdelta(deltas, &nalloc, ndeltas,
681 alloc_chunk_size, 1, block->offset, blocklen);
687 * This block needs to be sourced from the file itself.
689 err = emitdelta(deltas, &nalloc, ndeltas,
690 alloc_chunk_size, 0, fileoffset - offset0, blocklen);
694 fileoffset += blocklen;
695 if (fseeko(f, fileoffset, SEEK_SET) == -1) {
696 err = got_error_from_errno("fseeko");
709 const struct got_error *
710 got_deltify_file_mem(struct got_delta_instruction **deltas, int *ndeltas,
711 FILE *f, off_t fileoffset, off_t filesize,
712 struct got_delta_table *dt, uint8_t *basedata,
713 off_t basefile_offset0, off_t basefile_size)
715 const struct got_error *err = NULL;
716 const off_t offset0 = fileoffset;
718 const size_t alloc_chunk_size = 64;
724 * offset0 indicates where data to be deltified begins.
725 * For example, we want to avoid deltifying a Git object header at
726 * the beginning of the file.
728 if (fseeko(f, offset0, SEEK_SET) == -1)
729 return got_error_from_errno("fseeko");
731 *deltas = reallocarray(NULL, alloc_chunk_size,
732 sizeof(struct got_delta_instruction));
734 return got_error_from_errno("reallocarray");
735 nalloc = alloc_chunk_size;
737 while (fileoffset < filesize) {
738 uint8_t buf[GOT_DELTIFY_MAXCHUNK];
740 struct got_delta_block *block;
741 err = nextblk(buf, &blocklen, f);
745 /* Source remainder from the file itself. */
746 if (fileoffset < filesize) {
747 err = emitdelta(deltas, &nalloc, ndeltas,
748 alloc_chunk_size, 0, fileoffset - offset0,
749 filesize - fileoffset);
753 err = lookupblk_mem(&block, dt, buf, blocklen, basedata,
759 * We have found a matching block in the delta base.
760 * Attempt to stretch the block as far as possible and
761 * generate a copy instruction.
763 err = stretchblk_file_mem(basedata, basefile_offset0,
764 basefile_size, block, f, filesize, &blocklen);
767 err = emitdelta(deltas, &nalloc, ndeltas,
768 alloc_chunk_size, 1, block->offset, blocklen);
774 * This block needs to be sourced from the file itself.
776 err = emitdelta(deltas, &nalloc, ndeltas,
777 alloc_chunk_size, 0, fileoffset - offset0, blocklen);
781 fileoffset += blocklen;
782 if (fseeko(f, fileoffset, SEEK_SET) == -1) {
783 err = got_error_from_errno("fseeko");
796 const struct got_error *
797 got_deltify_mem_file(struct got_delta_instruction **deltas, int *ndeltas,
798 uint8_t *data, off_t fileoffset, off_t filesize,
799 struct got_delta_table *dt, FILE *basefile,
800 off_t basefile_offset0, off_t basefile_size)
802 const struct got_error *err = NULL;
803 const off_t offset0 = fileoffset;
805 const size_t alloc_chunk_size = 64;
810 *deltas = reallocarray(NULL, alloc_chunk_size,
811 sizeof(struct got_delta_instruction));
813 return got_error_from_errno("reallocarray");
814 nalloc = alloc_chunk_size;
816 while (fileoffset < filesize) {
818 struct got_delta_block *block;
819 err = nextblk_mem(&blocklen, data, fileoffset, filesize);
823 /* Source remainder from the file itself. */
824 if (fileoffset < filesize) {
825 err = emitdelta(deltas, &nalloc, ndeltas,
826 alloc_chunk_size, 0, fileoffset - offset0,
827 filesize - fileoffset);
831 err = lookupblk(&block, dt, data + fileoffset, blocklen,
832 basefile, basefile_offset0);
837 * We have found a matching block in the delta base.
838 * Attempt to stretch the block as far as possible and
839 * generate a copy instruction.
841 err = stretchblk_mem_file(basefile, basefile_offset0,
842 block, data, fileoffset + blocklen, filesize,
846 err = emitdelta(deltas, &nalloc, ndeltas,
847 alloc_chunk_size, 1, block->offset, blocklen);
853 * This block needs to be sourced from the file itself.
855 err = emitdelta(deltas, &nalloc, ndeltas,
856 alloc_chunk_size, 0, fileoffset - offset0, blocklen);
860 fileoffset += blocklen;
871 const struct got_error *
872 got_deltify_mem_mem(struct got_delta_instruction **deltas, int *ndeltas,
873 uint8_t *data, off_t fileoffset, off_t filesize,
874 struct got_delta_table *dt, uint8_t *basedata,
875 off_t basefile_offset0, off_t basefile_size)
877 const struct got_error *err = NULL;
878 const off_t offset0 = fileoffset;
880 const size_t alloc_chunk_size = 64;
885 *deltas = reallocarray(NULL, alloc_chunk_size,
886 sizeof(struct got_delta_instruction));
888 return got_error_from_errno("reallocarray");
889 nalloc = alloc_chunk_size;
891 while (fileoffset < filesize) {
893 struct got_delta_block *block;
894 err = nextblk_mem(&blocklen, data, fileoffset, filesize);
898 /* Source remainder from the file itself. */
899 if (fileoffset < filesize) {
900 err = emitdelta(deltas, &nalloc, ndeltas,
901 alloc_chunk_size, 0, fileoffset - offset0,
902 filesize - fileoffset);
906 err = lookupblk_mem(&block, dt, data + fileoffset, blocklen,
907 basedata, basefile_offset0);
912 * We have found a matching block in the delta base.
913 * Attempt to stretch the block as far as possible and
914 * generate a copy instruction.
916 err = stretchblk_mem_mem(basedata, basefile_offset0,
917 basefile_size, block, data, fileoffset + blocklen,
918 filesize, &blocklen);
921 err = emitdelta(deltas, &nalloc, ndeltas,
922 alloc_chunk_size, 1, block->offset, blocklen);
928 * This block needs to be sourced from the file itself.
930 err = emitdelta(deltas, &nalloc, ndeltas,
931 alloc_chunk_size, 0, fileoffset - offset0, blocklen);
935 fileoffset += blocklen;