2 56e7a15f 2021-03-10 stsp * Copyright (c) 2020 Ori Bernstein
3 69aa0e90 2021-03-10 stsp * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
5 2a0e67f7 2021-03-10 stsp * Permission to use, copy, modify, and distribute this software for any
6 2a0e67f7 2021-03-10 stsp * purpose with or without fee is hereby granted, provided that the above
7 2a0e67f7 2021-03-10 stsp * copyright notice and this permission notice appear in all copies.
9 2a0e67f7 2021-03-10 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 2a0e67f7 2021-03-10 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 2a0e67f7 2021-03-10 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 2a0e67f7 2021-03-10 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 2a0e67f7 2021-03-10 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 2a0e67f7 2021-03-10 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 2a0e67f7 2021-03-10 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18 f364801d 2021-03-10 stsp #include <sys/types.h>
19 f364801d 2021-03-10 stsp #include <sys/queue.h>
21 f364801d 2021-03-10 stsp #include <assert.h>
22 69aa0e90 2021-03-10 stsp #include <errno.h>
23 f364801d 2021-03-10 stsp #include <limits.h>
24 f364801d 2021-03-10 stsp #include <stdint.h>
25 f364801d 2021-03-10 stsp #include <stdio.h>
26 f364801d 2021-03-10 stsp #include <stdlib.h>
27 f364801d 2021-03-10 stsp #include <string.h>
28 f364801d 2021-03-10 stsp #include <sha1.h>
30 f364801d 2021-03-10 stsp #include "got_error.h"
32 69aa0e90 2021-03-10 stsp #include "got_lib_deltify.h"
35 69aa0e90 2021-03-10 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
38 f364801d 2021-03-10 stsp static uint32_t geartab[] = {
39 69aa0e90 2021-03-10 stsp 0x67ed26b7, 0x32da500c, 0x53d0fee0, 0xce387dc7, 0xcd406d90, 0x2e83a4d4,
40 69aa0e90 2021-03-10 stsp 0x9fc9a38d, 0xb67259dc, 0xca6b1722, 0x6d2ea08c, 0x235cea2e, 0x3149bb5f,
41 69aa0e90 2021-03-10 stsp 0x1beda787, 0x2a6b77d5, 0x2f22d9ac, 0x91fc0544, 0xe413acfa, 0x5a30ff7a,
42 69aa0e90 2021-03-10 stsp 0xad6fdde0, 0x444fd0f5, 0x7ad87864, 0x58c5ff05, 0x8d2ec336, 0x2371f853,
43 69aa0e90 2021-03-10 stsp 0x550f8572, 0x6aa448dd, 0x7c9ddbcf, 0x95221e14, 0x2a82ec33, 0xcbec5a78,
44 69aa0e90 2021-03-10 stsp 0xc6795a0d, 0x243995b7, 0x1c909a2f, 0x4fded51c, 0x635d334b, 0x0e2b9999,
45 69aa0e90 2021-03-10 stsp 0x2702968d, 0x856de1d5, 0x3325d60e, 0xeb6a7502, 0xec2a9844, 0x0905835a,
46 69aa0e90 2021-03-10 stsp 0xa1820375, 0xa4be5cab, 0x96a6c058, 0x2c2ccd70, 0xba40fce3, 0xd794c46b,
47 69aa0e90 2021-03-10 stsp 0x8fbae83e, 0xc3aa7899, 0x3d3ff8ed, 0xa0d42b5b, 0x571c0c97, 0xd2811516,
48 69aa0e90 2021-03-10 stsp 0xf7e7b96c, 0x4fd2fcbd, 0xe2fdec94, 0x282cc436, 0x78e8e95c, 0x80a3b613,
49 69aa0e90 2021-03-10 stsp 0xcfbee20c, 0xd4a32d1c, 0x2a12ff13, 0x6af82936, 0xe5630258, 0x8efa6a98,
50 69aa0e90 2021-03-10 stsp 0x294fb2d1, 0xdeb57086, 0x5f0fddb3, 0xeceda7ce, 0x4c87305f, 0x3a6d3307,
51 69aa0e90 2021-03-10 stsp 0xe22d2942, 0x9d060217, 0x1e42ed02, 0xb6f63b52, 0x4367f39f, 0x055cf262,
52 69aa0e90 2021-03-10 stsp 0x03a461b2, 0x5ef9e382, 0x386bc03a, 0x2a1e79c7, 0xf1a0058b, 0xd4d2dea9,
53 69aa0e90 2021-03-10 stsp 0x56baf37d, 0x5daff6cc, 0xf03a951d, 0xaef7de45, 0xa8f4581e, 0x3960b555,
54 69aa0e90 2021-03-10 stsp 0xffbfff6d, 0xbe702a23, 0x8f5b6d6f, 0x061739fb, 0x98696f47, 0x3fd596d4,
55 69aa0e90 2021-03-10 stsp 0x151eac6b, 0xa9fcc4f5, 0x69181a12, 0x3ac5a107, 0xb5198fe7, 0x96bcb1da,
56 69aa0e90 2021-03-10 stsp 0x1b5ddf8e, 0xc757d650, 0x65865c3a, 0x8fc0a41a, 0x87435536, 0x99eda6f2,
57 69aa0e90 2021-03-10 stsp 0x41874794, 0x29cff4e8, 0xb70efd9a, 0x3103f6e7, 0x84d2453b, 0x15a450bd,
58 69aa0e90 2021-03-10 stsp 0x74f49af1, 0x60f664b1, 0xa1c86935, 0xfdafbce1, 0xe36353e3, 0x5d9ba739,
59 69aa0e90 2021-03-10 stsp 0xbc0559ba, 0x708b0054, 0xd41d808c, 0xb2f31723, 0x9027c41f, 0xf136d165,
60 69aa0e90 2021-03-10 stsp 0xb5374b12, 0x9420a6ac, 0x273958b6, 0xe6c2fad0, 0xebdc1f21, 0xfb33af8b,
61 69aa0e90 2021-03-10 stsp 0xc71c25cd, 0xe9a2d8e5, 0xbeb38a50, 0xbceb7cc2, 0x4e4e73f0, 0xcd6c251d,
62 69aa0e90 2021-03-10 stsp 0xde4c032c, 0x4b04ac30, 0x725b8b21, 0x4eb8c33b, 0x20d07b75, 0x0567aa63,
63 69aa0e90 2021-03-10 stsp 0xb56b2bb7, 0xc1f5fd3a, 0xcafd35ca, 0x470dd4da, 0xfe4f94cd, 0xfb8de424,
64 69aa0e90 2021-03-10 stsp 0xe8dbcf40, 0xfe50a37a, 0x62db5b5d, 0xf32f4ab6, 0x2c4a8a51, 0x18473dc0,
65 69aa0e90 2021-03-10 stsp 0xfe0cbb6e, 0xfe399efd, 0xdf34ecc9, 0x6ccd5055, 0x46097073, 0x139135c2,
66 69aa0e90 2021-03-10 stsp 0x721c76f6, 0x1c6a94b4, 0x6eee014d, 0x8a508e02, 0x3da538f5, 0x280d394f,
67 69aa0e90 2021-03-10 stsp 0x5248a0c4, 0x3ce94c6c, 0x9a71ad3a, 0x8493dd05, 0xe43f0ab6, 0x18e4ed42,
68 69aa0e90 2021-03-10 stsp 0x6c5c0e09, 0x42b06ec9, 0x8d330343, 0xa45b6f59, 0x2a573c0c, 0xd7fd3de6,
69 69aa0e90 2021-03-10 stsp 0xeedeab68, 0x5c84dafc, 0xbbd1b1a8, 0xa3ce1ad1, 0x85b70bed, 0xb6add07f,
70 69aa0e90 2021-03-10 stsp 0xa531309c, 0x8f8ab852, 0x564de332, 0xeac9ed0c, 0x73da402c, 0x3ec52761,
71 69aa0e90 2021-03-10 stsp 0x43af2f4d, 0xd6ff45c8, 0x4c367462, 0xd553bd6a, 0x44724855, 0x3b2aa728,
72 69aa0e90 2021-03-10 stsp 0x56e5eb65, 0xeaf16173, 0x33fa42ff, 0xd714bb5d, 0xfbd0a3b9, 0xaf517134,
73 69aa0e90 2021-03-10 stsp 0x9416c8cd, 0x534cf94f, 0x548947c2, 0x34193569, 0x32f4389a, 0xfe7028bc,
74 69aa0e90 2021-03-10 stsp 0xed73b1ed, 0x9db95770, 0x468e3922, 0x0440c3cd, 0x60059a62, 0x33504562,
75 69aa0e90 2021-03-10 stsp 0x2b229fbd, 0x5174dca5, 0xf7028752, 0xd63c6aa8, 0x31276f38, 0x0646721c,
76 69aa0e90 2021-03-10 stsp 0xb0191da8, 0xe00e6de0, 0x9eac1a6e, 0x9f7628a5, 0xed6c06ea, 0x0bb8af15,
77 69aa0e90 2021-03-10 stsp 0xf119fb12, 0x38693c1c, 0x732bc0fe, 0x84953275, 0xb82ec888, 0x33a4f1b3,
78 69aa0e90 2021-03-10 stsp 0x3099835e, 0x028a8782, 0x5fdd51d7, 0xc6c717b3, 0xb06caf71, 0x17c8c111,
79 69aa0e90 2021-03-10 stsp 0x61bad754, 0x9fd03061, 0xe09df1af, 0x3bc9eb73, 0x85878413, 0x9889aaf2,
80 69aa0e90 2021-03-10 stsp 0x3f5a9e46, 0x42c9f01f, 0x9984a4f4, 0xd5de43cc, 0xd294daed, 0xbecba2d2,
81 69aa0e90 2021-03-10 stsp 0xf1f6e72c, 0x5551128a, 0x83af87e2, 0x6f0342ba,
84 f364801d 2021-03-10 stsp static uint64_t
85 69aa0e90 2021-03-10 stsp hashblk(const unsigned char *p, off_t n)
87 f364801d 2021-03-10 stsp unsigned char buf[SHA1_DIGEST_LENGTH];
89 69aa0e90 2021-03-10 stsp SHA1_CTX ctx;
90 69aa0e90 2021-03-10 stsp SHA1Init(&ctx);
91 69aa0e90 2021-03-10 stsp SHA1Update(&ctx, p, n);
92 69aa0e90 2021-03-10 stsp SHA1Final(buf, &ctx);
93 f364801d 2021-03-10 stsp memcpy(&h, buf, sizeof(h));
94 f364801d 2021-03-10 stsp return be64toh(h);
97 f364801d 2021-03-10 stsp static const struct got_error *
98 69aa0e90 2021-03-10 stsp addblk(struct got_delta_table *dt, FILE *f, off_t len, off_t offset, uint64_t h)
100 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
101 69aa0e90 2021-03-10 stsp int i, nalloc;
102 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
103 69aa0e90 2021-03-10 stsp size_t r = 0;
105 69aa0e90 2021-03-10 stsp if (len == 0)
106 69aa0e90 2021-03-10 stsp return NULL;
108 69aa0e90 2021-03-10 stsp i = h % dt->nalloc;
109 69aa0e90 2021-03-10 stsp while (dt->blocks[i].len != 0) {
110 69aa0e90 2021-03-10 stsp if (r == 0) {
111 69aa0e90 2021-03-10 stsp if (fseeko(f, offset, SEEK_SET) == -1)
112 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
113 69aa0e90 2021-03-10 stsp r = fread(buf, 1, len, f);
114 69aa0e90 2021-03-10 stsp if (r != len) {
115 69aa0e90 2021-03-10 stsp if (ferror(f))
116 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
117 69aa0e90 2021-03-10 stsp return NULL;
121 69aa0e90 2021-03-10 stsp * Avoid adding duplicate blocks.
122 69aa0e90 2021-03-10 stsp * NB: A matching hash is insufficient for detecting equality.
123 69aa0e90 2021-03-10 stsp * The hash can only detect inequality, so only check 'len'.
125 69aa0e90 2021-03-10 stsp if (len == dt->blocks[i].len) {
126 69aa0e90 2021-03-10 stsp uint8_t buf2[GOT_DELTIFY_MAXCHUNK];
127 69aa0e90 2021-03-10 stsp if (fseeko(f, dt->blocks[i].offset, SEEK_SET) == -1)
128 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
129 69aa0e90 2021-03-10 stsp if (fread(buf2, 1, len, f) != len)
130 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
131 69aa0e90 2021-03-10 stsp if (memcmp(buf, buf2, len) == 0)
132 69aa0e90 2021-03-10 stsp return NULL;
135 69aa0e90 2021-03-10 stsp i = (i + 1) % dt->nalloc;
137 69aa0e90 2021-03-10 stsp assert(dt->blocks[i].len == 0);
138 69aa0e90 2021-03-10 stsp dt->blocks[i].len = len;
139 69aa0e90 2021-03-10 stsp dt->blocks[i].offset = offset;
140 69aa0e90 2021-03-10 stsp dt->blocks[i].hash = h;
141 69aa0e90 2021-03-10 stsp dt->nblocks++;
142 69aa0e90 2021-03-10 stsp if (dt->nalloc < 2 * dt->nblocks) {
143 69aa0e90 2021-03-10 stsp struct got_delta_block *db;
144 69aa0e90 2021-03-10 stsp nalloc = dt->nalloc * 2;
145 69aa0e90 2021-03-10 stsp db = dt->blocks;
146 69aa0e90 2021-03-10 stsp dt->blocks = calloc(nalloc, sizeof(struct got_delta_block));
147 69aa0e90 2021-03-10 stsp if (dt->blocks == NULL) {
148 69aa0e90 2021-03-10 stsp err = got_error_from_errno("calloc");
149 69aa0e90 2021-03-10 stsp dt->blocks = db;
150 69aa0e90 2021-03-10 stsp return err;
152 69aa0e90 2021-03-10 stsp dt->nalloc = nalloc;
154 69aa0e90 2021-03-10 stsp * Recompute all block positions. Hash-based indices of blocks
155 69aa0e90 2021-03-10 stsp * in the array depend on the allocated length of the array.
157 69aa0e90 2021-03-10 stsp dt->nblocks = 0;
158 69aa0e90 2021-03-10 stsp for (i = 0; i < nalloc; i++) {
159 69aa0e90 2021-03-10 stsp if (db[i].len == 0)
161 69aa0e90 2021-03-10 stsp err = addblk(dt, f, db[i].len, db[i].offset,
162 69aa0e90 2021-03-10 stsp db[i].hash);
169 69aa0e90 2021-03-10 stsp return err;
172 69aa0e90 2021-03-10 stsp static const struct got_error *
173 69aa0e90 2021-03-10 stsp lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
174 69aa0e90 2021-03-10 stsp unsigned char *p, off_t len, FILE *basefile)
177 f364801d 2021-03-10 stsp uint64_t h;
178 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
181 69aa0e90 2021-03-10 stsp *block = NULL;
183 69aa0e90 2021-03-10 stsp h = hashblk(p, len);
184 69aa0e90 2021-03-10 stsp for (i = h % dt->nalloc; dt->blocks[i].len != 0;
185 69aa0e90 2021-03-10 stsp i = (i + 1) % dt->nalloc) {
186 69aa0e90 2021-03-10 stsp if (dt->blocks[i].hash != h ||
187 69aa0e90 2021-03-10 stsp dt->blocks[i].len != len)
189 69aa0e90 2021-03-10 stsp if (fseeko(basefile, dt->blocks[i].offset, SEEK_SET) == -1)
190 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
191 69aa0e90 2021-03-10 stsp r = fread(buf, 1, len, basefile);
192 69aa0e90 2021-03-10 stsp if (r != len)
193 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
194 69aa0e90 2021-03-10 stsp if (memcmp(p, buf, len) == 0) {
195 69aa0e90 2021-03-10 stsp *block = &dt->blocks[i];
199 f364801d 2021-03-10 stsp return NULL;
202 69aa0e90 2021-03-10 stsp static const struct got_error *
203 69aa0e90 2021-03-10 stsp nextblk(uint8_t *buf, off_t *blocklen, FILE *f)
205 f364801d 2021-03-10 stsp uint32_t gh;
206 69aa0e90 2021-03-10 stsp const unsigned char *p;
208 69aa0e90 2021-03-10 stsp off_t pos = ftello(f);
210 69aa0e90 2021-03-10 stsp *blocklen = 0;
212 69aa0e90 2021-03-10 stsp r = fread(buf, 1, GOT_DELTIFY_MAXCHUNK, f);
213 69aa0e90 2021-03-10 stsp if (r == 0 && ferror(f))
214 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
215 69aa0e90 2021-03-10 stsp if (r < GOT_DELTIFY_MINCHUNK)
216 69aa0e90 2021-03-10 stsp return NULL; /* no more delta-worthy blocks left */
218 69aa0e90 2021-03-10 stsp /* Got a deltifiable block. Find the split-point where it ends. */
219 69aa0e90 2021-03-10 stsp p = buf + GOT_DELTIFY_MINCHUNK;
221 69aa0e90 2021-03-10 stsp while (p != buf + r) {
222 f364801d 2021-03-10 stsp gh = (gh << 1) + geartab[*p++];
223 69aa0e90 2021-03-10 stsp if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
227 69aa0e90 2021-03-10 stsp *blocklen = (p - buf);
228 69aa0e90 2021-03-10 stsp if (fseeko(f, pos + *blocklen, SEEK_SET) == -1)
229 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
231 69aa0e90 2021-03-10 stsp return NULL;
234 f364801d 2021-03-10 stsp const struct got_error *
235 69aa0e90 2021-03-10 stsp got_deltify_init(struct got_delta_table **dt, FILE *f, off_t fileoffset,
236 69aa0e90 2021-03-10 stsp off_t filesize)
238 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
239 f364801d 2021-03-10 stsp uint64_t h;
241 69aa0e90 2021-03-10 stsp *dt = calloc(1, sizeof(**dt));
242 69aa0e90 2021-03-10 stsp if (*dt == NULL)
243 f364801d 2021-03-10 stsp return got_error_from_errno("calloc");
245 69aa0e90 2021-03-10 stsp (*dt)->nblocks = 0;
246 69aa0e90 2021-03-10 stsp (*dt)->nalloc = 128;
247 69aa0e90 2021-03-10 stsp (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
248 69aa0e90 2021-03-10 stsp if ((*dt)->blocks == NULL) {
249 69aa0e90 2021-03-10 stsp err = got_error_from_errno("calloc");
253 69aa0e90 2021-03-10 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
254 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
256 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
257 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
258 69aa0e90 2021-03-10 stsp off_t blocklen;
259 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
262 69aa0e90 2021-03-10 stsp if (blocklen == 0)
264 69aa0e90 2021-03-10 stsp h = hashblk(buf, blocklen);
265 69aa0e90 2021-03-10 stsp err = addblk(*dt, f, blocklen, fileoffset, h);
268 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
272 69aa0e90 2021-03-10 stsp free((*dt)->blocks);
274 69aa0e90 2021-03-10 stsp *dt = NULL;
277 69aa0e90 2021-03-10 stsp return err;
281 69aa0e90 2021-03-10 stsp got_deltify_free(struct got_delta_table *dt)
283 69aa0e90 2021-03-10 stsp free(dt->blocks);
287 f364801d 2021-03-10 stsp static const struct got_error *
288 69aa0e90 2021-03-10 stsp emitdelta(struct got_delta_instruction **deltas, int *ndeltas, int copy,
289 69aa0e90 2021-03-10 stsp off_t offset, off_t len)
291 69aa0e90 2021-03-10 stsp struct got_delta_instruction *d, *p;
293 69aa0e90 2021-03-10 stsp *ndeltas += 1;
294 69aa0e90 2021-03-10 stsp p = reallocarray(*deltas, *ndeltas,
295 69aa0e90 2021-03-10 stsp sizeof(struct got_delta_instruction));
296 f364801d 2021-03-10 stsp if (p == NULL)
297 f364801d 2021-03-10 stsp return got_error_from_errno("realloc");
298 69aa0e90 2021-03-10 stsp *deltas = p;
299 69aa0e90 2021-03-10 stsp d = &(*deltas)[*ndeltas - 1];
300 69aa0e90 2021-03-10 stsp d->copy = copy;
301 69aa0e90 2021-03-10 stsp d->offset = offset;
302 56e7a15f 2021-03-10 stsp d->len = len;
303 f364801d 2021-03-10 stsp return NULL;
306 69aa0e90 2021-03-10 stsp static const struct got_error *
307 69aa0e90 2021-03-10 stsp stretchblk(FILE *basefile, struct got_delta_block *block, FILE *f,
308 69aa0e90 2021-03-10 stsp off_t filesize, off_t *blocklen)
310 69aa0e90 2021-03-10 stsp uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
311 69aa0e90 2021-03-10 stsp size_t base_r, r, i;
312 69aa0e90 2021-03-10 stsp off_t orig_blocklen = *blocklen;
313 69aa0e90 2021-03-10 stsp off_t pos = ftello(f);
314 69aa0e90 2021-03-10 stsp int buf_equal = 1;
316 69aa0e90 2021-03-10 stsp if (fseeko(basefile, block->offset, SEEK_SET) == -1)
317 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
319 69aa0e90 2021-03-10 stsp while (buf_equal && *blocklen < (1 << 24) - 1) {
320 69aa0e90 2021-03-10 stsp base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
321 69aa0e90 2021-03-10 stsp if (base_r == 0) {
322 69aa0e90 2021-03-10 stsp if (ferror(basefile))
323 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
326 69aa0e90 2021-03-10 stsp r = fread(buf, 1, sizeof(buf), f);
327 69aa0e90 2021-03-10 stsp if (r == 0) {
328 69aa0e90 2021-03-10 stsp if (ferror(f))
329 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
332 69aa0e90 2021-03-10 stsp for (i = 0; i < MIN(base_r, r); i++) {
333 69aa0e90 2021-03-10 stsp if (buf[i] != basebuf[i]) {
334 69aa0e90 2021-03-10 stsp buf_equal = 0;
337 69aa0e90 2021-03-10 stsp (*blocklen)++;
341 69aa0e90 2021-03-10 stsp if (fseeko(f, pos + *blocklen - orig_blocklen, SEEK_SET) == -1)
342 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
344 69aa0e90 2021-03-10 stsp return NULL;
347 69aa0e90 2021-03-10 stsp const struct got_error *
348 69aa0e90 2021-03-10 stsp got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
349 69aa0e90 2021-03-10 stsp FILE *f, off_t fileoffset, off_t filesize,
350 69aa0e90 2021-03-10 stsp struct got_delta_table *dt, FILE *basefile,
351 69aa0e90 2021-03-10 stsp off_t basefile_size)
353 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
354 69aa0e90 2021-03-10 stsp const off_t offset0 = fileoffset;
356 69aa0e90 2021-03-10 stsp *deltas = NULL;
357 69aa0e90 2021-03-10 stsp *ndeltas = 0;
360 69aa0e90 2021-03-10 stsp * offset0 indicates where data to be deltified begins.
361 69aa0e90 2021-03-10 stsp * For example, we want to avoid deltifying a Git object header at
362 69aa0e90 2021-03-10 stsp * the beginning of the file.
364 69aa0e90 2021-03-10 stsp if (fseeko(f, offset0, SEEK_SET) == -1)
365 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
367 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
368 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
369 69aa0e90 2021-03-10 stsp off_t blocklen;
370 69aa0e90 2021-03-10 stsp struct got_delta_block *block;
371 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
374 69aa0e90 2021-03-10 stsp if (blocklen == 0)
376 69aa0e90 2021-03-10 stsp err = lookupblk(&block, dt, buf, blocklen, basefile);
379 69aa0e90 2021-03-10 stsp if (block != NULL) {
381 69aa0e90 2021-03-10 stsp * We have found a matching block in the delta base.
382 69aa0e90 2021-03-10 stsp * Attempt to stretch the block as far as possible and
383 69aa0e90 2021-03-10 stsp * generate a copy instruction.
385 69aa0e90 2021-03-10 stsp err = stretchblk(basefile, block, f, filesize,
386 69aa0e90 2021-03-10 stsp &blocklen);
389 69aa0e90 2021-03-10 stsp emitdelta(deltas, ndeltas, 1, block->offset, blocklen);
392 69aa0e90 2021-03-10 stsp * No match.
393 69aa0e90 2021-03-10 stsp * This block needs to be sourced from the file itself.
395 69aa0e90 2021-03-10 stsp emitdelta(deltas, ndeltas, 0, fileoffset - offset0,
398 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
402 69aa0e90 2021-03-10 stsp free(*deltas);
403 69aa0e90 2021-03-10 stsp *deltas = NULL;
404 69aa0e90 2021-03-10 stsp *ndeltas = 0;
406 69aa0e90 2021-03-10 stsp return err;
