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>
20 f364801d 2021-03-10 stsp #include <assert.h>
21 69aa0e90 2021-03-10 stsp #include <errno.h>
22 f364801d 2021-03-10 stsp #include <limits.h>
23 f364801d 2021-03-10 stsp #include <stdint.h>
24 f364801d 2021-03-10 stsp #include <stdio.h>
25 f364801d 2021-03-10 stsp #include <stdlib.h>
26 f364801d 2021-03-10 stsp #include <string.h>
28 f364801d 2021-03-10 stsp #include "got_error.h"
30 69aa0e90 2021-03-10 stsp #include "got_lib_deltify.h"
31 b7bf7afc 2022-02-12 thomas #include "murmurhash2.h"
34 69aa0e90 2021-03-10 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
38 2539ccf2 2021-10-08 thomas * The algorihm used here is FastCDC (Fast Content-Defined Chunking)
39 2539ccf2 2021-10-08 thomas * https://www.usenix.org/conference/atc16/technical-sessions/presentation/xia
42 fa8129f7 2022-03-22 thomas static const uint32_t geartab[256] = {
43 69aa0e90 2021-03-10 stsp 0x67ed26b7, 0x32da500c, 0x53d0fee0, 0xce387dc7, 0xcd406d90, 0x2e83a4d4,
44 69aa0e90 2021-03-10 stsp 0x9fc9a38d, 0xb67259dc, 0xca6b1722, 0x6d2ea08c, 0x235cea2e, 0x3149bb5f,
45 69aa0e90 2021-03-10 stsp 0x1beda787, 0x2a6b77d5, 0x2f22d9ac, 0x91fc0544, 0xe413acfa, 0x5a30ff7a,
46 69aa0e90 2021-03-10 stsp 0xad6fdde0, 0x444fd0f5, 0x7ad87864, 0x58c5ff05, 0x8d2ec336, 0x2371f853,
47 69aa0e90 2021-03-10 stsp 0x550f8572, 0x6aa448dd, 0x7c9ddbcf, 0x95221e14, 0x2a82ec33, 0xcbec5a78,
48 69aa0e90 2021-03-10 stsp 0xc6795a0d, 0x243995b7, 0x1c909a2f, 0x4fded51c, 0x635d334b, 0x0e2b9999,
49 69aa0e90 2021-03-10 stsp 0x2702968d, 0x856de1d5, 0x3325d60e, 0xeb6a7502, 0xec2a9844, 0x0905835a,
50 69aa0e90 2021-03-10 stsp 0xa1820375, 0xa4be5cab, 0x96a6c058, 0x2c2ccd70, 0xba40fce3, 0xd794c46b,
51 69aa0e90 2021-03-10 stsp 0x8fbae83e, 0xc3aa7899, 0x3d3ff8ed, 0xa0d42b5b, 0x571c0c97, 0xd2811516,
52 69aa0e90 2021-03-10 stsp 0xf7e7b96c, 0x4fd2fcbd, 0xe2fdec94, 0x282cc436, 0x78e8e95c, 0x80a3b613,
53 69aa0e90 2021-03-10 stsp 0xcfbee20c, 0xd4a32d1c, 0x2a12ff13, 0x6af82936, 0xe5630258, 0x8efa6a98,
54 69aa0e90 2021-03-10 stsp 0x294fb2d1, 0xdeb57086, 0x5f0fddb3, 0xeceda7ce, 0x4c87305f, 0x3a6d3307,
55 69aa0e90 2021-03-10 stsp 0xe22d2942, 0x9d060217, 0x1e42ed02, 0xb6f63b52, 0x4367f39f, 0x055cf262,
56 69aa0e90 2021-03-10 stsp 0x03a461b2, 0x5ef9e382, 0x386bc03a, 0x2a1e79c7, 0xf1a0058b, 0xd4d2dea9,
57 69aa0e90 2021-03-10 stsp 0x56baf37d, 0x5daff6cc, 0xf03a951d, 0xaef7de45, 0xa8f4581e, 0x3960b555,
58 69aa0e90 2021-03-10 stsp 0xffbfff6d, 0xbe702a23, 0x8f5b6d6f, 0x061739fb, 0x98696f47, 0x3fd596d4,
59 69aa0e90 2021-03-10 stsp 0x151eac6b, 0xa9fcc4f5, 0x69181a12, 0x3ac5a107, 0xb5198fe7, 0x96bcb1da,
60 69aa0e90 2021-03-10 stsp 0x1b5ddf8e, 0xc757d650, 0x65865c3a, 0x8fc0a41a, 0x87435536, 0x99eda6f2,
61 69aa0e90 2021-03-10 stsp 0x41874794, 0x29cff4e8, 0xb70efd9a, 0x3103f6e7, 0x84d2453b, 0x15a450bd,
62 69aa0e90 2021-03-10 stsp 0x74f49af1, 0x60f664b1, 0xa1c86935, 0xfdafbce1, 0xe36353e3, 0x5d9ba739,
63 69aa0e90 2021-03-10 stsp 0xbc0559ba, 0x708b0054, 0xd41d808c, 0xb2f31723, 0x9027c41f, 0xf136d165,
64 69aa0e90 2021-03-10 stsp 0xb5374b12, 0x9420a6ac, 0x273958b6, 0xe6c2fad0, 0xebdc1f21, 0xfb33af8b,
65 69aa0e90 2021-03-10 stsp 0xc71c25cd, 0xe9a2d8e5, 0xbeb38a50, 0xbceb7cc2, 0x4e4e73f0, 0xcd6c251d,
66 69aa0e90 2021-03-10 stsp 0xde4c032c, 0x4b04ac30, 0x725b8b21, 0x4eb8c33b, 0x20d07b75, 0x0567aa63,
67 69aa0e90 2021-03-10 stsp 0xb56b2bb7, 0xc1f5fd3a, 0xcafd35ca, 0x470dd4da, 0xfe4f94cd, 0xfb8de424,
68 69aa0e90 2021-03-10 stsp 0xe8dbcf40, 0xfe50a37a, 0x62db5b5d, 0xf32f4ab6, 0x2c4a8a51, 0x18473dc0,
69 69aa0e90 2021-03-10 stsp 0xfe0cbb6e, 0xfe399efd, 0xdf34ecc9, 0x6ccd5055, 0x46097073, 0x139135c2,
70 69aa0e90 2021-03-10 stsp 0x721c76f6, 0x1c6a94b4, 0x6eee014d, 0x8a508e02, 0x3da538f5, 0x280d394f,
71 69aa0e90 2021-03-10 stsp 0x5248a0c4, 0x3ce94c6c, 0x9a71ad3a, 0x8493dd05, 0xe43f0ab6, 0x18e4ed42,
72 69aa0e90 2021-03-10 stsp 0x6c5c0e09, 0x42b06ec9, 0x8d330343, 0xa45b6f59, 0x2a573c0c, 0xd7fd3de6,
73 69aa0e90 2021-03-10 stsp 0xeedeab68, 0x5c84dafc, 0xbbd1b1a8, 0xa3ce1ad1, 0x85b70bed, 0xb6add07f,
74 69aa0e90 2021-03-10 stsp 0xa531309c, 0x8f8ab852, 0x564de332, 0xeac9ed0c, 0x73da402c, 0x3ec52761,
75 69aa0e90 2021-03-10 stsp 0x43af2f4d, 0xd6ff45c8, 0x4c367462, 0xd553bd6a, 0x44724855, 0x3b2aa728,
76 69aa0e90 2021-03-10 stsp 0x56e5eb65, 0xeaf16173, 0x33fa42ff, 0xd714bb5d, 0xfbd0a3b9, 0xaf517134,
77 69aa0e90 2021-03-10 stsp 0x9416c8cd, 0x534cf94f, 0x548947c2, 0x34193569, 0x32f4389a, 0xfe7028bc,
78 69aa0e90 2021-03-10 stsp 0xed73b1ed, 0x9db95770, 0x468e3922, 0x0440c3cd, 0x60059a62, 0x33504562,
79 69aa0e90 2021-03-10 stsp 0x2b229fbd, 0x5174dca5, 0xf7028752, 0xd63c6aa8, 0x31276f38, 0x0646721c,
80 69aa0e90 2021-03-10 stsp 0xb0191da8, 0xe00e6de0, 0x9eac1a6e, 0x9f7628a5, 0xed6c06ea, 0x0bb8af15,
81 69aa0e90 2021-03-10 stsp 0xf119fb12, 0x38693c1c, 0x732bc0fe, 0x84953275, 0xb82ec888, 0x33a4f1b3,
82 69aa0e90 2021-03-10 stsp 0x3099835e, 0x028a8782, 0x5fdd51d7, 0xc6c717b3, 0xb06caf71, 0x17c8c111,
83 69aa0e90 2021-03-10 stsp 0x61bad754, 0x9fd03061, 0xe09df1af, 0x3bc9eb73, 0x85878413, 0x9889aaf2,
84 69aa0e90 2021-03-10 stsp 0x3f5a9e46, 0x42c9f01f, 0x9984a4f4, 0xd5de43cc, 0xd294daed, 0xbecba2d2,
85 69aa0e90 2021-03-10 stsp 0xf1f6e72c, 0x5551128a, 0x83af87e2, 0x6f0342ba,
88 b7bf7afc 2022-02-12 thomas static uint32_t
89 cc524d36 2022-05-31 thomas hashblk(const unsigned char *p, off_t n, uint32_t seed)
91 cc524d36 2022-05-31 thomas return murmurhash2(p, n, seed);
94 f364801d 2021-03-10 stsp static const struct got_error *
95 740bba1c 2021-06-18 stsp addblk(struct got_delta_table *dt, FILE *f, off_t file_offset0, off_t len,
96 90dfa2bf 2022-02-13 thomas off_t offset, uint32_t h)
98 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
100 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
101 0d15f6dc 2021-06-13 stsp uint8_t buf2[GOT_DELTIFY_MAXCHUNK];
102 69aa0e90 2021-03-10 stsp size_t r = 0;
104 69aa0e90 2021-03-10 stsp if (len == 0)
105 69aa0e90 2021-03-10 stsp return NULL;
107 69aa0e90 2021-03-10 stsp i = h % dt->nalloc;
108 69aa0e90 2021-03-10 stsp while (dt->blocks[i].len != 0) {
110 69aa0e90 2021-03-10 stsp * Avoid adding duplicate blocks.
111 69aa0e90 2021-03-10 stsp * NB: A matching hash is insufficient for detecting equality.
112 51a494da 2021-06-11 stsp * The hash can only detect inequality.
114 51a494da 2021-06-11 stsp if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
115 0d15f6dc 2021-06-13 stsp if (r == 0) {
116 740bba1c 2021-06-18 stsp if (fseeko(f, file_offset0 + offset, SEEK_SET) == -1)
117 0d15f6dc 2021-06-13 stsp return got_error_from_errno("fseeko");
118 0d15f6dc 2021-06-13 stsp r = fread(buf, 1, len, f);
119 0d15f6dc 2021-06-13 stsp if (r != len) {
120 0d15f6dc 2021-06-13 stsp if (!ferror(f))
121 0d15f6dc 2021-06-13 stsp return NULL;
122 0d15f6dc 2021-06-13 stsp return got_ferror(f, GOT_ERR_IO);
125 740bba1c 2021-06-18 stsp if (fseeko(f, file_offset0 + dt->blocks[i].offset,
126 740bba1c 2021-06-18 stsp SEEK_SET) == -1)
127 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
128 69aa0e90 2021-03-10 stsp if (fread(buf2, 1, len, f) != len)
129 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
130 69aa0e90 2021-03-10 stsp if (memcmp(buf, buf2, len) == 0)
131 69aa0e90 2021-03-10 stsp return NULL;
134 69aa0e90 2021-03-10 stsp i = (i + 1) % dt->nalloc;
136 69aa0e90 2021-03-10 stsp assert(dt->blocks[i].len == 0);
137 69aa0e90 2021-03-10 stsp dt->blocks[i].len = len;
138 69aa0e90 2021-03-10 stsp dt->blocks[i].offset = offset;
139 69aa0e90 2021-03-10 stsp dt->blocks[i].hash = h;
140 69aa0e90 2021-03-10 stsp dt->nblocks++;
141 e89540a9 2021-06-11 stsp if (dt->nalloc < dt->nblocks + 64) {
142 69aa0e90 2021-03-10 stsp struct got_delta_block *db;
143 e89540a9 2021-06-11 stsp size_t old_size = dt->nalloc;
144 69aa0e90 2021-03-10 stsp db = dt->blocks;
145 e89540a9 2021-06-11 stsp dt->blocks = calloc(dt->nalloc + 64,
146 e89540a9 2021-06-11 stsp 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 e89540a9 2021-06-11 stsp dt->nalloc += 64;
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 a893025f 2021-06-11 stsp for (i = 0; i < old_size; i++) {
159 69aa0e90 2021-03-10 stsp if (db[i].len == 0)
161 740bba1c 2021-06-18 stsp err = addblk(dt, f, file_offset0, db[i].len,
162 2b0ae357 2022-01-10 thomas db[i].offset, db[i].hash);
166 2b0ae357 2022-01-10 thomas free(db);
169 2b0ae357 2022-01-10 thomas return err;
172 2b0ae357 2022-01-10 thomas static const struct got_error *
173 2b0ae357 2022-01-10 thomas addblk_mem(struct got_delta_table *dt, uint8_t *data, off_t file_offset0,
174 90dfa2bf 2022-02-13 thomas off_t len, off_t offset, uint32_t h)
176 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
178 2b0ae357 2022-01-10 thomas uint8_t *block1;
179 2b0ae357 2022-01-10 thomas uint8_t *block2;
181 2b0ae357 2022-01-10 thomas if (len == 0)
182 2b0ae357 2022-01-10 thomas return NULL;
184 2b0ae357 2022-01-10 thomas i = h % dt->nalloc;
185 2b0ae357 2022-01-10 thomas while (dt->blocks[i].len != 0) {
187 2b0ae357 2022-01-10 thomas * Avoid adding duplicate blocks.
188 2b0ae357 2022-01-10 thomas * NB: A matching hash is insufficient for detecting equality.
189 2b0ae357 2022-01-10 thomas * The hash can only detect inequality.
191 2b0ae357 2022-01-10 thomas if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
192 2b0ae357 2022-01-10 thomas block1 = data + file_offset0 + dt->blocks[i].offset;
193 2b0ae357 2022-01-10 thomas block2 = data + file_offset0 + offset;
194 2b0ae357 2022-01-10 thomas if (memcmp(block1, block2, len) == 0)
195 2b0ae357 2022-01-10 thomas return NULL;
198 2b0ae357 2022-01-10 thomas i = (i + 1) % dt->nalloc;
200 2b0ae357 2022-01-10 thomas assert(dt->blocks[i].len == 0);
201 2b0ae357 2022-01-10 thomas dt->blocks[i].len = len;
202 2b0ae357 2022-01-10 thomas dt->blocks[i].offset = offset;
203 2b0ae357 2022-01-10 thomas dt->blocks[i].hash = h;
204 2b0ae357 2022-01-10 thomas dt->nblocks++;
205 2b0ae357 2022-01-10 thomas if (dt->nalloc < dt->nblocks + 64) {
206 2b0ae357 2022-01-10 thomas struct got_delta_block *db;
207 2b0ae357 2022-01-10 thomas size_t old_size = dt->nalloc;
208 2b0ae357 2022-01-10 thomas db = dt->blocks;
209 2b0ae357 2022-01-10 thomas dt->blocks = calloc(dt->nalloc + 64,
210 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_block));
211 2b0ae357 2022-01-10 thomas if (dt->blocks == NULL) {
212 2b0ae357 2022-01-10 thomas err = got_error_from_errno("calloc");
213 2b0ae357 2022-01-10 thomas dt->blocks = db;
214 2b0ae357 2022-01-10 thomas return err;
216 2b0ae357 2022-01-10 thomas dt->nalloc += 64;
218 2b0ae357 2022-01-10 thomas * Recompute all block positions. Hash-based indices of blocks
219 2b0ae357 2022-01-10 thomas * in the array depend on the allocated length of the array.
221 2b0ae357 2022-01-10 thomas dt->nblocks = 0;
222 2b0ae357 2022-01-10 thomas for (i = 0; i < old_size; i++) {
223 2b0ae357 2022-01-10 thomas if (db[i].len == 0)
224 2b0ae357 2022-01-10 thomas continue;
225 2b0ae357 2022-01-10 thomas err = addblk_mem(dt, data, file_offset0, db[i].len,
226 740bba1c 2021-06-18 stsp db[i].offset, db[i].hash);
233 69aa0e90 2021-03-10 stsp return err;
236 69aa0e90 2021-03-10 stsp static const struct got_error *
237 69aa0e90 2021-03-10 stsp lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
238 cc524d36 2022-05-31 thomas unsigned char *p, off_t len, uint32_t seed, FILE *basefile,
239 cc524d36 2022-05-31 thomas off_t basefile_offset0)
242 b7bf7afc 2022-02-12 thomas uint32_t h;
243 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
246 69aa0e90 2021-03-10 stsp *block = NULL;
248 cc524d36 2022-05-31 thomas h = hashblk(p, len, seed);
249 69aa0e90 2021-03-10 stsp for (i = h % dt->nalloc; dt->blocks[i].len != 0;
250 69aa0e90 2021-03-10 stsp i = (i + 1) % dt->nalloc) {
251 69aa0e90 2021-03-10 stsp if (dt->blocks[i].hash != h ||
252 69aa0e90 2021-03-10 stsp dt->blocks[i].len != len)
254 740bba1c 2021-06-18 stsp if (fseeko(basefile, basefile_offset0 + dt->blocks[i].offset,
255 740bba1c 2021-06-18 stsp SEEK_SET) == -1)
256 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
257 69aa0e90 2021-03-10 stsp r = fread(buf, 1, len, basefile);
258 69aa0e90 2021-03-10 stsp if (r != len)
259 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
260 69aa0e90 2021-03-10 stsp if (memcmp(p, buf, len) == 0) {
261 69aa0e90 2021-03-10 stsp *block = &dt->blocks[i];
265 f364801d 2021-03-10 stsp return NULL;
268 69aa0e90 2021-03-10 stsp static const struct got_error *
269 2b0ae357 2022-01-10 thomas lookupblk_mem(struct got_delta_block **block, struct got_delta_table *dt,
270 cc524d36 2022-05-31 thomas unsigned char *p, off_t len, uint32_t seed, uint8_t *basedata,
271 cc524d36 2022-05-31 thomas off_t basefile_offset0)
274 b7bf7afc 2022-02-12 thomas uint32_t h;
275 2b0ae357 2022-01-10 thomas uint8_t *b;
277 2b0ae357 2022-01-10 thomas *block = NULL;
279 cc524d36 2022-05-31 thomas h = hashblk(p, len, seed);
280 2b0ae357 2022-01-10 thomas for (i = h % dt->nalloc; dt->blocks[i].len != 0;
281 2b0ae357 2022-01-10 thomas i = (i + 1) % dt->nalloc) {
282 2b0ae357 2022-01-10 thomas if (dt->blocks[i].hash != h ||
283 2b0ae357 2022-01-10 thomas dt->blocks[i].len != len)
284 2b0ae357 2022-01-10 thomas continue;
285 2b0ae357 2022-01-10 thomas b = basedata + basefile_offset0 + dt->blocks[i].offset;
286 2b0ae357 2022-01-10 thomas if (memcmp(p, b, len) == 0) {
287 2b0ae357 2022-01-10 thomas *block = &dt->blocks[i];
291 2b0ae357 2022-01-10 thomas return NULL;
294 2b0ae357 2022-01-10 thomas static const struct got_error *
295 69aa0e90 2021-03-10 stsp nextblk(uint8_t *buf, off_t *blocklen, FILE *f)
297 f364801d 2021-03-10 stsp uint32_t gh;
298 69aa0e90 2021-03-10 stsp const unsigned char *p;
300 69aa0e90 2021-03-10 stsp off_t pos = ftello(f);
302 69aa0e90 2021-03-10 stsp *blocklen = 0;
304 69aa0e90 2021-03-10 stsp r = fread(buf, 1, GOT_DELTIFY_MAXCHUNK, f);
305 69aa0e90 2021-03-10 stsp if (r == 0 && ferror(f))
306 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
307 69aa0e90 2021-03-10 stsp if (r < GOT_DELTIFY_MINCHUNK)
308 69aa0e90 2021-03-10 stsp return NULL; /* no more delta-worthy blocks left */
310 69aa0e90 2021-03-10 stsp /* Got a deltifiable block. Find the split-point where it ends. */
311 69aa0e90 2021-03-10 stsp p = buf + GOT_DELTIFY_MINCHUNK;
313 69aa0e90 2021-03-10 stsp while (p != buf + r) {
314 f364801d 2021-03-10 stsp gh = (gh << 1) + geartab[*p++];
315 69aa0e90 2021-03-10 stsp if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
319 69aa0e90 2021-03-10 stsp *blocklen = (p - buf);
320 69aa0e90 2021-03-10 stsp if (fseeko(f, pos + *blocklen, SEEK_SET) == -1)
321 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
323 69aa0e90 2021-03-10 stsp return NULL;
326 2b0ae357 2022-01-10 thomas static const struct got_error *
327 2b0ae357 2022-01-10 thomas nextblk_mem(off_t *blocklen, uint8_t *data, off_t fileoffset, off_t filesize)
329 2b0ae357 2022-01-10 thomas uint32_t gh;
330 2b0ae357 2022-01-10 thomas const unsigned char *p;
332 2b0ae357 2022-01-10 thomas *blocklen = 0;
334 2b0ae357 2022-01-10 thomas if (fileoffset >= filesize ||
335 2b0ae357 2022-01-10 thomas filesize - fileoffset < GOT_DELTIFY_MINCHUNK)
336 2b0ae357 2022-01-10 thomas return NULL; /* no more delta-worthy blocks left */
338 2b0ae357 2022-01-10 thomas /* Got a deltifiable block. Find the split-point where it ends. */
339 2b0ae357 2022-01-10 thomas p = data + fileoffset + GOT_DELTIFY_MINCHUNK;
341 2b0ae357 2022-01-10 thomas while (p != data + MIN(fileoffset + GOT_DELTIFY_MAXCHUNK, filesize)) {
342 2b0ae357 2022-01-10 thomas gh = (gh << 1) + geartab[*p++];
343 2b0ae357 2022-01-10 thomas if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
347 2b0ae357 2022-01-10 thomas *blocklen = (p - (data + fileoffset));
348 2b0ae357 2022-01-10 thomas return NULL;
351 f364801d 2021-03-10 stsp const struct got_error *
352 69aa0e90 2021-03-10 stsp got_deltify_init(struct got_delta_table **dt, FILE *f, off_t fileoffset,
353 cc524d36 2022-05-31 thomas off_t filesize, uint32_t seed)
355 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
356 b7bf7afc 2022-02-12 thomas uint32_t h;
357 740bba1c 2021-06-18 stsp const off_t offset0 = fileoffset;
359 69aa0e90 2021-03-10 stsp *dt = calloc(1, sizeof(**dt));
360 69aa0e90 2021-03-10 stsp if (*dt == NULL)
361 f364801d 2021-03-10 stsp return got_error_from_errno("calloc");
363 69aa0e90 2021-03-10 stsp (*dt)->nblocks = 0;
364 69aa0e90 2021-03-10 stsp (*dt)->nalloc = 128;
365 69aa0e90 2021-03-10 stsp (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
366 69aa0e90 2021-03-10 stsp if ((*dt)->blocks == NULL) {
367 69aa0e90 2021-03-10 stsp err = got_error_from_errno("calloc");
371 69aa0e90 2021-03-10 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
372 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
374 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
375 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
376 69aa0e90 2021-03-10 stsp off_t blocklen;
377 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
380 69aa0e90 2021-03-10 stsp if (blocklen == 0)
382 cc524d36 2022-05-31 thomas h = hashblk(buf, blocklen, seed);
383 740bba1c 2021-06-18 stsp err = addblk(*dt, f, offset0, blocklen,
384 740bba1c 2021-06-18 stsp fileoffset - offset0, h);
387 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
388 68bdcdc2 2021-06-11 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
389 68bdcdc2 2021-06-11 stsp return got_error_from_errno("fseeko");
392 2b0ae357 2022-01-10 thomas if (err) {
393 2b0ae357 2022-01-10 thomas free((*dt)->blocks);
394 2b0ae357 2022-01-10 thomas free(*dt);
395 2b0ae357 2022-01-10 thomas *dt = NULL;
398 2b0ae357 2022-01-10 thomas return err;
401 2b0ae357 2022-01-10 thomas const struct got_error *
402 2b0ae357 2022-01-10 thomas got_deltify_init_mem(struct got_delta_table **dt, uint8_t *data,
403 cc524d36 2022-05-31 thomas off_t fileoffset, off_t filesize, uint32_t seed)
405 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
406 90dfa2bf 2022-02-13 thomas uint32_t h;
407 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
409 2b0ae357 2022-01-10 thomas *dt = calloc(1, sizeof(**dt));
410 2b0ae357 2022-01-10 thomas if (*dt == NULL)
411 2b0ae357 2022-01-10 thomas return got_error_from_errno("calloc");
413 2b0ae357 2022-01-10 thomas (*dt)->nblocks = 0;
414 2b0ae357 2022-01-10 thomas (*dt)->nalloc = 128;
415 2b0ae357 2022-01-10 thomas (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
416 2b0ae357 2022-01-10 thomas if ((*dt)->blocks == NULL) {
417 2b0ae357 2022-01-10 thomas err = got_error_from_errno("calloc");
418 2b0ae357 2022-01-10 thomas goto done;
421 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
422 2b0ae357 2022-01-10 thomas off_t blocklen;
423 2b0ae357 2022-01-10 thomas err = nextblk_mem(&blocklen, data, fileoffset, filesize);
425 2b0ae357 2022-01-10 thomas goto done;
426 2b0ae357 2022-01-10 thomas if (blocklen == 0)
428 cc524d36 2022-05-31 thomas h = hashblk(data + fileoffset, blocklen, seed);
429 2b0ae357 2022-01-10 thomas err = addblk_mem(*dt, data, offset0, blocklen,
430 2b0ae357 2022-01-10 thomas fileoffset - offset0, h);
432 2b0ae357 2022-01-10 thomas goto done;
433 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
437 69aa0e90 2021-03-10 stsp free((*dt)->blocks);
439 69aa0e90 2021-03-10 stsp *dt = NULL;
442 69aa0e90 2021-03-10 stsp return err;
446 69aa0e90 2021-03-10 stsp got_deltify_free(struct got_delta_table *dt)
448 dbbf4a5f 2021-05-20 stsp if (dt == NULL)
450 69aa0e90 2021-03-10 stsp free(dt->blocks);
454 f364801d 2021-03-10 stsp static const struct got_error *
455 dd29967c 2021-08-22 stsp emitdelta(struct got_delta_instruction **deltas, size_t *nalloc, int *ndeltas,
456 dd29967c 2021-08-22 stsp const size_t alloc_chunk_size, int copy, off_t offset, off_t len)
458 69aa0e90 2021-03-10 stsp struct got_delta_instruction *d, *p;
460 dd29967c 2021-08-22 stsp if (*nalloc < *ndeltas + alloc_chunk_size) {
461 dd29967c 2021-08-22 stsp p = reallocarray(*deltas, *nalloc + alloc_chunk_size,
462 dd29967c 2021-08-22 stsp sizeof(struct got_delta_instruction));
463 dd29967c 2021-08-22 stsp if (p == NULL)
464 7d76c4e7 2021-10-15 thomas return got_error_from_errno("reallocarray");
465 dd29967c 2021-08-22 stsp *deltas = p;
466 dd29967c 2021-08-22 stsp *nalloc += alloc_chunk_size;
468 69aa0e90 2021-03-10 stsp *ndeltas += 1;
469 69aa0e90 2021-03-10 stsp d = &(*deltas)[*ndeltas - 1];
470 69aa0e90 2021-03-10 stsp d->copy = copy;
471 69aa0e90 2021-03-10 stsp d->offset = offset;
472 56e7a15f 2021-03-10 stsp d->len = len;
473 f364801d 2021-03-10 stsp return NULL;
476 69aa0e90 2021-03-10 stsp static const struct got_error *
477 740bba1c 2021-06-18 stsp stretchblk(FILE *basefile, off_t base_offset0, struct got_delta_block *block,
478 740bba1c 2021-06-18 stsp FILE *f, off_t filesize, off_t *blocklen)
480 69aa0e90 2021-03-10 stsp uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
481 69aa0e90 2021-03-10 stsp size_t base_r, r, i;
482 69aa0e90 2021-03-10 stsp int buf_equal = 1;
484 5de743f8 2021-08-29 stsp if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
485 5de743f8 2021-08-29 stsp SEEK_SET) == -1)
486 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
488 69aa0e90 2021-03-10 stsp while (buf_equal && *blocklen < (1 << 24) - 1) {
489 69aa0e90 2021-03-10 stsp base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
490 69aa0e90 2021-03-10 stsp if (base_r == 0) {
491 69aa0e90 2021-03-10 stsp if (ferror(basefile))
492 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
495 69aa0e90 2021-03-10 stsp r = fread(buf, 1, sizeof(buf), f);
496 69aa0e90 2021-03-10 stsp if (r == 0) {
497 69aa0e90 2021-03-10 stsp if (ferror(f))
498 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
501 69aa0e90 2021-03-10 stsp for (i = 0; i < MIN(base_r, r); i++) {
502 69aa0e90 2021-03-10 stsp if (buf[i] != basebuf[i]) {
503 69aa0e90 2021-03-10 stsp buf_equal = 0;
506 69aa0e90 2021-03-10 stsp (*blocklen)++;
510 2b0ae357 2022-01-10 thomas return NULL;
513 2b0ae357 2022-01-10 thomas static const struct got_error *
514 2b0ae357 2022-01-10 thomas stretchblk_file_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
515 2b0ae357 2022-01-10 thomas struct got_delta_block *block, FILE *f, off_t filesize, off_t *blocklen)
517 2b0ae357 2022-01-10 thomas uint8_t buf[GOT_DELTIFY_MAXCHUNK];
518 2b0ae357 2022-01-10 thomas size_t r, i;
519 2b0ae357 2022-01-10 thomas int buf_equal = 1;
520 2b0ae357 2022-01-10 thomas off_t base_offset = base_offset0 + block->offset + *blocklen;
522 2b0ae357 2022-01-10 thomas if (base_offset > basefile_size) {
523 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
524 2b0ae357 2022-01-10 thomas "read beyond the size of delta base at offset %llu",
525 2b0ae357 2022-01-10 thomas base_offset);
528 2b0ae357 2022-01-10 thomas while (buf_equal && *blocklen < (1 << 24) - 1) {
529 2b0ae357 2022-01-10 thomas if (base_offset + *blocklen >= basefile_size)
531 2b0ae357 2022-01-10 thomas r = fread(buf, 1, sizeof(buf), f);
532 2b0ae357 2022-01-10 thomas if (r == 0) {
533 2b0ae357 2022-01-10 thomas if (ferror(f))
534 2b0ae357 2022-01-10 thomas return got_ferror(f, GOT_ERR_IO);
537 2b0ae357 2022-01-10 thomas for (i = 0; i < MIN(basefile_size - base_offset, r); i++) {
538 2b0ae357 2022-01-10 thomas if (buf[i] != *(basedata + base_offset + i)) {
539 2b0ae357 2022-01-10 thomas buf_equal = 0;
542 2b0ae357 2022-01-10 thomas (*blocklen)++;
546 2b0ae357 2022-01-10 thomas return NULL;
549 2b0ae357 2022-01-10 thomas static const struct got_error *
550 2b0ae357 2022-01-10 thomas stretchblk_mem_file(FILE *basefile, off_t base_offset0,
551 2b0ae357 2022-01-10 thomas struct got_delta_block *block, uint8_t *data, off_t fileoffset,
552 2b0ae357 2022-01-10 thomas off_t filesize, off_t *blocklen)
554 2b0ae357 2022-01-10 thomas uint8_t basebuf[GOT_DELTIFY_MAXCHUNK];
555 2b0ae357 2022-01-10 thomas size_t base_r, i;
556 2b0ae357 2022-01-10 thomas int buf_equal = 1;
558 2b0ae357 2022-01-10 thomas if (fileoffset > filesize) {
559 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
560 2b0ae357 2022-01-10 thomas "read beyond the size of deltify file at offset %llu",
561 2b0ae357 2022-01-10 thomas fileoffset);
564 2b0ae357 2022-01-10 thomas if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
565 2b0ae357 2022-01-10 thomas SEEK_SET) == -1)
566 2b0ae357 2022-01-10 thomas return got_error_from_errno("fseeko");
568 2b0ae357 2022-01-10 thomas while (buf_equal && *blocklen < (1 << 24) - 1) {
569 2b0ae357 2022-01-10 thomas if (fileoffset + *blocklen >= filesize)
571 2b0ae357 2022-01-10 thomas base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
572 2b0ae357 2022-01-10 thomas if (base_r == 0) {
573 2b0ae357 2022-01-10 thomas if (ferror(basefile))
574 2b0ae357 2022-01-10 thomas return got_ferror(basefile, GOT_ERR_IO);
577 2b0ae357 2022-01-10 thomas for (i = 0; i < MIN(base_r, filesize - fileoffset); i++) {
578 2b0ae357 2022-01-10 thomas if (*(data + fileoffset + i) != basebuf[i]) {
579 2b0ae357 2022-01-10 thomas buf_equal = 0;
582 2b0ae357 2022-01-10 thomas (*blocklen)++;
586 2b0ae357 2022-01-10 thomas return NULL;
589 2b0ae357 2022-01-10 thomas static const struct got_error *
590 2b0ae357 2022-01-10 thomas stretchblk_mem_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
591 2b0ae357 2022-01-10 thomas struct got_delta_block *block, uint8_t *data, off_t fileoffset,
592 2b0ae357 2022-01-10 thomas off_t filesize, off_t *blocklen)
594 2b0ae357 2022-01-10 thomas off_t i, maxlen;
595 2b0ae357 2022-01-10 thomas off_t base_offset = base_offset0 + block->offset + *blocklen;
596 2b0ae357 2022-01-10 thomas uint8_t *p, *q;
598 2b0ae357 2022-01-10 thomas if (base_offset > basefile_size) {
599 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
600 2b0ae357 2022-01-10 thomas "read beyond the size of delta base at offset %llu",
601 2b0ae357 2022-01-10 thomas base_offset);
604 2b0ae357 2022-01-10 thomas if (fileoffset > filesize) {
605 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
606 2b0ae357 2022-01-10 thomas "read beyond the size of deltify file at offset %llu",
607 2b0ae357 2022-01-10 thomas fileoffset);
610 2b0ae357 2022-01-10 thomas p = data + fileoffset;
611 2b0ae357 2022-01-10 thomas q = basedata + base_offset;
612 2b0ae357 2022-01-10 thomas maxlen = MIN(basefile_size - base_offset, filesize - fileoffset);
613 2b0ae357 2022-01-10 thomas for (i = 0; i < maxlen && *blocklen < (1 << 24) - 1; i++) {
614 2b0ae357 2022-01-10 thomas if (p[i] != q[i])
616 2b0ae357 2022-01-10 thomas (*blocklen)++;
619 69aa0e90 2021-03-10 stsp return NULL;
622 69aa0e90 2021-03-10 stsp const struct got_error *
623 69aa0e90 2021-03-10 stsp got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
624 cc524d36 2022-05-31 thomas FILE *f, off_t fileoffset, off_t filesize, uint32_t seed,
625 69aa0e90 2021-03-10 stsp struct got_delta_table *dt, FILE *basefile,
626 f34b169e 2021-06-18 stsp off_t basefile_offset0, off_t basefile_size)
628 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
629 69aa0e90 2021-03-10 stsp const off_t offset0 = fileoffset;
630 dd29967c 2021-08-22 stsp size_t nalloc = 0;
631 dd29967c 2021-08-22 stsp const size_t alloc_chunk_size = 64;
633 69aa0e90 2021-03-10 stsp *deltas = NULL;
634 69aa0e90 2021-03-10 stsp *ndeltas = 0;
637 69aa0e90 2021-03-10 stsp * offset0 indicates where data to be deltified begins.
638 69aa0e90 2021-03-10 stsp * For example, we want to avoid deltifying a Git object header at
639 69aa0e90 2021-03-10 stsp * the beginning of the file.
641 69aa0e90 2021-03-10 stsp if (fseeko(f, offset0, SEEK_SET) == -1)
642 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
644 dd29967c 2021-08-22 stsp *deltas = reallocarray(NULL, alloc_chunk_size,
645 dd29967c 2021-08-22 stsp sizeof(struct got_delta_instruction));
646 dd29967c 2021-08-22 stsp if (*deltas == NULL)
647 dd29967c 2021-08-22 stsp return got_error_from_errno("reallocarray");
648 dd29967c 2021-08-22 stsp nalloc = alloc_chunk_size;
650 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
651 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
652 69aa0e90 2021-03-10 stsp off_t blocklen;
653 69aa0e90 2021-03-10 stsp struct got_delta_block *block;
654 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
657 9a8dc2b3 2021-06-18 stsp if (blocklen == 0) {
658 9a8dc2b3 2021-06-18 stsp /* Source remainder from the file itself. */
659 9a8dc2b3 2021-06-18 stsp if (fileoffset < filesize) {
660 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
661 dd29967c 2021-08-22 stsp alloc_chunk_size, 0, fileoffset - offset0,
662 9a8dc2b3 2021-06-18 stsp filesize - fileoffset);
666 cc524d36 2022-05-31 thomas err = lookupblk(&block, dt, buf, blocklen, seed, basefile,
667 740bba1c 2021-06-18 stsp basefile_offset0);
670 69aa0e90 2021-03-10 stsp if (block != NULL) {
672 69aa0e90 2021-03-10 stsp * We have found a matching block in the delta base.
673 69aa0e90 2021-03-10 stsp * Attempt to stretch the block as far as possible and
674 69aa0e90 2021-03-10 stsp * generate a copy instruction.
676 740bba1c 2021-06-18 stsp err = stretchblk(basefile, basefile_offset0, block,
677 740bba1c 2021-06-18 stsp f, filesize, &blocklen);
680 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
681 dd29967c 2021-08-22 stsp alloc_chunk_size, 1, block->offset, blocklen);
686 69aa0e90 2021-03-10 stsp * No match.
687 69aa0e90 2021-03-10 stsp * This block needs to be sourced from the file itself.
689 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
690 dd29967c 2021-08-22 stsp alloc_chunk_size, 0, fileoffset - offset0, blocklen);
694 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
695 0af64e86 2021-08-22 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1) {
696 0af64e86 2021-08-22 stsp err = got_error_from_errno("fseeko");
702 69aa0e90 2021-03-10 stsp free(*deltas);
703 69aa0e90 2021-03-10 stsp *deltas = NULL;
704 69aa0e90 2021-03-10 stsp *ndeltas = 0;
706 69aa0e90 2021-03-10 stsp return err;
709 2b0ae357 2022-01-10 thomas const struct got_error *
710 2b0ae357 2022-01-10 thomas got_deltify_file_mem(struct got_delta_instruction **deltas, int *ndeltas,
711 cc524d36 2022-05-31 thomas FILE *f, off_t fileoffset, off_t filesize, uint32_t seed,
712 2b0ae357 2022-01-10 thomas struct got_delta_table *dt, uint8_t *basedata,
713 2b0ae357 2022-01-10 thomas off_t basefile_offset0, off_t basefile_size)
715 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
716 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
717 2b0ae357 2022-01-10 thomas size_t nalloc = 0;
718 2b0ae357 2022-01-10 thomas const size_t alloc_chunk_size = 64;
720 2b0ae357 2022-01-10 thomas *deltas = NULL;
721 2b0ae357 2022-01-10 thomas *ndeltas = 0;
724 2b0ae357 2022-01-10 thomas * offset0 indicates where data to be deltified begins.
725 2b0ae357 2022-01-10 thomas * For example, we want to avoid deltifying a Git object header at
726 2b0ae357 2022-01-10 thomas * the beginning of the file.
728 2b0ae357 2022-01-10 thomas if (fseeko(f, offset0, SEEK_SET) == -1)
729 2b0ae357 2022-01-10 thomas return got_error_from_errno("fseeko");
731 2b0ae357 2022-01-10 thomas *deltas = reallocarray(NULL, alloc_chunk_size,
732 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_instruction));
733 2b0ae357 2022-01-10 thomas if (*deltas == NULL)
734 2b0ae357 2022-01-10 thomas return got_error_from_errno("reallocarray");
735 2b0ae357 2022-01-10 thomas nalloc = alloc_chunk_size;
737 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
738 2b0ae357 2022-01-10 thomas uint8_t buf[GOT_DELTIFY_MAXCHUNK];
739 2b0ae357 2022-01-10 thomas off_t blocklen;
740 2b0ae357 2022-01-10 thomas struct got_delta_block *block;
741 2b0ae357 2022-01-10 thomas err = nextblk(buf, &blocklen, f);
744 2b0ae357 2022-01-10 thomas if (blocklen == 0) {
745 2b0ae357 2022-01-10 thomas /* Source remainder from the file itself. */
746 2b0ae357 2022-01-10 thomas if (fileoffset < filesize) {
747 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
748 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0,
749 2b0ae357 2022-01-10 thomas filesize - fileoffset);
753 cc524d36 2022-05-31 thomas err = lookupblk_mem(&block, dt, buf, blocklen, seed, basedata,
754 2b0ae357 2022-01-10 thomas basefile_offset0);
757 2b0ae357 2022-01-10 thomas if (block != NULL) {
759 2b0ae357 2022-01-10 thomas * We have found a matching block in the delta base.
760 2b0ae357 2022-01-10 thomas * Attempt to stretch the block as far as possible and
761 2b0ae357 2022-01-10 thomas * generate a copy instruction.
763 2b0ae357 2022-01-10 thomas err = stretchblk_file_mem(basedata, basefile_offset0,
764 2b0ae357 2022-01-10 thomas basefile_size, block, f, filesize, &blocklen);
767 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
768 2b0ae357 2022-01-10 thomas alloc_chunk_size, 1, block->offset, blocklen);
773 2b0ae357 2022-01-10 thomas * No match.
774 2b0ae357 2022-01-10 thomas * This block needs to be sourced from the file itself.
776 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
777 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0, blocklen);
781 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
782 2b0ae357 2022-01-10 thomas if (fseeko(f, fileoffset, SEEK_SET) == -1) {
783 2b0ae357 2022-01-10 thomas err = got_error_from_errno("fseeko");
788 2b0ae357 2022-01-10 thomas if (err) {
789 2b0ae357 2022-01-10 thomas free(*deltas);
790 2b0ae357 2022-01-10 thomas *deltas = NULL;
791 2b0ae357 2022-01-10 thomas *ndeltas = 0;
793 2b0ae357 2022-01-10 thomas return err;
796 2b0ae357 2022-01-10 thomas const struct got_error *
797 2b0ae357 2022-01-10 thomas got_deltify_mem_file(struct got_delta_instruction **deltas, int *ndeltas,
798 cc524d36 2022-05-31 thomas uint8_t *data, off_t fileoffset, off_t filesize, uint32_t seed,
799 2b0ae357 2022-01-10 thomas struct got_delta_table *dt, FILE *basefile,
800 2b0ae357 2022-01-10 thomas off_t basefile_offset0, off_t basefile_size)
802 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
803 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
804 2b0ae357 2022-01-10 thomas size_t nalloc = 0;
805 2b0ae357 2022-01-10 thomas const size_t alloc_chunk_size = 64;
807 2b0ae357 2022-01-10 thomas *deltas = NULL;
808 2b0ae357 2022-01-10 thomas *ndeltas = 0;
810 2b0ae357 2022-01-10 thomas *deltas = reallocarray(NULL, alloc_chunk_size,
811 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_instruction));
812 2b0ae357 2022-01-10 thomas if (*deltas == NULL)
813 2b0ae357 2022-01-10 thomas return got_error_from_errno("reallocarray");
814 2b0ae357 2022-01-10 thomas nalloc = alloc_chunk_size;
816 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
817 2b0ae357 2022-01-10 thomas off_t blocklen;
818 2b0ae357 2022-01-10 thomas struct got_delta_block *block;
819 2b0ae357 2022-01-10 thomas err = nextblk_mem(&blocklen, data, fileoffset, filesize);
822 2b0ae357 2022-01-10 thomas if (blocklen == 0) {
823 2b0ae357 2022-01-10 thomas /* Source remainder from the file itself. */
824 2b0ae357 2022-01-10 thomas if (fileoffset < filesize) {
825 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
826 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0,
827 2b0ae357 2022-01-10 thomas filesize - fileoffset);
831 cc524d36 2022-05-31 thomas err = lookupblk(&block, dt, data + fileoffset, blocklen, seed,
832 2b0ae357 2022-01-10 thomas basefile, basefile_offset0);
835 2b0ae357 2022-01-10 thomas if (block != NULL) {
837 2b0ae357 2022-01-10 thomas * We have found a matching block in the delta base.
838 2b0ae357 2022-01-10 thomas * Attempt to stretch the block as far as possible and
839 2b0ae357 2022-01-10 thomas * generate a copy instruction.
841 2b0ae357 2022-01-10 thomas err = stretchblk_mem_file(basefile, basefile_offset0,
842 2b0ae357 2022-01-10 thomas block, data, fileoffset + blocklen, filesize,
843 2b0ae357 2022-01-10 thomas &blocklen);
846 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
847 2b0ae357 2022-01-10 thomas alloc_chunk_size, 1, block->offset, blocklen);
852 2b0ae357 2022-01-10 thomas * No match.
853 2b0ae357 2022-01-10 thomas * This block needs to be sourced from the file itself.
855 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
856 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0, blocklen);
860 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
863 2b0ae357 2022-01-10 thomas if (err) {
864 2b0ae357 2022-01-10 thomas free(*deltas);
865 2b0ae357 2022-01-10 thomas *deltas = NULL;
866 2b0ae357 2022-01-10 thomas *ndeltas = 0;
868 2b0ae357 2022-01-10 thomas return err;
871 2b0ae357 2022-01-10 thomas const struct got_error *
872 2b0ae357 2022-01-10 thomas got_deltify_mem_mem(struct got_delta_instruction **deltas, int *ndeltas,
873 cc524d36 2022-05-31 thomas uint8_t *data, off_t fileoffset, off_t filesize, uint32_t seed,
874 2b0ae357 2022-01-10 thomas struct got_delta_table *dt, uint8_t *basedata,
875 2b0ae357 2022-01-10 thomas off_t basefile_offset0, off_t basefile_size)
877 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
878 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
879 2b0ae357 2022-01-10 thomas size_t nalloc = 0;
880 2b0ae357 2022-01-10 thomas const size_t alloc_chunk_size = 64;
882 2b0ae357 2022-01-10 thomas *deltas = NULL;
883 2b0ae357 2022-01-10 thomas *ndeltas = 0;
885 2b0ae357 2022-01-10 thomas *deltas = reallocarray(NULL, alloc_chunk_size,
886 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_instruction));
887 2b0ae357 2022-01-10 thomas if (*deltas == NULL)
888 2b0ae357 2022-01-10 thomas return got_error_from_errno("reallocarray");
889 2b0ae357 2022-01-10 thomas nalloc = alloc_chunk_size;
891 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
892 2b0ae357 2022-01-10 thomas off_t blocklen;
893 2b0ae357 2022-01-10 thomas struct got_delta_block *block;
894 2b0ae357 2022-01-10 thomas err = nextblk_mem(&blocklen, data, fileoffset, filesize);
897 2b0ae357 2022-01-10 thomas if (blocklen == 0) {
898 2b0ae357 2022-01-10 thomas /* Source remainder from the file itself. */
899 2b0ae357 2022-01-10 thomas if (fileoffset < filesize) {
900 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
901 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0,
902 2b0ae357 2022-01-10 thomas filesize - fileoffset);
906 2b0ae357 2022-01-10 thomas err = lookupblk_mem(&block, dt, data + fileoffset, blocklen,
907 cc524d36 2022-05-31 thomas seed, basedata, basefile_offset0);
910 2b0ae357 2022-01-10 thomas if (block != NULL) {
912 2b0ae357 2022-01-10 thomas * We have found a matching block in the delta base.
913 2b0ae357 2022-01-10 thomas * Attempt to stretch the block as far as possible and
914 2b0ae357 2022-01-10 thomas * generate a copy instruction.
916 2b0ae357 2022-01-10 thomas err = stretchblk_mem_mem(basedata, basefile_offset0,
917 2b0ae357 2022-01-10 thomas basefile_size, block, data, fileoffset + blocklen,
918 2b0ae357 2022-01-10 thomas filesize, &blocklen);
921 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
922 2b0ae357 2022-01-10 thomas alloc_chunk_size, 1, block->offset, blocklen);
927 2b0ae357 2022-01-10 thomas * No match.
928 2b0ae357 2022-01-10 thomas * This block needs to be sourced from the file itself.
930 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
931 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0, blocklen);
935 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
938 2b0ae357 2022-01-10 thomas if (err) {
939 2b0ae357 2022-01-10 thomas free(*deltas);
940 2b0ae357 2022-01-10 thomas *deltas = NULL;
941 2b0ae357 2022-01-10 thomas *ndeltas = 0;
943 2b0ae357 2022-01-10 thomas return err;