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 8b925c6c 2022-07-16 thomas #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>
29 f364801d 2021-03-10 stsp #include "got_error.h"
31 69aa0e90 2021-03-10 stsp #include "got_lib_deltify.h"
32 b7bf7afc 2022-02-12 thomas #include "murmurhash2.h"
35 69aa0e90 2021-03-10 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
39 2539ccf2 2021-10-08 thomas * The algorihm used here is FastCDC (Fast Content-Defined Chunking)
40 2539ccf2 2021-10-08 thomas * https://www.usenix.org/conference/atc16/technical-sessions/presentation/xia
43 fa8129f7 2022-03-22 thomas static const uint32_t geartab[256] = {
44 69aa0e90 2021-03-10 stsp 0x67ed26b7, 0x32da500c, 0x53d0fee0, 0xce387dc7, 0xcd406d90, 0x2e83a4d4,
45 69aa0e90 2021-03-10 stsp 0x9fc9a38d, 0xb67259dc, 0xca6b1722, 0x6d2ea08c, 0x235cea2e, 0x3149bb5f,
46 69aa0e90 2021-03-10 stsp 0x1beda787, 0x2a6b77d5, 0x2f22d9ac, 0x91fc0544, 0xe413acfa, 0x5a30ff7a,
47 69aa0e90 2021-03-10 stsp 0xad6fdde0, 0x444fd0f5, 0x7ad87864, 0x58c5ff05, 0x8d2ec336, 0x2371f853,
48 69aa0e90 2021-03-10 stsp 0x550f8572, 0x6aa448dd, 0x7c9ddbcf, 0x95221e14, 0x2a82ec33, 0xcbec5a78,
49 69aa0e90 2021-03-10 stsp 0xc6795a0d, 0x243995b7, 0x1c909a2f, 0x4fded51c, 0x635d334b, 0x0e2b9999,
50 69aa0e90 2021-03-10 stsp 0x2702968d, 0x856de1d5, 0x3325d60e, 0xeb6a7502, 0xec2a9844, 0x0905835a,
51 69aa0e90 2021-03-10 stsp 0xa1820375, 0xa4be5cab, 0x96a6c058, 0x2c2ccd70, 0xba40fce3, 0xd794c46b,
52 69aa0e90 2021-03-10 stsp 0x8fbae83e, 0xc3aa7899, 0x3d3ff8ed, 0xa0d42b5b, 0x571c0c97, 0xd2811516,
53 69aa0e90 2021-03-10 stsp 0xf7e7b96c, 0x4fd2fcbd, 0xe2fdec94, 0x282cc436, 0x78e8e95c, 0x80a3b613,
54 69aa0e90 2021-03-10 stsp 0xcfbee20c, 0xd4a32d1c, 0x2a12ff13, 0x6af82936, 0xe5630258, 0x8efa6a98,
55 69aa0e90 2021-03-10 stsp 0x294fb2d1, 0xdeb57086, 0x5f0fddb3, 0xeceda7ce, 0x4c87305f, 0x3a6d3307,
56 69aa0e90 2021-03-10 stsp 0xe22d2942, 0x9d060217, 0x1e42ed02, 0xb6f63b52, 0x4367f39f, 0x055cf262,
57 69aa0e90 2021-03-10 stsp 0x03a461b2, 0x5ef9e382, 0x386bc03a, 0x2a1e79c7, 0xf1a0058b, 0xd4d2dea9,
58 69aa0e90 2021-03-10 stsp 0x56baf37d, 0x5daff6cc, 0xf03a951d, 0xaef7de45, 0xa8f4581e, 0x3960b555,
59 69aa0e90 2021-03-10 stsp 0xffbfff6d, 0xbe702a23, 0x8f5b6d6f, 0x061739fb, 0x98696f47, 0x3fd596d4,
60 69aa0e90 2021-03-10 stsp 0x151eac6b, 0xa9fcc4f5, 0x69181a12, 0x3ac5a107, 0xb5198fe7, 0x96bcb1da,
61 69aa0e90 2021-03-10 stsp 0x1b5ddf8e, 0xc757d650, 0x65865c3a, 0x8fc0a41a, 0x87435536, 0x99eda6f2,
62 69aa0e90 2021-03-10 stsp 0x41874794, 0x29cff4e8, 0xb70efd9a, 0x3103f6e7, 0x84d2453b, 0x15a450bd,
63 69aa0e90 2021-03-10 stsp 0x74f49af1, 0x60f664b1, 0xa1c86935, 0xfdafbce1, 0xe36353e3, 0x5d9ba739,
64 69aa0e90 2021-03-10 stsp 0xbc0559ba, 0x708b0054, 0xd41d808c, 0xb2f31723, 0x9027c41f, 0xf136d165,
65 69aa0e90 2021-03-10 stsp 0xb5374b12, 0x9420a6ac, 0x273958b6, 0xe6c2fad0, 0xebdc1f21, 0xfb33af8b,
66 69aa0e90 2021-03-10 stsp 0xc71c25cd, 0xe9a2d8e5, 0xbeb38a50, 0xbceb7cc2, 0x4e4e73f0, 0xcd6c251d,
67 69aa0e90 2021-03-10 stsp 0xde4c032c, 0x4b04ac30, 0x725b8b21, 0x4eb8c33b, 0x20d07b75, 0x0567aa63,
68 69aa0e90 2021-03-10 stsp 0xb56b2bb7, 0xc1f5fd3a, 0xcafd35ca, 0x470dd4da, 0xfe4f94cd, 0xfb8de424,
69 69aa0e90 2021-03-10 stsp 0xe8dbcf40, 0xfe50a37a, 0x62db5b5d, 0xf32f4ab6, 0x2c4a8a51, 0x18473dc0,
70 69aa0e90 2021-03-10 stsp 0xfe0cbb6e, 0xfe399efd, 0xdf34ecc9, 0x6ccd5055, 0x46097073, 0x139135c2,
71 69aa0e90 2021-03-10 stsp 0x721c76f6, 0x1c6a94b4, 0x6eee014d, 0x8a508e02, 0x3da538f5, 0x280d394f,
72 69aa0e90 2021-03-10 stsp 0x5248a0c4, 0x3ce94c6c, 0x9a71ad3a, 0x8493dd05, 0xe43f0ab6, 0x18e4ed42,
73 69aa0e90 2021-03-10 stsp 0x6c5c0e09, 0x42b06ec9, 0x8d330343, 0xa45b6f59, 0x2a573c0c, 0xd7fd3de6,
74 69aa0e90 2021-03-10 stsp 0xeedeab68, 0x5c84dafc, 0xbbd1b1a8, 0xa3ce1ad1, 0x85b70bed, 0xb6add07f,
75 69aa0e90 2021-03-10 stsp 0xa531309c, 0x8f8ab852, 0x564de332, 0xeac9ed0c, 0x73da402c, 0x3ec52761,
76 69aa0e90 2021-03-10 stsp 0x43af2f4d, 0xd6ff45c8, 0x4c367462, 0xd553bd6a, 0x44724855, 0x3b2aa728,
77 69aa0e90 2021-03-10 stsp 0x56e5eb65, 0xeaf16173, 0x33fa42ff, 0xd714bb5d, 0xfbd0a3b9, 0xaf517134,
78 69aa0e90 2021-03-10 stsp 0x9416c8cd, 0x534cf94f, 0x548947c2, 0x34193569, 0x32f4389a, 0xfe7028bc,
79 69aa0e90 2021-03-10 stsp 0xed73b1ed, 0x9db95770, 0x468e3922, 0x0440c3cd, 0x60059a62, 0x33504562,
80 69aa0e90 2021-03-10 stsp 0x2b229fbd, 0x5174dca5, 0xf7028752, 0xd63c6aa8, 0x31276f38, 0x0646721c,
81 69aa0e90 2021-03-10 stsp 0xb0191da8, 0xe00e6de0, 0x9eac1a6e, 0x9f7628a5, 0xed6c06ea, 0x0bb8af15,
82 69aa0e90 2021-03-10 stsp 0xf119fb12, 0x38693c1c, 0x732bc0fe, 0x84953275, 0xb82ec888, 0x33a4f1b3,
83 69aa0e90 2021-03-10 stsp 0x3099835e, 0x028a8782, 0x5fdd51d7, 0xc6c717b3, 0xb06caf71, 0x17c8c111,
84 69aa0e90 2021-03-10 stsp 0x61bad754, 0x9fd03061, 0xe09df1af, 0x3bc9eb73, 0x85878413, 0x9889aaf2,
85 69aa0e90 2021-03-10 stsp 0x3f5a9e46, 0x42c9f01f, 0x9984a4f4, 0xd5de43cc, 0xd294daed, 0xbecba2d2,
86 69aa0e90 2021-03-10 stsp 0xf1f6e72c, 0x5551128a, 0x83af87e2, 0x6f0342ba,
89 b7bf7afc 2022-02-12 thomas static uint32_t
90 cc524d36 2022-05-31 thomas hashblk(const unsigned char *p, off_t n, uint32_t seed)
92 cc524d36 2022-05-31 thomas return murmurhash2(p, n, seed);
95 f364801d 2021-03-10 stsp static const struct got_error *
96 740bba1c 2021-06-18 stsp addblk(struct got_delta_table *dt, FILE *f, off_t file_offset0, off_t len,
97 90dfa2bf 2022-02-13 thomas off_t offset, uint32_t h)
99 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
101 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
102 0d15f6dc 2021-06-13 stsp uint8_t buf2[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) {
111 69aa0e90 2021-03-10 stsp * Avoid adding duplicate blocks.
112 69aa0e90 2021-03-10 stsp * NB: A matching hash is insufficient for detecting equality.
113 51a494da 2021-06-11 stsp * The hash can only detect inequality.
115 51a494da 2021-06-11 stsp if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
116 0d15f6dc 2021-06-13 stsp if (r == 0) {
117 740bba1c 2021-06-18 stsp if (fseeko(f, file_offset0 + offset, SEEK_SET) == -1)
118 0d15f6dc 2021-06-13 stsp return got_error_from_errno("fseeko");
119 0d15f6dc 2021-06-13 stsp r = fread(buf, 1, len, f);
120 0d15f6dc 2021-06-13 stsp if (r != len) {
121 0d15f6dc 2021-06-13 stsp if (!ferror(f))
122 0d15f6dc 2021-06-13 stsp return NULL;
123 0d15f6dc 2021-06-13 stsp return got_ferror(f, GOT_ERR_IO);
126 740bba1c 2021-06-18 stsp if (fseeko(f, file_offset0 + dt->blocks[i].offset,
127 740bba1c 2021-06-18 stsp 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 e89540a9 2021-06-11 stsp if (dt->nalloc < dt->nblocks + 64) {
143 69aa0e90 2021-03-10 stsp struct got_delta_block *db;
144 e89540a9 2021-06-11 stsp size_t old_size = dt->nalloc;
145 69aa0e90 2021-03-10 stsp db = dt->blocks;
146 e89540a9 2021-06-11 stsp dt->blocks = calloc(dt->nalloc + 64,
147 e89540a9 2021-06-11 stsp sizeof(struct got_delta_block));
148 69aa0e90 2021-03-10 stsp if (dt->blocks == NULL) {
149 69aa0e90 2021-03-10 stsp err = got_error_from_errno("calloc");
150 69aa0e90 2021-03-10 stsp dt->blocks = db;
151 69aa0e90 2021-03-10 stsp return err;
153 e89540a9 2021-06-11 stsp dt->nalloc += 64;
155 69aa0e90 2021-03-10 stsp * Recompute all block positions. Hash-based indices of blocks
156 69aa0e90 2021-03-10 stsp * in the array depend on the allocated length of the array.
158 69aa0e90 2021-03-10 stsp dt->nblocks = 0;
159 a893025f 2021-06-11 stsp for (i = 0; i < old_size; i++) {
160 69aa0e90 2021-03-10 stsp if (db[i].len == 0)
162 740bba1c 2021-06-18 stsp err = addblk(dt, f, file_offset0, db[i].len,
163 2b0ae357 2022-01-10 thomas db[i].offset, db[i].hash);
167 2b0ae357 2022-01-10 thomas free(db);
170 2b0ae357 2022-01-10 thomas return err;
173 2b0ae357 2022-01-10 thomas static const struct got_error *
174 2b0ae357 2022-01-10 thomas addblk_mem(struct got_delta_table *dt, uint8_t *data, off_t file_offset0,
175 90dfa2bf 2022-02-13 thomas off_t len, off_t offset, uint32_t h)
177 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
179 2b0ae357 2022-01-10 thomas uint8_t *block1;
180 2b0ae357 2022-01-10 thomas uint8_t *block2;
182 2b0ae357 2022-01-10 thomas if (len == 0)
183 2b0ae357 2022-01-10 thomas return NULL;
185 2b0ae357 2022-01-10 thomas i = h % dt->nalloc;
186 2b0ae357 2022-01-10 thomas while (dt->blocks[i].len != 0) {
188 2b0ae357 2022-01-10 thomas * Avoid adding duplicate blocks.
189 2b0ae357 2022-01-10 thomas * NB: A matching hash is insufficient for detecting equality.
190 2b0ae357 2022-01-10 thomas * The hash can only detect inequality.
192 2b0ae357 2022-01-10 thomas if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
193 2b0ae357 2022-01-10 thomas block1 = data + file_offset0 + dt->blocks[i].offset;
194 2b0ae357 2022-01-10 thomas block2 = data + file_offset0 + offset;
195 2b0ae357 2022-01-10 thomas if (memcmp(block1, block2, len) == 0)
196 2b0ae357 2022-01-10 thomas return NULL;
199 2b0ae357 2022-01-10 thomas i = (i + 1) % dt->nalloc;
201 2b0ae357 2022-01-10 thomas assert(dt->blocks[i].len == 0);
202 2b0ae357 2022-01-10 thomas dt->blocks[i].len = len;
203 2b0ae357 2022-01-10 thomas dt->blocks[i].offset = offset;
204 2b0ae357 2022-01-10 thomas dt->blocks[i].hash = h;
205 2b0ae357 2022-01-10 thomas dt->nblocks++;
206 2b0ae357 2022-01-10 thomas if (dt->nalloc < dt->nblocks + 64) {
207 2b0ae357 2022-01-10 thomas struct got_delta_block *db;
208 2b0ae357 2022-01-10 thomas size_t old_size = dt->nalloc;
209 2b0ae357 2022-01-10 thomas db = dt->blocks;
210 2b0ae357 2022-01-10 thomas dt->blocks = calloc(dt->nalloc + 64,
211 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_block));
212 2b0ae357 2022-01-10 thomas if (dt->blocks == NULL) {
213 2b0ae357 2022-01-10 thomas err = got_error_from_errno("calloc");
214 2b0ae357 2022-01-10 thomas dt->blocks = db;
215 2b0ae357 2022-01-10 thomas return err;
217 2b0ae357 2022-01-10 thomas dt->nalloc += 64;
219 2b0ae357 2022-01-10 thomas * Recompute all block positions. Hash-based indices of blocks
220 2b0ae357 2022-01-10 thomas * in the array depend on the allocated length of the array.
222 2b0ae357 2022-01-10 thomas dt->nblocks = 0;
223 2b0ae357 2022-01-10 thomas for (i = 0; i < old_size; i++) {
224 2b0ae357 2022-01-10 thomas if (db[i].len == 0)
225 2b0ae357 2022-01-10 thomas continue;
226 2b0ae357 2022-01-10 thomas err = addblk_mem(dt, data, file_offset0, db[i].len,
227 740bba1c 2021-06-18 stsp db[i].offset, db[i].hash);
234 69aa0e90 2021-03-10 stsp return err;
237 69aa0e90 2021-03-10 stsp static const struct got_error *
238 69aa0e90 2021-03-10 stsp lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
239 cc524d36 2022-05-31 thomas unsigned char *p, off_t len, uint32_t seed, FILE *basefile,
240 cc524d36 2022-05-31 thomas off_t basefile_offset0)
243 b7bf7afc 2022-02-12 thomas uint32_t h;
244 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
247 69aa0e90 2021-03-10 stsp *block = NULL;
249 cc524d36 2022-05-31 thomas h = hashblk(p, len, seed);
250 69aa0e90 2021-03-10 stsp for (i = h % dt->nalloc; dt->blocks[i].len != 0;
251 69aa0e90 2021-03-10 stsp i = (i + 1) % dt->nalloc) {
252 69aa0e90 2021-03-10 stsp if (dt->blocks[i].hash != h ||
253 69aa0e90 2021-03-10 stsp dt->blocks[i].len != len)
255 740bba1c 2021-06-18 stsp if (fseeko(basefile, basefile_offset0 + dt->blocks[i].offset,
256 740bba1c 2021-06-18 stsp SEEK_SET) == -1)
257 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
258 69aa0e90 2021-03-10 stsp r = fread(buf, 1, len, basefile);
259 69aa0e90 2021-03-10 stsp if (r != len)
260 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
261 69aa0e90 2021-03-10 stsp if (memcmp(p, buf, len) == 0) {
262 69aa0e90 2021-03-10 stsp *block = &dt->blocks[i];
266 f364801d 2021-03-10 stsp return NULL;
269 69aa0e90 2021-03-10 stsp static const struct got_error *
270 2b0ae357 2022-01-10 thomas lookupblk_mem(struct got_delta_block **block, struct got_delta_table *dt,
271 cc524d36 2022-05-31 thomas unsigned char *p, off_t len, uint32_t seed, uint8_t *basedata,
272 cc524d36 2022-05-31 thomas off_t basefile_offset0)
275 b7bf7afc 2022-02-12 thomas uint32_t h;
276 2b0ae357 2022-01-10 thomas uint8_t *b;
278 2b0ae357 2022-01-10 thomas *block = NULL;
280 cc524d36 2022-05-31 thomas h = hashblk(p, len, seed);
281 2b0ae357 2022-01-10 thomas for (i = h % dt->nalloc; dt->blocks[i].len != 0;
282 2b0ae357 2022-01-10 thomas i = (i + 1) % dt->nalloc) {
283 2b0ae357 2022-01-10 thomas if (dt->blocks[i].hash != h ||
284 2b0ae357 2022-01-10 thomas dt->blocks[i].len != len)
285 2b0ae357 2022-01-10 thomas continue;
286 2b0ae357 2022-01-10 thomas b = basedata + basefile_offset0 + dt->blocks[i].offset;
287 2b0ae357 2022-01-10 thomas if (memcmp(p, b, len) == 0) {
288 2b0ae357 2022-01-10 thomas *block = &dt->blocks[i];
292 2b0ae357 2022-01-10 thomas return NULL;
295 2b0ae357 2022-01-10 thomas static const struct got_error *
296 69aa0e90 2021-03-10 stsp nextblk(uint8_t *buf, off_t *blocklen, FILE *f)
298 f364801d 2021-03-10 stsp uint32_t gh;
299 69aa0e90 2021-03-10 stsp const unsigned char *p;
301 69aa0e90 2021-03-10 stsp off_t pos = ftello(f);
303 69aa0e90 2021-03-10 stsp *blocklen = 0;
305 69aa0e90 2021-03-10 stsp r = fread(buf, 1, GOT_DELTIFY_MAXCHUNK, f);
306 69aa0e90 2021-03-10 stsp if (r == 0 && ferror(f))
307 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
308 69aa0e90 2021-03-10 stsp if (r < GOT_DELTIFY_MINCHUNK)
309 69aa0e90 2021-03-10 stsp return NULL; /* no more delta-worthy blocks left */
311 69aa0e90 2021-03-10 stsp /* Got a deltifiable block. Find the split-point where it ends. */
312 69aa0e90 2021-03-10 stsp p = buf + GOT_DELTIFY_MINCHUNK;
314 69aa0e90 2021-03-10 stsp while (p != buf + r) {
315 f364801d 2021-03-10 stsp gh = (gh << 1) + geartab[*p++];
316 69aa0e90 2021-03-10 stsp if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
320 69aa0e90 2021-03-10 stsp *blocklen = (p - buf);
321 69aa0e90 2021-03-10 stsp if (fseeko(f, pos + *blocklen, SEEK_SET) == -1)
322 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
324 69aa0e90 2021-03-10 stsp return NULL;
327 2b0ae357 2022-01-10 thomas static const struct got_error *
328 2b0ae357 2022-01-10 thomas nextblk_mem(off_t *blocklen, uint8_t *data, off_t fileoffset, off_t filesize)
330 2b0ae357 2022-01-10 thomas uint32_t gh;
331 2b0ae357 2022-01-10 thomas const unsigned char *p;
333 2b0ae357 2022-01-10 thomas *blocklen = 0;
335 2b0ae357 2022-01-10 thomas if (fileoffset >= filesize ||
336 2b0ae357 2022-01-10 thomas filesize - fileoffset < GOT_DELTIFY_MINCHUNK)
337 2b0ae357 2022-01-10 thomas return NULL; /* no more delta-worthy blocks left */
339 2b0ae357 2022-01-10 thomas /* Got a deltifiable block. Find the split-point where it ends. */
340 2b0ae357 2022-01-10 thomas p = data + fileoffset + GOT_DELTIFY_MINCHUNK;
342 2b0ae357 2022-01-10 thomas while (p != data + MIN(fileoffset + GOT_DELTIFY_MAXCHUNK, filesize)) {
343 2b0ae357 2022-01-10 thomas gh = (gh << 1) + geartab[*p++];
344 2b0ae357 2022-01-10 thomas if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
348 2b0ae357 2022-01-10 thomas *blocklen = (p - (data + fileoffset));
349 2b0ae357 2022-01-10 thomas return NULL;
352 f364801d 2021-03-10 stsp const struct got_error *
353 69aa0e90 2021-03-10 stsp got_deltify_init(struct got_delta_table **dt, FILE *f, off_t fileoffset,
354 cc524d36 2022-05-31 thomas off_t filesize, uint32_t seed)
356 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
357 b7bf7afc 2022-02-12 thomas uint32_t h;
358 740bba1c 2021-06-18 stsp const off_t offset0 = fileoffset;
360 69aa0e90 2021-03-10 stsp *dt = calloc(1, sizeof(**dt));
361 69aa0e90 2021-03-10 stsp if (*dt == NULL)
362 f364801d 2021-03-10 stsp return got_error_from_errno("calloc");
364 69aa0e90 2021-03-10 stsp (*dt)->nblocks = 0;
365 69aa0e90 2021-03-10 stsp (*dt)->nalloc = 128;
366 69aa0e90 2021-03-10 stsp (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
367 69aa0e90 2021-03-10 stsp if ((*dt)->blocks == NULL) {
368 69aa0e90 2021-03-10 stsp err = got_error_from_errno("calloc");
372 69aa0e90 2021-03-10 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
373 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
375 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
376 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
377 69aa0e90 2021-03-10 stsp off_t blocklen;
378 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
381 69aa0e90 2021-03-10 stsp if (blocklen == 0)
383 cc524d36 2022-05-31 thomas h = hashblk(buf, blocklen, seed);
384 740bba1c 2021-06-18 stsp err = addblk(*dt, f, offset0, blocklen,
385 740bba1c 2021-06-18 stsp fileoffset - offset0, h);
388 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
389 68bdcdc2 2021-06-11 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
390 68bdcdc2 2021-06-11 stsp return got_error_from_errno("fseeko");
393 2b0ae357 2022-01-10 thomas if (err) {
394 2b0ae357 2022-01-10 thomas free((*dt)->blocks);
395 2b0ae357 2022-01-10 thomas free(*dt);
396 2b0ae357 2022-01-10 thomas *dt = NULL;
399 2b0ae357 2022-01-10 thomas return err;
402 2b0ae357 2022-01-10 thomas const struct got_error *
403 2b0ae357 2022-01-10 thomas got_deltify_init_mem(struct got_delta_table **dt, uint8_t *data,
404 cc524d36 2022-05-31 thomas off_t fileoffset, off_t filesize, uint32_t seed)
406 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
407 90dfa2bf 2022-02-13 thomas uint32_t h;
408 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
410 2b0ae357 2022-01-10 thomas *dt = calloc(1, sizeof(**dt));
411 2b0ae357 2022-01-10 thomas if (*dt == NULL)
412 2b0ae357 2022-01-10 thomas return got_error_from_errno("calloc");
414 2b0ae357 2022-01-10 thomas (*dt)->nblocks = 0;
415 2b0ae357 2022-01-10 thomas (*dt)->nalloc = 128;
416 2b0ae357 2022-01-10 thomas (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
417 2b0ae357 2022-01-10 thomas if ((*dt)->blocks == NULL) {
418 2b0ae357 2022-01-10 thomas err = got_error_from_errno("calloc");
419 2b0ae357 2022-01-10 thomas goto done;
422 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
423 2b0ae357 2022-01-10 thomas off_t blocklen;
424 2b0ae357 2022-01-10 thomas err = nextblk_mem(&blocklen, data, fileoffset, filesize);
426 2b0ae357 2022-01-10 thomas goto done;
427 2b0ae357 2022-01-10 thomas if (blocklen == 0)
429 cc524d36 2022-05-31 thomas h = hashblk(data + fileoffset, blocklen, seed);
430 2b0ae357 2022-01-10 thomas err = addblk_mem(*dt, data, offset0, blocklen,
431 2b0ae357 2022-01-10 thomas fileoffset - offset0, h);
433 2b0ae357 2022-01-10 thomas goto done;
434 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
438 69aa0e90 2021-03-10 stsp free((*dt)->blocks);
440 69aa0e90 2021-03-10 stsp *dt = NULL;
443 69aa0e90 2021-03-10 stsp return err;
447 69aa0e90 2021-03-10 stsp got_deltify_free(struct got_delta_table *dt)
449 dbbf4a5f 2021-05-20 stsp if (dt == NULL)
451 69aa0e90 2021-03-10 stsp free(dt->blocks);
455 f364801d 2021-03-10 stsp static const struct got_error *
456 dd29967c 2021-08-22 stsp emitdelta(struct got_delta_instruction **deltas, size_t *nalloc, int *ndeltas,
457 dd29967c 2021-08-22 stsp const size_t alloc_chunk_size, int copy, off_t offset, off_t len)
459 69aa0e90 2021-03-10 stsp struct got_delta_instruction *d, *p;
461 dd29967c 2021-08-22 stsp if (*nalloc < *ndeltas + alloc_chunk_size) {
462 dd29967c 2021-08-22 stsp p = reallocarray(*deltas, *nalloc + alloc_chunk_size,
463 dd29967c 2021-08-22 stsp sizeof(struct got_delta_instruction));
464 dd29967c 2021-08-22 stsp if (p == NULL)
465 7d76c4e7 2021-10-15 thomas return got_error_from_errno("reallocarray");
466 dd29967c 2021-08-22 stsp *deltas = p;
467 dd29967c 2021-08-22 stsp *nalloc += alloc_chunk_size;
469 69aa0e90 2021-03-10 stsp *ndeltas += 1;
470 69aa0e90 2021-03-10 stsp d = &(*deltas)[*ndeltas - 1];
471 69aa0e90 2021-03-10 stsp d->copy = copy;
472 69aa0e90 2021-03-10 stsp d->offset = offset;
473 56e7a15f 2021-03-10 stsp d->len = len;
474 f364801d 2021-03-10 stsp return NULL;
477 69aa0e90 2021-03-10 stsp static const struct got_error *
478 740bba1c 2021-06-18 stsp stretchblk(FILE *basefile, off_t base_offset0, struct got_delta_block *block,
479 740bba1c 2021-06-18 stsp FILE *f, off_t filesize, off_t *blocklen)
481 69aa0e90 2021-03-10 stsp uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
482 69aa0e90 2021-03-10 stsp size_t base_r, r, i;
483 69aa0e90 2021-03-10 stsp int buf_equal = 1;
485 5de743f8 2021-08-29 stsp if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
486 5de743f8 2021-08-29 stsp SEEK_SET) == -1)
487 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
489 69aa0e90 2021-03-10 stsp while (buf_equal && *blocklen < (1 << 24) - 1) {
490 69aa0e90 2021-03-10 stsp base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
491 69aa0e90 2021-03-10 stsp if (base_r == 0) {
492 69aa0e90 2021-03-10 stsp if (ferror(basefile))
493 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
496 69aa0e90 2021-03-10 stsp r = fread(buf, 1, sizeof(buf), f);
497 69aa0e90 2021-03-10 stsp if (r == 0) {
498 69aa0e90 2021-03-10 stsp if (ferror(f))
499 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
502 69aa0e90 2021-03-10 stsp for (i = 0; i < MIN(base_r, r); i++) {
503 69aa0e90 2021-03-10 stsp if (buf[i] != basebuf[i]) {
504 69aa0e90 2021-03-10 stsp buf_equal = 0;
507 69aa0e90 2021-03-10 stsp (*blocklen)++;
511 2b0ae357 2022-01-10 thomas return NULL;
514 2b0ae357 2022-01-10 thomas static const struct got_error *
515 2b0ae357 2022-01-10 thomas stretchblk_file_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
516 2b0ae357 2022-01-10 thomas struct got_delta_block *block, FILE *f, off_t filesize, off_t *blocklen)
518 2b0ae357 2022-01-10 thomas uint8_t buf[GOT_DELTIFY_MAXCHUNK];
519 2b0ae357 2022-01-10 thomas size_t r, i;
520 2b0ae357 2022-01-10 thomas int buf_equal = 1;
521 2b0ae357 2022-01-10 thomas off_t base_offset = base_offset0 + block->offset + *blocklen;
523 2b0ae357 2022-01-10 thomas if (base_offset > basefile_size) {
524 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
525 e4a6dbc6 2022-07-24 thomas "read beyond the size of delta base at offset %lld",
526 e4a6dbc6 2022-07-24 thomas (long long)base_offset);
529 2b0ae357 2022-01-10 thomas while (buf_equal && *blocklen < (1 << 24) - 1) {
530 2b0ae357 2022-01-10 thomas if (base_offset + *blocklen >= basefile_size)
532 2b0ae357 2022-01-10 thomas r = fread(buf, 1, sizeof(buf), f);
533 2b0ae357 2022-01-10 thomas if (r == 0) {
534 2b0ae357 2022-01-10 thomas if (ferror(f))
535 2b0ae357 2022-01-10 thomas return got_ferror(f, GOT_ERR_IO);
538 2b0ae357 2022-01-10 thomas for (i = 0; i < MIN(basefile_size - base_offset, r); i++) {
539 2b0ae357 2022-01-10 thomas if (buf[i] != *(basedata + base_offset + i)) {
540 2b0ae357 2022-01-10 thomas buf_equal = 0;
543 2b0ae357 2022-01-10 thomas (*blocklen)++;
547 2b0ae357 2022-01-10 thomas return NULL;
550 2b0ae357 2022-01-10 thomas static const struct got_error *
551 2b0ae357 2022-01-10 thomas stretchblk_mem_file(FILE *basefile, off_t base_offset0,
552 2b0ae357 2022-01-10 thomas struct got_delta_block *block, uint8_t *data, off_t fileoffset,
553 2b0ae357 2022-01-10 thomas off_t filesize, off_t *blocklen)
555 2b0ae357 2022-01-10 thomas uint8_t basebuf[GOT_DELTIFY_MAXCHUNK];
556 2b0ae357 2022-01-10 thomas size_t base_r, i;
557 2b0ae357 2022-01-10 thomas int buf_equal = 1;
559 2b0ae357 2022-01-10 thomas if (fileoffset > filesize) {
560 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
561 e4a6dbc6 2022-07-24 thomas "read beyond the size of deltify file at offset %lld",
562 e4a6dbc6 2022-07-24 thomas (long long)fileoffset);
565 2b0ae357 2022-01-10 thomas if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
566 2b0ae357 2022-01-10 thomas SEEK_SET) == -1)
567 2b0ae357 2022-01-10 thomas return got_error_from_errno("fseeko");
569 2b0ae357 2022-01-10 thomas while (buf_equal && *blocklen < (1 << 24) - 1) {
570 2b0ae357 2022-01-10 thomas if (fileoffset + *blocklen >= filesize)
572 2b0ae357 2022-01-10 thomas base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
573 2b0ae357 2022-01-10 thomas if (base_r == 0) {
574 2b0ae357 2022-01-10 thomas if (ferror(basefile))
575 2b0ae357 2022-01-10 thomas return got_ferror(basefile, GOT_ERR_IO);
578 2b0ae357 2022-01-10 thomas for (i = 0; i < MIN(base_r, filesize - fileoffset); i++) {
579 2b0ae357 2022-01-10 thomas if (*(data + fileoffset + i) != basebuf[i]) {
580 2b0ae357 2022-01-10 thomas buf_equal = 0;
583 2b0ae357 2022-01-10 thomas (*blocklen)++;
587 2b0ae357 2022-01-10 thomas return NULL;
590 2b0ae357 2022-01-10 thomas static const struct got_error *
591 2b0ae357 2022-01-10 thomas stretchblk_mem_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
592 2b0ae357 2022-01-10 thomas struct got_delta_block *block, uint8_t *data, off_t fileoffset,
593 2b0ae357 2022-01-10 thomas off_t filesize, off_t *blocklen)
595 2b0ae357 2022-01-10 thomas off_t i, maxlen;
596 2b0ae357 2022-01-10 thomas off_t base_offset = base_offset0 + block->offset + *blocklen;
597 2b0ae357 2022-01-10 thomas uint8_t *p, *q;
599 2b0ae357 2022-01-10 thomas if (base_offset > basefile_size) {
600 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
601 e4a6dbc6 2022-07-24 thomas "read beyond the size of delta base at offset %lld",
602 e4a6dbc6 2022-07-24 thomas (long long)base_offset);
605 2b0ae357 2022-01-10 thomas if (fileoffset > filesize) {
606 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
607 e4a6dbc6 2022-07-24 thomas "read beyond the size of deltify file at offset %lld",
608 e4a6dbc6 2022-07-24 thomas (long long)fileoffset);
611 2b0ae357 2022-01-10 thomas p = data + fileoffset;
612 2b0ae357 2022-01-10 thomas q = basedata + base_offset;
613 2b0ae357 2022-01-10 thomas maxlen = MIN(basefile_size - base_offset, filesize - fileoffset);
614 2b0ae357 2022-01-10 thomas for (i = 0; i < maxlen && *blocklen < (1 << 24) - 1; i++) {
615 2b0ae357 2022-01-10 thomas if (p[i] != q[i])
617 2b0ae357 2022-01-10 thomas (*blocklen)++;
620 69aa0e90 2021-03-10 stsp return NULL;
623 69aa0e90 2021-03-10 stsp const struct got_error *
624 69aa0e90 2021-03-10 stsp got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
625 cc524d36 2022-05-31 thomas FILE *f, off_t fileoffset, off_t filesize, uint32_t seed,
626 69aa0e90 2021-03-10 stsp struct got_delta_table *dt, FILE *basefile,
627 f34b169e 2021-06-18 stsp off_t basefile_offset0, off_t basefile_size)
629 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
630 69aa0e90 2021-03-10 stsp const off_t offset0 = fileoffset;
631 dd29967c 2021-08-22 stsp size_t nalloc = 0;
632 dd29967c 2021-08-22 stsp const size_t alloc_chunk_size = 64;
634 69aa0e90 2021-03-10 stsp *deltas = NULL;
635 69aa0e90 2021-03-10 stsp *ndeltas = 0;
638 69aa0e90 2021-03-10 stsp * offset0 indicates where data to be deltified begins.
639 69aa0e90 2021-03-10 stsp * For example, we want to avoid deltifying a Git object header at
640 69aa0e90 2021-03-10 stsp * the beginning of the file.
642 69aa0e90 2021-03-10 stsp if (fseeko(f, offset0, SEEK_SET) == -1)
643 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
645 dd29967c 2021-08-22 stsp *deltas = reallocarray(NULL, alloc_chunk_size,
646 dd29967c 2021-08-22 stsp sizeof(struct got_delta_instruction));
647 dd29967c 2021-08-22 stsp if (*deltas == NULL)
648 dd29967c 2021-08-22 stsp return got_error_from_errno("reallocarray");
649 dd29967c 2021-08-22 stsp nalloc = alloc_chunk_size;
651 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
652 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
653 69aa0e90 2021-03-10 stsp off_t blocklen;
654 69aa0e90 2021-03-10 stsp struct got_delta_block *block;
655 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
658 9a8dc2b3 2021-06-18 stsp if (blocklen == 0) {
659 9a8dc2b3 2021-06-18 stsp /* Source remainder from the file itself. */
660 9a8dc2b3 2021-06-18 stsp if (fileoffset < filesize) {
661 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
662 dd29967c 2021-08-22 stsp alloc_chunk_size, 0, fileoffset - offset0,
663 9a8dc2b3 2021-06-18 stsp filesize - fileoffset);
667 cc524d36 2022-05-31 thomas err = lookupblk(&block, dt, buf, blocklen, seed, basefile,
668 740bba1c 2021-06-18 stsp basefile_offset0);
671 69aa0e90 2021-03-10 stsp if (block != NULL) {
673 69aa0e90 2021-03-10 stsp * We have found a matching block in the delta base.
674 69aa0e90 2021-03-10 stsp * Attempt to stretch the block as far as possible and
675 69aa0e90 2021-03-10 stsp * generate a copy instruction.
677 740bba1c 2021-06-18 stsp err = stretchblk(basefile, basefile_offset0, block,
678 740bba1c 2021-06-18 stsp f, filesize, &blocklen);
681 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
682 dd29967c 2021-08-22 stsp alloc_chunk_size, 1, block->offset, blocklen);
687 69aa0e90 2021-03-10 stsp * No match.
688 69aa0e90 2021-03-10 stsp * This block needs to be sourced from the file itself.
690 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
691 dd29967c 2021-08-22 stsp alloc_chunk_size, 0, fileoffset - offset0, blocklen);
695 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
696 0af64e86 2021-08-22 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1) {
697 0af64e86 2021-08-22 stsp err = got_error_from_errno("fseeko");
703 69aa0e90 2021-03-10 stsp free(*deltas);
704 69aa0e90 2021-03-10 stsp *deltas = NULL;
705 69aa0e90 2021-03-10 stsp *ndeltas = 0;
707 69aa0e90 2021-03-10 stsp return err;
710 2b0ae357 2022-01-10 thomas const struct got_error *
711 2b0ae357 2022-01-10 thomas got_deltify_file_mem(struct got_delta_instruction **deltas, int *ndeltas,
712 cc524d36 2022-05-31 thomas FILE *f, off_t fileoffset, off_t filesize, uint32_t seed,
713 2b0ae357 2022-01-10 thomas struct got_delta_table *dt, uint8_t *basedata,
714 2b0ae357 2022-01-10 thomas off_t basefile_offset0, off_t basefile_size)
716 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
717 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
718 2b0ae357 2022-01-10 thomas size_t nalloc = 0;
719 2b0ae357 2022-01-10 thomas const size_t alloc_chunk_size = 64;
721 2b0ae357 2022-01-10 thomas *deltas = NULL;
722 2b0ae357 2022-01-10 thomas *ndeltas = 0;
725 2b0ae357 2022-01-10 thomas * offset0 indicates where data to be deltified begins.
726 2b0ae357 2022-01-10 thomas * For example, we want to avoid deltifying a Git object header at
727 2b0ae357 2022-01-10 thomas * the beginning of the file.
729 2b0ae357 2022-01-10 thomas if (fseeko(f, offset0, SEEK_SET) == -1)
730 2b0ae357 2022-01-10 thomas return got_error_from_errno("fseeko");
732 2b0ae357 2022-01-10 thomas *deltas = reallocarray(NULL, alloc_chunk_size,
733 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_instruction));
734 2b0ae357 2022-01-10 thomas if (*deltas == NULL)
735 2b0ae357 2022-01-10 thomas return got_error_from_errno("reallocarray");
736 2b0ae357 2022-01-10 thomas nalloc = alloc_chunk_size;
738 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
739 2b0ae357 2022-01-10 thomas uint8_t buf[GOT_DELTIFY_MAXCHUNK];
740 2b0ae357 2022-01-10 thomas off_t blocklen;
741 2b0ae357 2022-01-10 thomas struct got_delta_block *block;
742 2b0ae357 2022-01-10 thomas err = nextblk(buf, &blocklen, f);
745 2b0ae357 2022-01-10 thomas if (blocklen == 0) {
746 2b0ae357 2022-01-10 thomas /* Source remainder from the file itself. */
747 2b0ae357 2022-01-10 thomas if (fileoffset < filesize) {
748 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
749 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0,
750 2b0ae357 2022-01-10 thomas filesize - fileoffset);
754 cc524d36 2022-05-31 thomas err = lookupblk_mem(&block, dt, buf, blocklen, seed, basedata,
755 2b0ae357 2022-01-10 thomas basefile_offset0);
758 2b0ae357 2022-01-10 thomas if (block != NULL) {
760 2b0ae357 2022-01-10 thomas * We have found a matching block in the delta base.
761 2b0ae357 2022-01-10 thomas * Attempt to stretch the block as far as possible and
762 2b0ae357 2022-01-10 thomas * generate a copy instruction.
764 2b0ae357 2022-01-10 thomas err = stretchblk_file_mem(basedata, basefile_offset0,
765 2b0ae357 2022-01-10 thomas basefile_size, block, f, filesize, &blocklen);
768 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
769 2b0ae357 2022-01-10 thomas alloc_chunk_size, 1, block->offset, blocklen);
774 2b0ae357 2022-01-10 thomas * No match.
775 2b0ae357 2022-01-10 thomas * This block needs to be sourced from the file itself.
777 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
778 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0, blocklen);
782 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
783 2b0ae357 2022-01-10 thomas if (fseeko(f, fileoffset, SEEK_SET) == -1) {
784 2b0ae357 2022-01-10 thomas err = got_error_from_errno("fseeko");
789 2b0ae357 2022-01-10 thomas if (err) {
790 2b0ae357 2022-01-10 thomas free(*deltas);
791 2b0ae357 2022-01-10 thomas *deltas = NULL;
792 2b0ae357 2022-01-10 thomas *ndeltas = 0;
794 2b0ae357 2022-01-10 thomas return err;
797 2b0ae357 2022-01-10 thomas const struct got_error *
798 2b0ae357 2022-01-10 thomas got_deltify_mem_file(struct got_delta_instruction **deltas, int *ndeltas,
799 cc524d36 2022-05-31 thomas uint8_t *data, off_t fileoffset, off_t filesize, uint32_t seed,
800 2b0ae357 2022-01-10 thomas struct got_delta_table *dt, FILE *basefile,
801 2b0ae357 2022-01-10 thomas off_t basefile_offset0, off_t basefile_size)
803 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
804 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
805 2b0ae357 2022-01-10 thomas size_t nalloc = 0;
806 2b0ae357 2022-01-10 thomas const size_t alloc_chunk_size = 64;
808 2b0ae357 2022-01-10 thomas *deltas = NULL;
809 2b0ae357 2022-01-10 thomas *ndeltas = 0;
811 2b0ae357 2022-01-10 thomas *deltas = reallocarray(NULL, alloc_chunk_size,
812 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_instruction));
813 2b0ae357 2022-01-10 thomas if (*deltas == NULL)
814 2b0ae357 2022-01-10 thomas return got_error_from_errno("reallocarray");
815 2b0ae357 2022-01-10 thomas nalloc = alloc_chunk_size;
817 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
818 2b0ae357 2022-01-10 thomas off_t blocklen;
819 2b0ae357 2022-01-10 thomas struct got_delta_block *block;
820 2b0ae357 2022-01-10 thomas err = nextblk_mem(&blocklen, data, fileoffset, filesize);
823 2b0ae357 2022-01-10 thomas if (blocklen == 0) {
824 2b0ae357 2022-01-10 thomas /* Source remainder from the file itself. */
825 2b0ae357 2022-01-10 thomas if (fileoffset < filesize) {
826 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
827 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0,
828 2b0ae357 2022-01-10 thomas filesize - fileoffset);
832 cc524d36 2022-05-31 thomas err = lookupblk(&block, dt, data + fileoffset, blocklen, seed,
833 2b0ae357 2022-01-10 thomas basefile, basefile_offset0);
836 2b0ae357 2022-01-10 thomas if (block != NULL) {
838 2b0ae357 2022-01-10 thomas * We have found a matching block in the delta base.
839 2b0ae357 2022-01-10 thomas * Attempt to stretch the block as far as possible and
840 2b0ae357 2022-01-10 thomas * generate a copy instruction.
842 2b0ae357 2022-01-10 thomas err = stretchblk_mem_file(basefile, basefile_offset0,
843 2b0ae357 2022-01-10 thomas block, data, fileoffset + blocklen, filesize,
844 2b0ae357 2022-01-10 thomas &blocklen);
847 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
848 2b0ae357 2022-01-10 thomas alloc_chunk_size, 1, block->offset, blocklen);
853 2b0ae357 2022-01-10 thomas * No match.
854 2b0ae357 2022-01-10 thomas * This block needs to be sourced from the file itself.
856 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
857 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0, blocklen);
861 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
864 2b0ae357 2022-01-10 thomas if (err) {
865 2b0ae357 2022-01-10 thomas free(*deltas);
866 2b0ae357 2022-01-10 thomas *deltas = NULL;
867 2b0ae357 2022-01-10 thomas *ndeltas = 0;
869 2b0ae357 2022-01-10 thomas return err;
872 2b0ae357 2022-01-10 thomas const struct got_error *
873 2b0ae357 2022-01-10 thomas got_deltify_mem_mem(struct got_delta_instruction **deltas, int *ndeltas,
874 cc524d36 2022-05-31 thomas uint8_t *data, off_t fileoffset, off_t filesize, uint32_t seed,
875 2b0ae357 2022-01-10 thomas struct got_delta_table *dt, uint8_t *basedata,
876 2b0ae357 2022-01-10 thomas off_t basefile_offset0, off_t basefile_size)
878 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
879 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
880 2b0ae357 2022-01-10 thomas size_t nalloc = 0;
881 2b0ae357 2022-01-10 thomas const size_t alloc_chunk_size = 64;
883 2b0ae357 2022-01-10 thomas *deltas = NULL;
884 2b0ae357 2022-01-10 thomas *ndeltas = 0;
886 2b0ae357 2022-01-10 thomas *deltas = reallocarray(NULL, alloc_chunk_size,
887 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_instruction));
888 2b0ae357 2022-01-10 thomas if (*deltas == NULL)
889 2b0ae357 2022-01-10 thomas return got_error_from_errno("reallocarray");
890 2b0ae357 2022-01-10 thomas nalloc = alloc_chunk_size;
892 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
893 2b0ae357 2022-01-10 thomas off_t blocklen;
894 2b0ae357 2022-01-10 thomas struct got_delta_block *block;
895 2b0ae357 2022-01-10 thomas err = nextblk_mem(&blocklen, data, fileoffset, filesize);
898 2b0ae357 2022-01-10 thomas if (blocklen == 0) {
899 2b0ae357 2022-01-10 thomas /* Source remainder from the file itself. */
900 2b0ae357 2022-01-10 thomas if (fileoffset < filesize) {
901 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
902 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0,
903 2b0ae357 2022-01-10 thomas filesize - fileoffset);
907 2b0ae357 2022-01-10 thomas err = lookupblk_mem(&block, dt, data + fileoffset, blocklen,
908 cc524d36 2022-05-31 thomas seed, basedata, basefile_offset0);
911 2b0ae357 2022-01-10 thomas if (block != NULL) {
913 2b0ae357 2022-01-10 thomas * We have found a matching block in the delta base.
914 2b0ae357 2022-01-10 thomas * Attempt to stretch the block as far as possible and
915 2b0ae357 2022-01-10 thomas * generate a copy instruction.
917 2b0ae357 2022-01-10 thomas err = stretchblk_mem_mem(basedata, basefile_offset0,
918 2b0ae357 2022-01-10 thomas basefile_size, block, data, fileoffset + blocklen,
919 2b0ae357 2022-01-10 thomas filesize, &blocklen);
922 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
923 2b0ae357 2022-01-10 thomas alloc_chunk_size, 1, block->offset, blocklen);
928 2b0ae357 2022-01-10 thomas * No match.
929 2b0ae357 2022-01-10 thomas * This block needs to be sourced from the file itself.
931 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
932 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0, blocklen);
936 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
939 2b0ae357 2022-01-10 thomas if (err) {
940 2b0ae357 2022-01-10 thomas free(*deltas);
941 2b0ae357 2022-01-10 thomas *deltas = NULL;
942 2b0ae357 2022-01-10 thomas *ndeltas = 0;
944 2b0ae357 2022-01-10 thomas return err;
