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 4fccd2fe 2023-03-08 thomas #include "got_compat.h"
20 f364801d 2021-03-10 stsp #include <sys/types.h>
21 8b925c6c 2022-07-16 thomas #include <sys/queue.h>
23 f364801d 2021-03-10 stsp #include <assert.h>
24 69aa0e90 2021-03-10 stsp #include <errno.h>
25 f364801d 2021-03-10 stsp #include <limits.h>
26 f364801d 2021-03-10 stsp #include <stdint.h>
27 f364801d 2021-03-10 stsp #include <stdio.h>
28 f364801d 2021-03-10 stsp #include <stdlib.h>
29 f364801d 2021-03-10 stsp #include <string.h>
31 f364801d 2021-03-10 stsp #include "got_error.h"
33 69aa0e90 2021-03-10 stsp #include "got_lib_deltify.h"
34 b7bf7afc 2022-02-12 thomas #include "murmurhash2.h"
37 69aa0e90 2021-03-10 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
41 2539ccf2 2021-10-08 thomas * The algorihm used here is FastCDC (Fast Content-Defined Chunking)
42 2539ccf2 2021-10-08 thomas * https://www.usenix.org/conference/atc16/technical-sessions/presentation/xia
45 fa8129f7 2022-03-22 thomas static const uint32_t geartab[256] = {
46 69aa0e90 2021-03-10 stsp 0x67ed26b7, 0x32da500c, 0x53d0fee0, 0xce387dc7, 0xcd406d90, 0x2e83a4d4,
47 69aa0e90 2021-03-10 stsp 0x9fc9a38d, 0xb67259dc, 0xca6b1722, 0x6d2ea08c, 0x235cea2e, 0x3149bb5f,
48 69aa0e90 2021-03-10 stsp 0x1beda787, 0x2a6b77d5, 0x2f22d9ac, 0x91fc0544, 0xe413acfa, 0x5a30ff7a,
49 69aa0e90 2021-03-10 stsp 0xad6fdde0, 0x444fd0f5, 0x7ad87864, 0x58c5ff05, 0x8d2ec336, 0x2371f853,
50 69aa0e90 2021-03-10 stsp 0x550f8572, 0x6aa448dd, 0x7c9ddbcf, 0x95221e14, 0x2a82ec33, 0xcbec5a78,
51 69aa0e90 2021-03-10 stsp 0xc6795a0d, 0x243995b7, 0x1c909a2f, 0x4fded51c, 0x635d334b, 0x0e2b9999,
52 69aa0e90 2021-03-10 stsp 0x2702968d, 0x856de1d5, 0x3325d60e, 0xeb6a7502, 0xec2a9844, 0x0905835a,
53 69aa0e90 2021-03-10 stsp 0xa1820375, 0xa4be5cab, 0x96a6c058, 0x2c2ccd70, 0xba40fce3, 0xd794c46b,
54 69aa0e90 2021-03-10 stsp 0x8fbae83e, 0xc3aa7899, 0x3d3ff8ed, 0xa0d42b5b, 0x571c0c97, 0xd2811516,
55 69aa0e90 2021-03-10 stsp 0xf7e7b96c, 0x4fd2fcbd, 0xe2fdec94, 0x282cc436, 0x78e8e95c, 0x80a3b613,
56 69aa0e90 2021-03-10 stsp 0xcfbee20c, 0xd4a32d1c, 0x2a12ff13, 0x6af82936, 0xe5630258, 0x8efa6a98,
57 69aa0e90 2021-03-10 stsp 0x294fb2d1, 0xdeb57086, 0x5f0fddb3, 0xeceda7ce, 0x4c87305f, 0x3a6d3307,
58 69aa0e90 2021-03-10 stsp 0xe22d2942, 0x9d060217, 0x1e42ed02, 0xb6f63b52, 0x4367f39f, 0x055cf262,
59 69aa0e90 2021-03-10 stsp 0x03a461b2, 0x5ef9e382, 0x386bc03a, 0x2a1e79c7, 0xf1a0058b, 0xd4d2dea9,
60 69aa0e90 2021-03-10 stsp 0x56baf37d, 0x5daff6cc, 0xf03a951d, 0xaef7de45, 0xa8f4581e, 0x3960b555,
61 69aa0e90 2021-03-10 stsp 0xffbfff6d, 0xbe702a23, 0x8f5b6d6f, 0x061739fb, 0x98696f47, 0x3fd596d4,
62 69aa0e90 2021-03-10 stsp 0x151eac6b, 0xa9fcc4f5, 0x69181a12, 0x3ac5a107, 0xb5198fe7, 0x96bcb1da,
63 69aa0e90 2021-03-10 stsp 0x1b5ddf8e, 0xc757d650, 0x65865c3a, 0x8fc0a41a, 0x87435536, 0x99eda6f2,
64 69aa0e90 2021-03-10 stsp 0x41874794, 0x29cff4e8, 0xb70efd9a, 0x3103f6e7, 0x84d2453b, 0x15a450bd,
65 69aa0e90 2021-03-10 stsp 0x74f49af1, 0x60f664b1, 0xa1c86935, 0xfdafbce1, 0xe36353e3, 0x5d9ba739,
66 69aa0e90 2021-03-10 stsp 0xbc0559ba, 0x708b0054, 0xd41d808c, 0xb2f31723, 0x9027c41f, 0xf136d165,
67 69aa0e90 2021-03-10 stsp 0xb5374b12, 0x9420a6ac, 0x273958b6, 0xe6c2fad0, 0xebdc1f21, 0xfb33af8b,
68 69aa0e90 2021-03-10 stsp 0xc71c25cd, 0xe9a2d8e5, 0xbeb38a50, 0xbceb7cc2, 0x4e4e73f0, 0xcd6c251d,
69 69aa0e90 2021-03-10 stsp 0xde4c032c, 0x4b04ac30, 0x725b8b21, 0x4eb8c33b, 0x20d07b75, 0x0567aa63,
70 69aa0e90 2021-03-10 stsp 0xb56b2bb7, 0xc1f5fd3a, 0xcafd35ca, 0x470dd4da, 0xfe4f94cd, 0xfb8de424,
71 69aa0e90 2021-03-10 stsp 0xe8dbcf40, 0xfe50a37a, 0x62db5b5d, 0xf32f4ab6, 0x2c4a8a51, 0x18473dc0,
72 69aa0e90 2021-03-10 stsp 0xfe0cbb6e, 0xfe399efd, 0xdf34ecc9, 0x6ccd5055, 0x46097073, 0x139135c2,
73 69aa0e90 2021-03-10 stsp 0x721c76f6, 0x1c6a94b4, 0x6eee014d, 0x8a508e02, 0x3da538f5, 0x280d394f,
74 69aa0e90 2021-03-10 stsp 0x5248a0c4, 0x3ce94c6c, 0x9a71ad3a, 0x8493dd05, 0xe43f0ab6, 0x18e4ed42,
75 69aa0e90 2021-03-10 stsp 0x6c5c0e09, 0x42b06ec9, 0x8d330343, 0xa45b6f59, 0x2a573c0c, 0xd7fd3de6,
76 69aa0e90 2021-03-10 stsp 0xeedeab68, 0x5c84dafc, 0xbbd1b1a8, 0xa3ce1ad1, 0x85b70bed, 0xb6add07f,
77 69aa0e90 2021-03-10 stsp 0xa531309c, 0x8f8ab852, 0x564de332, 0xeac9ed0c, 0x73da402c, 0x3ec52761,
78 69aa0e90 2021-03-10 stsp 0x43af2f4d, 0xd6ff45c8, 0x4c367462, 0xd553bd6a, 0x44724855, 0x3b2aa728,
79 69aa0e90 2021-03-10 stsp 0x56e5eb65, 0xeaf16173, 0x33fa42ff, 0xd714bb5d, 0xfbd0a3b9, 0xaf517134,
80 69aa0e90 2021-03-10 stsp 0x9416c8cd, 0x534cf94f, 0x548947c2, 0x34193569, 0x32f4389a, 0xfe7028bc,
81 69aa0e90 2021-03-10 stsp 0xed73b1ed, 0x9db95770, 0x468e3922, 0x0440c3cd, 0x60059a62, 0x33504562,
82 69aa0e90 2021-03-10 stsp 0x2b229fbd, 0x5174dca5, 0xf7028752, 0xd63c6aa8, 0x31276f38, 0x0646721c,
83 69aa0e90 2021-03-10 stsp 0xb0191da8, 0xe00e6de0, 0x9eac1a6e, 0x9f7628a5, 0xed6c06ea, 0x0bb8af15,
84 69aa0e90 2021-03-10 stsp 0xf119fb12, 0x38693c1c, 0x732bc0fe, 0x84953275, 0xb82ec888, 0x33a4f1b3,
85 69aa0e90 2021-03-10 stsp 0x3099835e, 0x028a8782, 0x5fdd51d7, 0xc6c717b3, 0xb06caf71, 0x17c8c111,
86 69aa0e90 2021-03-10 stsp 0x61bad754, 0x9fd03061, 0xe09df1af, 0x3bc9eb73, 0x85878413, 0x9889aaf2,
87 69aa0e90 2021-03-10 stsp 0x3f5a9e46, 0x42c9f01f, 0x9984a4f4, 0xd5de43cc, 0xd294daed, 0xbecba2d2,
88 69aa0e90 2021-03-10 stsp 0xf1f6e72c, 0x5551128a, 0x83af87e2, 0x6f0342ba,
91 b7bf7afc 2022-02-12 thomas static uint32_t
92 cc524d36 2022-05-31 thomas hashblk(const unsigned char *p, off_t n, uint32_t seed)
94 cc524d36 2022-05-31 thomas return murmurhash2(p, n, seed);
97 f364801d 2021-03-10 stsp static const struct got_error *
98 740bba1c 2021-06-18 stsp addblk(struct got_delta_table *dt, FILE *f, off_t file_offset0, off_t len,
99 90dfa2bf 2022-02-13 thomas off_t offset, uint32_t h)
101 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
103 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
104 0d15f6dc 2021-06-13 stsp uint8_t buf2[GOT_DELTIFY_MAXCHUNK];
105 69aa0e90 2021-03-10 stsp size_t r = 0;
107 69aa0e90 2021-03-10 stsp if (len == 0)
108 69aa0e90 2021-03-10 stsp return NULL;
110 69aa0e90 2021-03-10 stsp i = h % dt->nalloc;
111 69aa0e90 2021-03-10 stsp while (dt->blocks[i].len != 0) {
113 69aa0e90 2021-03-10 stsp * Avoid adding duplicate blocks.
114 69aa0e90 2021-03-10 stsp * NB: A matching hash is insufficient for detecting equality.
115 51a494da 2021-06-11 stsp * The hash can only detect inequality.
117 51a494da 2021-06-11 stsp if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
118 0d15f6dc 2021-06-13 stsp if (r == 0) {
119 740bba1c 2021-06-18 stsp if (fseeko(f, file_offset0 + offset, SEEK_SET) == -1)
120 0d15f6dc 2021-06-13 stsp return got_error_from_errno("fseeko");
121 0d15f6dc 2021-06-13 stsp r = fread(buf, 1, len, f);
122 0d15f6dc 2021-06-13 stsp if (r != len) {
123 0d15f6dc 2021-06-13 stsp if (!ferror(f))
124 0d15f6dc 2021-06-13 stsp return NULL;
125 0d15f6dc 2021-06-13 stsp return got_ferror(f, GOT_ERR_IO);
128 740bba1c 2021-06-18 stsp if (fseeko(f, file_offset0 + dt->blocks[i].offset,
129 740bba1c 2021-06-18 stsp SEEK_SET) == -1)
130 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
131 69aa0e90 2021-03-10 stsp if (fread(buf2, 1, len, f) != len)
132 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
133 69aa0e90 2021-03-10 stsp if (memcmp(buf, buf2, len) == 0)
134 69aa0e90 2021-03-10 stsp return NULL;
137 69aa0e90 2021-03-10 stsp i = (i + 1) % dt->nalloc;
139 69aa0e90 2021-03-10 stsp assert(dt->blocks[i].len == 0);
140 69aa0e90 2021-03-10 stsp dt->blocks[i].len = len;
141 69aa0e90 2021-03-10 stsp dt->blocks[i].offset = offset;
142 69aa0e90 2021-03-10 stsp dt->blocks[i].hash = h;
143 69aa0e90 2021-03-10 stsp dt->nblocks++;
144 c0604191 2023-07-29 thomas if (dt->nalloc < dt->nblocks + 256) {
145 69aa0e90 2021-03-10 stsp struct got_delta_block *db;
146 e89540a9 2021-06-11 stsp size_t old_size = dt->nalloc;
147 69aa0e90 2021-03-10 stsp db = dt->blocks;
148 c0604191 2023-07-29 thomas dt->blocks = calloc(dt->nalloc + 256,
149 e89540a9 2021-06-11 stsp sizeof(struct got_delta_block));
150 69aa0e90 2021-03-10 stsp if (dt->blocks == NULL) {
151 69aa0e90 2021-03-10 stsp err = got_error_from_errno("calloc");
152 69aa0e90 2021-03-10 stsp dt->blocks = db;
153 69aa0e90 2021-03-10 stsp return err;
155 c0604191 2023-07-29 thomas dt->nalloc += 256;
157 69aa0e90 2021-03-10 stsp * Recompute all block positions. Hash-based indices of blocks
158 69aa0e90 2021-03-10 stsp * in the array depend on the allocated length of the array.
160 69aa0e90 2021-03-10 stsp dt->nblocks = 0;
161 a893025f 2021-06-11 stsp for (i = 0; i < old_size; i++) {
162 69aa0e90 2021-03-10 stsp if (db[i].len == 0)
164 740bba1c 2021-06-18 stsp err = addblk(dt, f, file_offset0, db[i].len,
165 2b0ae357 2022-01-10 thomas db[i].offset, db[i].hash);
169 2b0ae357 2022-01-10 thomas free(db);
172 2b0ae357 2022-01-10 thomas return err;
175 2b0ae357 2022-01-10 thomas static const struct got_error *
176 2b0ae357 2022-01-10 thomas addblk_mem(struct got_delta_table *dt, uint8_t *data, off_t file_offset0,
177 90dfa2bf 2022-02-13 thomas off_t len, off_t offset, uint32_t h)
179 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
181 2b0ae357 2022-01-10 thomas uint8_t *block1;
182 2b0ae357 2022-01-10 thomas uint8_t *block2;
184 2b0ae357 2022-01-10 thomas if (len == 0)
185 2b0ae357 2022-01-10 thomas return NULL;
187 2b0ae357 2022-01-10 thomas i = h % dt->nalloc;
188 2b0ae357 2022-01-10 thomas while (dt->blocks[i].len != 0) {
190 2b0ae357 2022-01-10 thomas * Avoid adding duplicate blocks.
191 2b0ae357 2022-01-10 thomas * NB: A matching hash is insufficient for detecting equality.
192 2b0ae357 2022-01-10 thomas * The hash can only detect inequality.
194 2b0ae357 2022-01-10 thomas if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
195 2b0ae357 2022-01-10 thomas block1 = data + file_offset0 + dt->blocks[i].offset;
196 2b0ae357 2022-01-10 thomas block2 = data + file_offset0 + offset;
197 2b0ae357 2022-01-10 thomas if (memcmp(block1, block2, len) == 0)
198 2b0ae357 2022-01-10 thomas return NULL;
201 2b0ae357 2022-01-10 thomas i = (i + 1) % dt->nalloc;
203 2b0ae357 2022-01-10 thomas assert(dt->blocks[i].len == 0);
204 2b0ae357 2022-01-10 thomas dt->blocks[i].len = len;
205 2b0ae357 2022-01-10 thomas dt->blocks[i].offset = offset;
206 2b0ae357 2022-01-10 thomas dt->blocks[i].hash = h;
207 2b0ae357 2022-01-10 thomas dt->nblocks++;
208 c0604191 2023-07-29 thomas if (dt->nalloc < dt->nblocks + 256) {
209 2b0ae357 2022-01-10 thomas struct got_delta_block *db;
210 2b0ae357 2022-01-10 thomas size_t old_size = dt->nalloc;
211 2b0ae357 2022-01-10 thomas db = dt->blocks;
212 c0604191 2023-07-29 thomas dt->blocks = calloc(dt->nalloc + 256,
213 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_block));
214 2b0ae357 2022-01-10 thomas if (dt->blocks == NULL) {
215 2b0ae357 2022-01-10 thomas err = got_error_from_errno("calloc");
216 2b0ae357 2022-01-10 thomas dt->blocks = db;
217 2b0ae357 2022-01-10 thomas return err;
219 c0604191 2023-07-29 thomas dt->nalloc += 256;
221 2b0ae357 2022-01-10 thomas * Recompute all block positions. Hash-based indices of blocks
222 2b0ae357 2022-01-10 thomas * in the array depend on the allocated length of the array.
224 2b0ae357 2022-01-10 thomas dt->nblocks = 0;
225 2b0ae357 2022-01-10 thomas for (i = 0; i < old_size; i++) {
226 2b0ae357 2022-01-10 thomas if (db[i].len == 0)
227 2b0ae357 2022-01-10 thomas continue;
228 2b0ae357 2022-01-10 thomas err = addblk_mem(dt, data, file_offset0, db[i].len,
229 740bba1c 2021-06-18 stsp db[i].offset, db[i].hash);
236 69aa0e90 2021-03-10 stsp return err;
239 69aa0e90 2021-03-10 stsp static const struct got_error *
240 69aa0e90 2021-03-10 stsp lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
241 cc524d36 2022-05-31 thomas unsigned char *p, off_t len, uint32_t seed, FILE *basefile,
242 cc524d36 2022-05-31 thomas off_t basefile_offset0)
245 b7bf7afc 2022-02-12 thomas uint32_t h;
246 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
249 69aa0e90 2021-03-10 stsp *block = NULL;
251 cc524d36 2022-05-31 thomas h = hashblk(p, len, seed);
252 69aa0e90 2021-03-10 stsp for (i = h % dt->nalloc; dt->blocks[i].len != 0;
253 69aa0e90 2021-03-10 stsp i = (i + 1) % dt->nalloc) {
254 69aa0e90 2021-03-10 stsp if (dt->blocks[i].hash != h ||
255 69aa0e90 2021-03-10 stsp dt->blocks[i].len != len)
257 740bba1c 2021-06-18 stsp if (fseeko(basefile, basefile_offset0 + dt->blocks[i].offset,
258 740bba1c 2021-06-18 stsp SEEK_SET) == -1)
259 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
260 69aa0e90 2021-03-10 stsp r = fread(buf, 1, len, basefile);
261 69aa0e90 2021-03-10 stsp if (r != len)
262 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
263 69aa0e90 2021-03-10 stsp if (memcmp(p, buf, len) == 0) {
264 69aa0e90 2021-03-10 stsp *block = &dt->blocks[i];
268 f364801d 2021-03-10 stsp return NULL;
271 69aa0e90 2021-03-10 stsp static const struct got_error *
272 2b0ae357 2022-01-10 thomas lookupblk_mem(struct got_delta_block **block, struct got_delta_table *dt,
273 cc524d36 2022-05-31 thomas unsigned char *p, off_t len, uint32_t seed, uint8_t *basedata,
274 cc524d36 2022-05-31 thomas off_t basefile_offset0)
277 b7bf7afc 2022-02-12 thomas uint32_t h;
278 2b0ae357 2022-01-10 thomas uint8_t *b;
280 2b0ae357 2022-01-10 thomas *block = NULL;
282 cc524d36 2022-05-31 thomas h = hashblk(p, len, seed);
283 2b0ae357 2022-01-10 thomas for (i = h % dt->nalloc; dt->blocks[i].len != 0;
284 2b0ae357 2022-01-10 thomas i = (i + 1) % dt->nalloc) {
285 2b0ae357 2022-01-10 thomas if (dt->blocks[i].hash != h ||
286 2b0ae357 2022-01-10 thomas dt->blocks[i].len != len)
287 2b0ae357 2022-01-10 thomas continue;
288 2b0ae357 2022-01-10 thomas b = basedata + basefile_offset0 + dt->blocks[i].offset;
289 2b0ae357 2022-01-10 thomas if (memcmp(p, b, len) == 0) {
290 2b0ae357 2022-01-10 thomas *block = &dt->blocks[i];
294 2b0ae357 2022-01-10 thomas return NULL;
297 2b0ae357 2022-01-10 thomas static const struct got_error *
298 69aa0e90 2021-03-10 stsp nextblk(uint8_t *buf, off_t *blocklen, FILE *f)
300 f364801d 2021-03-10 stsp uint32_t gh;
301 69aa0e90 2021-03-10 stsp const unsigned char *p;
303 69aa0e90 2021-03-10 stsp off_t pos = ftello(f);
305 69aa0e90 2021-03-10 stsp *blocklen = 0;
307 69aa0e90 2021-03-10 stsp r = fread(buf, 1, GOT_DELTIFY_MAXCHUNK, f);
308 69aa0e90 2021-03-10 stsp if (r == 0 && ferror(f))
309 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
310 69aa0e90 2021-03-10 stsp if (r < GOT_DELTIFY_MINCHUNK)
311 69aa0e90 2021-03-10 stsp return NULL; /* no more delta-worthy blocks left */
313 69aa0e90 2021-03-10 stsp /* Got a deltifiable block. Find the split-point where it ends. */
314 69aa0e90 2021-03-10 stsp p = buf + GOT_DELTIFY_MINCHUNK;
316 69aa0e90 2021-03-10 stsp while (p != buf + r) {
317 f364801d 2021-03-10 stsp gh = (gh << 1) + geartab[*p++];
318 69aa0e90 2021-03-10 stsp if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
322 69aa0e90 2021-03-10 stsp *blocklen = (p - buf);
323 69aa0e90 2021-03-10 stsp if (fseeko(f, pos + *blocklen, SEEK_SET) == -1)
324 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
326 69aa0e90 2021-03-10 stsp return NULL;
329 2b0ae357 2022-01-10 thomas static const struct got_error *
330 2b0ae357 2022-01-10 thomas nextblk_mem(off_t *blocklen, uint8_t *data, off_t fileoffset, off_t filesize)
332 2b0ae357 2022-01-10 thomas uint32_t gh;
333 2b0ae357 2022-01-10 thomas const unsigned char *p;
335 2b0ae357 2022-01-10 thomas *blocklen = 0;
337 2b0ae357 2022-01-10 thomas if (fileoffset >= filesize ||
338 2b0ae357 2022-01-10 thomas filesize - fileoffset < GOT_DELTIFY_MINCHUNK)
339 2b0ae357 2022-01-10 thomas return NULL; /* no more delta-worthy blocks left */
341 2b0ae357 2022-01-10 thomas /* Got a deltifiable block. Find the split-point where it ends. */
342 2b0ae357 2022-01-10 thomas p = data + fileoffset + GOT_DELTIFY_MINCHUNK;
344 2b0ae357 2022-01-10 thomas while (p != data + MIN(fileoffset + GOT_DELTIFY_MAXCHUNK, filesize)) {
345 2b0ae357 2022-01-10 thomas gh = (gh << 1) + geartab[*p++];
346 2b0ae357 2022-01-10 thomas if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
350 2b0ae357 2022-01-10 thomas *blocklen = (p - (data + fileoffset));
351 2b0ae357 2022-01-10 thomas return NULL;
354 f364801d 2021-03-10 stsp const struct got_error *
355 69aa0e90 2021-03-10 stsp got_deltify_init(struct got_delta_table **dt, FILE *f, off_t fileoffset,
356 cc524d36 2022-05-31 thomas off_t filesize, uint32_t seed)
358 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
359 b7bf7afc 2022-02-12 thomas uint32_t h;
360 740bba1c 2021-06-18 stsp const off_t offset0 = fileoffset;
362 69aa0e90 2021-03-10 stsp *dt = calloc(1, sizeof(**dt));
363 69aa0e90 2021-03-10 stsp if (*dt == NULL)
364 f364801d 2021-03-10 stsp return got_error_from_errno("calloc");
366 69aa0e90 2021-03-10 stsp (*dt)->nblocks = 0;
367 69aa0e90 2021-03-10 stsp (*dt)->nalloc = 128;
368 69aa0e90 2021-03-10 stsp (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
369 69aa0e90 2021-03-10 stsp if ((*dt)->blocks == NULL) {
370 69aa0e90 2021-03-10 stsp err = got_error_from_errno("calloc");
374 69aa0e90 2021-03-10 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
375 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
377 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
378 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
379 69aa0e90 2021-03-10 stsp off_t blocklen;
380 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
383 69aa0e90 2021-03-10 stsp if (blocklen == 0)
385 cc524d36 2022-05-31 thomas h = hashblk(buf, blocklen, seed);
386 740bba1c 2021-06-18 stsp err = addblk(*dt, f, offset0, blocklen,
387 740bba1c 2021-06-18 stsp fileoffset - offset0, h);
390 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
391 68bdcdc2 2021-06-11 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
392 68bdcdc2 2021-06-11 stsp return got_error_from_errno("fseeko");
395 2b0ae357 2022-01-10 thomas if (err) {
396 2b0ae357 2022-01-10 thomas free((*dt)->blocks);
397 2b0ae357 2022-01-10 thomas free(*dt);
398 2b0ae357 2022-01-10 thomas *dt = NULL;
401 2b0ae357 2022-01-10 thomas return err;
404 2b0ae357 2022-01-10 thomas const struct got_error *
405 2b0ae357 2022-01-10 thomas got_deltify_init_mem(struct got_delta_table **dt, uint8_t *data,
406 cc524d36 2022-05-31 thomas off_t fileoffset, off_t filesize, uint32_t seed)
408 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
409 90dfa2bf 2022-02-13 thomas uint32_t h;
410 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
412 2b0ae357 2022-01-10 thomas *dt = calloc(1, sizeof(**dt));
413 2b0ae357 2022-01-10 thomas if (*dt == NULL)
414 2b0ae357 2022-01-10 thomas return got_error_from_errno("calloc");
416 2b0ae357 2022-01-10 thomas (*dt)->nblocks = 0;
417 2b0ae357 2022-01-10 thomas (*dt)->nalloc = 128;
418 2b0ae357 2022-01-10 thomas (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
419 2b0ae357 2022-01-10 thomas if ((*dt)->blocks == NULL) {
420 2b0ae357 2022-01-10 thomas err = got_error_from_errno("calloc");
421 2b0ae357 2022-01-10 thomas goto done;
424 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
425 2b0ae357 2022-01-10 thomas off_t blocklen;
426 2b0ae357 2022-01-10 thomas err = nextblk_mem(&blocklen, data, fileoffset, filesize);
428 2b0ae357 2022-01-10 thomas goto done;
429 2b0ae357 2022-01-10 thomas if (blocklen == 0)
431 cc524d36 2022-05-31 thomas h = hashblk(data + fileoffset, blocklen, seed);
432 2b0ae357 2022-01-10 thomas err = addblk_mem(*dt, data, offset0, blocklen,
433 2b0ae357 2022-01-10 thomas fileoffset - offset0, h);
435 2b0ae357 2022-01-10 thomas goto done;
436 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
440 69aa0e90 2021-03-10 stsp free((*dt)->blocks);
442 69aa0e90 2021-03-10 stsp *dt = NULL;
445 69aa0e90 2021-03-10 stsp return err;
449 69aa0e90 2021-03-10 stsp got_deltify_free(struct got_delta_table *dt)
451 dbbf4a5f 2021-05-20 stsp if (dt == NULL)
453 69aa0e90 2021-03-10 stsp free(dt->blocks);
457 f364801d 2021-03-10 stsp static const struct got_error *
458 dd29967c 2021-08-22 stsp emitdelta(struct got_delta_instruction **deltas, size_t *nalloc, int *ndeltas,
459 dd29967c 2021-08-22 stsp const size_t alloc_chunk_size, int copy, off_t offset, off_t len)
461 69aa0e90 2021-03-10 stsp struct got_delta_instruction *d, *p;
463 dd29967c 2021-08-22 stsp if (*nalloc < *ndeltas + alloc_chunk_size) {
464 dd29967c 2021-08-22 stsp p = reallocarray(*deltas, *nalloc + alloc_chunk_size,
465 dd29967c 2021-08-22 stsp sizeof(struct got_delta_instruction));
466 dd29967c 2021-08-22 stsp if (p == NULL)
467 7d76c4e7 2021-10-15 thomas return got_error_from_errno("reallocarray");
468 dd29967c 2021-08-22 stsp *deltas = p;
469 dd29967c 2021-08-22 stsp *nalloc += alloc_chunk_size;
471 69aa0e90 2021-03-10 stsp *ndeltas += 1;
472 69aa0e90 2021-03-10 stsp d = &(*deltas)[*ndeltas - 1];
473 69aa0e90 2021-03-10 stsp d->copy = copy;
474 69aa0e90 2021-03-10 stsp d->offset = offset;
475 56e7a15f 2021-03-10 stsp d->len = len;
476 f364801d 2021-03-10 stsp return NULL;
479 69aa0e90 2021-03-10 stsp static const struct got_error *
480 740bba1c 2021-06-18 stsp stretchblk(FILE *basefile, off_t base_offset0, struct got_delta_block *block,
481 740bba1c 2021-06-18 stsp FILE *f, off_t filesize, off_t *blocklen)
483 69aa0e90 2021-03-10 stsp uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
484 69aa0e90 2021-03-10 stsp size_t base_r, r, i;
485 69aa0e90 2021-03-10 stsp int buf_equal = 1;
487 5de743f8 2021-08-29 stsp if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
488 5de743f8 2021-08-29 stsp SEEK_SET) == -1)
489 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
491 69aa0e90 2021-03-10 stsp while (buf_equal && *blocklen < (1 << 24) - 1) {
492 69aa0e90 2021-03-10 stsp base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
493 69aa0e90 2021-03-10 stsp if (base_r == 0) {
494 69aa0e90 2021-03-10 stsp if (ferror(basefile))
495 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
498 69aa0e90 2021-03-10 stsp r = fread(buf, 1, sizeof(buf), f);
499 69aa0e90 2021-03-10 stsp if (r == 0) {
500 69aa0e90 2021-03-10 stsp if (ferror(f))
501 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
504 69aa0e90 2021-03-10 stsp for (i = 0; i < MIN(base_r, r); i++) {
505 69aa0e90 2021-03-10 stsp if (buf[i] != basebuf[i]) {
506 69aa0e90 2021-03-10 stsp buf_equal = 0;
509 69aa0e90 2021-03-10 stsp (*blocklen)++;
513 2b0ae357 2022-01-10 thomas return NULL;
516 2b0ae357 2022-01-10 thomas static const struct got_error *
517 2b0ae357 2022-01-10 thomas stretchblk_file_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
518 2b0ae357 2022-01-10 thomas struct got_delta_block *block, FILE *f, off_t filesize, off_t *blocklen)
520 2b0ae357 2022-01-10 thomas uint8_t buf[GOT_DELTIFY_MAXCHUNK];
521 2b0ae357 2022-01-10 thomas size_t r, i;
522 2b0ae357 2022-01-10 thomas int buf_equal = 1;
523 2b0ae357 2022-01-10 thomas off_t base_offset = base_offset0 + block->offset + *blocklen;
525 2b0ae357 2022-01-10 thomas if (base_offset > basefile_size) {
526 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
527 e4a6dbc6 2022-07-24 thomas "read beyond the size of delta base at offset %lld",
528 e4a6dbc6 2022-07-24 thomas (long long)base_offset);
531 2b0ae357 2022-01-10 thomas while (buf_equal && *blocklen < (1 << 24) - 1) {
532 2b0ae357 2022-01-10 thomas if (base_offset + *blocklen >= basefile_size)
534 2b0ae357 2022-01-10 thomas r = fread(buf, 1, sizeof(buf), f);
535 2b0ae357 2022-01-10 thomas if (r == 0) {
536 2b0ae357 2022-01-10 thomas if (ferror(f))
537 2b0ae357 2022-01-10 thomas return got_ferror(f, GOT_ERR_IO);
540 2b0ae357 2022-01-10 thomas for (i = 0; i < MIN(basefile_size - base_offset, r); i++) {
541 2b0ae357 2022-01-10 thomas if (buf[i] != *(basedata + base_offset + i)) {
542 2b0ae357 2022-01-10 thomas buf_equal = 0;
545 2b0ae357 2022-01-10 thomas (*blocklen)++;
549 2b0ae357 2022-01-10 thomas return NULL;
552 2b0ae357 2022-01-10 thomas static const struct got_error *
553 2b0ae357 2022-01-10 thomas stretchblk_mem_file(FILE *basefile, off_t base_offset0,
554 2b0ae357 2022-01-10 thomas struct got_delta_block *block, uint8_t *data, off_t fileoffset,
555 2b0ae357 2022-01-10 thomas off_t filesize, off_t *blocklen)
557 2b0ae357 2022-01-10 thomas uint8_t basebuf[GOT_DELTIFY_MAXCHUNK];
558 2b0ae357 2022-01-10 thomas size_t base_r, i;
559 2b0ae357 2022-01-10 thomas int buf_equal = 1;
561 2b0ae357 2022-01-10 thomas if (fileoffset > filesize) {
562 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
563 e4a6dbc6 2022-07-24 thomas "read beyond the size of deltify file at offset %lld",
564 e4a6dbc6 2022-07-24 thomas (long long)fileoffset);
567 2b0ae357 2022-01-10 thomas if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
568 2b0ae357 2022-01-10 thomas SEEK_SET) == -1)
569 2b0ae357 2022-01-10 thomas return got_error_from_errno("fseeko");
571 2b0ae357 2022-01-10 thomas while (buf_equal && *blocklen < (1 << 24) - 1) {
572 2b0ae357 2022-01-10 thomas if (fileoffset + *blocklen >= filesize)
574 2b0ae357 2022-01-10 thomas base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
575 2b0ae357 2022-01-10 thomas if (base_r == 0) {
576 2b0ae357 2022-01-10 thomas if (ferror(basefile))
577 2b0ae357 2022-01-10 thomas return got_ferror(basefile, GOT_ERR_IO);
580 2b0ae357 2022-01-10 thomas for (i = 0; i < MIN(base_r, filesize - fileoffset); i++) {
581 2b0ae357 2022-01-10 thomas if (*(data + fileoffset + i) != basebuf[i]) {
582 2b0ae357 2022-01-10 thomas buf_equal = 0;
585 2b0ae357 2022-01-10 thomas (*blocklen)++;
589 2b0ae357 2022-01-10 thomas return NULL;
592 2b0ae357 2022-01-10 thomas static const struct got_error *
593 2b0ae357 2022-01-10 thomas stretchblk_mem_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
594 2b0ae357 2022-01-10 thomas struct got_delta_block *block, uint8_t *data, off_t fileoffset,
595 2b0ae357 2022-01-10 thomas off_t filesize, off_t *blocklen)
597 2b0ae357 2022-01-10 thomas off_t i, maxlen;
598 2b0ae357 2022-01-10 thomas off_t base_offset = base_offset0 + block->offset + *blocklen;
599 2b0ae357 2022-01-10 thomas uint8_t *p, *q;
601 2b0ae357 2022-01-10 thomas if (base_offset > basefile_size) {
602 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
603 e4a6dbc6 2022-07-24 thomas "read beyond the size of delta base at offset %lld",
604 e4a6dbc6 2022-07-24 thomas (long long)base_offset);
607 2b0ae357 2022-01-10 thomas if (fileoffset > filesize) {
608 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
609 e4a6dbc6 2022-07-24 thomas "read beyond the size of deltify file at offset %lld",
610 e4a6dbc6 2022-07-24 thomas (long long)fileoffset);
613 2b0ae357 2022-01-10 thomas p = data + fileoffset;
614 2b0ae357 2022-01-10 thomas q = basedata + base_offset;
615 2b0ae357 2022-01-10 thomas maxlen = MIN(basefile_size - base_offset, filesize - fileoffset);
616 2b0ae357 2022-01-10 thomas for (i = 0; i < maxlen && *blocklen < (1 << 24) - 1; i++) {
617 2b0ae357 2022-01-10 thomas if (p[i] != q[i])
619 2b0ae357 2022-01-10 thomas (*blocklen)++;
622 69aa0e90 2021-03-10 stsp return NULL;
625 69aa0e90 2021-03-10 stsp const struct got_error *
626 69aa0e90 2021-03-10 stsp got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
627 cc524d36 2022-05-31 thomas FILE *f, off_t fileoffset, off_t filesize, uint32_t seed,
628 69aa0e90 2021-03-10 stsp struct got_delta_table *dt, FILE *basefile,
629 f34b169e 2021-06-18 stsp off_t basefile_offset0, off_t basefile_size)
631 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
632 69aa0e90 2021-03-10 stsp const off_t offset0 = fileoffset;
633 dd29967c 2021-08-22 stsp size_t nalloc = 0;
634 dd29967c 2021-08-22 stsp const size_t alloc_chunk_size = 64;
636 69aa0e90 2021-03-10 stsp *deltas = NULL;
637 69aa0e90 2021-03-10 stsp *ndeltas = 0;
640 69aa0e90 2021-03-10 stsp * offset0 indicates where data to be deltified begins.
641 69aa0e90 2021-03-10 stsp * For example, we want to avoid deltifying a Git object header at
642 69aa0e90 2021-03-10 stsp * the beginning of the file.
644 69aa0e90 2021-03-10 stsp if (fseeko(f, offset0, SEEK_SET) == -1)
645 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
647 dd29967c 2021-08-22 stsp *deltas = reallocarray(NULL, alloc_chunk_size,
648 dd29967c 2021-08-22 stsp sizeof(struct got_delta_instruction));
649 dd29967c 2021-08-22 stsp if (*deltas == NULL)
650 dd29967c 2021-08-22 stsp return got_error_from_errno("reallocarray");
651 dd29967c 2021-08-22 stsp nalloc = alloc_chunk_size;
653 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
654 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
655 69aa0e90 2021-03-10 stsp off_t blocklen;
656 69aa0e90 2021-03-10 stsp struct got_delta_block *block;
657 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
660 9a8dc2b3 2021-06-18 stsp if (blocklen == 0) {
661 9a8dc2b3 2021-06-18 stsp /* Source remainder from the file itself. */
662 9a8dc2b3 2021-06-18 stsp if (fileoffset < filesize) {
663 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
664 dd29967c 2021-08-22 stsp alloc_chunk_size, 0, fileoffset - offset0,
665 9a8dc2b3 2021-06-18 stsp filesize - fileoffset);
669 cc524d36 2022-05-31 thomas err = lookupblk(&block, dt, buf, blocklen, seed, basefile,
670 740bba1c 2021-06-18 stsp basefile_offset0);
673 69aa0e90 2021-03-10 stsp if (block != NULL) {
675 69aa0e90 2021-03-10 stsp * We have found a matching block in the delta base.
676 69aa0e90 2021-03-10 stsp * Attempt to stretch the block as far as possible and
677 69aa0e90 2021-03-10 stsp * generate a copy instruction.
679 740bba1c 2021-06-18 stsp err = stretchblk(basefile, basefile_offset0, block,
680 740bba1c 2021-06-18 stsp f, filesize, &blocklen);
683 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
684 dd29967c 2021-08-22 stsp alloc_chunk_size, 1, block->offset, blocklen);
689 69aa0e90 2021-03-10 stsp * No match.
690 69aa0e90 2021-03-10 stsp * This block needs to be sourced from the file itself.
692 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
693 dd29967c 2021-08-22 stsp alloc_chunk_size, 0, fileoffset - offset0, blocklen);
697 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
698 0af64e86 2021-08-22 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1) {
699 0af64e86 2021-08-22 stsp err = got_error_from_errno("fseeko");
705 69aa0e90 2021-03-10 stsp free(*deltas);
706 69aa0e90 2021-03-10 stsp *deltas = NULL;
707 69aa0e90 2021-03-10 stsp *ndeltas = 0;
709 69aa0e90 2021-03-10 stsp return err;
712 2b0ae357 2022-01-10 thomas const struct got_error *
713 2b0ae357 2022-01-10 thomas got_deltify_file_mem(struct got_delta_instruction **deltas, int *ndeltas,
714 cc524d36 2022-05-31 thomas FILE *f, off_t fileoffset, off_t filesize, uint32_t seed,
715 2b0ae357 2022-01-10 thomas struct got_delta_table *dt, uint8_t *basedata,
716 2b0ae357 2022-01-10 thomas off_t basefile_offset0, off_t basefile_size)
718 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
719 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
720 2b0ae357 2022-01-10 thomas size_t nalloc = 0;
721 2b0ae357 2022-01-10 thomas const size_t alloc_chunk_size = 64;
723 2b0ae357 2022-01-10 thomas *deltas = NULL;
724 2b0ae357 2022-01-10 thomas *ndeltas = 0;
727 2b0ae357 2022-01-10 thomas * offset0 indicates where data to be deltified begins.
728 2b0ae357 2022-01-10 thomas * For example, we want to avoid deltifying a Git object header at
729 2b0ae357 2022-01-10 thomas * the beginning of the file.
731 2b0ae357 2022-01-10 thomas if (fseeko(f, offset0, SEEK_SET) == -1)
732 2b0ae357 2022-01-10 thomas return got_error_from_errno("fseeko");
734 2b0ae357 2022-01-10 thomas *deltas = reallocarray(NULL, alloc_chunk_size,
735 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_instruction));
736 2b0ae357 2022-01-10 thomas if (*deltas == NULL)
737 2b0ae357 2022-01-10 thomas return got_error_from_errno("reallocarray");
738 2b0ae357 2022-01-10 thomas nalloc = alloc_chunk_size;
740 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
741 2b0ae357 2022-01-10 thomas uint8_t buf[GOT_DELTIFY_MAXCHUNK];
742 2b0ae357 2022-01-10 thomas off_t blocklen;
743 2b0ae357 2022-01-10 thomas struct got_delta_block *block;
744 2b0ae357 2022-01-10 thomas err = nextblk(buf, &blocklen, f);
747 2b0ae357 2022-01-10 thomas if (blocklen == 0) {
748 2b0ae357 2022-01-10 thomas /* Source remainder from the file itself. */
749 2b0ae357 2022-01-10 thomas if (fileoffset < filesize) {
750 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
751 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0,
752 2b0ae357 2022-01-10 thomas filesize - fileoffset);
756 cc524d36 2022-05-31 thomas err = lookupblk_mem(&block, dt, buf, blocklen, seed, basedata,
757 2b0ae357 2022-01-10 thomas basefile_offset0);
760 2b0ae357 2022-01-10 thomas if (block != NULL) {
762 2b0ae357 2022-01-10 thomas * We have found a matching block in the delta base.
763 2b0ae357 2022-01-10 thomas * Attempt to stretch the block as far as possible and
764 2b0ae357 2022-01-10 thomas * generate a copy instruction.
766 2b0ae357 2022-01-10 thomas err = stretchblk_file_mem(basedata, basefile_offset0,
767 2b0ae357 2022-01-10 thomas basefile_size, block, f, filesize, &blocklen);
770 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
771 2b0ae357 2022-01-10 thomas alloc_chunk_size, 1, block->offset, blocklen);
776 2b0ae357 2022-01-10 thomas * No match.
777 2b0ae357 2022-01-10 thomas * This block needs to be sourced from the file itself.
779 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
780 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0, blocklen);
784 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
785 2b0ae357 2022-01-10 thomas if (fseeko(f, fileoffset, SEEK_SET) == -1) {
786 2b0ae357 2022-01-10 thomas err = got_error_from_errno("fseeko");
791 2b0ae357 2022-01-10 thomas if (err) {
792 2b0ae357 2022-01-10 thomas free(*deltas);
793 2b0ae357 2022-01-10 thomas *deltas = NULL;
794 2b0ae357 2022-01-10 thomas *ndeltas = 0;
796 2b0ae357 2022-01-10 thomas return err;
799 2b0ae357 2022-01-10 thomas const struct got_error *
800 2b0ae357 2022-01-10 thomas got_deltify_mem_file(struct got_delta_instruction **deltas, int *ndeltas,
801 cc524d36 2022-05-31 thomas uint8_t *data, off_t fileoffset, off_t filesize, uint32_t seed,
802 2b0ae357 2022-01-10 thomas struct got_delta_table *dt, FILE *basefile,
803 2b0ae357 2022-01-10 thomas off_t basefile_offset0, off_t basefile_size)
805 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
806 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
807 2b0ae357 2022-01-10 thomas size_t nalloc = 0;
808 2b0ae357 2022-01-10 thomas const size_t alloc_chunk_size = 64;
810 2b0ae357 2022-01-10 thomas *deltas = NULL;
811 2b0ae357 2022-01-10 thomas *ndeltas = 0;
813 2b0ae357 2022-01-10 thomas *deltas = reallocarray(NULL, alloc_chunk_size,
814 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_instruction));
815 2b0ae357 2022-01-10 thomas if (*deltas == NULL)
816 2b0ae357 2022-01-10 thomas return got_error_from_errno("reallocarray");
817 2b0ae357 2022-01-10 thomas nalloc = alloc_chunk_size;
819 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
820 2b0ae357 2022-01-10 thomas off_t blocklen;
821 2b0ae357 2022-01-10 thomas struct got_delta_block *block;
822 2b0ae357 2022-01-10 thomas err = nextblk_mem(&blocklen, data, fileoffset, filesize);
825 2b0ae357 2022-01-10 thomas if (blocklen == 0) {
826 2b0ae357 2022-01-10 thomas /* Source remainder from the file itself. */
827 2b0ae357 2022-01-10 thomas if (fileoffset < filesize) {
828 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
829 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0,
830 2b0ae357 2022-01-10 thomas filesize - fileoffset);
834 cc524d36 2022-05-31 thomas err = lookupblk(&block, dt, data + fileoffset, blocklen, seed,
835 2b0ae357 2022-01-10 thomas basefile, basefile_offset0);
838 2b0ae357 2022-01-10 thomas if (block != NULL) {
840 2b0ae357 2022-01-10 thomas * We have found a matching block in the delta base.
841 2b0ae357 2022-01-10 thomas * Attempt to stretch the block as far as possible and
842 2b0ae357 2022-01-10 thomas * generate a copy instruction.
844 2b0ae357 2022-01-10 thomas err = stretchblk_mem_file(basefile, basefile_offset0,
845 2b0ae357 2022-01-10 thomas block, data, fileoffset + blocklen, filesize,
846 2b0ae357 2022-01-10 thomas &blocklen);
849 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
850 2b0ae357 2022-01-10 thomas alloc_chunk_size, 1, block->offset, blocklen);
855 2b0ae357 2022-01-10 thomas * No match.
856 2b0ae357 2022-01-10 thomas * This block needs to be sourced from the file itself.
858 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
859 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0, blocklen);
863 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
866 2b0ae357 2022-01-10 thomas if (err) {
867 2b0ae357 2022-01-10 thomas free(*deltas);
868 2b0ae357 2022-01-10 thomas *deltas = NULL;
869 2b0ae357 2022-01-10 thomas *ndeltas = 0;
871 2b0ae357 2022-01-10 thomas return err;
874 2b0ae357 2022-01-10 thomas const struct got_error *
875 2b0ae357 2022-01-10 thomas got_deltify_mem_mem(struct got_delta_instruction **deltas, int *ndeltas,
876 cc524d36 2022-05-31 thomas uint8_t *data, off_t fileoffset, off_t filesize, uint32_t seed,
877 2b0ae357 2022-01-10 thomas struct got_delta_table *dt, uint8_t *basedata,
878 2b0ae357 2022-01-10 thomas off_t basefile_offset0, off_t basefile_size)
880 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
881 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
882 2b0ae357 2022-01-10 thomas size_t nalloc = 0;
883 2b0ae357 2022-01-10 thomas const size_t alloc_chunk_size = 64;
885 2b0ae357 2022-01-10 thomas *deltas = NULL;
886 2b0ae357 2022-01-10 thomas *ndeltas = 0;
888 2b0ae357 2022-01-10 thomas *deltas = reallocarray(NULL, alloc_chunk_size,
889 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_instruction));
890 2b0ae357 2022-01-10 thomas if (*deltas == NULL)
891 2b0ae357 2022-01-10 thomas return got_error_from_errno("reallocarray");
892 2b0ae357 2022-01-10 thomas nalloc = alloc_chunk_size;
894 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
895 2b0ae357 2022-01-10 thomas off_t blocklen;
896 2b0ae357 2022-01-10 thomas struct got_delta_block *block;
897 2b0ae357 2022-01-10 thomas err = nextblk_mem(&blocklen, data, fileoffset, filesize);
900 2b0ae357 2022-01-10 thomas if (blocklen == 0) {
901 2b0ae357 2022-01-10 thomas /* Source remainder from the file itself. */
902 2b0ae357 2022-01-10 thomas if (fileoffset < filesize) {
903 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
904 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0,
905 2b0ae357 2022-01-10 thomas filesize - fileoffset);
909 2b0ae357 2022-01-10 thomas err = lookupblk_mem(&block, dt, data + fileoffset, blocklen,
910 cc524d36 2022-05-31 thomas seed, basedata, basefile_offset0);
913 2b0ae357 2022-01-10 thomas if (block != NULL) {
915 2b0ae357 2022-01-10 thomas * We have found a matching block in the delta base.
916 2b0ae357 2022-01-10 thomas * Attempt to stretch the block as far as possible and
917 2b0ae357 2022-01-10 thomas * generate a copy instruction.
919 2b0ae357 2022-01-10 thomas err = stretchblk_mem_mem(basedata, basefile_offset0,
920 2b0ae357 2022-01-10 thomas basefile_size, block, data, fileoffset + blocklen,
921 2b0ae357 2022-01-10 thomas filesize, &blocklen);
924 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
925 2b0ae357 2022-01-10 thomas alloc_chunk_size, 1, block->offset, blocklen);
930 2b0ae357 2022-01-10 thomas * No match.
931 2b0ae357 2022-01-10 thomas * This block needs to be sourced from the file itself.
933 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
934 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0, blocklen);
938 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
941 2b0ae357 2022-01-10 thomas if (err) {
942 2b0ae357 2022-01-10 thomas free(*deltas);
943 2b0ae357 2022-01-10 thomas *deltas = NULL;
944 2b0ae357 2022-01-10 thomas *ndeltas = 0;
946 2b0ae357 2022-01-10 thomas return err;