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 2633cf30 2024-03-02 thomas static const struct got_error *addblk(struct got_delta_table *, FILE *,
92 2633cf30 2024-03-02 thomas uint8_t *, off_t, off_t, off_t, uint32_t);
94 b7bf7afc 2022-02-12 thomas static uint32_t
95 cc524d36 2022-05-31 thomas hashblk(const unsigned char *p, off_t n, uint32_t seed)
97 cc524d36 2022-05-31 thomas return murmurhash2(p, n, seed);
100 f364801d 2021-03-10 stsp static const struct got_error *
101 2633cf30 2024-03-02 thomas lookup(int *n, struct got_delta_table *dt, FILE *f, off_t file_offset0,
102 2633cf30 2024-03-02 thomas off_t len, off_t offset, uint32_t h)
104 2633cf30 2024-03-02 thomas struct got_delta_block *block;
105 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
106 0d15f6dc 2021-06-13 stsp uint8_t buf2[GOT_DELTIFY_MAXCHUNK];
107 69aa0e90 2021-03-10 stsp size_t r = 0;
110 2633cf30 2024-03-02 thomas for (i = h % dt->size; dt->offs[i] != 0; i = (i + 1) % dt->size) {
111 2633cf30 2024-03-02 thomas block = &dt->blocks[dt->offs[i] - 1];
112 2633cf30 2024-03-02 thomas if (block->len != len || block->hash != h)
113 2633cf30 2024-03-02 thomas continue;
115 2633cf30 2024-03-02 thomas if (r == 0) {
116 2633cf30 2024-03-02 thomas if (fseeko(f, file_offset0 + offset, SEEK_SET) == -1)
117 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
118 2633cf30 2024-03-02 thomas r = fread(buf, 1, len, f);
119 2633cf30 2024-03-02 thomas if (r != len) {
120 2633cf30 2024-03-02 thomas if (!ferror(f))
121 2633cf30 2024-03-02 thomas break; /* why? */
122 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
126 2633cf30 2024-03-02 thomas if (fseeko(f, file_offset0 + block->offset, SEEK_SET) == -1)
127 2633cf30 2024-03-02 thomas return got_error_from_errno("fseeko");
128 2633cf30 2024-03-02 thomas if (fread(buf2, 1, len, f) != len)
129 2633cf30 2024-03-02 thomas return got_ferror(f, GOT_ERR_IO);
130 2633cf30 2024-03-02 thomas if (memcmp(buf, buf2, len) == 0)
135 2633cf30 2024-03-02 thomas return NULL;
138 2633cf30 2024-03-02 thomas static const struct got_error *
139 2633cf30 2024-03-02 thomas lookup_mem(int *n, struct got_delta_table *dt, uint8_t *basedata,
140 2633cf30 2024-03-02 thomas off_t basefile_offset0, uint8_t *p, off_t len, uint32_t h)
142 2633cf30 2024-03-02 thomas struct got_delta_block *block;
143 2633cf30 2024-03-02 thomas uint8_t *d;
146 2633cf30 2024-03-02 thomas for (i = h % dt->size; dt->offs[i] != 0; i = (i + 1) % dt->size) {
147 2633cf30 2024-03-02 thomas block = &dt->blocks[dt->offs[i] - 1];
148 2633cf30 2024-03-02 thomas if (len == block->len && h == block->hash) {
149 2633cf30 2024-03-02 thomas d = basedata + basefile_offset0 + block->offset;
150 2633cf30 2024-03-02 thomas if (memcmp(d, p, len) == 0)
156 2633cf30 2024-03-02 thomas return NULL;
159 2b0ae357 2022-01-10 thomas static const struct got_error *
160 2633cf30 2024-03-02 thomas resizeht(struct got_delta_table *dt, FILE *f, off_t offset0, uint8_t *data,
161 2633cf30 2024-03-02 thomas off_t len)
163 2633cf30 2024-03-02 thomas const struct got_error *err;
164 2633cf30 2024-03-02 thomas struct got_delta_block *b;
165 2633cf30 2024-03-02 thomas size_t newsize, oldsize;
166 2633cf30 2024-03-02 thomas int i, n;
168 2633cf30 2024-03-02 thomas if (dt->nblocks == dt->nalloc) {
169 2633cf30 2024-03-02 thomas newsize = dt->nalloc + 256;
170 2633cf30 2024-03-02 thomas b = reallocarray(dt->blocks, newsize, sizeof(*dt->blocks));
171 2633cf30 2024-03-02 thomas if (b == NULL)
172 2633cf30 2024-03-02 thomas return got_error_from_errno("reallocarray");
173 2633cf30 2024-03-02 thomas dt->blocks = b;
174 2633cf30 2024-03-02 thomas dt->nalloc = newsize;
177 2633cf30 2024-03-02 thomas if (dt->blocks == NULL || dt->len * 100 / dt->size > 70) {
178 2633cf30 2024-03-02 thomas oldsize = dt->size;
179 2633cf30 2024-03-02 thomas dt->size *= 2;
180 2633cf30 2024-03-02 thomas free(dt->offs);
181 2633cf30 2024-03-02 thomas dt->offs = calloc(dt->size, sizeof(*dt->offs));
182 2633cf30 2024-03-02 thomas if (dt->offs == NULL) {
183 2633cf30 2024-03-02 thomas dt->size = oldsize;
184 2633cf30 2024-03-02 thomas return got_error_from_errno("calloc");
187 2633cf30 2024-03-02 thomas for (i = 0; i < dt->nblocks; ++i) {
188 2633cf30 2024-03-02 thomas b = &dt->blocks[i];
190 2633cf30 2024-03-02 thomas err = lookup(&n, dt, f, offset0, b->len,
191 2633cf30 2024-03-02 thomas b->offset, b->hash);
193 2633cf30 2024-03-02 thomas err = lookup_mem(&n, dt, data, offset0,
194 2633cf30 2024-03-02 thomas data + b->offset, b->len, b->hash);
195 2633cf30 2024-03-02 thomas if (err) {
196 2633cf30 2024-03-02 thomas free(dt->offs);
197 2633cf30 2024-03-02 thomas dt->offs = NULL;
198 2633cf30 2024-03-02 thomas dt->size = oldsize;
199 2633cf30 2024-03-02 thomas return err;
201 2633cf30 2024-03-02 thomas assert(dt->offs[n] == 0);
202 2633cf30 2024-03-02 thomas dt->offs[n] = i + 1;
206 2633cf30 2024-03-02 thomas return NULL;
209 69aa0e90 2021-03-10 stsp static const struct got_error *
210 2633cf30 2024-03-02 thomas addblk(struct got_delta_table *dt, FILE *f, uint8_t *data, off_t file_offset0,
211 2633cf30 2024-03-02 thomas off_t len, off_t offset, uint32_t h)
213 2633cf30 2024-03-02 thomas const struct got_error *err;
214 2633cf30 2024-03-02 thomas struct got_delta_block *block;
217 2633cf30 2024-03-02 thomas if (len == 0)
218 2633cf30 2024-03-02 thomas return NULL;
220 2633cf30 2024-03-02 thomas err = resizeht(dt, f, file_offset0, data, len);
222 2633cf30 2024-03-02 thomas return err;
225 2633cf30 2024-03-02 thomas err = lookup(&i, dt, f, file_offset0, len, offset, h);
227 2633cf30 2024-03-02 thomas err = lookup_mem(&i, dt, data, file_offset0, data + offset,
230 2633cf30 2024-03-02 thomas return err;
232 2633cf30 2024-03-02 thomas if (dt->offs[i] != 0)
233 2633cf30 2024-03-02 thomas return NULL; /* found */
235 2633cf30 2024-03-02 thomas dt->offs[i] = dt->nblocks + 1;
236 2633cf30 2024-03-02 thomas block = &dt->blocks[dt->nblocks];
237 2633cf30 2024-03-02 thomas block->len = len;
238 2633cf30 2024-03-02 thomas block->offset = offset;
239 2633cf30 2024-03-02 thomas block->hash = h;
241 2633cf30 2024-03-02 thomas dt->nblocks++;
242 2633cf30 2024-03-02 thomas dt->len++;
244 2633cf30 2024-03-02 thomas return NULL;
247 2633cf30 2024-03-02 thomas static const struct got_error *
248 69aa0e90 2021-03-10 stsp lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
249 cc524d36 2022-05-31 thomas unsigned char *p, off_t len, uint32_t seed, FILE *basefile,
250 cc524d36 2022-05-31 thomas off_t basefile_offset0)
253 b7bf7afc 2022-02-12 thomas uint32_t h;
254 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
255 2633cf30 2024-03-02 thomas struct got_delta_block *b;
258 69aa0e90 2021-03-10 stsp *block = NULL;
260 cc524d36 2022-05-31 thomas h = hashblk(p, len, seed);
262 2633cf30 2024-03-02 thomas for (i = h % dt->size; dt->offs[i] != 0; i = (i + 1) % dt->size) {
263 2633cf30 2024-03-02 thomas b = &dt->blocks[dt->offs[i] - 1];
264 2633cf30 2024-03-02 thomas if (b->hash != h || b->len != len)
266 2633cf30 2024-03-02 thomas if (fseeko(basefile, basefile_offset0 + b->offset,
267 740bba1c 2021-06-18 stsp SEEK_SET) == -1)
268 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
269 69aa0e90 2021-03-10 stsp r = fread(buf, 1, len, basefile);
270 69aa0e90 2021-03-10 stsp if (r != len)
271 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
272 69aa0e90 2021-03-10 stsp if (memcmp(p, buf, len) == 0) {
273 2633cf30 2024-03-02 thomas *block = b;
277 f364801d 2021-03-10 stsp return NULL;
280 69aa0e90 2021-03-10 stsp static const struct got_error *
281 2b0ae357 2022-01-10 thomas lookupblk_mem(struct got_delta_block **block, struct got_delta_table *dt,
282 cc524d36 2022-05-31 thomas unsigned char *p, off_t len, uint32_t seed, uint8_t *basedata,
283 cc524d36 2022-05-31 thomas off_t basefile_offset0)
285 2633cf30 2024-03-02 thomas const struct got_error *err;
287 b7bf7afc 2022-02-12 thomas uint32_t h;
289 2b0ae357 2022-01-10 thomas *block = NULL;
290 cc524d36 2022-05-31 thomas h = hashblk(p, len, seed);
291 2633cf30 2024-03-02 thomas err = lookup_mem(&n, dt, basedata, basefile_offset0, p, len, h);
293 2633cf30 2024-03-02 thomas return err;
294 2633cf30 2024-03-02 thomas if (dt->offs[n] != 0)
295 2633cf30 2024-03-02 thomas *block = &dt->blocks[dt->offs[n] - 1];
296 2b0ae357 2022-01-10 thomas return NULL;
299 2b0ae357 2022-01-10 thomas static const struct got_error *
300 69aa0e90 2021-03-10 stsp nextblk(uint8_t *buf, off_t *blocklen, FILE *f)
302 f364801d 2021-03-10 stsp uint32_t gh;
303 69aa0e90 2021-03-10 stsp const unsigned char *p;
305 69aa0e90 2021-03-10 stsp off_t pos = ftello(f);
307 69aa0e90 2021-03-10 stsp *blocklen = 0;
309 69aa0e90 2021-03-10 stsp r = fread(buf, 1, GOT_DELTIFY_MAXCHUNK, f);
310 69aa0e90 2021-03-10 stsp if (r == 0 && ferror(f))
311 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
312 69aa0e90 2021-03-10 stsp if (r < GOT_DELTIFY_MINCHUNK)
313 69aa0e90 2021-03-10 stsp return NULL; /* no more delta-worthy blocks left */
315 69aa0e90 2021-03-10 stsp /* Got a deltifiable block. Find the split-point where it ends. */
316 69aa0e90 2021-03-10 stsp p = buf + GOT_DELTIFY_MINCHUNK;
318 69aa0e90 2021-03-10 stsp while (p != buf + r) {
319 f364801d 2021-03-10 stsp gh = (gh << 1) + geartab[*p++];
320 69aa0e90 2021-03-10 stsp if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
324 69aa0e90 2021-03-10 stsp *blocklen = (p - buf);
325 69aa0e90 2021-03-10 stsp if (fseeko(f, pos + *blocklen, SEEK_SET) == -1)
326 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
328 69aa0e90 2021-03-10 stsp return NULL;
331 2b0ae357 2022-01-10 thomas static const struct got_error *
332 2b0ae357 2022-01-10 thomas nextblk_mem(off_t *blocklen, uint8_t *data, off_t fileoffset, off_t filesize)
334 2b0ae357 2022-01-10 thomas uint32_t gh;
335 2b0ae357 2022-01-10 thomas const unsigned char *p;
337 2b0ae357 2022-01-10 thomas *blocklen = 0;
339 2b0ae357 2022-01-10 thomas if (fileoffset >= filesize ||
340 2b0ae357 2022-01-10 thomas filesize - fileoffset < GOT_DELTIFY_MINCHUNK)
341 2b0ae357 2022-01-10 thomas return NULL; /* no more delta-worthy blocks left */
343 2b0ae357 2022-01-10 thomas /* Got a deltifiable block. Find the split-point where it ends. */
344 2b0ae357 2022-01-10 thomas p = data + fileoffset + GOT_DELTIFY_MINCHUNK;
346 2b0ae357 2022-01-10 thomas while (p != data + MIN(fileoffset + GOT_DELTIFY_MAXCHUNK, filesize)) {
347 2b0ae357 2022-01-10 thomas gh = (gh << 1) + geartab[*p++];
348 2b0ae357 2022-01-10 thomas if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
352 2b0ae357 2022-01-10 thomas *blocklen = (p - (data + fileoffset));
353 2b0ae357 2022-01-10 thomas return NULL;
356 f364801d 2021-03-10 stsp const struct got_error *
357 69aa0e90 2021-03-10 stsp got_deltify_init(struct got_delta_table **dt, FILE *f, off_t fileoffset,
358 cc524d36 2022-05-31 thomas off_t filesize, uint32_t seed)
360 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
361 b7bf7afc 2022-02-12 thomas uint32_t h;
362 740bba1c 2021-06-18 stsp const off_t offset0 = fileoffset;
364 69aa0e90 2021-03-10 stsp *dt = calloc(1, sizeof(**dt));
365 69aa0e90 2021-03-10 stsp if (*dt == NULL)
366 f364801d 2021-03-10 stsp return got_error_from_errno("calloc");
368 69aa0e90 2021-03-10 stsp (*dt)->nblocks = 0;
369 69aa0e90 2021-03-10 stsp (*dt)->nalloc = 128;
370 69aa0e90 2021-03-10 stsp (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
371 69aa0e90 2021-03-10 stsp if ((*dt)->blocks == NULL) {
372 69aa0e90 2021-03-10 stsp err = got_error_from_errno("calloc");
376 2633cf30 2024-03-02 thomas (*dt)->size = 128;
377 2633cf30 2024-03-02 thomas (*dt)->offs = calloc((*dt)->size, sizeof(*(*dt)->offs));
378 2633cf30 2024-03-02 thomas if ((*dt)->offs == NULL) {
379 2633cf30 2024-03-02 thomas err = got_error_from_errno("calloc");
380 2633cf30 2024-03-02 thomas goto done;
383 69aa0e90 2021-03-10 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
384 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
386 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
387 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
388 69aa0e90 2021-03-10 stsp off_t blocklen;
389 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
392 69aa0e90 2021-03-10 stsp if (blocklen == 0)
394 cc524d36 2022-05-31 thomas h = hashblk(buf, blocklen, seed);
395 2633cf30 2024-03-02 thomas err = addblk(*dt, f, NULL, offset0, blocklen,
396 740bba1c 2021-06-18 stsp fileoffset - offset0, h);
399 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
400 68bdcdc2 2021-06-11 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
401 68bdcdc2 2021-06-11 stsp return got_error_from_errno("fseeko");
404 2b0ae357 2022-01-10 thomas if (err) {
405 2b0ae357 2022-01-10 thomas free((*dt)->blocks);
406 2b0ae357 2022-01-10 thomas free(*dt);
407 2b0ae357 2022-01-10 thomas *dt = NULL;
410 2b0ae357 2022-01-10 thomas return err;
413 2b0ae357 2022-01-10 thomas const struct got_error *
414 2b0ae357 2022-01-10 thomas got_deltify_init_mem(struct got_delta_table **dt, uint8_t *data,
415 cc524d36 2022-05-31 thomas off_t fileoffset, off_t filesize, uint32_t seed)
417 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
418 90dfa2bf 2022-02-13 thomas uint32_t h;
419 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
421 2b0ae357 2022-01-10 thomas *dt = calloc(1, sizeof(**dt));
422 2b0ae357 2022-01-10 thomas if (*dt == NULL)
423 2b0ae357 2022-01-10 thomas return got_error_from_errno("calloc");
425 2b0ae357 2022-01-10 thomas (*dt)->nblocks = 0;
426 2b0ae357 2022-01-10 thomas (*dt)->nalloc = 128;
427 2b0ae357 2022-01-10 thomas (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
428 2b0ae357 2022-01-10 thomas if ((*dt)->blocks == NULL) {
429 2633cf30 2024-03-02 thomas err = got_error_from_errno("calloc");
430 2633cf30 2024-03-02 thomas goto done;
433 2633cf30 2024-03-02 thomas (*dt)->size = 128;
434 2633cf30 2024-03-02 thomas (*dt)->offs = calloc((*dt)->size, sizeof(*(*dt)->offs));
435 2633cf30 2024-03-02 thomas if ((*dt)->offs == NULL) {
436 2b0ae357 2022-01-10 thomas err = got_error_from_errno("calloc");
437 2b0ae357 2022-01-10 thomas goto done;
440 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
441 2b0ae357 2022-01-10 thomas off_t blocklen;
442 2b0ae357 2022-01-10 thomas err = nextblk_mem(&blocklen, data, fileoffset, filesize);
444 2b0ae357 2022-01-10 thomas goto done;
445 2b0ae357 2022-01-10 thomas if (blocklen == 0)
447 cc524d36 2022-05-31 thomas h = hashblk(data + fileoffset, blocklen, seed);
448 2633cf30 2024-03-02 thomas err = addblk(*dt, NULL, data, offset0, blocklen,
449 2b0ae357 2022-01-10 thomas fileoffset - offset0, h);
451 2b0ae357 2022-01-10 thomas goto done;
452 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
456 69aa0e90 2021-03-10 stsp free((*dt)->blocks);
458 69aa0e90 2021-03-10 stsp *dt = NULL;
461 69aa0e90 2021-03-10 stsp return err;
465 69aa0e90 2021-03-10 stsp got_deltify_free(struct got_delta_table *dt)
467 dbbf4a5f 2021-05-20 stsp if (dt == NULL)
469 69aa0e90 2021-03-10 stsp free(dt->blocks);
470 2633cf30 2024-03-02 thomas free(dt->offs);
474 f364801d 2021-03-10 stsp static const struct got_error *
475 dd29967c 2021-08-22 stsp emitdelta(struct got_delta_instruction **deltas, size_t *nalloc, int *ndeltas,
476 dd29967c 2021-08-22 stsp const size_t alloc_chunk_size, int copy, off_t offset, off_t len)
478 69aa0e90 2021-03-10 stsp struct got_delta_instruction *d, *p;
480 dd29967c 2021-08-22 stsp if (*nalloc < *ndeltas + alloc_chunk_size) {
481 dd29967c 2021-08-22 stsp p = reallocarray(*deltas, *nalloc + alloc_chunk_size,
482 dd29967c 2021-08-22 stsp sizeof(struct got_delta_instruction));
483 dd29967c 2021-08-22 stsp if (p == NULL)
484 7d76c4e7 2021-10-15 thomas return got_error_from_errno("reallocarray");
485 dd29967c 2021-08-22 stsp *deltas = p;
486 dd29967c 2021-08-22 stsp *nalloc += alloc_chunk_size;
488 69aa0e90 2021-03-10 stsp *ndeltas += 1;
489 69aa0e90 2021-03-10 stsp d = &(*deltas)[*ndeltas - 1];
490 69aa0e90 2021-03-10 stsp d->copy = copy;
491 69aa0e90 2021-03-10 stsp d->offset = offset;
492 56e7a15f 2021-03-10 stsp d->len = len;
493 f364801d 2021-03-10 stsp return NULL;
496 69aa0e90 2021-03-10 stsp static const struct got_error *
497 740bba1c 2021-06-18 stsp stretchblk(FILE *basefile, off_t base_offset0, struct got_delta_block *block,
498 740bba1c 2021-06-18 stsp FILE *f, off_t filesize, off_t *blocklen)
500 69aa0e90 2021-03-10 stsp uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
501 69aa0e90 2021-03-10 stsp size_t base_r, r, i;
502 69aa0e90 2021-03-10 stsp int buf_equal = 1;
504 5de743f8 2021-08-29 stsp if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
505 5de743f8 2021-08-29 stsp SEEK_SET) == -1)
506 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
508 69aa0e90 2021-03-10 stsp while (buf_equal && *blocklen < (1 << 24) - 1) {
509 69aa0e90 2021-03-10 stsp base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
510 69aa0e90 2021-03-10 stsp if (base_r == 0) {
511 69aa0e90 2021-03-10 stsp if (ferror(basefile))
512 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
515 69aa0e90 2021-03-10 stsp r = fread(buf, 1, sizeof(buf), f);
516 69aa0e90 2021-03-10 stsp if (r == 0) {
517 69aa0e90 2021-03-10 stsp if (ferror(f))
518 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
521 69aa0e90 2021-03-10 stsp for (i = 0; i < MIN(base_r, r); i++) {
522 69aa0e90 2021-03-10 stsp if (buf[i] != basebuf[i]) {
523 69aa0e90 2021-03-10 stsp buf_equal = 0;
526 69aa0e90 2021-03-10 stsp (*blocklen)++;
530 2b0ae357 2022-01-10 thomas return NULL;
533 2b0ae357 2022-01-10 thomas static const struct got_error *
534 2b0ae357 2022-01-10 thomas stretchblk_file_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
535 2b0ae357 2022-01-10 thomas struct got_delta_block *block, FILE *f, off_t filesize, off_t *blocklen)
537 2b0ae357 2022-01-10 thomas uint8_t buf[GOT_DELTIFY_MAXCHUNK];
538 2b0ae357 2022-01-10 thomas size_t r, i;
539 2b0ae357 2022-01-10 thomas int buf_equal = 1;
540 2b0ae357 2022-01-10 thomas off_t base_offset = base_offset0 + block->offset + *blocklen;
542 2b0ae357 2022-01-10 thomas if (base_offset > basefile_size) {
543 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
544 e4a6dbc6 2022-07-24 thomas "read beyond the size of delta base at offset %lld",
545 e4a6dbc6 2022-07-24 thomas (long long)base_offset);
548 2b0ae357 2022-01-10 thomas while (buf_equal && *blocklen < (1 << 24) - 1) {
549 2b0ae357 2022-01-10 thomas if (base_offset + *blocklen >= basefile_size)
551 2b0ae357 2022-01-10 thomas r = fread(buf, 1, sizeof(buf), f);
552 2b0ae357 2022-01-10 thomas if (r == 0) {
553 2b0ae357 2022-01-10 thomas if (ferror(f))
554 2b0ae357 2022-01-10 thomas return got_ferror(f, GOT_ERR_IO);
557 2b0ae357 2022-01-10 thomas for (i = 0; i < MIN(basefile_size - base_offset, r); i++) {
558 2b0ae357 2022-01-10 thomas if (buf[i] != *(basedata + base_offset + i)) {
559 2b0ae357 2022-01-10 thomas buf_equal = 0;
562 2b0ae357 2022-01-10 thomas (*blocklen)++;
566 2b0ae357 2022-01-10 thomas return NULL;
569 2b0ae357 2022-01-10 thomas static const struct got_error *
570 2b0ae357 2022-01-10 thomas stretchblk_mem_file(FILE *basefile, off_t base_offset0,
571 2b0ae357 2022-01-10 thomas struct got_delta_block *block, uint8_t *data, off_t fileoffset,
572 2b0ae357 2022-01-10 thomas off_t filesize, off_t *blocklen)
574 2b0ae357 2022-01-10 thomas uint8_t basebuf[GOT_DELTIFY_MAXCHUNK];
575 2b0ae357 2022-01-10 thomas size_t base_r, i;
576 2b0ae357 2022-01-10 thomas int buf_equal = 1;
578 2b0ae357 2022-01-10 thomas if (fileoffset > filesize) {
579 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
580 e4a6dbc6 2022-07-24 thomas "read beyond the size of deltify file at offset %lld",
581 e4a6dbc6 2022-07-24 thomas (long long)fileoffset);
584 2b0ae357 2022-01-10 thomas if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
585 2b0ae357 2022-01-10 thomas SEEK_SET) == -1)
586 2b0ae357 2022-01-10 thomas return got_error_from_errno("fseeko");
588 2b0ae357 2022-01-10 thomas while (buf_equal && *blocklen < (1 << 24) - 1) {
589 2b0ae357 2022-01-10 thomas if (fileoffset + *blocklen >= filesize)
591 2b0ae357 2022-01-10 thomas base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
592 2b0ae357 2022-01-10 thomas if (base_r == 0) {
593 2b0ae357 2022-01-10 thomas if (ferror(basefile))
594 2b0ae357 2022-01-10 thomas return got_ferror(basefile, GOT_ERR_IO);
597 2b0ae357 2022-01-10 thomas for (i = 0; i < MIN(base_r, filesize - fileoffset); i++) {
598 2b0ae357 2022-01-10 thomas if (*(data + fileoffset + i) != basebuf[i]) {
599 2b0ae357 2022-01-10 thomas buf_equal = 0;
602 2b0ae357 2022-01-10 thomas (*blocklen)++;
606 2b0ae357 2022-01-10 thomas return NULL;
609 2b0ae357 2022-01-10 thomas static const struct got_error *
610 2b0ae357 2022-01-10 thomas stretchblk_mem_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
611 2b0ae357 2022-01-10 thomas struct got_delta_block *block, uint8_t *data, off_t fileoffset,
612 2b0ae357 2022-01-10 thomas off_t filesize, off_t *blocklen)
614 2b0ae357 2022-01-10 thomas off_t i, maxlen;
615 2b0ae357 2022-01-10 thomas off_t base_offset = base_offset0 + block->offset + *blocklen;
616 2b0ae357 2022-01-10 thomas uint8_t *p, *q;
618 2b0ae357 2022-01-10 thomas if (base_offset > basefile_size) {
619 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
620 e4a6dbc6 2022-07-24 thomas "read beyond the size of delta base at offset %lld",
621 e4a6dbc6 2022-07-24 thomas (long long)base_offset);
624 2b0ae357 2022-01-10 thomas if (fileoffset > filesize) {
625 2b0ae357 2022-01-10 thomas return got_error_fmt(GOT_ERR_RANGE,
626 e4a6dbc6 2022-07-24 thomas "read beyond the size of deltify file at offset %lld",
627 e4a6dbc6 2022-07-24 thomas (long long)fileoffset);
630 2b0ae357 2022-01-10 thomas p = data + fileoffset;
631 2b0ae357 2022-01-10 thomas q = basedata + base_offset;
632 2b0ae357 2022-01-10 thomas maxlen = MIN(basefile_size - base_offset, filesize - fileoffset);
633 2b0ae357 2022-01-10 thomas for (i = 0; i < maxlen && *blocklen < (1 << 24) - 1; i++) {
634 2b0ae357 2022-01-10 thomas if (p[i] != q[i])
636 2b0ae357 2022-01-10 thomas (*blocklen)++;
639 69aa0e90 2021-03-10 stsp return NULL;
642 69aa0e90 2021-03-10 stsp const struct got_error *
643 69aa0e90 2021-03-10 stsp got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
644 cc524d36 2022-05-31 thomas FILE *f, off_t fileoffset, off_t filesize, uint32_t seed,
645 69aa0e90 2021-03-10 stsp struct got_delta_table *dt, FILE *basefile,
646 f34b169e 2021-06-18 stsp off_t basefile_offset0, off_t basefile_size)
648 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
649 69aa0e90 2021-03-10 stsp const off_t offset0 = fileoffset;
650 dd29967c 2021-08-22 stsp size_t nalloc = 0;
651 dd29967c 2021-08-22 stsp const size_t alloc_chunk_size = 64;
653 69aa0e90 2021-03-10 stsp *deltas = NULL;
654 69aa0e90 2021-03-10 stsp *ndeltas = 0;
657 69aa0e90 2021-03-10 stsp * offset0 indicates where data to be deltified begins.
658 69aa0e90 2021-03-10 stsp * For example, we want to avoid deltifying a Git object header at
659 69aa0e90 2021-03-10 stsp * the beginning of the file.
661 69aa0e90 2021-03-10 stsp if (fseeko(f, offset0, SEEK_SET) == -1)
662 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
664 dd29967c 2021-08-22 stsp *deltas = reallocarray(NULL, alloc_chunk_size,
665 dd29967c 2021-08-22 stsp sizeof(struct got_delta_instruction));
666 dd29967c 2021-08-22 stsp if (*deltas == NULL)
667 dd29967c 2021-08-22 stsp return got_error_from_errno("reallocarray");
668 dd29967c 2021-08-22 stsp nalloc = alloc_chunk_size;
670 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
671 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
672 69aa0e90 2021-03-10 stsp off_t blocklen;
673 69aa0e90 2021-03-10 stsp struct got_delta_block *block;
674 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
677 9a8dc2b3 2021-06-18 stsp if (blocklen == 0) {
678 9a8dc2b3 2021-06-18 stsp /* Source remainder from the file itself. */
679 9a8dc2b3 2021-06-18 stsp if (fileoffset < filesize) {
680 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
681 dd29967c 2021-08-22 stsp alloc_chunk_size, 0, fileoffset - offset0,
682 9a8dc2b3 2021-06-18 stsp filesize - fileoffset);
686 cc524d36 2022-05-31 thomas err = lookupblk(&block, dt, buf, blocklen, seed, basefile,
687 740bba1c 2021-06-18 stsp basefile_offset0);
690 69aa0e90 2021-03-10 stsp if (block != NULL) {
692 69aa0e90 2021-03-10 stsp * We have found a matching block in the delta base.
693 69aa0e90 2021-03-10 stsp * Attempt to stretch the block as far as possible and
694 69aa0e90 2021-03-10 stsp * generate a copy instruction.
696 740bba1c 2021-06-18 stsp err = stretchblk(basefile, basefile_offset0, block,
697 740bba1c 2021-06-18 stsp f, filesize, &blocklen);
700 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
701 dd29967c 2021-08-22 stsp alloc_chunk_size, 1, block->offset, blocklen);
706 69aa0e90 2021-03-10 stsp * No match.
707 69aa0e90 2021-03-10 stsp * This block needs to be sourced from the file itself.
709 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
710 dd29967c 2021-08-22 stsp alloc_chunk_size, 0, fileoffset - offset0, blocklen);
714 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
715 0af64e86 2021-08-22 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1) {
716 0af64e86 2021-08-22 stsp err = got_error_from_errno("fseeko");
722 69aa0e90 2021-03-10 stsp free(*deltas);
723 69aa0e90 2021-03-10 stsp *deltas = NULL;
724 69aa0e90 2021-03-10 stsp *ndeltas = 0;
726 69aa0e90 2021-03-10 stsp return err;
729 2b0ae357 2022-01-10 thomas const struct got_error *
730 2b0ae357 2022-01-10 thomas got_deltify_file_mem(struct got_delta_instruction **deltas, int *ndeltas,
731 cc524d36 2022-05-31 thomas FILE *f, off_t fileoffset, off_t filesize, uint32_t seed,
732 2b0ae357 2022-01-10 thomas struct got_delta_table *dt, uint8_t *basedata,
733 2b0ae357 2022-01-10 thomas off_t basefile_offset0, off_t basefile_size)
735 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
736 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
737 2b0ae357 2022-01-10 thomas size_t nalloc = 0;
738 2b0ae357 2022-01-10 thomas const size_t alloc_chunk_size = 64;
740 2b0ae357 2022-01-10 thomas *deltas = NULL;
741 2b0ae357 2022-01-10 thomas *ndeltas = 0;
744 2b0ae357 2022-01-10 thomas * offset0 indicates where data to be deltified begins.
745 2b0ae357 2022-01-10 thomas * For example, we want to avoid deltifying a Git object header at
746 2b0ae357 2022-01-10 thomas * the beginning of the file.
748 2b0ae357 2022-01-10 thomas if (fseeko(f, offset0, SEEK_SET) == -1)
749 2b0ae357 2022-01-10 thomas return got_error_from_errno("fseeko");
751 2b0ae357 2022-01-10 thomas *deltas = reallocarray(NULL, alloc_chunk_size,
752 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_instruction));
753 2b0ae357 2022-01-10 thomas if (*deltas == NULL)
754 2b0ae357 2022-01-10 thomas return got_error_from_errno("reallocarray");
755 2b0ae357 2022-01-10 thomas nalloc = alloc_chunk_size;
757 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
758 2b0ae357 2022-01-10 thomas uint8_t buf[GOT_DELTIFY_MAXCHUNK];
759 2b0ae357 2022-01-10 thomas off_t blocklen;
760 2b0ae357 2022-01-10 thomas struct got_delta_block *block;
761 2b0ae357 2022-01-10 thomas err = nextblk(buf, &blocklen, f);
764 2b0ae357 2022-01-10 thomas if (blocklen == 0) {
765 2b0ae357 2022-01-10 thomas /* Source remainder from the file itself. */
766 2b0ae357 2022-01-10 thomas if (fileoffset < filesize) {
767 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
768 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0,
769 2b0ae357 2022-01-10 thomas filesize - fileoffset);
773 cc524d36 2022-05-31 thomas err = lookupblk_mem(&block, dt, buf, blocklen, seed, basedata,
774 2b0ae357 2022-01-10 thomas basefile_offset0);
777 2b0ae357 2022-01-10 thomas if (block != NULL) {
779 2b0ae357 2022-01-10 thomas * We have found a matching block in the delta base.
780 2b0ae357 2022-01-10 thomas * Attempt to stretch the block as far as possible and
781 2b0ae357 2022-01-10 thomas * generate a copy instruction.
783 2b0ae357 2022-01-10 thomas err = stretchblk_file_mem(basedata, basefile_offset0,
784 2b0ae357 2022-01-10 thomas basefile_size, block, f, filesize, &blocklen);
787 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
788 2b0ae357 2022-01-10 thomas alloc_chunk_size, 1, block->offset, blocklen);
793 2b0ae357 2022-01-10 thomas * No match.
794 2b0ae357 2022-01-10 thomas * This block needs to be sourced from the file itself.
796 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
797 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0, blocklen);
801 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
802 2b0ae357 2022-01-10 thomas if (fseeko(f, fileoffset, SEEK_SET) == -1) {
803 2b0ae357 2022-01-10 thomas err = got_error_from_errno("fseeko");
808 2b0ae357 2022-01-10 thomas if (err) {
809 2b0ae357 2022-01-10 thomas free(*deltas);
810 2b0ae357 2022-01-10 thomas *deltas = NULL;
811 2b0ae357 2022-01-10 thomas *ndeltas = 0;
813 2b0ae357 2022-01-10 thomas return err;
816 2b0ae357 2022-01-10 thomas const struct got_error *
817 2b0ae357 2022-01-10 thomas got_deltify_mem_file(struct got_delta_instruction **deltas, int *ndeltas,
818 cc524d36 2022-05-31 thomas uint8_t *data, off_t fileoffset, off_t filesize, uint32_t seed,
819 2b0ae357 2022-01-10 thomas struct got_delta_table *dt, FILE *basefile,
820 2b0ae357 2022-01-10 thomas off_t basefile_offset0, off_t basefile_size)
822 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
823 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
824 2b0ae357 2022-01-10 thomas size_t nalloc = 0;
825 2b0ae357 2022-01-10 thomas const size_t alloc_chunk_size = 64;
827 2b0ae357 2022-01-10 thomas *deltas = NULL;
828 2b0ae357 2022-01-10 thomas *ndeltas = 0;
830 2b0ae357 2022-01-10 thomas *deltas = reallocarray(NULL, alloc_chunk_size,
831 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_instruction));
832 2b0ae357 2022-01-10 thomas if (*deltas == NULL)
833 2b0ae357 2022-01-10 thomas return got_error_from_errno("reallocarray");
834 2b0ae357 2022-01-10 thomas nalloc = alloc_chunk_size;
836 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
837 2b0ae357 2022-01-10 thomas off_t blocklen;
838 2b0ae357 2022-01-10 thomas struct got_delta_block *block;
839 2b0ae357 2022-01-10 thomas err = nextblk_mem(&blocklen, data, fileoffset, filesize);
842 2b0ae357 2022-01-10 thomas if (blocklen == 0) {
843 2b0ae357 2022-01-10 thomas /* Source remainder from the file itself. */
844 2b0ae357 2022-01-10 thomas if (fileoffset < filesize) {
845 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
846 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0,
847 2b0ae357 2022-01-10 thomas filesize - fileoffset);
851 cc524d36 2022-05-31 thomas err = lookupblk(&block, dt, data + fileoffset, blocklen, seed,
852 2b0ae357 2022-01-10 thomas basefile, basefile_offset0);
855 2b0ae357 2022-01-10 thomas if (block != NULL) {
857 2b0ae357 2022-01-10 thomas * We have found a matching block in the delta base.
858 2b0ae357 2022-01-10 thomas * Attempt to stretch the block as far as possible and
859 2b0ae357 2022-01-10 thomas * generate a copy instruction.
861 2b0ae357 2022-01-10 thomas err = stretchblk_mem_file(basefile, basefile_offset0,
862 2b0ae357 2022-01-10 thomas block, data, fileoffset + blocklen, filesize,
863 2b0ae357 2022-01-10 thomas &blocklen);
866 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
867 2b0ae357 2022-01-10 thomas alloc_chunk_size, 1, block->offset, blocklen);
872 2b0ae357 2022-01-10 thomas * No match.
873 2b0ae357 2022-01-10 thomas * This block needs to be sourced from the file itself.
875 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
876 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0, blocklen);
880 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
883 2b0ae357 2022-01-10 thomas if (err) {
884 2b0ae357 2022-01-10 thomas free(*deltas);
885 2b0ae357 2022-01-10 thomas *deltas = NULL;
886 2b0ae357 2022-01-10 thomas *ndeltas = 0;
888 2b0ae357 2022-01-10 thomas return err;
891 2b0ae357 2022-01-10 thomas const struct got_error *
892 2b0ae357 2022-01-10 thomas got_deltify_mem_mem(struct got_delta_instruction **deltas, int *ndeltas,
893 cc524d36 2022-05-31 thomas uint8_t *data, off_t fileoffset, off_t filesize, uint32_t seed,
894 2b0ae357 2022-01-10 thomas struct got_delta_table *dt, uint8_t *basedata,
895 2b0ae357 2022-01-10 thomas off_t basefile_offset0, off_t basefile_size)
897 2b0ae357 2022-01-10 thomas const struct got_error *err = NULL;
898 2b0ae357 2022-01-10 thomas const off_t offset0 = fileoffset;
899 2b0ae357 2022-01-10 thomas size_t nalloc = 0;
900 2b0ae357 2022-01-10 thomas const size_t alloc_chunk_size = 64;
902 2b0ae357 2022-01-10 thomas *deltas = NULL;
903 2b0ae357 2022-01-10 thomas *ndeltas = 0;
905 2b0ae357 2022-01-10 thomas *deltas = reallocarray(NULL, alloc_chunk_size,
906 2b0ae357 2022-01-10 thomas sizeof(struct got_delta_instruction));
907 2b0ae357 2022-01-10 thomas if (*deltas == NULL)
908 2b0ae357 2022-01-10 thomas return got_error_from_errno("reallocarray");
909 2b0ae357 2022-01-10 thomas nalloc = alloc_chunk_size;
911 2b0ae357 2022-01-10 thomas while (fileoffset < filesize) {
912 2b0ae357 2022-01-10 thomas off_t blocklen;
913 2b0ae357 2022-01-10 thomas struct got_delta_block *block;
914 2b0ae357 2022-01-10 thomas err = nextblk_mem(&blocklen, data, fileoffset, filesize);
917 2b0ae357 2022-01-10 thomas if (blocklen == 0) {
918 2b0ae357 2022-01-10 thomas /* Source remainder from the file itself. */
919 2b0ae357 2022-01-10 thomas if (fileoffset < filesize) {
920 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
921 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0,
922 2b0ae357 2022-01-10 thomas filesize - fileoffset);
926 2b0ae357 2022-01-10 thomas err = lookupblk_mem(&block, dt, data + fileoffset, blocklen,
927 cc524d36 2022-05-31 thomas seed, basedata, basefile_offset0);
930 2b0ae357 2022-01-10 thomas if (block != NULL) {
932 2b0ae357 2022-01-10 thomas * We have found a matching block in the delta base.
933 2b0ae357 2022-01-10 thomas * Attempt to stretch the block as far as possible and
934 2b0ae357 2022-01-10 thomas * generate a copy instruction.
936 2b0ae357 2022-01-10 thomas err = stretchblk_mem_mem(basedata, basefile_offset0,
937 2b0ae357 2022-01-10 thomas basefile_size, block, data, fileoffset + blocklen,
938 2b0ae357 2022-01-10 thomas filesize, &blocklen);
941 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
942 2b0ae357 2022-01-10 thomas alloc_chunk_size, 1, block->offset, blocklen);
947 2b0ae357 2022-01-10 thomas * No match.
948 2b0ae357 2022-01-10 thomas * This block needs to be sourced from the file itself.
950 2b0ae357 2022-01-10 thomas err = emitdelta(deltas, &nalloc, ndeltas,
951 2b0ae357 2022-01-10 thomas alloc_chunk_size, 0, fileoffset - offset0, blocklen);
955 2b0ae357 2022-01-10 thomas fileoffset += blocklen;
958 2b0ae357 2022-01-10 thomas if (err) {
959 2b0ae357 2022-01-10 thomas free(*deltas);
960 2b0ae357 2022-01-10 thomas *deltas = NULL;
961 2b0ae357 2022-01-10 thomas *ndeltas = 0;
963 2b0ae357 2022-01-10 thomas return err;