2 56e7a15f 2021-03-10 stsp * Copyright (c) 2020 Ori Bernstein
3 69aa0e90 2021-03-10 stsp * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
5 2a0e67f7 2021-03-10 stsp * Permission to use, copy, modify, and distribute this software for any
6 2a0e67f7 2021-03-10 stsp * purpose with or without fee is hereby granted, provided that the above
7 2a0e67f7 2021-03-10 stsp * copyright notice and this permission notice appear in all copies.
9 2a0e67f7 2021-03-10 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 2a0e67f7 2021-03-10 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 2a0e67f7 2021-03-10 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 2a0e67f7 2021-03-10 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 2a0e67f7 2021-03-10 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 2a0e67f7 2021-03-10 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 2a0e67f7 2021-03-10 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18 f364801d 2021-03-10 stsp #include <sys/types.h>
19 f364801d 2021-03-10 stsp #include <sys/queue.h>
21 f364801d 2021-03-10 stsp #include <assert.h>
22 22403ab7 2021-03-22 naddy #include <endian.h>
23 69aa0e90 2021-03-10 stsp #include <errno.h>
24 f364801d 2021-03-10 stsp #include <limits.h>
25 f364801d 2021-03-10 stsp #include <stdint.h>
26 f364801d 2021-03-10 stsp #include <stdio.h>
27 f364801d 2021-03-10 stsp #include <stdlib.h>
28 f364801d 2021-03-10 stsp #include <string.h>
30 f364801d 2021-03-10 stsp #include "got_error.h"
32 69aa0e90 2021-03-10 stsp #include "got_lib_deltify.h"
33 2b474c25 2022-02-11 stsp #include "murmurhash2.h"
36 69aa0e90 2021-03-10 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
40 f736d93a 2021-10-07 stsp * The algorihm used here is FastCDC (Fast Content-Defined Chunking)
41 f736d93a 2021-10-07 stsp * https://www.usenix.org/conference/atc16/technical-sessions/presentation/xia
44 d58ddaf3 2022-03-17 naddy static const uint32_t geartab[256] = {
45 69aa0e90 2021-03-10 stsp 0x67ed26b7, 0x32da500c, 0x53d0fee0, 0xce387dc7, 0xcd406d90, 0x2e83a4d4,
46 69aa0e90 2021-03-10 stsp 0x9fc9a38d, 0xb67259dc, 0xca6b1722, 0x6d2ea08c, 0x235cea2e, 0x3149bb5f,
47 69aa0e90 2021-03-10 stsp 0x1beda787, 0x2a6b77d5, 0x2f22d9ac, 0x91fc0544, 0xe413acfa, 0x5a30ff7a,
48 69aa0e90 2021-03-10 stsp 0xad6fdde0, 0x444fd0f5, 0x7ad87864, 0x58c5ff05, 0x8d2ec336, 0x2371f853,
49 69aa0e90 2021-03-10 stsp 0x550f8572, 0x6aa448dd, 0x7c9ddbcf, 0x95221e14, 0x2a82ec33, 0xcbec5a78,
50 69aa0e90 2021-03-10 stsp 0xc6795a0d, 0x243995b7, 0x1c909a2f, 0x4fded51c, 0x635d334b, 0x0e2b9999,
51 69aa0e90 2021-03-10 stsp 0x2702968d, 0x856de1d5, 0x3325d60e, 0xeb6a7502, 0xec2a9844, 0x0905835a,
52 69aa0e90 2021-03-10 stsp 0xa1820375, 0xa4be5cab, 0x96a6c058, 0x2c2ccd70, 0xba40fce3, 0xd794c46b,
53 69aa0e90 2021-03-10 stsp 0x8fbae83e, 0xc3aa7899, 0x3d3ff8ed, 0xa0d42b5b, 0x571c0c97, 0xd2811516,
54 69aa0e90 2021-03-10 stsp 0xf7e7b96c, 0x4fd2fcbd, 0xe2fdec94, 0x282cc436, 0x78e8e95c, 0x80a3b613,
55 69aa0e90 2021-03-10 stsp 0xcfbee20c, 0xd4a32d1c, 0x2a12ff13, 0x6af82936, 0xe5630258, 0x8efa6a98,
56 69aa0e90 2021-03-10 stsp 0x294fb2d1, 0xdeb57086, 0x5f0fddb3, 0xeceda7ce, 0x4c87305f, 0x3a6d3307,
57 69aa0e90 2021-03-10 stsp 0xe22d2942, 0x9d060217, 0x1e42ed02, 0xb6f63b52, 0x4367f39f, 0x055cf262,
58 69aa0e90 2021-03-10 stsp 0x03a461b2, 0x5ef9e382, 0x386bc03a, 0x2a1e79c7, 0xf1a0058b, 0xd4d2dea9,
59 69aa0e90 2021-03-10 stsp 0x56baf37d, 0x5daff6cc, 0xf03a951d, 0xaef7de45, 0xa8f4581e, 0x3960b555,
60 69aa0e90 2021-03-10 stsp 0xffbfff6d, 0xbe702a23, 0x8f5b6d6f, 0x061739fb, 0x98696f47, 0x3fd596d4,
61 69aa0e90 2021-03-10 stsp 0x151eac6b, 0xa9fcc4f5, 0x69181a12, 0x3ac5a107, 0xb5198fe7, 0x96bcb1da,
62 69aa0e90 2021-03-10 stsp 0x1b5ddf8e, 0xc757d650, 0x65865c3a, 0x8fc0a41a, 0x87435536, 0x99eda6f2,
63 69aa0e90 2021-03-10 stsp 0x41874794, 0x29cff4e8, 0xb70efd9a, 0x3103f6e7, 0x84d2453b, 0x15a450bd,
64 69aa0e90 2021-03-10 stsp 0x74f49af1, 0x60f664b1, 0xa1c86935, 0xfdafbce1, 0xe36353e3, 0x5d9ba739,
65 69aa0e90 2021-03-10 stsp 0xbc0559ba, 0x708b0054, 0xd41d808c, 0xb2f31723, 0x9027c41f, 0xf136d165,
66 69aa0e90 2021-03-10 stsp 0xb5374b12, 0x9420a6ac, 0x273958b6, 0xe6c2fad0, 0xebdc1f21, 0xfb33af8b,
67 69aa0e90 2021-03-10 stsp 0xc71c25cd, 0xe9a2d8e5, 0xbeb38a50, 0xbceb7cc2, 0x4e4e73f0, 0xcd6c251d,
68 69aa0e90 2021-03-10 stsp 0xde4c032c, 0x4b04ac30, 0x725b8b21, 0x4eb8c33b, 0x20d07b75, 0x0567aa63,
69 69aa0e90 2021-03-10 stsp 0xb56b2bb7, 0xc1f5fd3a, 0xcafd35ca, 0x470dd4da, 0xfe4f94cd, 0xfb8de424,
70 69aa0e90 2021-03-10 stsp 0xe8dbcf40, 0xfe50a37a, 0x62db5b5d, 0xf32f4ab6, 0x2c4a8a51, 0x18473dc0,
71 69aa0e90 2021-03-10 stsp 0xfe0cbb6e, 0xfe399efd, 0xdf34ecc9, 0x6ccd5055, 0x46097073, 0x139135c2,
72 69aa0e90 2021-03-10 stsp 0x721c76f6, 0x1c6a94b4, 0x6eee014d, 0x8a508e02, 0x3da538f5, 0x280d394f,
73 69aa0e90 2021-03-10 stsp 0x5248a0c4, 0x3ce94c6c, 0x9a71ad3a, 0x8493dd05, 0xe43f0ab6, 0x18e4ed42,
74 69aa0e90 2021-03-10 stsp 0x6c5c0e09, 0x42b06ec9, 0x8d330343, 0xa45b6f59, 0x2a573c0c, 0xd7fd3de6,
75 69aa0e90 2021-03-10 stsp 0xeedeab68, 0x5c84dafc, 0xbbd1b1a8, 0xa3ce1ad1, 0x85b70bed, 0xb6add07f,
76 69aa0e90 2021-03-10 stsp 0xa531309c, 0x8f8ab852, 0x564de332, 0xeac9ed0c, 0x73da402c, 0x3ec52761,
77 69aa0e90 2021-03-10 stsp 0x43af2f4d, 0xd6ff45c8, 0x4c367462, 0xd553bd6a, 0x44724855, 0x3b2aa728,
78 69aa0e90 2021-03-10 stsp 0x56e5eb65, 0xeaf16173, 0x33fa42ff, 0xd714bb5d, 0xfbd0a3b9, 0xaf517134,
79 69aa0e90 2021-03-10 stsp 0x9416c8cd, 0x534cf94f, 0x548947c2, 0x34193569, 0x32f4389a, 0xfe7028bc,
80 69aa0e90 2021-03-10 stsp 0xed73b1ed, 0x9db95770, 0x468e3922, 0x0440c3cd, 0x60059a62, 0x33504562,
81 69aa0e90 2021-03-10 stsp 0x2b229fbd, 0x5174dca5, 0xf7028752, 0xd63c6aa8, 0x31276f38, 0x0646721c,
82 69aa0e90 2021-03-10 stsp 0xb0191da8, 0xe00e6de0, 0x9eac1a6e, 0x9f7628a5, 0xed6c06ea, 0x0bb8af15,
83 69aa0e90 2021-03-10 stsp 0xf119fb12, 0x38693c1c, 0x732bc0fe, 0x84953275, 0xb82ec888, 0x33a4f1b3,
84 69aa0e90 2021-03-10 stsp 0x3099835e, 0x028a8782, 0x5fdd51d7, 0xc6c717b3, 0xb06caf71, 0x17c8c111,
85 69aa0e90 2021-03-10 stsp 0x61bad754, 0x9fd03061, 0xe09df1af, 0x3bc9eb73, 0x85878413, 0x9889aaf2,
86 69aa0e90 2021-03-10 stsp 0x3f5a9e46, 0x42c9f01f, 0x9984a4f4, 0xd5de43cc, 0xd294daed, 0xbecba2d2,
87 69aa0e90 2021-03-10 stsp 0xf1f6e72c, 0x5551128a, 0x83af87e2, 0x6f0342ba,
90 2b474c25 2022-02-11 stsp static uint32_t
91 d6a28ffe 2022-05-20 op hashblk(const unsigned char *p, off_t n, uint32_t seed)
93 d6a28ffe 2022-05-20 op return murmurhash2(p, n, seed);
96 f364801d 2021-03-10 stsp static const struct got_error *
97 740bba1c 2021-06-18 stsp addblk(struct got_delta_table *dt, FILE *f, off_t file_offset0, off_t len,
98 f6027426 2022-02-12 naddy off_t offset, uint32_t h)
100 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
102 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
103 0d15f6dc 2021-06-13 stsp uint8_t buf2[GOT_DELTIFY_MAXCHUNK];
104 69aa0e90 2021-03-10 stsp size_t r = 0;
106 69aa0e90 2021-03-10 stsp if (len == 0)
107 69aa0e90 2021-03-10 stsp return NULL;
109 69aa0e90 2021-03-10 stsp i = h % dt->nalloc;
110 69aa0e90 2021-03-10 stsp while (dt->blocks[i].len != 0) {
112 69aa0e90 2021-03-10 stsp * Avoid adding duplicate blocks.
113 69aa0e90 2021-03-10 stsp * NB: A matching hash is insufficient for detecting equality.
114 51a494da 2021-06-11 stsp * The hash can only detect inequality.
116 51a494da 2021-06-11 stsp if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
117 0d15f6dc 2021-06-13 stsp if (r == 0) {
118 740bba1c 2021-06-18 stsp if (fseeko(f, file_offset0 + offset, SEEK_SET) == -1)
119 0d15f6dc 2021-06-13 stsp return got_error_from_errno("fseeko");
120 0d15f6dc 2021-06-13 stsp r = fread(buf, 1, len, f);
121 0d15f6dc 2021-06-13 stsp if (r != len) {
122 0d15f6dc 2021-06-13 stsp if (!ferror(f))
123 0d15f6dc 2021-06-13 stsp return NULL;
124 0d15f6dc 2021-06-13 stsp return got_ferror(f, GOT_ERR_IO);
127 740bba1c 2021-06-18 stsp if (fseeko(f, file_offset0 + dt->blocks[i].offset,
128 740bba1c 2021-06-18 stsp SEEK_SET) == -1)
129 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
130 69aa0e90 2021-03-10 stsp if (fread(buf2, 1, len, f) != len)
131 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
132 69aa0e90 2021-03-10 stsp if (memcmp(buf, buf2, len) == 0)
133 69aa0e90 2021-03-10 stsp return NULL;
136 69aa0e90 2021-03-10 stsp i = (i + 1) % dt->nalloc;
138 69aa0e90 2021-03-10 stsp assert(dt->blocks[i].len == 0);
139 69aa0e90 2021-03-10 stsp dt->blocks[i].len = len;
140 69aa0e90 2021-03-10 stsp dt->blocks[i].offset = offset;
141 69aa0e90 2021-03-10 stsp dt->blocks[i].hash = h;
142 69aa0e90 2021-03-10 stsp dt->nblocks++;
143 e89540a9 2021-06-11 stsp if (dt->nalloc < dt->nblocks + 64) {
144 69aa0e90 2021-03-10 stsp struct got_delta_block *db;
145 e89540a9 2021-06-11 stsp size_t old_size = dt->nalloc;
146 69aa0e90 2021-03-10 stsp db = dt->blocks;
147 e89540a9 2021-06-11 stsp dt->blocks = calloc(dt->nalloc + 64,
148 e89540a9 2021-06-11 stsp sizeof(struct got_delta_block));
149 69aa0e90 2021-03-10 stsp if (dt->blocks == NULL) {
150 69aa0e90 2021-03-10 stsp err = got_error_from_errno("calloc");
151 69aa0e90 2021-03-10 stsp dt->blocks = db;
152 69aa0e90 2021-03-10 stsp return err;
154 e89540a9 2021-06-11 stsp dt->nalloc += 64;
156 69aa0e90 2021-03-10 stsp * Recompute all block positions. Hash-based indices of blocks
157 69aa0e90 2021-03-10 stsp * in the array depend on the allocated length of the array.
159 69aa0e90 2021-03-10 stsp dt->nblocks = 0;
160 a893025f 2021-06-11 stsp for (i = 0; i < old_size; i++) {
161 69aa0e90 2021-03-10 stsp if (db[i].len == 0)
163 740bba1c 2021-06-18 stsp err = addblk(dt, f, file_offset0, db[i].len,
164 64a8571e 2022-01-07 stsp db[i].offset, db[i].hash);
171 64a8571e 2022-01-07 stsp return err;
174 64a8571e 2022-01-07 stsp static const struct got_error *
175 64a8571e 2022-01-07 stsp addblk_mem(struct got_delta_table *dt, uint8_t *data, off_t file_offset0,
176 f6027426 2022-02-12 naddy off_t len, off_t offset, uint32_t h)
178 64a8571e 2022-01-07 stsp const struct got_error *err = NULL;
180 64a8571e 2022-01-07 stsp uint8_t *block1;
181 64a8571e 2022-01-07 stsp uint8_t *block2;
183 64a8571e 2022-01-07 stsp if (len == 0)
184 64a8571e 2022-01-07 stsp return NULL;
186 64a8571e 2022-01-07 stsp i = h % dt->nalloc;
187 64a8571e 2022-01-07 stsp while (dt->blocks[i].len != 0) {
189 64a8571e 2022-01-07 stsp * Avoid adding duplicate blocks.
190 64a8571e 2022-01-07 stsp * NB: A matching hash is insufficient for detecting equality.
191 64a8571e 2022-01-07 stsp * The hash can only detect inequality.
193 64a8571e 2022-01-07 stsp if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
194 64a8571e 2022-01-07 stsp block1 = data + file_offset0 + dt->blocks[i].offset;
195 64a8571e 2022-01-07 stsp block2 = data + file_offset0 + offset;
196 64a8571e 2022-01-07 stsp if (memcmp(block1, block2, len) == 0)
197 64a8571e 2022-01-07 stsp return NULL;
200 64a8571e 2022-01-07 stsp i = (i + 1) % dt->nalloc;
202 64a8571e 2022-01-07 stsp assert(dt->blocks[i].len == 0);
203 64a8571e 2022-01-07 stsp dt->blocks[i].len = len;
204 64a8571e 2022-01-07 stsp dt->blocks[i].offset = offset;
205 64a8571e 2022-01-07 stsp dt->blocks[i].hash = h;
206 64a8571e 2022-01-07 stsp dt->nblocks++;
207 64a8571e 2022-01-07 stsp if (dt->nalloc < dt->nblocks + 64) {
208 64a8571e 2022-01-07 stsp struct got_delta_block *db;
209 64a8571e 2022-01-07 stsp size_t old_size = dt->nalloc;
210 64a8571e 2022-01-07 stsp db = dt->blocks;
211 64a8571e 2022-01-07 stsp dt->blocks = calloc(dt->nalloc + 64,
212 64a8571e 2022-01-07 stsp sizeof(struct got_delta_block));
213 64a8571e 2022-01-07 stsp if (dt->blocks == NULL) {
214 64a8571e 2022-01-07 stsp err = got_error_from_errno("calloc");
215 64a8571e 2022-01-07 stsp dt->blocks = db;
216 64a8571e 2022-01-07 stsp return err;
218 64a8571e 2022-01-07 stsp dt->nalloc += 64;
220 64a8571e 2022-01-07 stsp * Recompute all block positions. Hash-based indices of blocks
221 64a8571e 2022-01-07 stsp * in the array depend on the allocated length of the array.
223 64a8571e 2022-01-07 stsp dt->nblocks = 0;
224 64a8571e 2022-01-07 stsp for (i = 0; i < old_size; i++) {
225 64a8571e 2022-01-07 stsp if (db[i].len == 0)
227 64a8571e 2022-01-07 stsp err = addblk_mem(dt, data, file_offset0, db[i].len,
228 740bba1c 2021-06-18 stsp db[i].offset, db[i].hash);
235 69aa0e90 2021-03-10 stsp return err;
238 69aa0e90 2021-03-10 stsp static const struct got_error *
239 69aa0e90 2021-03-10 stsp lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
240 d6a28ffe 2022-05-20 op unsigned char *p, off_t len, uint32_t seed, FILE *basefile,
241 d6a28ffe 2022-05-20 op off_t basefile_offset0)
244 2b474c25 2022-02-11 stsp uint32_t h;
245 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
248 69aa0e90 2021-03-10 stsp *block = NULL;
250 d6a28ffe 2022-05-20 op h = hashblk(p, len, seed);
251 69aa0e90 2021-03-10 stsp for (i = h % dt->nalloc; dt->blocks[i].len != 0;
252 69aa0e90 2021-03-10 stsp i = (i + 1) % dt->nalloc) {
253 69aa0e90 2021-03-10 stsp if (dt->blocks[i].hash != h ||
254 69aa0e90 2021-03-10 stsp dt->blocks[i].len != len)
256 740bba1c 2021-06-18 stsp if (fseeko(basefile, basefile_offset0 + dt->blocks[i].offset,
257 740bba1c 2021-06-18 stsp SEEK_SET) == -1)
258 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
259 69aa0e90 2021-03-10 stsp r = fread(buf, 1, len, basefile);
260 69aa0e90 2021-03-10 stsp if (r != len)
261 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
262 69aa0e90 2021-03-10 stsp if (memcmp(p, buf, len) == 0) {
263 69aa0e90 2021-03-10 stsp *block = &dt->blocks[i];
267 f364801d 2021-03-10 stsp return NULL;
270 69aa0e90 2021-03-10 stsp static const struct got_error *
271 64a8571e 2022-01-07 stsp lookupblk_mem(struct got_delta_block **block, struct got_delta_table *dt,
272 d6a28ffe 2022-05-20 op unsigned char *p, off_t len, uint32_t seed, uint8_t *basedata,
273 d6a28ffe 2022-05-20 op off_t basefile_offset0)
276 2b474c25 2022-02-11 stsp uint32_t h;
277 64a8571e 2022-01-07 stsp uint8_t *b;
279 64a8571e 2022-01-07 stsp *block = NULL;
281 d6a28ffe 2022-05-20 op h = hashblk(p, len, seed);
282 64a8571e 2022-01-07 stsp for (i = h % dt->nalloc; dt->blocks[i].len != 0;
283 64a8571e 2022-01-07 stsp i = (i + 1) % dt->nalloc) {
284 64a8571e 2022-01-07 stsp if (dt->blocks[i].hash != h ||
285 64a8571e 2022-01-07 stsp dt->blocks[i].len != len)
287 64a8571e 2022-01-07 stsp b = basedata + basefile_offset0 + dt->blocks[i].offset;
288 64a8571e 2022-01-07 stsp if (memcmp(p, b, len) == 0) {
289 64a8571e 2022-01-07 stsp *block = &dt->blocks[i];
293 64a8571e 2022-01-07 stsp return NULL;
296 64a8571e 2022-01-07 stsp static const struct got_error *
297 69aa0e90 2021-03-10 stsp nextblk(uint8_t *buf, off_t *blocklen, FILE *f)
299 f364801d 2021-03-10 stsp uint32_t gh;
300 69aa0e90 2021-03-10 stsp const unsigned char *p;
302 69aa0e90 2021-03-10 stsp off_t pos = ftello(f);
304 69aa0e90 2021-03-10 stsp *blocklen = 0;
306 69aa0e90 2021-03-10 stsp r = fread(buf, 1, GOT_DELTIFY_MAXCHUNK, f);
307 69aa0e90 2021-03-10 stsp if (r == 0 && ferror(f))
308 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
309 69aa0e90 2021-03-10 stsp if (r < GOT_DELTIFY_MINCHUNK)
310 69aa0e90 2021-03-10 stsp return NULL; /* no more delta-worthy blocks left */
312 69aa0e90 2021-03-10 stsp /* Got a deltifiable block. Find the split-point where it ends. */
313 69aa0e90 2021-03-10 stsp p = buf + GOT_DELTIFY_MINCHUNK;
315 69aa0e90 2021-03-10 stsp while (p != buf + r) {
316 f364801d 2021-03-10 stsp gh = (gh << 1) + geartab[*p++];
317 69aa0e90 2021-03-10 stsp if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
321 69aa0e90 2021-03-10 stsp *blocklen = (p - buf);
322 69aa0e90 2021-03-10 stsp if (fseeko(f, pos + *blocklen, SEEK_SET) == -1)
323 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
325 69aa0e90 2021-03-10 stsp return NULL;
328 64a8571e 2022-01-07 stsp static const struct got_error *
329 64a8571e 2022-01-07 stsp nextblk_mem(off_t *blocklen, uint8_t *data, off_t fileoffset, off_t filesize)
331 64a8571e 2022-01-07 stsp uint32_t gh;
332 64a8571e 2022-01-07 stsp const unsigned char *p;
334 64a8571e 2022-01-07 stsp *blocklen = 0;
336 64a8571e 2022-01-07 stsp if (fileoffset >= filesize ||
337 64a8571e 2022-01-07 stsp filesize - fileoffset < GOT_DELTIFY_MINCHUNK)
338 64a8571e 2022-01-07 stsp return NULL; /* no more delta-worthy blocks left */
340 64a8571e 2022-01-07 stsp /* Got a deltifiable block. Find the split-point where it ends. */
341 64a8571e 2022-01-07 stsp p = data + fileoffset + GOT_DELTIFY_MINCHUNK;
343 64a8571e 2022-01-07 stsp while (p != data + MIN(fileoffset + GOT_DELTIFY_MAXCHUNK, filesize)) {
344 64a8571e 2022-01-07 stsp gh = (gh << 1) + geartab[*p++];
345 64a8571e 2022-01-07 stsp if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
349 64a8571e 2022-01-07 stsp *blocklen = (p - (data + fileoffset));
350 64a8571e 2022-01-07 stsp return NULL;
353 f364801d 2021-03-10 stsp const struct got_error *
354 69aa0e90 2021-03-10 stsp got_deltify_init(struct got_delta_table **dt, FILE *f, off_t fileoffset,
355 d6a28ffe 2022-05-20 op off_t filesize, uint32_t seed)
357 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
358 2b474c25 2022-02-11 stsp uint32_t h;
359 740bba1c 2021-06-18 stsp const off_t offset0 = fileoffset;
361 69aa0e90 2021-03-10 stsp *dt = calloc(1, sizeof(**dt));
362 69aa0e90 2021-03-10 stsp if (*dt == NULL)
363 f364801d 2021-03-10 stsp return got_error_from_errno("calloc");
365 69aa0e90 2021-03-10 stsp (*dt)->nblocks = 0;
366 69aa0e90 2021-03-10 stsp (*dt)->nalloc = 128;
367 69aa0e90 2021-03-10 stsp (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
368 69aa0e90 2021-03-10 stsp if ((*dt)->blocks == NULL) {
369 69aa0e90 2021-03-10 stsp err = got_error_from_errno("calloc");
373 69aa0e90 2021-03-10 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
374 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
376 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
377 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
378 69aa0e90 2021-03-10 stsp off_t blocklen;
379 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
382 69aa0e90 2021-03-10 stsp if (blocklen == 0)
384 d6a28ffe 2022-05-20 op h = hashblk(buf, blocklen, seed);
385 740bba1c 2021-06-18 stsp err = addblk(*dt, f, offset0, blocklen,
386 740bba1c 2021-06-18 stsp fileoffset - offset0, h);
389 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
390 68bdcdc2 2021-06-11 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
391 68bdcdc2 2021-06-11 stsp return got_error_from_errno("fseeko");
395 64a8571e 2022-01-07 stsp free((*dt)->blocks);
397 64a8571e 2022-01-07 stsp *dt = NULL;
400 64a8571e 2022-01-07 stsp return err;
403 64a8571e 2022-01-07 stsp const struct got_error *
404 64a8571e 2022-01-07 stsp got_deltify_init_mem(struct got_delta_table **dt, uint8_t *data,
405 d6a28ffe 2022-05-20 op off_t fileoffset, off_t filesize, uint32_t seed)
407 64a8571e 2022-01-07 stsp const struct got_error *err = NULL;
408 f6027426 2022-02-12 naddy uint32_t h;
409 64a8571e 2022-01-07 stsp const off_t offset0 = fileoffset;
411 64a8571e 2022-01-07 stsp *dt = calloc(1, sizeof(**dt));
412 64a8571e 2022-01-07 stsp if (*dt == NULL)
413 64a8571e 2022-01-07 stsp return got_error_from_errno("calloc");
415 64a8571e 2022-01-07 stsp (*dt)->nblocks = 0;
416 64a8571e 2022-01-07 stsp (*dt)->nalloc = 128;
417 64a8571e 2022-01-07 stsp (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
418 64a8571e 2022-01-07 stsp if ((*dt)->blocks == NULL) {
419 64a8571e 2022-01-07 stsp err = got_error_from_errno("calloc");
423 64a8571e 2022-01-07 stsp while (fileoffset < filesize) {
424 64a8571e 2022-01-07 stsp off_t blocklen;
425 64a8571e 2022-01-07 stsp err = nextblk_mem(&blocklen, data, fileoffset, filesize);
428 64a8571e 2022-01-07 stsp if (blocklen == 0)
430 d6a28ffe 2022-05-20 op h = hashblk(data + fileoffset, blocklen, seed);
431 64a8571e 2022-01-07 stsp err = addblk_mem(*dt, data, offset0, blocklen,
432 64a8571e 2022-01-07 stsp fileoffset - offset0, h);
435 64a8571e 2022-01-07 stsp fileoffset += blocklen;
439 69aa0e90 2021-03-10 stsp free((*dt)->blocks);
441 69aa0e90 2021-03-10 stsp *dt = NULL;
444 69aa0e90 2021-03-10 stsp return err;
448 69aa0e90 2021-03-10 stsp got_deltify_free(struct got_delta_table *dt)
450 dbbf4a5f 2021-05-20 stsp if (dt == NULL)
452 69aa0e90 2021-03-10 stsp free(dt->blocks);
456 f364801d 2021-03-10 stsp static const struct got_error *
457 dd29967c 2021-08-22 stsp emitdelta(struct got_delta_instruction **deltas, size_t *nalloc, int *ndeltas,
458 dd29967c 2021-08-22 stsp const size_t alloc_chunk_size, int copy, off_t offset, off_t len)
460 69aa0e90 2021-03-10 stsp struct got_delta_instruction *d, *p;
462 dd29967c 2021-08-22 stsp if (*nalloc < *ndeltas + alloc_chunk_size) {
463 dd29967c 2021-08-22 stsp p = reallocarray(*deltas, *nalloc + alloc_chunk_size,
464 dd29967c 2021-08-22 stsp sizeof(struct got_delta_instruction));
465 dd29967c 2021-08-22 stsp if (p == NULL)
466 2d467c6d 2021-10-13 stsp return got_error_from_errno("reallocarray");
467 dd29967c 2021-08-22 stsp *deltas = p;
468 dd29967c 2021-08-22 stsp *nalloc += alloc_chunk_size;
470 69aa0e90 2021-03-10 stsp *ndeltas += 1;
471 69aa0e90 2021-03-10 stsp d = &(*deltas)[*ndeltas - 1];
472 69aa0e90 2021-03-10 stsp d->copy = copy;
473 69aa0e90 2021-03-10 stsp d->offset = offset;
474 56e7a15f 2021-03-10 stsp d->len = len;
475 f364801d 2021-03-10 stsp return NULL;
478 69aa0e90 2021-03-10 stsp static const struct got_error *
479 740bba1c 2021-06-18 stsp stretchblk(FILE *basefile, off_t base_offset0, struct got_delta_block *block,
480 740bba1c 2021-06-18 stsp FILE *f, off_t filesize, off_t *blocklen)
482 69aa0e90 2021-03-10 stsp uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
483 69aa0e90 2021-03-10 stsp size_t base_r, r, i;
484 69aa0e90 2021-03-10 stsp int buf_equal = 1;
486 5de743f8 2021-08-29 stsp if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
487 5de743f8 2021-08-29 stsp SEEK_SET) == -1)
488 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
490 69aa0e90 2021-03-10 stsp while (buf_equal && *blocklen < (1 << 24) - 1) {
491 69aa0e90 2021-03-10 stsp base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
492 69aa0e90 2021-03-10 stsp if (base_r == 0) {
493 69aa0e90 2021-03-10 stsp if (ferror(basefile))
494 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
497 69aa0e90 2021-03-10 stsp r = fread(buf, 1, sizeof(buf), f);
498 69aa0e90 2021-03-10 stsp if (r == 0) {
499 69aa0e90 2021-03-10 stsp if (ferror(f))
500 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
503 69aa0e90 2021-03-10 stsp for (i = 0; i < MIN(base_r, r); i++) {
504 69aa0e90 2021-03-10 stsp if (buf[i] != basebuf[i]) {
505 69aa0e90 2021-03-10 stsp buf_equal = 0;
508 69aa0e90 2021-03-10 stsp (*blocklen)++;
512 64a8571e 2022-01-07 stsp return NULL;
515 64a8571e 2022-01-07 stsp static const struct got_error *
516 64a8571e 2022-01-07 stsp stretchblk_file_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
517 64a8571e 2022-01-07 stsp struct got_delta_block *block, FILE *f, off_t filesize, off_t *blocklen)
519 64a8571e 2022-01-07 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
520 64a8571e 2022-01-07 stsp size_t r, i;
521 64a8571e 2022-01-07 stsp int buf_equal = 1;
522 64a8571e 2022-01-07 stsp off_t base_offset = base_offset0 + block->offset + *blocklen;
524 64a8571e 2022-01-07 stsp if (base_offset > basefile_size) {
525 64a8571e 2022-01-07 stsp return got_error_fmt(GOT_ERR_RANGE,
526 04aed155 2022-07-24 naddy "read beyond the size of delta base at offset %lld",
527 04aed155 2022-07-24 naddy (long long)base_offset);
530 64a8571e 2022-01-07 stsp while (buf_equal && *blocklen < (1 << 24) - 1) {
531 64a8571e 2022-01-07 stsp if (base_offset + *blocklen >= basefile_size)
533 64a8571e 2022-01-07 stsp r = fread(buf, 1, sizeof(buf), f);
534 64a8571e 2022-01-07 stsp if (r == 0) {
535 64a8571e 2022-01-07 stsp if (ferror(f))
536 64a8571e 2022-01-07 stsp return got_ferror(f, GOT_ERR_IO);
539 64a8571e 2022-01-07 stsp for (i = 0; i < MIN(basefile_size - base_offset, r); i++) {
540 64a8571e 2022-01-07 stsp if (buf[i] != *(basedata + base_offset + i)) {
541 64a8571e 2022-01-07 stsp buf_equal = 0;
544 64a8571e 2022-01-07 stsp (*blocklen)++;
548 64a8571e 2022-01-07 stsp return NULL;
551 64a8571e 2022-01-07 stsp static const struct got_error *
552 64a8571e 2022-01-07 stsp stretchblk_mem_file(FILE *basefile, off_t base_offset0,
553 64a8571e 2022-01-07 stsp struct got_delta_block *block, uint8_t *data, off_t fileoffset,
554 64a8571e 2022-01-07 stsp off_t filesize, off_t *blocklen)
556 64a8571e 2022-01-07 stsp uint8_t basebuf[GOT_DELTIFY_MAXCHUNK];
557 64a8571e 2022-01-07 stsp size_t base_r, i;
558 64a8571e 2022-01-07 stsp int buf_equal = 1;
560 64a8571e 2022-01-07 stsp if (fileoffset > filesize) {
561 64a8571e 2022-01-07 stsp return got_error_fmt(GOT_ERR_RANGE,
562 04aed155 2022-07-24 naddy "read beyond the size of deltify file at offset %lld",
563 04aed155 2022-07-24 naddy (long long)fileoffset);
566 64a8571e 2022-01-07 stsp if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
567 64a8571e 2022-01-07 stsp SEEK_SET) == -1)
568 64a8571e 2022-01-07 stsp return got_error_from_errno("fseeko");
570 64a8571e 2022-01-07 stsp while (buf_equal && *blocklen < (1 << 24) - 1) {
571 64a8571e 2022-01-07 stsp if (fileoffset + *blocklen >= filesize)
573 64a8571e 2022-01-07 stsp base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
574 64a8571e 2022-01-07 stsp if (base_r == 0) {
575 64a8571e 2022-01-07 stsp if (ferror(basefile))
576 64a8571e 2022-01-07 stsp return got_ferror(basefile, GOT_ERR_IO);
579 64a8571e 2022-01-07 stsp for (i = 0; i < MIN(base_r, filesize - fileoffset); i++) {
580 64a8571e 2022-01-07 stsp if (*(data + fileoffset + i) != basebuf[i]) {
581 64a8571e 2022-01-07 stsp buf_equal = 0;
584 64a8571e 2022-01-07 stsp (*blocklen)++;
588 64a8571e 2022-01-07 stsp return NULL;
591 64a8571e 2022-01-07 stsp static const struct got_error *
592 64a8571e 2022-01-07 stsp stretchblk_mem_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
593 64a8571e 2022-01-07 stsp struct got_delta_block *block, uint8_t *data, off_t fileoffset,
594 64a8571e 2022-01-07 stsp off_t filesize, off_t *blocklen)
596 64a8571e 2022-01-07 stsp off_t i, maxlen;
597 64a8571e 2022-01-07 stsp off_t base_offset = base_offset0 + block->offset + *blocklen;
598 64a8571e 2022-01-07 stsp uint8_t *p, *q;
600 64a8571e 2022-01-07 stsp if (base_offset > basefile_size) {
601 64a8571e 2022-01-07 stsp return got_error_fmt(GOT_ERR_RANGE,
602 04aed155 2022-07-24 naddy "read beyond the size of delta base at offset %lld",
603 04aed155 2022-07-24 naddy (long long)base_offset);
606 64a8571e 2022-01-07 stsp if (fileoffset > filesize) {
607 64a8571e 2022-01-07 stsp return got_error_fmt(GOT_ERR_RANGE,
608 04aed155 2022-07-24 naddy "read beyond the size of deltify file at offset %lld",
609 04aed155 2022-07-24 naddy (long long)fileoffset);
612 64a8571e 2022-01-07 stsp p = data + fileoffset;
613 64a8571e 2022-01-07 stsp q = basedata + base_offset;
614 64a8571e 2022-01-07 stsp maxlen = MIN(basefile_size - base_offset, filesize - fileoffset);
615 64a8571e 2022-01-07 stsp for (i = 0; i < maxlen && *blocklen < (1 << 24) - 1; i++) {
616 64a8571e 2022-01-07 stsp if (p[i] != q[i])
618 64a8571e 2022-01-07 stsp (*blocklen)++;
621 69aa0e90 2021-03-10 stsp return NULL;
624 69aa0e90 2021-03-10 stsp const struct got_error *
625 69aa0e90 2021-03-10 stsp got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
626 d6a28ffe 2022-05-20 op FILE *f, off_t fileoffset, off_t filesize, uint32_t seed,
627 69aa0e90 2021-03-10 stsp struct got_delta_table *dt, FILE *basefile,
628 f34b169e 2021-06-18 stsp off_t basefile_offset0, off_t basefile_size)
630 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
631 69aa0e90 2021-03-10 stsp const off_t offset0 = fileoffset;
632 dd29967c 2021-08-22 stsp size_t nalloc = 0;
633 dd29967c 2021-08-22 stsp const size_t alloc_chunk_size = 64;
635 69aa0e90 2021-03-10 stsp *deltas = NULL;
636 69aa0e90 2021-03-10 stsp *ndeltas = 0;
639 69aa0e90 2021-03-10 stsp * offset0 indicates where data to be deltified begins.
640 69aa0e90 2021-03-10 stsp * For example, we want to avoid deltifying a Git object header at
641 69aa0e90 2021-03-10 stsp * the beginning of the file.
643 69aa0e90 2021-03-10 stsp if (fseeko(f, offset0, SEEK_SET) == -1)
644 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
646 dd29967c 2021-08-22 stsp *deltas = reallocarray(NULL, alloc_chunk_size,
647 dd29967c 2021-08-22 stsp sizeof(struct got_delta_instruction));
648 dd29967c 2021-08-22 stsp if (*deltas == NULL)
649 dd29967c 2021-08-22 stsp return got_error_from_errno("reallocarray");
650 dd29967c 2021-08-22 stsp nalloc = alloc_chunk_size;
652 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
653 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
654 69aa0e90 2021-03-10 stsp off_t blocklen;
655 69aa0e90 2021-03-10 stsp struct got_delta_block *block;
656 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
659 9a8dc2b3 2021-06-18 stsp if (blocklen == 0) {
660 9a8dc2b3 2021-06-18 stsp /* Source remainder from the file itself. */
661 9a8dc2b3 2021-06-18 stsp if (fileoffset < filesize) {
662 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
663 dd29967c 2021-08-22 stsp alloc_chunk_size, 0, fileoffset - offset0,
664 9a8dc2b3 2021-06-18 stsp filesize - fileoffset);
668 d6a28ffe 2022-05-20 op err = lookupblk(&block, dt, buf, blocklen, seed, basefile,
669 740bba1c 2021-06-18 stsp basefile_offset0);
672 69aa0e90 2021-03-10 stsp if (block != NULL) {
674 69aa0e90 2021-03-10 stsp * We have found a matching block in the delta base.
675 69aa0e90 2021-03-10 stsp * Attempt to stretch the block as far as possible and
676 69aa0e90 2021-03-10 stsp * generate a copy instruction.
678 740bba1c 2021-06-18 stsp err = stretchblk(basefile, basefile_offset0, block,
679 740bba1c 2021-06-18 stsp f, filesize, &blocklen);
682 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
683 dd29967c 2021-08-22 stsp alloc_chunk_size, 1, block->offset, blocklen);
688 69aa0e90 2021-03-10 stsp * No match.
689 69aa0e90 2021-03-10 stsp * This block needs to be sourced from the file itself.
691 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
692 dd29967c 2021-08-22 stsp alloc_chunk_size, 0, fileoffset - offset0, blocklen);
696 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
697 0af64e86 2021-08-22 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1) {
698 0af64e86 2021-08-22 stsp err = got_error_from_errno("fseeko");
704 69aa0e90 2021-03-10 stsp free(*deltas);
705 69aa0e90 2021-03-10 stsp *deltas = NULL;
706 69aa0e90 2021-03-10 stsp *ndeltas = 0;
708 69aa0e90 2021-03-10 stsp return err;
711 64a8571e 2022-01-07 stsp const struct got_error *
712 64a8571e 2022-01-07 stsp got_deltify_file_mem(struct got_delta_instruction **deltas, int *ndeltas,
713 d6a28ffe 2022-05-20 op FILE *f, off_t fileoffset, off_t filesize, uint32_t seed,
714 64a8571e 2022-01-07 stsp struct got_delta_table *dt, uint8_t *basedata,
715 64a8571e 2022-01-07 stsp off_t basefile_offset0, off_t basefile_size)
717 64a8571e 2022-01-07 stsp const struct got_error *err = NULL;
718 64a8571e 2022-01-07 stsp const off_t offset0 = fileoffset;
719 64a8571e 2022-01-07 stsp size_t nalloc = 0;
720 64a8571e 2022-01-07 stsp const size_t alloc_chunk_size = 64;
722 64a8571e 2022-01-07 stsp *deltas = NULL;
723 64a8571e 2022-01-07 stsp *ndeltas = 0;
726 64a8571e 2022-01-07 stsp * offset0 indicates where data to be deltified begins.
727 64a8571e 2022-01-07 stsp * For example, we want to avoid deltifying a Git object header at
728 64a8571e 2022-01-07 stsp * the beginning of the file.
730 64a8571e 2022-01-07 stsp if (fseeko(f, offset0, SEEK_SET) == -1)
731 64a8571e 2022-01-07 stsp return got_error_from_errno("fseeko");
733 64a8571e 2022-01-07 stsp *deltas = reallocarray(NULL, alloc_chunk_size,
734 64a8571e 2022-01-07 stsp sizeof(struct got_delta_instruction));
735 64a8571e 2022-01-07 stsp if (*deltas == NULL)
736 64a8571e 2022-01-07 stsp return got_error_from_errno("reallocarray");
737 64a8571e 2022-01-07 stsp nalloc = alloc_chunk_size;
739 64a8571e 2022-01-07 stsp while (fileoffset < filesize) {
740 64a8571e 2022-01-07 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
741 64a8571e 2022-01-07 stsp off_t blocklen;
742 64a8571e 2022-01-07 stsp struct got_delta_block *block;
743 64a8571e 2022-01-07 stsp err = nextblk(buf, &blocklen, f);
746 64a8571e 2022-01-07 stsp if (blocklen == 0) {
747 64a8571e 2022-01-07 stsp /* Source remainder from the file itself. */
748 64a8571e 2022-01-07 stsp if (fileoffset < filesize) {
749 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
750 64a8571e 2022-01-07 stsp alloc_chunk_size, 0, fileoffset - offset0,
751 64a8571e 2022-01-07 stsp filesize - fileoffset);
755 d6a28ffe 2022-05-20 op err = lookupblk_mem(&block, dt, buf, blocklen, seed, basedata,
756 64a8571e 2022-01-07 stsp basefile_offset0);
759 64a8571e 2022-01-07 stsp if (block != NULL) {
761 64a8571e 2022-01-07 stsp * We have found a matching block in the delta base.
762 64a8571e 2022-01-07 stsp * Attempt to stretch the block as far as possible and
763 64a8571e 2022-01-07 stsp * generate a copy instruction.
765 64a8571e 2022-01-07 stsp err = stretchblk_file_mem(basedata, basefile_offset0,
766 64a8571e 2022-01-07 stsp basefile_size, block, f, filesize, &blocklen);
769 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
770 64a8571e 2022-01-07 stsp alloc_chunk_size, 1, block->offset, blocklen);
775 64a8571e 2022-01-07 stsp * No match.
776 64a8571e 2022-01-07 stsp * This block needs to be sourced from the file itself.
778 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
779 64a8571e 2022-01-07 stsp alloc_chunk_size, 0, fileoffset - offset0, blocklen);
783 64a8571e 2022-01-07 stsp fileoffset += blocklen;
784 64a8571e 2022-01-07 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1) {
785 64a8571e 2022-01-07 stsp err = got_error_from_errno("fseeko");
791 64a8571e 2022-01-07 stsp free(*deltas);
792 64a8571e 2022-01-07 stsp *deltas = NULL;
793 64a8571e 2022-01-07 stsp *ndeltas = 0;
795 64a8571e 2022-01-07 stsp return err;
798 64a8571e 2022-01-07 stsp const struct got_error *
799 64a8571e 2022-01-07 stsp got_deltify_mem_file(struct got_delta_instruction **deltas, int *ndeltas,
800 d6a28ffe 2022-05-20 op uint8_t *data, off_t fileoffset, off_t filesize, uint32_t seed,
801 64a8571e 2022-01-07 stsp struct got_delta_table *dt, FILE *basefile,
802 64a8571e 2022-01-07 stsp off_t basefile_offset0, off_t basefile_size)
804 64a8571e 2022-01-07 stsp const struct got_error *err = NULL;
805 64a8571e 2022-01-07 stsp const off_t offset0 = fileoffset;
806 64a8571e 2022-01-07 stsp size_t nalloc = 0;
807 64a8571e 2022-01-07 stsp const size_t alloc_chunk_size = 64;
809 64a8571e 2022-01-07 stsp *deltas = NULL;
810 64a8571e 2022-01-07 stsp *ndeltas = 0;
812 64a8571e 2022-01-07 stsp *deltas = reallocarray(NULL, alloc_chunk_size,
813 64a8571e 2022-01-07 stsp sizeof(struct got_delta_instruction));
814 64a8571e 2022-01-07 stsp if (*deltas == NULL)
815 64a8571e 2022-01-07 stsp return got_error_from_errno("reallocarray");
816 64a8571e 2022-01-07 stsp nalloc = alloc_chunk_size;
818 64a8571e 2022-01-07 stsp while (fileoffset < filesize) {
819 64a8571e 2022-01-07 stsp off_t blocklen;
820 64a8571e 2022-01-07 stsp struct got_delta_block *block;
821 64a8571e 2022-01-07 stsp err = nextblk_mem(&blocklen, data, fileoffset, filesize);
824 64a8571e 2022-01-07 stsp if (blocklen == 0) {
825 64a8571e 2022-01-07 stsp /* Source remainder from the file itself. */
826 64a8571e 2022-01-07 stsp if (fileoffset < filesize) {
827 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
828 64a8571e 2022-01-07 stsp alloc_chunk_size, 0, fileoffset - offset0,
829 64a8571e 2022-01-07 stsp filesize - fileoffset);
833 d6a28ffe 2022-05-20 op err = lookupblk(&block, dt, data + fileoffset, blocklen, seed,
834 64a8571e 2022-01-07 stsp basefile, basefile_offset0);
837 64a8571e 2022-01-07 stsp if (block != NULL) {
839 64a8571e 2022-01-07 stsp * We have found a matching block in the delta base.
840 64a8571e 2022-01-07 stsp * Attempt to stretch the block as far as possible and
841 64a8571e 2022-01-07 stsp * generate a copy instruction.
843 64a8571e 2022-01-07 stsp err = stretchblk_mem_file(basefile, basefile_offset0,
844 64a8571e 2022-01-07 stsp block, data, fileoffset + blocklen, filesize,
845 64a8571e 2022-01-07 stsp &blocklen);
848 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
849 64a8571e 2022-01-07 stsp alloc_chunk_size, 1, block->offset, blocklen);
854 64a8571e 2022-01-07 stsp * No match.
855 64a8571e 2022-01-07 stsp * This block needs to be sourced from the file itself.
857 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
858 64a8571e 2022-01-07 stsp alloc_chunk_size, 0, fileoffset - offset0, blocklen);
862 64a8571e 2022-01-07 stsp fileoffset += blocklen;
866 64a8571e 2022-01-07 stsp free(*deltas);
867 64a8571e 2022-01-07 stsp *deltas = NULL;
868 64a8571e 2022-01-07 stsp *ndeltas = 0;
870 64a8571e 2022-01-07 stsp return err;
873 64a8571e 2022-01-07 stsp const struct got_error *
874 64a8571e 2022-01-07 stsp got_deltify_mem_mem(struct got_delta_instruction **deltas, int *ndeltas,
875 d6a28ffe 2022-05-20 op uint8_t *data, off_t fileoffset, off_t filesize, uint32_t seed,
876 64a8571e 2022-01-07 stsp struct got_delta_table *dt, uint8_t *basedata,
877 64a8571e 2022-01-07 stsp off_t basefile_offset0, off_t basefile_size)
879 64a8571e 2022-01-07 stsp const struct got_error *err = NULL;
880 64a8571e 2022-01-07 stsp const off_t offset0 = fileoffset;
881 64a8571e 2022-01-07 stsp size_t nalloc = 0;
882 64a8571e 2022-01-07 stsp const size_t alloc_chunk_size = 64;
884 64a8571e 2022-01-07 stsp *deltas = NULL;
885 64a8571e 2022-01-07 stsp *ndeltas = 0;
887 64a8571e 2022-01-07 stsp *deltas = reallocarray(NULL, alloc_chunk_size,
888 64a8571e 2022-01-07 stsp sizeof(struct got_delta_instruction));
889 64a8571e 2022-01-07 stsp if (*deltas == NULL)
890 64a8571e 2022-01-07 stsp return got_error_from_errno("reallocarray");
891 64a8571e 2022-01-07 stsp nalloc = alloc_chunk_size;
893 64a8571e 2022-01-07 stsp while (fileoffset < filesize) {
894 64a8571e 2022-01-07 stsp off_t blocklen;
895 64a8571e 2022-01-07 stsp struct got_delta_block *block;
896 64a8571e 2022-01-07 stsp err = nextblk_mem(&blocklen, data, fileoffset, filesize);
899 64a8571e 2022-01-07 stsp if (blocklen == 0) {
900 64a8571e 2022-01-07 stsp /* Source remainder from the file itself. */
901 64a8571e 2022-01-07 stsp if (fileoffset < filesize) {
902 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
903 64a8571e 2022-01-07 stsp alloc_chunk_size, 0, fileoffset - offset0,
904 64a8571e 2022-01-07 stsp filesize - fileoffset);
908 64a8571e 2022-01-07 stsp err = lookupblk_mem(&block, dt, data + fileoffset, blocklen,
909 d6a28ffe 2022-05-20 op seed, basedata, basefile_offset0);
912 64a8571e 2022-01-07 stsp if (block != NULL) {
914 64a8571e 2022-01-07 stsp * We have found a matching block in the delta base.
915 64a8571e 2022-01-07 stsp * Attempt to stretch the block as far as possible and
916 64a8571e 2022-01-07 stsp * generate a copy instruction.
918 64a8571e 2022-01-07 stsp err = stretchblk_mem_mem(basedata, basefile_offset0,
919 64a8571e 2022-01-07 stsp basefile_size, block, data, fileoffset + blocklen,
920 64a8571e 2022-01-07 stsp filesize, &blocklen);
923 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
924 64a8571e 2022-01-07 stsp alloc_chunk_size, 1, block->offset, blocklen);
929 64a8571e 2022-01-07 stsp * No match.
930 64a8571e 2022-01-07 stsp * This block needs to be sourced from the file itself.
932 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
933 64a8571e 2022-01-07 stsp alloc_chunk_size, 0, fileoffset - offset0, blocklen);
937 64a8571e 2022-01-07 stsp fileoffset += blocklen;
941 64a8571e 2022-01-07 stsp free(*deltas);
942 64a8571e 2022-01-07 stsp *deltas = NULL;
943 64a8571e 2022-01-07 stsp *ndeltas = 0;
945 64a8571e 2022-01-07 stsp return err;