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>
29 f364801d 2021-03-10 stsp #include <sha1.h>
31 f364801d 2021-03-10 stsp #include "got_error.h"
33 69aa0e90 2021-03-10 stsp #include "got_lib_deltify.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 6eab69f7 2021-10-07 stsp static 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 f364801d 2021-03-10 stsp static uint64_t
91 69aa0e90 2021-03-10 stsp hashblk(const unsigned char *p, off_t n)
93 f364801d 2021-03-10 stsp unsigned char buf[SHA1_DIGEST_LENGTH];
95 69aa0e90 2021-03-10 stsp SHA1_CTX ctx;
96 69aa0e90 2021-03-10 stsp SHA1Init(&ctx);
97 69aa0e90 2021-03-10 stsp SHA1Update(&ctx, p, n);
98 69aa0e90 2021-03-10 stsp SHA1Final(buf, &ctx);
99 f364801d 2021-03-10 stsp memcpy(&h, buf, sizeof(h));
100 f364801d 2021-03-10 stsp return be64toh(h);
103 f364801d 2021-03-10 stsp static const struct got_error *
104 740bba1c 2021-06-18 stsp addblk(struct got_delta_table *dt, FILE *f, off_t file_offset0, off_t len,
105 740bba1c 2021-06-18 stsp off_t offset, uint64_t h)
107 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
109 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
110 0d15f6dc 2021-06-13 stsp uint8_t buf2[GOT_DELTIFY_MAXCHUNK];
111 69aa0e90 2021-03-10 stsp size_t r = 0;
113 69aa0e90 2021-03-10 stsp if (len == 0)
114 69aa0e90 2021-03-10 stsp return NULL;
116 69aa0e90 2021-03-10 stsp i = h % dt->nalloc;
117 69aa0e90 2021-03-10 stsp while (dt->blocks[i].len != 0) {
119 69aa0e90 2021-03-10 stsp * Avoid adding duplicate blocks.
120 69aa0e90 2021-03-10 stsp * NB: A matching hash is insufficient for detecting equality.
121 51a494da 2021-06-11 stsp * The hash can only detect inequality.
123 51a494da 2021-06-11 stsp if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
124 0d15f6dc 2021-06-13 stsp if (r == 0) {
125 740bba1c 2021-06-18 stsp if (fseeko(f, file_offset0 + offset, SEEK_SET) == -1)
126 0d15f6dc 2021-06-13 stsp return got_error_from_errno("fseeko");
127 0d15f6dc 2021-06-13 stsp r = fread(buf, 1, len, f);
128 0d15f6dc 2021-06-13 stsp if (r != len) {
129 0d15f6dc 2021-06-13 stsp if (!ferror(f))
130 0d15f6dc 2021-06-13 stsp return NULL;
131 0d15f6dc 2021-06-13 stsp return got_ferror(f, GOT_ERR_IO);
134 740bba1c 2021-06-18 stsp if (fseeko(f, file_offset0 + dt->blocks[i].offset,
135 740bba1c 2021-06-18 stsp SEEK_SET) == -1)
136 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
137 69aa0e90 2021-03-10 stsp if (fread(buf2, 1, len, f) != len)
138 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
139 69aa0e90 2021-03-10 stsp if (memcmp(buf, buf2, len) == 0)
140 69aa0e90 2021-03-10 stsp return NULL;
143 69aa0e90 2021-03-10 stsp i = (i + 1) % dt->nalloc;
145 69aa0e90 2021-03-10 stsp assert(dt->blocks[i].len == 0);
146 69aa0e90 2021-03-10 stsp dt->blocks[i].len = len;
147 69aa0e90 2021-03-10 stsp dt->blocks[i].offset = offset;
148 69aa0e90 2021-03-10 stsp dt->blocks[i].hash = h;
149 69aa0e90 2021-03-10 stsp dt->nblocks++;
150 e89540a9 2021-06-11 stsp if (dt->nalloc < dt->nblocks + 64) {
151 69aa0e90 2021-03-10 stsp struct got_delta_block *db;
152 e89540a9 2021-06-11 stsp size_t old_size = dt->nalloc;
153 69aa0e90 2021-03-10 stsp db = dt->blocks;
154 e89540a9 2021-06-11 stsp dt->blocks = calloc(dt->nalloc + 64,
155 e89540a9 2021-06-11 stsp sizeof(struct got_delta_block));
156 69aa0e90 2021-03-10 stsp if (dt->blocks == NULL) {
157 69aa0e90 2021-03-10 stsp err = got_error_from_errno("calloc");
158 69aa0e90 2021-03-10 stsp dt->blocks = db;
159 69aa0e90 2021-03-10 stsp return err;
161 e89540a9 2021-06-11 stsp dt->nalloc += 64;
163 69aa0e90 2021-03-10 stsp * Recompute all block positions. Hash-based indices of blocks
164 69aa0e90 2021-03-10 stsp * in the array depend on the allocated length of the array.
166 69aa0e90 2021-03-10 stsp dt->nblocks = 0;
167 a893025f 2021-06-11 stsp for (i = 0; i < old_size; i++) {
168 69aa0e90 2021-03-10 stsp if (db[i].len == 0)
170 740bba1c 2021-06-18 stsp err = addblk(dt, f, file_offset0, db[i].len,
171 64a8571e 2022-01-07 stsp db[i].offset, db[i].hash);
178 64a8571e 2022-01-07 stsp return err;
181 64a8571e 2022-01-07 stsp static const struct got_error *
182 64a8571e 2022-01-07 stsp addblk_mem(struct got_delta_table *dt, uint8_t *data, off_t file_offset0,
183 64a8571e 2022-01-07 stsp off_t len, off_t offset, uint64_t h)
185 64a8571e 2022-01-07 stsp const struct got_error *err = NULL;
187 64a8571e 2022-01-07 stsp uint8_t *block1;
188 64a8571e 2022-01-07 stsp uint8_t *block2;
190 64a8571e 2022-01-07 stsp if (len == 0)
191 64a8571e 2022-01-07 stsp return NULL;
193 64a8571e 2022-01-07 stsp i = h % dt->nalloc;
194 64a8571e 2022-01-07 stsp while (dt->blocks[i].len != 0) {
196 64a8571e 2022-01-07 stsp * Avoid adding duplicate blocks.
197 64a8571e 2022-01-07 stsp * NB: A matching hash is insufficient for detecting equality.
198 64a8571e 2022-01-07 stsp * The hash can only detect inequality.
200 64a8571e 2022-01-07 stsp if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
201 64a8571e 2022-01-07 stsp block1 = data + file_offset0 + dt->blocks[i].offset;
202 64a8571e 2022-01-07 stsp block2 = data + file_offset0 + offset;
203 64a8571e 2022-01-07 stsp if (memcmp(block1, block2, len) == 0)
204 64a8571e 2022-01-07 stsp return NULL;
207 64a8571e 2022-01-07 stsp i = (i + 1) % dt->nalloc;
209 64a8571e 2022-01-07 stsp assert(dt->blocks[i].len == 0);
210 64a8571e 2022-01-07 stsp dt->blocks[i].len = len;
211 64a8571e 2022-01-07 stsp dt->blocks[i].offset = offset;
212 64a8571e 2022-01-07 stsp dt->blocks[i].hash = h;
213 64a8571e 2022-01-07 stsp dt->nblocks++;
214 64a8571e 2022-01-07 stsp if (dt->nalloc < dt->nblocks + 64) {
215 64a8571e 2022-01-07 stsp struct got_delta_block *db;
216 64a8571e 2022-01-07 stsp size_t old_size = dt->nalloc;
217 64a8571e 2022-01-07 stsp db = dt->blocks;
218 64a8571e 2022-01-07 stsp dt->blocks = calloc(dt->nalloc + 64,
219 64a8571e 2022-01-07 stsp sizeof(struct got_delta_block));
220 64a8571e 2022-01-07 stsp if (dt->blocks == NULL) {
221 64a8571e 2022-01-07 stsp err = got_error_from_errno("calloc");
222 64a8571e 2022-01-07 stsp dt->blocks = db;
223 64a8571e 2022-01-07 stsp return err;
225 64a8571e 2022-01-07 stsp dt->nalloc += 64;
227 64a8571e 2022-01-07 stsp * Recompute all block positions. Hash-based indices of blocks
228 64a8571e 2022-01-07 stsp * in the array depend on the allocated length of the array.
230 64a8571e 2022-01-07 stsp dt->nblocks = 0;
231 64a8571e 2022-01-07 stsp for (i = 0; i < old_size; i++) {
232 64a8571e 2022-01-07 stsp if (db[i].len == 0)
234 64a8571e 2022-01-07 stsp err = addblk_mem(dt, data, file_offset0, db[i].len,
235 740bba1c 2021-06-18 stsp db[i].offset, db[i].hash);
242 69aa0e90 2021-03-10 stsp return err;
245 69aa0e90 2021-03-10 stsp static const struct got_error *
246 69aa0e90 2021-03-10 stsp lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
247 740bba1c 2021-06-18 stsp unsigned char *p, off_t len, FILE *basefile, off_t basefile_offset0)
250 f364801d 2021-03-10 stsp uint64_t h;
251 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
254 69aa0e90 2021-03-10 stsp *block = NULL;
256 69aa0e90 2021-03-10 stsp h = hashblk(p, len);
257 69aa0e90 2021-03-10 stsp for (i = h % dt->nalloc; dt->blocks[i].len != 0;
258 69aa0e90 2021-03-10 stsp i = (i + 1) % dt->nalloc) {
259 69aa0e90 2021-03-10 stsp if (dt->blocks[i].hash != h ||
260 69aa0e90 2021-03-10 stsp dt->blocks[i].len != len)
262 740bba1c 2021-06-18 stsp if (fseeko(basefile, basefile_offset0 + dt->blocks[i].offset,
263 740bba1c 2021-06-18 stsp SEEK_SET) == -1)
264 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
265 69aa0e90 2021-03-10 stsp r = fread(buf, 1, len, basefile);
266 69aa0e90 2021-03-10 stsp if (r != len)
267 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
268 69aa0e90 2021-03-10 stsp if (memcmp(p, buf, len) == 0) {
269 69aa0e90 2021-03-10 stsp *block = &dt->blocks[i];
273 f364801d 2021-03-10 stsp return NULL;
276 69aa0e90 2021-03-10 stsp static const struct got_error *
277 64a8571e 2022-01-07 stsp lookupblk_mem(struct got_delta_block **block, struct got_delta_table *dt,
278 64a8571e 2022-01-07 stsp unsigned char *p, off_t len, uint8_t *basedata, off_t basefile_offset0)
281 64a8571e 2022-01-07 stsp uint64_t h;
282 64a8571e 2022-01-07 stsp uint8_t *b;
284 64a8571e 2022-01-07 stsp *block = NULL;
286 64a8571e 2022-01-07 stsp h = hashblk(p, len);
287 64a8571e 2022-01-07 stsp for (i = h % dt->nalloc; dt->blocks[i].len != 0;
288 64a8571e 2022-01-07 stsp i = (i + 1) % dt->nalloc) {
289 64a8571e 2022-01-07 stsp if (dt->blocks[i].hash != h ||
290 64a8571e 2022-01-07 stsp dt->blocks[i].len != len)
292 64a8571e 2022-01-07 stsp b = basedata + basefile_offset0 + dt->blocks[i].offset;
293 64a8571e 2022-01-07 stsp if (memcmp(p, b, len) == 0) {
294 64a8571e 2022-01-07 stsp *block = &dt->blocks[i];
298 64a8571e 2022-01-07 stsp return NULL;
301 64a8571e 2022-01-07 stsp static const struct got_error *
302 69aa0e90 2021-03-10 stsp nextblk(uint8_t *buf, off_t *blocklen, FILE *f)
304 f364801d 2021-03-10 stsp uint32_t gh;
305 69aa0e90 2021-03-10 stsp const unsigned char *p;
307 69aa0e90 2021-03-10 stsp off_t pos = ftello(f);
309 69aa0e90 2021-03-10 stsp *blocklen = 0;
311 69aa0e90 2021-03-10 stsp r = fread(buf, 1, GOT_DELTIFY_MAXCHUNK, f);
312 69aa0e90 2021-03-10 stsp if (r == 0 && ferror(f))
313 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
314 69aa0e90 2021-03-10 stsp if (r < GOT_DELTIFY_MINCHUNK)
315 69aa0e90 2021-03-10 stsp return NULL; /* no more delta-worthy blocks left */
317 69aa0e90 2021-03-10 stsp /* Got a deltifiable block. Find the split-point where it ends. */
318 69aa0e90 2021-03-10 stsp p = buf + GOT_DELTIFY_MINCHUNK;
320 69aa0e90 2021-03-10 stsp while (p != buf + r) {
321 f364801d 2021-03-10 stsp gh = (gh << 1) + geartab[*p++];
322 69aa0e90 2021-03-10 stsp if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
326 69aa0e90 2021-03-10 stsp *blocklen = (p - buf);
327 69aa0e90 2021-03-10 stsp if (fseeko(f, pos + *blocklen, SEEK_SET) == -1)
328 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
330 69aa0e90 2021-03-10 stsp return NULL;
333 64a8571e 2022-01-07 stsp static const struct got_error *
334 64a8571e 2022-01-07 stsp nextblk_mem(off_t *blocklen, uint8_t *data, off_t fileoffset, off_t filesize)
336 64a8571e 2022-01-07 stsp uint32_t gh;
337 64a8571e 2022-01-07 stsp const unsigned char *p;
339 64a8571e 2022-01-07 stsp *blocklen = 0;
341 64a8571e 2022-01-07 stsp if (fileoffset >= filesize ||
342 64a8571e 2022-01-07 stsp filesize - fileoffset < GOT_DELTIFY_MINCHUNK)
343 64a8571e 2022-01-07 stsp return NULL; /* no more delta-worthy blocks left */
345 64a8571e 2022-01-07 stsp /* Got a deltifiable block. Find the split-point where it ends. */
346 64a8571e 2022-01-07 stsp p = data + fileoffset + GOT_DELTIFY_MINCHUNK;
348 64a8571e 2022-01-07 stsp while (p != data + MIN(fileoffset + GOT_DELTIFY_MAXCHUNK, filesize)) {
349 64a8571e 2022-01-07 stsp gh = (gh << 1) + geartab[*p++];
350 64a8571e 2022-01-07 stsp if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
354 64a8571e 2022-01-07 stsp *blocklen = (p - (data + fileoffset));
355 64a8571e 2022-01-07 stsp return NULL;
358 f364801d 2021-03-10 stsp const struct got_error *
359 69aa0e90 2021-03-10 stsp got_deltify_init(struct got_delta_table **dt, FILE *f, off_t fileoffset,
360 69aa0e90 2021-03-10 stsp off_t filesize)
362 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
363 f364801d 2021-03-10 stsp uint64_t h;
364 740bba1c 2021-06-18 stsp const off_t offset0 = fileoffset;
366 69aa0e90 2021-03-10 stsp *dt = calloc(1, sizeof(**dt));
367 69aa0e90 2021-03-10 stsp if (*dt == NULL)
368 f364801d 2021-03-10 stsp return got_error_from_errno("calloc");
370 69aa0e90 2021-03-10 stsp (*dt)->nblocks = 0;
371 69aa0e90 2021-03-10 stsp (*dt)->nalloc = 128;
372 69aa0e90 2021-03-10 stsp (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
373 69aa0e90 2021-03-10 stsp if ((*dt)->blocks == NULL) {
374 69aa0e90 2021-03-10 stsp err = got_error_from_errno("calloc");
378 69aa0e90 2021-03-10 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
379 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
381 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
382 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
383 69aa0e90 2021-03-10 stsp off_t blocklen;
384 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
387 69aa0e90 2021-03-10 stsp if (blocklen == 0)
389 69aa0e90 2021-03-10 stsp h = hashblk(buf, blocklen);
390 740bba1c 2021-06-18 stsp err = addblk(*dt, f, offset0, blocklen,
391 740bba1c 2021-06-18 stsp fileoffset - offset0, h);
394 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
395 68bdcdc2 2021-06-11 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
396 68bdcdc2 2021-06-11 stsp return got_error_from_errno("fseeko");
400 64a8571e 2022-01-07 stsp free((*dt)->blocks);
402 64a8571e 2022-01-07 stsp *dt = NULL;
405 64a8571e 2022-01-07 stsp return err;
408 64a8571e 2022-01-07 stsp const struct got_error *
409 64a8571e 2022-01-07 stsp got_deltify_init_mem(struct got_delta_table **dt, uint8_t *data,
410 64a8571e 2022-01-07 stsp off_t fileoffset, off_t filesize)
412 64a8571e 2022-01-07 stsp const struct got_error *err = NULL;
413 64a8571e 2022-01-07 stsp uint64_t h;
414 64a8571e 2022-01-07 stsp const off_t offset0 = fileoffset;
416 64a8571e 2022-01-07 stsp *dt = calloc(1, sizeof(**dt));
417 64a8571e 2022-01-07 stsp if (*dt == NULL)
418 64a8571e 2022-01-07 stsp return got_error_from_errno("calloc");
420 64a8571e 2022-01-07 stsp (*dt)->nblocks = 0;
421 64a8571e 2022-01-07 stsp (*dt)->nalloc = 128;
422 64a8571e 2022-01-07 stsp (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
423 64a8571e 2022-01-07 stsp if ((*dt)->blocks == NULL) {
424 64a8571e 2022-01-07 stsp err = got_error_from_errno("calloc");
428 64a8571e 2022-01-07 stsp while (fileoffset < filesize) {
429 64a8571e 2022-01-07 stsp off_t blocklen;
430 64a8571e 2022-01-07 stsp err = nextblk_mem(&blocklen, data, fileoffset, filesize);
433 64a8571e 2022-01-07 stsp if (blocklen == 0)
435 64a8571e 2022-01-07 stsp h = hashblk(data + fileoffset, blocklen);
436 64a8571e 2022-01-07 stsp err = addblk_mem(*dt, data, offset0, blocklen,
437 64a8571e 2022-01-07 stsp fileoffset - offset0, h);
440 64a8571e 2022-01-07 stsp fileoffset += blocklen;
444 69aa0e90 2021-03-10 stsp free((*dt)->blocks);
446 69aa0e90 2021-03-10 stsp *dt = NULL;
449 69aa0e90 2021-03-10 stsp return err;
453 69aa0e90 2021-03-10 stsp got_deltify_free(struct got_delta_table *dt)
455 dbbf4a5f 2021-05-20 stsp if (dt == NULL)
457 69aa0e90 2021-03-10 stsp free(dt->blocks);
461 f364801d 2021-03-10 stsp static const struct got_error *
462 dd29967c 2021-08-22 stsp emitdelta(struct got_delta_instruction **deltas, size_t *nalloc, int *ndeltas,
463 dd29967c 2021-08-22 stsp const size_t alloc_chunk_size, int copy, off_t offset, off_t len)
465 69aa0e90 2021-03-10 stsp struct got_delta_instruction *d, *p;
467 dd29967c 2021-08-22 stsp if (*nalloc < *ndeltas + alloc_chunk_size) {
468 dd29967c 2021-08-22 stsp p = reallocarray(*deltas, *nalloc + alloc_chunk_size,
469 dd29967c 2021-08-22 stsp sizeof(struct got_delta_instruction));
470 dd29967c 2021-08-22 stsp if (p == NULL)
471 2d467c6d 2021-10-13 stsp return got_error_from_errno("reallocarray");
472 dd29967c 2021-08-22 stsp *deltas = p;
473 dd29967c 2021-08-22 stsp *nalloc += alloc_chunk_size;
475 69aa0e90 2021-03-10 stsp *ndeltas += 1;
476 69aa0e90 2021-03-10 stsp d = &(*deltas)[*ndeltas - 1];
477 69aa0e90 2021-03-10 stsp d->copy = copy;
478 69aa0e90 2021-03-10 stsp d->offset = offset;
479 56e7a15f 2021-03-10 stsp d->len = len;
480 f364801d 2021-03-10 stsp return NULL;
483 69aa0e90 2021-03-10 stsp static const struct got_error *
484 740bba1c 2021-06-18 stsp stretchblk(FILE *basefile, off_t base_offset0, struct got_delta_block *block,
485 740bba1c 2021-06-18 stsp FILE *f, off_t filesize, off_t *blocklen)
487 69aa0e90 2021-03-10 stsp uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
488 69aa0e90 2021-03-10 stsp size_t base_r, r, i;
489 69aa0e90 2021-03-10 stsp int buf_equal = 1;
491 5de743f8 2021-08-29 stsp if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
492 5de743f8 2021-08-29 stsp SEEK_SET) == -1)
493 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
495 69aa0e90 2021-03-10 stsp while (buf_equal && *blocklen < (1 << 24) - 1) {
496 69aa0e90 2021-03-10 stsp base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
497 69aa0e90 2021-03-10 stsp if (base_r == 0) {
498 69aa0e90 2021-03-10 stsp if (ferror(basefile))
499 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
502 69aa0e90 2021-03-10 stsp r = fread(buf, 1, sizeof(buf), f);
503 69aa0e90 2021-03-10 stsp if (r == 0) {
504 69aa0e90 2021-03-10 stsp if (ferror(f))
505 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
508 69aa0e90 2021-03-10 stsp for (i = 0; i < MIN(base_r, r); i++) {
509 69aa0e90 2021-03-10 stsp if (buf[i] != basebuf[i]) {
510 69aa0e90 2021-03-10 stsp buf_equal = 0;
513 69aa0e90 2021-03-10 stsp (*blocklen)++;
517 64a8571e 2022-01-07 stsp return NULL;
520 64a8571e 2022-01-07 stsp static const struct got_error *
521 64a8571e 2022-01-07 stsp stretchblk_file_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
522 64a8571e 2022-01-07 stsp struct got_delta_block *block, FILE *f, off_t filesize, off_t *blocklen)
524 64a8571e 2022-01-07 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
525 64a8571e 2022-01-07 stsp size_t r, i;
526 64a8571e 2022-01-07 stsp int buf_equal = 1;
527 64a8571e 2022-01-07 stsp off_t base_offset = base_offset0 + block->offset + *blocklen;
529 64a8571e 2022-01-07 stsp if (base_offset > basefile_size) {
530 64a8571e 2022-01-07 stsp return got_error_fmt(GOT_ERR_RANGE,
531 64a8571e 2022-01-07 stsp "read beyond the size of delta base at offset %llu",
532 64a8571e 2022-01-07 stsp base_offset);
535 64a8571e 2022-01-07 stsp while (buf_equal && *blocklen < (1 << 24) - 1) {
536 64a8571e 2022-01-07 stsp if (base_offset + *blocklen >= basefile_size)
538 64a8571e 2022-01-07 stsp r = fread(buf, 1, sizeof(buf), f);
539 64a8571e 2022-01-07 stsp if (r == 0) {
540 64a8571e 2022-01-07 stsp if (ferror(f))
541 64a8571e 2022-01-07 stsp return got_ferror(f, GOT_ERR_IO);
544 64a8571e 2022-01-07 stsp for (i = 0; i < MIN(basefile_size - base_offset, r); i++) {
545 64a8571e 2022-01-07 stsp if (buf[i] != *(basedata + base_offset + i)) {
546 64a8571e 2022-01-07 stsp buf_equal = 0;
549 64a8571e 2022-01-07 stsp (*blocklen)++;
553 64a8571e 2022-01-07 stsp return NULL;
556 64a8571e 2022-01-07 stsp static const struct got_error *
557 64a8571e 2022-01-07 stsp stretchblk_mem_file(FILE *basefile, off_t base_offset0,
558 64a8571e 2022-01-07 stsp struct got_delta_block *block, uint8_t *data, off_t fileoffset,
559 64a8571e 2022-01-07 stsp off_t filesize, off_t *blocklen)
561 64a8571e 2022-01-07 stsp uint8_t basebuf[GOT_DELTIFY_MAXCHUNK];
562 64a8571e 2022-01-07 stsp size_t base_r, i;
563 64a8571e 2022-01-07 stsp int buf_equal = 1;
565 64a8571e 2022-01-07 stsp if (fileoffset > filesize) {
566 64a8571e 2022-01-07 stsp return got_error_fmt(GOT_ERR_RANGE,
567 64a8571e 2022-01-07 stsp "read beyond the size of deltify file at offset %llu",
568 64a8571e 2022-01-07 stsp fileoffset);
571 64a8571e 2022-01-07 stsp if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
572 64a8571e 2022-01-07 stsp SEEK_SET) == -1)
573 64a8571e 2022-01-07 stsp return got_error_from_errno("fseeko");
575 64a8571e 2022-01-07 stsp while (buf_equal && *blocklen < (1 << 24) - 1) {
576 64a8571e 2022-01-07 stsp if (fileoffset + *blocklen >= filesize)
578 64a8571e 2022-01-07 stsp base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
579 64a8571e 2022-01-07 stsp if (base_r == 0) {
580 64a8571e 2022-01-07 stsp if (ferror(basefile))
581 64a8571e 2022-01-07 stsp return got_ferror(basefile, GOT_ERR_IO);
584 64a8571e 2022-01-07 stsp for (i = 0; i < MIN(base_r, filesize - fileoffset); i++) {
585 64a8571e 2022-01-07 stsp if (*(data + fileoffset + i) != basebuf[i]) {
586 64a8571e 2022-01-07 stsp buf_equal = 0;
589 64a8571e 2022-01-07 stsp (*blocklen)++;
593 64a8571e 2022-01-07 stsp return NULL;
596 64a8571e 2022-01-07 stsp static const struct got_error *
597 64a8571e 2022-01-07 stsp stretchblk_mem_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
598 64a8571e 2022-01-07 stsp struct got_delta_block *block, uint8_t *data, off_t fileoffset,
599 64a8571e 2022-01-07 stsp off_t filesize, off_t *blocklen)
601 64a8571e 2022-01-07 stsp off_t i, maxlen;
602 64a8571e 2022-01-07 stsp off_t base_offset = base_offset0 + block->offset + *blocklen;
603 64a8571e 2022-01-07 stsp uint8_t *p, *q;
605 64a8571e 2022-01-07 stsp if (base_offset > basefile_size) {
606 64a8571e 2022-01-07 stsp return got_error_fmt(GOT_ERR_RANGE,
607 64a8571e 2022-01-07 stsp "read beyond the size of delta base at offset %llu",
608 64a8571e 2022-01-07 stsp base_offset);
611 64a8571e 2022-01-07 stsp if (fileoffset > filesize) {
612 64a8571e 2022-01-07 stsp return got_error_fmt(GOT_ERR_RANGE,
613 64a8571e 2022-01-07 stsp "read beyond the size of deltify file at offset %llu",
614 64a8571e 2022-01-07 stsp fileoffset);
617 64a8571e 2022-01-07 stsp p = data + fileoffset;
618 64a8571e 2022-01-07 stsp q = basedata + base_offset;
619 64a8571e 2022-01-07 stsp maxlen = MIN(basefile_size - base_offset, filesize - fileoffset);
620 64a8571e 2022-01-07 stsp for (i = 0; i < maxlen && *blocklen < (1 << 24) - 1; i++) {
621 64a8571e 2022-01-07 stsp if (p[i] != q[i])
623 64a8571e 2022-01-07 stsp (*blocklen)++;
626 69aa0e90 2021-03-10 stsp return NULL;
629 69aa0e90 2021-03-10 stsp const struct got_error *
630 69aa0e90 2021-03-10 stsp got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
631 69aa0e90 2021-03-10 stsp FILE *f, off_t fileoffset, off_t filesize,
632 69aa0e90 2021-03-10 stsp struct got_delta_table *dt, FILE *basefile,
633 f34b169e 2021-06-18 stsp off_t basefile_offset0, off_t basefile_size)
635 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
636 69aa0e90 2021-03-10 stsp const off_t offset0 = fileoffset;
637 dd29967c 2021-08-22 stsp size_t nalloc = 0;
638 dd29967c 2021-08-22 stsp const size_t alloc_chunk_size = 64;
640 69aa0e90 2021-03-10 stsp *deltas = NULL;
641 69aa0e90 2021-03-10 stsp *ndeltas = 0;
644 69aa0e90 2021-03-10 stsp * offset0 indicates where data to be deltified begins.
645 69aa0e90 2021-03-10 stsp * For example, we want to avoid deltifying a Git object header at
646 69aa0e90 2021-03-10 stsp * the beginning of the file.
648 69aa0e90 2021-03-10 stsp if (fseeko(f, offset0, SEEK_SET) == -1)
649 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
651 dd29967c 2021-08-22 stsp *deltas = reallocarray(NULL, alloc_chunk_size,
652 dd29967c 2021-08-22 stsp sizeof(struct got_delta_instruction));
653 dd29967c 2021-08-22 stsp if (*deltas == NULL)
654 dd29967c 2021-08-22 stsp return got_error_from_errno("reallocarray");
655 dd29967c 2021-08-22 stsp nalloc = alloc_chunk_size;
657 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
658 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
659 69aa0e90 2021-03-10 stsp off_t blocklen;
660 69aa0e90 2021-03-10 stsp struct got_delta_block *block;
661 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
664 9a8dc2b3 2021-06-18 stsp if (blocklen == 0) {
665 9a8dc2b3 2021-06-18 stsp /* Source remainder from the file itself. */
666 9a8dc2b3 2021-06-18 stsp if (fileoffset < filesize) {
667 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
668 dd29967c 2021-08-22 stsp alloc_chunk_size, 0, fileoffset - offset0,
669 9a8dc2b3 2021-06-18 stsp filesize - fileoffset);
673 740bba1c 2021-06-18 stsp err = lookupblk(&block, dt, buf, blocklen, basefile,
674 740bba1c 2021-06-18 stsp basefile_offset0);
677 69aa0e90 2021-03-10 stsp if (block != NULL) {
679 69aa0e90 2021-03-10 stsp * We have found a matching block in the delta base.
680 69aa0e90 2021-03-10 stsp * Attempt to stretch the block as far as possible and
681 69aa0e90 2021-03-10 stsp * generate a copy instruction.
683 740bba1c 2021-06-18 stsp err = stretchblk(basefile, basefile_offset0, block,
684 740bba1c 2021-06-18 stsp f, filesize, &blocklen);
687 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
688 dd29967c 2021-08-22 stsp alloc_chunk_size, 1, block->offset, blocklen);
693 69aa0e90 2021-03-10 stsp * No match.
694 69aa0e90 2021-03-10 stsp * This block needs to be sourced from the file itself.
696 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
697 dd29967c 2021-08-22 stsp alloc_chunk_size, 0, fileoffset - offset0, blocklen);
701 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
702 0af64e86 2021-08-22 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1) {
703 0af64e86 2021-08-22 stsp err = got_error_from_errno("fseeko");
709 69aa0e90 2021-03-10 stsp free(*deltas);
710 69aa0e90 2021-03-10 stsp *deltas = NULL;
711 69aa0e90 2021-03-10 stsp *ndeltas = 0;
713 69aa0e90 2021-03-10 stsp return err;
716 64a8571e 2022-01-07 stsp const struct got_error *
717 64a8571e 2022-01-07 stsp got_deltify_file_mem(struct got_delta_instruction **deltas, int *ndeltas,
718 64a8571e 2022-01-07 stsp FILE *f, off_t fileoffset, off_t filesize,
719 64a8571e 2022-01-07 stsp struct got_delta_table *dt, uint8_t *basedata,
720 64a8571e 2022-01-07 stsp off_t basefile_offset0, off_t basefile_size)
722 64a8571e 2022-01-07 stsp const struct got_error *err = NULL;
723 64a8571e 2022-01-07 stsp const off_t offset0 = fileoffset;
724 64a8571e 2022-01-07 stsp size_t nalloc = 0;
725 64a8571e 2022-01-07 stsp const size_t alloc_chunk_size = 64;
727 64a8571e 2022-01-07 stsp *deltas = NULL;
728 64a8571e 2022-01-07 stsp *ndeltas = 0;
731 64a8571e 2022-01-07 stsp * offset0 indicates where data to be deltified begins.
732 64a8571e 2022-01-07 stsp * For example, we want to avoid deltifying a Git object header at
733 64a8571e 2022-01-07 stsp * the beginning of the file.
735 64a8571e 2022-01-07 stsp if (fseeko(f, offset0, SEEK_SET) == -1)
736 64a8571e 2022-01-07 stsp return got_error_from_errno("fseeko");
738 64a8571e 2022-01-07 stsp *deltas = reallocarray(NULL, alloc_chunk_size,
739 64a8571e 2022-01-07 stsp sizeof(struct got_delta_instruction));
740 64a8571e 2022-01-07 stsp if (*deltas == NULL)
741 64a8571e 2022-01-07 stsp return got_error_from_errno("reallocarray");
742 64a8571e 2022-01-07 stsp nalloc = alloc_chunk_size;
744 64a8571e 2022-01-07 stsp while (fileoffset < filesize) {
745 64a8571e 2022-01-07 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
746 64a8571e 2022-01-07 stsp off_t blocklen;
747 64a8571e 2022-01-07 stsp struct got_delta_block *block;
748 64a8571e 2022-01-07 stsp err = nextblk(buf, &blocklen, f);
751 64a8571e 2022-01-07 stsp if (blocklen == 0) {
752 64a8571e 2022-01-07 stsp /* Source remainder from the file itself. */
753 64a8571e 2022-01-07 stsp if (fileoffset < filesize) {
754 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
755 64a8571e 2022-01-07 stsp alloc_chunk_size, 0, fileoffset - offset0,
756 64a8571e 2022-01-07 stsp filesize - fileoffset);
760 64a8571e 2022-01-07 stsp err = lookupblk_mem(&block, dt, buf, blocklen, basedata,
761 64a8571e 2022-01-07 stsp basefile_offset0);
764 64a8571e 2022-01-07 stsp if (block != NULL) {
766 64a8571e 2022-01-07 stsp * We have found a matching block in the delta base.
767 64a8571e 2022-01-07 stsp * Attempt to stretch the block as far as possible and
768 64a8571e 2022-01-07 stsp * generate a copy instruction.
770 64a8571e 2022-01-07 stsp err = stretchblk_file_mem(basedata, basefile_offset0,
771 64a8571e 2022-01-07 stsp basefile_size, block, f, filesize, &blocklen);
774 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
775 64a8571e 2022-01-07 stsp alloc_chunk_size, 1, block->offset, blocklen);
780 64a8571e 2022-01-07 stsp * No match.
781 64a8571e 2022-01-07 stsp * This block needs to be sourced from the file itself.
783 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
784 64a8571e 2022-01-07 stsp alloc_chunk_size, 0, fileoffset - offset0, blocklen);
788 64a8571e 2022-01-07 stsp fileoffset += blocklen;
789 64a8571e 2022-01-07 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1) {
790 64a8571e 2022-01-07 stsp err = got_error_from_errno("fseeko");
796 64a8571e 2022-01-07 stsp free(*deltas);
797 64a8571e 2022-01-07 stsp *deltas = NULL;
798 64a8571e 2022-01-07 stsp *ndeltas = 0;
800 64a8571e 2022-01-07 stsp return err;
803 64a8571e 2022-01-07 stsp const struct got_error *
804 64a8571e 2022-01-07 stsp got_deltify_mem_file(struct got_delta_instruction **deltas, int *ndeltas,
805 64a8571e 2022-01-07 stsp uint8_t *data, off_t fileoffset, off_t filesize,
806 64a8571e 2022-01-07 stsp struct got_delta_table *dt, FILE *basefile,
807 64a8571e 2022-01-07 stsp off_t basefile_offset0, off_t basefile_size)
809 64a8571e 2022-01-07 stsp const struct got_error *err = NULL;
810 64a8571e 2022-01-07 stsp const off_t offset0 = fileoffset;
811 64a8571e 2022-01-07 stsp size_t nalloc = 0;
812 64a8571e 2022-01-07 stsp const size_t alloc_chunk_size = 64;
814 64a8571e 2022-01-07 stsp *deltas = NULL;
815 64a8571e 2022-01-07 stsp *ndeltas = 0;
817 64a8571e 2022-01-07 stsp *deltas = reallocarray(NULL, alloc_chunk_size,
818 64a8571e 2022-01-07 stsp sizeof(struct got_delta_instruction));
819 64a8571e 2022-01-07 stsp if (*deltas == NULL)
820 64a8571e 2022-01-07 stsp return got_error_from_errno("reallocarray");
821 64a8571e 2022-01-07 stsp nalloc = alloc_chunk_size;
823 64a8571e 2022-01-07 stsp while (fileoffset < filesize) {
824 64a8571e 2022-01-07 stsp off_t blocklen;
825 64a8571e 2022-01-07 stsp struct got_delta_block *block;
826 64a8571e 2022-01-07 stsp err = nextblk_mem(&blocklen, data, fileoffset, filesize);
829 64a8571e 2022-01-07 stsp if (blocklen == 0) {
830 64a8571e 2022-01-07 stsp /* Source remainder from the file itself. */
831 64a8571e 2022-01-07 stsp if (fileoffset < filesize) {
832 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
833 64a8571e 2022-01-07 stsp alloc_chunk_size, 0, fileoffset - offset0,
834 64a8571e 2022-01-07 stsp filesize - fileoffset);
838 64a8571e 2022-01-07 stsp err = lookupblk(&block, dt, data + fileoffset, blocklen,
839 64a8571e 2022-01-07 stsp basefile, basefile_offset0);
842 64a8571e 2022-01-07 stsp if (block != NULL) {
844 64a8571e 2022-01-07 stsp * We have found a matching block in the delta base.
845 64a8571e 2022-01-07 stsp * Attempt to stretch the block as far as possible and
846 64a8571e 2022-01-07 stsp * generate a copy instruction.
848 64a8571e 2022-01-07 stsp err = stretchblk_mem_file(basefile, basefile_offset0,
849 64a8571e 2022-01-07 stsp block, data, fileoffset + blocklen, filesize,
850 64a8571e 2022-01-07 stsp &blocklen);
853 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
854 64a8571e 2022-01-07 stsp alloc_chunk_size, 1, block->offset, blocklen);
859 64a8571e 2022-01-07 stsp * No match.
860 64a8571e 2022-01-07 stsp * This block needs to be sourced from the file itself.
862 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
863 64a8571e 2022-01-07 stsp alloc_chunk_size, 0, fileoffset - offset0, blocklen);
867 64a8571e 2022-01-07 stsp fileoffset += blocklen;
871 64a8571e 2022-01-07 stsp free(*deltas);
872 64a8571e 2022-01-07 stsp *deltas = NULL;
873 64a8571e 2022-01-07 stsp *ndeltas = 0;
875 64a8571e 2022-01-07 stsp return err;
878 64a8571e 2022-01-07 stsp const struct got_error *
879 64a8571e 2022-01-07 stsp got_deltify_mem_mem(struct got_delta_instruction **deltas, int *ndeltas,
880 64a8571e 2022-01-07 stsp uint8_t *data, off_t fileoffset, off_t filesize,
881 64a8571e 2022-01-07 stsp struct got_delta_table *dt, uint8_t *basedata,
882 64a8571e 2022-01-07 stsp off_t basefile_offset0, off_t basefile_size)
884 64a8571e 2022-01-07 stsp const struct got_error *err = NULL;
885 64a8571e 2022-01-07 stsp const off_t offset0 = fileoffset;
886 64a8571e 2022-01-07 stsp size_t nalloc = 0;
887 64a8571e 2022-01-07 stsp const size_t alloc_chunk_size = 64;
889 64a8571e 2022-01-07 stsp *deltas = NULL;
890 64a8571e 2022-01-07 stsp *ndeltas = 0;
892 64a8571e 2022-01-07 stsp *deltas = reallocarray(NULL, alloc_chunk_size,
893 64a8571e 2022-01-07 stsp sizeof(struct got_delta_instruction));
894 64a8571e 2022-01-07 stsp if (*deltas == NULL)
895 64a8571e 2022-01-07 stsp return got_error_from_errno("reallocarray");
896 64a8571e 2022-01-07 stsp nalloc = alloc_chunk_size;
898 64a8571e 2022-01-07 stsp while (fileoffset < filesize) {
899 64a8571e 2022-01-07 stsp off_t blocklen;
900 64a8571e 2022-01-07 stsp struct got_delta_block *block;
901 64a8571e 2022-01-07 stsp err = nextblk_mem(&blocklen, data, fileoffset, filesize);
904 64a8571e 2022-01-07 stsp if (blocklen == 0) {
905 64a8571e 2022-01-07 stsp /* Source remainder from the file itself. */
906 64a8571e 2022-01-07 stsp if (fileoffset < filesize) {
907 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
908 64a8571e 2022-01-07 stsp alloc_chunk_size, 0, fileoffset - offset0,
909 64a8571e 2022-01-07 stsp filesize - fileoffset);
913 64a8571e 2022-01-07 stsp err = lookupblk_mem(&block, dt, data + fileoffset, blocklen,
914 64a8571e 2022-01-07 stsp basedata, basefile_offset0);
917 64a8571e 2022-01-07 stsp if (block != NULL) {
919 64a8571e 2022-01-07 stsp * We have found a matching block in the delta base.
920 64a8571e 2022-01-07 stsp * Attempt to stretch the block as far as possible and
921 64a8571e 2022-01-07 stsp * generate a copy instruction.
923 64a8571e 2022-01-07 stsp err = stretchblk_mem_mem(basedata, basefile_offset0,
924 64a8571e 2022-01-07 stsp basefile_size, block, data, fileoffset + blocklen,
925 64a8571e 2022-01-07 stsp filesize, &blocklen);
928 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
929 64a8571e 2022-01-07 stsp alloc_chunk_size, 1, block->offset, blocklen);
934 64a8571e 2022-01-07 stsp * No match.
935 64a8571e 2022-01-07 stsp * This block needs to be sourced from the file itself.
937 64a8571e 2022-01-07 stsp err = emitdelta(deltas, &nalloc, ndeltas,
938 64a8571e 2022-01-07 stsp alloc_chunk_size, 0, fileoffset - offset0, blocklen);
942 64a8571e 2022-01-07 stsp fileoffset += blocklen;
946 64a8571e 2022-01-07 stsp free(*deltas);
947 64a8571e 2022-01-07 stsp *deltas = NULL;
948 64a8571e 2022-01-07 stsp *ndeltas = 0;
950 64a8571e 2022-01-07 stsp return err;