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>
20 f364801d 2021-03-10 stsp #include <assert.h>
21 69aa0e90 2021-03-10 stsp #include <errno.h>
22 f364801d 2021-03-10 stsp #include <limits.h>
23 f364801d 2021-03-10 stsp #include <stdint.h>
24 f364801d 2021-03-10 stsp #include <stdio.h>
25 f364801d 2021-03-10 stsp #include <stdlib.h>
26 f364801d 2021-03-10 stsp #include <string.h>
28 f364801d 2021-03-10 stsp #include "got_error.h"
30 69aa0e90 2021-03-10 stsp #include "got_lib_deltify.h"
33 69aa0e90 2021-03-10 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
37 2539ccf2 2021-10-08 thomas * The algorihm used here is FastCDC (Fast Content-Defined Chunking)
38 2539ccf2 2021-10-08 thomas * https://www.usenix.org/conference/atc16/technical-sessions/presentation/xia
41 82e16997 2021-10-08 thomas static uint32_t geartab[256] = {
42 69aa0e90 2021-03-10 stsp 0x67ed26b7, 0x32da500c, 0x53d0fee0, 0xce387dc7, 0xcd406d90, 0x2e83a4d4,
43 69aa0e90 2021-03-10 stsp 0x9fc9a38d, 0xb67259dc, 0xca6b1722, 0x6d2ea08c, 0x235cea2e, 0x3149bb5f,
44 69aa0e90 2021-03-10 stsp 0x1beda787, 0x2a6b77d5, 0x2f22d9ac, 0x91fc0544, 0xe413acfa, 0x5a30ff7a,
45 69aa0e90 2021-03-10 stsp 0xad6fdde0, 0x444fd0f5, 0x7ad87864, 0x58c5ff05, 0x8d2ec336, 0x2371f853,
46 69aa0e90 2021-03-10 stsp 0x550f8572, 0x6aa448dd, 0x7c9ddbcf, 0x95221e14, 0x2a82ec33, 0xcbec5a78,
47 69aa0e90 2021-03-10 stsp 0xc6795a0d, 0x243995b7, 0x1c909a2f, 0x4fded51c, 0x635d334b, 0x0e2b9999,
48 69aa0e90 2021-03-10 stsp 0x2702968d, 0x856de1d5, 0x3325d60e, 0xeb6a7502, 0xec2a9844, 0x0905835a,
49 69aa0e90 2021-03-10 stsp 0xa1820375, 0xa4be5cab, 0x96a6c058, 0x2c2ccd70, 0xba40fce3, 0xd794c46b,
50 69aa0e90 2021-03-10 stsp 0x8fbae83e, 0xc3aa7899, 0x3d3ff8ed, 0xa0d42b5b, 0x571c0c97, 0xd2811516,
51 69aa0e90 2021-03-10 stsp 0xf7e7b96c, 0x4fd2fcbd, 0xe2fdec94, 0x282cc436, 0x78e8e95c, 0x80a3b613,
52 69aa0e90 2021-03-10 stsp 0xcfbee20c, 0xd4a32d1c, 0x2a12ff13, 0x6af82936, 0xe5630258, 0x8efa6a98,
53 69aa0e90 2021-03-10 stsp 0x294fb2d1, 0xdeb57086, 0x5f0fddb3, 0xeceda7ce, 0x4c87305f, 0x3a6d3307,
54 69aa0e90 2021-03-10 stsp 0xe22d2942, 0x9d060217, 0x1e42ed02, 0xb6f63b52, 0x4367f39f, 0x055cf262,
55 69aa0e90 2021-03-10 stsp 0x03a461b2, 0x5ef9e382, 0x386bc03a, 0x2a1e79c7, 0xf1a0058b, 0xd4d2dea9,
56 69aa0e90 2021-03-10 stsp 0x56baf37d, 0x5daff6cc, 0xf03a951d, 0xaef7de45, 0xa8f4581e, 0x3960b555,
57 69aa0e90 2021-03-10 stsp 0xffbfff6d, 0xbe702a23, 0x8f5b6d6f, 0x061739fb, 0x98696f47, 0x3fd596d4,
58 69aa0e90 2021-03-10 stsp 0x151eac6b, 0xa9fcc4f5, 0x69181a12, 0x3ac5a107, 0xb5198fe7, 0x96bcb1da,
59 69aa0e90 2021-03-10 stsp 0x1b5ddf8e, 0xc757d650, 0x65865c3a, 0x8fc0a41a, 0x87435536, 0x99eda6f2,
60 69aa0e90 2021-03-10 stsp 0x41874794, 0x29cff4e8, 0xb70efd9a, 0x3103f6e7, 0x84d2453b, 0x15a450bd,
61 69aa0e90 2021-03-10 stsp 0x74f49af1, 0x60f664b1, 0xa1c86935, 0xfdafbce1, 0xe36353e3, 0x5d9ba739,
62 69aa0e90 2021-03-10 stsp 0xbc0559ba, 0x708b0054, 0xd41d808c, 0xb2f31723, 0x9027c41f, 0xf136d165,
63 69aa0e90 2021-03-10 stsp 0xb5374b12, 0x9420a6ac, 0x273958b6, 0xe6c2fad0, 0xebdc1f21, 0xfb33af8b,
64 69aa0e90 2021-03-10 stsp 0xc71c25cd, 0xe9a2d8e5, 0xbeb38a50, 0xbceb7cc2, 0x4e4e73f0, 0xcd6c251d,
65 69aa0e90 2021-03-10 stsp 0xde4c032c, 0x4b04ac30, 0x725b8b21, 0x4eb8c33b, 0x20d07b75, 0x0567aa63,
66 69aa0e90 2021-03-10 stsp 0xb56b2bb7, 0xc1f5fd3a, 0xcafd35ca, 0x470dd4da, 0xfe4f94cd, 0xfb8de424,
67 69aa0e90 2021-03-10 stsp 0xe8dbcf40, 0xfe50a37a, 0x62db5b5d, 0xf32f4ab6, 0x2c4a8a51, 0x18473dc0,
68 69aa0e90 2021-03-10 stsp 0xfe0cbb6e, 0xfe399efd, 0xdf34ecc9, 0x6ccd5055, 0x46097073, 0x139135c2,
69 69aa0e90 2021-03-10 stsp 0x721c76f6, 0x1c6a94b4, 0x6eee014d, 0x8a508e02, 0x3da538f5, 0x280d394f,
70 69aa0e90 2021-03-10 stsp 0x5248a0c4, 0x3ce94c6c, 0x9a71ad3a, 0x8493dd05, 0xe43f0ab6, 0x18e4ed42,
71 69aa0e90 2021-03-10 stsp 0x6c5c0e09, 0x42b06ec9, 0x8d330343, 0xa45b6f59, 0x2a573c0c, 0xd7fd3de6,
72 69aa0e90 2021-03-10 stsp 0xeedeab68, 0x5c84dafc, 0xbbd1b1a8, 0xa3ce1ad1, 0x85b70bed, 0xb6add07f,
73 69aa0e90 2021-03-10 stsp 0xa531309c, 0x8f8ab852, 0x564de332, 0xeac9ed0c, 0x73da402c, 0x3ec52761,
74 69aa0e90 2021-03-10 stsp 0x43af2f4d, 0xd6ff45c8, 0x4c367462, 0xd553bd6a, 0x44724855, 0x3b2aa728,
75 69aa0e90 2021-03-10 stsp 0x56e5eb65, 0xeaf16173, 0x33fa42ff, 0xd714bb5d, 0xfbd0a3b9, 0xaf517134,
76 69aa0e90 2021-03-10 stsp 0x9416c8cd, 0x534cf94f, 0x548947c2, 0x34193569, 0x32f4389a, 0xfe7028bc,
77 69aa0e90 2021-03-10 stsp 0xed73b1ed, 0x9db95770, 0x468e3922, 0x0440c3cd, 0x60059a62, 0x33504562,
78 69aa0e90 2021-03-10 stsp 0x2b229fbd, 0x5174dca5, 0xf7028752, 0xd63c6aa8, 0x31276f38, 0x0646721c,
79 69aa0e90 2021-03-10 stsp 0xb0191da8, 0xe00e6de0, 0x9eac1a6e, 0x9f7628a5, 0xed6c06ea, 0x0bb8af15,
80 69aa0e90 2021-03-10 stsp 0xf119fb12, 0x38693c1c, 0x732bc0fe, 0x84953275, 0xb82ec888, 0x33a4f1b3,
81 69aa0e90 2021-03-10 stsp 0x3099835e, 0x028a8782, 0x5fdd51d7, 0xc6c717b3, 0xb06caf71, 0x17c8c111,
82 69aa0e90 2021-03-10 stsp 0x61bad754, 0x9fd03061, 0xe09df1af, 0x3bc9eb73, 0x85878413, 0x9889aaf2,
83 69aa0e90 2021-03-10 stsp 0x3f5a9e46, 0x42c9f01f, 0x9984a4f4, 0xd5de43cc, 0xd294daed, 0xbecba2d2,
84 69aa0e90 2021-03-10 stsp 0xf1f6e72c, 0x5551128a, 0x83af87e2, 0x6f0342ba,
87 f364801d 2021-03-10 stsp static uint64_t
88 69aa0e90 2021-03-10 stsp hashblk(const unsigned char *p, off_t n)
90 f364801d 2021-03-10 stsp unsigned char buf[SHA1_DIGEST_LENGTH];
92 69aa0e90 2021-03-10 stsp SHA1_CTX ctx;
93 69aa0e90 2021-03-10 stsp SHA1Init(&ctx);
94 69aa0e90 2021-03-10 stsp SHA1Update(&ctx, p, n);
95 69aa0e90 2021-03-10 stsp SHA1Final(buf, &ctx);
96 f364801d 2021-03-10 stsp memcpy(&h, buf, sizeof(h));
97 f364801d 2021-03-10 stsp return be64toh(h);
100 f364801d 2021-03-10 stsp static const struct got_error *
101 740bba1c 2021-06-18 stsp addblk(struct got_delta_table *dt, FILE *f, off_t file_offset0, off_t len,
102 740bba1c 2021-06-18 stsp off_t offset, uint64_t h)
104 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
106 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
107 0d15f6dc 2021-06-13 stsp uint8_t buf2[GOT_DELTIFY_MAXCHUNK];
108 69aa0e90 2021-03-10 stsp size_t r = 0;
110 69aa0e90 2021-03-10 stsp if (len == 0)
111 69aa0e90 2021-03-10 stsp return NULL;
113 69aa0e90 2021-03-10 stsp i = h % dt->nalloc;
114 69aa0e90 2021-03-10 stsp while (dt->blocks[i].len != 0) {
116 69aa0e90 2021-03-10 stsp * Avoid adding duplicate blocks.
117 69aa0e90 2021-03-10 stsp * NB: A matching hash is insufficient for detecting equality.
118 51a494da 2021-06-11 stsp * The hash can only detect inequality.
120 51a494da 2021-06-11 stsp if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
121 0d15f6dc 2021-06-13 stsp if (r == 0) {
122 740bba1c 2021-06-18 stsp if (fseeko(f, file_offset0 + offset, SEEK_SET) == -1)
123 0d15f6dc 2021-06-13 stsp return got_error_from_errno("fseeko");
124 0d15f6dc 2021-06-13 stsp r = fread(buf, 1, len, f);
125 0d15f6dc 2021-06-13 stsp if (r != len) {
126 0d15f6dc 2021-06-13 stsp if (!ferror(f))
127 0d15f6dc 2021-06-13 stsp return NULL;
128 0d15f6dc 2021-06-13 stsp return got_ferror(f, GOT_ERR_IO);
131 740bba1c 2021-06-18 stsp if (fseeko(f, file_offset0 + dt->blocks[i].offset,
132 740bba1c 2021-06-18 stsp SEEK_SET) == -1)
133 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
134 69aa0e90 2021-03-10 stsp if (fread(buf2, 1, len, f) != len)
135 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
136 69aa0e90 2021-03-10 stsp if (memcmp(buf, buf2, len) == 0)
137 69aa0e90 2021-03-10 stsp return NULL;
140 69aa0e90 2021-03-10 stsp i = (i + 1) % dt->nalloc;
142 69aa0e90 2021-03-10 stsp assert(dt->blocks[i].len == 0);
143 69aa0e90 2021-03-10 stsp dt->blocks[i].len = len;
144 69aa0e90 2021-03-10 stsp dt->blocks[i].offset = offset;
145 69aa0e90 2021-03-10 stsp dt->blocks[i].hash = h;
146 69aa0e90 2021-03-10 stsp dt->nblocks++;
147 e89540a9 2021-06-11 stsp if (dt->nalloc < dt->nblocks + 64) {
148 69aa0e90 2021-03-10 stsp struct got_delta_block *db;
149 e89540a9 2021-06-11 stsp size_t old_size = dt->nalloc;
150 69aa0e90 2021-03-10 stsp db = dt->blocks;
151 e89540a9 2021-06-11 stsp dt->blocks = calloc(dt->nalloc + 64,
152 e89540a9 2021-06-11 stsp sizeof(struct got_delta_block));
153 69aa0e90 2021-03-10 stsp if (dt->blocks == NULL) {
154 69aa0e90 2021-03-10 stsp err = got_error_from_errno("calloc");
155 69aa0e90 2021-03-10 stsp dt->blocks = db;
156 69aa0e90 2021-03-10 stsp return err;
158 e89540a9 2021-06-11 stsp dt->nalloc += 64;
160 69aa0e90 2021-03-10 stsp * Recompute all block positions. Hash-based indices of blocks
161 69aa0e90 2021-03-10 stsp * in the array depend on the allocated length of the array.
163 69aa0e90 2021-03-10 stsp dt->nblocks = 0;
164 a893025f 2021-06-11 stsp for (i = 0; i < old_size; i++) {
165 69aa0e90 2021-03-10 stsp if (db[i].len == 0)
167 740bba1c 2021-06-18 stsp err = addblk(dt, f, file_offset0, db[i].len,
168 740bba1c 2021-06-18 stsp db[i].offset, db[i].hash);
175 69aa0e90 2021-03-10 stsp return err;
178 69aa0e90 2021-03-10 stsp static const struct got_error *
179 69aa0e90 2021-03-10 stsp lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
180 740bba1c 2021-06-18 stsp unsigned char *p, off_t len, FILE *basefile, off_t basefile_offset0)
183 f364801d 2021-03-10 stsp uint64_t h;
184 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
187 69aa0e90 2021-03-10 stsp *block = NULL;
189 69aa0e90 2021-03-10 stsp h = hashblk(p, len);
190 69aa0e90 2021-03-10 stsp for (i = h % dt->nalloc; dt->blocks[i].len != 0;
191 69aa0e90 2021-03-10 stsp i = (i + 1) % dt->nalloc) {
192 69aa0e90 2021-03-10 stsp if (dt->blocks[i].hash != h ||
193 69aa0e90 2021-03-10 stsp dt->blocks[i].len != len)
195 740bba1c 2021-06-18 stsp if (fseeko(basefile, basefile_offset0 + dt->blocks[i].offset,
196 740bba1c 2021-06-18 stsp SEEK_SET) == -1)
197 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
198 69aa0e90 2021-03-10 stsp r = fread(buf, 1, len, basefile);
199 69aa0e90 2021-03-10 stsp if (r != len)
200 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
201 69aa0e90 2021-03-10 stsp if (memcmp(p, buf, len) == 0) {
202 69aa0e90 2021-03-10 stsp *block = &dt->blocks[i];
206 f364801d 2021-03-10 stsp return NULL;
209 69aa0e90 2021-03-10 stsp static const struct got_error *
210 69aa0e90 2021-03-10 stsp nextblk(uint8_t *buf, off_t *blocklen, FILE *f)
212 f364801d 2021-03-10 stsp uint32_t gh;
213 69aa0e90 2021-03-10 stsp const unsigned char *p;
215 69aa0e90 2021-03-10 stsp off_t pos = ftello(f);
217 69aa0e90 2021-03-10 stsp *blocklen = 0;
219 69aa0e90 2021-03-10 stsp r = fread(buf, 1, GOT_DELTIFY_MAXCHUNK, f);
220 69aa0e90 2021-03-10 stsp if (r == 0 && ferror(f))
221 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
222 69aa0e90 2021-03-10 stsp if (r < GOT_DELTIFY_MINCHUNK)
223 69aa0e90 2021-03-10 stsp return NULL; /* no more delta-worthy blocks left */
225 69aa0e90 2021-03-10 stsp /* Got a deltifiable block. Find the split-point where it ends. */
226 69aa0e90 2021-03-10 stsp p = buf + GOT_DELTIFY_MINCHUNK;
228 69aa0e90 2021-03-10 stsp while (p != buf + r) {
229 f364801d 2021-03-10 stsp gh = (gh << 1) + geartab[*p++];
230 69aa0e90 2021-03-10 stsp if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
234 69aa0e90 2021-03-10 stsp *blocklen = (p - buf);
235 69aa0e90 2021-03-10 stsp if (fseeko(f, pos + *blocklen, SEEK_SET) == -1)
236 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
238 69aa0e90 2021-03-10 stsp return NULL;
241 f364801d 2021-03-10 stsp const struct got_error *
242 69aa0e90 2021-03-10 stsp got_deltify_init(struct got_delta_table **dt, FILE *f, off_t fileoffset,
243 69aa0e90 2021-03-10 stsp off_t filesize)
245 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
246 f364801d 2021-03-10 stsp uint64_t h;
247 740bba1c 2021-06-18 stsp const off_t offset0 = fileoffset;
249 69aa0e90 2021-03-10 stsp *dt = calloc(1, sizeof(**dt));
250 69aa0e90 2021-03-10 stsp if (*dt == NULL)
251 f364801d 2021-03-10 stsp return got_error_from_errno("calloc");
253 69aa0e90 2021-03-10 stsp (*dt)->nblocks = 0;
254 69aa0e90 2021-03-10 stsp (*dt)->nalloc = 128;
255 69aa0e90 2021-03-10 stsp (*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
256 69aa0e90 2021-03-10 stsp if ((*dt)->blocks == NULL) {
257 69aa0e90 2021-03-10 stsp err = got_error_from_errno("calloc");
261 69aa0e90 2021-03-10 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
262 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
264 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
265 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
266 69aa0e90 2021-03-10 stsp off_t blocklen;
267 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
270 69aa0e90 2021-03-10 stsp if (blocklen == 0)
272 69aa0e90 2021-03-10 stsp h = hashblk(buf, blocklen);
273 740bba1c 2021-06-18 stsp err = addblk(*dt, f, offset0, blocklen,
274 740bba1c 2021-06-18 stsp fileoffset - offset0, h);
277 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
278 68bdcdc2 2021-06-11 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1)
279 68bdcdc2 2021-06-11 stsp return got_error_from_errno("fseeko");
283 69aa0e90 2021-03-10 stsp free((*dt)->blocks);
285 69aa0e90 2021-03-10 stsp *dt = NULL;
288 69aa0e90 2021-03-10 stsp return err;
292 69aa0e90 2021-03-10 stsp got_deltify_free(struct got_delta_table *dt)
294 dbbf4a5f 2021-05-20 stsp if (dt == NULL)
296 69aa0e90 2021-03-10 stsp free(dt->blocks);
300 f364801d 2021-03-10 stsp static const struct got_error *
301 dd29967c 2021-08-22 stsp emitdelta(struct got_delta_instruction **deltas, size_t *nalloc, int *ndeltas,
302 dd29967c 2021-08-22 stsp const size_t alloc_chunk_size, int copy, off_t offset, off_t len)
304 69aa0e90 2021-03-10 stsp struct got_delta_instruction *d, *p;
306 dd29967c 2021-08-22 stsp if (*nalloc < *ndeltas + alloc_chunk_size) {
307 dd29967c 2021-08-22 stsp p = reallocarray(*deltas, *nalloc + alloc_chunk_size,
308 dd29967c 2021-08-22 stsp sizeof(struct got_delta_instruction));
309 dd29967c 2021-08-22 stsp if (p == NULL)
310 dd29967c 2021-08-22 stsp return got_error_from_errno("realloc");
311 dd29967c 2021-08-22 stsp *deltas = p;
312 dd29967c 2021-08-22 stsp *nalloc += alloc_chunk_size;
314 69aa0e90 2021-03-10 stsp *ndeltas += 1;
315 69aa0e90 2021-03-10 stsp d = &(*deltas)[*ndeltas - 1];
316 69aa0e90 2021-03-10 stsp d->copy = copy;
317 69aa0e90 2021-03-10 stsp d->offset = offset;
318 56e7a15f 2021-03-10 stsp d->len = len;
319 f364801d 2021-03-10 stsp return NULL;
322 69aa0e90 2021-03-10 stsp static const struct got_error *
323 740bba1c 2021-06-18 stsp stretchblk(FILE *basefile, off_t base_offset0, struct got_delta_block *block,
324 740bba1c 2021-06-18 stsp FILE *f, off_t filesize, off_t *blocklen)
326 69aa0e90 2021-03-10 stsp uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
327 69aa0e90 2021-03-10 stsp size_t base_r, r, i;
328 69aa0e90 2021-03-10 stsp int buf_equal = 1;
330 5de743f8 2021-08-29 stsp if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
331 5de743f8 2021-08-29 stsp SEEK_SET) == -1)
332 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
334 69aa0e90 2021-03-10 stsp while (buf_equal && *blocklen < (1 << 24) - 1) {
335 69aa0e90 2021-03-10 stsp base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
336 69aa0e90 2021-03-10 stsp if (base_r == 0) {
337 69aa0e90 2021-03-10 stsp if (ferror(basefile))
338 69aa0e90 2021-03-10 stsp return got_ferror(basefile, GOT_ERR_IO);
341 69aa0e90 2021-03-10 stsp r = fread(buf, 1, sizeof(buf), f);
342 69aa0e90 2021-03-10 stsp if (r == 0) {
343 69aa0e90 2021-03-10 stsp if (ferror(f))
344 69aa0e90 2021-03-10 stsp return got_ferror(f, GOT_ERR_IO);
347 69aa0e90 2021-03-10 stsp for (i = 0; i < MIN(base_r, r); i++) {
348 69aa0e90 2021-03-10 stsp if (buf[i] != basebuf[i]) {
349 69aa0e90 2021-03-10 stsp buf_equal = 0;
352 69aa0e90 2021-03-10 stsp (*blocklen)++;
356 69aa0e90 2021-03-10 stsp return NULL;
359 69aa0e90 2021-03-10 stsp const struct got_error *
360 69aa0e90 2021-03-10 stsp got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
361 69aa0e90 2021-03-10 stsp FILE *f, off_t fileoffset, off_t filesize,
362 69aa0e90 2021-03-10 stsp struct got_delta_table *dt, FILE *basefile,
363 f34b169e 2021-06-18 stsp off_t basefile_offset0, off_t basefile_size)
365 69aa0e90 2021-03-10 stsp const struct got_error *err = NULL;
366 69aa0e90 2021-03-10 stsp const off_t offset0 = fileoffset;
367 dd29967c 2021-08-22 stsp size_t nalloc = 0;
368 dd29967c 2021-08-22 stsp const size_t alloc_chunk_size = 64;
370 69aa0e90 2021-03-10 stsp *deltas = NULL;
371 69aa0e90 2021-03-10 stsp *ndeltas = 0;
374 69aa0e90 2021-03-10 stsp * offset0 indicates where data to be deltified begins.
375 69aa0e90 2021-03-10 stsp * For example, we want to avoid deltifying a Git object header at
376 69aa0e90 2021-03-10 stsp * the beginning of the file.
378 69aa0e90 2021-03-10 stsp if (fseeko(f, offset0, SEEK_SET) == -1)
379 69aa0e90 2021-03-10 stsp return got_error_from_errno("fseeko");
381 dd29967c 2021-08-22 stsp *deltas = reallocarray(NULL, alloc_chunk_size,
382 dd29967c 2021-08-22 stsp sizeof(struct got_delta_instruction));
383 dd29967c 2021-08-22 stsp if (*deltas == NULL)
384 dd29967c 2021-08-22 stsp return got_error_from_errno("reallocarray");
385 dd29967c 2021-08-22 stsp nalloc = alloc_chunk_size;
387 69aa0e90 2021-03-10 stsp while (fileoffset < filesize) {
388 69aa0e90 2021-03-10 stsp uint8_t buf[GOT_DELTIFY_MAXCHUNK];
389 69aa0e90 2021-03-10 stsp off_t blocklen;
390 69aa0e90 2021-03-10 stsp struct got_delta_block *block;
391 69aa0e90 2021-03-10 stsp err = nextblk(buf, &blocklen, f);
394 9a8dc2b3 2021-06-18 stsp if (blocklen == 0) {
395 9a8dc2b3 2021-06-18 stsp /* Source remainder from the file itself. */
396 9a8dc2b3 2021-06-18 stsp if (fileoffset < filesize) {
397 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
398 dd29967c 2021-08-22 stsp alloc_chunk_size, 0, fileoffset - offset0,
399 9a8dc2b3 2021-06-18 stsp filesize - fileoffset);
403 740bba1c 2021-06-18 stsp err = lookupblk(&block, dt, buf, blocklen, basefile,
404 740bba1c 2021-06-18 stsp basefile_offset0);
407 69aa0e90 2021-03-10 stsp if (block != NULL) {
409 69aa0e90 2021-03-10 stsp * We have found a matching block in the delta base.
410 69aa0e90 2021-03-10 stsp * Attempt to stretch the block as far as possible and
411 69aa0e90 2021-03-10 stsp * generate a copy instruction.
413 740bba1c 2021-06-18 stsp err = stretchblk(basefile, basefile_offset0, block,
414 740bba1c 2021-06-18 stsp f, filesize, &blocklen);
417 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
418 dd29967c 2021-08-22 stsp alloc_chunk_size, 1, block->offset, blocklen);
423 69aa0e90 2021-03-10 stsp * No match.
424 69aa0e90 2021-03-10 stsp * This block needs to be sourced from the file itself.
426 dd29967c 2021-08-22 stsp err = emitdelta(deltas, &nalloc, ndeltas,
427 dd29967c 2021-08-22 stsp alloc_chunk_size, 0, fileoffset - offset0, blocklen);
431 69aa0e90 2021-03-10 stsp fileoffset += blocklen;
432 0af64e86 2021-08-22 stsp if (fseeko(f, fileoffset, SEEK_SET) == -1) {
433 0af64e86 2021-08-22 stsp err = got_error_from_errno("fseeko");
439 69aa0e90 2021-03-10 stsp free(*deltas);
440 69aa0e90 2021-03-10 stsp *deltas = NULL;
441 69aa0e90 2021-03-10 stsp *ndeltas = 0;
443 69aa0e90 2021-03-10 stsp return err;
