1 /*
2  * Copyright (c) 2020 Ori Bernstein
3  * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
4  *
5  * Permission to use, copy, modify, and distribute this software for any
6  * purpose with or without fee is hereby granted, provided that the above
7  * copyright notice and this permission notice appear in all copies.
8  *
9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16  */
17 
18 #include <sys/types.h>
19 
20 #include <assert.h>
21 #include <errno.h>
22 #include <limits.h>
23 #include <stdint.h>
24 #include <stdio.h>
25 #include <stdlib.h>
26 #include <string.h>
27 
28 #include "got_error.h"
29 
30 #include "got_lib_deltify.h"
31 
32 #ifndef MIN
33 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
34 #endif
35 
36 /*
37  * The algorihm used here is FastCDC (Fast Content-Defined Chunking)
38  * https://www.usenix.org/conference/atc16/technical-sessions/presentation/xia
39  */
40 
41 static uint32_t geartab[256] = {
42     0x67ed26b7, 0x32da500c, 0x53d0fee0, 0xce387dc7, 0xcd406d90, 0x2e83a4d4,
43     0x9fc9a38d, 0xb67259dc, 0xca6b1722, 0x6d2ea08c, 0x235cea2e, 0x3149bb5f,
44     0x1beda787, 0x2a6b77d5, 0x2f22d9ac, 0x91fc0544, 0xe413acfa, 0x5a30ff7a,
45     0xad6fdde0, 0x444fd0f5, 0x7ad87864, 0x58c5ff05, 0x8d2ec336, 0x2371f853,
46     0x550f8572, 0x6aa448dd, 0x7c9ddbcf, 0x95221e14, 0x2a82ec33, 0xcbec5a78,
47     0xc6795a0d, 0x243995b7, 0x1c909a2f, 0x4fded51c, 0x635d334b, 0x0e2b9999,
48     0x2702968d, 0x856de1d5, 0x3325d60e, 0xeb6a7502, 0xec2a9844, 0x0905835a,
49     0xa1820375, 0xa4be5cab, 0x96a6c058, 0x2c2ccd70, 0xba40fce3, 0xd794c46b,
50     0x8fbae83e, 0xc3aa7899, 0x3d3ff8ed, 0xa0d42b5b, 0x571c0c97, 0xd2811516,
51     0xf7e7b96c, 0x4fd2fcbd, 0xe2fdec94, 0x282cc436, 0x78e8e95c, 0x80a3b613,
52     0xcfbee20c, 0xd4a32d1c, 0x2a12ff13, 0x6af82936, 0xe5630258, 0x8efa6a98,
53     0x294fb2d1, 0xdeb57086, 0x5f0fddb3, 0xeceda7ce, 0x4c87305f, 0x3a6d3307,
54     0xe22d2942, 0x9d060217, 0x1e42ed02, 0xb6f63b52, 0x4367f39f, 0x055cf262,
55     0x03a461b2, 0x5ef9e382, 0x386bc03a, 0x2a1e79c7, 0xf1a0058b, 0xd4d2dea9,
56     0x56baf37d, 0x5daff6cc, 0xf03a951d, 0xaef7de45, 0xa8f4581e, 0x3960b555,
57     0xffbfff6d, 0xbe702a23, 0x8f5b6d6f, 0x061739fb, 0x98696f47, 0x3fd596d4,
58     0x151eac6b, 0xa9fcc4f5, 0x69181a12, 0x3ac5a107, 0xb5198fe7, 0x96bcb1da,
59     0x1b5ddf8e, 0xc757d650, 0x65865c3a, 0x8fc0a41a, 0x87435536, 0x99eda6f2,
60     0x41874794, 0x29cff4e8, 0xb70efd9a, 0x3103f6e7, 0x84d2453b, 0x15a450bd,
61     0x74f49af1, 0x60f664b1, 0xa1c86935, 0xfdafbce1, 0xe36353e3, 0x5d9ba739,
62     0xbc0559ba, 0x708b0054, 0xd41d808c, 0xb2f31723, 0x9027c41f, 0xf136d165,
63     0xb5374b12, 0x9420a6ac, 0x273958b6, 0xe6c2fad0, 0xebdc1f21, 0xfb33af8b,
64     0xc71c25cd, 0xe9a2d8e5, 0xbeb38a50, 0xbceb7cc2, 0x4e4e73f0, 0xcd6c251d,
65     0xde4c032c, 0x4b04ac30, 0x725b8b21, 0x4eb8c33b, 0x20d07b75, 0x0567aa63,
66     0xb56b2bb7, 0xc1f5fd3a, 0xcafd35ca, 0x470dd4da, 0xfe4f94cd, 0xfb8de424,
67     0xe8dbcf40, 0xfe50a37a, 0x62db5b5d, 0xf32f4ab6, 0x2c4a8a51, 0x18473dc0,
68     0xfe0cbb6e, 0xfe399efd, 0xdf34ecc9, 0x6ccd5055, 0x46097073, 0x139135c2,
69     0x721c76f6, 0x1c6a94b4, 0x6eee014d, 0x8a508e02, 0x3da538f5, 0x280d394f,
70     0x5248a0c4, 0x3ce94c6c, 0x9a71ad3a, 0x8493dd05, 0xe43f0ab6, 0x18e4ed42,
71     0x6c5c0e09, 0x42b06ec9, 0x8d330343, 0xa45b6f59, 0x2a573c0c, 0xd7fd3de6,
72     0xeedeab68, 0x5c84dafc, 0xbbd1b1a8, 0xa3ce1ad1, 0x85b70bed, 0xb6add07f,
73     0xa531309c, 0x8f8ab852, 0x564de332, 0xeac9ed0c, 0x73da402c, 0x3ec52761,
74     0x43af2f4d, 0xd6ff45c8, 0x4c367462, 0xd553bd6a, 0x44724855, 0x3b2aa728,
75     0x56e5eb65, 0xeaf16173, 0x33fa42ff, 0xd714bb5d, 0xfbd0a3b9, 0xaf517134,
76     0x9416c8cd, 0x534cf94f, 0x548947c2, 0x34193569, 0x32f4389a, 0xfe7028bc,
77     0xed73b1ed, 0x9db95770, 0x468e3922, 0x0440c3cd, 0x60059a62, 0x33504562,
78     0x2b229fbd, 0x5174dca5, 0xf7028752, 0xd63c6aa8, 0x31276f38, 0x0646721c,
79     0xb0191da8, 0xe00e6de0, 0x9eac1a6e, 0x9f7628a5, 0xed6c06ea, 0x0bb8af15,
80     0xf119fb12, 0x38693c1c, 0x732bc0fe, 0x84953275, 0xb82ec888, 0x33a4f1b3,
81     0x3099835e, 0x028a8782, 0x5fdd51d7, 0xc6c717b3, 0xb06caf71, 0x17c8c111,
82     0x61bad754, 0x9fd03061, 0xe09df1af, 0x3bc9eb73, 0x85878413, 0x9889aaf2,
83     0x3f5a9e46, 0x42c9f01f, 0x9984a4f4, 0xd5de43cc, 0xd294daed, 0xbecba2d2,
84     0xf1f6e72c, 0x5551128a, 0x83af87e2, 0x6f0342ba,
85 };
86 
87 static uint64_t
88 hashblk(const unsigned char *p, off_t n)
89 {
90 	unsigned char buf[SHA1_DIGEST_LENGTH];
91 	uint64_t h;
92 	SHA1_CTX ctx;
93 	SHA1Init(&ctx);
94 	SHA1Update(&ctx, p, n);
95 	SHA1Final(buf, &ctx);
96 	memcpy(&h, buf, sizeof(h));
97 	return be64toh(h);
98 }
99 
100 static const struct got_error *
101 addblk(struct got_delta_table *dt, FILE *f, off_t file_offset0, off_t len,
102     off_t offset, uint64_t h)
103 {
104 	const struct got_error *err = NULL;
105 	int i;
106 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
107 	uint8_t buf2[GOT_DELTIFY_MAXCHUNK];
108 	size_t r = 0;
109 
110 	if (len == 0)
111 		return NULL;
112 
113 	i = h % dt->nalloc;
114 	while (dt->blocks[i].len != 0) {
115 		/*
116 		 * Avoid adding duplicate blocks.
117 		 * NB: A matching hash is insufficient for detecting equality.
118 		 * The hash can only detect inequality.
119 		 */
120 		if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
121 			if (r == 0) {
122 				if (fseeko(f, file_offset0 + offset, SEEK_SET) == -1)
123 					return got_error_from_errno("fseeko");
124 				r = fread(buf, 1, len, f);
125 				if (r != len) {
126 					if (!ferror(f))
127 						return NULL;
128 					return got_ferror(f, GOT_ERR_IO);
129 				}
130 			}
131 			if (fseeko(f, file_offset0 + dt->blocks[i].offset,
132 			    SEEK_SET) == -1)
133 				return got_error_from_errno("fseeko");
134 			if (fread(buf2, 1, len, f) != len)
135 				return got_ferror(f, GOT_ERR_IO);
136 			if (memcmp(buf, buf2, len) == 0)
137 				return NULL;
138 		}
139 
140 		i = (i + 1) % dt->nalloc;
141 	}
142 	assert(dt->blocks[i].len == 0);
143 	dt->blocks[i].len = len;
144 	dt->blocks[i].offset = offset;
145 	dt->blocks[i].hash = h;
146 	dt->nblocks++;
147 	if (dt->nalloc < dt->nblocks + 64) {
148 		struct got_delta_block *db;
149 		size_t old_size = dt->nalloc;
150 		db = dt->blocks;
151 		dt->blocks = calloc(dt->nalloc + 64,
152 		    sizeof(struct got_delta_block));
153 		if (dt->blocks == NULL) {
154 			err = got_error_from_errno("calloc");
155 			dt->blocks = db;
156 			return err;
157 		}
158 		dt->nalloc += 64;
159 		/*
160 		 * Recompute all block positions. Hash-based indices of blocks
161 		 * in the array depend on the allocated length of the array.
162 		 */
163 		dt->nblocks = 0;
164 		for (i = 0; i < old_size; i++) {
165 			if (db[i].len == 0)
166 				continue;
167 			err = addblk(dt, f, file_offset0, db[i].len,
168 			    db[i].offset, db[i].hash);
169 			if (err)
170 				break;
171 		}
172 		free(db);
173 	}
174 
175 	return err;
176 }
177 
178 static const struct got_error *
179 addblk_mem(struct got_delta_table *dt, uint8_t *data, off_t file_offset0,
180     off_t len, off_t offset, uint64_t h)
181 {
182 	const struct got_error *err = NULL;
183 	int i;
184 	uint8_t *block1;
185 	uint8_t *block2;
186 
187 	if (len == 0)
188 		return NULL;
189 
190 	i = h % dt->nalloc;
191 	while (dt->blocks[i].len != 0) {
192 		/*
193 		 * Avoid adding duplicate blocks.
194 		 * NB: A matching hash is insufficient for detecting equality.
195 		 * The hash can only detect inequality.
196 		 */
197 		if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
198 			block1 = data + file_offset0 + dt->blocks[i].offset;
199 			block2 = data + file_offset0 + offset;
200 			if (memcmp(block1, block2, len) == 0)
201 				return NULL;
202 		}
203 
204 		i = (i + 1) % dt->nalloc;
205 	}
206 	assert(dt->blocks[i].len == 0);
207 	dt->blocks[i].len = len;
208 	dt->blocks[i].offset = offset;
209 	dt->blocks[i].hash = h;
210 	dt->nblocks++;
211 	if (dt->nalloc < dt->nblocks + 64) {
212 		struct got_delta_block *db;
213 		size_t old_size = dt->nalloc;
214 		db = dt->blocks;
215 		dt->blocks = calloc(dt->nalloc + 64,
216 		    sizeof(struct got_delta_block));
217 		if (dt->blocks == NULL) {
218 			err = got_error_from_errno("calloc");
219 			dt->blocks = db;
220 			return err;
221 		}
222 		dt->nalloc += 64;
223 		/*
224 		 * Recompute all block positions. Hash-based indices of blocks
225 		 * in the array depend on the allocated length of the array.
226 		 */
227 		dt->nblocks = 0;
228 		for (i = 0; i < old_size; i++) {
229 			if (db[i].len == 0)
230 				continue;
231 			err = addblk_mem(dt, data, file_offset0, db[i].len,
232 			    db[i].offset, db[i].hash);
233 			if (err)
234 				break;
235 		}
236 		free(db);
237 	}
238 
239 	return err;
240 }
241 
242 static const struct got_error *
243 lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
244     unsigned char *p, off_t len, FILE *basefile, off_t basefile_offset0)
245 {
246 	int i;
247 	uint64_t h;
248 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
249 	size_t r;
250 
251 	*block = NULL;
252 
253 	h = hashblk(p, len);
254 	for (i = h % dt->nalloc; dt->blocks[i].len != 0;
255 	     i = (i + 1) % dt->nalloc) {
256 		if (dt->blocks[i].hash != h ||
257 		    dt->blocks[i].len != len)
258 			continue;
259 		if (fseeko(basefile, basefile_offset0 + dt->blocks[i].offset,
260 		    SEEK_SET) == -1)
261 			return got_error_from_errno("fseeko");
262 		r = fread(buf, 1, len, basefile);
263 		if (r != len)
264 			return got_ferror(basefile, GOT_ERR_IO);
265 		if (memcmp(p, buf, len) == 0) {
266 			*block = &dt->blocks[i];
267 			break;
268 		}
269 	}
270 	return NULL;
271 }
272 
273 static const struct got_error *
274 lookupblk_mem(struct got_delta_block **block, struct got_delta_table *dt,
275     unsigned char *p, off_t len, uint8_t *basedata, off_t basefile_offset0)
276 {
277 	int i;
278 	uint64_t h;
279 	uint8_t *b;
280 
281 	*block = NULL;
282 
283 	h = hashblk(p, len);
284 	for (i = h % dt->nalloc; dt->blocks[i].len != 0;
285 	     i = (i + 1) % dt->nalloc) {
286 		if (dt->blocks[i].hash != h ||
287 		    dt->blocks[i].len != len)
288 			continue;
289 		b = basedata + basefile_offset0 + dt->blocks[i].offset;
290 		if (memcmp(p, b, len) == 0) {
291 			*block = &dt->blocks[i];
292 			break;
293 		}
294 	}
295 	return NULL;
296 }
297 
298 static const struct got_error *
299 nextblk(uint8_t *buf, off_t *blocklen, FILE *f)
300 {
301 	uint32_t gh;
302 	const unsigned char *p;
303 	size_t r;
304 	off_t pos = ftello(f);
305 
306 	*blocklen = 0;
307 
308 	r = fread(buf, 1, GOT_DELTIFY_MAXCHUNK, f);
309 	if (r == 0 && ferror(f))
310 		return got_ferror(f, GOT_ERR_IO);
311 	if (r < GOT_DELTIFY_MINCHUNK)
312 		return NULL; /* no more delta-worthy blocks left */
313 
314 	/* Got a deltifiable block. Find the split-point where it ends. */
315 	p = buf + GOT_DELTIFY_MINCHUNK;
316 	gh = 0;
317 	while (p != buf + r) {
318 		gh = (gh << 1) + geartab[*p++];
319 		if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
320 			break;
321 	}
322 
323 	*blocklen = (p - buf);
324 	if (fseeko(f, pos + *blocklen, SEEK_SET) == -1)
325 		return got_error_from_errno("fseeko");
326 
327 	return NULL;
328 }
329 
330 static const struct got_error *
331 nextblk_mem(off_t *blocklen, uint8_t *data, off_t fileoffset, off_t filesize)
332 {
333 	uint32_t gh;
334 	const unsigned char *p;
335 
336 	*blocklen = 0;
337 
338 	if (fileoffset >= filesize ||
339 	    filesize - fileoffset < GOT_DELTIFY_MINCHUNK)
340 		return NULL; /* no more delta-worthy blocks left */
341 
342 	/* Got a deltifiable block. Find the split-point where it ends. */
343 	p = data + fileoffset + GOT_DELTIFY_MINCHUNK;
344 	gh = 0;
345 	while (p != data + MIN(fileoffset + GOT_DELTIFY_MAXCHUNK, filesize)) {
346 		gh = (gh << 1) + geartab[*p++];
347 		if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
348 			break;
349 	}
350 
351 	*blocklen = (p - (data + fileoffset));
352 	return NULL;
353 }
354 
355 const struct got_error *
356 got_deltify_init(struct got_delta_table **dt, FILE *f, off_t fileoffset,
357     off_t filesize)
358 {
359 	const struct got_error *err = NULL;
360 	uint64_t h;
361 	const off_t offset0 = fileoffset;
362 
363 	*dt = calloc(1, sizeof(**dt));
364 	if (*dt == NULL)
365 		return got_error_from_errno("calloc");
366 
367 	(*dt)->nblocks = 0;
368 	(*dt)->nalloc = 128;
369 	(*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
370 	if ((*dt)->blocks == NULL) {
371 		err = got_error_from_errno("calloc");
372 		goto done;
373 	}
374 
375 	if (fseeko(f, fileoffset, SEEK_SET) == -1)
376 		return got_error_from_errno("fseeko");
377 
378 	while (fileoffset < filesize) {
379 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
380 		off_t blocklen;
381 		err = nextblk(buf, &blocklen, f);
382 		if (err)
383 			goto done;
384 		if (blocklen == 0)
385 			break;
386 		h = hashblk(buf, blocklen);
387 		err = addblk(*dt, f, offset0, blocklen,
388 		    fileoffset - offset0, h);
389 		if (err)
390 			goto done;
391 		fileoffset += blocklen;
392 		if (fseeko(f, fileoffset, SEEK_SET) == -1)
393 			return got_error_from_errno("fseeko");
394 	}
395 done:
396 	if (err) {
397 		free((*dt)->blocks);
398 		free(*dt);
399 		*dt = NULL;
400 	}
401 
402 	return err;
403 }
404 
405 const struct got_error *
406 got_deltify_init_mem(struct got_delta_table **dt, uint8_t *data,
407     off_t fileoffset, off_t filesize)
408 {
409 	const struct got_error *err = NULL;
410 	uint64_t h;
411 	const off_t offset0 = fileoffset;
412 
413 	*dt = calloc(1, sizeof(**dt));
414 	if (*dt == NULL)
415 		return got_error_from_errno("calloc");
416 
417 	(*dt)->nblocks = 0;
418 	(*dt)->nalloc = 128;
419 	(*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
420 	if ((*dt)->blocks == NULL) {
421 		err = got_error_from_errno("calloc");
422 		goto done;
423 	}
424 
425 	while (fileoffset < filesize) {
426 		off_t blocklen;
427 		err = nextblk_mem(&blocklen, data, fileoffset, filesize);
428 		if (err)
429 			goto done;
430 		if (blocklen == 0)
431 			break;
432 		h = hashblk(data + fileoffset, blocklen);
433 		err = addblk_mem(*dt, data, offset0, blocklen,
434 		    fileoffset - offset0, h);
435 		if (err)
436 			goto done;
437 		fileoffset += blocklen;
438 	}
439 done:
440 	if (err) {
441 		free((*dt)->blocks);
442 		free(*dt);
443 		*dt = NULL;
444 	}
445 
446 	return err;
447 }
448 
449 void
450 got_deltify_free(struct got_delta_table *dt)
451 {
452 	if (dt == NULL)
453 		return;
454 	free(dt->blocks);
455 	free(dt);
456 }
457 
458 static const struct got_error *
459 emitdelta(struct got_delta_instruction **deltas, size_t *nalloc, int *ndeltas,
460     const size_t alloc_chunk_size, int copy, off_t offset, off_t len)
461 {
462 	struct got_delta_instruction *d, *p;
463 
464 	if (*nalloc < *ndeltas + alloc_chunk_size) {
465 		p = reallocarray(*deltas, *nalloc + alloc_chunk_size,
466 		    sizeof(struct got_delta_instruction));
467 		if (p == NULL)
468 			return got_error_from_errno("reallocarray");
469 		*deltas = p;
470 		*nalloc += alloc_chunk_size;
471 	}
472 	*ndeltas += 1;
473 	d = &(*deltas)[*ndeltas - 1];
474 	d->copy = copy;
475 	d->offset = offset;
476 	d->len = len;
477 	return NULL;
478 }
479 
480 static const struct got_error *
481 stretchblk(FILE *basefile, off_t base_offset0, struct got_delta_block *block,
482     FILE *f, off_t filesize, off_t *blocklen)
483 {
484 	uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
485 	size_t base_r, r, i;
486 	int buf_equal = 1;
487 
488 	if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
489 	    SEEK_SET) == -1)
490 		return got_error_from_errno("fseeko");
491 
492 	while (buf_equal && *blocklen < (1 << 24) - 1) {
493 		base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
494 		if (base_r == 0) {
495 			if (ferror(basefile))
496 				return got_ferror(basefile, GOT_ERR_IO);
497 			break;
498 		}
499 		r = fread(buf, 1, sizeof(buf), f);
500 		if (r == 0) {
501 			if (ferror(f))
502 				return got_ferror(f, GOT_ERR_IO);
503 			break;
504 		}
505 		for (i = 0; i < MIN(base_r, r); i++) {
506 			if (buf[i] != basebuf[i]) {
507 				buf_equal = 0;
508 				break;
509 			}
510 			(*blocklen)++;
511 		}
512 	}
513 
514 	return NULL;
515 }
516 
517 static const struct got_error *
518 stretchblk_file_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
519      struct got_delta_block *block, FILE *f, off_t filesize, off_t *blocklen)
520 {
521 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
522 	size_t r, i;
523 	int buf_equal = 1;
524 	off_t base_offset = base_offset0 + block->offset + *blocklen;
525 
526 	if (base_offset > basefile_size) {
527 		return got_error_fmt(GOT_ERR_RANGE,
528 		    "read beyond the size of delta base at offset %llu",
529 		    base_offset);
530 	}
531 
532 	while (buf_equal && *blocklen < (1 << 24) - 1) {
533 		if (base_offset + *blocklen >= basefile_size)
534 			break;
535 		r = fread(buf, 1, sizeof(buf), f);
536 		if (r == 0) {
537 			if (ferror(f))
538 				return got_ferror(f, GOT_ERR_IO);
539 			break;
540 		}
541 		for (i = 0; i < MIN(basefile_size - base_offset, r); i++) {
542 			if (buf[i] != *(basedata + base_offset + i)) {
543 				buf_equal = 0;
544 				break;
545 			}
546 			(*blocklen)++;
547 		}
548 	}
549 
550 	return NULL;
551 }
552 
553 static const struct got_error *
554 stretchblk_mem_file(FILE *basefile, off_t base_offset0,
555     struct got_delta_block *block, uint8_t *data, off_t fileoffset,
556     off_t filesize, off_t *blocklen)
557 {
558 	uint8_t basebuf[GOT_DELTIFY_MAXCHUNK];
559 	size_t base_r, i;
560 	int buf_equal = 1;
561 
562 	if (fileoffset > filesize) {
563 		return got_error_fmt(GOT_ERR_RANGE,
564 		    "read beyond the size of deltify file at offset %llu",
565 		    fileoffset);
566 	}
567 
568 	if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
569 	    SEEK_SET) == -1)
570 		return got_error_from_errno("fseeko");
571 
572 	while (buf_equal && *blocklen < (1 << 24) - 1) {
573 		if (fileoffset + *blocklen >= filesize)
574 			break;
575 		base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
576 		if (base_r == 0) {
577 			if (ferror(basefile))
578 				return got_ferror(basefile, GOT_ERR_IO);
579 			break;
580 		}
581 		for (i = 0; i < MIN(base_r, filesize - fileoffset); i++) {
582 			if (*(data + fileoffset + i) != basebuf[i]) {
583 				buf_equal = 0;
584 				break;
585 			}
586 			(*blocklen)++;
587 		}
588 	}
589 
590 	return NULL;
591 }
592 
593 static const struct got_error *
594 stretchblk_mem_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
595     struct got_delta_block *block, uint8_t *data, off_t fileoffset,
596     off_t filesize, off_t *blocklen)
597 {
598 	off_t i, maxlen;
599 	off_t base_offset = base_offset0 + block->offset + *blocklen;
600 	uint8_t *p, *q;
601 
602 	if (base_offset > basefile_size) {
603 		return got_error_fmt(GOT_ERR_RANGE,
604 		    "read beyond the size of delta base at offset %llu",
605 		    base_offset);
606 	}
607 
608 	if (fileoffset > filesize) {
609 		return got_error_fmt(GOT_ERR_RANGE,
610 		    "read beyond the size of deltify file at offset %llu",
611 		    fileoffset);
612 	}
613 
614 	p = data + fileoffset;
615 	q = basedata + base_offset;
616 	maxlen = MIN(basefile_size - base_offset, filesize - fileoffset);
617 	for (i = 0; i < maxlen && *blocklen < (1 << 24) - 1; i++) {
618 		if (p[i] != q[i])
619 			break;
620 		(*blocklen)++;
621 	}
622 
623 	return NULL;
624 }
625 
626 const struct got_error *
627 got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
628     FILE *f, off_t fileoffset, off_t filesize,
629     struct got_delta_table *dt, FILE *basefile,
630     off_t basefile_offset0, off_t basefile_size)
631 {
632 	const struct got_error *err = NULL;
633 	const off_t offset0 = fileoffset;
634 	size_t nalloc = 0;
635 	const size_t alloc_chunk_size = 64;
636 
637 	*deltas = NULL;
638 	*ndeltas = 0;
639 
640 	/*
641 	 * offset0 indicates where data to be deltified begins.
642 	 * For example, we want to avoid deltifying a Git object header at
643 	 * the beginning of the file.
644 	 */
645 	if (fseeko(f, offset0, SEEK_SET) == -1)
646 		return got_error_from_errno("fseeko");
647 
648 	*deltas = reallocarray(NULL, alloc_chunk_size,
649 	    sizeof(struct got_delta_instruction));
650 	if (*deltas == NULL)
651 		return got_error_from_errno("reallocarray");
652 	nalloc = alloc_chunk_size;
653 
654 	while (fileoffset < filesize) {
655 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
656 		off_t blocklen;
657 		struct got_delta_block *block;
658 		err = nextblk(buf, &blocklen, f);
659 		if (err)
660 			break;
661 		if (blocklen == 0) {
662 			/* Source remainder from the file itself. */
663 			if (fileoffset < filesize) {
664 				err = emitdelta(deltas, &nalloc, ndeltas,
665 				    alloc_chunk_size, 0, fileoffset - offset0,
666 				    filesize - fileoffset);
667 			}
668 			break;
669 		}
670 		err = lookupblk(&block, dt, buf, blocklen, basefile,
671 		    basefile_offset0);
672 		if (err)
673 			break;
674 		if (block != NULL) {
675 			/*
676 			 * We have found a matching block in the delta base.
677 			 * Attempt to stretch the block as far as possible and
678 			 * generate a copy instruction.
679 			 */
680 			err = stretchblk(basefile, basefile_offset0, block,
681 			    f, filesize, &blocklen);
682 			if (err)
683 				break;
684 			err = emitdelta(deltas, &nalloc, ndeltas,
685 			    alloc_chunk_size, 1, block->offset, blocklen);
686 			if (err)
687 				break;
688 		} else {
689 			/*
690 			 * No match.
691 			 * This block needs to be sourced from the file itself.
692 			 */
693 			err = emitdelta(deltas, &nalloc, ndeltas,
694 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
695 			if (err)
696 				break;
697 		}
698 		fileoffset += blocklen;
699 		if (fseeko(f, fileoffset, SEEK_SET) == -1) {
700 			err = got_error_from_errno("fseeko");
701 			break;
702 		}
703 	}
704 
705 	if (err) {
706 		free(*deltas);
707 		*deltas = NULL;
708 		*ndeltas = 0;
709 	}
710 	return err;
711 }
712 
713 const struct got_error *
714 got_deltify_file_mem(struct got_delta_instruction **deltas, int *ndeltas,
715     FILE *f, off_t fileoffset, off_t filesize,
716     struct got_delta_table *dt, uint8_t *basedata,
717     off_t basefile_offset0, off_t basefile_size)
718 {
719 	const struct got_error *err = NULL;
720 	const off_t offset0 = fileoffset;
721 	size_t nalloc = 0;
722 	const size_t alloc_chunk_size = 64;
723 
724 	*deltas = NULL;
725 	*ndeltas = 0;
726 
727 	/*
728 	 * offset0 indicates where data to be deltified begins.
729 	 * For example, we want to avoid deltifying a Git object header at
730 	 * the beginning of the file.
731 	 */
732 	if (fseeko(f, offset0, SEEK_SET) == -1)
733 		return got_error_from_errno("fseeko");
734 
735 	*deltas = reallocarray(NULL, alloc_chunk_size,
736 	    sizeof(struct got_delta_instruction));
737 	if (*deltas == NULL)
738 		return got_error_from_errno("reallocarray");
739 	nalloc = alloc_chunk_size;
740 
741 	while (fileoffset < filesize) {
742 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
743 		off_t blocklen;
744 		struct got_delta_block *block;
745 		err = nextblk(buf, &blocklen, f);
746 		if (err)
747 			break;
748 		if (blocklen == 0) {
749 			/* Source remainder from the file itself. */
750 			if (fileoffset < filesize) {
751 				err = emitdelta(deltas, &nalloc, ndeltas,
752 				    alloc_chunk_size, 0, fileoffset - offset0,
753 				    filesize - fileoffset);
754 			}
755 			break;
756 		}
757 		err = lookupblk_mem(&block, dt, buf, blocklen, basedata,
758 		    basefile_offset0);
759 		if (err)
760 			break;
761 		if (block != NULL) {
762 			/*
763 			 * We have found a matching block in the delta base.
764 			 * Attempt to stretch the block as far as possible and
765 			 * generate a copy instruction.
766 			 */
767 			err = stretchblk_file_mem(basedata, basefile_offset0,
768 			    basefile_size, block, f, filesize, &blocklen);
769 			if (err)
770 				break;
771 			err = emitdelta(deltas, &nalloc, ndeltas,
772 			    alloc_chunk_size, 1, block->offset, blocklen);
773 			if (err)
774 				break;
775 		} else {
776 			/*
777 			 * No match.
778 			 * This block needs to be sourced from the file itself.
779 			 */
780 			err = emitdelta(deltas, &nalloc, ndeltas,
781 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
782 			if (err)
783 				break;
784 		}
785 		fileoffset += blocklen;
786 		if (fseeko(f, fileoffset, SEEK_SET) == -1) {
787 			err = got_error_from_errno("fseeko");
788 			break;
789 		}
790 	}
791 
792 	if (err) {
793 		free(*deltas);
794 		*deltas = NULL;
795 		*ndeltas = 0;
796 	}
797 	return err;
798 }
799 
800 const struct got_error *
801 got_deltify_mem_file(struct got_delta_instruction **deltas, int *ndeltas,
802     uint8_t *data, off_t fileoffset, off_t filesize,
803     struct got_delta_table *dt, FILE *basefile,
804     off_t basefile_offset0, off_t basefile_size)
805 {
806 	const struct got_error *err = NULL;
807 	const off_t offset0 = fileoffset;
808 	size_t nalloc = 0;
809 	const size_t alloc_chunk_size = 64;
810 
811 	*deltas = NULL;
812 	*ndeltas = 0;
813 
814 	*deltas = reallocarray(NULL, alloc_chunk_size,
815 	    sizeof(struct got_delta_instruction));
816 	if (*deltas == NULL)
817 		return got_error_from_errno("reallocarray");
818 	nalloc = alloc_chunk_size;
819 
820 	while (fileoffset < filesize) {
821 		off_t blocklen;
822 		struct got_delta_block *block;
823 		err = nextblk_mem(&blocklen, data, fileoffset, filesize);
824 		if (err)
825 			break;
826 		if (blocklen == 0) {
827 			/* Source remainder from the file itself. */
828 			if (fileoffset < filesize) {
829 				err = emitdelta(deltas, &nalloc, ndeltas,
830 				    alloc_chunk_size, 0, fileoffset - offset0,
831 				    filesize - fileoffset);
832 			}
833 			break;
834 		}
835 		err = lookupblk(&block, dt, data + fileoffset, blocklen,
836 		    basefile, basefile_offset0);
837 		if (err)
838 			break;
839 		if (block != NULL) {
840 			/*
841 			 * We have found a matching block in the delta base.
842 			 * Attempt to stretch the block as far as possible and
843 			 * generate a copy instruction.
844 			 */
845 			err = stretchblk_mem_file(basefile, basefile_offset0,
846 			    block, data, fileoffset + blocklen, filesize,
847 			    &blocklen);
848 			if (err)
849 				break;
850 			err = emitdelta(deltas, &nalloc, ndeltas,
851 			    alloc_chunk_size, 1, block->offset, blocklen);
852 			if (err)
853 				break;
854 		} else {
855 			/*
856 			 * No match.
857 			 * This block needs to be sourced from the file itself.
858 			 */
859 			err = emitdelta(deltas, &nalloc, ndeltas,
860 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
861 			if (err)
862 				break;
863 		}
864 		fileoffset += blocklen;
865 	}
866 
867 	if (err) {
868 		free(*deltas);
869 		*deltas = NULL;
870 		*ndeltas = 0;
871 	}
872 	return err;
873 }
874 
875 const struct got_error *
876 got_deltify_mem_mem(struct got_delta_instruction **deltas, int *ndeltas,
877     uint8_t *data, off_t fileoffset, off_t filesize,
878     struct got_delta_table *dt, uint8_t *basedata,
879     off_t basefile_offset0, off_t basefile_size)
880 {
881 	const struct got_error *err = NULL;
882 	const off_t offset0 = fileoffset;
883 	size_t nalloc = 0;
884 	const size_t alloc_chunk_size = 64;
885 
886 	*deltas = NULL;
887 	*ndeltas = 0;
888 
889 	*deltas = reallocarray(NULL, alloc_chunk_size,
890 	    sizeof(struct got_delta_instruction));
891 	if (*deltas == NULL)
892 		return got_error_from_errno("reallocarray");
893 	nalloc = alloc_chunk_size;
894 
895 	while (fileoffset < filesize) {
896 		off_t blocklen;
897 		struct got_delta_block *block;
898 		err = nextblk_mem(&blocklen, data, fileoffset, filesize);
899 		if (err)
900 			break;
901 		if (blocklen == 0) {
902 			/* Source remainder from the file itself. */
903 			if (fileoffset < filesize) {
904 				err = emitdelta(deltas, &nalloc, ndeltas,
905 				    alloc_chunk_size, 0, fileoffset - offset0,
906 				    filesize - fileoffset);
907 			}
908 			break;
909 		}
910 		err = lookupblk_mem(&block, dt, data + fileoffset, blocklen,
911 		    basedata, basefile_offset0);
912 		if (err)
913 			break;
914 		if (block != NULL) {
915 			/*
916 			 * We have found a matching block in the delta base.
917 			 * Attempt to stretch the block as far as possible and
918 			 * generate a copy instruction.
919 			 */
920 			err = stretchblk_mem_mem(basedata, basefile_offset0,
921 			    basefile_size, block, data, fileoffset + blocklen,
922 			    filesize, &blocklen);
923 			if (err)
924 				break;
925 			err = emitdelta(deltas, &nalloc, ndeltas,
926 			    alloc_chunk_size, 1, block->offset, blocklen);
927 			if (err)
928 				break;
929 		} else {
930 			/*
931 			 * No match.
932 			 * This block needs to be sourced from the file itself.
933 			 */
934 			err = emitdelta(deltas, &nalloc, ndeltas,
935 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
936 			if (err)
937 				break;
938 		}
939 		fileoffset += blocklen;
940 	}
941 
942 	if (err) {
943 		free(*deltas);
944 		*deltas = NULL;
945 		*ndeltas = 0;
946 	}
947 	return err;
948 }