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 "got_compat.h"
19 
20 #include <sys/types.h>
21 #include <sys/queue.h>
22 
23 #include <assert.h>
24 #include <errno.h>
25 #include <limits.h>
26 #include <stdint.h>
27 #include <stdio.h>
28 #include <stdlib.h>
29 #include <string.h>
30 
31 #include "got_error.h"
32 
33 #include "got_lib_deltify.h"
34 #include "murmurhash2.h"
35 
36 #ifndef MIN
37 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
38 #endif
39 
40 /*
41  * The algorihm used here is FastCDC (Fast Content-Defined Chunking)
42  * https://www.usenix.org/conference/atc16/technical-sessions/presentation/xia
43  */
44 
45 static const uint32_t geartab[256] = {
46     0x67ed26b7, 0x32da500c, 0x53d0fee0, 0xce387dc7, 0xcd406d90, 0x2e83a4d4,
47     0x9fc9a38d, 0xb67259dc, 0xca6b1722, 0x6d2ea08c, 0x235cea2e, 0x3149bb5f,
48     0x1beda787, 0x2a6b77d5, 0x2f22d9ac, 0x91fc0544, 0xe413acfa, 0x5a30ff7a,
49     0xad6fdde0, 0x444fd0f5, 0x7ad87864, 0x58c5ff05, 0x8d2ec336, 0x2371f853,
50     0x550f8572, 0x6aa448dd, 0x7c9ddbcf, 0x95221e14, 0x2a82ec33, 0xcbec5a78,
51     0xc6795a0d, 0x243995b7, 0x1c909a2f, 0x4fded51c, 0x635d334b, 0x0e2b9999,
52     0x2702968d, 0x856de1d5, 0x3325d60e, 0xeb6a7502, 0xec2a9844, 0x0905835a,
53     0xa1820375, 0xa4be5cab, 0x96a6c058, 0x2c2ccd70, 0xba40fce3, 0xd794c46b,
54     0x8fbae83e, 0xc3aa7899, 0x3d3ff8ed, 0xa0d42b5b, 0x571c0c97, 0xd2811516,
55     0xf7e7b96c, 0x4fd2fcbd, 0xe2fdec94, 0x282cc436, 0x78e8e95c, 0x80a3b613,
56     0xcfbee20c, 0xd4a32d1c, 0x2a12ff13, 0x6af82936, 0xe5630258, 0x8efa6a98,
57     0x294fb2d1, 0xdeb57086, 0x5f0fddb3, 0xeceda7ce, 0x4c87305f, 0x3a6d3307,
58     0xe22d2942, 0x9d060217, 0x1e42ed02, 0xb6f63b52, 0x4367f39f, 0x055cf262,
59     0x03a461b2, 0x5ef9e382, 0x386bc03a, 0x2a1e79c7, 0xf1a0058b, 0xd4d2dea9,
60     0x56baf37d, 0x5daff6cc, 0xf03a951d, 0xaef7de45, 0xa8f4581e, 0x3960b555,
61     0xffbfff6d, 0xbe702a23, 0x8f5b6d6f, 0x061739fb, 0x98696f47, 0x3fd596d4,
62     0x151eac6b, 0xa9fcc4f5, 0x69181a12, 0x3ac5a107, 0xb5198fe7, 0x96bcb1da,
63     0x1b5ddf8e, 0xc757d650, 0x65865c3a, 0x8fc0a41a, 0x87435536, 0x99eda6f2,
64     0x41874794, 0x29cff4e8, 0xb70efd9a, 0x3103f6e7, 0x84d2453b, 0x15a450bd,
65     0x74f49af1, 0x60f664b1, 0xa1c86935, 0xfdafbce1, 0xe36353e3, 0x5d9ba739,
66     0xbc0559ba, 0x708b0054, 0xd41d808c, 0xb2f31723, 0x9027c41f, 0xf136d165,
67     0xb5374b12, 0x9420a6ac, 0x273958b6, 0xe6c2fad0, 0xebdc1f21, 0xfb33af8b,
68     0xc71c25cd, 0xe9a2d8e5, 0xbeb38a50, 0xbceb7cc2, 0x4e4e73f0, 0xcd6c251d,
69     0xde4c032c, 0x4b04ac30, 0x725b8b21, 0x4eb8c33b, 0x20d07b75, 0x0567aa63,
70     0xb56b2bb7, 0xc1f5fd3a, 0xcafd35ca, 0x470dd4da, 0xfe4f94cd, 0xfb8de424,
71     0xe8dbcf40, 0xfe50a37a, 0x62db5b5d, 0xf32f4ab6, 0x2c4a8a51, 0x18473dc0,
72     0xfe0cbb6e, 0xfe399efd, 0xdf34ecc9, 0x6ccd5055, 0x46097073, 0x139135c2,
73     0x721c76f6, 0x1c6a94b4, 0x6eee014d, 0x8a508e02, 0x3da538f5, 0x280d394f,
74     0x5248a0c4, 0x3ce94c6c, 0x9a71ad3a, 0x8493dd05, 0xe43f0ab6, 0x18e4ed42,
75     0x6c5c0e09, 0x42b06ec9, 0x8d330343, 0xa45b6f59, 0x2a573c0c, 0xd7fd3de6,
76     0xeedeab68, 0x5c84dafc, 0xbbd1b1a8, 0xa3ce1ad1, 0x85b70bed, 0xb6add07f,
77     0xa531309c, 0x8f8ab852, 0x564de332, 0xeac9ed0c, 0x73da402c, 0x3ec52761,
78     0x43af2f4d, 0xd6ff45c8, 0x4c367462, 0xd553bd6a, 0x44724855, 0x3b2aa728,
79     0x56e5eb65, 0xeaf16173, 0x33fa42ff, 0xd714bb5d, 0xfbd0a3b9, 0xaf517134,
80     0x9416c8cd, 0x534cf94f, 0x548947c2, 0x34193569, 0x32f4389a, 0xfe7028bc,
81     0xed73b1ed, 0x9db95770, 0x468e3922, 0x0440c3cd, 0x60059a62, 0x33504562,
82     0x2b229fbd, 0x5174dca5, 0xf7028752, 0xd63c6aa8, 0x31276f38, 0x0646721c,
83     0xb0191da8, 0xe00e6de0, 0x9eac1a6e, 0x9f7628a5, 0xed6c06ea, 0x0bb8af15,
84     0xf119fb12, 0x38693c1c, 0x732bc0fe, 0x84953275, 0xb82ec888, 0x33a4f1b3,
85     0x3099835e, 0x028a8782, 0x5fdd51d7, 0xc6c717b3, 0xb06caf71, 0x17c8c111,
86     0x61bad754, 0x9fd03061, 0xe09df1af, 0x3bc9eb73, 0x85878413, 0x9889aaf2,
87     0x3f5a9e46, 0x42c9f01f, 0x9984a4f4, 0xd5de43cc, 0xd294daed, 0xbecba2d2,
88     0xf1f6e72c, 0x5551128a, 0x83af87e2, 0x6f0342ba,
89 };
90 
91 static uint32_t
92 hashblk(const unsigned char *p, off_t n, uint32_t seed)
93 {
94 	return murmurhash2(p, n, seed);
95 }
96 
97 static const struct got_error *
98 addblk(struct got_delta_table *dt, FILE *f, off_t file_offset0, off_t len,
99     off_t offset, uint32_t h)
100 {
101 	const struct got_error *err = NULL;
102 	int i;
103 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
104 	uint8_t buf2[GOT_DELTIFY_MAXCHUNK];
105 	size_t r = 0;
106 
107 	if (len == 0)
108 		return NULL;
109 
110 	i = h % dt->nalloc;
111 	while (dt->blocks[i].len != 0) {
112 		/*
113 		 * Avoid adding duplicate blocks.
114 		 * NB: A matching hash is insufficient for detecting equality.
115 		 * The hash can only detect inequality.
116 		 */
117 		if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
118 			if (r == 0) {
119 				if (fseeko(f, file_offset0 + offset, SEEK_SET) == -1)
120 					return got_error_from_errno("fseeko");
121 				r = fread(buf, 1, len, f);
122 				if (r != len) {
123 					if (!ferror(f))
124 						return NULL;
125 					return got_ferror(f, GOT_ERR_IO);
126 				}
127 			}
128 			if (fseeko(f, file_offset0 + dt->blocks[i].offset,
129 			    SEEK_SET) == -1)
130 				return got_error_from_errno("fseeko");
131 			if (fread(buf2, 1, len, f) != len)
132 				return got_ferror(f, GOT_ERR_IO);
133 			if (memcmp(buf, buf2, len) == 0)
134 				return NULL;
135 		}
136 
137 		i = (i + 1) % dt->nalloc;
138 	}
139 	assert(dt->blocks[i].len == 0);
140 	dt->blocks[i].len = len;
141 	dt->blocks[i].offset = offset;
142 	dt->blocks[i].hash = h;
143 	dt->nblocks++;
144 	if (dt->nalloc < dt->nblocks + 64) {
145 		struct got_delta_block *db;
146 		size_t old_size = dt->nalloc;
147 		db = dt->blocks;
148 		dt->blocks = calloc(dt->nalloc + 64,
149 		    sizeof(struct got_delta_block));
150 		if (dt->blocks == NULL) {
151 			err = got_error_from_errno("calloc");
152 			dt->blocks = db;
153 			return err;
154 		}
155 		dt->nalloc += 64;
156 		/*
157 		 * Recompute all block positions. Hash-based indices of blocks
158 		 * in the array depend on the allocated length of the array.
159 		 */
160 		dt->nblocks = 0;
161 		for (i = 0; i < old_size; i++) {
162 			if (db[i].len == 0)
163 				continue;
164 			err = addblk(dt, f, file_offset0, db[i].len,
165 			    db[i].offset, db[i].hash);
166 			if (err)
167 				break;
168 		}
169 		free(db);
170 	}
171 
172 	return err;
173 }
174 
175 static const struct got_error *
176 addblk_mem(struct got_delta_table *dt, uint8_t *data, off_t file_offset0,
177     off_t len, off_t offset, uint32_t h)
178 {
179 	const struct got_error *err = NULL;
180 	int i;
181 	uint8_t *block1;
182 	uint8_t *block2;
183 
184 	if (len == 0)
185 		return NULL;
186 
187 	i = h % dt->nalloc;
188 	while (dt->blocks[i].len != 0) {
189 		/*
190 		 * Avoid adding duplicate blocks.
191 		 * NB: A matching hash is insufficient for detecting equality.
192 		 * The hash can only detect inequality.
193 		 */
194 		if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
195 			block1 = data + file_offset0 + dt->blocks[i].offset;
196 			block2 = data + file_offset0 + offset;
197 			if (memcmp(block1, block2, len) == 0)
198 				return NULL;
199 		}
200 
201 		i = (i + 1) % dt->nalloc;
202 	}
203 	assert(dt->blocks[i].len == 0);
204 	dt->blocks[i].len = len;
205 	dt->blocks[i].offset = offset;
206 	dt->blocks[i].hash = h;
207 	dt->nblocks++;
208 	if (dt->nalloc < dt->nblocks + 64) {
209 		struct got_delta_block *db;
210 		size_t old_size = dt->nalloc;
211 		db = dt->blocks;
212 		dt->blocks = calloc(dt->nalloc + 64,
213 		    sizeof(struct got_delta_block));
214 		if (dt->blocks == NULL) {
215 			err = got_error_from_errno("calloc");
216 			dt->blocks = db;
217 			return err;
218 		}
219 		dt->nalloc += 64;
220 		/*
221 		 * Recompute all block positions. Hash-based indices of blocks
222 		 * in the array depend on the allocated length of the array.
223 		 */
224 		dt->nblocks = 0;
225 		for (i = 0; i < old_size; i++) {
226 			if (db[i].len == 0)
227 				continue;
228 			err = addblk_mem(dt, data, file_offset0, db[i].len,
229 			    db[i].offset, db[i].hash);
230 			if (err)
231 				break;
232 		}
233 		free(db);
234 	}
235 
236 	return err;
237 }
238 
239 static const struct got_error *
240 lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
241     unsigned char *p, off_t len, uint32_t seed, FILE *basefile,
242     off_t basefile_offset0)
243 {
244 	int i;
245 	uint32_t h;
246 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
247 	size_t r;
248 
249 	*block = NULL;
250 
251 	h = hashblk(p, len, seed);
252 	for (i = h % dt->nalloc; dt->blocks[i].len != 0;
253 	     i = (i + 1) % dt->nalloc) {
254 		if (dt->blocks[i].hash != h ||
255 		    dt->blocks[i].len != len)
256 			continue;
257 		if (fseeko(basefile, basefile_offset0 + dt->blocks[i].offset,
258 		    SEEK_SET) == -1)
259 			return got_error_from_errno("fseeko");
260 		r = fread(buf, 1, len, basefile);
261 		if (r != len)
262 			return got_ferror(basefile, GOT_ERR_IO);
263 		if (memcmp(p, buf, len) == 0) {
264 			*block = &dt->blocks[i];
265 			break;
266 		}
267 	}
268 	return NULL;
269 }
270 
271 static const struct got_error *
272 lookupblk_mem(struct got_delta_block **block, struct got_delta_table *dt,
273     unsigned char *p, off_t len, uint32_t seed, uint8_t *basedata,
274     off_t basefile_offset0)
275 {
276 	int i;
277 	uint32_t h;
278 	uint8_t *b;
279 
280 	*block = NULL;
281 
282 	h = hashblk(p, len, seed);
283 	for (i = h % dt->nalloc; dt->blocks[i].len != 0;
284 	     i = (i + 1) % dt->nalloc) {
285 		if (dt->blocks[i].hash != h ||
286 		    dt->blocks[i].len != len)
287 			continue;
288 		b = basedata + basefile_offset0 + dt->blocks[i].offset;
289 		if (memcmp(p, b, len) == 0) {
290 			*block = &dt->blocks[i];
291 			break;
292 		}
293 	}
294 	return NULL;
295 }
296 
297 static const struct got_error *
298 nextblk(uint8_t *buf, off_t *blocklen, FILE *f)
299 {
300 	uint32_t gh;
301 	const unsigned char *p;
302 	size_t r;
303 	off_t pos = ftello(f);
304 
305 	*blocklen = 0;
306 
307 	r = fread(buf, 1, GOT_DELTIFY_MAXCHUNK, f);
308 	if (r == 0 && ferror(f))
309 		return got_ferror(f, GOT_ERR_IO);
310 	if (r < GOT_DELTIFY_MINCHUNK)
311 		return NULL; /* no more delta-worthy blocks left */
312 
313 	/* Got a deltifiable block. Find the split-point where it ends. */
314 	p = buf + GOT_DELTIFY_MINCHUNK;
315 	gh = 0;
316 	while (p != buf + r) {
317 		gh = (gh << 1) + geartab[*p++];
318 		if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
319 			break;
320 	}
321 
322 	*blocklen = (p - buf);
323 	if (fseeko(f, pos + *blocklen, SEEK_SET) == -1)
324 		return got_error_from_errno("fseeko");
325 
326 	return NULL;
327 }
328 
329 static const struct got_error *
330 nextblk_mem(off_t *blocklen, uint8_t *data, off_t fileoffset, off_t filesize)
331 {
332 	uint32_t gh;
333 	const unsigned char *p;
334 
335 	*blocklen = 0;
336 
337 	if (fileoffset >= filesize ||
338 	    filesize - fileoffset < GOT_DELTIFY_MINCHUNK)
339 		return NULL; /* no more delta-worthy blocks left */
340 
341 	/* Got a deltifiable block. Find the split-point where it ends. */
342 	p = data + fileoffset + GOT_DELTIFY_MINCHUNK;
343 	gh = 0;
344 	while (p != data + MIN(fileoffset + GOT_DELTIFY_MAXCHUNK, filesize)) {
345 		gh = (gh << 1) + geartab[*p++];
346 		if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
347 			break;
348 	}
349 
350 	*blocklen = (p - (data + fileoffset));
351 	return NULL;
352 }
353 
354 const struct got_error *
355 got_deltify_init(struct got_delta_table **dt, FILE *f, off_t fileoffset,
356     off_t filesize, uint32_t seed)
357 {
358 	const struct got_error *err = NULL;
359 	uint32_t h;
360 	const off_t offset0 = fileoffset;
361 
362 	*dt = calloc(1, sizeof(**dt));
363 	if (*dt == NULL)
364 		return got_error_from_errno("calloc");
365 
366 	(*dt)->nblocks = 0;
367 	(*dt)->nalloc = 128;
368 	(*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
369 	if ((*dt)->blocks == NULL) {
370 		err = got_error_from_errno("calloc");
371 		goto done;
372 	}
373 
374 	if (fseeko(f, fileoffset, SEEK_SET) == -1)
375 		return got_error_from_errno("fseeko");
376 
377 	while (fileoffset < filesize) {
378 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
379 		off_t blocklen;
380 		err = nextblk(buf, &blocklen, f);
381 		if (err)
382 			goto done;
383 		if (blocklen == 0)
384 			break;
385 		h = hashblk(buf, blocklen, seed);
386 		err = addblk(*dt, f, offset0, blocklen,
387 		    fileoffset - offset0, h);
388 		if (err)
389 			goto done;
390 		fileoffset += blocklen;
391 		if (fseeko(f, fileoffset, SEEK_SET) == -1)
392 			return got_error_from_errno("fseeko");
393 	}
394 done:
395 	if (err) {
396 		free((*dt)->blocks);
397 		free(*dt);
398 		*dt = NULL;
399 	}
400 
401 	return err;
402 }
403 
404 const struct got_error *
405 got_deltify_init_mem(struct got_delta_table **dt, uint8_t *data,
406     off_t fileoffset, off_t filesize, uint32_t seed)
407 {
408 	const struct got_error *err = NULL;
409 	uint32_t h;
410 	const off_t offset0 = fileoffset;
411 
412 	*dt = calloc(1, sizeof(**dt));
413 	if (*dt == NULL)
414 		return got_error_from_errno("calloc");
415 
416 	(*dt)->nblocks = 0;
417 	(*dt)->nalloc = 128;
418 	(*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
419 	if ((*dt)->blocks == NULL) {
420 		err = got_error_from_errno("calloc");
421 		goto done;
422 	}
423 
424 	while (fileoffset < filesize) {
425 		off_t blocklen;
426 		err = nextblk_mem(&blocklen, data, fileoffset, filesize);
427 		if (err)
428 			goto done;
429 		if (blocklen == 0)
430 			break;
431 		h = hashblk(data + fileoffset, blocklen, seed);
432 		err = addblk_mem(*dt, data, offset0, blocklen,
433 		    fileoffset - offset0, h);
434 		if (err)
435 			goto done;
436 		fileoffset += blocklen;
437 	}
438 done:
439 	if (err) {
440 		free((*dt)->blocks);
441 		free(*dt);
442 		*dt = NULL;
443 	}
444 
445 	return err;
446 }
447 
448 void
449 got_deltify_free(struct got_delta_table *dt)
450 {
451 	if (dt == NULL)
452 		return;
453 	free(dt->blocks);
454 	free(dt);
455 }
456 
457 static const struct got_error *
458 emitdelta(struct got_delta_instruction **deltas, size_t *nalloc, int *ndeltas,
459     const size_t alloc_chunk_size, int copy, off_t offset, off_t len)
460 {
461 	struct got_delta_instruction *d, *p;
462 
463 	if (*nalloc < *ndeltas + alloc_chunk_size) {
464 		p = reallocarray(*deltas, *nalloc + alloc_chunk_size,
465 		    sizeof(struct got_delta_instruction));
466 		if (p == NULL)
467 			return got_error_from_errno("reallocarray");
468 		*deltas = p;
469 		*nalloc += alloc_chunk_size;
470 	}
471 	*ndeltas += 1;
472 	d = &(*deltas)[*ndeltas - 1];
473 	d->copy = copy;
474 	d->offset = offset;
475 	d->len = len;
476 	return NULL;
477 }
478 
479 static const struct got_error *
480 stretchblk(FILE *basefile, off_t base_offset0, struct got_delta_block *block,
481     FILE *f, off_t filesize, off_t *blocklen)
482 {
483 	uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
484 	size_t base_r, r, i;
485 	int buf_equal = 1;
486 
487 	if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
488 	    SEEK_SET) == -1)
489 		return got_error_from_errno("fseeko");
490 
491 	while (buf_equal && *blocklen < (1 << 24) - 1) {
492 		base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
493 		if (base_r == 0) {
494 			if (ferror(basefile))
495 				return got_ferror(basefile, GOT_ERR_IO);
496 			break;
497 		}
498 		r = fread(buf, 1, sizeof(buf), f);
499 		if (r == 0) {
500 			if (ferror(f))
501 				return got_ferror(f, GOT_ERR_IO);
502 			break;
503 		}
504 		for (i = 0; i < MIN(base_r, r); i++) {
505 			if (buf[i] != basebuf[i]) {
506 				buf_equal = 0;
507 				break;
508 			}
509 			(*blocklen)++;
510 		}
511 	}
512 
513 	return NULL;
514 }
515 
516 static const struct got_error *
517 stretchblk_file_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
518      struct got_delta_block *block, FILE *f, off_t filesize, off_t *blocklen)
519 {
520 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
521 	size_t r, i;
522 	int buf_equal = 1;
523 	off_t base_offset = base_offset0 + block->offset + *blocklen;
524 
525 	if (base_offset > basefile_size) {
526 		return got_error_fmt(GOT_ERR_RANGE,
527 		    "read beyond the size of delta base at offset %lld",
528 		    (long long)base_offset);
529 	}
530 
531 	while (buf_equal && *blocklen < (1 << 24) - 1) {
532 		if (base_offset + *blocklen >= basefile_size)
533 			break;
534 		r = fread(buf, 1, sizeof(buf), f);
535 		if (r == 0) {
536 			if (ferror(f))
537 				return got_ferror(f, GOT_ERR_IO);
538 			break;
539 		}
540 		for (i = 0; i < MIN(basefile_size - base_offset, r); i++) {
541 			if (buf[i] != *(basedata + base_offset + i)) {
542 				buf_equal = 0;
543 				break;
544 			}
545 			(*blocklen)++;
546 		}
547 	}
548 
549 	return NULL;
550 }
551 
552 static const struct got_error *
553 stretchblk_mem_file(FILE *basefile, off_t base_offset0,
554     struct got_delta_block *block, uint8_t *data, off_t fileoffset,
555     off_t filesize, off_t *blocklen)
556 {
557 	uint8_t basebuf[GOT_DELTIFY_MAXCHUNK];
558 	size_t base_r, i;
559 	int buf_equal = 1;
560 
561 	if (fileoffset > filesize) {
562 		return got_error_fmt(GOT_ERR_RANGE,
563 		    "read beyond the size of deltify file at offset %lld",
564 		    (long long)fileoffset);
565 	}
566 
567 	if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
568 	    SEEK_SET) == -1)
569 		return got_error_from_errno("fseeko");
570 
571 	while (buf_equal && *blocklen < (1 << 24) - 1) {
572 		if (fileoffset + *blocklen >= filesize)
573 			break;
574 		base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
575 		if (base_r == 0) {
576 			if (ferror(basefile))
577 				return got_ferror(basefile, GOT_ERR_IO);
578 			break;
579 		}
580 		for (i = 0; i < MIN(base_r, filesize - fileoffset); i++) {
581 			if (*(data + fileoffset + i) != basebuf[i]) {
582 				buf_equal = 0;
583 				break;
584 			}
585 			(*blocklen)++;
586 		}
587 	}
588 
589 	return NULL;
590 }
591 
592 static const struct got_error *
593 stretchblk_mem_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
594     struct got_delta_block *block, uint8_t *data, off_t fileoffset,
595     off_t filesize, off_t *blocklen)
596 {
597 	off_t i, maxlen;
598 	off_t base_offset = base_offset0 + block->offset + *blocklen;
599 	uint8_t *p, *q;
600 
601 	if (base_offset > basefile_size) {
602 		return got_error_fmt(GOT_ERR_RANGE,
603 		    "read beyond the size of delta base at offset %lld",
604 		    (long long)base_offset);
605 	}
606 
607 	if (fileoffset > filesize) {
608 		return got_error_fmt(GOT_ERR_RANGE,
609 		    "read beyond the size of deltify file at offset %lld",
610 		    (long long)fileoffset);
611 	}
612 
613 	p = data + fileoffset;
614 	q = basedata + base_offset;
615 	maxlen = MIN(basefile_size - base_offset, filesize - fileoffset);
616 	for (i = 0; i < maxlen && *blocklen < (1 << 24) - 1; i++) {
617 		if (p[i] != q[i])
618 			break;
619 		(*blocklen)++;
620 	}
621 
622 	return NULL;
623 }
624 
625 const struct got_error *
626 got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
627     FILE *f, off_t fileoffset, off_t filesize, uint32_t seed,
628     struct got_delta_table *dt, FILE *basefile,
629     off_t basefile_offset0, off_t basefile_size)
630 {
631 	const struct got_error *err = NULL;
632 	const off_t offset0 = fileoffset;
633 	size_t nalloc = 0;
634 	const size_t alloc_chunk_size = 64;
635 
636 	*deltas = NULL;
637 	*ndeltas = 0;
638 
639 	/*
640 	 * offset0 indicates where data to be deltified begins.
641 	 * For example, we want to avoid deltifying a Git object header at
642 	 * the beginning of the file.
643 	 */
644 	if (fseeko(f, offset0, SEEK_SET) == -1)
645 		return got_error_from_errno("fseeko");
646 
647 	*deltas = reallocarray(NULL, alloc_chunk_size,
648 	    sizeof(struct got_delta_instruction));
649 	if (*deltas == NULL)
650 		return got_error_from_errno("reallocarray");
651 	nalloc = alloc_chunk_size;
652 
653 	while (fileoffset < filesize) {
654 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
655 		off_t blocklen;
656 		struct got_delta_block *block;
657 		err = nextblk(buf, &blocklen, f);
658 		if (err)
659 			break;
660 		if (blocklen == 0) {
661 			/* Source remainder from the file itself. */
662 			if (fileoffset < filesize) {
663 				err = emitdelta(deltas, &nalloc, ndeltas,
664 				    alloc_chunk_size, 0, fileoffset - offset0,
665 				    filesize - fileoffset);
666 			}
667 			break;
668 		}
669 		err = lookupblk(&block, dt, buf, blocklen, seed, basefile,
670 		    basefile_offset0);
671 		if (err)
672 			break;
673 		if (block != NULL) {
674 			/*
675 			 * We have found a matching block in the delta base.
676 			 * Attempt to stretch the block as far as possible and
677 			 * generate a copy instruction.
678 			 */
679 			err = stretchblk(basefile, basefile_offset0, block,
680 			    f, filesize, &blocklen);
681 			if (err)
682 				break;
683 			err = emitdelta(deltas, &nalloc, ndeltas,
684 			    alloc_chunk_size, 1, block->offset, blocklen);
685 			if (err)
686 				break;
687 		} else {
688 			/*
689 			 * No match.
690 			 * This block needs to be sourced from the file itself.
691 			 */
692 			err = emitdelta(deltas, &nalloc, ndeltas,
693 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
694 			if (err)
695 				break;
696 		}
697 		fileoffset += blocklen;
698 		if (fseeko(f, fileoffset, SEEK_SET) == -1) {
699 			err = got_error_from_errno("fseeko");
700 			break;
701 		}
702 	}
703 
704 	if (err) {
705 		free(*deltas);
706 		*deltas = NULL;
707 		*ndeltas = 0;
708 	}
709 	return err;
710 }
711 
712 const struct got_error *
713 got_deltify_file_mem(struct got_delta_instruction **deltas, int *ndeltas,
714     FILE *f, off_t fileoffset, off_t filesize, uint32_t seed,
715     struct got_delta_table *dt, uint8_t *basedata,
716     off_t basefile_offset0, off_t basefile_size)
717 {
718 	const struct got_error *err = NULL;
719 	const off_t offset0 = fileoffset;
720 	size_t nalloc = 0;
721 	const size_t alloc_chunk_size = 64;
722 
723 	*deltas = NULL;
724 	*ndeltas = 0;
725 
726 	/*
727 	 * offset0 indicates where data to be deltified begins.
728 	 * For example, we want to avoid deltifying a Git object header at
729 	 * the beginning of the file.
730 	 */
731 	if (fseeko(f, offset0, SEEK_SET) == -1)
732 		return got_error_from_errno("fseeko");
733 
734 	*deltas = reallocarray(NULL, alloc_chunk_size,
735 	    sizeof(struct got_delta_instruction));
736 	if (*deltas == NULL)
737 		return got_error_from_errno("reallocarray");
738 	nalloc = alloc_chunk_size;
739 
740 	while (fileoffset < filesize) {
741 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
742 		off_t blocklen;
743 		struct got_delta_block *block;
744 		err = nextblk(buf, &blocklen, f);
745 		if (err)
746 			break;
747 		if (blocklen == 0) {
748 			/* Source remainder from the file itself. */
749 			if (fileoffset < filesize) {
750 				err = emitdelta(deltas, &nalloc, ndeltas,
751 				    alloc_chunk_size, 0, fileoffset - offset0,
752 				    filesize - fileoffset);
753 			}
754 			break;
755 		}
756 		err = lookupblk_mem(&block, dt, buf, blocklen, seed, basedata,
757 		    basefile_offset0);
758 		if (err)
759 			break;
760 		if (block != NULL) {
761 			/*
762 			 * We have found a matching block in the delta base.
763 			 * Attempt to stretch the block as far as possible and
764 			 * generate a copy instruction.
765 			 */
766 			err = stretchblk_file_mem(basedata, basefile_offset0,
767 			    basefile_size, block, f, filesize, &blocklen);
768 			if (err)
769 				break;
770 			err = emitdelta(deltas, &nalloc, ndeltas,
771 			    alloc_chunk_size, 1, block->offset, blocklen);
772 			if (err)
773 				break;
774 		} else {
775 			/*
776 			 * No match.
777 			 * This block needs to be sourced from the file itself.
778 			 */
779 			err = emitdelta(deltas, &nalloc, ndeltas,
780 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
781 			if (err)
782 				break;
783 		}
784 		fileoffset += blocklen;
785 		if (fseeko(f, fileoffset, SEEK_SET) == -1) {
786 			err = got_error_from_errno("fseeko");
787 			break;
788 		}
789 	}
790 
791 	if (err) {
792 		free(*deltas);
793 		*deltas = NULL;
794 		*ndeltas = 0;
795 	}
796 	return err;
797 }
798 
799 const struct got_error *
800 got_deltify_mem_file(struct got_delta_instruction **deltas, int *ndeltas,
801     uint8_t *data, off_t fileoffset, off_t filesize, uint32_t seed,
802     struct got_delta_table *dt, FILE *basefile,
803     off_t basefile_offset0, off_t basefile_size)
804 {
805 	const struct got_error *err = NULL;
806 	const off_t offset0 = fileoffset;
807 	size_t nalloc = 0;
808 	const size_t alloc_chunk_size = 64;
809 
810 	*deltas = NULL;
811 	*ndeltas = 0;
812 
813 	*deltas = reallocarray(NULL, alloc_chunk_size,
814 	    sizeof(struct got_delta_instruction));
815 	if (*deltas == NULL)
816 		return got_error_from_errno("reallocarray");
817 	nalloc = alloc_chunk_size;
818 
819 	while (fileoffset < filesize) {
820 		off_t blocklen;
821 		struct got_delta_block *block;
822 		err = nextblk_mem(&blocklen, data, fileoffset, filesize);
823 		if (err)
824 			break;
825 		if (blocklen == 0) {
826 			/* Source remainder from the file itself. */
827 			if (fileoffset < filesize) {
828 				err = emitdelta(deltas, &nalloc, ndeltas,
829 				    alloc_chunk_size, 0, fileoffset - offset0,
830 				    filesize - fileoffset);
831 			}
832 			break;
833 		}
834 		err = lookupblk(&block, dt, data + fileoffset, blocklen, seed,
835 		    basefile, basefile_offset0);
836 		if (err)
837 			break;
838 		if (block != NULL) {
839 			/*
840 			 * We have found a matching block in the delta base.
841 			 * Attempt to stretch the block as far as possible and
842 			 * generate a copy instruction.
843 			 */
844 			err = stretchblk_mem_file(basefile, basefile_offset0,
845 			    block, data, fileoffset + blocklen, filesize,
846 			    &blocklen);
847 			if (err)
848 				break;
849 			err = emitdelta(deltas, &nalloc, ndeltas,
850 			    alloc_chunk_size, 1, block->offset, blocklen);
851 			if (err)
852 				break;
853 		} else {
854 			/*
855 			 * No match.
856 			 * This block needs to be sourced from the file itself.
857 			 */
858 			err = emitdelta(deltas, &nalloc, ndeltas,
859 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
860 			if (err)
861 				break;
862 		}
863 		fileoffset += blocklen;
864 	}
865 
866 	if (err) {
867 		free(*deltas);
868 		*deltas = NULL;
869 		*ndeltas = 0;
870 	}
871 	return err;
872 }
873 
874 const struct got_error *
875 got_deltify_mem_mem(struct got_delta_instruction **deltas, int *ndeltas,
876     uint8_t *data, off_t fileoffset, off_t filesize, uint32_t seed,
877     struct got_delta_table *dt, uint8_t *basedata,
878     off_t basefile_offset0, off_t basefile_size)
879 {
880 	const struct got_error *err = NULL;
881 	const off_t offset0 = fileoffset;
882 	size_t nalloc = 0;
883 	const size_t alloc_chunk_size = 64;
884 
885 	*deltas = NULL;
886 	*ndeltas = 0;
887 
888 	*deltas = reallocarray(NULL, alloc_chunk_size,
889 	    sizeof(struct got_delta_instruction));
890 	if (*deltas == NULL)
891 		return got_error_from_errno("reallocarray");
892 	nalloc = alloc_chunk_size;
893 
894 	while (fileoffset < filesize) {
895 		off_t blocklen;
896 		struct got_delta_block *block;
897 		err = nextblk_mem(&blocklen, data, fileoffset, filesize);
898 		if (err)
899 			break;
900 		if (blocklen == 0) {
901 			/* Source remainder from the file itself. */
902 			if (fileoffset < filesize) {
903 				err = emitdelta(deltas, &nalloc, ndeltas,
904 				    alloc_chunk_size, 0, fileoffset - offset0,
905 				    filesize - fileoffset);
906 			}
907 			break;
908 		}
909 		err = lookupblk_mem(&block, dt, data + fileoffset, blocklen,
910 		    seed, basedata, basefile_offset0);
911 		if (err)
912 			break;
913 		if (block != NULL) {
914 			/*
915 			 * We have found a matching block in the delta base.
916 			 * Attempt to stretch the block as far as possible and
917 			 * generate a copy instruction.
918 			 */
919 			err = stretchblk_mem_mem(basedata, basefile_offset0,
920 			    basefile_size, block, data, fileoffset + blocklen,
921 			    filesize, &blocklen);
922 			if (err)
923 				break;
924 			err = emitdelta(deltas, &nalloc, ndeltas,
925 			    alloc_chunk_size, 1, block->offset, blocklen);
926 			if (err)
927 				break;
928 		} else {
929 			/*
930 			 * No match.
931 			 * This block needs to be sourced from the file itself.
932 			 */
933 			err = emitdelta(deltas, &nalloc, ndeltas,
934 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
935 			if (err)
936 				break;
937 		}
938 		fileoffset += blocklen;
939 	}
940 
941 	if (err) {
942 		free(*deltas);
943 		*deltas = NULL;
944 		*ndeltas = 0;
945 	}
946 	return err;
947 }