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 #include <sys/queue.h>
20 
21 #include <assert.h>
22 #include <endian.h>
23 #include <errno.h>
24 #include <limits.h>
25 #include <stdint.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
29 
30 #include "got_error.h"
31 
32 #include "got_lib_deltify.h"
33 #include "murmurhash2.h"
34 
35 #ifndef MIN
36 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
37 #endif
38 
39 /*
40  * The algorihm used here is FastCDC (Fast Content-Defined Chunking)
41  * https://www.usenix.org/conference/atc16/technical-sessions/presentation/xia
42  */
43 
44 static const uint32_t geartab[256] = {
45     0x67ed26b7, 0x32da500c, 0x53d0fee0, 0xce387dc7, 0xcd406d90, 0x2e83a4d4,
46     0x9fc9a38d, 0xb67259dc, 0xca6b1722, 0x6d2ea08c, 0x235cea2e, 0x3149bb5f,
47     0x1beda787, 0x2a6b77d5, 0x2f22d9ac, 0x91fc0544, 0xe413acfa, 0x5a30ff7a,
48     0xad6fdde0, 0x444fd0f5, 0x7ad87864, 0x58c5ff05, 0x8d2ec336, 0x2371f853,
49     0x550f8572, 0x6aa448dd, 0x7c9ddbcf, 0x95221e14, 0x2a82ec33, 0xcbec5a78,
50     0xc6795a0d, 0x243995b7, 0x1c909a2f, 0x4fded51c, 0x635d334b, 0x0e2b9999,
51     0x2702968d, 0x856de1d5, 0x3325d60e, 0xeb6a7502, 0xec2a9844, 0x0905835a,
52     0xa1820375, 0xa4be5cab, 0x96a6c058, 0x2c2ccd70, 0xba40fce3, 0xd794c46b,
53     0x8fbae83e, 0xc3aa7899, 0x3d3ff8ed, 0xa0d42b5b, 0x571c0c97, 0xd2811516,
54     0xf7e7b96c, 0x4fd2fcbd, 0xe2fdec94, 0x282cc436, 0x78e8e95c, 0x80a3b613,
55     0xcfbee20c, 0xd4a32d1c, 0x2a12ff13, 0x6af82936, 0xe5630258, 0x8efa6a98,
56     0x294fb2d1, 0xdeb57086, 0x5f0fddb3, 0xeceda7ce, 0x4c87305f, 0x3a6d3307,
57     0xe22d2942, 0x9d060217, 0x1e42ed02, 0xb6f63b52, 0x4367f39f, 0x055cf262,
58     0x03a461b2, 0x5ef9e382, 0x386bc03a, 0x2a1e79c7, 0xf1a0058b, 0xd4d2dea9,
59     0x56baf37d, 0x5daff6cc, 0xf03a951d, 0xaef7de45, 0xa8f4581e, 0x3960b555,
60     0xffbfff6d, 0xbe702a23, 0x8f5b6d6f, 0x061739fb, 0x98696f47, 0x3fd596d4,
61     0x151eac6b, 0xa9fcc4f5, 0x69181a12, 0x3ac5a107, 0xb5198fe7, 0x96bcb1da,
62     0x1b5ddf8e, 0xc757d650, 0x65865c3a, 0x8fc0a41a, 0x87435536, 0x99eda6f2,
63     0x41874794, 0x29cff4e8, 0xb70efd9a, 0x3103f6e7, 0x84d2453b, 0x15a450bd,
64     0x74f49af1, 0x60f664b1, 0xa1c86935, 0xfdafbce1, 0xe36353e3, 0x5d9ba739,
65     0xbc0559ba, 0x708b0054, 0xd41d808c, 0xb2f31723, 0x9027c41f, 0xf136d165,
66     0xb5374b12, 0x9420a6ac, 0x273958b6, 0xe6c2fad0, 0xebdc1f21, 0xfb33af8b,
67     0xc71c25cd, 0xe9a2d8e5, 0xbeb38a50, 0xbceb7cc2, 0x4e4e73f0, 0xcd6c251d,
68     0xde4c032c, 0x4b04ac30, 0x725b8b21, 0x4eb8c33b, 0x20d07b75, 0x0567aa63,
69     0xb56b2bb7, 0xc1f5fd3a, 0xcafd35ca, 0x470dd4da, 0xfe4f94cd, 0xfb8de424,
70     0xe8dbcf40, 0xfe50a37a, 0x62db5b5d, 0xf32f4ab6, 0x2c4a8a51, 0x18473dc0,
71     0xfe0cbb6e, 0xfe399efd, 0xdf34ecc9, 0x6ccd5055, 0x46097073, 0x139135c2,
72     0x721c76f6, 0x1c6a94b4, 0x6eee014d, 0x8a508e02, 0x3da538f5, 0x280d394f,
73     0x5248a0c4, 0x3ce94c6c, 0x9a71ad3a, 0x8493dd05, 0xe43f0ab6, 0x18e4ed42,
74     0x6c5c0e09, 0x42b06ec9, 0x8d330343, 0xa45b6f59, 0x2a573c0c, 0xd7fd3de6,
75     0xeedeab68, 0x5c84dafc, 0xbbd1b1a8, 0xa3ce1ad1, 0x85b70bed, 0xb6add07f,
76     0xa531309c, 0x8f8ab852, 0x564de332, 0xeac9ed0c, 0x73da402c, 0x3ec52761,
77     0x43af2f4d, 0xd6ff45c8, 0x4c367462, 0xd553bd6a, 0x44724855, 0x3b2aa728,
78     0x56e5eb65, 0xeaf16173, 0x33fa42ff, 0xd714bb5d, 0xfbd0a3b9, 0xaf517134,
79     0x9416c8cd, 0x534cf94f, 0x548947c2, 0x34193569, 0x32f4389a, 0xfe7028bc,
80     0xed73b1ed, 0x9db95770, 0x468e3922, 0x0440c3cd, 0x60059a62, 0x33504562,
81     0x2b229fbd, 0x5174dca5, 0xf7028752, 0xd63c6aa8, 0x31276f38, 0x0646721c,
82     0xb0191da8, 0xe00e6de0, 0x9eac1a6e, 0x9f7628a5, 0xed6c06ea, 0x0bb8af15,
83     0xf119fb12, 0x38693c1c, 0x732bc0fe, 0x84953275, 0xb82ec888, 0x33a4f1b3,
84     0x3099835e, 0x028a8782, 0x5fdd51d7, 0xc6c717b3, 0xb06caf71, 0x17c8c111,
85     0x61bad754, 0x9fd03061, 0xe09df1af, 0x3bc9eb73, 0x85878413, 0x9889aaf2,
86     0x3f5a9e46, 0x42c9f01f, 0x9984a4f4, 0xd5de43cc, 0xd294daed, 0xbecba2d2,
87     0xf1f6e72c, 0x5551128a, 0x83af87e2, 0x6f0342ba,
88 };
89 
90 static const struct got_error *addblk(struct got_delta_table *, FILE *,
91     uint8_t *, off_t, off_t, off_t, uint32_t);
92 
93 static uint32_t
94 hashblk(const unsigned char *p, off_t n, uint32_t seed)
95 {
96 	return murmurhash2(p, n, seed);
97 }
98 
99 static const struct got_error *
100 lookup(int *n, struct got_delta_table *dt, FILE *f, off_t file_offset0,
101     off_t len, off_t offset, uint32_t h)
102 {
103 	struct got_delta_block *block;
104 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
105 	uint8_t buf2[GOT_DELTIFY_MAXCHUNK];
106 	size_t r = 0;
107 	int i;
108 
109 	for (i = h % dt->size; dt->offs[i] != 0; i = (i + 1) % dt->size) {
110 		block = &dt->blocks[dt->offs[i] - 1];
111 		if (block->len != len || block->hash != h)
112 			continue;
113 
114 		if (r == 0) {
115 			if (fseeko(f, file_offset0 + offset, SEEK_SET) == -1)
116 				return got_error_from_errno("fseeko");
117 			r = fread(buf, 1, len, f);
118 			if (r != len) {
119 				if (!ferror(f))
120 					break;		/* why? */
121 				return got_ferror(f, GOT_ERR_IO);
122 			}
123 		}
124 
125 		if (fseeko(f, file_offset0 + block->offset, SEEK_SET) == -1)
126 			return got_error_from_errno("fseeko");
127 		if (fread(buf2, 1, len, f) != len)
128 			return got_ferror(f, GOT_ERR_IO);
129 		if (memcmp(buf, buf2, len) == 0)
130 			break;
131 	}
132 
133 	*n = i;
134 	return NULL;
135 }
136 
137 static const struct got_error *
138 lookup_mem(int *n, struct got_delta_table *dt, uint8_t *basedata,
139     off_t basefile_offset0, uint8_t *p, off_t len, uint32_t h)
140 {
141 	struct got_delta_block *block;
142 	uint8_t *d;
143 	int i;
144 
145 	for (i = h % dt->size; dt->offs[i] != 0; i = (i + 1) % dt->size) {
146 		block = &dt->blocks[dt->offs[i] - 1];
147 		if (len == block->len && h == block->hash) {
148 			d = basedata + basefile_offset0 + block->offset;
149 			if (memcmp(d, p, len) == 0)
150 				break;
151 		}
152 	}
153 
154 	*n = i;
155 	return NULL;
156 }
157 
158 static const struct got_error *
159 resizeht(struct got_delta_table *dt, FILE *f, off_t offset0, uint8_t *data,
160     off_t len)
161 {
162 	const struct got_error *err;
163 	struct got_delta_block *b;
164 	size_t newsize, oldsize;
165 	int i, n;
166 
167 	if (dt->nblocks == dt->nalloc) {
168 		newsize = dt->nalloc + 256;
169 		b = reallocarray(dt->blocks, newsize, sizeof(*dt->blocks));
170 		if (b == NULL)
171 			return got_error_from_errno("reallocarray");
172 		dt->blocks = b;
173 		dt->nalloc = newsize;
174 	}
175 
176 	if (dt->blocks == NULL || dt->len * 100 / dt->size > 70) {
177 		oldsize = dt->size;
178 		dt->size *= 2;
179 		free(dt->offs);
180 		dt->offs = calloc(dt->size, sizeof(*dt->offs));
181 		if (dt->offs == NULL) {
182 			dt->size = oldsize;
183 			return got_error_from_errno("calloc");
184 		}
185 
186 		for (i = 0; i < dt->nblocks; ++i) {
187 			b = &dt->blocks[i];
188 			if (f)
189 				err = lookup(&n, dt, f, offset0, b->len,
190 				    b->offset, b->hash);
191 			else
192 				err = lookup_mem(&n, dt, data, offset0,
193 				    data + b->offset, b->len, b->hash);
194 			if (err) {
195 				free(dt->offs);
196 				dt->offs = NULL;
197 				dt->size = oldsize;
198 				return err;
199 			}
200 			assert(dt->offs[n] == 0);
201 			dt->offs[n] = i + 1;
202 		}
203 	}
204 
205 	return NULL;
206 }
207 
208 static const struct got_error *
209 addblk(struct got_delta_table *dt, FILE *f, uint8_t *data, off_t file_offset0,
210     off_t len, off_t offset, uint32_t h)
211 {
212 	const struct got_error *err;
213 	struct got_delta_block *block;
214 	int i;
215 
216 	if (len == 0)
217 		return NULL;
218 
219 	err = resizeht(dt, f, file_offset0, data, len);
220 	if (err)
221 		return err;
222 
223 	if (f)
224 		err = lookup(&i, dt, f, file_offset0, len, offset, h);
225 	else
226 		err = lookup_mem(&i, dt, data, file_offset0, data + offset,
227 		    len, h);
228 	if (err)
229 		return err;
230 
231 	if (dt->offs[i] != 0)
232 		return NULL;	/* found */
233 
234 	dt->offs[i] = dt->nblocks + 1;
235 	block = &dt->blocks[dt->nblocks];
236 	block->len = len;
237 	block->offset = offset;
238 	block->hash = h;
239 
240 	dt->nblocks++;
241 	dt->len++;
242 
243 	return NULL;
244 }
245 
246 static const struct got_error *
247 lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
248     unsigned char *p, off_t len, uint32_t seed, FILE *basefile,
249     off_t basefile_offset0)
250 {
251 	int i;
252 	uint32_t h;
253 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
254 	struct got_delta_block *b;
255 	size_t r;
256 
257 	*block = NULL;
258 
259 	h = hashblk(p, len, seed);
260 	
261 	for (i = h % dt->size; dt->offs[i] != 0; i = (i + 1) % dt->size) {
262 		b = &dt->blocks[dt->offs[i] - 1];
263 		if (b->hash != h || b->len != len)
264 			continue;
265 		if (fseeko(basefile, basefile_offset0 + b->offset,
266 		    SEEK_SET) == -1)
267 			return got_error_from_errno("fseeko");
268 		r = fread(buf, 1, len, basefile);
269 		if (r != len)
270 			return got_ferror(basefile, GOT_ERR_IO);
271 		if (memcmp(p, buf, len) == 0) {
272 			*block = b;
273 			break;
274 		}
275 	}
276 	return NULL;
277 }
278 
279 static const struct got_error *
280 lookupblk_mem(struct got_delta_block **block, struct got_delta_table *dt,
281     unsigned char *p, off_t len, uint32_t seed, uint8_t *basedata,
282     off_t basefile_offset0)
283 {
284 	const struct got_error *err;
285 	int n;
286 	uint32_t h;
287 
288 	*block = NULL;
289 	h = hashblk(p, len, seed);
290 	err = lookup_mem(&n, dt, basedata, basefile_offset0, p, len, h);
291 	if (err)
292 		return err;
293 	if (dt->offs[n] != 0)
294 		*block = &dt->blocks[dt->offs[n] - 1];
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, uint32_t seed)
358 {
359 	const struct got_error *err = NULL;
360 	uint32_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 	(*dt)->size = 128;
376 	(*dt)->offs = calloc((*dt)->size, sizeof(*(*dt)->offs));
377 	if ((*dt)->offs == NULL) {
378 		err = got_error_from_errno("calloc");
379 		goto done;
380 	}
381 
382 	if (fseeko(f, fileoffset, SEEK_SET) == -1)
383 		return got_error_from_errno("fseeko");
384 
385 	while (fileoffset < filesize) {
386 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
387 		off_t blocklen;
388 		err = nextblk(buf, &blocklen, f);
389 		if (err)
390 			goto done;
391 		if (blocklen == 0)
392 			break;
393 		h = hashblk(buf, blocklen, seed);
394 		err = addblk(*dt, f, NULL, offset0, blocklen,
395 		    fileoffset - offset0, h);
396 		if (err)
397 			goto done;
398 		fileoffset += blocklen;
399 		if (fseeko(f, fileoffset, SEEK_SET) == -1)
400 			return got_error_from_errno("fseeko");
401 	}
402 done:
403 	if (err) {
404 		free((*dt)->blocks);
405 		free(*dt);
406 		*dt = NULL;
407 	}
408 
409 	return err;
410 }
411 
412 const struct got_error *
413 got_deltify_init_mem(struct got_delta_table **dt, uint8_t *data,
414     off_t fileoffset, off_t filesize, uint32_t seed)
415 {
416 	const struct got_error *err = NULL;
417 	uint32_t h;
418 	const off_t offset0 = fileoffset;
419 
420 	*dt = calloc(1, sizeof(**dt));
421 	if (*dt == NULL)
422 		return got_error_from_errno("calloc");
423 
424 	(*dt)->nblocks = 0;
425 	(*dt)->nalloc = 128;
426 	(*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
427 	if ((*dt)->blocks == NULL) {
428 		err = got_error_from_errno("calloc");
429 		goto done;
430 	}
431 
432 	(*dt)->size = 128;
433 	(*dt)->offs = calloc((*dt)->size, sizeof(*(*dt)->offs));
434 	if ((*dt)->offs == NULL) {
435 		err = got_error_from_errno("calloc");
436 		goto done;
437 	}
438 
439 	while (fileoffset < filesize) {
440 		off_t blocklen;
441 		err = nextblk_mem(&blocklen, data, fileoffset, filesize);
442 		if (err)
443 			goto done;
444 		if (blocklen == 0)
445 			break;
446 		h = hashblk(data + fileoffset, blocklen, seed);
447 		err = addblk(*dt, NULL, data, offset0, blocklen,
448 		    fileoffset - offset0, h);
449 		if (err)
450 			goto done;
451 		fileoffset += blocklen;
452 	}
453 done:
454 	if (err) {
455 		free((*dt)->blocks);
456 		free(*dt);
457 		*dt = NULL;
458 	}
459 
460 	return err;
461 }
462 
463 void
464 got_deltify_free(struct got_delta_table *dt)
465 {
466 	if (dt == NULL)
467 		return;
468 	free(dt->blocks);
469 	free(dt->offs);
470 	free(dt);
471 }
472 
473 static const struct got_error *
474 emitdelta(struct got_delta_instruction **deltas, size_t *nalloc, int *ndeltas,
475     const size_t alloc_chunk_size, int copy, off_t offset, off_t len)
476 {
477 	struct got_delta_instruction *d, *p;
478 
479 	if (*nalloc < *ndeltas + alloc_chunk_size) {
480 		p = reallocarray(*deltas, *nalloc + alloc_chunk_size,
481 		    sizeof(struct got_delta_instruction));
482 		if (p == NULL)
483 			return got_error_from_errno("reallocarray");
484 		*deltas = p;
485 		*nalloc += alloc_chunk_size;
486 	}
487 	*ndeltas += 1;
488 	d = &(*deltas)[*ndeltas - 1];
489 	d->copy = copy;
490 	d->offset = offset;
491 	d->len = len;
492 	return NULL;
493 }
494 
495 static const struct got_error *
496 stretchblk(FILE *basefile, off_t base_offset0, struct got_delta_block *block,
497     FILE *f, off_t filesize, off_t *blocklen)
498 {
499 	uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
500 	size_t base_r, r, i;
501 	int buf_equal = 1;
502 
503 	if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
504 	    SEEK_SET) == -1)
505 		return got_error_from_errno("fseeko");
506 
507 	while (buf_equal && *blocklen < (1 << 24) - 1) {
508 		base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
509 		if (base_r == 0) {
510 			if (ferror(basefile))
511 				return got_ferror(basefile, GOT_ERR_IO);
512 			break;
513 		}
514 		r = fread(buf, 1, sizeof(buf), f);
515 		if (r == 0) {
516 			if (ferror(f))
517 				return got_ferror(f, GOT_ERR_IO);
518 			break;
519 		}
520 		for (i = 0; i < MIN(base_r, r); i++) {
521 			if (buf[i] != basebuf[i]) {
522 				buf_equal = 0;
523 				break;
524 			}
525 			(*blocklen)++;
526 		}
527 	}
528 
529 	return NULL;
530 }
531 
532 static const struct got_error *
533 stretchblk_file_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
534      struct got_delta_block *block, FILE *f, off_t filesize, off_t *blocklen)
535 {
536 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
537 	size_t r, i;
538 	int buf_equal = 1;
539 	off_t base_offset = base_offset0 + block->offset + *blocklen;
540 
541 	if (base_offset > basefile_size) {
542 		return got_error_fmt(GOT_ERR_RANGE,
543 		    "read beyond the size of delta base at offset %lld",
544 		    (long long)base_offset);
545 	}
546 
547 	while (buf_equal && *blocklen < (1 << 24) - 1) {
548 		if (base_offset + *blocklen >= basefile_size)
549 			break;
550 		r = fread(buf, 1, sizeof(buf), f);
551 		if (r == 0) {
552 			if (ferror(f))
553 				return got_ferror(f, GOT_ERR_IO);
554 			break;
555 		}
556 		for (i = 0; i < MIN(basefile_size - base_offset, r); i++) {
557 			if (buf[i] != *(basedata + base_offset + i)) {
558 				buf_equal = 0;
559 				break;
560 			}
561 			(*blocklen)++;
562 		}
563 	}
564 
565 	return NULL;
566 }
567 
568 static const struct got_error *
569 stretchblk_mem_file(FILE *basefile, off_t base_offset0,
570     struct got_delta_block *block, uint8_t *data, off_t fileoffset,
571     off_t filesize, off_t *blocklen)
572 {
573 	uint8_t basebuf[GOT_DELTIFY_MAXCHUNK];
574 	size_t base_r, i;
575 	int buf_equal = 1;
576 
577 	if (fileoffset > filesize) {
578 		return got_error_fmt(GOT_ERR_RANGE,
579 		    "read beyond the size of deltify file at offset %lld",
580 		    (long long)fileoffset);
581 	}
582 
583 	if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
584 	    SEEK_SET) == -1)
585 		return got_error_from_errno("fseeko");
586 
587 	while (buf_equal && *blocklen < (1 << 24) - 1) {
588 		if (fileoffset + *blocklen >= filesize)
589 			break;
590 		base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
591 		if (base_r == 0) {
592 			if (ferror(basefile))
593 				return got_ferror(basefile, GOT_ERR_IO);
594 			break;
595 		}
596 		for (i = 0; i < MIN(base_r, filesize - fileoffset); i++) {
597 			if (*(data + fileoffset + i) != basebuf[i]) {
598 				buf_equal = 0;
599 				break;
600 			}
601 			(*blocklen)++;
602 		}
603 	}
604 
605 	return NULL;
606 }
607 
608 static const struct got_error *
609 stretchblk_mem_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
610     struct got_delta_block *block, uint8_t *data, off_t fileoffset,
611     off_t filesize, off_t *blocklen)
612 {
613 	off_t i, maxlen;
614 	off_t base_offset = base_offset0 + block->offset + *blocklen;
615 	uint8_t *p, *q;
616 
617 	if (base_offset > basefile_size) {
618 		return got_error_fmt(GOT_ERR_RANGE,
619 		    "read beyond the size of delta base at offset %lld",
620 		    (long long)base_offset);
621 	}
622 
623 	if (fileoffset > filesize) {
624 		return got_error_fmt(GOT_ERR_RANGE,
625 		    "read beyond the size of deltify file at offset %lld",
626 		    (long long)fileoffset);
627 	}
628 
629 	p = data + fileoffset;
630 	q = basedata + base_offset;
631 	maxlen = MIN(basefile_size - base_offset, filesize - fileoffset);
632 	for (i = 0; i < maxlen && *blocklen < (1 << 24) - 1; i++) {
633 		if (p[i] != q[i])
634 			break;
635 		(*blocklen)++;
636 	}
637 
638 	return NULL;
639 }
640 
641 const struct got_error *
642 got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
643     FILE *f, off_t fileoffset, off_t filesize, uint32_t seed,
644     struct got_delta_table *dt, FILE *basefile,
645     off_t basefile_offset0, off_t basefile_size)
646 {
647 	const struct got_error *err = NULL;
648 	const off_t offset0 = fileoffset;
649 	size_t nalloc = 0;
650 	const size_t alloc_chunk_size = 64;
651 
652 	*deltas = NULL;
653 	*ndeltas = 0;
654 
655 	/*
656 	 * offset0 indicates where data to be deltified begins.
657 	 * For example, we want to avoid deltifying a Git object header at
658 	 * the beginning of the file.
659 	 */
660 	if (fseeko(f, offset0, SEEK_SET) == -1)
661 		return got_error_from_errno("fseeko");
662 
663 	*deltas = reallocarray(NULL, alloc_chunk_size,
664 	    sizeof(struct got_delta_instruction));
665 	if (*deltas == NULL)
666 		return got_error_from_errno("reallocarray");
667 	nalloc = alloc_chunk_size;
668 
669 	while (fileoffset < filesize) {
670 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
671 		off_t blocklen;
672 		struct got_delta_block *block;
673 		err = nextblk(buf, &blocklen, f);
674 		if (err)
675 			break;
676 		if (blocklen == 0) {
677 			/* Source remainder from the file itself. */
678 			if (fileoffset < filesize) {
679 				err = emitdelta(deltas, &nalloc, ndeltas,
680 				    alloc_chunk_size, 0, fileoffset - offset0,
681 				    filesize - fileoffset);
682 			}
683 			break;
684 		}
685 		err = lookupblk(&block, dt, buf, blocklen, seed, basefile,
686 		    basefile_offset0);
687 		if (err)
688 			break;
689 		if (block != NULL) {
690 			/*
691 			 * We have found a matching block in the delta base.
692 			 * Attempt to stretch the block as far as possible and
693 			 * generate a copy instruction.
694 			 */
695 			err = stretchblk(basefile, basefile_offset0, block,
696 			    f, filesize, &blocklen);
697 			if (err)
698 				break;
699 			err = emitdelta(deltas, &nalloc, ndeltas,
700 			    alloc_chunk_size, 1, block->offset, blocklen);
701 			if (err)
702 				break;
703 		} else {
704 			/*
705 			 * No match.
706 			 * This block needs to be sourced from the file itself.
707 			 */
708 			err = emitdelta(deltas, &nalloc, ndeltas,
709 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
710 			if (err)
711 				break;
712 		}
713 		fileoffset += blocklen;
714 		if (fseeko(f, fileoffset, SEEK_SET) == -1) {
715 			err = got_error_from_errno("fseeko");
716 			break;
717 		}
718 	}
719 
720 	if (err) {
721 		free(*deltas);
722 		*deltas = NULL;
723 		*ndeltas = 0;
724 	}
725 	return err;
726 }
727 
728 const struct got_error *
729 got_deltify_file_mem(struct got_delta_instruction **deltas, int *ndeltas,
730     FILE *f, off_t fileoffset, off_t filesize, uint32_t seed,
731     struct got_delta_table *dt, uint8_t *basedata,
732     off_t basefile_offset0, off_t basefile_size)
733 {
734 	const struct got_error *err = NULL;
735 	const off_t offset0 = fileoffset;
736 	size_t nalloc = 0;
737 	const size_t alloc_chunk_size = 64;
738 
739 	*deltas = NULL;
740 	*ndeltas = 0;
741 
742 	/*
743 	 * offset0 indicates where data to be deltified begins.
744 	 * For example, we want to avoid deltifying a Git object header at
745 	 * the beginning of the file.
746 	 */
747 	if (fseeko(f, offset0, SEEK_SET) == -1)
748 		return got_error_from_errno("fseeko");
749 
750 	*deltas = reallocarray(NULL, alloc_chunk_size,
751 	    sizeof(struct got_delta_instruction));
752 	if (*deltas == NULL)
753 		return got_error_from_errno("reallocarray");
754 	nalloc = alloc_chunk_size;
755 
756 	while (fileoffset < filesize) {
757 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
758 		off_t blocklen;
759 		struct got_delta_block *block;
760 		err = nextblk(buf, &blocklen, f);
761 		if (err)
762 			break;
763 		if (blocklen == 0) {
764 			/* Source remainder from the file itself. */
765 			if (fileoffset < filesize) {
766 				err = emitdelta(deltas, &nalloc, ndeltas,
767 				    alloc_chunk_size, 0, fileoffset - offset0,
768 				    filesize - fileoffset);
769 			}
770 			break;
771 		}
772 		err = lookupblk_mem(&block, dt, buf, blocklen, seed, basedata,
773 		    basefile_offset0);
774 		if (err)
775 			break;
776 		if (block != NULL) {
777 			/*
778 			 * We have found a matching block in the delta base.
779 			 * Attempt to stretch the block as far as possible and
780 			 * generate a copy instruction.
781 			 */
782 			err = stretchblk_file_mem(basedata, basefile_offset0,
783 			    basefile_size, block, f, filesize, &blocklen);
784 			if (err)
785 				break;
786 			err = emitdelta(deltas, &nalloc, ndeltas,
787 			    alloc_chunk_size, 1, block->offset, blocklen);
788 			if (err)
789 				break;
790 		} else {
791 			/*
792 			 * No match.
793 			 * This block needs to be sourced from the file itself.
794 			 */
795 			err = emitdelta(deltas, &nalloc, ndeltas,
796 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
797 			if (err)
798 				break;
799 		}
800 		fileoffset += blocklen;
801 		if (fseeko(f, fileoffset, SEEK_SET) == -1) {
802 			err = got_error_from_errno("fseeko");
803 			break;
804 		}
805 	}
806 
807 	if (err) {
808 		free(*deltas);
809 		*deltas = NULL;
810 		*ndeltas = 0;
811 	}
812 	return err;
813 }
814 
815 const struct got_error *
816 got_deltify_mem_file(struct got_delta_instruction **deltas, int *ndeltas,
817     uint8_t *data, off_t fileoffset, off_t filesize, uint32_t seed,
818     struct got_delta_table *dt, FILE *basefile,
819     off_t basefile_offset0, off_t basefile_size)
820 {
821 	const struct got_error *err = NULL;
822 	const off_t offset0 = fileoffset;
823 	size_t nalloc = 0;
824 	const size_t alloc_chunk_size = 64;
825 
826 	*deltas = NULL;
827 	*ndeltas = 0;
828 
829 	*deltas = reallocarray(NULL, alloc_chunk_size,
830 	    sizeof(struct got_delta_instruction));
831 	if (*deltas == NULL)
832 		return got_error_from_errno("reallocarray");
833 	nalloc = alloc_chunk_size;
834 
835 	while (fileoffset < filesize) {
836 		off_t blocklen;
837 		struct got_delta_block *block;
838 		err = nextblk_mem(&blocklen, data, fileoffset, filesize);
839 		if (err)
840 			break;
841 		if (blocklen == 0) {
842 			/* Source remainder from the file itself. */
843 			if (fileoffset < filesize) {
844 				err = emitdelta(deltas, &nalloc, ndeltas,
845 				    alloc_chunk_size, 0, fileoffset - offset0,
846 				    filesize - fileoffset);
847 			}
848 			break;
849 		}
850 		err = lookupblk(&block, dt, data + fileoffset, blocklen, seed,
851 		    basefile, basefile_offset0);
852 		if (err)
853 			break;
854 		if (block != NULL) {
855 			/*
856 			 * We have found a matching block in the delta base.
857 			 * Attempt to stretch the block as far as possible and
858 			 * generate a copy instruction.
859 			 */
860 			err = stretchblk_mem_file(basefile, basefile_offset0,
861 			    block, data, fileoffset + blocklen, filesize,
862 			    &blocklen);
863 			if (err)
864 				break;
865 			err = emitdelta(deltas, &nalloc, ndeltas,
866 			    alloc_chunk_size, 1, block->offset, blocklen);
867 			if (err)
868 				break;
869 		} else {
870 			/*
871 			 * No match.
872 			 * This block needs to be sourced from the file itself.
873 			 */
874 			err = emitdelta(deltas, &nalloc, ndeltas,
875 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
876 			if (err)
877 				break;
878 		}
879 		fileoffset += blocklen;
880 	}
881 
882 	if (err) {
883 		free(*deltas);
884 		*deltas = NULL;
885 		*ndeltas = 0;
886 	}
887 	return err;
888 }
889 
890 const struct got_error *
891 got_deltify_mem_mem(struct got_delta_instruction **deltas, int *ndeltas,
892     uint8_t *data, off_t fileoffset, off_t filesize, uint32_t seed,
893     struct got_delta_table *dt, uint8_t *basedata,
894     off_t basefile_offset0, off_t basefile_size)
895 {
896 	const struct got_error *err = NULL;
897 	const off_t offset0 = fileoffset;
898 	size_t nalloc = 0;
899 	const size_t alloc_chunk_size = 64;
900 
901 	*deltas = NULL;
902 	*ndeltas = 0;
903 
904 	*deltas = reallocarray(NULL, alloc_chunk_size,
905 	    sizeof(struct got_delta_instruction));
906 	if (*deltas == NULL)
907 		return got_error_from_errno("reallocarray");
908 	nalloc = alloc_chunk_size;
909 
910 	while (fileoffset < filesize) {
911 		off_t blocklen;
912 		struct got_delta_block *block;
913 		err = nextblk_mem(&blocklen, data, fileoffset, filesize);
914 		if (err)
915 			break;
916 		if (blocklen == 0) {
917 			/* Source remainder from the file itself. */
918 			if (fileoffset < filesize) {
919 				err = emitdelta(deltas, &nalloc, ndeltas,
920 				    alloc_chunk_size, 0, fileoffset - offset0,
921 				    filesize - fileoffset);
922 			}
923 			break;
924 		}
925 		err = lookupblk_mem(&block, dt, data + fileoffset, blocklen,
926 		    seed, basedata, basefile_offset0);
927 		if (err)
928 			break;
929 		if (block != NULL) {
930 			/*
931 			 * We have found a matching block in the delta base.
932 			 * Attempt to stretch the block as far as possible and
933 			 * generate a copy instruction.
934 			 */
935 			err = stretchblk_mem_mem(basedata, basefile_offset0,
936 			    basefile_size, block, data, fileoffset + blocklen,
937 			    filesize, &blocklen);
938 			if (err)
939 				break;
940 			err = emitdelta(deltas, &nalloc, ndeltas,
941 			    alloc_chunk_size, 1, block->offset, blocklen);
942 			if (err)
943 				break;
944 		} else {
945 			/*
946 			 * No match.
947 			 * This block needs to be sourced from the file itself.
948 			 */
949 			err = emitdelta(deltas, &nalloc, ndeltas,
950 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
951 			if (err)
952 				break;
953 		}
954 		fileoffset += blocklen;
955 	}
956 
957 	if (err) {
958 		free(*deltas);
959 		*deltas = NULL;
960 		*ndeltas = 0;
961 	}
962 	return err;
963 }