op public repos

Blob

Date:: Fri Jan 7 23:32:27 2022 UTC
Message:: map raw object files into memory while packing if possible
Actions:: History | Blame | Raw File
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 #include <sha1.h>
30 
31 #include "got_error.h"
32 
33 #include "got_lib_deltify.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 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 uint64_t
91 hashblk(const unsigned char *p, off_t n)
92 {
93 	unsigned char buf[SHA1_DIGEST_LENGTH];
94 	uint64_t h;
95 	SHA1_CTX ctx;
96 	SHA1Init(&ctx);
97 	SHA1Update(&ctx, p, n);
98 	SHA1Final(buf, &ctx);
99 	memcpy(&h, buf, sizeof(h));
100 	return be64toh(h);
101 }
102 
103 static const struct got_error *
104 addblk(struct got_delta_table *dt, FILE *f, off_t file_offset0, off_t len,
105     off_t offset, uint64_t h)
106 {
107 	const struct got_error *err = NULL;
108 	int i;
109 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
110 	uint8_t buf2[GOT_DELTIFY_MAXCHUNK];
111 	size_t r = 0;
112 
113 	if (len == 0)
114 		return NULL;
115 
116 	i = h % dt->nalloc;
117 	while (dt->blocks[i].len != 0) {
118 		/*
119 		 * Avoid adding duplicate blocks.
120 		 * NB: A matching hash is insufficient for detecting equality.
121 		 * The hash can only detect inequality.
122 		 */
123 		if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
124 			if (r == 0) {
125 				if (fseeko(f, file_offset0 + offset, SEEK_SET) == -1)
126 					return got_error_from_errno("fseeko");
127 				r = fread(buf, 1, len, f);
128 				if (r != len) {
129 					if (!ferror(f))
130 						return NULL;
131 					return got_ferror(f, GOT_ERR_IO);
132 				}
133 			}
134 			if (fseeko(f, file_offset0 + dt->blocks[i].offset,
135 			    SEEK_SET) == -1)
136 				return got_error_from_errno("fseeko");
137 			if (fread(buf2, 1, len, f) != len)
138 				return got_ferror(f, GOT_ERR_IO);
139 			if (memcmp(buf, buf2, len) == 0)
140 				return NULL;
141 		}
142 
143 		i = (i + 1) % dt->nalloc;
144 	}
145 	assert(dt->blocks[i].len == 0);
146 	dt->blocks[i].len = len;
147 	dt->blocks[i].offset = offset;
148 	dt->blocks[i].hash = h;
149 	dt->nblocks++;
150 	if (dt->nalloc < dt->nblocks + 64) {
151 		struct got_delta_block *db;
152 		size_t old_size = dt->nalloc;
153 		db = dt->blocks;
154 		dt->blocks = calloc(dt->nalloc + 64,
155 		    sizeof(struct got_delta_block));
156 		if (dt->blocks == NULL) {
157 			err = got_error_from_errno("calloc");
158 			dt->blocks = db;
159 			return err;
160 		}
161 		dt->nalloc += 64;
162 		/*
163 		 * Recompute all block positions. Hash-based indices of blocks
164 		 * in the array depend on the allocated length of the array.
165 		 */
166 		dt->nblocks = 0;
167 		for (i = 0; i < old_size; i++) {
168 			if (db[i].len == 0)
169 				continue;
170 			err = addblk(dt, f, file_offset0, db[i].len,
171 			    db[i].offset, db[i].hash);
172 			if (err)
173 				break;
174 		}
175 		free(db);
176 	}
177 
178 	return err;
179 }
180 
181 static const struct got_error *
182 addblk_mem(struct got_delta_table *dt, uint8_t *data, off_t file_offset0,
183     off_t len, off_t offset, uint64_t h)
184 {
185 	const struct got_error *err = NULL;
186 	int i;
187 	uint8_t *block1;
188 	uint8_t *block2;
189 
190 	if (len == 0)
191 		return NULL;
192 
193 	i = h % dt->nalloc;
194 	while (dt->blocks[i].len != 0) {
195 		/*
196 		 * Avoid adding duplicate blocks.
197 		 * NB: A matching hash is insufficient for detecting equality.
198 		 * The hash can only detect inequality.
199 		 */
200 		if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
201 			block1 = data + file_offset0 + dt->blocks[i].offset;
202 			block2 = data + file_offset0 + offset;
203 			if (memcmp(block1, block2, len) == 0)
204 				return NULL;
205 		}
206 
207 		i = (i + 1) % dt->nalloc;
208 	}
209 	assert(dt->blocks[i].len == 0);
210 	dt->blocks[i].len = len;
211 	dt->blocks[i].offset = offset;
212 	dt->blocks[i].hash = h;
213 	dt->nblocks++;
214 	if (dt->nalloc < dt->nblocks + 64) {
215 		struct got_delta_block *db;
216 		size_t old_size = dt->nalloc;
217 		db = dt->blocks;
218 		dt->blocks = calloc(dt->nalloc + 64,
219 		    sizeof(struct got_delta_block));
220 		if (dt->blocks == NULL) {
221 			err = got_error_from_errno("calloc");
222 			dt->blocks = db;
223 			return err;
224 		}
225 		dt->nalloc += 64;
226 		/*
227 		 * Recompute all block positions. Hash-based indices of blocks
228 		 * in the array depend on the allocated length of the array.
229 		 */
230 		dt->nblocks = 0;
231 		for (i = 0; i < old_size; i++) {
232 			if (db[i].len == 0)
233 				continue;
234 			err = addblk_mem(dt, data, file_offset0, db[i].len,
235 			    db[i].offset, db[i].hash);
236 			if (err)
237 				break;
238 		}
239 		free(db);
240 	}
241 
242 	return err;
243 }
244 
245 static const struct got_error *
246 lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
247     unsigned char *p, off_t len, FILE *basefile, off_t basefile_offset0)
248 {
249 	int i;
250 	uint64_t h;
251 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
252 	size_t r;
253 
254 	*block = NULL;
255 
256 	h = hashblk(p, len);
257 	for (i = h % dt->nalloc; dt->blocks[i].len != 0;
258 	     i = (i + 1) % dt->nalloc) {
259 		if (dt->blocks[i].hash != h ||
260 		    dt->blocks[i].len != len)
261 			continue;
262 		if (fseeko(basefile, basefile_offset0 + dt->blocks[i].offset,
263 		    SEEK_SET) == -1)
264 			return got_error_from_errno("fseeko");
265 		r = fread(buf, 1, len, basefile);
266 		if (r != len)
267 			return got_ferror(basefile, GOT_ERR_IO);
268 		if (memcmp(p, buf, len) == 0) {
269 			*block = &dt->blocks[i];
270 			break;
271 		}
272 	}
273 	return NULL;
274 }
275 
276 static const struct got_error *
277 lookupblk_mem(struct got_delta_block **block, struct got_delta_table *dt,
278     unsigned char *p, off_t len, uint8_t *basedata, off_t basefile_offset0)
279 {
280 	int i;
281 	uint64_t h;
282 	uint8_t *b;
283 
284 	*block = NULL;
285 
286 	h = hashblk(p, len);
287 	for (i = h % dt->nalloc; dt->blocks[i].len != 0;
288 	     i = (i + 1) % dt->nalloc) {
289 		if (dt->blocks[i].hash != h ||
290 		    dt->blocks[i].len != len)
291 			continue;
292 		b = basedata + basefile_offset0 + dt->blocks[i].offset;
293 		if (memcmp(p, b, len) == 0) {
294 			*block = &dt->blocks[i];
295 			break;
296 		}
297 	}
298 	return NULL;
299 }
300 
301 static const struct got_error *
302 nextblk(uint8_t *buf, off_t *blocklen, FILE *f)
303 {
304 	uint32_t gh;
305 	const unsigned char *p;
306 	size_t r;
307 	off_t pos = ftello(f);
308 
309 	*blocklen = 0;
310 
311 	r = fread(buf, 1, GOT_DELTIFY_MAXCHUNK, f);
312 	if (r == 0 && ferror(f))
313 		return got_ferror(f, GOT_ERR_IO);
314 	if (r < GOT_DELTIFY_MINCHUNK)
315 		return NULL; /* no more delta-worthy blocks left */
316 
317 	/* Got a deltifiable block. Find the split-point where it ends. */
318 	p = buf + GOT_DELTIFY_MINCHUNK;
319 	gh = 0;
320 	while (p != buf + r) {
321 		gh = (gh << 1) + geartab[*p++];
322 		if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
323 			break;
324 	}
325 
326 	*blocklen = (p - buf);
327 	if (fseeko(f, pos + *blocklen, SEEK_SET) == -1)
328 		return got_error_from_errno("fseeko");
329 
330 	return NULL;
331 }
332 
333 static const struct got_error *
334 nextblk_mem(off_t *blocklen, uint8_t *data, off_t fileoffset, off_t filesize)
335 {
336 	uint32_t gh;
337 	const unsigned char *p;
338 
339 	*blocklen = 0;
340 
341 	if (fileoffset >= filesize ||
342 	    filesize - fileoffset < GOT_DELTIFY_MINCHUNK)
343 		return NULL; /* no more delta-worthy blocks left */
344 
345 	/* Got a deltifiable block. Find the split-point where it ends. */
346 	p = data + fileoffset + GOT_DELTIFY_MINCHUNK;
347 	gh = 0;
348 	while (p != data + MIN(fileoffset + GOT_DELTIFY_MAXCHUNK, filesize)) {
349 		gh = (gh << 1) + geartab[*p++];
350 		if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
351 			break;
352 	}
353 
354 	*blocklen = (p - (data + fileoffset));
355 	return NULL;
356 }
357 
358 const struct got_error *
359 got_deltify_init(struct got_delta_table **dt, FILE *f, off_t fileoffset,
360     off_t filesize)
361 {
362 	const struct got_error *err = NULL;
363 	uint64_t h;
364 	const off_t offset0 = fileoffset;
365 
366 	*dt = calloc(1, sizeof(**dt));
367 	if (*dt == NULL)
368 		return got_error_from_errno("calloc");
369 
370 	(*dt)->nblocks = 0;
371 	(*dt)->nalloc = 128;
372 	(*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
373 	if ((*dt)->blocks == NULL) {
374 		err = got_error_from_errno("calloc");
375 		goto done;
376 	}
377 
378 	if (fseeko(f, fileoffset, SEEK_SET) == -1)
379 		return got_error_from_errno("fseeko");
380 
381 	while (fileoffset < filesize) {
382 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
383 		off_t blocklen;
384 		err = nextblk(buf, &blocklen, f);
385 		if (err)
386 			goto done;
387 		if (blocklen == 0)
388 			break;
389 		h = hashblk(buf, blocklen);
390 		err = addblk(*dt, f, offset0, blocklen,
391 		    fileoffset - offset0, h);
392 		if (err)
393 			goto done;
394 		fileoffset += blocklen;
395 		if (fseeko(f, fileoffset, SEEK_SET) == -1)
396 			return got_error_from_errno("fseeko");
397 	}
398 done:
399 	if (err) {
400 		free((*dt)->blocks);
401 		free(*dt);
402 		*dt = NULL;
403 	}
404 
405 	return err;
406 }
407 
408 const struct got_error *
409 got_deltify_init_mem(struct got_delta_table **dt, uint8_t *data,
410     off_t fileoffset, off_t filesize)
411 {
412 	const struct got_error *err = NULL;
413 	uint64_t h;
414 	const off_t offset0 = fileoffset;
415 
416 	*dt = calloc(1, sizeof(**dt));
417 	if (*dt == NULL)
418 		return got_error_from_errno("calloc");
419 
420 	(*dt)->nblocks = 0;
421 	(*dt)->nalloc = 128;
422 	(*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
423 	if ((*dt)->blocks == NULL) {
424 		err = got_error_from_errno("calloc");
425 		goto done;
426 	}
427 
428 	while (fileoffset < filesize) {
429 		off_t blocklen;
430 		err = nextblk_mem(&blocklen, data, fileoffset, filesize);
431 		if (err)
432 			goto done;
433 		if (blocklen == 0)
434 			break;
435 		h = hashblk(data + fileoffset, blocklen);
436 		err = addblk_mem(*dt, data, offset0, blocklen,
437 		    fileoffset - offset0, h);
438 		if (err)
439 			goto done;
440 		fileoffset += blocklen;
441 	}
442 done:
443 	if (err) {
444 		free((*dt)->blocks);
445 		free(*dt);
446 		*dt = NULL;
447 	}
448 
449 	return err;
450 }
451 
452 void
453 got_deltify_free(struct got_delta_table *dt)
454 {
455 	if (dt == NULL)
456 		return;
457 	free(dt->blocks);
458 	free(dt);
459 }
460 
461 static const struct got_error *
462 emitdelta(struct got_delta_instruction **deltas, size_t *nalloc, int *ndeltas,
463     const size_t alloc_chunk_size, int copy, off_t offset, off_t len)
464 {
465 	struct got_delta_instruction *d, *p;
466 
467 	if (*nalloc < *ndeltas + alloc_chunk_size) {
468 		p = reallocarray(*deltas, *nalloc + alloc_chunk_size,
469 		    sizeof(struct got_delta_instruction));
470 		if (p == NULL)
471 			return got_error_from_errno("reallocarray");
472 		*deltas = p;
473 		*nalloc += alloc_chunk_size;
474 	}
475 	*ndeltas += 1;
476 	d = &(*deltas)[*ndeltas - 1];
477 	d->copy = copy;
478 	d->offset = offset;
479 	d->len = len;
480 	return NULL;
481 }
482 
483 static const struct got_error *
484 stretchblk(FILE *basefile, off_t base_offset0, struct got_delta_block *block,
485     FILE *f, off_t filesize, off_t *blocklen)
486 {
487 	uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
488 	size_t base_r, r, i;
489 	int buf_equal = 1;
490 
491 	if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
492 	    SEEK_SET) == -1)
493 		return got_error_from_errno("fseeko");
494 
495 	while (buf_equal && *blocklen < (1 << 24) - 1) {
496 		base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
497 		if (base_r == 0) {
498 			if (ferror(basefile))
499 				return got_ferror(basefile, GOT_ERR_IO);
500 			break;
501 		}
502 		r = fread(buf, 1, sizeof(buf), f);
503 		if (r == 0) {
504 			if (ferror(f))
505 				return got_ferror(f, GOT_ERR_IO);
506 			break;
507 		}
508 		for (i = 0; i < MIN(base_r, r); i++) {
509 			if (buf[i] != basebuf[i]) {
510 				buf_equal = 0;
511 				break;
512 			}
513 			(*blocklen)++;
514 		}
515 	}
516 
517 	return NULL;
518 }
519 
520 static const struct got_error *
521 stretchblk_file_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
522      struct got_delta_block *block, FILE *f, off_t filesize, off_t *blocklen)
523 {
524 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
525 	size_t r, i;
526 	int buf_equal = 1;
527 	off_t base_offset = base_offset0 + block->offset + *blocklen;
528 
529 	if (base_offset > basefile_size) {
530 		return got_error_fmt(GOT_ERR_RANGE,
531 		    "read beyond the size of delta base at offset %llu",
532 		    base_offset);
533 	}
534 
535 	while (buf_equal && *blocklen < (1 << 24) - 1) {
536 		if (base_offset + *blocklen >= basefile_size)
537 			break;
538 		r = fread(buf, 1, sizeof(buf), f);
539 		if (r == 0) {
540 			if (ferror(f))
541 				return got_ferror(f, GOT_ERR_IO);
542 			break;
543 		}
544 		for (i = 0; i < MIN(basefile_size - base_offset, r); i++) {
545 			if (buf[i] != *(basedata + base_offset + i)) {
546 				buf_equal = 0;
547 				break;
548 			}
549 			(*blocklen)++;
550 		}
551 	}
552 
553 	return NULL;
554 }
555 
556 static const struct got_error *
557 stretchblk_mem_file(FILE *basefile, off_t base_offset0,
558     struct got_delta_block *block, uint8_t *data, off_t fileoffset,
559     off_t filesize, off_t *blocklen)
560 {
561 	uint8_t basebuf[GOT_DELTIFY_MAXCHUNK];
562 	size_t base_r, i;
563 	int buf_equal = 1;
564 
565 	if (fileoffset > filesize) {
566 		return got_error_fmt(GOT_ERR_RANGE,
567 		    "read beyond the size of deltify file at offset %llu",
568 		    fileoffset);
569 	}
570 
571 	if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
572 	    SEEK_SET) == -1)
573 		return got_error_from_errno("fseeko");
574 
575 	while (buf_equal && *blocklen < (1 << 24) - 1) {
576 		if (fileoffset + *blocklen >= filesize)
577 			break;
578 		base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
579 		if (base_r == 0) {
580 			if (ferror(basefile))
581 				return got_ferror(basefile, GOT_ERR_IO);
582 			break;
583 		}
584 		for (i = 0; i < MIN(base_r, filesize - fileoffset); i++) {
585 			if (*(data + fileoffset + i) != basebuf[i]) {
586 				buf_equal = 0;
587 				break;
588 			}
589 			(*blocklen)++;
590 		}
591 	}
592 
593 	return NULL;
594 }
595 
596 static const struct got_error *
597 stretchblk_mem_mem(uint8_t *basedata, off_t base_offset0, off_t basefile_size,
598     struct got_delta_block *block, uint8_t *data, off_t fileoffset,
599     off_t filesize, off_t *blocklen)
600 {
601 	off_t i, maxlen;
602 	off_t base_offset = base_offset0 + block->offset + *blocklen;
603 	uint8_t *p, *q;
604 
605 	if (base_offset > basefile_size) {
606 		return got_error_fmt(GOT_ERR_RANGE,
607 		    "read beyond the size of delta base at offset %llu",
608 		    base_offset);
609 	}
610 
611 	if (fileoffset > filesize) {
612 		return got_error_fmt(GOT_ERR_RANGE,
613 		    "read beyond the size of deltify file at offset %llu",
614 		    fileoffset);
615 	}
616 
617 	p = data + fileoffset;
618 	q = basedata + base_offset;
619 	maxlen = MIN(basefile_size - base_offset, filesize - fileoffset);
620 	for (i = 0; i < maxlen && *blocklen < (1 << 24) - 1; i++) {
621 		if (p[i] != q[i])
622 			break;
623 		(*blocklen)++;
624 	}
625 
626 	return NULL;
627 }
628 
629 const struct got_error *
630 got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
631     FILE *f, off_t fileoffset, off_t filesize,
632     struct got_delta_table *dt, FILE *basefile,
633     off_t basefile_offset0, off_t basefile_size)
634 {
635 	const struct got_error *err = NULL;
636 	const off_t offset0 = fileoffset;
637 	size_t nalloc = 0;
638 	const size_t alloc_chunk_size = 64;
639 
640 	*deltas = NULL;
641 	*ndeltas = 0;
642 
643 	/*
644 	 * offset0 indicates where data to be deltified begins.
645 	 * For example, we want to avoid deltifying a Git object header at
646 	 * the beginning of the file.
647 	 */
648 	if (fseeko(f, offset0, SEEK_SET) == -1)
649 		return got_error_from_errno("fseeko");
650 
651 	*deltas = reallocarray(NULL, alloc_chunk_size,
652 	    sizeof(struct got_delta_instruction));
653 	if (*deltas == NULL)
654 		return got_error_from_errno("reallocarray");
655 	nalloc = alloc_chunk_size;
656 
657 	while (fileoffset < filesize) {
658 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
659 		off_t blocklen;
660 		struct got_delta_block *block;
661 		err = nextblk(buf, &blocklen, f);
662 		if (err)
663 			break;
664 		if (blocklen == 0) {
665 			/* Source remainder from the file itself. */
666 			if (fileoffset < filesize) {
667 				err = emitdelta(deltas, &nalloc, ndeltas,
668 				    alloc_chunk_size, 0, fileoffset - offset0,
669 				    filesize - fileoffset);
670 			}
671 			break;
672 		}
673 		err = lookupblk(&block, dt, buf, blocklen, basefile,
674 		    basefile_offset0);
675 		if (err)
676 			break;
677 		if (block != NULL) {
678 			/*
679 			 * We have found a matching block in the delta base.
680 			 * Attempt to stretch the block as far as possible and
681 			 * generate a copy instruction.
682 			 */
683 			err = stretchblk(basefile, basefile_offset0, block,
684 			    f, filesize, &blocklen);
685 			if (err)
686 				break;
687 			err = emitdelta(deltas, &nalloc, ndeltas,
688 			    alloc_chunk_size, 1, block->offset, blocklen);
689 			if (err)
690 				break;
691 		} else {
692 			/*
693 			 * No match.
694 			 * This block needs to be sourced from the file itself.
695 			 */
696 			err = emitdelta(deltas, &nalloc, ndeltas,
697 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
698 			if (err)
699 				break;
700 		}
701 		fileoffset += blocklen;
702 		if (fseeko(f, fileoffset, SEEK_SET) == -1) {
703 			err = got_error_from_errno("fseeko");
704 			break;
705 		}
706 	}
707 
708 	if (err) {
709 		free(*deltas);
710 		*deltas = NULL;
711 		*ndeltas = 0;
712 	}
713 	return err;
714 }
715 
716 const struct got_error *
717 got_deltify_file_mem(struct got_delta_instruction **deltas, int *ndeltas,
718     FILE *f, off_t fileoffset, off_t filesize,
719     struct got_delta_table *dt, uint8_t *basedata,
720     off_t basefile_offset0, off_t basefile_size)
721 {
722 	const struct got_error *err = NULL;
723 	const off_t offset0 = fileoffset;
724 	size_t nalloc = 0;
725 	const size_t alloc_chunk_size = 64;
726 
727 	*deltas = NULL;
728 	*ndeltas = 0;
729 
730 	/*
731 	 * offset0 indicates where data to be deltified begins.
732 	 * For example, we want to avoid deltifying a Git object header at
733 	 * the beginning of the file.
734 	 */
735 	if (fseeko(f, offset0, SEEK_SET) == -1)
736 		return got_error_from_errno("fseeko");
737 
738 	*deltas = reallocarray(NULL, alloc_chunk_size,
739 	    sizeof(struct got_delta_instruction));
740 	if (*deltas == NULL)
741 		return got_error_from_errno("reallocarray");
742 	nalloc = alloc_chunk_size;
743 
744 	while (fileoffset < filesize) {
745 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
746 		off_t blocklen;
747 		struct got_delta_block *block;
748 		err = nextblk(buf, &blocklen, f);
749 		if (err)
750 			break;
751 		if (blocklen == 0) {
752 			/* Source remainder from the file itself. */
753 			if (fileoffset < filesize) {
754 				err = emitdelta(deltas, &nalloc, ndeltas,
755 				    alloc_chunk_size, 0, fileoffset - offset0,
756 				    filesize - fileoffset);
757 			}
758 			break;
759 		}
760 		err = lookupblk_mem(&block, dt, buf, blocklen, basedata,
761 		    basefile_offset0);
762 		if (err)
763 			break;
764 		if (block != NULL) {
765 			/*
766 			 * We have found a matching block in the delta base.
767 			 * Attempt to stretch the block as far as possible and
768 			 * generate a copy instruction.
769 			 */
770 			err = stretchblk_file_mem(basedata, basefile_offset0,
771 			    basefile_size, block, f, filesize, &blocklen);
772 			if (err)
773 				break;
774 			err = emitdelta(deltas, &nalloc, ndeltas,
775 			    alloc_chunk_size, 1, block->offset, blocklen);
776 			if (err)
777 				break;
778 		} else {
779 			/*
780 			 * No match.
781 			 * This block needs to be sourced from the file itself.
782 			 */
783 			err = emitdelta(deltas, &nalloc, ndeltas,
784 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
785 			if (err)
786 				break;
787 		}
788 		fileoffset += blocklen;
789 		if (fseeko(f, fileoffset, SEEK_SET) == -1) {
790 			err = got_error_from_errno("fseeko");
791 			break;
792 		}
793 	}
794 
795 	if (err) {
796 		free(*deltas);
797 		*deltas = NULL;
798 		*ndeltas = 0;
799 	}
800 	return err;
801 }
802 
803 const struct got_error *
804 got_deltify_mem_file(struct got_delta_instruction **deltas, int *ndeltas,
805     uint8_t *data, off_t fileoffset, off_t filesize,
806     struct got_delta_table *dt, FILE *basefile,
807     off_t basefile_offset0, off_t basefile_size)
808 {
809 	const struct got_error *err = NULL;
810 	const off_t offset0 = fileoffset;
811 	size_t nalloc = 0;
812 	const size_t alloc_chunk_size = 64;
813 
814 	*deltas = NULL;
815 	*ndeltas = 0;
816 
817 	*deltas = reallocarray(NULL, alloc_chunk_size,
818 	    sizeof(struct got_delta_instruction));
819 	if (*deltas == NULL)
820 		return got_error_from_errno("reallocarray");
821 	nalloc = alloc_chunk_size;
822 
823 	while (fileoffset < filesize) {
824 		off_t blocklen;
825 		struct got_delta_block *block;
826 		err = nextblk_mem(&blocklen, data, fileoffset, filesize);
827 		if (err)
828 			break;
829 		if (blocklen == 0) {
830 			/* Source remainder from the file itself. */
831 			if (fileoffset < filesize) {
832 				err = emitdelta(deltas, &nalloc, ndeltas,
833 				    alloc_chunk_size, 0, fileoffset - offset0,
834 				    filesize - fileoffset);
835 			}
836 			break;
837 		}
838 		err = lookupblk(&block, dt, data + fileoffset, blocklen,
839 		    basefile, basefile_offset0);
840 		if (err)
841 			break;
842 		if (block != NULL) {
843 			/*
844 			 * We have found a matching block in the delta base.
845 			 * Attempt to stretch the block as far as possible and
846 			 * generate a copy instruction.
847 			 */
848 			err = stretchblk_mem_file(basefile, basefile_offset0,
849 			    block, data, fileoffset + blocklen, filesize,
850 			    &blocklen);
851 			if (err)
852 				break;
853 			err = emitdelta(deltas, &nalloc, ndeltas,
854 			    alloc_chunk_size, 1, block->offset, blocklen);
855 			if (err)
856 				break;
857 		} else {
858 			/*
859 			 * No match.
860 			 * This block needs to be sourced from the file itself.
861 			 */
862 			err = emitdelta(deltas, &nalloc, ndeltas,
863 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
864 			if (err)
865 				break;
866 		}
867 		fileoffset += blocklen;
868 	}
869 
870 	if (err) {
871 		free(*deltas);
872 		*deltas = NULL;
873 		*ndeltas = 0;
874 	}
875 	return err;
876 }
877 
878 const struct got_error *
879 got_deltify_mem_mem(struct got_delta_instruction **deltas, int *ndeltas,
880     uint8_t *data, off_t fileoffset, off_t filesize,
881     struct got_delta_table *dt, uint8_t *basedata,
882     off_t basefile_offset0, off_t basefile_size)
883 {
884 	const struct got_error *err = NULL;
885 	const off_t offset0 = fileoffset;
886 	size_t nalloc = 0;
887 	const size_t alloc_chunk_size = 64;
888 
889 	*deltas = NULL;
890 	*ndeltas = 0;
891 
892 	*deltas = reallocarray(NULL, alloc_chunk_size,
893 	    sizeof(struct got_delta_instruction));
894 	if (*deltas == NULL)
895 		return got_error_from_errno("reallocarray");
896 	nalloc = alloc_chunk_size;
897 
898 	while (fileoffset < filesize) {
899 		off_t blocklen;
900 		struct got_delta_block *block;
901 		err = nextblk_mem(&blocklen, data, fileoffset, filesize);
902 		if (err)
903 			break;
904 		if (blocklen == 0) {
905 			/* Source remainder from the file itself. */
906 			if (fileoffset < filesize) {
907 				err = emitdelta(deltas, &nalloc, ndeltas,
908 				    alloc_chunk_size, 0, fileoffset - offset0,
909 				    filesize - fileoffset);
910 			}
911 			break;
912 		}
913 		err = lookupblk_mem(&block, dt, data + fileoffset, blocklen,
914 		    basedata, basefile_offset0);
915 		if (err)
916 			break;
917 		if (block != NULL) {
918 			/*
919 			 * We have found a matching block in the delta base.
920 			 * Attempt to stretch the block as far as possible and
921 			 * generate a copy instruction.
922 			 */
923 			err = stretchblk_mem_mem(basedata, basefile_offset0,
924 			    basefile_size, block, data, fileoffset + blocklen,
925 			    filesize, &blocklen);
926 			if (err)
927 				break;
928 			err = emitdelta(deltas, &nalloc, ndeltas,
929 			    alloc_chunk_size, 1, block->offset, blocklen);
930 			if (err)
931 				break;
932 		} else {
933 			/*
934 			 * No match.
935 			 * This block needs to be sourced from the file itself.
936 			 */
937 			err = emitdelta(deltas, &nalloc, ndeltas,
938 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
939 			if (err)
940 				break;
941 		}
942 		fileoffset += blocklen;
943 	}
944 
945 	if (err) {
946 		free(*deltas);
947 		*deltas = NULL;
948 		*ndeltas = 0;
949 	}
950 	return err;
951 }