1 /*
2  * Copyright (c) 2018 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/stat.h>
19 #include <sys/queue.h>
20 #include <sys/mman.h>
21 
22 #include <dirent.h>
23 #include <fcntl.h>
24 #include <errno.h>
25 #include <stdio.h>
26 #include <stdint.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <limits.h>
30 #include <sha1.h>
31 #include <endian.h>
32 #include <zlib.h>
33 
34 #include "got_error.h"
35 #include "got_object.h"
36 #include "got_repository.h"
37 #include "got_opentemp.h"
38 
39 #include "got_lib_sha1.h"
40 #include "got_lib_pack.h"
41 #include "got_lib_path.h"
42 #include "got_lib_delta.h"
43 #include "got_lib_inflate.h"
44 #include "got_lib_object.h"
45 #include "got_lib_repository.h"
46 
47 #ifndef nitems
48 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
49 #endif
50 
51 #define GOT_PACK_PREFIX		"pack-"
52 #define GOT_PACKFILE_SUFFIX	".pack"
53 #define GOT_PACKIDX_SUFFIX		".idx"
54 #define GOT_PACKFILE_NAMELEN	(strlen(GOT_PACK_PREFIX) + \
55 				SHA1_DIGEST_STRING_LENGTH - 1 + \
56 				strlen(GOT_PACKFILE_SUFFIX))
57 #define GOT_PACKIDX_NAMELEN	(strlen(GOT_PACK_PREFIX) + \
58 				SHA1_DIGEST_STRING_LENGTH - 1 + \
59 				strlen(GOT_PACKIDX_SUFFIX))
60 
61 #ifndef MIN
62 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
63 #endif
64 
65 static const struct got_error *
66 verify_fanout_table(uint32_t *fanout_table)
67 {
68 	int i;
69 
70 	for (i = 0; i < 0xff - 1; i++) {
71 		if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
72 			return got_error(GOT_ERR_BAD_PACKIDX);
73 	}
74 
75 	return NULL;
76 }
77 
78 static const struct got_error *
79 get_packfile_size(size_t *size, const char *path)
80 {
81 	struct stat sb;
82 	char *dot;
83 
84 	*size = 0;
85 
86 	dot = strrchr(path, '.');
87 	if (dot == NULL)
88 		return got_error(GOT_ERR_BAD_PATH);
89 
90 	/* Path must point to a pack index or to a pack file. */
91 	if (strcmp(dot, GOT_PACKIDX_SUFFIX) == 0) {
92 		const struct got_error *err = NULL;
93 		char *path_pack;
94 		char base_path[PATH_MAX];
95 
96 		/* Convert pack index path to pack file path. */
97 		if (strlcpy(base_path, path, PATH_MAX) > PATH_MAX)
98 			return got_error(GOT_ERR_NO_SPACE);
99 		dot = strrchr(base_path, '.');
100 		if (dot == NULL)
101 			return got_error(GOT_ERR_BAD_PATH);
102 		*dot = '\0';
103 		if (asprintf(&path_pack, "%s.pack", base_path) == -1)
104 			return got_error_from_errno();
105 
106 		if (stat(path_pack, &sb) != 0)
107 			err = got_error_from_errno();
108 		free(path_pack);
109 		if (err)
110 			return err;
111 	} else if (strcmp(dot, GOT_PACKFILE_SUFFIX) == 0) {
112 		if (stat(path, &sb) != 0)
113 			return got_error_from_errno();
114 	} else
115 		return got_error(GOT_ERR_BAD_PATH);
116 
117 	*size = sb.st_size;
118 	return 0;
119 }
120 
121 const struct got_error *
122 got_packidx_init_hdr(struct got_packidx *p, int verify)
123 {
124 	const struct got_error *err = NULL;
125 	struct got_packidx_v2_hdr *h;
126 	SHA1_CTX ctx;
127 	uint8_t sha1[SHA1_DIGEST_LENGTH];
128 	size_t nobj, len_fanout, len_ids, offset, remain;
129 	ssize_t n;
130 
131 	SHA1Init(&ctx);
132 
133 	h = &p->hdr;
134 	offset = 0;
135 	remain = p->len;
136 
137 	if (remain < sizeof(*h->magic)) {
138 		err = got_error(GOT_ERR_BAD_PACKIDX);
139 		goto done;
140 	}
141 	if (p->map)
142 		h->magic = (uint32_t *)(p->map + offset);
143 	else {
144 		h->magic = malloc(sizeof(*h->magic));
145 		if (h->magic == NULL) {
146 			err = got_error_from_errno();
147 			goto done;
148 		}
149 		n = read(p->fd, h->magic, sizeof(*h->magic));
150 		if (n != sizeof(*h->magic)) {
151 			err = got_error(GOT_ERR_BAD_PACKIDX);
152 			goto done;
153 		}
154 	}
155 	if (betoh32(*h->magic) != GOT_PACKIDX_V2_MAGIC) {
156 		err = got_error(GOT_ERR_BAD_PACKIDX);
157 		goto done;
158 	}
159 	offset += sizeof(*h->magic);
160 	remain -= sizeof(*h->magic);
161 
162 	if (verify)
163 		SHA1Update(&ctx, (uint8_t *)h->magic, sizeof(*h->magic));
164 
165 	if (remain < sizeof(*h->version)) {
166 		err = got_error(GOT_ERR_BAD_PACKIDX);
167 		goto done;
168 	}
169 	if (p->map)
170 		h->version = (uint32_t *)(p->map + offset);
171 	else {
172 		h->version = malloc(sizeof(*h->version));
173 		if (h->version == NULL) {
174 			err = got_error_from_errno();
175 			goto done;
176 		}
177 		n = read(p->fd, h->version, sizeof(*h->version));
178 		if (n != sizeof(*h->version)) {
179 			err = got_error(GOT_ERR_BAD_PACKIDX);
180 			goto done;
181 		}
182 	}
183 	if (betoh32(*h->version) != GOT_PACKIDX_VERSION) {
184 		err = got_error(GOT_ERR_BAD_PACKIDX);
185 		goto done;
186 	}
187 	offset += sizeof(*h->version);
188 	remain -= sizeof(*h->version);
189 
190 	if (verify)
191 		SHA1Update(&ctx, (uint8_t *)h->version, sizeof(*h->version));
192 
193 	len_fanout =
194 	    sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
195 	if (remain < len_fanout) {
196 		err = got_error(GOT_ERR_BAD_PACKIDX);
197 		goto done;
198 	}
199 	if (p->map)
200 		h->fanout_table = (uint32_t *)(p->map + offset);
201 	else {
202 		h->fanout_table = malloc(len_fanout);
203 		if (h->fanout_table == NULL) {
204 			err = got_error_from_errno();
205 			goto done;
206 		}
207 		n = read(p->fd, h->fanout_table, len_fanout);
208 		if (n != len_fanout) {
209 			err = got_error(GOT_ERR_BAD_PACKIDX);
210 			goto done;
211 		}
212 	}
213 	err = verify_fanout_table(h->fanout_table);
214 	if (err)
215 		goto done;
216 	if (verify)
217 		SHA1Update(&ctx, (uint8_t *)h->fanout_table, len_fanout);
218 	offset += len_fanout;
219 	remain -= len_fanout;
220 
221 	nobj = betoh32(h->fanout_table[0xff]);
222 	len_ids = nobj * sizeof(*h->sorted_ids);
223 	if (len_ids <= nobj || len_ids > remain) {
224 		err = got_error(GOT_ERR_BAD_PACKIDX);
225 		goto done;
226 	}
227 	if (p->map)
228 		h->sorted_ids =
229 		    (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
230 	else {
231 		h->sorted_ids = malloc(len_ids);
232 		if (h->sorted_ids == NULL) {
233 			err = got_error(GOT_ERR_BAD_PACKIDX);
234 			goto done;
235 		}
236 		n = read(p->fd, h->sorted_ids, len_ids);
237 		if (n != len_ids) {
238 			err = got_error(GOT_ERR_BAD_PACKIDX);
239 			goto done;
240 		}
241 	}
242 	if (verify)
243 		SHA1Update(&ctx, (uint8_t *)h->sorted_ids, len_ids);
244 	offset += len_ids;
245 	remain -= len_ids;
246 
247 	if (remain < nobj * sizeof(*h->crc32)) {
248 		err = got_error(GOT_ERR_BAD_PACKIDX);
249 		goto done;
250 	}
251 	if (p->map)
252 		h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
253 	else {
254 		h->crc32 = malloc(nobj * sizeof(*h->crc32));
255 		if (h->crc32 == NULL) {
256 			err = got_error_from_errno();
257 			goto done;
258 		}
259 		n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
260 		if (n != nobj * sizeof(*h->crc32)) {
261 			err = got_error(GOT_ERR_BAD_PACKIDX);
262 			goto done;
263 		}
264 	}
265 	if (verify)
266 		SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
267 	remain -= nobj * sizeof(*h->crc32);
268 	offset += nobj * sizeof(*h->crc32);
269 
270 	if (remain < nobj * sizeof(*h->offsets)) {
271 		err = got_error(GOT_ERR_BAD_PACKIDX);
272 		goto done;
273 	}
274 	if (p->map)
275 		h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
276 	else {
277 		h->offsets = malloc(nobj * sizeof(*h->offsets));
278 		if (h->offsets == NULL) {
279 			err = got_error_from_errno();
280 			goto done;
281 		}
282 		n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
283 		if (n != nobj * sizeof(*h->offsets)) {
284 			err = got_error(GOT_ERR_BAD_PACKIDX);
285 			goto done;
286 		}
287 	}
288 	if (verify)
289 		SHA1Update(&ctx, (uint8_t *)h->offsets,
290 		    nobj * sizeof(*h->offsets));
291 	remain -= nobj * sizeof(*h->offsets);
292 	offset += nobj * sizeof(*h->offsets);
293 
294 	/* Large file offsets are contained only in files > 2GB. */
295 	if (p->len <= 0x80000000)
296 		goto checksum;
297 
298 	if (remain < nobj * sizeof(*h->large_offsets)) {
299 		err = got_error(GOT_ERR_BAD_PACKIDX);
300 		goto done;
301 	}
302 	if (p->map)
303 		h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
304 	else {
305 		h->offsets = malloc(nobj * sizeof(*h->large_offsets));
306 		if (h->offsets == NULL) {
307 			err = got_error_from_errno();
308 			goto done;
309 		}
310 		n = read(p->fd, h->large_offsets,
311 		    nobj * sizeof(*h->large_offsets));
312 		if (n != nobj * sizeof(*h->large_offsets)) {
313 			err = got_error(GOT_ERR_BAD_PACKIDX);
314 			goto done;
315 		}
316 	}
317 	if (verify)
318 		SHA1Update(&ctx, (uint8_t*)h->large_offsets,
319 		    nobj * sizeof(*h->large_offsets));
320 	remain -= nobj * sizeof(*h->large_offsets);
321 	offset += nobj * sizeof(*h->large_offsets);
322 
323 checksum:
324 	if (remain < sizeof(*h->trailer)) {
325 		err = got_error(GOT_ERR_BAD_PACKIDX);
326 		goto done;
327 	}
328 	if (p->map)
329 		h->trailer =
330 		    (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
331 	else {
332 		h->trailer = malloc(sizeof(*h->trailer));
333 		if (h->trailer == NULL) {
334 			err = got_error_from_errno();
335 			goto done;
336 		}
337 		n = read(p->fd, h->trailer, sizeof(*h->trailer));
338 		if (n != sizeof(*h->trailer)) {
339 			err = got_error(GOT_ERR_BAD_PACKIDX);
340 			goto done;
341 		}
342 	}
343 	if (verify) {
344 		SHA1Update(&ctx, h->trailer->packfile_sha1, SHA1_DIGEST_LENGTH);
345 		SHA1Final(sha1, &ctx);
346 		if (memcmp(h->trailer->packidx_sha1, sha1,
347 		    SHA1_DIGEST_LENGTH) != 0)
348 			err = got_error(GOT_ERR_PACKIDX_CSUM);
349 	}
350 done:
351 	return err;
352 }
353 
354 const struct got_error *
355 got_packidx_open(struct got_packidx **packidx, const char *path, int verify)
356 {
357 	const struct got_error *err = NULL;
358 	struct got_packidx *p;
359 
360 	*packidx = NULL;
361 
362 	p = calloc(1, sizeof(*p));
363 	if (p == NULL)
364 		return got_error_from_errno();
365 
366 	p->fd = open(path, O_RDONLY | O_NOFOLLOW, GOT_DEFAULT_FILE_MODE);
367 	if (p->fd == -1)
368 		return got_error_from_errno();
369 
370 	err = get_packfile_size(&p->len, path);
371 	if (err) {
372 		close(p->fd);
373 		free(p);
374 		return err;
375 	}
376 	if (p->len < sizeof(p->hdr)) {
377 		err = got_error(GOT_ERR_BAD_PACKIDX);
378 		close(p->fd);
379 		free(p);
380 		return err;
381 	}
382 
383 	p->path_packidx = strdup(path);
384 	if (p->path_packidx == NULL) {
385 		err = got_error_from_errno();
386 		goto done;
387 	}
388 
389 #ifndef GOT_PACK_NO_MMAP
390 	p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
391 	if (p->map == MAP_FAILED)
392 		p->map = NULL; /* fall back to read(2) */
393 #endif
394 
395 	err = got_packidx_init_hdr(p, verify);
396 done:
397 	if (err)
398 		got_packidx_close(p);
399 	else
400 		*packidx = p;
401 
402 	return err;
403 }
404 
405 const struct got_error *
406 got_packidx_close(struct got_packidx *packidx)
407 {
408 	const struct got_error *err = NULL;
409 
410 	free(packidx->path_packidx);
411 	if (packidx->map) {
412 		if (munmap(packidx->map, packidx->len) == -1)
413 			err = got_error_from_errno();
414 	} else {
415 		free(packidx->hdr.magic);
416 		free(packidx->hdr.version);
417 		free(packidx->hdr.fanout_table);
418 		free(packidx->hdr.sorted_ids);
419 		free(packidx->hdr.crc32);
420 		free(packidx->hdr.offsets);
421 		free(packidx->hdr.large_offsets);
422 		free(packidx->hdr.trailer);
423 	}
424 	close(packidx->fd);
425 	free(packidx);
426 
427 	return err;
428 }
429 
430 static int
431 is_packidx_filename(const char *name, size_t len)
432 {
433 	if (len != GOT_PACKIDX_NAMELEN)
434 		return 0;
435 
436 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
437 		return 0;
438 
439 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
440 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
441 		return 0;
442 
443 	return 1;
444 }
445 
446 static off_t
447 get_object_offset(struct got_packidx *packidx, int idx)
448 {
449 	uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
450 	uint32_t offset = betoh32(packidx->hdr.offsets[idx]);
451 	if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
452 		uint64_t loffset;
453 		idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
454 		if (idx < 0 || idx > totobj ||
455 		    packidx->hdr.large_offsets == NULL)
456 			return -1;
457 		loffset = betoh64(packidx->hdr.large_offsets[idx]);
458 		return (loffset > INT64_MAX ? -1 : (off_t)loffset);
459 	}
460 	return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
461 }
462 
463 static int
464 get_object_idx(struct got_packidx *packidx, struct got_object_id *id)
465 {
466 	u_int8_t id0 = id->sha1[0];
467 	uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
468 	int left = 0, right = totobj - 1;
469 
470 	if (id0 > 0)
471 		left = betoh32(packidx->hdr.fanout_table[id0 - 1]);
472 
473 	while (left <= right) {
474 		struct got_packidx_object_id *oid;
475 		int i, cmp;
476 
477 		i = ((left + right) / 2);
478 		oid = &packidx->hdr.sorted_ids[i];
479 		cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
480 		if (cmp == 0)
481 			return i;
482 		else if (cmp > 0)
483 			left = i + 1;
484 		else if (cmp < 0)
485 			right = i - 1;
486 	}
487 
488 	return -1;
489 }
490 
491 static const struct got_error *
492 cache_packidx(struct got_packidx *packidx, struct got_repository *repo)
493 {
494 	const struct got_error *err = NULL;
495 	int i;
496 
497 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
498 		if (repo->packidx_cache[i] == NULL)
499 			break;
500 	}
501 
502 	if (i == nitems(repo->packidx_cache)) {
503 		err = got_packidx_close(repo->packidx_cache[i - 1]);
504 		if (err)
505 			return err;
506 		memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
507 		    sizeof(repo->packidx_cache) -
508 		    sizeof(repo->packidx_cache[0]));
509 		i = 0;
510 	}
511 
512 	repo->packidx_cache[i] = packidx;
513 	return NULL;
514 }
515 
516 static const struct got_error *
517 search_packidx(struct got_packidx **packidx, int *idx,
518     struct got_repository *repo, struct got_object_id *id)
519 {
520 	const struct got_error *err;
521 	char *path_packdir;
522 	DIR *packdir;
523 	struct dirent *dent;
524 	char *path_packidx;
525 	int i;
526 
527 	/* Search pack index cache. */
528 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
529 		if (repo->packidx_cache[i] == NULL)
530 			break;
531 		*idx = get_object_idx(repo->packidx_cache[i], id);
532 		if (*idx != -1) {
533 			*packidx = repo->packidx_cache[i];
534 			return NULL;
535 		}
536 	}
537 	/* No luck. Search the filesystem. */
538 
539 	path_packdir = got_repo_get_path_objects_pack(repo);
540 	if (path_packdir == NULL)
541 		return got_error_from_errno();
542 
543 	packdir = opendir(path_packdir);
544 	if (packdir == NULL) {
545 		err = got_error_from_errno();
546 		goto done;
547 	}
548 
549 	while ((dent = readdir(packdir)) != NULL) {
550 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
551 			continue;
552 
553 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
554 		    dent->d_name) == -1) {
555 			err = got_error_from_errno();
556 			goto done;
557 		}
558 
559 		err = got_packidx_open(packidx, path_packidx, 0);
560 		free(path_packidx);
561 		if (err)
562 			goto done;
563 
564 		*idx = get_object_idx(*packidx, id);
565 		if (*idx != -1) {
566 			err = NULL; /* found the object */
567 			err = cache_packidx(*packidx, repo);
568 			goto done;
569 		}
570 
571 		err = got_packidx_close(*packidx);
572 		*packidx = NULL;
573 		if (err)
574 			goto done;
575 	}
576 
577 	err = got_error(GOT_ERR_NO_OBJ);
578 done:
579 	free(path_packdir);
580 	if (packdir && closedir(packdir) != 0 && err == 0)
581 		err = got_error_from_errno();
582 	return err;
583 }
584 
585 static const struct got_error *
586 get_packfile_path(char **path_packfile, struct got_packidx *packidx)
587 {
588 	size_t size;
589 
590 	/* Packfile path contains ".pack" instead of ".idx", so add one byte. */
591 	size = strlen(packidx->path_packidx) + 2;
592 	if (size < GOT_PACKFILE_NAMELEN + 1)
593 		return got_error(GOT_ERR_BAD_PATH);
594 
595 	*path_packfile = calloc(size, sizeof(**path_packfile));
596 	if (*path_packfile == NULL)
597 		return got_error_from_errno();
598 
599 	/* Copy up to and excluding ".idx". */
600 	if (strlcpy(*path_packfile, packidx->path_packidx,
601 	    size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
602 		return got_error(GOT_ERR_NO_SPACE);
603 
604 	if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
605 		return got_error(GOT_ERR_NO_SPACE);
606 
607 	return NULL;
608 }
609 
610 static const struct got_error *
611 read_packfile_hdr(int fd, struct got_packidx *packidx)
612 {
613 	const struct got_error *err = NULL;
614 	uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
615 	struct got_packfile_hdr hdr;
616 	ssize_t n;
617 
618 	n = read(fd, &hdr, sizeof(hdr));
619 	if (n < 0)
620 		return got_error_from_errno();
621 	if (n != sizeof(hdr))
622 		return got_error(GOT_ERR_BAD_PACKFILE);
623 
624 	if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
625 	    betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
626 	    betoh32(hdr.nobjects) != totobj)
627 		err = got_error(GOT_ERR_BAD_PACKFILE);
628 
629 	return err;
630 }
631 
632 static const struct got_error *
633 open_packfile(int *fd, const char *path_packfile, struct got_packidx *packidx)
634 {
635 	const struct got_error *err = NULL;
636 
637 	*fd = open(path_packfile, O_RDONLY | O_NOFOLLOW, GOT_DEFAULT_FILE_MODE);
638 	if (*fd == -1)
639 		return got_error_from_errno();
640 
641 	if (packidx) {
642 		err = read_packfile_hdr(*fd, packidx);
643 		if (err) {
644 			close(*fd);
645 			*fd = -1;
646 		}
647 	}
648 	return err;
649 }
650 
651 const struct got_error *
652 got_pack_close(struct got_pack *pack)
653 {
654 	const struct got_error *err = NULL;
655 
656 	if (pack->map && munmap(pack->map, pack->filesize) == -1)
657 		err = got_error_from_errno();
658 	close(pack->fd);
659 	pack->fd = -1;
660 	free(pack->path_packfile);
661 	pack->path_packfile = NULL;
662 	pack->filesize = 0;
663 
664 	return err;
665 }
666 
667 static const struct got_error *
668 cache_pack(struct got_pack **packp, const char *path_packfile,
669     struct got_packidx *packidx, struct got_repository *repo)
670 {
671 	const struct got_error *err = NULL;
672 	struct got_pack *pack = NULL;
673 	int i;
674 
675 	if (packp)
676 		*packp = NULL;
677 
678 	for (i = 0; i < nitems(repo->packs); i++) {
679 		pack = &repo->packs[i];
680 		if (pack->path_packfile == NULL)
681 			break;
682 		if (strcmp(pack->path_packfile, path_packfile) == 0)
683 			return NULL;
684 	}
685 
686 	if (i == nitems(repo->packs) - 1) {
687 		err = got_pack_close(&repo->packs[i - 1]);
688 		if (err)
689 			return err;
690 		memmove(&repo->packs[1], &repo->packs[0],
691 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
692 		i = 0;
693 	}
694 
695 	pack = &repo->packs[i];
696 
697 	pack->path_packfile = strdup(path_packfile);
698 	if (pack->path_packfile == NULL) {
699 		err = got_error_from_errno();
700 		goto done;
701 	}
702 
703 	err = open_packfile(&pack->fd, path_packfile, packidx);
704 	if (err)
705 		goto done;
706 
707 	err = get_packfile_size(&pack->filesize, path_packfile);
708 	if (err)
709 		goto done;
710 
711 #ifndef GOT_PACK_NO_MMAP
712 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
713 	    pack->fd, 0);
714 	if (pack->map == MAP_FAILED)
715 		pack->map = NULL; /* fall back to read(2) */
716 #endif
717 done:
718 	if (err) {
719 		if (pack) {
720 			free(pack->path_packfile);
721 			memset(pack, 0, sizeof(*pack));
722 		}
723 	} else if (packp)
724 		*packp = pack;
725 	return err;
726 }
727 
728 struct got_pack *
729 get_cached_pack(const char *path_packfile, struct got_repository *repo)
730 {
731 	struct got_pack *pack = NULL;
732 	int i;
733 
734 	for (i = 0; i < nitems(repo->packs); i++) {
735 		pack = &repo->packs[i];
736 		if (pack->path_packfile == NULL)
737 			break;
738 		if (strcmp(pack->path_packfile, path_packfile) == 0)
739 			return pack;
740 	}
741 
742 	return NULL;
743 }
744 
745 static const struct got_error *
746 parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
747     struct got_pack *pack, off_t offset)
748 {
749 	uint8_t t = 0;
750 	uint64_t s = 0;
751 	uint8_t sizeN;
752 	size_t mapoff = 0;
753 	int i = 0;
754 
755 	*len = 0;
756 
757 	if (offset >= pack->filesize)
758 		return got_error(GOT_ERR_PACK_OFFSET);
759 
760 	if (pack->map) {
761 		mapoff = (size_t)offset;
762 	} else {
763 		if (lseek(pack->fd, offset, SEEK_SET) == -1)
764 			return got_error_from_errno();
765 	}
766 
767 	do {
768 		/* We do not support size values which don't fit in 64 bit. */
769 		if (i > 9)
770 			return got_error(GOT_ERR_NO_SPACE);
771 
772 		if (pack->map) {
773 			sizeN = *(pack->map + mapoff);
774 			mapoff += sizeof(sizeN);
775 		} else {
776 			ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
777 			if (n < 0)
778 				return got_error_from_errno();
779 			if (n != sizeof(sizeN))
780 				return got_error(GOT_ERR_BAD_PACKFILE);
781 		}
782 		*len += sizeof(sizeN);
783 
784 		if (i == 0) {
785 			t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
786 			    GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
787 			s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
788 		} else {
789 			size_t shift = 4 + 7 * (i - 1);
790 			s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
791 		}
792 		i++;
793 	} while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
794 
795 	*type = t;
796 	*size = s;
797 	return NULL;
798 }
799 
800 static const struct got_error *
801 open_plain_object(struct got_object **obj, const char *path_packfile,
802     struct got_object_id *id, uint8_t type, off_t offset, size_t size)
803 {
804 	*obj = calloc(1, sizeof(**obj));
805 	if (*obj == NULL)
806 		return got_error_from_errno();
807 
808 	(*obj)->path_packfile = strdup(path_packfile);
809 	if ((*obj)->path_packfile == NULL) {
810 		const struct got_error *err = got_error_from_errno();
811 		free(*obj);
812 		*obj = NULL;
813 		return err;
814 	}
815 
816 	(*obj)->type = type;
817 	(*obj)->flags = GOT_OBJ_FLAG_PACKED;
818 	(*obj)->hdrlen = 0;
819 	(*obj)->size = size;
820 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
821 	(*obj)->pack_offset = offset;
822 
823 	return NULL;
824 }
825 
826 static const struct got_error *
827 parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
828     off_t delta_offset)
829 {
830 	int64_t o = 0;
831 	uint8_t offN;
832 	int i = 0;
833 
834 	*len = 0;
835 
836 	do {
837 		/* We do not support offset values which don't fit in 64 bit. */
838 		if (i > 8)
839 			return got_error(GOT_ERR_NO_SPACE);
840 
841 		if (pack->map) {
842 			size_t mapoff;
843 			if (delta_offset >= pack->filesize)
844 				return got_error(GOT_ERR_PACK_OFFSET);
845 			mapoff = (size_t)delta_offset + *len;
846 			offN = *(pack->map + mapoff);
847 		} else {
848 			ssize_t n;
849 			n = read(pack->fd, &offN, sizeof(offN));
850 			if (n < 0)
851 				return got_error_from_errno();
852 			if (n != sizeof(offN))
853 				return got_error(GOT_ERR_BAD_PACKFILE);
854 		}
855 		*len += sizeof(offN);
856 
857 		if (i == 0)
858 			o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
859 		else {
860 			o++;
861 			o <<= 7;
862 			o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
863 		}
864 		i++;
865 	} while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
866 
867 	*offset = o;
868 	return NULL;
869 }
870 
871 static const struct got_error *
872 parse_offset_delta(off_t *base_offset, size_t *len, struct got_pack *pack,
873     off_t offset, int tslen)
874 {
875 	const struct got_error *err;
876 	int64_t negoffset;
877 	size_t negofflen;
878 
879 	*len = 0;
880 
881 	err = parse_negative_offset(&negoffset, &negofflen, pack,
882 	    offset + tslen);
883 	if (err)
884 		return err;
885 
886 	/* Compute the base object's offset (must be in the same pack file). */
887 	*base_offset = (offset - negoffset);
888 	if (*base_offset <= 0)
889 		return got_error(GOT_ERR_BAD_PACKFILE);
890 
891 	*len = negofflen;
892 	return NULL;
893 }
894 
895 static const struct got_error *
896 resolve_delta_chain(struct got_delta_chain *, struct got_repository *,
897     struct got_packidx *, struct got_pack *, off_t, size_t, int,
898     size_t, unsigned int);
899 
900 static const struct got_error *
901 add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
902     int delta_type, size_t delta_size, size_t delta_data_offset,
903     uint8_t *delta_buf, size_t delta_len)
904 {
905 	struct got_delta *delta;
906 
907 	delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
908 	    delta_data_offset, delta_buf,
909 	    delta_len);
910 	if (delta == NULL)
911 		return got_error_from_errno();
912 	/* delta is freed in got_object_close() */
913 	deltas->nentries++;
914 	SIMPLEQ_INSERT_HEAD(&deltas->entries, delta, entry);
915 	return NULL;
916 }
917 
918 static const struct got_error *
919 resolve_offset_delta(struct got_delta_chain *deltas,
920     struct got_repository *repo, struct got_packidx *packidx,
921     struct got_pack *pack, off_t delta_offset, size_t tslen,
922     int delta_type, size_t delta_size, unsigned int recursion)
923 
924 {
925 	const struct got_error *err;
926 	off_t base_offset;
927 	uint8_t base_type;
928 	uint64_t base_size;
929 	size_t base_tslen;
930 	off_t delta_data_offset;
931 	uint8_t *delta_buf;
932 	size_t delta_len, consumed;
933 
934 	err = parse_offset_delta(&base_offset, &consumed, pack,
935 	    delta_offset, tslen);
936 	if (err)
937 		return err;
938 
939 	delta_data_offset = delta_offset + tslen + consumed;
940 	if (delta_data_offset >= pack->filesize)
941 		return got_error(GOT_ERR_PACK_OFFSET);
942 
943 	if (pack->map == NULL) {
944 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
945 		if (delta_data_offset == -1)
946 			return got_error_from_errno();
947 	}
948 
949 	if (pack->map) {
950 		size_t mapoff = (size_t)delta_data_offset;
951 		err = got_inflate_to_mem_mmap(&delta_buf, &delta_len, pack->map,
952 		    mapoff, pack->filesize - mapoff);
953 		if (err)
954 			return err;
955 	} else {
956 
957 		err = got_inflate_to_mem_fd(&delta_buf, &delta_len, pack->fd);
958 		if (err)
959 			return err;
960 	}
961 
962 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
963 	    delta_data_offset, delta_buf, delta_len);
964 	if (err)
965 		goto done;
966 
967 	/* An offset delta must be in the same packfile. */
968 	if (base_offset >= pack->filesize) {
969 		err = got_error(GOT_ERR_PACK_OFFSET);
970 		goto done;
971 	}
972 
973 	err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
974 	    pack, base_offset);
975 	if (err)
976 		goto done;
977 
978 	err = resolve_delta_chain(deltas, repo, packidx, pack, base_offset,
979 	    base_tslen, base_type, base_size, recursion - 1);
980 done:
981 	if (err)
982 		free(delta_buf);
983 	return err;
984 }
985 
986 static const struct got_error *
987 resolve_ref_delta(struct got_delta_chain *deltas, struct got_repository *repo,
988     struct got_packidx *packidx, struct got_pack *pack,
989     off_t delta_offset, size_t tslen, int delta_type, size_t delta_size,
990     unsigned int recursion)
991 {
992 	const struct got_error *err;
993 	struct got_object_id id;
994 	int idx;
995 	off_t base_offset;
996 	uint8_t base_type;
997 	uint64_t base_size;
998 	size_t base_tslen;
999 	off_t delta_data_offset;
1000 	uint8_t *delta_buf = NULL;
1001 	size_t delta_len;
1002 
1003 	if (delta_offset >= pack->filesize)
1004 		return got_error(GOT_ERR_PACK_OFFSET);
1005 	delta_data_offset = delta_offset + tslen;
1006 	if (delta_data_offset >= pack->filesize)
1007 		return got_error(GOT_ERR_PACK_OFFSET);
1008 
1009 	if (pack->map == NULL) {
1010 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1011 		if (delta_data_offset == -1)
1012 			return got_error_from_errno();
1013 	}
1014 
1015 
1016 	if (pack->map) {
1017 		size_t mapoff = (size_t)delta_data_offset;
1018 		memcpy(&id, pack->map + mapoff, sizeof(id));
1019 		mapoff += sizeof(id);
1020 		err = got_inflate_to_mem_mmap(&delta_buf, &delta_len, pack->map,
1021 		    mapoff, pack->filesize - mapoff);
1022 		if (err)
1023 			goto done;
1024 	} else {
1025 		ssize_t n = read(pack->fd, &id, sizeof(id));
1026 		if (n < 0)
1027 			return got_error_from_errno();
1028 		if (n != sizeof(id))
1029 			return got_error(GOT_ERR_BAD_PACKFILE);
1030 		err = got_inflate_to_mem_fd(&delta_buf, &delta_len, pack->fd);
1031 		if (err)
1032 			goto done;
1033 	}
1034 
1035 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1036 	    delta_data_offset, delta_buf, delta_len);
1037 	if (err)
1038 		goto done;
1039 
1040 	/* Delta base must be in the same pack file. */
1041 	idx = get_object_idx(packidx, &id);
1042 	if (idx == -1) {
1043 		err = got_error(GOT_ERR_BAD_PACKFILE);
1044 		goto done;
1045 	}
1046 
1047 	base_offset = get_object_offset(packidx, idx);
1048 	if (base_offset == (uint64_t)-1) {
1049 		err = got_error(GOT_ERR_BAD_PACKIDX);
1050 		goto done;
1051 	}
1052 
1053 	if (base_offset >= pack->filesize) {
1054 		err = got_error(GOT_ERR_PACK_OFFSET);
1055 		goto done;
1056 	}
1057 
1058 	err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
1059 	    pack, base_offset);
1060 	if (err)
1061 		goto done;
1062 
1063 	err = resolve_delta_chain(deltas, repo, packidx, pack, base_offset,
1064 	    base_tslen, base_type, base_size, recursion - 1);
1065 done:
1066 	if (err)
1067 		free(delta_buf);
1068 	return err;
1069 }
1070 
1071 static const struct got_error *
1072 resolve_delta_chain(struct got_delta_chain *deltas, struct got_repository *repo,
1073     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1074     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1075 {
1076 	const struct got_error *err = NULL;
1077 
1078 	if (--recursion == 0)
1079 		return got_error(GOT_ERR_RECURSION);
1080 
1081 	switch (delta_type) {
1082 	case GOT_OBJ_TYPE_COMMIT:
1083 	case GOT_OBJ_TYPE_TREE:
1084 	case GOT_OBJ_TYPE_BLOB:
1085 	case GOT_OBJ_TYPE_TAG:
1086 		/* Plain types are the final delta base. Recursion ends. */
1087 		err = add_delta(deltas, delta_offset, tslen, delta_type,
1088 		    delta_size, 0, NULL, 0);
1089 		break;
1090 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1091 		err = resolve_offset_delta(deltas, repo, packidx, pack,
1092 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1093 		break;
1094 	case GOT_OBJ_TYPE_REF_DELTA:
1095 		err = resolve_ref_delta(deltas, repo, packidx, pack,
1096 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1097 		break;
1098 	default:
1099 		return got_error(GOT_ERR_OBJ_TYPE);
1100 	}
1101 
1102 	return err;
1103 }
1104 
1105 static const struct got_error *
1106 open_delta_object(struct got_object **obj, struct got_repository *repo,
1107     struct got_packidx *packidx, struct got_pack *pack,
1108     struct got_object_id *id, off_t offset, size_t tslen,
1109     int delta_type, size_t delta_size)
1110 {
1111 	const struct got_error *err = NULL;
1112 	int resolved_type;
1113 
1114 	*obj = calloc(1, sizeof(**obj));
1115 	if (*obj == NULL)
1116 		return got_error_from_errno();
1117 
1118 	(*obj)->flags = 0;
1119 	(*obj)->hdrlen = 0;
1120 	(*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1121 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1122 	(*obj)->pack_offset = offset + tslen;
1123 
1124 	(*obj)->path_packfile = strdup(pack->path_packfile);
1125 	if ((*obj)->path_packfile == NULL) {
1126 		err = got_error_from_errno();
1127 		goto done;
1128 	}
1129 	(*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1130 
1131 	SIMPLEQ_INIT(&(*obj)->deltas.entries);
1132 	(*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1133 
1134 	err = resolve_delta_chain(&(*obj)->deltas, repo, packidx, pack, offset,
1135 	    tslen, delta_type, delta_size, GOT_DELTA_CHAIN_RECURSION_MAX);
1136 	if (err)
1137 		goto done;
1138 
1139 	err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1140 	if (err)
1141 		goto done;
1142 	(*obj)->type = resolved_type;
1143 
1144 done:
1145 	if (err) {
1146 		got_object_close(*obj);
1147 		*obj = NULL;
1148 	}
1149 	return err;
1150 }
1151 
1152 static const struct got_error *
1153 open_packed_object(struct got_object **obj, struct got_repository *repo,
1154     struct got_packidx *packidx, int idx, struct got_object_id *id)
1155 {
1156 	const struct got_error *err = NULL;
1157 	off_t offset;
1158 	char *path_packfile;
1159 	struct got_pack *pack;
1160 	uint8_t type;
1161 	uint64_t size;
1162 	size_t tslen;
1163 
1164 	*obj = NULL;
1165 
1166 	offset = get_object_offset(packidx, idx);
1167 	if (offset == (uint64_t)-1)
1168 		return got_error(GOT_ERR_BAD_PACKIDX);
1169 
1170 	err = get_packfile_path(&path_packfile, packidx);
1171 	if (err)
1172 		return err;
1173 
1174 	pack = get_cached_pack(path_packfile, repo);
1175 	if (pack == NULL) {
1176 		err = cache_pack(&pack, path_packfile, packidx, repo);
1177 		if (err)
1178 			goto done;
1179 	}
1180 
1181 	err = parse_object_type_and_size(&type, &size, &tslen, pack, offset);
1182 	if (err)
1183 		goto done;
1184 
1185 	switch (type) {
1186 	case GOT_OBJ_TYPE_COMMIT:
1187 	case GOT_OBJ_TYPE_TREE:
1188 	case GOT_OBJ_TYPE_BLOB:
1189 	case GOT_OBJ_TYPE_TAG:
1190 		err = open_plain_object(obj, path_packfile, id, type,
1191 		    offset + tslen, size);
1192 		break;
1193 
1194 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1195 	case GOT_OBJ_TYPE_REF_DELTA:
1196 		err = open_delta_object(obj, repo, packidx, pack, id, offset,
1197 		    tslen, type, size);
1198 		break;
1199 
1200 	default:
1201 		err = got_error(GOT_ERR_OBJ_TYPE);
1202 		goto done;
1203 	}
1204 done:
1205 	free(path_packfile);
1206 	return err;
1207 }
1208 
1209 const struct got_error *
1210 got_packfile_open_object(struct got_object **obj, struct got_object_id *id,
1211     struct got_repository *repo)
1212 {
1213 	const struct got_error *err = NULL;
1214 	struct got_packidx *packidx = NULL;
1215 	int idx;
1216 
1217 	err = search_packidx(&packidx, &idx, repo, id);
1218 	if (err)
1219 		return err;
1220 
1221 	err = open_packed_object(obj, repo, packidx, idx, id);
1222 	if (err)
1223 		return err;
1224 
1225 	err = cache_pack(NULL, (*obj)->path_packfile, packidx, repo);
1226 	return err;
1227 }
1228 
1229 static const struct got_error *
1230 get_delta_chain_max_size(uint64_t *max_size, struct got_delta_chain *deltas)
1231 {
1232 	struct got_delta *delta;
1233 	uint64_t base_size = 0, result_size = 0;
1234 
1235 	*max_size = 0;
1236 	SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1237 		/* Plain object types are the delta base. */
1238 		if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1239 		    delta->type != GOT_OBJ_TYPE_TREE &&
1240 		    delta->type != GOT_OBJ_TYPE_BLOB &&
1241 		    delta->type != GOT_OBJ_TYPE_TAG) {
1242 			const struct got_error *err;
1243 			err = got_delta_get_sizes(&base_size, &result_size,
1244 			    delta->delta_buf, delta->delta_len);
1245 			if (err)
1246 				return err;
1247 		} else
1248 			base_size = delta->size;
1249 		if (base_size > *max_size)
1250 			*max_size = base_size;
1251 		if (result_size > *max_size)
1252 			*max_size = result_size;
1253 	}
1254 
1255 	return NULL;
1256 }
1257 
1258 static const struct got_error *
1259 dump_delta_chain_to_file(size_t *result_size, struct got_delta_chain *deltas,
1260     const char *path_packfile, FILE *outfile, struct got_repository *repo)
1261 {
1262 	const struct got_error *err = NULL;
1263 	struct got_delta *delta;
1264 	FILE *base_file = NULL, *accum_file = NULL;
1265 	uint8_t *base_buf = NULL, *accum_buf = NULL;
1266 	size_t accum_size = 0;
1267 	uint64_t max_size;
1268 	int n = 0;
1269 
1270 	*result_size = 0;
1271 
1272 	if (SIMPLEQ_EMPTY(&deltas->entries))
1273 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1274 
1275 	/* We process small enough files entirely in memory for speed. */
1276 	err = get_delta_chain_max_size(&max_size, deltas);
1277 	if (err)
1278 		return err;
1279 	if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1280 		accum_buf = malloc(max_size);
1281 		if (accum_buf == NULL)
1282 			return got_error_from_errno();
1283 	} else {
1284 		base_file = got_opentemp();
1285 		if (base_file == NULL)
1286 			return got_error_from_errno();
1287 
1288 		accum_file = got_opentemp();
1289 		if (accum_file == NULL) {
1290 			err = got_error_from_errno();
1291 			fclose(base_file);
1292 			return err;
1293 		}
1294 	}
1295 
1296 	/* Deltas are ordered in ascending order. */
1297 	SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1298 		if (n == 0) {
1299 			struct got_pack *pack;
1300 			size_t base_len, mapoff;
1301 			off_t delta_data_offset;
1302 
1303 			/* Plain object types are the delta base. */
1304 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1305 			    delta->type != GOT_OBJ_TYPE_TREE &&
1306 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1307 			    delta->type != GOT_OBJ_TYPE_TAG) {
1308 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1309 				goto done;
1310 			}
1311 
1312 			pack = get_cached_pack(path_packfile, repo);
1313 			if (pack == NULL) {
1314 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1315 				goto done;
1316 			}
1317 
1318 			delta_data_offset = delta->offset + delta->tslen;
1319 			if (delta_data_offset >= pack->filesize) {
1320 				err = got_error(GOT_ERR_PACK_OFFSET);
1321 				goto done;
1322 			}
1323 			if (pack->map == NULL) {
1324 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1325 				    == -1) {
1326 					err = got_error_from_errno();
1327 					goto done;
1328 				}
1329 			}
1330 			if (base_file) {
1331 				if (pack->map) {
1332 					mapoff = (size_t)delta_data_offset;
1333 					err = got_inflate_to_file_mmap(
1334 					    &base_len, pack->map, mapoff,
1335 					    pack->filesize - mapoff, base_file);
1336 				} else
1337 					err = got_inflate_to_file_fd(&base_len,
1338 					    pack->fd, base_file);
1339 			} else {
1340 				if (pack->map) {
1341 					mapoff = (size_t)delta_data_offset;
1342 					err = got_inflate_to_mem_mmap(&base_buf,
1343 					    &base_len, pack->map, mapoff,
1344 					    pack->filesize - mapoff);
1345 				} else
1346 					err = got_inflate_to_mem_fd(&base_buf,
1347 					    &base_len, pack->fd);
1348 				if (base_len < max_size) {
1349 					uint8_t *p;
1350 					p = reallocarray(base_buf, 1, max_size);
1351 					if (p == NULL) {
1352 						err = got_error_from_errno();
1353 						goto done;
1354 					}
1355 					base_buf = p;
1356 				}
1357 			}
1358 			if (err)
1359 				goto done;
1360 			n++;
1361 			if (base_file)
1362 				rewind(base_file);
1363 			continue;
1364 		}
1365 
1366 		if (base_buf) {
1367 			err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1368 			    delta->delta_len, accum_buf, &accum_size);
1369 			n++;
1370 		} else {
1371 			err = got_delta_apply(base_file, delta->delta_buf,
1372 			    delta->delta_len,
1373 			    /* Final delta application writes to output file. */
1374 			    ++n < deltas->nentries ? accum_file : outfile,
1375 			    &accum_size);
1376 		}
1377 		if (err)
1378 			goto done;
1379 
1380 		if (n < deltas->nentries) {
1381 			/* Accumulated delta becomes the new base. */
1382 			if (base_buf) {
1383 				uint8_t *tmp = accum_buf;
1384 				accum_buf = base_buf;
1385 				base_buf = tmp;
1386 			} else {
1387 				FILE *tmp = accum_file;
1388 				accum_file = base_file;
1389 				base_file = tmp;
1390 				rewind(base_file);
1391 				rewind(accum_file);
1392 			}
1393 		}
1394 	}
1395 
1396 done:
1397 	free(base_buf);
1398 	if (accum_buf) {
1399 		size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1400 		free(accum_buf);
1401 		if (len != accum_size)
1402 			err = got_ferror(outfile, GOT_ERR_IO);
1403 	}
1404 	if (base_file)
1405 		fclose(base_file);
1406 	if (accum_file)
1407 		fclose(accum_file);
1408 	rewind(outfile);
1409 	if (err == NULL)
1410 		*result_size = accum_size;
1411 	return err;
1412 }
1413 
1414 static const struct got_error *
1415 dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1416     struct got_delta_chain *deltas, const char *path_packfile,
1417     struct got_repository *repo)
1418 {
1419 	const struct got_error *err = NULL;
1420 	struct got_delta *delta;
1421 	uint8_t *base_buf = NULL, *accum_buf = NULL;
1422 	size_t accum_size;
1423 	uint64_t max_size;
1424 	int n = 0;
1425 
1426 	*outbuf = NULL;
1427 	*outlen = 0;
1428 
1429 	if (SIMPLEQ_EMPTY(&deltas->entries))
1430 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1431 
1432 	err = get_delta_chain_max_size(&max_size, deltas);
1433 	if (err)
1434 		return err;
1435 	accum_buf = malloc(max_size);
1436 	if (accum_buf == NULL)
1437 		return got_error_from_errno();
1438 
1439 	/* Deltas are ordered in ascending order. */
1440 	SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1441 		if (n == 0) {
1442 			struct got_pack *pack;
1443 			size_t base_len;
1444 			size_t delta_data_offset;
1445 
1446 			/* Plain object types are the delta base. */
1447 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1448 			    delta->type != GOT_OBJ_TYPE_TREE &&
1449 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1450 			    delta->type != GOT_OBJ_TYPE_TAG) {
1451 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1452 				goto done;
1453 			}
1454 
1455 			pack = get_cached_pack(path_packfile, repo);
1456 			if (pack == NULL) {
1457 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1458 				goto done;
1459 			}
1460 
1461 			delta_data_offset = delta->offset + delta->tslen;
1462 			if (delta_data_offset >= pack->filesize) {
1463 				err = got_error(GOT_ERR_PACK_OFFSET);
1464 				goto done;
1465 			}
1466 			if (pack->map) {
1467 				size_t mapoff = (size_t)delta_data_offset;
1468 				err = got_inflate_to_mem_mmap(&base_buf,
1469 				    &base_len, pack->map, mapoff,
1470 				    pack->filesize - mapoff);
1471 			} else {
1472 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1473 				    == -1) {
1474 					err = got_error_from_errno();
1475 					goto done;
1476 				}
1477 				err = got_inflate_to_mem_fd(&base_buf,
1478 				    &base_len, pack->fd);
1479 			}
1480 			if (err)
1481 				goto done;
1482 			if (base_len < max_size) {
1483 				uint8_t *p;
1484 				p = reallocarray(base_buf, 1, max_size);
1485 				if (p == NULL) {
1486 					err = got_error_from_errno();
1487 					goto done;
1488 				}
1489 				base_buf = p;
1490 			}
1491 			n++;
1492 			continue;
1493 		}
1494 
1495 		err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1496 		    delta->delta_len, accum_buf, &accum_size);
1497 		n++;
1498 		if (err)
1499 			goto done;
1500 
1501 		if (n < deltas->nentries) {
1502 			/* Accumulated delta becomes the new base. */
1503 			uint8_t *tmp = accum_buf;
1504 			accum_buf = base_buf;
1505 			base_buf = tmp;
1506 		}
1507 	}
1508 
1509 done:
1510 	free(base_buf);
1511 	if (err) {
1512 		free(accum_buf);
1513 		*outbuf = NULL;
1514 		*outlen = 0;
1515 	} else {
1516 		*outbuf = accum_buf;
1517 		*outlen = accum_size;
1518 	}
1519 	return err;
1520 }
1521 
1522 const struct got_error *
1523 got_packfile_extract_object(FILE **f, struct got_object *obj,
1524     struct got_repository *repo)
1525 {
1526 	const struct got_error *err = NULL;
1527 
1528 	*f = NULL;
1529 
1530 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1531 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1532 
1533 	*f = got_opentemp();
1534 	if (*f == NULL) {
1535 		err = got_error_from_errno();
1536 		goto done;
1537 	}
1538 
1539 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1540 		struct got_pack *pack;
1541 
1542 		pack = get_cached_pack(obj->path_packfile, repo);
1543 		if (pack == NULL) {
1544 			err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1545 			if (err)
1546 				goto done;
1547 		}
1548 
1549 		if (obj->pack_offset >= pack->filesize) {
1550 			err = got_error(GOT_ERR_PACK_OFFSET);
1551 			goto done;
1552 		}
1553 
1554 		if (pack->map) {
1555 			size_t mapoff = (size_t)obj->pack_offset;
1556 			err = got_inflate_to_file_mmap(&obj->size, pack->map,
1557 			    mapoff, pack->filesize - mapoff, *f);
1558 		} else {
1559 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1) {
1560 				err = got_error_from_errno();
1561 				goto done;
1562 			}
1563 			err = got_inflate_to_file_fd(&obj->size, pack->fd, *f);
1564 		}
1565 	} else
1566 		err = dump_delta_chain_to_file(&obj->size,
1567 		    &obj->deltas, obj->path_packfile, *f, repo);
1568 done:
1569 	if (err && *f) {
1570 		fclose(*f);
1571 		*f = NULL;
1572 	}
1573 	return err;
1574 }
1575 
1576 const struct got_error *
1577 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1578     struct got_object *obj, struct got_repository *repo)
1579 {
1580 	const struct got_error *err = NULL;
1581 
1582 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1583 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1584 
1585 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1586 		struct got_pack *pack;
1587 
1588 		pack = get_cached_pack(obj->path_packfile, repo);
1589 		if (pack == NULL) {
1590 			err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1591 			if (err)
1592 				goto done;
1593 		}
1594 
1595 		if (obj->pack_offset >= pack->filesize) {
1596 			err = got_error(GOT_ERR_PACK_OFFSET);
1597 			goto done;
1598 		}
1599 		if (pack->map) {
1600 			size_t mapoff = (size_t)obj->pack_offset;
1601 			err = got_inflate_to_mem_mmap(buf, len, pack->map,
1602 			    mapoff, pack->filesize - mapoff);
1603 		} else {
1604 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1) {
1605 				err = got_error_from_errno();
1606 				goto done;
1607 			}
1608 			err = got_inflate_to_mem_fd(buf, len, pack->fd);
1609 		}
1610 	} else
1611 		err = dump_delta_chain_to_mem(buf, len, &obj->deltas,
1612 		    obj->path_packfile, repo);
1613 done:
1614 	return err;
1615 }