op public repos

Blob

Date:: Fri Apr 28 07:52:31 2023 UTC
Message:: cache delta base fulltexts when dumping a delta chain to a file
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2018, 2019, 2020 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/uio.h>
21 #include <sys/mman.h>
22 #include <sys/resource.h>
23 #include <sys/socket.h>
24 
25 #include <fcntl.h>
26 #include <errno.h>
27 #include <stdio.h>
28 #include <stdint.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <limits.h>
32 #include <sha1.h>
33 #include <sha2.h>
34 #include <endian.h>
35 #include <unistd.h>
36 #include <zlib.h>
37 #include <imsg.h>
38 
39 #include "got_error.h"
40 #include "got_object.h"
41 #include "got_path.h"
42 
43 #include "got_lib_hash.h"
44 #include "got_lib_delta.h"
45 #include "got_lib_delta_cache.h"
46 #include "got_lib_inflate.h"
47 #include "got_lib_object.h"
48 #include "got_lib_object_qid.h"
49 #include "got_lib_object_parse.h"
50 #include "got_lib_privsep.h"
51 #include "got_lib_pack.h"
52 
53 #ifndef nitems
54 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
55 #endif
56 
57 #ifndef MIN
58 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
59 #endif
60 
61 static const struct got_error *
62 verify_fanout_table(uint32_t *fanout_table)
63 {
64 	int i;
65 
66 	for (i = 0; i < 0xff - 1; i++) {
67 		if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
68 			return got_error(GOT_ERR_BAD_PACKIDX);
69 	}
70 
71 	return NULL;
72 }
73 
74 const struct got_error *
75 got_packidx_init_hdr(struct got_packidx *p, int verify, off_t packfile_size)
76 {
77 	const struct got_error *err = NULL;
78 	enum got_hash_algorithm algo = GOT_HASH_SHA1;
79 	struct got_packidx_v2_hdr *h;
80 	struct got_hash ctx;
81 	uint8_t hash[GOT_HASH_DIGEST_MAXLEN];
82 	size_t nobj, len_fanout, len_ids, offset, remain;
83 	ssize_t n;
84 	int i;
85 
86 	got_hash_init(&ctx, algo);
87 
88 	h = &p->hdr;
89 	offset = 0;
90 	remain = p->len;
91 
92 	if (remain < sizeof(*h->magic)) {
93 		err = got_error(GOT_ERR_BAD_PACKIDX);
94 		goto done;
95 	}
96 	if (p->map)
97 		h->magic = (uint32_t *)(p->map + offset);
98 	else {
99 		h->magic = malloc(sizeof(*h->magic));
100 		if (h->magic == NULL) {
101 			err = got_error_from_errno("malloc");
102 			goto done;
103 		}
104 		n = read(p->fd, h->magic, sizeof(*h->magic));
105 		if (n < 0) {
106 			err = got_error_from_errno("read");
107 			goto done;
108 		} else if (n != sizeof(*h->magic)) {
109 			err = got_error(GOT_ERR_BAD_PACKIDX);
110 			goto done;
111 		}
112 	}
113 	if (*h->magic != htobe32(GOT_PACKIDX_V2_MAGIC)) {
114 		err = got_error(GOT_ERR_BAD_PACKIDX);
115 		goto done;
116 	}
117 	offset += sizeof(*h->magic);
118 	remain -= sizeof(*h->magic);
119 
120 	if (verify)
121 		got_hash_update(&ctx, h->magic, sizeof(*h->magic));
122 
123 	if (remain < sizeof(*h->version)) {
124 		err = got_error(GOT_ERR_BAD_PACKIDX);
125 		goto done;
126 	}
127 	if (p->map)
128 		h->version = (uint32_t *)(p->map + offset);
129 	else {
130 		h->version = malloc(sizeof(*h->version));
131 		if (h->version == NULL) {
132 			err = got_error_from_errno("malloc");
133 			goto done;
134 		}
135 		n = read(p->fd, h->version, sizeof(*h->version));
136 		if (n < 0) {
137 			err = got_error_from_errno("read");
138 			goto done;
139 		} else if (n != sizeof(*h->version)) {
140 			err = got_error(GOT_ERR_BAD_PACKIDX);
141 			goto done;
142 		}
143 	}
144 	if (*h->version != htobe32(GOT_PACKIDX_VERSION)) {
145 		err = got_error(GOT_ERR_BAD_PACKIDX);
146 		goto done;
147 	}
148 	offset += sizeof(*h->version);
149 	remain -= sizeof(*h->version);
150 
151 	if (verify)
152 		got_hash_update(&ctx, h->version, sizeof(*h->version));
153 
154 	len_fanout =
155 	    sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
156 	if (remain < len_fanout) {
157 		err = got_error(GOT_ERR_BAD_PACKIDX);
158 		goto done;
159 	}
160 	if (p->map)
161 		h->fanout_table = (uint32_t *)(p->map + offset);
162 	else {
163 		h->fanout_table = malloc(len_fanout);
164 		if (h->fanout_table == NULL) {
165 			err = got_error_from_errno("malloc");
166 			goto done;
167 		}
168 		n = read(p->fd, h->fanout_table, len_fanout);
169 		if (n < 0) {
170 			err = got_error_from_errno("read");
171 			goto done;
172 		} else if (n != len_fanout) {
173 			err = got_error(GOT_ERR_BAD_PACKIDX);
174 			goto done;
175 		}
176 	}
177 	err = verify_fanout_table(h->fanout_table);
178 	if (err)
179 		goto done;
180 	if (verify)
181 		got_hash_update(&ctx, h->fanout_table, len_fanout);
182 	offset += len_fanout;
183 	remain -= len_fanout;
184 
185 	nobj = be32toh(h->fanout_table[0xff]);
186 	len_ids = nobj * sizeof(*h->sorted_ids);
187 	if (len_ids <= nobj || len_ids > remain) {
188 		err = got_error(GOT_ERR_BAD_PACKIDX);
189 		goto done;
190 	}
191 	if (p->map)
192 		h->sorted_ids =
193 		    (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
194 	else {
195 		h->sorted_ids = malloc(len_ids);
196 		if (h->sorted_ids == NULL) {
197 			err = got_error(GOT_ERR_BAD_PACKIDX);
198 			goto done;
199 		}
200 		n = read(p->fd, h->sorted_ids, len_ids);
201 		if (n < 0)
202 			err = got_error_from_errno("read");
203 		else if (n != len_ids) {
204 			err = got_error(GOT_ERR_BAD_PACKIDX);
205 			goto done;
206 		}
207 	}
208 	if (verify)
209 		got_hash_update(&ctx, h->sorted_ids, len_ids);
210 	offset += len_ids;
211 	remain -= len_ids;
212 
213 	if (remain < nobj * sizeof(*h->crc32)) {
214 		err = got_error(GOT_ERR_BAD_PACKIDX);
215 		goto done;
216 	}
217 	if (p->map)
218 		h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
219 	else {
220 		h->crc32 = malloc(nobj * sizeof(*h->crc32));
221 		if (h->crc32 == NULL) {
222 			err = got_error_from_errno("malloc");
223 			goto done;
224 		}
225 		n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
226 		if (n < 0)
227 			err = got_error_from_errno("read");
228 		else if (n != nobj * sizeof(*h->crc32)) {
229 			err = got_error(GOT_ERR_BAD_PACKIDX);
230 			goto done;
231 		}
232 	}
233 	if (verify)
234 		got_hash_update(&ctx, h->crc32, nobj * sizeof(*h->crc32));
235 	remain -= nobj * sizeof(*h->crc32);
236 	offset += nobj * sizeof(*h->crc32);
237 
238 	if (remain < nobj * sizeof(*h->offsets)) {
239 		err = got_error(GOT_ERR_BAD_PACKIDX);
240 		goto done;
241 	}
242 	if (p->map)
243 		h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
244 	else {
245 		h->offsets = malloc(nobj * sizeof(*h->offsets));
246 		if (h->offsets == NULL) {
247 			err = got_error_from_errno("malloc");
248 			goto done;
249 		}
250 		n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
251 		if (n < 0)
252 			err = got_error_from_errno("read");
253 		else if (n != nobj * sizeof(*h->offsets)) {
254 			err = got_error(GOT_ERR_BAD_PACKIDX);
255 			goto done;
256 		}
257 	}
258 	if (verify)
259 		got_hash_update(&ctx, h->offsets, nobj * sizeof(*h->offsets));
260 	remain -= nobj * sizeof(*h->offsets);
261 	offset += nobj * sizeof(*h->offsets);
262 
263 	/* Large file offsets are contained only in files > 2GB. */
264 	if (verify || packfile_size > 0x7fffffff) {
265 		for (i = 0; i < nobj; i++) {
266 			uint32_t o = h->offsets[i];
267 			if (o & htobe32(GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX))
268 				p->nlargeobj++;
269 		}
270 	}
271 	if (p->nlargeobj == 0)
272 		goto checksum;
273 	else if (packfile_size <= 0x7fffffff) {
274 		err = got_error(GOT_ERR_BAD_PACKIDX);
275 		goto done;
276 	}
277 
278 	if (remain < p->nlargeobj * sizeof(*h->large_offsets)) {
279 		err = got_error(GOT_ERR_BAD_PACKIDX);
280 		goto done;
281 	}
282 	if (p->map)
283 		h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
284 	else {
285 		h->large_offsets = malloc(p->nlargeobj *
286 		    sizeof(*h->large_offsets));
287 		if (h->large_offsets == NULL) {
288 			err = got_error_from_errno("malloc");
289 			goto done;
290 		}
291 		n = read(p->fd, h->large_offsets,
292 		    p->nlargeobj * sizeof(*h->large_offsets));
293 		if (n < 0)
294 			err = got_error_from_errno("read");
295 		else if (n != p->nlargeobj * sizeof(*h->large_offsets)) {
296 			err = got_error(GOT_ERR_BAD_PACKIDX);
297 			goto done;
298 		}
299 	}
300 	if (verify)
301 		got_hash_update(&ctx, h->large_offsets,
302 		    p->nlargeobj * sizeof(*h->large_offsets));
303 	remain -= p->nlargeobj * sizeof(*h->large_offsets);
304 	offset += p->nlargeobj * sizeof(*h->large_offsets);
305 
306 checksum:
307 	if (remain < sizeof(*h->trailer)) {
308 		err = got_error(GOT_ERR_BAD_PACKIDX);
309 		goto done;
310 	}
311 	if (p->map)
312 		h->trailer =
313 		    (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
314 	else {
315 		h->trailer = malloc(sizeof(*h->trailer));
316 		if (h->trailer == NULL) {
317 			err = got_error_from_errno("malloc");
318 			goto done;
319 		}
320 		n = read(p->fd, h->trailer, sizeof(*h->trailer));
321 		if (n < 0)
322 			err = got_error_from_errno("read");
323 		else if (n != sizeof(*h->trailer)) {
324 			err = got_error(GOT_ERR_BAD_PACKIDX);
325 			goto done;
326 		}
327 	}
328 	if (verify) {
329 		got_hash_update(&ctx, h->trailer->packfile_sha1,
330 		    SHA1_DIGEST_LENGTH);
331 		got_hash_final(&ctx, hash);
332 		if (got_hash_cmp(ctx.algo, hash, h->trailer->packidx_sha1) != 0)
333 			err = got_error(GOT_ERR_PACKIDX_CSUM);
334 	}
335 done:
336 	return err;
337 }
338 
339 const struct got_error *
340 got_packidx_open(struct got_packidx **packidx,
341     int dir_fd, const char *relpath, int verify)
342 {
343 	const struct got_error *err = NULL;
344 	struct got_packidx *p = NULL;
345 	char *pack_relpath;
346 	struct stat idx_sb, pack_sb;
347 
348 	*packidx = NULL;
349 
350 	err = got_packidx_get_packfile_path(&pack_relpath, relpath);
351 	if (err)
352 		return err;
353 
354 	/*
355 	 * Ensure that a corresponding pack file exists.
356 	 * Some Git repositories have this problem. Git seems to ignore
357 	 * the existence of lonely pack index files but we do not.
358 	 */
359 	if (fstatat(dir_fd, pack_relpath, &pack_sb, 0) == -1) {
360 		if (errno == ENOENT) {
361 			err = got_error_fmt(GOT_ERR_LONELY_PACKIDX,
362 			    "%s", relpath);
363 		} else
364 			err = got_error_from_errno2("fstatat", pack_relpath);
365 		goto done;
366 	}
367 
368 	p = calloc(1, sizeof(*p));
369 	if (p == NULL) {
370 		err = got_error_from_errno("calloc");
371 		goto done;
372 	}
373 
374 	p->fd = openat(dir_fd, relpath, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
375 	if (p->fd == -1) {
376 		err = got_error_from_errno2("openat", relpath);
377 		goto done;
378 	}
379 
380 	if (fstat(p->fd, &idx_sb) != 0) {
381 		err = got_error_from_errno2("fstat", relpath);
382 		goto done;
383 	}
384 	p->len = idx_sb.st_size;
385 	if (p->len < sizeof(p->hdr)) {
386 		err = got_error(GOT_ERR_BAD_PACKIDX);
387 		goto done;
388 	}
389 
390 	p->path_packidx = strdup(relpath);
391 	if (p->path_packidx == NULL) {
392 		err = got_error_from_errno("strdup");
393 		goto done;
394 	}
395 
396 #ifndef GOT_PACK_NO_MMAP
397 	if (p->len > 0 && p->len <= SIZE_MAX) {
398 		p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
399 		if (p->map == MAP_FAILED) {
400 			if (errno != ENOMEM) {
401 				err = got_error_from_errno("mmap");
402 				goto done;
403 			}
404 			p->map = NULL; /* fall back to read(2) */
405 		}
406 	}
407 #endif
408 
409 	err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
410 done:
411 	if (err) {
412 		if (p)
413 			got_packidx_close(p);
414 	} else
415 		*packidx = p;
416 	free(pack_relpath);
417 	return err;
418 }
419 
420 const struct got_error *
421 got_packidx_close(struct got_packidx *packidx)
422 {
423 	const struct got_error *err = NULL;
424 
425 	free(packidx->path_packidx);
426 	if (packidx->map) {
427 		if (munmap(packidx->map, packidx->len) == -1)
428 			err = got_error_from_errno("munmap");
429 	} else {
430 		free(packidx->hdr.magic);
431 		free(packidx->hdr.version);
432 		free(packidx->hdr.fanout_table);
433 		free(packidx->hdr.sorted_ids);
434 		free(packidx->hdr.crc32);
435 		free(packidx->hdr.offsets);
436 		free(packidx->hdr.large_offsets);
437 		free(packidx->hdr.trailer);
438 	}
439 	if (close(packidx->fd) == -1 && err == NULL)
440 		err = got_error_from_errno("close");
441 	free(packidx->sorted_offsets);
442 	free(packidx->sorted_large_offsets);
443 	free(packidx);
444 
445 	return err;
446 }
447 
448 const struct got_error *
449 got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
450 {
451 	size_t size;
452 
453 	/* Packfile path contains ".pack" instead of ".idx", so add one byte. */
454 	size = strlen(path_packidx) + 2;
455 	if (size < GOT_PACKFILE_NAMELEN + 1)
456 		return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
457 
458 	*path_packfile = malloc(size);
459 	if (*path_packfile == NULL)
460 		return got_error_from_errno("malloc");
461 
462 	/* Copy up to and excluding ".idx". */
463 	if (strlcpy(*path_packfile, path_packidx,
464 	    size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
465 		return got_error(GOT_ERR_NO_SPACE);
466 
467 	if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
468 		return got_error(GOT_ERR_NO_SPACE);
469 
470 	return NULL;
471 }
472 
473 off_t
474 got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
475 {
476 	uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
477 	if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
478 		uint64_t loffset;
479 		idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
480 		if (idx < 0 || idx >= packidx->nlargeobj ||
481 		    packidx->hdr.large_offsets == NULL)
482 			return -1;
483 		loffset = be64toh(packidx->hdr.large_offsets[idx]);
484 		return (loffset > INT64_MAX ? -1 : (off_t)loffset);
485 	}
486 	return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
487 }
488 
489 int
490 got_packidx_get_object_idx(struct got_packidx *packidx,
491     struct got_object_id *id)
492 {
493 	u_int8_t id0 = id->sha1[0];
494 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
495 	int left = 0, right = totobj - 1;
496 
497 	if (id0 > 0)
498 		left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
499 
500 	while (left <= right) {
501 		struct got_packidx_object_id *oid;
502 		int i, cmp;
503 
504 		i = ((left + right) / 2);
505 		oid = &packidx->hdr.sorted_ids[i];
506 		cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
507 		if (cmp == 0)
508 			return i;
509 		else if (cmp > 0)
510 			left = i + 1;
511 		else if (cmp < 0)
512 			right = i - 1;
513 	}
514 
515 	return -1;
516 }
517 
518 static int
519 offset_cmp(const void *pa, const void *pb)
520 {
521 	const struct got_pack_offset_index *a, *b;
522 
523 	a = (const struct got_pack_offset_index *)pa;
524 	b = (const struct got_pack_offset_index *)pb;
525 
526 	if (a->offset < b->offset)
527 		return -1;
528 	else if (a->offset > b->offset)
529 		return 1;
530 
531 	return 0;
532 }
533 
534 static int
535 large_offset_cmp(const void *pa, const void *pb)
536 {
537 	const struct got_pack_large_offset_index *a, *b;
538 
539 	a = (const struct got_pack_large_offset_index *)pa;
540 	b = (const struct got_pack_large_offset_index *)pb;
541 
542 	if (a->offset < b->offset)
543 		return -1;
544 	else if (a->offset > b->offset)
545 		return 1;
546 
547 	return 0;
548 }
549 
550 static const struct got_error *
551 build_offset_index(struct got_packidx *p)
552 {
553 	uint32_t nobj = be32toh(p->hdr.fanout_table[0xff]);
554 	unsigned int i, j, k;
555 
556 	p->sorted_offsets = calloc(nobj - p->nlargeobj,
557 	    sizeof(p->sorted_offsets[0]));
558 	if (p->sorted_offsets == NULL)
559 		return got_error_from_errno("calloc");
560 
561 	if (p->nlargeobj > 0) {
562 		p->sorted_large_offsets = calloc(p->nlargeobj,
563 		    sizeof(p->sorted_large_offsets[0]));
564 		if (p->sorted_large_offsets == NULL)
565 			return got_error_from_errno("calloc");
566 	}
567 
568 	j = 0;
569 	k = 0;
570 	for (i = 0; i < nobj; i++) {
571 		uint32_t offset = be32toh(p->hdr.offsets[i]);
572 		if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
573 			uint64_t loffset;
574 			uint32_t idx;
575 			idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
576 			if (idx >= p->nlargeobj ||
577 			    p->nlargeobj == 0 ||
578 			    p->hdr.large_offsets == NULL)
579 				return got_error(GOT_ERR_BAD_PACKIDX);
580 			loffset = be64toh(p->hdr.large_offsets[idx]);
581 			p->sorted_large_offsets[j].offset = loffset;
582 			p->sorted_large_offsets[j].idx = i;
583 			j++;
584 		} else {
585 			p->sorted_offsets[k].offset = offset;
586 			p->sorted_offsets[k].idx = i;
587 			k++;
588 		}
589 	}
590 	if (j != p->nlargeobj || k != nobj - p->nlargeobj)
591 		return got_error(GOT_ERR_BAD_PACKIDX);
592 
593 	qsort(p->sorted_offsets, nobj - p->nlargeobj,
594 	    sizeof(p->sorted_offsets[0]), offset_cmp);
595 
596 	if (p->sorted_large_offsets)
597 		qsort(p->sorted_large_offsets, p->nlargeobj,
598 		    sizeof(p->sorted_large_offsets[0]), large_offset_cmp);
599 
600 	return NULL;
601 }
602 
603 const struct got_error *
604 got_packidx_get_offset_idx(int *idx, struct got_packidx *packidx, off_t offset)
605 {
606 	const struct got_error *err;
607 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
608 	int i, left, right;
609 
610 	*idx = -1;
611 
612 	if (packidx->sorted_offsets == NULL) {
613 		err = build_offset_index(packidx);
614 		if (err)
615 			return err;
616 	}
617 
618 	if (offset >= 0x7fffffff) {
619 		uint64_t lo;
620 		left = 0, right = packidx->nlargeobj - 1;
621 		while (left <= right) {
622 			i = ((left + right) / 2);
623 			lo = packidx->sorted_large_offsets[i].offset;
624 			if (lo == offset) {
625 				*idx = packidx->sorted_large_offsets[i].idx;
626 				break;
627 			} else if (offset > lo)
628 				left = i + 1;
629 			else if (offset < lo)
630 				right = i - 1;
631 		}
632 	} else {
633 		uint32_t o;
634 		left = 0, right = totobj - packidx->nlargeobj - 1;
635 		while (left <= right) {
636 			i = ((left + right) / 2);
637 			o = packidx->sorted_offsets[i].offset;
638 			if (o == offset) {
639 				*idx = packidx->sorted_offsets[i].idx;
640 				break;
641 			} else if (offset > o)
642 				left = i + 1;
643 			else if (offset < o)
644 				right = i - 1;
645 		}
646 	}
647 
648 	return NULL;
649 }
650 
651 const struct got_error *
652 got_packidx_get_object_id(struct got_object_id *id,
653     struct got_packidx *packidx, int idx)
654 {
655 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
656 	struct got_packidx_object_id *oid;
657 
658 	if (idx < 0 || idx >= totobj)
659 		return got_error(GOT_ERR_NO_OBJ);
660 
661 	oid = &packidx->hdr.sorted_ids[idx];
662 	memcpy(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
663 	return NULL;
664 }
665 
666 const struct got_error *
667 got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
668     struct got_packidx *packidx, const char *id_str_prefix)
669 {
670 	const struct got_error *err = NULL;
671 	u_int8_t id0;
672 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
673 	char hex[3];
674 	size_t prefix_len = strlen(id_str_prefix);
675 	struct got_packidx_object_id *oid;
676 	uint32_t i = 0;
677 
678 	STAILQ_INIT(matched_ids);
679 
680 	if (prefix_len < 2)
681 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
682 
683 	hex[0] = id_str_prefix[0];
684 	hex[1] = id_str_prefix[1];
685 	hex[2] = '\0';
686 	if (!got_parse_xdigit(&id0, hex))
687 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
688 
689 	if (id0 > 0)
690 		i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
691 	oid = &packidx->hdr.sorted_ids[i];
692 	while (i < totobj && oid->sha1[0] == id0) {
693 		char id_str[SHA1_DIGEST_STRING_LENGTH];
694 		struct got_object_qid *qid;
695 		int cmp;
696 
697 		if (!got_sha1_digest_to_str(oid->sha1, id_str, sizeof(id_str)))
698 		        return got_error(GOT_ERR_NO_SPACE);
699 
700 		cmp = strncmp(id_str, id_str_prefix, prefix_len);
701 		if (cmp < 0) {
702 			oid = &packidx->hdr.sorted_ids[++i];
703 			continue;
704 		} else if (cmp > 0)
705 			break;
706 
707 		err = got_object_qid_alloc_partial(&qid);
708 		if (err)
709 			break;
710 		memcpy(qid->id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
711 		STAILQ_INSERT_TAIL(matched_ids, qid, entry);
712 
713 		oid = &packidx->hdr.sorted_ids[++i];
714 	}
715 
716 	if (err)
717 		got_object_id_queue_free(matched_ids);
718 	return err;
719 }
720 
721 static void
722 set_max_datasize(void)
723 {
724 	struct rlimit rl;
725 
726 	if (getrlimit(RLIMIT_DATA, &rl) != 0)
727 		return;
728 
729 	rl.rlim_cur = rl.rlim_max;
730 	setrlimit(RLIMIT_DATA, &rl);
731 }
732 
733 const struct got_error *
734 got_pack_start_privsep_child(struct got_pack *pack, struct got_packidx *packidx)
735 {
736 	const struct got_error *err = NULL;
737 	int imsg_fds[2];
738 	pid_t pid;
739 	struct imsgbuf *ibuf;
740 
741 	ibuf = calloc(1, sizeof(*ibuf));
742 	if (ibuf == NULL)
743 		return got_error_from_errno("calloc");
744 
745 	pack->privsep_child = calloc(1, sizeof(*pack->privsep_child));
746 	if (pack->privsep_child == NULL) {
747 		err = got_error_from_errno("calloc");
748 		free(ibuf);
749 		return err;
750 	}
751 	pack->child_has_tempfiles = 0;
752 	pack->child_has_delta_outfd = 0;
753 
754 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
755 		err = got_error_from_errno("socketpair");
756 		goto done;
757 	}
758 
759 	pid = fork();
760 	if (pid == -1) {
761 		err = got_error_from_errno("fork");
762 		goto done;
763 	} else if (pid == 0) {
764 		set_max_datasize();
765 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_PACK,
766 		    pack->path_packfile);
767 		/* not reached */
768 	}
769 
770 	if (close(imsg_fds[1]) == -1)
771 		return got_error_from_errno("close");
772 	pack->privsep_child->imsg_fd = imsg_fds[0];
773 	pack->privsep_child->pid = pid;
774 	imsg_init(ibuf, imsg_fds[0]);
775 	pack->privsep_child->ibuf = ibuf;
776 
777 	err = got_privsep_init_pack_child(ibuf, pack, packidx);
778 	if (err) {
779 		const struct got_error *child_err;
780 		err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
781 		child_err = got_privsep_wait_for_child(
782 		    pack->privsep_child->pid);
783 		if (child_err && err == NULL)
784 			err = child_err;
785 	}
786 done:
787 	if (err) {
788 		free(ibuf);
789 		free(pack->privsep_child);
790 		pack->privsep_child = NULL;
791 	}
792 	return err;
793 }
794 
795 static const struct got_error *
796 pack_stop_privsep_child(struct got_pack *pack)
797 {
798 	const struct got_error *err = NULL, *close_err = NULL;
799 
800 	if (pack->privsep_child == NULL)
801 		return NULL;
802 
803 	err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
804 	if (err)
805 		return err;
806 	if (close(pack->privsep_child->imsg_fd) == -1)
807 		close_err = got_error_from_errno("close");
808 	err = got_privsep_wait_for_child(pack->privsep_child->pid);
809 	if (close_err && err == NULL)
810 		err = close_err;
811 	imsg_clear(pack->privsep_child->ibuf);
812 	free(pack->privsep_child->ibuf);
813 	free(pack->privsep_child);
814 	pack->privsep_child = NULL;
815 	return err;
816 }
817 
818 const struct got_error *
819 got_pack_close(struct got_pack *pack)
820 {
821 	const struct got_error *err = NULL;
822 
823 	err = pack_stop_privsep_child(pack);
824 	if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
825 		err = got_error_from_errno("munmap");
826 	if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
827 		err = got_error_from_errno("close");
828 	pack->fd = -1;
829 	free(pack->path_packfile);
830 	pack->path_packfile = NULL;
831 	pack->filesize = 0;
832 	if (pack->delta_cache) {
833 		got_delta_cache_free(pack->delta_cache);
834 		pack->delta_cache = NULL;
835 	}
836 
837 	/*
838 	 * Leave accumfd and basefd alone. They are managed by the
839 	 * repository layer and can be reused.
840 	 */
841 
842 	return err;
843 }
844 
845 const struct got_error *
846 got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
847     struct got_pack *pack, off_t offset)
848 {
849 	uint8_t t = 0;
850 	uint64_t s = 0;
851 	uint8_t sizeN;
852 	size_t mapoff = 0;
853 	int i = 0;
854 
855 	*len = 0;
856 
857 	if (offset >= pack->filesize)
858 		return got_error(GOT_ERR_PACK_OFFSET);
859 
860 	if (pack->map) {
861 		if (offset > SIZE_MAX) {
862 			return got_error_fmt(GOT_ERR_PACK_OFFSET,
863 			    "offset %lld overflows size_t",
864 			    (long long)offset);
865 		}
866 
867 		mapoff = (size_t)offset;
868 	} else {
869 		if (lseek(pack->fd, offset, SEEK_SET) == -1)
870 			return got_error_from_errno("lseek");
871 	}
872 
873 	do {
874 		/* We do not support size values which don't fit in 64 bit. */
875 		if (i > 9)
876 			return got_error_fmt(GOT_ERR_OBJ_TOO_LARGE,
877 			    "packfile offset %lld", (long long)offset);
878 
879 		if (pack->map) {
880 			if (mapoff + sizeof(sizeN) >= pack->filesize)
881 				return got_error(GOT_ERR_BAD_PACKFILE);
882 			sizeN = *(pack->map + mapoff);
883 			mapoff += sizeof(sizeN);
884 		} else {
885 			ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
886 			if (n < 0)
887 				return got_error_from_errno("read");
888 			if (n != sizeof(sizeN))
889 				return got_error(GOT_ERR_BAD_PACKFILE);
890 		}
891 		*len += sizeof(sizeN);
892 
893 		if (i == 0) {
894 			t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
895 			    GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
896 			s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
897 		} else {
898 			size_t shift = 4 + 7 * (i - 1);
899 			s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
900 		}
901 		i++;
902 	} while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
903 
904 	*type = t;
905 	*size = s;
906 	return NULL;
907 }
908 
909 static const struct got_error *
910 open_plain_object(struct got_object **obj, struct got_object_id *id,
911     uint8_t type, off_t offset, size_t size, int idx)
912 {
913 	*obj = calloc(1, sizeof(**obj));
914 	if (*obj == NULL)
915 		return got_error_from_errno("calloc");
916 
917 	(*obj)->type = type;
918 	(*obj)->flags = GOT_OBJ_FLAG_PACKED;
919 	(*obj)->pack_idx = idx;
920 	(*obj)->hdrlen = 0;
921 	(*obj)->size = size;
922 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
923 	(*obj)->pack_offset = offset;
924 
925 	return NULL;
926 }
927 
928 static const struct got_error *
929 parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
930     off_t delta_offset)
931 {
932 	int64_t o = 0;
933 	uint8_t offN;
934 	int i = 0;
935 
936 	*offset = 0;
937 	*len = 0;
938 
939 	do {
940 		/* We do not support offset values which don't fit in 64 bit. */
941 		if (i > 8)
942 			return got_error(GOT_ERR_NO_SPACE);
943 
944 		if (pack->map) {
945 			size_t mapoff;
946 
947 			if (delta_offset + *len > SIZE_MAX) {
948 				return got_error_fmt(GOT_ERR_PACK_OFFSET,
949 				    "mapoff %lld would overflow size_t",
950 				    (long long)delta_offset + *len);
951 			}
952 
953 			mapoff = (size_t)delta_offset + *len;
954 			if (mapoff + sizeof(offN) >= pack->filesize)
955 				return got_error(GOT_ERR_PACK_OFFSET);
956 			offN = *(pack->map + mapoff);
957 		} else {
958 			ssize_t n;
959 			n = read(pack->fd, &offN, sizeof(offN));
960 			if (n < 0)
961 				return got_error_from_errno("read");
962 			if (n != sizeof(offN))
963 				return got_error(GOT_ERR_BAD_PACKFILE);
964 		}
965 		*len += sizeof(offN);
966 
967 		if (i == 0)
968 			o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
969 		else {
970 			o++;
971 			o <<= 7;
972 			o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
973 		}
974 		i++;
975 	} while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
976 
977 	*offset = o;
978 	return NULL;
979 }
980 
981 const struct got_error *
982 got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
983     struct got_pack *pack, off_t offset, size_t tslen)
984 {
985 	const struct got_error *err;
986 	int64_t negoffset;
987 	size_t negofflen;
988 
989 	*len = 0;
990 
991 	err = parse_negative_offset(&negoffset, &negofflen, pack,
992 	    offset + tslen);
993 	if (err)
994 		return err;
995 
996 	/* Compute the base object's offset (must be in the same pack file). */
997 	*base_offset = (offset - negoffset);
998 	if (*base_offset <= 0)
999 		return got_error(GOT_ERR_BAD_PACKFILE);
1000 
1001 	*len = negofflen;
1002 	return NULL;
1003 }
1004 
1005 static const struct got_error *
1006 read_delta_data(uint8_t **delta_buf, size_t *delta_len,
1007     size_t *delta_compressed_len, size_t delta_data_offset,
1008     struct got_pack *pack)
1009 {
1010 	const struct got_error *err = NULL;
1011 	size_t consumed = 0;
1012 
1013 	if (pack->map) {
1014 		if (delta_data_offset >= pack->filesize)
1015 			return got_error(GOT_ERR_PACK_OFFSET);
1016 		err = got_inflate_to_mem_mmap(delta_buf, delta_len,
1017 		    &consumed, NULL, pack->map, delta_data_offset,
1018 		    pack->filesize - delta_data_offset);
1019 		if (err)
1020 			return err;
1021 	} else {
1022 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
1023 			return got_error_from_errno("lseek");
1024 		err = got_inflate_to_mem_fd(delta_buf, delta_len,
1025 		    &consumed, NULL, 0, pack->fd);
1026 		if (err)
1027 			return err;
1028 	}
1029 
1030 	if (delta_compressed_len)
1031 		*delta_compressed_len = consumed;
1032 
1033 	return NULL;
1034 }
1035 
1036 static const struct got_error *
1037 add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
1038     int delta_type, size_t delta_size, off_t delta_data_offset)
1039 {
1040 	struct got_delta *delta;
1041 
1042 	delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
1043 	    delta_data_offset);
1044 	if (delta == NULL)
1045 		return got_error_from_errno("got_delta_open");
1046 	/* delta is freed in got_object_close() */
1047 	deltas->nentries++;
1048 	STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
1049 	return NULL;
1050 }
1051 
1052 static const struct got_error *
1053 resolve_offset_delta(struct got_delta_chain *deltas,
1054     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1055     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1056 {
1057 	const struct got_error *err;
1058 	off_t base_offset;
1059 	uint8_t base_type;
1060 	uint64_t base_size;
1061 	size_t base_tslen;
1062 	off_t delta_data_offset;
1063 	size_t consumed;
1064 
1065 	err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
1066 	    delta_offset, tslen);
1067 	if (err)
1068 		return err;
1069 
1070 	delta_data_offset = delta_offset + tslen + consumed;
1071 	if (delta_data_offset >= pack->filesize)
1072 		return got_error(GOT_ERR_PACK_OFFSET);
1073 
1074 	if (pack->map == NULL) {
1075 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1076 		if (delta_data_offset == -1)
1077 			return got_error_from_errno("lseek");
1078 	}
1079 
1080 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1081 	    delta_data_offset);
1082 	if (err)
1083 		return err;
1084 
1085 	/* An offset delta must be in the same packfile. */
1086 	if (base_offset >= pack->filesize)
1087 		return got_error(GOT_ERR_PACK_OFFSET);
1088 
1089 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1090 	    &base_tslen, pack, base_offset);
1091 	if (err)
1092 		return err;
1093 
1094 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1095 	    base_tslen, base_type, base_size, recursion - 1);
1096 }
1097 
1098 const struct got_error *
1099 got_pack_parse_ref_delta(struct got_object_id *id,
1100     struct got_pack *pack, off_t delta_offset, int tslen)
1101 {
1102 	if (pack->map) {
1103 		size_t mapoff;
1104 
1105 		if (delta_offset + tslen > SIZE_MAX) {
1106 			return got_error_fmt(GOT_ERR_PACK_OFFSET,
1107 			    "mapoff %lld would overflow size_t",
1108 			    (long long)delta_offset + tslen);
1109 		}
1110 
1111 		mapoff = delta_offset + tslen;
1112 		if (mapoff + sizeof(*id) >= pack->filesize)
1113 			return got_error(GOT_ERR_PACK_OFFSET);
1114 		memcpy(id, pack->map + mapoff, sizeof(*id));
1115 	} else {
1116 		ssize_t n;
1117 		n = read(pack->fd, id, sizeof(*id));
1118 		if (n < 0)
1119 			return got_error_from_errno("read");
1120 		if (n != sizeof(*id))
1121 			return got_error(GOT_ERR_BAD_PACKFILE);
1122 	}
1123 
1124 	return NULL;
1125 }
1126 
1127 static const struct got_error *
1128 resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1129     struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1130     size_t delta_size, unsigned int recursion)
1131 {
1132 	const struct got_error *err;
1133 	struct got_object_id id;
1134 	int idx;
1135 	off_t base_offset;
1136 	uint8_t base_type;
1137 	uint64_t base_size;
1138 	size_t base_tslen;
1139 	off_t delta_data_offset;
1140 
1141 	if (delta_offset + tslen >= pack->filesize)
1142 		return got_error(GOT_ERR_PACK_OFFSET);
1143 
1144 	err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1145 	if (err)
1146 		return err;
1147 	if (pack->map) {
1148 		delta_data_offset = delta_offset + tslen + SHA1_DIGEST_LENGTH;
1149 	} else {
1150 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1151 		if (delta_data_offset == -1)
1152 			return got_error_from_errno("lseek");
1153 	}
1154 
1155 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1156 	    delta_data_offset);
1157 	if (err)
1158 		return err;
1159 
1160 	/* Delta base must be in the same pack file. */
1161 	idx = got_packidx_get_object_idx(packidx, &id);
1162 	if (idx == -1)
1163 		return got_error(GOT_ERR_NO_OBJ);
1164 
1165 	base_offset = got_packidx_get_object_offset(packidx, idx);
1166 	if (base_offset == -1)
1167 		return got_error(GOT_ERR_BAD_PACKIDX);
1168 
1169 	if (base_offset >= pack->filesize)
1170 		return got_error(GOT_ERR_PACK_OFFSET);
1171 
1172 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1173 	    &base_tslen, pack, base_offset);
1174 	if (err)
1175 		return err;
1176 
1177 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1178 	    base_tslen, base_type, base_size, recursion - 1);
1179 }
1180 
1181 const struct got_error *
1182 got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1183     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1184     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1185 {
1186 	const struct got_error *err = NULL;
1187 
1188 	if (--recursion == 0)
1189 		return got_error(GOT_ERR_RECURSION);
1190 
1191 	switch (delta_type) {
1192 	case GOT_OBJ_TYPE_COMMIT:
1193 	case GOT_OBJ_TYPE_TREE:
1194 	case GOT_OBJ_TYPE_BLOB:
1195 	case GOT_OBJ_TYPE_TAG:
1196 		/* Plain types are the final delta base. Recursion ends. */
1197 		err = add_delta(deltas, delta_offset, tslen, delta_type,
1198 		    delta_size, 0);
1199 		break;
1200 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1201 		err = resolve_offset_delta(deltas, packidx, pack,
1202 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1203 		break;
1204 	case GOT_OBJ_TYPE_REF_DELTA:
1205 		err = resolve_ref_delta(deltas, packidx, pack,
1206 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1207 		break;
1208 	default:
1209 		return got_error(GOT_ERR_OBJ_TYPE);
1210 	}
1211 
1212 	return err;
1213 }
1214 
1215 static const struct got_error *
1216 open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1217     struct got_pack *pack, struct got_object_id *id, off_t offset,
1218     size_t tslen, int delta_type, size_t delta_size, int idx)
1219 {
1220 	const struct got_error *err = NULL;
1221 	int resolved_type;
1222 
1223 	*obj = calloc(1, sizeof(**obj));
1224 	if (*obj == NULL)
1225 		return got_error_from_errno("calloc");
1226 
1227 	(*obj)->flags = 0;
1228 	(*obj)->hdrlen = 0;
1229 	(*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1230 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1231 	(*obj)->pack_offset = offset + tslen;
1232 
1233 	STAILQ_INIT(&(*obj)->deltas.entries);
1234 	(*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1235 	(*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1236 	(*obj)->pack_idx = idx;
1237 
1238 	err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1239 	    offset, tslen, delta_type, delta_size,
1240 	    GOT_DELTA_CHAIN_RECURSION_MAX);
1241 	if (err)
1242 		goto done;
1243 
1244 	err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1245 	if (err)
1246 		goto done;
1247 	(*obj)->type = resolved_type;
1248 done:
1249 	if (err) {
1250 		got_object_close(*obj);
1251 		*obj = NULL;
1252 	}
1253 	return err;
1254 }
1255 
1256 const struct got_error *
1257 got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1258     struct got_packidx *packidx, int idx, struct got_object_id *id)
1259 {
1260 	const struct got_error *err = NULL;
1261 	off_t offset;
1262 	uint8_t type;
1263 	uint64_t size;
1264 	size_t tslen;
1265 
1266 	*obj = NULL;
1267 
1268 	offset = got_packidx_get_object_offset(packidx, idx);
1269 	if (offset == -1)
1270 		return got_error(GOT_ERR_BAD_PACKIDX);
1271 
1272 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1273 	    pack, offset);
1274 	if (err)
1275 		return err;
1276 
1277 	switch (type) {
1278 	case GOT_OBJ_TYPE_COMMIT:
1279 	case GOT_OBJ_TYPE_TREE:
1280 	case GOT_OBJ_TYPE_BLOB:
1281 	case GOT_OBJ_TYPE_TAG:
1282 		err = open_plain_object(obj, id, type, offset + tslen,
1283 		    size, idx);
1284 		break;
1285 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1286 	case GOT_OBJ_TYPE_REF_DELTA:
1287 		err = open_delta_object(obj, packidx, pack, id, offset,
1288 		    tslen, type, size, idx);
1289 		break;
1290 	default:
1291 		err = got_error(GOT_ERR_OBJ_TYPE);
1292 		break;
1293 	}
1294 
1295 	return err;
1296 }
1297 
1298 const struct got_error *
1299 got_pack_get_delta_chain_max_size(uint64_t *max_size,
1300     struct got_delta_chain *deltas, struct got_pack *pack)
1301 {
1302 	struct got_delta *delta;
1303 	uint64_t base_size = 0, result_size = 0;
1304 
1305 	*max_size = 0;
1306 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1307 		/* Plain object types are the delta base. */
1308 		if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1309 		    delta->type != GOT_OBJ_TYPE_TREE &&
1310 		    delta->type != GOT_OBJ_TYPE_BLOB &&
1311 		    delta->type != GOT_OBJ_TYPE_TAG) {
1312 			const struct got_error *err;
1313 			uint8_t *delta_buf = NULL;
1314 			size_t delta_len;
1315 			int cached = 1;
1316 
1317 			if (pack->delta_cache) {
1318 				got_delta_cache_get(&delta_buf, &delta_len,
1319 				    NULL, NULL, pack->delta_cache,
1320 				    delta->data_offset);
1321 			}
1322 			if (delta_buf == NULL) {
1323 				cached = 0;
1324 				err = read_delta_data(&delta_buf, &delta_len,
1325 				    NULL, delta->data_offset, pack);
1326 				if (err)
1327 					return err;
1328 			}
1329 			if (pack->delta_cache && !cached) {
1330 				err = got_delta_cache_add(pack->delta_cache,
1331 				    delta->data_offset, delta_buf, delta_len);
1332 				if (err == NULL)
1333 					cached = 1;
1334 				else if (err->code != GOT_ERR_NO_SPACE) {
1335 					free(delta_buf);
1336 					return err;
1337 				}
1338 			}
1339 			err = got_delta_get_sizes(&base_size, &result_size,
1340 			    delta_buf, delta_len);
1341 			if (!cached)
1342 				free(delta_buf);
1343 			if (err)
1344 				return err;
1345 		} else
1346 			base_size = delta->size;
1347 		if (base_size > *max_size)
1348 			*max_size = base_size;
1349 		if (result_size > *max_size)
1350 			*max_size = result_size;
1351 	}
1352 
1353 	return NULL;
1354 }
1355 
1356 const struct got_error *
1357 got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1358     struct got_pack *pack)
1359 {
1360 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1361 		return got_error(GOT_ERR_OBJ_TYPE);
1362 
1363 	return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1364 }
1365 
1366 const struct got_error *
1367 got_pack_dump_delta_chain_to_file(size_t *result_size,
1368     struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1369     FILE *base_file, FILE *accum_file)
1370 {
1371 	const struct got_error *err = NULL;
1372 	struct got_delta *delta;
1373 	uint8_t *base_buf = NULL, *accum_buf = NULL;
1374 	size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0;
1375 	/* We process small enough files entirely in memory for speed. */
1376 	const size_t max_bufsize = GOT_DELTA_RESULT_SIZE_CACHED_MAX;
1377 	uint64_t max_size = 0;
1378 	int n = 0;
1379 
1380 	*result_size = 0;
1381 
1382 	if (STAILQ_EMPTY(&deltas->entries))
1383 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1384 
1385 	if (pack->delta_cache) {
1386 		uint8_t *delta_buf = NULL, *fulltext = NULL;
1387 		size_t delta_len, fulltext_len;
1388 
1389 		delta = STAILQ_LAST(&deltas->entries, got_delta, entry);
1390 		got_delta_cache_get(&delta_buf, &delta_len,
1391 		    &fulltext, &fulltext_len,
1392 		    pack->delta_cache, delta->data_offset);
1393 		if (fulltext) {
1394 			size_t w;
1395 
1396 			w = fwrite(fulltext, 1, fulltext_len, outfile);
1397 			if (w != fulltext_len)
1398 				return got_ferror(outfile, GOT_ERR_IO);
1399 			if (fflush(outfile) != 0)
1400 				return got_error_from_errno("fflush");
1401 			*result_size = fulltext_len;
1402 			return NULL;
1403 		}
1404 	}
1405 
1406 	if (fseeko(base_file, 0L, SEEK_SET) == -1)
1407 		return got_error_from_errno("fseeko");
1408 	if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1409 		return got_error_from_errno("fseeko");
1410 
1411 	/* Deltas are ordered in ascending order. */
1412 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1413 		uint8_t *delta_buf = NULL, *fulltext = NULL;
1414 		size_t delta_len, fulltext_len;
1415 		uint64_t base_size, result_size = 0;
1416 		int cached = 1;
1417 		if (n == 0) {
1418 			size_t mapoff;
1419 			off_t delta_data_offset;
1420 
1421 			/* Plain object types are the delta base. */
1422 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1423 			    delta->type != GOT_OBJ_TYPE_TREE &&
1424 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1425 			    delta->type != GOT_OBJ_TYPE_TAG) {
1426 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1427 				goto done;
1428 			}
1429 
1430 			delta_data_offset = delta->offset + delta->tslen;
1431 			if (delta_data_offset >= pack->filesize) {
1432 				err = got_error(GOT_ERR_PACK_OFFSET);
1433 				goto done;
1434 			}
1435 			if (pack->map == NULL) {
1436 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1437 				    == -1) {
1438 					err = got_error_from_errno("lseek");
1439 					goto done;
1440 				}
1441 			}
1442 			if (delta->size > max_size)
1443 				max_size = delta->size;
1444 			if (max_size > max_bufsize) {
1445 				if (pack->map) {
1446 					if (delta_data_offset > SIZE_MAX) {
1447 						return got_error_fmt(
1448 						    GOT_ERR_RANGE,
1449 						    "delta offset %lld "
1450 						    "overflows size_t",
1451 						    (long long)
1452 						    delta_data_offset);
1453 					}
1454 
1455 					mapoff = delta_data_offset;
1456 					err = got_inflate_to_file_mmap(
1457 					    &base_bufsz, NULL, NULL, pack->map,
1458 					    mapoff, pack->filesize - mapoff,
1459 					    base_file);
1460 				} else
1461 					err = got_inflate_to_file_fd(
1462 					    &base_bufsz, NULL, NULL, pack->fd,
1463 					    base_file);
1464 			} else {
1465 				accum_buf = malloc(max_size);
1466 				if (accum_buf == NULL) {
1467 					err = got_error_from_errno("malloc");
1468 					goto done;
1469 				}
1470 				accum_bufsz = max_size;
1471 				if (pack->map) {
1472 					if (delta_data_offset > SIZE_MAX) {
1473 						return got_error_fmt(
1474 						    GOT_ERR_RANGE,
1475 						    "delta offset %lld "
1476 						    "overflows size_t",
1477 						    (long long)
1478 						    delta_data_offset);
1479 					}
1480 
1481 					mapoff = delta_data_offset;
1482 					err = got_inflate_to_mem_mmap(&base_buf,
1483 					    &base_bufsz, NULL, NULL,
1484 					    pack->map, mapoff,
1485 					    pack->filesize - mapoff);
1486 				} else
1487 					err = got_inflate_to_mem_fd(&base_buf,
1488 					    &base_bufsz, NULL, NULL, max_size,
1489 					    pack->fd);
1490 			}
1491 			if (err)
1492 				goto done;
1493 			n++;
1494 			if (base_buf == NULL)
1495 				rewind(base_file);
1496 			else if (pack->delta_cache && fulltext == NULL) {
1497 				err = got_delta_cache_add(pack->delta_cache,
1498 				    delta_data_offset, NULL, 0);
1499 				if (err) {
1500 					if (err->code != GOT_ERR_NO_SPACE)
1501 						goto done;
1502 					err = NULL;
1503 				} else {
1504 					err = got_delta_cache_add_fulltext(
1505 					    pack->delta_cache,
1506 					    delta_data_offset,
1507 					    base_buf, base_bufsz);
1508 					if (err &&
1509 					    err->code != GOT_ERR_NO_SPACE)
1510 						goto done;
1511 					err = NULL;
1512 				}
1513 			}
1514 			continue;
1515 		}
1516 
1517 		if (pack->delta_cache) {
1518 			got_delta_cache_get(&delta_buf, &delta_len,
1519 			    &fulltext, &fulltext_len,
1520 			    pack->delta_cache, delta->data_offset);
1521 		}
1522 		if (delta_buf == NULL) {
1523 			cached = 0;
1524 			err = read_delta_data(&delta_buf, &delta_len, NULL,
1525 			    delta->data_offset, pack);
1526 			if (err)
1527 				goto done;
1528 		}
1529 		if (pack->delta_cache && !cached) {
1530 			err = got_delta_cache_add(pack->delta_cache,
1531 			    delta->data_offset, delta_buf, delta_len);
1532 			if (err == NULL)
1533 				cached = 1;
1534 			else if (err->code != GOT_ERR_NO_SPACE) {
1535 				free(delta_buf);
1536 				goto done;
1537 			}
1538 		}
1539 
1540 		err = got_delta_get_sizes(&base_size, &result_size,
1541 		    delta_buf, delta_len);
1542 		if (err) {
1543 			if (!cached)
1544 				free(delta_buf);
1545 			goto done;
1546 		}
1547 		if (base_size > max_size)
1548 			max_size = base_size;
1549 		if (result_size > max_size)
1550 			max_size = result_size;
1551 		if (fulltext_len > max_size)
1552 			max_size = fulltext_len;
1553 
1554 		if (base_buf && max_size > max_bufsize) {
1555 			/* Switch from buffers to temporary files. */
1556 			size_t w = fwrite(base_buf, 1, base_bufsz,
1557 			    base_file);
1558 			if (w != base_bufsz) {
1559 				err = got_ferror(outfile, GOT_ERR_IO);
1560 				if (!cached)
1561 					free(delta_buf);
1562 				goto done;
1563 			}
1564 			free(base_buf);
1565 			base_buf = NULL;
1566 			free(accum_buf);
1567 			accum_buf = NULL;
1568 		}
1569 
1570 		if (base_buf && max_size > base_bufsz) {
1571 			uint8_t *p = realloc(base_buf, max_size);
1572 			if (p == NULL) {
1573 				err = got_error_from_errno("realloc");
1574 				if (!cached)
1575 					free(delta_buf);
1576 				goto done;
1577 			}
1578 			base_buf = p;
1579 			base_bufsz = max_size;
1580 		}
1581 
1582 		if (accum_buf && max_size > accum_bufsz) {
1583 			uint8_t *p = realloc(accum_buf, max_size);
1584 			if (p == NULL) {
1585 				err = got_error_from_errno("realloc");
1586 				if (!cached)
1587 					free(delta_buf);
1588 				goto done;
1589 			}
1590 			accum_buf = p;
1591 			accum_bufsz = max_size;
1592 		}
1593 
1594 		if (base_buf) {
1595 			if (fulltext) {
1596 				memcpy(accum_buf, fulltext, fulltext_len);
1597 				accum_size = fulltext_len;
1598 				err = NULL;
1599 			} else {
1600 				err = got_delta_apply_in_mem(base_buf,
1601 				    base_bufsz, delta_buf, delta_len,
1602 				    accum_buf, &accum_size, max_size);
1603 			}
1604 			n++;
1605 			if (!cached)
1606 				free(delta_buf);
1607 			if (err)
1608 				goto done;
1609 			if (fulltext == NULL) {
1610 				err = got_delta_cache_add_fulltext(
1611 				    pack->delta_cache, delta->data_offset,
1612 				    accum_buf, accum_size);
1613 				if (err) {
1614 					if (err->code != GOT_ERR_NO_SPACE)
1615 						goto done;
1616 					err = NULL;
1617 				}
1618 			}
1619 		} else {
1620 			err = got_delta_apply(base_file, delta_buf,
1621 			    delta_len,
1622 			    /* Final delta application writes to output file. */
1623 			    ++n < deltas->nentries ? accum_file : outfile,
1624 			    &accum_size);
1625 			if (!cached)
1626 				free(delta_buf);
1627 			if (err)
1628 				goto done;
1629 		}
1630 
1631 		if (n < deltas->nentries) {
1632 			/* Accumulated delta becomes the new base. */
1633 			if (base_buf) {
1634 				uint8_t *tmp = accum_buf;
1635 				size_t tmp_size = accum_bufsz;
1636 				accum_buf = base_buf;
1637 				accum_bufsz = base_bufsz;
1638 				base_buf = tmp;
1639 				base_bufsz = tmp_size;
1640 			} else {
1641 				FILE *tmp = accum_file;
1642 				accum_file = base_file;
1643 				base_file = tmp;
1644 				rewind(base_file);
1645 				rewind(accum_file);
1646 			}
1647 		}
1648 	}
1649 
1650 done:
1651 	free(base_buf);
1652 	if (accum_buf) {
1653 		size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1654 		free(accum_buf);
1655 		if (len != accum_size)
1656 			err = got_ferror(outfile, GOT_ERR_IO);
1657 	}
1658 	rewind(outfile);
1659 	if (err == NULL)
1660 		*result_size = accum_size;
1661 	return err;
1662 }
1663 
1664 const struct got_error *
1665 got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1666     struct got_delta_chain *deltas, struct got_pack *pack)
1667 {
1668 	const struct got_error *err = NULL;
1669 	struct got_delta *delta;
1670 	uint8_t *base_buf = NULL, *accum_buf = NULL;
1671 	size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0;
1672 	uint64_t max_size = 0;
1673 	int n = 0;
1674 
1675 	*outbuf = NULL;
1676 	*outlen = 0;
1677 
1678 	if (STAILQ_EMPTY(&deltas->entries))
1679 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1680 
1681 	if (pack->delta_cache) {
1682 		uint8_t *delta_buf = NULL, *fulltext = NULL;
1683 		size_t delta_len, fulltext_len;
1684 
1685 		delta = STAILQ_LAST(&deltas->entries, got_delta, entry);
1686 		got_delta_cache_get(&delta_buf, &delta_len,
1687 		    &fulltext, &fulltext_len,
1688 		    pack->delta_cache, delta->data_offset);
1689 		if (fulltext) {
1690 			*outbuf = malloc(fulltext_len);
1691 			if (*outbuf == NULL)
1692 				return got_error_from_errno("malloc");
1693 			memcpy(*outbuf, fulltext, fulltext_len);
1694 			*outlen = fulltext_len;
1695 			return NULL;
1696 		}
1697 	}
1698 
1699 	/* Deltas are ordered in ascending order. */
1700 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1701 		uint8_t *delta_buf = NULL, *fulltext = NULL;
1702 		size_t delta_len, fulltext_len = 0;
1703 		uint64_t base_size, result_size = 0;
1704 		int cached = 1;
1705 		if (n == 0) {
1706 			off_t delta_data_offset;
1707 
1708 			/* Plain object types are the delta base. */
1709 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1710 			    delta->type != GOT_OBJ_TYPE_TREE &&
1711 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1712 			    delta->type != GOT_OBJ_TYPE_TAG) {
1713 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1714 				goto done;
1715 			}
1716 
1717 			delta_data_offset = delta->offset + delta->tslen;
1718 			if (delta_data_offset >= pack->filesize) {
1719 				err = got_error(GOT_ERR_PACK_OFFSET);
1720 				goto done;
1721 			}
1722 
1723 			if (pack->delta_cache) {
1724 				got_delta_cache_get(&delta_buf, &delta_len,
1725 				    &fulltext, &fulltext_len,
1726 				    pack->delta_cache, delta_data_offset);
1727 			}
1728 
1729 			if (delta->size > max_size)
1730 				max_size = delta->size;
1731 			if (delta->size > fulltext_len)
1732 				max_size = fulltext_len;
1733 
1734 			if (fulltext) {
1735 				base_buf = malloc(fulltext_len);
1736 				if (base_buf == NULL) {
1737 					err = got_error_from_errno("malloc");
1738 					goto done;
1739 				}
1740 				memcpy(base_buf, fulltext, fulltext_len);
1741 				base_bufsz = fulltext_len;
1742 			} else if (pack->map) {
1743 				size_t mapoff;
1744 
1745 				if (delta_data_offset > SIZE_MAX) {
1746 					return got_error_fmt(GOT_ERR_RANGE,
1747 					    "delta %lld offset would "
1748 					    "overflow size_t",
1749 					    (long long)delta_data_offset);
1750 				}
1751 
1752 				mapoff = delta_data_offset;
1753 				err = got_inflate_to_mem_mmap(&base_buf,
1754 				    &base_bufsz, NULL, NULL, pack->map,
1755 				    mapoff, pack->filesize - mapoff);
1756 			} else {
1757 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1758 				    == -1) {
1759 					err = got_error_from_errno("lseek");
1760 					goto done;
1761 				}
1762 				err = got_inflate_to_mem_fd(&base_buf,
1763 				    &base_bufsz, NULL, NULL, max_size,
1764 				    pack->fd);
1765 			}
1766 			if (err)
1767 				goto done;
1768 			n++;
1769 
1770 			if (pack->delta_cache && fulltext == NULL) {
1771 				err = got_delta_cache_add(pack->delta_cache,
1772 				    delta_data_offset, NULL, 0);
1773 				if (err) {
1774 					if (err->code != GOT_ERR_NO_SPACE)
1775 						goto done;
1776 					err = NULL;
1777 				} else {
1778 					err = got_delta_cache_add_fulltext(
1779 					    pack->delta_cache,
1780 					    delta_data_offset,
1781 					    base_buf, base_bufsz);
1782 					if (err &&
1783 					    err->code != GOT_ERR_NO_SPACE)
1784 						goto done;
1785 					err = NULL;
1786 				}
1787 			}
1788 			continue;
1789 		}
1790 
1791 		if (pack->delta_cache) {
1792 			got_delta_cache_get(&delta_buf, &delta_len,
1793 			    &fulltext, &fulltext_len,
1794 			    pack->delta_cache, delta->data_offset);
1795 		}
1796 		if (delta_buf == NULL) {
1797 			cached = 0;
1798 			err = read_delta_data(&delta_buf, &delta_len, NULL,
1799 			    delta->data_offset, pack);
1800 			if (err)
1801 				goto done;
1802 		}
1803 		if (pack->delta_cache && !cached) {
1804 			err = got_delta_cache_add(pack->delta_cache,
1805 			    delta->data_offset, delta_buf, delta_len);
1806 			if (err == NULL)
1807 				cached = 1;
1808 			else if (err->code != GOT_ERR_NO_SPACE) {
1809 				free(delta_buf);
1810 				goto done;
1811 			}
1812 		}
1813 
1814 		err = got_delta_get_sizes(&base_size, &result_size,
1815 		    delta_buf, delta_len);
1816 		if (err) {
1817 			if (!cached)
1818 				free(delta_buf);
1819 			goto done;
1820 		}
1821 		if (base_size > max_size)
1822 			max_size = base_size;
1823 		if (result_size > max_size)
1824 			max_size = result_size;
1825 		if (fulltext_len > max_size)
1826 			max_size = fulltext_len;
1827 
1828 		if (max_size > base_bufsz) {
1829 			uint8_t *p = realloc(base_buf, max_size);
1830 			if (p == NULL) {
1831 				err = got_error_from_errno("realloc");
1832 				if (!cached)
1833 					free(delta_buf);
1834 				goto done;
1835 			}
1836 			base_buf = p;
1837 			base_bufsz = max_size;
1838 		}
1839 
1840 		if (max_size > accum_bufsz) {
1841 			uint8_t *p = realloc(accum_buf, max_size);
1842 			if (p == NULL) {
1843 				err = got_error_from_errno("realloc");
1844 				if (!cached)
1845 					free(delta_buf);
1846 				goto done;
1847 			}
1848 			accum_buf = p;
1849 			accum_bufsz = max_size;
1850 		}
1851 
1852 		if (fulltext) {
1853 			memcpy(accum_buf, fulltext, fulltext_len);
1854 			accum_size = fulltext_len;
1855 			err = NULL;
1856 		} else {
1857 			err = got_delta_apply_in_mem(base_buf, base_bufsz,
1858 			    delta_buf, delta_len, accum_buf,
1859 			    &accum_size, max_size);
1860 		}
1861 		if (!cached)
1862 			free(delta_buf);
1863 		n++;
1864 		if (err)
1865 			goto done;
1866 
1867 		if (fulltext == NULL) {
1868 			err = got_delta_cache_add_fulltext(pack->delta_cache,
1869 			    delta->data_offset, accum_buf, accum_size);
1870 			if (err) {
1871 				if (err->code != GOT_ERR_NO_SPACE)
1872 					goto done;
1873 				err = NULL;
1874 			}
1875 		}
1876 
1877 		if (n < deltas->nentries) {
1878 			/* Accumulated delta becomes the new base. */
1879 			uint8_t *tmp = accum_buf;
1880 			size_t tmp_size = accum_bufsz;
1881 			accum_buf = base_buf;
1882 			accum_bufsz = base_bufsz;
1883 			base_buf = tmp;
1884 			base_bufsz = tmp_size;
1885 		}
1886 	}
1887 
1888 done:
1889 	free(base_buf);
1890 	if (err) {
1891 		free(accum_buf);
1892 		*outbuf = NULL;
1893 		*outlen = 0;
1894 	} else {
1895 		*outbuf = accum_buf;
1896 		*outlen = accum_size;
1897 	}
1898 	return err;
1899 }
1900 
1901 const struct got_error *
1902 got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1903     FILE *outfile, FILE *base_file, FILE *accum_file)
1904 {
1905 	const struct got_error *err = NULL;
1906 
1907 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1908 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1909 
1910 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1911 		if (obj->pack_offset >= pack->filesize)
1912 			return got_error(GOT_ERR_PACK_OFFSET);
1913 
1914 		if (pack->map) {
1915 			size_t mapoff;
1916 
1917 			if (obj->pack_offset > SIZE_MAX) {
1918 				return got_error_fmt(GOT_ERR_RANGE,
1919 				    "pack offset %lld would overflow size_t",
1920 				    (long long)obj->pack_offset);
1921 			}
1922 
1923 			mapoff = obj->pack_offset;
1924 			err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1925 			    pack->map, mapoff, pack->filesize - mapoff,
1926 			    outfile);
1927 		} else {
1928 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1929 				return got_error_from_errno("lseek");
1930 			err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1931 			    pack->fd, outfile);
1932 		}
1933 	} else
1934 		err = got_pack_dump_delta_chain_to_file(&obj->size,
1935 		    &obj->deltas, pack, outfile, base_file, accum_file);
1936 
1937 	return err;
1938 }
1939 
1940 const struct got_error *
1941 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1942     struct got_object *obj, struct got_pack *pack)
1943 {
1944 	const struct got_error *err = NULL;
1945 
1946 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1947 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1948 
1949 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1950 		if (obj->pack_offset >= pack->filesize)
1951 			return got_error(GOT_ERR_PACK_OFFSET);
1952 		if (pack->map) {
1953 			size_t mapoff;
1954 
1955 			if (obj->pack_offset > SIZE_MAX) {
1956 				return got_error_fmt(GOT_ERR_RANGE,
1957 				    "pack offset %lld would overflow size_t",
1958 				    (long long)obj->pack_offset);
1959 			}
1960 
1961 			mapoff = obj->pack_offset;
1962 			err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1963 			    pack->map, mapoff, pack->filesize - mapoff);
1964 		} else {
1965 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1966 				return got_error_from_errno("lseek");
1967 			err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1968 			    obj->size, pack->fd);
1969 		}
1970 	} else
1971 		err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1972 		    pack);
1973 
1974 	return err;
1975 }
1976 
1977 static const struct got_error *
1978 read_raw_delta_data(uint8_t **delta_buf, size_t *delta_len,
1979     size_t *delta_len_compressed, uint64_t *base_size, uint64_t *result_size,
1980     off_t delta_data_offset, struct got_pack *pack, struct got_packidx *packidx)
1981 {
1982 	const struct got_error *err = NULL;
1983 
1984 	/* Validate decompression and obtain the decompressed size. */
1985 	err = read_delta_data(delta_buf, delta_len, delta_len_compressed,
1986 	    delta_data_offset, pack);
1987 	if (err)
1988 		return err;
1989 
1990 	/* Read delta base/result sizes from head of delta stream. */
1991 	err = got_delta_get_sizes(base_size, result_size,
1992 	    *delta_buf, *delta_len);
1993 	if (err)
1994 		goto done;
1995 
1996 	/* Discard decompressed delta and read it again in compressed form. */
1997 	free(*delta_buf);
1998 	*delta_buf = malloc(*delta_len_compressed);
1999 	if (*delta_buf == NULL) {
2000 		err = got_error_from_errno("malloc");
2001 		goto done;
2002 	}
2003 	if (pack->map) {
2004 		if (delta_data_offset >= pack->filesize) {
2005 			err = got_error(GOT_ERR_PACK_OFFSET);
2006 			goto done;
2007 		}
2008 		memcpy(*delta_buf, pack->map + delta_data_offset,
2009 		    *delta_len_compressed);
2010 	} else {
2011 		ssize_t n;
2012 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
2013 			err = got_error_from_errno("lseek");
2014 			goto done;
2015 		}
2016 		n = read(pack->fd, *delta_buf, *delta_len_compressed);
2017 		if (n < 0) {
2018 			err = got_error_from_errno("read");
2019 			goto done;
2020 		} else if (n != *delta_len_compressed) {
2021 			err = got_error(GOT_ERR_IO);
2022 			goto done;
2023 		}
2024 	}
2025 done:
2026 	if (err) {
2027 		free(*delta_buf);
2028 		*delta_buf = NULL;
2029 		*delta_len = 0;
2030 		*delta_len_compressed = 0;
2031 		*base_size = 0;
2032 		*result_size = 0;
2033 	}
2034 	return err;
2035 }
2036 
2037 const struct got_error *
2038 got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
2039     size_t *delta_compressed_size, off_t *delta_offset,
2040     off_t *delta_data_offset, off_t *base_offset,
2041     struct got_object_id *base_id, uint64_t *base_size, uint64_t *result_size,
2042     struct got_pack *pack, struct got_packidx *packidx, int idx)
2043 {
2044 	const struct got_error *err = NULL;
2045 	off_t offset;
2046 	uint8_t type;
2047 	uint64_t size;
2048 	size_t tslen, delta_hdrlen;
2049 
2050 	*delta_buf = NULL;
2051 	*delta_size = 0;
2052 	*delta_compressed_size = 0;
2053 	*delta_offset = 0;
2054 	*delta_data_offset = 0;
2055 	*base_offset = 0;
2056 	*base_size = 0;
2057 	*result_size = 0;
2058 
2059 	offset = got_packidx_get_object_offset(packidx, idx);
2060 	if (offset == -1)
2061 		return got_error(GOT_ERR_BAD_PACKIDX);
2062 
2063 	if (offset >= pack->filesize)
2064 		return got_error(GOT_ERR_PACK_OFFSET);
2065 
2066 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
2067 	    pack, offset);
2068 	if (err)
2069 		return err;
2070 
2071 	if (tslen + size < tslen || offset + size < size ||
2072 	    tslen + offset < tslen)
2073 		return got_error(GOT_ERR_PACK_OFFSET);
2074 
2075 	switch (type) {
2076 	case GOT_OBJ_TYPE_OFFSET_DELTA:
2077 		err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
2078 		    pack, offset, tslen);
2079 		if (err)
2080 			return err;
2081 		break;
2082 	case GOT_OBJ_TYPE_REF_DELTA:
2083 		err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
2084 		if (err)
2085 			return err;
2086 		delta_hdrlen = SHA1_DIGEST_LENGTH;
2087 		break;
2088 	default:
2089 		return got_error_fmt(GOT_ERR_OBJ_TYPE,
2090 		    "non-delta object type %d found at offset %lld",
2091 		    type, (long long)offset);
2092 	}
2093 
2094 	if (tslen + delta_hdrlen < delta_hdrlen ||
2095 	    offset + delta_hdrlen < delta_hdrlen)
2096 		return got_error(GOT_ERR_BAD_DELTA);
2097 
2098 	*delta_data_offset = offset + tslen + delta_hdrlen;
2099 	err = read_raw_delta_data(delta_buf, delta_size, delta_compressed_size,
2100 	    base_size, result_size, *delta_data_offset, pack, packidx);
2101 	if (err)
2102 		return err;
2103 
2104 	if (*delta_size != size) {
2105 		err = got_error(GOT_ERR_BAD_DELTA);
2106 		goto done;
2107 	}
2108 
2109 	*delta_offset = offset;
2110 done:
2111 	if (err) {
2112 		free(*delta_buf);
2113 		*delta_buf = NULL;
2114 		*delta_size = 0;
2115 		*delta_compressed_size = 0;
2116 		*delta_offset = 0;
2117 		*base_offset = 0;
2118 		*base_size = 0;
2119 		*result_size = 0;
2120 	}
2121 	return err;
2122 }