Got Mirror

Blob

Date:: Tue Oct 25 07:37:58 2022 UTC
Message:: check size before calling mmap(2) It's only a preparatory step, as checking whether a size_t is less than SIZE_MAX is moot. In a follow-up commit, however, the `filesize' field of the struct got_pack will become off_t and these checks will kick in. This also makes consistent how we guard mmap(2) against empty files. ok and improvements stsp@
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 <endian.h>
34 #include <unistd.h>
35 #include <zlib.h>
36 #include <imsg.h>
37 
38 #include "got_error.h"
39 #include "got_object.h"
40 #include "got_path.h"
41 
42 #include "got_lib_sha1.h"
43 #include "got_lib_delta.h"
44 #include "got_lib_delta_cache.h"
45 #include "got_lib_inflate.h"
46 #include "got_lib_object.h"
47 #include "got_lib_object_parse.h"
48 #include "got_lib_privsep.h"
49 #include "got_lib_pack.h"
50 
51 #ifndef nitems
52 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
53 #endif
54 
55 #ifndef MIN
56 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
57 #endif
58 
59 static const struct got_error *
60 verify_fanout_table(uint32_t *fanout_table)
61 {
62 	int i;
63 
64 	for (i = 0; i < 0xff - 1; i++) {
65 		if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
66 			return got_error(GOT_ERR_BAD_PACKIDX);
67 	}
68 
69 	return NULL;
70 }
71 
72 const struct got_error *
73 got_packidx_init_hdr(struct got_packidx *p, int verify, off_t packfile_size)
74 {
75 	const struct got_error *err = NULL;
76 	struct got_packidx_v2_hdr *h;
77 	SHA1_CTX ctx;
78 	uint8_t sha1[SHA1_DIGEST_LENGTH];
79 	size_t nobj, len_fanout, len_ids, offset, remain;
80 	ssize_t n;
81 	int i;
82 
83 	SHA1Init(&ctx);
84 
85 	h = &p->hdr;
86 	offset = 0;
87 	remain = p->len;
88 
89 	if (remain < sizeof(*h->magic)) {
90 		err = got_error(GOT_ERR_BAD_PACKIDX);
91 		goto done;
92 	}
93 	if (p->map)
94 		h->magic = (uint32_t *)(p->map + offset);
95 	else {
96 		h->magic = malloc(sizeof(*h->magic));
97 		if (h->magic == NULL) {
98 			err = got_error_from_errno("malloc");
99 			goto done;
100 		}
101 		n = read(p->fd, h->magic, sizeof(*h->magic));
102 		if (n < 0) {
103 			err = got_error_from_errno("read");
104 			goto done;
105 		} else if (n != sizeof(*h->magic)) {
106 			err = got_error(GOT_ERR_BAD_PACKIDX);
107 			goto done;
108 		}
109 	}
110 	if (*h->magic != htobe32(GOT_PACKIDX_V2_MAGIC)) {
111 		err = got_error(GOT_ERR_BAD_PACKIDX);
112 		goto done;
113 	}
114 	offset += sizeof(*h->magic);
115 	remain -= sizeof(*h->magic);
116 
117 	if (verify)
118 		SHA1Update(&ctx, (uint8_t *)h->magic, sizeof(*h->magic));
119 
120 	if (remain < sizeof(*h->version)) {
121 		err = got_error(GOT_ERR_BAD_PACKIDX);
122 		goto done;
123 	}
124 	if (p->map)
125 		h->version = (uint32_t *)(p->map + offset);
126 	else {
127 		h->version = malloc(sizeof(*h->version));
128 		if (h->version == NULL) {
129 			err = got_error_from_errno("malloc");
130 			goto done;
131 		}
132 		n = read(p->fd, h->version, sizeof(*h->version));
133 		if (n < 0) {
134 			err = got_error_from_errno("read");
135 			goto done;
136 		} else if (n != sizeof(*h->version)) {
137 			err = got_error(GOT_ERR_BAD_PACKIDX);
138 			goto done;
139 		}
140 	}
141 	if (*h->version != htobe32(GOT_PACKIDX_VERSION)) {
142 		err = got_error(GOT_ERR_BAD_PACKIDX);
143 		goto done;
144 	}
145 	offset += sizeof(*h->version);
146 	remain -= sizeof(*h->version);
147 
148 	if (verify)
149 		SHA1Update(&ctx, (uint8_t *)h->version, sizeof(*h->version));
150 
151 	len_fanout =
152 	    sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
153 	if (remain < len_fanout) {
154 		err = got_error(GOT_ERR_BAD_PACKIDX);
155 		goto done;
156 	}
157 	if (p->map)
158 		h->fanout_table = (uint32_t *)(p->map + offset);
159 	else {
160 		h->fanout_table = malloc(len_fanout);
161 		if (h->fanout_table == NULL) {
162 			err = got_error_from_errno("malloc");
163 			goto done;
164 		}
165 		n = read(p->fd, h->fanout_table, len_fanout);
166 		if (n < 0) {
167 			err = got_error_from_errno("read");
168 			goto done;
169 		} else if (n != len_fanout) {
170 			err = got_error(GOT_ERR_BAD_PACKIDX);
171 			goto done;
172 		}
173 	}
174 	err = verify_fanout_table(h->fanout_table);
175 	if (err)
176 		goto done;
177 	if (verify)
178 		SHA1Update(&ctx, (uint8_t *)h->fanout_table, len_fanout);
179 	offset += len_fanout;
180 	remain -= len_fanout;
181 
182 	nobj = be32toh(h->fanout_table[0xff]);
183 	len_ids = nobj * sizeof(*h->sorted_ids);
184 	if (len_ids <= nobj || len_ids > remain) {
185 		err = got_error(GOT_ERR_BAD_PACKIDX);
186 		goto done;
187 	}
188 	if (p->map)
189 		h->sorted_ids =
190 		    (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
191 	else {
192 		h->sorted_ids = malloc(len_ids);
193 		if (h->sorted_ids == NULL) {
194 			err = got_error(GOT_ERR_BAD_PACKIDX);
195 			goto done;
196 		}
197 		n = read(p->fd, h->sorted_ids, len_ids);
198 		if (n < 0)
199 			err = got_error_from_errno("read");
200 		else if (n != len_ids) {
201 			err = got_error(GOT_ERR_BAD_PACKIDX);
202 			goto done;
203 		}
204 	}
205 	if (verify)
206 		SHA1Update(&ctx, (uint8_t *)h->sorted_ids, len_ids);
207 	offset += len_ids;
208 	remain -= len_ids;
209 
210 	if (remain < nobj * sizeof(*h->crc32)) {
211 		err = got_error(GOT_ERR_BAD_PACKIDX);
212 		goto done;
213 	}
214 	if (p->map)
215 		h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
216 	else {
217 		h->crc32 = malloc(nobj * sizeof(*h->crc32));
218 		if (h->crc32 == NULL) {
219 			err = got_error_from_errno("malloc");
220 			goto done;
221 		}
222 		n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
223 		if (n < 0)
224 			err = got_error_from_errno("read");
225 		else if (n != nobj * sizeof(*h->crc32)) {
226 			err = got_error(GOT_ERR_BAD_PACKIDX);
227 			goto done;
228 		}
229 	}
230 	if (verify)
231 		SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
232 	remain -= nobj * sizeof(*h->crc32);
233 	offset += nobj * sizeof(*h->crc32);
234 
235 	if (remain < nobj * sizeof(*h->offsets)) {
236 		err = got_error(GOT_ERR_BAD_PACKIDX);
237 		goto done;
238 	}
239 	if (p->map)
240 		h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
241 	else {
242 		h->offsets = malloc(nobj * sizeof(*h->offsets));
243 		if (h->offsets == NULL) {
244 			err = got_error_from_errno("malloc");
245 			goto done;
246 		}
247 		n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
248 		if (n < 0)
249 			err = got_error_from_errno("read");
250 		else if (n != nobj * sizeof(*h->offsets)) {
251 			err = got_error(GOT_ERR_BAD_PACKIDX);
252 			goto done;
253 		}
254 	}
255 	if (verify)
256 		SHA1Update(&ctx, (uint8_t *)h->offsets,
257 		    nobj * sizeof(*h->offsets));
258 	remain -= nobj * sizeof(*h->offsets);
259 	offset += nobj * sizeof(*h->offsets);
260 
261 	/* Large file offsets are contained only in files > 2GB. */
262 	if (verify || packfile_size > 0x7fffffff) {
263 		for (i = 0; i < nobj; i++) {
264 			uint32_t o = h->offsets[i];
265 			if (o & htobe32(GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX))
266 				p->nlargeobj++;
267 		}
268 	}
269 	if (p->nlargeobj == 0)
270 		goto checksum;
271 	else if (packfile_size <= 0x7fffffff) {
272 		err = got_error(GOT_ERR_BAD_PACKIDX);
273 		goto done;
274 	}
275 
276 	if (remain < p->nlargeobj * sizeof(*h->large_offsets)) {
277 		err = got_error(GOT_ERR_BAD_PACKIDX);
278 		goto done;
279 	}
280 	if (p->map)
281 		h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
282 	else {
283 		h->large_offsets = malloc(p->nlargeobj *
284 		    sizeof(*h->large_offsets));
285 		if (h->large_offsets == NULL) {
286 			err = got_error_from_errno("malloc");
287 			goto done;
288 		}
289 		n = read(p->fd, h->large_offsets,
290 		    p->nlargeobj * sizeof(*h->large_offsets));
291 		if (n < 0)
292 			err = got_error_from_errno("read");
293 		else if (n != p->nlargeobj * sizeof(*h->large_offsets)) {
294 			err = got_error(GOT_ERR_BAD_PACKIDX);
295 			goto done;
296 		}
297 	}
298 	if (verify)
299 		SHA1Update(&ctx, (uint8_t*)h->large_offsets,
300 		    p->nlargeobj * sizeof(*h->large_offsets));
301 	remain -= p->nlargeobj * sizeof(*h->large_offsets);
302 	offset += p->nlargeobj * sizeof(*h->large_offsets);
303 
304 checksum:
305 	if (remain < sizeof(*h->trailer)) {
306 		err = got_error(GOT_ERR_BAD_PACKIDX);
307 		goto done;
308 	}
309 	if (p->map)
310 		h->trailer =
311 		    (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
312 	else {
313 		h->trailer = malloc(sizeof(*h->trailer));
314 		if (h->trailer == NULL) {
315 			err = got_error_from_errno("malloc");
316 			goto done;
317 		}
318 		n = read(p->fd, h->trailer, sizeof(*h->trailer));
319 		if (n < 0)
320 			err = got_error_from_errno("read");
321 		else if (n != sizeof(*h->trailer)) {
322 			err = got_error(GOT_ERR_BAD_PACKIDX);
323 			goto done;
324 		}
325 	}
326 	if (verify) {
327 		SHA1Update(&ctx, h->trailer->packfile_sha1, SHA1_DIGEST_LENGTH);
328 		SHA1Final(sha1, &ctx);
329 		if (memcmp(h->trailer->packidx_sha1, sha1,
330 		    SHA1_DIGEST_LENGTH) != 0)
331 			err = got_error(GOT_ERR_PACKIDX_CSUM);
332 	}
333 done:
334 	return err;
335 }
336 
337 const struct got_error *
338 got_packidx_open(struct got_packidx **packidx,
339     int dir_fd, const char *relpath, int verify)
340 {
341 	const struct got_error *err = NULL;
342 	struct got_packidx *p = NULL;
343 	char *pack_relpath;
344 	struct stat idx_sb, pack_sb;
345 
346 	*packidx = NULL;
347 
348 	err = got_packidx_get_packfile_path(&pack_relpath, relpath);
349 	if (err)
350 		return err;
351 
352 	/*
353 	 * Ensure that a corresponding pack file exists.
354 	 * Some Git repositories have this problem. Git seems to ignore
355 	 * the existence of lonely pack index files but we do not.
356 	 */
357 	if (fstatat(dir_fd, pack_relpath, &pack_sb, 0) == -1) {
358 		if (errno == ENOENT) {
359 			err = got_error_fmt(GOT_ERR_LONELY_PACKIDX,
360 			    "%s", relpath);
361 		} else
362 			err = got_error_from_errno2("fstatat", pack_relpath);
363 		goto done;
364 	}
365 
366 	p = calloc(1, sizeof(*p));
367 	if (p == NULL) {
368 		err = got_error_from_errno("calloc");
369 		goto done;
370 	}
371 
372 	p->fd = openat(dir_fd, relpath, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
373 	if (p->fd == -1) {
374 		err = got_error_from_errno2("openat", relpath);
375 		goto done;
376 	}
377 
378 	if (fstat(p->fd, &idx_sb) != 0) {
379 		err = got_error_from_errno2("fstat", relpath);
380 		goto done;
381 	}
382 	p->len = idx_sb.st_size;
383 	if (p->len < sizeof(p->hdr)) {
384 		err = got_error(GOT_ERR_BAD_PACKIDX);
385 		goto done;
386 	}
387 
388 	p->path_packidx = strdup(relpath);
389 	if (p->path_packidx == NULL) {
390 		err = got_error_from_errno("strdup");
391 		goto done;
392 	}
393 
394 #ifndef GOT_PACK_NO_MMAP
395 	if (p->len > 0 && p->len <= SIZE_MAX) {
396 		p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
397 		if (p->map == MAP_FAILED) {
398 			if (errno != ENOMEM) {
399 				err = got_error_from_errno("mmap");
400 				goto done;
401 			}
402 			p->map = NULL; /* fall back to read(2) */
403 		}
404 	}
405 #endif
406 
407 	err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
408 done:
409 	if (err) {
410 		if (p)
411 			got_packidx_close(p);
412 	} else
413 		*packidx = p;
414 	free(pack_relpath);
415 	return err;
416 }
417 
418 const struct got_error *
419 got_packidx_close(struct got_packidx *packidx)
420 {
421 	const struct got_error *err = NULL;
422 
423 	free(packidx->path_packidx);
424 	if (packidx->map) {
425 		if (munmap(packidx->map, packidx->len) == -1)
426 			err = got_error_from_errno("munmap");
427 	} else {
428 		free(packidx->hdr.magic);
429 		free(packidx->hdr.version);
430 		free(packidx->hdr.fanout_table);
431 		free(packidx->hdr.sorted_ids);
432 		free(packidx->hdr.crc32);
433 		free(packidx->hdr.offsets);
434 		free(packidx->hdr.large_offsets);
435 		free(packidx->hdr.trailer);
436 	}
437 	if (close(packidx->fd) == -1 && err == NULL)
438 		err = got_error_from_errno("close");
439 	free(packidx->sorted_offsets);
440 	free(packidx->sorted_large_offsets);
441 	free(packidx);
442 
443 	return err;
444 }
445 
446 const struct got_error *
447 got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
448 {
449 	size_t size;
450 
451 	/* Packfile path contains ".pack" instead of ".idx", so add one byte. */
452 	size = strlen(path_packidx) + 2;
453 	if (size < GOT_PACKFILE_NAMELEN + 1)
454 		return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
455 
456 	*path_packfile = malloc(size);
457 	if (*path_packfile == NULL)
458 		return got_error_from_errno("malloc");
459 
460 	/* Copy up to and excluding ".idx". */
461 	if (strlcpy(*path_packfile, path_packidx,
462 	    size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
463 		return got_error(GOT_ERR_NO_SPACE);
464 
465 	if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
466 		return got_error(GOT_ERR_NO_SPACE);
467 
468 	return NULL;
469 }
470 
471 off_t
472 got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
473 {
474 	uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
475 	if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
476 		uint64_t loffset;
477 		idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
478 		if (idx < 0 || idx >= packidx->nlargeobj ||
479 		    packidx->hdr.large_offsets == NULL)
480 			return -1;
481 		loffset = be64toh(packidx->hdr.large_offsets[idx]);
482 		return (loffset > INT64_MAX ? -1 : (off_t)loffset);
483 	}
484 	return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
485 }
486 
487 int
488 got_packidx_get_object_idx(struct got_packidx *packidx,
489     struct got_object_id *id)
490 {
491 	u_int8_t id0 = id->sha1[0];
492 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
493 	int left = 0, right = totobj - 1;
494 
495 	if (id0 > 0)
496 		left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
497 
498 	while (left <= right) {
499 		struct got_packidx_object_id *oid;
500 		int i, cmp;
501 
502 		i = ((left + right) / 2);
503 		oid = &packidx->hdr.sorted_ids[i];
504 		cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
505 		if (cmp == 0)
506 			return i;
507 		else if (cmp > 0)
508 			left = i + 1;
509 		else if (cmp < 0)
510 			right = i - 1;
511 	}
512 
513 	return -1;
514 }
515 
516 static int
517 offset_cmp(const void *pa, const void *pb)
518 {
519 	const struct got_pack_offset_index *a, *b;
520 
521 	a = (const struct got_pack_offset_index *)pa;
522 	b = (const struct got_pack_offset_index *)pb;
523 
524 	if (a->offset < b->offset)
525 		return -1;
526 	else if (a->offset > b->offset)
527 		return 1;
528 
529 	return 0;
530 }
531 
532 static int
533 large_offset_cmp(const void *pa, const void *pb)
534 {
535 	const struct got_pack_large_offset_index *a, *b;
536 
537 	a = (const struct got_pack_large_offset_index *)pa;
538 	b = (const struct got_pack_large_offset_index *)pb;
539 
540 	if (a->offset < b->offset)
541 		return -1;
542 	else if (a->offset > b->offset)
543 		return 1;
544 
545 	return 0;
546 }
547 
548 static const struct got_error *
549 build_offset_index(struct got_packidx *p)
550 {
551 	uint32_t nobj = be32toh(p->hdr.fanout_table[0xff]);
552 	unsigned int i, j, k;
553 
554 	p->sorted_offsets = calloc(nobj - p->nlargeobj,
555 	    sizeof(p->sorted_offsets[0]));
556 	if (p->sorted_offsets == NULL)
557 		return got_error_from_errno("calloc");
558 
559 	if (p->nlargeobj > 0) {
560 		p->sorted_large_offsets = calloc(p->nlargeobj,
561 		    sizeof(p->sorted_large_offsets[0]));
562 		if (p->sorted_large_offsets == NULL)
563 			return got_error_from_errno("calloc");
564 	}
565 
566 	j = 0;
567 	k = 0;
568 	for (i = 0; i < nobj; i++) {
569 		uint32_t offset = be32toh(p->hdr.offsets[i]);
570 		if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
571 			uint64_t loffset;
572 			uint32_t idx;
573 			idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
574 			if (idx >= p->nlargeobj ||
575 			    p->nlargeobj == 0 ||
576 			    p->hdr.large_offsets == NULL)
577 				return got_error(GOT_ERR_BAD_PACKIDX);
578 			loffset = be64toh(p->hdr.large_offsets[idx]);
579 			p->sorted_large_offsets[j].offset = loffset;
580 			p->sorted_large_offsets[j].idx = i;
581 			j++;
582 		} else {
583 			p->sorted_offsets[k].offset = offset;
584 			p->sorted_offsets[k].idx = i;
585 			k++;
586 		}
587 	}
588 	if (j != p->nlargeobj || k != nobj - p->nlargeobj)
589 		return got_error(GOT_ERR_BAD_PACKIDX);
590 
591 	qsort(p->sorted_offsets, nobj - p->nlargeobj,
592 	    sizeof(p->sorted_offsets[0]), offset_cmp);
593 
594 	if (p->sorted_large_offsets)
595 		qsort(p->sorted_large_offsets, p->nlargeobj,
596 		    sizeof(p->sorted_large_offsets[0]), large_offset_cmp);
597 
598 	return NULL;
599 }
600 
601 const struct got_error *
602 got_packidx_get_offset_idx(int *idx, struct got_packidx *packidx, off_t offset)
603 {
604 	const struct got_error *err;
605 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
606 	int i, left, right;
607 
608 	*idx = -1;
609 
610 	if (packidx->sorted_offsets == NULL) {
611 		err = build_offset_index(packidx);
612 		if (err)
613 			return err;
614 	}
615 
616 	if (offset >= 0x7fffffff) {
617 		uint64_t lo;
618 		left = 0, right = packidx->nlargeobj - 1;
619 		while (left <= right) {
620 			i = ((left + right) / 2);
621 			lo = packidx->sorted_large_offsets[i].offset;
622 			if (lo == offset) {
623 				*idx = packidx->sorted_large_offsets[i].idx;
624 				break;
625 			} else if (offset > lo)
626 				left = i + 1;
627 			else if (offset < lo)
628 				right = i - 1;
629 		}
630 	} else {
631 		uint32_t o;
632 		left = 0, right = totobj - packidx->nlargeobj - 1;
633 		while (left <= right) {
634 			i = ((left + right) / 2);
635 			o = packidx->sorted_offsets[i].offset;
636 			if (o == offset) {
637 				*idx = packidx->sorted_offsets[i].idx;
638 				break;
639 			} else if (offset > o)
640 				left = i + 1;
641 			else if (offset < o)
642 				right = i - 1;
643 		}
644 	}
645 
646 	return NULL;
647 }
648 
649 const struct got_error *
650 got_packidx_get_object_id(struct got_object_id *id,
651     struct got_packidx *packidx, int idx)
652 {
653 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
654 	struct got_packidx_object_id *oid;
655 
656 	if (idx < 0 || idx >= totobj)
657 		return got_error(GOT_ERR_NO_OBJ);
658 
659 	oid = &packidx->hdr.sorted_ids[idx];
660 	memcpy(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
661 	return NULL;
662 }
663 
664 const struct got_error *
665 got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
666     struct got_packidx *packidx, const char *id_str_prefix)
667 {
668 	const struct got_error *err = NULL;
669 	u_int8_t id0;
670 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
671 	char hex[3];
672 	size_t prefix_len = strlen(id_str_prefix);
673 	struct got_packidx_object_id *oid;
674 	uint32_t i = 0;
675 
676 	STAILQ_INIT(matched_ids);
677 
678 	if (prefix_len < 2)
679 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
680 
681 	hex[0] = id_str_prefix[0];
682 	hex[1] = id_str_prefix[1];
683 	hex[2] = '\0';
684 	if (!got_parse_xdigit(&id0, hex))
685 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
686 
687 	if (id0 > 0)
688 		i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
689 	oid = &packidx->hdr.sorted_ids[i];
690 	while (i < totobj && oid->sha1[0] == id0) {
691 		char id_str[SHA1_DIGEST_STRING_LENGTH];
692 		struct got_object_qid *qid;
693 		int cmp;
694 
695 		if (!got_sha1_digest_to_str(oid->sha1, id_str, sizeof(id_str)))
696 		        return got_error(GOT_ERR_NO_SPACE);
697 
698 		cmp = strncmp(id_str, id_str_prefix, prefix_len);
699 		if (cmp < 0) {
700 			oid = &packidx->hdr.sorted_ids[++i];
701 			continue;
702 		} else if (cmp > 0)
703 			break;
704 
705 		err = got_object_qid_alloc_partial(&qid);
706 		if (err)
707 			break;
708 		memcpy(qid->id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
709 		STAILQ_INSERT_TAIL(matched_ids, qid, entry);
710 
711 		oid = &packidx->hdr.sorted_ids[++i];
712 	}
713 
714 	if (err)
715 		got_object_id_queue_free(matched_ids);
716 	return err;
717 }
718 
719 static void
720 set_max_datasize(void)
721 {
722 	struct rlimit rl;
723 
724 	if (getrlimit(RLIMIT_DATA, &rl) != 0)
725 		return;
726 
727 	rl.rlim_cur = rl.rlim_max;
728 	setrlimit(RLIMIT_DATA, &rl);
729 }
730 
731 const struct got_error *
732 got_pack_start_privsep_child(struct got_pack *pack, struct got_packidx *packidx)
733 {
734 	const struct got_error *err = NULL;
735 	int imsg_fds[2];
736 	pid_t pid;
737 	struct imsgbuf *ibuf;
738 
739 	ibuf = calloc(1, sizeof(*ibuf));
740 	if (ibuf == NULL)
741 		return got_error_from_errno("calloc");
742 
743 	pack->privsep_child = calloc(1, sizeof(*pack->privsep_child));
744 	if (pack->privsep_child == NULL) {
745 		err = got_error_from_errno("calloc");
746 		free(ibuf);
747 		return err;
748 	}
749 	pack->child_has_tempfiles = 0;
750 	pack->child_has_delta_outfd = 0;
751 
752 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
753 		err = got_error_from_errno("socketpair");
754 		goto done;
755 	}
756 
757 	pid = fork();
758 	if (pid == -1) {
759 		err = got_error_from_errno("fork");
760 		goto done;
761 	} else if (pid == 0) {
762 		set_max_datasize();
763 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_PACK,
764 		    pack->path_packfile);
765 		/* not reached */
766 	}
767 
768 	if (close(imsg_fds[1]) == -1)
769 		return got_error_from_errno("close");
770 	pack->privsep_child->imsg_fd = imsg_fds[0];
771 	pack->privsep_child->pid = pid;
772 	imsg_init(ibuf, imsg_fds[0]);
773 	pack->privsep_child->ibuf = ibuf;
774 
775 	err = got_privsep_init_pack_child(ibuf, pack, packidx);
776 	if (err) {
777 		const struct got_error *child_err;
778 		err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
779 		child_err = got_privsep_wait_for_child(
780 		    pack->privsep_child->pid);
781 		if (child_err && err == NULL)
782 			err = child_err;
783 	}
784 done:
785 	if (err) {
786 		free(ibuf);
787 		free(pack->privsep_child);
788 		pack->privsep_child = NULL;
789 	}
790 	return err;
791 }
792 
793 static const struct got_error *
794 pack_stop_privsep_child(struct got_pack *pack)
795 {
796 	const struct got_error *err = NULL, *close_err = NULL;
797 
798 	if (pack->privsep_child == NULL)
799 		return NULL;
800 
801 	err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
802 	if (err)
803 		return err;
804 	if (close(pack->privsep_child->imsg_fd) == -1)
805 		close_err = got_error_from_errno("close");
806 	err = got_privsep_wait_for_child(pack->privsep_child->pid);
807 	if (close_err && err == NULL)
808 		err = close_err;
809 	imsg_clear(pack->privsep_child->ibuf);
810 	free(pack->privsep_child->ibuf);
811 	free(pack->privsep_child);
812 	pack->privsep_child = NULL;
813 	return err;
814 }
815 
816 const struct got_error *
817 got_pack_close(struct got_pack *pack)
818 {
819 	const struct got_error *err = NULL;
820 
821 	err = pack_stop_privsep_child(pack);
822 	if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
823 		err = got_error_from_errno("munmap");
824 	if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
825 		err = got_error_from_errno("close");
826 	pack->fd = -1;
827 	free(pack->path_packfile);
828 	pack->path_packfile = NULL;
829 	pack->filesize = 0;
830 	if (pack->delta_cache) {
831 		got_delta_cache_free(pack->delta_cache);
832 		pack->delta_cache = NULL;
833 	}
834 
835 	/*
836 	 * Leave accumfd and basefd alone. They are managed by the
837 	 * repository layer and can be reused.
838 	 */
839 
840 	return err;
841 }
842 
843 const struct got_error *
844 got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
845     struct got_pack *pack, off_t offset)
846 {
847 	uint8_t t = 0;
848 	uint64_t s = 0;
849 	uint8_t sizeN;
850 	size_t mapoff = 0;
851 	int i = 0;
852 
853 	*len = 0;
854 
855 	if (offset >= pack->filesize)
856 		return got_error(GOT_ERR_PACK_OFFSET);
857 
858 	if (pack->map) {
859 		mapoff = (size_t)offset;
860 	} else {
861 		if (lseek(pack->fd, offset, SEEK_SET) == -1)
862 			return got_error_from_errno("lseek");
863 	}
864 
865 	do {
866 		/* We do not support size values which don't fit in 64 bit. */
867 		if (i > 9)
868 			return got_error_fmt(GOT_ERR_OBJ_TOO_LARGE,
869 			    "packfile offset %lld", (long long)offset);
870 
871 		if (pack->map) {
872 			if (mapoff + sizeof(sizeN) >= pack->filesize)
873 				return got_error(GOT_ERR_BAD_PACKFILE);
874 			sizeN = *(pack->map + mapoff);
875 			mapoff += sizeof(sizeN);
876 		} else {
877 			ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
878 			if (n < 0)
879 				return got_error_from_errno("read");
880 			if (n != sizeof(sizeN))
881 				return got_error(GOT_ERR_BAD_PACKFILE);
882 		}
883 		*len += sizeof(sizeN);
884 
885 		if (i == 0) {
886 			t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
887 			    GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
888 			s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
889 		} else {
890 			size_t shift = 4 + 7 * (i - 1);
891 			s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
892 		}
893 		i++;
894 	} while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
895 
896 	*type = t;
897 	*size = s;
898 	return NULL;
899 }
900 
901 static const struct got_error *
902 open_plain_object(struct got_object **obj, struct got_object_id *id,
903     uint8_t type, off_t offset, size_t size, int idx)
904 {
905 	*obj = calloc(1, sizeof(**obj));
906 	if (*obj == NULL)
907 		return got_error_from_errno("calloc");
908 
909 	(*obj)->type = type;
910 	(*obj)->flags = GOT_OBJ_FLAG_PACKED;
911 	(*obj)->pack_idx = idx;
912 	(*obj)->hdrlen = 0;
913 	(*obj)->size = size;
914 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
915 	(*obj)->pack_offset = offset;
916 
917 	return NULL;
918 }
919 
920 static const struct got_error *
921 parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
922     off_t delta_offset)
923 {
924 	int64_t o = 0;
925 	uint8_t offN;
926 	int i = 0;
927 
928 	*offset = 0;
929 	*len = 0;
930 
931 	do {
932 		/* We do not support offset values which don't fit in 64 bit. */
933 		if (i > 8)
934 			return got_error(GOT_ERR_NO_SPACE);
935 
936 		if (pack->map) {
937 			size_t mapoff;
938 			mapoff = (size_t)delta_offset + *len;
939 			if (mapoff + sizeof(offN) >= pack->filesize)
940 				return got_error(GOT_ERR_PACK_OFFSET);
941 			offN = *(pack->map + mapoff);
942 		} else {
943 			ssize_t n;
944 			n = read(pack->fd, &offN, sizeof(offN));
945 			if (n < 0)
946 				return got_error_from_errno("read");
947 			if (n != sizeof(offN))
948 				return got_error(GOT_ERR_BAD_PACKFILE);
949 		}
950 		*len += sizeof(offN);
951 
952 		if (i == 0)
953 			o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
954 		else {
955 			o++;
956 			o <<= 7;
957 			o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
958 		}
959 		i++;
960 	} while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
961 
962 	*offset = o;
963 	return NULL;
964 }
965 
966 const struct got_error *
967 got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
968     struct got_pack *pack, off_t offset, size_t tslen)
969 {
970 	const struct got_error *err;
971 	int64_t negoffset;
972 	size_t negofflen;
973 
974 	*len = 0;
975 
976 	err = parse_negative_offset(&negoffset, &negofflen, pack,
977 	    offset + tslen);
978 	if (err)
979 		return err;
980 
981 	/* Compute the base object's offset (must be in the same pack file). */
982 	*base_offset = (offset - negoffset);
983 	if (*base_offset <= 0)
984 		return got_error(GOT_ERR_BAD_PACKFILE);
985 
986 	*len = negofflen;
987 	return NULL;
988 }
989 
990 static const struct got_error *
991 read_delta_data(uint8_t **delta_buf, size_t *delta_len,
992     size_t *delta_compressed_len, size_t delta_data_offset,
993     struct got_pack *pack)
994 {
995 	const struct got_error *err = NULL;
996 	size_t consumed = 0;
997 
998 	if (pack->map) {
999 		if (delta_data_offset >= pack->filesize)
1000 			return got_error(GOT_ERR_PACK_OFFSET);
1001 		err = got_inflate_to_mem_mmap(delta_buf, delta_len,
1002 		    &consumed, NULL, pack->map, delta_data_offset,
1003 		    pack->filesize - delta_data_offset);
1004 		if (err)
1005 			return err;
1006 	} else {
1007 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
1008 			return got_error_from_errno("lseek");
1009 		err = got_inflate_to_mem_fd(delta_buf, delta_len,
1010 		    &consumed, NULL, 0, pack->fd);
1011 		if (err)
1012 			return err;
1013 	}
1014 
1015 	if (delta_compressed_len)
1016 		*delta_compressed_len = consumed;
1017 
1018 	return NULL;
1019 }
1020 
1021 static const struct got_error *
1022 add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
1023     int delta_type, size_t delta_size, size_t delta_data_offset)
1024 {
1025 	struct got_delta *delta;
1026 
1027 	delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
1028 	    delta_data_offset);
1029 	if (delta == NULL)
1030 		return got_error_from_errno("got_delta_open");
1031 	/* delta is freed in got_object_close() */
1032 	deltas->nentries++;
1033 	STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
1034 	return NULL;
1035 }
1036 
1037 static const struct got_error *
1038 resolve_offset_delta(struct got_delta_chain *deltas,
1039     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1040     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1041 {
1042 	const struct got_error *err;
1043 	off_t base_offset;
1044 	uint8_t base_type;
1045 	uint64_t base_size;
1046 	size_t base_tslen;
1047 	off_t delta_data_offset;
1048 	size_t consumed;
1049 
1050 	err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
1051 	    delta_offset, tslen);
1052 	if (err)
1053 		return err;
1054 
1055 	delta_data_offset = delta_offset + tslen + consumed;
1056 	if (delta_data_offset >= pack->filesize)
1057 		return got_error(GOT_ERR_PACK_OFFSET);
1058 
1059 	if (pack->map == NULL) {
1060 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1061 		if (delta_data_offset == -1)
1062 			return got_error_from_errno("lseek");
1063 	}
1064 
1065 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1066 	    delta_data_offset);
1067 	if (err)
1068 		return err;
1069 
1070 	/* An offset delta must be in the same packfile. */
1071 	if (base_offset >= pack->filesize)
1072 		return got_error(GOT_ERR_PACK_OFFSET);
1073 
1074 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1075 	    &base_tslen, pack, base_offset);
1076 	if (err)
1077 		return err;
1078 
1079 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1080 	    base_tslen, base_type, base_size, recursion - 1);
1081 }
1082 
1083 const struct got_error *
1084 got_pack_parse_ref_delta(struct got_object_id *id,
1085     struct got_pack *pack, off_t delta_offset, int tslen)
1086 {
1087 	if (pack->map) {
1088 		size_t mapoff = delta_offset + tslen;
1089 		if (mapoff + sizeof(*id) >= pack->filesize)
1090 			return got_error(GOT_ERR_PACK_OFFSET);
1091 		memcpy(id, pack->map + mapoff, sizeof(*id));
1092 	} else {
1093 		ssize_t n;
1094 		n = read(pack->fd, id, sizeof(*id));
1095 		if (n < 0)
1096 			return got_error_from_errno("read");
1097 		if (n != sizeof(*id))
1098 			return got_error(GOT_ERR_BAD_PACKFILE);
1099 	}
1100 
1101 	return NULL;
1102 }
1103 
1104 static const struct got_error *
1105 resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1106     struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1107     size_t delta_size, unsigned int recursion)
1108 {
1109 	const struct got_error *err;
1110 	struct got_object_id id;
1111 	int idx;
1112 	off_t base_offset;
1113 	uint8_t base_type;
1114 	uint64_t base_size;
1115 	size_t base_tslen;
1116 	off_t delta_data_offset;
1117 
1118 	if (delta_offset + tslen >= pack->filesize)
1119 		return got_error(GOT_ERR_PACK_OFFSET);
1120 
1121 	err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1122 	if (err)
1123 		return err;
1124 	if (pack->map) {
1125 		delta_data_offset = delta_offset + tslen + sizeof(id);
1126 	} else {
1127 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1128 		if (delta_data_offset == -1)
1129 			return got_error_from_errno("lseek");
1130 	}
1131 
1132 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1133 	    delta_data_offset);
1134 	if (err)
1135 		return err;
1136 
1137 	/* Delta base must be in the same pack file. */
1138 	idx = got_packidx_get_object_idx(packidx, &id);
1139 	if (idx == -1)
1140 		return got_error(GOT_ERR_NO_OBJ);
1141 
1142 	base_offset = got_packidx_get_object_offset(packidx, idx);
1143 	if (base_offset == -1)
1144 		return got_error(GOT_ERR_BAD_PACKIDX);
1145 
1146 	if (base_offset >= pack->filesize)
1147 		return got_error(GOT_ERR_PACK_OFFSET);
1148 
1149 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1150 	    &base_tslen, pack, base_offset);
1151 	if (err)
1152 		return err;
1153 
1154 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1155 	    base_tslen, base_type, base_size, recursion - 1);
1156 }
1157 
1158 const struct got_error *
1159 got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1160     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1161     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1162 {
1163 	const struct got_error *err = NULL;
1164 
1165 	if (--recursion == 0)
1166 		return got_error(GOT_ERR_RECURSION);
1167 
1168 	switch (delta_type) {
1169 	case GOT_OBJ_TYPE_COMMIT:
1170 	case GOT_OBJ_TYPE_TREE:
1171 	case GOT_OBJ_TYPE_BLOB:
1172 	case GOT_OBJ_TYPE_TAG:
1173 		/* Plain types are the final delta base. Recursion ends. */
1174 		err = add_delta(deltas, delta_offset, tslen, delta_type,
1175 		    delta_size, 0);
1176 		break;
1177 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1178 		err = resolve_offset_delta(deltas, packidx, pack,
1179 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1180 		break;
1181 	case GOT_OBJ_TYPE_REF_DELTA:
1182 		err = resolve_ref_delta(deltas, packidx, pack,
1183 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1184 		break;
1185 	default:
1186 		return got_error(GOT_ERR_OBJ_TYPE);
1187 	}
1188 
1189 	return err;
1190 }
1191 
1192 static const struct got_error *
1193 open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1194     struct got_pack *pack, struct got_object_id *id, off_t offset,
1195     size_t tslen, int delta_type, size_t delta_size, int idx)
1196 {
1197 	const struct got_error *err = NULL;
1198 	int resolved_type;
1199 
1200 	*obj = calloc(1, sizeof(**obj));
1201 	if (*obj == NULL)
1202 		return got_error_from_errno("calloc");
1203 
1204 	(*obj)->flags = 0;
1205 	(*obj)->hdrlen = 0;
1206 	(*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1207 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1208 	(*obj)->pack_offset = offset + tslen;
1209 
1210 	STAILQ_INIT(&(*obj)->deltas.entries);
1211 	(*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1212 	(*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1213 	(*obj)->pack_idx = idx;
1214 
1215 	err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1216 	    offset, tslen, delta_type, delta_size,
1217 	    GOT_DELTA_CHAIN_RECURSION_MAX);
1218 	if (err)
1219 		goto done;
1220 
1221 	err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1222 	if (err)
1223 		goto done;
1224 	(*obj)->type = resolved_type;
1225 done:
1226 	if (err) {
1227 		got_object_close(*obj);
1228 		*obj = NULL;
1229 	}
1230 	return err;
1231 }
1232 
1233 const struct got_error *
1234 got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1235     struct got_packidx *packidx, int idx, struct got_object_id *id)
1236 {
1237 	const struct got_error *err = NULL;
1238 	off_t offset;
1239 	uint8_t type;
1240 	uint64_t size;
1241 	size_t tslen;
1242 
1243 	*obj = NULL;
1244 
1245 	offset = got_packidx_get_object_offset(packidx, idx);
1246 	if (offset == -1)
1247 		return got_error(GOT_ERR_BAD_PACKIDX);
1248 
1249 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1250 	    pack, offset);
1251 	if (err)
1252 		return err;
1253 
1254 	switch (type) {
1255 	case GOT_OBJ_TYPE_COMMIT:
1256 	case GOT_OBJ_TYPE_TREE:
1257 	case GOT_OBJ_TYPE_BLOB:
1258 	case GOT_OBJ_TYPE_TAG:
1259 		err = open_plain_object(obj, id, type, offset + tslen,
1260 		    size, idx);
1261 		break;
1262 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1263 	case GOT_OBJ_TYPE_REF_DELTA:
1264 		err = open_delta_object(obj, packidx, pack, id, offset,
1265 		    tslen, type, size, idx);
1266 		break;
1267 	default:
1268 		err = got_error(GOT_ERR_OBJ_TYPE);
1269 		break;
1270 	}
1271 
1272 	return err;
1273 }
1274 
1275 const struct got_error *
1276 got_pack_get_delta_chain_max_size(uint64_t *max_size,
1277     struct got_delta_chain *deltas, struct got_pack *pack)
1278 {
1279 	struct got_delta *delta;
1280 	uint64_t base_size = 0, result_size = 0;
1281 
1282 	*max_size = 0;
1283 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1284 		/* Plain object types are the delta base. */
1285 		if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1286 		    delta->type != GOT_OBJ_TYPE_TREE &&
1287 		    delta->type != GOT_OBJ_TYPE_BLOB &&
1288 		    delta->type != GOT_OBJ_TYPE_TAG) {
1289 			const struct got_error *err;
1290 			uint8_t *delta_buf = NULL;
1291 			size_t delta_len;
1292 			int cached = 1;
1293 
1294 			if (pack->delta_cache) {
1295 				got_delta_cache_get(&delta_buf, &delta_len,
1296 				    pack->delta_cache, delta->data_offset);
1297 			}
1298 			if (delta_buf == NULL) {
1299 				cached = 0;
1300 				err = read_delta_data(&delta_buf, &delta_len,
1301 				    NULL, delta->data_offset, pack);
1302 				if (err)
1303 					return err;
1304 			}
1305 			if (pack->delta_cache && !cached) {
1306 				err = got_delta_cache_add(pack->delta_cache,
1307 				    delta->data_offset, delta_buf, delta_len);
1308 				if (err == NULL)
1309 					cached = 1;
1310 				else if (err->code != GOT_ERR_NO_SPACE) {
1311 					free(delta_buf);
1312 					return err;
1313 				}
1314 			}
1315 			err = got_delta_get_sizes(&base_size, &result_size,
1316 			    delta_buf, delta_len);
1317 			if (!cached)
1318 				free(delta_buf);
1319 			if (err)
1320 				return err;
1321 		} else
1322 			base_size = delta->size;
1323 		if (base_size > *max_size)
1324 			*max_size = base_size;
1325 		if (result_size > *max_size)
1326 			*max_size = result_size;
1327 	}
1328 
1329 	return NULL;
1330 }
1331 
1332 const struct got_error *
1333 got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1334     struct got_pack *pack)
1335 {
1336 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1337 		return got_error(GOT_ERR_OBJ_TYPE);
1338 
1339 	return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1340 }
1341 
1342 const struct got_error *
1343 got_pack_dump_delta_chain_to_file(size_t *result_size,
1344     struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1345     FILE *base_file, FILE *accum_file)
1346 {
1347 	const struct got_error *err = NULL;
1348 	struct got_delta *delta;
1349 	uint8_t *base_buf = NULL, *accum_buf = NULL;
1350 	size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1351 	/* We process small enough files entirely in memory for speed. */
1352 	const size_t max_bufsize = GOT_DELTA_RESULT_SIZE_CACHED_MAX;
1353 	uint64_t max_size = 0;
1354 	int n = 0;
1355 
1356 	*result_size = 0;
1357 
1358 	if (STAILQ_EMPTY(&deltas->entries))
1359 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1360 
1361 	if (fseeko(base_file, 0L, SEEK_SET) == -1)
1362 		return got_error_from_errno("fseeko");
1363 	if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1364 		return got_error_from_errno("fseeko");
1365 
1366 	/* Deltas are ordered in ascending order. */
1367 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1368 		uint8_t *delta_buf = NULL;
1369 		uint64_t base_size, result_size = 0;
1370 		int cached = 1;
1371 		if (n == 0) {
1372 			size_t mapoff;
1373 			off_t delta_data_offset;
1374 
1375 			/* Plain object types are the delta base. */
1376 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1377 			    delta->type != GOT_OBJ_TYPE_TREE &&
1378 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1379 			    delta->type != GOT_OBJ_TYPE_TAG) {
1380 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1381 				goto done;
1382 			}
1383 
1384 			delta_data_offset = delta->offset + delta->tslen;
1385 			if (delta_data_offset >= pack->filesize) {
1386 				err = got_error(GOT_ERR_PACK_OFFSET);
1387 				goto done;
1388 			}
1389 			if (pack->map == NULL) {
1390 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1391 				    == -1) {
1392 					err = got_error_from_errno("lseek");
1393 					goto done;
1394 				}
1395 			}
1396 			if (delta->size > max_size)
1397 				max_size = delta->size;
1398 			if (max_size > max_bufsize) {
1399 				if (pack->map) {
1400 					mapoff = (size_t)delta_data_offset;
1401 					err = got_inflate_to_file_mmap(
1402 					    &base_bufsz, NULL, NULL, pack->map,
1403 					    mapoff, pack->filesize - mapoff,
1404 					    base_file);
1405 				} else
1406 					err = got_inflate_to_file_fd(
1407 					    &base_bufsz, NULL, NULL, pack->fd,
1408 					    base_file);
1409 			} else {
1410 				accum_buf = malloc(max_size);
1411 				if (accum_buf == NULL) {
1412 					err = got_error_from_errno("malloc");
1413 					goto done;
1414 				}
1415 				accum_bufsz = max_size;
1416 				if (pack->map) {
1417 					mapoff = (size_t)delta_data_offset;
1418 					err = got_inflate_to_mem_mmap(&base_buf,
1419 					    &base_bufsz, NULL, NULL,
1420 					    pack->map, mapoff,
1421 					    pack->filesize - mapoff);
1422 				} else
1423 					err = got_inflate_to_mem_fd(&base_buf,
1424 					    &base_bufsz, NULL, NULL, max_size,
1425 					    pack->fd);
1426 			}
1427 			if (err)
1428 				goto done;
1429 			n++;
1430 			if (base_buf == NULL)
1431 				rewind(base_file);
1432 			continue;
1433 		}
1434 
1435 		if (pack->delta_cache) {
1436 			got_delta_cache_get(&delta_buf, &delta_len,
1437 			    pack->delta_cache, delta->data_offset);
1438 		}
1439 		if (delta_buf == NULL) {
1440 			cached = 0;
1441 			err = read_delta_data(&delta_buf, &delta_len, NULL,
1442 			    delta->data_offset, pack);
1443 			if (err)
1444 				goto done;
1445 		}
1446 		if (pack->delta_cache && !cached) {
1447 			err = got_delta_cache_add(pack->delta_cache,
1448 			    delta->data_offset, delta_buf, delta_len);
1449 			if (err == NULL)
1450 				cached = 1;
1451 			else if (err->code != GOT_ERR_NO_SPACE) {
1452 				free(delta_buf);
1453 				goto done;
1454 			}
1455 		}
1456 
1457 		err = got_delta_get_sizes(&base_size, &result_size,
1458 		    delta_buf, delta_len);
1459 		if (err)
1460 			goto done;
1461 		if (base_size > max_size)
1462 			max_size = base_size;
1463 		if (result_size > max_size)
1464 			max_size = result_size;
1465 
1466 		if (base_buf && max_size > max_bufsize) {
1467 			/* Switch from buffers to temporary files. */
1468 			size_t w = fwrite(base_buf, 1, base_bufsz,
1469 			    base_file);
1470 			if (w != base_bufsz) {
1471 				err = got_ferror(outfile, GOT_ERR_IO);
1472 				goto done;
1473 			}
1474 			free(base_buf);
1475 			base_buf = NULL;
1476 			free(accum_buf);
1477 			accum_buf = NULL;
1478 		}
1479 
1480 		if (base_buf && max_size > base_bufsz) {
1481 			uint8_t *p = realloc(base_buf, max_size);
1482 			if (p == NULL) {
1483 				err = got_error_from_errno("realloc");
1484 				goto done;
1485 			}
1486 			base_buf = p;
1487 			base_bufsz = max_size;
1488 		}
1489 
1490 		if (accum_buf && max_size > accum_bufsz) {
1491 			uint8_t *p = realloc(accum_buf, max_size);
1492 			if (p == NULL) {
1493 				err = got_error_from_errno("realloc");
1494 				goto done;
1495 			}
1496 			accum_buf = p;
1497 			accum_bufsz = max_size;
1498 		}
1499 
1500 		if (base_buf) {
1501 			err = got_delta_apply_in_mem(base_buf, base_bufsz,
1502 			    delta_buf, delta_len, accum_buf,
1503 			    &accum_size, max_size);
1504 			n++;
1505 		} else {
1506 			err = got_delta_apply(base_file, delta_buf,
1507 			    delta_len,
1508 			    /* Final delta application writes to output file. */
1509 			    ++n < deltas->nentries ? accum_file : outfile,
1510 			    &accum_size);
1511 		}
1512 		if (!cached)
1513 			free(delta_buf);
1514 		if (err)
1515 			goto done;
1516 
1517 		if (n < deltas->nentries) {
1518 			/* Accumulated delta becomes the new base. */
1519 			if (base_buf) {
1520 				uint8_t *tmp = accum_buf;
1521 				size_t tmp_size = accum_bufsz;
1522 				accum_buf = base_buf;
1523 				accum_bufsz = base_bufsz;
1524 				base_buf = tmp;
1525 				base_bufsz = tmp_size;
1526 			} else {
1527 				FILE *tmp = accum_file;
1528 				accum_file = base_file;
1529 				base_file = tmp;
1530 				rewind(base_file);
1531 				rewind(accum_file);
1532 			}
1533 		}
1534 	}
1535 
1536 done:
1537 	free(base_buf);
1538 	if (accum_buf) {
1539 		size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1540 		free(accum_buf);
1541 		if (len != accum_size)
1542 			err = got_ferror(outfile, GOT_ERR_IO);
1543 	}
1544 	rewind(outfile);
1545 	if (err == NULL)
1546 		*result_size = accum_size;
1547 	return err;
1548 }
1549 
1550 const struct got_error *
1551 got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1552     struct got_delta_chain *deltas, struct got_pack *pack)
1553 {
1554 	const struct got_error *err = NULL;
1555 	struct got_delta *delta;
1556 	uint8_t *base_buf = NULL, *accum_buf = NULL;
1557 	size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1558 	uint64_t max_size = 0;
1559 	int n = 0;
1560 
1561 	*outbuf = NULL;
1562 	*outlen = 0;
1563 
1564 	if (STAILQ_EMPTY(&deltas->entries))
1565 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1566 
1567 	/* Deltas are ordered in ascending order. */
1568 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1569 		uint8_t *delta_buf = NULL;
1570 		uint64_t base_size, result_size = 0;
1571 		int cached = 1;
1572 		if (n == 0) {
1573 			size_t delta_data_offset;
1574 
1575 			/* Plain object types are the delta base. */
1576 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1577 			    delta->type != GOT_OBJ_TYPE_TREE &&
1578 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1579 			    delta->type != GOT_OBJ_TYPE_TAG) {
1580 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1581 				goto done;
1582 			}
1583 
1584 			delta_data_offset = delta->offset + delta->tslen;
1585 			if (delta_data_offset >= pack->filesize) {
1586 				err = got_error(GOT_ERR_PACK_OFFSET);
1587 				goto done;
1588 			}
1589 
1590 			if (delta->size > max_size)
1591 				max_size = delta->size;
1592 
1593 			if (pack->map) {
1594 				size_t mapoff = (size_t)delta_data_offset;
1595 				err = got_inflate_to_mem_mmap(&base_buf,
1596 				    &base_bufsz, NULL, NULL, pack->map,
1597 				    mapoff, pack->filesize - mapoff);
1598 			} else {
1599 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1600 				    == -1) {
1601 					err = got_error_from_errno("lseek");
1602 					goto done;
1603 				}
1604 				err = got_inflate_to_mem_fd(&base_buf,
1605 				    &base_bufsz, NULL, NULL, max_size,
1606 				    pack->fd);
1607 			}
1608 			if (err)
1609 				goto done;
1610 			n++;
1611 			continue;
1612 		}
1613 
1614 		if (pack->delta_cache) {
1615 			got_delta_cache_get(&delta_buf, &delta_len,
1616 			    pack->delta_cache, delta->data_offset);
1617 		}
1618 		if (delta_buf == NULL) {
1619 			cached = 0;
1620 			err = read_delta_data(&delta_buf, &delta_len, NULL,
1621 			    delta->data_offset, pack);
1622 			if (err)
1623 				goto done;
1624 		}
1625 		if (pack->delta_cache && !cached) {
1626 			err = got_delta_cache_add(pack->delta_cache,
1627 			    delta->data_offset, delta_buf, delta_len);
1628 			if (err == NULL)
1629 				cached = 1;
1630 			else if (err->code != GOT_ERR_NO_SPACE) {
1631 				free(delta_buf);
1632 				goto done;
1633 			}
1634 		}
1635 
1636 		err = got_delta_get_sizes(&base_size, &result_size,
1637 		    delta_buf, delta_len);
1638 		if (err)
1639 			goto done;
1640 		if (base_size > max_size)
1641 			max_size = base_size;
1642 		if (result_size > max_size)
1643 			max_size = result_size;
1644 
1645 		if (max_size > base_bufsz) {
1646 			uint8_t *p = realloc(base_buf, max_size);
1647 			if (p == NULL) {
1648 				err = got_error_from_errno("realloc");
1649 				goto done;
1650 			}
1651 			base_buf = p;
1652 			base_bufsz = max_size;
1653 		}
1654 
1655 		if (max_size > accum_bufsz) {
1656 			uint8_t *p = realloc(accum_buf, max_size);
1657 			if (p == NULL) {
1658 				err = got_error_from_errno("realloc");
1659 				goto done;
1660 			}
1661 			accum_buf = p;
1662 			accum_bufsz = max_size;
1663 		}
1664 
1665 		err = got_delta_apply_in_mem(base_buf, base_bufsz,
1666 		    delta_buf, delta_len, accum_buf,
1667 		    &accum_size, max_size);
1668 		if (!cached)
1669 			free(delta_buf);
1670 		n++;
1671 		if (err)
1672 			goto done;
1673 
1674 		if (n < deltas->nentries) {
1675 			/* Accumulated delta becomes the new base. */
1676 			uint8_t *tmp = accum_buf;
1677 			size_t tmp_size = accum_bufsz;
1678 			accum_buf = base_buf;
1679 			accum_bufsz = base_bufsz;
1680 			base_buf = tmp;
1681 			base_bufsz = tmp_size;
1682 		}
1683 	}
1684 
1685 done:
1686 	free(base_buf);
1687 	if (err) {
1688 		free(accum_buf);
1689 		*outbuf = NULL;
1690 		*outlen = 0;
1691 	} else {
1692 		*outbuf = accum_buf;
1693 		*outlen = accum_size;
1694 	}
1695 	return err;
1696 }
1697 
1698 const struct got_error *
1699 got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1700     FILE *outfile, FILE *base_file, FILE *accum_file)
1701 {
1702 	const struct got_error *err = NULL;
1703 
1704 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1705 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1706 
1707 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1708 		if (obj->pack_offset >= pack->filesize)
1709 			return got_error(GOT_ERR_PACK_OFFSET);
1710 
1711 		if (pack->map) {
1712 			size_t mapoff = (size_t)obj->pack_offset;
1713 			err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1714 			    pack->map, mapoff, pack->filesize - mapoff,
1715 			    outfile);
1716 		} else {
1717 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1718 				return got_error_from_errno("lseek");
1719 			err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1720 			    pack->fd, outfile);
1721 		}
1722 	} else
1723 		err = got_pack_dump_delta_chain_to_file(&obj->size,
1724 		    &obj->deltas, pack, outfile, base_file, accum_file);
1725 
1726 	return err;
1727 }
1728 
1729 const struct got_error *
1730 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1731     struct got_object *obj, struct got_pack *pack)
1732 {
1733 	const struct got_error *err = NULL;
1734 
1735 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1736 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1737 
1738 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1739 		if (obj->pack_offset >= pack->filesize)
1740 			return got_error(GOT_ERR_PACK_OFFSET);
1741 		if (pack->map) {
1742 			size_t mapoff = (size_t)obj->pack_offset;
1743 			err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1744 			    pack->map, mapoff, pack->filesize - mapoff);
1745 		} else {
1746 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1747 				return got_error_from_errno("lseek");
1748 			err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1749 			    obj->size, pack->fd);
1750 		}
1751 	} else
1752 		err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1753 		    pack);
1754 
1755 	return err;
1756 }
1757 
1758 static const struct got_error *
1759 read_raw_delta_data(uint8_t **delta_buf, size_t *delta_len,
1760     size_t *delta_len_compressed, uint64_t *base_size, uint64_t *result_size,
1761     off_t delta_data_offset, struct got_pack *pack, struct got_packidx *packidx)
1762 {
1763 	const struct got_error *err = NULL;
1764 
1765 	/* Validate decompression and obtain the decompressed size. */
1766 	err = read_delta_data(delta_buf, delta_len, delta_len_compressed,
1767 	    delta_data_offset, pack);
1768 	if (err)
1769 		return err;
1770 
1771 	/* Read delta base/result sizes from head of delta stream. */
1772 	err = got_delta_get_sizes(base_size, result_size,
1773 	    *delta_buf, *delta_len);
1774 	if (err)
1775 		goto done;
1776 
1777 	/* Discard decompressed delta and read it again in compressed form. */
1778 	free(*delta_buf);
1779 	*delta_buf = malloc(*delta_len_compressed);
1780 	if (*delta_buf == NULL) {
1781 		err = got_error_from_errno("malloc");
1782 		goto done;
1783 	}
1784 	if (pack->map) {
1785 		if (delta_data_offset >= pack->filesize) {
1786 			err = got_error(GOT_ERR_PACK_OFFSET);
1787 			goto done;
1788 		}
1789 		memcpy(*delta_buf, pack->map + delta_data_offset,
1790 		    *delta_len_compressed);
1791 	} else {
1792 		ssize_t n;
1793 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1794 			err = got_error_from_errno("lseek");
1795 			goto done;
1796 		}
1797 		n = read(pack->fd, *delta_buf, *delta_len_compressed);
1798 		if (n < 0) {
1799 			err = got_error_from_errno("read");
1800 			goto done;
1801 		} else if (n != *delta_len_compressed) {
1802 			err = got_error(GOT_ERR_IO);
1803 			goto done;
1804 		}
1805 	}
1806 done:
1807 	if (err) {
1808 		free(*delta_buf);
1809 		*delta_buf = NULL;
1810 		*delta_len = 0;
1811 		*delta_len_compressed = 0;
1812 		*base_size = 0;
1813 		*result_size = 0;
1814 	}
1815 	return err;
1816 }
1817 
1818 const struct got_error *
1819 got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
1820     size_t *delta_compressed_size, off_t *delta_offset, off_t *base_offset,
1821     struct got_object_id *base_id, uint64_t *base_size, uint64_t *result_size,
1822     struct got_pack *pack, struct got_packidx *packidx, int idx)
1823 {
1824 	const struct got_error *err = NULL;
1825 	off_t offset;
1826 	uint8_t type;
1827 	uint64_t size;
1828 	size_t tslen, delta_hdrlen;
1829 	off_t delta_data_offset;
1830 
1831 	*delta_buf = NULL;
1832 	*delta_size = 0;
1833 	*delta_compressed_size = 0;
1834 	*delta_offset = 0;
1835 	*base_offset = 0;
1836 	*base_size = 0;
1837 	*result_size = 0;
1838 
1839 	offset = got_packidx_get_object_offset(packidx, idx);
1840 	if (offset == -1)
1841 		return got_error(GOT_ERR_BAD_PACKIDX);
1842 
1843 	if (offset >= pack->filesize)
1844 		return got_error(GOT_ERR_PACK_OFFSET);
1845 
1846 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1847 	    pack, offset);
1848 	if (err)
1849 		return err;
1850 
1851 	if (tslen + size < tslen || offset + size < size ||
1852 	    tslen + offset < tslen)
1853 		return got_error(GOT_ERR_PACK_OFFSET);
1854 
1855 	switch (type) {
1856 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1857 		err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
1858 		    pack, offset, tslen);
1859 		if (err)
1860 			return err;
1861 		break;
1862 	case GOT_OBJ_TYPE_REF_DELTA:
1863 		err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
1864 		if (err)
1865 			return err;
1866 		delta_hdrlen = SHA1_DIGEST_LENGTH;
1867 		break;
1868 	default:
1869 		return got_error_fmt(GOT_ERR_OBJ_TYPE,
1870 		    "non-delta object type %d found at offset %lld",
1871 		    type, (long long)offset);
1872 	}
1873 
1874 	if (tslen + delta_hdrlen < delta_hdrlen ||
1875 	    offset + delta_hdrlen < delta_hdrlen)
1876 		return got_error(GOT_ERR_BAD_DELTA);
1877 
1878 	delta_data_offset = offset + tslen + delta_hdrlen;
1879 	err = read_raw_delta_data(delta_buf, delta_size, delta_compressed_size,
1880 	    base_size, result_size, delta_data_offset, pack, packidx);
1881 	if (err)
1882 		return err;
1883 
1884 	if (*delta_size != size) {
1885 		err = got_error(GOT_ERR_BAD_DELTA);
1886 		goto done;
1887 	}
1888 
1889 	*delta_offset = offset;
1890 done:
1891 	if (err) {
1892 		free(*delta_buf);
1893 		*delta_buf = NULL;
1894 		*delta_size = 0;
1895 		*delta_compressed_size = 0;
1896 		*delta_offset = 0;
1897 		*base_offset = 0;
1898 		*base_size = 0;
1899 		*result_size = 0;
1900 	}
1901 	return err;
1902 }