Blob


1 /*
2 * Copyright (c) 2018 Stefan Sperling <stsp@openbsd.org>
3 *
4 * Permission to use, copy, modify, and distribute this software for any
5 * purpose with or without fee is hereby granted, provided that the above
6 * copyright notice and this permission notice appear in all copies.
7 *
8 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15 */
17 #include <sys/types.h>
18 #include <sys/stat.h>
19 #include <sys/queue.h>
20 #include <sys/mman.h>
22 #include <dirent.h>
23 #include <fcntl.h>
24 #include <errno.h>
25 #include <stdio.h>
26 #include <stdint.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <limits.h>
30 #include <sha1.h>
31 #include <endian.h>
32 #include <zlib.h>
34 #include "got_error.h"
35 #include "got_object.h"
36 #include "got_repository.h"
37 #include "got_opentemp.h"
39 #include "got_lib_sha1.h"
40 #include "got_lib_pack.h"
41 #include "got_lib_path.h"
42 #include "got_lib_delta.h"
43 #include "got_lib_zbuf.h"
44 #include "got_lib_object.h"
45 #include "got_lib_repository.h"
47 #ifndef nitems
48 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
49 #endif
51 #define GOT_PACK_PREFIX "pack-"
52 #define GOT_PACKFILE_SUFFIX ".pack"
53 #define GOT_PACKIDX_SUFFIX ".idx"
54 #define GOT_PACKFILE_NAMELEN (strlen(GOT_PACK_PREFIX) + \
55 SHA1_DIGEST_STRING_LENGTH - 1 + \
56 strlen(GOT_PACKFILE_SUFFIX))
57 #define GOT_PACKIDX_NAMELEN (strlen(GOT_PACK_PREFIX) + \
58 SHA1_DIGEST_STRING_LENGTH - 1 + \
59 strlen(GOT_PACKIDX_SUFFIX))
61 #ifndef MIN
62 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
63 #endif
65 static const struct got_error *
66 verify_fanout_table(uint32_t *fanout_table)
67 {
68 int i;
70 for (i = 0; i < 0xff - 1; i++) {
71 if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
72 return got_error(GOT_ERR_BAD_PACKIDX);
73 }
75 return NULL;
76 }
78 static const struct got_error *
79 get_packfile_size(size_t *size, const char *path)
80 {
81 struct stat sb;
82 char *dot;
84 *size = 0;
86 dot = strrchr(path, '.');
87 if (dot == NULL)
88 return got_error(GOT_ERR_BAD_PATH);
90 /* Path must point to a pack index or to a pack file. */
91 if (strcmp(dot, GOT_PACKIDX_SUFFIX) == 0) {
92 const struct got_error *err = NULL;
93 char *path_pack;
94 char base_path[PATH_MAX];
96 /* Convert pack index path to pack file path. */
97 if (strlcpy(base_path, path, PATH_MAX) > PATH_MAX)
98 return got_error(GOT_ERR_NO_SPACE);
99 dot = strrchr(base_path, '.');
100 if (dot == NULL)
101 return got_error(GOT_ERR_BAD_PATH);
102 *dot = '\0';
103 if (asprintf(&path_pack, "%s.pack", base_path) == -1)
104 return got_error_from_errno();
106 if (stat(path_pack, &sb) != 0)
107 err = got_error_from_errno();
108 free(path_pack);
109 if (err)
110 return err;
111 } else if (strcmp(dot, GOT_PACKFILE_SUFFIX) == 0) {
112 if (stat(path, &sb) != 0)
113 return got_error_from_errno();
114 } else
115 return got_error(GOT_ERR_BAD_PATH);
117 *size = sb.st_size;
118 return 0;
121 const struct got_error *
122 got_packidx_open(struct got_packidx **packidx, const char *path)
124 struct got_packidx *p;
125 struct got_packidx_v2_hdr *h;
126 const struct got_error *err = NULL;
127 size_t nobj, len_fanout, len_ids, offset, remain;
128 SHA1_CTX ctx;
129 uint8_t sha1[SHA1_DIGEST_LENGTH];
131 *packidx = NULL;
133 SHA1Init(&ctx);
135 p = calloc(1, sizeof(*p));
136 if (p == NULL)
137 return got_error_from_errno();
139 p->fd = open(path, O_RDONLY | O_NOFOLLOW, GOT_DEFAULT_FILE_MODE);
140 if (p->fd == -1)
141 return got_error_from_errno();
143 err = get_packfile_size(&p->len, path);
144 if (err) {
145 close(p->fd);
146 free(p);
147 return err;
149 if (p->len < sizeof(p->hdr)) {
150 err = got_error(GOT_ERR_BAD_PACKIDX);
151 close(p->fd);
152 free(p);
153 return err;
156 p->path_packidx = strdup(path);
157 if (p->path_packidx == NULL) {
158 err = got_error_from_errno();
159 goto done;
162 p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
163 if (p->map == MAP_FAILED) {
164 err = got_error_from_errno();
165 goto done;
167 h = &p->hdr;
168 offset = 0;
169 remain = p->len;
171 if (remain < sizeof(*h->magic)) {
172 err = got_error(GOT_ERR_BAD_PACKIDX);
173 goto done;
175 h->magic = (uint32_t *)(p->map + offset);
176 if (betoh32(*h->magic) != GOT_PACKIDX_V2_MAGIC) {
177 err = got_error(GOT_ERR_BAD_PACKIDX);
178 goto done;
180 offset += sizeof(*h->magic);
181 remain -= sizeof(*h->magic);
183 SHA1Update(&ctx, (uint8_t *)h->magic, sizeof(*h->magic));
185 if (remain < sizeof(*h->version)) {
186 err = got_error(GOT_ERR_BAD_PACKIDX);
187 goto done;
189 h->version = (uint32_t *)(p->map + offset);
190 if (betoh32(*h->version) != GOT_PACKIDX_VERSION) {
191 err = got_error(GOT_ERR_BAD_PACKIDX);
192 goto done;
194 offset += sizeof(*h->version);
195 remain -= sizeof(*h->version);
197 SHA1Update(&ctx, (uint8_t *)h->version, sizeof(*h->version));
199 len_fanout =
200 sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
201 if (remain < len_fanout) {
202 err = got_error(GOT_ERR_BAD_PACKIDX);
203 goto done;
205 h->fanout_table = (uint32_t *)(p->map + offset);
206 err = verify_fanout_table(h->fanout_table);
207 if (err)
208 goto done;
209 SHA1Update(&ctx, (uint8_t *)h->fanout_table, len_fanout);
210 offset += len_fanout;
211 remain -= len_fanout;
213 nobj = betoh32(h->fanout_table[0xff]);
214 len_ids = nobj * sizeof(*h->sorted_ids);
215 if (len_ids <= nobj || len_ids > remain) {
216 err = got_error(GOT_ERR_BAD_PACKIDX);
217 goto done;
219 h->sorted_ids =
220 (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
221 SHA1Update(&ctx, (uint8_t *)h->sorted_ids, len_ids);
222 offset += len_ids;
223 remain -= len_ids;
225 if (remain < nobj * sizeof(*h->crc32)) {
226 err = got_error(GOT_ERR_BAD_PACKIDX);
227 goto done;
229 h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
230 SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
231 remain -= nobj * sizeof(*h->crc32);
232 offset += nobj * sizeof(*h->crc32);
234 if (remain < nobj * sizeof(*h->offsets)) {
235 err = got_error(GOT_ERR_BAD_PACKIDX);
236 goto done;
238 h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
239 SHA1Update(&ctx, (uint8_t *)h->offsets, nobj * sizeof(*h->offsets));
240 remain -= nobj * sizeof(*h->offsets);
241 offset += nobj * sizeof(*h->offsets);
243 /* Large file offsets are contained only in files > 2GB. */
244 if (p->len <= 0x80000000)
245 goto checksum;
247 if (remain < nobj * sizeof(*h->large_offsets)) {
248 err = got_error(GOT_ERR_BAD_PACKIDX);
249 goto done;
251 h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
252 SHA1Update(&ctx, (uint8_t*)h->large_offsets,
253 nobj * sizeof(*h->large_offsets));
254 remain -= nobj * sizeof(*h->large_offsets);
255 offset += nobj * sizeof(*h->large_offsets);
257 checksum:
258 if (remain < sizeof(*h->trailer)) {
259 err = got_error(GOT_ERR_BAD_PACKIDX);
260 goto done;
262 h->trailer =
263 (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
264 SHA1Update(&ctx, h->trailer->packfile_sha1, SHA1_DIGEST_LENGTH);
265 SHA1Final(sha1, &ctx);
266 if (memcmp(h->trailer->packidx_sha1, sha1, SHA1_DIGEST_LENGTH) != 0)
267 err = got_error(GOT_ERR_PACKIDX_CSUM);
268 done:
269 if (err)
270 got_packidx_close(p);
271 else
272 *packidx = p;
273 return err;
276 const struct got_error *
277 got_packidx_close(struct got_packidx *packidx)
279 const struct got_error *err = NULL;
281 free(packidx->path_packidx);
282 if (packidx->map != NULL && packidx->map != MAP_FAILED) {
283 if (munmap(packidx->map, packidx->len) == -1)
284 err = got_error_from_errno();
286 close(packidx->fd);
287 free(packidx);
289 return err;
292 static int
293 is_packidx_filename(const char *name, size_t len)
295 if (len != GOT_PACKIDX_NAMELEN)
296 return 0;
298 if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
299 return 0;
301 if (strcmp(name + strlen(GOT_PACK_PREFIX) +
302 SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
303 return 0;
305 return 1;
308 static off_t
309 get_object_offset(struct got_packidx *packidx, int idx)
311 uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
312 uint32_t offset = betoh32(packidx->hdr.offsets[idx]);
313 if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
314 uint64_t loffset;
315 idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
316 if (idx < 0 || idx > totobj ||
317 packidx->hdr.large_offsets == NULL)
318 return -1;
319 loffset = betoh64(packidx->hdr.large_offsets[idx]);
320 return (loffset > INT64_MAX ? -1 : (off_t)loffset);
322 return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
325 static int
326 get_object_idx(struct got_packidx *packidx, struct got_object_id *id,
327 struct got_repository *repo)
329 u_int8_t id0 = id->sha1[0];
330 uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
331 int left = 0, right = totobj - 1;
333 if (id0 > 0)
334 left = betoh32(packidx->hdr.fanout_table[id0 - 1]);
336 while (left <= right) {
337 struct got_packidx_object_id *oid;
338 int i, cmp;
340 i = ((left + right) / 2);
341 oid = &packidx->hdr.sorted_ids[i];
342 cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
343 if (cmp == 0)
344 return i;
345 else if (cmp > 0)
346 left = i + 1;
347 else if (cmp < 0)
348 right = i - 1;
351 return -1;
354 static const struct got_error *
355 cache_packidx(struct got_packidx *packidx, struct got_repository *repo)
357 const struct got_error *err = NULL;
358 int i;
360 for (i = 0; i < nitems(repo->packidx_cache); i++) {
361 if (repo->packidx_cache[i] == NULL)
362 break;
365 if (i == nitems(repo->packidx_cache)) {
366 err = got_packidx_close(repo->packidx_cache[i - 1]);
367 if (err)
368 return err;
369 memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
370 sizeof(repo->packidx_cache) -
371 sizeof(repo->packidx_cache[0]));
372 i = 0;
375 repo->packidx_cache[i] = packidx;
376 return NULL;
379 static const struct got_error *
380 search_packidx(struct got_packidx **packidx, int *idx,
381 struct got_repository *repo, struct got_object_id *id)
383 const struct got_error *err;
384 char *path_packdir;
385 DIR *packdir;
386 struct dirent *dent;
387 char *path_packidx;
388 int i;
390 /* Search pack index cache. */
391 for (i = 0; i < nitems(repo->packidx_cache); i++) {
392 if (repo->packidx_cache[i] == NULL)
393 break;
394 *idx = get_object_idx(repo->packidx_cache[i], id, repo);
395 if (*idx != -1) {
396 *packidx = repo->packidx_cache[i];
397 return NULL;
400 /* No luck. Search the filesystem. */
402 path_packdir = got_repo_get_path_objects_pack(repo);
403 if (path_packdir == NULL)
404 return got_error_from_errno();
406 packdir = opendir(path_packdir);
407 if (packdir == NULL) {
408 err = got_error_from_errno();
409 goto done;
412 while ((dent = readdir(packdir)) != NULL) {
413 if (!is_packidx_filename(dent->d_name, dent->d_namlen))
414 continue;
416 if (asprintf(&path_packidx, "%s/%s", path_packdir,
417 dent->d_name) == -1) {
418 err = got_error_from_errno();
419 goto done;
422 err = got_packidx_open(packidx, path_packidx);
423 free(path_packidx);
424 if (err)
425 goto done;
427 *idx = get_object_idx(*packidx, id, repo);
428 if (*idx != -1) {
429 err = NULL; /* found the object */
430 err = cache_packidx(*packidx, repo);
431 goto done;
434 err = got_packidx_close(*packidx);
435 *packidx = NULL;
436 if (err)
437 goto done;
440 err = got_error(GOT_ERR_NO_OBJ);
441 done:
442 free(path_packdir);
443 if (packdir && closedir(packdir) != 0 && err == 0)
444 err = got_error_from_errno();
445 return err;
448 static const struct got_error *
449 get_packfile_path(char **path_packfile, struct got_repository *repo,
450 struct got_packidx *packidx)
452 size_t size;
454 /* Packfile path contains ".pack" instead of ".idx", so add one byte. */
455 size = strlen(packidx->path_packidx) + 2;
456 if (size < GOT_PACKFILE_NAMELEN + 1)
457 return got_error(GOT_ERR_BAD_PATH);
459 *path_packfile = calloc(size, sizeof(**path_packfile));
460 if (*path_packfile == NULL)
461 return got_error_from_errno();
463 /* Copy up to and excluding ".idx". */
464 if (strlcpy(*path_packfile, packidx->path_packidx,
465 size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
466 return got_error(GOT_ERR_NO_SPACE);
468 if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
469 return got_error(GOT_ERR_NO_SPACE);
471 return NULL;
474 const struct got_error *
475 read_packfile_hdr(int fd, struct got_packidx *packidx)
477 const struct got_error *err = NULL;
478 uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
479 struct got_packfile_hdr hdr;
480 ssize_t n;
482 n = read(fd, &hdr, sizeof(hdr));
483 if (n < 0)
484 return got_error_from_errno();
485 if (n != sizeof(hdr))
486 return got_error(GOT_ERR_BAD_PACKFILE);
488 if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
489 betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
490 betoh32(hdr.nobjects) != totobj)
491 err = got_error(GOT_ERR_BAD_PACKFILE);
493 return err;
496 static const struct got_error *
497 open_packfile(int *fd, const char *path_packfile,
498 struct got_repository *repo, struct got_packidx *packidx)
500 const struct got_error *err = NULL;
502 *fd = open(path_packfile, O_RDONLY | O_NOFOLLOW, GOT_DEFAULT_FILE_MODE);
503 if (*fd == -1)
504 return got_error_from_errno();
506 if (packidx) {
507 err = read_packfile_hdr(*fd, packidx);
508 if (err) {
509 close(*fd);
510 *fd = -1;
513 return err;
516 void
517 got_pack_close(struct got_pack *pack)
519 close(pack->fd);
520 pack->fd = -1;
521 free(pack->path_packfile);
522 pack->path_packfile = NULL;
523 pack->filesize = 0;
526 static const struct got_error *
527 cache_pack(struct got_pack **packp, const char *path_packfile,
528 struct got_packidx *packidx, struct got_repository *repo)
530 const struct got_error *err = NULL;
531 struct got_pack *pack = NULL;
532 int i;
534 if (packp)
535 *packp = NULL;
537 for (i = 0; i < nitems(repo->packs); i++) {
538 pack = &repo->packs[i];
539 if (pack->path_packfile == NULL)
540 break;
541 if (strcmp(pack->path_packfile, path_packfile) == 0)
542 return NULL;
545 if (i == nitems(repo->packs) - 1) {
546 got_pack_close(&repo->packs[i - 1]);
547 memmove(&repo->packs[1], &repo->packs[0],
548 sizeof(repo->packs) - sizeof(repo->packs[0]));
549 i = 0;
552 pack = &repo->packs[i];
554 pack->path_packfile = strdup(path_packfile);
555 if (pack->path_packfile == NULL) {
556 err = got_error_from_errno();
557 goto done;
560 err = open_packfile(&pack->fd, path_packfile, repo, packidx);
561 if (err)
562 goto done;
564 err = get_packfile_size(&pack->filesize, path_packfile);
565 done:
566 if (err) {
567 if (pack) {
568 free(pack->path_packfile);
569 memset(pack, 0, sizeof(*pack));
571 } else if (packp)
572 *packp = pack;
573 return err;
576 struct got_pack *
577 get_cached_pack(const char *path_packfile, struct got_repository *repo)
579 struct got_pack *pack = NULL;
580 int i;
582 for (i = 0; i < nitems(repo->packs); i++) {
583 pack = &repo->packs[i];
584 if (pack->path_packfile == NULL)
585 break;
586 if (strcmp(pack->path_packfile, path_packfile) == 0)
587 return pack;
590 return NULL;
593 static const struct got_error *
594 parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len, int fd)
596 uint8_t t = 0;
597 uint64_t s = 0;
598 uint8_t sizeN;
599 ssize_t n;
600 int i = 0;
602 do {
603 /* We do not support size values which don't fit in 64 bit. */
604 if (i > 9)
605 return got_error(GOT_ERR_NO_SPACE);
607 n = read(fd, &sizeN, sizeof(sizeN));
608 if (n < 0)
609 return got_error_from_errno();
610 if (n != sizeof(sizeN))
611 return got_error(GOT_ERR_BAD_PACKFILE);
613 if (i == 0) {
614 t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
615 GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
616 s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
617 } else {
618 size_t shift = 4 + 7 * (i - 1);
619 s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
621 i++;
622 } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
624 *type = t;
625 *size = s;
626 *len = i * sizeof(sizeN);
627 return NULL;
630 static const struct got_error *
631 open_plain_object(struct got_object **obj, const char *path_packfile,
632 struct got_object_id *id, uint8_t type, off_t offset, size_t size)
634 *obj = calloc(1, sizeof(**obj));
635 if (*obj == NULL)
636 return got_error_from_errno();
638 (*obj)->path_packfile = strdup(path_packfile);
639 if ((*obj)->path_packfile == NULL) {
640 const struct got_error *err = got_error_from_errno();
641 free(*obj);
642 *obj = NULL;
643 return err;
646 (*obj)->type = type;
647 (*obj)->flags = GOT_OBJ_FLAG_PACKED;
648 (*obj)->hdrlen = 0;
649 (*obj)->size = size;
650 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
651 (*obj)->pack_offset = offset;
653 return NULL;
656 static const struct got_error *
657 parse_negative_offset(int64_t *offset, size_t *len, int fd)
659 int64_t o = 0;
660 uint8_t offN;
661 ssize_t n;
662 int i = 0;
664 do {
665 /* We do not support offset values which don't fit in 64 bit. */
666 if (i > 8)
667 return got_error(GOT_ERR_NO_SPACE);
669 n = read(fd, &offN, sizeof(offN));
670 if (n < 0)
671 return got_error_from_errno();
672 if (n != sizeof(offN))
673 return got_error(GOT_ERR_BAD_PACKFILE);
675 if (i == 0)
676 o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
677 else {
678 o++;
679 o <<= 7;
680 o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
682 i++;
683 } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
685 *offset = o;
686 *len = i * sizeof(offN);
687 return NULL;
690 static const struct got_error *
691 parse_offset_delta(off_t *base_offset, int fd, off_t offset)
693 const struct got_error *err;
694 int64_t negoffset;
695 size_t negofflen;
697 err = parse_negative_offset(&negoffset, &negofflen, fd);
698 if (err)
699 return err;
701 /* Compute the base object's offset (must be in the same pack file). */
702 *base_offset = (offset - negoffset);
703 if (*base_offset <= 0)
704 return got_error(GOT_ERR_BAD_PACKFILE);
706 return NULL;
709 static const struct got_error *
710 resolve_delta_chain(struct got_delta_chain *, struct got_repository *,
711 struct got_packidx *, struct got_pack *, off_t, size_t, int,
712 size_t, unsigned int);
714 static const struct got_error *
715 add_delta(struct got_delta_chain *deltas, const char *path_packfile,
716 off_t delta_offset, size_t tslen, int delta_type, size_t delta_size,
717 size_t delta_data_offset, uint8_t *delta_buf, size_t delta_len)
719 struct got_delta *delta;
721 delta = got_delta_open(path_packfile, delta_offset, tslen,
722 delta_type, delta_size, delta_data_offset, delta_buf,
723 delta_len);
724 if (delta == NULL)
725 return got_error_from_errno();
726 /* delta is freed in got_object_close() */
727 deltas->nentries++;
728 SIMPLEQ_INSERT_HEAD(&deltas->entries, delta, entry);
729 return NULL;
732 static const struct got_error *
733 resolve_offset_delta(struct got_delta_chain *deltas,
734 struct got_repository *repo, struct got_packidx *packidx,
735 struct got_pack *pack, off_t delta_offset, size_t tslen,
736 int delta_type, size_t delta_size, unsigned int recursion)
739 const struct got_error *err;
740 off_t base_offset;
741 uint8_t base_type;
742 uint64_t base_size;
743 size_t base_tslen;
744 off_t delta_data_offset;
745 uint8_t *delta_buf;
746 size_t delta_len;
748 err = parse_offset_delta(&base_offset, pack->fd, delta_offset);
749 if (err)
750 return err;
752 delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
753 if (delta_data_offset == -1)
754 return got_error_from_errno();
756 err = got_inflate_to_mem_fd(&delta_buf, &delta_len, pack->fd);
757 if (err)
758 return err;
760 err = add_delta(deltas, pack->path_packfile, delta_offset, tslen,
761 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
762 if (err)
763 return err;
765 /* An offset delta must be in the same packfile. */
766 if (base_offset >= pack->filesize)
767 return got_error(GOT_ERR_PACK_OFFSET);
768 if (lseek(pack->fd, base_offset, SEEK_SET) == -1)
769 return got_error_from_errno();
771 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
772 pack->fd);
773 if (err)
774 return err;
776 return resolve_delta_chain(deltas, repo, packidx, pack, base_offset,
777 base_tslen, base_type, base_size, recursion - 1);
780 static const struct got_error *
781 resolve_ref_delta(struct got_delta_chain *deltas, struct got_repository *repo,
782 struct got_packidx *packidx, struct got_pack *pack,
783 off_t delta_offset, size_t tslen, int delta_type, size_t delta_size,
784 unsigned int recursion)
786 const struct got_error *err;
787 struct got_object_id id;
788 int idx;
789 off_t base_offset;
790 uint8_t base_type;
791 uint64_t base_size;
792 size_t base_tslen;
793 ssize_t n;
794 off_t delta_data_offset;
795 uint8_t *delta_buf;
796 size_t delta_len;
798 n = read(pack->fd, &id, sizeof(id));
799 if (n < 0)
800 return got_error_from_errno();
801 if (n != sizeof(id))
802 return got_error(GOT_ERR_BAD_PACKFILE);
804 delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
805 if (delta_data_offset == -1)
806 return got_error_from_errno();
808 err = got_inflate_to_mem_fd(&delta_buf, &delta_len, pack->fd);
809 if (err)
810 return err;
812 err = add_delta(deltas, pack->path_packfile, delta_offset, tslen,
813 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
814 if (err)
815 return err;
817 /* Delta base must be in the same pack file. */
818 idx = get_object_idx(packidx, &id, repo);
819 if (idx == -1)
820 return got_error(GOT_ERR_BAD_PACKFILE);
822 base_offset = get_object_offset(packidx, idx);
823 if (base_offset == (uint64_t)-1) {
824 return got_error(GOT_ERR_BAD_PACKIDX);
827 if (base_offset >= pack->filesize) {
828 err = got_error(GOT_ERR_PACK_OFFSET);
829 goto done;
831 if (lseek(pack->fd, base_offset, SEEK_SET) == -1) {
832 err = got_error_from_errno();
833 goto done;
836 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
837 pack->fd);
838 if (err)
839 goto done;
841 err = resolve_delta_chain(deltas, repo, packidx, pack, base_offset,
842 base_tslen, base_type, base_size, recursion - 1);
843 done:
844 return err;
847 static const struct got_error *
848 resolve_delta_chain(struct got_delta_chain *deltas, struct got_repository *repo,
849 struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
850 size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
852 const struct got_error *err = NULL;
854 if (--recursion == 0)
855 return got_error(GOT_ERR_RECURSION);
857 switch (delta_type) {
858 case GOT_OBJ_TYPE_COMMIT:
859 case GOT_OBJ_TYPE_TREE:
860 case GOT_OBJ_TYPE_BLOB:
861 case GOT_OBJ_TYPE_TAG:
862 /* Plain types are the final delta base. Recursion ends. */
863 err = add_delta(deltas, pack->path_packfile, delta_offset,
864 tslen, delta_type, delta_size, 0, NULL, 0);
865 break;
866 case GOT_OBJ_TYPE_OFFSET_DELTA:
867 err = resolve_offset_delta(deltas, repo, packidx, pack,
868 delta_offset, tslen, delta_type, delta_size, recursion - 1);
869 break;
870 case GOT_OBJ_TYPE_REF_DELTA:
871 err = resolve_ref_delta(deltas, repo, packidx, pack,
872 delta_offset, tslen, delta_type, delta_size, recursion - 1);
873 break;
874 default:
875 return got_error(GOT_ERR_OBJ_TYPE);
878 return err;
881 static const struct got_error *
882 open_delta_object(struct got_object **obj, struct got_repository *repo,
883 struct got_packidx *packidx, struct got_pack *pack,
884 struct got_object_id *id, off_t offset, size_t tslen,
885 int delta_type, size_t delta_size)
887 const struct got_error *err = NULL;
888 int resolved_type;
890 *obj = calloc(1, sizeof(**obj));
891 if (*obj == NULL)
892 return got_error_from_errno();
894 (*obj)->flags = 0;
895 (*obj)->hdrlen = 0;
896 (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
897 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
898 (*obj)->pack_offset = offset + tslen;
900 (*obj)->path_packfile = strdup(pack->path_packfile);
901 if ((*obj)->path_packfile == NULL) {
902 err = got_error_from_errno();
903 goto done;
905 (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
907 SIMPLEQ_INIT(&(*obj)->deltas.entries);
908 (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
910 err = resolve_delta_chain(&(*obj)->deltas, repo, packidx, pack, offset,
911 tslen, delta_type, delta_size, GOT_DELTA_CHAIN_RECURSION_MAX);
912 if (err)
913 goto done;
915 err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
916 if (err)
917 goto done;
918 (*obj)->type = resolved_type;
920 done:
921 if (err) {
922 got_object_close(*obj);
923 *obj = NULL;
925 return err;
928 static const struct got_error *
929 open_packed_object(struct got_object **obj, struct got_repository *repo,
930 struct got_packidx *packidx, int idx, struct got_object_id *id)
932 const struct got_error *err = NULL;
933 off_t offset;
934 char *path_packfile;
935 struct got_pack *pack;
936 uint8_t type;
937 uint64_t size;
938 size_t tslen;
940 *obj = NULL;
942 offset = get_object_offset(packidx, idx);
943 if (offset == (uint64_t)-1)
944 return got_error(GOT_ERR_BAD_PACKIDX);
946 err = get_packfile_path(&path_packfile, repo, packidx);
947 if (err)
948 return err;
950 pack = get_cached_pack(path_packfile, repo);
951 if (pack == NULL) {
952 err = cache_pack(&pack, path_packfile, packidx, repo);
953 if (err)
954 goto done;
957 if (offset >= pack->filesize) {
958 err = got_error(GOT_ERR_PACK_OFFSET);
959 goto done;
961 if (lseek(pack->fd, offset, SEEK_SET) == -1) {
962 err = got_error_from_errno();
963 goto done;
966 err = parse_object_type_and_size(&type, &size, &tslen, pack->fd);
967 if (err)
968 goto done;
970 switch (type) {
971 case GOT_OBJ_TYPE_COMMIT:
972 case GOT_OBJ_TYPE_TREE:
973 case GOT_OBJ_TYPE_BLOB:
974 case GOT_OBJ_TYPE_TAG:
975 err = open_plain_object(obj, path_packfile, id, type,
976 offset + tslen, size);
977 break;
979 case GOT_OBJ_TYPE_OFFSET_DELTA:
980 case GOT_OBJ_TYPE_REF_DELTA:
981 err = open_delta_object(obj, repo, packidx, pack, id, offset,
982 tslen, type, size);
983 break;
985 default:
986 err = got_error(GOT_ERR_OBJ_TYPE);
987 goto done;
989 done:
990 free(path_packfile);
991 return err;
994 const struct got_error *
995 got_packfile_open_object(struct got_object **obj, struct got_object_id *id,
996 struct got_repository *repo)
998 const struct got_error *err = NULL;
999 struct got_packidx *packidx = NULL;
1000 int idx;
1002 err = search_packidx(&packidx, &idx, repo, id);
1003 if (err)
1004 return err;
1006 err = open_packed_object(obj, repo, packidx, idx, id);
1007 if (err)
1008 return err;
1010 err = cache_pack(NULL, (*obj)->path_packfile, packidx, repo);
1011 return err;
1014 static const struct got_error *
1015 get_delta_chain_max_size(uint64_t *max_size, struct got_delta_chain *deltas)
1017 struct got_delta *delta;
1018 uint64_t base_size = 0, result_size = 0;
1020 *max_size = 0;
1021 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1022 /* Plain object types are the delta base. */
1023 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1024 delta->type != GOT_OBJ_TYPE_TREE &&
1025 delta->type != GOT_OBJ_TYPE_BLOB &&
1026 delta->type != GOT_OBJ_TYPE_TAG) {
1027 const struct got_error *err;
1028 err = got_delta_get_sizes(&base_size, &result_size,
1029 delta->delta_buf, delta->delta_len);
1030 if (err)
1031 return err;
1032 } else
1033 base_size = delta->size;
1034 if (base_size > *max_size)
1035 *max_size = base_size;
1036 if (result_size > *max_size)
1037 *max_size = result_size;
1040 return NULL;
1043 static const struct got_error *
1044 dump_delta_chain_to_file(size_t *result_size, struct got_delta_chain *deltas,
1045 FILE *outfile, struct got_repository *repo)
1047 const struct got_error *err = NULL;
1048 struct got_delta *delta;
1049 FILE *base_file = NULL, *accum_file = NULL;
1050 uint8_t *base_buf = NULL, *accum_buf = NULL;
1051 size_t accum_size = 0;
1052 uint64_t max_size;
1053 int n = 0;
1055 *result_size = 0;
1057 if (SIMPLEQ_EMPTY(&deltas->entries))
1058 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1060 /* We process small enough files entirely in memory for speed. */
1061 err = get_delta_chain_max_size(&max_size, deltas);
1062 if (err)
1063 return err;
1064 if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1065 accum_buf = malloc(max_size);
1066 if (accum_buf == NULL)
1067 return got_error_from_errno();
1068 } else {
1069 base_file = got_opentemp();
1070 if (base_file == NULL)
1071 return got_error_from_errno();
1073 accum_file = got_opentemp();
1074 if (accum_file == NULL) {
1075 err = got_error_from_errno();
1076 fclose(base_file);
1077 return err;
1081 /* Deltas are ordered in ascending order. */
1082 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1083 if (n == 0) {
1084 struct got_pack *pack;
1085 size_t base_len;
1086 off_t delta_data_offset;
1088 /* Plain object types are the delta base. */
1089 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1090 delta->type != GOT_OBJ_TYPE_TREE &&
1091 delta->type != GOT_OBJ_TYPE_BLOB &&
1092 delta->type != GOT_OBJ_TYPE_TAG) {
1093 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1094 goto done;
1097 pack = get_cached_pack(delta->path_packfile, repo);
1098 if (pack == NULL) {
1099 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1100 goto done;
1103 delta_data_offset = delta->offset + delta->tslen;
1104 if (delta_data_offset >= pack->filesize) {
1105 err = got_error(GOT_ERR_PACK_OFFSET);
1106 goto done;
1108 if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1109 == -1) {
1110 err = got_error_from_errno();
1111 goto done;
1113 if (base_file)
1114 err = got_inflate_to_file_fd(&base_len,
1115 pack->fd, base_file);
1116 else {
1117 err = got_inflate_to_mem_fd(&base_buf,
1118 &base_len, pack->fd);
1119 if (base_len < max_size) {
1120 uint8_t *p;
1121 p = reallocarray(base_buf, 1, max_size);
1122 if (p == NULL) {
1123 err = got_error_from_errno();
1124 goto done;
1126 base_buf = p;
1129 if (err)
1130 goto done;
1131 n++;
1132 if (base_file)
1133 rewind(base_file);
1134 continue;
1137 if (base_buf) {
1138 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1139 delta->delta_len, accum_buf, &accum_size);
1140 n++;
1141 } else {
1142 err = got_delta_apply(base_file, delta->delta_buf,
1143 delta->delta_len,
1144 /* Final delta application writes to output file. */
1145 ++n < deltas->nentries ? accum_file : outfile,
1146 &accum_size);
1148 if (err)
1149 goto done;
1151 if (n < deltas->nentries) {
1152 /* Accumulated delta becomes the new base. */
1153 if (base_buf) {
1154 uint8_t *tmp = accum_buf;
1155 accum_buf = base_buf;
1156 base_buf = tmp;
1157 } else {
1158 FILE *tmp = accum_file;
1159 accum_file = base_file;
1160 base_file = tmp;
1161 rewind(base_file);
1162 rewind(accum_file);
1167 done:
1168 free(base_buf);
1169 if (accum_buf) {
1170 size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1171 free(accum_buf);
1172 if (len != accum_size)
1173 return got_ferror(outfile, GOT_ERR_IO);
1175 if (base_file)
1176 fclose(base_file);
1177 if (accum_file)
1178 fclose(accum_file);
1179 rewind(outfile);
1180 if (err == NULL)
1181 *result_size = accum_size;
1182 return err;
1185 static const struct got_error *
1186 dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1187 struct got_delta_chain *deltas, struct got_repository *repo)
1189 const struct got_error *err = NULL;
1190 struct got_delta *delta;
1191 uint8_t *base_buf = NULL, *accum_buf = NULL;
1192 size_t accum_size;
1193 uint64_t max_size;
1194 int n = 0;
1196 *outbuf = NULL;
1197 *outlen = 0;
1199 if (SIMPLEQ_EMPTY(&deltas->entries))
1200 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1202 err = get_delta_chain_max_size(&max_size, deltas);
1203 if (err)
1204 return err;
1205 accum_buf = malloc(max_size);
1206 if (accum_buf == NULL)
1207 return got_error_from_errno();
1209 /* Deltas are ordered in ascending order. */
1210 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1211 if (n == 0) {
1212 struct got_pack *pack;
1213 size_t base_len;
1214 size_t delta_data_offset;
1216 /* Plain object types are the delta base. */
1217 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1218 delta->type != GOT_OBJ_TYPE_TREE &&
1219 delta->type != GOT_OBJ_TYPE_BLOB &&
1220 delta->type != GOT_OBJ_TYPE_TAG) {
1221 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1222 goto done;
1225 pack = get_cached_pack(delta->path_packfile, repo);
1226 if (pack == NULL) {
1227 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1228 goto done;
1231 delta_data_offset = delta->offset + delta->tslen;
1232 if (delta_data_offset >= pack->filesize) {
1233 err = got_error(GOT_ERR_PACK_OFFSET);
1234 goto done;
1236 if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1237 == -1) {
1238 err = got_error_from_errno();
1239 goto done;
1241 err = got_inflate_to_mem_fd(&base_buf, &base_len,
1242 pack->fd);
1243 if (base_len < max_size) {
1244 uint8_t *p;
1245 p = reallocarray(base_buf, 1, max_size);
1246 if (p == NULL) {
1247 err = got_error_from_errno();
1248 goto done;
1250 base_buf = p;
1252 if (err)
1253 goto done;
1254 n++;
1255 continue;
1258 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1259 delta->delta_len, accum_buf, &accum_size);
1260 n++;
1261 if (err)
1262 goto done;
1264 if (n < deltas->nentries) {
1265 /* Accumulated delta becomes the new base. */
1266 uint8_t *tmp = accum_buf;
1267 accum_buf = base_buf;
1268 base_buf = tmp;
1272 done:
1273 free(base_buf);
1274 if (err) {
1275 free(accum_buf);
1276 *outbuf = NULL;
1277 *outlen = 0;
1278 } else {
1279 *outbuf = accum_buf;
1280 *outlen = accum_size;
1282 return err;
1285 const struct got_error *
1286 got_packfile_extract_object(FILE **f, struct got_object *obj,
1287 struct got_repository *repo)
1289 const struct got_error *err = NULL;
1291 *f = NULL;
1293 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1294 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1296 *f = got_opentemp();
1297 if (*f == NULL) {
1298 err = got_error_from_errno();
1299 goto done;
1302 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1303 struct got_pack *pack;
1305 pack = get_cached_pack(obj->path_packfile, repo);
1306 if (pack == NULL) {
1307 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1308 if (err)
1309 goto done;
1312 if (obj->pack_offset >= pack->filesize) {
1313 err = got_error(GOT_ERR_PACK_OFFSET);
1314 goto done;
1316 if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1) {
1317 err = got_error_from_errno();
1318 goto done;
1321 err = got_inflate_to_file_fd(&obj->size, pack->fd, *f);
1322 } else
1323 err = dump_delta_chain_to_file(&obj->size,
1324 &obj->deltas, *f, repo);
1325 done:
1326 if (err && *f) {
1327 fclose(*f);
1328 *f = NULL;
1330 return err;
1333 const struct got_error *
1334 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1335 struct got_object *obj, struct got_repository *repo)
1337 const struct got_error *err = NULL;
1339 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1340 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1342 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1343 struct got_pack *pack;
1345 pack = get_cached_pack(obj->path_packfile, repo);
1346 if (pack == NULL) {
1347 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1348 if (err)
1349 goto done;
1352 if (obj->pack_offset >= pack->filesize) {
1353 err = got_error(GOT_ERR_PACK_OFFSET);
1354 goto done;
1356 if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1) {
1357 err = got_error_from_errno();
1358 goto done;
1361 err = got_inflate_to_mem_fd(buf, len, pack->fd);
1362 } else
1363 err = dump_delta_chain_to_mem(buf, len, &obj->deltas, repo);
1364 done:
1365 return err;