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>
21 #include <dirent.h>
22 #include <errno.h>
23 #include <stdio.h>
24 #include <stdint.h>
25 #include <stdlib.h>
26 #include <string.h>
27 #include <limits.h>
28 #include <sha1.h>
29 #include <endian.h>
30 #include <zlib.h>
32 #include "got_error.h"
33 #include "got_object.h"
34 #include "got_repository.h"
36 #include "got_sha1_priv.h"
37 #include "got_pack_priv.h"
38 #include "got_path_priv.h"
39 #include "got_delta_priv.h"
40 #include "got_zb_priv.h"
41 #include "got_object_priv.h"
42 #include "got_repository_priv.h"
44 #ifndef nitems
45 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
46 #endif
48 #define GOT_PACK_PREFIX "pack-"
49 #define GOT_PACKFILE_SUFFIX ".pack"
50 #define GOT_PACKIDX_SUFFIX ".idx"
51 #define GOT_PACKFILE_NAMELEN (strlen(GOT_PACK_PREFIX) + \
52 SHA1_DIGEST_STRING_LENGTH - 1 + \
53 strlen(GOT_PACKFILE_SUFFIX))
54 #define GOT_PACKIDX_NAMELEN (strlen(GOT_PACK_PREFIX) + \
55 SHA1_DIGEST_STRING_LENGTH - 1 + \
56 strlen(GOT_PACKIDX_SUFFIX))
58 #ifndef MIN
59 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
60 #endif
62 static const struct got_error *
63 verify_fanout_table(uint32_t *fanout_table)
64 {
65 int i;
67 for (i = 0; i < 0xff - 1; i++) {
68 if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
69 return got_error(GOT_ERR_BAD_PACKIDX);
70 }
72 return NULL;
73 }
75 static const struct got_error *
76 get_packfile_size(size_t *size, const char *path_idx)
77 {
78 struct stat sb;
79 char *path_pack;
80 char base_path[PATH_MAX];
81 char *dot;
83 if (strlcpy(base_path, path_idx, PATH_MAX) > PATH_MAX)
84 return got_error(GOT_ERR_NO_SPACE);
86 dot = strrchr(base_path, '.');
87 if (dot == NULL)
88 return got_error(GOT_ERR_BAD_PATH);
89 *dot = '\0';
90 if (asprintf(&path_pack, "%s.pack", base_path) == -1)
91 return got_error(GOT_ERR_NO_MEM);
93 if (stat(path_pack, &sb) != 0) {
94 free(path_pack);
95 return got_error_from_errno();
96 }
98 free(path_pack);
99 *size = sb.st_size;
100 return 0;
103 const struct got_error *
104 got_packidx_open(struct got_packidx_v2_hdr **packidx, const char *path)
106 struct got_packidx_v2_hdr *p;
107 FILE *f;
108 const struct got_error *err = NULL;
109 size_t n, nobj, packfile_size;
110 SHA1_CTX ctx;
111 uint8_t sha1[SHA1_DIGEST_LENGTH];
113 SHA1Init(&ctx);
115 f = fopen(path, "rb");
116 if (f == NULL)
117 return got_error(GOT_ERR_BAD_PATH);
119 err = get_packfile_size(&packfile_size, path);
120 if (err)
121 return err;
123 p = calloc(1, sizeof(*p));
124 if (p == NULL) {
125 err = got_error(GOT_ERR_NO_MEM);
126 goto done;
129 n = fread(&p->magic, sizeof(p->magic), 1, f);
130 if (n != 1) {
131 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
132 goto done;
135 if (betoh32(p->magic) != GOT_PACKIDX_V2_MAGIC) {
136 err = got_error(GOT_ERR_BAD_PACKIDX);
137 goto done;
140 SHA1Update(&ctx, (uint8_t *)&p->magic, sizeof(p->magic));
142 n = fread(&p->version, sizeof(p->version), 1, f);
143 if (n != 1) {
144 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
145 goto done;
148 if (betoh32(p->version) != GOT_PACKIDX_VERSION) {
149 err = got_error(GOT_ERR_BAD_PACKIDX);
150 goto done;
153 SHA1Update(&ctx, (uint8_t *)&p->version, sizeof(p->version));
155 n = fread(&p->fanout_table, sizeof(p->fanout_table), 1, f);
156 if (n != 1) {
157 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
158 goto done;
161 err = verify_fanout_table(p->fanout_table);
162 if (err)
163 goto done;
165 SHA1Update(&ctx, (uint8_t *)p->fanout_table, sizeof(p->fanout_table));
167 nobj = betoh32(p->fanout_table[0xff]);
169 p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
170 if (p->sorted_ids == NULL) {
171 err = got_error(GOT_ERR_NO_MEM);
172 goto done;
175 n = fread(p->sorted_ids, sizeof(*p->sorted_ids), nobj, f);
176 if (n != nobj) {
177 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
178 goto done;
181 SHA1Update(&ctx, (uint8_t *)p->sorted_ids,
182 nobj * sizeof(*p->sorted_ids));
184 p->crc32 = calloc(nobj, sizeof(*p->crc32));
185 if (p->crc32 == NULL) {
186 err = got_error(GOT_ERR_NO_MEM);
187 goto done;
190 n = fread(p->crc32, sizeof(*p->crc32), nobj, f);
191 if (n != nobj) {
192 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
193 goto done;
196 SHA1Update(&ctx, (uint8_t *)p->crc32, nobj * sizeof(*p->crc32));
198 p->offsets = calloc(nobj, sizeof(*p->offsets));
199 if (p->offsets == NULL) {
200 err = got_error(GOT_ERR_NO_MEM);
201 goto done;
204 n = fread(p->offsets, sizeof(*p->offsets), nobj, f);
205 if (n != nobj) {
206 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
207 goto done;
210 SHA1Update(&ctx, (uint8_t *)p->offsets, nobj * sizeof(*p->offsets));
212 /* Large file offsets are contained only in files > 2GB. */
213 if (packfile_size <= 0x80000000)
214 goto checksum;
216 p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
217 if (p->large_offsets == NULL) {
218 err = got_error(GOT_ERR_NO_MEM);
219 goto done;
222 n = fread(p->large_offsets, sizeof(*p->large_offsets), nobj, f);
223 if (n != nobj) {
224 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
225 goto done;
228 SHA1Update(&ctx, (uint8_t*)p->large_offsets,
229 nobj * sizeof(*p->large_offsets));
231 checksum:
232 n = fread(&p->trailer, sizeof(p->trailer), 1, f);
233 if (n != 1) {
234 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
235 goto done;
238 SHA1Update(&ctx, p->trailer.packfile_sha1, SHA1_DIGEST_LENGTH);
239 SHA1Final(sha1, &ctx);
240 if (memcmp(p->trailer.packidx_sha1, sha1, SHA1_DIGEST_LENGTH) != 0)
241 err = got_error(GOT_ERR_PACKIDX_CSUM);
242 done:
243 fclose(f);
244 if (err)
245 got_packidx_close(p);
246 else
247 *packidx = p;
248 return err;
251 void
252 got_packidx_close(struct got_packidx_v2_hdr *packidx)
254 free(packidx->sorted_ids);
255 free(packidx->offsets);
256 free(packidx->crc32);
257 free(packidx->large_offsets);
258 free(packidx);
261 static int
262 is_packidx_filename(const char *name, size_t len)
264 if (len != GOT_PACKIDX_NAMELEN)
265 return 0;
267 if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
268 return 0;
270 if (strcmp(name + strlen(GOT_PACK_PREFIX) +
271 SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
272 return 0;
274 return 1;
277 static off_t
278 get_object_offset(struct got_packidx_v2_hdr *packidx, int idx)
280 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
281 uint32_t offset = betoh32(packidx->offsets[idx]);
282 if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
283 uint64_t loffset;
284 idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
285 if (idx < 0 || idx > totobj || packidx->large_offsets == NULL)
286 return -1;
287 loffset = betoh64(packidx->large_offsets[idx]);
288 return (loffset > INT64_MAX ? -1 : (off_t)loffset);
290 return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
293 static int
294 get_object_idx(struct got_packidx_v2_hdr *packidx, struct got_object_id *id)
296 u_int8_t id0 = id->sha1[0];
297 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
298 int i = 0;
300 if (id0 > 0)
301 i = betoh32(packidx->fanout_table[id0 - 1]);
303 while (i < totobj) {
304 struct got_object_id *oid = &packidx->sorted_ids[i];
305 uint32_t offset;
306 int cmp = got_object_id_cmp(id, oid);
308 if (cmp == 0)
309 return i;
310 i++;
313 return -1;
316 static struct got_packidx_v2_hdr *
317 dup_packidx(struct got_packidx_v2_hdr *packidx)
319 struct got_packidx_v2_hdr *p;
320 size_t nobj;
322 p = calloc(1, sizeof(*p));
323 if (p == NULL)
324 return NULL;
326 memcpy(p, packidx, sizeof(*p));
327 p->sorted_ids = NULL;
328 p->crc32 = NULL;
329 p->offsets = NULL;
330 p->large_offsets = NULL;
332 nobj = betoh32(p->fanout_table[0xff]);
334 p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
335 if (p->sorted_ids == NULL)
336 goto err;
337 memcpy(p->sorted_ids, packidx->sorted_ids, nobj * sizeof(*p->sorted_ids));
339 p->crc32 = calloc(nobj, sizeof(*p->crc32));
340 if (p->crc32 == NULL)
341 goto err;
342 memcpy(p->crc32, packidx->crc32, nobj * sizeof(*p->crc32));
344 p->offsets = calloc(nobj, sizeof(*p->offsets));
345 if (p->offsets == NULL)
346 goto err;
347 memcpy(p->offsets, packidx->offsets, nobj * sizeof(*p->offsets));
349 if (p->large_offsets) {
350 p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
351 if (p->large_offsets == NULL)
352 goto err;
353 memcpy(p->large_offsets, packidx->large_offsets,
354 nobj * sizeof(*p->large_offsets));
357 return p;
359 err:
360 free(p->large_offsets);
361 free(p->offsets);
362 free(p->crc32);
363 free(p->sorted_ids);
364 free(p);
365 return NULL;
368 static void
369 cache_packidx(struct got_packidx_v2_hdr *packidx, struct got_repository *repo)
371 struct got_packidx_v2_hdr *p;
372 int i;
374 for (i = 0; i < nitems(repo->packidx_cache); i++) {
375 if (repo->packidx_cache[i] == NULL)
376 break;
379 if (i == nitems(repo->packidx_cache)) {
380 got_packidx_close(repo->packidx_cache[i - 1]);
381 memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
382 sizeof(repo->packidx_cache) -
383 sizeof(repo->packidx_cache[0]));
384 i = 0;
387 repo->packidx_cache[i] = dup_packidx(packidx);
390 static const struct got_error *
391 search_packidx(struct got_packidx_v2_hdr **packidx, int *idx,
392 struct got_repository *repo, struct got_object_id *id)
394 const struct got_error *err;
395 char *path_packdir;
396 DIR *packdir;
397 struct dirent *dent;
398 char *path_packidx;
399 int i;
401 /* Search pack index cache. */
402 for (i = 0; i < nitems(repo->packidx_cache); i++) {
403 if (repo->packidx_cache[i] == NULL)
404 break;
405 *idx = get_object_idx(repo->packidx_cache[i], id);
406 if (*idx != -1) {
407 *packidx = dup_packidx(repo->packidx_cache[i]);
408 if (*packidx == NULL)
409 return got_error(GOT_ERR_NO_MEM);
410 return NULL;
413 /* No luck. Search the filesystem. */
415 path_packdir = got_repo_get_path_objects_pack(repo);
416 if (path_packdir == NULL)
417 return got_error(GOT_ERR_NO_MEM);
419 packdir = opendir(path_packdir);
420 if (packdir == NULL) {
421 err = got_error_from_errno();
422 goto done;
425 while ((dent = readdir(packdir)) != NULL) {
426 if (!is_packidx_filename(dent->d_name, dent->d_namlen))
427 continue;
429 if (asprintf(&path_packidx, "%s/%s", path_packdir,
430 dent->d_name) == -1) {
431 err = got_error(GOT_ERR_NO_MEM);
432 goto done;
435 err = got_packidx_open(packidx, path_packidx);
436 free(path_packidx);
437 if (err)
438 goto done;
440 *idx = get_object_idx(*packidx, id);
441 if (*idx != -1) {
442 err = NULL; /* found the object */
443 cache_packidx(*packidx, repo);
444 goto done;
447 got_packidx_close(*packidx);
448 *packidx = NULL;
451 err = got_error(GOT_ERR_NO_OBJ);
452 done:
453 free(path_packdir);
454 if (packdir && closedir(packdir) != 0 && err == 0)
455 err = got_error_from_errno();
456 return err;
459 static const struct got_error *
460 get_packfile_path(char **path_packfile, struct got_repository *repo,
461 struct got_packidx_v2_hdr *packidx)
463 char *path_packdir;
464 char hex[SHA1_DIGEST_STRING_LENGTH];
465 char *sha1str;
466 char *path_packidx;
468 *path_packfile = NULL;
470 path_packdir = got_repo_get_path_objects_pack(repo);
471 if (path_packdir == NULL)
472 return got_error(GOT_ERR_NO_MEM);
474 sha1str = got_sha1_digest_to_str(packidx->trailer.packfile_sha1,
475 hex, sizeof(hex));
476 if (sha1str == NULL)
477 return got_error(GOT_ERR_PACKIDX_CSUM);
479 if (asprintf(path_packfile, "%s/%s%s%s", path_packdir,
480 GOT_PACK_PREFIX, sha1str, GOT_PACKFILE_SUFFIX) == -1) {
481 *path_packfile = NULL;
482 return got_error(GOT_ERR_NO_MEM);
485 return NULL;
488 const struct got_error *
489 read_packfile_hdr(FILE *f, struct got_packidx_v2_hdr *packidx)
491 const struct got_error *err = NULL;
492 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
493 struct got_packfile_hdr hdr;
494 size_t n;
496 n = fread(&hdr, sizeof(hdr), 1, f);
497 if (n != 1)
498 return got_ferror(f, GOT_ERR_BAD_PACKIDX);
500 if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
501 betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
502 betoh32(hdr.nobjects) != totobj)
503 err = got_error(GOT_ERR_BAD_PACKFILE);
505 return err;
508 static const struct got_error *
509 open_packfile(FILE **packfile, char **path_packfile,
510 struct got_repository *repo, struct got_packidx_v2_hdr *packidx)
512 const struct got_error *err;
514 *packfile = NULL;
516 err = get_packfile_path(path_packfile, repo, packidx);
517 if (err)
518 return err;
520 *packfile = fopen(*path_packfile, "rb");
521 if (*packfile == NULL) {
522 err = got_error_from_errno();
523 free(*path_packfile);
524 return err;
527 err = read_packfile_hdr(*packfile, packidx);
528 if (err) {
529 fclose(*packfile);
530 *packfile = NULL;
532 return err;
535 static const struct got_error *
536 parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
537 FILE *packfile)
539 uint8_t t = 0;
540 uint64_t s = 0;
541 uint8_t sizeN;
542 size_t n;
543 int i = 0;
545 do {
546 /* We do not support size values which don't fit in 64 bit. */
547 if (i > 9)
548 return got_error(GOT_ERR_NO_SPACE);
550 n = fread(&sizeN, sizeof(sizeN), 1, packfile);
551 if (n != 1)
552 return got_ferror(packfile, GOT_ERR_BAD_PACKIDX);
554 if (i == 0) {
555 t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
556 GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
557 s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
558 } else {
559 size_t shift = 4 + 7 * (i - 1);
560 s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
562 i++;
563 } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
565 *type = t;
566 *size = s;
567 *len = i * sizeof(sizeN);
568 return NULL;
571 static const struct got_error *
572 open_plain_object(struct got_object **obj, const char *path_packfile,
573 struct got_object_id *id, uint8_t type, off_t offset, size_t size)
575 *obj = calloc(1, sizeof(**obj));
576 if (*obj == NULL)
577 return got_error(GOT_ERR_NO_MEM);
579 (*obj)->path_packfile = strdup(path_packfile);
580 if ((*obj)->path_packfile == NULL) {
581 free(*obj);
582 *obj = NULL;
583 return got_error(GOT_ERR_NO_MEM);
586 (*obj)->type = type;
587 (*obj)->flags = GOT_OBJ_FLAG_PACKED;
588 (*obj)->hdrlen = 0;
589 (*obj)->size = size;
590 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
591 (*obj)->pack_offset = offset;
593 return NULL;
596 static const struct got_error *
597 parse_negative_offset(int64_t *offset, size_t *len, FILE *packfile)
599 int64_t o = 0;
600 uint8_t offN;
601 size_t n;
602 int i = 0;
604 do {
605 /* We do not support offset values which don't fit in 64 bit. */
606 if (i > 8)
607 return got_error(GOT_ERR_NO_SPACE);
609 n = fread(&offN, sizeof(offN), 1, packfile);
610 if (n != 1)
611 return got_ferror(packfile, GOT_ERR_BAD_PACKIDX);
613 if (i == 0)
614 o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
615 else {
616 o++;
617 o <<= 7;
618 o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
620 i++;
621 } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
623 *offset = o;
624 *len = i * sizeof(offN);
625 return NULL;
628 static const struct got_error *
629 parse_offset_delta(off_t *base_offset, FILE *packfile, off_t offset)
631 const struct got_error *err;
632 int64_t negoffset;
633 size_t negofflen;
635 err = parse_negative_offset(&negoffset, &negofflen, packfile);
636 if (err)
637 return err;
639 /* Compute the base object's offset (must be in the same pack file). */
640 *base_offset = (offset - negoffset);
641 if (*base_offset <= 0)
642 return got_error(GOT_ERR_BAD_PACKFILE);
644 return NULL;
647 static const struct got_error *
648 resolve_delta_chain(struct got_delta_chain *, struct got_repository *,
649 FILE *, const char *, off_t, size_t, int, size_t);
651 static const struct got_error *
652 add_delta(struct got_delta_chain *deltas, const char *path_packfile,
653 off_t delta_offset, size_t tslen, int delta_type, size_t delta_size,
654 size_t delta_data_offset)
656 struct got_delta *delta;
658 delta = got_delta_open(path_packfile, delta_offset, tslen,
659 delta_type, delta_size, delta_data_offset);
660 if (delta == NULL)
661 return got_error(GOT_ERR_NO_MEM);
662 /* delta is freed in got_object_close() */
663 deltas->nentries++;
664 SIMPLEQ_INSERT_HEAD(&deltas->entries, delta, entry);
665 return NULL;
668 static const struct got_error *
669 resolve_offset_delta(struct got_delta_chain *deltas,
670 struct got_repository *repo, FILE *packfile, const char *path_packfile,
671 off_t delta_offset,size_t tslen, int delta_type, size_t delta_size)
674 const struct got_error *err;
675 off_t base_offset;
676 uint8_t base_type;
677 uint64_t base_size;
678 size_t base_tslen;
679 off_t delta_data_offset;
681 err = parse_offset_delta(&base_offset, packfile, delta_offset);
682 if (err)
683 return err;
685 delta_data_offset = ftello(packfile);
686 if (delta_data_offset == -1)
687 return got_error_from_errno();
689 err = add_delta(deltas, path_packfile, delta_offset, tslen,
690 delta_type, delta_size, delta_data_offset);
691 if (err)
692 return err;
694 /* An offset delta must be in the same packfile. */
695 if (fseeko(packfile, base_offset, SEEK_SET) != 0)
696 return got_error_from_errno();
698 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
699 packfile);
700 if (err)
701 return err;
703 return resolve_delta_chain(deltas, repo, packfile, path_packfile,
704 base_offset, base_tslen, base_type, base_size);
707 static const struct got_error *
708 resolve_ref_delta(struct got_delta_chain *deltas, struct got_repository *repo,
709 FILE *packfile, const char *path_packfile, off_t delta_offset,
710 size_t tslen, int delta_type, size_t delta_size)
712 const struct got_error *err;
713 struct got_object_id id;
714 struct got_packidx_v2_hdr *packidx;
715 int idx;
716 off_t base_offset;
717 uint8_t base_type;
718 uint64_t base_size;
719 size_t base_tslen;
720 size_t n;
721 FILE *base_packfile;
722 char *path_base_packfile;
723 off_t delta_data_offset;
725 n = fread(&id, sizeof(id), 1, packfile);
726 if (n != 1)
727 return got_ferror(packfile, GOT_ERR_IO);
729 delta_data_offset = ftello(packfile);
730 if (delta_data_offset == -1)
731 return got_error_from_errno();
733 err = add_delta(deltas, path_packfile, delta_offset, tslen,
734 delta_type, delta_size, delta_data_offset);
735 if (err)
736 return err;
738 err = search_packidx(&packidx, &idx, repo, &id);
739 if (err)
740 return err;
742 base_offset = get_object_offset(packidx, idx);
743 if (base_offset == (uint64_t)-1) {
744 got_packidx_close(packidx);
745 return got_error(GOT_ERR_BAD_PACKIDX);
748 err = open_packfile(&base_packfile, &path_base_packfile, repo, packidx);
749 got_packidx_close(packidx);
750 if (err)
751 return err;
753 if (fseeko(base_packfile, base_offset, SEEK_SET) != 0) {
754 err = got_error_from_errno();
755 goto done;
758 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
759 base_packfile);
760 if (err)
761 goto done;
763 err = resolve_delta_chain(deltas, repo, base_packfile,
764 path_base_packfile, base_offset, base_tslen, base_type,
765 base_size);
766 done:
767 free(path_base_packfile);
768 if (base_packfile && fclose(base_packfile) == -1 && err == 0)
769 err = got_error_from_errno();
770 return err;
773 static const struct got_error *
774 resolve_delta_chain(struct got_delta_chain *deltas, struct got_repository *repo,
775 FILE *packfile, const char *path_packfile, off_t delta_offset, size_t tslen,
776 int delta_type, size_t delta_size)
778 const struct got_error *err = NULL;
780 switch (delta_type) {
781 case GOT_OBJ_TYPE_COMMIT:
782 case GOT_OBJ_TYPE_TREE:
783 case GOT_OBJ_TYPE_BLOB:
784 case GOT_OBJ_TYPE_TAG:
785 /* Plain types are the final delta base. Recursion ends. */
786 err = add_delta(deltas, path_packfile, delta_offset, tslen,
787 delta_type, delta_size, 0);
788 break;
789 case GOT_OBJ_TYPE_OFFSET_DELTA:
790 err = resolve_offset_delta(deltas, repo, packfile,
791 path_packfile, delta_offset, tslen, delta_type,
792 delta_size);
793 break;
794 case GOT_OBJ_TYPE_REF_DELTA:
795 err = resolve_ref_delta(deltas, repo, packfile,
796 path_packfile, delta_offset, tslen, delta_type,
797 delta_size);
798 break;
799 default:
800 return got_error(GOT_ERR_NOT_IMPL);
803 return err;
806 static const struct got_error *
807 open_delta_object(struct got_object **obj, struct got_repository *repo,
808 struct got_packidx_v2_hdr *packidx, const char *path_packfile,
809 FILE *packfile, struct got_object_id *id, off_t offset, size_t tslen,
810 int delta_type, size_t delta_size)
812 const struct got_error *err = NULL;
813 struct got_object_id base_id;
814 uint8_t base_type;
815 int resolved_type;
816 uint64_t base_size;
817 size_t base_tslen;
819 *obj = calloc(1, sizeof(**obj));
820 if (*obj == NULL)
821 return got_error(GOT_ERR_NO_MEM);
823 (*obj)->flags = 0;
824 (*obj)->hdrlen = 0;
825 (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
826 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
827 (*obj)->pack_offset = offset + tslen;
829 (*obj)->path_packfile = strdup(path_packfile);
830 if ((*obj)->path_packfile == NULL) {
831 err = got_error(GOT_ERR_NO_MEM);
832 goto done;
834 (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
836 SIMPLEQ_INIT(&(*obj)->deltas.entries);
837 (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
839 err = resolve_delta_chain(&(*obj)->deltas, repo, packfile,
840 path_packfile, offset, tslen, delta_type, delta_size);
841 if (err)
842 goto done;
844 err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
845 if (err)
846 goto done;
847 (*obj)->type = resolved_type;
849 done:
850 if (err) {
851 got_object_close(*obj);
852 *obj = NULL;
854 return err;
857 static const struct got_error *
858 open_packed_object(struct got_object **obj, struct got_repository *repo,
859 struct got_packidx_v2_hdr *packidx, int idx, struct got_object_id *id)
861 const struct got_error *err = NULL;
862 off_t offset;
863 char *path_packfile;
864 FILE *packfile;
865 uint8_t type;
866 uint64_t size;
867 size_t tslen;
869 *obj = NULL;
871 offset = get_object_offset(packidx, idx);
872 if (offset == (uint64_t)-1)
873 return got_error(GOT_ERR_BAD_PACKIDX);
875 err = open_packfile(&packfile, &path_packfile, repo, packidx);
876 if (err)
877 return err;
879 if (fseeko(packfile, offset, SEEK_SET) != 0) {
880 err = got_error_from_errno();
881 goto done;
884 err = parse_object_type_and_size(&type, &size, &tslen, packfile);
885 if (err)
886 goto done;
888 switch (type) {
889 case GOT_OBJ_TYPE_COMMIT:
890 case GOT_OBJ_TYPE_TREE:
891 case GOT_OBJ_TYPE_BLOB:
892 case GOT_OBJ_TYPE_TAG:
893 err = open_plain_object(obj, path_packfile, id, type,
894 offset + tslen, size);
895 break;
897 case GOT_OBJ_TYPE_OFFSET_DELTA:
898 case GOT_OBJ_TYPE_REF_DELTA:
899 err = open_delta_object(obj, repo, packidx, path_packfile,
900 packfile, id, offset, tslen, type, size);
901 break;
903 default:
904 err = got_error(GOT_ERR_NOT_IMPL);
905 goto done;
907 done:
908 free(path_packfile);
909 if (packfile && fclose(packfile) == -1 && err == 0)
910 err = got_error_from_errno();
911 return err;
914 const struct got_error *
915 got_packfile_open_object(struct got_object **obj, struct got_object_id *id,
916 struct got_repository *repo)
918 const struct got_error *err = NULL;
919 struct got_packidx_v2_hdr *packidx = NULL;
920 int idx;
922 err = search_packidx(&packidx, &idx, repo, id);
923 if (err)
924 return err;
926 err = open_packed_object(obj, repo, packidx, idx, id);
927 got_packidx_close(packidx);
928 return err;
931 static const struct got_error *
932 dump_delta_chain(struct got_delta_chain *deltas, FILE *outfile)
934 const struct got_error *err = NULL;
935 struct got_delta *delta;
936 FILE *base_file, *accum_file;
937 int n = 0;
939 if (SIMPLEQ_EMPTY(&deltas->entries))
940 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
942 base_file = got_opentemp();
943 if (base_file == NULL)
944 return got_error_from_errno();
946 accum_file = got_opentemp();
947 if (accum_file == NULL) {
948 err = got_error_from_errno();
949 fclose(base_file);
950 return err;
953 /* Deltas are ordered in ascending order. */
954 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
955 uint8_t *delta_buf = NULL;
956 size_t delta_len = 0;
957 FILE *delta_file;
959 delta_file = fopen(delta->path_packfile, "rb");
960 if (delta_file == NULL) {
961 err = got_error_from_errno();
962 goto done;
966 if (n == 0) {
967 /* Plain object types are the delta base. */
968 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
969 delta->type != GOT_OBJ_TYPE_TREE &&
970 delta->type != GOT_OBJ_TYPE_BLOB &&
971 delta->type != GOT_OBJ_TYPE_TAG) {
972 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
973 goto done;
976 if (fseeko(delta_file, delta->offset + delta->tslen,
977 SEEK_SET) != 0) {
978 fclose(delta_file);
979 err = got_error_from_errno();
980 goto done;
982 err = got_inflate_to_file(&delta_len, delta_file,
983 base_file);
984 fclose(delta_file);
985 if (err)
986 goto done;
987 n++;
988 rewind(base_file);
989 continue;
992 if (fseeko(delta_file, delta->data_offset, SEEK_CUR) != 0) {
993 fclose(delta_file);
994 err = got_error_from_errno();
995 goto done;
998 /* Delta streams should always fit in memory. */
999 err = got_inflate_to_mem(&delta_buf, &delta_len, delta_file);
1000 fclose(delta_file);
1001 if (err)
1002 goto done;
1004 err = got_delta_apply(base_file, delta_buf, delta_len,
1005 /* Final delta application writes to the output file. */
1006 ++n < deltas->nentries ? accum_file : outfile);
1007 free(delta_buf);
1008 if (err)
1009 goto done;
1011 if (n < deltas->nentries) {
1012 /* Accumulated delta becomes the new base. */
1013 FILE *tmp = accum_file;
1014 accum_file = base_file;
1015 base_file = tmp;
1016 rewind(base_file);
1017 rewind(accum_file);
1021 done:
1022 fclose(base_file);
1023 fclose(accum_file);
1024 rewind(outfile);
1025 return err;
1028 const struct got_error *
1029 got_packfile_extract_object(FILE **f, struct got_object *obj,
1030 struct got_repository *repo)
1032 const struct got_error *err = NULL;
1033 FILE *packfile = NULL;
1035 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1036 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1038 *f = got_opentemp();
1039 if (*f == NULL) {
1040 err = got_error(GOT_ERR_FILE_OPEN);
1041 goto done;
1044 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1045 packfile = fopen(obj->path_packfile, "rb");
1046 if (packfile == NULL) {
1047 err = got_error_from_errno();
1048 goto done;
1051 if (fseeko(packfile, obj->pack_offset, SEEK_SET) != 0) {
1052 err = got_error_from_errno();
1053 goto done;
1056 err = got_inflate_to_file(&obj->size, packfile, *f);
1057 } else
1058 err = dump_delta_chain(&obj->deltas, *f);
1059 done:
1060 if (packfile)
1061 fclose(packfile);
1062 if (err && *f)
1063 fclose(*f);
1064 return err;