2 * Copyright (c) 2018 Stefan Sperling <stsp@openbsd.org>
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.
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.
17 #include <sys/types.h>
19 #include <sys/queue.h>
33 #include "got_error.h"
34 #include "got_object.h"
35 #include "got_repository.h"
37 #include "got_lib_sha1.h"
38 #include "got_lib_pack.h"
39 #include "got_lib_path.h"
40 #include "got_lib_delta.h"
41 #include "got_lib_zbuf.h"
42 #include "got_lib_object.h"
43 #include "got_lib_repository.h"
46 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
49 #define GOT_PACK_PREFIX "pack-"
50 #define GOT_PACKFILE_SUFFIX ".pack"
51 #define GOT_PACKIDX_SUFFIX ".idx"
52 #define GOT_PACKFILE_NAMELEN (strlen(GOT_PACK_PREFIX) + \
53 SHA1_DIGEST_STRING_LENGTH - 1 + \
54 strlen(GOT_PACKFILE_SUFFIX))
55 #define GOT_PACKIDX_NAMELEN (strlen(GOT_PACK_PREFIX) + \
56 SHA1_DIGEST_STRING_LENGTH - 1 + \
57 strlen(GOT_PACKIDX_SUFFIX))
60 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
63 static const struct got_error *
64 verify_fanout_table(uint32_t *fanout_table)
68 for (i = 0; i < 0xff - 1; i++) {
69 if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
70 return got_error(GOT_ERR_BAD_PACKIDX);
76 static const struct got_error *
77 get_packfile_size(size_t *size, const char *path)
84 dot = strrchr(path, '.');
86 return got_error(GOT_ERR_BAD_PATH);
88 /* Path must point to a pack index or to a pack file. */
89 if (strcmp(dot, ".idx") == 0) {
90 const struct got_error *err = NULL;
92 char base_path[PATH_MAX];
94 /* Convert pack index path to pack file path. */
95 if (strlcpy(base_path, path, PATH_MAX) > PATH_MAX)
96 return got_error(GOT_ERR_NO_SPACE);
97 dot = strrchr(base_path, '.');
99 return got_error(GOT_ERR_BAD_PATH);
101 if (asprintf(&path_pack, "%s.pack", base_path) == -1)
102 return got_error_from_errno();
104 if (stat(path_pack, &sb) != 0)
105 err = got_error_from_errno();
109 } else if (strcmp(dot, ".pack") == 0) {
110 if (stat(path, &sb) != 0)
111 return got_error_from_errno();
113 return got_error(GOT_ERR_BAD_PATH);
119 const struct got_error *
120 got_packidx_open(struct got_packidx_v2_hdr **packidx, const char *path)
122 struct got_packidx_v2_hdr *p;
124 const struct got_error *err = NULL;
125 size_t n, nobj, packfile_size;
127 uint8_t sha1[SHA1_DIGEST_LENGTH];
131 f = fopen(path, "rb");
133 return got_error_from_errno();
135 err = get_packfile_size(&packfile_size, path);
139 p = calloc(1, sizeof(*p));
141 err = got_error_from_errno();
145 n = fread(&p->magic, sizeof(p->magic), 1, f);
147 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
151 if (betoh32(p->magic) != GOT_PACKIDX_V2_MAGIC) {
152 err = got_error(GOT_ERR_BAD_PACKIDX);
156 SHA1Update(&ctx, (uint8_t *)&p->magic, sizeof(p->magic));
158 n = fread(&p->version, sizeof(p->version), 1, f);
160 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
164 if (betoh32(p->version) != GOT_PACKIDX_VERSION) {
165 err = got_error(GOT_ERR_BAD_PACKIDX);
169 SHA1Update(&ctx, (uint8_t *)&p->version, sizeof(p->version));
171 n = fread(&p->fanout_table, sizeof(p->fanout_table), 1, f);
173 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
177 err = verify_fanout_table(p->fanout_table);
181 SHA1Update(&ctx, (uint8_t *)p->fanout_table, sizeof(p->fanout_table));
183 nobj = betoh32(p->fanout_table[0xff]);
185 p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
186 if (p->sorted_ids == NULL) {
187 err = got_error_from_errno();
191 n = fread(p->sorted_ids, sizeof(*p->sorted_ids), nobj, f);
193 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
197 SHA1Update(&ctx, (uint8_t *)p->sorted_ids,
198 nobj * sizeof(*p->sorted_ids));
200 p->crc32 = calloc(nobj, sizeof(*p->crc32));
201 if (p->crc32 == NULL) {
202 err = got_error_from_errno();
206 n = fread(p->crc32, sizeof(*p->crc32), nobj, f);
208 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
212 SHA1Update(&ctx, (uint8_t *)p->crc32, nobj * sizeof(*p->crc32));
214 p->offsets = calloc(nobj, sizeof(*p->offsets));
215 if (p->offsets == NULL) {
216 err = got_error_from_errno();
220 n = fread(p->offsets, sizeof(*p->offsets), nobj, f);
222 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
226 SHA1Update(&ctx, (uint8_t *)p->offsets, nobj * sizeof(*p->offsets));
228 /* Large file offsets are contained only in files > 2GB. */
229 if (packfile_size <= 0x80000000)
232 p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
233 if (p->large_offsets == NULL) {
234 err = got_error_from_errno();
238 n = fread(p->large_offsets, sizeof(*p->large_offsets), nobj, f);
240 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
244 SHA1Update(&ctx, (uint8_t*)p->large_offsets,
245 nobj * sizeof(*p->large_offsets));
248 n = fread(&p->trailer, sizeof(p->trailer), 1, f);
250 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
254 SHA1Update(&ctx, p->trailer.packfile_sha1, SHA1_DIGEST_LENGTH);
255 SHA1Final(sha1, &ctx);
256 if (memcmp(p->trailer.packidx_sha1, sha1, SHA1_DIGEST_LENGTH) != 0)
257 err = got_error(GOT_ERR_PACKIDX_CSUM);
261 got_packidx_close(p);
268 got_packidx_close(struct got_packidx_v2_hdr *packidx)
270 free(packidx->sorted_ids);
271 free(packidx->offsets);
272 free(packidx->crc32);
273 free(packidx->large_offsets);
278 is_packidx_filename(const char *name, size_t len)
280 if (len != GOT_PACKIDX_NAMELEN)
283 if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
286 if (strcmp(name + strlen(GOT_PACK_PREFIX) +
287 SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
294 get_object_offset(struct got_packidx_v2_hdr *packidx, int idx)
296 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
297 uint32_t offset = betoh32(packidx->offsets[idx]);
298 if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
300 idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
301 if (idx < 0 || idx > totobj || packidx->large_offsets == NULL)
303 loffset = betoh64(packidx->large_offsets[idx]);
304 return (loffset > INT64_MAX ? -1 : (off_t)loffset);
306 return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
309 static const struct got_error *
310 get_packfile_path(char **path_packfile, struct got_repository *repo,
311 struct got_packidx_v2_hdr *packidx);
314 get_object_idx(struct got_packidx_v2_hdr *packidx, struct got_object_id *id, struct got_repository *repo)
316 u_int8_t id0 = id->sha1[0];
317 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
321 i = betoh32(packidx->fanout_table[id0 - 1]);
324 struct got_object_id *oid = &packidx->sorted_ids[i];
325 int cmp = got_object_id_cmp(id, oid);
335 static struct got_packidx_v2_hdr *
336 dup_packidx(struct got_packidx_v2_hdr *packidx)
338 struct got_packidx_v2_hdr *p;
341 p = calloc(1, sizeof(*p));
345 memcpy(p, packidx, sizeof(*p));
346 p->sorted_ids = NULL;
349 p->large_offsets = NULL;
351 nobj = betoh32(p->fanout_table[0xff]);
353 p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
354 if (p->sorted_ids == NULL)
356 memcpy(p->sorted_ids, packidx->sorted_ids,
357 nobj * sizeof(*p->sorted_ids));
359 p->crc32 = calloc(nobj, sizeof(*p->crc32));
360 if (p->crc32 == NULL)
362 memcpy(p->crc32, packidx->crc32, nobj * sizeof(*p->crc32));
364 p->offsets = calloc(nobj, sizeof(*p->offsets));
365 if (p->offsets == NULL)
367 memcpy(p->offsets, packidx->offsets, nobj * sizeof(*p->offsets));
369 if (p->large_offsets) {
370 p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
371 if (p->large_offsets == NULL)
373 memcpy(p->large_offsets, packidx->large_offsets,
374 nobj * sizeof(*p->large_offsets));
380 free(p->large_offsets);
389 cache_packidx(struct got_packidx_v2_hdr *packidx, struct got_repository *repo)
393 for (i = 0; i < nitems(repo->packidx_cache); i++) {
394 if (repo->packidx_cache[i] == NULL)
398 if (i == nitems(repo->packidx_cache)) {
399 got_packidx_close(repo->packidx_cache[i - 1]);
400 memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
401 sizeof(repo->packidx_cache) -
402 sizeof(repo->packidx_cache[0]));
406 repo->packidx_cache[i] = dup_packidx(packidx);
409 static const struct got_error *
410 search_packidx(struct got_packidx_v2_hdr **packidx, int *idx,
411 struct got_repository *repo, struct got_object_id *id)
413 const struct got_error *err;
420 /* Search pack index cache. */
421 for (i = 0; i < nitems(repo->packidx_cache); i++) {
422 if (repo->packidx_cache[i] == NULL)
424 *idx = get_object_idx(repo->packidx_cache[i], id, repo);
426 *packidx = repo->packidx_cache[i];
430 /* No luck. Search the filesystem. */
432 path_packdir = got_repo_get_path_objects_pack(repo);
433 if (path_packdir == NULL)
434 return got_error_from_errno();
436 packdir = opendir(path_packdir);
437 if (packdir == NULL) {
438 err = got_error_from_errno();
442 while ((dent = readdir(packdir)) != NULL) {
443 if (!is_packidx_filename(dent->d_name, dent->d_namlen))
446 if (asprintf(&path_packidx, "%s/%s", path_packdir,
447 dent->d_name) == -1) {
448 err = got_error_from_errno();
452 err = got_packidx_open(packidx, path_packidx);
457 *idx = get_object_idx(*packidx, id, repo);
459 err = NULL; /* found the object */
460 cache_packidx(*packidx, repo);
464 got_packidx_close(*packidx);
468 err = got_error(GOT_ERR_NO_OBJ);
471 if (packdir && closedir(packdir) != 0 && err == 0)
472 err = got_error_from_errno();
476 static const struct got_error *
477 get_packfile_path(char **path_packfile, struct got_repository *repo,
478 struct got_packidx_v2_hdr *packidx)
481 char hex[SHA1_DIGEST_STRING_LENGTH];
484 path_packdir = got_repo_get_path_objects_pack(repo);
485 if (path_packdir == NULL)
486 return got_error_from_errno();
488 sha1str = got_sha1_digest_to_str(packidx->trailer.packfile_sha1,
491 return got_error(GOT_ERR_PACKIDX_CSUM);
493 if (asprintf(path_packfile, "%s/%s%s%s", path_packdir,
494 GOT_PACK_PREFIX, sha1str, GOT_PACKFILE_SUFFIX) == -1) {
495 *path_packfile = NULL;
496 return got_error_from_errno();
502 const struct got_error *
503 read_packfile_hdr(FILE *f, struct got_packidx_v2_hdr *packidx)
505 const struct got_error *err = NULL;
506 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
507 struct got_packfile_hdr hdr;
510 n = fread(&hdr, sizeof(hdr), 1, f);
512 return got_ferror(f, GOT_ERR_BAD_PACKIDX);
514 if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
515 betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
516 betoh32(hdr.nobjects) != totobj)
517 err = got_error(GOT_ERR_BAD_PACKFILE);
522 static const struct got_error *
523 open_packfile(FILE **packfile, const char *path_packfile,
524 struct got_repository *repo, struct got_packidx_v2_hdr *packidx)
526 const struct got_error *err = NULL;
530 *packfile = fopen(path_packfile, "rb");
531 if (*packfile == NULL) {
532 err = got_error_from_errno();
537 err = read_packfile_hdr(*packfile, packidx);
547 got_pack_close(struct got_pack *pack)
549 fclose(pack->packfile);
550 pack->packfile = NULL;
551 free(pack->path_packfile);
552 pack->path_packfile = NULL;
556 static const struct got_error *
557 cache_pack(struct got_pack **packp, const char *path_packfile,
558 struct got_packidx_v2_hdr *packidx, struct got_repository *repo)
560 const struct got_error *err = NULL;
561 struct got_pack *pack = NULL;
567 for (i = 0; i < nitems(repo->packs); i++) {
568 pack = &repo->packs[i];
569 if (pack->path_packfile == NULL)
571 if (strcmp(pack->path_packfile, path_packfile) == 0)
575 if (i == nitems(repo->packs) - 1) {
576 got_pack_close(&repo->packs[i - 1]);
577 memmove(&repo->packs[1], &repo->packs[0],
578 sizeof(repo->packs) - sizeof(repo->packs[0]));
582 pack = &repo->packs[i];
584 pack->path_packfile = strdup(path_packfile);
585 if (pack->path_packfile == NULL) {
586 err = got_error_from_errno();
590 err = open_packfile(&pack->packfile, path_packfile, repo, packidx);
594 err = get_packfile_size(&pack->filesize, path_packfile);
598 free(pack->path_packfile);
606 get_cached_pack(const char *path_packfile, struct got_repository *repo)
608 struct got_pack *pack = NULL;
611 for (i = 0; i < nitems(repo->packs); i++) {
612 pack = &repo->packs[i];
613 if (pack->path_packfile == NULL)
615 if (strcmp(pack->path_packfile, path_packfile) == 0)
622 static const struct got_error *
623 parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
633 /* We do not support size values which don't fit in 64 bit. */
635 return got_error(GOT_ERR_NO_SPACE);
637 n = fread(&sizeN, sizeof(sizeN), 1, packfile);
639 return got_ferror(packfile, GOT_ERR_BAD_PACKIDX);
642 t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
643 GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
644 s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
646 size_t shift = 4 + 7 * (i - 1);
647 s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
650 } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
654 *len = i * sizeof(sizeN);
658 static const struct got_error *
659 open_plain_object(struct got_object **obj, const char *path_packfile,
660 struct got_object_id *id, uint8_t type, off_t offset, size_t size)
662 *obj = calloc(1, sizeof(**obj));
664 return got_error_from_errno();
666 (*obj)->path_packfile = strdup(path_packfile);
667 if ((*obj)->path_packfile == NULL) {
668 const struct got_error *err = got_error_from_errno();
675 (*obj)->flags = GOT_OBJ_FLAG_PACKED;
678 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
679 (*obj)->pack_offset = offset;
684 static const struct got_error *
685 parse_negative_offset(int64_t *offset, size_t *len, FILE *packfile)
693 /* We do not support offset values which don't fit in 64 bit. */
695 return got_error(GOT_ERR_NO_SPACE);
697 n = fread(&offN, sizeof(offN), 1, packfile);
699 return got_ferror(packfile, GOT_ERR_BAD_PACKIDX);
702 o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
706 o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
709 } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
712 *len = i * sizeof(offN);
716 static const struct got_error *
717 parse_offset_delta(off_t *base_offset, FILE *packfile, off_t offset)
719 const struct got_error *err;
723 err = parse_negative_offset(&negoffset, &negofflen, packfile);
727 /* Compute the base object's offset (must be in the same pack file). */
728 *base_offset = (offset - negoffset);
729 if (*base_offset <= 0)
730 return got_error(GOT_ERR_BAD_PACKFILE);
735 static const struct got_error *
736 resolve_delta_chain(struct got_delta_chain *, struct got_repository *,
737 FILE *, const char *, off_t, size_t, int, size_t, unsigned int);
739 static const struct got_error *
740 add_delta(struct got_delta_chain *deltas, const char *path_packfile,
741 off_t delta_offset, size_t tslen, int delta_type, size_t delta_size,
742 size_t delta_data_offset, uint8_t *delta_buf, size_t delta_len)
744 struct got_delta *delta;
746 delta = got_delta_open(path_packfile, delta_offset, tslen,
747 delta_type, delta_size, delta_data_offset, delta_buf,
750 return got_error_from_errno();
751 /* delta is freed in got_object_close() */
753 SIMPLEQ_INSERT_HEAD(&deltas->entries, delta, entry);
757 static const struct got_error *
758 resolve_offset_delta(struct got_delta_chain *deltas,
759 struct got_repository *repo, FILE *packfile, const char *path_packfile,
760 off_t delta_offset,size_t tslen, int delta_type, size_t delta_size,
761 unsigned int recursion)
764 const struct got_error *err;
769 off_t delta_data_offset;
773 err = parse_offset_delta(&base_offset, packfile, delta_offset);
777 delta_data_offset = ftello(packfile);
778 if (delta_data_offset == -1)
779 return got_error_from_errno();
781 err = got_inflate_to_mem(&delta_buf, &delta_len, packfile);
785 err = add_delta(deltas, path_packfile, delta_offset, tslen,
786 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
790 /* An offset delta must be in the same packfile. */
791 if (fseeko(packfile, base_offset, SEEK_SET) != 0)
792 return got_error_from_errno();
794 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
799 return resolve_delta_chain(deltas, repo, packfile, path_packfile,
800 base_offset, base_tslen, base_type, base_size, recursion - 1);
803 static const struct got_error *
804 resolve_ref_delta(struct got_delta_chain *deltas, struct got_repository *repo,
805 FILE *packfile, const char *path_packfile, off_t delta_offset,
806 size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
808 const struct got_error *err;
809 struct got_object_id id;
810 struct got_packidx_v2_hdr *packidx;
817 char *path_base_packfile;
818 struct got_pack *base_pack;
819 off_t delta_data_offset;
823 n = fread(&id, sizeof(id), 1, packfile);
825 return got_ferror(packfile, GOT_ERR_IO);
827 delta_data_offset = ftello(packfile);
828 if (delta_data_offset == -1)
829 return got_error_from_errno();
831 err = got_inflate_to_mem(&delta_buf, &delta_len, packfile);
835 err = add_delta(deltas, path_packfile, delta_offset, tslen,
836 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
840 err = search_packidx(&packidx, &idx, repo, &id);
844 base_offset = get_object_offset(packidx, idx);
845 if (base_offset == (uint64_t)-1) {
846 return got_error(GOT_ERR_BAD_PACKIDX);
849 err = get_packfile_path(&path_base_packfile, repo, packidx);
853 base_pack = get_cached_pack(path_base_packfile, repo);
854 if (base_pack == NULL) {
855 err = cache_pack(&base_pack, path_base_packfile, NULL, repo);
860 if (fseeko(base_pack->packfile, base_offset, SEEK_SET) != 0) {
861 err = got_error_from_errno();
865 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
866 base_pack->packfile);
870 err = resolve_delta_chain(deltas, repo, base_pack->packfile,
871 path_base_packfile, base_offset, base_tslen, base_type,
872 base_size, recursion - 1);
874 free(path_base_packfile);
878 static const struct got_error *
879 resolve_delta_chain(struct got_delta_chain *deltas, struct got_repository *repo,
880 FILE *packfile, const char *path_packfile, off_t delta_offset, size_t tslen,
881 int delta_type, size_t delta_size, unsigned int recursion)
883 const struct got_error *err = NULL;
885 if (--recursion == 0)
886 return got_error(GOT_ERR_RECURSION);
888 switch (delta_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 /* Plain types are the final delta base. Recursion ends. */
894 err = add_delta(deltas, path_packfile, delta_offset, tslen,
895 delta_type, delta_size, 0, NULL, 0);
897 case GOT_OBJ_TYPE_OFFSET_DELTA:
898 err = resolve_offset_delta(deltas, repo, packfile,
899 path_packfile, delta_offset, tslen, delta_type,
900 delta_size, recursion - 1);
902 case GOT_OBJ_TYPE_REF_DELTA:
903 err = resolve_ref_delta(deltas, repo, packfile,
904 path_packfile, delta_offset, tslen, delta_type,
905 delta_size, recursion - 1);
908 return got_error(GOT_ERR_OBJ_TYPE);
914 static const struct got_error *
915 open_delta_object(struct got_object **obj, struct got_repository *repo,
916 struct got_packidx_v2_hdr *packidx, const char *path_packfile,
917 FILE *packfile, struct got_object_id *id, off_t offset, size_t tslen,
918 int delta_type, size_t delta_size)
920 const struct got_error *err = NULL;
923 *obj = calloc(1, sizeof(**obj));
925 return got_error_from_errno();
929 (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
930 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
931 (*obj)->pack_offset = offset + tslen;
933 (*obj)->path_packfile = strdup(path_packfile);
934 if ((*obj)->path_packfile == NULL) {
935 err = got_error_from_errno();
938 (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
940 SIMPLEQ_INIT(&(*obj)->deltas.entries);
941 (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
943 err = resolve_delta_chain(&(*obj)->deltas, repo, packfile,
944 path_packfile, offset, tslen, delta_type, delta_size,
945 GOT_DELTA_CHAIN_RECURSION_MAX);
949 err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
952 (*obj)->type = resolved_type;
956 got_object_close(*obj);
962 static const struct got_error *
963 open_packed_object(struct got_object **obj, struct got_repository *repo,
964 struct got_packidx_v2_hdr *packidx, int idx, struct got_object_id *id)
966 const struct got_error *err = NULL;
969 struct got_pack *pack;
976 offset = get_object_offset(packidx, idx);
977 if (offset == (uint64_t)-1)
978 return got_error(GOT_ERR_BAD_PACKIDX);
980 err = get_packfile_path(&path_packfile, repo, packidx);
984 pack = get_cached_pack(path_packfile, repo);
986 err = cache_pack(&pack, path_packfile, packidx, repo);
991 if (fseeko(pack->packfile, offset, SEEK_SET) != 0) {
992 err = got_error_from_errno();
996 err = parse_object_type_and_size(&type, &size, &tslen, pack->packfile);
1001 case GOT_OBJ_TYPE_COMMIT:
1002 case GOT_OBJ_TYPE_TREE:
1003 case GOT_OBJ_TYPE_BLOB:
1004 case GOT_OBJ_TYPE_TAG:
1005 err = open_plain_object(obj, path_packfile, id, type,
1006 offset + tslen, size);
1009 case GOT_OBJ_TYPE_OFFSET_DELTA:
1010 case GOT_OBJ_TYPE_REF_DELTA:
1011 err = open_delta_object(obj, repo, packidx, path_packfile,
1012 pack->packfile, id, offset, tslen, type, size);
1016 err = got_error(GOT_ERR_OBJ_TYPE);
1020 free(path_packfile);
1024 const struct got_error *
1025 got_packfile_open_object(struct got_object **obj, struct got_object_id *id,
1026 struct got_repository *repo)
1028 const struct got_error *err = NULL;
1029 struct got_packidx_v2_hdr *packidx = NULL;
1032 err = search_packidx(&packidx, &idx, repo, id);
1036 err = open_packed_object(obj, repo, packidx, idx, id);
1040 err = cache_pack(NULL, (*obj)->path_packfile, packidx, repo);
1044 static const struct got_error *
1045 get_delta_chain_max_size(uint64_t *max_size, struct got_delta_chain *deltas)
1047 struct got_delta *delta;
1048 uint64_t base_size = 0, result_size = 0;
1051 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1052 /* Plain object types are the delta base. */
1053 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1054 delta->type != GOT_OBJ_TYPE_TREE &&
1055 delta->type != GOT_OBJ_TYPE_BLOB &&
1056 delta->type != GOT_OBJ_TYPE_TAG) {
1057 const struct got_error *err;
1058 err = got_delta_get_sizes(&base_size, &result_size,
1059 delta->delta_buf, delta->delta_len);
1063 base_size = delta->size;
1064 if (base_size > *max_size)
1065 *max_size = base_size;
1066 if (result_size > *max_size)
1067 *max_size = result_size;
1073 static const struct got_error *
1074 dump_delta_chain_to_file(size_t *result_size, struct got_delta_chain *deltas,
1075 FILE *outfile, struct got_repository *repo)
1077 const struct got_error *err = NULL;
1078 struct got_delta *delta;
1079 FILE *base_file = NULL, *accum_file = NULL;
1080 uint8_t *base_buf = NULL, *accum_buf = NULL;
1081 size_t accum_size = 0;
1087 if (SIMPLEQ_EMPTY(&deltas->entries))
1088 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1090 /* We process small enough files entirely in memory for speed. */
1091 err = get_delta_chain_max_size(&max_size, deltas);
1094 if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1095 accum_buf = malloc(max_size);
1096 if (accum_buf == NULL)
1097 return got_error_from_errno();
1099 base_file = got_opentemp();
1100 if (base_file == NULL)
1101 return got_error_from_errno();
1103 accum_file = got_opentemp();
1104 if (accum_file == NULL) {
1105 err = got_error_from_errno();
1111 /* Deltas are ordered in ascending order. */
1112 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1114 struct got_pack *pack;
1117 /* Plain object types are the delta base. */
1118 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1119 delta->type != GOT_OBJ_TYPE_TREE &&
1120 delta->type != GOT_OBJ_TYPE_BLOB &&
1121 delta->type != GOT_OBJ_TYPE_TAG) {
1122 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1126 pack = get_cached_pack(delta->path_packfile, repo);
1128 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1132 if (fseeko(pack->packfile, delta->offset + delta->tslen,
1134 err = got_error_from_errno();
1138 err = got_inflate_to_file(&base_len,
1139 pack->packfile, base_file);
1141 err = got_inflate_to_mem(&base_buf, &base_len,
1143 if (base_len < max_size) {
1145 p = reallocarray(base_buf, 1, max_size);
1147 err = got_error_from_errno();
1162 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1163 delta->delta_len, accum_buf, &accum_size);
1166 err = got_delta_apply(base_file, delta->delta_buf,
1168 /* Final delta application writes to output file. */
1169 ++n < deltas->nentries ? accum_file : outfile,
1175 if (n < deltas->nentries) {
1176 /* Accumulated delta becomes the new base. */
1178 uint8_t *tmp = accum_buf;
1179 accum_buf = base_buf;
1182 FILE *tmp = accum_file;
1183 accum_file = base_file;
1194 size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1196 if (len != accum_size)
1197 return got_ferror(outfile, GOT_ERR_IO);
1205 *result_size = accum_size;
1209 static const struct got_error *
1210 dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1211 struct got_delta_chain *deltas, struct got_repository *repo)
1213 const struct got_error *err = NULL;
1214 struct got_delta *delta;
1215 uint8_t *base_buf = NULL, *accum_buf = NULL;
1223 if (SIMPLEQ_EMPTY(&deltas->entries))
1224 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1226 err = get_delta_chain_max_size(&max_size, deltas);
1229 accum_buf = malloc(max_size);
1230 if (accum_buf == NULL)
1231 return got_error_from_errno();
1233 /* Deltas are ordered in ascending order. */
1234 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1236 struct got_pack *pack;
1239 /* Plain object types are the delta base. */
1240 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1241 delta->type != GOT_OBJ_TYPE_TREE &&
1242 delta->type != GOT_OBJ_TYPE_BLOB &&
1243 delta->type != GOT_OBJ_TYPE_TAG) {
1244 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1248 pack = get_cached_pack(delta->path_packfile, repo);
1250 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1254 if (fseeko(pack->packfile, delta->offset + delta->tslen,
1256 err = got_error_from_errno();
1259 err = got_inflate_to_mem(&base_buf, &base_len,
1261 if (base_len < max_size) {
1263 p = reallocarray(base_buf, 1, max_size);
1265 err = got_error_from_errno();
1276 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1277 delta->delta_len, accum_buf, &accum_size);
1282 if (n < deltas->nentries) {
1283 /* Accumulated delta becomes the new base. */
1284 uint8_t *tmp = accum_buf;
1285 accum_buf = base_buf;
1297 *outbuf = accum_buf;
1298 *outlen = accum_size;
1303 const struct got_error *
1304 got_packfile_extract_object(FILE **f, struct got_object *obj,
1305 struct got_repository *repo)
1307 const struct got_error *err = NULL;
1309 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1310 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1312 *f = got_opentemp();
1314 err = got_error(GOT_ERR_FILE_OPEN);
1318 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1319 struct got_pack *pack;
1321 pack = get_cached_pack(obj->path_packfile, repo);
1323 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1328 if (fseeko(pack->packfile, obj->pack_offset, SEEK_SET) != 0) {
1329 err = got_error_from_errno();
1333 err = got_inflate_to_file(&obj->size, pack->packfile, *f);
1335 err = dump_delta_chain_to_file(&obj->size, &obj->deltas, *f,
1345 const struct got_error *
1346 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1347 struct got_object *obj, struct got_repository *repo)
1349 const struct got_error *err = NULL;
1351 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1352 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1354 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1355 struct got_pack *pack;
1357 pack = get_cached_pack(obj->path_packfile, repo);
1359 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1364 if (fseeko(pack->packfile, obj->pack_offset, SEEK_SET) != 0) {
1365 err = got_error_from_errno();
1369 err = got_inflate_to_mem(buf, len, pack->packfile);
1371 err = dump_delta_chain_to_mem(buf, len, &obj->deltas, repo);