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))
59 #define GOT_DELTA_CHAIN_RECURSION_MAX 100
62 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
65 static const struct got_error *
66 verify_fanout_table(uint32_t *fanout_table)
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);
78 static const struct got_error *
79 get_packfile_size(size_t *size, const char *path)
86 dot = strrchr(path, '.');
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, ".idx") == 0) {
92 const struct got_error *err = NULL;
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, '.');
101 return got_error(GOT_ERR_BAD_PATH);
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();
111 } else if (strcmp(dot, ".pack") == 0) {
112 if (stat(path, &sb) != 0)
113 return got_error_from_errno();
115 return got_error(GOT_ERR_BAD_PATH);
121 const struct got_error *
122 got_packidx_open(struct got_packidx_v2_hdr **packidx, const char *path)
124 struct got_packidx_v2_hdr *p;
126 const struct got_error *err = NULL;
127 size_t n, nobj, packfile_size;
129 uint8_t sha1[SHA1_DIGEST_LENGTH];
133 f = fopen(path, "rb");
135 return got_error_from_errno();
137 err = get_packfile_size(&packfile_size, path);
141 p = calloc(1, sizeof(*p));
143 err = got_error_from_errno();
147 n = fread(&p->magic, sizeof(p->magic), 1, f);
149 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
153 if (betoh32(p->magic) != GOT_PACKIDX_V2_MAGIC) {
154 err = got_error(GOT_ERR_BAD_PACKIDX);
158 SHA1Update(&ctx, (uint8_t *)&p->magic, sizeof(p->magic));
160 n = fread(&p->version, sizeof(p->version), 1, f);
162 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
166 if (betoh32(p->version) != GOT_PACKIDX_VERSION) {
167 err = got_error(GOT_ERR_BAD_PACKIDX);
171 SHA1Update(&ctx, (uint8_t *)&p->version, sizeof(p->version));
173 n = fread(&p->fanout_table, sizeof(p->fanout_table), 1, f);
175 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
179 err = verify_fanout_table(p->fanout_table);
183 SHA1Update(&ctx, (uint8_t *)p->fanout_table, sizeof(p->fanout_table));
185 nobj = betoh32(p->fanout_table[0xff]);
187 p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
188 if (p->sorted_ids == NULL) {
189 err = got_error_from_errno();
193 n = fread(p->sorted_ids, sizeof(*p->sorted_ids), nobj, f);
195 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
199 SHA1Update(&ctx, (uint8_t *)p->sorted_ids,
200 nobj * sizeof(*p->sorted_ids));
202 p->crc32 = calloc(nobj, sizeof(*p->crc32));
203 if (p->crc32 == NULL) {
204 err = got_error_from_errno();
208 n = fread(p->crc32, sizeof(*p->crc32), nobj, f);
210 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
214 SHA1Update(&ctx, (uint8_t *)p->crc32, nobj * sizeof(*p->crc32));
216 p->offsets = calloc(nobj, sizeof(*p->offsets));
217 if (p->offsets == NULL) {
218 err = got_error_from_errno();
222 n = fread(p->offsets, sizeof(*p->offsets), nobj, f);
224 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
228 SHA1Update(&ctx, (uint8_t *)p->offsets, nobj * sizeof(*p->offsets));
230 /* Large file offsets are contained only in files > 2GB. */
231 if (packfile_size <= 0x80000000)
234 p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
235 if (p->large_offsets == NULL) {
236 err = got_error_from_errno();
240 n = fread(p->large_offsets, sizeof(*p->large_offsets), nobj, f);
242 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
246 SHA1Update(&ctx, (uint8_t*)p->large_offsets,
247 nobj * sizeof(*p->large_offsets));
250 n = fread(&p->trailer, sizeof(p->trailer), 1, f);
252 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
256 SHA1Update(&ctx, p->trailer.packfile_sha1, SHA1_DIGEST_LENGTH);
257 SHA1Final(sha1, &ctx);
258 if (memcmp(p->trailer.packidx_sha1, sha1, SHA1_DIGEST_LENGTH) != 0)
259 err = got_error(GOT_ERR_PACKIDX_CSUM);
263 got_packidx_close(p);
270 got_packidx_close(struct got_packidx_v2_hdr *packidx)
272 free(packidx->sorted_ids);
273 free(packidx->offsets);
274 free(packidx->crc32);
275 free(packidx->large_offsets);
280 is_packidx_filename(const char *name, size_t len)
282 if (len != GOT_PACKIDX_NAMELEN)
285 if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
288 if (strcmp(name + strlen(GOT_PACK_PREFIX) +
289 SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
296 get_object_offset(struct got_packidx_v2_hdr *packidx, int idx)
298 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
299 uint32_t offset = betoh32(packidx->offsets[idx]);
300 if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
302 idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
303 if (idx < 0 || idx > totobj || packidx->large_offsets == NULL)
305 loffset = betoh64(packidx->large_offsets[idx]);
306 return (loffset > INT64_MAX ? -1 : (off_t)loffset);
308 return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
311 static const struct got_error *
312 get_packfile_path(char **path_packfile, struct got_repository *repo,
313 struct got_packidx_v2_hdr *packidx);
316 get_object_idx(struct got_packidx_v2_hdr *packidx, struct got_object_id *id, struct got_repository *repo)
318 u_int8_t id0 = id->sha1[0];
319 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
323 i = betoh32(packidx->fanout_table[id0 - 1]);
326 struct got_object_id *oid = &packidx->sorted_ids[i];
327 int cmp = got_object_id_cmp(id, oid);
337 static struct got_packidx_v2_hdr *
338 dup_packidx(struct got_packidx_v2_hdr *packidx)
340 struct got_packidx_v2_hdr *p;
343 p = calloc(1, sizeof(*p));
347 memcpy(p, packidx, sizeof(*p));
348 p->sorted_ids = NULL;
351 p->large_offsets = NULL;
353 nobj = betoh32(p->fanout_table[0xff]);
355 p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
356 if (p->sorted_ids == NULL)
358 memcpy(p->sorted_ids, packidx->sorted_ids,
359 nobj * sizeof(*p->sorted_ids));
361 p->crc32 = calloc(nobj, sizeof(*p->crc32));
362 if (p->crc32 == NULL)
364 memcpy(p->crc32, packidx->crc32, nobj * sizeof(*p->crc32));
366 p->offsets = calloc(nobj, sizeof(*p->offsets));
367 if (p->offsets == NULL)
369 memcpy(p->offsets, packidx->offsets, nobj * sizeof(*p->offsets));
371 if (p->large_offsets) {
372 p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
373 if (p->large_offsets == NULL)
375 memcpy(p->large_offsets, packidx->large_offsets,
376 nobj * sizeof(*p->large_offsets));
382 free(p->large_offsets);
391 cache_packidx(struct got_packidx_v2_hdr *packidx, struct got_repository *repo)
395 for (i = 0; i < nitems(repo->packidx_cache); i++) {
396 if (repo->packidx_cache[i] == NULL)
400 if (i == nitems(repo->packidx_cache)) {
401 got_packidx_close(repo->packidx_cache[i - 1]);
402 memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
403 sizeof(repo->packidx_cache) -
404 sizeof(repo->packidx_cache[0]));
408 repo->packidx_cache[i] = dup_packidx(packidx);
411 static const struct got_error *
412 search_packidx(struct got_packidx_v2_hdr **packidx, int *idx,
413 struct got_repository *repo, struct got_object_id *id)
415 const struct got_error *err;
422 /* Search pack index cache. */
423 for (i = 0; i < nitems(repo->packidx_cache); i++) {
424 if (repo->packidx_cache[i] == NULL)
426 *idx = get_object_idx(repo->packidx_cache[i], id, repo);
428 *packidx = repo->packidx_cache[i];
432 /* No luck. Search the filesystem. */
434 path_packdir = got_repo_get_path_objects_pack(repo);
435 if (path_packdir == NULL)
436 return got_error_from_errno();
438 packdir = opendir(path_packdir);
439 if (packdir == NULL) {
440 err = got_error_from_errno();
444 while ((dent = readdir(packdir)) != NULL) {
445 if (!is_packidx_filename(dent->d_name, dent->d_namlen))
448 if (asprintf(&path_packidx, "%s/%s", path_packdir,
449 dent->d_name) == -1) {
450 err = got_error_from_errno();
454 err = got_packidx_open(packidx, path_packidx);
459 *idx = get_object_idx(*packidx, id, repo);
461 err = NULL; /* found the object */
462 cache_packidx(*packidx, repo);
466 got_packidx_close(*packidx);
470 err = got_error(GOT_ERR_NO_OBJ);
473 if (packdir && closedir(packdir) != 0 && err == 0)
474 err = got_error_from_errno();
478 static const struct got_error *
479 get_packfile_path(char **path_packfile, struct got_repository *repo,
480 struct got_packidx_v2_hdr *packidx)
483 char hex[SHA1_DIGEST_STRING_LENGTH];
486 path_packdir = got_repo_get_path_objects_pack(repo);
487 if (path_packdir == NULL)
488 return got_error_from_errno();
490 sha1str = got_sha1_digest_to_str(packidx->trailer.packfile_sha1,
493 return got_error(GOT_ERR_PACKIDX_CSUM);
495 if (asprintf(path_packfile, "%s/%s%s%s", path_packdir,
496 GOT_PACK_PREFIX, sha1str, GOT_PACKFILE_SUFFIX) == -1) {
497 *path_packfile = NULL;
498 return got_error_from_errno();
504 const struct got_error *
505 read_packfile_hdr(FILE *f, struct got_packidx_v2_hdr *packidx)
507 const struct got_error *err = NULL;
508 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
509 struct got_packfile_hdr hdr;
512 n = fread(&hdr, sizeof(hdr), 1, f);
514 return got_ferror(f, GOT_ERR_BAD_PACKIDX);
516 if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
517 betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
518 betoh32(hdr.nobjects) != totobj)
519 err = got_error(GOT_ERR_BAD_PACKFILE);
524 static const struct got_error *
525 open_packfile(FILE **packfile, const char *path_packfile,
526 struct got_repository *repo, struct got_packidx_v2_hdr *packidx)
528 const struct got_error *err = NULL;
532 *packfile = fopen(path_packfile, "rb");
533 if (*packfile == NULL) {
534 err = got_error_from_errno();
539 err = read_packfile_hdr(*packfile, packidx);
549 got_pack_close(struct got_pack *pack)
551 fclose(pack->packfile);
552 pack->packfile = NULL;
553 free(pack->path_packfile);
554 pack->path_packfile = NULL;
558 static const struct got_error *
559 cache_pack(struct got_pack **packp, const char *path_packfile,
560 struct got_packidx_v2_hdr *packidx, struct got_repository *repo)
562 const struct got_error *err = NULL;
563 struct got_pack *pack = NULL;
569 for (i = 0; i < nitems(repo->packs); i++) {
570 pack = &repo->packs[i];
571 if (pack->path_packfile == NULL)
573 if (strcmp(pack->path_packfile, path_packfile) == 0)
577 if (i == nitems(repo->packs) - 1) {
578 got_pack_close(&repo->packs[i - 1]);
579 memmove(&repo->packs[1], &repo->packs[0],
580 sizeof(repo->packs) - sizeof(repo->packs[0]));
584 pack = &repo->packs[i];
586 pack->path_packfile = strdup(path_packfile);
587 if (pack->path_packfile == NULL) {
588 err = got_error_from_errno();
592 err = open_packfile(&pack->packfile, path_packfile, repo, packidx);
596 err = get_packfile_size(&pack->filesize, path_packfile);
600 free(pack->path_packfile);
608 get_cached_pack(const char *path_packfile, struct got_repository *repo)
610 struct got_pack *pack = NULL;
613 for (i = 0; i < nitems(repo->packs); i++) {
614 pack = &repo->packs[i];
615 if (pack->path_packfile == NULL)
617 if (strcmp(pack->path_packfile, path_packfile) == 0)
624 static const struct got_error *
625 parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
635 /* We do not support size values which don't fit in 64 bit. */
637 return got_error(GOT_ERR_NO_SPACE);
639 n = fread(&sizeN, sizeof(sizeN), 1, packfile);
641 return got_ferror(packfile, GOT_ERR_BAD_PACKIDX);
644 t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
645 GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
646 s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
648 size_t shift = 4 + 7 * (i - 1);
649 s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
652 } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
656 *len = i * sizeof(sizeN);
660 static const struct got_error *
661 open_plain_object(struct got_object **obj, const char *path_packfile,
662 struct got_object_id *id, uint8_t type, off_t offset, size_t size)
664 *obj = calloc(1, sizeof(**obj));
666 return got_error_from_errno();
668 (*obj)->path_packfile = strdup(path_packfile);
669 if ((*obj)->path_packfile == NULL) {
670 const struct got_error *err = got_error_from_errno();
677 (*obj)->flags = GOT_OBJ_FLAG_PACKED;
680 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
681 (*obj)->pack_offset = offset;
686 static const struct got_error *
687 parse_negative_offset(int64_t *offset, size_t *len, FILE *packfile)
695 /* We do not support offset values which don't fit in 64 bit. */
697 return got_error(GOT_ERR_NO_SPACE);
699 n = fread(&offN, sizeof(offN), 1, packfile);
701 return got_ferror(packfile, GOT_ERR_BAD_PACKIDX);
704 o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
708 o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
711 } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
714 *len = i * sizeof(offN);
718 static const struct got_error *
719 parse_offset_delta(off_t *base_offset, FILE *packfile, off_t offset)
721 const struct got_error *err;
725 err = parse_negative_offset(&negoffset, &negofflen, packfile);
729 /* Compute the base object's offset (must be in the same pack file). */
730 *base_offset = (offset - negoffset);
731 if (*base_offset <= 0)
732 return got_error(GOT_ERR_BAD_PACKFILE);
737 static const struct got_error *
738 resolve_delta_chain(struct got_delta_chain *, struct got_repository *,
739 FILE *, const char *, off_t, size_t, int, size_t, unsigned int);
741 static const struct got_error *
742 add_delta(struct got_delta_chain *deltas, const char *path_packfile,
743 off_t delta_offset, size_t tslen, int delta_type, size_t delta_size,
744 size_t delta_data_offset, uint8_t *delta_buf, size_t delta_len)
746 struct got_delta *delta;
748 delta = got_delta_open(path_packfile, delta_offset, tslen,
749 delta_type, delta_size, delta_data_offset, delta_buf,
752 return got_error_from_errno();
753 /* delta is freed in got_object_close() */
755 SIMPLEQ_INSERT_HEAD(&deltas->entries, delta, entry);
759 static const struct got_error *
760 resolve_offset_delta(struct got_delta_chain *deltas,
761 struct got_repository *repo, FILE *packfile, const char *path_packfile,
762 off_t delta_offset,size_t tslen, int delta_type, size_t delta_size,
763 unsigned int recursion)
766 const struct got_error *err;
771 off_t delta_data_offset;
775 err = parse_offset_delta(&base_offset, packfile, delta_offset);
779 delta_data_offset = ftello(packfile);
780 if (delta_data_offset == -1)
781 return got_error_from_errno();
783 err = got_inflate_to_mem(&delta_buf, &delta_len, packfile);
787 err = add_delta(deltas, path_packfile, delta_offset, tslen,
788 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
792 /* An offset delta must be in the same packfile. */
793 if (fseeko(packfile, base_offset, SEEK_SET) != 0)
794 return got_error_from_errno();
796 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
801 return resolve_delta_chain(deltas, repo, packfile, path_packfile,
802 base_offset, base_tslen, base_type, base_size, recursion - 1);
805 static const struct got_error *
806 resolve_ref_delta(struct got_delta_chain *deltas, struct got_repository *repo,
807 FILE *packfile, const char *path_packfile, off_t delta_offset,
808 size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
810 const struct got_error *err;
811 struct got_object_id id;
812 struct got_packidx_v2_hdr *packidx;
819 char *path_base_packfile;
820 struct got_pack *base_pack;
821 off_t delta_data_offset;
825 n = fread(&id, sizeof(id), 1, packfile);
827 return got_ferror(packfile, GOT_ERR_IO);
829 delta_data_offset = ftello(packfile);
830 if (delta_data_offset == -1)
831 return got_error_from_errno();
833 err = got_inflate_to_mem(&delta_buf, &delta_len, packfile);
837 err = add_delta(deltas, path_packfile, delta_offset, tslen,
838 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
842 err = search_packidx(&packidx, &idx, repo, &id);
846 base_offset = get_object_offset(packidx, idx);
847 if (base_offset == (uint64_t)-1) {
848 return got_error(GOT_ERR_BAD_PACKIDX);
851 err = get_packfile_path(&path_base_packfile, repo, packidx);
855 base_pack = get_cached_pack(path_base_packfile, repo);
856 if (base_pack == NULL) {
857 err = cache_pack(&base_pack, path_base_packfile, NULL, repo);
862 if (fseeko(base_pack->packfile, base_offset, SEEK_SET) != 0) {
863 err = got_error_from_errno();
867 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
868 base_pack->packfile);
872 err = resolve_delta_chain(deltas, repo, base_pack->packfile,
873 path_base_packfile, base_offset, base_tslen, base_type,
874 base_size, recursion - 1);
876 free(path_base_packfile);
880 static const struct got_error *
881 resolve_delta_chain(struct got_delta_chain *deltas, struct got_repository *repo,
882 FILE *packfile, const char *path_packfile, off_t delta_offset, size_t tslen,
883 int delta_type, size_t delta_size, unsigned int recursion)
885 const struct got_error *err = NULL;
887 if (--recursion == 0)
888 return got_error(GOT_ERR_RECURSION);
890 switch (delta_type) {
891 case GOT_OBJ_TYPE_COMMIT:
892 case GOT_OBJ_TYPE_TREE:
893 case GOT_OBJ_TYPE_BLOB:
894 case GOT_OBJ_TYPE_TAG:
895 /* Plain types are the final delta base. Recursion ends. */
896 err = add_delta(deltas, path_packfile, delta_offset, tslen,
897 delta_type, delta_size, 0, NULL, 0);
899 case GOT_OBJ_TYPE_OFFSET_DELTA:
900 err = resolve_offset_delta(deltas, repo, packfile,
901 path_packfile, delta_offset, tslen, delta_type,
902 delta_size, recursion - 1);
904 case GOT_OBJ_TYPE_REF_DELTA:
905 err = resolve_ref_delta(deltas, repo, packfile,
906 path_packfile, delta_offset, tslen, delta_type,
907 delta_size, recursion - 1);
910 return got_error(GOT_ERR_OBJ_TYPE);
916 static const struct got_error *
917 open_delta_object(struct got_object **obj, struct got_repository *repo,
918 struct got_packidx_v2_hdr *packidx, const char *path_packfile,
919 FILE *packfile, struct got_object_id *id, off_t offset, size_t tslen,
920 int delta_type, size_t delta_size)
922 const struct got_error *err = NULL;
925 *obj = calloc(1, sizeof(**obj));
927 return got_error_from_errno();
931 (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
932 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
933 (*obj)->pack_offset = offset + tslen;
935 (*obj)->path_packfile = strdup(path_packfile);
936 if ((*obj)->path_packfile == NULL) {
937 err = got_error_from_errno();
940 (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
942 SIMPLEQ_INIT(&(*obj)->deltas.entries);
943 (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
945 err = resolve_delta_chain(&(*obj)->deltas, repo, packfile,
946 path_packfile, offset, tslen, delta_type, delta_size,
947 GOT_DELTA_CHAIN_RECURSION_MAX);
951 err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
954 (*obj)->type = resolved_type;
958 got_object_close(*obj);
964 static const struct got_error *
965 open_packed_object(struct got_object **obj, struct got_repository *repo,
966 struct got_packidx_v2_hdr *packidx, int idx, struct got_object_id *id)
968 const struct got_error *err = NULL;
971 struct got_pack *pack;
978 offset = get_object_offset(packidx, idx);
979 if (offset == (uint64_t)-1)
980 return got_error(GOT_ERR_BAD_PACKIDX);
982 err = get_packfile_path(&path_packfile, repo, packidx);
986 pack = get_cached_pack(path_packfile, repo);
988 err = cache_pack(&pack, path_packfile, packidx, repo);
993 if (fseeko(pack->packfile, offset, SEEK_SET) != 0) {
994 err = got_error_from_errno();
998 err = parse_object_type_and_size(&type, &size, &tslen, pack->packfile);
1003 case GOT_OBJ_TYPE_COMMIT:
1004 case GOT_OBJ_TYPE_TREE:
1005 case GOT_OBJ_TYPE_BLOB:
1006 case GOT_OBJ_TYPE_TAG:
1007 err = open_plain_object(obj, path_packfile, id, type,
1008 offset + tslen, size);
1011 case GOT_OBJ_TYPE_OFFSET_DELTA:
1012 case GOT_OBJ_TYPE_REF_DELTA:
1013 err = open_delta_object(obj, repo, packidx, path_packfile,
1014 pack->packfile, id, offset, tslen, type, size);
1018 err = got_error(GOT_ERR_OBJ_TYPE);
1022 free(path_packfile);
1026 const struct got_error *
1027 got_packfile_open_object(struct got_object **obj, struct got_object_id *id,
1028 struct got_repository *repo)
1030 const struct got_error *err = NULL;
1031 struct got_packidx_v2_hdr *packidx = NULL;
1034 err = search_packidx(&packidx, &idx, repo, id);
1038 err = open_packed_object(obj, repo, packidx, idx, id);
1042 err = cache_pack(NULL, (*obj)->path_packfile, packidx, repo);
1046 static const struct got_error *
1047 get_delta_chain_max_size(uint64_t *max_size, struct got_delta_chain *deltas)
1049 struct got_delta *delta;
1050 uint64_t base_size = 0, result_size = 0;
1053 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1054 /* Plain object types are the delta base. */
1055 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1056 delta->type != GOT_OBJ_TYPE_TREE &&
1057 delta->type != GOT_OBJ_TYPE_BLOB &&
1058 delta->type != GOT_OBJ_TYPE_TAG) {
1059 const struct got_error *err;
1060 err = got_delta_get_sizes(&base_size, &result_size,
1061 delta->delta_buf, delta->delta_len);
1065 base_size = delta->size;
1066 if (base_size > *max_size)
1067 *max_size = base_size;
1068 if (result_size > *max_size)
1069 *max_size = result_size;
1075 static const struct got_error *
1076 dump_delta_chain_to_file(size_t *result_size, struct got_delta_chain *deltas,
1077 FILE *outfile, struct got_repository *repo)
1079 const struct got_error *err = NULL;
1080 struct got_delta *delta;
1081 FILE *base_file = NULL, *accum_file = NULL;
1082 uint8_t *base_buf = NULL, *accum_buf = NULL;
1083 size_t accum_size = 0;
1089 if (SIMPLEQ_EMPTY(&deltas->entries))
1090 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1092 /* We process small enough files entirely in memory for speed. */
1093 err = get_delta_chain_max_size(&max_size, deltas);
1096 if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1097 accum_buf = malloc(max_size);
1098 if (accum_buf == NULL)
1099 return got_error_from_errno();
1101 base_file = got_opentemp();
1102 if (base_file == NULL)
1103 return got_error_from_errno();
1105 accum_file = got_opentemp();
1106 if (accum_file == NULL) {
1107 err = got_error_from_errno();
1113 /* Deltas are ordered in ascending order. */
1114 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1116 struct got_pack *pack;
1119 /* Plain object types are the delta base. */
1120 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1121 delta->type != GOT_OBJ_TYPE_TREE &&
1122 delta->type != GOT_OBJ_TYPE_BLOB &&
1123 delta->type != GOT_OBJ_TYPE_TAG) {
1124 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1128 pack = get_cached_pack(delta->path_packfile, repo);
1130 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1134 if (fseeko(pack->packfile, delta->offset + delta->tslen,
1136 err = got_error_from_errno();
1140 err = got_inflate_to_file(&base_len,
1141 pack->packfile, base_file);
1143 err = got_inflate_to_mem(&base_buf, &base_len,
1145 if (base_len < max_size) {
1147 p = reallocarray(base_buf, 1, max_size);
1149 err = got_error_from_errno();
1164 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1165 delta->delta_len, accum_buf, &accum_size);
1168 err = got_delta_apply(base_file, delta->delta_buf,
1170 /* Final delta application writes to output file. */
1171 ++n < deltas->nentries ? accum_file : outfile,
1177 if (n < deltas->nentries) {
1178 /* Accumulated delta becomes the new base. */
1180 uint8_t *tmp = accum_buf;
1181 accum_buf = base_buf;
1184 FILE *tmp = accum_file;
1185 accum_file = base_file;
1196 size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1198 if (len != accum_size)
1199 return got_ferror(outfile, GOT_ERR_IO);
1207 *result_size = accum_size;
1211 static const struct got_error *
1212 dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1213 struct got_delta_chain *deltas, struct got_repository *repo)
1215 const struct got_error *err = NULL;
1216 struct got_delta *delta;
1217 uint8_t *base_buf = NULL, *accum_buf = NULL;
1225 if (SIMPLEQ_EMPTY(&deltas->entries))
1226 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1228 err = get_delta_chain_max_size(&max_size, deltas);
1231 accum_buf = malloc(max_size);
1232 if (accum_buf == NULL)
1233 return got_error_from_errno();
1235 /* Deltas are ordered in ascending order. */
1236 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1238 struct got_pack *pack;
1241 /* Plain object types are the delta base. */
1242 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1243 delta->type != GOT_OBJ_TYPE_TREE &&
1244 delta->type != GOT_OBJ_TYPE_BLOB &&
1245 delta->type != GOT_OBJ_TYPE_TAG) {
1246 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1250 pack = get_cached_pack(delta->path_packfile, repo);
1252 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1256 if (fseeko(pack->packfile, delta->offset + delta->tslen,
1258 err = got_error_from_errno();
1261 err = got_inflate_to_mem(&base_buf, &base_len,
1263 if (base_len < max_size) {
1265 p = reallocarray(base_buf, 1, max_size);
1267 err = got_error_from_errno();
1278 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1279 delta->delta_len, accum_buf, &accum_size);
1284 if (n < deltas->nentries) {
1285 /* Accumulated delta becomes the new base. */
1286 uint8_t *tmp = accum_buf;
1287 accum_buf = base_buf;
1299 *outbuf = accum_buf;
1300 *outlen = accum_size;
1305 const struct got_error *
1306 got_packfile_extract_object(FILE **f, struct got_object *obj,
1307 struct got_repository *repo)
1309 const struct got_error *err = NULL;
1311 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1312 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1314 *f = got_opentemp();
1316 err = got_error(GOT_ERR_FILE_OPEN);
1320 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1321 struct got_pack *pack;
1323 pack = get_cached_pack(obj->path_packfile, repo);
1325 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1330 if (fseeko(pack->packfile, obj->pack_offset, SEEK_SET) != 0) {
1331 err = got_error_from_errno();
1335 err = got_inflate_to_file(&obj->size, pack->packfile, *f);
1337 err = dump_delta_chain_to_file(&obj->size, &obj->deltas, *f,
1347 const struct got_error *
1348 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1349 struct got_object *obj, struct got_repository *repo)
1351 const struct got_error *err = NULL;
1353 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1354 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1356 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1357 struct got_pack *pack;
1359 pack = get_cached_pack(obj->path_packfile, repo);
1361 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1366 if (fseeko(pack->packfile, obj->pack_offset, SEEK_SET) != 0) {
1367 err = got_error_from_errno();
1371 err = got_inflate_to_mem(buf, len, pack->packfile);
1373 err = dump_delta_chain_to_mem(buf, len, &obj->deltas, repo);