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_idx)
81 char base_path[PATH_MAX];
84 if (strlcpy(base_path, path_idx, PATH_MAX) > PATH_MAX)
85 return got_error(GOT_ERR_NO_SPACE);
87 dot = strrchr(base_path, '.');
89 return got_error(GOT_ERR_BAD_PATH);
91 if (asprintf(&path_pack, "%s.pack", base_path) == -1)
92 return got_error_from_errno();
94 if (stat(path_pack, &sb) != 0) {
96 return got_error_from_errno();
104 const struct got_error *
105 got_packidx_open(struct got_packidx_v2_hdr **packidx, const char *path)
107 struct got_packidx_v2_hdr *p;
109 const struct got_error *err = NULL;
110 size_t n, nobj, packfile_size;
112 uint8_t sha1[SHA1_DIGEST_LENGTH];
116 f = fopen(path, "rb");
118 return got_error_from_errno();
120 err = get_packfile_size(&packfile_size, path);
124 p = calloc(1, sizeof(*p));
126 err = got_error_from_errno();
130 n = fread(&p->magic, sizeof(p->magic), 1, f);
132 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
136 if (betoh32(p->magic) != GOT_PACKIDX_V2_MAGIC) {
137 err = got_error(GOT_ERR_BAD_PACKIDX);
141 SHA1Update(&ctx, (uint8_t *)&p->magic, sizeof(p->magic));
143 n = fread(&p->version, sizeof(p->version), 1, f);
145 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
149 if (betoh32(p->version) != GOT_PACKIDX_VERSION) {
150 err = got_error(GOT_ERR_BAD_PACKIDX);
154 SHA1Update(&ctx, (uint8_t *)&p->version, sizeof(p->version));
156 n = fread(&p->fanout_table, sizeof(p->fanout_table), 1, f);
158 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
162 err = verify_fanout_table(p->fanout_table);
166 SHA1Update(&ctx, (uint8_t *)p->fanout_table, sizeof(p->fanout_table));
168 nobj = betoh32(p->fanout_table[0xff]);
170 p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
171 if (p->sorted_ids == NULL) {
172 err = got_error_from_errno();
176 n = fread(p->sorted_ids, sizeof(*p->sorted_ids), nobj, f);
178 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
182 SHA1Update(&ctx, (uint8_t *)p->sorted_ids,
183 nobj * sizeof(*p->sorted_ids));
185 p->crc32 = calloc(nobj, sizeof(*p->crc32));
186 if (p->crc32 == NULL) {
187 err = got_error_from_errno();
191 n = fread(p->crc32, sizeof(*p->crc32), nobj, f);
193 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
197 SHA1Update(&ctx, (uint8_t *)p->crc32, nobj * sizeof(*p->crc32));
199 p->offsets = calloc(nobj, sizeof(*p->offsets));
200 if (p->offsets == NULL) {
201 err = got_error_from_errno();
205 n = fread(p->offsets, sizeof(*p->offsets), nobj, f);
207 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
211 SHA1Update(&ctx, (uint8_t *)p->offsets, nobj * sizeof(*p->offsets));
213 /* Large file offsets are contained only in files > 2GB. */
214 if (packfile_size <= 0x80000000)
217 p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
218 if (p->large_offsets == NULL) {
219 err = got_error_from_errno();
223 n = fread(p->large_offsets, sizeof(*p->large_offsets), nobj, f);
225 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
229 SHA1Update(&ctx, (uint8_t*)p->large_offsets,
230 nobj * sizeof(*p->large_offsets));
233 n = fread(&p->trailer, sizeof(p->trailer), 1, f);
235 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
239 SHA1Update(&ctx, p->trailer.packfile_sha1, SHA1_DIGEST_LENGTH);
240 SHA1Final(sha1, &ctx);
241 if (memcmp(p->trailer.packidx_sha1, sha1, SHA1_DIGEST_LENGTH) != 0)
242 err = got_error(GOT_ERR_PACKIDX_CSUM);
246 got_packidx_close(p);
253 got_packidx_close(struct got_packidx_v2_hdr *packidx)
255 free(packidx->sorted_ids);
256 free(packidx->offsets);
257 free(packidx->crc32);
258 free(packidx->large_offsets);
263 is_packidx_filename(const char *name, size_t len)
265 if (len != GOT_PACKIDX_NAMELEN)
268 if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
271 if (strcmp(name + strlen(GOT_PACK_PREFIX) +
272 SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
279 get_object_offset(struct got_packidx_v2_hdr *packidx, int idx)
281 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
282 uint32_t offset = betoh32(packidx->offsets[idx]);
283 if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
285 idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
286 if (idx < 0 || idx > totobj || packidx->large_offsets == NULL)
288 loffset = betoh64(packidx->large_offsets[idx]);
289 return (loffset > INT64_MAX ? -1 : (off_t)loffset);
291 return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
294 static const struct got_error *
295 get_packfile_path(char **path_packfile, struct got_repository *repo,
296 struct got_packidx_v2_hdr *packidx);
299 get_object_idx(struct got_packidx_v2_hdr *packidx, struct got_object_id *id, struct got_repository *repo)
301 u_int8_t id0 = id->sha1[0];
302 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
306 i = betoh32(packidx->fanout_table[id0 - 1]);
309 struct got_object_id *oid = &packidx->sorted_ids[i];
310 int cmp = got_object_id_cmp(id, oid);
320 static struct got_packidx_v2_hdr *
321 dup_packidx(struct got_packidx_v2_hdr *packidx)
323 struct got_packidx_v2_hdr *p;
326 p = calloc(1, sizeof(*p));
330 memcpy(p, packidx, sizeof(*p));
331 p->sorted_ids = NULL;
334 p->large_offsets = NULL;
336 nobj = betoh32(p->fanout_table[0xff]);
338 p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
339 if (p->sorted_ids == NULL)
341 memcpy(p->sorted_ids, packidx->sorted_ids, nobj * sizeof(*p->sorted_ids));
343 p->crc32 = calloc(nobj, sizeof(*p->crc32));
344 if (p->crc32 == NULL)
346 memcpy(p->crc32, packidx->crc32, nobj * sizeof(*p->crc32));
348 p->offsets = calloc(nobj, sizeof(*p->offsets));
349 if (p->offsets == NULL)
351 memcpy(p->offsets, packidx->offsets, nobj * sizeof(*p->offsets));
353 if (p->large_offsets) {
354 p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
355 if (p->large_offsets == NULL)
357 memcpy(p->large_offsets, packidx->large_offsets,
358 nobj * sizeof(*p->large_offsets));
364 free(p->large_offsets);
373 cache_packidx(struct got_packidx_v2_hdr *packidx, struct got_repository *repo)
377 for (i = 0; i < nitems(repo->packidx_cache); i++) {
378 if (repo->packidx_cache[i] == NULL)
382 if (i == nitems(repo->packidx_cache)) {
383 got_packidx_close(repo->packidx_cache[i - 1]);
384 memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
385 sizeof(repo->packidx_cache) -
386 sizeof(repo->packidx_cache[0]));
390 repo->packidx_cache[i] = dup_packidx(packidx);
393 static const struct got_error *
394 search_packidx(struct got_packidx_v2_hdr **packidx, int *idx,
395 struct got_repository *repo, struct got_object_id *id)
397 const struct got_error *err;
404 /* Search pack index cache. */
405 for (i = 0; i < nitems(repo->packidx_cache); i++) {
406 if (repo->packidx_cache[i] == NULL)
408 *idx = get_object_idx(repo->packidx_cache[i], id, repo);
410 *packidx = repo->packidx_cache[i];
414 /* No luck. Search the filesystem. */
416 path_packdir = got_repo_get_path_objects_pack(repo);
417 if (path_packdir == NULL)
418 return got_error_from_errno();
420 packdir = opendir(path_packdir);
421 if (packdir == NULL) {
422 err = got_error_from_errno();
426 while ((dent = readdir(packdir)) != NULL) {
427 if (!is_packidx_filename(dent->d_name, dent->d_namlen))
430 if (asprintf(&path_packidx, "%s/%s", path_packdir,
431 dent->d_name) == -1) {
432 err = got_error_from_errno();
436 err = got_packidx_open(packidx, path_packidx);
441 *idx = get_object_idx(*packidx, id, repo);
443 err = NULL; /* found the object */
444 cache_packidx(*packidx, repo);
448 got_packidx_close(*packidx);
452 err = got_error(GOT_ERR_NO_OBJ);
455 if (packdir && closedir(packdir) != 0 && err == 0)
456 err = got_error_from_errno();
460 static const struct got_error *
461 get_packfile_path(char **path_packfile, struct got_repository *repo,
462 struct got_packidx_v2_hdr *packidx)
465 char hex[SHA1_DIGEST_STRING_LENGTH];
468 path_packdir = got_repo_get_path_objects_pack(repo);
469 if (path_packdir == NULL)
470 return got_error_from_errno();
472 sha1str = got_sha1_digest_to_str(packidx->trailer.packfile_sha1,
475 return got_error(GOT_ERR_PACKIDX_CSUM);
477 if (asprintf(path_packfile, "%s/%s%s%s", path_packdir,
478 GOT_PACK_PREFIX, sha1str, GOT_PACKFILE_SUFFIX) == -1) {
479 *path_packfile = NULL;
480 return got_error_from_errno();
486 const struct got_error *
487 read_packfile_hdr(FILE *f, struct got_packidx_v2_hdr *packidx)
489 const struct got_error *err = NULL;
490 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
491 struct got_packfile_hdr hdr;
494 n = fread(&hdr, sizeof(hdr), 1, f);
496 return got_ferror(f, GOT_ERR_BAD_PACKIDX);
498 if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
499 betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
500 betoh32(hdr.nobjects) != totobj)
501 err = got_error(GOT_ERR_BAD_PACKFILE);
506 static const struct got_error *
507 open_packfile(FILE **packfile, const char *path_packfile,
508 struct got_repository *repo, struct got_packidx_v2_hdr *packidx)
510 const struct got_error *err = NULL;
514 *packfile = fopen(path_packfile, "rb");
515 if (*packfile == NULL) {
516 err = got_error_from_errno();
521 err = read_packfile_hdr(*packfile, packidx);
531 got_pack_close(struct got_pack *pack)
533 fclose(pack->packfile);
534 pack->packfile = NULL;
535 free(pack->path_packfile);
536 pack->path_packfile = NULL;
540 static const struct got_error *
541 cache_pack(struct got_pack **packp, const char *path_packfile,
542 struct got_packidx_v2_hdr *packidx, struct got_repository *repo)
544 const struct got_error *err = NULL;
545 struct got_pack *pack = NULL;
551 for (i = 0; i < nitems(repo->packs); i++) {
552 pack = &repo->packs[i];
553 if (pack->path_packfile == NULL)
555 if (strcmp(pack->path_packfile, path_packfile) == 0)
559 if (i == nitems(repo->packs) - 1) {
560 got_pack_close(&repo->packs[i - 1]);
561 memmove(&repo->packs[1], &repo->packs[0],
562 sizeof(repo->packs) - sizeof(repo->packs[0]));
566 pack = &repo->packs[i];
568 pack->path_packfile = strdup(path_packfile);
569 if (pack->path_packfile == NULL) {
570 err = got_error_from_errno();
574 err = open_packfile(&pack->packfile, path_packfile, repo, packidx);
578 err = get_packfile_size(&pack->filesize, path_packfile);
582 free(pack->path_packfile);
590 get_cached_pack(const char *path_packfile, struct got_repository *repo)
592 struct got_pack *pack = NULL;
595 for (i = 0; i < nitems(repo->packs); i++) {
596 pack = &repo->packs[i];
597 if (pack->path_packfile == NULL)
599 if (strcmp(pack->path_packfile, path_packfile) == 0)
606 static const struct got_error *
607 parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
617 /* We do not support size values which don't fit in 64 bit. */
619 return got_error(GOT_ERR_NO_SPACE);
621 n = fread(&sizeN, sizeof(sizeN), 1, packfile);
623 return got_ferror(packfile, GOT_ERR_BAD_PACKIDX);
626 t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
627 GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
628 s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
630 size_t shift = 4 + 7 * (i - 1);
631 s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
634 } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
638 *len = i * sizeof(sizeN);
642 static const struct got_error *
643 open_plain_object(struct got_object **obj, const char *path_packfile,
644 struct got_object_id *id, uint8_t type, off_t offset, size_t size)
646 *obj = calloc(1, sizeof(**obj));
648 return got_error_from_errno();
650 (*obj)->path_packfile = strdup(path_packfile);
651 if ((*obj)->path_packfile == NULL) {
652 const struct got_error *err = got_error_from_errno();
659 (*obj)->flags = GOT_OBJ_FLAG_PACKED;
662 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
663 (*obj)->pack_offset = offset;
668 static const struct got_error *
669 parse_negative_offset(int64_t *offset, size_t *len, FILE *packfile)
677 /* We do not support offset values which don't fit in 64 bit. */
679 return got_error(GOT_ERR_NO_SPACE);
681 n = fread(&offN, sizeof(offN), 1, packfile);
683 return got_ferror(packfile, GOT_ERR_BAD_PACKIDX);
686 o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
690 o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
693 } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
696 *len = i * sizeof(offN);
700 static const struct got_error *
701 parse_offset_delta(off_t *base_offset, FILE *packfile, off_t offset)
703 const struct got_error *err;
707 err = parse_negative_offset(&negoffset, &negofflen, packfile);
711 /* Compute the base object's offset (must be in the same pack file). */
712 *base_offset = (offset - negoffset);
713 if (*base_offset <= 0)
714 return got_error(GOT_ERR_BAD_PACKFILE);
719 static const struct got_error *
720 resolve_delta_chain(struct got_delta_chain *, struct got_repository *,
721 FILE *, const char *, off_t, size_t, int, size_t);
723 static const struct got_error *
724 add_delta(struct got_delta_chain *deltas, const char *path_packfile,
725 off_t delta_offset, size_t tslen, int delta_type, size_t delta_size,
726 size_t delta_data_offset, uint8_t *delta_buf, size_t delta_len)
728 struct got_delta *delta;
730 delta = got_delta_open(path_packfile, delta_offset, tslen,
731 delta_type, delta_size, delta_data_offset, delta_buf,
734 return got_error_from_errno();
735 /* delta is freed in got_object_close() */
737 SIMPLEQ_INSERT_HEAD(&deltas->entries, delta, entry);
741 static const struct got_error *
742 resolve_offset_delta(struct got_delta_chain *deltas,
743 struct got_repository *repo, FILE *packfile, const char *path_packfile,
744 off_t delta_offset,size_t tslen, int delta_type, size_t delta_size)
747 const struct got_error *err;
752 off_t delta_data_offset;
756 err = parse_offset_delta(&base_offset, packfile, delta_offset);
760 delta_data_offset = ftello(packfile);
761 if (delta_data_offset == -1)
762 return got_error_from_errno();
764 err = got_inflate_to_mem(&delta_buf, &delta_len, packfile);
768 err = add_delta(deltas, path_packfile, delta_offset, tslen,
769 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
773 /* An offset delta must be in the same packfile. */
774 if (fseeko(packfile, base_offset, SEEK_SET) != 0)
775 return got_error_from_errno();
777 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
782 return resolve_delta_chain(deltas, repo, packfile, path_packfile,
783 base_offset, base_tslen, base_type, base_size);
786 static const struct got_error *
787 resolve_ref_delta(struct got_delta_chain *deltas, struct got_repository *repo,
788 FILE *packfile, const char *path_packfile, off_t delta_offset,
789 size_t tslen, int delta_type, size_t delta_size)
791 const struct got_error *err;
792 struct got_object_id id;
793 struct got_packidx_v2_hdr *packidx;
800 char *path_base_packfile;
801 struct got_pack *base_pack;
802 off_t delta_data_offset;
806 n = fread(&id, sizeof(id), 1, packfile);
808 return got_ferror(packfile, GOT_ERR_IO);
810 delta_data_offset = ftello(packfile);
811 if (delta_data_offset == -1)
812 return got_error_from_errno();
814 err = got_inflate_to_mem(&delta_buf, &delta_len, packfile);
818 err = add_delta(deltas, path_packfile, delta_offset, tslen,
819 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
823 err = search_packidx(&packidx, &idx, repo, &id);
827 base_offset = get_object_offset(packidx, idx);
828 if (base_offset == (uint64_t)-1) {
829 return got_error(GOT_ERR_BAD_PACKIDX);
832 err = get_packfile_path(&path_base_packfile, repo, packidx);
836 base_pack = get_cached_pack(path_base_packfile, repo);
837 if (base_pack == NULL) {
838 err = cache_pack(&base_pack, path_base_packfile, NULL, repo);
843 if (fseeko(base_pack->packfile, base_offset, SEEK_SET) != 0) {
844 err = got_error_from_errno();
848 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
849 base_pack->packfile);
853 err = resolve_delta_chain(deltas, repo, base_pack->packfile,
854 path_base_packfile, base_offset, base_tslen, base_type,
857 free(path_base_packfile);
861 static const struct got_error *
862 resolve_delta_chain(struct got_delta_chain *deltas, struct got_repository *repo,
863 FILE *packfile, const char *path_packfile, off_t delta_offset, size_t tslen,
864 int delta_type, size_t delta_size)
866 const struct got_error *err = NULL;
868 switch (delta_type) {
869 case GOT_OBJ_TYPE_COMMIT:
870 case GOT_OBJ_TYPE_TREE:
871 case GOT_OBJ_TYPE_BLOB:
872 case GOT_OBJ_TYPE_TAG:
873 /* Plain types are the final delta base. Recursion ends. */
874 err = add_delta(deltas, path_packfile, delta_offset, tslen,
875 delta_type, delta_size, 0, NULL, 0);
877 case GOT_OBJ_TYPE_OFFSET_DELTA:
878 err = resolve_offset_delta(deltas, repo, packfile,
879 path_packfile, delta_offset, tslen, delta_type,
882 case GOT_OBJ_TYPE_REF_DELTA:
883 err = resolve_ref_delta(deltas, repo, packfile,
884 path_packfile, delta_offset, tslen, delta_type,
888 return got_error(GOT_ERR_NOT_IMPL);
894 static const struct got_error *
895 open_delta_object(struct got_object **obj, struct got_repository *repo,
896 struct got_packidx_v2_hdr *packidx, const char *path_packfile,
897 FILE *packfile, struct got_object_id *id, off_t offset, size_t tslen,
898 int delta_type, size_t delta_size)
900 const struct got_error *err = NULL;
903 *obj = calloc(1, sizeof(**obj));
905 return got_error_from_errno();
909 (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
910 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
911 (*obj)->pack_offset = offset + tslen;
913 (*obj)->path_packfile = strdup(path_packfile);
914 if ((*obj)->path_packfile == NULL) {
915 err = got_error_from_errno();
918 (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
920 SIMPLEQ_INIT(&(*obj)->deltas.entries);
921 (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
923 err = resolve_delta_chain(&(*obj)->deltas, repo, packfile,
924 path_packfile, offset, tslen, delta_type, delta_size);
928 err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
931 (*obj)->type = resolved_type;
935 got_object_close(*obj);
941 static const struct got_error *
942 open_packed_object(struct got_object **obj, struct got_repository *repo,
943 struct got_packidx_v2_hdr *packidx, int idx, struct got_object_id *id)
945 const struct got_error *err = NULL;
948 struct got_pack *pack;
955 offset = get_object_offset(packidx, idx);
956 if (offset == (uint64_t)-1)
957 return got_error(GOT_ERR_BAD_PACKIDX);
959 err = get_packfile_path(&path_packfile, repo, packidx);
963 pack = get_cached_pack(path_packfile, repo);
965 err = cache_pack(&pack, path_packfile, packidx, repo);
970 if (fseeko(pack->packfile, offset, SEEK_SET) != 0) {
971 err = got_error_from_errno();
975 err = parse_object_type_and_size(&type, &size, &tslen, pack->packfile);
980 case GOT_OBJ_TYPE_COMMIT:
981 case GOT_OBJ_TYPE_TREE:
982 case GOT_OBJ_TYPE_BLOB:
983 case GOT_OBJ_TYPE_TAG:
984 err = open_plain_object(obj, path_packfile, id, type,
985 offset + tslen, size);
988 case GOT_OBJ_TYPE_OFFSET_DELTA:
989 case GOT_OBJ_TYPE_REF_DELTA:
990 err = open_delta_object(obj, repo, packidx, path_packfile,
991 pack->packfile, id, offset, tslen, type, size);
995 err = got_error(GOT_ERR_NOT_IMPL);
1003 const struct got_error *
1004 got_packfile_open_object(struct got_object **obj, struct got_object_id *id,
1005 struct got_repository *repo)
1007 const struct got_error *err = NULL;
1008 struct got_packidx_v2_hdr *packidx = NULL;
1011 err = search_packidx(&packidx, &idx, repo, id);
1015 err = open_packed_object(obj, repo, packidx, idx, id);
1019 err = cache_pack(NULL, (*obj)->path_packfile, packidx, repo);
1023 static const struct got_error *
1024 get_delta_chain_max_size(uint64_t *max_size, struct got_delta_chain *deltas)
1026 struct got_delta *delta;
1027 uint64_t base_size = 0, result_size = 0;
1030 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1031 /* Plain object types are the delta base. */
1032 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1033 delta->type != GOT_OBJ_TYPE_TREE &&
1034 delta->type != GOT_OBJ_TYPE_BLOB &&
1035 delta->type != GOT_OBJ_TYPE_TAG) {
1036 const struct got_error *err;
1037 err = got_delta_get_sizes(&base_size, &result_size,
1038 delta->delta_buf, delta->delta_len);
1042 base_size = delta->size;
1043 if (base_size > *max_size)
1044 *max_size = base_size;
1045 if (result_size > *max_size)
1046 *max_size = result_size;
1052 static const struct got_error *
1053 dump_delta_chain_to_file(size_t *result_size, struct got_delta_chain *deltas,
1054 FILE *outfile, struct got_repository *repo)
1056 const struct got_error *err = NULL;
1057 struct got_delta *delta;
1058 FILE *base_file = NULL, *accum_file = NULL;
1059 uint8_t *base_buf = NULL, *accum_buf = NULL;
1060 size_t accum_size = 0;
1066 if (SIMPLEQ_EMPTY(&deltas->entries))
1067 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1069 /* We process small enough files entirely in memory for speed. */
1070 err = get_delta_chain_max_size(&max_size, deltas);
1073 if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1074 accum_buf = malloc(max_size);
1075 if (accum_buf == NULL)
1076 return got_error_from_errno();
1078 base_file = got_opentemp();
1079 if (base_file == NULL)
1080 return got_error_from_errno();
1082 accum_file = got_opentemp();
1083 if (accum_file == NULL) {
1084 err = got_error_from_errno();
1090 /* Deltas are ordered in ascending order. */
1091 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1093 struct got_pack *pack;
1096 /* Plain object types are the delta base. */
1097 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1098 delta->type != GOT_OBJ_TYPE_TREE &&
1099 delta->type != GOT_OBJ_TYPE_BLOB &&
1100 delta->type != GOT_OBJ_TYPE_TAG) {
1101 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1105 pack = get_cached_pack(delta->path_packfile, repo);
1107 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1111 if (fseeko(pack->packfile, delta->offset + delta->tslen,
1113 err = got_error_from_errno();
1117 err = got_inflate_to_file(&base_len,
1118 pack->packfile, base_file);
1120 err = got_inflate_to_mem(&base_buf, &base_len,
1122 if (base_len < max_size) {
1124 p = reallocarray(base_buf, 1, max_size);
1126 err = got_error_from_errno();
1141 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1142 delta->delta_len, accum_buf, &accum_size);
1145 err = got_delta_apply(base_file, delta->delta_buf,
1147 /* Final delta application writes to output file. */
1148 ++n < deltas->nentries ? accum_file : outfile,
1154 if (n < deltas->nentries) {
1155 /* Accumulated delta becomes the new base. */
1157 uint8_t *tmp = accum_buf;
1158 accum_buf = base_buf;
1161 FILE *tmp = accum_file;
1162 accum_file = base_file;
1173 size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1175 if (len != accum_size)
1176 return got_ferror(outfile, GOT_ERR_IO);
1184 *result_size = accum_size;
1188 static const struct got_error *
1189 dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1190 struct got_delta_chain *deltas, struct got_repository *repo)
1192 const struct got_error *err = NULL;
1193 struct got_delta *delta;
1194 uint8_t *base_buf = NULL, *accum_buf = NULL;
1202 if (SIMPLEQ_EMPTY(&deltas->entries))
1203 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1205 err = get_delta_chain_max_size(&max_size, deltas);
1208 accum_buf = malloc(max_size);
1209 if (accum_buf == NULL)
1210 return got_error_from_errno();
1212 /* Deltas are ordered in ascending order. */
1213 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1215 struct got_pack *pack;
1218 /* Plain object types are the delta base. */
1219 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1220 delta->type != GOT_OBJ_TYPE_TREE &&
1221 delta->type != GOT_OBJ_TYPE_BLOB &&
1222 delta->type != GOT_OBJ_TYPE_TAG) {
1223 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1227 pack = get_cached_pack(delta->path_packfile, repo);
1229 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1233 if (fseeko(pack->packfile, delta->offset + delta->tslen,
1235 err = got_error_from_errno();
1238 err = got_inflate_to_mem(&base_buf, &base_len,
1240 if (base_len < max_size) {
1242 p = reallocarray(base_buf, 1, max_size);
1244 err = got_error_from_errno();
1255 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1256 delta->delta_len, accum_buf, &accum_size);
1261 if (n < deltas->nentries) {
1262 /* Accumulated delta becomes the new base. */
1263 uint8_t *tmp = accum_buf;
1264 accum_buf = base_buf;
1276 *outbuf = accum_buf;
1277 *outlen = accum_size;
1282 const struct got_error *
1283 got_packfile_extract_object(FILE **f, struct got_object *obj,
1284 struct got_repository *repo)
1286 const struct got_error *err = NULL;
1288 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1289 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1291 *f = got_opentemp();
1293 err = got_error(GOT_ERR_FILE_OPEN);
1297 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1298 struct got_pack *pack;
1300 pack = get_cached_pack(obj->path_packfile, repo);
1302 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1307 if (fseeko(pack->packfile, obj->pack_offset, SEEK_SET) != 0) {
1308 err = got_error_from_errno();
1312 err = got_inflate_to_file(&obj->size, pack->packfile, *f);
1314 err = dump_delta_chain_to_file(&obj->size, &obj->deltas, *f,
1324 const struct got_error *
1325 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1326 struct got_object *obj, struct got_repository *repo)
1328 const struct got_error *err = NULL;
1330 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1331 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1333 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1334 struct got_pack *pack;
1336 pack = get_cached_pack(obj->path_packfile, repo);
1338 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1343 if (fseeko(pack->packfile, obj->pack_offset, SEEK_SET) != 0) {
1344 err = got_error_from_errno();
1348 err = got_inflate_to_mem(buf, len, pack->packfile);
1350 err = dump_delta_chain_to_mem(buf, len, &obj->deltas, repo);