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"
36 #include "got_opentemp.h"
38 #include "got_lib_sha1.h"
39 #include "got_lib_pack.h"
40 #include "got_lib_path.h"
41 #include "got_lib_delta.h"
42 #include "got_lib_zbuf.h"
43 #include "got_lib_object.h"
44 #include "got_lib_repository.h"
47 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
50 #define GOT_PACK_PREFIX "pack-"
51 #define GOT_PACKFILE_SUFFIX ".pack"
52 #define GOT_PACKIDX_SUFFIX ".idx"
53 #define GOT_PACKFILE_NAMELEN (strlen(GOT_PACK_PREFIX) + \
54 SHA1_DIGEST_STRING_LENGTH - 1 + \
55 strlen(GOT_PACKFILE_SUFFIX))
56 #define GOT_PACKIDX_NAMELEN (strlen(GOT_PACK_PREFIX) + \
57 SHA1_DIGEST_STRING_LENGTH - 1 + \
58 strlen(GOT_PACKIDX_SUFFIX))
61 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
64 static const struct got_error *
65 verify_fanout_table(uint32_t *fanout_table)
69 for (i = 0; i < 0xff - 1; i++) {
70 if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
71 return got_error(GOT_ERR_BAD_PACKIDX);
77 static const struct got_error *
78 get_packfile_size(size_t *size, const char *path)
85 dot = strrchr(path, '.');
87 return got_error(GOT_ERR_BAD_PATH);
89 /* Path must point to a pack index or to a pack file. */
90 if (strcmp(dot, GOT_PACKIDX_SUFFIX) == 0) {
91 const struct got_error *err = NULL;
93 char base_path[PATH_MAX];
95 /* Convert pack index path to pack file path. */
96 if (strlcpy(base_path, path, PATH_MAX) > PATH_MAX)
97 return got_error(GOT_ERR_NO_SPACE);
98 dot = strrchr(base_path, '.');
100 return got_error(GOT_ERR_BAD_PATH);
102 if (asprintf(&path_pack, "%s.pack", base_path) == -1)
103 return got_error_from_errno();
105 if (stat(path_pack, &sb) != 0)
106 err = got_error_from_errno();
110 } else if (strcmp(dot, GOT_PACKFILE_SUFFIX) == 0) {
111 if (stat(path, &sb) != 0)
112 return got_error_from_errno();
114 return got_error(GOT_ERR_BAD_PATH);
120 const struct got_error *
121 got_packidx_open(struct got_packidx **packidx, const char *path)
123 struct got_packidx *p;
124 struct got_packidx_v2_hdr *h;
126 const struct got_error *err = NULL;
127 size_t n, nobj, packfile_size;
129 uint8_t sha1[SHA1_DIGEST_LENGTH];
135 f = fopen(path, "rb");
137 return got_error_from_errno();
139 err = get_packfile_size(&packfile_size, path);
143 p = calloc(1, sizeof(*p));
145 return got_error_from_errno();
146 p->path_packidx = strdup(path);
147 if (p->path_packidx == NULL) {
148 err = got_error_from_errno();
149 free(p->path_packidx);
155 n = fread(&h->magic, sizeof(h->magic), 1, f);
157 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
161 if (betoh32(h->magic) != GOT_PACKIDX_V2_MAGIC) {
162 err = got_error(GOT_ERR_BAD_PACKIDX);
166 SHA1Update(&ctx, (uint8_t *)&h->magic, sizeof(h->magic));
168 n = fread(&h->version, sizeof(h->version), 1, f);
170 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
174 if (betoh32(h->version) != GOT_PACKIDX_VERSION) {
175 err = got_error(GOT_ERR_BAD_PACKIDX);
179 SHA1Update(&ctx, (uint8_t *)&h->version, sizeof(h->version));
181 n = fread(&h->fanout_table, sizeof(h->fanout_table), 1, f);
183 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
187 err = verify_fanout_table(h->fanout_table);
191 SHA1Update(&ctx, (uint8_t *)h->fanout_table, sizeof(h->fanout_table));
193 nobj = betoh32(h->fanout_table[0xff]);
195 h->sorted_ids = calloc(nobj, sizeof(*h->sorted_ids));
196 if (h->sorted_ids == NULL) {
197 err = got_error_from_errno();
201 n = fread(h->sorted_ids, sizeof(*h->sorted_ids), nobj, f);
203 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
207 SHA1Update(&ctx, (uint8_t *)h->sorted_ids,
208 nobj * sizeof(*h->sorted_ids));
210 h->crc32 = calloc(nobj, sizeof(*h->crc32));
211 if (h->crc32 == NULL) {
212 err = got_error_from_errno();
216 n = fread(h->crc32, sizeof(*h->crc32), nobj, f);
218 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
222 SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
224 h->offsets = calloc(nobj, sizeof(*h->offsets));
225 if (h->offsets == NULL) {
226 err = got_error_from_errno();
230 n = fread(h->offsets, sizeof(*h->offsets), nobj, f);
232 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
236 SHA1Update(&ctx, (uint8_t *)h->offsets, nobj * sizeof(*h->offsets));
238 /* Large file offsets are contained only in files > 2GB. */
239 if (packfile_size <= 0x80000000)
242 h->large_offsets = calloc(nobj, sizeof(*h->large_offsets));
243 if (h->large_offsets == NULL) {
244 err = got_error_from_errno();
248 n = fread(h->large_offsets, sizeof(*h->large_offsets), nobj, f);
250 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
254 SHA1Update(&ctx, (uint8_t*)h->large_offsets,
255 nobj * sizeof(*h->large_offsets));
258 n = fread(&h->trailer, sizeof(h->trailer), 1, f);
260 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
264 SHA1Update(&ctx, h->trailer.packfile_sha1, SHA1_DIGEST_LENGTH);
265 SHA1Final(sha1, &ctx);
266 if (memcmp(h->trailer.packidx_sha1, sha1, SHA1_DIGEST_LENGTH) != 0)
267 err = got_error(GOT_ERR_PACKIDX_CSUM);
271 got_packidx_close(p);
278 got_packidx_close(struct got_packidx *packidx)
280 free(packidx->hdr.sorted_ids);
281 free(packidx->hdr.offsets);
282 free(packidx->hdr.crc32);
283 free(packidx->hdr.large_offsets);
284 free(packidx->path_packidx);
289 is_packidx_filename(const char *name, size_t len)
291 if (len != GOT_PACKIDX_NAMELEN)
294 if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
297 if (strcmp(name + strlen(GOT_PACK_PREFIX) +
298 SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
305 get_object_offset(struct got_packidx *packidx, int idx)
307 uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
308 uint32_t offset = betoh32(packidx->hdr.offsets[idx]);
309 if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
311 idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
312 if (idx < 0 || idx > totobj ||
313 packidx->hdr.large_offsets == NULL)
315 loffset = betoh64(packidx->hdr.large_offsets[idx]);
316 return (loffset > INT64_MAX ? -1 : (off_t)loffset);
318 return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
322 get_object_idx(struct got_packidx *packidx, struct got_object_id *id,
323 struct got_repository *repo)
325 u_int8_t id0 = id->sha1[0];
326 uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
330 i = betoh32(packidx->hdr.fanout_table[id0 - 1]);
333 struct got_object_id *oid = &packidx->hdr.sorted_ids[i];
334 int cmp = got_object_id_cmp(id, oid);
344 static struct got_packidx *
345 dup_packidx(struct got_packidx *packidx)
347 struct got_packidx *p;
350 p = calloc(1, sizeof(*p));
354 p->path_packidx = strdup(packidx->path_packidx);
355 if (p->path_packidx == NULL) {
359 memcpy(&p->hdr, &packidx->hdr, sizeof(p->hdr));
360 p->hdr.sorted_ids = NULL;
362 p->hdr.offsets = NULL;
363 p->hdr.large_offsets = NULL;
365 nobj = betoh32(p->hdr.fanout_table[0xff]);
367 p->hdr.sorted_ids = calloc(nobj, sizeof(*p->hdr.sorted_ids));
368 if (p->hdr.sorted_ids == NULL)
370 memcpy(p->hdr.sorted_ids, packidx->hdr.sorted_ids,
371 nobj * sizeof(*p->hdr.sorted_ids));
373 p->hdr.crc32 = calloc(nobj, sizeof(*p->hdr.crc32));
374 if (p->hdr.crc32 == NULL)
376 memcpy(p->hdr.crc32, packidx->hdr.crc32, nobj * sizeof(*p->hdr.crc32));
378 p->hdr.offsets = calloc(nobj, sizeof(*p->hdr.offsets));
379 if (p->hdr.offsets == NULL)
381 memcpy(p->hdr.offsets, packidx->hdr.offsets,
382 nobj * sizeof(*p->hdr.offsets));
384 if (p->hdr.large_offsets) {
385 p->hdr.large_offsets = calloc(nobj,
386 sizeof(*p->hdr.large_offsets));
387 if (p->hdr.large_offsets == NULL)
389 memcpy(p->hdr.large_offsets, packidx->hdr.large_offsets,
390 nobj * sizeof(*p->hdr.large_offsets));
396 free(p->hdr.large_offsets);
397 free(p->hdr.offsets);
399 free(p->hdr.sorted_ids);
400 free(p->path_packidx);
406 cache_packidx(struct got_packidx *packidx, struct got_repository *repo)
410 for (i = 0; i < nitems(repo->packidx_cache); i++) {
411 if (repo->packidx_cache[i] == NULL)
415 if (i == nitems(repo->packidx_cache)) {
416 got_packidx_close(repo->packidx_cache[i - 1]);
417 memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
418 sizeof(repo->packidx_cache) -
419 sizeof(repo->packidx_cache[0]));
423 repo->packidx_cache[i] = dup_packidx(packidx);
426 static const struct got_error *
427 search_packidx(struct got_packidx **packidx, int *idx,
428 struct got_repository *repo, struct got_object_id *id)
430 const struct got_error *err;
437 /* Search pack index cache. */
438 for (i = 0; i < nitems(repo->packidx_cache); i++) {
439 if (repo->packidx_cache[i] == NULL)
441 *idx = get_object_idx(repo->packidx_cache[i], id, repo);
443 *packidx = repo->packidx_cache[i];
447 /* No luck. Search the filesystem. */
449 path_packdir = got_repo_get_path_objects_pack(repo);
450 if (path_packdir == NULL)
451 return got_error_from_errno();
453 packdir = opendir(path_packdir);
454 if (packdir == NULL) {
455 err = got_error_from_errno();
459 while ((dent = readdir(packdir)) != NULL) {
460 if (!is_packidx_filename(dent->d_name, dent->d_namlen))
463 if (asprintf(&path_packidx, "%s/%s", path_packdir,
464 dent->d_name) == -1) {
465 err = got_error_from_errno();
469 err = got_packidx_open(packidx, path_packidx);
474 *idx = get_object_idx(*packidx, id, repo);
476 err = NULL; /* found the object */
477 cache_packidx(*packidx, repo);
481 got_packidx_close(*packidx);
485 err = got_error(GOT_ERR_NO_OBJ);
488 if (packdir && closedir(packdir) != 0 && err == 0)
489 err = got_error_from_errno();
493 static const struct got_error *
494 get_packfile_path(char **path_packfile, struct got_repository *repo,
495 struct got_packidx *packidx)
499 /* Packfile path contains ".pack" instead of ".idx", so add one byte. */
500 size = strlen(packidx->path_packidx) + 2;
501 if (size < GOT_PACKFILE_NAMELEN + 1)
502 return got_error(GOT_ERR_BAD_PATH);
504 *path_packfile = calloc(size, sizeof(**path_packfile));
505 if (*path_packfile == NULL)
506 return got_error_from_errno();
508 /* Copy up to and excluding ".idx". */
509 if (strlcpy(*path_packfile, packidx->path_packidx,
510 size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
511 return got_error(GOT_ERR_NO_SPACE);
513 if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
514 return got_error(GOT_ERR_NO_SPACE);
519 const struct got_error *
520 read_packfile_hdr(int fd, struct got_packidx *packidx)
522 const struct got_error *err = NULL;
523 uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
524 struct got_packfile_hdr hdr;
527 n = read(fd, &hdr, sizeof(hdr));
529 return got_error_from_errno();
530 if (n != sizeof(hdr))
531 return got_error(GOT_ERR_BAD_PACKFILE);
533 if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
534 betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
535 betoh32(hdr.nobjects) != totobj)
536 err = got_error(GOT_ERR_BAD_PACKFILE);
541 static const struct got_error *
542 open_packfile(int *fd, const char *path_packfile,
543 struct got_repository *repo, struct got_packidx *packidx)
545 const struct got_error *err = NULL;
547 *fd = open(path_packfile, O_RDONLY | O_NOFOLLOW, GOT_DEFAULT_FILE_MODE);
549 return got_error_from_errno();
552 err = read_packfile_hdr(*fd, packidx);
562 got_pack_close(struct got_pack *pack)
566 free(pack->path_packfile);
567 pack->path_packfile = NULL;
571 static const struct got_error *
572 cache_pack(struct got_pack **packp, const char *path_packfile,
573 struct got_packidx *packidx, struct got_repository *repo)
575 const struct got_error *err = NULL;
576 struct got_pack *pack = NULL;
582 for (i = 0; i < nitems(repo->packs); i++) {
583 pack = &repo->packs[i];
584 if (pack->path_packfile == NULL)
586 if (strcmp(pack->path_packfile, path_packfile) == 0)
590 if (i == nitems(repo->packs) - 1) {
591 got_pack_close(&repo->packs[i - 1]);
592 memmove(&repo->packs[1], &repo->packs[0],
593 sizeof(repo->packs) - sizeof(repo->packs[0]));
597 pack = &repo->packs[i];
599 pack->path_packfile = strdup(path_packfile);
600 if (pack->path_packfile == NULL) {
601 err = got_error_from_errno();
605 err = open_packfile(&pack->fd, path_packfile, repo, packidx);
609 err = get_packfile_size(&pack->filesize, path_packfile);
613 free(pack->path_packfile);
614 memset(pack, 0, sizeof(*pack));
622 get_cached_pack(const char *path_packfile, struct got_repository *repo)
624 struct got_pack *pack = NULL;
627 for (i = 0; i < nitems(repo->packs); i++) {
628 pack = &repo->packs[i];
629 if (pack->path_packfile == NULL)
631 if (strcmp(pack->path_packfile, path_packfile) == 0)
638 static const struct got_error *
639 parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len, int fd)
648 /* We do not support size values which don't fit in 64 bit. */
650 return got_error(GOT_ERR_NO_SPACE);
652 n = read(fd, &sizeN, sizeof(sizeN));
654 return got_error_from_errno();
655 if (n != sizeof(sizeN))
656 return got_error(GOT_ERR_BAD_PACKFILE);
659 t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
660 GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
661 s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
663 size_t shift = 4 + 7 * (i - 1);
664 s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
667 } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
671 *len = i * sizeof(sizeN);
675 static const struct got_error *
676 open_plain_object(struct got_object **obj, const char *path_packfile,
677 struct got_object_id *id, uint8_t type, off_t offset, size_t size)
679 *obj = calloc(1, sizeof(**obj));
681 return got_error_from_errno();
683 (*obj)->path_packfile = strdup(path_packfile);
684 if ((*obj)->path_packfile == NULL) {
685 const struct got_error *err = got_error_from_errno();
692 (*obj)->flags = GOT_OBJ_FLAG_PACKED;
695 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
696 (*obj)->pack_offset = offset;
701 static const struct got_error *
702 parse_negative_offset(int64_t *offset, size_t *len, int fd)
710 /* We do not support offset values which don't fit in 64 bit. */
712 return got_error(GOT_ERR_NO_SPACE);
714 n = read(fd, &offN, sizeof(offN));
716 return got_error_from_errno();
717 if (n != sizeof(offN))
718 return got_error(GOT_ERR_BAD_PACKFILE);
721 o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
725 o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
728 } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
731 *len = i * sizeof(offN);
735 static const struct got_error *
736 parse_offset_delta(off_t *base_offset, int fd, off_t offset)
738 const struct got_error *err;
742 err = parse_negative_offset(&negoffset, &negofflen, fd);
746 /* Compute the base object's offset (must be in the same pack file). */
747 *base_offset = (offset - negoffset);
748 if (*base_offset <= 0)
749 return got_error(GOT_ERR_BAD_PACKFILE);
754 static const struct got_error *
755 resolve_delta_chain(struct got_delta_chain *, struct got_repository *,
756 int, size_t, const char *, off_t, size_t, int, size_t, unsigned int);
758 static const struct got_error *
759 add_delta(struct got_delta_chain *deltas, const char *path_packfile,
760 off_t delta_offset, size_t tslen, int delta_type, size_t delta_size,
761 size_t delta_data_offset, uint8_t *delta_buf, size_t delta_len)
763 struct got_delta *delta;
765 delta = got_delta_open(path_packfile, delta_offset, tslen,
766 delta_type, delta_size, delta_data_offset, delta_buf,
769 return got_error_from_errno();
770 /* delta is freed in got_object_close() */
772 SIMPLEQ_INSERT_HEAD(&deltas->entries, delta, entry);
776 static const struct got_error *
777 resolve_offset_delta(struct got_delta_chain *deltas,
778 struct got_repository *repo, int fd, size_t packfile_size,
779 const char *path_packfile, off_t delta_offset, size_t tslen,
780 int delta_type, size_t delta_size, unsigned int recursion)
783 const struct got_error *err;
788 off_t delta_data_offset;
792 err = parse_offset_delta(&base_offset, fd, delta_offset);
796 delta_data_offset = lseek(fd, 0, SEEK_CUR);
797 if (delta_data_offset == -1)
798 return got_error_from_errno();
800 err = got_inflate_to_mem_fd(&delta_buf, &delta_len, fd);
804 err = add_delta(deltas, path_packfile, delta_offset, tslen,
805 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
809 /* An offset delta must be in the same packfile. */
810 if (base_offset >= packfile_size)
811 return got_error(GOT_ERR_PACK_OFFSET);
812 if (lseek(fd, base_offset, SEEK_SET) == -1)
813 return got_error_from_errno();
815 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
820 return resolve_delta_chain(deltas, repo, fd, packfile_size,
821 path_packfile, base_offset, base_tslen, base_type, base_size,
825 static const struct got_error *
826 resolve_ref_delta(struct got_delta_chain *deltas, struct got_repository *repo,
827 int fd, const char *path_packfile, off_t delta_offset,
828 size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
830 const struct got_error *err;
831 struct got_object_id id;
832 struct got_packidx *packidx;
839 char *path_base_packfile;
840 struct got_pack *base_pack;
841 off_t delta_data_offset;
845 n = read(fd, &id, sizeof(id));
847 return got_error_from_errno();
849 return got_error(GOT_ERR_BAD_PACKFILE);
851 delta_data_offset = lseek(fd, 0, SEEK_CUR);
852 if (delta_data_offset == -1)
853 return got_error_from_errno();
855 err = got_inflate_to_mem_fd(&delta_buf, &delta_len, fd);
859 err = add_delta(deltas, path_packfile, delta_offset, tslen,
860 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
864 err = search_packidx(&packidx, &idx, repo, &id);
868 base_offset = get_object_offset(packidx, idx);
869 if (base_offset == (uint64_t)-1) {
870 return got_error(GOT_ERR_BAD_PACKIDX);
873 err = get_packfile_path(&path_base_packfile, repo, packidx);
877 base_pack = get_cached_pack(path_base_packfile, repo);
878 if (base_pack == NULL) {
879 err = cache_pack(&base_pack, path_base_packfile, NULL, repo);
884 if (base_offset >= base_pack->filesize) {
885 err = got_error(GOT_ERR_PACK_OFFSET);
888 if (lseek(base_pack->fd, base_offset, SEEK_SET) == -1) {
889 err = got_error_from_errno();
893 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
898 err = resolve_delta_chain(deltas, repo, base_pack->fd,
899 base_pack->filesize, path_base_packfile, base_offset,
900 base_tslen, base_type, base_size, recursion - 1);
902 free(path_base_packfile);
906 static const struct got_error *
907 resolve_delta_chain(struct got_delta_chain *deltas, struct got_repository *repo,
908 int fd, size_t packfile_size, const char *path_packfile, off_t delta_offset,
909 size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
911 const struct got_error *err = NULL;
913 if (--recursion == 0)
914 return got_error(GOT_ERR_RECURSION);
916 switch (delta_type) {
917 case GOT_OBJ_TYPE_COMMIT:
918 case GOT_OBJ_TYPE_TREE:
919 case GOT_OBJ_TYPE_BLOB:
920 case GOT_OBJ_TYPE_TAG:
921 /* Plain types are the final delta base. Recursion ends. */
922 err = add_delta(deltas, path_packfile, delta_offset, tslen,
923 delta_type, delta_size, 0, NULL, 0);
925 case GOT_OBJ_TYPE_OFFSET_DELTA:
926 err = resolve_offset_delta(deltas, repo, fd, packfile_size,
927 path_packfile, delta_offset, tslen, delta_type, delta_size,
930 case GOT_OBJ_TYPE_REF_DELTA:
931 err = resolve_ref_delta(deltas, repo, fd, path_packfile,
932 delta_offset, tslen, delta_type, delta_size, recursion - 1);
935 return got_error(GOT_ERR_OBJ_TYPE);
941 static const struct got_error *
942 open_delta_object(struct got_object **obj, struct got_repository *repo,
943 const char *path_packfile, int fd, size_t packfile_size,
944 struct got_object_id *id, off_t offset, size_t tslen,
945 int delta_type, size_t delta_size)
947 const struct got_error *err = NULL;
950 *obj = calloc(1, sizeof(**obj));
952 return got_error_from_errno();
956 (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
957 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
958 (*obj)->pack_offset = offset + tslen;
960 (*obj)->path_packfile = strdup(path_packfile);
961 if ((*obj)->path_packfile == NULL) {
962 err = got_error_from_errno();
965 (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
967 SIMPLEQ_INIT(&(*obj)->deltas.entries);
968 (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
970 err = resolve_delta_chain(&(*obj)->deltas, repo, fd, packfile_size,
971 path_packfile, offset, tslen, delta_type, delta_size,
972 GOT_DELTA_CHAIN_RECURSION_MAX);
976 err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
979 (*obj)->type = resolved_type;
983 got_object_close(*obj);
989 static const struct got_error *
990 open_packed_object(struct got_object **obj, struct got_repository *repo,
991 struct got_packidx *packidx, int idx, struct got_object_id *id)
993 const struct got_error *err = NULL;
996 struct got_pack *pack;
1003 offset = get_object_offset(packidx, idx);
1004 if (offset == (uint64_t)-1)
1005 return got_error(GOT_ERR_BAD_PACKIDX);
1007 err = get_packfile_path(&path_packfile, repo, packidx);
1011 pack = get_cached_pack(path_packfile, repo);
1013 err = cache_pack(&pack, path_packfile, packidx, repo);
1018 if (offset >= pack->filesize) {
1019 err = got_error(GOT_ERR_PACK_OFFSET);
1022 if (lseek(pack->fd, offset, SEEK_SET) == -1) {
1023 err = got_error_from_errno();
1027 err = parse_object_type_and_size(&type, &size, &tslen, pack->fd);
1032 case GOT_OBJ_TYPE_COMMIT:
1033 case GOT_OBJ_TYPE_TREE:
1034 case GOT_OBJ_TYPE_BLOB:
1035 case GOT_OBJ_TYPE_TAG:
1036 err = open_plain_object(obj, path_packfile, id, type,
1037 offset + tslen, size);
1040 case GOT_OBJ_TYPE_OFFSET_DELTA:
1041 case GOT_OBJ_TYPE_REF_DELTA:
1042 err = open_delta_object(obj, repo, path_packfile, pack->fd,
1043 pack->filesize, id, offset, tslen, type, size);
1047 err = got_error(GOT_ERR_OBJ_TYPE);
1051 free(path_packfile);
1055 const struct got_error *
1056 got_packfile_open_object(struct got_object **obj, struct got_object_id *id,
1057 struct got_repository *repo)
1059 const struct got_error *err = NULL;
1060 struct got_packidx *packidx = NULL;
1063 err = search_packidx(&packidx, &idx, repo, id);
1067 err = open_packed_object(obj, repo, packidx, idx, id);
1071 err = cache_pack(NULL, (*obj)->path_packfile, packidx, repo);
1075 static const struct got_error *
1076 get_delta_chain_max_size(uint64_t *max_size, struct got_delta_chain *deltas)
1078 struct got_delta *delta;
1079 uint64_t base_size = 0, result_size = 0;
1082 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1083 /* Plain object types are the delta base. */
1084 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1085 delta->type != GOT_OBJ_TYPE_TREE &&
1086 delta->type != GOT_OBJ_TYPE_BLOB &&
1087 delta->type != GOT_OBJ_TYPE_TAG) {
1088 const struct got_error *err;
1089 err = got_delta_get_sizes(&base_size, &result_size,
1090 delta->delta_buf, delta->delta_len);
1094 base_size = delta->size;
1095 if (base_size > *max_size)
1096 *max_size = base_size;
1097 if (result_size > *max_size)
1098 *max_size = result_size;
1104 static const struct got_error *
1105 dump_delta_chain_to_file(size_t *result_size, struct got_delta_chain *deltas,
1106 FILE *outfile, struct got_repository *repo)
1108 const struct got_error *err = NULL;
1109 struct got_delta *delta;
1110 FILE *base_file = NULL, *accum_file = NULL;
1111 uint8_t *base_buf = NULL, *accum_buf = NULL;
1112 size_t accum_size = 0;
1118 if (SIMPLEQ_EMPTY(&deltas->entries))
1119 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1121 /* We process small enough files entirely in memory for speed. */
1122 err = get_delta_chain_max_size(&max_size, deltas);
1125 if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1126 accum_buf = malloc(max_size);
1127 if (accum_buf == NULL)
1128 return got_error_from_errno();
1130 base_file = got_opentemp();
1131 if (base_file == NULL)
1132 return got_error_from_errno();
1134 accum_file = got_opentemp();
1135 if (accum_file == NULL) {
1136 err = got_error_from_errno();
1142 /* Deltas are ordered in ascending order. */
1143 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1145 struct got_pack *pack;
1147 off_t delta_data_offset;
1149 /* Plain object types are the delta base. */
1150 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1151 delta->type != GOT_OBJ_TYPE_TREE &&
1152 delta->type != GOT_OBJ_TYPE_BLOB &&
1153 delta->type != GOT_OBJ_TYPE_TAG) {
1154 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1158 pack = get_cached_pack(delta->path_packfile, repo);
1160 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1164 delta_data_offset = delta->offset + delta->tslen;
1165 if (delta_data_offset >= pack->filesize) {
1166 err = got_error(GOT_ERR_PACK_OFFSET);
1169 if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1170 err = got_error_from_errno();
1174 err = got_inflate_to_file_fd(&base_len,
1175 pack->fd, base_file);
1177 err = got_inflate_to_mem_fd(&base_buf,
1178 &base_len, pack->fd);
1179 if (base_len < max_size) {
1181 p = reallocarray(base_buf, 1, max_size);
1183 err = got_error_from_errno();
1198 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1199 delta->delta_len, accum_buf, &accum_size);
1202 err = got_delta_apply(base_file, delta->delta_buf,
1204 /* Final delta application writes to output file. */
1205 ++n < deltas->nentries ? accum_file : outfile,
1211 if (n < deltas->nentries) {
1212 /* Accumulated delta becomes the new base. */
1214 uint8_t *tmp = accum_buf;
1215 accum_buf = base_buf;
1218 FILE *tmp = accum_file;
1219 accum_file = base_file;
1230 size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1232 if (len != accum_size)
1233 return got_ferror(outfile, GOT_ERR_IO);
1241 *result_size = accum_size;
1245 static const struct got_error *
1246 dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1247 struct got_delta_chain *deltas, struct got_repository *repo)
1249 const struct got_error *err = NULL;
1250 struct got_delta *delta;
1251 uint8_t *base_buf = NULL, *accum_buf = NULL;
1259 if (SIMPLEQ_EMPTY(&deltas->entries))
1260 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1262 err = get_delta_chain_max_size(&max_size, deltas);
1265 accum_buf = malloc(max_size);
1266 if (accum_buf == NULL)
1267 return got_error_from_errno();
1269 /* Deltas are ordered in ascending order. */
1270 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1272 struct got_pack *pack;
1274 size_t delta_data_offset;
1276 /* Plain object types are the delta base. */
1277 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1278 delta->type != GOT_OBJ_TYPE_TREE &&
1279 delta->type != GOT_OBJ_TYPE_BLOB &&
1280 delta->type != GOT_OBJ_TYPE_TAG) {
1281 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1285 pack = get_cached_pack(delta->path_packfile, repo);
1287 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1291 delta_data_offset = delta->offset + delta->tslen;
1292 if (delta_data_offset >= pack->filesize) {
1293 err = got_error(GOT_ERR_PACK_OFFSET);
1296 if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1297 err = got_error_from_errno();
1300 err = got_inflate_to_mem_fd(&base_buf, &base_len,
1302 if (base_len < max_size) {
1304 p = reallocarray(base_buf, 1, max_size);
1306 err = got_error_from_errno();
1317 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1318 delta->delta_len, accum_buf, &accum_size);
1323 if (n < deltas->nentries) {
1324 /* Accumulated delta becomes the new base. */
1325 uint8_t *tmp = accum_buf;
1326 accum_buf = base_buf;
1338 *outbuf = accum_buf;
1339 *outlen = accum_size;
1344 const struct got_error *
1345 got_packfile_extract_object(FILE **f, struct got_object *obj,
1346 struct got_repository *repo)
1348 const struct got_error *err = NULL;
1352 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1353 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1355 *f = got_opentemp();
1357 err = got_error_from_errno();
1361 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1362 struct got_pack *pack;
1364 pack = get_cached_pack(obj->path_packfile, repo);
1366 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1371 if (obj->pack_offset >= pack->filesize) {
1372 err = got_error(GOT_ERR_PACK_OFFSET);
1375 if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1) {
1376 err = got_error_from_errno();
1380 err = got_inflate_to_file_fd(&obj->size, pack->fd, *f);
1382 err = dump_delta_chain_to_file(&obj->size,
1383 &obj->deltas, *f, repo);
1392 const struct got_error *
1393 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1394 struct got_object *obj, struct got_repository *repo)
1396 const struct got_error *err = NULL;
1398 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1399 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1401 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1402 struct got_pack *pack;
1404 pack = get_cached_pack(obj->path_packfile, repo);
1406 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1411 if (obj->pack_offset >= pack->filesize) {
1412 err = got_error(GOT_ERR_PACK_OFFSET);
1415 if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1) {
1416 err = got_error_from_errno();
1420 err = got_inflate_to_mem_fd(buf, len, pack->fd);
1422 err = dump_delta_chain_to_mem(buf, len, &obj->deltas, repo);