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]);
327 int left = 0, right = totobj - 1;
330 left = betoh32(packidx->hdr.fanout_table[id0 - 1]);
332 while (left <= right) {
333 struct got_object_id *oid;
336 i = ((left + right) / 2);
337 oid = &packidx->hdr.sorted_ids[i];
338 cmp = got_object_id_cmp(id, oid);
351 cache_packidx(struct got_packidx *packidx, struct got_repository *repo)
355 for (i = 0; i < nitems(repo->packidx_cache); i++) {
356 if (repo->packidx_cache[i] == NULL)
360 if (i == nitems(repo->packidx_cache)) {
361 got_packidx_close(repo->packidx_cache[i - 1]);
362 memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
363 sizeof(repo->packidx_cache) -
364 sizeof(repo->packidx_cache[0]));
368 repo->packidx_cache[i] = packidx;
371 static const struct got_error *
372 search_packidx(struct got_packidx **packidx, int *idx,
373 struct got_repository *repo, struct got_object_id *id)
375 const struct got_error *err;
382 /* Search pack index cache. */
383 for (i = 0; i < nitems(repo->packidx_cache); i++) {
384 if (repo->packidx_cache[i] == NULL)
386 *idx = get_object_idx(repo->packidx_cache[i], id, repo);
388 *packidx = repo->packidx_cache[i];
392 /* No luck. Search the filesystem. */
394 path_packdir = got_repo_get_path_objects_pack(repo);
395 if (path_packdir == NULL)
396 return got_error_from_errno();
398 packdir = opendir(path_packdir);
399 if (packdir == NULL) {
400 err = got_error_from_errno();
404 while ((dent = readdir(packdir)) != NULL) {
405 if (!is_packidx_filename(dent->d_name, dent->d_namlen))
408 if (asprintf(&path_packidx, "%s/%s", path_packdir,
409 dent->d_name) == -1) {
410 err = got_error_from_errno();
414 err = got_packidx_open(packidx, path_packidx);
419 *idx = get_object_idx(*packidx, id, repo);
421 err = NULL; /* found the object */
422 cache_packidx(*packidx, repo);
426 got_packidx_close(*packidx);
430 err = got_error(GOT_ERR_NO_OBJ);
433 if (packdir && closedir(packdir) != 0 && err == 0)
434 err = got_error_from_errno();
438 static const struct got_error *
439 get_packfile_path(char **path_packfile, struct got_repository *repo,
440 struct got_packidx *packidx)
444 /* Packfile path contains ".pack" instead of ".idx", so add one byte. */
445 size = strlen(packidx->path_packidx) + 2;
446 if (size < GOT_PACKFILE_NAMELEN + 1)
447 return got_error(GOT_ERR_BAD_PATH);
449 *path_packfile = calloc(size, sizeof(**path_packfile));
450 if (*path_packfile == NULL)
451 return got_error_from_errno();
453 /* Copy up to and excluding ".idx". */
454 if (strlcpy(*path_packfile, packidx->path_packidx,
455 size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
456 return got_error(GOT_ERR_NO_SPACE);
458 if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
459 return got_error(GOT_ERR_NO_SPACE);
464 const struct got_error *
465 read_packfile_hdr(int fd, struct got_packidx *packidx)
467 const struct got_error *err = NULL;
468 uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
469 struct got_packfile_hdr hdr;
472 n = read(fd, &hdr, sizeof(hdr));
474 return got_error_from_errno();
475 if (n != sizeof(hdr))
476 return got_error(GOT_ERR_BAD_PACKFILE);
478 if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
479 betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
480 betoh32(hdr.nobjects) != totobj)
481 err = got_error(GOT_ERR_BAD_PACKFILE);
486 static const struct got_error *
487 open_packfile(int *fd, const char *path_packfile,
488 struct got_repository *repo, struct got_packidx *packidx)
490 const struct got_error *err = NULL;
492 *fd = open(path_packfile, O_RDONLY | O_NOFOLLOW, GOT_DEFAULT_FILE_MODE);
494 return got_error_from_errno();
497 err = read_packfile_hdr(*fd, packidx);
507 got_pack_close(struct got_pack *pack)
511 free(pack->path_packfile);
512 pack->path_packfile = NULL;
516 static const struct got_error *
517 cache_pack(struct got_pack **packp, const char *path_packfile,
518 struct got_packidx *packidx, struct got_repository *repo)
520 const struct got_error *err = NULL;
521 struct got_pack *pack = NULL;
527 for (i = 0; i < nitems(repo->packs); i++) {
528 pack = &repo->packs[i];
529 if (pack->path_packfile == NULL)
531 if (strcmp(pack->path_packfile, path_packfile) == 0)
535 if (i == nitems(repo->packs) - 1) {
536 got_pack_close(&repo->packs[i - 1]);
537 memmove(&repo->packs[1], &repo->packs[0],
538 sizeof(repo->packs) - sizeof(repo->packs[0]));
542 pack = &repo->packs[i];
544 pack->path_packfile = strdup(path_packfile);
545 if (pack->path_packfile == NULL) {
546 err = got_error_from_errno();
550 err = open_packfile(&pack->fd, path_packfile, repo, packidx);
554 err = get_packfile_size(&pack->filesize, path_packfile);
558 free(pack->path_packfile);
559 memset(pack, 0, sizeof(*pack));
567 get_cached_pack(const char *path_packfile, struct got_repository *repo)
569 struct got_pack *pack = NULL;
572 for (i = 0; i < nitems(repo->packs); i++) {
573 pack = &repo->packs[i];
574 if (pack->path_packfile == NULL)
576 if (strcmp(pack->path_packfile, path_packfile) == 0)
583 static const struct got_error *
584 parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len, int fd)
593 /* We do not support size values which don't fit in 64 bit. */
595 return got_error(GOT_ERR_NO_SPACE);
597 n = read(fd, &sizeN, sizeof(sizeN));
599 return got_error_from_errno();
600 if (n != sizeof(sizeN))
601 return got_error(GOT_ERR_BAD_PACKFILE);
604 t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
605 GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
606 s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
608 size_t shift = 4 + 7 * (i - 1);
609 s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
612 } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
616 *len = i * sizeof(sizeN);
620 static const struct got_error *
621 open_plain_object(struct got_object **obj, const char *path_packfile,
622 struct got_object_id *id, uint8_t type, off_t offset, size_t size)
624 *obj = calloc(1, sizeof(**obj));
626 return got_error_from_errno();
628 (*obj)->path_packfile = strdup(path_packfile);
629 if ((*obj)->path_packfile == NULL) {
630 const struct got_error *err = got_error_from_errno();
637 (*obj)->flags = GOT_OBJ_FLAG_PACKED;
640 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
641 (*obj)->pack_offset = offset;
646 static const struct got_error *
647 parse_negative_offset(int64_t *offset, size_t *len, int fd)
655 /* We do not support offset values which don't fit in 64 bit. */
657 return got_error(GOT_ERR_NO_SPACE);
659 n = read(fd, &offN, sizeof(offN));
661 return got_error_from_errno();
662 if (n != sizeof(offN))
663 return got_error(GOT_ERR_BAD_PACKFILE);
666 o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
670 o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
673 } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
676 *len = i * sizeof(offN);
680 static const struct got_error *
681 parse_offset_delta(off_t *base_offset, int fd, off_t offset)
683 const struct got_error *err;
687 err = parse_negative_offset(&negoffset, &negofflen, fd);
691 /* Compute the base object's offset (must be in the same pack file). */
692 *base_offset = (offset - negoffset);
693 if (*base_offset <= 0)
694 return got_error(GOT_ERR_BAD_PACKFILE);
699 static const struct got_error *
700 resolve_delta_chain(struct got_delta_chain *, struct got_repository *,
701 int, size_t, const char *, off_t, size_t, int, size_t, unsigned int);
703 static const struct got_error *
704 add_delta(struct got_delta_chain *deltas, const char *path_packfile,
705 off_t delta_offset, size_t tslen, int delta_type, size_t delta_size,
706 size_t delta_data_offset, uint8_t *delta_buf, size_t delta_len)
708 struct got_delta *delta;
710 delta = got_delta_open(path_packfile, delta_offset, tslen,
711 delta_type, delta_size, delta_data_offset, delta_buf,
714 return got_error_from_errno();
715 /* delta is freed in got_object_close() */
717 SIMPLEQ_INSERT_HEAD(&deltas->entries, delta, entry);
721 static const struct got_error *
722 resolve_offset_delta(struct got_delta_chain *deltas,
723 struct got_repository *repo, int fd, size_t packfile_size,
724 const char *path_packfile, off_t delta_offset, size_t tslen,
725 int delta_type, size_t delta_size, unsigned int recursion)
728 const struct got_error *err;
733 off_t delta_data_offset;
737 err = parse_offset_delta(&base_offset, fd, delta_offset);
741 delta_data_offset = lseek(fd, 0, SEEK_CUR);
742 if (delta_data_offset == -1)
743 return got_error_from_errno();
745 err = got_inflate_to_mem_fd(&delta_buf, &delta_len, fd);
749 err = add_delta(deltas, path_packfile, delta_offset, tslen,
750 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
754 /* An offset delta must be in the same packfile. */
755 if (base_offset >= packfile_size)
756 return got_error(GOT_ERR_PACK_OFFSET);
757 if (lseek(fd, base_offset, SEEK_SET) == -1)
758 return got_error_from_errno();
760 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
765 return resolve_delta_chain(deltas, repo, fd, packfile_size,
766 path_packfile, base_offset, base_tslen, base_type, base_size,
770 static const struct got_error *
771 resolve_ref_delta(struct got_delta_chain *deltas, struct got_repository *repo,
772 int fd, const char *path_packfile, off_t delta_offset,
773 size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
775 const struct got_error *err;
776 struct got_object_id id;
777 struct got_packidx *packidx;
784 char *path_base_packfile;
785 struct got_pack *base_pack;
786 off_t delta_data_offset;
790 n = read(fd, &id, sizeof(id));
792 return got_error_from_errno();
794 return got_error(GOT_ERR_BAD_PACKFILE);
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 err = search_packidx(&packidx, &idx, repo, &id);
813 base_offset = get_object_offset(packidx, idx);
814 if (base_offset == (uint64_t)-1) {
815 return got_error(GOT_ERR_BAD_PACKIDX);
818 err = get_packfile_path(&path_base_packfile, repo, packidx);
822 base_pack = get_cached_pack(path_base_packfile, repo);
823 if (base_pack == NULL) {
824 err = cache_pack(&base_pack, path_base_packfile, NULL, repo);
829 if (base_offset >= base_pack->filesize) {
830 err = got_error(GOT_ERR_PACK_OFFSET);
833 if (lseek(base_pack->fd, base_offset, SEEK_SET) == -1) {
834 err = got_error_from_errno();
838 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
843 err = resolve_delta_chain(deltas, repo, base_pack->fd,
844 base_pack->filesize, path_base_packfile, base_offset,
845 base_tslen, base_type, base_size, recursion - 1);
847 free(path_base_packfile);
851 static const struct got_error *
852 resolve_delta_chain(struct got_delta_chain *deltas, struct got_repository *repo,
853 int fd, size_t packfile_size, const char *path_packfile, off_t delta_offset,
854 size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
856 const struct got_error *err = NULL;
858 if (--recursion == 0)
859 return got_error(GOT_ERR_RECURSION);
861 switch (delta_type) {
862 case GOT_OBJ_TYPE_COMMIT:
863 case GOT_OBJ_TYPE_TREE:
864 case GOT_OBJ_TYPE_BLOB:
865 case GOT_OBJ_TYPE_TAG:
866 /* Plain types are the final delta base. Recursion ends. */
867 err = add_delta(deltas, path_packfile, delta_offset, tslen,
868 delta_type, delta_size, 0, NULL, 0);
870 case GOT_OBJ_TYPE_OFFSET_DELTA:
871 err = resolve_offset_delta(deltas, repo, fd, packfile_size,
872 path_packfile, delta_offset, tslen, delta_type, delta_size,
875 case GOT_OBJ_TYPE_REF_DELTA:
876 err = resolve_ref_delta(deltas, repo, fd, path_packfile,
877 delta_offset, tslen, delta_type, delta_size, recursion - 1);
880 return got_error(GOT_ERR_OBJ_TYPE);
886 static const struct got_error *
887 open_delta_object(struct got_object **obj, struct got_repository *repo,
888 const char *path_packfile, int fd, size_t packfile_size,
889 struct got_object_id *id, off_t offset, size_t tslen,
890 int delta_type, size_t delta_size)
892 const struct got_error *err = NULL;
895 *obj = calloc(1, sizeof(**obj));
897 return got_error_from_errno();
901 (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
902 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
903 (*obj)->pack_offset = offset + tslen;
905 (*obj)->path_packfile = strdup(path_packfile);
906 if ((*obj)->path_packfile == NULL) {
907 err = got_error_from_errno();
910 (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
912 SIMPLEQ_INIT(&(*obj)->deltas.entries);
913 (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
915 err = resolve_delta_chain(&(*obj)->deltas, repo, fd, packfile_size,
916 path_packfile, offset, tslen, delta_type, delta_size,
917 GOT_DELTA_CHAIN_RECURSION_MAX);
921 err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
924 (*obj)->type = resolved_type;
928 got_object_close(*obj);
934 static const struct got_error *
935 open_packed_object(struct got_object **obj, struct got_repository *repo,
936 struct got_packidx *packidx, int idx, struct got_object_id *id)
938 const struct got_error *err = NULL;
941 struct got_pack *pack;
948 offset = get_object_offset(packidx, idx);
949 if (offset == (uint64_t)-1)
950 return got_error(GOT_ERR_BAD_PACKIDX);
952 err = get_packfile_path(&path_packfile, repo, packidx);
956 pack = get_cached_pack(path_packfile, repo);
958 err = cache_pack(&pack, path_packfile, packidx, repo);
963 if (offset >= pack->filesize) {
964 err = got_error(GOT_ERR_PACK_OFFSET);
967 if (lseek(pack->fd, offset, SEEK_SET) == -1) {
968 err = got_error_from_errno();
972 err = parse_object_type_and_size(&type, &size, &tslen, pack->fd);
977 case GOT_OBJ_TYPE_COMMIT:
978 case GOT_OBJ_TYPE_TREE:
979 case GOT_OBJ_TYPE_BLOB:
980 case GOT_OBJ_TYPE_TAG:
981 err = open_plain_object(obj, path_packfile, id, type,
982 offset + tslen, size);
985 case GOT_OBJ_TYPE_OFFSET_DELTA:
986 case GOT_OBJ_TYPE_REF_DELTA:
987 err = open_delta_object(obj, repo, path_packfile, pack->fd,
988 pack->filesize, id, offset, tslen, type, size);
992 err = got_error(GOT_ERR_OBJ_TYPE);
1000 const struct got_error *
1001 got_packfile_open_object(struct got_object **obj, struct got_object_id *id,
1002 struct got_repository *repo)
1004 const struct got_error *err = NULL;
1005 struct got_packidx *packidx = NULL;
1008 err = search_packidx(&packidx, &idx, repo, id);
1012 err = open_packed_object(obj, repo, packidx, idx, id);
1016 err = cache_pack(NULL, (*obj)->path_packfile, packidx, repo);
1020 static const struct got_error *
1021 get_delta_chain_max_size(uint64_t *max_size, struct got_delta_chain *deltas)
1023 struct got_delta *delta;
1024 uint64_t base_size = 0, result_size = 0;
1027 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1028 /* Plain object types are the delta base. */
1029 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1030 delta->type != GOT_OBJ_TYPE_TREE &&
1031 delta->type != GOT_OBJ_TYPE_BLOB &&
1032 delta->type != GOT_OBJ_TYPE_TAG) {
1033 const struct got_error *err;
1034 err = got_delta_get_sizes(&base_size, &result_size,
1035 delta->delta_buf, delta->delta_len);
1039 base_size = delta->size;
1040 if (base_size > *max_size)
1041 *max_size = base_size;
1042 if (result_size > *max_size)
1043 *max_size = result_size;
1049 static const struct got_error *
1050 dump_delta_chain_to_file(size_t *result_size, struct got_delta_chain *deltas,
1051 FILE *outfile, struct got_repository *repo)
1053 const struct got_error *err = NULL;
1054 struct got_delta *delta;
1055 FILE *base_file = NULL, *accum_file = NULL;
1056 uint8_t *base_buf = NULL, *accum_buf = NULL;
1057 size_t accum_size = 0;
1063 if (SIMPLEQ_EMPTY(&deltas->entries))
1064 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1066 /* We process small enough files entirely in memory for speed. */
1067 err = get_delta_chain_max_size(&max_size, deltas);
1070 if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1071 accum_buf = malloc(max_size);
1072 if (accum_buf == NULL)
1073 return got_error_from_errno();
1075 base_file = got_opentemp();
1076 if (base_file == NULL)
1077 return got_error_from_errno();
1079 accum_file = got_opentemp();
1080 if (accum_file == NULL) {
1081 err = got_error_from_errno();
1087 /* Deltas are ordered in ascending order. */
1088 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1090 struct got_pack *pack;
1092 off_t delta_data_offset;
1094 /* Plain object types are the delta base. */
1095 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1096 delta->type != GOT_OBJ_TYPE_TREE &&
1097 delta->type != GOT_OBJ_TYPE_BLOB &&
1098 delta->type != GOT_OBJ_TYPE_TAG) {
1099 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1103 pack = get_cached_pack(delta->path_packfile, repo);
1105 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1109 delta_data_offset = delta->offset + delta->tslen;
1110 if (delta_data_offset >= pack->filesize) {
1111 err = got_error(GOT_ERR_PACK_OFFSET);
1114 if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1115 err = got_error_from_errno();
1119 err = got_inflate_to_file_fd(&base_len,
1120 pack->fd, base_file);
1122 err = got_inflate_to_mem_fd(&base_buf,
1123 &base_len, pack->fd);
1124 if (base_len < max_size) {
1126 p = reallocarray(base_buf, 1, max_size);
1128 err = got_error_from_errno();
1143 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1144 delta->delta_len, accum_buf, &accum_size);
1147 err = got_delta_apply(base_file, delta->delta_buf,
1149 /* Final delta application writes to output file. */
1150 ++n < deltas->nentries ? accum_file : outfile,
1156 if (n < deltas->nentries) {
1157 /* Accumulated delta becomes the new base. */
1159 uint8_t *tmp = accum_buf;
1160 accum_buf = base_buf;
1163 FILE *tmp = accum_file;
1164 accum_file = base_file;
1175 size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1177 if (len != accum_size)
1178 return got_ferror(outfile, GOT_ERR_IO);
1186 *result_size = accum_size;
1190 static const struct got_error *
1191 dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1192 struct got_delta_chain *deltas, struct got_repository *repo)
1194 const struct got_error *err = NULL;
1195 struct got_delta *delta;
1196 uint8_t *base_buf = NULL, *accum_buf = NULL;
1204 if (SIMPLEQ_EMPTY(&deltas->entries))
1205 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1207 err = get_delta_chain_max_size(&max_size, deltas);
1210 accum_buf = malloc(max_size);
1211 if (accum_buf == NULL)
1212 return got_error_from_errno();
1214 /* Deltas are ordered in ascending order. */
1215 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1217 struct got_pack *pack;
1219 size_t delta_data_offset;
1221 /* Plain object types are the delta base. */
1222 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1223 delta->type != GOT_OBJ_TYPE_TREE &&
1224 delta->type != GOT_OBJ_TYPE_BLOB &&
1225 delta->type != GOT_OBJ_TYPE_TAG) {
1226 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1230 pack = get_cached_pack(delta->path_packfile, repo);
1232 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1236 delta_data_offset = delta->offset + delta->tslen;
1237 if (delta_data_offset >= pack->filesize) {
1238 err = got_error(GOT_ERR_PACK_OFFSET);
1241 if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1242 err = got_error_from_errno();
1245 err = got_inflate_to_mem_fd(&base_buf, &base_len,
1247 if (base_len < max_size) {
1249 p = reallocarray(base_buf, 1, max_size);
1251 err = got_error_from_errno();
1262 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1263 delta->delta_len, accum_buf, &accum_size);
1268 if (n < deltas->nentries) {
1269 /* Accumulated delta becomes the new base. */
1270 uint8_t *tmp = accum_buf;
1271 accum_buf = base_buf;
1283 *outbuf = accum_buf;
1284 *outlen = accum_size;
1289 const struct got_error *
1290 got_packfile_extract_object(FILE **f, struct got_object *obj,
1291 struct got_repository *repo)
1293 const struct got_error *err = NULL;
1297 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1298 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1300 *f = got_opentemp();
1302 err = got_error_from_errno();
1306 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1307 struct got_pack *pack;
1309 pack = get_cached_pack(obj->path_packfile, repo);
1311 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1316 if (obj->pack_offset >= pack->filesize) {
1317 err = got_error(GOT_ERR_PACK_OFFSET);
1320 if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1) {
1321 err = got_error_from_errno();
1325 err = got_inflate_to_file_fd(&obj->size, pack->fd, *f);
1327 err = dump_delta_chain_to_file(&obj->size,
1328 &obj->deltas, *f, repo);
1337 const struct got_error *
1338 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1339 struct got_object *obj, struct got_repository *repo)
1341 const struct got_error *err = NULL;
1343 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1344 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1346 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1347 struct got_pack *pack;
1349 pack = get_cached_pack(obj->path_packfile, repo);
1351 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1356 if (obj->pack_offset >= pack->filesize) {
1357 err = got_error(GOT_ERR_PACK_OFFSET);
1360 if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1) {
1361 err = got_error_from_errno();
1365 err = got_inflate_to_mem_fd(buf, len, pack->fd);
1367 err = dump_delta_chain_to_mem(buf, len, &obj->deltas, repo);