2 * Copyright (c) 2018, 2019, 2020 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>
18 #include <sys/queue.h>
21 #include <sys/socket.h>
24 #include <sys/resource.h>
48 #include "got_error.h"
49 #include "got_reference.h"
50 #include "got_repository.h"
52 #include "got_cancel.h"
53 #include "got_object.h"
54 #include "got_opentemp.h"
56 #include "got_lib_delta.h"
57 #include "got_lib_delta_cache.h"
58 #include "got_lib_inflate.h"
59 #include "got_lib_object.h"
60 #include "got_lib_object_parse.h"
61 #include "got_lib_object_create.h"
62 #include "got_lib_pack.h"
63 #include "got_lib_privsep.h"
64 #include "got_lib_hash.h"
65 #include "got_lib_object_cache.h"
66 #include "got_lib_repository.h"
67 #include "got_lib_gotconfig.h"
70 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
73 #define GOT_PACK_NUM_TEMPFILES GOT_PACK_CACHE_SIZE * 2
75 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
76 got_packidx_bloom_filter_cmp);
79 is_boolean_val(const char *val)
81 return (strcasecmp(val, "true") == 0 ||
82 strcasecmp(val, "false") == 0 ||
83 strcasecmp(val, "on") == 0 ||
84 strcasecmp(val, "off") == 0 ||
85 strcasecmp(val, "yes") == 0 ||
86 strcasecmp(val, "no") == 0 ||
87 strcasecmp(val, "1") == 0 ||
88 strcasecmp(val, "0") == 0);
92 get_boolean_val(const char *val)
94 return (strcasecmp(val, "true") == 0 ||
95 strcasecmp(val, "on") == 0 ||
96 strcasecmp(val, "yes") == 0 ||
97 strcasecmp(val, "1") == 0);
101 got_repo_get_path(struct got_repository *repo)
107 got_repo_get_path_git_dir(struct got_repository *repo)
109 return repo->path_git_dir;
113 got_repo_get_fd(struct got_repository *repo)
115 return repo->gitdir_fd;
118 enum got_hash_algorithm
119 got_repo_get_object_format(struct got_repository *repo)
125 got_repo_get_gitconfig_author_name(struct got_repository *repo)
127 return repo->gitconfig_author_name;
131 got_repo_get_gitconfig_author_email(struct got_repository *repo)
133 return repo->gitconfig_author_email;
137 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
139 return repo->global_gitconfig_author_name;
143 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
145 return repo->global_gitconfig_author_email;
149 got_repo_get_gitconfig_owner(struct got_repository *repo)
151 return repo->gitconfig_owner;
155 got_repo_has_extension(struct got_repository *repo, const char *ext)
159 for (i = 0; i < repo->nextensions; ++i) {
160 if (!strcasecmp(ext, repo->extnames[i]))
161 return get_boolean_val(repo->extvals[i]);
168 got_repo_is_bare(struct got_repository *repo)
170 return (strcmp(repo->path, repo->path_git_dir) == 0);
174 get_path_git_child(struct got_repository *repo, const char *basename)
178 if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
186 got_repo_get_path_objects(struct got_repository *repo)
188 return get_path_git_child(repo, GOT_OBJECTS_DIR);
192 got_repo_get_path_objects_pack(struct got_repository *repo)
194 return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
198 got_repo_get_path_refs(struct got_repository *repo)
200 return get_path_git_child(repo, GOT_REFS_DIR);
204 got_repo_get_path_packed_refs(struct got_repository *repo)
206 return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
210 get_path_head(struct got_repository *repo)
212 return get_path_git_child(repo, GOT_HEAD_FILE);
216 got_repo_get_path_gitconfig(struct got_repository *repo)
218 return get_path_git_child(repo, GOT_GITCONFIG);
222 got_repo_get_path_gotconfig(struct got_repository *repo)
224 return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
227 const struct got_gotconfig *
228 got_repo_get_gotconfig(struct got_repository *repo)
230 return repo->gotconfig;
234 got_repo_get_gitconfig_remotes(int *nremotes,
235 const struct got_remote_repo **remotes, struct got_repository *repo)
237 *nremotes = repo->ngitconfig_remotes;
238 *remotes = repo->gitconfig_remotes;
242 is_git_repo(struct got_repository *repo)
244 const char *path_git = got_repo_get_path_git_dir(repo);
245 char *path_objects = got_repo_get_path_objects(repo);
246 char *path_refs = got_repo_get_path_refs(repo);
247 char *path_head = get_path_head(repo);
250 struct got_reference *head_ref;
252 if (lstat(path_git, &sb) == -1)
254 if (!S_ISDIR(sb.st_mode))
257 if (lstat(path_objects, &sb) == -1)
259 if (!S_ISDIR(sb.st_mode))
262 if (lstat(path_refs, &sb) == -1)
264 if (!S_ISDIR(sb.st_mode))
267 if (lstat(path_head, &sb) == -1)
269 if (!S_ISREG(sb.st_mode))
272 /* Check if the HEAD reference can be opened. */
273 if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
275 got_ref_close(head_ref);
286 static const struct got_error *
287 close_tempfiles(int *fds, size_t nfds)
289 const struct got_error *err = NULL;
292 for (i = 0; i < nfds; i++) {
295 if (close(fds[i]) == -1) {
296 err = got_error_from_errno("close");
304 static const struct got_error *
305 open_tempfiles(int **fds, size_t array_size, size_t nfds)
307 const struct got_error *err = NULL;
310 *fds = calloc(array_size, sizeof(**fds));
312 return got_error_from_errno("calloc");
314 for (i = 0; i < array_size; i++)
317 for (i = 0; i < nfds; i++) {
318 (*fds)[i] = got_opentempfd();
319 if ((*fds)[i] == -1) {
320 err = got_error_from_errno("got_opentempfd");
321 close_tempfiles(*fds, nfds);
330 static const struct got_error *
331 get_pack_cache_size(int *pack_cache_size)
335 if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
336 return got_error_from_errno("getrlimit");
338 *pack_cache_size = GOT_PACK_CACHE_SIZE;
339 if (*pack_cache_size > rl.rlim_cur / 8)
340 *pack_cache_size = rl.rlim_cur / 8;
345 const struct got_error *
346 got_repo_pack_fds_open(int **pack_fds)
348 const struct got_error *err;
351 err = get_pack_cache_size(&nfds);
356 * We need one basefd and one accumfd per cached pack.
357 * Our constants should be set up in a way such that
358 * this error never triggers.
360 if (nfds * 2 > GOT_PACK_NUM_TEMPFILES)
361 return got_error(GOT_ERR_NO_SPACE);
363 return open_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES, nfds * 2);
366 const struct got_error *
367 got_repo_pack_fds_close(int *pack_fds)
369 return close_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES);
372 const struct got_error *
373 got_repo_temp_fds_open(int **temp_fds)
375 return open_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES,
376 GOT_REPO_NUM_TEMPFILES);
380 got_repo_temp_fds_set(struct got_repository *repo, int *temp_fds)
384 for (i = 0; i < GOT_REPO_NUM_TEMPFILES; i++)
385 repo->tempfiles[i] = temp_fds[i];
388 const struct got_error *
389 got_repo_temp_fds_get(int *fd, int *idx, struct got_repository *repo)
396 for (i = 0; i < nitems(repo->tempfiles); i++) {
397 if (repo->tempfile_use_mask & (1 << i))
399 if (repo->tempfiles[i] != -1) {
400 if (ftruncate(repo->tempfiles[i], 0L) == -1)
401 return got_error_from_errno("ftruncate");
402 *fd = repo->tempfiles[i];
404 repo->tempfile_use_mask |= (1 << i);
409 return got_error(GOT_ERR_REPO_TEMPFILE);
413 got_repo_temp_fds_put(int idx, struct got_repository *repo)
415 repo->tempfile_use_mask &= ~(1 << idx);
418 const struct got_error *
419 got_repo_temp_fds_close(int *temp_fds)
421 return close_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES);
424 const struct got_error *
425 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
426 struct got_object *obj)
428 #ifndef GOT_NO_OBJ_CACHE
429 const struct got_error *err = NULL;
430 err = got_object_cache_add(&repo->objcache, id, obj);
432 if (err->code == GOT_ERR_OBJ_EXISTS ||
433 err->code == GOT_ERR_OBJ_TOO_LARGE)
443 got_repo_get_cached_object(struct got_repository *repo,
444 struct got_object_id *id)
446 return (struct got_object *)got_object_cache_get(&repo->objcache, id);
449 const struct got_error *
450 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
451 struct got_tree_object *tree)
453 #ifndef GOT_NO_OBJ_CACHE
454 const struct got_error *err = NULL;
455 err = got_object_cache_add(&repo->treecache, id, tree);
457 if (err->code == GOT_ERR_OBJ_EXISTS ||
458 err->code == GOT_ERR_OBJ_TOO_LARGE)
467 struct got_tree_object *
468 got_repo_get_cached_tree(struct got_repository *repo,
469 struct got_object_id *id)
471 return (struct got_tree_object *)got_object_cache_get(
472 &repo->treecache, id);
475 const struct got_error *
476 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
477 struct got_commit_object *commit)
479 #ifndef GOT_NO_OBJ_CACHE
480 const struct got_error *err = NULL;
481 err = got_object_cache_add(&repo->commitcache, id, commit);
483 if (err->code == GOT_ERR_OBJ_EXISTS ||
484 err->code == GOT_ERR_OBJ_TOO_LARGE)
493 struct got_commit_object *
494 got_repo_get_cached_commit(struct got_repository *repo,
495 struct got_object_id *id)
497 return (struct got_commit_object *)got_object_cache_get(
498 &repo->commitcache, id);
501 const struct got_error *
502 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
503 struct got_tag_object *tag)
505 #ifndef GOT_NO_OBJ_CACHE
506 const struct got_error *err = NULL;
507 err = got_object_cache_add(&repo->tagcache, id, tag);
509 if (err->code == GOT_ERR_OBJ_EXISTS ||
510 err->code == GOT_ERR_OBJ_TOO_LARGE)
519 struct got_tag_object *
520 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
522 return (struct got_tag_object *)got_object_cache_get(
523 &repo->tagcache, id);
526 const struct got_error *
527 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
528 struct got_raw_object *raw)
530 #ifndef GOT_NO_OBJ_CACHE
531 const struct got_error *err = NULL;
532 err = got_object_cache_add(&repo->rawcache, id, raw);
534 if (err->code == GOT_ERR_OBJ_EXISTS ||
535 err->code == GOT_ERR_OBJ_TOO_LARGE)
545 struct got_raw_object *
546 got_repo_get_cached_raw_object(struct got_repository *repo,
547 struct got_object_id *id)
549 return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
553 static const struct got_error *
554 open_repo(struct got_repository *repo, const char *path)
556 const struct got_error *err = NULL;
558 repo->gitdir_fd = -1;
560 /* bare git repository? */
561 repo->path_git_dir = strdup(path);
562 if (repo->path_git_dir == NULL)
563 return got_error_from_errno("strdup");
564 if (is_git_repo(repo)) {
565 repo->path = strdup(repo->path_git_dir);
566 if (repo->path == NULL) {
567 err = got_error_from_errno("strdup");
570 repo->gitdir_fd = open(repo->path_git_dir,
571 O_DIRECTORY | O_CLOEXEC);
572 if (repo->gitdir_fd == -1) {
573 err = got_error_from_errno2("open",
580 /* git repository with working tree? */
581 free(repo->path_git_dir);
582 repo->path_git_dir = NULL;
583 if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
584 err = got_error_from_errno("asprintf");
587 if (is_git_repo(repo)) {
588 repo->path = strdup(path);
589 if (repo->path == NULL) {
590 err = got_error_from_errno("strdup");
593 repo->gitdir_fd = open(repo->path_git_dir,
594 O_DIRECTORY | O_CLOEXEC);
595 if (repo->gitdir_fd == -1) {
596 err = got_error_from_errno2("open",
603 err = got_error(GOT_ERR_NOT_GIT_REPO);
608 free(repo->path_git_dir);
609 repo->path_git_dir = NULL;
610 if (repo->gitdir_fd != -1)
611 close(repo->gitdir_fd);
612 repo->gitdir_fd = -1;
618 static const struct got_error *
619 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
621 const struct got_error *err = NULL;
622 char *repo_gitconfig_path = NULL;
624 if (global_gitconfig_path) {
625 /* Read settings from ~/.gitconfig. */
626 int dummy_repo_version;
627 err = got_repo_read_gitconfig(&dummy_repo_version,
628 &repo->global_gitconfig_author_name,
629 &repo->global_gitconfig_author_email,
630 NULL, NULL, NULL, NULL, NULL, NULL,
631 global_gitconfig_path);
636 /* Read repository's .git/config file. */
637 repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
638 if (repo_gitconfig_path == NULL)
639 return got_error_from_errno("got_repo_get_path_gitconfig");
641 err = got_repo_read_gitconfig(
642 &repo->gitconfig_repository_format_version,
643 &repo->gitconfig_author_name, &repo->gitconfig_author_email,
644 &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
645 &repo->gitconfig_owner, &repo->extnames, &repo->extvals,
646 &repo->nextensions, repo_gitconfig_path);
650 if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
653 for (i = 0; i < repo->ngitconfig_remotes; i++) {
654 got_repo_free_remote_repo_data(
655 &repo->gitconfig_remotes[i]);
657 free(repo->gitconfig_remotes);
658 repo->gitconfig_remotes = NULL;
659 repo->ngitconfig_remotes = 0;
661 free(repo->gitconfig_author_name);
662 repo->gitconfig_author_name = NULL;
663 free(repo->gitconfig_author_email);
664 repo->gitconfig_author_email = NULL;
666 free(repo->global_gitconfig_author_name);
667 repo->global_gitconfig_author_name = NULL;
668 free(repo->global_gitconfig_author_email);
669 repo->global_gitconfig_author_email = NULL;
673 free(repo_gitconfig_path);
677 static const struct got_error *
678 read_gotconfig(struct got_repository *repo)
680 const struct got_error *err = NULL;
681 char *gotconfig_path;
683 gotconfig_path = got_repo_get_path_gotconfig(repo);
684 if (gotconfig_path == NULL)
685 return got_error_from_errno("got_repo_get_path_gotconfig");
687 err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
688 free(gotconfig_path);
692 /* Supported repository format extensions. */
693 static const char *const repo_extensions[] = {
694 "noop", /* Got supports repository format version 1. */
695 "preciousObjects", /* Supported by gotadmin cleanup. */
696 "worktreeConfig", /* Got does not care about Git work trees. */
699 const struct got_error *
700 got_repo_open(struct got_repository **repop, const char *path,
701 const char *global_gitconfig_path, int *pack_fds)
703 struct got_repository *repo = NULL;
704 const struct got_error *err = NULL;
705 char *repo_path = NULL;
710 repo = calloc(1, sizeof(*repo));
712 return got_error_from_errno("calloc");
714 RB_INIT(&repo->packidx_bloom_filters);
715 TAILQ_INIT(&repo->packidx_paths);
717 for (i = 0; i < nitems(repo->privsep_children); i++) {
718 memset(&repo->privsep_children[i], 0,
719 sizeof(repo->privsep_children[0]));
720 repo->privsep_children[i].imsg_fd = -1;
723 err = got_object_cache_init(&repo->objcache,
724 GOT_OBJECT_CACHE_TYPE_OBJ);
727 err = got_object_cache_init(&repo->treecache,
728 GOT_OBJECT_CACHE_TYPE_TREE);
731 err = got_object_cache_init(&repo->commitcache,
732 GOT_OBJECT_CACHE_TYPE_COMMIT);
735 err = got_object_cache_init(&repo->tagcache,
736 GOT_OBJECT_CACHE_TYPE_TAG);
739 err = got_object_cache_init(&repo->rawcache,
740 GOT_OBJECT_CACHE_TYPE_RAW);
744 err = get_pack_cache_size(&repo->pack_cache_size);
747 for (i = 0; i < nitems(repo->packs); i++) {
748 if (pack_fds != NULL && i < repo->pack_cache_size) {
749 repo->packs[i].basefd = pack_fds[j++];
750 repo->packs[i].accumfd = pack_fds[j++];
752 repo->packs[i].basefd = -1;
753 repo->packs[i].accumfd = -1;
756 for (i = 0; i < nitems(repo->tempfiles); i++)
757 repo->tempfiles[i] = -1;
758 repo->pinned_pack = -1;
759 repo->pinned_packidx = -1;
760 repo->pinned_pid = 0;
762 repo_path = realpath(path, NULL);
763 if (repo_path == NULL) {
764 err = got_error_from_errno2("realpath", path);
771 err = open_repo(repo, repo_path);
774 if (err->code != GOT_ERR_NOT_GIT_REPO)
776 if (repo_path[0] == '/' && repo_path[1] == '\0') {
777 err = got_error(GOT_ERR_NOT_GIT_REPO);
780 err = got_path_dirname(&parent_path, repo_path);
784 repo_path = parent_path;
787 err = read_gotconfig(repo);
791 err = read_gitconfig(repo, global_gitconfig_path);
794 if (repo->gitconfig_repository_format_version > 1) {
795 err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
798 for (i = 0; i < repo->nextensions; i++) {
799 char *ext = repo->extnames[i];
800 char *val = repo->extvals[i];
801 int j, supported = 0;
803 if (repo->gitconfig_repository_format_version == 1 &&
804 strcasecmp(ext, "objectformat") == 0) {
805 if (strcmp(val, "sha1") == 0)
807 if (strcmp(val, "sha256") == 0) {
808 repo->algo = GOT_HASH_SHA256;
811 err = got_error_path(val, GOT_ERR_OBJECT_FORMAT);
815 if (!is_boolean_val(val)) {
816 err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
820 if (!get_boolean_val(val))
823 for (j = 0; j < nitems(repo_extensions); j++) {
824 if (strcmp(ext, repo_extensions[j]) == 0) {
830 err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
835 err = got_repo_list_packidx(&repo->packidx_paths, repo);
838 got_repo_close(repo);
845 const struct got_error *
846 got_repo_close(struct got_repository *repo)
848 const struct got_error *err = NULL, *child_err;
849 struct got_packidx_bloom_filter *bf;
852 for (i = 0; i < repo->pack_cache_size; i++) {
853 if (repo->packidx_cache[i] == NULL)
855 got_packidx_close(repo->packidx_cache[i]);
858 while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
859 &repo->packidx_bloom_filters))) {
860 RB_REMOVE(got_packidx_bloom_filter_tree,
861 &repo->packidx_bloom_filters, bf);
862 bloom_free(bf->bloom);
867 for (i = 0; i < repo->pack_cache_size; i++)
868 if (repo->packs[i].path_packfile)
869 if (repo->packs[i].path_packfile)
870 got_pack_close(&repo->packs[i]);
873 free(repo->path_git_dir);
875 got_object_cache_close(&repo->objcache);
876 got_object_cache_close(&repo->treecache);
877 got_object_cache_close(&repo->commitcache);
878 got_object_cache_close(&repo->tagcache);
879 got_object_cache_close(&repo->rawcache);
881 for (i = 0; i < nitems(repo->privsep_children); i++) {
882 if (repo->privsep_children[i].imsg_fd == -1)
884 imsg_clear(repo->privsep_children[i].ibuf);
885 free(repo->privsep_children[i].ibuf);
886 err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
887 child_err = got_privsep_wait_for_child(
888 repo->privsep_children[i].pid);
889 if (child_err && err == NULL)
891 if (close(repo->privsep_children[i].imsg_fd) == -1 &&
893 err = got_error_from_errno("close");
896 if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
898 err = got_error_from_errno("close");
901 got_gotconfig_free(repo->gotconfig);
902 free(repo->gitconfig_author_name);
903 free(repo->gitconfig_author_email);
904 for (i = 0; i < repo->ngitconfig_remotes; i++)
905 got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
906 free(repo->gitconfig_remotes);
907 for (i = 0; i < repo->nextensions; i++) {
908 free(repo->extnames[i]);
909 free(repo->extvals[i]);
911 free(repo->extnames);
914 got_pathlist_free(&repo->packidx_paths, GOT_PATHLIST_FREE_PATH);
921 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
927 free(repo->fetch_url);
928 repo->fetch_url = NULL;
929 free(repo->send_url);
930 repo->send_url = NULL;
931 for (i = 0; i < repo->nfetch_branches; i++)
932 free(repo->fetch_branches[i]);
933 free(repo->fetch_branches);
934 repo->fetch_branches = NULL;
935 repo->nfetch_branches = 0;
936 for (i = 0; i < repo->nsend_branches; i++)
937 free(repo->send_branches[i]);
938 free(repo->send_branches);
939 repo->send_branches = NULL;
940 repo->nsend_branches = 0;
943 const struct got_error *
944 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
945 const char *input_path)
947 const struct got_error *err = NULL;
948 const char *repo_abspath = NULL;
950 char *canonpath, *path = NULL;
952 *in_repo_path = NULL;
954 canonpath = strdup(input_path);
955 if (canonpath == NULL) {
956 err = got_error_from_errno("strdup");
959 err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
963 repo_abspath = got_repo_get_path(repo);
965 if (canonpath[0] == '\0') {
966 path = strdup(canonpath);
968 err = got_error_from_errno("strdup");
972 path = realpath(canonpath, NULL);
974 if (errno != ENOENT) {
975 err = got_error_from_errno2("realpath",
980 * Path is not on disk.
981 * Assume it is already relative to repository root.
983 path = strdup(canonpath);
985 err = got_error_from_errno("strdup");
990 repolen = strlen(repo_abspath);
994 if (strcmp(path, repo_abspath) == 0) {
998 err = got_error_from_errno("strdup");
1001 } else if (len > repolen &&
1002 got_path_is_child(path, repo_abspath, repolen)) {
1003 /* Matched an on-disk path inside repository. */
1004 if (got_repo_is_bare(repo)) {
1006 * Matched an on-disk path inside repository
1007 * database. Treat input as repository-relative.
1014 /* Strip common prefix with repository path. */
1015 err = got_path_skip_common_ancestor(&child,
1016 repo_abspath, path);
1024 * Matched unrelated on-disk path.
1025 * Treat input as repository-relative.
1033 /* Make in-repository path absolute */
1034 if (path[0] != '/') {
1036 if (asprintf(&abspath, "/%s", path) == -1) {
1037 err = got_error_from_errno("asprintf");
1049 *in_repo_path = path;
1053 static const struct got_error *
1054 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1055 const char *path_packidx)
1057 const struct got_error *err = NULL;
1060 for (i = 0; i < repo->pack_cache_size; i++) {
1061 if (repo->packidx_cache[i] == NULL)
1063 if (strcmp(repo->packidx_cache[i]->path_packidx,
1064 path_packidx) == 0) {
1065 return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1068 if (i == repo->pack_cache_size) {
1071 } while (i > 0 && repo->pinned_packidx >= 0 &&
1072 i == repo->pinned_packidx);
1073 err = got_packidx_close(repo->packidx_cache[i]);
1078 repo->packidx_cache[i] = packidx;
1084 got_repo_is_packidx_filename(const char *name, size_t len,
1085 enum got_hash_algorithm algo)
1089 idlen = got_hash_digest_string_length(algo);
1091 if (len != GOT_PACKIDX_NAMELEN(idlen))
1094 if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1097 if (strcmp(name + strlen(GOT_PACK_PREFIX) + idlen - 1,
1098 GOT_PACKIDX_SUFFIX) != 0)
1104 static struct got_packidx_bloom_filter *
1105 get_packidx_bloom_filter(struct got_repository *repo,
1106 const char *path, size_t path_len)
1108 struct got_packidx_bloom_filter key;
1110 if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1111 return NULL; /* XXX */
1112 key.path_len = path_len;
1114 return RB_FIND(got_packidx_bloom_filter_tree,
1115 &repo->packidx_bloom_filters, &key);
1119 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1120 const char *path_packidx, struct got_object_id *id)
1122 struct got_packidx_bloom_filter *bf;
1124 bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1126 return bloom_check(bf->bloom, id->hash,
1127 got_hash_digest_length(id->algo));
1129 /* No bloom filter means this pack index must be searched. */
1133 static const struct got_error *
1134 add_packidx_bloom_filter(struct got_repository *repo,
1135 struct got_packidx *packidx, const char *path_packidx)
1137 int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1138 struct got_packidx_bloom_filter *bf;
1141 idlen = got_hash_digest_length(repo->algo);
1144 * Don't use bloom filters for very large pack index files.
1145 * Large pack files will contain a relatively large fraction
1146 * of our objects so we will likely need to visit them anyway.
1147 * The more objects a pack file contains the higher the probability
1148 * of a false-positive match from the bloom filter. And reading
1149 * all object IDs from a large pack index file can be expensive.
1151 if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1154 /* Do we already have a filter for this pack index? */
1155 if (get_packidx_bloom_filter(repo, path_packidx,
1156 strlen(path_packidx)) != NULL)
1159 bf = calloc(1, sizeof(*bf));
1161 return got_error_from_errno("calloc");
1162 bf->bloom = calloc(1, sizeof(*bf->bloom));
1163 if (bf->bloom == NULL) {
1165 return got_error_from_errno("calloc");
1168 len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1169 if (len >= sizeof(bf->path)) {
1172 return got_error(GOT_ERR_NO_SPACE);
1176 /* Minimum size supported by our bloom filter is 1000 entries. */
1177 bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1178 for (i = 0; i < nobjects; i++) {
1179 struct got_packidx_object_id *id;
1180 id = packidx->hdr.sorted_ids + i * idlen;
1181 bloom_add(bf->bloom, id->hash, idlen);
1184 RB_INSERT(got_packidx_bloom_filter_tree,
1185 &repo->packidx_bloom_filters, bf);
1190 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1192 struct got_pathlist_entry *pe;
1194 while (!TAILQ_EMPTY(packidx_paths)) {
1195 pe = TAILQ_FIRST(packidx_paths);
1196 TAILQ_REMOVE(packidx_paths, pe, entry);
1197 free((char *)pe->path);
1202 static const struct got_error *
1203 refresh_packidx_paths(struct got_repository *repo)
1205 const struct got_error *err = NULL;
1206 char *objects_pack_dir = NULL;
1209 objects_pack_dir = got_repo_get_path_objects_pack(repo);
1210 if (objects_pack_dir == NULL)
1211 return got_error_from_errno("got_repo_get_path_objects_pack");
1213 if (stat(objects_pack_dir, &sb) == -1) {
1214 if (errno != ENOENT) {
1215 err = got_error_from_errno2("stat", objects_pack_dir);
1218 } else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1219 sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1220 sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1221 purge_packidx_paths(&repo->packidx_paths);
1222 err = got_repo_list_packidx(&repo->packidx_paths, repo);
1227 free(objects_pack_dir);
1231 const struct got_error *
1232 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1233 struct got_repository *repo, struct got_object_id *id)
1235 const struct got_error *err;
1236 struct got_pathlist_entry *pe;
1239 /* Search pack index cache. */
1240 for (i = 0; i < repo->pack_cache_size; i++) {
1241 if (repo->packidx_cache[i] == NULL)
1243 if (!got_repo_check_packidx_bloom_filter(repo,
1244 repo->packidx_cache[i]->path_packidx, id))
1245 continue; /* object will not be found in this index */
1246 *idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1248 *packidx = repo->packidx_cache[i];
1250 * Move this cache entry to the front. Repeatedly
1251 * searching a wrong pack index can be expensive.
1254 memmove(&repo->packidx_cache[1],
1255 &repo->packidx_cache[0],
1256 i * sizeof(repo->packidx_cache[0]));
1257 repo->packidx_cache[0] = *packidx;
1258 if (repo->pinned_packidx >= 0 &&
1259 repo->pinned_packidx < i)
1260 repo->pinned_packidx++;
1261 else if (repo->pinned_packidx == i)
1262 repo->pinned_packidx = 0;
1267 /* No luck. Search the filesystem. */
1269 err = refresh_packidx_paths(repo);
1273 TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1274 const char *path_packidx = pe->path;
1277 if (!got_repo_check_packidx_bloom_filter(repo,
1279 continue; /* object will not be found in this index */
1281 for (i = 0; i < repo->pack_cache_size; i++) {
1282 if (repo->packidx_cache[i] == NULL)
1284 if (strcmp(repo->packidx_cache[i]->path_packidx,
1285 path_packidx) == 0) {
1291 continue; /* already searched */
1293 err = got_packidx_open(packidx, got_repo_get_fd(repo),
1294 path_packidx, 0, repo->algo);
1298 err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1302 err = cache_packidx(repo, *packidx, path_packidx);
1306 *idx = got_packidx_get_object_idx(*packidx, id);
1308 err = NULL; /* found the object */
1313 err = got_error_no_obj(id);
1318 const struct got_error *
1319 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1320 struct got_repository *repo)
1322 const struct got_error *err = NULL;
1323 DIR *packdir = NULL;
1324 struct dirent *dent;
1325 char *path_packidx = NULL;
1329 packdir_fd = openat(got_repo_get_fd(repo),
1330 GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1331 if (packdir_fd == -1) {
1332 return got_error_from_errno_fmt("openat: %s/%s",
1333 got_repo_get_path_git_dir(repo),
1334 GOT_OBJECTS_PACK_DIR);
1337 packdir = fdopendir(packdir_fd);
1338 if (packdir == NULL) {
1339 err = got_error_from_errno("fdopendir");
1343 if (fstat(packdir_fd, &sb) == -1) {
1344 err = got_error_from_errno("fstat");
1347 repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1348 repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1350 while ((dent = readdir(packdir)) != NULL) {
1351 if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen,
1355 if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1356 dent->d_name) == -1) {
1357 err = got_error_from_errno("asprintf");
1358 path_packidx = NULL;
1362 err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1369 if (packdir && closedir(packdir) != 0 && err == NULL)
1370 err = got_error_from_errno("closedir");
1374 const struct got_error *
1375 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1376 struct got_repository *repo)
1378 const struct got_error *err;
1383 /* Search pack index cache. */
1384 for (i = 0; i < repo->pack_cache_size; i++) {
1385 if (repo->packidx_cache[i] == NULL)
1387 if (strcmp(repo->packidx_cache[i]->path_packidx,
1388 path_packidx) == 0) {
1389 *packidx = repo->packidx_cache[i];
1393 /* No luck. Search the filesystem. */
1395 err = got_packidx_open(packidx, got_repo_get_fd(repo),
1396 path_packidx, 0, repo->algo);
1400 err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1404 err = cache_packidx(repo, *packidx, path_packidx);
1407 got_packidx_close(*packidx);
1413 static const struct got_error *
1414 read_packfile_hdr(int fd, struct got_packidx *packidx)
1416 const struct got_error *err = NULL;
1417 uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1418 struct got_packfile_hdr hdr;
1421 n = read(fd, &hdr, sizeof(hdr));
1423 return got_error_from_errno("read");
1424 if (n != sizeof(hdr))
1425 return got_error(GOT_ERR_BAD_PACKFILE);
1427 if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1428 be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1429 be32toh(hdr.nobjects) != totobj)
1430 err = got_error(GOT_ERR_BAD_PACKFILE);
1435 static const struct got_error *
1436 open_packfile(int *fd, struct got_repository *repo,
1437 const char *relpath, struct got_packidx *packidx)
1439 const struct got_error *err = NULL;
1441 *fd = openat(got_repo_get_fd(repo), relpath,
1442 O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1444 return got_error_from_errno_fmt("openat: %s/%s",
1445 got_repo_get_path_git_dir(repo), relpath);
1448 err = read_packfile_hdr(*fd, packidx);
1458 const struct got_error *
1459 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1460 const char *path_packfile, struct got_packidx *packidx)
1462 const struct got_error *err = NULL;
1463 struct got_pack *pack = NULL;
1470 for (i = 0; i < repo->pack_cache_size; i++) {
1471 pack = &repo->packs[i];
1472 if (pack->path_packfile == NULL)
1474 if (strcmp(pack->path_packfile, path_packfile) == 0)
1475 return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1478 if (i == repo->pack_cache_size) {
1479 struct got_pack tmp;
1482 } while (i > 0 && repo->pinned_pack >= 0 &&
1483 i == repo->pinned_pack);
1484 err = got_pack_close(&repo->packs[i]);
1487 if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1488 return got_error_from_errno("ftruncate");
1489 if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1490 return got_error_from_errno("ftruncate");
1491 memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1492 memcpy(&repo->packs[i], &repo->packs[0],
1493 sizeof(repo->packs[i]));
1494 memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1495 if (repo->pinned_pack == 0)
1496 repo->pinned_pack = i;
1497 else if (repo->pinned_pack == i)
1498 repo->pinned_pack = 0;
1502 pack = &repo->packs[i];
1504 pack->path_packfile = strdup(path_packfile);
1505 if (pack->path_packfile == NULL) {
1506 err = got_error_from_errno("strdup");
1510 err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1514 if (fstat(pack->fd, &sb) != 0) {
1515 err = got_error_from_errno("fstat");
1518 pack->filesize = sb.st_size;
1519 pack->algo = got_repo_get_object_format(repo);
1521 pack->privsep_child = NULL;
1523 err = got_delta_cache_alloc(&pack->delta_cache);
1527 #ifndef GOT_PACK_NO_MMAP
1528 if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1529 pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1531 if (pack->map == MAP_FAILED) {
1532 if (errno != ENOMEM) {
1533 err = got_error_from_errno("mmap");
1536 pack->map = NULL; /* fall back to read(2) */
1543 got_pack_close(pack);
1550 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1552 struct got_pack *pack = NULL;
1555 for (i = 0; i < repo->pack_cache_size; i++) {
1556 pack = &repo->packs[i];
1557 if (pack->path_packfile == NULL)
1559 if (strcmp(pack->path_packfile, path_packfile) == 0)
1566 const struct got_error *
1567 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1568 struct got_pack *pack)
1571 int pinned_pack = -1, pinned_packidx = -1;
1573 for (i = 0; i < repo->pack_cache_size; i++) {
1574 if (repo->packidx_cache[i] &&
1575 strcmp(repo->packidx_cache[i]->path_packidx,
1576 packidx->path_packidx) == 0)
1578 if (repo->packs[i].path_packfile &&
1579 strcmp(repo->packs[i].path_packfile,
1580 pack->path_packfile) == 0)
1584 if (pinned_packidx == -1 || pinned_pack == -1)
1585 return got_error(GOT_ERR_PIN_PACK);
1587 repo->pinned_pack = pinned_pack;
1588 repo->pinned_packidx = pinned_packidx;
1589 if (repo->packs[pinned_pack].privsep_child)
1590 repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1595 got_repo_get_pinned_pack(struct got_repository *repo)
1597 if (repo->pinned_pack >= 0 &&
1598 repo->pinned_pack < repo->pack_cache_size)
1599 return &repo->packs[repo->pinned_pack];
1605 got_repo_unpin_pack(struct got_repository *repo)
1607 repo->pinned_packidx = -1;
1608 repo->pinned_pack = -1;
1609 repo->pinned_pid = 0;
1612 const struct got_error *
1613 got_repo_init(const char *repo_path, const char *head_name)
1615 const struct got_error *err = NULL;
1616 const char *dirnames[] = {
1618 GOT_OBJECTS_PACK_DIR,
1621 const char *description_str = "Unnamed repository; "
1622 "edit this file 'description' to name the repository.";
1623 const char *headref = "ref: refs/heads/";
1624 const char *gitconfig_str = "[core]\n"
1625 "\trepositoryformatversion = 0\n"
1626 "\tfilemode = true\n"
1628 char *headref_str, *path;
1631 if (!got_path_dir_is_empty(repo_path))
1632 return got_error(GOT_ERR_DIR_NOT_EMPTY);
1634 for (i = 0; i < nitems(dirnames); i++) {
1635 if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1636 return got_error_from_errno("asprintf");
1638 err = got_path_mkdir(path);
1644 if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1645 return got_error_from_errno("asprintf");
1646 err = got_path_create_file(path, description_str);
1651 if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1652 return got_error_from_errno("asprintf");
1653 if (asprintf(&headref_str, "%s%s", headref,
1654 head_name ? head_name : "main") == -1) {
1656 return got_error_from_errno("asprintf");
1658 err = got_path_create_file(path, headref_str);
1664 if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1665 return got_error_from_errno("asprintf");
1666 err = got_path_create_file(path, gitconfig_str);
1674 static const struct got_error *
1675 match_packed_object(struct got_object_id **unique_id,
1676 struct got_repository *repo, const char *id_str_prefix, int obj_type)
1678 const struct got_error *err = NULL;
1679 struct got_object_id_queue matched_ids;
1680 struct got_pathlist_entry *pe;
1682 STAILQ_INIT(&matched_ids);
1684 err = refresh_packidx_paths(repo);
1688 TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1689 const char *path_packidx = pe->path;
1690 struct got_packidx *packidx;
1691 struct got_object_qid *qid;
1693 err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1694 path_packidx, 0, repo->algo);
1698 err = got_packidx_match_id_str_prefix(&matched_ids,
1699 packidx, id_str_prefix);
1701 got_packidx_close(packidx);
1704 err = got_packidx_close(packidx);
1708 STAILQ_FOREACH(qid, &matched_ids, entry) {
1709 if (obj_type != GOT_OBJ_TYPE_ANY) {
1711 err = got_object_get_type(&matched_type, repo,
1715 if (matched_type != obj_type)
1718 if (*unique_id == NULL) {
1719 *unique_id = got_object_id_dup(&qid->id);
1720 if (*unique_id == NULL) {
1721 err = got_error_from_errno("malloc");
1725 if (got_object_id_cmp(*unique_id,
1727 continue; /* packed multiple times */
1728 err = got_error(GOT_ERR_AMBIGUOUS_ID);
1734 got_object_id_queue_free(&matched_ids);
1742 static const struct got_error *
1743 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1744 const char *object_dir, const char *id_str_prefix, int obj_type,
1745 struct got_repository *repo)
1747 const struct got_error *err = NULL;
1748 char *path, *id_str = NULL;
1750 struct dirent *dent;
1751 struct got_object_id id;
1753 if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1754 err = got_error_from_errno("asprintf");
1758 dir = opendir(path);
1760 if (errno == ENOENT) {
1764 err = got_error_from_errno2("opendir", path);
1767 while ((dent = readdir(dir)) != NULL) {
1773 if (strcmp(dent->d_name, ".") == 0 ||
1774 strcmp(dent->d_name, "..") == 0)
1777 if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1778 err = got_error_from_errno("asprintf");
1783 memset(&id, 0, sizeof(id));
1784 id.algo = repo->algo;
1785 if (!got_parse_hash_digest(id.hash, id_str, repo->algo))
1789 * Directory entries do not necessarily appear in
1790 * sorted order, so we must iterate over all of them.
1792 cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1796 if (*unique_id == NULL) {
1797 if (obj_type != GOT_OBJ_TYPE_ANY) {
1799 err = got_object_get_type(&matched_type, repo,
1803 if (matched_type != obj_type)
1806 *unique_id = got_object_id_dup(&id);
1807 if (*unique_id == NULL) {
1808 err = got_error_from_errno("got_object_id_dup");
1812 if (got_object_id_cmp(*unique_id, &id) == 0)
1813 continue; /* both packed and loose */
1814 err = got_error(GOT_ERR_AMBIGUOUS_ID);
1819 if (dir && closedir(dir) != 0 && err == NULL)
1820 err = got_error_from_errno("closedir");
1830 const struct got_error *
1831 got_repo_match_object_id_prefix(struct got_object_id **id,
1832 const char *id_str_prefix, int obj_type, struct got_repository *repo)
1834 const struct got_error *err = NULL;
1835 char *path_objects = NULL, *object_dir = NULL;
1841 path_objects = got_repo_get_path_objects(repo);
1843 len = strlen(id_str_prefix);
1844 if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1845 err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1849 for (i = 0; i < len; i++) {
1850 if (isxdigit((unsigned char)id_str_prefix[i]))
1852 err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1857 err = match_packed_object(id, repo, id_str_prefix, obj_type);
1860 object_dir = strndup(id_str_prefix, 2);
1861 if (object_dir == NULL) {
1862 err = got_error_from_errno("strdup");
1865 err = match_loose_object(id, path_objects, object_dir,
1866 id_str_prefix, obj_type, repo);
1867 } else if (len == 1) {
1869 for (i = 0; i < 0xf; i++) {
1870 if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1872 err = got_error_from_errno("asprintf");
1875 err = match_packed_object(id, repo, object_dir,
1879 err = match_loose_object(id, path_objects, object_dir,
1880 id_str_prefix, obj_type, repo);
1885 err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1894 } else if (*id == NULL) {
1896 case GOT_OBJ_TYPE_BLOB:
1897 err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1898 GOT_OBJ_LABEL_BLOB, id_str_prefix);
1900 case GOT_OBJ_TYPE_TREE:
1901 err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1902 GOT_OBJ_LABEL_TREE, id_str_prefix);
1904 case GOT_OBJ_TYPE_COMMIT:
1905 err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1906 GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1908 case GOT_OBJ_TYPE_TAG:
1909 err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1910 GOT_OBJ_LABEL_TAG, id_str_prefix);
1913 err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1921 const struct got_error *
1922 got_repo_match_object_id(struct got_object_id **id, char **label,
1923 const char *id_str, int obj_type, struct got_reflist_head *refs,
1924 struct got_repository *repo)
1926 const struct got_error *err;
1927 struct got_tag_object *tag;
1928 struct got_reference *ref = NULL;
1935 err = got_repo_object_match_tag(&tag, id_str, obj_type,
1938 *id = got_object_id_dup(
1939 got_object_tag_get_object_id(tag));
1941 err = got_error_from_errno("got_object_id_dup");
1942 else if (label && asprintf(label, "refs/tags/%s",
1943 got_object_tag_get_name(tag)) == -1) {
1944 err = got_error_from_errno("asprintf");
1948 got_object_tag_close(tag);
1950 } else if (err->code != GOT_ERR_OBJ_TYPE &&
1951 err->code != GOT_ERR_NO_OBJ)
1955 err = got_ref_open(&ref, repo, id_str, 0);
1957 err = got_ref_resolve(id, repo, ref);
1961 *label = strdup(got_ref_get_name(ref));
1962 if (*label == NULL) {
1963 err = got_error_from_errno("strdup");
1968 if (err->code != GOT_ERR_NOT_REF &&
1969 err->code != GOT_ERR_BAD_REF_NAME)
1971 err = got_repo_match_object_id_prefix(id, id_str,
1974 if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1975 err = got_error_not_ref(id_str);
1979 err = got_object_id_str(label, *id);
1980 if (*label == NULL) {
1981 err = got_error_from_errno("strdup");
1992 const struct got_error *
1993 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1994 int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1996 const struct got_error *err = NULL;
1997 struct got_reflist_entry *re;
1998 struct got_object_id *tag_id;
1999 int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
2003 TAILQ_FOREACH(re, refs, entry) {
2004 const char *refname;
2005 refname = got_ref_get_name(re->ref);
2006 if (got_ref_is_symbolic(re->ref))
2008 if (strncmp(refname, "refs/tags/", 10) != 0)
2010 if (!name_is_absolute)
2011 refname += strlen("refs/tags/");
2012 if (strcmp(refname, name) != 0)
2014 err = got_ref_resolve(&tag_id, repo, re->ref);
2017 err = got_object_open_as_tag(tag, repo, tag_id);
2021 if (obj_type == GOT_OBJ_TYPE_ANY ||
2022 got_object_tag_get_object_type(*tag) == obj_type)
2024 got_object_tag_close(*tag);
2028 if (err == NULL && *tag == NULL)
2029 err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
2030 GOT_OBJ_LABEL_TAG, name);
2034 static const struct got_error *
2035 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
2036 const char *name, mode_t mode, struct got_object_id *blob_id)
2038 const struct got_error *err = NULL;
2042 *new_te = calloc(1, sizeof(**new_te));
2043 if (*new_te == NULL)
2044 return got_error_from_errno("calloc");
2046 if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2047 sizeof((*new_te)->name)) {
2048 err = got_error(GOT_ERR_NO_SPACE);
2052 if (S_ISLNK(mode)) {
2053 (*new_te)->mode = S_IFLNK;
2055 (*new_te)->mode = S_IFREG;
2056 (*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2058 memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2060 if (err && *new_te) {
2067 static const struct got_error *
2068 import_file(struct got_tree_entry **new_te, struct dirent *de,
2069 const char *path, struct got_repository *repo)
2071 const struct got_error *err;
2072 struct got_object_id *blob_id = NULL;
2076 if (asprintf(&filepath, "%s%s%s", path,
2077 path[0] == '\0' ? "" : "/", de->d_name) == -1)
2078 return got_error_from_errno("asprintf");
2080 if (lstat(filepath, &sb) != 0) {
2081 err = got_error_from_errno2("lstat", path);
2085 err = got_object_blob_create(&blob_id, filepath, repo);
2089 err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2098 static const struct got_error *
2099 insert_tree_entry(struct got_tree_entry *new_te,
2100 struct got_pathlist_head *paths)
2102 const struct got_error *err = NULL;
2103 struct got_pathlist_entry *new_pe;
2105 err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2109 return got_error(GOT_ERR_TREE_DUP_ENTRY);
2113 static const struct got_error *write_tree(struct got_object_id **,
2114 const char *, struct got_pathlist_head *, struct got_repository *,
2115 got_repo_import_cb progress_cb, void *progress_arg);
2117 static const struct got_error *
2118 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2119 const char *path, struct got_pathlist_head *ignores,
2120 struct got_repository *repo,
2121 got_repo_import_cb progress_cb, void *progress_arg)
2123 const struct got_error *err;
2124 struct got_object_id *id = NULL;
2127 if (asprintf(&subdirpath, "%s%s%s", path,
2128 path[0] == '\0' ? "" : "/", de->d_name) == -1)
2129 return got_error_from_errno("asprintf");
2131 (*new_te) = calloc(1, sizeof(**new_te));
2132 if (*new_te == NULL)
2133 return got_error_from_errno("calloc");
2134 (*new_te)->mode = S_IFDIR;
2135 if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2136 sizeof((*new_te)->name)) {
2137 err = got_error(GOT_ERR_NO_SPACE);
2140 err = write_tree(&id, subdirpath, ignores, repo,
2141 progress_cb, progress_arg);
2144 memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2156 static const struct got_error *
2157 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2158 struct got_pathlist_head *ignores, struct got_repository *repo,
2159 got_repo_import_cb progress_cb, void *progress_arg)
2161 const struct got_error *err = NULL;
2165 struct got_tree_entry *new_te = NULL;
2166 struct got_pathlist_head paths;
2167 struct got_pathlist_entry *pe;
2169 *new_tree_id = NULL;
2173 dir = opendir(path_dir);
2175 err = got_error_from_errno2("opendir", path_dir);
2180 while ((de = readdir(dir)) != NULL) {
2184 if (strcmp(de->d_name, ".") == 0 ||
2185 strcmp(de->d_name, "..") == 0)
2188 err = got_path_dirent_type(&type, path_dir, de);
2192 TAILQ_FOREACH(pe, ignores, entry) {
2193 if (type == DT_DIR && pe->path_len > 0 &&
2194 pe->path[pe->path_len - 1] == '/') {
2195 char stripped[PATH_MAX];
2197 if (strlcpy(stripped, pe->path,
2198 sizeof(stripped)) >= sizeof(stripped)) {
2199 err = got_error(GOT_ERR_NO_SPACE);
2202 got_path_strip_trailing_slashes(stripped);
2203 if (fnmatch(stripped, de->d_name, 0) == 0) {
2207 } else if (fnmatch(pe->path, de->d_name, 0) == 0) {
2215 if (type == DT_DIR) {
2216 err = import_subdir(&new_te, de, path_dir,
2217 ignores, repo, progress_cb, progress_arg);
2219 if (err->code != GOT_ERR_NO_TREE_ENTRY)
2224 } else if (type == DT_REG || type == DT_LNK) {
2225 err = import_file(&new_te, de, path_dir, repo);
2231 err = insert_tree_entry(new_te, &paths);
2237 if (TAILQ_EMPTY(&paths)) {
2238 err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2239 "cannot create tree without any entries");
2243 TAILQ_FOREACH(pe, &paths, entry) {
2244 struct got_tree_entry *te = pe->data;
2246 if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2248 if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2249 err = got_error_from_errno("asprintf");
2252 err = (*progress_cb)(progress_arg, path);
2258 err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2262 got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2266 const struct got_error *
2267 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2268 const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2269 struct got_repository *repo, got_repo_import_cb progress_cb,
2272 const struct got_error *err;
2273 struct got_object_id *new_tree_id;
2275 err = write_tree(&new_tree_id, path_dir, ignores, repo,
2276 progress_cb, progress_arg);
2280 err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2281 author, time(NULL), author, time(NULL), logmsg, repo);
2286 const struct got_error *
2287 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2288 struct got_repository *repo)
2290 const struct got_error *err = NULL;
2291 char *path_objects = NULL, *path = NULL;
2293 struct got_object_id id;
2299 path_objects = got_repo_get_path_objects(repo);
2300 if (path_objects == NULL)
2301 return got_error_from_errno("got_repo_get_path_objects");
2303 for (i = 0; i <= 0xff; i++) {
2304 struct dirent *dent;
2306 if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2307 err = got_error_from_errno("asprintf");
2311 dir = opendir(path);
2313 if (errno == ENOENT) {
2317 err = got_error_from_errno2("opendir", path);
2321 while ((dent = readdir(dir)) != NULL) {
2326 if (strcmp(dent->d_name, ".") == 0 ||
2327 strcmp(dent->d_name, "..") == 0)
2330 if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2331 err = got_error_from_errno("asprintf");
2335 memset(&id, 0, sizeof(id));
2336 id.algo = repo->algo;
2337 if (!got_parse_hash_digest(id.hash, id_str,
2344 err = got_object_open_loose_fd(&fd, &id, repo);
2348 if (fstat(fd, &sb) == -1) {
2349 err = got_error_from_errno("fstat");
2354 (*ondisk_size) += sb.st_size;
2356 if (close(fd) == -1) {
2357 err = got_error_from_errno("close");
2362 if (closedir(dir) != 0) {
2363 err = got_error_from_errno("closedir");
2372 if (dir && closedir(dir) != 0 && err == NULL)
2373 err = got_error_from_errno("closedir");
2384 const struct got_error *
2385 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2386 off_t *total_packsize, struct got_repository *repo)
2388 const struct got_error *err = NULL;
2389 DIR *packdir = NULL;
2390 struct dirent *dent;
2391 struct got_packidx *packidx = NULL;
2393 char *path_packfile;
2399 *total_packsize = 0;
2401 packdir_fd = openat(got_repo_get_fd(repo),
2402 GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2403 if (packdir_fd == -1) {
2404 return got_error_from_errno_fmt("openat: %s/%s",
2405 got_repo_get_path_git_dir(repo),
2406 GOT_OBJECTS_PACK_DIR);
2409 packdir = fdopendir(packdir_fd);
2410 if (packdir == NULL) {
2411 err = got_error_from_errno("fdopendir");
2415 while ((dent = readdir(packdir)) != NULL) {
2416 if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen,
2420 if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2421 dent->d_name) == -1) {
2422 err = got_error_from_errno("asprintf");
2426 err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2427 path_packidx, 0, repo->algo);
2432 if (fstat(packidx->fd, &sb) == -1)
2434 *total_packsize += sb.st_size;
2436 err = got_packidx_get_packfile_path(&path_packfile,
2437 packidx->path_packidx);
2441 if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2443 free(path_packfile);
2446 free(path_packfile);
2447 *total_packsize += sb.st_size;
2449 *nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2453 got_packidx_close(packidx);
2458 got_packidx_close(packidx);
2459 if (packdir && closedir(packdir) != 0 && err == NULL)
2460 err = got_error_from_errno("closedir");
2464 *total_packsize = 0;
2469 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2470 got_packidx_bloom_filter_cmp);